Failure Report- Plains Pipeline, LP 5/19/15

Page 1Official PDFFailure Investigation Report Plains Pipeline, LP, Line 901 Crude Oil Release, May 19, 2015 Santa Barbara County, California May 2016#

Page 2Plains Pipeline, LP - Failure Investigation Report Santa Barbara County, California Crude Oil Release - May 19, 2015 Table of Contents Executive Summary .......................................................................................................................... 3 Final Report Methodology ................................................................................................................ 4 Facility Background ......................................................................................................................... 4 Events Immediately Prior to and During the Crude Oil Release ...................................................... 6 Plains’ Field Response and National Response Center Notifications .............................................. 7 PHMSA’s Corrective Action Order ................................................................................................. 9 Pipeline Alignment ........................................................................................................................... 9 Las Flores Station to Gaviota Station Line 901 Elevation Description .......................................... 9 Gaviota to Pentland Station Line 903 Elevation Description ....................................................... 10 Post-Incident Investigation Results ................................................................................................ 11 Metallurgical Evaluation of Failed Pipe ....................................................................................... 11 In-Line Inspection Survey Review ............................................................................................... 12 Number of Anomalies ............................................................................................................ 13 Cathodic Protection Findings ........................................................................................................ 13 Spill Volume Estimate from Plains’ Third-Party Consultant ....................................................... 13 Investigation Findings and Conclusions ......................................................................................... 14 Proximate or Direct Cause ............................................................................................................ 14 Contributory Causes ...................................................................................................................... 14 PHMSA Post-Incident Action Chronology .................................................................................... 18 Appendices ..................................................................................................................................... 20 ii#

Page 3Plains Pipeline, LP - Failure Investigation Report Santa Barbara County, California Crude Oil Release - May 19, 2015 Executive Summary At approximately 10:55 a.m. Pacific Daylight Time (PDT) on May 19, 2015, the Plains Pipeline, LP (Plains), Line 901 pipeline in Santa Barbara County, CA, ruptured, resulting in the release of approximately 2,934 barrels (bbl) of heavy crude oil.i An estimated 500 bbl of crude oil entered the Pacific Ocean. Line 901 is a 24-inch diameter buried, insulated pipeline which extends approximately 10.7 miles in length and transports heated crude oil from Exxon Mobil’s storage tanks in Las Flores Canyon westward to Plains’ Gaviota Pumping Station. On May 21, 2015, the Pipeline and Hazardous Materials Safety Administration (PHMSA), a regulatory agency within the U.S. Department of Transportation, issued a Corrective Action Order (CAO) that required the operator to shut down Line 901. Concurrent with the issuance and implementation of the CAO, PHMSA conducted an investigation to identify causal factors that contributed to the occurrence and size of the crude oil release. As the failure investigation progressed, the CAO was amended to address additional safety concerns that were identified. On June 18, 2015, Line 901 was purged and filled with inert nitrogen to enhance safety during the investigation and development of a remedial action plan.ii No fatalities or injuries occurred as a result of this rupture and release. The spill resulted in substantial damage to natural habitats and wildlife. PHMSA’s findings indicate that the proximate or direct cause of the Line 901 failure was external corrosion that thinned the pipe wall to a level where it ruptured suddenly and released heavy crude oil. PHMSA’s investigation identified numerous contributory causes of the rupture, including: 1) Ineffective protection against external corrosion of the pipeline  The condition of the pipeline’s coating and insulation system fostered an environment that led to the external corrosion.  The pipeline’s cathodic protection (CP) system was not effective in preventing corrosion from occurring beneath the pipeline’s coating/insulation system. 2) Failure by Plains to detect and mitigate the corrosion  The in-line inspection (ILI) tool and subsequent analysis of ILI data did not characterize the extent and depth of the external corrosion accurately. 3) Lack of timely detection of and response to the rupture  The pipeline supervisory control and data acquisition (SCADA) system did not have safety-related alarms established at values sufficient to alert the control room staff to the release at this location.  Control room staff did not detect the abnormal conditions in regards to the release as they occurred. This resulted in a delayed shutdown of the pipeline.  The pipeline controller restarted the Line 901 pipeline after the release occurred.  The pipeline’s leak detection system lacked instrumentation and associated calculations to monitor line pack (the total volume of liquid present in a pipeline section) along all portions of the pipeline when it was operating or shut down.  Control room staff training lacked formalized and succinct requirements, including emergency shutdown and leak detection system functions such as Page 3 of 21#

Page 4Plains Pipeline, LP - Failure Investigation Report Santa Barbara County, California Crude Oil Release - May 19, 2015 alarms. The consequences of the spill were additionally aggravated by an oil spill response plan that did not identify the culvert near the release site as a spill pathway to the Pacific Ocean. This report contains factual information and analysis regarding the events leading up to the release, information collected during PHMSA’s failure investigation to date, and the technical analysis of that information known at the time of the completion of this report. PHMSA used this information to mandate remedial measures on Line 901, Line 903, and associated stations and tankage. PHMSA will also use the information to determine whether violations of the federal pipeline safety regulations occurred. Final Report Methodology PHMSA conducted relevant interviews, gathered and reviewed numerous historical documents and available records, and performed a thorough review of the Plains Control Room in Midland, TX. An ILI subject matter expert (SME) was hired to review the raw magnetic flux leakage (MFL) data and final vendor reports from the MFL surveys, and evaluated Plains actions as a result of their review of the vendor reports. PHMSA issued a CAO which in part instructed Plains to have the failed pipe examined by a PHMSA-approved metallurgical laboratory and to have a root cause failure analysis (RCFA) performed by a third party independent consultant. The factual evidence reviewed includes: the Plains Integrity Management Plan (IMP), CP records, ILI reports, anomaly dig information, SCADA event and alarm logs, pressure and flow trends, procedures and reports obtained from the pipeline operator and PHMSA SMEs. The arrangement of this report provides a general description of the pipeline system, the events that occurred on the day of the release, and acts or omissions of the operator that led to this failure and release of crude oil. Specific evidence is supplied and pertinent statements from each report are excerpted where appropriate. Facility Background Plains transports crude oil produced in federal and state waters off the coast of Santa Barbara, CA to inland refineries. Plains’ pipeline is composed of two major pipeline sections: (1) Line 901, and (2) Line 903. Lines 901 and 903 were constructed in the late 1980s, hydrostatically tested in 1990, and went into crude oil service in 1992 and 1991, respectively. The pipelines are coated with coal tar urethane and covered with foam insulation which in turn is covered by a tape wrap over the insulation. Shrink wrap sleeves, which provide a barrier between the steel pipeline and soil for corrosion prevention, are present at all of the pipeline joints on Line 901 and multiple locations on Line 903. The pipelines carry high viscosity crude oil at a temperature of approximately 135 degrees Fahrenheit to facilitate transport. Lines 901 and 903 are controlled from the Plains Control Room’s (PCR) California console in Midland, TX. (1) Line 901 is a 24-inch diameter pipeline that extends approximately 10.7 miles in length from the Las Flores Pump Station to the Gaviota Pump Station; and (2) Line 903 is a 30-inch diameter pipeline that extends approximately 128 miles in length from the Gaviota Pump Station to the Emidio Pump Station, with intermediate stations at Sisquoc Mile Post (MP) 38.5 and Pentland (MP 114.57). There is a delivery point into Line 901 from Venoco’s Line 96 located approximately 2 miles downstream of the Las Flores Station. All of Line 901 crude oil throughput enters Line 903. Line 901 was manufactured of low carbon steel by Nippon Steel Page 4 of 21#

Page 5Plains Pipeline, LP - Failure Investigation Report Santa Barbara County, California Crude Oil Release - May 19, 2015 in Japan in 1986. Line 901’s pipe specifications are API 5L, Grade X-65 pipe, 0.344-inch wall thickness, with a high frequency-electric resistance welded (HF-ERW) long seam. The line was hydrotested to 1,686 pounds per square inch gauge (psig) on November 25, 1990. Figure 1. Map of Plains’ Western Division Pipelines. The arrow points to the approximate release site on Line 901. At Sisquoc Station, crude oil can be pumped to one of two locations: a nearby refinery via a 12- inch diameter pipeline operated by Phillips 66, or continue down Line 903 to Pentland Station. There are additional crude oil lines coming in and out of Pentland Station with numerous tanks at that station used to blend different crude oils for delivery further downstream. At Emidio Station crude oil is delivered to above-ground storage tanks for future delivery to Los Angeles refineries in a separate pipeline system. Prior to the May 19, 2015 release, there had been four small releases meeting PHMSA reportable criteria at pump stations on Lines 901 and 903. No releases were reported to PHMSA on the pipelines outside of pump stations prior to 2015. The operator reported maximum operating pressure (MOP) of Line 901 is 1,341 psig. At the time of the spill, Plains All American Pipeline (PAAPL) operated Line 901 and Line 903 under a Federal Energy Regulatory Commission (FERC) certificate of economic regulatory jurisdiction that was issued in 1987. Plains Pipeline, LP, is a subsidiary of PAAPL. Based on the FERC filing, Lines 901 and 903 were classified as interstate pipelines, pursuant to 49 U.S.C. § 60101(7), as facilities used to transport hazardous liquid in interstate or foreign commerce, and as such, were regulated by PHMSA as interstate pipelines. Plains cancelled the FERC certificates for Lines 901 and 903 on February 12, 2016 and April 29, 2016, Page 5 of 21#

Page 6Plains Pipeline, LP - Failure Investigation Report Santa Barbara County, California Crude Oil Release - May 19, 2015 respectively, stating that the transportation service was no longer available in interstate commerce. Line 903 from Gaviota to Sisquoc to Pentland Stations was purged with nitrogen in accordance with Amendment No. 2 to the CAO, and remains shut down between these stations. The Pentland to Emidio segment of Line 903 is active and operating intermittently at low pressures. This section of pipe between Pentland and Emidio is not directly connected to the Gaviota to Pentland segment and is used to transport crude product from breakout tanks in Pentland Station. Events Immediately Prior to and During the Crude Oil Release On the morning of May 19, 2015, Lines 901 and 903 were transporting crude oil with a flow rate setpoint of 1,240 bbl per hour (BPH) leaving the Las Flores Station, and the discharge pressure was approximately 575 psig. Pumps were operating at the Las Flores Station on Line 901 and Sisquoc Station on Line 903. A Plains instrumentation and electrical technician was dispatched that morning to disconnect and remove a motor from a non-operational pump at the Sisquoc Station. While the technician was performing his work, the operational pump (Pump 401) at the Sisquoc Station was shut down unintentionally (i.e., “uncommanded”). When Pump 401 on Line 903 stopped operating, the pressure in Line 901 increased. The pressure rose to a maximum of 696 psig at the Las Flores Station discharge. The controller shut down the pump at Las Flores Station and the pressure remained at 677 psig. Approximately four minutes later, the pump at Las Flores Station was restarted. At approximately 10:55 a.m. PDT, the flow rate at Las Flores Station climbed from zero to 2,042 BPH. Concurrently, the line pressure rose to a high of 721 psig, then dropped to 199 psig, and then slightly increased to approximately 210 psig until the Las Flores pump was shut down a second and final time. Generally, a sudden increase in flow rate accompanied by a decrease in pressure is indicative of a release. PHMSA has determined that Pump 401 going offline in an “uncommanded” manner on the morning of May 19, 2015, was an abnormal event, but that this in itself should not have caused Line 901 to rupture. PHMSA performed a detailed review of the SCADA event and alarm logs, and pressure and flow records. The review indicated that there was information reported by the SCADA system that indicated a release had occurred by approximately 10:58 a.m., and an alarm was generated on low pressure. The alarm was not set at an appropriate value. The alarm also did not have a major priority/severity or safety-related alarm status. The controller did not recognize the information he received as indicative of an abnormal operation. Evidence indicates that the controller was focused on the events at Sisquoc Station (i.e., restarting the Sisquoc pump that had gone down once uncommanded, and a second time on high case temperature along with other duties).iii Due to the Sisquoc Station maintenance activity resulting in an unplanned pump shutdown, the controller anticipated alarms would be activated from the pipeline leak monitoring (PLM) system. According to interviews and a review of the alarm log, the PLM inhibit was requested by the controller to the step-up shift supervisor between 11:15 and 11:22 a.m.iv The step-up shift supervisor then inhibited (shut off) the PLM system alarms.v Also, during this time, the controller started an investigation of the SCADA data in an attempt to understand the operational abnormalities that were occurring. After attempting to restart the Sisquoc pump twice, the controller shut down the pipeline. PHMSA requested the operator review the flow imbalance calculations and provide a time when the PLM system would have generated an alarm if not inhibited, and it was determined that alarms would have been generated Page 6 of 21#

Page 7Plains Pipeline, LP - Failure Investigation Report Santa Barbara County, California Crude Oil Release - May 19, 2015 approximately two minutes before the controller shut down the pipeline.vi Figure 2. Schematic of Plains Pipeline, LP, Line 901 and spill path. Plains’ Field Response and National Response Center Notifications The following is a timeline of Plains and emergency responder activities conducted immediately prior to locating the leak site:vii  At 11:42 a.m. a call reporting a petroleum smell was received at Santa Barbara Fire Department (SBFD) Station 18. Engine 18 left the station to investigate the odor complaint near Refugio State Beach.  At approximately 12:15 p.m., prior to a scheduled tabletop spill drill required by federal regulations 49 C.F.R. §194, the pre-drill meeting was completed and adjourned. A representative from the Santa Barbara Office of Emergency Management (SB-OEM) received a call from the SBFD reporting that there was oil on Refugio Beach. The SB- OEM representative and the Plains representatives left the spill drill and drove separately to Highway 101 at Refugio Beach.  The Santa Barbara Dispatch notified the National Response Center (NRC #1116950) at 12:43 p.m. PDT of an unknown sheen in the ocean at Highway 101 and Refugio Beach.viii  At approximately 12:55 p.m., the two Plains representatives arrived at the south side of Highway 101 where the SBFD personnel were. They noted oil in the ocean but could not determine the source of the oil. One of the Plains representatives told the assembled group that he did not think the oil was coming from Line 901 because the pipeline is located on the other side of Highway 101, and there would be oil flowing across Highway 101 if Line 901 was leaking. Page 7 of 21#

Page 8Plains Pipeline, LP - Failure Investigation Report Santa Barbara County, California Crude Oil Release - May 19, 2015  The Plains representatives drove to the company’s pipeline right-of-way (ROW). At approximately 1:27 p.m., the Plains representatives located the leak site on the Plains ROW. They called the controller to report the leak and to tell the controller to leave Line 901 shut down and to close the Refugio gate valve. The Plains representatives used their cell phones to contact other Plains personnel, the landowner where the leak occurred, Plains’ oil spill response contractors, and others. The Plains representatives noted that crude oil from the release site had entered a culvert that crosses under the Highway 101 and railroad tracks and discharges to Refugio Beach. The Plains representatives, along with Fire Department personnel, attempted to stop the flow of oil into the culvert. However, the culvert was too large to stop the flow with shovels, and sand bags were not readily available, so their immediate efforts were unsuccessful. At approximately 3:00 p.m., additional equipment and personnel arrived, the culvert was dammed and oil was prevented from entering the culvert.  At 2:56 p.m., a representative from Plains called the NRC to report (NRC #1116972) the release of crude oil at 2:56 p.m. PDT. This report indicated that the release was at Latitude: 34° 27' 43" N; and Longitude: 120° 05' 24" W. This NRC report was made 89 minutes after the release site was found by Plains field personnel.ix Figure 3. Spill location relative to Refugio Beach in Santa Barbara County, CA. Photo: John L. Wiley http://flickr.com/jw4pix Federal pipeline safety regulations, (49 C.F.R. § 195.52), require that the NRC be notified at the earliest practicable moment following discovery of a release of a hazardous liquid, including “[a]ny failure that resulted in pollution of any stream, river, lake, reservoir, or other similar body of water that violated applicable water quality stands, caused a discoloration of the surface of the water or adjoining shoreline, or deposited a sludge or emulsion beneath the surface of the water or upon adjoining shorelines.” On January 30, 2013, PHMSA issued an Page 8 of 21#

Page 9Plains Pipeline, LP - Failure Investigation Report Santa Barbara County, California Crude Oil Release - May 19, 2015 Advisory Bulletin clarifying that this was to be interpreted as within one hour of discovery. Plains reported the rupture to the NRC approximately 89 minutes after discovery, thus notifying the NRC 29 minutes late. The estimated costs reported by the operator as of December 23, 2015, were $142,931,884. This figure includes all costs the operator spent as a result of this release through the date reported, including commodity lost, the operator’s property damage and repairs, operator’s emergency response, environmental remediation, and estimated other costs spent including government agency costs and media relations expenses.x PHMSA’s Corrective Action Order On May 21, 2015, PHMSA issued a CAO, CPF No. 5-2015-5011H, to Plains. The CAO required Plains to purge Line 901; review the pipeline’s construction, operating, maintenance, and integrity management history; expedite the review of data from the May 5, 2015, ILI tool run; conduct metallurgical evaluation of the failed pipe; repair any integrity-threatening anomalies identified by the ILI survey; and conduct a root cause failure analysis. The CAO requires Plains to purge Line 901 and to keep Line 901 shut down until PHMSA approves the restart of the pipeline. Plains’ Line 901 was purged and filled with an inert nitrogen gas on June 18, 2015. On June 3, 2015, PHMSA issued Amendment No. 1 to the CAO. The amendment was issued to address preliminary findings from the early stages of PHMSA’s investigation, and the possibility that the conditions on Line 901 also existed on Plains Line 903. The amendment to the CAO required Plains to conduct additional non-destructive testing of ILI anomalies on Lines 901 and 903; review the construction, operating, maintenance, integrity management, and ILI history of Line 903; and reduce the operating pressure of Line 903 to 80% of the highest pressure sustained for a continuous 8-hour period during the month before the May 19 failure. This pressure reduction was intended to enhance safety until all facets of the line’s integrity could be evaluated. On November 12, 2015, PHMSA issued Amendment No. 2 to the CAO. The amendment required Plains to empty and purge Line 903 between Gaviota and Pentland Stations and fill it with an inert gas. Line 903 was purged between Gaviota and Pentland Stations and filled with inert nitrogen. The complex purging operations began in December 2015, and were completed on April 18, 2016. Both Line 901 and the purged sections of Line 903 will remain shut down until all actions required by PHMSA’s CAO and subsequent amendments have been completed. PHMSA may continue to issue additional amendments to the CAO as necessary. Pipeline Alignment Las Flores Station to Gaviota Station Line 901 Elevation Description To fully understand the Line 901 release, it is vital to understand the elevation profile of Line 901 and Line 903 from the Las Flores Canyon to Pentland Station. Line 901 starts at the Las Flores Station at an elevation of approximately 180 feet. There are two large hills downstream of the originating pump station. The first hill has a peak elevation of approximately 740 feet and the second hill has an elevation of approximately 600 feet. The release occurred downstream of the second hill at an elevation of approximately 80 feet. Immediately downstream of the release point, the pipeline rises slightly and then runs relatively level approaching the Gaviota station. This fact is important because as soon as the pump at Las Page 9 of 21#

Page 10Plains Pipeline, LP - Failure Investigation Report Santa Barbara County, California Crude Oil Release - May 19, 2015 Flores Pump Station was turned off the second time, the only crude oil that could be released was the height of oil in the pipeline above the release site and not the amount located between the two aforementioned hills. Gaviota to Pentland Station Line 903 Elevation Description Line 903 receives all of the crude oil delivered by Line 901. The line elevation at Gaviota is approximately 150 feet. The elevation at Sisquoc is approximately 880 feet. Downstream of Sisquoc, Line 903 rises to 2,420 feet and then to a height of approximately 2,750 feet and ultimately to an elevation of close to 3,000 feet before dropping into Pentland Station at an elevation of approximately 690 feet. Line 903 exhibits many of the same construction and operation conditions as Line 901 and was addressed by the amendments to the CAO. Pump 401 at Sisquoc Station has adequate capacity to push the oil up and over the downstream hills and into Pentland Station but only if it has full suction pressure and full flow coming into the pump. Because of the release, the pump could not push the oil over the downstream hills, and so the oil in the pump became hot and the pump shut down to prevent overheating. Page 10 of 21#

Page 11Plains Pipeline, LP - Failure Investigation Report Santa Barbara County, California Crude Oil Release - May 19, 2015 Post-Incident Investigation Results Metallurgical Evaluation of Failed Pipe The failed pipe segment has been analyzed by third-party metallurgical experts, Det Norske Veritas (U.S.A.), Inc.’s (DNV-GL) in Dublin, OH. The failed pipe assessment and testing was witnessed by PHMSA, the California Department of Fish and Wildlife, and the U.S. Department of Justice. Figure 4. The failed pipe and surrounding insulation and coating. Figure 5. Pipe External Surface at the Line 901 failure site after cleaning. Page 11 of 21#

Page 12Plains Pipeline, LP - Failure Investigation Report Santa Barbara County, California Crude Oil Release - May 19, 2015 DNV-GL’s draft report was completed and disseminated to Plains and PHMSA on August 6, 2015. The draft report was reviewed by PHMSA engineers, and a number of comments and clarification requests were made. DNV-GL reviewed the comments and revised the report. The Final Report was issued on September 18, 2015. The Final Report provides a summary of findings, including the following excerpt: “The results of the metallurgical analysis indicate that the leak occurred at an area of external corrosion that ultimately failed in ductile overload under the imposed operating pressure. The morphology of the external corrosion observed on the pipe section is consistent with corrosion under insulation facilitated by wet-dry cycling.”xi In-Line Inspection Survey Review Plains conducted ILI surveys on Line 901 (10.7 miles in length) to assess the integrity of the pipeline in accordance with PHMSA regulations in 2007, 2012, and 2015. According to 49 C.F.R. § 195.452(j)(3), the pipeline is required to be surveyed at intervals commensurate with the pipeline’s risk of integrity threats, but at least every 5 years. Plains changed Line 901 from a 5-year assessment cycle to a 3-year assessment cycle after the 2012 ILI survey. The data collected during these surveys must be fully evaluated within 180 days of the ILI, and an operator must take action upon discovery of any “immediate repair conditions” as defined in 49 C.F.R. § 195.452(h) unless the operator can demonstrate that the 180-day period is impracticable. The most recent ILI survey for Line 901 was completed on May 6, 2015. The 2015 ILI survey data for the first 2 miles of Line 901, as measured from the Las Flores Station, was found to be incomplete and not useable for ILI analysis. For the rest of the ILI survey, the correlation digs, which are used to gauge survey data accuracy in the ILI vendor’s preliminary report, had not been finished at the time of the May 19, 2015 failure. PHMSA’s independent third-party ILI SME also performed an analysis of the data from past ILI surveys of Line 901. Preliminary data from the results of each of the ILI surveys are summarized below and show a growing number of corrosion anomalies on Line 901. Page 12 of 21#

Page 13Plains Pipeline, LP - Failure Investigation Report Santa Barbara County, California Crude Oil Release - May 19, 2015 Number of Anomalies Metal loss June 19, 2007 July 3, 2012 May 6, 2015 Greater than 80% 0 0 2 60-79% 2 5 12 12 54 80 40-59% The May 6, 2015 ILI survey data and subsequent analysis by the ILI vendor predicted external corrosion at the failure site with an area of 5.38 inches by 5.45 inches, and a maximum depth of 47% of the original pipe wall thickness. After the failure, the DNV-GL metallurgical investigators physically measured external corrosion at the failure site to have a maximum depth of 89%.xii The dimensions of the corrosion feature were 12.1 inches axially by 7.4 inches in circumference. The maximum depth, as measured using laser scan data, was 0.318 inches or 89% of the measured wall thickness (0.359 inches). The ILI summary report prepared by PHMSA’s SME also examined the “as-called” (ILI- predicted) versus as-found (field measured) lengths, widths and area for the excavated anomalies on Line 901. The report demonstrates that the lengths and widths of the anomalies were under-called (underestimated) in many cases, however many were also over-called. Plains submitted little documentation concerning their analysis of how the field measured anomalies compared to the ILI vendor analysis. Furthermore, Plains did not provide documentation showing that discrepancies between the originally reported anomaly sizes predicted by the ILI vendor and Plain’s actual field-measured sizing of the corrosion anomalies were subsequently discussed with the ILI vendor, as required by Plains’ IMP.xiii Cathodic Protection Findings According to 49 C.F.R. § 195.563, CP is required under the federal Pipeline Safety Regulations to prevent external corrosion of buried pipelines. Historical CP records for line 901 have been reviewed and reveal protection levels that typically are sufficient to protect non-insulated, coated steel pipe. Line 901 and Line 903, however, are insulated. An increasing frequency and extent of corrosion anomalies were noted on both Lines 901 and 903 in ILI survey results, anomaly excavations, and repairs. PHMSA inspectors noted moisture entrained in the insulation at four excavations performed by Plains on Line 901 after the May 19 spill and prior to the PHMSA-mandated purging of the pipelines. Spill Volume Estimate from Plains’ Third-Party Consultant Plains initially estimated the volume of spilled crude oil to be approximately 2,400 bbl, of which 500 bbl was estimated to have reached the ocean. On August 4, 2015, Plains reported to the Unified Command that the 2,400 bbl release estimate was still accurate. However, after Plains completed the PHMSA-mandated purge, the company’s calculations indicated that up to 3,400 bbl had possibly been released from the pipeline. Plains notified the Unified Command Page 13 of 21#

Page 14Plains Pipeline, LP - Failure Investigation Report Santa Barbara County, California Crude Oil Release - May 19, 2015 that RPS Knowledge Reservoir (RPS), a third-party investigator hired by Plains, was still trying to reconcile the difference. On November 24, 2015, Plains informed PHMSA that RPS had completed their analysis regarding the release volume and produced a report of findings. RPS used the OLGA simulation software tool to model the behavioral dynamics of the pipeline prior to, during, and immediately after the May 19, 2015 leak. The report concluded that the discharge leak volume was 2,934 bbl. The RPS report was dated November 11, 2015. Plains has reported 1,100 bbl of crude oil have been recovered. Investigation Findings and Conclusions Line 901 pipeline ruptured at approximately 56% of the MOP. Although the operational events that occurred on the morning of the release were abnormal, this should not have caused the release if the pipeline’s integrity had been maintained to federal standards. Proximate or Direct Cause PHMSA determined that the proximate or direct cause of the release was progressive external corrosion of the insulated, 24-inch diameter steel pipeline. The corrosion occurred under the pipeline’s coating system, which consisted of a urethane coal tar coating applied directly to the bare pipe, covered by foam thermal insulation with an overlying Polyken tape wrap. Water has been noted in the foam insulation at a number of digs, indicating that the integrity of the coating system had been compromised. The external corrosion was facilitated by the environment’s wet/dry cycling, as determined by the PHMSA-approved, third-party metallurgical laboratory. The release was a single event caused at an area where external corrosion had thinned the pipeline wall. There is no evidence that the pipeline leaked before the rupture. There was a telltale “fish mouth” (a split due to over-pressurization) at the release site indicating the line failed in a single event. PHMSA’s investigation identified numerous contributory causes of the rupture. The contributory causes can be grouped into three categories: 1) ineffective protection against external corrosion of the pipeline; 2) failure by Plains to detect and mitigate the corrosion;, and 3) lack of timely detection of the rupture. Below is a summary of the key contributory causes: Contributory Causes 1) Ineffective protection against external corrosion of the pipeline  Plains’ CP system was ineffective in protecting thermally insulated underground pipeline systems from external corrosion. Industry practices recognize that an impressed current system like the one utilized on Line 901 cannot protect an insulated steel pipeline should the coating (tape wrap over insulation) become compromised. The external coating in the area of the rupture had allowed moisture to enter the insulation adjacent to the steel pipe.xiv Corrosion under insulation (CUI) cannot be prevented on insulated lines where the coating system has been compromised.xv 2) Failure by Plains to detect and mitigate external corrosion  Plains did not identify CUI as a risk-driving threat in their federally-mandated integrity management program (IMP). Page 14 of 21#

Page 15Plains Pipeline, LP - Failure Investigation Report Santa Barbara County, California Crude Oil Release - May 19, 2015  Plains’ did not fully implement their IMP. o Plains did not perform suitable analysis of the field measurements of the excavated corrosion anomalies that occurred after ILI surveys were completed in 2007 and 2012. o The data reported by the ILI vendor were inconsistent (and did not meet the published accuracy of the ILI tools of +/- 10%, 80% of the time for depth) when compared to the results of the field-measured corrosion anomalies. o Plains’ as-found field measurements of corrosion anomalies were inconsistent with the as-called vendor-provided ILI data and analytical reports. ILI surveys conducted in 2007 and 2012 revealed inconsistencies in the character of the anomalies. In both of these cases, Plains did not consult the ILI vendor to help resolve the inconsistency. o Plains failed to follow written procedures directing the IMP group to perform appropriate statistical analysis after the anomaly dig reports were received from the field, and to discuss any inconsistencies with the ILI vendor.xvi  Plains’ Pipeline Integrity group created a unity plot for depth after the 2012 ILI survey and anomaly digs. There is no documentation detailing what was done with the information from the unity plot. o Plains incorrectly added the over-called anomalies in the close-out reports.  The close-out reports should have only reported the anomalies that were within the reported accuracy of the ILI tool. The reported tool accuracy is +/- 10 %, 80 % of the time. Adding the overcalled anomalies outside of the tool accuracy skews the data.  Plains’ Pipeline Integrity group was historically focused on pitting corrosion under “shrink sleeves” at the pipeline girth welds (circumferential welds to join pipe segments). o The release location was within 6 feet of a corrosion anomaly that was exposed and repaired after the 2012 ILI survey. There was evidence of corrosion and degraded coating systems between the 2012 repair site and the 2015 rupture site. o The anomaly that ruptured was called out by the ILI tool at 45% depth in 2012. Plains’ IMP specified adding 10% to all anomalies (55% depth in this case) then “growing them” to predicted failure using an anticipated corrosion growth rate. This analysis would provide a predicted failure time. Plains did not excavate the anomaly that failed. 3) Lack of timely detection of and response to the rupture  The controller did not have information communicated from the SCADA system in such a manner to be successful in detecting abnormal operations. The pipeline SCADA system did not have safety-related alarms on low pressure configured at the Page 15 of 21#

Page 16Plains Pipeline, LP - Failure Investigation Report Santa Barbara County, California Crude Oil Release - May 19, 2015 correct value or priority to alert the control room staff of the rupture. When this alarm was provided to the controller, the discharge pressure at Las Flores was 199 psig but, within a minute, pressure elevated above 210 psig, the alarm status cleared, and the discharge pressure remained above 200 psig (approximately 210-211 psig) until the pipeline was purged. The pipeline was still leaking when the discharge pressure at Las Flores was above 200 psig, and continued to do so without additional alarm indications. When the pipeline was down, isolated but still leaking, the minimum pipeline discharge pressure at Las Flores remained at 210-211 psig. The low discharge pressure alarm setpoint value was not set properly as it should have been above 211 psig. This type of alarm should be identified as a high priority safety related alarm. While the controllers and shift supervisors can access historical trend data or continue to monitor a given pressure or flow, when the pipeline was ultimately shut down at 11:30 a.m., neither the controller nor step-up shift supervisor detected any drop of pressure at the specific failure location that would indicate that oil was being released.  Neither the pipeline controller nor step-up shift supervisor detected the initial abnormal conditions as the release occurred. There was an indication of decreased pressure and increased flow between 10:53 and 10:58 a.m., which is consistent with a pipeline release. This resulted in a delayed shutdown of the pipeline. Adequate alarm setpoint values with correct priorities are essential to controller and shift supervisor recognition of abnormal operations, especially when many pipeline systems are operated from the same console.  The pipeline controller restarted Line 901 after the release occurred.  The pipeline leak detection system lacked instrumentation and associated calculations to monitor line pack. o The function of the PLM system was a simple line balance calculation based on flow meter values without line pack considerations. The PLM relies on comparing “meter in – meter out” calculations over time. This type of leak detection system without the use of safety-related, high-priority, low-pressure alarms does not provide the controller or shift supervisors with adequate information when the pipeline is down. o When the pipeline is not running, even if only due to scheduling and not required maintenance activities, flows will be close to zero and the imbalance calculation will provide little if any value as currently configured. Leak detection on a down pipeline requires a robust system of planned and accurate high-priority alarm types and alarm setpoint values in order for response to occur on critical low pressures. o The leak detection system for Lines 901 and 903 consists of two leak detection segments. Additional instrumentation such as pressure and temperature transmitters located at Refugio Gate and Cuyama valve settings (both transmitter types on each side of the valves) would allow additional information about the operating status of the pipeline to be presented and pack calculations pursued. o Plains utilizes the SimSuite application for other pipelines in the control Page 16 of 21#

Page 17Plains Pipeline, LP - Failure Investigation Report Santa Barbara County, California Crude Oil Release - May 19, 2015 center. This application does allow for pack calculations to be utilized in the leak detection system. According to information obtained during meetings with Plains hydraulic specialists, Lines 901 and 903 were pipeline systems with a low to medium priority defined for future modeling efforts compared to other assets in the Plains operations. The approach utilized by Plains for prioritizing which systems should be modeled first did not appear to take into account all appropriate consequence-based asset impacts (such as culverts providing a pathway to the ocean) associated with these two systems. Existing instrumentation and the need for added instrumentation would factor into this prioritization decision.  Control room staff training lacked formalized and succinct requirements, including emergency shutdown and leak detection system functions such as alarms. o Interviews determined that the step-up shift supervisor and shift supervisor training lacked formalized and succinct requirements, including that for leak detection system functions such as “inhibit” options. The interviews determined that different shift supervisors performed PLM inhibit functions without contacting the console supervisor first as required by procedure. o Step-up and shift supervisor responsibilities include emergency shutdown of any pipeline. However, training does not cover a means by which to accomplish this for all relevant pipelines. A general emergency shutdown provision has not been programed for supervisory use on all systems.  The oil spill response plan required by 49 C.F.R. §194 did not account for a culvert near the release site that traversed the Pacific Coast Highway and Amtrak railroad tracks. This culvert provided a quick flow path between the pipeline ROW and the Pacific Ocean, thereby allowing crude oil to flow easily towards Refugio State Beach and the ocean. The response plan did not have a response strategy that considered the presence of the culverts. Page 17 of 21#

Page 18Plains Pipeline, LP - Failure Investigation Report Santa Barbara County, California Crude Oil Release - May 19, 2015 PHMSA Post-Incident Action Chronology Following the May 19, 2015 Plains Pipeline, LP, Line 901 rupture in Santa Barbara County, CA, PHMSA took the following actions:  On May 19, 2015, PHMSA deployed inspectors to investigate the Plains Pipeline LP Line 901 pipeline failure in Santa Barbara County, CA. PHMSA also provided information updates to the Unified Command (UC), US Coast Guard, the Federal on Scene Coordinator (FOSC), State Fish and Wildlife, and other agencies on site.  On May 21, 2015: o PHMSA issued a Corrective Action Order (CAO), CPF No. 5-2015-5011H, to Plains Pipeline LP ordering it to suspend operations and to specific safety actions to further protect the public, property, and the environment from potential hazards associated with the recent failure. PHMSA staff reviewed the CAO with the operator and briefed the California State Attorney on the CAO and provided an overview of PHMSA’s regulations. o PHMSA sent an inspector to Plains’ control room in Midland, Texas to collect operational data and interview the control room operators on duty at the time of the incident and their supervisors. The inspector gathered any pertinent logs and information, including electronic copies of relevant data from the Supervisory Control and Data Acquisition (SCADA) system. o PHMSA staff worked with the operator to review their plan to expose the pipe and to cold tap it to ensure there was no pressure or crude left in the line at a low spot immediately downstream of the release point. The plan was signed off by the UC at approximately 5 pm PDT.  On May 22, 2015: o PHMSA staff met with representatives from the Assistant U.S. Attorney, DOT Inspector General, EPA Criminal Investigation Division, California Attorney General, and others to brief them on PHMSA’s process for securing and transporting the failed pipe to a metallurgical lab for evaluation. o PHMSA staff remained on the scene as the operator exposed, tapped, removed any remaining product, and excavated the pipeline downstream of the release site.  On May 25, 2015: o PHMSA issued an approval letter for Plains to excavate, remove and secure the failed joint of pipe under the supervision of two DNV metallurgists (third party contractor) but requested that the coating and insulation not be touched until the failed pipe has been removed because the DNV personnel were interested in in gathering available samples there as well. o A PHMSA inspector returned to Midland, TX to interview the controller and the Operations Control Center supervisor and to obtain any handwritten logs created by the controller on the morning of the release.  On May 28, 2015: o A PHMSA investigator was on site when affected pipeline was removed, crated, and transported to secure location for metallurgical evaluation. PHMSA retained a third-party ILI expert to examine the 2012 and 2015 ILI runs. DNV personnel took soil and insulation samples.  On June 3, 2015, PHMSA amended the CAO to address preliminary findings from the early stages of the investigation (Amendment No. 1). The amended CAO mandated Page 18 of 21#

Page 19Plains Pipeline, LP - Failure Investigation Report Santa Barbara County, California Crude Oil Release - May 19, 2015 additional safety requirements on Line 901 and expanded the scope of the CAO to include the 128-mile long Line 903, which is located downstream of Line 901. The amendment reduced the operating pressure of the Lone 903 by 80% of the highest 8 hour continuous pressure between April 19, 2015 and May 19, 2015. On May 30, 2015, Plains voluntarily shutdown Line 903.  On June 18, 2015, PHMSA staff monitored the Line 901 purge to ensure safety during the purging process. Plains completed the purge and injected inert gas in Line 901.  On September 18, 2015, PHMSA received the DNV Final Mechanical and Metallurgical Report. PHMSA staff reviewed the document and provided comments.  On November 12, 2015, PHMSA issued Amendment No. 2 to the CAO, which ordered Plains to purge and shutdown Line 903 from Gaviota to Pentland.  On December 1, 2015, PHMSA staff monitored Plains moving Freeport McMoRan crude oil from their offshore platforms into Line 903 from Gaviota Station to Sisquoc Station. Movement of the Freeport McMoRan oil was completed on December 10, 2015.  On December 4, 2015, PHMSA staff received the DNV Root Cause Failure Analysis Report. PHMSA reviewed and commented on the report.  On December 14, 2015, PHMSA staff monitored the purge process on Line 903 from Gaviota Station to Sisquoc Station. The purge was completed on December 18, 2015 and the line was filled with inert gas.  On February 17, 2016, PHMSA issued a Preliminary Factual Final Report.  On April 2, 2016, PHMSA staff monitored the Line 903 Sisquoc to Pentland portion purge that was completed on April 18, 2016. Line 901 and 903 are shutdown, except for the Pentland to Emidio section of Line 903, which is not connected to 903 any longer. Page 19 of 21#

Page 20Plains Pipeline, LP - Failure Investigation Report Santa Barbara County, California Crude Oil Release - May 19, 2015 APPENDICES A. Investigation Summary Detail B. Supervisory Control and Data Acquisition (SCADA) Log Excerpts C. Pipeline Leak Monitoring Details D. Excerpts and Discussion of Plains Integrity Management Plan (IMP) Requirements E. Corrosion Control and Pipeline Conditions F. Industry Standards and General Requirements for In-Line Inspection G. In-Line Inspection Report H. PHMSA’s Independent Analysis of In-Line Inspection Data I. Maps and Photographs J. National Response Center Report #1 K. National Response Center Report #2 L. Form PHMSA F 7000.1: Accident Report for Hazardous Liquid Pipeline Systems M. Det Norske Veritas (U.S.A.), Inc. (DNV GL): Line 901 Release (5/19/15) Mechanical and Metallurgical Testing N. Det Norske Veritas (U.S.A.), Inc. (DNV GL): Line 901 Release (5/19/15) Technical Root Cause Analysis O. NACE International: Effectiveness of Cathodic Protection on Thermally Insulated Underground Metallic Structures i According to the FRACTURE CONTROL TECHNOLOGY FOR NATURAL GAS PIPELINES CIRCA 2001 (the PRCI report superseding NG-18 Report 208): “The distinction between leak and rupture for the pipeline community is based on the size and configuration of the breach, not how it develops.” Based on these calculations and visual observations, the length of the feature is consistent with a leak, arresting within the corrosion feature, and did not propagate outside of the feature into nominal wall-thickness pipe. According to the instructions for completing PHMSA Accident Form 7000-1, this type of accident would be classified as a rupture since PHMSA defines a “rupture” as a “loss of containment that immediate impairs the operation of the pipeline”. ii The remedial action plan requires: a) investigation and remediation of anomalies on Line 901 (including anomalies requiring repair per 49 C.F.R. § 195.452(h) and similar anomalies); b) analysis of field measurements taken from anomaly investigations; c) re-grade of previous in-line inspection (ILI) data from 2012 and 2015 ILI surveys using an expanded set of interaction criteria; d) additional integrity assessments using a circumferential magnetic flux leakage (MFL-C) ILI tool and integration of MFL-C ILI data with previous ILI survey results; e) investigation and remediation of anomalies that are identified in the MFL-C tool run (if any); f) based on information collected from remedial work plan and root cause analysis report released by Det Norske Veritas (U.S.A.), Inc., improving the integrity management program; and g) integrity studies to reduce spill volumes, including an emergency flow restriction device evaluation and a surge study. Completion of the remedial work plan is required prior to the PHMSA Western Region Director approving a restart plan and return to service for Line 901. iii High case temperature refers to the oil temperature inside the pump cavity. The case holds the pump impeller Page 20 of 21#

Page 21Plains Pipeline, LP - Failure Investigation Report Santa Barbara County, California Crude Oil Release - May 19, 2015 where oil passes through. This was a centrifugal pump that continues spinning whether there is product in the pump or not. When the rupture occurred, there was not enough pressure or flow rate to allow the pump to continue pumping the oil over the hills and into Pentland Station. Therefore, the oil that was in the pump remained in place and as the pump continued to spin, and temperature was reported to the SCADA system. If the pump reaches the high temperature setpoint, the pump shuts itself off to protect itself from burning up. iv The PCR utilizes two shift supervisors to cover the entire set of 22 consoles. The California Console is handled by shift supervisor B. The shift supervisor B position at the time of the failure was filled by a step-up shift supervisor. A step-up shift supervisor is a controller who is currently qualified on a specific console in the PCR and has received some informal training by working on shift with other shift supervisors. Step-up shift supervisors are used to cover the shift supervisor positions when additional personnel are needed due to illness, vacation, training, etc. Plains has indicated that two step-up shift supervisors are not allowed to be on duty at the same time so one shift supervisor is paired with a step-up shift supervisor when additional personnel is needed. v PLM is the SCADA vendor software tool that serves as the leak detection system for PCR. vi See Appendix B. vii SCADA Data/Plains Control Room time is local to the Central Time Zone. A two-hour time difference separates Central Time from Pacific Time, with Central Time falling two hours ahead. The release occurred in the Pacific Time Zone which is two (2) hours earlier. All times in this report have been adjusted to Pacific Time. viii See Appendix J. ix See Appendix K. x See Appendix L. xi See Appendix M. xii PHMSA has access to this data through a view-only web portal. xiii See Appendix G. xiv The inability of an impressed cathodic protection system to protect insulated pipelines was most recently reaffirmed in the National Association of Corrosion Engineers (NACE) Publication 10A392 (2006 Edition) – “Effectiveness of Cathodic Protection (CP) on Thermally Insulated Underground Metallic Structures.” xv See NACE Report at Appendix O, Background section stating that “[o] n most thermally insulated oil and gas transmission pipelines installed prior to 1980 to 1981, a shop mold-formed thermal insulation was placed directly over the bare steel pipe, with an outer jacket applied to moisture-proof the system. At the field joint, preformed insulation half shells were applied over the joint area to fit between the ends of the shop-applied insulation. After the insulation was fitted, a heat shrink sleeve or a tape wrap was applied over the insulation. When the integrity of the outer moisture barrier was compromised, the space, gap, or void between the edges of the preformed half shells and the shop-applied insulation allowed oxygenated water to diffuse to the bare steel beneath. Damage to the outer moisture barrier has also occurred remote from the joint, allowing oxygenated ground water ingress. “Thermally insulated pipelines have experienced relatively aggressive corrosion, with some failures occurring within three years of service, although acceptable industry standards of CP had been applied and maintained shortly after line construction. The most predominant failures have been those occurring at joints; however, moisture has migrated along the pipeline steel surface to create electrochemical corrosion cells remote from the field joint, culminating in extensive replacements of substantial lengths of line. An article titled ‘Corrosion of Underground Insulated Pipelines’ supports this committee's conclusions that sufficient CP current from an external source may not reach the insulated metallic surface in sufficient quantity to establish adequate corrosion control.” xvi See Appendix D. Page 21 of 21#

Page 22Appendix A Investigation Summary Detail#

Page 23Appendix A: Investigation Summary Detail DOT US Department of Transportation PHMSA Pipelines and Hazardous Materials Safety Administration OPS Office of Pipeline Safety Western Region Principal Investigator Regional Accident Coordinator Peter J. Katchmar Peter J. Katchmar Region Director Chris Hoidal Date of Report 5/5/2016 Subject Failure Investigation Report – HL Santa Barbara County CA Crude Oil Release Operator, Location, & Consequences Date of Failure 5/19/2015 Commodity Released Crude Oil City/County & State Refugio State Beach, Santa Barbara County, CA OpID & Operator Name 300 – Plains Pipeline, LP Unit # & Unit Name 33175 - CSFM #1050A SMART Activity # 150537 Milepost / Location MP 4.16 Type of Failure External Corrosion Fatalities 0 Injuries 0 Description of area impacted Ranch land ¼ mile east of the Pacific Ocean, Refugio State Beach and the Pacific Ocean. Oil flowed to a water drainage culvert that ran under California State Highway 101 (Pacific Coast Highway) and the Amtrak Railroad embankment and into the Pacific Ocean. $ 142,931,884 (through December 23, 2015) Property Damage and Cleanup Cost Page 1 of 1#

Page 24Appendix B Supervisory Control and Data Acquisition (SCADA) Log Excerpts#

Page 25Appendix B: Supervisory Control and Data Acquisition (SCADA) Log Excerpts Listed below is a chronology of events, as obtained from the Plains Control Room (PCR) Supervisory Control and Data Acquisition (SCADA)1 logs. The SCADA log records alarms and events that occur per pipeline system for each line operated from the console. Due to the significant volume of entries and information occurring at the time of this release, only those data points relevant to the CA30 system (901 and portions of 903) have been included  At 10:42:06, Pump 401 at the Sisquoc Station shut down uncommanded due to maintenance activities.  At 10:48:44, the Plains controller at the PCR issued a command to shut down Pump 102 at the Las Flores Station as the result of pump problems at Sisquoc.  At 10:48:52, the SCADA system reported that the Pump 102 at Las Flores had successfully shut down.  The discharge pressure at the Las Flores Station immediately prior to shutdown was recorded by the SCADA to have reached ~677 psig at a flow setpoint of ~1220 Barrels per Hour (BPH).  At 10:49, Tech 2 called the controller and notified him that he could restart Pump 401 at Sisquoc Station.  At 10:52:52, the controller issued a command to restart Pump102 at Las Flores PS.  At 10:53:01, the SCADA system reported Pump 102 successfully started.  Between 10:53 and 10:56 the Pressure and Flow Data from the SCADA indicated the discharge pressure at the Las Flores PS reached ~721 psig and the flow rate reached as high as ~2042 barrels per hour (BPH). Pressure and Flow Trends confirm that 10:55 is approximately when the release occurred.  At 10:55:52, the controller commanded the Pump 401 at the Sisquoc Station to start.  At 10:56:52, the SCADA system reported that Pump 401 at Sisquoc Station was running.  At 10:57:59, the SCADA system reported the discharge pressure at the Las Flores Station dropped to 199 psig and the SCADA system reported a low pressure alarm to the controller.  At 10:58:48 the discharge pressure rises to 210 psig. This automatically resets the low pressure alarm.  At 10:58:58 the controller acknowledges the 210 psig discharge pressure notification. 1 SCADA systems are used to remotely control and monitor pipeline operations. Page 1 of 3#

Page 26 At 11:00:00 the SCADA system reported the flow rate was at 1458 BPH – (a soft high state)  At 11:00:05 controller acknowledges the soft high flow rate.  At 11:00:14 the SCADA system reported flow rate at Las Flores was 1254 BPH = Normal State.  At 11:09:20, the SCADA System recorded that Sisquoc Pump 402 had a high case temperature. However, Sisquoc Pump 402 was not running.  At 11:12, Venoco personnel called the controller and notified him they wanted to start a delivery into line 901 through their line 96. Venoco’s line 96 ties into line 901 about 2.83 miles downstream of the Las Flores Station between the two hills.  At 11:14, controller called the I&E Tech at Sisquoc Station to tell him of the high temperature on Pump 402.  At 11:15:14, the SCADA System recorded that Sisquoc Pump 401 shut down on High Temperature.  At 11:15:48, Venoco started their pump to start a delivery into line 901.  At 11:20, Venoco personnel called the Plains controller and told him the pressure in line 901 was too low to run their line 96 pump.  At 11:20:12, Venoco turned off their pump and closed their valve.  At 11:22:58, the SCADA log states “PLM inhibited.” The Pipeline Leak Monitoring System, or PLM, calculates the imbalance between volumetric meters along the pipeline.  At 11:26:43, the controller issued a command to start Pump 401 at Sisquoc PS.  At 11:27:50, the pump start command timed out. Pump 401 did not start.  At 11:28:12 the controller again issued a command to start Pump 401 at Sisquoc PS.  At 11:29:20, the pump start command timed out. Pump 401 did not start.  At 11:29:56, the controller issued a stop command to the Pump 102 at Las Flores PS. [2 minutes after the PLM would have alarmed according to the calculation presented in Appendix C.]  At 11:30:05, the SCADA system reports that Pump 102 at Las Flores PS is stopped. Mainline Valve 102B at Las Flores closes automatically upon Las Flores Pump 102 shutdown. The pressure at Las Flores is recorded by the SCADA to be between 211 and 213 psig.  At 1:27, the PCR was notified of the line 901 release near Refugio Beach, approximately 4.16 miles from the Las Flores PS. The static pressure immediately downstream of the Las Flores PS is recorded by SCADA to be 211 psig.  At 1:27:23, the controller at the PCR issues a command to close the Refugio Creek mainline valve. [This and the following actions were in response to the controller being informed of oil on the ground at MP 4.16.] Page 2 of 3#

Page 27 At 1:28:31, the controller Issues Command to close Valve 108 at Las Flores PS.  At 1:29:34, SCADA reports the mainline valve at Refugio Creek, approximately 2.83 miles downstream of the Las Flores PS and 1.2 miles upstream of the release site, had successfully closed.  At 1:30:34, SCADA reports Las Flores PS Valve 108 successfully closed.  Between 3:47:14 and 3:48:13, the controller issues commands to close valves 208A, 208 C, and 209A at Gaviota Station.  Between 3:49:51 and 3:51:11, SCADA reports successful closure of valves 208A, 208C, and 209A at Gaviota PS.  At 3:57:48, controller issues command to close valve 209B at Gaviota PS. At 4:00:49, SCADA reports successful closure of Valve 209B at Gaviota PS and the pipeline remained down. Page 3 of 3#

Page 28Appendix C Pipeline Leak Monitoring Details#

Page 29Appendix C: Pipeline Leak Monitoring Details Plains submitted documentation showing the parameters of the PLM and extrapolated what would have occurred if the PLM system had not been inhibited. The submitted documentation shows that the PLM would have alarmed at approximately two minutes before the controller issued the command to shut down the pump at Las Flores PS at approximately 11:30am PDT. The graphical representation is shown on this page. Ma view Add Edit Delete Version 2.0.5317.16227 Segment: CA_LSFCA_PNTC From: 4/12/20157:00:00 AM To 5/29/20157:00:00 AM Observation Times71 - Observation Sper: hes Low Base: / -150,00 High Bose: / 150 00 Over/Snort-Thresholds Trigger Chart | Meter Dats | Packing Rate | Tank Date Quality Bar Legend Question Inhibited Excluded Ois Data Graph Legend Low Threshold 800 800 Trigger Enable Over/Short 200 LD System woul have : Las Fores pump shutdoven commaad asued at 1329.56 -800 : LD System inhibited 12:40 13:00 13:20 13:40 14:00 14:20 14:39 Date Analyze Use Historical OS Start Time: 05/18/2015 12:20:33 PM * Span: /00.0221:35 End Time: | SS18/201524209 PH rUse Line Pack Graphical Representation of when the PLM System would have alarmed had it not been "inhibited". Times in the graph and explanation are in Central Time. Pacific Time is two hours earlier. Plains provided the following explanation along with the graphical representation above. It is quoted as all content is there. [A few changes were made for emphasis and readability that do not compromise the integrity of the explanation.] "This is an explanation of how the program calculated the time when the PLM would have alarmed. • There are 5 meters that receive oil into the line 901/903 PLM. The SCADA tags are: • There are 6 meters that deliver oil out of the PLM. The SCADA tags are:#

Page 30     o Each of these meters are running accumulators, like the odometer on your car, that count barrels (BBLs) through them. The BBLs through them in the last hour is the current value of the accumulator minus the value of the accumulator from 1 hour ago. If we designate the specific value of the accumulator by appending the time parenthetically, the volume of oil through the first Las Flores meter in the hour before 5/19/15 13:27 would be:  (5/19/15 13:27) - 5/19/15 12:27) o The one hour over/short is the sum of all the oil through the out meters for the past hour minus the sum through the in meters for the same time period o (5/19/15 13:27) =  5/19/15 13:27) (5/19/15 12:27) +  5/19/15 13:27) (5/19/15 12:27) +  (5/19/15 13:27) (5/19/15 12:27) +  (5/19/15 13:27) (5/19/15 12:27) +  (5/19/15 13:27) (5/19/15 12:27) +  (5/19/15 13:27) (5/19/15 12:27) -  (5/19/15 13:27) 5/19/15 12:27) +  /19/15 13:27) - /19/15 12:27) +  (5/19/15 13:27) 5/19/15 12:27) +  (5/19/15 13:27) (5/19/15 12:27) +  (5/19/15 13:27) (5/19/15 12:27) ) o The estimated value o 5/19/15 13:26) was -585.6 BBLs. The value of (5/19/15 13:27) was -607.5 BBLs. The alarm limit was at -600 BBLs so the alarm would have been issued at 13:27. [This equals 11:27am Pacific Time]#

Page 31Appendix D Excerpts and Discussion of Plains Integrity Management Plan (IMP) Requirements#

Page 32Appendix D: Excerpts and Discussion of Plains Integrity Management Plan (IMP) Requirements Plains submitted a copy of their IMP dated, December 18, 2003. Applicable sections from that IMP are copied below. “Section 6.0 Procedures for Conducting Assessments and Processing Results Rule 49 CFR §195.452 (f)(8) and (f)(4) requirements: (f)(8) - A process for review of integrity assessment results and information analysis by a person qualified to evaluate the results and information. (f)(4) Criteria for remedial actions to address integrity issues raised by the assessment methods and information analysis.” On page 6-4 of the Plains’ IMP, there is a flowchart, “Figure 6-1 Pipeline In-Line Inspection (ILI) Assessment and Repair – Sequencing of Tasks.” Page 1 of 4#

Page 33Plains Marketing, L.P./ Plains Pipeline, LP START Line Pipe Baseline and Re-Assessment (continue for all Company pipelines with HCA segments ) Schedules Division Pipeline Specialist develops list and schedule of Integrity Coordinator oversees ILl tool vendor activities Integrity Management Group review and assists with setup and of preliminary findings: year that will have an In-Line pipelines for the upcoming execution of tool runs. - Determine necessary pressure reduction for Immediate Repair Condiions Inspection (ILI). (LI tool (or tools) is removed from • Prepare Dig Sheets for field found data in Anomaly Integrity Specialist analyzes pig receiver after a successful Immediate Repairs condions. issues revised Dig Integrity Specialst Tracker to validate the lU Integrity Coordinator requests run: 180-day time limit cook results and integrates data Maintenance Supervisor to begins for anomaly repair complete Pipeline categorization of all tool anomaly Asst. Division Manager receives Dig Sheets Sheets of Immediate with other pipeline data Questionnaire indications pressure reduction and repairs are pertormed for Immediate Repair conditions. Ensures Dent Repairs. as soon as possible and documented preliminary findings on Dents ILl Tool Vendor issues 1 LI Tool Vendor issues Division PHMSA Resords that could prevent MFL tool run. F required, EH&S fles "Safety- re-graded Preliminary i Specialist inputs field found Related Condiion Report findings on Immediate ! anomaly data from Pipeline Repair conditiong Insection Recorts into Anomaly Tracker issues schedule for year's ILls Pipeline Integity Manager I preliminary findings on metal loss ILl Tool Vendor issues spreadsheet. Integrity Specialist determines lor crack like anomales that may i required repair methods) inates ILl tool vendor Integrity Specialist be immediate repair conditions. re-grading of raw tool data. Integrity Coordinator interfaces Coordinate Large Division PHMSA Records with Operations to ensure I Report on line pipe deformation ILl Tool Vendor issues Final completes oscrepances betweer LI Project pig cals and actual size of YES Pipeline Inspection Reports Specialist comples al Imetal loss or crack like anomales dents, metal loss or crac with assistance from Div Fle anomalies? Poeline Specialist and Division Engineer Integrity Coordinator interfaces Integnity Management Group with Operations and ILI Tool completes Corrosion Growth reviews Final Ll Report and Division Pipeline Specialist oversees Vendor to finalize schedule exposed pipe and re Immediate Repairs and records Division Pipeline Specialst oversees repairs Analysis Report (CGAR) on on Pipeline Insection Report nair informatior information on Pipeine Inspection Report. and records exposed pipe and repair anomalies (Optional) Asst. Div. Mgr. receives request for metal loss + issues DOT Compliance Report Integrity Management Group Asst. Div. Mgr. receives DOT Compliance Report with → ensures pressure reductions additional repairs and EHBS fles Safety- Specialist validation digs transmitted to validation digs. Results of with additional Immediate repairs, and repairs are performed Related determ nes Pipeline interty Specialist F→ 60-day repairs, and 180-day within fime limits and are Condition required repairs properly documented Report YES FIGURE 6-1 Pipeline In-Line Inspection (LI) Assessment Integnity Specialist revises DOT Compliance Repor ¡ IL Tool Vendor issues -re-graded Final Report ] Integrity Spesialist inates LI too and Repair - Sequencing of Tasks Date of Revision: 10 July 2008 Page 6-4 Integrity Management Plan Page 2 of 4#

Page 34An enlarged portion of Figure 6-1, from the bottom right quadrant is copied below. The two diamond shapes in the flowchart state the same decision point: “Large discrepancy between pig calls and actual size of dents, metal loss or crack like anomalies?” If “yes” the next box in both cases is: “Integrity Specialist initiates ILI tool vendor re-grading of raw tool data.” PHMSA requested all documentation between the Plains IMP Group and their ILI vendor with respect to their line 901 and 903 before March 19, 2015. PHMSA was provided access to three email strings between the vendor and Plains IMP Group. The first email string had to do with discrepancies noted by Plains IMP group for “clustering” on the Pentland to Emidio Page 3 of 4#

Page 35segment on line 903. The second and third email strings discuss an anomaly called out as a 66% wall loss by the vendor which was found to be 95% wall loss when excavated and measured in the field. This anomaly was on line 903 between the Gaviota PS and Sisquoc PS and was excavated after the 2013 ILI survey on that line segment. This event was described as a “close call” by the Plains representative. He asked the vendor what the cause of this under reporting might be. The ILI vendor responded: “The anomaly in the 2008 run had a lower calculated wall loss of 28% (A neighboring anomaly had a wall loss of 32%, which ended up being assigned to the cluster) because the lower resolution DHD sensors capturing the signal as one anomaly with a wide profile, which resulted in a low wall loss calculation. For the 2013 run, although the tool captured a better profile, with two peaks at that same spot, the anomaly sized a bit wider, encompassing part of the neighboring peak (which had the lower amplitude), which resulted in the 66% wall loss. After adjusting the width to only account for the higher peak, the resulting wall loss was 76%.” The vendor also requested additional dig results from this Gaviota to Sisquoc survey via email. Plains apparently sent them additional digs results at a later date via email attachment. This interaction demonstrates that the ILI vendor is able to reanalyze data and did come closer to the actual anomaly depth. Even after re-analyzing the anomaly, the vendor still under- called the anomaly by 19%. This should have led to increased conversation. When provided additional information from the operator, the vendor uses the “new” information to reanalyze the specific anomaly to better provide a more accurate characterization of the anomaly. Also, the vendor analyst requested additional data from the digs that were being performed. Page 4 of 4#

Page 36Appendix E Corrosion Control and Pipeline Conditions#

Page 37Appendix E: Corrosion Control and Pipeline Conditions Corrosion Control All interstate pipelines regulated by PHMSA on which construction was begun after March 31, 1970 are required to be coated and cathodically protected. Cathodic protection (CP) is a process by which bare steel is protected from corrosion by introducing a small electric current from a rectifier through an anode bed into the earth and back to the rectifier through the pipe (the cathode). A pipe will corrode if steel is allowed to leave the pipe at bare spots called “holidays” in the coating. CP forces electricity toward the pipe at holidays which counters the corrosion process. Pipeline Coatings The first line of protection from pipeline corrosion is a good coating. Line 901 was installed with a coal tar urethane coating in intimate contact with the bare steel 24-inch pipe. Approximately 1.5-inches of urethane foam insulation were then sprayed onto the pipe over the coal tar urethane coating. The pipe was then finally wrapped with a polyethylene tape as a moisture barrier and to hold and protect the insulation on the pipe. The girth welds, where each joint of pipe is welded to the next joint, were coated with shrink sleeves which are made of a thermoplastic that shrinks when heat is applied with a torch which then adheres the sleeve tightly to the pipe. CP on Line 901 Operators are required to install and monitor a CP system within a year of constructing a pipeline. This was done for Line 901. Periodic testing and evaluations are required to ensure the CP system is functioning properly. Bimonthly inspections of rectifiers and annual inspections of pipe-to-soil potentials at each test station along the pipeline are required and reports are kept. PHMSA reviewed CP reports for Line 901 with a focus on 2003 to the present. The operator conducted a close-interval-survey (CIS) in December 2008 and again in April 2015 on Line 901. A CIS is an effort where the operator reports an “on” potential and an “off” potential at approximate three-foot intervals. These reports showed that the CP system appeared to be working well and that the pipe-to-soil potentials were within accepted criteria. The CIS in 2008 showed that the polarized potential of the pipeline was generally around a volt (-1,000mV). In 2015, the polarized potential had moved in the more negative direction towards the maximum polarized potential of steel or ~1,200mV. The off readings in 2015 were generally more negative than -1,100mV. There are two explanations for the movement of the polarized potential on Line 901. One would be that the operator turned up the output on the rectifiers that supply the current to the pipe or they installed additional rectifiers. The second would be that the operator removed some of the protected steel from the CP circuit. PHMSA reviewed the rectifier inspections and found that they were not “turned up” during this time period. The rectifiers had generally consistent output. This meant that the only other possibility would be the removal of a significant amount of steel from the protected pipeline system. PHMSA requested that the operator provide documentation of the amount of pipe removed Page 1 of 4#

Page 38from the system between 2008 and 2015. Plains provided a statement to PHMSA indicating that between 2008 and 2015, approximately 2120 feet of 20-inch and 24-inch piping was disconnected from or removed from the cathodically protected pipeline system. CP is Ineffective on Buried Insulated Pipelines After the release, PHMSA personnel visited Plains offices in Houston, TX, to continue the investigation. During this first visit, one of the first questions concerned external corrosion and cathodic protection because this appeared to be the apparent cause of the release. Plains personnel showed PHMSA a Technical Committee Report from the National Association of Corrosion Engineers (NACE International), titled, “Effectiveness of Cathodic Protection (CP) on Thermally Insulated Underground Metallic Structures” - NACE International Publication 10A392 (2006 Edition) – originally prepared in 1992 by NACE Task Group (TG) T-10A-19, a component of Unit Committee T-10A on Cathodic Protection and was reaffirmed with editorial changes in 2006 by Specific Technology Group (STG) 35 on Pipelines, Tanks, and Well Casings. It is published by NACE under the auspices of STG 35.” This report details the reasons that CP is not effective on buried insulated underground structures. In the “Background” section the report states, “Thermally insulated pipelines have experienced relatively aggressive corrosion, with some failures occurring within three years of service, although acceptable industry standards of CP had been applied and maintained shortly after line construction. The most predominant failures have been those occurring at joints; however, moisture has migrated along the pipeline steel surface to create electrochemical corrosion cells remote from the field joint, culminating in extensive replacements of substantial lengths of line.” Ultimately, it appears that moisture migrated along Line 901 to the lowest local elevation point and created an electrochemical corrosion cell approximately six (6) feet from the nearest girth weld. Discussion of Corrosion Under Insulation (CUI) On non-insulated buried pipelines, external corrosion is normally able to be mitigated by Cathodic Protection (CP). Generally, external corrosion cannot occur as long as CP current is getting onto the pipe. CP current creates an oxygen-free environment around the pipe which will stop the electrochemical process of corrosion, barring additional circumstances. Where external corrosion does occur, current is allowed to get off the pipe and migrate into the surrounding soil. When this occurs, the current takes metal ions with it causing the wall loss or external corrosion. There is little to no “corrosion product” that remains at the pipe surface. In a buried insulated line, the coatings and insulation do not allow the metal ions that result from the electrochemical process of corrosion to migrate away from the pipe surface. Thus, the “corrosion product” will remain close to the pipe and it will become dormant when the electrochemical process depletes all of the oxygen in the moisture. This is known as the dry cycle. When fresh “oxygenated” moisture infiltrates the coating and reaches the area of external corrosion on the pipe, the corrosion process reactivates and again continues until the oxygen is depleted. This is known as the wet cycle. This process is described in detail in the attached metallurgical report as Corrosion Under Insulation (CUI) facilitated by wet/dry cycling which was determined to be the actual cause of the wall thinning at the release site. The metallurgical report contained descriptions of the “corrosion product” as being dense and Page 2 of 4#

Page 39tightly adhered to the pipe. The structure of the “corrosion product” was alternating layers of magnetite and goethite; both have magnetic properties. Due to the composition and density, PHMSA requested additional testing to better quantify the parameters of density and magnetic permeability of the “corrosion product”. This was done and the results were presented in the final root cause failure analysis (RCFA) report also attached to this report. The results came back that the density of the “corrosion product” was 25% of steel and the magnetic permeability was 5% that of steel. While 5% magnetic permeability is small, the large volume of the corrosion product compared with that of the remaining pipe wall led, in part to the MFL tool’s inconsistent reporting. This phenomenon is discussed below and in more detail in the ILI SME Report. Magnetic Flux Leakage (MFL) Technology and Under-Calling the Failed Anomaly In simple terms, the MFL tools used are comprised of magnets that apply a magnetic flux into the pipe steel in the longitudinal direction. The amount of magnetic flux put into the pipe is calibrated to saturate the full wall thickness. There are numerous sensors placed circumferentially around the tool and central to the induced flux field so as to measure and record variances in the magnetic flux that remains in the pipe wall. Any volumetric metal loss that the magnetic field encounters will cause the magnetic flux to “leak” from the pipe wall. The amount of this leakage is then recorded by any number of the sensors in its proximity. When this data is processed, the leakage can be measured to infer the depth, length and width of the metal loss in the pipe wall. As discussed above, when external corrosion is allowed to leave the pipe and migrate into the surrounding soil, the anomaly that is left is usually only the remaining steel. Slight corrosion product might be discovered but not to the extent encountered under insulated coated buried pipe. On coated, insulated and buried pipe, the “corrosion product” grows and remains in close proximity to the pipe steel. This is similar to the type of corrosion on vehicles, in which the corrosion under bubbled paint can be easily flaked off. The corrosion-related paint bubbling on vehicles is similar to what occurred on Line 901. There is a pinhole in the paint where oxygenated moisture can get in and allow the corrosion to occur. The remaining paint has enough integrity to keep the moisture in, which allows the corrosion to occur and corrosion product to grow. The corrosion product gets thicker and thicker until the paint fails entirely. This is similar to the mechanism of CUI that occurred on Line 901. The following picture is excerpted from the metallurgical report. Page 3 of 4#

Page 40This picture is excerpted from the final metallurgical report. “Figure 16. Photograph showing a piece of insulation removed from adjacent to the failure location; near 4:30 orientation.” Page 4 of 4#

Page 41Appendix F Industry Standards and General Requirements for In- Line Inspection#

Page 42Appendix F: Industry Standards and General Requirements for In-Line Inspections 49 CFR Part 195.452(b)(6) requires that operators, “Follow recognized industry practices in carrying out this section, unless – (i) This section specifies otherwise; or (ii) The operator demonstrates that an alternative practice is supported by a reliable engineering evaluation and provides an equivalent level of public safety and environmental protection.” The following discusses the three current accepted industry standards for In-Line Inspections (ILI). The American Petroleum Institute (API) developed “API Standard 1163, “In-line Inspection Systems Qualification Standard” in 2005. A portion of the forward states that this document, “…serves as an umbrella document to be used with and complement companion standards. NACE RP 0102 Standard Recommended Practice, In-Line Inspection of Pipelines; and ASNT ILI-PQ In-Line Inspection Personnel Qualification & Certification all have been developed enabling service providers and pipeline operators to provide rigorous processes that will consistently qualify the equipment, people, processes and software utilized in the in-line inspection industry.” Section 1.2 Guiding Principles of API 1163 goes on to state, “Personnel and equipment used to perform in-line inspections and analyze the results shall be qualified according to this Standard and its companions, ASNT In-Line Personnel Qualification and Certification Standard No. ILL-PQ, and NACE Standard Recommended Practice In-Line Inspection of Pipelines RP0102. Combined, these three standards provide requirements and processes for the qualification of inline inspection systems, including the in-line inspection tools, their software, and the personnel to operate the systems and analyze the results. This Standard is an umbrella document covering all aspects of in-line inspection systems, incorporating the requirements of ASNT ILI-PQ and NACE RP 0102 by reference. Section 9 System Results Verification and Section 9.2.4 – Verification Measurements requires in part, “When verification digs are performed, information from the measurements shall be given to the service provider to confirm and continuously refine the data analysis processes. The information to be collected from the verification measurements and given to the service provider shall be agreed upon by both the operator and the service provider and shall include the measurement techniques used and their accuracies. Information to be provided by the service provider to the operator should include the measurement threshold, reporting threshold, and interaction criteria, if any. Appendix D lists types of information that should be provided to the service provider. Any discrepancies between the reported inspection results and verification measurements that are outside of performance specifications shall be documented. The source of the discrepancies should be identified through discussions between the service provider and the operator and through analyses of essential variables, the dig verification process, and data analysis process. Based on the source and extent of the identified and analyzed discrepancies, one of the following courses of action may be taken: a. The inspection data may be reanalyzed taking into account the detailed correlations between anomaly characteristics and the inspection data. b. All or part of the inspection results may be invalidated. c. The performance specification may be revised for all or part of the inspection results.” Generally, the pipeline operator will contract with an ILI vendor to provide an assessment of Page 1 of 2#

Page 43their pipeline. It must be stated that even though MFL ILI devices are known as “Smart Pigs” they only report what they record. It is up to the pipeline operator to establish defined parameters for what they want the ILI vendor to do with the raw data. The operator, by contract, establishes operational parameters, sets interaction criteria, and must work intimately with the ILI vendor to obtain useable information about their pipeline system. After a tool is removed from the pipeline, the vendor converts the raw data into useable, measurable data. They provide a final report to the operator that provides their best analysis of the data obtained from the tool within the operator’s defined parameters. It is then the operator’s responsibility to review the final report and create a dig list and perform the excavations. A vital step in the overall process is feedback to the vendor with respect to the accuracy of their tool calls. Section “8.7 Correlation of ILI Reported Results with Field Measurements from Section 8: Data Analysis in the NACE Standard RP0102 – “In-Line Inspection of Pipelines” is excerpted below: “8.7.1 An important part of “closing the loop” is the feedback of the field inspection results to the ILI service provider. Using this information, the ILI vendor can continuously improve the validity and accuracy of the data analysis.” Page 2 of 2#

Page 44Appendix G In-Line Inspection Report#

Page 45In Line Inspection Review For NPS 24 Pipeline Plains All American Pipeline; Line 901 – Las Flores to Gaviota Last Surveyed May 6, 2015 Final Report March 4, 2016 Contains Confidential Information Provided By Plains All American Pipeline LP 121 Lippincott St Toronto, ON M5S 2P2 Cell: 289.259.6277 Canada email: mlamontagne@pipe-life.com Private and Confidential; Client/Attorney Privileged#

Page 46ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Disclaimer Lamontagne Pipeline Assessment Corporation is not responsible for errors in calculation as a result of third party inaccuracies in information provided by Plains All American Pipelines. The evaluations provided are estimates calculated on a best efforts basis. Oak Ridge National Laboratory and the Pipeline and Hazardous Materials Administration must be aware of the inaccuracies of in-line inspection tool data and their subsequent effect on data interpretation and evaluation and heed any suggested limitations provided in the following document. Lamontagne Pipeline Assessment Corporation is to be held wholly harmless as a result of any inaccuracies, misrepresentations, misinterpretations or anomalies not interpreted at all from the in-line inspection data or other consultant reports used to prepare this report. At no time should the data provided herein be used as reason to ignore, violate, or alter any law, regulation, or published industry standard. In no event shall Lamontagne Pipeline Assessment Corporation be liable for any special, incidental, indirect, or consequential damages whatsoever including, but not limited to damage to any reservoir or pipeline, pipeline failure, blowout, explosion, pollution (whether surface or subsurface), damages for loss of business profits, business interruption, loss of business information, or any other pecuniary loss arising out of the use of, or inability to use, the data provided herein. The information contained in this document is CONFIDENTIAL information intended for the use of the individual or entity named herein. If the reader of this document is not the intended recipient, or the employee or agent responsible to deliver it to the intended recipient, you are hereby notified that any dissemination, distribution or copying of this document is strictly prohibited. Contains Confidential Information Provided By Plains All American Pipeline LP 1 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 47ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Executive Summary An ILI review has been completed on the Plains All American Pipeline, 10.87 mile, 24” OD Line 901 - Las Flores to Gaviota based on the comparison of the June 19, 2007, July 3, 2012 and May 6, 2015 magnetic flux leakage (MFL) and associated deformation inspections. The focus of this report was to examine the veracity of the inspections and to estimate appropriate growth rates within the segment then apply those rates to the metal loss anomalies as delineated in the most recent 2015 MFL inspection. An excavation prioritization for the segment was then investigated. A discussion on the MFL characterization of the failed anomaly is also presented. 2 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 48ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota The in-line inspection results from the 2007, 2012, and 2015 MFL runs were examined. The vast majority of the corrosion is external and distributed throughout the length of the segment. The distribution of the external metal loss, in general terms, can be said to predominate in localized low elevations. Previous to the 2015 inspection, the majority of the internal anomalies were found in the first 3000’. This data was not collected in 2015. All three inspections were completed by Rosen USA with different tool designations and modifications employed for each run, either in hardware or software. # Ext. # Int. # Mill Total Metal Loss # Dents with Inspection Metal Metal Metal Metal in First # Dents Metal Loss or Loss Loss Loss Loss 9450’ on Weld 2007 MFL (≥10%) 386 237 88 711 277 0 0 2012 MFL (≥10%) 1578 6 2 1586 469 22 2 (repaired) 2015 MFL* (≥10%) 1747 0 21 1768 N/A 6 1 (repaired) *First 9450’ of 2015 data did not record metal loss There is a trend indicating an increase in the number of metal loss anomalies greater than 10% depth. The 2007 inspection had an ID/OD discrimination fault defining many external anomalies as internal. This discrimination error would not compromise excavation prioritization. An anomaly matching analysis was conducted between the 2007, 2012, and 2015 MFL inspections by aligning each of the runs by distance and orientation. The following table describes the number of metal loss anomalies that were aligned (considered the same anomaly) between particular inspections. The “percent possible” noted represents the percentage aligned of the maximum possible. It is intuitive that the greater the number of matches, the more informed is the determination of growth. Total # Anomaly ILI Runs # of Matches for # of Matches for # of Matches for Matches (% of Compared External Metal Loss Internal Metal Loss Mill Metal Loss possible matches) 1 2 2007-2012 488 491 (70%) 2007-2015* 306 0 12 318 (73%) 2012-2015* 802 0 18 820 (73%) *Consideration given to missing data area Corrosion growth rates were investigated by analyzing the growth of matched metal loss anomalies between the 2007, 2012, and 2015 MFL inspections. The best statistical fit came from the 2007 to 2015 comparison. A growth rate could only be established for external corrosion as no internal anomalies were delineated in the 2015 inspection and very few in 2012 as well. The corrosion rate for the external anomalies was calculated as the 99th percentile with a 95% confidence interval and was determined to be 0.0166 in/yr. 3 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 49ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota By applying this estimated growth rate to the anomalies delineated in the first 9450’ of the 2012 inspection and the 2015 data, a suggested excavation timeline based on a 50% depth limit and 139% MOP pressure limit was investigated. This is a conservative approach but is considered necessary as a result of the errant depth reported by the ILI at the failed defect. Based on the above limits, and taking into account excavations already completed, the following excavation timeline for metal loss was delineated; Anomalies 'Failing' Number of Dig Date 50% Depth Criteria Excavations Jan-15 3 3 May-15 18 17 Nov-15 7 5 Jan-16 3 2 May-16 12 6 Jul-16 2 1 Nov-16 8 5 Jan-17 5 2 May-17 11 5 Jul-17 4 2 Nov-17 7 3 Jan-18 10 6 May-18 20 10 Jul-18 20 8 Nov-18 34 9 Based on the ILI sizing and these growth estimates, all of the anomalies will fail by the 50% depth criterion prior to being concerned with the burst pressure approaching 139% MOP. The locations with the 2015 excavation timeline are, GW Dig Start Dig End Length Dig Date GW Dig Start Dig End Length Dig Date 260 1370 4608.59 4610.89 2.3 Jan-15 1570 5382.03 5382.84 0.81 Jan-15 4150/4160/ 4160.01/4160.02 May-15 4210/4220 15049.79 15076.28 26.49 Nov-15 4220/4230 636.44 14945.13 15086.08 22033.77 23006.72 29453.57 29742.02 31555.39 636.47 14968.53 15106.27 22035.46 23032.43 29471.01 31558.11 0.03 23.4 20.19 25.71 17.44 2.72 Jan-15 May-15 May-15 May-15 Nov-15 May-15 9280/9290 33469.53 33482.42 12.89 May-15 9420 33999 34026.28 27.28 May-15 12420/12430 44745.1 44774.5 29.4 Nov-15 6100/6110 1.69 May-15 6350/6360 7990 29171.19 29171.21 0.02 May-15 8060 9270/9280 11060 12410/12420 12820/12830 12880 33431.52 39810.31 44708.24 46183.26 46415.43 47412.91 33460.75 44725.74 46208.59 46424.28 47413.4 48882.36 29.23 17.5 25.33 8.85 0.49 May-15 Nov-15 May-15 May-15 May-15 May-15 13200/13210 47373.47 47402.54 29.07 May-15 13210 13700 48882.16 0.2 May-15 8140 8280/8290 30307.15 30333.12 25.97 May-15 8640/8650 The growth rates, excavations required and re-inspection frequency should be re-examined after every future in line inspection. 4 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 50ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota The depth sizing accuracy stated by Rosen is ±10% with 80% certainty for pitting and general corrosion. With respect to depth measured during excavations, the 2015 inspection was within ±10%, 57% of the time, the 2012 inspection was within ±10%, 58% of the time, and the 2007 inspection was within ±10%, 33% of the time. If overcalled anomalies were considered (i.e. ILI depth>10% over actual) then in all years the unities would be ±10%, >70% of the time. Likewise, employing API 1163, the tool performance was not within stated specifications. The length and width dimensions of metal loss anomalies also play a key part in the sentencing of metal loss with respect to the remaining strength. The depth and axial length of metal loss are primary factors in the remaining strength evaluations, whilst the width estimates can affect the estimated depth of an anomaly during grading by the ILI vendor. Parameters that may affect the accuracy of the sizing estimate are the aspect ratio of the corrosion, corrosion geometry, corrosion complexity, defect spacing, tool velocity, and pipe line magnetic permeability amongst others. The length and width sizing specification given by Rosen is ±0.59” for general corrosion and better for pitting. The importance of interacting “boxes” appropriately to form “clusters” of an area as closely approximating the actual corrosion area dimensions cannot be emphasized enough. Plains specifies an interaction rule that is one of the most commonly employed throughout industry. But Plains requires that only metal loss with depths 15% or greater, are to be included for “clustering”. This differs from the usual. Typically all ILI delineated corrosion is interacted to define “clusters”. The vast majority of all excavated anomalies have been undercalled in length and to a lesser extent in width. A recommendation is provided to review and possibly alter the present interaction criteria for both the in-line inspection analysis and the field measurement process. Deformation or dents were examined with consideration to depth, location to welds and their association to corrosion. The 2012 inspection delineated 1 dent on a weld that was subsequently repaired and the 2015 ILI reported 6 dents. In order to expedite the May 2015 deformation report after the rupture, Plains asked for the report with graded metal loss only. As a result, the report did not provide sizing of the dents. Consideration should be given to reviewing this further. For further delineation of possible dents with metal loss, ILI anomaly alignment was also completed between the 2007, 2012, and 2015 MFL and deformation runs. To which, no locations of a dent with metal loss were found. The documented procedure used by Plains entitled “Procedure for the Assessment of In-Line Inspection Results; DOC NO: PAALP-INT-PRC-NJP- 001” was provided as part of the review process. The document outlines the steps Plains personnel are required to take following the receipt of preliminary and final ILI reports. According to this document they comply with the requirements of the Code of Federal Regulations 49 Part 195.452 with respect to addressing MFL detectable anomalies. Besides the complex shape of the corrosion, it is surmised that the tightly adhered magnetically susceptible corrosion product may have had some influence in the MFL sizing of the failed anomaly. This segment should be re-inspected with an ultrasonic wall loss tool. The ultrasonic inspection will provide a measure of the remaining wall thickness and length without being influenced by the corrosion product and less by shape. A circumferential MFL may delineate the corrosion lengths more accurately but there is still the issue of depth determination by that magnetic tool. 5 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 51ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Table of Contents Executive Summary ................................................................................................................................................ 2 Introduction ............................................................................................................................................................. 7 Results and Discussion ............................................................................................................................................ 7 Review of the Inspection Metal Loss Data ............................................................................................................. 7 In-Line Inspection Comparison and Growth Rate Estimation .............................................................................. 11 In-Line Inspection Tool Accuracy ........................................................................................................................ 13 Plains Anomaly Mitigation Strategy ..................................................................................................................... 20 ILI Details for the Failed Anomaly ....................................................................................................................... 22 Future Anomaly Mitigation ................................................................................................................................... 28 Deformation Discussion ........................................................................................................................................ 29 Discussion to Note ................................................................................................................................................ 30 Appendix A – ILI growth variance in depth between all inspections ................................................................... 31 Appendix B – Individual Anomaly Excavation Timeline ..................................................................................... 33 Appendix C – Full Excavation Timeline (30’ limit) ............................................................................................. 89 List of Tables Table 1. Percentage of metal loss by depth under shrink sleeves. ........................................................................ 10 Table 2. In line inspection results.......................................................................................................................... 11 Table 3. Anomaly matches between inspections. ................................................................................................. 11 Table 4. Distribution of metal loss lengths for boxes and clusters. ....................................................................... 17 Table 5. Distribution of metal loss widths for boxes and clusters. ....................................................................... 17 Table 6. Close-Out summary for the 2012 inspection of Las Flores to Gaviota. .................................................. 21 Table 7. ILI reported dimensions of the failed area. ............................................................................................. 22 Table 8. Estimated failure pressure from profile in Figure 18. ............................................................................. 24 Table 9. Magnetic properties of the pipeline steel and corrosion product (EWI). ................................................ 27 Table 10. Suggested excavation timeline. ............................................................................................................ 29 Table 11. Suggested excavation locations. ............................................................................................................ 29 Table 12. Dent summary from previous deformation inspections. ....................................................................... 30 List of Figures Figure 1. Distribution of external metal loss anomalies. ........................................................................................ 8 Figure 2. Distribution of internal metal loss anomalies. ........................................................................................ 8 Figure 3. Distribution of external metal loss anomalies by clock position. ........................................................... 9 Figure 4. Distribution of internal metal loss anomalies by clock position. ............................................................ 9 Figure 5. Distance of metal loss to the nearest girth weld in 2012. ...................................................................... 10 Figure 6. Distribution of growth rates for matching metal loss. .......................................................................... 12 Figure 7. Probability plot of growth rates for a linear rate assumption. ................................................................ 12 Figure 8. Distribution of external growth by ILI depth variance. ........................................................................ 13 Figure 9. Unity plot for the 2015 MFL inspection. ............................................................................................... 13 Figure 10. Unity plot for the 2012 MFL inspection. ............................................................................................. 14 Figure 11. Unity plot for the 2007 MFL inspection. ............................................................................................. 14 Figure 12. Interacted MFL metal loss in failed anomaly. ..................................................................................... 16 Figure 13. ILI estimated length compared to field measured. ............................................................................... 18 Figure 14. ILI estimated width compared to field measured.#

Page 51, passage 2............................................................................... 18 Figure 15. ILI estimated depth compared to field measured. ................................................................................ 19 Figure 16. Metal loss area; ILI vs field measurement. .......................................................................................... 19 Figure 17. C-Scans with boxes of the failed location as detailed by the 2007, 2012 and 2015 inspections ......... 22 Figure 18. The failed area, A) the Rosen 2015 A-Scan, B) the Laser Scan. ......................................................... 23 Figure 19. Depth profile of failed anomaly (DNV)............................................................................................... 24 Figure 20. Tightly adhered corrosion product (Fig. 58 from metallurgical report). ............................................. 26 Figure 21. X-Ray Diffraction (XRD) of the corrosion product indicating layering of Magnetite and Goethite... 27 Figure 22. Amplitude magnetic permeability of the pipeline steel and corrosion product (EWI). ....................... 27 6 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 52ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Introduction Data was provided by Plains All American Pipelines (Plains) for three magnetic flux leakage (MFL) and deformation in-line inspections (ILI) that have been conducted on the 10.87 mile, 24” OD Line 901 - Las Flores to Gaviota segment. This line is reported to transport crude oil at high temperatures (135ᵒF+) and is comprised of 0.344” API5L X-65 HF-ERW and 0.500” API5L X-60 HF-ERW pipe. The inspection runs reviewed were all axially oriented magnetic flux leakage tools by ROSEN USA (Rosen). The inspections were conducted on June 19, 2007, July 3, 2012 and May 6, 2015. The 2007 inspection employed the CDG (corrosion detection and mapping tool) and EGP (Electronic Geometry Pig) in two separate runs. The 2012 and 2015 inspections used tools having both metal loss and geometry capabilities. The 2012 inspection utilized the CXG (corrosion detection and extended geometry) tool and the 2015 inspection was made using the A/XT (Axial extended geometry) tool. The aim of this report is to review the findings of the in-line inspections with the focus on anomalies requiring excavation and further evaluation that may lead to repair. This report will review the caliper and corrosion inspections and recommended excavation evaluation for those anomalies to be examined in short order and based on an estimated growth rate applied to the 2012/2015 inspection to determine future excavation dates. The growth rates will be estimated based on the differences found by comparing the 2007 MFL inspection to the most recent 2012 and 2015 MFL inspections. There are also brief discussions on the Plains mitigation strategy and details surrounding the MFL interpretation of the failed anomaly. Results and Discussion Review of the Inspection Metal Loss Data The service provider, Rosen, has stated within the 2007 and 2012 reports received by Plains that all data was accepted and used for evaluation purposes. The 2015 inspection data from ~ 9450’ to the end of the inspection was accepted. At the time of the release Plains and Rosen were in discussions around scheduling a re-inspection of this segment to capture the initial 9450’. The distribution of the metal loss anomalies is detailed in Figures 1 and 2. The vast majority of the corrosion is external and distributed throughout the length of the segment. The distribution of the external metal loss, in general terms, can be said to predominate in localized low elevations. The internal anomalies are seen primarily in the first 3000’ and are most likely the result of the incline of the pipeline. There was no internal metal loss delineated in the 2015 inspection, which may be due to the data quality or classification, it also did not have any information on the first 9450’ of pipe. 7 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 53ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Figure 1. Distribution of external metal loss anomalies. Figure 2. Distribution of internal metal loss anomalies. A check of the distribution of the corrosion anomalies by clock position in Figures 3 and 4 showed some preference for external metal loss around 4:00 to 8:00 (bottom of pipe) but may be found in all orientations. The internal metal loss in the first 3000’ can be found at any 8 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 54ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota orientation. The remainder of the internal metal loss is shown to be between the 4:00 to 8:00 (bottom of pipe) o’clock orientations. The internal anomalies identified in 2007 in the first 3000’ may be external due to an ID/OD discrimination error. Figure 3. Distribution of external metal loss anomalies by clock position. Figure 4. Distribution of internal metal loss anomalies by clock position. 9 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 55ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Plains had previously identified the shrink sleeve coating applied over the girth welds as a priority corrosion issue. Figure 5 shows the distance of each metal loss >20% depth in relation to the nearest girth weld as identified in the 2012 inspection. 2012 Inspection Data Figure 5. Distance of metal loss to the nearest girth weld in 2012. The percentage of anomalies by depth within 18” of the nearest girth weld is presented in Table 1. This distance was examined in consideration of the 34” length of shrink sleeve employed (1” greater due to coating interface). The depths were found to have greater criticality nearer the girth welds in the 2012 data than in either of the 2007 or 2015 data. The 2007 and 2015 data approximate an even spread of depth whether under a shrink sleeve or not. Table 1. Percentage of metal loss by depth under shrink sleeves. Percentage of Anomalies Within 18" of Girth Weld Anomaly Depth 2007 2012 2015 20% to 39% 53% 36% 35% 40% to 59% 50% 56% 50% >60% 50% 100% 57% 10 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 56ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota In-Line Inspection Comparison and Growth Rate Estimation The in-line inspection results from the 2007, 2012, and 2015 MFL runs were examined. All three inspections were completed by Rosen USA with different tool designations and modifications employed for each run, either in hardware or software. Table 2 details the inspection results. There is a trend indicating an increase in the number of metal loss anomalies greater than 10% depth and therefore active corrosion. Table 2. In line inspection results. # Ext. # Int. # Mill Total Metal Loss # Dents with Inspection Metal Metal Metal Metal in First # Dents Metal Loss or Loss Loss Loss Loss 9450’ on Weld 2007 MFL (≥10%) 386 237 88 711 277 0 0 2012 MFL (≥10%) 1578 6 2 1586 469 22 1 2015 MFL* (≥10%) 1747 0 21 1768 N/A 6 1 (sleeved) *First 9450’ of 2015 data did not record metal loss An anomaly matching analysis was conducted between the 2007, 2012, and 2015 MFL inspections by aligning each of the runs. Table 3 summarizes the number of metal loss anomalies that have been matched between inspections (considered the same anomaly). The “percent possible” noted represents the percentage aligned of the maximum possible. It is intuitive that the greater the number of matches, the more informed is the determination of growth. Table 3. Anomaly matches between inspections. # of Matches for Total # Anomaly ILI Runs # of Matches for # of Matches for External Metal Matches (% of Compared Internal Metal Loss Mill Metal Loss Loss possible matches) 2007-2012 2007-2015* 2012-2015* 488 306 802 1 2 0 12 0 18 491 (70%) 318 (73%) 820 (73%) *Consideration given to missing data area Corrosion growth rates were investigated by analyzing the growth of matched metal loss anomalies between the 2007, 2012, and 2015 MFL inspections. The best statistical fit came from the 2007 to 2015 comparison. A growth rate could only be established for external corrosion as no internal anomalies were delineated in the 2015 inspection and very few in 2012. Figure 6 displays the frequency of growth by percentage from 2007 to 2015. Figure 7 provides a probability plot of the absolute percentage growth. The corrosion rate for the external anomalies was determined to be 0.0166 in/yr by the 99th percentile having a 95% confidence interval. In some instances the growth rate of pitting may be higher than the growth rate of general corrosion. Unfortunately this cannot be delineated as the interaction 11 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 57ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota rules applied to cluster metal loss in the MFL analysis do not appear appropriate as will be discussed later. Figure 6. Distribution of growth rates for matching metal loss. Figure 7. Probability plot of growth rates for a linear rate assumption. Figure 8 details the 99th percentile growth rate with respect the absolute variance in the estimated depths from the 2007 and 2015 inspections. 12 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 58ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Failed anomaly Figure 8. Distribution of external growth by ILI depth variance. Appendix A examines the ILI growth variance in depth between all inspections similar to Figure 8, except with a finer odometer. It clearly illustrates the greater potential for corrosion and corrosion growth within localized low elevations. In-Line Inspection Tool Accuracy The Rosen stated depth sizing accuracy is ±10% with 80 % certainty for pitting and general corrosion. The unity plot in Figure 9 examines the 2015 MFL inspection tool accuracy. The 2015 ILI estimated depths are compared to field measured depths either from the 4 excavations following the failure or the areas recoated after the 2007 and 2012 inspections. The unity plot shows that the 2015 Rosen inspection is within ±10%, 57% of the time. It may be seen that the failure location has an uncharacteristically high deviation from the ILI estimate. Figure 9. Unity plot for the 2015 MFL inspection. Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016 13#

Page 59ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota The unity plot for the 2012 inspection is provided in Figure 10. The 2012 Rosen inspection is within ±10%, 58% of the time with respect to the 2012 excavations (blue) and 2007 excavation recoats (violet). When comparing to the 2015 field excavated results based on the 2012 ILI data, growth may have occurred, causing the comparisons between field and ILI to be undercalled (orange). The 2015 digs were not considered in the above stated accuracy. Figure 10. Unity plot for the 2012 MFL inspection. The unity plot for the 2007 inspection is provided in Figure 11. The 2007 Rosen inspection is within ±10%, 33% of the time with respect to the 2007 excavations. Figure 11. Unity plot for the 2007 MFL inspection. Likewise, employing API 1163, the tool performance was not within stated specifications. If overcalled anomalies were considered (i.e. >10% over actual) then in all years the unities would be ±10%, >70% of the time. 14 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 60ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota The length and width dimensions of metal loss anomalies also play a key part in the sentencing of metal loss with respect to the remaining strength. The depth and axial length of metal loss are primary factors in the remaining strength evaluations, whilst the width estimates can affect the estimated depth of an anomaly during grading by the ILI vendor. Parameters that affect the accuracy of the sizing estimate are the aspect ratio of the corrosion, corrosion geometry, corrosion complexity, defect spacing, tool velocity, and pipe line magnetic permeability amongst others. All ILI vendors employ software that examine the flux leakage characteristics and amplitude then automatically “box” the metal loss anomalies. The automated boxing determines the depth, length and width for each anomaly based on proprietary algorithms developed by each vendor. These algorithms are created for each model and diameter of inspection tool by pulling (i.e. pull test) the instrument through many known metal loss sizes under controlled conditions. From the signal response the algorithms are created or calibrated. Vendors may create algorithms specifically for particular metal loss characteristics such as general metal loss (large area) or pitting (small and isolated). This is done to more accurately size anomalies as the signal strength and characteristics can and do vary. During the process of characterization the vendor’s proprietary software extracts specific signals from the inspection by an automated algorithm, then classifies the “metal loss”, then quantifies the depth, length and width by the algorithm. The proprietary algorithms must take into account the signal dimensions and typically follow the generic relationship 𝑑𝑑𝑑𝑑𝑑𝑑𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 = � 𝑤𝑤𝑤𝑤𝑑𝑑𝑤𝑤ℎ 𝑙𝑙𝑑𝑑𝑙𝑙𝑙𝑙𝑤𝑤ℎ�𝑎𝑎 𝑎𝑎𝑎𝑎𝑑𝑑𝑙𝑙𝑤𝑤𝑤𝑤𝑎𝑎𝑑𝑑𝑑𝑑𝑏𝑏 𝑏𝑏𝑎𝑎𝑏𝑏𝑏𝑏𝑙𝑙𝑏𝑏𝑏𝑏𝑎𝑎𝑙𝑙𝑑𝑑𝑝𝑝 (K. Reber, A. Belanger, Reliability of Flaw Size Calculation based on Magnetic Flux Leakage Inspection of Pipelines, ECNDT 2006 - Tu.3.1.1, pp 1-11) This characterization “boxes” individual metal loss anomalies. Once the metal loss is individually “boxed”, interaction routines are applied to “cluster” individual indications into a more realistic representation of the corrosion area. Clusters can also be grouped; however, Plains did not request that Rosen do any grouping. Generally, the interaction criteria are specified by the operator (Plains) as part of the inspection contract. Internal and external corrosion must be considered at the same time. If they are at a coincident location, they should be considered additive. There are five general categories of interaction criteria to “cluster” and/or “group” the “boxed” anomalies 1) Length and/or width dependence 2) Absolute value 3) Wall thickness dependent 4) Combinations 5) Sector defined The choice of interaction criteria is important as it may need to be varied depending on the characteristics of the metal loss in the segment being inspected. Plains specifies an interaction criteria to be a combination of absolute value for the length component (1”) and wall thickness dependence for the width component (6t). The 1” x 6t interaction rule is one of the most commonly employed throughout industry and is the example given in ASME B31.4. 15 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 61ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota To form a metal loss “cluster” from “boxes”, two or more “boxes” must be within 1” of axial separation or within 6 wall thicknesses circumferentially. An example of this process may be seen in Figure 12 which shows the boxes and clusters delineated at the failed anomaly in the 2015 inspection. Solid yellow boxes are metal loss with depths 10%-20%. Solid green boxes are 20-40% depth and solid blue boxes have depths greater than 40%. As requested by Plains, only 15%, and greater metal loss, are to be included for “clustering”. The dashed boxes represent the metal loss “cluster” formed by employing the above interaction rule to the boxes (>15%). The resulting size of a clustered anomaly is the length and width extent with a depth represented by the deepest metal loss box within the cluster. The clusters formed by Rosen (green and blue dashed) in Figure 12 by the interaction process overlap and do not accurately represent the extent of the actual corrosion area. The two clusters identified overlap due to clustering of metal loss ≥15% depth only, as per Plains. If all “boxes” down to 10% depth were included in the interaction parameter then the cluster would have been represented as per the orange dashed box in Figure 12. Consideration of all metal loss would have defined the actual area much more accurately or alternatively, grouping of clusters could be considered. The importance of interacting “boxes” appropriately to form “clusters” of an area as closely approximating the actual corrosion area cannot be emphasized enough. The importance of the depth and length measurement will be explored in more detail in the discussion to follow. 47% ~7.78”L ~8.38”W 47% 1.10”L 1.22”W 47% 5.38”L 5.45”W 23% 3.0”L 3.44”W Figure 12. Interacted MFL metal loss in failed anomaly. During this review process a variance was seen in the length and width sizing of anomalies between inspections as detailed in Tables 4 and 5. The 2007 inspection delineated generally larger metal loss features in length and width dimensions. The 2012 inspection defined the smallest anomalies. The 2012 inspection greatly undercalled the length and width of the 16 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 62ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota failed defect. All 3 inspections were to have used the 1” x 6t interaction of the boxes although it is unknown if there were changes in the minimum depth requirements. All three inspections were carried out using different MFL tool generations. There have been no details provided as to what changes were made in the proprietary sizing algorithms between tool generations or analysis processes. Table 4. Distribution of metal loss lengths for boxes and clusters. Length 2007 2012 2015 Length 2007 2012 2015 (in) Boxes Boxes Boxes (in) Clusters Clusters Clusters 0-0.5 16 379 166 0-5 263 83 121 0.5-1 151 683 907 5-10 90 0 1 1-1.5 59 238 304 10-15 20 0 0 1.5-2 21 154 206 15-20 4 0 0 2-2.5 13 49 63 20-25 3 0 0 2.5-3 9 0 0 25-30 3 0 0 3-3.5 20 0 0 30-35 2 0 0 3.5-4 19 0 0 35-40 1 0 0 4-4.5 10 0 0 40-45 1 0 0 4.5-5 1 0 0 5-5.5 3 0 0 5.5-6 1 0 0 6-6.5 1 0 0 Table 5. Distribution of metal loss widths for boxes and clusters. Width 2007 2012 2015 Width 2007 2012 2015 (in) Boxes Boxes Boxes (in) Clusters Clusters Clusters 0-1 192 632 357 0-5 168 58 90 1-2 84 683 976 5-10 117 21 29 2-3 19 113 198 10-15 68 3 2 3-4 7 40 50 15-20 19 1 1 4-5 5 19 29 20-25 8 0 0 5-6 3 9 18 6-7 5 5 8 25-30 3 0 0 7-8 3 1 6 30-35 2 0 0 8-9 4 1 2 35-40 1 0 0 9-10 2 0 0 40-45 1 0 0 10-11 0 0 1 11-12 0 0 1 Now consider the length and width sizing with respect to that measured in excavations. Figures 13 and 14 compare the tool estimates of length and width to measurements taken during a few field excavations and repair. There are only a few as these were all of the length and width measurements from the field that were provided. In both figures the solid line represents the ideal where the estimated tool sizing is equal to the field measurement and the dashed lines represent ± 0.59”, the tool sizing error for length and width specified by the vendor. Figure 13 shows the data to have a couple length estimates within specification but the remainder of length and width estimates were all under estimated. The excavations shown were done after the 2012 inspection and only the locations that were called by all three tool runs are included. The red markers representing the failure lengths and widths will most assuredly not be the same field measurement in 2007 or 2012, this only represents the dimensions as called by the ILI in that year. 17 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 63ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Figure 13. ILI estimated length compared to field measured. Figure 14. ILI estimated width compared to field measured. Figure 15 details the depth comparison of the same anomalies. The 2007 depths are primarily undercalled but this could be a result of the 5 years growth from the time of inspection to excavation. The 2012 and 2015 inspections had 56% and 63% of the anomalies overcalled or within specification, respectively. The red markers representing the failure depths will most assuredly not be 89% in 2007 or 2012, this only represents the depth as called by the ILI in that year. 18 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 64ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Figure 15. ILI estimated depth compared to field measured. The issue of underestimating the length and width of a corrosion anomaly will lead to gross underestimations of the corrosion area. Figure 16 delineates all of the Line 901 anomalies with width and length reported from ILI estimates versus excavations made, on a logarithmic scale. As an example, it is showing that 38% of the anomalies had an area stated by the ILI of ≤ 1.5 in2 when in fact the corrosion areas were between 2.5 in2 and 7300 in2. This being said, there may be a difference in the field measurement technique to consider. It is important that the techniques used in the field be comparable to that required by the ILI analysis to enable a proper assessment of the ILI performance. Figure 16. Metal loss area; ILI vs field measurement. 19 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 65ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Plains Anomaly Mitigation Strategy The documented procedure used by Plains entitled “Procedure for the Assessment of In-Line Inspection Results; DOC NO: PAALP-INT-PRC-NJP- 001” was provided as part of the review process. The document outlines the steps Plains personnel are required to take following the receipt of preliminary and final ILI reports. According to this document they comply to the requirements of Code of Federal Regulations 49 Part 195.452 with respect to addressing MFL detectable anomalies. Part of the Plains document process is a “Close-Out” report that is created following the reception and repair of anomalies related to any ILI Final Report. The most recent “Close- Out” report for the segment in question relates to the 2012 inspection as the 2015 inspection was run only 13 days before the failure. Table 6 is the summary that was provided within the 2012 Las Flores to Gaviota Close-Out report. The table shows that this segment had 382 anomalies addressed. The worst anomalies remaining after repairs, based on the ILI estimated sizing was a 52% deep anomaly and one with an estimated failure pressure of 1608 psi (1.57 factor of safety). It is not clear in the document whether these are one in the same anomaly. It is also unclear as to why an anomaly greater than 50% depth was left as Plains repairs to a minimum of 50% and try to repair to 40% depth (i.e. To attain 50% with the 10% tool tolerance). But the regulations state a ≥50% deep area of general corrosion need be repaired, this does not included pitting. Assuming the remaining >50% deep anomaly was considered to be pitting then Plains by all accounts met 49 CFR 192.452 requirements as per the 2012 ILI information. Note: the close out report for the 2012 inspection has a later date than the May 19, 2015 release. Plains has noted in their response to PHMSA on November 23, 2015 with respect to CPF 5- 2015-5011H Correction Action Order Amendment 2, page 3, that “…Furthermore, Plains’ focus on the depth of anomalies, rather than length and width, is supported by the industry standard API 1160, Annex D, Managing System Integrity for Hazardous Liquid Pipelines, which states on p. 87: “Growth of an anomaly in depth has a much greater deleterious effect on failure pressure than growth in length, so much so that growth in length can be safely ignored.” “ Although this response is with respect to Line 903, it is misleading and incorrect. The comment quoted above from API 1160 is out of context. Having the most accurate length is very important to the calculation of the remaining strength of every type of anomaly. The length must be defined as accurately as possible. The comment quoted from API 1160 above refers only to the known fact that when corrosion is growing, the depth aspect will be much more influential than the length. This occurs because the percentage depth increases much more rapidly due to the thin wall of the pipe. 20 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 66ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Table 6. Close-Out summary for the 2012 inspection of Las Flores to Gaviota. CLOSE-OUT REPORT ILI run date: 7/3/2012 Line Name: L901 - Las Flores - Gaviota - 24" Summary of In-Line Inspection Indications 6 6 0 0 1,247 998 Date: 6/22/2015 Metal Loss Anomalies Ext Int Mfg Total ILI After ILI After ILI After ILI After d/t < 20% WT 1,241 992 20% WT < d/t < 30% WT 182 137 0 0 2 2 184 139 30% WT < d/t < 40% WT 99 57 0 0 0 0 99 57 40% WT < d/t < 50% WT 36 9 0 0 0 0 36 9 50% WT < d/t < 60% WT 15 1 0 0 0 0 15 1 60% WT < d/t < 70% WT 4 0 0 0 0 0 4 0 70% WT < d/t < 80% WT 1 0 0 0 0 0 1 0 d/t > 80% WT 0 0 0 0 0 0 0 0 Internal ML consistent with internal corrosion 0 0 0 0 0 0 0 0 Selective Seam Corrosion 0 0 0 0 0 0 0 0 Total 1,578 1,196 6 6 2 2 1,586 1,204 Failure Pressures and Deepest Pits ILI After Deformation Anomalies Total After Reported deepest external metal loss (%WT) 78% 52% Dent Depth > 6% OD 0 0 Reported deepest internal metal loss (%WT) 18% 18% Dent Depth < 6% OD 22 5 Calculated lowest Safe_ pressure (based on CGAR) 1,090 1,158 Dent Depth > 2% OD with metal loss/crack 0 0 Calculated lowest P_Burst (based on CGAR) 1,515 1,608 Dent Depth < 2% OD with metal loss/crack 0 0 Dent Depth > 2% OD affecting weld 0 0 Seam Weld Anomalies Total After Dent Depth < 2% OD affecting weld 2 0 SWA-A 0 0 Girth weld anomalies 0 0 SWA-B 0 0 Wrinkle bends 16 0 SWA 0 0 Crack Anomalies (Depth) Crack-Like Crack Field Notch-Like Mid Wall (Lamination/ inclusion) Total ILI After ILI After ILI After ILI After ILI After 0.040" - 0.079" 0 0 0 0 0 0 0 0 0 0 0.08" - 0.119" 0 0 0 0 0 0 0 0 0 0 0.12" - 0.159" 0 0 0 0 0 0 0 0 0 0 > 0.16" 0 0 0 0 0 0 0 0 0 0 No depth 0 0 0 0 0 0 0 0 0 0 Total 0 0 0 0 0 0 0 0 0 0 Results/Comment/Recom mendation: 1. 2012 ILI - 49 anomalies repaired using Type B, 37 anomalies using composite sleeves, 211 anomalies using recoat, and 0 anomaly using pipe replacement. 2. The result shows that the ILI tool is within the tool's tolerance specification. No further anomalies need to be investigated. 3. The result shows that 73 % of the excavated anomalies were within tool tolerance or overcalled by the ILI tool and no anomalies meet conditions for further evaluations. 4. The earliest the remaining ML anomalies to have predicted depth >80%WT or calculated burst pressure < MOP (based on CGAR) is 3/19/2016. 21 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 67ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota ILI Details for the Failed Anomaly All three in-line inspections sized the eventual failure site. Table 7 details the failed anomaly as it was reported by the various inspections. Table 7. ILI reported dimensions of the failed area. Distance Length width (ft) (in) (in) Depth (%) Clock Comments 2007 21355.45 3 6.5 19 4:01 Ext Cluster 2012 21384.96 1.2 1.4 45 4:23 Ext metal loss 2015 21384.58 5.38 5.45 47 3:57 External cluster The anomaly as detailed by the C-Scan (color scan) for each inspection is given in Figure 17. 2007: 19%, 3”L, 6.5”W 2012: 45%, 1.2”L, 1.4”W 2015: 47%, 5.38”L, 5.45”W Figure 17. C-Scans with boxes of the failed location as detailed by the 2007, 2012 and 2015 inspections 22 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 68ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Figure 18 provides the in-line inspection A-scan in comparison to the in-the-lab laser scan as described in the “Draft Mechanical and Metallurgical Testing Report, Report OAPUS309DNOR (PP136049), DNV, August 6, 2015”. 47% 1.10”L 1.22”W 23% 3.0”L 3.44”W 47% 5.38”L 5.45”W A B Figure 18. The failed area, A) the Rosen 2015 A-Scan, B) the Laser Scan. 23 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 69ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Noted in Table 7 and Figure 17, the last known depth prior to failure was 45% in 2012. The verification of the depth estimate at that time is not available. It is highly likely that there was continued corrosion growth of the anomaly from that time to failure. The metallurgical report by DNV provided the following information for the depth profile of the failed anomaly in Figure 19. The remaining strength of the anomaly was determined using industry accepted and publicly available software (KAPA, Kiefner and Associates) and is provided in Table 8. Figure 19. Depth profile of failed anomaly (DNV). Table 8. Estimated failure pressure from profile in Figure 18. Predicted Failure Factor of Safety Effective Length Pressure (Pf, psi) (Pf/MOP) (in) Effective Area Method 684 0.67 7.94 Modified B31G 665 0.65 The estimated operating pressure at the time and location of failure was 737 psi, the estimated failure pressure as determined by the profile is 685 psi with an effective length of 7.94”. This effective length illustrates the difference between the ILI determined cluster length (5.38”) and actual. Albeit there was a depth prediction error, an appropriate representation of the corrosion area should be determined through appropriate interaction rules. 24 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 70ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Rosen stated in their final report (RoCombo Inspection Service, Line 901 Las Flores to Gaviota, May 2015, Project # 0-1000-12834, Rosen Group, June 4, 2015) presented to Plains, “The data recorded during the RoCombo MFL-A/XT inspection survey, performed on May 6. 2015, was accepted and used for evaluation purposes. During the RoCombo MFL-A/XT inspection survey, there was an area of incomplete data due to odometer slippage. The area starts at ROSEN log distance 111.52 ft and continues to 9412.95 ft totaling 9301.43 ft. The resulting data recorded is 83.79% of the total line length. The survey was correlated to the ROSEN 2012 inspection survey to aid in the evaluation process. During the survey, all sensors were operational in areas outside of the odometer slippage. An additional inspection for this line segment will be performed for coverage in the areas of odometer slippage. The tool velocity during RoCombo MFL-A/XT inspection survey was within the pre-agreed range. Generally, in all areas where the velocity is outside of the optimum range, the ROSEN standard accuracy might not be achieved. Over the complete line length of the RoCombo MFL-A/XT inspection survey, the magnetization level was within the pre-agreed specification of 10 - 30 kA/m. Generally, in all areas where the magnetization level is outside of the optimum range, the ROSEN standard accuracy might not be achieved.” Further, with respect to the tool velocity, “The RoCombo MFL-A/XT tool used during this survey was programmed to operate within a velocity range of 0.33 feet per second to 16.41 feet per second.” The velocity of the 2015 tool in the failed joint was reported to be 0.7 ft/s, which is within the accepted velocity range. Further, with respect to the magnetization level, “The magnetization level achieved during the RoCombo MFL-A/XT survey is typically between 10kA/m and 30kA/m in order to meet the Metal Loss Inspection Performance Specifications.” The magnetization level of the 2015 tool in the failed joint was reported to be approximately 23 kA/m at the failure location, which is within the accepted magnetization range. The reported maximum depth of the failed anomaly in the 2015 inspection was 47% of the wall thickness ±10%. The actual maximum depth was determined by the metallurgical examination to be 89%. The axial and circumferential (length and width) sizing in the 2015 MFL report, though not fully interacting throughout, provides a respectable representation of the actual anomaly. This feature does have complexity in the feature geometry that should be considered as well. The depth variance mentioned above raises some question. Since the inspection tool at the failure location was responding normally and the velocity and magnetization levels were within specification, the tool response is said to be acceptable and within optimal conditions. 25 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 71ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota When the pipe wall is saturated with magnetic flux there is a specific background signal attained by the MFL sensors. When there is a corrosion area that is free from ferromagnetic material, there will be a flux leakage that is attained and is relative to the length, width and depth of the missing metal. With a tightly adhered magnetic corrosion product such as described in the metallurgical report, the level of flux that “leaks” from the metal loss may have been reduced. This may in turn be partially responsible for the undercalling of the depth based on the observed length and width dimensions. The metallurgical report states that the thickness of the said corrosion product adjacent to the failure area was approximately 0.55” thick. The corrosion product as detailed in the metallurgical study, Figures 20 and 21, describe a layered strata of Magnetite (Fe3O4) and Goethite (FeO(OH)). Magnetite (aka. Lodestone) is highly magnetic, whereas Geothite has low magnetic properties but nonetheless is still magnetically susceptible. This magnetic acceptance of the corrosion products provides for the potential retarding of flux leakage. Further study into the actual magnetic properties of the corrosion product has determined that the corrosion product has a slightly increasing magnetic permeability as the magnetic field increases, Figure 22. In the region of the release the magnetization was noted by Rosen to be ~23 kA/m = ~288 Oe. At this level the amplitude permeability of the corrosion product is approximately 5% that of the steel pipe. Intuitively, the greater the permeability the greater the flux density allowed into a material. That being said, the maximum flux densities derived from testing, given in Table 9, show that the corrosion product will carry, at a maximum, ~5% of the flux density of the steel pipe. Consideration must also be given to the volume of the corrosion product with respect to the flux carrying capacity. A greater volume of corrosion product will carry a greater flux density. Therefore, intrinsically, there will be some “masking” of the flux leakage thereby interfering with an accurate determination of the corrosion depth (less flux leakage=shallower depth). To what degree is beyond the scope of this review. The magnetic study was performed by the Edison Welding Institute, EWI Project No. 56251CSP Final Report October 16, 2015. Pipe Corrosion Product Metallurgical Mount Substrate Figure 20. Tightly adhered corrosion product (Fig. 58 from metallurgical report). 26 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 72ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Figure 21. X-Ray Diffraction (XRD) of the corrosion product indicating layering of Magnetite and Goethite (Fig. 62 from metallurgical report). Figure 22. Amplitude magnetic permeability of the pipeline steel and corrosion product (EWI). Table 9. Magnetic properties of the pipeline steel and corrosion product (EWI). Specimen μin μmax Hc Br Hmax Bmax - - Oe G Oe G Longitudinal Steel Pipe 149 1467 6.66 12750 1018.2 22193 Transverse Steel Pipe 177.3 1863 6.50 14147 1011.5 22337 0.5" Corrosion Product-2A AB 1.87 2.545 95.16 280 1008.0 1802 0.5" Corrosion Product-2B CD 2.475 3.598 104.40 426 998.3 1987 0.3" Corrosion Product-3A AB 2.229 2.89 86.27 278 1020.6 1680 0.3" Corrosion Product-3B CD 2.511 3.325 92.31 352 1009.9 2050 27 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 73ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota It should be noted as well that each inspection was carried out by a different model of Rosen MFL inspection tool. The June 2007 inspection employed the Corrosion Detection and Mapping (CDG) tool. The July 2012 inspection utilized the Corrosion Detection and eXtended Geometry tool (CXG). The May 2015 inspection utilized the Axial/eXtended Geometry tool (A/XT). It is unclear by their publicly released specifications the exact differences in the technologies. It is also unclear what the differences in the sizing algorithms used through the years and on various tools may be. It is conjectured that there was a change in sizing algorithms as the length and width dimensions for the 2012 inspection were typically smaller than that for the 2007 and 2015 inspections. Rosen states that all reports used the 1” by 6t interaction rule, but as stated earlier this may have changed by the minimum depth required for interaction as specified by Plains. Future Anomaly Mitigation By using the above determined maximum rate of growth (0.0166 in/yr) and applying this rate to both the length and depth of the corrosion anomalies delineated in the 2012 (first 9450’) and 2015 inspection, an anomaly mitigation program can be developed. Two variables were examined with respect to the corrosion growth, the first being the depth and the second being the estimated burst pressure. To examine the effect of growth the rate was applied on a six month interval over a span of ten years. Depth and remaining strength (estimated burst pressure) limits were set to determine when an anomaly should be excavated. The depth criterion was set at 50% and the burst pressure criterion was set at 139% of the MOP of 1025 psi or 1425 psi. To determine the effects of growth on the estimated burst pressures of each anomaly the 0.85 dL technique, otherwise known as the modified B31G equation, was applied to the growing depth and length estimates. To be even more aggressive, anomalies were deemed to require excavation six months prior to their estimated burst pressures becoming less than or equal to 139% MOP or having an estimated depth greater than or equal to 50%. Employing these conservative limits, conservative growth rate and the six month buffer, allows for ILI sizing prediction error. The determination of excavation locations and their suggested date were made considering the 2012 inspection data for the first 9450’ (no 2015 data collected) and the 2015 data for the remainder. The results were combined and the suggested excavation timeline to the end of 2018 is given in Table 10. Table 11 lists the chainage of the recommended locations for 2015. Some of the locations listed in Table 11 and also in Appendix C may be combined into a single excavation. For a listing of all excavations to 2025 refer to Appendix B and C. The growth rates, excavations required and re-inspection frequency should be re-examined after every future in line inspection. 28 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 74ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Table 10. Suggested excavation timeline. Anomalies 'Failing' Number of Dig Date Criteria Excavations Jan-15 3 3 May-15 18 17 Nov-15 7 5 Jan-16 3 2 May-16 12 6 Jul-16 2 1 Nov-16 8 5 Jan-17 5 2 May-17 11 5 Jul-17 4 2 Nov-17 7 3 Jan-18 10 6 May-18 20 10 Jul-18 20 8 Nov-18 34 9 Table 11. Suggested excavation locations. GW Dig Start Dig End Length Dig Date GW Dig Start Dig End Length Dig Date 260 636.44 636.47 0.03 Jan-15 8640/8650 31555.39 31558.11 2.72 May-15 1370 4608.59 4610.89 2.3 Jan-15 9270/9280 33431.52 33460.75 29.23 May-15 1570 5382.03 5382.84 0.81 Jan-15 9280/9290 33469.53 33482.42 12.89 May-15 4150/4160/ 4160.01/4160.02 9420 33999 34026.28 27.28 May-15 14945.13 14968.53 23.4 May-15 11060 39810.31 Nov-15 4210/4220 15049.79 15076.28 26.49 Nov-15 12410/12420 44708.24 44725.74 17.5 May-15 4220/4230 15086.08 15106.27 20.19 May-15 12420/12430 44745.1 44774.5 29.4 Nov-15 6100/6110 22033.77 22035.46 1.69 May-15 12820/12830 46183.26 46208.59 25.33 May-15 6350/6360 23006.72 23032.43 25.71 May-15 12880 46415.43 46424.28 8.85 May-15 7990 29171.19 29171.21 0.02 May-15 13200/13210 47373.47 47402.54 29.07 May-15 8060 29453.57 29471.01 17.44 May-15 13210 47412.91 47413.4 0.49 May-15 8140 29742.02 Nov-15 13700 48882.16 48882.36 0.2 May-15 8280/8290 30307.15 30333.12 25.97 May-15 Deformation Discussion Deformation or dents were examined with consideration to depth, location to welds and their association to corrosion and/or cracking. Table 12 summarizes the details of the three previous deformation inspections. For further delineation of possible dents with metal loss, ILI anomaly alignment was also completed between the 2007, 2012, and 2015 MFL and deformation runs. To which, no locations of a dent with metal loss were found. In order to expedite the May 2015 deformation report after the rupture, Plains asked for the report with 29 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 75ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota metal loss only. As a result, the report did not provide dent sizing. Consideration should be given to reviewing this further. Table 12. Dent summary from previous deformation inspections. # Geometric # Dents on or # Dents with # Dents on Inspection # Dents Magnetic adjacent to Long Metal Loss Girth Welds Anomalies Seam 2007 Def (≥2%) 0 0 0 0 0 2012 Def (≥1.0%) 1 22 0 2 (repaired) 0 2015 Def (≥n/a) 6 0 0 1 (repaired) 0 Discussion to Note The following points should be considered: 1. This segment should be re-inspected with an ultrasonic wall loss tool. The ultrasonic inspection will provide a measure of the remaining wall thickness without being influenced by the corrosion product. A circumferential MFL may delineate the corrosion lengths more accurately but there is still the issue of depth determination. 2. Interaction rules should be reviewed and changed to provide for adequate sizing of the corrosion anomalies. 3. The field measurements should be comparable to the ILI interaction rules (i.e. the extent of the anomaly is to a depth of 10% with no other metal loss within the specified interaction distance). 4. The dent review found inadequate information from the 2015 inspection. Other details to consider with respect to this report: 1. Correlations were not made with respect to the agreement in location concerning tees and/or pipe supports or other appurtences. 2. The defects during the growth stage of this report are not examined for further interaction. 3. This report considers metal loss as delineated by the MFL and deformation tool; no other threats or areas of possible concern were considered. 30 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 76ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Appendix A – ILI growth variance in depth between all inspections 31 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 77ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota 32 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 78ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Appendix B – Individual Anomaly Excavation Timeline Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 70.01 125.59 X 70.01 125.66 X 70.01 125.79 X 70.01 128.04 X 70.01 128.12 X 70.01 128.13 X 80 142.7 X 80 142.82 X 250 597.18 X 260 607.51 X 260 636.44 X 260 636.47 X 290 668.45 X 290 668.65 X 290 668.69 X 310 678.73 X 310 678.73 X 310 678.96 X 310 678.96 X 420 956.64 X 420 956.7 X 470 1134.02 X 480 1164.9 X 480 1186.34 X 480 1197.72 X 490 1207.96 X 490 1224.65 X 500 1239.97 X 500 1262.12 X 500 1277.24 X 500 1278.03 X 510 1287.34 X 510 1296.79 X 510 1303.42 X 520 1320.27 X 520 1325.96 X 520 1326.71 X 520 1329.16 X 520 1329.9 X 520 1332.09 X 520 1332.19 X 520 1334.55 X 520 1336.38 X 520 1336.89 X 33 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 79ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 520 1340.58 X 520 1342.35 X 520 1342.85 X 520 1343.33 X 520 1346.06 X 520 1348.62 X 520 1350.34 X 520 1352.06 X 520 1352.4 X 520 1352.78 X 530 1360.52 X 530 1365.54 X 530 1366.72 X 530 1367.22 X 530 1369.08 X 530 1369.47 X 530 1371.74 X 530 1372.27 X 530 1372.95 X 530 1373.79 X 530 1373.85 X 530 1375.95 X 530 1380.45 X 530 1381.38 X 530 1383.55 X 530 1385.47 X 530 1386.43 X 530 1387.57 X 530 1388.73 X 530 1391.81 X 530 1397.07 X 530 1398.29 X 530 1398.49 X 540 1400.11 X 540 1402.31 X 540 1405.69 X 540 1412.65 X 540 1413.85 X 540 1414.07 X 540 1416.93 X 540 1417.69 X 540 1418.47 X 540 1418.65 X 540 1419.33 X 540 1419.71 X 540 1420.81 X 540 1424.87 X 540 1425.32 X 540 1427.32 X 34 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 80ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 540 1436.11 X 540 1437.2 X 550 1446.49 X 550 1448.82 X 550 1454.62 X 550 1464.27 X 550 1465.12 X 550 1466.98 X 550 1473.06 X 560 1477.26 X 560 1488.7 X 560 1491.34 X 560 1491.94 X 560 1492.65 X 560 1504.82 X 560 1509.34 X 560 1511.45 X 560 1512.66 X 560 1512.99 X 570 1515.58 X 570 1517.55 X 570 1525.05 X 570 1530.49 X 570 1531.75 X 570 1535.61 X 570 1538.27 X 570 1540.72 X 580 1555.81 X 580 1563.76 X 580 1572.03 X 580 1572.16 X 580 1576.22 X 580 1580.67 X 580 1581.6 X 580 1581.76 X 580 1582.47 X 580 1585.75 X 580 1586.42 X 580 1587.4 X 580 1588.19 X 580 1593.66 X 580 1593.8 X 590 1595.45 X 590 1595.48 X 590 1595.49 X 590 1597.15 X 590 1602.81 X 590 1604.13 X 600 1673.77 X 35 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 81ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 610 1675.46 X 610 1675.53 X 610 1690.52 X 610 1696.94 X 610 1700.71 X 610 1708.07 X 620 1752.17 X 620 1753.14 X 640 1817.13 X 650 1823.57 X 650 1823.67 X 650 1824.49 X 650 1824.58 X 650 1825.85 X 650 1826.27 X 650 1826.74 X 650 1829.55 X 710 2097.51 X 720 2104.99 X 720 2107.52 X 720 2107.6 X 720 2108.82 X 720 2112.04 X 720 2112.8 X 720 2114.02 X 720 2114.85 X 720 2116.16 X 720 2118.72 X 720 2119.01 X 720 2121.36 X 720 2122.67 X 720 2123.61 X 720 2124.95 X 720 2125.85 X 720 2125.86 X 720 2126.87 X 720 2127.56 X 720 2127.59 X 720 2130.47 X 720 2133.54 X 720 2134.62 X 720 2136.99 X 720 2138.56 X 720 2139.86 X 730 2146.91 X 730 2148.13 X 730 2148.65 X 730 2150.9 X 730 2153.36 X 36 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 82ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 730 2153.69 X 730 2153.96 X 730 2157.94 X 730 2160.08 X 730 2161.52 X 730 2162.49 X 730 2170.35 X 730 2175.05 X 730 2175.97 X 730 2177.25 X 730 2179.72 X 730 2179.92 X 730 2180.45 X 740 2182.2 X 740 2184.73 X 740 2193.05 X 740 2194.6 X 740 2194.96 X 740 2200.92 X 740 2209.22 X 740 2211.03 X 740 2219.16 X 740 2220.09 X 740 2220.66 X 740 2220.85 X 740 2220.87 X 750 2221.36 X 750 2221.77 X 750 2221.82 X 750 2222.91 X 750 2223.28 X 750 2227.11 X 750 2227.17 X 750 2228.41 X 750 2229.25 X 750 2233.5 X 750 2233.65 X 750 2233.74 X 750 2233.89 X 750 2234.83 X 750 2235.72 X 750 2237.39 X 750 2238.53 X 750 2238.53 X 750 2239.45 X 750 2240.02 X 750 2241.7 X 750 2244.41 X 750 2247.51 X 37 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 83ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 750 2249.24 X 750 2249.78 X 750 2250.65 X 750 2256.47 X 750 2257.86 X 750 2258.59 X 750 2259.85 X 750 2260.65 X 750 2260.74 X 760 2260.95 X 760 2261.29 X 760 2261.78 X 760 2262.28 X 760 2262.46 X 760 2266.49 X 760 2278.68 X 760 2286.7 X 760 2290.04 X 760 2290.15 X 760 2290.97 X 760 2291.48 X 760 2296.34 X 760 2299.62 X 760 2300.07 X 760 2300.63 X 760 2300.67 X 770 2310.45 X 770 2310.99 X 780 2314.34 X 800 2368.43 X 800 2368.66 X 800 2368.77 X 800 2368.81 X 970 2990.53 X 970 2991.4 X 970 2991.74 X 970 2992.27 X 970 2993.37 X 970 2994.13 X 970 2994.77 X 970 3002.23 X 970 3002.4 X 970 3020.01 X 970 3023.98 X 970 3027.15 X 970 3027.86 X 980 3032.13 X 980 3035.79 X 980 3035.9 X 38 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 84ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 980 3049.07 X 980 3051.26 X 980 3052.62 X 980 3052.71 X 980 3052.82 X 980 3052.94 X 980 3055.48 X 980 3056.12 X 980 3058.1 X 980 3058.21 X 980 3059.16 X 980 3059.32 X 980 3059.38 X 980 3059.51 X 980 3059.57 X 980 3059.81 X 980 3060.04 X 980 3060.15 X 980 3060.2 X 980 3060.42 X 980 3060.44 X 980 3060.54 X 980 3060.64 X 980 3060.97 X 980 3061.27 X 980 3061.34 X 980 3061.48 X 980 3061.72 X 980 3061.74 X 980 3061.83 X 980 3061.86 X 980 3061.9 X 980 3062.02 X 980 3062.14 X 980 3062.27 X 980 3062.36 X 980 3062.51 X 980 3062.52 X 980 3062.87 X 980 3062.95 X 980 3063.14 X 980 3063.42 X 980 3063.65 X 980 3063.74 X 980 3064.76 X 980 3065.41 X 980 3065.82 X 980 3066.31 X 980 3069.28 X 39 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 85ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 990 3074.62 X 1050 3347.6 X 1050 3347.6 X 1050 3347.71 X 1070 3406.47 X 1070 3427.76 X 1070 3427.81 X 1070 3427.82 X 1090 3489.91 X 1350 4530.98 X 1350 4530.99 X 1360 4532.7 X 1370 4608.59 X 1370 4608.6 X 1370 4608.73 X 1370 4610.87 X 1370 4610.88 X 1370 4610.89 X 1480 5026.35 X 1520 5186.17 X 1550 5295.01 X 1550 5295.13 X 1550 5296.73 X 1560 5318.53 X 1560 5318.54 X 1560 5319.35 X 1560 5319.85 X 1570 5382.03 X 1570 5382.15 X 1570 5382.72 X 1570 5382.74 X 1570 5382.84 X 1580 5425.09 X 1590 5427.07 X 1600 5449.97 X 1700 5833.49 X 1700 5833.51 X 1700 5833.55 X 1980 6902.95 X 1990 6927.55 X 1990 6928.82 X 2170 7618.29 X 2170 7618.32 X 2170 7618.79 X 2170 7618.86 X 2170 7618.95 X 2170 7639.79 X 2170 7640.31 X 2170 7640.88 X 40 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 86ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 2170 7641.34 X 2170 7641.56 X 2170 7641.78 X 2170 7641.99 X 2170 7642.1 X 2170 7642.23 X 2170 7642.57 X 2170 7642.69 X 2170 7643.22 X 2170 7643.34 X 2170 7643.37 X 2170 7643.57 X 2170 7643.58 X 2170 7643.61 X 2170 7643.74 X 2170 7644.11 X 2170 7644.12 X 2170 7644.21 X 2170 7644.64 X 2170 7645.18 X 2170 7646.5 X 2170 7647.12 X 2170 7647.26 X 2170 7647.55 X 2170 7647.91 X 2170 7648.22 X 2170 7654.47 X 2170 7654.78 X 2210 7786.93 X 2210 7787.14 X 2370 8428.95 X 2430 8654.45 X 2450 8733.29 X 2450 8733.3 X 2450 8733.33 X 2450 8733.39 X 2500 8898.57 X 2640 9452.13 X 2640 9452.84 X 2640 9453.04 X 2640 9458.19 X 2640 9458.78 X 2640 9459.51 X 2640 9459.88 X 2640 9460.01 X 2830 10119.9 9 X 2940 10515.1 5 X 41 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 87ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 2960 10584.1 2 X 2960 10584.2 X 2960 10584.2 X 2960 10584.7 5 X 2960 10586.5 7 X 2960 10586.9 7 X 2960 10587.1 5 X 2960 10587.3 4 X 2960 10587.5 6 X 2960 10587.7 5 X 2960 10587.9 2 X 2960 10587.9 8 X 2960 10588.2 X 2960 10588.4 7 X 2960 10588.7 5 X 2960 10588.9 X 2960 10589.0 3 X 2960 10589.4 3 X 3010 10755.9 2 X 3010 10756.3 5 X 3090 11060 X 3220 11615.8 5 X 3370 12195.9 1 X 3430 12427.1 6 X 3630 13155.9 1 X 3630 13155.9 3 X 3630 13156.1 8 X 3680 13394.4 4 X 3680 13394.4 4 X 3750 13645.5 6 X 3750 13645.7 5 X 3810 13852.5 X 42 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 88ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 4 3810 13853.7 9 X 3810 13853.9 3 X 3850 14002.0 2 X 4080 14751.6 2 X 4080 14751.8 4 X 4080 14753.1 X 4080 14753.3 5 X 4080 14753.6 5 X 4080 14753.9 1 X 4080 14754.9 8 X 4080 14757.5 7 X 4140 14909.4 3 X 4150 14932.1 9 X 4150 14934.6 9 X 4150 14945.1 3 X 4150 14945.3 3 X 4160 14960.0 9 X 4160. 01 14960.8 7 X 4160. 01 14966.4 1 X 4160. 01 14967.0 6 X 4160. 01 14967.4 9 X 4160. 01 14967.9 2 X 4160. 02 14968.5 3 X 4200 15015.8 2 X 4200 15024.4 7 X 4200 15024.4 8 X 4200 15024.4 9 X 4210 15026.1 6 X 4210 15049.7 9 X 43 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 89ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 4210 15052.4 2 X 4210 15053.3 7 X 4210 15053.8 9 X 4210 15054.9 1 X 4210 15054.9 5 X 4210 15055.1 7 X 4210 15055.6 7 X 4210 15055.7 X 4210 15056.0 4 X 4210 15056.1 5 X 4210 15059.4 9 X 4210 15061.4 4 X 4220 15069.4 7 X 4220 15069.7 4 X 4220 15073.6 1 X 4220 15073.8 5 X 4220 15074.0 2 X 4220 15074.1 8 X 4220 15076.1 1 X 4220 15076.2 8 X 4220 15086.0 8 X 4220 15086.2 2 X 4220 15086.4 7 X 4220 15086.6 1 X 4220 15092.7 8 X 4220 15092.9 7 X 4220 15092.9 9 X 4220 15093.8 9 X 4220 15093.9 7 X 4220 15094.8 6 X 44 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 90ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 4220 15094.9 5 X 4220 15095.0 7 X 4220 15095.2 9 X 4220 15095.6 3 X 4220 15095.9 4 X 4220 15096.3 1 X 4220 15096.8 8 X 4220 15097.0 9 X 4220 15098.3 7 X 4220 15098.8 6 X 4220 15104.5 9 X 4230 15106.2 7 X 4240 15184.5 X 4240 15184.5 X 4240 15184.5 4 X 4250 15186.2 X 4260 15263.7 9 X 4270 15275.4 3 X 4270 15275.4 3 X 4270 15275.4 4 X 4270 15275.4 5 X 4300 15295.6 3 X 4340 15366.0 8 X 4340 15366.0 8 X 4360 15377.4 3 X 4360 15377.5 9 X 4390 15454.7 2 X 4410 15505.1 2 X 4430 15585.1 9 X 4540 16038.8 7 X 4620 16377.7 X 45 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 91ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 1 4640 16458.4 9 X 4650 16460.6 2 X 4660 16532.1 2 X 4660 16532.6 4 X 4660 16538.5 5 X 4660 16538.6 5 X 4660 16538.6 5 X 4660 16538.6 7 X 4680 16618.7 9 X 4690 16620.5 4 X 4690 16620.5 4 X 4690 16620.5 4 X 4690 16620.5 7 X 4690 16620.6 6 X 4730 16780.4 5 X 4730 16780.7 7 X 4900 17473.1 7 X 5100 18203.4 7 X 5100 18203.6 2 X 5120 18213.5 2 X 5120 18213.5 8 X 5180 18453.3 6 X 5400 19323.5 1 X 5400 19324.2 3 X 5620 20203.6 2 X 5660 20363.6 3 X 5660 20363.8 2 X 5680 20442.4 8 X 5840 21048.9 X 46 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 92ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 5870 21150.0 1 X 5930 21367.1 1 X 5930 21367.6 7 X 5930 21368.4 3 X 5930 21368.8 8 X 5930 21369.2 6 X 5930 21369.3 2 X 5930 21369.4 9 X 5930 21369.5 6 X 5930 21369.5 7 X 5930 21369.9 9 X 5930 21370.1 3 X 5930 21370.3 8 X 5930 21370.4 8 X 5930 21371.2 1 X 5930 21382.4 X 5930 21382.4 X 5930 21383.8 7 X 5930 21384.1 7 X 5930 21384.3 8 X 5930 21384.3 8 X 5930 21384.3 9 X 5930 21384.5 4 X 5930 21384.5 8 X 5930 21384.6 3 X 5930 21385.3 9 X 5930 21385.6 9 X 5980 21552.3 3 X 6010 21710.2 8 X 6060 21845.4 3 X 47 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 93ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 6060 21873.5 3 X 6060 21873.5 3 X 6070 21875.2 1 X 6070 21875.2 3 X 6070 21875.2 6 X 6090 21955.2 7 X 6100 22033.7 7 X 6100 22033.7 9 X 6100 22034.0 4 X 6100 22034.0 5 X 6110 22035.4 1 X 6110 22035.4 6 X 6180 22354.2 X 6180 22354.2 X 6270 22682.6 6 X 6270 22682.9 9 X 6310 22812.9 1 X 6350 23006.7 2 X 6350 23007 X 6350 23007.0 5 X 6350 23007.4 9 X 6350 23007.7 1 X 6350 23008.5 4 X 6360 23018.5 9 X 6360 23018.9 5 X 6360 23019.2 3 X 6360 23019.6 X 6360 23020.8 3 X 6360 23021.3 1 X 6360 23021.7 9 X 6360 23022.8 5 X 48 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 94ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 6360 23022.8 9 X 6360 23023 X 6360 23023.3 X 6360 23023.7 X 6360 23023.8 X 6360 23024.3 9 X 6360 23024.8 2 X 6360 23024.8 8 X 6360 23028.3 8 X 6360 23028.6 6 X 6360 23028.8 6 X 6360 23029.7 6 X 6360 23030.1 4 X 6360 23030.4 1 X 6360 23031.7 6 X 6360 23032.4 3 X 6360 23040.2 5 X 6360 23040.5 7 X 6370 23053.1 4 X 6370 23060.5 5 X 6370 23063.4 8 X 6370 23074.6 X 6400 23198.6 7 X 6520 23639.8 4 X 6520 23667.0 4 X 6520 23667.3 1 X 6520 23667.5 X 6520 23668.1 7 X 6520 23668.7 X 6520 23669 X 6520 23669.0 9 X 6580 23867.6 X 6590 23945.6 8 X 49 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 95ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 6600 23947.6 2 X 6790 24696.3 2 X 6990 25487.1 8 X 6990 25525.0 5 X 7010 25570.0 1 X 7120 25875.1 3 X 7120 25912.1 3 X 7160 26035.2 1 X 7260 26444.2 5 X 7260 26454.2 6 X 7400 26969.9 4 X 7400 26970 X 7400 26970.7 3 X 7400 26970.7 4 X 7400 26970.8 4 X 7420 26985.1 X 7420 26985.1 1 X 7420 26985.1 3 X 7420 26985.3 1 X 7420 26985.4 3 X 7490 27284.6 7 X 7490 27285.0 6 X 7520 27356.4 8 X 7550 27515.0 6 X 7550 27515.2 9 X 7670 27985.6 1 X 7670 27987.8 2 X 7670 27992.0 4 X 7670 27992.3 3 X 7760 28302.3 7 X 50 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 96ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 7860 28674.2 8 X 7860 28674.4 3 X 7990 29171.1 9 X 7990 29171.2 1 X 8060 29453.5 7 X 8060 29454.2 9 X 8060 29454.6 1 X 8060 29455.3 6 X 8060 29455.3 6 X 8060 29455.9 1 X 8060 29458.6 5 X 8060 29464.3 2 X 8060 29464.3 3 X 8060 29464.9 7 X 8060 29465.5 X 8060 29465.6 4 X 8060 29466.7 5 X 8060 29466.7 7 X 8060 29467.9 7 X 8060 29469.0 6 X 8060 29470.9 6 X 8060 29471.0 1 X 8060 29485.3 3 X 8060 29485.3 4 X 8060 29485.6 3 X 8060 29486.1 5 X 8070 29496.6 7 X 8140 29742.0 2 X 8280 30307.1 5 X 8280 30308.3 9 X 51 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 97ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 8280 30308.6 3 X 8280 30308.9 1 X 8280 30309.2 2 X 8280 30309.4 5 X 8290 30333.1 2 X 8300 30357.4 9 X 8300 30395.9 5 X 8300 30396.3 2 X 8340 30517.6 5 X 8360 30621.8 3 X 8360 30624.3 5 X 8460 30970.7 9 X 8460 30970.8 8 X 8500 31060.9 X 8520 31153.5 3 X 8520 31153.8 6 X 8520 31154.3 X 8520 31154.9 8 X 8590 31450.8 8 X 8590 31450.9 6 X 8640 31555.3 9 X 8640 31555.4 4 X 8640 31555.7 1 X 8650 31558.0 6 X 8650 31558.0 6 X 8650 31558.1 1 X 8660 31597.7 5 X 8660 31598.1 2 X 8660 31598.1 4 X 8660 31598.1 7 X 52 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 98ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 8660 31598.1 8 X 8660 31598.2 1 X 8660 31598.2 8 X 8660 31599.8 9 X 8660 31600.2 3 X 8660 31612.7 7 X 8680 31632.1 9 X 8680 31632.3 4 X 8690 31634.2 5 X 8690 31635.4 X 8700 31656.4 1 X 8700 31695.0 7 X 8910 32268.1 1 X 8910 32268.2 2 X 8960 32400.7 X 8980 32410.9 9 X 9030 32563.9 5 X 9030 32564.0 3 X 9040 32564.8 X 9040 32564.8 5 X 9060 32644.8 9 X 9060 32644.9 3 X 9160 32962.2 2 X 9160 32975.4 8 X 9160 32975.7 5 X 9160 32975.7 6 X 9160 32976.0 5 X 9160 32976.0 6 X 9160 32976.0 6 X 9160 32976.0 6 X 9160 32976.0 X 53 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 99ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 7 9160 32976.1 9 X 9160 32976.2 3 X 9160 32976.6 9 X 9160 32977.1 3 X 9160 32977.1 4 X 9160 32978.3 5 X 9200 33161.1 6 X 9210 33162.9 3 X 9220 33240.9 1 X 9220 33241.3 3 X 9220 33241.5 3 X 9220 33241.5 8 X 9230 33242.2 1 X 9250 33322.9 6 X 9250 33354.5 9 X 9250 33355.1 5 X 9250 33356.7 9 X 9250 33356.8 X 9250 33359.4 2 X 9250 33360.4 7 X 9250 33360.8 2 X 9250 33361 X 9260 33371.6 8 X 9260 33371.7 9 X 9260 33371.8 3 X 9260 33372.1 4 X 9260 33400.7 1 X 9270 33402.6 9 X 9270 33425.2 2 X 9270 33431.5 X 54 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 100ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 2 9270 33436.2 5 X 9270 33440.3 9 X 9270 33440.6 2 X 9270 33441.1 1 X 9270 33441.2 4 X 9270 33441.5 9 X 9270 33441.6 X 9270 33441.6 6 X 9280 33443.9 6 X 9280 33452.4 9 X 9280 33453.3 8 X 9280 33454.0 4 X 9280 33454.2 8 X 9280 33455.7 7 X 9280 33456 X 9280 33456.2 5 X 9280 33457.7 8 X 9280 33458.6 8 X 9280 33458.7 X 9280 33459.2 8 X 9280 33460.0 7 X 9280 33460.7 5 X 9280 33469.5 3 X 9280 33469.6 8 X 9280 33469.8 6 X 9280 33470.4 3 X 9280 33470.6 1 X 9280 33472.3 X 9280 33472.4 X 9280 33473.4 7 X 9280 33473.9 4 X 55 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 101ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 9280 33474.0 8 X 9280 33474.4 X 9280 33474.4 X 9280 33474.4 6 X 9280 33474.4 8 X 9280 33474.4 9 X 9280 33474.5 5 X 9280 33474.6 1 X 9280 33474.8 5 X 9280 33474.9 4 X 9280 33474.9 6 X 9280 33475.1 8 X 9280 33475.4 1 X 9280 33475.4 3 X 9280 33475.6 X 9280 33476.0 2 X 9280 33476.0 8 X 9280 33476.2 6 X 9280 33476.3 2 X 9280 33476.5 7 X 9280 33476.7 5 X 9280 33476.9 X 9280 33477.1 6 X 9280 33477.3 6 X 9280 33477.3 7 X 9280 33477.5 X 9280 33477.5 X 9280 33477.5 6 X 9280 33477.6 6 X 9280 33477.7 5 X 9280 33478.0 2 X 9280 33478.3 7 X 56 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 102ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 9280 33479.8 5 X 9290 33482.3 7 X 9290 33482.3 8 X 9290 33482.4 2 X 9300 33527.6 9 X 9300 33527.9 9 X 9300 33528.7 5 X 9300 33528.8 3 X 9300 33529.1 8 X 9300 33529.5 X 9300 33529.5 1 X 9300 33529.9 6 X 9300 33530.3 9 X 9300 33530.4 7 X 9300 33530.5 6 X 9300 33530.9 9 X 9300 33531.0 2 X 9300 33531.0 4 X 9300 33531.1 9 X 9310 33562.5 6 X 9310 33562.6 1 X 9310 33562.8 X 9360 33764.3 8 X 9360 33764.6 5 X 9360 33764.8 9 X 9360 33765.2 X 9360 33766.3 4 X 9360 33766.6 2 X 9360 33766.8 9 X 9360 33788.6 5 X 9360 33789.4 X 57 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 103ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 7 9360 33789.5 4 X 9360 33789.7 1 X 9390 33867.5 2 X 9390 33903.0 9 X 9390 33903.3 X 9390 33903.3 7 X 9390 33903.6 X 9390 33903.9 X 9390 33904.0 8 X 9420 33999 X 9420 33999.0 7 X 9420 33999.3 2 X 9420 33999.5 9 X 9420 33999.7 3 X 9420 33999.8 6 X 9420 34000.1 8 X 9420 34000.4 X 9420 34000.5 8 X 9420 34026.2 1 X 9420 34026.2 2 X 9420 34026.2 2 X 9420 34026.2 4 X 9420 34026.2 8 X 9430 34059 X 9430 34063.6 1 X 9430 34063.7 6 X 9430 34063.9 3 X 9450 34139.9 2 X 9450 34140.7 4 X 9460 34156.5 8 X 9460 34157.5 X 9460 34158.2 X 58 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 104ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 4 9470 34188.4 1 X 9470 34188.4 3 X 9470 34188.4 4 X 9470 34188.4 4 X 9470 34188.4 5 X 9590 34670.2 6 X 9650 34912.2 3 X 9650 34913.7 5 X 9660 34962.4 9 X 9670 34971.8 9 X 9690 35051.4 6 X 9860 35635.8 5 X 9890 35794.2 9 X 9890 35794.3 2 X 9910 35853.4 8 X 9910 35853.5 6 X 9920 35875.9 5 X 9920 35875.9 6 X 9920 35875.9 8 X 9920 35875.9 8 X 9920 35876 X 9920 35876.0 2 X 9920 35876.0 2 X 1007 0 36496.6 8 X 1051 0 37951.3 1 X 1054 0 38046.7 1 X 1062 0 38363.7 X 1064 0 38444.4 1 X 1064 0 38444.4 2 X 59 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 105ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 1064 0 38444.4 3 X 1083 0 39159.9 6 X 1083 0 39176.5 7 X 1083 0 39176.5 7 X 1083 0 39176.6 6 X 1084 0 39205.6 6 X 1092 0 39375.2 5 X 1092 0 39375.3 1 X 1092 0 39396.1 1 X 1093 0 39425.0 7 X 1095 0 39466.9 6 X 1095 0 39467.0 4 X 1095 0 39467.0 6 X 1095 0 39467.0 8 X 1095 0 39467.0 9 X 1095 0 39467.2 3 X 1095 0 39470.9 3 X 1095 0 39471.5 6 X 1095 0 39485.5 1 X 1095 0 39492.7 1 X 1095 0 39493.0 3 X 1096 0 39506.9 2 X 1096 0 39540.9 2 X 1096 0 39541.3 4 X 1096 0 39541.5 7 X 1098 0 39551.4 1 X 1098 0 39551.9 8 X 1098 0 39554.3 9 X 1099 0 39592.1 7 X 1099 39592.4 X 60 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 106ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 0 4 1100 0 39614.6 8 X 1100 0 39615.1 7 X 1100 0 39615.2 2 X 1100 0 39615.3 6 X 1100 0 39615.3 6 X 1100 0 39615.4 7 X 1103 0 39703.1 2 X 1103 0 39703.3 6 X 1103 0 39703.4 4 X 1105 0 39768.1 9 X 1106 0 39810.3 1 X 1131 0 40732.5 3 X 1131 0 40732.6 3 X 1131 0 40732.6 4 X 1133 0 40751.9 9 X 1133 0 40752.0 6 X 1133 0 40752.1 3 X 1133 0 40752.1 3 X 1141 0 40991.6 6 X 1141 0 40991.6 7 X 1141 0 40992.2 5 X 1141 0 40993.5 5 X 1141 0 40993.5 9 X 1141 0 40999.3 X 1141 0 40999.5 4 X 1141 0 40999.6 5 X 1141 0 40999.6 8 X 1146 0 41205.1 8 X 1147 0 41230.3 2 X 61 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 107ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 1147 0 41234.8 9 X 1147 0 41234.9 2 X 1154 0 41506.7 8 X 1154 0 41507.8 3 X 1154 0 41508.2 6 X 1154 0 41508.8 2 X 1154 0 41508.8 8 X 1154 0 41509.0 6 X 1154 0 41509.2 6 X 1154 0 41509.6 6 X 1154 0 41509.7 9 X 1154 0 41509.9 3 X 1154 0 41509.9 7 X 1154 0 41510.0 1 X 1154 0 41510.0 5 X 1154 0 41510.1 7 X 1154 0 41510.3 2 X 1154 0 41510.4 7 X 1154 0 41510.6 5 X 1154 0 41510.8 5 X 1154 0 41510.8 6 X 1154 0 41511.1 8 X 1154 0 41511.1 9 X 1154 0 41511.2 8 X 1154 0 41511.3 3 X 1154 0 41511.4 X 1154 0 41511.5 7 X 1154 0 41511.6 X 1154 0 41511.8 1 X 1154 41512.0 X 62 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 108ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 0 7 1154 0 41512.1 6 X 1154 0 41512.2 4 X 1154 0 41512.2 6 X 1154 0 41512.3 7 X 1154 0 41512.4 2 X 1154 0 41512.4 4 X 1154 0 41512.4 6 X 1154 0 41512.5 X 1154 0 41512.7 4 X 1154 0 41513.0 1 X 1154 0 41513.0 1 X 1154 0 41513.0 4 X 1154 0 41513.0 4 X 1154 0 41513.1 X 1154 0 41513.1 X 1154 0 41530.5 X 1155 0 41531.4 8 X 1155 0 41531.5 2 X 1155 0 41531.5 2 X 1155 0 41531.5 3 X 1155 0 41531.5 3 X 1155 0 41531.5 3 X 1155 0 41531.5 4 X 1155 0 41531.6 9 X 1155 0 41531.7 1 X 1155 0 41531.7 1 X 1155 0 41538.6 5 X 1155 0 41539.1 5 X 1155 0 41550.9 3 X 63 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 109ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 1155 0 41551.0 6 X 1155 0 41551.2 6 X 1155 0 41551.4 8 X 1155 0 41551.6 X 1155 0 41551.8 X 1155 0 41551.9 5 X 1155 0 41551.9 7 X 1155 0 41551.9 7 X 1155 0 41551.9 8 X 1155 0 41552.6 6 X 1155 0 41552.7 2 X 1155 0 41553.2 1 X 1155 0 41553.2 5 X 1155 0 41560.5 5 X 1155 0 41562.4 3 X 1155 0 41562.4 4 X 1155 0 41563.3 7 X 1155 0 41565.0 1 X 1155 0 41565.3 X 1155 0 41565.6 2 X 1155 0 41566.2 6 X 1155 0 41566.3 4 X 1155 0 41569.5 5 X 1155 0 41569.6 2 X 1156 0 41570.7 8 X 1156 0 41571.3 8 X 1156 0 41571.3 8 X 1156 0 41571.6 8 X 1156 0 41572.0 1 X 1157 41641.6 X 64 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 110ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 0 8 1157 0 41642.6 4 X 1157 0 41645.6 7 X 1157 0 41646.3 5 X 1157 0 41646.5 6 X 1158 0 41678.6 5 X 1158 0 41679.4 8 X 1158 0 41689.8 5 X 1158 0 41690.6 8 X 1158 0 41690.7 7 X 1159 0 41690.9 6 X 1159 0 41691.1 5 X 1159 0 41691.1 5 X 1159 0 41691.7 5 X 1159 0 41691.8 3 X 1159 0 41691.8 6 X 1159 0 41700.4 3 X 1159 0 41700.7 4 X 1159 0 41700.9 8 X 1159 0 41701 X 1159 0 41701.3 6 X 1159 0 41701.4 1 X 1159 0 41701.4 5 X 1159 0 41701.5 X 1159 0 41701.5 1 X 1159 0 41701.5 6 X 1159 0 41701.7 3 X 1159 0 41701.8 6 X 1159 0 41701.9 2 X 1159 0 41702.0 5 X 65 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 111ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 1159 0 41702.1 2 X 1159 0 41702.1 3 X 1159 0 41702.3 7 X 1159 0 41702.3 9 X 1159 0 41702.4 5 X 1159 0 41702.6 6 X 1159 0 41702.7 X 1159 0 41702.7 2 X 1159 0 41702.7 4 X 1159 0 41702.7 5 X 1159 0 41702.7 6 X 1159 0 41702.8 5 X 1159 0 41702.8 9 X 1159 0 41703.0 6 X 1159 0 41703.0 6 X 1159 0 41703.0 7 X 1159 0 41703.3 3 X 1159 0 41703.8 X 1159 0 41707.2 7 X 1159 0 41707.7 1 X 1159 0 41707.7 2 X 1159 0 41707.9 7 X 1159 0 41708.0 4 X 1159 0 41709.2 4 X 1159 0 41709.9 6 X 1159 0 41711.1 8 X 1159 0 41711.8 6 X 1159 0 41711.9 4 X 1159 0 41711.9 5 X 1159 41712.1 X 66 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 112ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 0 2 1159 0 41712.2 X 1159 0 41712.2 1 X 1159 0 41712.5 4 X 1159 0 41712.7 1 X 1159 0 41712.8 X 1159 0 41712.9 1 X 1159 0 41713.0 9 X 1159 0 41713.3 8 X 1159 0 41713.6 5 X 1159 0 41713.6 7 X 1159 0 41713.6 9 X 1159 0 41714.5 7 X 1159 0 41714.7 1 X 1159 0 41714.7 5 X 1159 0 41714.7 5 X 1159 0 41715.2 7 X 1159 0 41715.4 5 X 1159 0 41715.8 4 X 1159 0 41715.8 4 X 1159 0 41715.9 8 X 1159 0 41716.1 8 X 1159 0 41716.4 1 X 1159 0 41716.4 9 X 1159 0 41716.7 2 X 1159 0 41716.9 4 X 1159 0 41716.9 8 X 1159 0 41717.0 6 X 1159 0 41717.4 X 1159 0 41717.6 1 X 67 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 113ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 1159 0 41717.6 5 X 1159 0 41717.7 3 X 1159 0 41717.7 6 X 1159 0 41717.9 X 1159 0 41717.9 5 X 1159 0 41718.0 1 X 1159 0 41718.7 1 X 1159 0 41719.2 6 X 1159 0 41719.4 6 X 1159 0 41719.8 9 X 1159 0 41719.9 7 X 1159 0 41720.0 6 X 1159 0 41720.3 5 X 1159 0 41720.5 2 X 1159 0 41720.6 3 X 1159 0 41720.7 2 X 1159 0 41721.1 8 X 1159 0 41721.3 5 X 1159 0 41721.5 4 X 1159 0 41721.6 4 X 1159 0 41721.6 8 X 1159 0 41721.7 X 1159 0 41721.9 1 X 1159 0 41721.9 6 X 1159 0 41722 X 1159 0 41722.0 6 X 1159 0 41722.1 X 1159 0 41722.1 7 X 1159 0 41722.1 7 X 1159 41722.2 X 68 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 114ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 0 1159 0 41722.3 7 X 1159 0 41722.4 3 X 1159 0 41722.6 1 X 1159 0 41722.7 6 X 1159 0 41722.8 2 X 1159 0 41722.8 6 X 1159 0 41723.2 5 X 1159 0 41723.2 8 X 1159 0 41723.5 4 X 1159 0 41723.6 2 X 1159 0 41723.8 9 X 1159 0 41723.9 1 X 1159 0 41724.0 8 X 1159 0 41724.1 2 X 1159 0 41724.4 8 X 1159 0 41724.5 9 X 1159 0 41724.7 5 X 1159 0 41725.0 3 X 1159 0 41725.3 7 X 1159 0 41725.4 6 X 1159 0 41725.9 8 X 1159 0 41726.2 X 1160 0 41741.6 X 1160 0 41741.7 7 X 1160 0 41741.9 2 X 1160 0 41742.0 1 X 1160 0 41742.0 3 X 1160 0 41742.0 7 X 1160 0 41742.0 9 X 69 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 115ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 1160 0 41742.1 X 1160 0 41742.1 3 X 1160 0 41742.2 2 X 1160 0 41742.2 3 X 1161 0 41744.2 5 X 1161 0 41746.6 7 X 1161 0 41747.3 1 X 1161 0 41747.7 8 X 1162 0 41804.1 8 X 1162 0 41804.4 4 X 1162 0 41805.0 7 X 1163 0 41812.0 7 X 1163 0 41812.9 3 X 1163 0 41813.2 1 X 1163 0 41839.8 2 X 1163 0 41839.8 5 X 1164 0 41851.2 4 X 1164 0 41852.0 7 X 1164 0 41852.3 3 X 1165 0 41930.2 X 1166 0 41931.8 5 X 1166 0 41931.8 6 X 1166 0 41931.8 9 X 1166 0 41931.8 9 X 1167 0 41985.3 9 X 1167 0 41985.4 X 1167 0 42010.3 5 X 1193 0 42952.4 9 X 1199 0 43171.6 1 X 1212 43585.0 X 70 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 116ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 0 1 1213 0 43608.1 9 X 1213 0 43608.4 3 X 1215 0 43666.9 6 X 1215 0 43666.9 8 X 1215 0 43667.0 3 X 1216 0 43716.9 4 X 1216 0 43734.4 7 X 1216 0 43735.8 2 X 1216 0 43736.5 6 X 1216 0 43736.6 1 X 1216 0 43736.7 6 X 1216 0 43736.9 X 1216 0 43745.0 1 X 1216 0 43745.0 4 X 1216 0 43745.1 5 X 1217 0 43746.6 5 X 1217 0 43746.7 2 X 1217 0 43746.7 4 X 1217 0 43748.0 2 X 1217 0 43749.8 4 X 1217 0 43750.3 2 X 1217 0 43750.6 9 X 1217 0 43751.2 1 X 1217 0 43752.0 6 X 1217 0 43752.4 8 X 1217 0 43756.8 3 X 1219 0 43814.1 3 X 1223 0 44011.9 3 X 1223 0 44012.3 9 X 71 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 117ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 1223 0 44013.5 X 1224 0 44018.5 9 X 1224 0 44019.6 4 X 1224 0 44019.8 3 X 1224 0 44020.1 2 X 1224 0 44020.9 1 X 1224 0 44021.0 3 X 1224 0 44021.3 4 X 1224 0 44021.8 X 1224 0 44023.6 X 1224 0 44023.6 1 X 1224 0 44023.6 7 X 1224 0 44024.0 2 X 1224 0 44024.6 2 X 1224 0 44024.6 3 X 1224 0 44026.7 1 X 1224 0 44026.7 7 X 1224 0 44027.2 8 X 1224 0 44028.1 X 1224 0 44029.8 9 X 1224 0 44031.6 1 X 1224 0 44032.5 3 X 1224 0 44033.1 7 X 1224 0 44033.3 9 X 1224 0 44033.8 2 X 1224 0 44033.9 7 X 1224 0 44035.6 9 X 1224 0 44037.1 5 X 1224 0 44038.1 9 X 1224 44041.0 X 72 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 118ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 0 6 1224 0 44046.6 8 X 1224 0 44047.5 7 X 1224 0 44048.1 X 1224 0 44048.5 8 X 1224 0 44049.1 3 X 1224 0 44049.3 8 X 1224 0 44051.3 X 1224 0 44053.3 4 X 1224 0 44053.3 5 X 1224 0 44053.3 6 X 1224 0 44053.3 8 X 1224 0 44053.3 8 X 1224 0 44053.4 X 1224 0 44053.4 3 X 1225 0 44055.1 6 X 1225 0 44055.1 7 X 1225 0 44055.1 7 X 1225 0 44055.1 8 X 1225 0 44055.2 X 1227 0 44126.3 9 X 1227 0 44126.4 1 X 1228 0 44170.4 8 X 1228 0 44171.0 7 X 1228 0 44171.1 8 X 1228 0 44171.2 9 X 1228 0 44173.1 8 X 1228 0 44204.7 X 1228 0 44204.7 5 X 1231 0 44286.7 2 X 73 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 119ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 1241 0 44708.2 4 X 1241 0 44708.5 6 X 1241 0 44708.6 4 X 1241 0 44708.9 9 X 1241 0 44709.3 6 X 1241 0 44709.3 9 X 1242 0 44709.7 6 X 1242 0 44710 X 1242 0 44710.6 6 X 1242 0 44710.9 3 X 1242 0 44710.9 4 X 1242 0 44711.2 8 X 1242 0 44712.1 7 X 1242 0 44712.3 2 X 1242 0 44712.4 4 X 1242 0 44712.7 3 X 1242 0 44712.7 8 X 1242 0 44713.3 1 X 1242 0 44725.7 4 X 1242 0 44745.1 X 1242 0 44745.6 5 X 1242 0 44745.8 4 X 1242 0 44747.5 1 X 1242 0 44747.5 3 X 1242 0 44747.5 9 X 1242 0 44747.7 4 X 1242 0 44747.7 8 X 1242 0 44747.8 X 1242 0 44747.8 X 1242 44747.8 X 74 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 120ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 0 2 1242 0 44748.5 5 X 1243 0 44748.7 8 X 1243 0 44748.9 6 X 1243 0 44749.0 4 X 1243 0 44749.6 2 X 1243 0 44749.7 2 X 1243 0 44750.1 8 X 1243 0 44750.3 1 X 1243 0 44750.4 5 X 1243 0 44750.6 3 X 1243 0 44753.3 1 X 1243 0 44753.7 7 X 1243 0 44765.6 8 X 1243 0 44766.2 3 X 1243 0 44766.7 X 1243 0 44766.8 2 X 1243 0 44767.0 3 X 1243 0 44767.3 4 X 1243 0 44768.0 4 X 1243 0 44768.4 7 X 1243 0 44768.7 9 X 1243 0 44769.4 2 X 1243 0 44769.9 2 X 1243 0 44770.9 5 X 1243 0 44770.9 9 X 1243 0 44771.6 2 X 1243 0 44774.5 X 1243 0 44784.0 3 X 1243 0 44784.3 8 X 75 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 121ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 1243 0 44784.6 8 X 1243 0 44788.3 X 1243 0 44788.4 4 X 1243 0 44788.6 7 X 1246 0 44908.1 3 X 1247 0 44909.8 2 X 1248 0 44987.9 7 X 1249 0 44989.6 X 1249 0 44989.6 X 1249 0 44989.6 8 X 1249 0 45010.9 2 X 1250 0 45034.2 1 X 1250 0 45039.7 8 X 1250 0 45044.4 9 X 1250 0 45044.5 4 X 1250 0 45068 X 1250 0 45068.1 1 X 1251 0 45070 X 1251 0 45070.0 4 X 1251 0 45072.8 9 X 1251 0 45073.1 4 X 1251 0 45074.1 8 X 1251 0 45074.4 5 X 1251 0 45075.2 9 X 1251 0 45076.5 3 X 1251 0 45077.5 X 1251 0 45083.1 6 X 1251 0 45083.2 9 X 1251 0 45083.6 9 X 1251 45084.0 X 76 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 122ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 0 3 1251 0 45084.2 X 1251 0 45084.4 9 X 1251 0 45085.7 6 X 1251 0 45086.7 4 X 1251 0 45088.2 X 1251 0 45088.7 7 X 1251 0 45089.1 1 X 1251 0 45089.7 9 X 1251 0 45090.1 1 X 1251 0 45090.2 9 X 1251 0 45090.4 3 X 1251 0 45090.5 4 X 1253 0 45150.0 7 X 1253 0 45178.2 4 X 1253 0 45178.4 7 X 1254 0 45197.1 8 X 1254 0 45197.6 2 X 1254 0 45200.0 5 X 1254 0 45200.5 7 X 1254 0 45204.0 9 X 1254 0 45204.4 X 1255 0 45205.0 1 X 1255 0 45205.0 6 X 1255 0 45205.4 7 X 1255 0 45205.8 2 X 1255 0 45206.2 1 X 1255 0 45206.6 9 X 1255 0 45222.4 8 X 1255 0 45238.9 4 X 77 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 123ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 1259 0 45370.3 8 X 1259 0 45370.4 6 X 1259 0 45370.6 2 X 1259 0 45371.0 2 X 1271 0 45748.3 4 X 1271 0 45748.3 6 X 1271 0 45748.3 8 X 1271 0 45748.4 2 X 1271 0 45748.4 4 X 1272 0 45812.8 7 X 1273 0 45814.2 1 X 1278 0 46022.4 8 X 1279 0 46024.2 9 X 1279 0 46044.1 8 X 1280 0 46082.3 7 X 1280 0 46082.7 2 X 1280 0 46082.8 4 X 1280 0 46083.0 7 X 1280 0 46083.5 6 X 1280 0 46083.5 7 X 1280 0 46083.6 4 X 1280 0 46083.8 9 X 1280 0 46084.4 2 X 1280 0 46084.4 7 X 1280 0 46084.8 4 X 1280 0 46085 X 1280 0 46085.4 2 X 1280 0 46085.6 3 X 1280 0 46085.7 2 X 1280 46085.9 X 78 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 124ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 0 8 1280 0 46086.6 2 X 1280 0 46086.6 4 X 1280 0 46087.2 8 X 1280 0 46087.3 2 X 1280 0 46087.6 1 X 1280 0 46087.7 3 X 1280 0 46087.9 7 X 1280 0 46088.3 3 X 1280 0 46088.5 2 X 1280 0 46088.5 5 X 1280 0 46088.9 1 X 1280 0 46089.1 7 X 1280 0 46089.4 5 X 1280 0 46089.5 1 X 1280 0 46089.8 3 X 1280 0 46090.0 3 X 1280 0 46090.2 X 1280 0 46091.3 8 X 1280 0 46091.9 3 X 1280 0 46092.6 4 X 1280 0 46092.7 7 X 1280 0 46093.1 1 X 1280 0 46093.5 6 X 1280 0 46093.6 3 X 1280 0 46093.8 8 X 1280 0 46094.0 4 X 1280 0 46094.4 6 X 1280 0 46094.6 4 X 1280 0 46102.9 6 X 79 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 125ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 1280 0 46103.0 1 X 1280 0 46103.0 7 X 1282 0 46183.2 6 X 1282 0 46183.3 3 X 1282 0 46183.3 4 X 1282 0 46183.3 5 X 1282 0 46183.3 9 X 1282 0 46183.4 2 X 1283 0 46204.8 5 X 1283 0 46204.9 5 X 1283 0 46205.4 3 X 1283 0 46207.4 3 X 1283 0 46207.4 3 X 1283 0 46208.0 9 X 1283 0 46208.5 X 1283 0 46208.5 9 X 1284 0 46226.6 1 X 1284 0 46237.9 6 X 1284 0 46238.5 6 X 1284 0 46239 X 1284 0 46239.2 3 X 1284 0 46239.2 8 X 1284 0 46239.9 5 X 1284 0 46239.9 9 X 1284 0 46240.1 8 X 1284 0 46240.8 4 X 1284 0 46240.9 2 X 1284 0 46241.1 X 1284 0 46241.4 1 X 1284 46250.2 X 80 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 126ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 0 3 1284 0 46251.8 X 1284 0 46251.9 7 X 1284 0 46252.1 6 X 1284 0 46252.3 4 X 1284 0 46252.5 6 X 1284 0 46252.6 9 X 1284 0 46253.2 8 X 1284 0 46253.4 7 X 1284 0 46257.0 4 X 1284 0 46257.0 6 X 1285 0 46272.6 8 X 1285 0 46272.7 5 X 1285 0 46272.8 1 X 1285 0 46273.2 X 1285 0 46273.7 6 X 1285 0 46274.2 X 1285 0 46274.7 9 X 1285 0 46284.5 8 X 1285 0 46284.6 7 X 1285 0 46285.4 X 1285 0 46285.7 X 1285 0 46286.1 2 X 1285 0 46286.2 1 X 1285 0 46286.7 5 X 1285 0 46286.8 4 X 1285 0 46287.1 8 X 1285 0 46288.4 9 X 1285 0 46293.8 9 X 1285 0 46294.3 5 X 81 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 127ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 1285 0 46295.8 7 X 1286 0 46342.5 6 X 1286 0 46342.7 1 X 1286 0 46343.0 1 X 1286 0 46343.5 7 X 1286 0 46343.9 1 X 1287 0 46345.5 8 X 1287 0 46345.6 X 1287 0 46345.6 X 1287 0 46345.6 X 1288 0 46415.4 3 X 1288 0 46415.8 3 X 1288 0 46423.8 6 X 1288 0 46423.9 2 X 1288 0 46423.9 2 X 1288 0 46423.9 8 X 1288 0 46423.9 8 X 1288 0 46424.0 3 X 1288 0 46424.0 6 X 1288 0 46424.0 6 X 1288 0 46424.2 8 X 1289 0 46449.2 7 X 1289 0 46449.5 4 X 1289 0 46449.8 8 X 1289 0 46451.5 X 1290 0 46504.2 3 X 1291 0 46505.9 2 X 1291 0 46511.1 4 X 1297 0 46721.3 4 X 1298 46735.9 X 82 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 128ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 0 1 1298 0 46738.9 4 X 1300 0 46796.3 8 X 1300 0 46796.4 3 X 1300 0 46798.6 9 X 1300 0 46800.2 6 X 1300 0 46806.4 7 X 1300 0 46806.7 4 X 1300 0 46806.7 6 X 1300 0 46807.3 8 X 1300 0 46807.5 9 X 1300 0 46807.7 1 X 1301 0 46837.7 9 X 1301 0 46837.8 1 X 1302 0 46838.6 8 X 1302 0 46838.9 7 X 1304 0 46944.3 8 X 1311 0 47203.5 3 X 1311 0 47206.4 7 X 1311 0 47206.5 7 X 1311 0 47206.9 2 X 1311 0 47207.1 9 X 1314 0 47272.5 6 X 1317 0 47319.3 1 X 1317 0 47329.3 9 X 1317 0 47329.5 8 X 1317 0 47329.5 8 X 1317 0 47329.5 9 X 1318 0 47330.8 4 X 1320 0 47373.4 7 X 83 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 129ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 1320 0 47373.5 5 X 1320 0 47373.9 7 X 1320 0 47374.1 9 X 1320 0 47400.3 6 X 1320 0 47400.4 4 X 1320 0 47400.4 6 X 1320 0 47400.4 9 X 1320 0 47400.5 2 X 1320 0 47400.5 9 X 1320 0 47400.6 6 X 1320 0 47400.6 7 X 1320 0 47400.7 X 1320 0 47400.7 X 1320 0 47400.7 1 X 1320 0 47400.7 1 X 1320 0 47400.7 1 X 1321 0 47402.3 5 X 1321 0 47402.3 6 X 1321 0 47402.3 8 X 1321 0 47402.3 9 X 1321 0 47402.3 9 X 1321 0 47402.5 4 X 1321 0 47412.9 1 X 1321 0 47413.4 X 1322 0 47448.5 7 X 1322 0 47480.6 1 X 1323 0 47482.3 3 X 1326 0 47623.5 5 X 1326 0 47623.6 7 X 1326 47623.6 X 84 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 130ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 0 7 1326 0 47623.7 2 X 1326 0 47623.7 3 X 1332 0 47780.0 8 X 1357 0 48607.3 X 1357 0 48607.8 9 X 1357 0 48607.9 2 X 1370 0 48882.1 6 X 1370 0 48882.1 7 X 1370 0 48882.2 6 X 1370 0 48882.3 6 X 1402 0 50168.3 8 X 1402 0 50168.3 8 X 1404 0 50246.6 6 X 1404 0 50246.8 5 X 1404 0 50246.8 8 X 1404 0 50246.9 7 X 1406 0 50259.5 9 X 1406 0 50264.7 7 X 1406 0 50266.1 9 X 1406 0 50267.7 7 X 1406 0 50269.1 1 X 1406 0 50269.2 8 X 1406 0 50269.5 7 X 1406 0 50269.5 7 X 1406 0 50269.6 7 X 1406 0 50269.8 7 X 1406 0 50270.4 4 X 1406 0 50270.5 5 X 1406 0 50270.9 8 X 85 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 131ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 1406 0 50271.2 8 X 1406 0 50271.8 X 1406 0 50271.8 2 X 1406 0 50272.0 7 X 1406 0 50272.3 9 X 1406 0 50272.5 6 X 1406 0 50272.6 X 1406 0 50273.1 9 X 1406 0 50273.1 9 X 1406 0 50273.2 7 X 1406 0 50273.2 8 X 1411 0 50475.5 1 X 1420 0 50820.9 X 1420 0 50821.0 8 X 1430 0 51203.9 6 X 1431 0 51271.2 1 X 1431 0 51271.2 2 X 1431 0 51271.4 4 X 1431 0 51271.6 7 X 1431 0 51271.7 8 X 1431 0 51271.8 4 X 1431 0 51271.8 4 X 1431 0 51271.9 5 X 1447 0 51642.6 3 X 1450 0 51691.6 4 X 1459 0 51974.4 6 X 1465 0 52251.8 7 X 1465 0 52251.8 8 X 1489 0 53192.1 2 X 1490 53193.8 X 86 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 132ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 0 4 1490 0 53193.8 6 X 1491 0 53234.2 8 X 1492 0 53274.9 4 X 1492 0 53275.2 4 X 1492 0 53313.0 9 X 1505 0 53823.9 9 X 1553 0 55653.5 5 X 1554 0 55696.3 4 X 1577 0 56478.4 5 X 1577 0 56478.4 7 X 1578 0 56480.1 1 X 1578 0 56480.1 3 X 1585 0 56711.6 4 X 1585 0 56712.9 7 X 1585 0 56713.4 1 X 1585 0 56713.5 9 X 1585 0 56713.6 1 X 1585 0 56713.7 1 X 1585 0 56714.4 3 X 1585 0 56714.4 9 X 1585 0 56716.3 6 X 1585 0 56717.4 5 X 1588 0 56809.4 X 1588 0 56809.4 1 X 1590 0 56850.4 5 X 1590 0 56850.9 4 X 1590 0 56851.1 8 X 1590 0 56858.6 8 X 1590 0 56868 X 87 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 133ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Distance Jan- May - Nov - Jan- May - Jul- Nov - Jan- May - Jul - Nov - Jan- May - Jul- Nov- Jan- May- Jul- Nov- Jan- May- Jul- Nov- May- Nov- May- Nov- May- May- GW (ft) 15 15 15 16 16 16 16 17 17 17 17 18 18 18 18 19 19 19 19 20 20 20 20 21 21 22 22 23 24 1590 0 56878.5 X 1590 0 56882.6 5 X 1591 0 56895.8 7 X 1591 0 56895.9 5 X 1591 0 56902.7 9 X 1591 0 56902.9 8 X 1591 0 56903.9 2 X 1591 0 56914.1 9 X 1591 0 56916.5 4 X 1591 0 56922.0 3 X 1591 0 56922.0 8 X 1595 0 56972.5 1 X 1595 0 56974.2 4 X 88 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016#

Page 134ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota Appendix C – Full Excavation Timeline (30’ limit) GW Dig Start (ft) Dig End (ft) Length (ft) Dig Date GW Dig Start (ft) 70.01/80 125.59 142.82 17.23 Jul-17 750 2223.28 250/260 597.18 607.51 10.33 Jan-20 750/760 2256.47 260 636.44 636.47 0.03 Jan-15 760/770/780 2286.7 290/310 668.45 678.96 10.51 Jul-17 800 2368.43 420 956.64 956.7 0.06 Jan-16 970 2990.53 470 1134.02 Jul-20 970/980 3023.98 480 1164.9 1186.34 21.44 Jan-20 980/990 3055.48 480/490 1197.72 1224.65 26.93 Jul-20 1050 3347.6 500 1239.97 1262.12 22.15 Jul-20 1070 3406.47 500/510 1277.24 1303.42 26.18 Jul-19 1090 3489.91 520 1320.27 1348.62 28.35 Jan-20 1350/1360 4530.98 530 1350.34 1375.95 25.61 Jul-19 1370 4608.59 530/540 1380.45 1405.69 25.24 Jul-20 1480 5026.35 540 1412.65 1437.2 24.55 Jul-20 1520 5186.17 550 1446.49 1473.06 26.57 Jul-20 1550/1560 5295.01 560 1477.26 1504.82 27.56 Jan-20 1570 5382.03 560/570 1509.34 1538.27 28.93 Jul-20 1580/1590/1600 5425.09 570/580 1540.72 1563.76 23.04 Jul-20 1700 5833.49 580/590 1572.03 1604.13 32.1 Jan-18 1980/1990 6902.95 600/610 1673.77 1700.71 26.94 Jan-18 2170 7618.29 610 1708.07 Jul-20 2170 7654.47 620 1752.17 1753.14 0.97 Jul-20 2210 7786.93 640/650 1817.13 1829.55 12.42 Jan-20 2370 8428.95 710/720 2097.51 2126.87 29.36 Jan-19 2430 8654.45 720/730 2127.56 2153.96 26.4 Jul-18 2450 8733.29 730/740 2157.94 2184.73 26.79 Jul-18 2500 8898.57 740/750 2193.05 2222.91 29.86 Jul-18 2640 9452.13 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016 Dig End (ft) 2250.65 2278.68 2314.34 2368.81 3020.01 3052.94 3074.62 3347.71 3427.82 4532.7 4610.89 5319.85 5382.84 5449.97 5833.55 6928.82 7648.22 7654.78 7787.14 8733.39 9460.01 Length (ft) Dig Date 27.37 Jan-18 22.21 Jul-18 27.64 Jan-18 0.38 Jul-19 29.48 Jul-19 28.96 Jul-20 19.14 Jul-19 0.11 Jul-18 21.35 Jul-18 Jan-20 1.72 Jul-16 2.3 Jan-15 Jul-19 Jan-20 24.84 Jul-18 0.81 Jan-15 24.88 Jan-20 0.06 Jan-17 25.87 Jul-18 29.93 Jan-16 0.31 Jan-17 0.21 Jan-18 Jul-20 Jul-20 0.1 Jan-18 Jul-20 7.88 May-17 89#

Page 135ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota GW Dig Start (ft) Dig End (ft) Length (ft) Dig Date GW Dig Start (ft) 2830 10119.99 May-21 4680/4690 16618.79 2940 10515.15 May-23 4730 16780.45 2960 10584.12 10589.43 5.31 May-16 4900 17473.17 3010 10755.92 10756.35 0.43 May-21 5100/5120 18203.47 3090 11060 May-22 5180 18453.36 3220 11615.85 May-23 5400 19323.51 3370 12195.91 May-23 5620 20203.62 3430 12427.16 May-22 5660 20363.63 3630 13155.91 13156.18 0.27 May-20 5680 20442.48 3680 13394.44 May-20 5840 21048.9 3750 13645.56 13645.75 0.19 May-22 5870 21150.01 3810 13852.54 13853.93 1.39 May-22 5930 21367.11 3850 14002.02 Nov-22 5980 21552.33 4080 14751.62 14757.57 5.95 May-19 6010 21710.28 4140/4150 14909.43 14934.69 25.26 Nov-19 6060/6070 21845.43 4150/4160/4160.01/4160.02 14945.13 14968.53 23.4 May-15 6090 21955.27 4200/4210 15015.82 15026.16 10.34 May-22 6100/6110 22033.77 4210/4220 15049.79 15076.28 26.49 Nov-15 6180 22354.2 4220/4230 15086.08 15106.27 20.19 May-15 6270 22682.66 4240/4250 15184.5 15186.2 1.7 May-20 6310 22812.91 4260/4270 15263.79 15275.45 11.66 Nov-21 6350/6360 23006.72 4300 15295.63 May-23 6360/6370 23040.25 4340/4360 15366.08 15377.59 11.51 May-21 6370 23074.6 4390 15454.72 May-21 6400 23198.67 4410 15505.12 Nov-22 6520 23639.84 4430 15585.19 May-21 6580 23867.6 4540 16038.87 Nov-22 6590/6600 23945.68 4620 16377.71 May-23 6790 24696.32 4640/4650 16458.49 16460.62 2.13 Nov-17 6990 25487.18 4660 16532.12 16538.67 6.55 May-18 6990 25525.05 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016 Dig End (ft) 16620.66 16780.77 18213.58 19324.23 20363.82 21385.69 21875.26 22035.46 22682.99 23032.43 23063.48 23669.09 23947.62 Length (ft) 1.87 0.32 10.11 0.72 0.19 18.58 29.83 1.69 0.33 25.71 23.23 29.25 1.94 Dig Date May-22 Nov-22 May-23 Nov-18 Nov-20 May-22 May-23 May-18 May-21 May-20 May-23 Nov-15 May-23 May-22 May-18 May-23 May-15 May-22 May-22 May-23 May-15 Nov-17 May-23 May-23 Nov-18 Nov-20 May-21 May-21 May-22 May-23 90#

Page 136ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota GW Dig Start (ft) Dig End (ft) Length (ft) Dig Date GW Dig Start (ft) 7010 25570.01 Nov-22 8960/8980 32400.7 7120 25875.13 May-22 9030/9040 32563.95 7120 25912.13 May-23 9060 32644.89 7160 26035.21 Nov-22 9160 32962.22 7260 26444.25 26454.26 10.01 May-22 9200/9210 33161.16 7400/7420 26969.94 26985.43 15.49 May-19 9220/9230 33240.91 7490 27284.67 27285.06 0.39 Nov-21 9250 33322.96 7520 27356.48 May-23 9250/9260 33354.59 7550 27515.06 27515.29 0.23 May-21 9260/9270 33400.71 7670 27985.61 27992.33 6.72 May-20 9270/9280 33431.52 7760 28302.37 Nov-22 9280/9290 33469.53 7860 28674.28 28674.43 0.15 Nov-22 9300 33527.69 7990 29171.19 29171.21 0.02 May-15 9310 33562.56 8060 29453.57 29471.01 17.44 May-15 9360 33764.38 8060/8070 29485.33 29496.67 11.34 Nov-20 9390 33867.52 8140 29742.02 Nov-15 9390 33903.09 8280/8290 30307.15 30333.12 25.97 May-15 9420 33999 8300 30357.49 May-23 9430 34059 8300 30395.95 30396.32 0.37 Nov-22 9450/9460 34139.92 8340 30517.65 May-23 9470 34188.41 8360 30621.83 30624.35 2.52 Nov-22 9590 34670.26 8460 30970.79 30970.88 0.09 Nov-21 9650 34912.23 8500 31060.9 May-22 9660/9670 34962.49 8520 31153.53 31154.98 1.45 May-21 9690 35051.46 8590 31450.88 31450.96 0.08 May-21 9860 35635.85 8640/8650 31555.39 31558.11 2.72 May-15 9890 35794.29 8660 31597.75 31612.77 15.02 May-19 9910/9920 35853.48 8680/8690/8700 31632.19 31656.41 24.22 Nov-20 10070 36496.68 8700 31695.07 May-22 10510 37951.31 8910 32268.11 32268.22 0.11 May-22 10540 38046.71 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016 Dig End (ft) 32410.99 32564.85 32644.93 32978.35 33162.93 33242.21 33372.14 33425.22 33460.75 33482.42 33531.19 33562.8 33789.71 33904.08 34026.28 34063.93 34158.24 34188.45 34913.75 34971.89 35794.32 35876.02 Length (ft) 10.29 0.9 0.04 16.13 1.77 1.3 17.55 24.51 29.23 12.89 3.5 0.24 25.33 0.99 27.28 4.93 18.32 0.04 1.52 9.4 0.03 22.54 Dig Date May-21 Nov-22 May-22 Nov-16 May-21 Nov-21 May-22 Nov-16 Nov-18 May-15 May-15 May-18 May-18 May-19 May-18 May-17 May-15 May-16 Nov-16 May-18 Nov-22 May-21 May-22 May-23 May-18 Nov-19 Nov-19 May-23 May-23 Nov-21 91#

Page 137ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota GW Dig Start (ft) Dig End (ft) Length (ft) Dig Date GW Dig Start (ft) 10620 38363.7 May-23 12150 43666.96 10640 38444.41 38444.43 0.02 Nov-21 12160/12170 43716.94 10830 39159.96 39176.66 16.7 May-21 12170 43748.02 10840 39205.66 Nov-22 12190 43814.13 10920 39375.25 39396.11 20.86 May-22 12230/12240 44011.93 10930 39425.07 May-22 12240/12250 44046.68 10950 39466.96 39493.03 26.07 May-18 12270 44126.39 10960 39506.92 May-23 12280 44170.48 10960/10980 39540.92 39554.39 13.47 May-21 12280 44204.7 10990/11000 39592.17 39615.47 23.3 May-21 12310 44286.72 11030 39703.12 39703.44 0.32 May-22 12410/12420 44708.24 11050 39768.19 Nov-22 12420/12430 44745.1 11060 39810.31 Nov-15 12430 44784.03 11310/11330 40732.53 40752.13 19.6 Nov-19 12460/12470 44908.13 11410 40991.66 40999.68 8.02 Nov-18 12480/12490 44987.97 11460/11470 41205.18 41234.92 29.74 Nov-21 12500 45034.21 11540/11550 41506.78 41531.71 24.93 Nov-16 12510 45068 11550 41538.65 41566.34 27.69 Nov-19 12530 45150.07 11550/11560 41569.55 41572.01 2.46 May-19 12540/12550 45197.18 11570 41641.68 41646.56 4.88 May-20 12550 45238.94 11580/11590 41678.65 41708.04 29.39 May-17 12590 45370.38 11590 41709.24 41726.2 16.96 Nov-17 12710 45748.34 11600/11610 41741.6 41747.78 6.18 Nov-18 12720/12730 45812.87 11620/11630 41804.18 41813.21 9.03 May-21 12780/12790 46022.48 11630/11640 41839.82 41852.33 12.51 May-22 12800 46082.37 11650/11660 41930.2 41931.89 1.69 Nov-19 12820/12830 46183.26 11670 41985.39 42010.35 24.96 Nov-20 12840 46226.61 11930 42952.49 May-22 12840/12850 46257.04 11990 43171.61 May-20 12850 46287.18 12120/12130 43585.01 43608.43 23.42 Nov-21 12860/12870 46342.56 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016 Dig End (ft) 43667.03 43746.74 43756.83 44041.06 44055.2 44126.41 44173.18 44204.75 44725.74 44774.5 44788.67 44909.82 45010.92 45044.54 45090.54 45178.47 45222.48 45371.02 45748.44 45814.21 46044.18 46103.07 46208.59 46253.47 46286.84 46295.87 46345.6 Length (ft) 0.07 29.8 8.81 29.13 8.52 0.02 2.7 0.05 17.5 29.4 4.64 1.69 22.95 10.33 22.54 28.4 25.3 0.64 0.1 1.34 21.7 20.7 25.33 26.86 29.8 8.69 3.04 Dig Date May-19 May-17 May-20 May-22 May-16 Nov-19 Nov-16 May-20 Nov-19 May-23 May-15 Nov-15 May-21 May-21 May-16 May-22 May-16 May-21 May-17 May-23 May-20 Nov-19 Nov-22 Nov-19 May-18 May-15 Nov-18 Nov-18 Nov-18 May-16 92#

Page 138ILI Evaluation Report – PAAPL Line 901; Las Flores to Gaviota GW Dig Start (ft) Dig End (ft) Length (ft) Dig Date GW Dig Start (ft) 12880 46415.43 46424.28 8.85 May-15 14200 50820.9 12890 46449.27 46451.5 2.23 May-21 14300 51203.96 12900/12910 46504.23 46511.14 6.91 May-19 14310 51271.21 12970/12980 46721.34 46738.94 17.6 Nov-18 14470 51642.63 13000 46796.38 46807.71 11.33 May-19 14500 51691.64 13010/13020 46837.79 46838.97 1.18 May-20 14590 51974.46 13040 46944.38 May-22 14650 52251.87 13110 47203.53 47207.19 3.66 May-22 14890/14900 53192.12 13140 47272.56 Nov-21 14910 53234.28 13170/13180 47319.31 47330.84 11.53 May-21 14920 53274.94 13200/13210 47373.47 47402.54 29.07 May-15 14920 53313.09 13210 47412.91 47413.4 0.49 May-15 15050 53823.99 13220 47448.57 May-23 15530 55653.55 13220/13230 47480.61 47482.33 1.72 May-22 15540 55696.34 13260 47623.55 47623.73 0.18 Nov-19 15770/15780 56478.45 13320 47780.08 May-23 15850 56711.64 13570 48607.3 48607.92 0.62 Nov-21 15880 56809.4 13700 48882.16 48882.36 0.2 May-15 15900 56850.45 14020 50168.38 Nov-20 15900/15910 56882.65 14040/14060 50246.66 50273.28 26.62 May-20 15910 56914.19 14110 50475.51 Nov-22 15950 56972.51 Pipeline Hazardous Materials Private and Confidential; Final Report and Safety Administration Client/Attorney Privileged March 4, 2016 Dig End (ft) 50821.08 51271.95 52251.88 53193.86 53275.24 56480.13 56717.45 56809.41 56878.5 56903.92 56922.08 56974.24 Length (ft) 0.18 0.74 0.01 1.74 0.3 1.68 5.81 0.01 28.05 21.27 7.89 1.73 Dig Date May-21 May-22 May-20 May-23 May-21 May-23 May-22 Nov-20 Nov-22 Nov-22 May-23 May-22 May-23 Nov-21 May-22 May-20 May-23 May-22 Nov-19 May-22 Nov-21 93#

Page 139Appendix H PHMSA’s Independent Analysis of In-Line Inspection Data#

Page 140Appendix H: PHMSA’s Independent Analysis of ILI Data Plains’ IMP provides written procedures for reviewing an ILI vendor’s final report and describes how they are to analyze the data provided to create a dig list. The corrosion growth rate is assumed to be a linear growth that has taken place over 75% of the time since construction. This is considered similarly for both the depth and length growth. Plains’ IMP Group then considers two failure modes, leak (80% depth limit) and rupture at maximum operating pressure (MOP). The rupture date is set to the date that is 70% of the estimated time taken to reach failure at the MOP. An anomaly is scheduled for excavation when the nearest date from either mode occurs prior to the next proposed ILI assessment survey date. Anomaly dig sheets are then created in Houston and they are sent to the field for execution. The field obtains appropriate permits, conducts the digs and hires a company to come in and perform the NDE on each anomaly. A dig package is then created for each anomaly and includes pictures, data forms, NDE measurements, etc. Once an anomaly dig is completed, the dig package is sent back to the IMP Group in Houston, TX. Plains’ IMP has procedures directing the IMP Group how to analyze the data once the dig information arrives back in Houston from the field. Their procedures are contained in Section 6 and Appendix E1 Magnetic Flux Leakage In-Line Inspection Tool Specification of Plains’ IMP. A short section from page 6-17 and 6-18 of Section 6 of the Plains IMP [Date of Revision: 10 July 2008] are excerpted below. “Validation of ILI Results To validate the ILI results, Plains will record field found anomaly data on the Anomaly Tracker spreadsheet for the anomalies selected for investigation from the PHMSA Compliance Report. A list of data columns of the Anomaly Tracker spreadsheet is included at the end of this Section. The PHMSA Records Specialists will be responsible for inputting the data into this spreadsheet using data from the Form 501 Pipeline Inspection Reports. Once the data on a pipeline segment is compiled, it will be analyzed by various methods, such as, plotting unity graphs and performing statistical analysis. The field found anomaly data will also be entered into a database, where it can be integrated with other pipeline data for additional analysis.” PHMSA requested all records and analysis performed after each of the ILI Surveys on line 901. Plains’ submitted a Unity Plot for as-found versus as-called depth that was created in 2013 – after the 2012 survey and the ILI digs. The plot is shown here. Page 1 of 11#

Page 141The black line is the one to one line where any “as-called equal as-found” anomalies would be plotted. The green lines are +/- 10% lines and the red lines are +/- 15% lines. ~57% of the plotted anomalies are within the +/- 10% lines. Reported tool accuracy is +/- 10% 80% of the time. There is no documentation regarding any further analysis or discussion which may have ensued after the creation of this Unity Plot for wall loss (depth). Appendix E-1 has more specific written procedures concerning contracting with an MFL Tool vendor. This information is excerpted from the Plains IMP with a Date of Revision: “20 December, 2005 APPENDIX E Integrity Management Plan”. On page E1-7 and E1-8 it states: “6.2 Detection and Anomaly Sizing Specification The MFL tool shall meet the minimum detection and anomaly sizing specifications listed in Table E1-1. The Tool Vendor will submit their MFL tool’s actual specifications with their bid. The Company may modify these specifications. Page 2 of 11#

Page 1426.3 Interaction Criteria The Tool Vendor shall use the 1”x 6t interaction criteria in data analysis. For the 1”x 6t interaction criteria, two anomalies will interact if the distance between them is less than or equal to 1” in the axial direction and the circumferential distance between them is less than or equal to 6 times the nominal wall thickness.” The Plains’ IMP also includes a section on verification of tool data as follows: “8.0 Verification of the Inspection Data 8.1 Selection of Verification Digs The investigation of selected anomalies that the Company and Tool Vendor agree upon will be used to compare actual vs. predicted dimensions to provide anomaly verification data to the Tool Vendor. The Tool Vendor’s bid must contain a provision for adjusting the anomaly grading based on the verification data at no cost to Company. 8.2 Verification of External Anomalies External anomalies will be measured by Company field personnel who are qualified to perform API Covered Tasks 8.1 and 8.3. Measurements will be made and recorded for the depth, length and width of the anomaly, as well as the location of the anomaly relative to the reference girth weld. Digital photographs and a sketch or etchings of the anomalies will be made and included in the record. Length of the affected joint and its location relative to the reference marker will be included for comparison to information provided by Tool Vendor. The Company will provide copies of all information obtained from the selected anomalies to the Tool Vendor as soon as possible. The Tool Vendor will review the field data for any corrections to the data analysis for the Final Report.” Independent Review of Smart Pig Data and Field Found Data PHMSA contracted with Oak Ridge National Laboratories (ORNL) to provide a Subject Matter Expert (SME) to assist in the investigation by performing an analysis of the MFL smart Page 3 of 11#

Page 143pig in-line-inspection (ILI) data and by comparing that data with the digs made and the information gathered by the non-destructive-examination (NDE) of the anomalies in the field. The analysis included a review of the raw ILI data from the 2007, 2012 and 2015 ILI surveys and comparing that data to the as found data when each anomaly was excavated and measured in the field. All of the data used in the ILI-SME report was provided by the Plains’ IMP Group. Unity plots (as-found versus as-called) were made for length, width and depth. Generally, if a point is called a certain value and the field measured value is the same value, the point will fall on a line that runs at a 45 degree angle from zero up and to the right on a Cartesian coordinate graph. Dotted lines are added parallel to the unity line, placed at +/- 10% which is the reported tool tolerance. The smart pigs utilized for each of the surveys in 2007, 2012, and 2015 were from the same vendor and were high resolution magnetic flux leakage (MFL) smart pigs. Note: The tools differed slightly but utilized the same MFL technology. See the full report for the full discussion. The first item of note in the ILI-SME Report is that external corrosion was active on line 901. “Table 2. In line inspection results” from that report is copied below. The table shows that from survey to survey (5 years then 3 years), the number of external corrosion anomalies greater than or equal to 10% increased by 1192 and 169 from 2007 to 2012 and 2012 to 2015 respectively. The ILI-SME report goes on to describe the accuracy of the data presented by the ILI tool vendor compared with the actual measurements found when excavated and measured. The stated accuracy of the tool in the vendor-operator contract was not met for any of the ILI surveys. Also, the tool accuracy using the accepted industry standard, API 1163, was not within stated specifications either. Page 4 of 11#

Page 144Note: The report does conclude that: “If overcalled anomalies were considered (i.e. >10% over actual) then in all years the unities would be ±10%, >70% of the time for depth.” The report concludes the following with respect to the accuracies reported by the ILI Tool vendor after the field reported measurements for the same anomalies. “The unity plot for the 2007 inspection is within ±10%, 33% of the time with respect to the 2007 excavations.” “The unity plot for the 2012 inspection is within ±10%, 58% of the time with respect to the 2012 excavations (blue) and 2007 excavation recoats (violet). When comparing to the 2015 field excavated results based on the 2012 ILI data, growth may have occurred, causing the comparisons between field and ILI to be under-called (orange). The 2015 digs were not considered in the above stated accuracy.” Page 5 of 11#

Page 145“The 2015 ILI estimated depths are compared to field measured depths either from the 4 excavations following the failure or the areas recoated after the 2007 and 2012 inspections. The unity plot shows that the 2015 Rosen inspection is within ±10%, 57% of the time. It may be seen that the failure location has an uncharacteristically high deviation from the ILI estimate.” The ILI-SME describes the process for calculating the remaining strength of a pipe based on the length, width and depth of an anomaly. He also describes the manner in which ILI vendors’ interact individual pits into boxes and then how the boxes interact to form clusters and how clusters can be grouped. Suffice it to say that there is a defined process for Page 6 of 11#

Page 146interacting metal loss anomalies. The only interaction criterion requested by Plains IMP Group was the industry standard one inch by 6 wall thicknesses (1” X 6t) which is normally used for isolated pitting. From the ILI-SME report: “Plains specifies an interaction criteria to be a combination of absolute value for the length component (1”) and wall thickness dependence for the width component (6t). The 1” x 6t interaction rule is one of the most commonly employed throughout industry and is the example given in ASME B31.4.” The ILI-SME report goes on to describe why accurate length and width measurements are important when analyzing external corrosion anomalies. From the ILI-SME report: “The issue of underestimating the length and width of a corrosion anomaly will lead to gross underestimations of the corrosion area. Figure 16 delineates all of the line 901 anomalies with width and length reported from ILI estimates versus excavations made, on a logarithmic scale. As an example, it is showing that 38% of the anomalies had an area stated by the ILI of ≤ 1.5 in2 when in fact the corrosion areas were between 2.5 in2 and 7300 in2. This being said, there may be a difference in the field measurement technique to consider. It is important that the techniques used in the field be comparable to that required by the ILI analysis to enable a proper assessment of the ILI performance.” Figure 16. Metal loss area; ILI vs field measurement. The following are two “Close-Out Reports” provided by the Plains IMP Group. The first one is for the 2007 ILI survey and anomaly digs and the second is for the 2012 ILI survey and Page 7 of 11#

Page 147anomaly digs. The 2007 Close-Out Report states in the, “Results/Comments/Recommendations”, at the bottom of the form, “2. The results show that 86% of the excavated anomalies were within tool tolerance or over-called by the ILI tool and no anomalies meet conditions for further evaluations.” This report was completed on 6/21/2015. The 2012 Close-Out Report states in the “Results/Comments/Recommendations”, at the bottom of the form: “2. The results show that the ILI tool is within the tool’s tolerance specification. 3. The results show that 73% of the excavated anomalies were within tool tolerance or over- called by the ILI tool and no anomalies meet conditions for further evaluations.” Page 8 of 11#

Page 148Two additional analyses were performed by the ILI-SME which was not included in the final filed report. One data set was the number of anomaly digs that were within one and a half feet of a girth weld. Below is the analysis stated and presented graphically. “Within the 2007 excavation locations approximately 50% were within 1.5’ of a GW (blue diamonds). Within the 2012 excavation locations approximately 76% were within 1.5’ of a GW (yellow circles). (The shrink sleeve utilized was 34” total, therefore the length from each side of the GW is app 1.5’). The depth of metal loss found within these excavations relative to distance from the girth weld is shown below.” Page 9 of 11#

Page 149The second analysis had to do with the estimated cubic yards of dirt excavated during each anomaly dig. Plains’ personnel explained to PHMSA that they were required to keep their excavations below 100 cubic feet. This is important because PHMSA was told by the Plains’ IMP Group that Santa Barbara County has strict requirements for excavators and that obtaining a permit for larger excavations would take from six months or more to obtain a permit. However, Plains’ IMP Group reported that there is an exception for excavations made that are less than 100 cubic yards of dirt. Below is an excerpt from the “Santa Barbara County, Planning and Development, Building and Safety Division Grading Plan Submittal Requirements for Projects (Other than Subdivisions)” delineating exception #4. The following spreadsheet is a calculated estimate of the amount of dirt excavated during a number of the anomaly digs in 2007 and 2012. On the right of the figure there are some noted assumptions including: “* Assuming 10’ width and 8’ depth ** Does not include side or end wall terracing Lengths taken from individual “Pipeline Inspection and Repair Reports” *** Lengths in Repair Reports are inconsistent -Some refer to the full dig opening and others refer to the repaired/recoat length only.” Page 10 of 11#

Page 150This spreadsheet was created to estimate excavated soil volumes for each anomaly dig. Dig numbers are provided as well as volume estimates and assumptions used. Volume was calculated in Cubic Feet and converted to Cubic Yards. If the volume estimates are doubled, they all still come in under the 100 cubic yard threshold. Dig #13 in 2012, was located only six feet downstream of the failure location. Page 11 of 11#

Page 151Appendix Maps and Photographs#

Page 152Appendix I: Maps and Photographs Map of Plains’ Western Division Pipelines. The arrow in the ocean is pointing to the approximate release site on line 901. Overview from Santa Barbara Spill Web Site Page 1 of 4#

Page 153Release Site with Culvert in the foreground. Vacuum Truck sucking up pooled oil in the background. Culvert Under Highway and RR Tracks to Ocean Page 2 of 4#

Page 154Release Location Flow Direction 2012 Dig #13 Recoat This picture shows the release site wrapped in plastic 6 feet upstream from girth weld 5940 where the coating repair is visible. The repair was identified as Dig #13 from the Post 2012 ILI Survey Anomaly Digs. Page 3 of 4#

Page 155This is one of the first pictures of the release location after removal from the ditch. This picture was copied from the Final Metallurgical Report. One can see the bare pipe where the insulation and other coatings were removed to allow the pipe to be cut. Page 4 of 4#

Page 156Appendix J National Response Center Report #1#

Page 157TeleDetail http://hmis.phmsa.dot.gov/hmis/telephonics/Teledetail.aspx?showresult... HMIS->INCIDENTS->TELEPHONICS (Version 4.0.0 PROD ) Rules of Behavior Home Logout Menu [Return to Search] NRC Number: 1116950 Call Date: 05/19/2015 Call Time: 15:43:00 Caller Information First Name: Last Name: Company Name: Address: City: State: Country: Zip: Phone 1: Phone 2: Organization Type: Confidential: Yes No No Response Is caller the spiller? Yes No No Response Discharger Information First Name: Last Name: Company Name: Address: City: State: Country: Zip: Phone 1: Phone 2: Organization Type: Spill Information State: County: Nearest City: Zip Code: Location Spill Date: (mm/dd/yyyy) Spill Time: (24hh:mm:ss) DTG Type: Incident Type Reported Incident Type Description Materials Involved Material / Chris Name Chris Code Total Qty. Water Qty. UNKNOWN OIL OUN 0 UNKNOWN AMOUNT 0 UNKNOWN AMOUNT Medium Type: Additional Medium Information: Injuries: Fatalites: Evacuations: Yes No Unknown No. of Evacuations: Damages: Yes No Unknown Damage Amount: Federal Agency Notified: Yes No Unknown State Agency Notified: Yes No Unknown Other Agency Notified: Yes No Unknown Remedial Actions 1 of 2 1/6/2016 7:23 AM#

Page 158TeleDetail http://hmis.phmsa.dot.gov/hmis/telephonics/Teledetail.aspx?showresult... Additional Info Latitude Degrees: Minutes: Seconds: Quadrant: Longitude Degrees: Minutes: Seconds: Quadrant: Distance from City: Direction: Section: Township: Range: Milepost: Rescinded Comments (max 250 characters) 11..11 of 26 2 of 2 1/6/2016 7:23 AM#

Page 159Appendix K National Response Center Report #2#

Page 160TeleDetail http://hmis.phmsa.dot.gov/hmis/telephonics/Teledetail.aspx?showresult... HMIS->INCIDENTS->TELEPHONICS (Version 4.0.0 PROD ) Rules of Behavior Home Logout Menu [Return to Search] NRC Number: 1116972 Call Date: 05/19/2015 Call Time: 17:56:00 Caller Information First Name: Last Name: Company Name: Address: City: State: Country: Zip: Phone 1: Phone 2: Organization Type: Confidential: Yes No No Response Is caller the spiller? Yes No No Response Discharger Information First Name: Last Name: Company Name: Address: City: State: Country: Zip: Phone 1: Phone 2: Organization Type: Spill Information State: County: Nearest City: Zip Code: Location Spill Date: (mm/dd/yyyy) Spill Time: (24hh:mm:ss) DTG Type: Incident Type Reported Incident Type Description Materials Involved Material / Chris Name Chris Code Total Qty. Water Qty. OIL: CRUDE OIL 500 BARREL(S) 0 UNKNOWN AMOUNT Medium Type: Additional Medium Information: Injuries: Fatalites: Evacuations: Yes No Unknown No. of Evacuations: Damages: Yes No Unknown Damage Amount: Federal Agency Notified: Yes No Unknown State Agency Notified: Yes No Unknown Other Agency Notified: Yes No Unknown Remedial Actions 1 of 2 1/6/2016 7:18 AM#

Page 161TeleDetail http://hmis.phmsa.dot.gov/hmis/telephonics/Teledetail.aspx?showresult... Additional Info Latitude Degrees: Minutes: Seconds: Quadrant: Longitude Degrees: Minutes: Seconds: Quadrant: Distance from City: Direction: Section: Township: Range: Milepost: Rescinded Comments (max 250 characters) 1..1 of 1 2 of 2 1/6/2016 7:18 AM#

Page 162Appendix L Form PHMSA F 7000.1: Accident Report for Hazardous Liquid Pipeline Systems#

Page 163NOTICE: This report is required by 49 CFR Part 195. Failure to report can result in a civil penalty not to exceed $100,000 for each violation for each day that such violation persists except that the maximum civil penalty shall not exceed $1,000,000 as provided in 49 USC 60122. OMB NO: 2137-0047 EXPIRATION DATE: 12/31/2016 Original Report Date: 06/17/2015 U.S Department of Transportation Pipeline and Hazardous Materials Safety Administration No. 20150224 - 21010 -------------------------- (DOT Use Only) ACCIDENT REPORT - HAZARDOUS LIQUID PIPELINE SYSTEMS A federal agency may not conduct or sponsor, and a person is not required to respond to, nor shall a person be subject to a penalty for failure to comply with a collection of information subject to the requirements of the Paperwork Reduction Act unless that collection of information displays a current valid OMB Control Number. The OMB Control Number for this information collection is 2137-0047. All responses to the collection of information are mandatory. Send comments regarding this burden or any other aspect of this collection of information, including suggestions for reducing the burden to: Information Collection Clearance Officer, PHMSA, Office of Pipeline Safety (PHP-30) 1200 New Jersey Avenue, SE, Washington, D.C. 20590. INSTRUCTIONS Important: Please read the separate instructions for completing this form before you begin. They clarify the information requested and provide specific examples. If you do not have a copy of the instructions, you can obtain one from the PHMSA Pipeline Safety Community Web Page at http://www.phmsa.dot.gov/pipeline/library/forms. PART A - KEY REPORT INFORMATION Report Type: (select all that apply) Original: Supplemental: Yes Final: Last Revision Date: 12/23/2015 1. Operator's OPS-issued Operator Identification Number (OPID): 300 2. Name of Operator PLAINS PIPELINE, L.P. 3. Address of Operator: 3a. Street Address 333 CLAY STREET, SUITE 1600 3b. City HOUSTON 3c. State Texas 3d. Zip Code 77002 4. Local time (24-hr clock) and date of the Accident: 05/19/2015 10:57 5. Location of Accident: Latitude: 34.462434 Longitude: -120.086714 6. National Response Center Report Number (if applicable): 1116972 7. Local time (24-hr clock) and date of initial telephonic report to the National Response Center (if applicable): 05/19/2015 14:56 8. Commodity released: (select only one, based on predominant volume released) Crude Oil - Specify Commodity Subtype: - If "Other" Subtype, Describe: - If Biofuel/Alternative Fuel and Commodity Subtype is Ethanol Blend, then % Ethanol Blend: - If Biofuel/Alternative Fuel and Commodity Subtype is Biodiesel, then Biodiesel Blend e.g. B2, B20, B100 9. Estimated volume of commodity released unintentionally (Barrels): 2,934.00 10. Estimated volume of intentional and/or controlled release/blowdown (Barrels): 11. Estimated volume of commodity recovered (Barrels): 1,100.00 12. Were there fatalities? No - If Yes, specify the number in each category: 12a. Operator employees 12b. Contractor employees working for the Operator 12c. Non-Operator emergency responders 12d. Workers working on the right-of-way, but NOT associated with this Operator 12e. General public 12f. Total fatalities (sum of above) 13. Were there injuries requiring inpatient hospitalization? No - If Yes, specify the number in each category: 13a. Operator employees 13b. Contractor employees working for the Operator 13c. Non-Operator emergency responders 13d. Workers working on the right-of-way, but NOT associated with this Operator 13e. General public Form PHMSA F 7000.1#

Page 16413f. Total injuries (sum of above) 14. Was the pipeline/facility shut down due to the Accident? Yes - If No, Explain: - If Yes, complete Questions 14a and 14b: (use local time, 24-hr clock) 14a. Local time and date of shutdown: 05/19/2015 11:30 14b. Local time pipeline/facility restarted: - Still shut down? (* Supplemental Report Required) Yes 15. Did the commodity ignite? No 16. Did the commodity explode? No 17. Number of general public evacuated: 1 18. Time sequence (use local time, 24-hour clock): 18a. Local time Operator identified Accident - effective 7- 2014 changed to "Local time Operator identified failure": 05/19/2015 13:27 18b. Local time Operator resources arrived on site: 05/19/2015 13:27 PART B - ADDITIONAL LOCATION INFORMATION 1. Was the origin of the Accident onshore? Yes If Yes, Complete Questions (2-12) If No, Complete Questions (13-15) - If Onshore: 2. State: California 3. Zip Code: 93117 4. City Goleta 5. County or Parish Santa Barbara 6. Operator-designated location: Milepost/Valve Station Specify: 4 7. Pipeline/Facility name: Las Flores to Gaviota 24" 8. Segment name/ID: Line 901 9. Was Accident on Federal land, other than the Outer Continental Shelf (OCS)? No 10. Location of Accident: Pipeline Right-of-way 11. Area of Accident (as found): Underground Specify: Under soil - If Other, Describe: Depth-of-Cover (in): 56 12. Did Accident occur in a crossing? No - If Yes, specify type below: - If Bridge crossing – Cased/ Uncased: - If Railroad crossing – Cased/ Uncased/ Bored/drilled - If Road crossing – Cased/ Uncased/ Bored/drilled - If Water crossing – Cased/ Uncased - Name of body of water, if commonly known: - Approx. water depth (ft) at the point of the Accident: - Select: - If Offshore: 13. Approximate water depth (ft) at the point of the Accident: 14. Origin of Accident: - In State waters - Specify: - State: - Area: - Block/Tract #: - Nearest County/Parish: - On the Outer Continental Shelf (OCS) - Specify: - Area: - Block #: 15. Area of Accident: PART C - ADDITIONAL FACILITY INFORMATION 1. Is the pipeline or facility: Interstate 2. Part of system involved in Accident: Onshore Pipeline, Including Valve Sites - If Onshore Breakout Tank or Storage Vessel, Including Attached Appurtenances, specify: 3. Item involved in Accident: Pipe - If Pipe, specify: Pipe Body 3a. Nominal diameter of pipe (in): 24 Form PHMSA F 7000.1#

Page 1653b. Wall thickness (in): 3c. SMYS (Specified Minimum Yield Strength) of pipe (psi): 3d. Pipe specification: 3e. Pipe Seam , specify: - If Other, Describe: 3f. Pipe manufacturer: 3g. Year of manufacture: 3h. Pipeline coating type at point of Accident, specify: - If Other, Describe: - If Weld, including heat-affected zone, specify. 3a through 3h above are required: If Pipe Girth Weld, - If Other, Describe: - If Valve, specify: - If Mainline, specify: - If Other, Describe: 3i. Manufactured by: 3j. Year of manufacture: - If Tank/Vessel, specify: - If Other - Describe: - If Other, describe: 4. Year item involved in Accident was installed: 5. Material involved in Accident: - If Material other than Carbon Steel, specify: 6. Type of Accident Involved: - If Mechanical Puncture – Specify Approx. size: in. (axial) by in. (circumferential) - If Leak - Select Type: - If Other, Describe: - If Rupture - Select Orientation: - If Other, Describe: Approx. size: in. (widest opening) by in. (length circumferentially or axially) - If Other – Describe: PART D - ADDITIONAL CONSEQUENCE INFORMATION 1. Wildlife impact: 1a. If Yes, specify all that apply: - Fish/aquatic - Birds - Terrestrial 2. Soil contamination: 3. Long term impact assessment performed or planned: 4. Anticipated remediation: 4a. If Yes, specify all that apply: - Surface water - Groundwater - Soil - Vegetation - Wildlife 5. Water contamination: 5a. If Yes, specify all that apply: - Ocean/Seawater - Surface - Groundwater - Drinking water: (Select one or both) - Private Well - Public Water Intake 5b. Estimated amount released in or reaching water (Barrels): 5c. Name of body of water, if commonly known: 6. At the location of this Accident, had the pipeline segment or facility been identified as one that "could affect" a High Consequence Area (HCA) as determined in the Operator's Integrity Management Program? 7. Did the released commodity reach or occur in one or more High Consequence Area (HCA)? 7a. If Yes, specify HCA type(s): (Select all that apply) - Commercially Navigable Waterway: Was this HCA identified in the "could affect" determination for this Accident site in the Operator's Form PHMSA F 7000.1 .344 65,000 X-65 Longitudinal ERW - High Frequency Nippon Steel 1986 Coal Tar 1990 Carbon Steel Leak Other Narrow slit opening. Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes 500.00 Pacific Ocean. Yes Yes Yes Yes#

Page 166Integrity Management Program? - High Population Area: Was this HCA identified in the "could affect" determination for this Accident site in the Operator's Integrity Management Program? - Other Populated Area Was this HCA identified in the "could affect" determination for this Accident site in the Operator's Integrity Management Program? - Unusually Sensitive Area (USA) - Drinking Water Was this HCA identified in the "could affect" determination for this Accident site in the Operator's Integrity Management Program? - Unusually Sensitive Area (USA) - Ecological Yes Was this HCA identified in the "could affect" determination for this Accident site in the Operator's Integrity Management Program? Yes 8. Estimated cost to Operator – effective 12-2012, changed to "Estimated Property Damage": 8a. Estimated cost of public and non-Operator private property damage paid/reimbursed by the Operator – effective 12-2012, "paid/reimbursed by the Operator" removed $ 0 8b. Estimated cost of commodity lost $ 144,000 8c. Estimated cost of Operator's property damage & repairs $ 9,868,173 8d. Estimated cost of Operator's emergency response $ 90,701,042 8e. Estimated cost of Operator's environmental remediation $ 22,421,933 8f. Estimated other costs $ 19,796,736 Describe: Goverment Agency Costs and Media Relations. 8g. Estimated total costs (sum of above) – effective 12-2012, changed to "Total estimated property damage (sum of above)" $ 142,931,884 PART E - ADDITIONAL OPERATING INFORMATION 1. Estimated pressure at the point and time of the Accident (psig): 750.00 2. Maximum Operating Pressure (MOP) at the point and time of the Accident (psig): 1,056.00 3. Describe the pressure on the system or facility relating to the Accident (psig): Pressure did not exceed MOP 4. Not including pressure reductions required by PHMSA regulations (such as for repairs and pipe movement), was the system or facility relating to the Accident operating under an established pressure restriction with pressure limits below those normally allowed by the MOP? No - If Yes, Complete 4.a and 4.b below: 4a. restriction? Did the pressure exceed this established pressure 4b. State? Was this pressure restriction mandated by PHMSA or the 5. 2? Was "Onshore Pipeline, Including Valve Sites" OR "Offshore Pipeline, Including Riser and Riser Bend" selected in PART C, Question Yes - If Yes - (Complete 5a. – 5f below) effective 12-2012, changed to "(Complete 5.a – 5.e below)" 5a. Type of upstream valve used to initially isolate release source: Remotely Controlled 5b. Type of downstream valve used to initially isolate release source: Check Valve 5c. Length of segment isolated between valves (ft): 56,752 5d. Is the pipeline configured to accommodate internal inspection tools? Yes - If No, Which physical features limit tool accommodation? (select all that apply) - Changes in line pipe diameter - Presence of unsuitable mainline valves - Tight or mitered pipe bends - Other passage restrictions (i.e. unbarred tee's, projecting instrumentation, etc.) - Extra thick pipe wall (applicable only for magnetic flux leakage internal inspection tools) - Other - - If Other, Describe: 5e. For this pipeline, are there operational factors which significantly complicate the execution of an internal inspection tool run? No - If Yes, Which operational factors complicate execution? (select all that apply) Form PHMSA F 7000.1#

Page 167- Excessive debris or scale, wax, or other wall buildup - Low operating pressure(s) - Low flow or absence of flow - Incompatible commodity - Other - - If Other, Describe: 5f. Function of pipeline system: 6. Was a Supervisory Control and Data Acquisition (SCADA)-based system in place on the pipeline or facility involved in the Accident? If Yes - 6a. Was it operating at the time of the Accident? 6b. Was it fully functional at the time of the Accident? 6c. Did SCADA-based information (such as alarm(s), alert(s), event(s), and/or volume calculations) assist with the detection of the Accident? 6d. Did SCADA-based information (such as alarm(s), alert(s), event(s), and/or volume calculations) assist with the confirmation of the Accident? 7. Was a CPM leak detection system in place on the pipeline or facility involved in the Accident? - If Yes: 7a. Was it operating at the time of the Accident? 7b. Was it fully functional at the time of the Accident? 7c. Did CPM leak detection system information (such as alarm(s), alert(s), event(s), and/or volume calculations) assist with the detection of the Accident? 7d. Did CPM leak detection system information (such as alarm(s), alert(s), event(s), and/or volume calculations) assist with the confirmation of the Accident? 8. How was the Accident initially identified for the Operator? - If Other, Specify: 8a. If "Controller", "Local Operating Personnel", including contractors", "Air Patrol", or "Ground Patrol by Operator or its contractor" is selected in Question 8, specify: 9. Was an investigation initiated into whether or not the controller(s) or control room issues were the cause of or a contributing factor to the Accident? - If No, the Operator did not find that an investigation of the controller(s) actions or control room issues was necessary due to: (provide an explanation for why the operator did not investigate) - If Yes, specify investigation result(s): (select all that apply) - Investigation reviewed work schedule rotations, continuous hours of service (while working for the Operator), and other factors associated with fatigue - Investigation did NOT review work schedule rotations, continuous hours of service (while working for the Operator), and other factors associated with fatigue Provide an explanation for why not: - Investigation identified no control room issues - Investigation identified no controller issues - Investigation identified incorrect controller action or controller error - Investigation identified that fatigue may have affected the controller(s) involved or impacted the involved controller(s) response - Investigation identified incorrect procedures - Investigation identified incorrect control room equipment operation - Investigation identified maintenance activities that affected control room operations, procedures, and/or controller response - Investigation identified areas other than those above: Describe: PART F - DRUG & ALCOHOL TESTING INFORMATION Form PHMSA F 7000.1 > 20% SMYS Regulated Trunkline/Transmission Yes Yes Yes Yes Yes Yes Yes Yes No No Local Operating Personnel, including contractors Operator employee Yes, specify investigation result(s): (select all that apply) Yes Yes Yes Yes Yes Investigation identified that a minor procedure was not followed. This failure was not a cause of or contributing factor to the Accident. Additional training on this procedure has been provided.#

Page 1681. As a result of this Accident, were any Operator employees tested under the post-accident drug and alcohol testing requirements of DOT's Drug & Alcohol Testing regulations? Yes - If Yes: 1a. Specify how many were tested: 1 1b. Specify how many failed: 0 2. As a result of this Accident, were any Operator contractor employees tested under the post-accident drug and alcohol testing requirements of DOT's Drug & Alcohol Testing regulations? No - If Yes: 2a. Specify how many were tested: 2b. Specify how many failed: PART G – APPARENT CAUSE Select only one box from PART G in shaded column on left representing the APPARENT Cause of the Accident, and answer the questions on the right. Describe secondary, contributing or root causes of the Accident in the narrative (PART H). Apparent Cause: G1 - Corrosion Failure G1 - Corrosion Failure - only one sub-cause can be picked from shaded left-hand column Corrosion Failure – Sub-Cause: External Corrosion - If External Corrosion: 1. Results of visual examination: Other - If Other, Describe: Corrosion under insulation. 2. Type of corrosion: (select all that apply) - Galvanic - Atmospheric - Stray Current - Microbiological - Selective Seam - Other: Yes - If Other, Describe: Corrosion under insulation. 3. The type(s) of corrosion selected in Question 2 is based on the following: (select all that apply) - Field examination - Determined by metallurgical analysis Yes - Other: - If Other, Describe: 4. Was the failed item buried under the ground? Yes - If Yes : 4a. Was failed item considered to be under cathodic protection at the time of the Accident? Yes If Yes - Year protection started: 1990 4b. Was shielding, tenting, or disbonding of coating evident at the point of the Accident? Yes 4c. Has one or more Cathodic Protection Survey been conducted at the point of the Accident? Yes If "Yes, CP Annual Survey" – Most recent year conducted: 2015 If "Yes, Close Interval Survey" – Most recent year conducted: 2015 If "Yes, Other CP Survey" – Most recent year conducted: - If No: 4d. Was the failed item externally coated or painted? 5. Was there observable damage to the coating or paint in the vicinity of the corrosion? Yes - If Internal Corrosion: 6. Results of visual examination: - Other: 7. Type of corrosion (select all that apply): - - Corrosive Commodity - Water drop-out/Acid - Microbiological - Erosion - Other: - If Other, Describe: 8. The cause(s) of corrosion selected in Question 7 is based on the following (select all that apply): - - Field examination - Determined by metallurgical analysis - Other: Form PHMSA F 7000.1#

Page 169- If Other, Describe: 9. Location of corrosion (select all that apply): - - Low point in pipe - Elbow - Other: - If Other, Describe: 10. Was the commodity treated with corrosion inhibitors or biocides? 11. Was the interior coated or lined with protective coating? 12. Were cleaning/dewatering pigs (or other operations) routinely utilized? 13. Were corrosion coupons routinely utilized? Complete the following if any Corrosion Failure sub-cause is selected AND the "Item Involved in Accident" (from PART C, Question 3) is Tank/Vessel. 14. List the year of the most recent inspections: 14a. API Std 653 Out-of-Service Inspection - No Out-of-Service Inspection completed 14b. API Std 653 In-Service Inspection - No In-Service Inspection completed Complete the following if any Corrosion Failure sub-cause is selected AND the "Item Involved in Accident" (from PART C, Question 3) is Pipe or Weld. 15. Has one or more internal inspection tool collected data at the point of the Accident? Yes 15a. If Yes, for each tool used, select type of internal inspection tool and indicate most recent year run: - - Magnetic Flux Leakage Tool Yes Most recent year: 2015 - Ultrasonic Most recent year: - Geometry Most recent year: - Caliper Yes Most recent year: 2015 - Crack Most recent year: - Hard Spot Most recent year: - Combination Tool Yes Most recent year: 2015 - Transverse Field/Triaxial Most recent year: - Other Most recent year: Describe: 16. Has one or more hydrotest or other pressure test been conducted since original construction at the point of the Accident? No If Yes - Most recent year tested: Test pressure: 17. Has one or more Direct Assessment been conducted on this segment? No - If Yes, and an investigative dig was conducted at the point of the Accident:: Most recent year conducted: - If Yes, but the point of the Accident was not identified as a dig site: Most recent year conducted: 18. Has one or more non-destructive examination been conducted at the point of the Accident since January 1, 2002? No 18a. If Yes, for each examination conducted since January 1, 2002, select type of non-destructive examination and indicate most recent year the examination was conducted: - Radiography Most recent year conducted: - Guided Wave Ultrasonic Most recent year conducted: - Handheld Ultrasonic Tool Most recent year conducted: - Wet Magnetic Particle Test Most recent year conducted: - Dry Magnetic Particle Test Most recent year conducted: - Other Most recent year conducted: Describe: Form PHMSA F 7000.1#

Page 170G2 - Natural Force Damage - only one sub-cause can be picked from shaded left-handed column Natural Force Damage – Sub-Cause: - If Earth Movement, NOT due to Heavy Rains/Floods: 1. Specify: - If Other, Describe: - If Heavy Rains/Floods: 2. Specify: - If Other, Describe: - If Lightning: 3. Specify: - If Temperature: 4. Specify: - If Other, Describe: - If Other Natural Force Damage: 5. Describe: Complete the following if any Natural Force Damage sub-cause is selected. 6. Were the natural forces causing the Accident generated in conjunction with an extreme weather event? 6a. If Yes, specify: (select all that apply) - Hurricane - Tropical Storm - Tornado - Other - If Other, Describe: G3 - Excavation Damage - only one sub-cause can be picked from shaded left-hand column Excavation Damage – Sub-Cause: - If Previous Damage due to Excavation Activity: Complete Questions 1-5 ONLY IF the "Item Involved in Accident" (from PART C, Question 3) is Pipe or Weld. 1. Has one or more internal inspection tool collected data at the point of the Accident? 1a. If Yes, for each tool used, select type of internal inspection tool and indicate most recent year run: - - Magnetic Flux Leakage Most recent year conducted: - Ultrasonic Most recent year conducted: - Geometry Most recent year conducted: - Caliper Most recent year conducted: - Crack Most recent year conducted: - Hard Spot Most recent year conducted: - Combination Tool Most recent year conducted: - Transverse Field/Triaxial Most recent year conducted: - Other Most recent year conducted: Describe: 2. Do you have reason to believe that the internal inspection was completed BEFORE the damage was sustained? 3. Has one or more hydrotest or other pressure test been conducted since original construction at the point of the Accident? - If Yes: Most recent year tested: Test pressure (psig): 4. Has one or more Direct Assessment been conducted on the pipeline segment? - If Yes, and an investigative dig was conducted at the point of the Accident: Most recent year conducted: - If Yes, but the point of the Accident was not identified as a dig site: Most recent year conducted: 5. Has one or more non-destructive examination been conducted at the point of the Accident since January 1, 2002? Form PHMSA F 7000.1#

Page 1715a. If Yes, for each examination, conducted since January 1, 2002, select type of non-destructive examination and indicate most recent year the examination was conducted: - Radiography Most recent year conducted: - Guided Wave Ultrasonic Most recent year conducted: - Handheld Ultrasonic Tool Most recent year conducted: - Wet Magnetic Particle Test Most recent year conducted: - Dry Magnetic Particle Test Most recent year conducted: - Other Most recent year conducted: Describe: Complete the following if Excavation Damage by Third Party is selected as the sub-cause. 6. Did the operator get prior notification of the excavation activity? 6a. If Yes, Notification received from: (select all that apply) - - One-Call System - Excavator - Contractor - Landowner Complete the following mandatory CGA-DIRT Program questions if any Excavation Damage sub-cause is selected. 7. Do you want PHMSA to upload the following information to CGA- DIRT (www.cga-dirt.com)? 8. Right-of-Way where event occurred: (select all that apply) - - Public - If "Public", Specify: - Private - If "Private", Specify: - Pipeline Property/Easement - Power/Transmission Line - Railroad - Dedicated Public Utility Easement - Federal Land - Data not collected - Unknown/Other 9. Type of excavator: 10. Type of excavation equipment: 11. Type of work performed: 12. Was the One-Call Center notified? 12a. If Yes, specify ticket number: 12b. If this is a State where more than a single One-Call Center exists, list the name of the One-Call Center notified: 13. Type of Locator: 14. Were facility locate marks visible in the area of excavation? 15. Were facilities marked correctly? 16. Did the damage cause an interruption in service? 16a. If Yes, specify duration of the interruption (hours) 17. Description of the CGA-DIRT Root Cause (select only the one predominant first level CGA-DIRT Root Cause and then, where available as a choice, the one predominant second level CGA-DIRT Root Cause as well): Root Cause: - If One-Call Notification Practices Not Sufficient, specify: - If Locating Practices Not Sufficient, specify: - If Excavation Practices Not Sufficient, specify: - If Other/None of the Above, explain: G4 - Other Outside Force Damage - only one sub-cause can be selected from the shaded left-hand column Other Outside Force Damage – Sub-Cause: - If Damage by Car, Truck, or Other Motorized Vehicle/Equipment NOT Engaged in Excavation: 1. Vehicle/Equipment operated by: - If Damage by Boats, Barges, Drilling Rigs, or Other Maritime Equipment or Vessels Set Adrift or Which Have Otherwise Lost Their Mooring: 2. Select one or more of the following IF an extreme weather event was a factor: - Hurricane - Tropical Storm - Tornado Form PHMSA F 7000.1#

Page 172- Heavy Rains/Flood - Other - If Other, Describe: - If Previous Mechanical Damage NOT Related to Excavation: Complete Questions 3-7 ONLY IF the "Item Involved in Accident" (from PART C, Question 3) is Pipe or Weld. 3. Has one or more internal inspection tool collected data at the point of the Accident? 3a. If Yes, for each tool used, select type of internal inspection tool and indicate most recent year run: - Magnetic Flux Leakage Most recent year conducted: - Ultrasonic Most recent year conducted: - Geometry Most recent year conducted: - Caliper Most recent year conducted: - Crack Most recent year conducted: - Hard Spot Most recent year conducted: - Combination Tool Most recent year conducted: - Transverse Field/Triaxial Most recent year conducted: - Other Most recent year conducted: Describe: 4. Do you have reason to believe that the internal inspection was completed BEFORE the damage was sustained? 5. Has one or more hydrotest or other pressure test been conducted since original construction at the point of the Accident? - If Yes: Most recent year tested: Test pressure (psig): 6. Has one or more Direct Assessment been conducted on the pipeline segment? - If Yes, and an investigative dig was conducted at the point of the Accident: Most recent year conducted: - If Yes, but the point of the Accident was not identified as a dig site: Most recent year conducted: 7. Has one or more non-destructive examination been conducted at the point of the Accident since January 1, 2002? 7a. If Yes, for each examination conducted since January 1, 2002, select type of non-destructive examination and indicate most recent year the examination was conducted: - Radiography Most recent year conducted: - Guided Wave Ultrasonic Most recent year conducted: - Handheld Ultrasonic Tool Most recent year conducted: - Wet Magnetic Particle Test Most recent year conducted: - Dry Magnetic Particle Test Most recent year conducted: - Other Most recent year conducted: Describe: - If Intentional Damage: 8. Specify: - If Other, Describe: - If Other Outside Force Damage: 9. Describe: G5 - Material Failure of Pipe or Weld - only one sub-cause can be selected from the shaded left-hand column Use this section to report material failures ONLY IF the "Item Involved in Accident" (from PART C, Question 3) is "Pipe" or "Weld." Material Failure of Pipe or Weld – Sub-Cause: 1. The sub-cause shown above is based on the following: (select all that apply) Form PHMSA F 7000.1#

Page 173- Field Examination - Determined by Metallurgical Analysis - Other Analysis - If "Other Analysis", Describe: - Sub-cause is Tentative or Suspected; Still Under Investigation (Supplemental Report required) - If Construction, Installation, or Fabrication-related Or If Original Manufacturing-related: 2. List contributing factors: (select all that apply) - Fatigue or Vibration-related Specify: - If Other, Describe: - Mechanical Stress: - Other - If Other, Describe: - If Environmental Cracking-related: 3. Specify: - If Other - Describe: Complete the following if any Material Failure of Pipe or Weld sub-cause is selected. 4. Additional factors: (select all that apply): - Dent - Gouge - Pipe Bend - Arc Burn - Crack - Lack of Fusion - Lamination - Buckle - Wrinkle - Misalignment - Burnt Steel - Other: - If Other, Describe: 5. Has one or more internal inspection tool collected data at the point of the Accident? 5a. If Yes, for each tool used, select type of internal inspection tool and indicate most recent year run: - Magnetic Flux Leakage Most recent year run: - Ultrasonic Most recent year run: - Geometry Most recent year run: - Caliper Most recent year run: - Crack Most recent year run: - Hard Spot Most recent year run: - Combination Tool Most recent year run: - Transverse Field/Triaxial Most recent year run: - Other Most recent year run: Describe: 6. Has one or more hydrotest or other pressure test been conducted since original construction at the point of the Accident? - If Yes: Most recent year tested: Test pressure (psig): 7. Has one or more Direct Assessment been conducted on the pipeline segment? - If Yes, and an investigative dig was conducted at the point of the Accident - Most recent year conducted: - If Yes, but the point of the Accident was not identified as a dig site - Most recent year conducted: 8. Has one or more non-destructive examination(s) been conducted at the point of the Accident since January 1, 2002? 8a. If Yes, for each examination conducted since January 1, 2002, select type of non-destructive examination and indicate most recent year the examination was conducted: - Form PHMSA F 7000.1#

Page 174- Radiography Most recent year conducted: - Guided Wave Ultrasonic Most recent year conducted: - Handheld Ultrasonic Tool Most recent year conducted: - Wet Magnetic Particle Test Most recent year conducted: - Dry Magnetic Particle Test Most recent year conducted: - Other Most recent year conducted: Describe: G6 – Equipment Failure - only one sub-cause can be selected from the shaded left-hand column Equipment Failure – Sub-Cause: - If Malfunction of Control/Relief Equipment: 1. Specify: (select all that apply) - - Control Valve - Instrumentation - SCADA - Communications - Block Valve - Check Valve - Relief Valve - Power Failure - Stopple/Control Fitting - ESD System Failure - Other - If Other – Describe: - If Pump or Pump-related Equipment: 2. Specify: - If Other – Describe: - If Threaded Connection/Coupling Failure: 3. Specify: - If Other – Describe: - If Non-threaded Connection Failure: 4. Specify: - If Other – Describe: - If Other Equipment Failure: 5. Describe: Complete the following if any Equipment Failure sub-cause is selected. 6. Additional factors that contributed to the equipment failure: (select all that apply) - Excessive vibration - Overpressurization - No support or loss of support - Manufacturing defect - Loss of electricity - Improper installation - Mismatched items (different manufacturer for tubing and tubing fittings) - Dissimilar metals - Breakdown of soft goods due to compatibility issues with transported commodity - Valve vault or valve can contributed to the release - Alarm/status failure - Misalignment - Thermal stress - Other - If Other, Describe: G7 - Incorrect Operation - only one sub-cause can be selected from the shaded left-hand column Incorrect Operation – Sub-Cause: Form PHMSA F 7000.1#

Page 175- If Tank, Vessel, or Sump/Separator Allowed or Caused to Overfill or Overflow 1. Specify: - If Other, Describe: - If Other Incorrect Operation 2. Describe: Complete the following if any Incorrect Operation sub-cause is selected. 3. Was this Accident related to (select all that apply): - - Inadequate procedure - No procedure established - Failure to follow procedure - Other: - If Other, Describe: 4. What category type was the activity that caused the Accident? 5. Was the task(s) that led to the Accident identified as a covered task in your Operator Qualification Program? 5a. If Yes, were the individuals performing the task(s) qualified for the task(s)? G8 - Other Accident Cause - only one sub-cause can be selected from the shaded left-hand column Other Accident Cause – Sub-Cause: - If Miscellaneous: 1. Describe: - If Unknown: 2. Specify: PART H - NARRATIVE DESCRIPTION OF THE ACCIDENT Crude oil was released from a 24-inch pipeline, located along Highway 101 in Santa Barbara County, California. The released crude reached a culvert which leads to the Pacific Ocean and, as a result, impacted the shoreline and ocean water. The cause of the release is currently under investigation. The pipe has been excavated. The affected portion of pipe was securely packaged to preserve its condition and has been transported to a secure, independent facility for an independent third-party analysis and investigation. A supplemental report will be submitted upon receipt of the third party, metallurgical analysis. In the meantime, Plains personnel are actively engaged in cleanup and environmental remediation efforts. Part A. Question 7. - 14:56 is the time Operator notified the National Response Center (NRC). The NRC was first notified at 12:43 by an unrelated third party. Part A. Question 9. - Answer is a best-estimate as of 6/17/2015. Part A. Question 11.- Response reflects current estimate as of 6/17/2015. The volume of recovered commodity will be revised upward in the supplemental report as more information becomes available. Part A. Question 17. -The number of people evacuated from local State Park campsites is currently undetermined as no estimates are included in the initial first responder reports we have received. We are investigating this further and will revise the Supplemental report as more information becomes available. Part D. Question 8. - Answer reflects estimated costs incurred through 6/16/2015. Supplemental Narratives: Part A, Number 11 and Part D, Number 8 have also been updated to reflect new information as of 7/10/2015. As of 8/4/15 the current estimated release volume remains approx. 2,400 bbls. Preliminary data from the purge activity estimates the release could be potentially 3,400 bbls. While Plains believes the volume estimate listed in Part A, Question 9 best represents the potential discharge volume, we are working with an outside expert to reconcile the differences and will provide additional updates as appropriate. As of 11/24/2015, based on the work performed by our independent third party consultant (i.e. the 'outside expert' mentioned above), our best estimate of the spill volume is 2,934 barrels. The results of the metallurgical analysis of the pipeline segment indicate that the failure occurred at an area of wall thinning from external corrosion that ultimately failed by ductile overload under the imposed operating pressure. The morphology of the external corrosion observed on the pipe section is consistent with corrosion under insulation facilitated by wet-dry cycling. Line 901 remains shut down and subject to Corrective Action Order CPF No. 5-2015-5011H and Amendments. Updated costs for the repair and restart of this line, remains the only outstanding item in order to finalize this 7000-1 form. PART I - PREPARER AND AUTHORIZED SIGNATURE Preparer's Name Chrystah Carter Preparer's Title Compliance Specialist Preparer's Telephone Number 713-993-5080 Preparer's E-mail Address crcarter@paalp.com Preparer's Facsimile Number 713-646-4310 Authorized Signer Name Troy E Valenzuela Form PHMSA F 7000.1#

Page 176Authorized Signer Title VP of Environmental Health and Safety Authorized Signer Telephone Number 713-646-4614 Authorized Signer Email tevalenzuela@paalp.com Date 12/23/2015 Form PHMSA F 7000.1#

Page 177Appendix M Det Norske Veritas (U.S.A.), Inc. (DNV GL): Line 901 Release (5/19/15) Mechanical and Metallurgical Testing#

Page 178Final Report Line 901 Release (5/19/15): Mechanical and Metallurgical Testing Plains All American Pipeline, L.P. Houston, Texas Report No.: OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 179Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Project Name: Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Plains All American Pipeline, P.L. Customer: Contact Person: Date of Issue: September 18, 2015 Project No.: PP136049 Organization Unit: Incident Investigation Report No.: OAPUS309DNOR DET NORSKE VERITAS (U.S.A.), INC. (DNV GL) Materials & Corrosion Technology Center Incident Investigation 5777 Frantz Road Dublin, OH 43017-1886 United States Tel: (614) 761-1214 Fax: (614) 761-1633 www.dnvgl.com Task and Objective: Please see Executive Summary. Prepared by Verified by Approved by David M. Norfleet, Ph.D., P.E. Principal Engineer John A. Beavers, Ph.D., FNACE Director – Incident Investigation Neil G. Thompson, Ph.D., FNACE Senior VP, Pipeline Services ☐ Unrestricted Distribution (internal and external) ☐ Unrestricted Distribution within DNV GL ☐ Limited Distribution within DNV GL after 3 years ☒ No Distribution (confidential) ☐ Secret Keywords Rev. No. Date Reason for Issue: Prepared by: Verified by: Approved by: 0 2015-08-06 First Issue 1 2015-09-18 Final Copyright © DNV GL 2015. All rights reserved. This publication or parts thereof may not be copied, reproduced, or transmitted in any form, or by any means, whether digitally or otherwise without the prior written consent of DNV GL. DNV GL and the Horizon Graphic are trademarks of DNV GL AS. The content of this publication shall be kept confidential by the customer, unless otherwise agreed in writing. Reference to part of this publication, which may lead to misinterpretation, is prohibited. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP ii#

Page 180Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Executive Summary Plains All American Pipeline, L.P. (Plains) retained Det Norske Veritas (U.S.A.), Inc. (DNV GL) to perform a metallurgical analysis and mechanical testing on a section of pipe from Line 901 - Las Flores to Gaviota (L901), 24-inch nominal diameter crude oil pipeline that failued while in service. The failure occurred on May 19, 2015 in Goleta (Santa Barbara County), California at milepost (MP) 4, 33.5 feet downstream (D/S) of the nearest upstream (U/S) girth weld and 4.05 miles D/S of the nearest U/S pump station. A failure of a pipe segment can be characterized either as a leak or a rupture; the failure on L901 is characterized as a leak.1 The section of the pipeline that failed is comprised of 24-inch diameter by 0.344-inch wall thickness, API 5L Grade X65 line pipe steel that contains a high frequency electric resistance welded (ERW) longitudinal seam and was manufactured by Nippon Steel in 1986. The maximum operating pressure (MOP) is 1,3412 pounds per square inch gauge (psig) (72% of the specified minimum yield strength [SMYS]). The pressure at the time of failure was reported by Plains to be 737 psig (39.6% of SMYS) at the failure location and time of failure. The pipeline was installed in 1990 and constructed using pipe that was externally coated with a coal tar urethane coating on the steel substrate, 1.5-inch thick rigid polyurethane foam, and an external polyethylene tape. The pipeline has an impressed current cathodic protection (CP) system with the nearest rectifier located 4.05 miles U/S of the failure location, at the Las Flores Pump Station. A hydrostatic test was performed at the time of commissioning for 8 hours at 1719 psi (Gaviota Station) on November 25th, 1990. In-line inspection (ILI) runs, consisting of deformation and magnetic flux leakage (MFL) tools, were performed in 2007, 2012, and 2015. The failed pipe joint and 5 feet of the U/S and D/S joints were removed from the failure location and delivered to DNV GL in two pipe sections for analysis. Pipe Section 1 (PS 1) was 19.05 feet in length and contained 5.05 feet of the U/S joint, the U/S girth weld, and 1 According to the FRACTURE CONTROL TECHNOLOGY FOR NATURAL GAS PIPELINES CIRCA 2001 (the PRCI report superseding NG-18 Report 208), “The distinction between leak and rupture for the pipeline community is based on the size and configuration of the breach, not how it develops. A “leak” is characterized by a narrow slit-like hole with length less than the diameter, which limits the fluid volume that escapes through the breach. In contrast, a “rupture” involves a longer, open hole that can be bulged over its length, which is on the order of a diameter or longer and can permit escape of a significant fluid volume.” Similarly, the research performed as part of the historical NG-18 work identified empirical equations to predict the length at which a feature will propagate versus pop through and arrest; the leak/rupture length. Based on these calculations and visual observations, the length of the feature is consistent with a leak, arresting within the corrosion feature, and did not propagate outside of the feature into nominal wall-thickness pipe. 2 Theoretical maximum operating pressure at the lowest elevation using the lowest pressure of either 80% of the commissioning hydro-test pressure, the 72% of SMYS, or the lowest component rating along the line segment. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP iii#

Page 181Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing 15 feet of the failure joint located U/S of the failure. Pipe Section 2 (PS 2) was 31.06 feet in length and contained the failure location, the D/S girth weld, a 2013 composite repair sleeve, and 5 feet of the D/S joint. The objective of the analysis was to determine the metallurgical (or immediate) cause of the failure. Metallurgical Cause: The results of the metallurgical analysis indicate that the failure occurred at an area of wall thinning from external corrosion that ultimately failed by ductile overload under the imposed operating pressure. The morphology of the external corrosion observed on the pipe section is consistent with corrosion under insulation facilitated by wet-dry cycling. The following steps were performed for this analysis. The pipe sections were visually inspected and photographed. The external polyethylene (PE) tape was removed from PS 1 and PS 2 and visually inspected and photographed. The external pipe surfaces (with insulation) were laser scanned using a FaroArmTM to produce digital maps. The insulation from PS 2 was then removed and the pipe was visually inspected and photographed. The coal tar coating was then removed around the failure location, areas of corrosion, and at the ends of each pipe section. Wall thicknesses, diameters, and circumferences were measured at various locations on PS 1 and PS 2 where coating was removed and there was no measurable corrosion. Corrosion products were collected from PS 2 for characterization. Analyses performed on these products included: (1) pH testing using litmus paper, (2) spot tests for carbonates and sulfides using 2-normal hydrochloric acid (2N HCl), (3) elemental analyses using energy dispersive spectroscopy (EDS) with a scanning electron microscope (SEM) and (4) compound identification using x-ray diffraction (XRD). Swab samples were also obtained for bacteria analyses at two locations; an area of external corrosion and an area where the coating was disbonded but there was negligible external corrosion. Separate swab samples were taken for serial dilution and microscopic analysis. Liquid culture media for acid-producing bacteria (APB), sulfate-reducing bacteria (SRB), nitrate-reducing bacteria (NRB), aerobic bacteria (AERO), anaerobic bacteria (ANA), and iron-related bacteria (IRB) was used for the serial dilutions to evaluate growth of various types of bacteria. A five vial serial dilution (1:10,000) was performed using each type of media. Coupons containing the failure location and areas of corrosion were cut from PS 2 using cold-cutting techniques. Coupon 1 contained the failure location and was a full ring section removed between 30.66 and 35.95 feet from the U/S GW. Coupon 2 contained external corrosion features further U/S from the failure location and was removed between 14.00 and 20.60 feet from the U/S GW; between the 4- and 8-o’clock orientations. The internal DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP iv#

Page 182Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing and external surfaces were visually inspected and photographed. Where necessary, the samples were cleaned using a degreaser (LPS Presolve®) and acetone. Ultrasonic testing (UT) was performed on the samples removed from PS 2, using a 1-inch by 1-inch grid spacing, to produce a thickness map. The external and internal pipe surfaces of these coupons were laser scanned to produce a thickness contour dataset. Magnetic particle inspection (MPI) was performed on the external and internal pipe surfaces of the coupon containing the failure location. The fracture surfaces were cleaned with methanol and acetone, optically examined, and photographed. Samples were then removed from one of the mating fracture surfaces, cleaned with Rhodine inhibited HCl solution and ENPREP® 214 to remove corrosion products, and examined at high magnifications in an SEM to document the fracture morphology. Transverse cross sections were removed from the suspected failure origin, an area of corrosion further U/S, and across the longitudinal seam weld of the failure joint. The transverse cross sections were mounted, polished, and etched. Light photomicrographs were taken to document the fracture and corrosion morphologies and steel microstructure. In addition, corrosion products collected from an area adjacent to the failure location and from areas of corrosion further U/S of the failure were mounted in cross-section and polished. Light photomicrographs were taken to document the corrosion product morphologies. Elemental analysis using EDS was performed to identify the elemental constituents of each. Soil analyses were conducted on a sample removed (in the field) approximately 8 feet U/S of the U/S girth weld (GW). The soil was tested for resistivity, moisture content, pH, total acidity, total alkalinity, concentration of soluble anions and cations, total dissolved solids, and linear polarization resistance. Mechanical (duplicate tensile tests and full Charpy V-notch [CVN]) curves) testing was performed on specimens removed from the failed pipe joint and U/S and D/S joints to determine the tensile and fracture toughness properties. Chemical analyses were performed on a steel sample removed from the failed pipe joint and U/S and D/S joints to determine the compositions. CorLASTM calculations were performed to estimate the failure pressure based on the pipe geometry, base metal mechanical properties, and the measured flaw profile. This value was compared with the estimated pressure at the failure location. External corrosion was identified at several locations along the bottom of the failed pipe section, including the corrosion feature that ultimately failed on May 19, 2015. The corrosion features were associated with thick layered deposits and areas of compression and water saturation of the thermal insulation. The characteristics of the failure are consistent DNV GL – OAPUS309DNOR (PP136049) v September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 183Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing with corrosion under insulation in the presence of wet-dry cycling. Summary of Observations • The failure was associated with an external corrosion feature located 33.50 feet from the upstream girth weld, at the 4:24 orientation (center of corrosion feature). • The dimensions of the corrosion feature were 12.1 inches axially by 7.4 inches circumferentially. The maximum depth, as measured using laser scan data, was 0.318 inches or 89% of the measured wall thickness (0.359 inches). • The failure opening was 6.6 inches in axial length, with the upstream and downstream ends located 33.35 and 33.9 feet from the upstream girth weld. • The maximum circumferential dimension of the failure opening was 1.14 inches, approximately 33.45 feet from the upstream girth weld, at the 4:15 orientation. • The fracture surfaces exhibited ductile overload. • Cracking and wrinkling were observed within the polyethylene tape. • Compression was observed within the polyurethane insulation at areas on the bottom of the pipe. These areas were saturated with moisture. • Disbondment of the coal tar coating was observed on the bottom of the pipe along the length of Pipe Section 2. • External corrosion features, including the feature associated with the failure, were identified at or adjacent to areas of saturated, compressed insulation. • The corrosion products were rigid, non-friable, and, at some locations, well adhered to the pipe section. The products consist of alternating layers of goethite and magnetite. • There is no strong evidence to indicate that microbiological influenced corrosion (MIC) contributed to the observed corrosion. • No evidence of internal corrosion was observed along the length of the pipe sections inspected. • The average yield strength (YS) for the failure joint is marginally lower than the minimum YS requirements for API 5L X65 line pipe steel of 65.0 ksi. The average is based on two tests values; one slightly higher (65.2 ksi) and one slightly lower (64.4 ksi) than the requirement. The average ultimate tensile strength (UTS) of the failure joint meets the minimum UTS requirements for API 5L X65 line pipe steel of 80 ksi. • The Charpy V-notch (CVN) properties of the base metal are typical for the vintage and grade of line pipe steel. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP vi#

Page 184Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing • The chemical composition of the base metal meets requirements for the vintage and grade of line pipe steel. • The microstructure of the base metal is typical for the vintage and grade of line pipe steel. • The CorLAS™ predicted failure pressure for the failed joint was calculated to be approximately 760 psig, which is in very good agreement with reported pressure at the failure location and time of failure (737 psig). DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP vii#

Page 185Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Table of Contents 1.0 BACKGROUND ........................................................................................ 1 2.0 TECHNICAL APPROACH ............................................................................ 2 3.0 RESULTS AND DISCUSSION...................................................................... 4 3.1 Visual Examination .................................................................................. 4 3.1.1 External Polyethylene Tape ....................................................................... 5 3.1.2 Rigid Polyurethane Insulation .................................................................... 6 3.1.3 Coal Tar Urethane Coating ........................................................................ 7 3.1.4 Carbon Steel Line Pipe ............................................................................. 7 3.1.5 Composite “Armor Plate” Sleeve ................................................................ 8 3.2 3D Laser Scanning .................................................................................. 9 3.3 Ultrasonic Testing ...................................................................................10 3.4 Magnetic Particle Inspection .....................................................................10 3.5 Fractographic Examination .......................................................................10 3.5.1 Optical ..................................................................................................10 3.5.2 Scanning Electron Microscopy ..................................................................11 3.5.3 Fracture/Corrosion Profile ........................................................................11 3.6 Metallographic Examination .....................................................................12 3.7 Solid Sampling of Corrosion Products ........................................................13 3.7.1 pH Testing and Qualitative Spot Testing ....................................................13 3.7.2 X-ray Diffraction ....................................................................................14 3.7.3 Energy Dispersive Spectroscopy ...............................................................14 3.8 Microbiological Analyses ..........................................................................15 3.8.1 Serial Dilution – Liquid Culture Media ........................................................15 3.8.2 Microscopic Examination for Total Bacteria .................................................15 3.9 Soil Testing ...........................................................................................15 3.10 Mechanical Testing .................................................................................16 3.11 Chemical Analysis ..................................................................................17 3.12 Failure Pressure Analysis .........................................................................17 4.0 DISCUSSION AND CONCLUSIONS ............................................................18 DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP viii#

Page 186Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Appendices Appendix A – CorLAS™ DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP ix#

Page 187Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing List of Tables Table 1. Summary of the locations and dimensions of corrosion features identified during the laboratory examination on the external surface of PS 2. ................................................................................................21 Table 2. Results of diameter measurements performed on PS 1 and PS 2 using Pi tape and a tape measure. ............................................................22 Table 3. Results of wall thickness measurements performed on PS 1 and PS 2. ........................................................................................................22 Table 4. Results of thickness measurements performed adjacent (~0.100 inches circumferentially) to the failure opening using the laser scan dataset. See Figure 32. ..........................................................................23 Table 5. Results of elemental analyses, using EDS, performed on corrosion products from Feature 1 and Feature 4 compared to ideal chemistry compositions of goethite and magnetite; values presented in mass percent (wt.%). .....................................................................................24 Table 6. Results of bacteria analyses performed on swabs taken, over an ~1 cm2 area, from the external surface of Pipe Section 2 at Feature 1 on the failure joint and at an area of disbonded coating away from significant corrosion. ...............................................................................25 Table 7. Results of optical microscopy examination for fixed internal swab samples taken, over a ~1 cm2 area, from the external surface of the pipe section at Feature 1 and at an area away from significant corrosion. .............................................................................................25 Table 8. Summary of soluble cation and anion concentrations for soil sample 10000151761. .......................................................................................26 Table 9. Summary of various chemical properties for soil sample 10000151761. .......................................................................................26 Table 10. Summary of various electrochemical properties for soil sample 10000151761. .......................................................................................26 Table 11. Results of tensile tests performed on transverse base metal specimens from the failure and the U/S and D/S joints compared with requirements for API 5L Grade X65 line pipe steel. ...............................27 Table 12. Results of Charpy V-notch impact tests for transverse base metal specimens removed from the joint that failed (Joint 5930). ..........................27 Table 13. Results of Charpy V-notch impact tests for transverse base metal specimens removed from the U/S joint (Joint 5920). ...................................28 Table 14. Results of Charpy V-notch impact tests for transverse base metal specimens removed from the D/S joint (Joint 5940). ...................................28 DNV GL – OAPUS309DNOR (PP136049) x September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 188Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Table 15. Results of analyses of the Charpy V-notch impact energy and percent shear plots for base metal specimens removed from the three pipe joints. ....................................................................................29 Table 16. Table 17. Results of chemical analyses of samples removed from the joint that failed and the U/S and D/S joints compared with composition requirements (product analysis) for API 5L Grade X65 line pipe steel.1 ..................................................................................................30 Results of failure pressure analyses using CorLASTM. The pressure at the failure site was estimated at 737 psig. .................................................31 DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP xi#

Page 189Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing List of Figures Figure 1. Photograph showing the failure location and the locations of the two pipe sections, during excavation. ..............................................................32 Figure 2. Photograph showing Pipe Section 1 following removal from the ditch. ............33 Figure 3. Photograph showing Pipe Section 2 being removed from the ditch. ................34 Figure 4. Schematic showing the location of the failure and where samples were removed for various analyses. ..........................................................35 Figure 5. Photograph showing the cargo container in the as-received condition. ..............................................................................................36 Figure 6. Photographs showing the pipe sections in the as-received condition, within the cargo container. ......................................................................36 Figure 7. Photographs showing failure location on PS 2 a) before and b) after evidence tape and a clear protective wrapping was removed. .......................37 Figure 8. Photographs showing the failure location a) while on site (May 28, 2015) and b) after transit to DNV GL’s facility (June 15, 2015). ....................38 Figure 9. Photograph showing a stamp on the internal surface of the failure joint, near the D/S GW (GW 5940). “NIPPON” and “24” are legible. ...............39 Figure 10. Photographs showing the condition of the external tape on the failure joint. Tape measure indicates distance to upstream girth weld. ....................................................................................................40 Figure 11. Photograph showing the internal surface of the external tape from the failure joint. .....................................................................................41 Figure 12. Photograph showing the internal surface of the external tape at the failure location. Tape measure indicates distance to upstream girth weld. ....................................................................................................42 Figure 13. Photograph showing the external surface of the external tape at the failure location. ......................................................................................43 Figure 14. Photographs showing the external surface of the PU insulation at a) the U/S end of PS 2 (14’ to 20’ from U/S GW) and b) the failure location (31.5’ to 36.4’ from U/S GW). Tape measure indicates distance to upstream girth weld. ..............................................................44 Figure 15. Photograph showing a crack in the PU insulation within a wrinkle. White contrast paint was applied to the surface to facilitate laser scanning and visual inspection. Area shown in Figure 14; scale in mm. .....................................................................................................45 Figure 16. Photograph showing a piece of insulation removed from adjacent to the failure location; near 4:30 orientation. .................................................45 DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP xii#

Page 190Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Figure 17. Photograph showing corrosion product that was wedged between the pipe surface and polyurethane insulation. Location indicated in Figure 16; scale in mm. .......................................................................46 Figure 18. Photographs showing the amount of compression in the insulation adjacent to the failure location; near 6:00 orientation. Scale in mm. .............47 Figure 19. Photograph showing the insulation and coal tar coating separating from the pipe in large sheets on the underside of the pipe; approximately 29’ from U/S GW. ..............................................................48 Figure 20. Photograph showing the insulation and coal tar coating separating from the pipe in large sheets on the underside of the pipe; approximately 17’ from U/S GW. ..............................................................48 Figure 21. Photograph showing blistered coal tar coating along the 12:00 orientation on PS 2; approximately 28’ from U/S GW. ..................................49 Figure 22. Photograph showing external corrosion features on PS 2; Feature 1 and Feature 2. Tape measure indicates distance to upstream girth weld. ....................................................................................................49 Figure 23. Photograph showing an external corrosion feature on PS 2; Feature 3. Tape measure indicates distance to U/S GW. ..........................................50 Figure 24. Photograph showing external corrosion features on PS 2; Feature 4 and Feature 6. Tape measure indicates distance to U/S GW (Note: tape slipped 0.1’ to the right). ..................................................................50 Figure 25. Photograph showing an external corrosion feature on PS 2; Feature 5. Tape measure indicates distance to U/S GW. ..........................................51 Figure 26. Photographs showing the a) external and b) internal surfaces of the pipe section at the failure location (Coupon 1). ...........................................52 Figure 27. Photographs showing Pipe Section 2 from 37.7 feet to 40.4 feet from GW 5930 a) before and b) after the composite repair sleeve was removed. Tape measure indicates distance to U/S GW (GW 5930). ...............53 Figure 28. Photograph the primary feature repaired on May 13, 2013. Area shown in Figure 27. Tape measure indicates distance to U/S GW. ..................54 Figure 29. Renderings of PS 1 and PS 2 (viewed from the OD surface) from laser scanning data showing Feature 1 and Feature 2; Coupon 2. The transparency of the insulation was changed to show a correlation between the features observed on the insulation with corrosion features observed on the pipe. The scale on the right is from 0.000 to 0.400 inches. .....................................................................55 Figure 30. Renderings of PS 1 and PS 2 (viewed from the OD surface) from laser scanning data showing the failure location (Feature 4) and Feature 3, 5, and 6; Coupon 1. The transparency of the insulation was changed to show a correlation between the features observed DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP xiii#

Page 191Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing on the insulation with corrosion features observed on the pipe. The scale on the right is from 0.000 to 0.400 inches. ........................................56 Figure 31. Rendering of the failure location (viewed from the OD surface) from laser scanning data showing the remaining wall thickness along the fracture surface. The scale on the right is from 0.000 to 0.400 inches. Location indicated in Figure 30. .....................................................57 Figure 32. Rendering of the failure location showing various thickness measurements made along the corrosion feature and failure opening. The scale on the right is from 0.000 to 0.400 inches. Measurements are tabulated in Table 4. ....................................................58 Figure 33. Renderings of the area under the composite repair sleeve showing the thickness profile for the feature repaired on May 13, 2013. The scale on the right is from 0.000 to 0.400 inches. ........................................59 Figure 34. Photograph showing the internal surface of the pipe section at the failure location with the 1 × 1 inch grid painted on the surface. Tape measure indicates distance to U/S GW. .....................................................60 Figure 35. Results of the UT measurements performed on 1” × 1” grid near the failure location; Feature 4 (Coupon 1). ......................................................61 Figure 36. Results of the UT measurements performed on 1” × 1” grid near Feature 1 (Coupon 2). ............................................................................62 Figure 37. Results of the UT measurements performed on 1” × 1” grid near Feature 2 (Coupon 2). ............................................................................63 Figure 38. Photographs showing the a) clockwise and b) counterclockwise fractures surfaces following cleaning with a degreaser and acetone and/or methanol. Tape measure indicates distance to U/S GW. .....................64 Figure 39. Stereo light photomicrographs of representative locations along the clockwise fracture surface following cleaning with a), b), c) a degreaser and acetone and/or methanol and d), e) an inhibited HCl acid and ENPREP®. Photomicrographs b) and d) are from the same location. Tape measure indicates distance to U/S GW. .................................65 Figure 40. SEM image showing the fracture surface at the location identified in Figure 39e. ........................................................................................66 Figure 41. SEM image showing the transition between Region 1 and Region 2. Area indicated in Figure 40. .....................................................................66 Figure 42. SEM image showing the ductile fracture morphology of Region 1. Area indicated in Figure 41 ......................................................................67 Figure 43. SEM image showing a nondescript/corroded morphology of Region 2. Area indicated in Figure 41 ......................................................................67 DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP xiv#

Page 192Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Figure 44. SEM image showing the fracture surface at the location identified in Figure 39c. ........................................................................................68 Figure 45. SEM image showing a representative ductile fracture morphology for Region 1. Area indicated in Figure 44. ......................................................68 Figure 46. Photograph of the clockwise fracture surface showing thickness measurements of the Region 1 along the fracture surface at 5 mm intervals. Scale in mm. ...........................................................................69 Figure 47. Fracture/Corrosion profile based on three measurement techniques. .............70 Figure 48. Photograph of the mounted transverse cross-section, Mount 195367- 1b removed from the suspected failure origin; Feature 4. Mount location indicated in Figure 39. .................................................................71 Figure 49. Photomicrograph showing the suspected failure origin in cross- section; Feature 4. Area indicated in Figure 48. (Mount 195367-1b; 4% Nital Etch) .......................................................................................71 Figure 50. Photomicrograph showing the negligible grain elongation and plasticity along the CW fracture surface. Area indicated in Figure 49. (Mount 195367-1b; 4% Nital Etch) ...........................................................72 Figure 51. Photomicrograph showing the grain elongation and plasticity along the CCW fracture surface. Area indicated in Figure 49. (Mount 195367-1b; 4% Nital Etch) ......................................................................72 Figure 52. Photomicrograph showing corrosion products near the CCW fracture surface. Area indicated in Figure 48. (Mount 195367-1b; 4% Nital Etch) ....................................................................................................73 Figure 53. Photomicrograph showing corrosion products with some undercutting near the CCW fracture surface. Area indicated in Figure 49. (Mount 195367-1b; 4% Nital Etch) ........................................73 Figure 54. Photomicrograph showing base metal microstructure of Mount 195367-1b; 4% Nital Etch) ......................................................................74 Figure 55. Photograph of the mounted cross-section of corrosion products, Mount 195331-1 removed from Feature 4. .................................................74 Figure 56. Photomicrographs of the mounted cross-section, Mount 195331-1, from corrosion products removed from Feature 4; location identified in Figure 55. ..........................................................................................75 Figure 57. Photograph of the transverse mounted cross-section, Mount 195370- 1 removed from Feature 1. Mount location indicated in Figure 22. .................76 Figure 58. Photomicrograph of the mounted cross-section, Mount 195370-1 removed from Feature 1 showing the corrosion morphology at the external surface; mirror image of location indicated in Figure 57. ..................76 DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP xv#

Page 193Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Figure 59. Photograph of the mounted cross-section of corrosion products, Mount 195322-1, removed from Feature 1. ................................................77 Figure 60. Photomicrographs of the mounted cross-section, Mount 195322-1, from the corrosion product removed from Feature 1; location indicated in Figure 59. ............................................................................77 Figure 61. Photograph of the mounted cross-section, Mount 195365-1, removed from across the longitudinal seam weld. Location indicated in Figure 4. ...........................................................................................78 Figure 62. XRD spectrum acquired from the corrosion products collected from Feature 1, identifying Goethite and Magnetite as the compounds. .................79 Figure 63. XRD spectrum acquired from the corrosion products collected from Feature 2, identifying Goethite and Magnetite as the compounds. .................80 Figure 64. EDS data collected from Mount 195370-1, mirror image of area indicated in Figure 58. ............................................................................81 Figure 65. EDS data collected from Mount 195331-1, area indicated in Figure 56. ..........................................................................................82 Figure 66. EDS line scan collected from Mount 195331-1, area indicated in Figure 56. ..........................................................................................83 Figure 67. Photograph showing the soil samples collected from below the pipe, 8 feet U/S of GW 5930. ...........................................................................84 Figure 68. Percent shear from Charpy V-notch tests as a function of temperature for transverse base metal specimens removed from the failure joint (Joint 5930). .........................................................................85 Figure 69. Charpy V-notch impact energy as a function of temperature for transverse base metal specimens removed from the failure joint (Joint 5930). .........................................................................................85 Figure 70. Percent shear from Charpy V-notch tests as a function of temperature for transverse base metal specimens removed from U/S joint (Joint 5920). ..................................................................................86 Figure 71. Charpy V-notch impact energy as a function of temperature for transverse base metal specimens removed from the U/S joint (Joint 5920). ..................................................................................................86 Figure 72. Percent shear from Charpy V-notch tests as a function of temperature for transverse base metal specimens removed from D/S joint (Joint 5940). ..................................................................................87 Figure 73. Charpy V-notch impact energy as a function of temperature for transverse base metal specimens removed from the D/S joint (Joint 5940). ..................................................................................................87 DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP xvi#

Page 194Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Figure 74. Color profile showing the results of average thickness measurements performed on the laser scan dataset following discretizing the data into ½-inch cells. The resulting profile is highlighted in blue and plotted in Figure 47. ...............................................................................88 DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP xvii#

Page 195Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing 1.0 BACKGROUND Plains All American Pipeline, L.P. (Plains) retained Det Norske Veritas (U.S.A.), Inc. (DNV GL) to perform a metallurgical analysis and mechanical testing on a section of pipe from Line 901 - Las Flores to Gaviota (L901), 24-inch nominal diameter crude oil pipeline that failed while in service. The failure occurred on May 19, 2015 in Goleta (Santa Barbara County), California at milepost (MP) 4, 33.5 feet downstream (D/S) of the nearest upstream (U/S) girth weld and 4.05 miles D/S of the nearest U/S pump station. A failure of a pipe segment can be characterized either as a leak or a rupture; the failure on L901 is characterized as a leak.1 The section of the pipeline that failed is comprised of 24-inch diameter by 0.344-inch wall thickness, API 5L Grade X65 line pipe steel that contains a high frequency electric resistance welded (ERW) longitudinal seam and was manufactured by Nippon Steel in 1986. The maximum operating pressure (MOP) is 1,3412 pounds per square inch gauge (psig) (72% of the specified minimum yield strength [SMYS]). The pressure at the time of failure was reported by Plains to be 737 psig (39.6% of SMYS) at the failure location and time of failure. The pipeline was installed in 1990 and constructed using pipe that was externally coated with a coal tar urethane coating on the steel substrate, 1.5-inch thick rigid polyurethane foam, and an external polyethylene tape. The pipeline has an impressed current cathodic protection (CP) system with the nearest rectifier located 4.05 miles U/S of the failure location, at the Las Flores Pump Station. A hydrostatic test was performed at the time of commissioning for 8 hours at 1719 psi (Gaviota Station) on November 25, 1990. In-line inspection (ILI) runs, consisting of deformation and magnetic flux leakage (MFL) tools, were performed in 2007, 2012, and 2015. The failed pipe joint and 5 feet of the U/S and D/S joints were removed from the failure location and delivered to DNV GL in two pipe sections for analysis. Figure 1 is a photograph showing the failed pipe section at the failure site, while Figure 2 and Figure 3 are 1 According to the FRACTURE CONTROL TECHNOLOGY FOR NATURAL GAS PIPELINES CIRCA 2001 (the PRCI report superseding NG-18 Report 208), “The distinction between leak and rupture for the pipeline community is based on the size and configuration of the breach, not how it develops. A “leak” is characterized by a narrow slit-like hole with length less than the diameter, which limits the fluid volume that escapes through the breach. In contrast, a “rupture” involves a longer, open hole that can be bulged over its length, which is on the order of a diameter or longer and can permit escape of a significant fluid volume.” Similarly, the research performed as part of the historical NG-18 work identified empirical equations to predict the length at which a feature will propagate versus pop through and arrest; the leak/rupture length. Based on these calculations and visual observations, the length of the feature is consistent with a leak, arresting within the corrosion feature, and did not propagate outside of the feature into nominal wall-thickness pipe. 2 Theoretical maximum operating pressure at the lowest elevation using the lowest pressure of either 80% of the commissioning hydro-test pressure, the 72% of SMYS, or the lowest component rating along the line segment. DNV GL – OAPUS309DNOR (PP136049) 1 September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 196Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing photographs showing the pipe sections after removal from the ditch. Pipe Section 1 (PS 1) was 19.05 feet in length and contained 5.05 feet of the U/S joint, the U/S girth weld, and 15 feet of the failure joint located U/S of the failure. Pipe Section 2 (PS 2) was 31.06 feet in length and contained the failure location, the D/S girth weld, a 2013 composite repair sleeve, and 5 feet of the D/S joint. The objective of the analysis was to determine the metallurgical (or immediate) cause of the failure. 2.0 TECHNICAL APPROACH The procedures used in the analysis were in accordance with industry-accepted standards. Five of the general standards governing terminology, specific metallographic procedures, mechanical testing, and chemical analysis used are as follows: • ASTM E7, “Standard Terminology Relating to Metallography.” • ASTM E3, “Standard Methods of Preparation of Metallographic Specimens.” • ASTM E8, “Test Methods for Tension Testing of Metallic Materials.” • ASTM E23, “Standard Test Methods for Notched Bar Impact Testing of Metallic Materials.” • ASTM A751, “Standard Test Methods, Practices, and Terminology for Chemical Analysis of Steel Products.” The following steps were performed for this analysis. The protective shipping wrap was removed and the pipe sections were visually inspected and photographed. The external polyethylene (PE) tape was removed from PS 1(PS 1-ID 100001522513) and PS 2 (PS 2-ID 10000152251) and visually inspected and photographed. The external pipe surfaces (with insulation) were laser scanned using a FaroArmTM to produce digital maps. Laser scanning is a non-destructive technique that uses light, in the form of a laser, to make very accurate three-dimensional (3D) data sets, which capture the x, y, and z coordinates from millions of measurements along the scanned surface. The datasets can then be used to generate 3D renderings of the scanned object(s) that can be rotated, manipulated, and measured. The insulation from PS 2 was then removed and the pipe was visually inspected and photographed. The coal tar coating was then removed around the failure location, areas of corrosion, and at the ends of each pipe section using brass mallets, putty knives, and methyl ethyl ketone (MEK) and/or acetone. 3 Unique DNV GL barcode assigned to each piece of evidence. DNV GL – OAPUS309DNOR (PP136049) 2 September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 197Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Wall thicknesses, diameters, and circumferences were measured at various locations on PS 1 and PS 2 where coating was removed and there was no measurable corrosion. Corrosion products were collected from PS 2 for characterization. Analyses performed on these products included: (1) pH testing using litmus paper, (2) spot tests for carbonates and sulfides using 2-normal hydrochloric acid (2N HCl), (3) elemental analyses using energy dispersive spectroscopy (EDS) with a scanning electron microscope (SEM) and (4) compound identification using x-ray diffraction (XRD). The pH measurements were obtained by placing a few drops of deionized (DI) water on the pH test paper and then the wetted paper was placed in contact with the surface. The pH test paper was examined for color changes and compared to pH color charts. Swab samples were also obtained for bacteria analyses, over a standard area of 1 cm2, at the two locations; at an area of corrosion and area where the coating was disbonded but there was negligible external corrosion. Separate swab samples were taken for serial dilution and microscopic analysis. Liquid culture media for acid-producing bacteria (APB), sulfate-reducing bacteria (SRB), nitrate-reducing bacteria (NRB), aerobic bacteria (AERO), anaerobic bacteria (ANA), and iron-related bacteria (IRB) was used for the serial dilutions to evaluate growth of various types of bacteria. A five vial serial dilution (1:10,000) was performed using each type of media. The swab obtained for the microscopic analysis was fixed in 1% glutaraldehyde. A five microliter spot was removed from the fixed sample and prepared for examination by drying on a microscope slide and staining with 0.1% fluorescein isothiocyanate (FITC). The sample was examined using a CFI PLAN FLUOR 100X oil immersion objective on a Nikon Eclipse 50i epifluorescent microscope equipped with a FITC filter set to determine bacteria cell counts and morphology. Coupons containing the failure location and areas of corrosion were cut from PS 2 using cold-cutting techniques. Coupon 1 contained the failure location and was full ring section removed between 30.66 and 35.95 feet from the U/S GW. Coupon 2 contained external corrosion features further U/S from the failure location and was removed between 14.00 and 20.60 feet from the U/S GW; between the 4- and 8-o’clock orientations. The internal and external surfaces were visually inspected and photographed. Where necessary, the samples were cleaned using a degreaser (LPS Presolve®) and acetone. Ultrasonic testing (UT) was performed on the samples removed from PS 2, using a 1-inch by 1-inch grid spacing, to produce a thickness map. The external and internal pipe surfaces of these coupons were laser scanned to produce a thickness contour dataset. Magnetic particle inspection (MPI) was performed on the external and internal pipe surfaces of the coupon containing the failure location. The fracture surfaces were cleaned with methanol and acetone, optically examined, and photographed. Samples were then removed from one of the mating fracture surfaces, DNV GL – OAPUS309DNOR (PP136049) 3 September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 198Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing cleaned with Rhodine inhibited HCl solution and ENPREP® 214 to remove corrosion products, and examined at high magnifications in an SEM to document the fracture morphology. Transverse cross sections were removed from the suspected failure origin, an area of corrosion further U/S, and across the longitudinal seam weld. The transverse cross sections were mounted, polished, and etched; see Figure 4 for locations. Light photomicrographs were taken to document the fracture and corrosion morphologies and steel microstructure. In addition, corrosion products collected from an area adjacent to the failure location and from areas of corrosion further U/S of the failure were mounted in cross-section and polished. Light photomicrographs were taken to document the corrosion morphologies. Elemental analysis using EDS was performed to identify the elemental constituents of each. Soil analyses were conducted on a sample removed (in the field) approximately 8 feet U/S of the U/S girth weld (GW). The soil was tested for resistivity, moisture content, pH, total acidity, total alkalinity, concentration of soluble anions and cations, total dissolved solids, and linear polarization resistance. Mechanical (duplicate tensile tests and full Charpy V-notch [CVN]) curves) testing was performed on specimens removed from the failed pipe joint and U/S and D/S joints to determine the tensile and fracture toughness properties. Chemical analyses were performed on a steel sample removed from the failed pipe joint and U/S and D/S joints to determine the compositions. CorLASTM calculations were performed to estimate the failure pressure based on the pipe geometry, base metal mechanical properties, and the measured flaw profile. This value was compared with the estimated pressure at the failure location. 3.0 RESULTS AND DISCUSSION 3.1 Visual Examination The pipe sections were transported to DNV GL’s facility near Columbus, Ohio in a sealed cargo container on a flatbed semi-truck. The cargo container was locked and secured with three keyed padlocks, a serialized cargo lock, and evidence tape prior to transport. The corresponding keys for the locks were distributed amongst the interested parties, such that no one person had access to all of the keys. The container was then driven non-stop to a DNV GL storage facility. Upon receipt, the locks and evidence tape were inspected. Figure 5 is a photograph showing the container being loaded into a DNV GL storage facility; the four locks are identified in the figure with yellow circles. Figure 6 contains photographs showing the two pipe sections in the as-received condition. The pipe sections were wrapped in opaque plastic wrap and boxed. Evidence tape was DNV GL – OAPUS309DNOR (PP136049) 4 September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 199Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing applied in the field on top of the plastic wrap approximately 1 foot U/S and D/S of the failure on PS 2. Figure 7 contains photographs showing the failure location before and after removal of the evidence tape and protective plastic wrap at DNV GL’s facility. The fracture surfaces were protected with foam insulation that was place around the clockwise (CW) fracture surface; shown in Figure 7b. Figure 8 contains photographs showing the failure location while on site ( Figure 8a - May 28, 2015) and at DNV GL’s facility ( Figure 8b – June 15, 2015). Some red corrosion products were observed near the failure location, as shown in Figure 8b. The failure opening was 6.6 inches in length and located at the 4:15 orientation, consisting of an irregular fracture path that opened in the clockwise (CW) direction looking D/S. The U/S and D/S ends of the fracture path were located at 33.35 and 33.9 feet, respectively, from the U/S GW. The maximum opening measured 1.14 inches, approximately 33.45 feet from the U/S GW. The failure is associated with a corrosion feature that measured approximately 12.1 inches axially in length by 7.4 inches circumferentially in width. Additional data are presented in Section 3.1.4. PS 1 was 19.05 feet in length and contained 5.05 feet of the U/S joint, the U/S girth weld (GW) and 14 feet of the failure joint U/S of the failure. Pipe Section 2 was 31.06 feet in length and contained 26.06 feet of the failure joint, including the failure location, the D/S GW, a 2013 composite repair sleeve, and 5 feet of the D/S joint. PS 2 was 31.06 feet in length and contained the failure location, the D/S GW, a composite repair sleeve, and 5 feet of the D/S joint. Reference markings were identified on each pipe section noting the top- dead-center (TDC) and the location of each girth weld. A stamp was identified on the internal surface of the failure joint towards the D/S end, adjacent to GW 5940. Figure 9 is a photograph showing the stamp; “NIPPON”, “24”, and other indiscernible characters were observed. 3.1.1 External Polyethylene Tape An external polyethylene tape (external surface of the rigid polyurethane insulation) was present on PS 1 and PS 2, Figure 2 and Figure 10 respectively. The tape is installed in a white condition; however, exposure to the soil and released product discolored the tape to varying shades of brown. Areas of decohesion from the insulation substrate were observed in varying degrees along the length of the two pipe sections. The most pronounced areas were located near the failure location, as shown in the photographs presented at the bottom of Figure 10. Cracks were also observed in the PE tape, primarily at the 12- and 6-o’clock orientations; some of the cracks are identified in Figure 10. Wrinkles in the tape were observed along the entire length of both pipe sections on the bottom half of the pipe (2:00 to 10:00 orientation). DNV GL – OAPUS309DNOR (PP136049) 5 September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 200Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing The PE tape was removed from each pipe section, aligned as it was on the pipe, and visually inspected and photographed. Figure 11 is a photograph of the internal surface of the tape looking D/S. The original white coloration of the PE tape is apparent in the figure along with bands of product at the 4:00 and 7:00 orientations, which appear to correlate with the wrinkle bands in the tape along the pipe section. In general, the areas exhibiting wrinkles were disbonded or partially disbonded from the insulation and were much easier to remove from the pipe section. Figure 12 and Figure 13 are photographs showing the internal and external surfaces of the PE tape at the failure location, respectively. Similar to other areas along the pipeline, cracking and wrinkles in the PE tape were observed near the failure location. The cracks in the tape were located at the 6:00 orientation, while the wrinkles were located at the 4:00 and 7:00 orientations. 3.1.2 Rigid Polyurethane Insulation The polyurethane (PU) insulation was visually inspected after the PE tape was removed. Figure 14 contains photographs from two location along PS 2; the U/S end and the failure location. The insulation exhibited impressions corresponding to the wrinkles observed in the PE tape and, at one location a small crack in the insulation was identified within a wrinkle impression, refer to Figure 15. The white contrast paint evident in the figure was applied to the insulation to facilitate laser scanning and visual inspection. Compression of the insulation was also observed at locations along the 6:00 orientation; two locations are identified Figure 14. Additional detail is provided in Section 3.2. The insulation adjacent to the failure location was removed at the 6:00 orientation, refer to Figure 16. Wedged between the insulation and the pipe surface was a piece of corrosion product, which was collected and bagged. Figure 17 is a photograph showing the corrosion product. The corrosion product is dark, saturated with oil, and rigid. The insulation was partially saturated with a clear liquid. Figure 18 contains photographs showing the insulation in cross-section. The liquid line is evident in the lower-left photograph. At this location, the insulation is saturated near the external surface, while the middle photograph shows saturation that is through the full thickness of the insulation. In addition, signification compression of the insulation was observed at this location (center photo). The compressed thickness measured 0.276 inches as compared to the nominal thickness of 1.5 inches, which corresponds to over 80% compression. In general, the compressed insulation was located on the bottom of the pipe and areas of saturation were within the compressed areas. A pH measurement was also made at a saturated location along the insulation using pH paper; location identified in Figure 17. The pH was between 6 and 7. DNV GL – OAPUS309DNOR (PP136049) 6 September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 201Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing 3.1.3 Coal Tar Urethane Coating The thickness of the coal tar coating was measured using micrometers on a piece that disbonded near the failure location; area shown in Figure 16. The average of four measurements was 0.043 inches, which corresponds closely with measurements performed further U/S (~20 feet U/S from failure location) that averaged 0.040 inches. Figure 19 is a photograph showing the removal of the PU insulation and coal tar coating. Along the underside of the pipe, the insulation and coal tar coating came off together, such that the coal tar coating had disbonded from the pipe surface. At this location, released product can be seen along the mating surfaces of the pipe and coating. Disbondment of the coal tar coating was also observed further U/S on PS 2 at areas where released product did not reach, refer to Figure 20. Disbondment was associated with large corrosion cells, evident in the figure, and in areas where no deep corrosion was observed, but exhibited a layer of fine corrosion products on the pipe surface. Approximately 28 feet from the U/S GW on PS 2 an area of blistered coal tar coating was observed; refer to Figure 21. The insulation against this area was moist, but there was no significant corrosion associated with this location. A syringe was used to extract fluid contained within one of the blisters from which a pH measurement was made using pH paper. The pH was between 6 and 7, consistent with the pH measurement performed on a piece of saturated insulation adjacent to the failure location described above. 3.1.4 Carbon Steel Line Pipe Following removal of the PE tape, PU insulation, and areas of coal tar coating that had disbonded from PS 2, the pipe section was visually inspected. Areas of corrosion were observed on the external surface surrounding the failure location and approximately 14 to 20 feet U/S of the failure location on the bottom of the pipe. The larger features are identified in Figure 22 through Figure 25; a summary of the feature dimensions and locations is presented in Table 1. The corrosion features were located on the bottom of the pipe section in or adjacent to areas that exhibited disbondment of the coal tar coating and compression in the adjacent PU insulation. The corrosion products were dry, rigid, and magnetic. At some locations, a putty knife was required to separate the corrosion products from the pipe body. For the most part, the products associated with each corroded area came off in one piece that was non-friable in nature. The products were dark brown to black or charcoal in appearance and could be handled while remaining intact. The products also appeared to be layered. A Dremel® rotary tool was ultimately used to cut through some of the products for the metallography presented in Section 3.6. The corrosion products from each of the features identified in Table 1 were collected for subsequent analyses. DNV GL – OAPUS309DNOR (PP136049) 7 September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 202Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing A 5.3-foot ring coupon (Coupon 1) containing the failure location and Features 3 through 6 was cut from PS 2. The cuts were made at 30.66 and 35.95 feet from the U/S GW. Similarly, a 6.6-foot coupon (Coupon 2) was cut from U/S end of PS 2 capturing Feature 1 and Feature 2. The longitudinal cuts were made at approximately the 4:00 and 8:00 o’clock orientations. Circumferences/diameters and wall thicknesses were measured on the U/S end of PS 1 (U/S Joint), D/S end of PS 2 (D/S Joint), and on the U/S and D/S ends of the ring section containing the failure location. The measurements were made in areas with no coating or measurable corrosion. The diameters were measured using a Pi tape and are shown in Table 2. The diameter meets API 5L tolerances for 24-inch nominal diameter pipe. The diameters were measured with a tape measure from the 3 to 9 o’clock and 12 to 6 o’clock orientations. The diameters varied from 24.0 to 24.1 inches, indicating no significant ovality, as shown in Table 2 . The wall thickness was measured at the 12, 3, 6, and 9 o’clock orientations at the same locations described above; see Table 3 for details. The wall thickness values ranged from 0.356 to 0.362 inches. These wall thickness values meet API 5L tolerances for a nominal wall thickness of 0.344 inches.4 The 5.3-foot long ring coupon was cut longitudinally at the 3:00 and 9:00 o’clock orientations to facilitate examination of the internal surface. Figure 26 contains photographs showing the external and internal surfaces of the bottom-half of the ring coupon. There was no observable corrosion on the internal surface. A small, superficial, mill anomaly was identified approximately 6 inches D/S from the failure opening. 3.1.5 Composite “Armor Plate” Sleeve The composite repair sleeve installed on May 13, 2013, was comprised of composite Armor Fiber® and cured resin, overlaid with a green two-part epoxy. There were no indications of water ingress or disbondment of the two-part epoxy. The repair was removed by cutting, chiseling, and ultimately sand blasting. Figure 27 contains photographs showing the pipe before and after removal of the repair sleeve. Throughout the removal process, the pipe section was visually inspected for indications of discoloration and corrosion to determine if additional corrosion had occurred following installation of the sleeve in 2013. Figure 28 is a photograph showing the primary feature that was repaired in 2013 after the composite sleeve was removed. There was no evidence of discoloration or additional corrosion associated with the feature; additional discussion and depth measurements are presented in Section 3.2. 4 API 5LX, 35th Edition, May 1986. DNV GL – OAPUS309DNOR (PP136049) 8 September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 203Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing The two-part epoxy and resin were well-adhered, precluding electrolyte from reaching the pipe surface and thus, eliminating additional corrosion of the feature. Similarly, there is no evidence that the repair influenced corrosion of the feature that ultimately failed, i.e. galvanic couple. Without an ionic pathway through the electrolyte there is no means by which to setup such a cell. 3.2 3D Laser Scanning The external surfaces of PS 1 and PS 2 were laser scanned, using a FaroArmTM , following removal of the external polyethylene tape. Similarly, the failure opening and corrosion features along the external surfaces of the pipe section (and the corresponding internal surfaces) were also laser scanned once the polyurethane insulation was removed. The datasets were aligned using reference magnets placed along the pipe sections prior to scanning. With the exception of Feature 1, the features were cleaned with a soft-bristled brush, brass-bristles brush, and methanol and/or acetone. The corrosion products within Feature 1 were left intact for metallographic examination (Section 3.6). As a result of scanning the internal and external surfaces of the pipe at and around the corrosion features, a remaining wall thickness profile was generated to show the extent of corrosion for each feature. Figure 29 contains renderings of the aligned dataset, highlighting the areas of corrosion on the U/S end of PS 2 (Feature 1 and Feature 2). The pipe was rotated such that the viewing direction is normal to the 6:00 orientation. From this perspective, areas of corrosion are clearly visible along the 6:00 orientation. The transparency of the polyurethane dataset on PS 2 was set to 30% providing an opportunity to identify any correlation between features on the insulation and areas of corrosion. It is clear from these data that the corrosion features are located at or adjacent to areas of compressed insulation. Similarly, Figure 30 contains renderings highlighting the areas of corrosion at or near the failure location (Features 3-6). Consistent with the observations above, the corrosion features are located at or adjacent to areas of compressed insulation. Figure 31 is a rendering showing Feature 4; the failure location. The maximum depth of each feature was determined, based on a measured nominal wall thickness of 0.359 inches, from the laser scan data and are presented in Table 1. Various thickness measurements were made slightly offset (~0.100 inches circumferentially) from the fracture path to provide data that would not contain necking, providing a better representation of the wall thickness just prior to the failure. A rendering showing the measurement locations is provided in Figure 32, while the data is given in Table 4. Based on this, the maximum depth of Feature 4 was 0.318 inches or 89% of the measured wall thickness. The failure opening measured 6.55 DNV GL – OAPUS309DNOR (PP136049) 9 September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 204Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing inches in length with a maximum opening of 1.10 inches, which are consistent with the measurements made during the visual examination. The corrosion areas associated with the 2013 repair were also laser scanned. Figure 33 is a rendering showing the thickness profile of the deepest feature. The maximum corrosion depth was 0.220 inches, which corresponds closely to the maximum depth (0.228 inches) and location within the feature identified in 2013 prior to sleeving the pipe section. It’s possible that the discrepancy is due to remnant resin still present within the feature prior to laser scanning. 3.3 Ultrasonic Testing Ultrasonic testing was performed on the coupons identified in Figure 29 and Figure 30. Thickness measurements were made at 1-inch intervals along a 1 × 1-inch grid that was applied to the internal surfaces of each coupon. Figure 34 is a photograph of the grid applied to the coupon containing the failure location. The corresponding measurements are provided in Figure 35; however, the data or the photograph in Figure 34 would need to be mirrored to match one another, as the data is provided as observed from the external surface. Similar measurements were made on the coupon containing Feature 1 and Feature 2. These data are presented in Figure 35 and Figure 36, respectively. The data are provide with a color overlay; dark red being the thinnest remaining wall thickness. Similar measurements were made on the coupon containing Feature 1 and Feature 2. The results from Feature 2 are provided in Figure 37. The UT data agreed very well with the laser scan data, and given the increased lateral resolution of the laser scan data, subsequent discussions and depth data presented in this report are based on the laser scan datasets. 3.4 Magnetic Particle Inspection Magnetic particle inspection was performed on the internal and external surfaces surrounding the failure location. There were no features or anomalies identified. 3.5 Fractographic Examination 3.5.1 Optical Figure 38 contains photographs of the clockwise (CW) and counterclockwise (CCW) fracture surfaces, Figure 38a and Figure 38b, respectively following cleaning with a degreaser and methanol and/or acetone. The fracture surfaces are brown in color and slanted with respect to the radial direction. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 10#

Page 205Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Figure 39 contains stereo light photomicrographs of representative locations along the CW fracture surface following cleaning with a degreaser and methanol and/or acetone [(a), (b), (c)] and one location following cleaning with Rhodine inhibited HCl acid and ENPREP® 214 [(d), (e)]. As shown in Figure 39a and Figure 39c, the fracture surface is tapered or slanted through the thickness. This is a typical characteristic of ductile overload. Similarly, the fracture surface along its length is wavy and has characteristics of ductile tearing and overload. Figure 39d and Figure 39e are micrographs of the regions shown in Figure 39b following cleaning with Rhodine inhibited hydrochloric (HCl) acid and ENPREP® 214. At this location, two unique morphologies are present; Region 1, which has a dull/matte finish and is associated with the areas along the slanted fracture surface, and Region 2, which is more reflective and at a shallower angle with respect to the outer or inner surfaces. These areas are identified in Figure 39e. Region 2 extended the deepest at this location, approximately 33.76 feet from the U/S GW, along the fracture surface. 3.5.2 Scanning Electron Microscopy Figure 40 is an SEM image of the area identified in Figure 39e along the CW fracture surface (Sample 195367-1; 33.76’ from the U/S GW). The white dashed line indicates the interface between Region 1 and Region 2. Figure 41 is a higher magnification SEM image showing the transition between Region 1 and Region 2. Region 2 is relatively smooth with spherical- shaped impressions, while Region 1 appears rough with smaller topographical features. Figure 42 contains a high magnification SEM image in Region 1, near the ID. The fracture surface at this location exhibits dimples, which are characteristic of ductile overload. Figure 43 contains a high magnification SEM image in Region 2, near the ID surface. The fracture surface is nondescript having a corroded appearance and is inconsistent with a typical fracture morphology, indicating that this region was present prior to the failure. This observation coupled with the oblique angle of the surface and visual appearance indicates that Region 2 is associated with external corrosion. Figure 44 is an SEM image from a representative location along the fracture surface exhibiting a shear or slanted fracture surface (Sample 195367-2, 33.55’ from the U/S GW). At higher magnification ( Figure 45), a rough-dimpled appearance, consistent with ductile overload, was observed. The dimples are elongated in the vertical direction as shown in the figure, which is consistent with the orientation of the sheared fracture plane. 3.5.3 Fracture/Corrosion Profile Using the observations of the optical and SEM fractographic examinations, a fracture profile was generated showing the boundary of Region 1. Measurements were made along the DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 11#

Page 206Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing fracture surface at 5 mm intervals, refer to Figure 46. The resulting data are plotted in Figure 47 as measured (remaining) wall thickness versus distance to U/S GW. Given that plasticity/ductility was observed along the fracture surface, the measured thickness of Region 1 is not necessarily representative of the remaining wall thickness prior to failure. This is due to necking of the material, a process governed by the Poisson effect, whereby tensile strain on one direction (i.e. circumferential) results in compressive strain in the other two perpendicular directions (i.e. radial and longitudinal) for isotropic materials; such as steel. Therefore, the thickness of Region 1 was larger than the measured values prior to failure. For this reason, two additional profiles are presented in the figure using the laser scan data; one based on a ½-inch by ½-inch grid, which will be discussed in Section 3.12 and one based on the measurements made adjacent to the fracture surface ( Figure 32), approximately 0.100 inches circumferentially, presumably at locations not heavily influenced by necking. 3.6 Metallographic Examination Figure 48 is a photograph of the transverse metallographic cross-section (Mount 195367-1b) removed from across the fracture surface at approximately 33.76 feet D/S of the U/S GW; same location identified in Figure 41. The corrosion profile near the failure opening is relatively uniform; however, transitions sharply with a steep side wall approximately 15 mm CCW from the opening. Figure 49 is a photomicrograph showing the two mating fracture surface in the etched condition. Figure 50 is a photomicrograph showing the mating CW fracture surface. With the exception of a small ligament at the internal surface, there were no obvious indications of plasticity corresponding to Region 1. This suggests that the preexisting corrosion feature was almost through-wall at this location just prior to failure. In comparison, the mating CCW fracture surface, presented in Figure 51, exhibited grain elongation and deformation, consistent with plasticity and the results obtained from the SEM examination showing dimpled failure in Region 1. The discrepancy between microstructural characteristics of the CW and CCW surfaces at this location is due to a small misalignment between the two mating fracture surfaces when the transverse cuts were made to produce the metallographic cross-section. Figure 52 and Figure 53 are representative photomicrographs showing the corrosion morphology along the external surfaces of Mount 195367-1b. The corrosion is scalloped in most cases ( Figure 52), and the remaining corrosion products exhibit a layered texture with alternating light and dark bands. However, at the base of some of these scallops, some undercutting was also observed, as shown in Figure 53. Figure 54 is a photomicrograph showing the typical microstructure of the base metal. The microstructure consists of ferrite DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 12#

Page 207Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing (white areas) and fine pearlite (gray areas). This microstructure is consistent with the vintage and grade of the steel. Figure 55 is a photograph of the transverse metallographic cross-section (Mount 195331-1) removed from the corrosion products associated with Feature 4; collected adjacent to the failure location. At this location, the thickness of the product is approximately 0.55 inches. Droplets of oil can be seen on the surface of the mount, as the photograph was taken following the SEM examination in which the mount was pumped down to low pressures to facilitate observation in the SEM. Figure 56 contains photomicrographs through the thickness of the corrosion product. The morphology of the corrosion products are similar throughout, consisting of alternating light and dark layers. Figure 57 is a photograph of the transverse metallographic cross-section (Mount 195370-1) removed through Feature 1, at approximately 16.65 feet D/S of the U/S GW; feature identified in Figure 22. The corrosion depth is much less severe at this location and the profile is relatively uniform. At higher magnification ( Figure 58), the corrosion products exhibit a similar layered morphology as those observed near the failure location; Figure 52. Figure 59 is a photograph showing the transverse metallographic cross- section (Mount 195322-1) removed from the corrosion products associated with Feature 1. At this location, the thickness of the product is approximately 0.40 inches. Consistent with the other corrosion products, the morphology contains alternating dark and light layers; refer to Figure 60. Figure 61 is a photograph showing the transverse metallographic cross-section remove from the longitudinal seam weld of the failure joint. At higher magnification, an hourglass shape (associated with a heat affected zone) can be observed, characteristic of a high-frequency electric resistance weld (ERW). 3.7 Solid Sampling of Corrosion Products 3.7.1 pH Testing and Qualitative Spot Testing The pH of the external corrosion products collected from Feature 5 was determined using deionized (DI) water and pH test paper. The pH of the deposits was 5 to 6 and the pH of the DI water used in the analysis was also 5 to 6. Qualitative spot testing, using 2N HCl, was performed on three external corrosion products collected from Feature 1, Feature 2, and Feature 5. Portions of the samples were placed in vials with lead acetate tape and a few drops of the HCl were placed on the products. Vigorous bubbling is a positive indication for the presence of carbonates. A rotten egg odor DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 13#

Page 208Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing and/or discoloration of lead acetate tape are positive indications for the presence of sulfides. All samples tested negative for both carbonates and sulfides. 3.7.2 X-ray Diffraction X-ray diffraction was performed on corrosion products collected from Feature 1 and Feature 2. The resulting spectrum for each is presented in Figure 62 and Figure 63, respectively. The compounds identified for each were goethite (FeOOH) and magnetite (Fe3O4). Goethite is one of the most thermodynamically stable iron oxides under aerobic (high oxygen) conditions. Conversely, magnetite is metastable phase formed under low oxygen conditions. 3.7.3 Energy Dispersive Spectroscopy Energy dispersive spectroscopy was performed on the corrosion products captured in the metallographic cross-sections. Figure 64 contains the results of EDS scans performed on the corrosion products in Mount 195370-1, associated with Feature 1. EDS scans were performed within the layered regions identified in the metallographic examination. The results are summarized in Table 5. The two primary constituents are iron (Fe) and oxygen (O), which are characteristic of iron oxides. Small quantities of chlorine (Cl) were identified, likely associated with chlorides, while most of the other constituents are elements common to line pipe steels. A relatively high concentration of copper (Cu) was identified in Scan 1 (8 wt.%), which is atypical of line pipe steel and may be associated with deposits from groundwater. The other three scans identified typical concentrations of Cu. The light area, Scan 3, has an O content of 29.24 wt.%, while the darker bands, Scan 2 and Scan 4, have an average oxygen content of 36.88 wt.%. Given that the XRD analyses identified goethite and magnetite as the two compounds associated with the corrosion products, these values were compared to the calculated oxygen content for goethite (36 wt.%) and magnetite (28 wt.%). These values correlate very closely, indicating that the light areas are likely magnetite and the darker areas are likely goethite. Figure 65 contains the results of EDS scans performed on the corrosion products on Mount 195331-1, associated with Feature 4. Similar results were obtained for the layers identified in the cross-section. The results are summarized in Table 5, which again shows that the compositions of the light layers correspond to magnetite and the compositions of the darker layers correspond to goethite. The variation in oxygen content is apparent in the line scan presented in Figure 66, which illustrated the decreased oxygen content of the lighter layer. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 14#

Page 209Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing 3.8 Microbiological Analyses The external surface of the pipe section was swabbed (at DNV GL) over a standard area of approximately 1 cm2 for bacterial analysis. The swabs were taken from the bottom of Feature 1, 16.65 feet D/S from the U/S GW and from an area beneath disbonded coal tar coating with no significant corrosion; approximately 5 inches CCW from Feature 1. Separate swab samples were taken from each location for the serial dilution and microscopic examination analyses. The results of the microbiological analyses are discussed below. 3.8.1 Serial Dilution – Liquid Culture Media Table 6 shows the results of the bacteria serial dilution testing for the swab samples collected from the pipe section. The results reveal a positive indication for five types of bacteria (APB, AERO, ANA, IRB, and NRB) at the corrosion feature, while there were no positive indications for bacteria at the area away with no significant corrosion. As seen in the table, the highest concentration of bacteria detected was 100 bacteria per cm2, which is a relatively low value. 3.8.2 Microscopic Examination for Total Bacteria The swabs collected from Feature 4 and from the area away were fixed in 1% glutaraldehyde and examined using epifluorescent microscopy. The practical minimum detection limit for this method is approximately 103 cells/ml of fixed sample. The results of the analysis are provided in Table 7. As seen in the table, rod-shaped cells were detected for the swab samples removed at Feature 1 and an area of no apparent corrosion. The calculated concentration of cells for the swab samples were 1.70 × 104 cells/mL and 2.8 × 104 cells/mL. This type of microscopic examination does not differentiate between living and non-living organisms. 3.9 Soil Testing DNV GL collected six (6) soil samples from the dig site at the failure location. Two samples were collected from under the pipe at each of the three locations: 8 feet U/S of GW 5930 (IDs 10000151761 & 10000151762), 2 feet D/S of failure location (IDs 10000151753 & 1000151759), and 12.5 feet D/S of GW 5940 (IDs 10000151754 & 10000151755). The only samples not contaminated with product were the samples collected 8 feet U/S of GW 5930. Figure 67 is a photograph showing the soil samples collected 8 feet US of GW 5930. The samples were placed in a cooler with ice packs and shipped to DNV GL’s laboratory for testing. One of the uncontaminated samples (Sample 10000151761) was sieved in order to conduct the testing. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 15#

Page 210Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Testing was conducted to determine the physical and chemical properties of the sample, including: (1) resistivity, (2) moisture content, (3) pH, (4) soluble anions [Cl-, SO4 2-, S2 - , NO2 -, NO3 -, CO3 2-, HCO3 -], (5) soluble cations [K+, Ca2+, Mg2+, Na+], (6) total alkalinity, (7) total acidity, (8) linear polarization resistance, and (9) total dissolved solids (TDS). The results of the analyses are provided in Table 8 through Table 10. The sample exhibited relatively low levels of nitrate (NO3 -), chloride (Cl-) and carbonate (HCO3 -) and high levels of sulfate (SO4 2-) anions; 115, 117, 204, and 3600 mg/L, respectively. The soil resistivity decreased from 3,800 ohm-cm in the as-received condition to 400 ohm-cm when saturated. Based on these data, the soil is considered corrosive.5 Corrosion rates were determined for the soil sample in the as-received and saturated condition using linear polarization resistance (LPR). The resulting corrosion rates were 2.5 and 2.7 mils per year (mpy), respectively. 3.10 Mechanical Testing The results of tensile testing of duplicate, transverse base metal specimens removed from the pipe joint that failed and the U/S and D/S joints are shown in Table 11. The average yield strength (YS) and ultimate tensile strength (UTS) of Joint 5930 were 64.8 ksi and 84.0 ksi, respectively. The average YS of the base-metal samples is marginally lower than the minimum YS requirements for API 5L X65 line pipe steel of 65.0 ksi. The average is based on two tests values of 65.2 and 64.4 ksi. The average UTS of the base-metal samples meets the minimum UTS requirements for API 5L X65 line pipe steel of 80 ksi. The tensile properties of the U/S and D/S joints meet the requirements for API 5L X65 line pipe steel, as shown in Table 11. Table 12 - Table 14 summarize the results of the Charpy testing for the transverse base metal samples while Figure 68 through Figure 73 show the corresponding Charpy percent shear and impact energy curves. An analysis of the data for the base metal specimens from the failure joint, Joint 5930, indicates that the 85% fracture appearance transition temperature (FATT) is -58.5°F and the upper shelf Charpy energy is 164.8-ft·lbs, full size. These are very good values and typical for modern line pipe steels. The CVN test results can be adjusted to determine the 85% FATT that would be expected for full-scale pipe by applying temperature shifts to the data. This method (full-scale) adjusts the 85% FATT obtained from the Charpy tests to a predicted FATT from the Battelle Drop-Weight Tear Test (BDWTT). The predicted 85% FATT from the BDWTT test most closely represents the 5 Peabody’s Control of Pipeline Corrosion, 2nd Edition, Table 5.5 and Table 5.7. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 16#

Page 211Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing expected FATT for full-scale pipe wall material.6 The full-scale brittle to ductile transition temperature for the failure joint, based on a pipe wall thickness of 0.359 inches, is shown in Table 15. The pipe joint is expected to exhibit ductile fracture behavior above -78.4°F.7 The values for the U/S and D/S joints are also provided in Table 15. The toughness properties of these joints also are very good but not quite as good as the failure joint. 3.11 Chemical Analysis The results of the chemical analyses performed on samples removed from the failure joint and the U/S and D/S pipe joints are shown in Table 16. The results show that the pipe joints meet the composition specifications for API 5L X65 line pipe steel at the time of manufacture. 3.12 Failure Pressure Analysis CorLAS™ (Version 3.02) was used to perform Remaining Strength (RSTRENG) calculations to estimate the failure pressure incorporating the effective-area methodology. The calculations were based on the measured mechanical properties and dimensions of the failure joint, and the measured flaw profile. Three flaw profiles were considered for the analysis: • Case 1: Measurements made along the fracture surface combined with laser scan data on either end of the failure opening, within Feature 4; the black profile presented in Figure 47. • Case 2: The laser scan data and measurements made adjacent to the fracture surfaces, presumably in areas where necking/plasticity was minimized; the blue profile presented in Figure 47. • Case 3: The laser scan data and discretizing corrosion Feature 4 into ½-inch cells; columns running axial and rows running circumferential. The average depth for each cell was determined and the lowest values identified within each column were then used to generate the flaw profile; refer to Figure 74 and Figure 47 (green profile). 8 The measured flaw profiles, presented in Figure 47, were fed into CorLAS™ whereby an algorithm converted each profile into an equivalent semi-elliptical flaw. These effective (or 6 W. A. Maxey, J. F. Kiefner, R. J. Eiber, Brittle Fracture Arrest in Gas Pipelines,” NG-18 Report No. 135, A.G.A. Catalog No. L51436, April 1983, Battelle Columbus Laboratories. 7 Rosenfeld, M.J., “A Simple Procedure for Synthesizing Charpy Impact Energy Transition Curves from Limited Test Data,” International Pipeline Conference, Volume 1, ASME, 1996, Equation 1. 8 The average thickness for each cell was used (instead of minimum values) due to meshing effects along the fracture surface that provided unrealistic minimum wall thickness values as a result of the slanted fracture surfaces. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 17#

Page 212Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing equivalent) flaws were used to estimate the failure pressure. The results of the analysis are presented in Table 17. The calculated failure pressure for Case 1, Case 2, and Case 3 are 474 psig, 759 psig, and 763 psig, respectively. The estimated pressure at the failure location at the time of the failure was reportedly 737 psig, which is in very good agreement with the estimated failure pressures for Case 2 and Case 3. As discussed previously, the estimated failure pressure for Case 1 is underestimated due to the presence of necking that resulted from the overload event. Additional results of the analysis, and a description of CorLAS™, are summarized in Appendix A. 4.0 DISCUSSION AND CONCLUSIONS External corrosion was identified at several locations along the failed pipe section, including the corrosion feature that ultimately failed on May 19, 2015. The corrosion features are located in areas where the external polyethylene tape, thermal polyurethane insulation, and coal tar enamel were compromised, allowing the ingress of moisture to facilitate aqueous corrosion. Cracking in the polyethylene tape, as well as wrinkles, provided pathways for water to collect against the bottom of the pipe section. This in turn may have initiated and/or accelerated the breakdown of the thermal insulation that resulted in compression of the insulation and breakdown of the cellular structure, causing water absorption and retention. The presence of goethite and magnetite in a layered morphology is consistent with aqueous corrosion under wet-dry cycling.9 When oxygen transport is high, such as during the drying stages when the electrolyte is relatively thin, goethite is the stable oxide. However, during saturated conditions or when oxygen is limited (i.e. thick products), magnetite is predominant. The alternating nature of the layers suggests that external variables, such as rain, drainage, and operating temperature contributed to the corrosion process. The term corrosion under insulation (CUI) may be defined as external corrosion of carbon steel piping, pressure vessels, and structural components resulting from water trapped under insulation.10 Although typically associated with above-ground piping, CUI is the appropriate corrosion mechanism for this particular failure given the contributing role of the thermal insulation to the corrosion process. Thus, the results of the analyses indicate that CUI was the primary corrosion mechanism, facilitated by wet-dry cycling. Although bacteria were identified at a corrosion feature sampled U/S of the failure location, the quantities were low and the layered morphology within the corrosion products is not necessarily 9 Nasrazadani, S. and Raman, A., Formation and Transformation of Magnetite (Fe3O4) on steel surfaces under Continuous and Cyclic Water Fog Testing, Corrosion, 1993. 10 API Recommended Practice 583, Corrosion Under Insulation and Fireproofing, May 2014. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 18#

Page 213Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing consistent with MIC, particularly given that the corrosion products are extremely rigid. Therefore, there is no strong evidence to indicate that MIC contributed to the observed corrosion. A summary of our observations are provided below. Summary of Observations • The failure was associated with an external corrosion feature located 33.50 feet from the upstream girth weld, at the 4:24 orientation (center of corrosion feature). • The dimensions of the corrosion feature were 12.1 inches axially by 7.4 inches circumferentially. The maximum depth, as measured using laser scan data, was 0.318 inches or 89% of the measured wall thickness (0.359 inches). • The failure opening was 6.6 inches in axial length, with the upstream and downstream ends located 33.35 and 33.9 feet from the upstream girth weld. • The maximum circumferential dimension of the failure opening was 1.14 inches, approximately 33.45 feet from the upstream girth weld, at the 4:15 orientation. • The fracture surfaces exhibited ductile overload. • Cracking and wrinkling were observed within the polyethylene tape. • Compression was observed within the polyurethane insulation at areas on the bottom of the pipe. These areas were saturated with moisture. • Disbondment of the coal tar coating was observed on the bottom of the pipe along the length of Pipe Section 2. • External corrosion features, including the feature associated with the failure, were identified at or adjacent to areas of saturated, compressed insulation. • The corrosion products were rigid, non-friable, and, at some locations, well adhered to the pipe section. The products consist of alternating layers of goethite and magnetite. • No evidence of internal corrosion was observed along the length of the pipe sections inspected. • The average yield strength (YS) for the failure joint is marginally lower than the minimum YS requirements for API 5L X65 line pipe steel of 65.0 ksi. The average is based on two tests values; one slightly higher (65.2 ksi) and one slightly lower (64.4 ksi) than the requirement. The average ultimate tensile strength (UTS) of the failure joint meets the minimum UTS requirements for API 5L X65 line pipe steel of 80 ksi. • The Charpy V-notch (CVN) properties of the base metal are typical for the vintage and grade of line pipe steel. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 19#

Page 214Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing • The chemical composition of the base metal meets requirements for the vintage and grade of line pipe steel. • The microstructure of the base metal is typical for the vintage and grade of line pipe steel. • The CorLAS™ predicted failure pressure for the failed joint was calculated to be approximately 760 psig, which is in very good agreement with reported pressure at the failure location and time of failure (737 psig). DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 20#

Page 215Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Table 1. Summary of the locations and dimensions of corrosion features identified during the laboratory examination on the external surface of PS 2. Corrosion Feature (2015 ILI “Log Dist.”) Distance from U/S GW to Center of Feature (feet) Axial Length (inches) Distance from TDC to Center of Feature, Clockwise (inches) Circumferential Length (inches) O’clock Orientation (TDC to Center of Feature) Maximum Depth from Laser Scan Data11 (inches) Feature 1 (21367.67) 16.55 8.3 43.00 7.7 6:50 0.112 (31%) Feature 2 (21368.88) (21369.26) (21369.49) (21369.56) (21369.57) (21369.99) (21370.13) (21370.48) 18.61 21.9 40.30 9.8 6:24 0.199 (55%) Feature 3 (21382.40) 31.52 7.7 35.60 17.2 5:40 0.208 (58%) Feature 4 (21384.17) (21384.38) (21484.39) (21484.54) (21484.58) (21384.63) 33.50 12.1 27.75 7.4 4:24 0.318 (89%) Feature 5 33.83 1.8 44.80 2.2 7:00 0.025 (7%) Feature 6 (21385.39) 34.32 2.8 30.10 2.8 4:48 0.122 (34%) 11 Based on measured nominal wall thickness of 0.359 inches. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 21 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 216Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Table 2. Results of diameter measurements performed on PS 1 and PS 2 using Pi tape and a tape measure. Location Diameter Using Pi Tape (inches) Diameter 3 to 9 o’clock (inches) Diameter 6 to 12 o’clock (inches) PS 1 - U/S end; Joint 5920 24.059 24.0 24.0 PS 2 - 30.7’ from U/S GW; Joint 5930 24.059 24.0 24.0 PS 2 - 36’ from U/S GW; Joint 5930 24.055 24.1 24.0 PS 2 - D/S end; Joint 5940 24.048 24.1 24.0 Table 3. Results of wall thickness measurements performed on PS 1 and PS 2. O’clock Orientations Wall Thickness, PS 1 U/S End Joint 5920 (inches) Wall Thickness, PS 2 30.7’ from U/S GW Joint 5930 (inches) Wall Thickness, PS 2 36’ from U/S GW Joint 5930 (inches) Wall Thickness, PS 2 D/S End Joint 5940 (inches) 12:00 0.356 0.359 0.359 0.358 3:00 0.360 0.362 0.362 0.359 6:00 0.356 0.359 0.358 0.359 9:00 0.357 0.359 0.357 0.358 Average 0.357 0.360 0.359 0.359 DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 22#

Page 217Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Table 4. Results of thickness measurements performed adjacent (~0.100 inches circumferentially) to the failure opening using the laser scan dataset. See Figure 32. Distance to U/S GW (feet) Measured Wall Thickness (inches) 33.02 0.354 33.07 0.344 33.12 0.340 33.15 0.249 33.17 0.189 33.21 0.118 33.26 0.138 33.32 0.124 33.36 0.111 33.41 0.096 33.44 0.072 33.47 0.051 33.48 0.043 33.48 0.043 33.51 0.066 33.53 0.067 33.55 0.091 33.60 0.119 33.62 0.074 33.64 0.049 33.67 0.073 33.70 0.085 33.73 0.079 33.76 0.072 33.79 0.073 33.84 0.042 33.89 0.072 33.92 0.106 33.95 0.242 33.98 0.352 DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 23#

Page 218Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Table 5. Results of elemental analyses, using EDS, performed on corrosion products from Feature 1 and Feature 4 compared to ideal chemistry compositions of goethite and magnetite; values presented in mass percent (wt.%). Mount 195370-1 (Feature 1 Products) Mount 195331-1 (Feature 4 Products) Element Scan 1 (Mixed Layers) Scan 2 (Dark Layer) Scan 3 (Light Layer) Scan 4 (Light Layer) Scan 1 (Light Layer) Scan 2 (Light Layer) Scan 3 (Dark Layer) Goethite (FeOOH) Magnetite (Fe3O4) Oxygen (O) 33.52 35.78 29.24 37.97 27.85 29.45 37.83 36.01 27.64 Silicon (Si) 0.19 0.40 0.41 0.26 0.21 0.20 0.22 – – Chlorine (Cl) 0.19 0.06 0.07 0.15 – – – – – Sulfur (S) – – – – – – 0.18 – – Manganese (Mn) 0.79 0.80 0.68 0.73 0.59 0.93 0.98 – – Magnesium (Mg) – – – – – – 0.48 – – Iron (Fe) 56.87 62.46 69.60 60.71 71.35 69.41 60.30 62.85 72.36 Copper (Cu) 8.44 0.49 – 0.18 – – – – – DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 24 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 219Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Table 6. Results of bacteria analyses performed on swabs taken, over an ~1 cm2 area, from the external surface of Pipe Section 2 at Feature 1 on the failure joint and at an area of disbonded coating away from significant corrosion. Feature 1 (16.65 ft D/S from U/S GW) Area of No Significant Corrosion (Outside of Feature 1; ~ 5 inches CCW) Bacteria Type Test Result Number of Positive Vials Test Result Number of Positive Vials Aerobic (AERO) Positive 2 Not detected – Anaerobic (ANA) Positive 2 Not detected – Acid-Producing (APB) Positive 2 Not detected – Sulfate-Reducing (SRB) Not detected – Not detected – Iron-Related (IRB) Positive 2 Not detected – Nitrate-Reducing (NRB) Positive 2 Not detected – Bacteria Concentration Key: 1 10 bacteria per cm2 2 100 bacteria per cm2 , 3 1,000 bacteria per cm2 , 4 10,000 bacteria per cm2 , 5 100,000 bacteria per cm2 Table 7. Results of optical microscopy examination for fixed internal swab samples taken, over a ~1 cm2 area, from the external surface of the pipe section at Feature 1 and at an area away from significant corrosion. Sample Identification Aliquot Volume, uL Total Cells Observed Calculated № cells/mL Morphology Feature 1 (16.65 ft D/S from U/S GW) 5 12 1.70 × 104 Rod Area of No Apparent Corrosion (Outside of Feature 1; ~ 5 inches CCW) 5 >20 2.80 × 104 Rod DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 25#

Page 220Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Table 8. Summary of soluble cation and anion concentrations for soil sample 10000151761. Soluble cations mg/L Sample ID Na+ K+ Ca2+ Mg2+ NO2 - NO3 - Cl- 10000151761 (8’ U/S of GW 5930; below pipe) 898 320. 495 9.64 < 2.1 115 117 Soluble anions mg/L SO4 - S2 - CO3 2- < 0.67 HCO3 - 3600 < 13.3 204 Table 9. Summary of various chemical properties for soil sample 10000151761. Sample ID pH soil Total Acidity mg CaCO3/kg Total Alkalinity mg CaCO3/kg As-Received Moisture Content % Total Dissolved Solids (mg/L) 10000151761 (8’ U/S of GW 5930; below pipe) 7.95 < 66.5 204 (a) 27.59% (b) 21.62% 6350 a – Percent moisture per AASHTO T265 & ASTM D2216 b – Percent moisture per EPA Method 1684, Eq. 2. Table 10. Summary of various electrochemical properties for soil sample 10000151761. Sample ID Resistivity Ohm-cm (as-received) Resistivity Ohm-cm (saturated) LPR mpy (as-received) LPR mpy (saturated) 10000151761 (8’ U/S of GW 5930; below pipe) 3,800 400 2.5 2.7 DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 26 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 221Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Table 11. Results of tensile tests performed on transverse base metal specimens from the failure and the U/S and D/S joints compared with requirements for API 5L Grade X65 line pipe steel. Failure Joint (10000151970) U/S Joint (10000151968) D/S Joint (10000151969) API 5L Grade X52 (Minimum Values) 3 Yield Strength, ksi 1 64.82 65.9 68.4 65 Tensile Strength, ksi 1 84.0 84.6 87.7 80 Elongation in 2 inches, % 1 35.0 33.6 32.8 21.25 Reduction of Area, % 1 60.1 62.8 57.6 – 1 – Average of duplicate tests. 2 – Average of 65.2 ksi (extensometer on OD) and 64.4 ksi (extensometer on ID) 3 – API 5LX, 35th Edition, May 1986. Table 12. Results of Charpy V-notch impact tests for transverse base metal specimens removed from the joint that failed (Joint 5930). Sample ID Temperature, °F Sub-size Impact Energy, ft-lbs Full Size Impact Energy, ft-lbs Shear, % Lateral Expansion, mils 15446-1-6 -238 2 2 0 0.006 15446-1-10 -189 4 5 1 0.006 15446-1-4 -148 24 28 8 0.017 15446-1-2 -103 33 38 29 0.022 15446-1-8 -65 103 120 83 0.076 15446-1-7 -29 124 144 91 0.084 15446-1-1 -4 119 138 99 0.081 15446-1-3 32 130 151 100 0.082 15446-1-5 68 158 184 100 0.083 15446-1-9 100 142 165 100 0.080 DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 27#

Page 222Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Table 13. Results of Charpy V-notch impact tests for transverse base metal specimens removed from the U/S joint (Joint 5920). Sample ID Temperature, °F Sub-size Impact Energy, ft-lbs Full Size Impact Energy, ft-lbs Shear, % Lateral Expansion, mils 15446-2-9 -312 1 1 0 0.001 15446-2-3 -238 2 2 0 0.001 15446-2-2 -193 4 5 3 0.001 15446-2-5 -148 5 6 15 0.002 15446-2-1 -103 18 22 20 0.012 15446-2-10 -51 48 58 41 0.045 15446-2-4 -4 101 122 92 0.075 15446-2-6 32 104 126 100 0.073 15446-2-7 75 111 135 100 0.078 15446-2-8 100 117 142 100 0.079 Table 14. Results of Charpy V-notch impact tests for transverse base metal specimens removed from the D/S joint (Joint 5940). Sample ID Temperature, °F Sub-size Impact Energy, ft-lbs Full Size Impact Energy, ft-lbs Shear, % Lateral Expansion, mils 15446-3-9 -312 1 1 0 0.000 15446-3-3 -238 2 2 1 0.002 15446-3-2 -193 3 4 3 0.002 15446-3-5 -148 5 6 15 0.003 15446-3-1 -103 12 15 15 0.007 15446-3-10 -51 40 48 36 0.040 15446-3-4 -4 83 101 83 0.068 15446-3-6 32 92 112 100 0.077 15446-3-7 75 96 116 100 0.079 15446-3-8 100 103 125 100 0.080 DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 28#

Page 223Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Table 15. Results of analyses of the Charpy V-notch impact energy and percent shear plots for base metal specimens removed from the three pipe joints. Failure Joint; Joint 5930 (10000151970) U/S Joint; Joint 5920 (10000151968) D/S Joint; Joint 5940 (10000151969) Upper Shelf Impact Energy (Full Size), Ft-lbs 164.8 138.9 121.3 85% FATT, °F -58.5 -1.6 4.8 85% FATT, °F (Full Scale Pipe) 1 -78.4 -19.4 -12.7 1 – Full Scale Pipe FATT = 85% FATT + ((66*(tw 0.55/tc 0.7)-100) where tw = pipe wall thickness and tc = width of the CVN specimen. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 29#

Page 224Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Table 16. Results of chemical analyses of samples removed from the joint that failed and the U/S and D/S joints compared with composition requirements (product analysis) for API 5L Grade X65 line pipe steel.1 Element Failure Joint; 5930 (10000151970) (Wt. %) U/S Joint; 5920 (10000151968) (Wt. %) D/S Joint; 5940 (10000151969) (Wt. %) API 5L Grade X65 Spec (Wt. %) 1 C (Carbon) 0.082 0.083 0.078 0.30 (max) Mn (Manganese) 1.110 1.160 1.120 1.50 (max) P (Phosphorus) 0.011 0.010 0.010 S (Sulfur) 0.007 0.007 Si (Silicon) 0.170 0.190 0.160 – Cu (Copper) 0.268 0.270 0.274 – Sn (Tin) 0.000 0.000 0.000 – Ni (Nickel) 0.008 0.008 0.006 – Cr (Chromium) 0.035 0.027 0.028 – Mo (Molybdenum) 0.000 0.000 0.000 – Al (Aluminum) 0.010 0.016 0.012 – V (Vanadium) 0.022 0.024 0.050 (max) 0.060 (max) 0.028 Nb (Niobium) 0.063 0.065 Zr (Zirconium) 0.000 0.000 Ti (Titanium) 0.011 0.016 B (Boron) 0.0006 0.0005 W (Tungsten) 0.000 0.000 Co (Cobalt) 0.000 0.000 Fe (Iron) 98.200 98.100 0.007 0.010 (min) 0.005 (min) 98.200 0.000 – 0.015 – 0.0005 – 0.062 Balance 0.000 – 0.000 – 1 – API 5L, 35th Edition, May 1986. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 30#

Page 225Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Table 17. Results of failure pressure analyses using CorLASTM . The pressure at the failure site was estimated at 737 psig. Case № Equivalent Flaw Profile Properties Estimated Failure Pressure (psig) 1 As-measured along fracture surface (includes necking) Measured 474 2 Laser scan data adjacent to fracture surface Measured 759 3 Laser scan data ½ × ½ inch grid (average) Measured 763 DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 31#

Page 226Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Failure Location GW 5940 Figure 1. Photograph showing the failure location and the locations of the two pipe sections, during excavation. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 32 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 227Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing GW 5930 Wrinkle band Figure 2. Photograph showing Pipe Section 1 following removal from the ditch. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 33 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 228Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Failure Location GW 5940 31.06 feet Figure 3. Photograph showing Pipe Section 2 being removed from the ditch. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 34#

Page 229Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Figure 4. Schematic showing the location of the failure and where samples were removed for various analyses. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 35#

Page 230Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Figure 5. Photograph showing the cargo container in the as-received condition. Pipe Section 1 Pipe Section 2 Figure 6. Photographs showing the pipe sections in the as-received condition, within the cargo container. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 36#

Page 231Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Evidence Tape (a) Polyethylene Foam Figure 7. (b) Photographs showing failure location on PS 2 a) before and b) after evidence tape and a clear protective wrapping was removed. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 37#

Page 232Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing (a) ` Figure 8. (b) Photographs showing the failure location a) while on site (May 28, 2015) and b) after transit to DNV GL’s facility (June 15, 2015). DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 38#

Page 233Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing GW 5940 Figure 9. Photograph showing a stamp on the internal surface of the failure joint, near the D/S GW (GW 5940). “NIPPON” and “24” are legible. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 39#

Page 234Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Cracking Wrinkle band in tape Defined Wrinkles Cracking Figure 10. Photographs showing the condition of the external tape on the failure joint. Tape measure indicates distance to upstream girth weld. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 40 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 235Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Failure Location GW 5930 6:00 Figure 11. Photograph showing the internal surface of the external tape from the failure joint. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 41#

Page 236Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing 26’ 38.5’ 12:00 6:00 Failure Location Figure 12. Photograph showing the internal surface of the external tape at the failure location. Tape measure indicates distance to upstream girth weld. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 42 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 237Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Failure Location Cracking 6:00 Wrinkles 12:00 29’ 36’ Figure 13. Photograph showing the external surface of the external tape at the failure location. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 43#

Page 238Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing 15’ 19’ Wrinkle Impressions 6:00 Compression (a) 35’ 32’ Wrinkle Impressions Figure 15 6:00 Compression Figure 14. (b) Photographs showing the external surface of the PU insulation at a) the U/S end of PS 2 (14’ to 20’ from U/S GW) and b) the failure location (31.5’ to 36.4’ from U/S GW). Tape measure indicates distance to upstream girth weld. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 44#

Page 239Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Wrinkle Impressions Crack Figure 15. Photograph showing a crack in the PU insulation within a wrinkle. White contrast paint was applied to the surface to facilitate laser scanning and visual inspection. Area shown in Figure 14; scale in mm. Failure Location Disbonded Coal Tar Coating 32’ Figure 17 35.5’ Figure 16. Photograph showing a piece of insulation removed from adjacent to the failure location; near 4:30 orientation. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 45#

Page 240Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing pH Measurement Location Mount 195331-1 Figure 17. Photograph showing corrosion product that was wedged between the pipe surface and polyurethane insulation. Location indicated in Figure 16; scale in mm. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 46#

Page 241Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing 33.35’ to U/S GW Flow 4:30 Nominal Thickness (1.5 inches) Liquid Line Nominal Thickness (1.5 inches) Nominal Thickness Figure 18. Photographs showing the amount of compression in the insulation adjacent to the failure location; near 6:00 orientation. Scale in mm. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 47 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 242Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Flow Area of Disbonded Coating Figure 19. Photograph showing the insulation and coal tar coating separating from the pipe in large sheets on the underside of the pipe; approximately 29’ from U/S GW. Flow Released Product Staining Area of Disbonded Coating 6:00 6 in Figure 20. Photograph showing the insulation and coal tar coating separating from the pipe in large sheets on the underside of the pipe; approximately 17’ from U/S GW. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 48#

Page 243Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing 12:00 2 in Figure 21. Photograph showing blistered coal tar coating along the 12:00 orientation on PS 2; approximately 28’ from U/S GW. Feature 1 16.2-16.9’ from U/S GW Feature 2 17.7-19.5’ from U/S GW 6:00 Mount 195370-1 17’ 19’ Flow Figure 22. Photograph showing external corrosion features on PS 2; Feature 1 and Feature 2. Tape measure indicates distance to upstream girth weld. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 49#

Page 244Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Flow Feature 3 31.2-31.85’ from U/S GW 6:00 Figure 23. Photograph showing an external corrosion feature on PS 2; Feature 3. Tape measure indicates distance to U/S GW. Flow 4:22 Feature 4 (Failure) 33.0-34.0’ from U/S GW Feature 6 34.2-34.42’ from U/S GW Figure 24. Photograph showing external corrosion features on PS 2; Feature 4 and Feature 6. Tape measure indicates distance to U/S GW (Note: tape slipped 0.1’ to the right). DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 50#

Page 245Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Flow 33’ 34’ Feature 5 33.75-34.0’ from U/S GW Figure 25. Photograph showing an external corrosion feature on PS 2; Feature 5. Tape measure indicates distance to U/S GW. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 51#

Page 246Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Flow 32’ 33’ 34’ 35’ (a) Flow 7:00 5:00 3:00 32’ 33’ 34’ 35’ Figure 26. (b) Photographs showing the a) external and b) internal surfaces of the pipe section at the failure location (Coupon 1). DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 52#

Page 247Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing (a) 7:30 38’ 39’ 40’ Figure 28 Figure 27. (b) Photographs showing Pipe Section 2 from 37.7 feet to 40.4 feet from GW 5930 a) before and b) after the composite repair sleeve was removed. Tape measure indicates distance to U/S GW (GW 5930). DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 53#

Page 248Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing 7:30 Figure 28. Photograph the primary feature repaired on May 13, 2013. Area shown in Figure 27. Tape measure indicates distance to U/S GW. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 54#

Page 249Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing PS 1 PS 2 Flow 6:00 Feature 1 16.2-16.9’ Feature 2 17.7-19.5’ 1 ft Figure 29. Renderings of PS 1 and PS 2 (viewed from the OD surface) from laser scanning data showing Feature 1 and Feature 2; Coupon 2. The transparency of the insulation was changed to show a correlation between the features observed on the insulation with corrosion features observed on the pipe. The scale on the right is from 0.000 to 0.400 inches. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 55 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 250Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing PS 1 PS 2 Flow Failure Location 6:00 Figure 31 Feature 6 34.2-34.42’ Feature 4 33.0-34.0’ Feature 3 31.2-31.85’ Feature 5 33.75-34.0’ 6 in Figure 30. Renderings of PS 1 and PS 2 (viewed from the OD surface) from laser scanning data showing the failure location (Feature 4) and Feature 3, 5, and 6; Coupon 1. The transparency of the insulation was changed to show a correlation between the features observed on the insulation with corrosion features observed on the pipe. The scale on the right is from 0.000 to 0.400 inches. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 56 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 251Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Flow Feature 6 34.2-34.42’ Feature 4 33.0-34.0’ Figure 31. Rendering of the failure location (viewed from the OD surface) from laser scanning data showing the remaining wall thickness along the fracture surface. The scale on the right is from 0.000 to 0.400 inches. Location indicated in Figure 30. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 57 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 252Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Flow 12.99 Figure 32. Rendering of the failure location showing various thickness measurements made along the corrosion feature and failure opening. The scale on the right is from 0.000 to 0.400 inches. Measurements are tabulated in Table 4. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 58 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 253Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Flow GW 5940 7:30 2 in Figure 33. Renderings of the area under the composite repair sleeve showing the thickness profile for the feature repaired on May 13, 2013. The scale on the right is from 0.000 to 0.400 inches. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 59#

Page 254Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Flow 4:00 33’ 31’ 35’ 32’ 34’ Figure 34. Photograph showing the internal surface of the pipe section at the failure location with the 1 × 1 inch grid painted on the surface. Tape measure indicates distance to U/S GW. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 60 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 255Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing DNV GL – OAPUS309DNOR (PP136049) 61 September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP Figure 35. Results of the UT measurements performed on 1” × 1” grid near the failure location; Feature 4 (Coupon 1). C-7 C-6 C-5 C-4 C-3 C-2 C-1 C0 C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 C12 C13 C14 C15 C16 C17 C18 C19 C20 C21 C22 C23 C24 C25 C26 C27 C28 C29 C30 C31 C32 C33 C34 C35 C36 C37 C38 C39 C40 C41 L0 0.364 0.364 0.364 0.364 0.364 0.364 0.364 0.364 0.363 L1 0.365 0.365 0.364 0.363 0.366 0.363 0.363 0.364 0.364 0.364 0.364 0.364 0.363 0.363 0.362 0.363 L2 0.364 0.364 0.365 0.364 0.363 0.364 0.364 0.363 0.364 0.364 0.365 0.364 0.364 0.364 0.365 0.364 0.364 0.365 0.363 0.364 0.361 0.363 0.362 0.362 0.363 0.364 0.364 0.363 0.362 L3 0.364 0.365 0.365 0.365 0.365 0.365 0.365 0.363 0.363 0.364 0.362 0.362 0.362 0.362 0.362 0.363 0.363 0.364 0.363 0.362 0.361 0.363 0.363 0.364 0.364 0.363 0.358 0.340 0.344 0.346 0.336 0.348 0.355 0.356 0.363 0.362 0.363 0.364 0.363 0.363 0.362 0.363 0.363 0.363 0.363 0.363 L4 0.363 0.363 0.363 0.363 0.363 0.364 0.363 0.360 0.361 0.361 0.360 0.361 0.360 0.358 0.355 0.353 0.361 0.362 0.351 0.353 0.349 0.360 0.361 0.361 0.360 0.310 0.116 0.108 0.202 0.073 0.289 0.282 0.321 0.303 0.362 0.361 0.360 0.362 0.360 0.360 0.360 0.363 0.360 0.361 0.360 0.360 0.362 0.361 L5 0.362 0.362 0.361 0.360 0.361 0.362 0.362 0.359 0.361 0.360 0.358 0.358 0.358 0.344 0.346 0.352 0.356 0.352 0.352 0.332 0.353 0.360 0.359 0.358 0.345 0.212 0.220 0.131 0.101 0.078 0.064 0.157 0.317 0.359 0.360 0.359 0.358 0.360 0.359 0.361 0.359 0.358 0.359 0.360 0.360 0.358 L6 0.363 0.362 0.363 0.362 0.355 0.316 0.324 0.351 0.358 0.360 0.360 0.360 0.357 0.355 0.352 0.352 0.348 0.350 0.344 0.348 0.344 0.358 0.343 0.350 0.332 0.335 0.188 0.168 0.147 0.124 0.102 0.127 0.134 0.230 0.357 0.359 0.359 0.358 0.359 0.358 0.359 0.355 0.360 0.361 0.361 0.362 0.361 L7 0.364 0.362 0.362 0.326 0.275 0.305 0.337 0.329 0.362 0.362 0.362 0.361 0.363 0.358 0.358 0.340 0.353 0.348 0.363 0.345 0.326 0.333 0.310 0.314 0.292 0.326 0.262 0.191 0.148 0.141 0.136 0.168 0.168 0.103 0.190 0.360 0.362 0.362 0.360 0.359 0.361 0.360 0.360 0.360 0.353 0.360 0.365 0.365 0.364 L8 0.365 0.363 0.363 0.293 0.309 0.341 0.355 0.358 0.362 0.363 0.361 0.356 0.346 0.339 0.346 0.343 0.322 0.348 0.336 0.359 0.338 0.333 0.338 0.336 0.345 0.332 0.263 0.259 0.243 0.248 0.261 0.231 0.226 0.250 0.297 0.362 0.363 0.361 0.358 0.295 0.361 0.356 0.296 0.326 0.353 0.312 0.357 0.364 0.365 L9 0.365 0.363 0.362 0.286 0.330 0.348 0.362 0.365 0.363 0.363 0.355 0.350 0.354 0.360 0.353 0.358 0.360 0.356 0.362 0.357 0.346 0.350 0.348 0.349 0.351 0.332 0.314 0.231 0.240 0.216 0.266 0.240 0.213 0.222 0.362 0.360 0.362 0.358 0.352 0.333 0.361 0.360 0.363 0.355 0.356 0.324 0.350 0.365 0.363 L10 0.366 0.363 0.366 0.318 0.358 0.364 0.365 0.366 0.364 0.363 0.356 0.351 0.350 0.363 0.363 0.363 0.364 0.367 0.362 0.364 0.360 0.360 0.356 0.348 0.347 0.362 0.289 0.333 0.357 0.360 0.361 0.360 0.306 0.262 0.300 0.314 0.360 0.360 0.334 0.358 0.363 0.364 0.360 0.366 0.366 0.366 0.366 0.365 0.365 L11 0.363 0.362 0.348 0.304 0.341 0.352 0.360 0.360 0.360 0.359 0.358 0.350 0.348 0.359 0.363 0.337 0.360 0.336 0.360 0.360 0.354 0.333 0.329 0.337 0.324 0.350 0.354 0.357 0.362 0.367 0.362 0.362 0.361 0.355 0.356 0.357 0.358 0.360 0.360 0.362 0.362 0.362 0.362 0.363 0.362 0.363 0.362 0.364 0.362 L12 0.363 0.361 0.362 0.318 0.336 0.361 0.362 0.361 0.361 0.361 0.359 0.358 0.358 0.361 0.361 0.360 0.361 0.360 0.361 0.358 0.333 0.333 0.320 0.337 0.355 0.355 0.359 0.362 0.362 0.362 0.363 0.360 0.362 0.363 0.361 0.362 0.362 0.362 0.362 0.362 0.362 0.363 0.363 0.363 0.363 L13 0.362 0.363 0.362 0.352 0.356 0.363 0.364 0.359 0.360 0.359 0.360 0.361 0.358 0.359 0.360 0.360 0.359 0.359 0.360 0.354 0.332 0.324 0.351 0.348 0.354 0.348 0.359 0.362 0.362 0.361 0.361 0.362 0.361 0.362 0.363 0.362 0.361 L14 0.363 0.363 0.362 0.290 0.305 0.354 0.363 0.360 0.361 0.361 0.360 0.361 0.360 0.360 0.360 0.360 0.360 0.359 0.358 0.344 0.339 0.356 0.352 0.355 0.346 0.356 0.361 0.362 0.362 0.361 0.364 0.363 0.362 L15 0.363 0.362 0.267 0.274 0.282 0.289 0.304 0.344 0.361 0.360 0.360 0.361 0.360 0.361 0.360 0.360 0.362 0.358 0.355 0.343 0.357 0.329 0.353 0.359 0.349 0.357 0.360 0.361 0.362 L16 0.362 0.362 0.255 0.227 0.232 0.249 0.273 0.284 0.334 0.361 0.360 0.360 0.360 0.352 0.353 0.333 0.322 0.358 0.358 0.358 0.358 L17 0.362 0.362 0.362 0.241 0.213 0.295 0.297 0.341 0.360 0.360 0.358 0.359 0.360 0.350 0.358 0.353 0.352 0.351 0.360 0.355 0.362 L18 0.362 0.362 0.338 0.294 0.231 0.239 0.260 0.278 0.354 0.359 0.360 0.360 0.357 0.358 0.358 0.360 0.359 0.360 0.360 0.353 0.353 L19 0.364 0.362 0.310 0.165 0.196 0.229 0.250 0.268 0.305 0.361 0.360 0.362 0.360 0.362 0.361 0.361 0.360 0.360 0.362 0.361 0.361 L20 0.364 0.364 0.356 0.298 0.218 0.234 0.264 0.273 0.314 0.361 0.363 0.361 0.361 0.361 0.359 0.361 0.361 0.361 0.363 L21 0.364 0.362 0.363 0.362 0.196 0.222 0.265 0.273 0.303 0.361 0.363 0.362 0.361 0.361 0.360 L22 0.363 0.362 0.362 0.362 0.360 0.246 0.264 0.327 0.361 0.362 L23 0.362 0.363 0.362 0.363 0.229 0.244 0.252 0.299 0.356 0.362 L24 0.363 0.364 0.362 0.361 0.356 0.275 0.363 0.360 0.362 0.360 L25 0.362 0.362 0.363 0.363 0.362 0.363 0.362 0.362 0.361 0.362 C17 C18 C19 C20 C21 C22 C23 C24 C25 C26 C27 C28 C29 L0 0.363 L1 0.364 0.364 0.364 0.364 0.363 0.363 0.362 0.363 L2 0.364 0.364 0.365 0.363 0.364 0.361 0.363 0.362 0.362 0.363 0.364 0.364 L3 0.364 0.363 0.358 0.340 0.344 0.346 0.336 0.348 0.355 0.356 0.363 0.362 0.363 L4 0.360 0.310 0.116 0.108 0.202 0.073 0.289 0.282 0.321 0.303 0.362 0.361 0.360 L5 0.345 0.212 0.220 0.131 0.101 0.078 0.064 0.157 0.317 0.359 0.360 L6 0.332 0.335 0.188 0.168 0.147 0.124 0.102 0.127 0.134 0.230 0.357 0.359 L7 0.292 0.326 0.262 0.191 0.148 0.141 0.136 0.168 0.168 0.103 0.190 0.360 0.362 L8 0.345 0.332 0.263 0.259 0.243 0.248 0.261 0.231 0.226 0.250#

Page 255, passage 20.297 0.362 0.363 L9 0.351 0.332 0.314 0.231 0.240 0.216 0.266 0.240 0.213 0.222 0.362 0.360 0.362 L10 0.347 0.362 0.289 0.333 0.357 0.360 0.361 0.360 0.306 0.262 0.300 0.314 0.360 L11 0.324 0.350 0.354 0.357 0.362 0.367 0.362 0.362 0.361 0.355 0.356 0.357 0.358 L12 0.355 0.355 0.359 0.362 0.362 0.362 0.363 0.360 0.362 0.363 0.361 0.362 0.362 L13 0.354 0.348 0.359 0.362 0.362 0.361 0.361 0.362 0.361 0.362 0.363 L14 0.346 0.356 0.361 0.362 0.362 0.361 0.364 0.363 L15 0.349 0.357 0.360 0.361 L16 0.358 0.358 0.358 L17 0.360 0.355 0.362 L18 0.360 0.353 0.353 L19 0.362 0.361 0.361 L20 0.361 0.363 L21 33.3’ Flow#

Page 256Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing DNV GL – OAPUS309DNOR (PP136049) 62 September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP Figure 36. Results of the UT measurements performed on 1” × 1” grid near Feature 1 (Coupon 2). C0 C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 C12 C13 C14 C15 C16 C17 C18 C19 C20 C21 C22 C23 C24 C25 C26 C27 C28 C29 C30 C31 C32 C33 C34 C35 C36 C37 C38 C39 C40 C41 C42 C43 C44 C45 C46 C47 C48 C49 C50 C51 C52 C53 C54 C55 C56 C57 C58 C59 C60 C61 C62 C63 C64 C65 C66 C67 C68 C69 C70 C71 C72 C73 C74 C75 C76 C77 C78 C79 L0 0.361 0.362 0.362 0.363 0.362 0.362 0.361 0.361 0.362 0.362 0.362 0.361 0.362 0.362 0.362 0.362 0.360 0.362 0.362 0.362 0.362 0.361 0.362 0.362 0.362 0.362 0.361 0.360 0.361 0.362 0.362 0.361 0.361 0.360 0.362 0.361 0.359 0.361 0.362 0.361 0.361 0.360 0.361 0.361 0.361 0.362 0.360 0.361 0.362 0.362 0.362 0.360 0.361 0.362 0.362 0.362 0.361 0.362 0.362 0.362 0.362 0.361 0.363 0.363 0.362 0.362 0.362 0.362 0.362 0.362 0.363 0.362 0.363 0.363 0.363 0.362 0.361 0.364 0.364 0.365 L1 0.363 0.363 0.364 0.365 0.364 0.364 0.364 0.362 0.364 0.363 0.362 0.362 0.363 0.363 0.364 0.364 0.362 0.362 0.363 0.363 0.363 0.362 0.363 0.364 0.363 0.363 0.362 0.362 0.363 0.363 0.364 0.362 0.362 0.363 0.364 0.363 0.361 0.362 0.363 0.363 0.363 0.362 0.361 0.362 0.362 0.362 0.362 0.362 0.363 0.363 0.364 0.362 0.362 0.362 0.363 0.363 0.361 0.364 0.363 0.363 0.363 0.362 0.363 0.362 0.363 0.363 0.364 0.363 0.363 0.364 0.364 0.363 0.364 0.364 0.363 0.364 0.363 0.365 0.365 0.366 L2 0.364 0.365 0.365 0.365 0.365 0.366 0.365 0.365 0.365 0.365 0.365 0.364 0.365 0.365 0.365 0.365 0.363 0.364 0.365 0.365 0.364 0.363 0.364 0.365 0.364 0.362 0.362 0.364 0.362 0.364 0.364 0.362 0.364 0.364 0.364 0.364 0.362 0.363 0.362 0.363 0.363 0.365 0.363 0.364 0.364 0.364 0.364 0.364 0.364 0.364 0.364 0.363 0.364 0.362 0.363 0.363 0.363 0.364 0.363 0.364 0.364 0.363 0.363 0.363 0.364 0.364 0.364 0.365 0.365 0.366 0.366 0.366 0.365 0.365 0.365 0.365 0.365 0.365 0.366 0.365 L3 0.364 0.365 0.365 0.365 0.365 0.365 0.365 0.365 0.365 0.365 0.365 0.364 0.365 0.365 0.365 0.365 0.364 0.364 0.365 0.365 0.362 0.363 0.362 0.364 0.363 0.364 0.365 0.365 0.360 0.362 0.364 0.363 0.363 0.362 0.364 0.363 0.361 0.360 0.356 0.362 0.363 0.363 0.364 0.363 0.363 0.363 0.363 0.363 0.362 0.358 0.362 0.360 0.358 0.360 0.361 0.363 0.363 0.362 0.363 0.363 0.363 0.362 0.361 0.357 0.324 0.360 0.364 0.365 0.365 0.367 0.366 0.365 0.365 0.365 0.365 0.365 0.365 0.365 0.365 0.365 L4 0.363 0.364 0.363 0.363 0.363 0.362 0.361 0.363 0.363 0.361 0.362 0.359 0.364 0.365 0.362 0.360 0.362 0.360 0.363 0.360 0.356 0.359 0.360 0.363 0.360 0.365 0.343 0.352 0.356 0.360 0.362 0.360 0.354 0.360 0.361 0.361 0.363 0.349 0.329 0.353 0.363 0.360 0.364 0.361 0.362 0.362 0.360 0.358 0.361 0.362 0.360 0.358 0.360 0.362 0.361 0.362 0.360 0.358 0.360 0.358 0.361 0.355 0.338 0.334 0.340 0.361 0.364 0.359 0.360 0.361 0.362 0.361 0.361 0.360 0.360 0.361 0.360 0.361 0.361 0.362 L5 0.363 0.364 0.365 0.365 0.365 0.365 0.364 0.365 0.365 0.365 0.364 0.364 0.365 0.365 0.364 0.364 0.363 0.363 0.363 0.362 0.365 0.364 0.366 0.365 0.364 0.365 0.364 0.364 0.363 0.361 0.365 0.362 0.360 0.326 0.327 0.325 0.364 0.363 0.359 0.364 0.361 0.365 0.363 0.362 0.350 0.363 0.364 0.362 0.359 0.363 0.300 0.308 0.361 0.366 0.362 0.362 0.361 0.359 0.363 0.359 0.360 0.363 0.357 0.343 0.363 0.365 0.366 0.364 0.361 0.361 0.363 0.362 0.362 0.364 0.365 0.365 0.366 0.365 0.365 0.364 L6 0.363 0.362 0.362 0.363 0.362 0.364 0.363 0.361 0.360 0.361 0.361 0.358 0.360 0.360 0.358 0.356 0.347 0.356 0.362 0.363 0.362 0.363 0.364 0.364 0.363 0.364 0.358 0.363 0.363 0.362 0.362 0.357 0.358 0.359 0.361 0.359 0.350 0.349 0.359 0.360 0.361 0.361 0.361 0.362 0.361 0.358 0.360 0.362 0.359 0.355 0.308 0.278 0.359 0.364 0.362 0.360 0.355 0.355 0.353 0.362 0.358 0.355 0.323 0.314 0.359 0.358 0.363 0.343 0.329 0.334 0.361 0.355 0.358 0.357 0.361 0.363 0.362 0.363 0.363 0.363 L7 0.361 0.361 0.362 0.362 0.362 0.363 0.361 0.360 0.362 0.362 0.363 0.361 0.362 0.362 0.361 0.357 0.361 0.362 0.363 0.362 0.363 0.361 0.362 0.363 0.343 0.360 0.361 0.362 0.363 0.362 0.362 0.360 0.359 0.359 0.362 0.359 0.359 0.324 0.359 0.364 0.361 0.360 0.360 0.361 0.361 0.362 0.359 0.357 0.362 0.358 0.358 0.355 0.327 0.330 0.322 0.330 0.347 0.350 0.354 0.363 0.362 0.338 0.311 0.303 0.360 0.351 0.361 0.362 0.359 0.361 0.340 0.337 0.325 0.359 0.360 0.361 0.362 0.362 0.363 0.362 L8 0.361 0.362 0.361 0.362 0.362 0.362 0.357 0.360 0.362 0.365 0.362 0.362 0.362 0.362 0.362 0.361 0.362 0.362 0.362 0.362 0.362 0.362 0.363 0.354 0.311 0.337 0.361 0.362 0.361 0.362 0.363 0.360 0.361 0.360 0.361 0.359 0.354 0.358 0.361 0.362 0.362 0.361 0.360 0.361 0.362 0.361 0.360 0.361 0.361 0.342 0.359 0.354 0.348 0.360 0.357 0.347 0.352 0.341 0.355 0.354 0.348 0.345 0.359 0.356 0.359 0.361 0.350 0.352 0.339 0.355 0.359 0.350 0.337 0.346 0.358 0.361 0.362 0.363 0.363 0.362 L9 0.361 0.360 0.363 0.361 0.362 0.364 0.361 0.362 0.362 0.366 0.363 0.362 0.362 0.363 0.363 0.362 0.361 0.362 0.363 0.363 0.363 0.363 0.363 0.360 0.334 0.354 0.362 0.362 0.355 0.350 0.355 0.359 0.361 0.360 0.361 0.358 0.358 0.360 0.362 0.362 0.362 0.361 0.362 0.362 0.362 0.362 0.361 0.362 0.362 0.361 0.358 0.356 0.344 0.357 0.362 0.359 0.357 0.352 0.360 0.358 0.329 0.351 0.360 0.305 0.355 0.361 0.343 0.359 0.360 0.357 0.352 0.336 0.338 0.311 0.289 0.360 0.359 0.362 0.364 0.364 L10 0.364 0.363 0.363 0.363 0.362 0.365 0.362 0.364 0.363 0.357 0.366 0.363 0.364 0.364 0.364 0.363 0.365 0.364 0.364 0.365 0.363 0.364 0.366 0.364 0.362 0.364 0.363 0.361 0.342 0.339 0.339 0.349 0.343 0.359 0.347 0.353 0.360 0.361 0.364 0.362 0.363 0.363 0.365 0.365 0.365 0.364 0.367 0.366 0.366 0.363 0.358 0.358 0.358 0.331 0.337 0.362 0.356 0.357 0.360 0.360 0.359 0.358 0.362 0.362 0.360 0.358 0.332 0.362 0.360 0.360 0.351 0.356 0.315 0.297 0.296 0.329 0.359 0.362 0.362 0.362 L11 0.361 0.362 0.363 0.363 0.363 0.361 0.362#

Page 256, passage 20.361 0.360 0.352 0.361 0.362 0.361 0.362 0.362 0.362 0.361 0.362 0.361 0.363 0.362 0.364 0.363 0.362 0.363 0.364 0.355 0.340 0.329 0.312 0.305 0.303 0.307 0.336 0.332 0.352 0.352 0.361 0.362 0.362 0.362 0.361 0.362 0.361 0.362 0.362 0.361 0.363 0.364 0.366 0.356 0.358 0.352 0.304 0.357 0.359 0.356 0.360 0.350 0.321 0.316 0.336 0.362 0.360 0.360 0.361 0.359 0.360 0.358 0.338 0.346 0.336 0.303 0.275 0.306 0.344 0.359 0.363 0.362 0.362 L12 0.363 0.361 0.362 0.362 0.363 0.361 0.361 0.362 0.362 0.360 0.362 0.360 0.361 0.362 0.361 0.362 0.360 0.361 0.362 0.362 0.363 0.363 0.362 0.362 0.363 0.362 0.348 0.336 0.324 0.351 0.269 0.234 0.306 0.325 0.353 0.359 0.355 0.358 0.358 0.362 0.361 0.360 0.362 0.360 0.362 0.361 0.360 0.361 0.361 0.361 0.358 0.353 0.355 0.344 0.238 0.247 0.323 0.306 0.271 0.270 0.245 0.285 0.285 0.358 0.358 0.355 0.360 0.356 0.360 0.361 0.362 0.351 0.297 0.299 0.335 0.362 0.361 0.363 0.363 0.363 L13 0.360 0.362 0.364 0.362 0.362 0.362 0.360 0.362 0.361 0.331 0.361 0.361 0.361 0.361 0.362 0.361 0.360 0.362 0.362 0.362 0.363 0.362 0.362 0.361 0.362 0.362 0.351 0.337 0.349 0.358 0.264 0.263 0.295 0.324 0.356 0.362 0.361 0.360 0.358 0.361 0.361 0.360 0.361 0.362 0.362 0.361 0.361 0.361 0.362 0.360 0.357 0.352 0.258 0.252 0.255 0.281 0.357 0.291 0.237 0.274 0.282 0.271 0.263 0.291 0.251 0.300 0.360 0.362 0.362 0.362 0.362 0.355 0.329 0.356 0.335 0.361 0.362 0.363 0.363 0.363 L14 0.360 0.361 0.363 0.363 0.363 0.362 0.363 0.361 0.363 0.359 0.362 0.360 0.360 0.361 0.361 0.362 0.361 0.362 0.362 0.363 0.363 0.362 0.363 0.363 0.362 0.363 0.351 0.337 0.350 0.355 0.359 0.335 0.311 0.322 0.358 0.362 0.361 0.358 0.353 0.360 0.360 0.361 0.362 0.362 0.363 0.362 0.361 0.362 0.362 0.298 0.286 0.355 0.226 0.263 0.215 0.217 0.260 0.313 0.215 0.238 0.211 0.214 0.258 0.289 0.276 0.272 0.362 0.363 0.363 0.361 0.360 0.358 0.358 0.358 0.361 0.361 0.361 0.361 0.362 0.362 L15 0.361 0.363 0.363 0.363 0.363 0.363 0.362 0.362 0.363 0.362 0.359 0.360 0.362 0.356 0.360 0.362 0.328 0.360 0.362 0.362 0.363 0.362 0.362 0.362 0.363 0.361 0.357 0.345 0.357 0.357 0.361 0.359 0.343 0.342 0.363 0.359 0.354 0.360 0.344 0.347 0.360 0.360 0.359 0.362 0.363 0.362 0.362 0.363 0.353 0.311 0.275 0.257 0.242 0.247 0.252 0.237 0.267 0.362 0.344 0.262 0.274 0.289 0.269 0.298 0.316 0.317 0.362 0.361 0.364 0.363 0.362 0.358 0.359 0.359 0.362 0.362 0.362 0.362 0.363 0.363 L16 0.362 0.362 0.363 0.363 0.365 0.362 0.362 0.361 0.364 0.362 0.362 0.360 0.362 0.362 0.363 0.363 0.360 0.362 0.363 0.362 0.363 0.362 0.362 0.362 0.363 0.363 0.362 0.361 0.361 0.363 0.361 0.361 0.359 0.357 0.361 0.360 0.362 0.361 0.361 0.362 0.362 0.361 0.360 0.362 0.362 0.349 0.362 0.333 0.286 0.272 0.296 0.255 0.276 0.256 0.258 0.298 0.361 0.362 0.361 0.363 0.282 0.284 0.357 0.362 0.361 0.362 0.362 0.363 0.364 0.363 0.363 0.362 0.361 0.360 0.364 0.363 0.364 0.364 0.365 0.364 L17 0.362 0.363 0.363 0.363 0.363 0.363 0.362 0.362 0.362 0.362 0.363 0.361 0.362 0.362 0.362 0.362 0.361 0.362 0.362 0.362 0.364 0.363 0.363 0.363 0.363 0.363 0.361 0.348 0.322 0.361 0.361 0.361 0.361 0.357 0.357 0.362 0.361 0.361 0.361 0.361 0.361 0.361 0.360 0.360 0.352 0.260 0.264 0.267 0.266 0.251 0.278 0.266 0.270 0.255 0.290 0.362 0.364 0.362 0.363 0.362 0.361 0.361 0.361 0.363 0.362 0.363 0.361 0.361 0.362 0.362 0.364 0.363 0.362 0.362 0.363 0.365 0.364 0.363 0.365 0.363 L18 0.362 0.362 0.362 0.364 0.364 0.365 0.362 0.364 0.362 0.362 0.363 0.361 0.362 0.361 0.362 0.363 0.361 0.362 0.364 0.363 0.361 0.362 0.363 0.340 0.363 0.364 0.362 0.360 0.362 0.362 0.362 0.361 0.362 0.361 0.362 0.362 0.363 0.362 0.362 0.360 0.307 0.355 0.360 0.360 0.350 0.252 0.250 0.258 0.264 0.255 0.285 0.258 0.268 0.271 0.361 0.363 0.362 0.362 0.363 0.363 0.363 0.362 0.363 0.362 0.361 0.362 0.363 0.363 0.363 0.362 0.362 0.362 0.363 0.362 0.363 0.363 0.363 0.363 0.365 0.366 L19 0.361 0.362 0.363 0.362 0.364 0.363 0.362 0.362 0.363 0.362 0.362 0.362 0.361 0.361 0.362 0.362 0.361 0.362 0.363 0.364 0.361 0.363 0.363 0.362 0.363 0.362 0.363 0.361 0.362 0.362 0.361 0.363 0.360 0.362 0.362 0.361 0.363 0.361 0.361 0.361 0.361 0.360 0.361 0.360 0.347 0.312 0.233 0.256 0.264 0.261 0.269 0.281 0.269 0.360 0.363 0.362 0.363 0.362 0.364 0.362 0.363 0.363 0.363 0.362 0.363 0.362 0.362 0.364 0.364 0.364 0.363 0.362 0.363 0.362 0.362 0.364 0.363 0.363 0.363 0.362 L20 0.362 0.363 0.362 0.362 0.362 0.363 0.361 0.362 0.362 0.362 0.362 0.362 0.362 0.361 0.362 0.362 0.361 0.363 0.364 0.363 0.363 0.363 0.363 0.363 0.363 0.363 0.362 0.362 0.362 0.362 0.362 0.362 0.361 0.362 0.362 0.362 0.363 0.360 0.362 0.363 0.363 0.362 0.362 0.363 0.361 0.362 0.361 0.362 0.362 0.363 0.363 0.362 0.364 0.364 0.364 0.364 0.363 0.363 0.364 0.363 0.363 0.363 0.363 0.363 0.364 0.364 0.364 0.363 0.363 0.364 0.364 0.363 0.363 0.363 0.364 0.364 0.363 0.364 0.365 0.363 L21 0.363 0.363 0.363 0.363 0.363 0.362 0.362 0.362 0.362 0.362 0.362 0.362 0.362 0.362 0.363 0.362 0.361 0.362 0.364 0.363 0.363 0.362 0.362 0.363 0.363 0.363 0.363 0.362 0.362 0.363 0.363 0.361 0.362 0.362 0.362 0.362 0.362 0.362 0.363 0.363 0.363 0.362 0.363 0.363 0.362 0.363 0.362 0.362 0.362 0.363 0.363 0.362 0.365 0.363 0.364 0.363 0.363 0.363 0.363 0.363 0.364 0.364 0.364 0.364 0.364 0.364 0.364 0.364 0.364 0.363 0.364 0.363 0.363 0.364 0.363 0.364 0.364 0.363 0.365 0.363 L22 0.361 0.363 0.361 0.362 0.361 0.363 0.362 0.362 0.362 0.362 0.362 0.361 0.361 0.362 0.361 0.360 0.360 0.360 0.360 0.361 0.363 0.361 0.362 0.361 0.362 0.360 0.361 0.362 0.362 0.363 0.361 0.360 0.360 0.363 0.361 0.360 0.360 0.361 0.361 0.361 0.360 0.360 0.361 0.360 0.361 0.362 0.363 0.362 0.362 0.362 0.362 0.362 0.362 0.362 0.362 0.362 0.361 0.362 0.365 0.363 0.362 0.362 0.362 0.362 0.363 0.362 0.362 0.362 0.363 0.363 0.364 0.361 0.362 0.362 0.362 0.364 0.362 0.362 0.364 0.363 C25 C26 C27 C28 C29 C30 C31 C32 C33 C34 L0 0.362 0.361 0.360 0.361 0.362 0.362 0.361 0.361 0.360 0.362 L1 0.363 0.362 0.362 0.363 0.363 0.364 0.362 0.362 0.363 0.364 L2 0.362 0.362 0.364 0.362 0.364 0.364 0.362 0.364 0.364 0.364 L3 0.364#

Page 256, passage 30.365 0.365 0.360 0.362 0.364 0.363 0.363 0.362 0.364 L4 0.365 0.343 0.352 0.356 0.360 0.362 0.360 0.354 0.360 0.361 L5 0.365 0.364 0.364 0.363 0.361 0.365 0.362 0.360 0.326 0.327 L6 0.364 0.358 0.363 0.363 0.362 0.362 0.357 0.358 0.359 0.361 L7 0.360 0.361 0.362 0.363 0.362 0.362 0.360 0.359 0.359 0.362 L8 0.337 0.361 0.362 0.361 0.362 0.363 0.360 0.361 0.360 0.361 L9 0.354 0.362 0.362 0.355 0.350 0.355 0.359 0.361 0.360 0.361 L10 0.364 0.363 0.361 0.342 0.339 0.339 0.349 0.343 0.359 0.347 L11 0.364 0.355 0.340 0.329 0.312 0.305 0.303 0.307 0.336 0.332 L12 0.362 0.348 0.336 0.324 0.351 0.269 0.234 0.306 0.325 0.353 L13 0.362 0.351 0.337 0.349 0.358 0.264 0.263 0.295 0.324 0.356 L14 0.363 0.351 0.337 0.350 0.355 0.359 0.335 0.311 0.322 0.358 L15 0.361 0.357 0.345 0.357 0.357 0.361 0.359 0.343 0.342 0.363 L16 0.363 0.362 0.361 0.361 0.363 0.361 0.361 0.359 0.357 0.361 L17 0.363 0.361 0.348 0.322 0.361 0.361 0.361 0.361 0.357 0.357 L18 0.364 0.362 0.360 0.362 0.362 0.362 0.361 0.362 0.361 0.362 L19 0.362 0.363 0.361 0.362 0.362 0.361 0.363 0.360 0.362 0.362 L20 0.363 0.362 0.362 0.362 0.362 0.362 0.362 0.361 0.362 0.362 L21 0.363 0.363 0.362 0.362 0.363 0.363 0.361 0.362 0.362 0.362 L22 0.360 0.361 0.362 0.362 0.363 0.361 0.360 0.360 0.363 0.361 16.1’ Flow#

Page 257Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing DNV GL – OAPUS309DNOR (PP136049) 63 September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP Figure 37. Results of the UT measurements performed on 1” × 1” grid near Feature 2 (Coupon 2). C0 C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 C12 C13 C14 C15 C16 C17 C18 C19 C20 C21 C22 C23 C24 C25 C26 C27 C28 C29 C30 C31 C32 C33 C34 C35 C36 C37 C38 C39 C40 C41 C42 C43 C44 C45 C46 C47 C48 C49 C50 C51 C52 C53 C54 C55 C56 C57 C58 C59 C60 C61 C62 C63 C64 C65 C66 C67 C68 C69 C70 C71 C72 C73 C74 C75 C76 C77 C78 C79 L0 0.361 0.362 0.362 0.363 0.362 0.362 0.361 0.361 0.362 0.362 0.362 0.361 0.362 0.362 0.362 0.362 0.360 0.362 0.362 0.362 0.362 0.361 0.362 0.362 0.362 0.362 0.361 0.360 0.361 0.362 0.362 0.361 0.361 0.360 0.362 0.361 0.359 0.361 0.362 0.361 0.361 0.360 0.361 0.361 0.361 0.362 0.360 0.361 0.362 0.362 0.362 0.360 0.361 0.362 0.362 0.362 0.361 0.362 0.362 0.362 0.362 0.361 0.363 0.363 0.362 0.362 0.362 0.362 0.362 0.362 0.363 0.362 0.363 0.363 0.363 0.362 0.361 0.364 0.364 0.365 L1 0.363 0.363 0.364 0.365 0.364 0.364 0.364 0.362 0.364 0.363 0.362 0.362 0.363 0.363 0.364 0.364 0.362 0.362 0.363 0.363 0.363 0.362 0.363 0.364 0.363 0.363 0.362 0.362 0.363 0.363 0.364 0.362 0.362 0.363 0.364 0.363 0.361 0.362 0.363 0.363 0.363 0.362 0.361 0.362 0.362 0.362 0.362 0.362 0.363 0.363 0.364 0.362 0.362 0.362 0.363 0.363 0.361 0.364 0.363 0.363 0.363 0.362 0.363 0.362 0.363 0.363 0.364 0.363 0.363 0.364 0.364 0.363 0.364 0.364 0.363 0.364 0.363 0.365 0.365 0.366 L2 0.364 0.365 0.365 0.365 0.365 0.366 0.365 0.365 0.365 0.365 0.365 0.364 0.365 0.365 0.365 0.365 0.363 0.364 0.365 0.365 0.364 0.363 0.364 0.365 0.364 0.362 0.362 0.364 0.362 0.364 0.364 0.362 0.364 0.364 0.364 0.364 0.362 0.363 0.362 0.363 0.363 0.365 0.363 0.364 0.364 0.364 0.364 0.364 0.364 0.364 0.364 0.363 0.364 0.362 0.363 0.363 0.363 0.364 0.363 0.364 0.364 0.363 0.363 0.363 0.364 0.364 0.364 0.365 0.365 0.366 0.366 0.366 0.365 0.365 0.365 0.365 0.365 0.365 0.366 0.365 L3 0.364 0.365 0.365 0.365 0.365 0.365 0.365 0.365 0.365 0.365 0.365 0.364 0.365 0.365 0.365 0.365 0.364 0.364 0.365 0.365 0.362 0.363 0.362 0.364 0.363 0.364 0.365 0.365 0.360 0.362 0.364 0.363 0.363 0.362 0.364 0.363 0.361 0.360 0.356 0.362 0.363 0.363 0.364 0.363 0.363 0.363 0.363 0.363 0.362 0.358 0.362 0.360 0.358 0.360 0.361 0.363 0.363 0.362 0.363 0.363 0.363 0.362 0.361 0.357 0.324 0.360 0.364 0.365 0.365 0.367 0.366 0.365 0.365 0.365 0.365 0.365 0.365 0.365 0.365 0.365 L4 0.363 0.364 0.363 0.363 0.363 0.362 0.361 0.363 0.363 0.361 0.362 0.359 0.364 0.365 0.362 0.360 0.362 0.360 0.363 0.360 0.356 0.359 0.360 0.363 0.360 0.365 0.343 0.352 0.356 0.360 0.362 0.360 0.354 0.360 0.361 0.361 0.363 0.349 0.329 0.353 0.363 0.360 0.364 0.361 0.362 0.362 0.360 0.358 0.361 0.362 0.360 0.358 0.360 0.362 0.361 0.362 0.360 0.358 0.360 0.358 0.361 0.355 0.338 0.334 0.340 0.361 0.364 0.359 0.360 0.361 0.362 0.361 0.361 0.360 0.360 0.361 0.360 0.361 0.361 0.362 L5 0.363 0.364 0.365 0.365 0.365 0.365 0.364 0.365 0.365 0.365 0.364 0.364 0.365 0.365 0.364 0.364 0.363 0.363 0.363 0.362 0.365 0.364 0.366 0.365 0.364 0.365 0.364 0.364 0.363 0.361 0.365 0.362 0.360 0.326 0.327 0.325 0.364 0.363 0.359 0.364 0.361 0.365 0.363 0.362 0.350 0.363 0.364 0.362 0.359 0.363 0.300 0.308 0.361 0.366 0.362 0.362 0.361 0.359 0.363 0.359 0.360 0.363 0.357 0.343 0.363 0.365 0.366 0.364 0.361 0.361 0.363 0.362 0.362 0.364 0.365 0.365 0.366 0.365 0.365 0.364 L6 0.363 0.362 0.362 0.363 0.362 0.364 0.363 0.361 0.360 0.361 0.361 0.358 0.360 0.360 0.358 0.356 0.347 0.356 0.362 0.363 0.362 0.363 0.364 0.364 0.363 0.364 0.358 0.363 0.363 0.362 0.362 0.357 0.358 0.359 0.361 0.359 0.350 0.349 0.359 0.360 0.361 0.361 0.361 0.362 0.361 0.358 0.360 0.362 0.359 0.355 0.308 0.278 0.359 0.364 0.362 0.360 0.355 0.355 0.353 0.362 0.358 0.355 0.323 0.314 0.359 0.358 0.363 0.343 0.329 0.334 0.361 0.355 0.358 0.357 0.361 0.363 0.362 0.363 0.363 0.363 L7 0.361 0.361 0.362 0.362 0.362 0.363 0.361 0.360 0.362 0.362 0.363 0.361 0.362 0.362 0.361 0.357 0.361 0.362 0.363 0.362 0.363 0.361 0.362 0.363 0.343 0.360 0.361 0.362 0.363 0.362 0.362 0.360 0.359 0.359 0.362 0.359 0.359 0.324 0.359 0.364 0.361 0.360 0.360 0.361 0.361 0.362 0.359 0.357 0.362 0.358 0.358 0.355 0.327 0.330 0.322 0.330 0.347 0.350 0.354 0.363 0.362 0.338 0.311 0.303 0.360 0.351 0.361 0.362 0.359 0.361 0.340 0.337 0.325 0.359 0.360 0.361 0.362 0.362 0.363 0.362 L8 0.361 0.362 0.361 0.362 0.362 0.362 0.357 0.360 0.362 0.365 0.362 0.362 0.362 0.362 0.362 0.361 0.362 0.362 0.362 0.362 0.362 0.362 0.363 0.354 0.311 0.337 0.361 0.362 0.361 0.362 0.363 0.360 0.361 0.360 0.361 0.359 0.354 0.358 0.361 0.362 0.362 0.361 0.360 0.361 0.362 0.361 0.360 0.361 0.361 0.342 0.359 0.354 0.348 0.360 0.357 0.347 0.352 0.341 0.355 0.354 0.348 0.345 0.359 0.356 0.359 0.361 0.350 0.352 0.339 0.355 0.359 0.350 0.337 0.346 0.358 0.361 0.362 0.363 0.363 0.362 L9 0.361 0.360 0.363 0.361 0.362 0.364 0.361 0.362 0.362 0.366 0.363 0.362 0.362 0.363 0.363 0.362 0.361 0.362 0.363 0.363 0.363 0.363 0.363 0.360 0.334 0.354 0.362 0.362 0.355 0.350 0.355 0.359 0.361 0.360 0.361 0.358 0.358 0.360 0.362 0.362 0.362 0.361 0.362 0.362 0.362 0.362 0.361 0.362 0.362 0.361 0.358 0.356 0.344 0.357 0.362 0.359 0.357 0.352 0.360 0.358 0.329 0.351 0.360 0.305 0.355 0.361 0.343 0.359 0.360 0.357 0.352 0.336 0.338 0.311 0.289 0.360 0.359 0.362 0.364 0.364 L10 0.364 0.363 0.363 0.363 0.362 0.365 0.362 0.364 0.363 0.357 0.366 0.363 0.364 0.364 0.364 0.363 0.365 0.364 0.364 0.365 0.363 0.364 0.366 0.364 0.362 0.364 0.363 0.361 0.342 0.339 0.339 0.349 0.343 0.359 0.347 0.353 0.360 0.361 0.364 0.362 0.363 0.363 0.365 0.365 0.365 0.364 0.367 0.366 0.366 0.363 0.358 0.358 0.358 0.331 0.337 0.362 0.356 0.357 0.360 0.360 0.359 0.358 0.362 0.362 0.360 0.358 0.332 0.362 0.360 0.360 0.351 0.356 0.315 0.297 0.296 0.329 0.359 0.362 0.362 0.362 L11 0.361 0.362 0.363 0.363 0.363 0.361 0.362#

Page 257, passage 20.361 0.360 0.352 0.361 0.362 0.361 0.362 0.362 0.362 0.361 0.362 0.361 0.363 0.362 0.364 0.363 0.362 0.363 0.364 0.355 0.340 0.329 0.312 0.305 0.303 0.307 0.336 0.332 0.352 0.352 0.361 0.362 0.362 0.362 0.361 0.362 0.361 0.362 0.362 0.361 0.363 0.364 0.366 0.356 0.358 0.352 0.304 0.357 0.359 0.356 0.360 0.350 0.321 0.316 0.336 0.362 0.360 0.360 0.361 0.359 0.360 0.358 0.338 0.346 0.336 0.303 0.275 0.306 0.344 0.359 0.363 0.362 0.362 L12 0.363 0.361 0.362 0.362 0.363 0.361 0.361 0.362 0.362 0.360 0.362 0.360 0.361 0.362 0.361 0.362 0.360 0.361 0.362 0.362 0.363 0.363 0.362 0.362 0.363 0.362 0.348 0.336 0.324 0.351 0.269 0.234 0.306 0.325 0.353 0.359 0.355 0.358 0.358 0.362 0.361 0.360 0.362 0.360 0.362 0.361 0.360 0.361 0.361 0.361 0.358 0.353 0.355 0.344 0.238 0.247 0.323 0.306 0.271 0.270 0.245 0.285 0.285 0.358 0.358 0.355 0.360 0.356 0.360 0.361 0.362 0.351 0.297 0.299 0.335 0.362 0.361 0.363 0.363 0.363 L13 0.360 0.362 0.364 0.362 0.362 0.362 0.360 0.362 0.361 0.331 0.361 0.361 0.361 0.361 0.362 0.361 0.360 0.362 0.362 0.362 0.363 0.362 0.362 0.361 0.362 0.362 0.351 0.337 0.349 0.358 0.264 0.263 0.295 0.324 0.356 0.362 0.361 0.360 0.358 0.361 0.361 0.360 0.361 0.362 0.362 0.361 0.361 0.361 0.362 0.360 0.357 0.352 0.258 0.252 0.255 0.281 0.357 0.291 0.237 0.274 0.282 0.271 0.263 0.291 0.251 0.300 0.360 0.362 0.362 0.362 0.362 0.355 0.329 0.356 0.335 0.361 0.362 0.363 0.363 0.363 L14 0.360 0.361 0.363 0.363 0.363 0.362 0.363 0.361 0.363 0.359 0.362 0.360 0.360 0.361 0.361 0.362 0.361 0.362 0.362 0.363 0.363 0.362 0.363 0.363 0.362 0.363 0.351 0.337 0.350 0.355 0.359 0.335 0.311 0.322 0.358 0.362 0.361 0.358 0.353 0.360 0.360 0.361 0.362 0.362 0.363 0.362 0.361 0.362 0.362 0.298 0.286 0.355 0.226 0.263 0.215 0.217 0.260 0.313 0.215 0.238 0.211 0.214 0.258 0.289 0.276 0.272 0.362 0.363 0.363 0.361 0.360 0.358 0.358 0.358 0.361 0.361 0.361 0.361 0.362 0.362 L15 0.361 0.363 0.363 0.363 0.363 0.363 0.362 0.362 0.363 0.362 0.359 0.360 0.362 0.356 0.360 0.362 0.328 0.360 0.362 0.362 0.363 0.362 0.362 0.362 0.363 0.361 0.357 0.345 0.357 0.357 0.361 0.359 0.343 0.342 0.363 0.359 0.354 0.360 0.344 0.347 0.360 0.360 0.359 0.362 0.363 0.362 0.362 0.363 0.353 0.311 0.275 0.257 0.242 0.247 0.252 0.237 0.267 0.362 0.344 0.262 0.274 0.289 0.269 0.298 0.316 0.317 0.362 0.361 0.364 0.363 0.362 0.358 0.359 0.359 0.362 0.362 0.362 0.362 0.363 0.363 L16 0.362 0.362 0.363 0.363 0.365 0.362 0.362 0.361 0.364 0.362 0.362 0.360 0.362 0.362 0.363 0.363 0.360 0.362 0.363 0.362 0.363 0.362 0.362 0.362 0.363 0.363 0.362 0.361 0.361 0.363 0.361 0.361 0.359 0.357 0.361 0.360 0.362 0.361 0.361 0.362 0.362 0.361 0.360 0.362 0.362 0.349 0.362 0.333 0.286 0.272 0.296 0.255 0.276 0.256 0.258 0.298 0.361 0.362 0.361 0.363 0.282 0.284 0.357 0.362 0.361 0.362 0.362 0.363 0.364 0.363 0.363 0.362 0.361 0.360 0.364 0.363 0.364 0.364 0.365 0.364 L17 0.362 0.363 0.363 0.363 0.363 0.363 0.362 0.362 0.362 0.362 0.363 0.361 0.362 0.362 0.362 0.362 0.361 0.362 0.362 0.362 0.364 0.363 0.363 0.363 0.363 0.363 0.361 0.348 0.322 0.361 0.361 0.361 0.361 0.357 0.357 0.362 0.361 0.361 0.361 0.361 0.361 0.361 0.360 0.360 0.352 0.260 0.264 0.267 0.266 0.251 0.278 0.266 0.270 0.255 0.290 0.362 0.364 0.362 0.363 0.362 0.361 0.361 0.361 0.363 0.362 0.363 0.361 0.361 0.362 0.362 0.364 0.363 0.362 0.362 0.363 0.365 0.364 0.363 0.365 0.363 L18 0.362 0.362 0.362 0.364 0.364 0.365 0.362 0.364 0.362 0.362 0.363 0.361 0.362 0.361 0.362 0.363 0.361 0.362 0.364 0.363 0.361 0.362 0.363 0.340 0.363 0.364 0.362 0.360 0.362 0.362 0.362 0.361 0.362 0.361 0.362 0.362 0.363 0.362 0.362 0.360 0.307 0.355 0.360 0.360 0.350 0.252 0.250 0.258 0.264 0.255 0.285 0.258 0.268 0.271 0.361 0.363 0.362 0.362 0.363 0.363 0.363 0.362 0.363 0.362 0.361 0.362 0.363 0.363 0.363 0.362 0.362 0.362 0.363 0.362 0.363 0.363 0.363 0.363 0.365 0.366 L19 0.361 0.362 0.363 0.362 0.364 0.363 0.362 0.362 0.363 0.362 0.362 0.362 0.361 0.361 0.362 0.362 0.361 0.362 0.363 0.364 0.361 0.363 0.363 0.362 0.363 0.362 0.363 0.361 0.362 0.362 0.361 0.363 0.360 0.362 0.362 0.361 0.363 0.361 0.361 0.361 0.361 0.360 0.361 0.360 0.347 0.312 0.233 0.256 0.264 0.261 0.269 0.281 0.269 0.360 0.363 0.362 0.363 0.362 0.364 0.362 0.363 0.363 0.363 0.362 0.363 0.362 0.362 0.364 0.364 0.364 0.363 0.362 0.363 0.362 0.362 0.364 0.363 0.363 0.363 0.362 L20 0.362 0.363 0.362 0.362 0.362 0.363 0.361 0.362 0.362 0.362 0.362 0.362 0.362 0.361 0.362 0.362 0.361 0.363 0.364 0.363 0.363 0.363 0.363 0.363 0.363 0.363 0.362 0.362 0.362 0.362 0.362 0.362 0.361 0.362 0.362 0.362 0.363 0.360 0.362 0.363 0.363 0.362 0.362 0.363 0.361 0.362 0.361 0.362 0.362 0.363 0.363 0.362 0.364 0.364 0.364 0.364 0.363 0.363 0.364 0.363 0.363 0.363 0.363 0.363 0.364 0.364 0.364 0.363 0.363 0.364 0.364 0.363 0.363 0.363 0.364 0.364 0.363 0.364 0.365 0.363 L21 0.363 0.363 0.363 0.363 0.363 0.362 0.362 0.362 0.362 0.362 0.362 0.362 0.362 0.362 0.363 0.362 0.361 0.362 0.364 0.363 0.363 0.362 0.362 0.363 0.363 0.363 0.363 0.362 0.362 0.363 0.363 0.361 0.362 0.362 0.362 0.362 0.362 0.362 0.363 0.363 0.363 0.362 0.363 0.363 0.362 0.363 0.362 0.362 0.362 0.363 0.363 0.362 0.365 0.363 0.364 0.363 0.363 0.363 0.363 0.363 0.364 0.364 0.364 0.364 0.364 0.364 0.364 0.364 0.364 0.363 0.364 0.363 0.363 0.364 0.363 0.364 0.364 0.363 0.365 0.363 L22 0.361 0.363 0.361 0.362 0.361 0.363 0.362 0.362 0.362 0.362 0.362 0.361 0.361 0.362 0.361 0.360 0.360 0.360 0.360 0.361 0.363 0.361 0.362 0.361 0.362 0.360 0.361 0.362 0.362 0.363 0.361 0.360 0.360 0.363 0.361 0.360 0.360 0.361 0.361 0.361 0.360 0.360 0.361 0.360 0.361 0.362 0.363 0.362 0.362 0.362 0.362 0.362 0.362 0.362 0.362 0.362 0.361 0.362 0.365 0.363 0.362 0.362 0.362 0.362 0.363 0.362 0.362 0.362 0.363 0.363 0.364 0.361 0.362 0.362 0.362 0.364 0.362 0.362 0.364 0.363 C43 C44 C45 C46 C47 C48 C49 C50 C51 C52 C53 C54 C55 C56 C57 C58 C59 C60 C61 C62 C63 C64 C65 C66 L0 0.361 0.361 0.362 0.360 0.361 0.362 0.362 0.362 0.360 0.361 0.362 0.362 0.362 0.361 0.362 0.362 0.362 0.362 0.361 0.363 0.363 0.362 0.362#

Page 257, passage 30.362 L1 0.362 0.362 0.362 0.362 0.362 0.363 0.363 0.364 0.362 0.362 0.362 0.363 0.363 0.361 0.364 0.363 0.363 0.363 0.362 0.363 0.362 0.363 0.363 0.364 L2 0.364 0.364 0.364 0.364 0.364 0.364 0.364 0.364 0.363 0.364 0.362 0.363 0.363 0.363 0.364 0.363 0.364 0.364 0.363 0.363 0.363 0.364 0.364 0.364 L3 0.363 0.363 0.363 0.363 0.363 0.362 0.358 0.362 0.360 0.358 0.360 0.361 0.363 0.363 0.362 0.363 0.363 0.363 0.362 0.361 0.357 0.324 0.360 0.364 L4 0.361 0.362 0.362 0.360 0.358 0.361 0.362 0.360 0.358 0.360 0.362 0.361 0.362 0.360 0.358 0.360 0.358 0.361 0.355 0.338 0.334 0.340 0.361 0.364 L5 0.362 0.350 0.363 0.364 0.362 0.359 0.363 0.300 0.308 0.361 0.366 0.362 0.362 0.361 0.359 0.363 0.359 0.360 0.363 0.357 0.343 0.363 0.365 0.366 L6 0.362 0.361 0.358 0.360 0.362 0.359 0.355 0.308 0.278 0.359 0.364 0.362 0.360 0.355 0.355 0.353 0.362 0.358 0.355 0.323 0.314 0.359 0.358 0.363 L7 0.361 0.361 0.362 0.359 0.357 0.362 0.358 0.358 0.355 0.327 0.330 0.322 0.330 0.347 0.350 0.354 0.363 0.362 0.338 0.311 0.303 0.360 0.351 0.361 L8 0.361 0.362 0.361 0.360 0.361 0.361 0.342 0.359 0.354 0.348 0.360 0.357 0.347 0.352 0.341 0.355 0.354 0.348 0.345 0.359 0.356 0.359 0.361 0.350 L9 0.362 0.362 0.362 0.361 0.362 0.362 0.361 0.358 0.356 0.344 0.357 0.362 0.359 0.357 0.352 0.360 0.358 0.329 0.351 0.360 0.305 0.355 0.361 0.343 L10 0.365 0.365 0.364 0.367 0.366 0.366 0.363 0.358 0.358 0.358 0.331 0.337 0.362 0.356 0.357 0.360 0.360 0.359 0.358 0.362 0.362 0.360 0.358 0.332 L11 0.361 0.362 0.362 0.361 0.363 0.364 0.366 0.356 0.358 0.352 0.304 0.357 0.359 0.356 0.360 0.350 0.321 0.316 0.336 0.362 0.360 0.360 0.361 0.359 L12 0.360 0.362 0.361 0.360 0.361 0.361 0.361 0.358 0.353 0.355 0.344 0.238 0.247 0.323 0.306 0.271 0.270 0.245 0.285 0.285 0.358 0.358 0.355 0.360 L13 0.362 0.362 0.361 0.361 0.361 0.362 0.360 0.357 0.352 0.258 0.252 0.255 0.281 0.357 0.291 0.237 0.274 0.282 0.271 0.263 0.291 0.251 0.300 0.360 L14 0.362 0.363 0.362 0.361 0.362 0.362 0.298 0.286 0.355 0.226 0.263 0.215 0.217 0.260 0.313 0.215 0.238 0.211 0.214 0.258 0.289 0.276 0.272 0.362 L15 0.362 0.363 0.362 0.362 0.363 0.353 0.311 0.275 0.257 0.242 0.247 0.252 0.237 0.267 0.362 0.344 0.262 0.274 0.289 0.269 0.298 0.316 0.317 0.362 L16 0.362 0.362 0.349 0.362 0.333 0.286 0.272 0.296 0.255 0.276 0.256 0.258 0.298 0.361 0.362 0.361 0.363 0.282 0.284 0.357 0.362 0.361 0.362 0.362 L17 0.360 0.352 0.260 0.264 0.267 0.266 0.251 0.278 0.266 0.270 0.255 0.290 0.362 0.364 0.362 0.363 0.362 0.361 0.361 0.361 0.363 0.362 0.363 0.361 L18 0.360 0.350 0.252 0.250 0.258 0.264 0.255 0.285 0.258 0.268 0.271 0.361 0.363 0.362 0.362 0.363 0.363 0.363 0.362 0.363 0.362 0.361 0.362 0.363 L19 0.360 0.347 0.312 0.233 0.256 0.264 0.261 0.269 0.281 0.269 0.360 0.363 0.362 0.363 0.362 0.364 0.362 0.363 0.363 0.363 0.362 0.363 0.362 0.362 L20 0.363 0.361 0.362 0.361 0.362 0.362 0.363 0.363 0.362 0.364 0.364 0.364 0.364 0.363 0.363 0.364 0.363 0.363 0.363 0.363 0.363 0.364 0.364 0.364 L21 0.363 0.362 0.363 0.362 0.362 0.362 0.363 0.363 0.362 0.365 0.363 0.364 0.363 0.363 0.363 0.363 0.363 0.364 0.364 0.364 0.364 0.364 0.364 0.364 L22 0.360 0.361 0.362 0.363 0.362 0.362 0.362 0.362 0.362 0.362 0.362 0.362 0.362 0.361 0.362 0.365 0.363 0.362 0.362 0.362 0.362 0.363 0.362 0.362 17.6’ Flow#

Page 258Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Flow OD ID (a) Flow ID OD Figure 38. (b) Photographs showing the a) clockwise and b) counterclockwise fractures surfaces following cleaning with a degreaser and acetone and/or methanol. Tape measure indicates distance to U/S GW. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 64 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 259Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Flow OD ID a b c Figure 44 d e Region 2 Mount 195367-1b Region 1 Figure 40 Figure 39. Stereo light photomicrographs of representative locations along the clockwise fracture surface following cleaning with a), b), c) a degreaser and acetone and/or methanol and d), e) an inhibited HCl acid and ENPREP®. Photomicrographs b) and d) are from the same location. Tape measure indicates distance to U/S GW. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 65 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 260Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Flow Region 2 OD Figure 41 Region 1 ID Figure 40. SEM image showing the fracture surface at the location identified in Figure 39e. Figure 43; Region 2 Figure 42; Region 1 ID Figure 41. SEM image showing the transition between Region 1 and Region 2. Area indicated in Figure 40. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 66#

Page 261Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Figure 42. SEM image showing the ductile fracture morphology of Region 1. Area indicated in Figure 41 Figure 43. SEM image showing a nondescript/corroded morphology of Region 2. Area indicated in Figure 41 DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 67#

Page 262Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Flow OD Figure 45 ID Figure 44. SEM image showing the fracture surface at the location identified in Figure 39c. Figure 45. SEM image showing a representative ductile fracture morphology for Region 1. Area indicated in Figure 44. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 68#

Page 263Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Flow OD ID Figure 46. Photograph of the clockwise fracture surface showing thickness measurements of the Region 1 along the fracture surface at 5 mm intervals. Scale in mm. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 69 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 264Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Flow Region 2 (Corrosion) Region 1 Region 1 Figure 47. Fracture/Corrosion profile based on three measurement techniques. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 70 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 265Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Figure 52 Figure 53 Figure 49 Figure 48. Photograph of the mounted transverse cross-section, Mount 195367-1b removed from the suspected failure origin; Feature 4. Mount location indicated in Figure 39. CCW Fracture Surface CW Fracture Surface Preexisting Corrosion Preexisting Corrosion Overload Figure 51 Overload Ligament Figure 50 Figure 49. Photomicrograph showing the suspected failure origin in cross-section; Feature 4. Area indicated in Figure 48. (Mount 195367-1b; 4% Nital Etch) DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 71#

Page 266Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Preexisting Corrosion Figure 50. Overload Ligament ID Photomicrograph showing the negligible grain elongation and plasticity along the CW fracture surface. Area indicated in Figure 49. (Mount 195367-1b; 4% Nital Etch) Grain elongation Figure 51. ID Photomicrograph showing the grain elongation and plasticity along the CCW fracture surface. Area indicated in Figure 49. (Mount 195367-1b; 4% Nital Etch) DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 72#

Page 267Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Figure 52. Photomicrograph showing corrosion products near the CCW fracture surface. Area indicated in Figure 48. (Mount 195367-1b; 4% Nital Etch) Figure 53. Photomicrograph showing corrosion products with some undercutting near the CCW fracture surface. Area indicated in Figure 49. (Mount 195367-1b; 4% Nital Etch) DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 73#

Page 268Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Figure 54. Photomicrograph showing base metal microstructure of Mount 195367-1b; 4% Nital Etch) Figure 56 Figure 55. Photograph of the mounted cross-section of corrosion products, Mount 195331-1 removed from Feature 4. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 74#

Page 269Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing OD Figure 65 ID Figure 56. Photomicrographs of the mounted cross-section, Mount 195331-1, from corrosion products removed from Feature 4; location identified in Figure 55. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 75#

Page 270Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Figure 58 Figure 57. Photograph of the transverse mounted cross-section, Mount 195370-1 removed from Feature 1. Mount location indicated in Figure 22. Figure 64 Figure 58. Photomicrograph of the mounted cross-section, Mount 195370-1 removed from Feature 1 showing the corrosion morphology at the external surface; mirror image of location indicated in Figure 57. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 76#

Page 271Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Figure 60 Figure 59. Photograph of the mounted cross-section of corrosion products, Mount 195322-1, removed from Feature 1. Figure 60. Photomicrographs of the mounted cross-section, Mount 195322-1, from the corrosion product removed from Feature 1; location indicated in Figure 59. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 77#

Page 272Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Figure 61. Photograph of the mounted cross-section, Mount 195365-1, removed from across the longitudinal seam weld. Location indicated in Figure 4. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 78#

Page 273Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Figure 62. XRD spectrum acquired from the corrosion products collected from Feature 1, identifying Goethite and Magnetite as the compounds. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 79#

Page 274Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Figure 63. XRD spectrum acquired from the corrosion products collected from Feature 2, identifying Goethite and Magnetite as the compounds. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 80#

Page 275Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing cps/eV 30 O 25 20 15 Fe 10 Fe 5 C Cu Si Cl Mn Cu 0 1 2 3 4 5 6 7 8 keV Norm. mass percent (%) Spectrum C O Si Cl Mn Fe Cu ------------------------------------------------ # 1 # 2 # 4 0.00 33.52 0.19 0.19 0.79 56.87 8.44 0.00 35.78 0.40 0.06 0.80 62.46 0.49 # 3 0.00 29.24 0.41 0.07 0.68 69.60 - 0.00 37.97 0.26 0.15 0.73 60.71 0.18 ------------------------------------------------ Mean value: 0.00 34.13 0.32 0.12 0.75 62.41 3.04 Sigma: 0.00 3.73 0.11 0.06 0.06 5.33 4.68 Sigma mean: 0.00 1.87 0.05 0.03 0.03 2.67 2.34 Figure 64. EDS data collected from Mount 195370-1, mirror image of area indicated in Figure 58. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 81#

Page 276Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing cps/eV O 18 16 14 12 10 8 6 4 2 0 Fe Fe Mg Si S S Mn 1 2 3 4 5 6 7 keV Norm. mass percent (%) Spectrum O Mg Si S Mn Fe ---------------------------------------------------- 195331-area 2_1kx# 1 27.85 - 0.21 - 0.59 71.35 195331-area 2_1kx# 2 29.45 - 0.20 - 0.93 69.41 195331-area 2_1kx# 3 37.83 0.48 0.22 0.18 0.98 60.30 ---------------------------------------------------- Mean value: Sigma: Sigma mean: 31.71 0.48 0.21 0.18 0.83 67.02 5.36 0.00 0.01 0.00 0.21 5.90 3.09 0.00 0.01 0.00 0.12 3.41 Figure 65. EDS data collected from Mount 195331-1, area indicated in Figure 56. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 82#

Page 277Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Fe 16 cps/eV C-KA O-K Si-KA S-KA Mn-KA Fe-KA 14 12 10 8 6 4 2 0 C O Si S Mn Fe 1 2 3 4 5 6 7 keV Intensity Figure 66. EDS line scan collected from Mount 195331-1, area indicated in Figure 56. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 83#

Page 278Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Figure 67. Photograph showing the soil samples collected from below the pipe, 8 feet U/S of GW 5930. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 84#

Page 279Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Figure 68. Percent shear from Charpy V-notch tests as a function of temperature for transverse base metal specimens removed from the failure joint (Joint 5930). Figure 69. Charpy V-notch impact energy as a function of temperature for transverse base metal specimens removed from the failure joint (Joint 5930). DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 85#

Page 280Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Figure 70. Percent shear from Charpy V-notch tests as a function of temperature for transverse base metal specimens removed from U/S joint (Joint 5920). Figure 71. Charpy V-notch impact energy as a function of temperature for transverse base metal specimens removed from the U/S joint (Joint 5920). DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 86#

Page 281Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Figure 72. Percent shear from Charpy V-notch tests as a function of temperature for transverse base metal specimens removed from D/S joint (Joint 5940). Figure 73. Charpy V-notch impact energy as a function of temperature for transverse base metal specimens removed from the D/S joint (Joint 5940). DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP 87#

Page 282Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing 0.358 0.358 0.359 0.360 0.360 0.362 0.361 0.362 0.362 0.362 0.361 0.362 0.362 0.363 0.362 0.362 0.361 0.360 0.359 0.360 0.360 0.359 0.358 0.360 0.360 0.360 0.358 0.356 0.357 0.357 0.357 0.357 0.357 0.357 0.356 0.356 0.356 0.356 0.355 0.355 0.357 0.359 0.359 0.358 0.360 0.361 0.361 0.361 0.361 0.361 0.360 0.362 0.362 0.361 0.361 0.362 0.362 0.359 0.359 0.359 0.360 0.359 0.357 0.358 0.358 0.357 0.357 0.356 0.357 0.357 0.359 0.359 0.359 0.359 0.357 0.357 0.357 0.357 0.356 0.355 0.358 0.358 0.359 0.359 0.360 0.361 0.360 0.360 0.359 0.359 0.359 0.359 0.360 0.358 0.356 0.357 0.358 0.357 0.356 0.355 0.357 0.356 0.355 0.355 0.355 0.355 0.355 0.354 0.356 0.354 0.357 0.360 0.359 0.357 0.355 0.355 0.354 0.354 0.354 0.355 0.359 0.358 0.359 0.359 0.360 0.359 0.358 0.358 0.357 0.357 0.357 0.357 0.357 0.357 0.354 0.358 0.356 0.356 0.354 0.355 0.355 0.355 0.355 0.355 0.355 0.355 0.355 0.355 0.355 0.355 0.357 0.359 0.358 0.357 0.356 0.355 0.355 0.355 0.355 0.355 0.359 0.359 0.360 0.360 0.360 0.360 0.360 0.360 0.359 0.359 0.358 0.357 0.355 0.351 0.348 0.349 0.351 0.351 0.351 0.353 0.351 0.351 0.352 0.357 0.355 0.355 0.355 0.355 0.357 0.357 0.357 0.359 0.356 0.355 0.356 0.355 0.355 0.356 0.355 0.355 0.353 0.356 0.360 0.360 0.360 0.360 0.360 0.361 0.359 0.357 0.357 0.351 0.349 0.349 0.343 0.347 0.349 0.348 0.347 0.347 0.343 0.343 0.346 0.354 0.351 0.353 0.355 0.356 0.358 0.358 0.358 0.357 0.355 0.355 0.356 0.356 0.355 0.355 0.355 0.355 0.347 0.351 0.359 0.359 0.359 0.359 0.358 0.357 0.357 0.352 0.343 0.330 0.287 0.174 0.153 0.199 0.255 0.269 0.320 0.337 0.328 0.326 0.329 0.334 0.341 0.345 0.349 0.353 0.358 0.358 0.357 0.356 0.356 0.355 0.355 0.354 0.353 0.353 0.354 0.354 0.351 0.355 0.359 0.359 0.358 0.358 0.357 0.356 0.354 0.338 0.318 0.261 0.126 0.096 0.101 0.176 0.166 0.050 0.096 0.174 0.242 0.275 0.274 0.299 0.321 0.326 0.334 0.350 0.358 0.358 0.357 0.357 0.356 0.355 0.355 0.354 0.352 0.352 0.352 0.353 0.352 0.356 0.359 0.358 0.357 0.357 0.355 0.347 0.347 0.316 0.249 0.176 0.132 0.116 0.100 0.100 0.077 0.063 0.066 0.088 0.133 0.116 0.144 0.194 0.229 0.249 0.288 0.320 0.356 0.356 0.357 0.356 0.355 0.357 0.357 0.355 0.352 0.350 0.351 0.351 0.350 0.356 0.358 0.358 0.357 0.355 0.348 0.336 0.339 0.327 0.233 0.221 0.170 0.137 0.119 0.127 0.109 0.089 0.101 0.115 0.120 0.060 0.072 0.068 0.067 0.074 0.171 0.272 0.339 0.356 0.356 0.356 0.356 0.356 0.357 0.355 0.353 0.351 0.351 0.351 0.348 0.352 0.356 0.356 0.355 0.350 0.328 0.308 0.337 0.330 0.291 0.235 0.185 0.160 0.157 0.144 0.128 0.121 0.103 0.133 0.091 0.080 0.081 0.089 0.084 0.072 0.065 0.161 0.320 0.355 0.355 0.356 0.356 0.356 0.357 0.355 0.353 0.351 0.351 0.351 0.346 0.348 0.351 0.352 0.349 0.347 0.336 0.330 0.342 0.332 0.311 0.257 0.200 0.179 0.161 0.143 0.133 0.136 0.153 0.130 0.123 0.106 0.120 0.120 0.106 0.098 0.096 0.122 0.270 0.345 0.354 0.355 0.356 0.355 0.355 0.355 0.355 0.352 0.351 0.351 0.341 0.344 0.348 0.349 0.352 0.351 0.350 0.343 0.316 0.331 0.328 0.287 0.234 0.191 0.168 0.161 0.147 0.148 0.194 0.170 0.127 0.128 0.151 0.123 0.115 0.126 0.146 0.135 0.179 0.319 0.355 0.357 0.358 0.358 0.358 0.357 0.354 0.352 0.352 0.352 0.341 0.339 0.337 0.346 0.345 0.345 0.347 0.325 0.319 0.322 0.338 0.316 0.257 0.207 0.197 0.172 0.163 0.177 0.161 0.171 0.147 0.140 0.192 0.162 0.128 0.136 0.124 0.134 0.201 0.289 0.356 0.358 0.359 0.359 0.359 0.358 0.353 0.352 0.352 0.353 0.338 0.333 0.337 0.344 0.348 0.351 0.352 0.344 0.325 0.321 0.330 0.309 0.271 0.263 0.232 0.212 0.186 0.203 0.175 0.207 0.213 0.153 0.165 0.184 0.196 0.163 0.160 0.191 0.219 0.319 0.358 0.359 0.359 0.359 0.358 0.358 0.355 0.351 0.354 0.353 0.341 0.338 0.336 0.346 0.347 0.344 0.349 0.351 0.345 0.335 0.336 0.319 0.265 0.267 0.259 0.226 0.240 0.199 0.233 0.213 0.248 0.245 0.206 0.225 0.206 0.231 0.248 0.239 0.306 0.357 0.361 0.360 0.360 0.361 0.359 0.358 0.352 0.287 0.313 0.331 0.347 0.345 0.338 0.345 0.338 0.346 0.342 0.342 0.349 0.344 0.337 0.330 0.291 0.262 0.240 0.219 0.219 0.221 0.220 0.211 0.221 0.211 0.206 0.195 0.198 0.202 0.224 0.203 0.288 0.360 0.361 0.361 0.361 0.361 0.361 0.359 0.353 0.312 0.277 0.268 0.352 0.351 0.351 0.350 0.347 0.344 0.343 0.345 0.348 0.345 0.342 0.334 0.308 0.251 0.235 0.225 0.224 0.220 0.259 0.287 0.278 0.259 0.228 0.214 0.216 0.244 0.215 0.257 0.335 0.345 0.360 0.361 0.361 0.361 0.361 0.359 0.349 0.336 0.337 0.311 0.352 0.352 0.351 0.351 0.351 0.350 0.343 0.343 0.346 0.345 0.343 0.342 0.324 0.300 0.276 0.273 0.286 0.294 0.322 0.348 0.348 0.333 0.293 0.276 0.242 0.248 0.241 0.252 0.282 0.307 0.338 0.359 0.360 0.358 0.357 0.353 0.324 0.303 0.339 0.349 0.355 0.355 0.354 0.353 0.353 0.352 0.352 0.343 0.343 0.350 0.352 0.349 0.344 0.336 0.337 0.332 0.341 0.341 0.356 0.359 0.359 0.356 0.334 0.322 0.299 0.323 0.302 0.302 0.322 0.321 0.318 0.355 0.355 0.356 0.352 0.340 0.335 0.333 0.349 0.350 0.356 0.354 0.351 0.352 0.351 0.350 0.347 0.351 0.350 0.352 0.351 0.347 0.349 0.349 0.353 0.355 0.356 0.355 0.355 0.359 0.360 0.359 0.360 0.360 0.359 0.351 0.348 0.356 0.352 0.348 0.354 0.356 0.355 0.356 0.356 0.351 0.348 0.349 0.352 0.352 0.353 0.346 0.345 0.340 0.341 0.336 0.341 0.344 0.346 0.350 0.351 0.352 0.349 0.352 0.359 0.359 0.359 0.359 0.359 0.361 0.362 0.362 0.362 0.362 0.362 0.358 0.361 0.359 0.351 0.353 0.357 0.359 0.357 0.360 0.359 0.356 0.357 0.360 0.356 0.355 0.350 0.348 0.344 0.339 0.336 0.333 0.341 0.343 0.345 0.349 0.351 0.351 0.352 0.353 0.357 0.358 0.358 0.359 0.358 0.358 0.359 0.359 0.361 0.361 0.362 0.362 0.364 0.359 0.357 0.356 0.357 0.358 0.358 0.357 0.356 0.355 0.355 0.355 0.356 0.356 0.338 0.333 0.336 0.339 0.336 0.333 0.328 0.308 0.316 0.316 0.233 0.176 0.126 0.096 0.100 0.100 0.077 0.050 0.066 0.088 0.091 0.060 0.072 0.068 0.067 0.072 0.065 0.122 0.179 0.289 0.318 0.355 0.355 0.355 0.352 0.340 0.324 0.287 0.277 0.268 Figure 74. Color profile showing the results of average thickness measurements performed on the laser scan dataset following discretizing#

Page 282, passage 2the data into ½- inch cells. The resulting profile is highlighted in blue and plotted in Figure 47. DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 88 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 283Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing APPENDIX A CorLAS™ DNV GL – OAPUS309DNOR (PP136049) September 18, 2015 Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 284Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Appendix A Description of CorLAS™ The CorLAS™ computer program was developed by Det Norske Veritas (U.S.A.), Inc. (formerly CC Technologies) to evaluate crack-like flaws in pipelines based on inelastic fracture mechanics. Using the effective area of the actual, measured crack length-depth profile, an equivalent semi-elliptical surface flaw is modeled and used to compute the effective stress and the applied value of J for internal pressure loading. The effective stress and applied J are then compared with the flow strength ( s ) and fracture toughness (JC), fs respectively, to predict the failure pressure. The program also contains a similar inelastic fracture mechanics analysis for through-wall flaws. The fracture toughness of the steel can be estimated from Charpy data or measured by means of a JIC test. In the most recent version of CorLAS™, the fracture toughness analysis automatically checks for plastic instability and only the fracture toughness curve needs to be considered for crack-like flaws. The actual tensile and Charpy properties of the pipe joint, measured from the samples removed, can be used for the critical leak/rupture length calculation. DNV GL – OAPUS309DNOR (PP136049) August 6, 2015 A-1#

Page 285Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Case 1: Fracture Surface Measurements & Laser Scan Data Plains Line 901 Semi-Elliptical Flaw Profile API X65 - Joint 5930 Maximum Operating Pressure (psig) 1341 UTS (psi) 84000 YS (psi) 64800 FS (psi) 74800 E (ksi) 29500 nexp 0.098 Jc (lb/in) 12097 Thin-wall (OD) formula for hoop stress Tmat 208.9 OD (in.) 24 Wall Thickness (in.) 0.359 Summary of Results for Effective Area Method Flaw: Start (in.) 0.13 Length (in.) 5.9 Area (in.^2) 1.889 Depth (in.) Maximum 0.356 Equivalent Flaw 0.408 For Design Factor 0.72 Design Pressure (psig) 1395.79 Failure Stress (psi) 15523 Failure Pressure (psig) 464.41 For Design Factor 0.72 Maximum Safe Pressure (psig) 334.38 Summary of Results for 0.85dL Eff. Area Method Failure Stress (psi) 16205 Failure Pressure (psig) 484.79 For Design Factor 0.72 Maximum Safe Pressure (psig) 349.05 DNV GL – OAPUS309DNOR (PP136049) August 6, 2015 A-2#

Page 286Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Case 2: Laser Scan Data - Offset from Fracture Surface Plains Line 901 Semi-Elliptical Flaw Profile API X65 - Joint 5930 Maximum Operating Pressure (psig) 1341 UTS (psi) 84000 YS (psi) 64800 FS (psi) 74800 E (ksi) 29500 nexp 0.098 Jc (lb/in) 12097 Thin-wall (OD) formula for hoop stress Tmat 208.9 OD (in.) 24 Wall Thickness (in.) 0.359 Summary of Results for Effective Area Method Flaw: Start (in.) 4.076 Length (in.) 7.2 Area (in.^2) 2.026 Depth (in.) Maximum 0.318 Equivalent Flaw 0.358 For Design Factor 0.72 Design Pressure (psig) 1395.79 Failure Stress (psi) 25388 Failure Pressure (psig) 759.53 For Design Factor 0.72 Maximum Safe Pressure (psig) 546.86 Summary of Results for 0.85dL Eff. Area Method Failure Stress (psi) 23779 Failure Pressure (psig) 711.39 For Design Factor 0.72 Maximum Safe Pressure (psig) 512.2 DNV GL – OAPUS309DNOR (PP136049) August 6, 2015 A-3#

Page 287Plains All American Pipeline, L.P. Line 901 Release (05-19-15): Mechanical and Metallurgical Testing Case 3: Laser Scan Data - ½-Inch Grid (Average) Plains Line 901 Semi-Elliptical Flaw Profile API X65 - Joint 5930 Maximum Operating Pressure (psig) 1341 UTS (psi) 84000 YS (psi) 64800 FS (psi) 74800 E (ksi) 29500 nexp 0.098 Jc (lb/in) 12097 Thin-wall (OD) formula for hoop stress Tmat 208.9 OD (in.) 24 Wall Thickness (in.) 0.359 Summary of Results for Effective Area Method Flaw: Start (in.) 6 Length (in.) 7.5 Area (in.^2) 2.095 Depth (in.) Maximum 0.309 Equivalent Flaw 0.356 For Design Factor 0.72 Design Pressure (psig) 1395.79 Failure Stress (psi) 25516 Failure Pressure (psig) 763.35 For Design Factor 0.72 Maximum Safe Pressure (psig) 549.61 Summary of Results for 0.85dL Eff. Area Method Failure Stress (psi) 23715 Failure Pressure (psig) 709.46 For Design Factor 0.72 Maximum Safe Pressure (psig) 510.81 DNV GL – OAPUS309DNOR (PP136049) August 6, 2015 A-4#

Page 288ABOUT DNV GL Driven by our purpose of safeguarding life, property, and the environment, DNV GL enables organizations to advance the safety and sustainability of their business. We provide classification and technical assurance along with software and independent expert advisory services to the maritime, oil and gas, and energy industries. We also provide certification services to customers across a wide range of industries. Operating in more than 100 countries, our 16,000 professionals are dedicated to helping our customers make the world safer, smarter, and greener. Contains Confidential Information Provided By Plains All American Pipeline LP#

Page 289Appendix N Det Norske Veritas (U.S.A.), Inc. (DNV GL): Line 901 Release (5/19/15) Technical Root Cause Analysis#

Page 290Final Report Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. Houston, Texas Report No.: OAPUS307KKRA (PP136049) December 4, 2015#

Page 291Plains All American Pipeline, L.P Line 901 Release (5/19/15) Technical Root Cause Analysis Project Name: Line 901 Release (5/19/15) Technical Root Cause Analysis DET NORSKE VERITAS (U.S.A.), INC. (DNV GL) Materials & Corrosion Technology Center Customer: Plains All American Pipeline, L.P. Incident Investigation 5777 Frantz Road Contact Person: Dublin, OH 43017-1886 Date of Issue: December 4, 2015 Tel: (614) 761-1214 United States Project No.: PP136049 Fax: (614) 761-1633 Organization Unit: Incident Investigation www.dnvgl.com Report No.: OAPUS307KKRA Task and Objective: Please see Executive Summary. Prepared by Verified by Approved by Kathume MBuckingham Oh a Beaner Katherine M. Buckingham, Ph.D. Principal Engineer John A. Beavers, Ph.D., FACE Director - Incident Investigation Neil G. Thompson, PH.D., FACE Vice President, Pipeline Services Balan A Pafist Senior Engineer Barbara N. Padgett, Ph.D. Steven J. Polasik, M.S., P.E. Senior Engineer Angel Kowalski Head of Section - Fitness for Service • Unrestricted Distribution (internal and external) Keywords • Unrestricted Distribution within DNV GL • Limited Distribution within DNV GL after 3 years * No Distribution (confidential) Secret any form, or by any means, whether digitally or otherwise without the prior written consent of DNV GL. DNV GLand the Horizon Copyright © DNV GL 2015. All rights reserved. This publication or parts thereof may not be copied, reproduced, or transmitted in agreed in writing. Reference to part of this publication, which may lead to misinterpretation, is prohibited. Graphic are trademarks of DNV GL AS. The content of this publication shall be kept confidential by the customer, unless otherwise Rev. No. Date Reason for Issue: Prepared by: Verified by Approved byg 0 2015-09-21 First Issue 2015-12-04 Final Issue P:\AZA Law - Plains\ Root Cause Analysis\Produced \Report\Comments\RCA Draft 2 report.docx OAPUS307KKRA December 4, 2015 ii#

Page 292Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Executive Summary Plains All American Pipeline, L.P. (Plains) retained Det Norske Veritas (U.S.A.), Inc. (DNV GL) to perform a root cause analysis (RCA) of a failure that occurred on Line Segment 901, which transports heated crude oil from the outer continental shelf (OCS) of California. The failure occurred on May 19, 2015 in Goleta, California (Santa Barbara County) and was located near milepost (MP) 4. The location was approximately 4.05 miles downstream (D/S) from Las Flores Pump Station and approximately 6.2 feet upstream (U/S) from the nearest girth weld, identified as Girth Weld (GW) 5940. Approximately 2,9341 barrels of crude oil were released. The portion of the pipeline that contained the failure is comprised of 24-inch diameter by 0.344 inch wall thickness, API 5L Grade X65 line pipe steel that was manufactured by Nippon Steel and contains a high frequency (HF) electric resistance welded (ERW) longitudinal seam. The pipeline (Line 901) was installed in 1990 and is approximately 10.87 miles in length, spanning between Las Flores Station on the U/S end and Gaviota Station on the D/S end. The pipeline is externally covered with the following: (1) a protective coating of coal tar urethane (CTU) that is in intimate contact with the steel pipe, (2) a layer of rigid thermal polyurethane (PU) foam insulation, and (3) an outer layer of polyethylene (PE) tape. The pipeline has an impressed cathodic protection (CP) system that was energized at the time of installation. The normal operating pressure and maximum discharge pressure (MDP) for the line are 616 psig and 1,025 psig, respectively. These pressures correspond to 33% and 55% of the specified minimum yield strength (SMYS), respectively. The pressure at the time and location of the failure was reported by Plains to be 737 psig [Ref ‎2], which corresponds to 39.6% of the SMYS and 71.9% of the MDP. The leak occurred in a mostly rural area that runs along the coastline of the Pacific Ocean. The topography in the area is hilly, with the pipeline oriented uphill from the ocean. The failure was located near a local low point along the pipeline. Several road crossings, such as Highway 1, are present in the area with drainage toward the coast via culverts. It is via these culverts that the released oil reached the Pacific Ocean at Refugio State Beach. The objective of the RCA was to identify factors contributing to the failure and document the decisions made preceding the failure. The portion of the pipeline that contained the failure location was removed and sent to DNV GL to determine the metallurgical cause of the failure and to identify any contributing factors. The conclusions and recommendations for this RCA are based on the findings from the final metallurgical report as well as information 1 [Ref ‎6] The final volume estimate for the released oil at the time of this report. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 iii#

Page 293Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis provided and publically reported by Plains. Based on the findings of the analysis, recommendations for improvements also are identified. The methodology used by DNV GL for the RCA of the Line 901 release was based on the DNV GL Loss Causation Model (LCM). This model is built on the concept that incidents can be attributed to immediate causes, basic causes, and failures of management systems to control hazards. The analysis uses a systematic method of processing evidence gathered during an investigation in order to identify the factors that led to the incident. This methodology assists in the development of corrective and/or remedial measures. The LCM approach used by DNV GL is called a Barrier-based Systematic Causal Analysis Technique (BSCAT). BSCAT™ is a technique that applies a Systematic Causal Analysis Technique (SCAT) model to each barrier, as opposed to the incident as a whole. This method results in a thorough review of the effectiveness of individual barriers identified in the risk assessment. BSCAT provides a methodology that allows for the analysis of complex incidents that involve multiple barriers. The results of the metallurgical analysis indicated that the immediate metallurgical cause for the Line 901 failure was wall thinning from external corrosion that ultimately failed by ductile overload under the imposed operating pressure [Ref ‎1]. The flaw that failed was not through wall prior to ductile overload and, therefore, the failure event was sudden in nature. The morphology of the external corrosion was determined to be consistent with corrosion under insulation (CUI), facilitated by wet-dry cycling. The results of the root cause analysis presented below are based on the provided documentation referenced in Appendix B. DNV GL reserves the right to modify or supplement these conclusions should new information become available. DNV GL identified four c basic root causes of the failure: 1. The external coating system failed to prevent moisture from reaching the pipe steel, allowing the external corrosion process to occur. Basis: Based on the metallurgical analysis, the protective coal tar urethane coating, thermal polyurethane foam insulation, and polyethylene tape were compromised at the failure location. The damage included wrinkles, cracks, staining, and decohesion of the polyethylene tape; staining, water saturation and retention, and compression of the polyurethane foam; and disbondment of the coal tar urethane. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 iv#

Page 294Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 2. The cathodic protection system was ineffective due to shielding by the thermal polyurethane insulation and external polyethylene wrap. Basis: Based on the provided documentation, Plains met the regulatory requirements for monitoring the cathodic protection (CP) system on Line 901, and the measured pipe to soil potential values met the required levels for protection. However, the presence of the polyurethane insulation and the polyethylene wrap shielded the cathodic protection current and prevented voltage monitoring of the shielded portions of the pipe. As a result, the CP current did not reach the pipe surface and the measured potentials did not represent the potentials at the areas of corrosion under the insulation. 3. The contracted in-line inspection significantly undersized the external corrosion feature that failed on Line 901. Basis: Based on the provided documentation, the 2015 MFL tool significantly undersized the external corrosion feature that ultimately leaked (i.e. a tool determined depth of 47% of the nominal wall thickness vs. a laboratory measured depth of 89% of the nominal wall thickness). The MFL tool likely also undersized the same feature in the 2012 ILI run based on a review and comparison of the 2007, 2012, and 2015 raw signal data for the feature that failed. 4. The mitigative actions taken by Plains on Line 901 did not adequately address the elevated integrity threat of corrosion under insulation. Basis: The results of the metallurgical analysis indicated that the immediate metallurgical cause of the failure was CUI. Corrosion under insulation is a unique corrosion mechanism that necessitates its own integrity risk assessment. Plains did not apply sufficient mitigative strategies specific to CUI to prevent this anomaly from failing. The measures could include enhancement of existing barriers and additional preventative barriers. DNV GL – OAPUS307KKRA (PP136049) v December 4, 2015#

Page 295Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Additional observations for improvement in the Integrity Management Program The following provides perspective on Plains’ integrity management plan (IMP) as related to the failure on Line 901. Coating systems, as a barrier to external corrosion related integrity threats, (i) are never perfect and (ii) age over time, thereby decreasing the effectiveness of the barrier. The cathodic protection (CP) system is another barrier to external corrosion integrity threats. Cathodic protection can be effective for many external corrosion related integrity threats; e.g., corrosion at holidays (holes in the coating) and microbiological influenced corrosion (MIC). There are limits to the effectiveness of CP for corrosion related integrity threats and mitigation barriers can be strengthened and/or other mitigation barriers can be employed in conjunction with CP; e.g., stray current enhanced corrosion, AC induced corrosion, stress corrosion cracking, and corrosion beneath disbonded coatings that shield CP current. For these, multiple barriers may be used depending on the individual integrity threat, but the ILI program in conjunction with a dig program becomes a more important barrier since it is known that the other barriers of coating and CP are not always effective. In the case of Line 901, Plains targeted 70 metal loss features in 2012, which included 31 features beyond those required by code and used for validation of the ILI program. These additional digs constitute a strengthened barrier in the prevention of a pipe failure due to a corrosion related integrity threat. Several of the digs were based on the strengthening of the ILI/dig barrier for the purpose of identifying and repairing corrosion under shrink sleeves used at girth welds; a known corrosion related integrity threat involving coatings that shield CP. In addition, the ILI re-inspection interval was decreased from a minimum of 5 years to 3 years (performed at 2.8 years). This also is a strengthening of a barrier in the prevention of a pipe failure due to a corrosion related integrity threat. Plains IMP aggressively addressed several of the corrosion related integrity threats; but, as mentioned under contributing causes, Plains did not apply sufficient mitigative strategies to prevent the CUI anomaly from failing. In addition, an IMP is only as good as the data that are utilized to monitor and measure its performance. As addressed as a contributing cause, the ILI significantly undersized (47% versus an actual value of 89% through wall) the feature that eventually failed. The RCA identified improvements that could be made within the integrity management program, which were not direct causes of the failure. These observations are given below. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 vi#

Page 296Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 1. Based on the information provided, Plains could adopt additional practices to identify and address any inaccuracies in future ILI runs. Basis: DNV GL performed an analysis of the 2012 ILI and dig data using API 1163, which is not a regulatory requirement or part of the IMP, and determined that the tool performance was not within the stated specifications. There was no produced documentation to indicate that Plains communicated with the ILI vendor, such as requesting a re-grade, following production of the unity plot[s] to account for the scatter observed within the data. However, using the recalculated tool tolerance would still result in a similar re-inspection interval as that used by Plains for the 2015 ILI run. 2. Based on the provided information, Plains could better incorporate the results from multiple ILI runs into their corrosion growth rate calculations. Basis: Plains IMP Section 9.2.2 states, “External and internal corrosion growth rates are estimated from multiple ILI runs, field observations, and observed historical growth rates.” The procedure specifies calculation of a corrosion growth rate in mils per year using the increase in corrosion depth during the time between consecutive ILI runs. There is no documentation provided to indicate that Plains performed such calculations using the historical ILI data. DNV GL calculated a corrosion growth rate for the feature that failed based on data from the 2007 and 2012 ILI runs. Although a higher corrosion rate was calculated than that determined using the CGAR process, this rate results in a similar re- inspection interval to that performed by Plains. Additional analyses that go beyond the IMP, codes, and standards, include:  Statistically active corrosion (SAC) analysis performed on Line 901 resulted in a similar re-inspection interval as that used by Plains (2.8 years) for the 2015 ILI run. The analysis identified a remaining life for the feature that failed that is greater than the re-inspection interval used by Plains. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 vii#

Page 297Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 3. Based on the provided information, Plains should improve their documentation and/or record-keeping of their decision-making processes related to actions taken. Basis: Over the course of the investigation, DNV GL identified areas within the integrity management process that were not sufficiently documented. For example, no justification (i.e. assumptions, analyses, etc.) was provided for determining the reassessment interval of 3 years based on the 2012 ILI data. Although a form explicitly identifying the justification for the reduction of their re- inspection interval from 5 years to 3 years was not provided, DNV GL’s assessments and calculations resulted in a similar re-inspection interval as that used by Plains. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 viii#

Page 298Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Table of Contents 1.0 INTRODUCTION ........................................................................................ 1 2.0 TECHNICAL APPROACH .............................................................................. 2 2.1 Methodology ............................................................................................. 2 2.2 Approach ................................................................................................. 2 3.0 TIMELINE OF EVENTS ................................................................................ 3 3.1 Key Events on Line 901 from 1990 to May 19, 2015 ...................................... 3 3.2 Key Events on Line 901 on May 19, 2015 ..................................................... 5 3.3 Probable Time of Failure ............................................................................. 5 3.3.1 Pressure Data ........................................................................................... 5 3.3.2 Leak Detection .......................................................................................... 6 4.0 IMMEDIATE / METALLURGICAL CAUSE ......................................................... 8 4.1.1 Summary of Metallurgical Findings .............................................................. 8 4.1.2 Immediate Cause Conclusion ...................................................................... 9 5.0 SUPPLEMENTAL ANALYSES ....................................................................... 10 6.0 BASIC ROOT CAUSES .............................................................................. 12 6.1 External Corrosion Control System ............................................................ 13 6.1.1 External Protective Coating System ........................................................... 13 6.1.2 Cathodic Protection System ...................................................................... 16 6.1.2.1 External Corrosion Data Review and Analysis .............................................. 18 6.2 Integrity Program .................................................................................... 21 6.2.1 Summary of Processes / Procedures Pertaining to Risk Assessments.............. 21 6.2.2 Summary of Processes / Procedures Pertaining to ILI Assessments ............... 22 6.2.2.1 Conducting Assessments and Processing Results ......................................... 22 6.2.2.2 Pipeline Repair Requirements.................................................................... 23 6.2.2.3 Continual Assessment and Evaluation of Pipeline Integrity ............................ 23 6.2.2.4 Identification of Preventive and Mitigative Measures .................................... 24 6.2.3 Available In-Line Inspection Data .............................................................. 25 6.2.4 Summary of Events Following the 2012 and 2015 In-Line Inspection Final Reports .......................................................................................... 25 6.2.5 Description and Review of the 2012 ILI CGAR Analysis as Applied to Joint 5930 .............................................................................................. 26 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 ix#

Page 299Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Table of Contents (Cont'd) 6.2.6 Description and Review of 2012 DOT Compliance Report (2012 Final Repair List) ............................................................................................ 29 6.2.7 Description and Review of the Validation of ILI Results ................................ 30 6.2.8 Description and Review of Re-Assessment Interval Determination ................. 32 6.2.9 Continual Evaluation and Assessment of Pipeline Integrity ............................ 35 7.0 SUMMARY AND CONCLUSIONS ................................................................. 35 DNV GL – OAPUS307KKRA (PP136049) x December 4, 2015#

Page 300Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Appendices Appendix A – BSCAT™ Methodology Appendix B – Documents Reviewed Appendix C – Supplemental Analyses for 2015 Digs Appendix D – Corrosion Product Supplemental Analyses - Density Testing Appendix E – Corrosion Product Supplemental Analyses - Magnetic Permeability Appendix F – Statistically Active Corrosion Assessment DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 xi#

Page 301Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis List of Tables Table 1. Summary of inspection data from aerial patrols of Line Segment 901 between January 7, 2015 and May 11, 2015. .............................................. 40 Table 2. Assessments on potential external corrosion mechanisms for the failure. ................................................................................................... 41 Table 3. Historical moisture conditions, based on 2007 and 2012 ILI dig information, for pipe joints near the 2015 failure location. ............................ 42 Table 4. Timeline of external corrosion control monitoring and inspection data. ..................................................................................................... 43 Table 5. Comparison of Rate Estimation Methods between 2007 and 2012 ILI. ............ 44 Table 6. Comparison of estimated time to reach 80% WT for features on Joint 5930. .................................................................................................... 45 Table 7. Summary of features selected for excavation following the 2012 ILI. ............. 46 Table 8. Comparison of re-assessment intervals for the feature associated with the 2015 Failure. .............................................................................. 47 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 xii#

Page 302Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis List of Figures Figure 1. Photographs showing the topography in the vicinity of the failure location. ................................................................................................. 48 Figure 2. Topographical map and plot showing the elevation profile for Line 901. The white triangles on the map correspond to mile post markers. The green star on the map and the green line on the plot identify the location of the May 19, 2015 failure. ......................................... 49 Figure 3. Photographs showing two of the culverts through which released product flowed. ....................................................................................... 50 Figure 4. Schematic showing the Loss Causation Model. ............................................ 51 Figure 5. Timeline showing key events for Line 901 from the time of construction to the day of the incident (May 19, 2015). ............................... 52 Figure 6. Timeline showing key events for Line 901 on the day of the incident (May 19, 2015). ...................................................................................... 53 Figure 7. Plot of pressure versus time showing the discharge pressure for Las Flores (red), the incoming pressure for Gaviota (blue), and the calculated pressure for Joint 5930 (green) on May 19, 2015 [Ref ‎227]. .............................................................................................. 54 Figure 8. Plot of pressure versus time showing the discharge pressure for Las Flores (red), the incoming pressure for Gaviota (blue), and the calculated pressure for Joint 5930 (green) on May 19, 2015 between 10:00 am and 12:00 pm [Ref ‎227]. ........................................................... 55 Figure 9. Schematic showing Line 901 from Las Flores to Gaviota showing the approximate location of the pipeline and the elevation profile of the pipeline. The red arrows indicate the locations of flow meters [Ref ‎248]. .............................................................................................. 56 Figure 10. Schematic showing Line 903 from Gaviota to Sisquoc the approximate location of the pipeline and the elevation profile of the pipeline. The red arrows indicate the locations of flow meters [Ref ‎249]. .............................................................................................. 57 Figure 11. Schematic showing Line 903 from Sisquoc to Station Number 1595 + 16 the approximate location of the pipeline and the elevation profile of the pipeline. The red arrows indicate the locations of flow meters [Ref ‎250]. ................................................................................... 58 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 xiii#

Page 303Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis List of Figures (Cont'd) Figure 12. Schematic showing Line 903 from Station Number 1595 + 16 to Pentland the approximate location of the pipeline and the elevation profile of the pipeline. The red arrows indicate locations of flow meters [Ref ‎251]. ................................................................................... 59 Figure 13. Plot showing volume versus time data for the Las Flores to Pentland line segment between May 18, 2015 at 5:00am and May 20, 2015 at 12:00am. ............................................................................................... 60 Figure 14. Plot showing volume versus time data for the Las Flores to Pentland line segment between May 19, 2015 at 1:00am and May 20, 2015 at 12:00am. ............................................................................................... 60 Figure 15. Photograph showing the failure location before (Top) and after (Bottom) cleaning. Tape measure indicates distance to U/S GW (Note: tape slipped 0.1’ to the right). ........................................................ 61 Figure 16. BowTie diagram, generated using the BSCA T™ methodology, summarizing the preventative barriers in place for the Line 901 Release. The barriers shown as two rectangles on either side of the horizontal line correspond to failed barriers, while the thin rectangles that span the horizontal line correspond to ineffective barriers. ..................... 62 Figure 17. Schematic and photograph showing the protective external coating and the PU foam and PE tape layers present on the Line 901 pipeline. ................................................................................................. 63 Figure 18. Photographs showing compromised protective coating and PU foam/PE tape layers at the failure location.................................................. 64 Figure 19. Photograph of wrinkles in the PE tape, located away from the failure location. ................................................................................................. 65 Figure 20. Photographs showing compromised coating and PU foam away from the failure location. ................................................................................. 66 Figure 21. Plot showing the distribution of external metal loss features vs. o’clock orientation identified for Line 901 during the 2007, 2012, and 2015 ILI runs. ........................................................................................ 67 Figure 22. Photographs showing recoated pipe on Line 901 after: (a) 2007 ILI Dig #5 and (b) 2012 ILI Dig #13 [Ref ‎147 & ‎169] ....................................... 68 Figure 23. Plot of temperature data, provided by Plains, for Las Flores Station between May 2014 and May 2015 [Ref ‎226]. .............................................. 69 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 xiv#

Page 304Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis List of Figures (Cont'd) Figure 24. Plains All American Line 901 "IRF" pipe-to-soil potential annual test point survey data years 2008, 2014, and 2015. Note: IRF: free of IR error. ..................................................................................................... 70 Figure 25. Plains All American Line 901 In-Line Inspection 2007 maximum external metal loss depth reported aligned with 2008 Close Interval Survey (- Pipe Nominal Wall Thickness). .................................................... 71 Figure 26. Plains All American Line 901 In-Line Inspection 2012 maximum external metal loss depth reported aligned with 2015 Close Interval Survey (- Pipe Nominal Wall Thickness). .................................................... 72 Figure 27. Plains All American Pipeline Las Flores I Rectifier: Direct current (DC) output recorded between 2005 - 2015. ...................................................... 73 Figure 28. Plains All American Pipeline Las Flores II Rectifier: Direct current (DC) output recorded between 2005 - 2015. .............................................. 74 Figure 29. Plains All American Pipeline Gaviota I Rectifier: Direct current (DC) output recorded between 2005 - 2015. ...................................................... 75 Figure 30. Plains All American Pipeline Gaviota II Rectifier: Direct current (DC) output recorded between 2005 - 2015. ...................................................... 76 Figure 31. Plains All American Pipeline L 901 Cathodic Protection Rectifiers: Average direct current (DC) output recorded between 2005 - 2015. .............. 77 Figure 32. Plains All American Line 901 2015 close interval potential survey and annual test point survey data recorded between 2009 and 2014. .................. 78 Figure 33. Timeline of events associated with Line 901, following the 2012 ILI. .............. 79 Figure 34. Excerpt from IMP Fig 9-2 illustrating process to estimate corrosion growth rates [Ref ‎22]. ............................................................................. 80 Figure 35. Representation of reported metal loss features on Joint 5930 ....................... 80 Figure 36. Depths of ILI-reported metal loss features on Joint 5930 ............................. 81 Figure 37. Metal loss depth unity plot using Plains data. .............................................. 82 Figure 38. Metal loss depth unity plot using Plains data. Light blue diamonds correspond to features located greater than 2 feet from a girth weld. Purple diamonds correspond to features within 2 feet of a girth weld. ............ 83 Figure 39. DNV GL-produced metal loss depth unity plot for the 2015 ILI of the Las Flores to Gaviota line segment. ........................................................... 84 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 xv#

Page 305Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis List of Figures (Cont'd) Figure 40. Excerpt from API 1163 used to establish consistency with performance specification (Table 8 in Appendix E, [Ref ‎309]). ...................... 85 Figure 41. Snapshot showing portion of Figure 6-1 from Section 6.2 of Plains’ IMP, regarding regrading [Ref ‎20]. ............................................................ 85 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 xvi#

Page 306Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Acronyms BSCAT™ Barrier-based Systematic Cause Analysis Technique CGAR Corrosion Growth Analysis Report COF Consequence of failure CP Cathodic Protection CPM Computation Pipeline Monitoring CTU Coal tar urethane D/S Downstream ERW Electric resistance weld GIS Geographic information system GW Girth weld HF High frequency ICCP Impressed Current Cathodic Protection ILI In-line Inspection IMP Integrity Management Plan LCM Lost Causation Model LOF Likelihood of failure LDS Leak Detection System MDP Maximum Discharge Pressure ML Metal loss MFL Magnetic Flux Leakage MOP Maximum Operating Pressure MP Mile Post MPI Magnetic Particle Inspection NWT Nominal wall thickness OCS Outer Continental Shelf OEM Office of Emergency Management P&M Preventative & Mitigative PE Polyethylene PHMSA Pipeline and Hazardous Materials Safety Administration PLM Pipeline Monitor PU Polyurethane RCA Root Cause Analysis RGW Reference Girth Weld ROF Risk of failure SBC Santa Barbara County SBCFD Santa Barbara County Fire Department SCADA Supervisory Control and Data Acquisition SCAT Systematic Causal Analysis Technique SMYS Specified Minimum Yield Strength U/S Upstream WT Wall Thickness DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 xvii#

Page 307Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 1.0 INTRODUCTION Plains All American Pipeline, L.P. (Plains) retained Det Norske Veritas (U.S.A.), Inc. (DNV GL) to perform a root cause analysis (RCA) of a failure that occurred on Line Segment 901, which transports heated crude oil from the outer continental shelf (OCS). The failure occurred on May 19, 2015 in Goleta, California (Santa Barbara County) and was located near milepost (MP) 4. The location was approximately 4.05 miles downstream (D/S) from Las Flores Pump Station and approximately 6.2 feet upstream (U/S) from the nearest girth weld, identified as Girth Weld (GW) 5940. As a result of the failure, approximately 2,9342 barrels of crude oil were estimated to have been released. The leak occurred in a mostly rural area that runs along the coastline of the Pacific Ocean. The topography in the area is hilly, with the pipeline oriented uphill from the ocean. ‎Figure 1 contains photographs showing the topography in the vicinity of the failure. ‎Figure 2 contains a topographical map and elevation plot of Line 901. As shown in the figure, the failure was located near a local low point along the pipeline. Several road crossings, such as Highway 1, are present in the area with drainage toward the coast via culverts. It is via these culverts that the released oil reached the Pacific Ocean at Refugio State Beach. ‎Figure 3 contains photographs showing the first two culvert through which the released product flowed. The photograph to the left in the figure corresponds to the culvert closest to the release site. A makeshift berm was created at this culvert to prevent any additional product from flowing through the culvert. The photograph to the right in the figure corresponds to the second culvert through which product flowed. This culvert ran beneath Highway 101. The portion of the pipeline that contained the failure is comprised of 24-inch diameter by 0.344 inch wall thickness, API 5L Grade X65 line pipe steel that was manufactured by Nippon Steel and contains a high frequency (HF) electric resistance welded (ERW) longitudinal seam. The pipeline (Line 901) was installed in 1990 and is approximately 10.87 miles in length, spanning between Las Flores Station on the U/S end and Gaviota Station on the D/S end. The pipeline is externally covered with the following: (1) a protective coating of coal tar urethane (CTU) that is in intimate contact with the steel pipe, (2) a layer of rigid thermal polyurethane (PU) foam insulation, and (3) an outer layer of polyethylene (PE) tape. The pipeline has an impressed cathodic protection (CP) system that was energized at the time of installation. The normal operating pressure and maximum discharge pressure (MDP) for the line are 616 psig and 1,025 psig, respectively. These pressures correspond to 33% and 55% of the 2 [Ref ‎6] The final volume estimate for the released oil at the time of this report. DNV GL – OAPUS307KKRA (PP136049) 1 December 4, 2015#

Page 308Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis specified minimum yield strength (SMYS), respectively. The pressure at the time and location of the failure was reported by Plains to be 737 psig [Ref ‎2], which corresponds to 39.6% of the SMYS and 71.9% of the MDP. The portion of the pipeline that contained the failure location was removed and sent to DNV GL to determine the metallurgical cause of the failure and to identify any contributing factors. The conclusions and recommendations for this RCA are based on the findings from the final metallurgical report as well as information provided and publically reported by Plains. The objective of the RCA was to identify factors contributing to the failure and document the decision-making process. Based on the findings of the analysis, recommendations for improvements also are identified. 2.0 TECHNICAL APPROACH 2.1 Methodology The methodology used by DNV GL for the RCA of the Line 901 release was based on DNV GL’s Loss Causation Model (LCM). The DNV GL LCM used in the analysis is shown in ‎Figure 4. This model is built on the concept that incidents can be attributed to immediate causes, basic causes, and failures of management systems to control hazards. The analysis uses a systematic method of processing evidence gathered during an investigation in order to identify the factors that led to the incident. This methodology assists in the development of corrective and/or remedial measures. The LCM approach used by DNV GL is called a Barrier-based Systematic Causal Analysis Technique (BSCAT™). BSCAT™ is a technique that applies a Systematic Causal Analysis Technique (SCAT) model to each barrier, as opposed to the incident as a whole. This method results in a thorough review of the effectiveness of individual barriers identified in the risk assessment. BSCAT™ provides a methodology that allows for the analysis of complex incidents that involve multiple barriers. Detailed information about the BSCAT™ methodology and its application is provided in Appendix A. 2.2 Approach DNV GL reviewed various materials provided and publically reported by Plains (i.e. technical documents, manuals, maps, and data) and produced by DNV GL. The materials are grouped into the following categories: (1) incident related documents - References 1 – 7, (2) integrity-related documents (i.e. integrity management plan, cathodic protection surveys, in-line inspections, and excavation reports and digs) - References ‎8 – ‎200, (3) leak detection documents - References ‎201 – ‎223, (4) operations documents - References ‎224 – ‎242, (5) historical documents - References ‎243 – ‎246, (6) drawings, maps, and diagrams - DNV GL – OAPUS307KKRA (PP136049) 2 December 4, 2015#

Page 309Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis References ‎247 - ‎293, (7) public reports issued by Plains - Reference ‎294 – ‎298 and (8) standards, papers, etc. - References ‎299 – ‎316. A complete list of the materials reviewed for the RCA is provided in Appendix B. The documents listed above were used for the following tasks: 1. Timeline creation of events leading up to the incident. 2. Immediate (Metallurgical) cause determination for the incident. 3. Basic cause(s) determination for the incident. 4. Technical root cause(s) determination for the incident. It is important to note that the analyses described within this report were only performed for the segment of the pipeline affected by the incident (i.e. Line 901). The findings and discussion presented in this report are not representative or indicative of the entire pipeline system and programs covered by Plains and Plains subsidiaries. The results and analysis incorporated herein are based on the provided documentation listed in Appendix B. DNV GL reserves the right to modify or supplement the report should new information become available. 3.0 TIMELINE OF EVENTS Two timelines were developed to help visualize the events that occurred leading up to the incident. Documents provided by and public reports issued by Plains were used to populate the timelines with relevant information. The first timeline incorporates key events that occurred on Line 901 between the time of construction to the day of the incident (May 19, 2015). The second timeline incorporates key events that occurred on the day of the incident up until the identification of the failure. These timelines were used to identify the barriers in place to prevent the incident and to identify the probable time of failure. 3.1 Key Events on Line 901 from 1990 to May 19, 2015 ‎Figure 5 is a timeline showing key events for Line 901 from the time of construction to the day of the incident. The timeline includes dates for (1) construction (olive green circles), (2) system ownership change (green circle), (3) in-line inspections (ILIs) (purple triangles), (4) ILI excavation digs (red lines), (5) close-interval surveys (blue lines), and (6) the May 19, 2015 failure (teal square). Five key events were identified relating to the construction and ownership of Line 901. The line pipe was manufactured in 1985 [Ref ‎246], but it was not installed until 1990 by All American Pipeline [Ref ‎2]. It was coated with mill-applied coal tar urethane and insulated DNV GL – OAPUS307KKRA (PP136049) 3 December 4, 2015#

Page 310Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis with 1.5 inches of mill-applied polyurethane foam in a double-joint configuration [Ref ‎244]. Girth welds performed during construction were coated using Raychem WPC M100-27000 x 34/A/Uni shrink sleeves combined with Raychem #S-1142 primer kits [Ref ‎244]. Cathodic protection in the form of impressed current was installed in 1990, the same year as the pipeline installation [Ref ‎2]. A hydrostatic test of the line was performed at Gaviota Station on Nov. 25, 1990. The test pressure of 1719 psig was held for 8 hours [Ref ‎2]. In 1994, the Las Flores Canyon Pump Station was constructed [Ref ‎296]. Four years later, All American Pipeline was acquired by Plains and Line 901 became part of Plains assets [Ref ‎297]. Four ILIs were performed on Line 901 between 1996 and May 19, 2015 [Ref ‎294]. Details about the vendor, tool(s) used, and the results for the 1996 ILI (ILI 1) were not available for review. The ILI vendor in the 2007 (ILI 2), 2012 (ILI 3), and 2015 (ILI 4) inspections was ROSEN, located in Houston, TX [Refs. ‎50, ‎91, ‎138]. For ILI 2, two tools were run– a geometry tool and a metal loss tool (magnetic flux leakage [MFL]). The tool type used for ILI 3 and ILI 4 was a combination MFL and deformation tool. Based on the results of the ILIs, digs were initiated in prioritized areas identified by Plains’ integrity management plan (IMP) within one year of the ILI tool runs. Thirteen digs3 were conducted between February 21, 2008 and March 3, 2009. Between October 15, 2012 and October 3, 2013, 44 digs were performed4 . After the 2015 ILI and the failure, 4 digs were performed in prioritized areas.5 Cathodic Protection Close-Interval Criteria Survey (CIS) assessments were conducted in December of 2008 (CIS1) and April of 2015 (CIS2) [Ref ‎31 – ‎44]. The CIS vendor was Hanson Survey & Design, from Houston, TX. As part of the monitoring program utilized by Plains for leak detection, aerial patrols were conducted routinely on Line 901 (on a weekly basis, approximately). The patrols were conducted by Kern Charter Inc. (Kern) of Line 901 from Las Flores to Gaviota and Line 902 from Gaviota Station to the Gaviota Booster [Refs. ‎216 – ‎220]. ‎Table 1 summarizes the inspection data from aerial patrols of Line Segment 901 between January 7, 2015 and May 11, 2015. Between these dates, 18 reports were completed. Three to twelve days separated the inspection dates. On three occasions (January 16, April 1, and April 17, 2015), weather prevented the inspection of Line 901. No leaks were identified by these aerial patrols. Surface patrols of the right of way (ROW) were not performed as part of 3 2007: Digs 3 (WC5365.72), Dig 3 (WC 5342.18), Digs 4 – Dig 11, Dig 11B, Dig 12, & Dig 13. [Refs. ‎144 - ‎156] 4 2012: Digs 1 – 19, Dig 20, Dig 20A, Dig 21, Dig 21A, Digs 22 – 33, Dig 33A, Digs 34 – Dig 41. [Refs. ‎157 -‎199] 5 2015: Digs 1 through 4. [Ref ‎200] DNV GL – OAPUS307KKRA (PP136049) 4 December 4, 2015#

Page 311Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Plains IMP of Line 901. The last aerial patrol prior to the failure was performed on May 11, 2015 by Kern. 3.2 Key Events on Line 901 on May 19, 2015 ‎Figure 6 is a timeline showing key events for Line 901 on the day of the incident (May 19, 2015). The timeline includes times for (1) operational events (blue triangles), (2) calls to the National Response Center [NRC] (purple “X”s), (3) responses by local personnel to NRC calls (green triangles), and (4) and failure confirmation by Plains (teal square). At approximately 10:55 am [Ref ‎294], an unplanned pump shutdown at the Sisquoc station occurred. The pump was successfully restarted. At 11:15 am, the pump at Sisquoc station was shut down [Ref ‎294]. Fifteen minutes later, the pump at Las Flores Station was shut down by Midland Control to prevent packing of the line [Ref ‎298].6 Line 901 was isolated at this time. Three calls were placed to emergency response entities, one call was placed to the Santa Barbara County (SBC) Fire Department and two calls were placed to the National Response Center (NRC). The SBC Fire Department was first notified of an odor near Refugio Beach at 11:42 am by an unidentified member of the public. State Parks staff were alerted to the 911 call and attempted to locate the source of the odor around 12:00 pm. SBC Emergency Management was then notified of the presence of oil on Refugio State Beach at 12:30 pm by SBC Fire Department. A call was placed, by an unidentified caller, to the NRC at 12:43 pm (1116950) reporting an oil sheen on Refugio State Beach. Around 1:30 pm, Plains confirmed a failure on Line 901 near Refugio State Beach. A call was placed by Plains to the NRC at 2:56 pm (1116972). 3.3 Probable Time of Failure 3.3.1 Pressure Data ‎Figure 7 is a plot of pressure versus time data for the discharge pressure for Las Flores (red, Ref ‎227), the incoming pressure for Gaviota (blue, [Ref ‎227]), and the calculated pressure for Joint 5930 (green) on May 19, 2015.7 The maximum recorded discharge pressure for Las Flores and incoming pressure for Gaviota on May 19 was 721 psig and 707 psig, respectively. These pressures were recorded at 12:55 pm and 12:54 pm, respectively. The maximum pressure data from Las Flores and Gaviota correspond to a pressure of 814 psig8 6 The remaining times referenced in this paragraph are from [Ref ‎298]. 7 Calculated pressures determined by DNV GL. 8 Calculated value based on OPS TTO5 – Low Frequency ERW and Lap Welded Longitudinal Seam Evaluation (p. 23), April 2004. Discrepancy with the value reported by Plains may be associated with the equation used to calculate value. This equation used by DNV GL: DNV GL – OAPUS307KKRA (PP136049) 5 December 4, 2015#

Page 312Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis at the location of the failure. The time at which the maximum pressure was recorded occurred between the first call to the NRC and before Plains confirmed the failure on Line 901. ‎Figure 8 plot of pressure versus time data for the discharge pressure for Las Flores (red), the incoming pressure for Gaviota (blue), and the calculated pressure for Joint 5930 (green) on May 19, 2015 between 10:00 am and 12:00 pm. The corresponding times for key events associated with operational events (blue triangles shown in ‎Figure 6) are indicated. Based on the pressure data, the unplanned and planned shutdowns at Sisquoc did not cause an increase in pressure, which would be expected due to line packing. There is a slight increase in pressure after the Las Flores Pump was shut down. Twelve minutes after the Las Flores pump was shutdown, a 911 call was placed to the SBC Fire Department notifying of the odor near Refugio State Beach. 3.3.2 Leak Detection Leak detection is performed on the Plains pipeline system using computation pipeline monitoring (CPM). Plains uses two systems for CPM: (1) Pipeline Monitor (PLM) and (2) SimSuite Leak Detection System (LDS) [Ref ‎201]. For the affected line segment, the PLM approach was utilized. PLM compares the metered in to the metered out using SCADA at all inlet and outlet connections. ‎Figure 9 through ‎Figure 12 contain alignment sheets for Line 901 from Las Flores to Gaviota, Line 903 from Gaviota to Sisquoc, Line 903 from Sisquoc to Station Number 1596+16, and Line 903 from Station Number 1596+16 to Pentland, respectively. Calculations performed using the PLM were done using all of the inlet and outlet metered data between Las Flores and Pentland. In total, there are eleven locations that are part of the calculation, five inlets and six outlets (locations shown as red arrows in the figures). There are six rolling time periods that are examined as part of PLM: (1) LT1 – 1 hour, (2) LT2 – 5 hour, and (3) LT3 – 24 hour, (4) ST1, (5) ST2, and (6) ST3.9 For Line 901 between Las Flores and Pentland, LT2 and LT3 were utilized in calculating the metered amount in that portion of the line segment in barrels. Plains calculated the overshort in two ways (1) historical and (2) estimated. The historical data are based on real-time data from SCADA and the estimated data are based on an approximation of total metered amount if the real- time data were not available. The calculated overshort data are monitored in the by a Leak Detection Engineer in the Plains’ Control Center located in Midland, Texas. 9 Acronyms LT and ST are not defined in provided documentation. DNV GL – OAPUS307KKRA (PP136049) 6 December 4, 2015#

Page 313Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Threshold alarm set points are selected by the Leak Detection Engineer based on the historical operating data for the pipeline and the events taking place on the pipeline. For instance, when product is flowing, an overshort value of 150 bbls is typical; however, when there is a pump shutdown (like the one preceding the detection of product outside the pipeline on May 19) a threshold value of 600 bbls is used. When the upper or lower threshold limits are violated, a PLM alarm indicating the over or the short is recorded in the SCADA. These instances are recorded as “critical” and have an audible sound associated with the event. An investigation into these types of events is immediately launched by the Leak Detection Controller. ‎Figure 13 and ‎Figure 14 are plots showing volume versus time data for the Las Flores to Pentland line segment for the for LT2 and LT3 rolling time calculations [Ref ‎226]. These span the time frames of May 18, 2015 at 5:00 am and May 20, 2015 at 12:00 am and between May 19, 2015 at 1:00 am and May 20, 2015 at 12:00am, respectively. As shown in the figures, there is a downward trend in the total metered amount around 12:30 pm on May 19. The estimated and historical lines then diverge at ~1:23 pm around a short of 600 bbls. In the SCADA between May 5, 2015 and May 19, 2015, twelve PLM alarms associated with Las Flores to Pentland segment of the pipeline were logged [Ref ‎206]. Ten of the PLM alarms were associated with events on May 6, 2015. These were associated with the ILI of the line pipe by ROSEN on that date. The two remaining PLMs took place on May 19, 2015 at 1:22:58 pm – the first was an alarm event and the second was the corresponding control description. The alarm event was associated with a violation of the “short” threshold (600 bbls) of the PML. The PLM was inhibited10 as a control by the leak detection engineer. By inhibiting the line, real-time recording of the inlet and outlet meters stopped. Hence, historical data were used to estimate the overshort values starting at 1:23 pm on May 19 (see ‎Figure 13 and ‎Figure 14). The pipeline was not shut-in at this time; however, an investigation into the alarm was initiated per Plains’ requirements outlined in Chapter 100-8 [Ref ‎201]. Based on a review of Plains Leak Detection methodologies, the overshort plots from the day of the event, and the SCADA from the two weeks prior, there is no evidence to suggest a slow leak was present within the system, which is consistent with the findings of the metallurgical report that indicated a sudden failure event. 10 The term “inhibited” means that the alarm was acknowledged by the leak detection engineer, and then silenced in order to begin an investigation in the alarm. DNV GL – OAPUS307KKRA (PP136049) 7 December 4, 2015#

Page 314Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 4.0 IMMEDIATE / METALLURGICAL CAUSE 4.1.1 Summary of Metallurgical Findings DNV GL performed a metallurgical analysis on the portion of the pipeline that failed and concluded that “the failure occurred at an area of wall thinning from external corrosion that ultimately failed by ductile overload under the imposed operating pressure. The morphology of the external corrosion observed on the pipe section is consistent with corrosion under insulation facilitated by wet-dry cycling.”[Ref ‎1] ‎Figure 15 contains photographs of the failure location, provided in the metallurgical report, before and after cleaning. The failure opening was determined to be 6.6 inches in length axially with a maximum opening of 1.14 inches. The failure was located at the 4:15 o’clock orientation within an area of external corrosion that extended 12.1 inches in the longitudinal direction and 7.4 inches in the circumferential direction. The maximum depth of the external corrosion was 89% of the measured wall thickness at the failure location. No portion of the flaw was through wall prior to the ductile overload failure and, therefore, the failure event was sudden in nature. During the investigation, several external corrosion features were identified along the bottom of the joint that failed. These features were in addition to the corrosion feature associated with the failure and were covered by thick, layered deposits that were magnetic. Chemical analyses performed on the deposits revealed that they were primarily comprised of layers of goethite and magnetite11, two forms of iron oxide. No evidence of calcareous deposits was detected within the deposits, indicating that CP likely did not reach these areas. The areas where the external corrosion features were located corresponded to areas of compromised coating. The coating at these locations consisted of a combination of disbonded coal tar urethane, compressed and water saturated insulation, and wrinkled polyethylene tape. The nature of the coating damage allowed for the ingress of water to the pipe surface, which facilitated the corrosion. Examination of the fracture surfaces from the failure location revealed the presence of two regions. The region near the external surface was nondescript and consistent with corrosion, while the region near the internal surface was dimpled and consistent with ductile overload. No evidence of in-service growth was identified on the fracture surface, indicating that the failure corresponded to a single sudden event. Chemical and mechanical testing was performed on the pipe joint that failed. The results of those tests revealed that the steel was consistent with the vintage and grade of steel. No 11 The chemical formula for goethite and magnetite are FeO(OH) and Fe3O4, respectively. DNV GL – OAPUS307KKRA (PP136049) 8 December 4, 2015#

Page 315Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis evidence of any metallurgical defects that may have played a role in the failure was identified within the steel. 4.1.2 Immediate Cause Conclusion The potential for various mechanisms that may have caused the external corrosion at the failure location were considered during the metallurgical investigation. The mechanisms considered included the following: (1) AC stray current corrosion, (2) DC stray current corrosion, (3) galvanic corrosion, (4) microbiologically influenced corrosion (MIC), and (5) corrosion under insulation (CUI). ‎Table 2 summarizes assessments for the potential external corrosion mechanisms at the failure location. The table is broken into three columns. The first column lists potential mechanisms (i.e. AC stray current corrosion, galvanic corrosion, etc.) that may have caused the corrosion. The second column contains the relevance of each mechanism to the corrosion observed at the failure location. The third column lists supporting evidence for the assessment given in column two. AC and DC stray current corrosion were both eliminated as potential mechanisms for several reasons. These phenomena do not occur beneath shielding coatings. The morphology of the corrosion and the associated corrosion products are not consistent with AC or DC stray current corrosion. Furthermore, field measurements indicated there was negligible AC voltages on the pipeline at the failure location and there was no high voltage AC (HVAC) lines or sources of DC stray current in the right of way (ROW). Galvanic corrosion was also eliminated as the primary cause of the corrosion. This is based on the fact that there was no evidence of dissimilar metals near the corrosion features observed on the failed pipe joint. MIC was eliminated as the primary cause of the corrosion, but may have played a contributing role. Bacteria were identified at a corrosion feature sampled U/S from the failure location. The levels of bacteria detected, however, were low. This finding coupled with the dense layered morphology of the corrosion products is not consistent with MIC. Based upon the results of the analysis, the most probable cause of the external corrosion is the mechanism of CUI. This conclusion is based upon (1) the morphology of the corrosion [i.e. mix of general corrosion and pits], (2) the thick layered morphology of the corrosion products, (3) the location of the corrosion [beneath saturated insulation], and (4) the association of the corrosion with compromised coating. The presence of wrinkling and cracks in the outer polyethylene tape coating likely allowed for the ingress of water to reach the pipe surface and facilitate corrosion. DNV GL – OAPUS307KKRA (PP136049) 9 December 4, 2015#

Page 316Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Thus, the immediate cause of the failure on Line 901 was determined to be external corrosion due to a CUI mechanism. Based on this finding, DNV GL reviewed historical documents regarding the service history of the line to identify contributing factors to the failure. 5.0 SUPPLEMENTAL ANALYSES Four priority digs, identified as Digs 1 - 4, were performed between May 29, 2015 and June 3, 2015, based on the preliminary findings of the 2015 ILI run. These locations were selected based on the maximum depths, identified by the tool, for external metal loss features on Line 901. DNV GL personnel were present during all four digs and collected various samples. The collected samples included the following: (1) corrosion products associated with the features, (2) swab samples for bacteria testing, (3) soil samples, and (4) coating insulation removed at the feature locations. The results of these analyses are summarized below and details are provided in Appendix C.  The corrosion products ♦ Are primarily dark brown in appearance with some areas that were rust-colored. ♦ Are dry, rigid, and magnetic. ♦ Consist of a layered morphology comprised primarily of goethite and magnetite.  There is no strong evidence to indicate that MIC played a primary role in the observed external corrosion observed for Digs 1 – 4.  The results of analyses performed on soil samples, removed near the failure and dig locations, revealed that the soil removed near the failure location exhibited higher corrosive properties.  Analyses of liquids extracted from insulation samples removed near the corrosion features from Digs 1 – 4 revealed higher concentrations of corrosive species (i.e. chlorides) than their respective soil samples. The corrosion products removed near the failure location were found to be tightly adhered to the surface of the pipe, such that mechanical means (i.e. hammer and chisel) were necessary to remove the products. The products were fairly rigid, coming off in sheets. Compound analyses performed on the products revealed that they are comprised of multiple alternating layers of magnetite and goethite. The products are also attracted to a magnet, indicating that the products may have affected the response seen by the tool. Based on these findings, analyses were performed on corrosion product samples removed from the DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 10#

Page 317Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis pipe joint that failed to assess the potential impact, if any, they had on the sizing capabilities of the MFL tool. The influence of corrosion products on MFL depth sizing has previously been noted in the literature. Bowerman et al. observed inaccuracies in pit depths, as reported by an MFL tool, when ferromagnetic debris was present within corrosion features [Ref ‎307]. Specific compounds identified within the debris included magnetite, iron sulfide, siderite, and hematite. These researchers tested deeper corrosion features that contained the products than had been reported by the tool. They speculated that the deposits decreased the induced magnetic flux and reduced the quality of the acquired data. Similar findings were observed by Kasai et al. [Ref ‎308]. These researchers observed that ferro- and semi- magnetic products within corrosion features caused distortion of the flux field pattern that impacted the ILI detection and sizing performance. In their cases, the features appeared smaller than their actual size. Based on the nature of the deposits, density and magnetic permeability measurements were performed on corrosion product samples removed near the 2015 failure location on Line 901. The results of the density testing are presented in Appendix D and revealed that a representative corrosion product, identified as Corrosion Product Sample 10000195318, had an approximate density of 3.53 g/cm3, which is approximately 45% of the density of low carbon steel. The product tested was removed from Feature 2 on the pipe joint that contained the 2015 failure (i.e. Pipe Joint 5930). The results of the magnetic permeability testing are presented in Appendix E and revealed the following:  The corrosion product specimens were less magnetic than the steel specimens.  No significant differences were determined for the magnetic properties of the specimens removed from the two corrosion product samples.  There were differences between the magnetic properties of the steel specimen in the axial (longitudinal) direction and the magnetic properties of the steel specimen in the transverse (circumferential direction).  At the field strengths typically associated with MFL tools, the magnetic permeability values of the corrosion product specimens were significantly lower than the magnetic permeability values of the steel specimens. The values for the corrosion product specimens were less than 5% of the values determined for the steel specimens. These results indicate that the magnetic nature of the deposits alone likely did not significantly impact the sizing capabilities of the MFL tool. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 11#

Page 318Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 6.0 BASIC ROOT CAUSES Basic root causes are contributing factors that are usually determined during the review of engineering controls and operational procedures. They may also be referred to as “indirect” causes. As shown in the schematic of the Loss Causation Model (See ‎Figure 4), basic causes lead to the immediate cause(s). There are a number of integrity assessment and integrity assurance methodologies that can be used on a pipeline. These methodologies are engineering controls that are typically used to prevent and/or assess for threats to pipeline integrity. The controls are considered “barriers” from the perspective of a root cause analysis. For this incident, the barriers fall into two main categories: (1) external corrosion control system and (2) integrity management program. Within each category, several areas that may have affected/contributed to the failure were considered. These areas are outlined below: 1. External Corrosion Control System External protective coating system – a method used to prevent moisture ingress to prevent corrosion. Cathodic protection (CP) system – an applied current used to counteract the natural electrochemistry of corrosion. 2. Integrity Program Contracted In-line inspection - a technology used to identify sections of metal loss in the pipeline. Mitigative actions – measures to address a specific threat that can include enhancement of existing barriers and/or the use of additional preventative barriers An analysis of these areas was performed using the BSCAT™ methodology. Ineffective, failed, and missing barriers related to the failure were identified. Effective, ineffective, failed, and/or missing barriers related to the failure were identified. The term “Effective” is used to describe a barrier that is performing in the manner as originally intended. “Ineffective” is a term used to describe a barrier that is in place and operating, but its performance is deficient. The term “Failed” is used to describe a barrier that was originally in place, but has degraded and no longer functions as originally intended. “Missing” is used to describe a barrier that was never in place. These barriers are graphically represented in ‎Figure 16 and discussed below by area. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 12#

Page 319Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 6.1 External Corrosion Control System The results of the metallurgical analysis [Ref ‎1] indicate that the leak occurred at an area of external metal loss due to corrosion that ultimately failed by ductile overload under the imposed operating pressure. Buried carbon steel pipelines are normally protected against external corrosion by a combination of an external coating and cathodic protection (CP). Plains’ Operations and Maintenance Manual O&M - 412 (OM412) [Ref ‎231] provides procedures to ensure the implementation of a sound corrosion control program to meet or exceed the minimum federal safety standards as defined by Title 49, Part 195 of the Code of Federal Regulations for Hazardous Liquids [Ref ‎316], developed by the U.S. Department of Transportation (DOT) Pipeline and Hazardous Materials Safety Administration (PHMSA). Both an external coating and CP system were in place on Line 901 to minimize the threat of external corrosion. Since the immediate cause of the failure is external corrosion, factors associated with one or both of these barriers failed and/or was ineffective. Details on both the external protective coating and CP system are described below. 6.1.1 External Protective Coating System The use of an external protective coating is one of the primary barriers used to prevent degradation of the external surface of a pipeline. The coating serves to prevent exposure of the external pipe surface to the surrounding soil environment and potentially corrosive conditions. When coating failure does occur, the remaining intact coating reduces the surface area of exposed metal, thereby decreasing the CP current requirements for protection. Line 901 is externally coated with a protective CTU. In addition to the protective coating, the external surface of the pipeline is also covered with a rigid PU foam and a white Polyken (PE) tape [Ref ‎244]. The use of the PU foam and PE tape was selected at the time of construction, by All American Pipeline, to maintain the temperature of the heated oil within the pipeline and minimize heat losses during transit. The PU foam was well bonded to the CTU coating and the PE tape was wrapped around the PU foam to reduce the ingress of water. ‎Figure 17 contains a schematic and a photograph showing the location of the CTU coating, the PU foam, and the PE tape with respect to the bare pipe steel. The CTU was identified as LAC-450 [Ref ‎131] and is in intimate contact with the steel. The average thickness of the coating ranged from 0.040 to 0.043 inches, as reported in the metallurgical report [Ref ‎1]. The outer PU foam was approximately 1.5 inches thick at the time of installation [Ref ‎244]. . The protective CTU coating, PU foam layer, and PE tape were compromised at the failure location, based on the evidence provided in the metallurgical report. The damage included DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 13#

Page 320Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis wrinkles, cracks, staining, and decohesion of the PE tape; staining, water saturation, and compression of the PU foam; and disbondment of the CTU [Ref ‎1]. The compression and saturation of the PU foam were found to be concentrated only along the bottom of the pipeline. The PU foam was found to exhibit minor to no evidence of compression and saturation along the top of the pipe. ‎Figure 18 contains representative photographs of the damage observed on the protective coating and the outer layers at the failure location. In addition to the damage, thick layers of corrosion products were found wedged between the protective CTU coating and the pipe steel. The presence of the corrosion products beneath the protective coating indicates that the coating had to have failed at this location such that water reached the pipe steel and established a corrosion cell. The morphology and location of the corrosion products are consistent with a CUI mechanism. The compromised coating on Line 901 was not isolated to just the failure location. Evidence of wrinkles and cracks were observed within the PE tape along the length of the excavated pipeline during the incident investigation. The wrinkles were concentrated at the bottom of the pipe along the 4:00 and 7:00 o’clock orientations [Ref ‎1], while the cracks were primarily along the 12:00 and 6:00 o’clock orientations. ‎Figure 19 is a photograph showing evidence of wrinkles within the PE tape layer, away from the failure location. Similarly, evidence of saturation/compression of the PU foam and thick deposits beneath the disbonded CTU coating were concentrated along the bottom of the pipe at Priority Dig 1 in 2015; see ‎Figure 20. These findings indicate that the environment along the bottom of the pipe is likely more corrosive than the environment along the top of the pipe. This conclusion is supported by the results of the 2007, 2012, and 2015 ILI runs, which show a higher distribution of external corrosion anomalies between the 3:00 and 9:00 o’clock orientations of the pipe; see ‎Figure 21. Thus, the protective external coating did not provide an effective barrier against the initiation and subsequent propagation of external corrosion. Repairs and excavations performed on Line 901 since 2007 have utilized a two part epoxy to recoat the pipeline. The recoat did not include the application of the PU foam insulation. ‎Figure 22 contains photographs showing two examples of recoats performed after representative 2007 and 2012 ILI digs [Ref ‎147 & ‎169]. The use of an epoxy protective coating with no PU foam helps to minimize the possibility of CP shielding in these areas. These steps increase the chance that CP can assist with mitigating external corrosion in areas where the two part epoxy coating is compromised. Probable contributing factors to the failure of the CTU protective coating are considered to be: DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 14#

Page 321Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis  Temperature of operation  Shearing stresses on the protective coating (insulation compression / land movement)  Design (outer coverings)  Wet / dry cycling The temperature of the product during operation of Line 901 averaged approximately 135 °F, based on data provided for a year prior to the failure [Ref ‎226]. Under typical operating conditions, this temperature was generally maintained between May 2014 and May 2015. During this time period, the temperature had a range between approximately 50 °F – 145 °F; see ‎Figure 23. The lower temperature excursions appear to be isolated events. Only two low temperature excursions were noted over the time period for which data were provided. One of the excursions corresponded to the time of the 2015 ILI run. In contrast, the high temperature excursions were a bit more frequent but shorter in duration. Hickey et al. [Ref ‎306] showed that, at these higher temperatures (i.e. ~ 150 °F), and when exposed to a chloride environment, CTU coatings exhibited poor cathodic disbondment properties. Thus, the operating temperature may have influenced the adhesion of the CTU coating to the pipeline steel. In addition to the effect that the operating temperature may have played on the CTU coating, the temperature may have also promoted CUI. Corrosion under insulation is a phenomenon that is well established in above ground piping facilities, like oil refineries and chemical process plants [Refs ‎301 and ‎310] and is known for underground pipelines [Ref ‎311]. CUI is identified as a concern in above-ground piping systems operating in a temperature range of 32 °F to 212 °F. The operating temperature of Line 901 falls within this range. Given the geometry of the CTU coating, PU insulation, and PE tape layer; the primary cause of failure from the metallurgical analysis; and the operating temperature of the line; the environment is consistent with circumstances conducive to CUI in above ground facilities. Thus, temperature may have been a contributing factor to the CUI. In combination with the temperatures discussed above, shearing stresses acting on the protective CTU coating likely contributed to the failure. In order for the corrosion to occur, the protective CTU coating had to disbond from the steel surface. Once the coating disbonded, the steel pipe was exposed to an electrolyte and corrosion could occur. Evidence of shearing due to soil stresses was observed along the pipeline, as evidenced by the presence of wrinkles and folds within the PE tape and compression of the PU foam. Based on the strong bond between the CTU coating and the PU foam, any soil stresses acting on the PE tape and PU foam were likely transferred to the CTU coating. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 15#

Page 322Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis The design of the insulating layers on Line 901 also contributed to the failure of the CTU coating. PE tape may have been selected to prevent water ingress to the PU foam and / or protection of the PU foam; however, tape is known to exhibit integrity issues in buried systems (i.e. wrinkling and poor corrosion control capabilities). Thus, when the PE tape was compromised, water was able to reach and saturate the PU foam. This water reached the CTU coating, which was absorbed by the PU foam. Wet / dry cycling is another probable contributor to the failure. Historical moisture data for the pipe joints at and adjacent to the May 19, 2015 failure location were reviewed due to: (1) the findings of the soil analyses in the metallurgical report and supplemental analyses (i.e. higher corrosive properties for saturated soils), (2) the presence of saturated PU foam adjacent to the failure location, and (3) the findings from the metallurgical report that indicate that the CUI was facilitated by wet/dry cycling. The data reviewed include the soil conditions reported in 2007 ILI Dig #5 and 6 [Ref ‎147 & ‎148] and reported in 2012 ILI Digs #12 and 13 [‎168 & ‎169]. The pipe joints excavated during these digs included Pipe Joints 5910 - 5950.12 These data were compared to historical average monthly precipitation reports for Santa Barbara, California and are presented in ‎Table 3. Both moist and dry soil conditions were encountered during the digs. The soils were found to be moist in February and March and dry in May. These findings correlate to the historic monthly rainfall patterns for Santa Barbara County, CA. The only pipe joint that was excavated during both a historically wet and dry month was Pipe Joint 5920. This pipe joint is directly adjacent to Pipe Joint 5930, which contained the failure location, on the U/S side. The fact that the soil adjacent to the pipe joint that failed exhibited wet-dry cycling indicates that wet-dry cycling likely occurred within the soil at the failure location and thus contributed to the failure. In addition, the location of Pipe Joint 5930 along Line 901 has the potential for extended periods of exposure to moisture as it falls within a low point along the line; see ‎Figure 2. Based on the metallurgical analysis, the protective coal tar urethane (CTU) coating, thermal polyurethane (PU) foam insulation, and polyethylene (PE) tape were compromised at the failure location. The damage included wrinkles, cracks, staining, and decohesion of the PE tape; staining, water saturation, and compression of the PU foam; and disbondment of the CTU. The damage to the external protective coating system allowed for water ingress, retention of water, and subsequent CUI. 6.1.2 Cathodic Protection System CP is intended to mitigate external corrosion at exposed coating holidays. OM412 indicates that all buried or submerged interstate hazardous liquid pipelines that are constructed, 12 The May 19, 2015 leak was associated with Pipe Joint 5930 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 16#

Page 323Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis relocated, replaced, or otherwise changed subsequent to March 1, 1970 must have CP installed [Ref ‎231]. OM412 also indicates that the CP system must be installed within one year after the pipeline is constructed, relocated, replaced, or otherwise changed. Both of these requirements were met for Line 901. OM412 indicates that all pipelines shall be electrically surveyed at least once each calendar year, but with intervals not exceeding 15 months, to determine whether the level of CP is adequate. The criteria for protection shall be a negative 0.850 volt with cathodic protection current applied. The pipe-to-soil potential shall be measured with reference to a copper- copper sulfate reference electrode (CSE) placed on the ground above the pipeline. Voltage (IR) drops other than those across the structure-to-electrolyte boundary shall be considered when evaluating the measured pipe-to-soil potentials. OM412 provides a second criterion for adequate CP, defined by a minimum of 100 millivolts of negative polarization voltage shift. The polarization voltage shift must be determined by interrupting the protective current (turning off all cathodic protection current sources, including those from any foreign system that may affect the pipeline pipe-to-soil potential) and measuring the polarization decay. The voltage reading after the immediate voltage shift occurs (when current is initially interrupted) shall be used as the base reading from which to measure the polarization decay. Section 6: Criteria and Other Considerations for Cathodic Protection of NACE International Standard Practice SP0169-2013 “Control of External Corrosion on Underground or Submerged Metallic Piping” [Ref ‎301], lists criteria for CP that indicate whether adequate CP of a metallic piping system has been achieved. The two criteria included in OM412 are included in SP0169, however, paragraph 6.2.1.4.2 indicates that at elevated temperatures (> 40 °C [104 °F]), the criteria listed in OM412 may not be sufficient, and also indicates that at temperatures greater than 60 °C (140 °F), the polarized potential of –0.950 volt CSE or more negative might be required. Experimental work performed by Jung-Gu and Yong- Wook [Ref ‎302] concluded that, for buried pipe under thermal insulation, adequate CP could not be obtained at -0.85 volt of polarization at temperatures greater than 25 °C [77 °F]. Paragraph 6.3.7 of SP0169-2013, indicates reliable measurement of potentials and therefore interpretation of CP criteria can be significantly affected by the presence of electrical shielding. Electrical shielding can be caused by disbonded coatings, thermal insulation, loose wrappers, high-resistivity rock or soils, metal structures or pipelines that are close to the structure being protected, and other man-made materials partially or completely surrounding the pipeline. The external coating system of L901 consists of a coal tar urethane coating on the steel substrate, 1.5-inch thick rigid polyurethane foam, and an external polyethylene tape. This type of coating systems has been reported to limit the DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 17#

Page 324Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis effectiveness of the cathodic protection in mitigating corrosion on areas where the electrolyte has reached the external surface of the steel pipe [Refs ‎302, ‎304, ‎305]. Pipe-to-soil potential data recorded at CP test stations located along L901 were provided between years 2005 and 2015 [Ref ‎46], for review and analysis. OM412 indicates all CP rectifiers shall be inspected at intervals not to exceed 2½ months, but at least 6 times each calendar year. The inspection shall include recording direct current (DC) output volts and amps, coarse and fine tap settings, and a visual inspection of rectifier components. Measurement of DC output volts and amps, and pipe-to-soil instant off potentials shall be completed as necessary to assure the rectifier is calibrated and adjusted properly. DC output volts and amperes, and taps settings of rectifiers Las Flores I, Las Flores II, Gaviota Station I, and Gaviota Station II, were provided for review and analysis between years 2005 and 2015 [Ref ‎47 and ‎49]. The analysis is discussed below. OM412 indicates a detailed potential survey, typically refer to as a close-interval potential survey (CIS) should be conducted where practicable and determined necessary by sound engineering practice, to accomplish the following objectives, established in paragraph 10.1.1.3 of NACE Standard SPO169-2007 [Ref ‎300]:  Assess the effectiveness of the CP system;  Provide base-line operating data;  Locate areas of inadequate protection levels;  Identify locations likely to be adversely affected by construction, stray currents, or other unusual environmental conditions; or  Select areas to be monitored periodically. CIS data recorded on L901 in years 2008 and 2015 were provided for review and analysis [Ref ‎33], [Ref ‎34]. External metal loss data from MFL ILI runs conducted in the years 2007, 2012 and 2015 were provided for analysis and review, [Ref ‎85], [Ref ‎126], and [Ref ‎139]. 6.1.2.1 External Corrosion Data Review and Analysis The purpose of the data review and analysis was to identify possible direct cause or causes that may have contributed to the failure that occurred on May 19th, 2015 in Goleta (Santa Barbara County), California at mile post (MP) 4, of pipeline L901. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 18#

Page 325Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis The first step when evaluating the performance of the external corrosion control system is to review the timeline of the data available. Based on the available data, some assumptions may be needed to establish operating conditions on the years where no data are available. The timeline of the data available to external corrosion is presented in ‎Table 4. As can be seen in ‎Table 4, the line started operation in 1990. Assuming that the external metal loss occurred at a constant rate since installation, a maximum wall loss of 0.318-inch reported at the leak [Ref ‎1], and 25 years of exposure (2015-1990), an average corrosion rate of 12.7 mils (1 mil = one thousandths of an inch) per year (mpy) is calculated. This corrosion rate value is consistent with the value provided in Appendix C3 of NACE International Standard SP0520-2010 [Ref ‎299] 12.2 mpy, which corresponds to the corrosion rate of a pipeline segment that had at least 40 mV of polarization (considering IR drop) for a significant fraction of the time since installation. This corrosion rate would not be expected on a pipeline segment with polarized annual pipe to soil potential values (IRF potentials stand for pipe-to-soil potentials free of IR error, also referred to as interrupted potentials) presented in ‎Figure 24. The IRF potentials recorded in the vicinity of the 2015 leak site meet both the criterion for adequate CP indicated in OM412 and the criterion suggested in NACE SP0502 for pipelines operating at temperatures higher than 60 °C (140 °F). However, data from only three years (recorded on one day of the specific year), of a pipeline that has been in operation for 25 years, may not be a good representation of the operational history of the external corrosion control system. Therefore additional data were aligned and analyzed. 2008 CIS data were aligned to 2007 ILI data, and 2015 CIS data were aligned to 2012 ILI, to check whether or not there was any correlation between external metal loss reported by the ILI runs and the pipe-to-soil potential profile along the pipeline route. The results are presented in ‎Figure 25 and ‎Figure 26, respectively. The interrupted pipe-to-soil potentials reported in 2008 and 2015 are more negative than -0.85 V CSE, and the 2015 interrupted potentials pipe-to-soil potentials are more negative than -0.95 V CSE along the entire length of L901. The locations where the 2008 CIS pipe-to-soil potential values were less negative than -0.95 V CSE (boxed in red rectangles in ‎Figure 25), don’t coincide with the locations where the deepest external metal loss were reported by the ILI tool. However, when ILI data are aligned and compared with CIS data, the validity (in time) of the CIS data needs to be checked. ILI data reports the cumulative metal loss that has occurred until the date of the inspection. CIS data report the pipe-to-soil potential values at the time of the survey and under the operating conditions of the CP system at the time of the survey. The CIS potential profile will only be valid on the days of the life of the pipeline in which the CP system was operating under the same conditions present at the time of the DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 19#

Page 326Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis survey (these conditions include the condition of the electrical insulators, foreign CP systems that affect the pipeline segment, rain fall, etc.). The operating conditions of the CP rectifiers Las Flores I, Las Flores II, Gaviota Station I, and Gaviota Station II, that provide CP current to L901 were plotted and analyzed between years 2005 and 2015. The results are presented in ‎Figure 27 – ‎Figure 30, respectively. To facilitate the analysis of the operating condition of the CP system, the yearly average of the DC current output was calculated for each year and for each rectifier. The total DC current outputs were plotted between years 2005 and 2015 and the ILI run and CIS inspections years were included in the plot presented in ‎Figure 31. As can be seen, the 2015 CIS data do not represent the operating conditions at which the CP system operated between 2008 and 2015. If it did, it could account for the external metal loss that occurred between 2007 and 2015 and could show consistency when compared to 2007 and 2015 ILI results. The limited validity of the CIS and ILI alignment is also evident in ‎Figure 32. The annual pipe- to-soil potential data recorded prior to the 2015 CIS show less polarization than the one recorded during the CIS. Despite the limitation of the CIS data, neither the annual test point data, nor the operating conditions of the rectifiers are consistent with the external metal loss reported by the ILI inspections. This inconsistency between the CP level and the external metal loss is likely a result of the electrical shielding produced by the coating system. The cathodic protection current cannot reach (or marginally reaches) the steel surface exposed to trapped electrolyte and the sensitivity of the electrical surveys used to monitor the condition of the buried pipe is significantly limited and not reliable. Probable contributing factors to the ineffectiveness of the CP system were considered and include:  Design of pipeline (insulation layers)  High resistive nature of the soil With the existing coating system, external corrosion will occur on the pipe surface at locations where the external polyethylene jacket allows the ingress of moisture, probably at field joints or areas where the topography of the right-of-way made it difficult to install the pipe. Areas where this moisture is trapped, together with seasonal changes that promote dry / humid cycles, may accelerate the degradation mechanism. This premise is validated by the preference of external metal loss on the bottom of the pipe where moisture will tend to accumulate due to gravity. ‎Figure 21 shows the distribution of the external metal loss anomalies around the circumference of the pipe. In 2015, more than 71% of the external DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 20#

Page 327Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis metal loss anomalies reported by the ILI were between the 3 and 9 o’clock position, i.e. the bottom of the pipe. Resistivity measurements were taken on soil samples removed near the failure location and at the four priority dig sites; see Appendix C. The resisitivities of the unsaturated (i.e. as- received) samples ranged from 2,500 – 78,000 Ohm-cm. Three of the five samples tested exhibited unsaturated resisitivities that were greater than 14,000 Ohm-cm, which is highly resistive. High resistivity soils can be detrimental to the effectiveness of the CP system. In summary, Plains met the regulatory requirements for monitoring cathodic protection system on Line 901, and based on the data provided, met the required levels for protection. However, the presence of polyurethane insulation and a polyethylene wrap shielded cathodic protection and the measured potentials are not representative of the electrochemical potentials at the areas of CUI. 6.2 Integrity Program Within its Integrity Management Plan (IMP), Plains implements a process of assessment and evaluation to maintain pipeline integrity. This investigation focused on the provided procedures to conduct a risk analysis and assess the integrity of the pipeline, including those used following the acceptance of the final ILI report related to assessments of internal and external corrosion. 6.2.1 Summary of Processes / Procedures Pertaining to Risk Assessments Plains utilize a relative risk indexing system (algorithm), which is described in “Risk Assessment Procedures” (Section 3 of the IMP). Nine likelihood of failure (LOF) types are identified and are consistent with general industry practices: external corrosion, internal corrosion, third party, equipment, construction, manufacturing, incorrect operations, weather and outside forces, and stress corrosion cracking. The description of the algorithm, including the weighting of each LOF type and the scoring mechanism for each variable category, is in Appendix D1 of IMP Section. For this investigation, external corrosion is the LOF type of interest. Plains identified this failure type in the relative risk model and it makes up 27% of the total likelihood score. This failure type has the highest weighting of all nine failure types identified. Plains provided DNV GL with their scoring mechanism for the failure type of external corrosion, which considered factors such as soil type, soil condition, asset age, coating type, the presence of insulation, and CP type [Ref ‎13]. For Line 901, the external corrosion risk “contribution” to the LOF algorithm remained relatively consistent from 2009-2014 (i.e. ranging between 0.94 and 1.16 according to Ref ‎14). DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 21#

Page 328Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Recommended Practice (RP) API 1160 provides guidance on managing system integrity to pipeline operators that transport hazardous liquids. Within the RP, a list of threats for underground pipelines is provided. All nine of the failure types (i.e. threats) identified by Plains are included in the practice. The specific threat of CUI is not addressed for underground pipelines in API 1160 or in Title 49, Part 195 of the Code of Federal Regulations for Hazardous Liquids [Ref ‎316]. Plains did identify external corrosion issues associated with field coatings (shrink sleeves) based on experience [Ref ‎90 and ‎131] and actions were taken to address this specific threat through the use of more stringent dig criteria and a shorter reassessment interval. Other than the accelerated re-inspection interval implemented on the line, Plains the mitigative actions taken by Plains on Line 901 did not adequately address the elevated integrity threat of CUI 6.2.2 Summary of Processes / Procedures Pertaining to ILI Assessments Portions of Sections 6, 8, and 9 of the IMP specify the procedures and guidance for “Conducting Assessments and Processing Results,” “Pipeline Repair Requirements,” and “Continual Assessment and Evaluation of Pipeline Integrity,” respectively [Ref ‎20, ‎21, and ‎22]. Section 11 of the IMP contains the procedure used for the “Identification of Preventive and Mitigative Measures” [Ref ‎23]. The relevant portions of each section are summarized below. 6.2.2.1 Conducting Assessments and Processing Results Section 6.3 “Review of New ILI Results – Repair Determinations and Schedules” includes the process used to evaluate ILI results and identify detected anomalies that require further evaluation and/or remediation. Two of the eight sub-sections are applicable to this review:  Tool Tolerance and Anomaly Classification: Specifies that the reported depths of “all significant corrosion anomalies” from the final ILI report are increased by a tool tolerance of 10% wall thickness. Anomalies are classified by comparing the Modified B31G burst pressure and Safe Operating Pressure to the MOP of the pipeline. The Corrosion Growth Analysis Report (CGAR)13 is used to calculate the estimated corrosion growth as part of the repair list generation.  Classification of Corrosion and Deformation Anomalies – Generate Initial Repair Lists: Specifies how corrosion and deformation anomalies are separated into Immediate, 60-Day, 180-Day, and other condition anomalies and the timeframes these conditions must be evaluated. 13 The CGAR process will be described in more detail later. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 22#

Page 329Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Section 6.4 “Data Integration of Pipeline ILI Results and Risk-Factor Data – Finalize Repair Scope and Schedule” describes the “procedures [to] be used to integrate other pipeline system information to finalize and supplement [§195.452(h)(4)-based] repair lists and set the repair schedule priorities.” Four of the five subsections are applicable to this review:  Manual Process for Data Integration: Integration of geographic information systems (GIS), current and previous ILI, previous repairs, cathodic protection data and estimated remaining lives are used to determine the final repair locations and schedule.  ILI Results Evaluation based on Data Integration: The compiled and integrated data are reviewed to identify subsequent actions. Results are documented on the “PHMSA Compliance Report”. 14  Repair Decisions based on Data Integration: Identification of additional repairs or evaluations, which may add additional repairs or exploratory digs to the repair schedule. When digs are performed in Santa Barbara County as part of the IMP, a grading plan for the dig has to be submitted to the County of Santa Barbara Planning and Development – Building and Safety Division. Specific requirements for grading on the dig are provided within the Santa Barbara County, California – Code of Ordinances in Chapter 14 [Ref ‎298].  Validation of ILI Results: Comparison of ILI-reported anomaly data and field- measured data, which is subject to analysis such as, plotting unity graphs and performing statistical analysis. 6.2.2.2 Pipeline Repair Requirements Section 8.3 “Repair Categorization” provides the definitions of repair categories (e.g., Immediate Condition) from §195.452(h)(4). These category definitions are also contained in the process schematic in Section 6.3 “Review of New ILI Results – Repair Determinations and Schedules”. 6.2.2.3 Continual Assessment and Evaluation of Pipeline Integrity The evaluation to determine a re-assessment interval for internal and external corrosion is presented in Section 9.2.2 “Procedures for Evaluating External and Internal Corrosion”. The external and internal corrosion procedures are intended to determine “the hypothetical time to failure (including safety factors) from internal and external corrosion growth and 14 The PHMSA Compliance report is also referred to as the “DOT Compliance Report” in the documentation provided to DNV GL DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 23#

Page 330Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis calculates an appropriate Re-Assessment interval to detect corrosion anomalies prior to the point at which the anomaly could potentially cause an operations failure.” Plains developed an Excel®-based program that performs the calculations described in Section 9.2.2 called “Corrosion Growth Analysis Report.” The CGAR program is also used during the procedures in Sections 6.3 and 6.4 to generate the final repair and evaluation schedule. As stated in Section 9.2.2, the procedure to determine a re-assessment interval for internal and external corrosion involves estimating the:  “Initial Corrosion Anomaly Size. The largest potential corrosion anomalies that could remain after the last assessment repairs were made are determined from ILI or hydrotest data.”  “Corrosion Growth Rates. External and internal corrosion growth rates are estimated from multiple ILI runs, field observations, and observed historical growth rates.”  “Time to Grow Corrosion Anomaly to Repair Condition. The time required to grow the initial corrosion anomaly size to failure is determined. The Reassessment interval based on corrosion growth is set at 70% of the predicted time to failure at the normal operating hoop stress of the system.” The recommended re-assessment interval is recorded on Form F11-2, Part A per Section 9.2.5 “Determination of the Re-Assessment Interval.” Changes to the re-assessment schedule are documented on the revision log for the assessment schedule per Section 9.3 “Revisions to Re-Assessment Schedule”. Periodic evaluations to assess overall pipeline integrity are required by §195.452(j)(2) and the procedural requirements for these evaluations are specified in Section 9.5 “Continual Evaluation and Assessment of Pipeline Integrity.” Evaluations occur at the midpoint between the last Preventative & Mitigative (P&M) evaluation and next scheduled assessment, after multiple leaks or failures by the same cause, following a “significant increase in risk analysis score” of a pipeline section and a “significant change in operations” of the pipeline section. The evaluations are documented on Form F9-1. 6.2.2.4 Identification of Preventive and Mitigative Measures Preventive and Mitigative Evaluation Meetings are defined in Section 11 of the IMP “Identification of Preventive and Mitigative Measures.” Section 11.3 specifies that “Division P&M Evaluation Teams meet yearly” and “P&M evaluations of assessments will occur within DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 24#

Page 331Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 15 months of the receipt of the final reports [to allow] time for reviewing the assessment results and investigating the worst anomalies to develop confidence in the validity of the assessment and to understand the pipeline segment’s condition.” 6.2.3 Available In-Line Inspection Data Plains provided results and documentation related to ILI assessments performed in 2007, 2012, and 2015. All three assessments were performed by Rosen using high resolution axial magnetic flux leakage (MFL) ILI tools. The 2015 MFL run was completed on May 6, 2015, approximately 13 days prior to the failure; however, the ILI data were still being analyzed by Rosen at the time of the failure. Plains received the preliminary ILI report on May 22, 2015 and the final ILI report on May 31, 2015. Although documentation was available for all three assessments, this review focused primarily on the information and analysis performed using the 2012 ILI as it pertains to processing the ILI results for excavations and determining an appropriate reassessment interval. The 2007 ILI is included in that process. The analysis is also supplemented with information from the 2015 ILI as appropriate. 6.2.4 Summary of Events Following the 2012 and 2015 In-Line Inspection Final Reports The timeline of events related to and following the receipt of the 2012 ILI final report on September 24, 2012 (including the 2015 ILI) is shown in ‎Figure 33. The CGAR analysis process began around September 26, 2012.15 Excavations were completed between October 18, 2012 and October 3, 2013. The DOT Compliance Report [Ref ‎124] was completed July 10, 2013. The Assessment Schedule [Ref ‎127] dated December 31, 2012 specified a three year reassessment interval for Line 901. As required in Section 9.3, the Assessment Plan revision log was updated. Form F11-2 [Ref ‎131], required as part of Section 9.2.5, for the 2012 ILI was completed on May 21, 2015. PHMSA conducted an inspection of procedures and records pertaining to Line 901 between August 19 and October 4, 2013 and provided Plains with the results of their inspection on September 11, 2015 [Ref ‎7]. On March 26, 2014, the highest pressure recorded at the Las Flores station between the 2012 ILI and May 18, 2015 (the day before the failure) was 888 psig. In April, a CIS and an aerial patrol were completed on the 9th and 28th, respectively. Between May 29, 2015 and June 3, 2015 four excavations were performed by Plains based on the 2015 ILI data. [Ref ‎200] 15 Plains provided an intermediate CGAR analysis file [Ref ‎128] dated September 26, 2012 indicating that the CGAR process began around this timeframe. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 25#

Page 332Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 6.2.5 Description and Review of the 2012 ILI CGAR Analysis as Applied to Joint 5930 The CGAR analysis is used within multiple steps during the analysis of ILI data and the reassessment interval determination. The initial and final repair lists are based on the results of this analysis and the calculations form the basis for the reassessment interval. In the 2012 CGAR analysis [Ref ‎128], the initial flaw size for all reported ILI features were increased by the ILI tool tolerances (equal to 10% of the nominal wall thickness (WT) for depth and 0.472-in (12 mm) for length). This is consistent with IMP Section 6.3 for depth; the addition of the length tolerance exceeds the requirements in Section 6.3. The MOP used was 1140 psig. (b) (4) This is consistent with the equation presented in Figure 9-2 of Section 9.2.2 for depth; the estimate of corrosion growth for length exceeds the requirements in Section 9.2.2. The CGAR analysis file [Ref ‎128] calculated the estimated dates features reach a depth of 80% WT, a modified B31G burst pressure less than MOP and the estimated reassessment date [Ref ‎312]. The estimated time to reach 80% of the WT is used as part of the requirements in Sections 6.3 and 6.4. The estimated reassessment date calculated by the CGAR analysis file [Ref ‎128] is consistent with Section 9.2.2 and is 70% of the estimated time for the features to reach a modified B31G burst pressure less than MOP. Section 9.2.2 states, “External and internal corrosion growth rates are estimated from multiple ILI runs, field observations, and observed historical growth rates.” An excerpt of IMP Figure 9-2 is presented in ‎Figure 34, which describes the requirements for estimating corrosion growth rates. For the case when multiple ILI runs that “allow depth comparisons of the same corrosion anomalies” are available, the procedure (see ‎Figure 34) specifies a corrosion growth rate in mils per year using the increase in corrosion depth during the time between ILI runs. It is DNV GL’s interpretation that, as presented in ‎Figure 34, the corrosion growth rate calculated using multiple ILI runs is then compared with the rate calculated using the CGAR analysis. The larger of the two values is intended to be used in the remainder of the CGAR analysis. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 26#

Page 333Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis ‎The 2012 CGAR analysis file [Ref ‎128] provides a column to enter the reported metal loss depths from previous assessments; this column is not referenced by existing equations or embedded macros. On Aug 20, 2015, Plains confirmed16 that the previous analysis results are not incorporated into the CGAR program calculations; instead the difference in depth is reviewed by the Integrity Specialist while finalizing the repair list. Therefore, the process followed by Plains to incorporate previous ILI results compared differences in reported depths, but did not directly calculate rates in mpy to compare them to the automated CGAR calculations. Evidence was not provided to indicate that the process was in strict adherence with the requirements of IMP Section 9.2.2. DNV GL performed a comparison of metal loss features reported in the 2007, 2012, and 2015 ILI runs on Joint 593017. The distance to the upstream girth weld, orientation, length, and width were compared. A graphical representation of this alignment is shown in ‎Figure 35. Blue and green boxes represent the locations of the metal loss features reported in the 2007 ILI and 2012 ILI, respectively. Black boxes represent the location of metal loss features in the 2015 ILI. The odometer location is presented in terms of the 2015 ILI to provide consistency with the Metallurgical Report [Ref ‎1]. Features identified within the Metallurgical Report [Ref ‎1] in the vicinity are shown as red boxes and the laser-scan measured depths are provided. The ILI-estimated depths of the reported metal loss features that are greater than 20% WT, and were not identified to be under a repair, are also included in the figure. In general, the locations of the ILI-reported metal loss and features found through physical examination correlate well. The 2015 ILI depths are less than the laser-scan measured depths as can be seen in ‎Figure 35 and ‎Figure 36. ‎Figure 36 contains a graphical representation of the reported metal loss depths from the 2007 (blue diamonds), 2012 (green squares) and 2015 (orange triangles) ILI runs in the region near the failure location. The failure location and the area recoated as part of the 2012, “Dig 13,” are also shown. The maximum depth of ILI-reported features undersize the depth at the failure location (measured to be 89% WT) for the 2015 ILI data. Defect characterization (i.e., depth sizing) is affected by the geometry of the anomaly. For the defect that led to the release, the edges were particularly ‘sharp’ meaning the depth profile changed rapidly from shallow to deep. To evaluate the potential impact of sharpness, DNV GL reviewed “Magnetic Flux Leakage (MFL) Technology for Natural Gas Pipeline Inspection”, prepared by J. B. Nestleroth and T. A. Bubenik, Battelle, for The Gas Research Institute, February 1999. This report along with data taken during the same time period show that a sharp defect can produce less flux leakage than a gradual defect. 16 Teleconference with AZA and Plains on August 20, 2015. 17 The 2015 failure location. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 27#

Page 334Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis However, the effect is modest (for the defects studied in the report, the leakage field strength is reduced by up to about 20%). Thus, sharpness could explain some, but not all, of the discrepancy between the defect depth reported by the ILI tool and the actual depth. Depth sizing by the ILI vendor is influenced by the defect depth and width relative to the pipe wall thickness (deep and/or narrow defects are difficult to size), length-to-width ratio (large length-to-width ratios are difficult to size), proximity to adjacent anomalies (overlapping inspection signals can complicate the analyses), and other parameters (e.g. magnetic nature of corrosion products, magnetization, and tool velocity). In this case, the most significant factor is probably the depth of the flaw relative to the pipe wall thickness, as it is especially difficult to size defects over 70% to 80% of the wall thickness. From an MFL inspection perspective, the defect is not particularly narrow and its length-to-width ratio is modest. In addition, the defect is away from other defects whose signals could have complicated the analysis. Nonetheless, each of these factors could have contributed to the undersizing. Given the fact that the 2012 ILI reported depth is within 2% WT of the 2015 ILI reported depth (compare 45% to 47%, respectively) and expected corrosion growth rates would not result in growth from ~47% WT to 89% WT in 13 days (the difference from the survey to the failure), it is conceivable that the actual depth in 2012 was much closer to 89% WT. The maximum pressure recorded at Las Flores station is 888 psig on March 26, 2014. The failure opening was measured at 6.6 in. If a feature of this length is assumed to exist on that date, then the depth needed to reach a modified B31G failure pressure equal to 888 psig is above 80% WT, suggesting that the depth of this feature in 2014 could have been up to 80% WT. If a flaw with a length of 12 inches is assumed, then the depth corresponding to a modified B31G failure pressure of 888 psig is 79% WT18, also suggesting that the depth of this feature could have been close to 80% WT. ‎Table 5 contains a listing of the metal loss features reported in the 2012 ILI data on Joint 5930, the CGAR estimated growth rate per Equation (1), and the rate estimated by the single anomaly comparison method (see excerpt of IMP Figure 9-2 in ‎Figure 34). For five out of the ten 2012 ILI features, the single anomaly-based rate between the 2007 and 2012 ILI is less than the 2012 CGAR estimated rate. The feature that corresponds to the release location is highlighted in bold in ‎Table 5. The estimated single anomaly-based rate for the feature associated with the 2015 failure is over two times faster (in mils per year) than is estimated by the CGAR process. 18 24-in OD, 0.344-in WT, API Grade X65 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 28#

Page 335Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis ‎Table 6 compares the time to reach 80% WT19 for each feature in Joint 5930 reported by the 2012 ILI. The initial depths in ‎Table 6 from the 2012 ILI were increased by the specified tool tolerance to be consistent with IMP Section 6.3 and the 2012 CGAR analysis file. The minimum time to reach 80% WT for the feature corresponding to the failure and is greater than the three-year re-assessment interval specified for this line as documented in the 2012 Assessment Plan [Ref ‎127]. Based solely on this criterion19, the feature corresponding to the failure would not have been selected for excavation. Discussions of the other failure criterion, other rate calculations, and initial flaw sizes are given below. The process followed by Plains to incorporate previous ILI results compared differences in reported depths, but did not directly calculate rates in mpy and compare them to the automated CGAR calculations. There was no evidence provided that their process was in strict adherence with the requirements of IMP Section 9.2.2. The feature corresponding to the release location was estimated to reach 80% WT after the specified reassessment interval using the process followed by Plains. The same conclusion would have been reached had Plains used the single anomaly comparison rate. 6.2.6 Description and Review of 2012 DOT Compliance Report (2012 Final Repair List) The final repair list is documented in the DOT Compliance Report [Ref ‎124] per the requirements of IMP Section 9.4. The 88 features (70 are metal loss) across 41 dig sites selected for excavation and repair are summarized in ‎Table 7. ‎Table 7 contains the documented selection criteria for the inclusion of the features in the repair list. The documented selection criterion for 21 (30% of 70 targeted metal loss) were based on their depth (greater than or equal to 40% WT) and close proximity (less than or equal to 2.0 feet) to a girth weld. The total number of targeted metal loss features that were within 2.0 feet of a girth weld is 50 (71% of 70 targeted metal loss). Only one feature in the 2012 ILI met the requirements for Immediate, 60-Day or 180-Day conditions in §195.452(h)(4)(i)-(iii). This feature, a top side dent, was included in the repair list and documented in the DOT Compliance report. The remaining selection criteria are Plains-specific criteria. 19 The estimated time to reach 80% of the WT is used as part of the requirements in Sections 6.3 and 6.4 to establish the final repair scope and schedule; features that are estimated to grow to 80% WT prior to the “due date” are selected for excavation and repair. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 29#

Page 336Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 6.2.7 Description and Review of the Validation of ILI Results Section 6.4 of the IMP, subsection “Validation of ILI Results” requires validation of the ILI results “by various methods, such as, plotting unity graphs and performing statistical analysis.” Fifteen field depth measurements were matched by Plains to 15 of the 2007 ILI-reported depths in a file [Ref ‎87]. DNV GL performed least squares linear regression on the field-ILI data for the 15 data points, as shown in ‎Figure 37. ‎ ‎Figure 37 presents the ILI-reported depth on the x-axis and the field-measured depth on the y-axis. The unity line and 10% WT tolerances are indicated. The upper region of the plot is where the ILI undersized the depths. The slope of the least squares regression equation is 0.1508 ± 0.4044 (95% confidence) and the R2 20 value is 0.0475. The 95% confidence interval on the slope includes 0, indicating that there is not enough statistical evidence at 95% confidence to support a relationship between the 2007 ILI-reported depth and the actual field-measured depth. Plains provided 52 field depth measurements matched to 52 of the 2012 ILI-reported depths in a file [Ref ‎129].21 DNV GL performed least squares linear regression on the field- ILI data for the 52 data points, as shown in ‎Figure 38. The slope of the least squares regression equation is 0.4988 ± 0.3113 (95% confidence) and the R2 value is 0.1716. The 95% confidence interval on the slope does not include 0, indicating that there is a relationship between the ILI-reported depth and the actual field-measured depth. ‎Figure 38 shows that the distribution of metal loss features more than 2.0 feet from a girth weld and those near a girth weld may be different. Those features near a girth weld exhibit depths both under and over the ILI-reported depths; whereas, those greater than 2.0 feet from a girth weld tend to be undersized by the 2012 ILI (none are over reported). The largest difference between the ILI-reported depth and the field-measured depth, when the ILI under-reports the field depth, is 24% WT. This difference is for a feature that was not within 2.0 feet of a girth weld. ‎Figure 38 suggests that some metal loss features away from the girth weld, like the feature associated with the 2015 failure, were under-reported by the 2012 ILI. Six field depth measurements were matched by Plains to six of the 2015 ILI-reported depths [Ref ‎‎200]. An additional five measurements, obtained using laser scanning, were matched to five 2015 ILI-reported depths in the Metallurgical Failure Report [Table 1 of 20 The R2 value ranges from 0.0 to 1.0 and measures how close the data are to the fitted regression line. The higher the R2 value, the better the linear model fits the data. 21 [Ref ‎129] is from 2015; the data are consistent with a unity plot generated by Plains in 2013 [Ref ‎130]. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 30#

Page 337Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Ref ‎1]. Twenty-two field measurements taken in 2013 and under recoat or composite sleeves, in response to the 2012 ILI, were matched to 22 of the 2015 ILI-reported depths by DNV GL by comparing the 2012 and 2015 ILI feature listings. The unity plot of the 33 total field to 2015 ILI correlations is shown in ‎Figure 39. DNV GL performed least squares linear regression on the field-ILI data for the 33 data points, as shown in ‎Figure 39. The green squares denote the field measurements correlated by comparing the 2012 and 2015 ILI, the purple diamonds are the laser scan measurements and the blue triangles are the measurements reported in the 45 Day CAO report. Features within two feet of a GW are indicated by purple circles. The slope of the least squares regression equation is 0.4815 ± 0.1999 (95% confidence) and the R2 value is 0.4402. The 95% confidence interval on the slope does not include 0, indicating that there is a relationship between the 2015 ILI-reported depth and the actual field-measured depth. Features matched by comparing the 2012 and 2015 ILI are both over and undersized. The tendency to over or undersize features does appear to be influenced by the measurement technique; features measured in the field are all over called and features measured in a laboratory using laser scanning are undersized. Comparing the 2012 and 2015 field-ILI unity plots demonstrates that the slope of the least squares regression equation is similar (close to, but below 0.5) for both with the 2015 ILI exhibiting less variability around the regression line (the R2 value is larger and the 95% interval on the slope has a smaller range). The intercepts are also similar (close to 25). The similarities in the least square regression equations for the 2012 and 2015 unity plots suggest that the mean (expected) field depth for a given ILI-reported depth in either 2012 or 2015 would be similar, but that the 2015 would have a smaller standard deviation around the mean. Although not a regulatory requirement, API Standard 1163 (API 1163) [Ref ‎309] provides guidelines for the qualification of in-line inspection systems used in gas and hazardous liquid in-line inspection system pipelines. In Appendix E of API 1163 the overall number of verification measurements, N, versus the number of verification measurements within tolerance, Nin, is used to establish consistency with performance specifications. ‎Figure 40 is an excerpt of API 1163 Appendix E containing a table that can be used to establish consistency with performance specifications. ‎Figure 40 was calculated assuming a tool performance specification of depths sized within a given tolerance with 80% certainty and a 95% confidence level. According to API 1163 (see ‎Figure 40, Ref ‎309) there must be at least 37 features within the specified tolerance with a sample size of 52 total features to establish consistency with DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 31#

Page 338Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis the stated performance specification. In ‎Figure 38, 29 measurements are within the tool tolerance used throughout the CGAR analysis (e.g., ±10% WT). Based on API 1163, there is not enough evidence to support that the 2012 ILI met the stated performance specification. Section 6.4 of the IMP requires validation of the ILI results “by various methods, such as, plotting unity graphs and performing statistical analysis,” but does not directly specify the requirements when the ILI does not meet the performance specifications. The only direct reference in the documents provided to DNV GL relating to discrepancies between the ILI and field measurements is in the process flowchart in Section 6.2 of the IMP (excerpt in ‎Figure 41). The text in this flow chart states that if there are “Large discrepancies between pig calls and actual size of dents, metal loss or crack like anomalies,” then the “Integrity Specialist initiates ILI tool vendor re-grading of raw tool data.” There is no guidance as to what constitutes a large discrepancy. There is no information or documentation indicating that Plains initiated a regrade of either the 2007 or 2012 inspection data. In order to evaluate the potential for other means to respond when the ILI tool does not meet the performance specifications, DNV GL redefined the assumed tool tolerance using API 1163. The intent was to determine a revised tolerance that would provide a similar confidence as the vendor-stated tolerance that is included in the requirements of IMP Section 6.3. For the 2012 ILI, a redefined tolerance of ±16% WT is needed to be consistent with API 1163 (i.e., 37 of 52 within tolerance per ‎Figure 40). The redefined tolerance is greater than the tolerance used by Plains in the CGAR analysis performed subsequent to the 2012 ILI. In Section 6.4 of the IMP, the data integration process is used to identify results requiring subsequent actions that “may include regrading the ILI anomaly tally; exploratory digs and repairs beyond those required for §195.452(h)(4)(i, ii & iii); special bellhole inspections (e.g. mag particle testing); and, similar efforts to resolve questions raised by the data integration analysis.” It is DNV GL’s opinion that the excavation results conducted as part of the ILI validation should be included in the data integration process. 6.2.8 Description and Review of Re-Assessment Interval Determination Plains based the re-assessment interval on the estimated time for the predicted burst pressure of any given feature to be less than MOP, specifically 70% of that time (refer to Section 9.2.2 of the IMP). In the Assessment Plan from 2012 [Ref ‎127] dated December 31, 2012, the reassessment interval is specified as three years; the “Change Inspection Interval” section states “Reduce DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 32#

Page 339Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis L901 Las Flores to Gaviota 24", L903 Gaviota to Sisquoc 30", L903 Sisquoc to Pentland 30" and L903 Pentland to Emidio 30" from 5 years to 3 years.” The date the decision was made to change the inspection interval and the reason for the change were not included in documents provided to DNV GL. DNV GL performed calculations, using the CGAR process, to evaluate whether the three year inspection interval was justified for the feature associated with the 2015 failure. ‎Table 8 compares the reassessment interval based on the tolerance and rate scenarios discussed previously. Specifically, the initial flaw size is based on a 10% and a 16% tolerance and the rates are based on either the CGAR methodology or the single anomaly comparison. The time to reach the 80% WT and modified B31G burst pressure (PFail) is calculated. For all cases, the feature associated with the 2015 failure is predicted to reach 80% WT before PFail ≤ MOP. Section 9.2.2 of the IMP defines the reassessment interval as “70% of the predicted time to failure at the normal operating hoop stress of the system.” It is unclear from the procedures provided by Plains how to handle these cases as modified B31G is not applicable for depths greater than 80% WT [Ref ‎315]. For the calculations in ‎Table 8, DNV GL has assumed that the reassessment interval is taken as 70% of the time to reach either 80% WT or PFail ≤ MOP. Based on the CGAR process, the estimated reassessment interval using a ±10% and ±16% tool tolerance are 7.3 and 5.5 years, respectively. Based on feature to feature matching, the estimated reassessment interval using a ±10% and ±16% tool tolerance are 3.4 and 2.5 years, respectively. While the most conservative reassessment interval of 2.5 years is less than the three year reassessment interval specified by Plains; the actual reassessment interval was 2.8 years22 and is similar when accounting for operational and logistical requirements for ILI. DNV GL applied the 2012 CGAR process using a 16% tool tolerance to all remaining unrepaired features from the 2012 ILI. The minimum predicted failure pressure for unrepaired features using the 2012 CGAR process after five years is 1452 psig, which is greater than the MOP used by Plains. If the 70% time frame per Section 9.2.2 is applied, then all features should have a predicted failure pressure above the MOP for at least 4.2 years to justify a three year assessment interval23. In addition, the minimum time to reach 80% WT for unrepaired features is 6.01 years (70% is 4.3 years). Therefore, the CGAR process as applied by Plains to the 2012 ILI supports a three year assessment interval. If a 16% “tolerance” is incorporated24 instead of the 10% used in 2012, then the minimum predicted failure pressure for unrepaired features using the procedure in the 2012 CGAR 22 Using a survey date of July 3, 2012 and May 6, 2015 23 Three years is approximately 70% of 4.2 years (i.e., 3.0 / 0.7 = 4.2) 24 This is the “redefined tolerance” needed to meet the requirements in API 1163, see Section ‎6.2.7 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 33#

Page 340Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis analysis file after five years is 1352 psig, which is also greater than the MOP used by Plains. The minimum time to reach 80% WT for unrepaired features is 3.9 years for a reported feature from the 2012 ILI25 (this corresponds to a 2.7 year inspection interval). The CGAR process using the larger tolerance resulted in a re-inspection that is of the same order that Plains used to initiate the 2015 ILI run. Thus, reevaluating the 2012 ILI data in this manner may not have prevented the failure. Multiple methods exist to compare ILI data and estimate corrosion growth rates. DNV GL performed an additional analysis comparing the 2007 and 2012 ILI data that is neither required in Plains’ IMP nor in the CFR. The analysis is termed statistically active corrosion (SAC). DNV GL developed the SAC methodology with the objective to identify pipeline locations for which ILI data indicates a likelihood of corrosion growth and predict corrosion rates. For selected joints with the potential for significant growth, a manual review of the ILI signal data was performed to determine whether the likely growth is evident in the ILI signal or a result of ILI sensitivity differences. Based on the results of the corrosion growth screening and probabilistic assessment, DNV GL manually reviewed 169 pipe joints and identified evidence of growth in 82 (49% of the total reviewed). As a result of the statistical analysis and manual review, DNV GL determined that the joint that failed in 2015 (Joint 5930) showed evidence of significant change in the signal data and is predicted to have a SAC growth rate (15 mpy), which is between the rate used in the CGAR process (8 mpy) and the rate obtained via pit-to-pit matching (18 mpy). With the SAC rate, the feature that led to the 2015 failure is estimated to reach 80% WT in 5.8 years (70% of that time is 4.0 years). Appendix F contains a description of the SAC methodology as well as the compiled summaries of the manual signal review and estimated rates. One of the minimum P&M measures that must be considered within the Preventive and Mitigative Evaluation Meeting is the potential for establishing shorter inspection intervals (see IMP Section 11.4 [Ref ‎23]). While the assessment interval was shortened from five to three years [Ref ‎127] prior to December 31, 2012, there is no documentation (e.g., Form F11-2) provided to DNV GL specifying the assumptions or calculations that were used to justify the three-year assessment interval. No information was provided documenting a Preventive and Mitigative Evaluation Meeting within the 15 month window of the receipt of the final report required in Section 11.3 “Forming Division Preventive and Mitigative Evaluation Teams.” On May 21, 2015 (after the 2015 failure) Form F11-2 [Ref ‎131] was completed. This form references the 2012 ILI data (not the 2015 ILI) and: 25 A 53% WT, 0.75-in metal loss feature on Joint 14470 (odometer 51640.14). DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 34#

Page 341Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis  States that the current inspection interval is three years (Part A-1: Review of Design, Operation and Risk Data)  References the release on 5/19/2015 (Part A-3: Leaks from Segment or Facility)  Recommends a reduction in the reassessment interval from three to two years to “ensure the control of growth of external corrosion under shrink sleeves” (Part B). 6.2.9 Continual Evaluation and Assessment of Pipeline Integrity Documents pertaining to the periodic evaluation process required in Section 9.5, specifically Form F9-1, were requested but were not provided to DNV GL. On September 11, 2015 Plains stated26 that Form F11-2 is similar to Form F9-1 and is therefore used in the place of Form F9-1. There is no documentation provided to DNV GL that a periodic evaluation process meeting took place prior to the 2015 release, as it pertains to the 2012 ILI data. Plains has not demonstrated that the requirements in Section 9.5 have been met. 7.0 SUMMARY AND CONCLUSIONS The results of the metallurgical analysis indicated that the immediate metallurgical cause for the Line 901 failure was wall thinning from external corrosion that ultimately failed by ductile overload under the imposed operating pressure [Ref ‎1]. The flaw that failed was not through wall prior to ductile overload and, therefore, the failure event was sudden in nature. The morphology of the external corrosion was determined to be consistent with corrosion under insulation (CUI), facilitated by wet-dry cycling. The results of the root cause analysis presented below are based on the provided documentation referenced in Appendix B. DNV GL reserves the right to modify or supplement these conclusions should new information become available. DNV GL identified four c basic root causes of the failure: 1. The external coating system failed to prevent moisture from reaching the pipe steel, allowing the external corrosion process to occur. Basis: Based on the metallurgical analysis, the protective coal tar urethane coating, thermal polyurethane foam insulation, and polyethylene tape were compromised at the failure location. The damage included wrinkles, cracks, staining, and decohesion of DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 35#

Page 342Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis the polyethylene tape; staining, water saturation and retention, and compression of the polyurethane foam; and disbondment of the coal tar urethane. 2. The cathodic protection system was ineffective due to shielding by the thermal polyurethane insulation and external polyethylene wrap. Basis: Based on the provided documentation, Plains met the regulatory requirements for monitoring the cathodic protection (CP) system on Line 901, and the measured pipe to soil potential values met the required levels for protection. However, the presence of the polyurethane insulation and the polyethylene wrap shielded the cathodic protection current and prevented voltage monitoring of the shielded portions of the pipe. As a result, the CP current did not reach the pipe surface and the measured potentials did not represent the potentials at the areas of corrosion under the insulation. 3. The contracted in-line inspection significantly undersized the external corrosion feature that failed on Line 901. Basis: Based on the provided documentation, the 2015 MFL tool significantly undersized the external corrosion feature that ultimately leaked (i.e. a tool determined depth of 47% of the nominal wall thickness vs. a laboratory measured depth of 89% of the nominal wall thickness). The MFL tool likely also undersized the same feature in the 2012 ILI run based on a review and comparison of the 2007, 2012, and 2015 raw signal data for the feature that failed. 4. The mitigative actions taken by Plains on Line 901 did not adequately address the elevated integrity threat of corrosion under insulation. Basis: The results of the metallurgical analysis indicated that the immediate metallurgical cause of the failure was CUI. Corrosion under insulation is a unique corrosion mechanism that necessitates its own integrity risk assessment. Plains did not apply sufficient mitigative strategies specific to CUI to prevent this anomaly from failing. The measures could include enhancement of existing barriers and additional preventative barriers. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 36#

Page 343Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Additional observations for improvement in the Integrity Management Program The following provides perspective on Plains’ integrity management plan (IMP) as related to the failure on Line 901. Coating systems, as a barrier to external corrosion related integrity threats, (i) are never perfect and (ii) age over time, thereby decreasing the effectiveness of the barrier. The cathodic protection (CP) system is another barrier to external corrosion integrity threats. Cathodic protection can be effective for many external corrosion related integrity threats; e.g., corrosion at holidays (holes in the coating) and microbiological influenced corrosion (MIC). There are limits to the effectiveness of CP for corrosion related integrity threats and mitigation barriers can be strengthened and/or other mitigation barriers can be employed in conjunction with CP; e.g., stray current enhanced corrosion, AC induced corrosion, stress corrosion cracking, and corrosion beneath disbonded coatings that shield CP current. For these, multiple barriers may be used depending on the individual integrity threat, but the ILI program in conjunction with a dig program becomes a more important barrier since it is known that the other barriers of coating and CP are not always effective. In the case of Line 901, Plains targeted 70 metal loss features in 2012, which included 31 features beyond those required by code and used for validation of the ILI program. These additional digs constitute a strengthened barrier in the prevention of a pipe failure due to a corrosion related integrity threat. Several of the digs were based on the strengthening of the ILI/dig barrier for the purpose of identifying and repairing corrosion under shrink sleeves used at girth welds; a known corrosion related integrity threat involving coatings that shield CP. In addition, the ILI re-inspection interval was decreased from a minimum of 5 years to 3 years (performed at 2.8 years). This also is a strengthening of a barrier in the prevention of a pipe failure due to a corrosion related integrity threat. Plains IMP aggressively addressed several of the corrosion related integrity threats; but, as mentioned under contributing causes, Plains did not apply sufficient mitigative strategies to prevent the CUI anomaly from failing. In addition, an IMP is only as good as the data that are utilized to monitor and measure its performance. As addressed as a contributing cause, the ILI significantly undersized (47% versus an actual value of 89% through wall) the feature that eventually failed. The RCA identified improvements that could be made within the integrity management program, which were not direct causes of the failure. These observations are given below. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 37#

Page 344Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 1. Based on the information provided, Plains could adopt additional practices to identify and address any inaccuracies in future ILI runs. Basis: DNV GL performed an analysis of the 2012 ILI and dig data using API 1163, which is not a regulatory requirement or part of the IMP, and determined that the tool performance was not within the stated specifications. There was no produced documentation to indicate that Plains communicated with the ILI vendor, such as requesting a re-grade, following production of the unity plot[s] to account for the scatter observed within the data. However, using the recalculated tool tolerance would still result in a similar re-inspection interval as that used by Plains for the 2015 ILI run. 2. Based on the provided information, Plains could better incorporate the results from multiple ILI runs into their corrosion growth rate calculations. Basis: Plains IMP Section 9.2.2 states, “External and internal corrosion growth rates are estimated from multiple ILI runs, field observations, and observed historical growth rates.” The procedure specifies calculation of a corrosion growth rate in mils per year using the increase in corrosion depth during the time between consecutive ILI runs. There is no documentation provided to indicate that Plains performed such calculations using the historical ILI data. DNV GL calculated a corrosion growth rate for the feature that failed based on data from the 2007 and 2012 ILI runs. Although a higher corrosion rate was calculated than that determined using the CGAR process, this rate results in a similar re- inspection interval to that performed by Plains. Additional analyses that go beyond the IMP, codes, and standards, include:  Statistically active corrosion (SAC) analysis performed on Line 901 resulted in a similar re-inspection interval as that used by Plains (2.8 years) for the 2015 ILI run. The analysis identified a remaining life for the feature that failed that is greater than the re-inspection interval used by Plains. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 38#

Page 345Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 3. Based on the provided information, Plains should improve their documentation and/or record-keeping of their decision-making processes related to actions taken. Basis: Over the course of the investigation, DNV GL identified areas within the integrity management process that were not sufficiently documented. For example, no justification (i.e. assumptions, analyses, etc.) was provided for determining the reassessment interval of 3 years based on the 2012 ILI data. Although a form explicitly identifying the justification for the reduction of their re- inspection interval from 5 years to 3 years was not provided, DNV GL’s assessments and calculations resulted in a similar re-inspection interval as that used by Plains. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 39#

Page 346Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Table 1. Summary of inspection data from aerial patrols of Line Segment 901 between January 7, 2015 and May 11, 2015. Patrol Date Inspection Data 7-Jan-15 Segment OK 16-Jan-15 UF due to Weather1 21-Jan-15 Segment OK 28-Jan-15 Segment OK 4-Feb-15 Segment OK 9-Feb-15 Segment OK 19-Feb-15 Segment OK 25-Feb-15 Segment OK 3-Mar-15 Segment OK 13-Mar-15 Segment OK 25-Mar-15 Segment OK 1-Apr-15 UF due to Weather1 6-Apr-15 Segment OK 17-Apr-15 UF due to Weather1 20-Apr-15 Segment OK 29-Apr-15 Segment OK 4-May-15 Segment OK 11-May-15 Segment OK 1 – UF: Unable to fly. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 40#

Page 347Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Table 2. Assessments on potential external corrosion mechanisms for the failure. Corrosion Mechanism Relevant to Line 901 Failure Assessment AC Stray Current Corrosion No AC field measurements were negligible and there was no HVAC lines in the ROW. DC Stray Current Corrosion No The corrosion was not characterized by sharp edged pitting and the absence of corrosion products around the pitted area (i.e. which is typical of DC stray current corrosion.) Also there were no foreign line crossings or parallel lines located in the ROW. Galvanic Corrosion No The corrosion was not associated with the coupling of two dissimilar materials. The corrosion features were found across the length of the line and were not isolated/concentrated to areas of previous armor plate repairs. Microbiologically Influenced Corrosion May have contributed to corrosion, but not cause Bacteria were identified at a corrosion feature sampled U/S of the failure location; however they were not preferentially flourishing within the corroded areas. Furthermore, the levels of bacteria were low and the layered morphology within the corrosion products is not consistent with MIC. Corrosion Under Insulation Yes Based on the morphology (general corrosion mixed with pits) and location (beneath damaged coating combined with wet, thermal insulation) of the corrosion associated with the failure, the corrosion is due to CUI. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 41#

Page 348Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Table 3. Historical moisture conditions, based on 2007 and 2012 ILI dig information, for pipe joints near the 2015 failure location. Dig Data Pipe Joint ILI Year Dig Number1 Dig Date Soil Condition1 Soil Description1 Average Monthly Precipitation (inches)2 5910 2012 Dig 12 5/9/13 Dry Clay, Sand, Rock 0.31 2007 Dig 5 3/3/09 Moist Loam 2.91 5920 2012 Dig 12 5/9/13 Dry Clay, Sand, Rock 0.31 59303 2012 Dig 13 5/10/13 Dry Clay, Sand, Rock 0.31 5940 2012 Dig 1 5/10/13 Dry Clay, Sand, Rock 0.31 5950 2007 Dig 6 2/21/08 Moist Loam 4.57 1 – [Ref ‎50 & Ref ‎91] 2 – [Ref ‎314] 3 - Pipe Joint 5930 contained the May 19, 2015 failure location. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 42#

Page 349Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Table 4. Timeline of external corrosion control monitoring and inspection data. Date # Event Year Month Start End Comments 1 Year Pipe in Service 1990 2 Year CP was commissioned Information not received 3 Earliest Annual Test Point Data provided 2005 January 4 1st ILI Metal Loss Run 2007 June 1 – Report date: August 15, 2007 5 1st CIS 2008 December 5 8 6 2nd ILI Metal Loss Run 2012 July 3 – Report date: September 26, 2012 7 2nd CIS 2015 April 8 9 8 3rd ILI Metal Loss 2015 May 6 – Report date: June 4, 2012 9 Failure 2015 May 19 – DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 43#

Page 350Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Table 5. Comparison of Rate Estimation Methods between 2007 and 2012 ILI. 2007 ILI 2012 ILI Rate Odometer (ft) Maximum Depth (% WT) Odometer (ft) Maximum Depth (% WT) CGAR † (mpy) Single Anomaly Comparison ‡, (mpy) 21367.68 12 2.20 1.38 21370.50 14 2.57 2.75 21341.79 11 21371.23 13 2.39 1.38 21353.03 23 21382.40 24 4.40 0.69 21384.36 38 6.97 19.26 21384.48 11 2.02 0.69 21384.80 21 3.85 7.57 21355.45 19 21384.96 45 8.26 17.89 21385.38 13 2.39 2.06 21360.90 26 21390.33 41 7.52 10.32 † Assumes a construction year of 1987, run year of 2012, 0.344-in WT ‡ Assumes five years between inspections, no tolerance added to either 2007 or 2012 reported depths, 0.344-inch WT and a 2007 feature depth of 10% WT for 2012 features without a match in 2007. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 44#

Page 351Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Table 6. Comparison of estimated time to reach 80% WT for features on Joint 5930. 2012 ILI Initial Flaw † Rate ‡ (mpy) Time to Reach 80% WT (yrs) Odometer (ft) Maximum Depth (% WT) Depth (% WT) CGAR Single Anomaly Comparison CGAR Single Anomaly Comparison 21367.68 12 22 2.20 1.38 90.63 145.00 21370.5 14 24 2.57 2.75 75.00 70.00 21371.23 13 23 2.39 1.38 82.21 142.50 21382.4 24 34 4.40 0.69 35.94 230.00 21384.36 38 48 6.97 19.26 15.79 5.71 21384.48 11 21 2.02 0.69 100.57 295.00 21384.8 21 31 3.85 7.57 43.75 22.27 21384.96 45 55 8.26 17.89 10.42 4.81 21385.38 13 23 2.39 2.06 82.21 95.00 21390.33 41 51 7.52 10.32 13.26 9.67 † The reported ILI depths are increased by a tool tolerance per Section 6.3 of the IMP ‡ Refer to ‎Table 5 for rate estimates DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 45#

Page 352Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Table 7. Summary of features selected for excavation following the 2012 ILI. Targeted Features < 2 feet DOT Compliance Report Selection Criteria > 2 feet from a girth weld 180 Day : Dent >2% on TOP1 from a girth weld 1 – 1 Total Calc Growth ≥ 80% : High Priority – 2 2 Calc Growth ≥ 80% 1 7 8 Additional : ≤ 2 ft from GW & ≥ 40% ML – 21 21 Additional : GMA near GW1 – 2 2 Additional : ML Validation – Freq. in Joint 18 – 18 Additional : ML Validation 1 14 15 Additional : ML Validation : Not Previously Reported – 6 6 Additional: Noted as Possible Wrinkle 1 15 – 15 Total Metal Loss 20 50 70 Total 36 52 88 1 – Geometric features. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 46#

Page 353Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Table 8. Comparison of re-assessment intervals for the feature associated with the 2015 Failure. Case ID Initial Flaw Size 1 Method for Rate Calculation 2 Time to Criteria 3 Assessment Interval 4 ≥ 80% WT PFail ≤ MOP PFail ≤ MOP Case 1 2012 ILI + 10% CGAR 10.4 10.4 7.28 Case 2 2012 ILI + 10% Single Anomaly Match 4.8 4.8 3.36 Case 3 2012 ILI + 16% CGAR 7.9 7.9 5.53 Case 4 2012 ILI + 16% Single Anomaly Match 3.6 3.6 2.52 1 10% is tool tolerance used by Plains during the 2012 CGAR process; 16% is the "redefined tolerance" based on API 1163. 2 CGAR uses Equation (1); Single Anomaly Match rate is based on pit-to-pit matching. 3 The estimated time to reach indicated criteria. The feature is predicted to reach 80% WT in depth before PFail drops below MOP. 4 IMP Section 9.2.2 defines the reassessment interval as “70% of the predicted time to failure at the normal operating hoop stress of the system.” Given that the feature is predicted to reach 80% WT before PFail ≤ MOP, the reassessment interval is taken as 70% time to reach 80% WT. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 47#

Page 354Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. location Failure Pacific Ocean Highway 101 (a) View parallel to pipeline showing slope of hill near failure location Pacific Ocean Highway - 101 a) View perpendicular to pipeline showing slope of hill near failure location Figure 1. Photographs showing the topography in the vicinity of the failure location. December 4, 2015 DNV GL - OAPUS307KKRA (PP136049) 48#

Page 355Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis b) (7)(F) Gaviota Las Fiores Googlem 0 005 06 Refugio State Beach Line 901 Elevation Profile Elevation [ft] — 2015 Failure location 1000 Gaviota Station FLOW Las Flores Station 800 600 400 200 50000 40000 30000 20000 10000 Distance Downstream from Las Flores Station (feet) Figure 2. Topographical map and plot showing the elevation profile for Line 901. The white triangles on the map correspond to mile post markers. The green star on the map and the green line on the plot identify the location of the May 19, 2015 failure. ONV GL - 49#

Page 356Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. Highway 101 Makeshift berm at first culvert (a) First culvert through which product flowed. (b) Second culvert through which product flowed. Figure 3. Photographs showing two of the culverts through which released product flowed. December 4, 2015 DNV GL - OAPUS307KKRA (PP136049) 50#

Page 357Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. Inadequate: Personal • Programme Factors • Standards • Compliance Job/System Factors Figure 4 Schematic showing the Loss Causation Model. December 4, 2015 DNV GL - OAPUS307KKRA (PP136049) 51#

Page 358Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Figure 5. Timeline showing key events for Line 901 from the time of construction to the day of the incident (May 19, 2015). DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 52#

Page 359Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 Figure 6. Timeline showing key events for Line 901 on the day of the incident (May 19, 2015). 53#

Page 360Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 900 Gaviota Incoming Pressure Las Flores Discharge Pressure 800 • Joint 5930 Pressure 700 - 1255 pm: Las Flores Pmax - 721 psig (recorded) 12:55 pm: Joint 5930P max - 814 psig (calculated) 12:54 pm: Gaviota Pmar - 707 psig (recorded) 100 - 00:00 T 06:00 12:00 18:00 00:00 Time (hour:minute) Figure 7. Plot of pressure versus time showing the discharge pressure for Las Flores (red), the incoming pressure for Gaviota (blue), and the calculated pressure for Joint 5930 (green) on May 19, 2015 [Ref 227]. December 4, 2015 DNV GL - OAPUS307KKRA (PP136049) 54#

Page 361Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 700 680 660 640 620- Pressure (psig) 600 - 580 - 560 - 10:55: Unplanned pump 540 - shutdown and restart at Sisquoc 11:30: Las Flores Pump Shutdowr 520 - 11:15: Sisquoc pump shutdown - Gaviota Incoming Pressure - Las Flores Discharge Pressure 500 - Joint 5930 Pressure 10:30 11:00 11:30 12:00 Time (hour:minute) Figure 8. Plot of pressure versus time showing the discharge pressure for Las Flores (red), the incoming pressure for Gaviota (blue), and the calculated pressure [Ref 227]. for Joint 5930 (green) on May 19, 2015 between 10:00 am and 12:00 pm December 4, 2015 DNV GL - OAPUS307KKRA (PP136049) 55#

Page 362Plains All American Pipeline, Line 901 Release (5/19/15) Technical Root Cause Analysis L.P. (b) (7)(F) MAP HCA MOP PIPE DATA PROFILE LEGEND I PLAINS WE All American Pipeline, L.F L901 LAS FLORES TO GAVIOTA - 24 ERECTIFER .903 GAVIOTA TO SISQUOC - 30 PLM Degment CA_LFCA_PNTCA "001 Figure 9. [Ref 248]. Schematic showing Line 901 from Las Flores to Gaviota showing the approximate location of the pipeline and the elevation profile of the pipeline. The red arrows indicate the locations of flow meters December 4, 201: 56#

Page 363Plains All American Pipeline, Line 901 Release (5/19/15) Technical Root Cause Analysis L.P. (b) (7X(F) HCA MOP PIPE DATA PROFILE LAINS PLAIN WEEN All American Pipeline, L. = RENTEER L903 GAVIOTA TO SISQUOC - 30" CA_LSFCA_PNTCA Mar 07 2012 Las Fores to Pertand Figure 10. Schematic showing Line 903 from Gaviota to Sisquoc the approximate location of the pipeline and the elevation profile of the pipeline. The red arrows indicate the locations of flow meters [Ref 249]. DNV GL - OAPUS307KKRA (PP136049 December 4, 2015 57#

Page 364Plains All American Pipeline, Line 901 Release (5/19/15) Technical Root Cause Analysis L.P. D)((F) HCA MOP PIPE DATA PROFILE HA DATING WIFE All American Pipeline, L.P. • RECTIFIER L903 SISQUOC TO PENTLAND - 30" PLM Segment CA_LSFCA_PNTCA Las Fores ta Fertiand 003 Figure 11. Schematic showing Line 903 from Sisquoc to Station Number 1595 + 16 the approximate location of the pipeline and the elevation profile of the pipeline. The red arrows indicate the locations of flow meters [Ref 250]. December 4, 2015 DNV GL - OAPUS307KKRA (PP136049)#

Page 365Plains All American Pipeline, Line 901 Release (5/19/15) Technical Root Cause Analysis L.P. (b) (7)(F) MAP HCA MOP PIPE DATA PROFILE LEGEND a NE All American Pipeline, LF ALAS PLAINS L903 SISQUOC TO PENTLAND - 30" CA_LFCA_PNTCA war 07 3013 Las Fores to Fentand 0D4 Figure 12. Seen es in Line 903 from Station Number 1595 + 16 to Pentand the approximate location of the pipeline and the elevation profile of the pipeline. The red arrows indicate locations of flow December 4, 201 59#

Page 366Plains All American Pipeline, L.P. Line 901 Release 5/19/15) Technical Root Cause Analysis —Historical — Estimated b) (4) Volume (bbls) -800 5/18/150:00 5/18/15 12:00 5/19/150:00 Date 5/19/15 12:00 5/20/150:00 5/20/15 12:00 Figure 13. Plot showing volume versus time data for the Las Flores to Pentland line segment between May 18, 2015 at 5:00am and May 20, 2015 at 12:00am. (b) (4) -Historical — Estimated (bbls) Volume 5/18/15 12:00 5/19/150:00 5/19/15 12:00 Date 5/20/150:00 5/20/15 12:00 Figure 14. Plot showing volume versus time data for the Las Flores to Pentland line segment between May 19, 2015 at 1:00am and May 20, 2015 at 12:00am. DNV GL - OAPUS307KKRA (PP136049) December 4, 2015 60#

Page 367Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. 4:22 33.35 33.9 ' (b) Flow 6 .8 .9 8 4:22 33.0-34.0' from U/S GW Feature 4 (Failure) 34.2-34.42' from U/S GW Feature 6 Figure 15. cleaning. Tape measure indicates distance to U/S GW (Note: tape slipped 0.1' to the Photograph showing the failure location before (Top) and after (Bottom) right). December 4, 2015 DNV GL - OAPUS307KKRA (PP136049) 61#

Page 368Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Transportation of Hazardous Liquid in Pipelines External Corrosion Control System Integrity Program External Loss of Corrosion - Containment Corrosion Under Insulation External Coating Cathodic Protection Contracted In-line System System Inspection Mitigative Actions Figure 16. BowTie diagram, generated using the BSCA T™ methodology, summarizing the preventative barriers in place for the Line 901 Release. The barriers shown as two rectangles on either side of the horizontal line correspond to failed barriers, while the thin rectangles that span the horizontal line correspond to ineffective barriers. December 4, 201 62#

Page 369Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. Polyethylene tape Polyethylene Tape Polyurethane foam 1.5 - inch thick rigid polyurethane foam Coal tar Coal tar urethane coating urethane API 5L Grade X65 line pipe steel Pipe steel Figure 17. Schematic and photograph showing the protective external coating and the PU foam and PE tape layers present on the Line 901 pipeline. DNV GL - OAPUS307KKRA (PP136049) 63 December 4, 2015#

Page 370Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. Decohesion PE tape Failure location Wrinkles PE tape Corrosion Saturation & Product Compression of Disbonded PU foam coal tar coating Figure 18. Photographs showing compromised protective coating and PU foam/PE tape layers at the failure location. DNV GL - OAPUS307KKRA (PP136049) 64 December 4, 2015#

Page 371Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Figure 19. Photograph of wrinkles in the PE tape, located away from the failure location. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 65#

Page 372Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. Corrosion product 100 110 120 130 140 150 15 Disbonded coal tar coating (24") Saturation of PU foam Figure 20. Photographs showing compromised coating and PU foam away from the failure location. December 4, 2015 DNV GL - OAPUS307KKRA (PP136049) 66#

Page 373Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Figure 21. Plot showing the distribution of external metal loss features vs. o’clock orientation identified for Line 901 during the 2007, 2012, and 2015 ILI runs. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 67#

Page 374Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. LF TO GAV Dig #05 12" Armor Plate 03/03/09 14 SAt 00/03/09 PIGT STA 213 + 46.66 STA 213 + 43.66 (a) 2007 ILI Dig #5 FLOW 5-13-13 Line 901 Dig 12 WC21311.88 Recoat @3000 V/OK val (b) 2012 ILI Dig #13 Figure 22. Photographs showing recoated pipe on Line 901 after: (a) 2007 ILI Dig #5 and (b) 2012 ILI Dig #13 [Ref 147 & 169] December 4, 2015 DNV GL - OAPUS307KKRA (PP136049) 68#

Page 375Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Figure 23. Plot of temperature data, provided by Plains, for Las Flores Station between May 2014 and May 2015 [Ref ‎226]. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 69#

Page 376Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. Plains All American Line 901 "IRF" Pipe-to-Soil Potential Annual Test Point Survey Data 2008 - 2014 - 2015 * 2015 IRF • 2014 IRF • 2008 IRF - -0.95 V Criterion - -0.85 V Criterion + 2015 Leak -2.000 -1.800 -1,600 -1.400 i -1.200 Soil Potential 1,000 7 g -0.800 Pipe -0.600 2015 Failure -0.400 -0.200 0.000 0+00 100+00 200÷00 300+00 400+00 500+00 600+00 Station (feet) Figure 24. Plains All American Line 901 "IRF" pipe-to-soil potential annual test point survey data years 2008, 2014, and 2015. Note: IRF: free of IR error. DNV GL - OAPUS307KKRA (PP136049) 70 December 4, 2015#

Page 377Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. Plains All American Line 901 In-Line Inspection 2007 Maximum External Metal Loss Depth Reported Aligned to 2008 Close Interval Survey 1000 800 Elevation (feet) 600 400 200 * On * Interrupted - Native • -0.95 V Criterion - -0.85 V Criterion — -1.2 V Reference -2.000 -1.500 1,200 -0 800 -0,400 0.000 0.500 0,400 2007 EML Depth (inch) 0,300 2015 Failure 0,200 0.100 0.000 10000 20000 30000 40000 50000 60000 Odometer (feet) Figure 25. Plains All American Line 901 In-Line Inspection 2007 maximum external metal loss depth reported aligned with 2008 Close Interval Survey (- Pipe Nominal Wall Thickness). DNV GL - OAPUS307KKRA (PP136049) 71 December 4, 2015#

Page 378Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Figure 26. Plains All American Line 901 In-Line Inspection 2012 maximum external metal loss depth reported aligned with 2015 Close Interval Survey (- Pipe Nominal Wall Thickness). DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 72#

Page 379Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Figure 27. Plains All American Pipeline Las Flores I Rectifier: Direct current (DC) output recorded between 2005 - 2015. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 73#

Page 380Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Figure 28. Plains All American Pipeline Las Flores II Rectifier: Direct current (DC) output recorded between 2005 - 2015. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 74#

Page 381Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Figure 29. Plains All American Pipeline Gaviota I Rectifier: Direct current (DC) output recorded between 2005 - 2015. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 75#

Page 382Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Figure 30. Plains All American Pipeline Gaviota II Rectifier: Direct current (DC) output recorded between 2005 - 2015. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 76#

Page 383Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Figure 31. Plains All American Pipeline L 901 Cathodic Protection Rectifiers: Average direct current (DC) output recorded between 2005 - 2015. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 77#

Page 384Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Figure 32. Plains All American Line 901 2015 close interval potential survey and annual test point survey data recorded between 2009 and 2014. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 78#

Page 385Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 5/21/2015 F11-2 (2012 ILI) 5/22/2015 Preliminary Report (2015 ILI) 5/6/2015 MFL-A/XT Survey (2015 ILI) 7/10/2013 DOT Compliance Report 7/3/2012 12/31/2012 8/19/2013 - 10/4/2013 CXG Survey 4/9/2015 Assessment Plan PHMSA Inspection† (3 yr Interval)* Close Interval Survey (CIS2) 6/4/2015 Final Report (2015 ILI) 9/24/2012 Final Report (2012 ILI) 9/26/2012 CGAR 7/3/2012 8/3/2015 10/18/2012 - 10/3/2013 Excavations 3/26/2014 Las Flores Discharge Pressure, 888 psig 4/28/2015 Aerial Patrol 5/19/2015 Failure 5/29/2015 - 6/3/2015 Excavations * Assessment plan specifies 3 year reassessment interval. Date indicated is the date of the Assessment Plan. † PHMSA inspection occurred on August 19-22, September 16-19 and September 30-October 4, 2013. Results of inspection provided to Plains September 11, 2015 Figure 33. Timeline of events associated with Line 901, following the 2012 ILI. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 79#

Page 386Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis (b) (4) Figure 34. Excerpt from IMP Fig 9-2 illustrating process to estimate corrosion growth rates [Ref ‎22]. Orientation, h:mm 12:00 11:00 10:00 9:00 8:00 7:00 6:00 5:00 4:00 3:00 2015 ILI – 23% 2007 ILI – 23% 2007 ILI – 19% 2012 ILI – 24% 2015 ILI – 47% 2015 ILI – 38% Feature 6– 34% 2015 ILI – 24% Feature 3– 58% Feature 4– 89% 2015 ILI – 25% 2012 ILI – 45% 2012 ILI – 21% 2:00 1:00 2012 ILI – 38% 2015 ILI – 47% 0:00 21378 21380 21382 21384 21386 21388 21390 21392 21394 2015 Odometer, ft Figure 35. Representation of reported metal loss features on Joint 5930 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 80#

Page 387Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. 100 Recoat 90 Failure Location 5/14/2013 89% WT Max. Depth (Dig 13) 80 70 60 %WT Depth, 02007 ILI Data • 2012 ILI Data 4 2015 ILI Data 30 20 10 21378 21384 LLL 2015 Odometer, ft 21386 21390 21392 21394 Figure 36. Depths of ILI-reported metal loss features on Joint 5930 December 4, 2015 DNV GL - OAPUS307KKRA (PP136049) 81#

Page 388Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 100 ILI under reported 90 80 70 Field-Measured Depth, %WT 60 50 y = 0.1508x + 33.165 R² = 0.0475 40 30 20 10 ILI over reported 0 0 10 20 30 40 50 60 70 80 90 100 2007 ILI-Reported Depth, %WT Figure 37. Metal loss depth unity plot using Plains data. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 82#

Page 389Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. 100 ILI under reported 90 80 = 0.4988x + 23.967 B$ = 0,1716 70 60 Field-Measured Depth, %WT 50 40 30 20 10 ILI over reported 0 10 20 60 70 80 90 100 2012 ILI-Reported Depth, %WT Figure 38. Metal loss depth unity plot using Plains data. Light blue diamonds correspond to features located greater than 2 feet from a girth weld. Purple diamonds correspond to features within 2 feet of a girth weld. December 4, 2015 DNV GL - OAPUS307KKRA (PP136049) 83#

Page 390Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 100 ILI under reported 90 80 70 60 Field-Measured Depth, %WT 50 y = 0.4815x + 25.589 R: = 0.4402 40 30 I2012-2015 Matching • Laser Scan [Ref 1] 20 A 45 Day Report [Ref 187] OML <= 2ft GW 10 ILI over reported 0 0 10 20 40 80 90 100 2015 ILI-Reported Depth, %WT Figure 39. DNV GL-produced metal loss depth unity plot for the 2015 ILI of the Las Flores to Gaviota line segment. December 4, 2015 DNV GL - OAPUS307KKRA (PP136049) 84#

Page 391Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Figure 40. Excerpt from API 1163 used to establish consistency with performance specification (Table 8 in Appendix E, [Ref ‎309]). Figure 41. Snapshot showing portion of Figure 6-1 from Section 6.2 of Plains’ IMP, regarding regrading [Ref ‎20]. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 85#

Page 392Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis APPENDIX A BSCAT™ Methodology DNV GL – OAPUS307KKRA (PP136049) December 4, 2015#

Page 393Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. BSCAT Methodology Causal analysis is the core of an incident investigation. The analysis uses a systematic method of processing evidence gathered during an investigation in order to identify the factors that led to the incident. This approach assists in the development of corrective and/or remedial measures. For the Line 901 Failure, DNV GL applied its standard Loss Causation Model to the incident. The DNV GL Loss Causation Model used in the analysis is shown in Figure A-1. As seen in the figure, the model involves a progression of factors that lead to an incident. In order to explain why and how the incident occurred, the progression would start at the box on the left-hand side, which is labeled "Lack of Control" and is often used interchangeably with "Root Cause." Typically, the root cause of an incident is related to weaknesses or gaps in the management system. The weaknesses or gaps may be related to programs, processes, standards, or compliance. Weaknesses in the management system then lead to a "Basic Cause." Typically, basic causes are related to engineering decisions, technical events, personal factors, or job/system factors. The basic cause in turn leads to the "Immediate Cause" of the incident. The immediate cause typically involves substandard conditions or acts/practices and is addressed in a metallurgical or materials analysis. Finally, the immediate cause progresses to the incident. Consequences of the incident are shown in the box to the far right. The consequences include a loss to people, property, equipment, a process, and/or the environment. When carrying out the analysis, the figure is applied in the reverse order (i.e. starting with the loss and working backwards toward the "root cause"). By identifying the root causes) of each incident, it is possible to derive process-related actions for improvement that can be implemented and managed throughout the site operations. Key lessons learned in this respect can also be shared with other sites exposed to similar conditions and programs. H • • lE 1 Inadequate: • Personal • Programme Factors • • IL I D • Standards • Compliance Job/System Factors • Figure A-1. Schematic showing loss causation model. DNV GL - OAPUS307KKRA (PP136049) December 4, 2015 A-1#

Page 394Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Two approaches typically used by DNV GL include the Systematic Causal Analysis Technique (SCAT™) and the Barrier-based Systematic Causal Analysis Technique (BSCAT). SCAT™ is an RCA approach that uses standardized causation descriptions to convey the immediate and basic causes of an incident. This technique helps incident investigators identify weak areas in the integrity management system. The standard causation descriptions help to categorize commonalities that can be tracked in order to prioritize the weak areas of the management system. BSCAT™ is a technique that applies the SCAT model to each barrier, as opposed to the incident as a whole. This method results in a thorough review of the effectiveness of the individual barriers identified in the risk assessment. BSCAT provides a methodology that allows for the analysis of complex incidents that involve multiple barriers. A summary of the steps involved in the BSCAT process are outlined in ‎Table A-1. BowTie diagrams are used in BSCAT™ to identify the barriers that are in place to prevent threats from escalating into an incident and the barriers that are in place to mitigate consequences following an incident. A BowTie analysis can be performed before an accident/incident to help assess the barriers that are in place and their current state. BowTies can also be created following an accident/incident to analyze the system’s barriers at the time of the accident/incident. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 A-2#

Page 395Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Table A-1. Summary of BSCAT process. The BSCAT process involves the following steps: 1. Evidence Capture – This includes collecting information pertaining to the incident through interviews of the people involved and reviews of documents related to the incident. 2. Timeline Development – The evidenced captured is used to create a timeline of the events leading up to the incident. 3. Barrier Identification – If a BowTie diagram of the incident has not been created, one is created using the threat that escalated to the main event. The barriers that are in place or could be in place are identified at this time. 4. Barrier State – The state of each barrier is determined. The barrier status descriptions include Effective, Ineffective, Failed, and Missing. The term “Effective” is used to describe a barrier that is performing in the manner as originally intended. “Ineffective” is a term used to describe a barrier that is in place and operating, but its performance is deficient. The term “Failed” is used to describe a barrier that was originally in place, but has degraded and no longer functions as originally intended. “Missing” is used to describe a barrier that was never in place. 5. Causal Analysis – The SCAT process is then applied to the barriers that are identified as Ineffective, Unreliable, or Missing. This process will show the immediate and basic causes of the barrier’s ineffective state, as well as where the gaps in the Management System Elements, as shown in Table A-1, exist. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 A-3#

Page 396Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis APPENDIX B References DNV GL – OAPUS307KKRA (PP136049) December 4, 2015#

Page 397Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. The following is a list of references that were used for the RCA. The reference numbers listed below are used throughout this report to identify the source of information. Nº of Reference Number Document Name Issued by Dated Page 1. Incident Related Documents 1. Metallurgical Report (PP13 6049) September 18 Plains All American Pipeline- Line 901 - Final DNV GL 9/18/15 111 2. Background Sheet-L901-6-3-15 American Plains All Pipeline, L.P. 6/3/15 3 Corrective_Action_Order_Plains_Pipeline_LP PHMSA 5/21/15 4. Order_06032015.pd 520155011H_Amendment_to_the_Corrective_Action_ PHMSA 6/3/15 7 5. 509pm National Response Center 2015 Current Data 7-16-15 NRC 7/16/15 L901 Supplemental (Rev 11.24.15) - PHMSA F 7000.1 American Plains All Pipeline, L.P. 11/24/15 14 520155019_NOPV PCO_09112015 PHMSA 9/11/15 6 2. Integrity Related Documents 8. ILI Review Process Procedure American Plains All 41 Pipeline, L.P 9. CGAR Checklist (2010) American Plains All Pipeline, L.P. 1 10. Repair Plan Checklist_2012 American Plains All Pipeline, L.P. 1 11. Repair Plan Checklist_6-16-15 American Plains All Pipeline, L.P. 1 12. Line 901 Las Flores to Gaviota Hydrotesting AKRI 1/10/91 34 13. ExternalCorrosionIndexFactors Plains All Pipeline, L.P. American 3 14. American Plains AlI RiskResults_2009-2014_23Jun2015 Pipeline, L.P. 1 15. 0646578 ROSOFT Data Management Version 6.70. Disc 1 of 1. Rosen 16. Summary Rpt_Las Flores to Gaviota_Final American Plains All Pipeline, L.P. 9/5/13 92 DNV GL - OAPUS307KKRA (PP136049) December 4, 2015 B-1#

Page 398Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. Reference N° of Number Document Name Issued by Dated Page S Integrity Management Plan 17. Intergrity Management Plan Table of Contents American Plains All Pipeline, L.P. 12/18/03 3 18. Section 3 - Risk Assessment Procedures American Plains All Pipeline, L.P. 2/7/07 28 19. Section 4 Pipeline Assessment Method Selection Proc American Plains All Pipeline, L.P. 9/11/14 16 20. IMP Section 6 Procedures for Conducting Plains All Assessments & Processing Results Pipeline, L.P. American 7/10/08 30 21. IMP Section 8 Pipeline Repair Requirements American Plains All Pipeline, L.P. 6/2/06 2 22. Section 9 Procedure for Continual Assessment_Eval American Plains All Pipeline, L.P 2/10/07 18 23. Section 11 - Plains IMP 2014 American Plains All Pipeline, L.P. 218/2007 30 24. Procedure Spec. No. 201-Pipeline Maintenance Welding & Repair American Plains All Pipeline, L.P. 6/22/07 26 Cathodic Protection Surveys 25. Foreign Line Crossing American Plains All Pipeline, L.P. 1 26. Las Flores Annual CP Survey 2012-15 Baker Hughes 5/21/15 3 American Plains All 27. Revised Rect inspection 5yr Pipeline, L.P. 8/6/15 6 28. REvisedTest Point Inspections 5yr American Plains All Pipeline, L.P. 8/6/15 4 29. Las Flores 24 In_2008_CRI Hanson Survey & Design 12/5/08 - 12/8/08 53 30. Las Flores 24In_2015_Book Hanson Survey & Design 4/8/15- 4/9/15 65 Close Interval Surveys 31. Gaviota to Emidio 30 In_2009_CRI Hanson Survey & Design, LLC 9/29/09 544 32. Gaviota to Emidio 30 In_2009_RAW Hanson Survey & Design, LLC 7/13/15 5 DNV GL - OAPUS307KKRA (PP136049) December 4, 2015 B-2#

Page 399Line 901 Release (5/19/15 Technical Root Cause Analysis Plains All American Pipeline, L.P. No of Reference Number Document Name Issued by Dated Page 33. Las Flores 24 In_2008_CRI Hanson Survey & Design, LLC 12/5/08 53 34. Las Flores 24In_2015_Book Hanson Survey & Design, LLC 65 35. Line 903 Hwy 101_2012_Book Hanson Survey & Design, LLC 86 36. Line 903 Hwy 101_2012_Depth Hanson Survey & Design, LLC 1 37. Line 903 Hwy 101_2012_Raw Hanson Survey & Design, LLC 496 38. Line 903 Hwy 101_2012_Rect Hanson Survey & Design, LLC 2012 July 39. Line 903 MP 54_2012_Book Hanson Survey & Design, LLC 21 40. Line 903 MP 54_2012_Raw Hanson Survey & Design, LLC 1 41. Line 903 MP 54_2012_Rect Hanson Survey & Design, LLC 1 42. Line 903 MP 60_2012_Book Hanson Survey & Design, LLC 49 43. Line 903 MP 60_2012_Raw Hanson Survey & Design, LLC 493 44. Line 903 MP 60_2012_Rect Hanson Survey & Design, LLC 1 Rectifier Reports 45. 2005-2015 Annaul TP Survey 903 Baker Hughes 6/4/15 35 46. 2005-2015 Annual TP Survey 901 Baker Hughes 6/11/15 8 47. L901 Rectifier Inspections 2005-2015 Baker Hughes 6/11/15 5 48. L-901-CCH224_COUPON Baker Hughes 5/28/15 1 49. L903 Rectifier Inspections 2005-2015 Baker Hughes 6/11/15 10 In Line Inspection 2007 In Line Inspection DNV GL - OAPUS307KKRA (PP136049) December 4, 2015 B-3#

Page 400Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Nº of Reference Number Document Name Issued by Dated Page 50. 2007 Rosen Report Rosen 8/15/07 28 51. anomaly counts_rtf_converted.docx Rosen 1 52. Anomaly Relative to Closest Weld Distance Rosen 8/15/07 1 53. Anomaly Type Distribution Chart_Pie Chart Rosen 8/15/07 1 54. CDG Magnetization Level Rosen 8/15/07 1 55. CDG Tool Rotation Rosen 8/15/07 1 56. CDG Tool Temperature Rosen 8/15/07 1 57. CDG Tool Velocity Rosen 8/15/07 1 58. Conclusions_itf_converted.docx Rosen 8/15/07 1 59. data quality summary_rtf_converted.docx Rosen 1 60. Depth Distribution of All Metal Loss Anomalies Rosen 8/15/07 1 61. Depth Distribution of Internal Metal Loss Anomalies Rosen 8/15/07 1 62. Anomalies_2 Depth Distribution of Non-Internal Metal Loss Rosen 8/15/07 1 63. Dig Sheets Las Flores to Gaviota Rosen 8/15/07 45 64. EGP Tool Rotation Rosen 8/15/07 1 65. EGP Tool Temperature Rosen 8/15/07 1 66. EGP Tool Velocity Rosen 8/15/07 1 67. ERF Distribution Graph Rosen 8/15/07 1 68. Given MAOP, Pdesign and Theoretical Safe Pressure Graph Rosen 8/15/07 1 DNV GL - OAPUS307KKRA (PP136049) December 4, 2015 B-4#

Page 401Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis No of Reference Number Document Name Dated Page Issued by S 69. ISFR_5362_02_ Rosen 8/15/07 3 70. ISFR_5365_72_ Rosen 8/15/07 3 71. ISFR_18814_52_ Rosen 8/15/07 3 72. ISFR_21438_78_ Rosen 8/15/07 3 73. ISFR_31574_91_ Rosen 8/15/07 3 74. List of Installations Rosen 8/15/07 4 75. List of Marker Positions Rosen 8/15/07 1 76. List of Most Severe Anomalies Rosen 8/15/07 2 77. List of Significances_2 Rosen 8/15/07 46 78. O'clock Position of All Metal Loss Anomalies_2 Rosen 8/15/07 1 79. O'clock Position of Internal Anomalies Rosen 8/15/07 1 80. O'clock Position of Non-Internal Anomalies Rosen 8/15/07 1 81. Pipe Tally 2007 Las Flores to Gaviota.xis Rosen 8/15/07 41 82. pipetally.xis Rosen 8/16/07 30 83. report Rosen 8/15/07 28 84. Signed 2007 ILI Summary Report Plains All American 8/23/07 6 85. 2007 Pipetally Rosen 8/15/07 86. American Plains All CGAR _LasFlores to Gaviota_08_17_07 Pipeline, L.P. 8/17/07 6 87. Close Out Report_LasFlores_Gav_2007 American Plains All Pipeline, L.P. 3 December 4, 2015 DNV GL - OAPUS307KKRA (PP136049) B-5#

Page 402Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Reference No of Number Document Name Issued by Dated Page 88. #_LCG-C129-A64. Rosen 2007. Disk 1 of 1 Discovery LCG >> S Experts. 89. #_LCG-C129-A65. Rosen 2007. Disk 2 of 2 Discovery LCG >> Experts. 90. L901 Form F11-2_2007 American Plains All Pipeline, L.P. 10/8/09 5 2012 In Line Inspection 91. 2012 Rosen Report Rosen 9/24/12 28 92. AGM sheets Rosen 9/24/12 24 93. Anomaly Relative to Closest Weld Distance_2 Rosen 9/24/12 1 94. Anomaly Type Distribution Chart Rosen 9/24/12 1 95. CG Magnetization Level Rosen 9/24/12 1 96. CG Tool Rotation Rosen 9/24/12 1 97. CXG Tool Temperature Rosen 9/24/12 1 98. CG Tool Velocity Rosen 9/24/12 1 99. Deformation Distribution Rosen 9/24/12 1 100. Deformation Orientation Rosen 9/24/12 1 101. Depth Distribution of All Metal Loss Anomalies_2 Rosen 9/24/12 1 102. Depth Distribution of Internal Metal Loss Anomalies_2 Rosen 9/24/12 1 103. Anomalies Depth Distribution of Non-Internal Metal Loss Rosen 9/24/12 1 104. Dig Sheets_P1 Rosen 9/24/12 3 105. Dig Sheets_P2 Rosen 9/24/12 38 DNV GL - OAPUS307KKRA (PP136049) December 4, 2015 B-6#

Page 403Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. Nº of Reference Number Document Name Dated Page Issued by S 106. ERF Distribution Rosen 9/24/12 1 107. Gauge Pig Spec Sheet Rosen 5/7/12 1 108. Given MAOP, Pdesign and Theoretical Safe Pressure Rosen 9/24/12 1 109. ISFR_23707_53- Rosen 9/24/12 3 110. ISFR_26907_13_ Rosen 9/24/12 3 111. ISFR_26907_36_ Rosen 9/24/12 3 112. ISFR_33480_86_ Rosen 9/24/12 3 113. ISFR_42570_35_ Rosen 9/24/12 3 114. List of Installations_2 Rosen 9/24/12 4 115. List of Markers Rosen 9/24/12 1 116. List of Significances Rosen 9/24/12 102 117. Metal Loss Distribution Rosen 9/24/12 1 118. Metal Loss Orientation Rosen 9/24/12 1 119. MFL tool Spec and calibration sheet Rosen 1/13/12 2 120. O'clock Position of All Metal Loss Anomalies Rosen 9/24/12 1 121. O'clock Position of Internal Metal Loss Anomalies Rosen 9/24/12 1 122. O'clock Position of Non-Internal Metal Loss Anomalies Rosen 9/24/12 1 123. Pipetally x/s Rosen 9/24/12 29 American Plains All 124. Signed 2012 ILl Summary Report Pipeline, L.P. 9/5/13 8 DNV GL - OAPUS307KKRA (PP136049) December 4, 2015 B-7#

Page 404Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Reference No of Number Document Name Issued by Dated Page 125. Site Survey Report Rosen 7/5/12 4 126. 2012 Pipetally Rosen 9/24/12 127. AssessmentPlan_31Dec2012 American Plains All Pipeline, L.P. 12/31/12 1 128. CGAR 2012_Las Flores to Gaviota_9-26-12 American Plains All Pipeline, L.P. 6 129. Close Out Report_LasFlores_Gav_2012_rev1 Pipeline, L.P. American Plains All 6/22/15 5 130. FW Unity Plots for Sisquoc to Pentland 2012 and Las American Plains All Flores to Gaviota 2008 and 2012.msg Pipeline, L.P. 9/18/15 131. L901 Las Flores to Gaviota Form F11-2 2015 rev American Plains All Approved 6-9-2015 Pipeline, L.P. 5/21/15 10 132. #_LCG-C129-A62. Rosen 2012-A. Disk 1 of 1 Discovery LCG >> LCG >> Experts 133. #_LCG-C129-A63. Rosen 2012-B. Disk 1 of 2 Discovery LCG >> Experts 134. #_LCG-C129-A71. Rosen 2012. Disk 2 of 2 Discovery Plains All Experts 135. CGAR_Line 901-Las Flores-Gaviota-2012 DAS Pipeline, L.P American 6/21/13 8 2015 In Line Inspection 136. anomaly counts Rosen 1 137. conclusions Rosen 1 138. Final Rosen Report May 2015 ILI Rosen 6/4/15 227 139. 2015 Pipetally Rosen 6/4/15 33 140. Pipetally.x/s Rosen 6/4/15 141. #PA84/A456. Rosen 2015. Disk 1 of 2 LCG»> Discovery Experts 142. #PA84/A457. Rosen 2015. Disk 2 of 2 Discovery LCG >> DNV GL - OAPUS307KKRA (PP136049) December 4, 2015 B-8#

Page 405Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. No of Reference Number Document Name Issued by Dated Page Experts 143. Preliminary Report 24in Line 901 Las Flores to Gaviota Rosen 5/22/15 13 Excavation and Repair Reports 2007 Digs 144. Las-Gav-WC 5342.18_Dig 3 American Plains All Pipeline, L.P 55 145. Plains All Las-Gav-WC 5365.72_Dig 3.PDF Pipeline, L.P. American 65 146. Las-Gav-WC 18814.52 Dig 4 American Plains All Pipeline, L.P. 55 147. Las-Gav-WC 21307.63 Dig 5 American Plains All Pipeline, L.P. 44 148. Las-Gav-WC 21438.79_Dig 6 American Plains All Pipeline, L.P. 149. Las-Gav-WC 22083.48 Dig 7 American Plains All Pipeline, L.P. 45 American Plains All 150. Las-Gav-WC 29052.98_Dig 8 Pipeline, L.P. 62 American Plains All 151. Las-Gav-WC 29206.61_Dig 9.PDF Pipeline, L.P. 152. Las-Gav-WC 31574.91_Dig 10.PDF American Plains All Pipeline, L.P. 54 153. Las-Gav-WC 34025.04_Dig 11.PDF American Plains All Pipeline, L.P. 65 154. Las-Gav-WC 34137.11_Dig 11B.PDF American Plains All Pipeline, L.P. 36 155. Las-Gav-WC 44100.91_Dig 12.PDF American Plains All 63 Pipeline, L.P. 156. Las-Gav-WC 45139.21_Dig 13 American Plains All 51 Pipeline, L.P. 2012 Digs American Plains All 157. Las-Gav_Dig 1 WC 26907.14 Pipeline, L.P. 56 DNV GL - OAPUS307KKRA (PP136049) December 4, 2015 B-9#

Page 406Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. N° of Reference Number Document Name Issued by Dated Page 158. Plains All S Las-Gav_Dig 2 WC 33480.86 Pipeline, L.P American 93 159. Las-Gav_Dig 3 WC 42570.35 American Plains All Pipeline, L.P. 99 160. Plains All Las-Gav_Dig 4 WC 269.14 Pipeline, L.P. American 43 161. Las-Gav_Dig 5 WC 678.93 American Plains All 31 Pipeline, L.P. 162. Las-Gav_Dig 6 WC 4324.14 American Plains All Pipeline, L.P. 75 163. Las-Gav_Dig 7 WC 6982.96 American Plains All Pipeline, L.P. 37 164. Plains All Las-Gav_Dig 8 WC 10595.52 Pipeline, L.P. American 43 165. American Plains All Las-Gav_Dig 9 WC 14970.32 Pipeline, L.P. 40 166. Las-Gav_Dig 10 WC 16669.12 American Plains All Pipeline, L.P. 39 167. American Plains All Las-Gav_Dig 11 WC 17474.02 Pipeline, L.P. 168. Las-Gav_Dig 12 WC 21311.88 American Plains All Pipeline, L.P. 40 169. Las-Gav_Dig 13 WC 21390.33 American Plains All 45 Pipeline, L.P. 170. Las-Gav_Dig 14 WC 23707.53 American Plains All Pipeline, L.P. 39 171. Las-Gav_Dig 15 WC 28857.22 American Plains All 37 Pipeline, L.P. 172. Las-Gav_Dig 16 WC 29490.43 American Plains All 36 Pipeline, L.P. 173. Las-Gav_Dig 17 WC 32732.65 American Plains All Pipeline, L.P. 66 174. Las-Gav_Dig 18 WC 32954.85 American Plains All 79 Pipeline, L.P. 175. Las-Gav_Dig 19 WC 33081.18 American Plains All Pipeline, L.P. 35 DNV GL - OAPUS307KKRA (PP136049) December 4, 2015 B-10#

Page 407Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. N° of Reference Number Document Name Issued by Dated Page 176. Plains All Las-Gav_Dig 20 WC 33800.97 Pipeline, L.P American 37 177. Las-Gav_Dig 20A WC 33820.56 American Plains All Pipeline, L.P. 59 178. Plains All Las-Gav_Dig 21 WC 33946.08 Pipeline, L.P. American 38 179. Las-Gav_Dig 21A WC 33993.62 American Plains All 65 Pipeline, L.P. 180. Las-Gav_Dig 22 WC 34106.54 American Plains All Pipeline, L.P. 33 181. Las-Gav_Dig 23 WC 34188.43 American Plains All Pipeline, L.P. 182. Plains All Las-Gav_Dig 24 WC 35975.10 Pipeline, L.P. American 77 183. American Plains All Las-Gav_Dig 25 WC 39655.36 Pipeline, L.P. 46 184. Plains All Las-Gav_Dig 26 WC 41609.94 Pipeline, L.P. American 93 185. Plains All Las-Gav_Dig 27 WC 41769.64 Pipeline, L.P. American 41 186. Las-Gav_Dig 28 WC 43990.86 American Plains All Pipeline, L.P. 50 187. Las-Gav_Dig 29 WC 44827.97 American Plains All Pipeline, L.P. 188. Las-Gav_Dig 30 WC 45183.16 American Plains All Pipeline, L.P. 42 189. Las-Gav_Dig 31 WC 45246.50 American Plains All Pipeline, L.P. 32 190. Las-Gav_Dig 32 WC 46263.62 American Plains All 54 Pipeline, L.P. 191. Las-Gav_Dig 33_33A WC 47191.95 American Plains All Pipeline, L.P. 101 192. Las-Gav_Dig 34 WC 47341.16 American Plains All 32 Pipeline, L.P. 193. Las-Gav_Dig 35 WC 47726.17 American Plains All Pipeline, L.P. 35 DNV GL - OAPUS307KKRA (PP136049) December 4, 2015 B-11#

Page 408Line 901 Release (5/19/15 Technical Root Cause Analysis Plains All American Pipeline, L.P. Reference No of Number Document Name Issued by Dated Page 194. Plains All Las-Gav_Dig 36 WC 50280.62 Pipeline, L.P. American 91 195. Plains All Las-Gav_Dig 37 WC 50941.59 Pipeline, L.P. American 63 196. Las-Gav_Dig 38 WC 54239.07 American Plains All Pipeline, L.P. 98 197. Las-Gav_Dig 39 WC 54358.73 American Plains All Pipeline, L.P 66 198. Las-Gav_Dig 40 WC 54472.60 American Plains All Pipeline, L.P. 34 199. Las-Gav_Dig 41 WC 54627.39 American Plains All Pipeline, L.P. 2015 Digs 200. 45 Day Report CAO 7-6-15 FINAL American Plains All Pipeline, L.P. 7/6/15 7 3. Leak Detection Documents 201. 100-8 Pipeline Leak Detection American Plains All Pipeline, L.P. 6 202. American Plains All OS_Data Pipeline, L.P. 1 203. American Plains All PAA_0000003 - PAA_0000028 Pipeline, L.P. 26 204. PAA_0000090 - PAA_0000095 American Plains All Pipeline, L.P. 6 205. PAA_0000104_CONFIDENTIAL American Plains All Pipeline, L.P. 464 206. PAA_0000105_CONFIDENTIAL American Plains All 82 Pipeline, L.P. 207. PAA_0000106 American Plains All Pipeline, L.P. 6 208. PAA_0000112 American Plains All Pipeline, L.P. 1 209. PAA_0000114 American Plains All Pipeline, L.P. 3 DNV GL - OAPUS307KKRA (PP136049) December 4, 2015 B-12#

Page 409Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis No of Reference Number Document Name Issued by Dated Page 210. PAA_0000119_CONFIDENTIAL American Plains All Pipeline, L.P 211. PAA_0000129_CONFIDENTIAL American Plains All Pipeline, L.P. 499 212. PAA00011284 American Plains All Pipeline, L.P. 1 213. Line 901 5-1-2014 to 5-19-2015 American Plains All Pipeline, L.P. 499 214. Line 903 5-1-2014 to 5-19-2015 American Plains All 499 Pipeline, L.P. 215. SCADA Tags American Plains All Pipeline, L.P. 1 Patrol Data 216. 01 - Jan 07, 2015 - Jan 28, 2015 Pipeline, L.P. American Plains All 1/21/15 - 2/2/15 4 217. 02 - Feb 04, 2015 - Feb 25, 2015 American Plains All 2/11/15- Pipeline, L.P. 3/4/15 4 218. 03 - Mar 04, 2015 - Mar 25, 2015 American Plains All 3/12/15- Pipeline, L.P. ? 3 219. 04 - Apr 01, 2015 - April 29, 2015 American Plains All 4/13/15 - Pipeline, L.P. 5/14/15 5 220. 05 - May 04, 2015 - May 11, 2015 American Plains All 5/14/15 - Pipeline, L.F 5/19/15 2 Written Responses to PHMSA 221. American Plains All PAA_0000102 - PAA_0000103 Pipeline, L.P 5/28/15 2 222. Responses to PHMSA Investigation Questions 7.2.15 American Plains All Pipeline, L.P 7/2/15 2 223. Information 6.2.15 Responses to PHMSA Request for Documents and American Plains All Pipeline, L.P. 6/2/15 2 4. Operations Documents 224. BKPlainsPipe American Plains All Pipeline, L.P 2/15 29 225. American Plains All 901_LasFlores_Gaviota_24in_7_2_15 7/2/15 Pipeline, L.P. 1 DNV GL - OAPUS307KKRA (PP136049) December 4, 2015 B-13#

Page 410Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. N° of Reference Number Document Name Issued by Dated Page 226. 901_pressuredata_summary American Plains All S 5/20/15 1/26/11 Pipeline, L.P. 7 227. Pressure Data 5-19-2015 American Plains All Pipeline, L.P. 5/19/15 1 228. O&M Manual Table of Contents American Plains All Pipeline, L.P. 9/10 11 229. O&M Table of Contents American Plains All Pipeline, L.P. 3/13 4 230. • & M - 405 System Start-up and Shutdown American Plains All Pipeline, L.P. 9/10 4 231. • & M - 412 Corrosion Control Plains All Pipeline, L.P. American 9/14 16 232. 0 & M - 415 Pipeline Repairs American Plains All Pipeline, L.P. 10/11 8 233. Conditions 0 & M - 425 Instruction for Recognizing Safety Related Plains All Pipeline, L.P. American 9/10 4 234. 0 & M - 501 Introduction Pipeline, L.P. American Plains All 10/13 4 235. • & M - 502 Unintended Valve Closure American Plains All Pipeline, L.P. 9/10 2 236. • & M - 503 Unintended Shutdown American Plains All Pipeline, L.P. 9/10 2 237. 0 & M - 504 Abnormal Pressure or Flow Rates American Plains All Pipeline, L.P. 9/10 6 238. 0 & M - 505 Complete Loss of Communications Plains All Pipeline, L.P. American 9/10 2 239. Failure of a Safety Device to Operate 0 & M - 506 Operation of a Safety Device or American Plains All Pipeline, L.P. 9/10 2 240. 0 & M - 507 Returning to Normal Operation American Plains All Pipeline, L.P. 9/10 2 241. Operations O & M - 508 Review of Response to Abnormal American Plains All 9/10 Pipeline, L.P. 2 242. O & M - 509 Tank Overfill Alarm Plains All Pipeline, L.P. American 9/10 2 5. Historical Documents DNV GL - OAPUS307KKRA (PP136049) December 4, 2015 B-14#

Page 411Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. N° of Reference Number Document Name Issued by Dated Page 243. _Binder 7 of 8_P.O., MTR's, O&M Manuals All American S Company Pipeline 90 244. Insulations specifications All American Company Pipeline 49 245. MTRs 24inch 312wt and 500wt Corporation Nippon Steel 12/10/85 3 246. MTRs 24inch 344wt Nippon Steel Corporation 12/10/85 7 6. Drawings, Maps, and Diagrams 247. 24in As Built Field Book Harold D. Hardin Land Surveyor 2/5/91 238 248. Plains All CA_LSFCA_PNTCA - 001 Pipeline, L.P American 317/13 1 249. Plains All CA_LSFCA_PNTCA - 002 Pipeline, L.P. American 3/7/13 1 250. CA_LSFCA_PNTCA - 003 American Plains All Pipeline, L.P. 3/7/13 1 251. Pipeline, L.P. American Plains All CA_LSFCA_PNTCA - 004 317/13 1 252. CA_LSFCA_PNTCA-RMS-001 American Plains All 6/10/15 Pipeline. L.P. 1 253. PAA_0000096 American Plains All Pipeline, L.P. 9/27/13 1 254. PAA_0000097 American Plains All Pipeline, L.P. - 1 255. SouthCuyamaGathering American Plains All 8/26/15 Pipeline, L.P. 1 Alignments Line 901 256. 01 CE-COV All American Company Pipeline 1 257. 02 CE-IND-1 All American Company Pipeline 1 258. 03 CE-IND-2 All American Company Pipeline 259. 04 CE-001A All American Pipeline Company 9/11/90 1 DNV GL - OAPUS307KKRA (PP136049) December 4, 2015 B-15#

Page 412Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. N° of Reference Number Document Name All American Issued by Dated Page S 260. 05 CE-001B Company Pipeline 4/11/90 1 261. 06 CE-001C All American Company Pipeline 9/11/90 1 262. 07 CE-001D All American Pipeline 9/11/90 All American Company 1 263. 08 CE-001E Pipeline 9/11/90 Plains AlI Company 1 264. 6/9/15 901-D-PP-ALL_REV3 Pipeline, L.P. American 6/16/15 13 Alignments Line 903 265. 09 CE-002 Celeron Pipeline Company of Celeron Pipeline California 7/13/87 1 266. 10 CE-003 Company of Celeron Pipeline California 7/13/87 1 267. 11 CE-004 Company of Celeron Pipeline California 7/13/87 1 268. 12 CE-005 Company of California 7/13/87 1 269. 13 CE-006 Celeron Pipeline Company of Celeron Pipeline California 7/13/87 1 270. 14 CE-007 Company of Celeron Pipeline California 7/13/87 1 271. 15 CE-008 Company of Celeron Pipeline California 7/13/87 1 272. 16 CE-009 Company of Celeron Pipeline California 7/13/87 1 273. 17 CE-010 Company of Celeron Pipeline California 7/13/87 1 274. 18 CE-011 Company of Celeron Pipeline California 7/13/87 1 275. 19 CE-012 Company of California 7/13/87 1 276. 20 CE-013 Celeron Pipeline Company of California 7/13/87 1 DNV GL - OAPUS307KKRA (PP136049) December 4, 2015 B-16#

Page 413Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. N° of Reference Number Document Name Dated Page Celeron Pipeline Issued by S 277. 21 CE-014 Company of Celeron Pipeline California 7/13/87 1 278. 22 CE-015 Company of Celeron Pipeline California 7/13/87 1 279. 23 CE-016 Company of Celeron Pipeline California 7/13/87 1 280. 24 CE-017 Company of Celeron Pipeline California 7/13/87 1 281. 25 CE-018 Company of Celeron Pipeline California 7/13/87 1 282. 26 CE-019 Company of Celeron Pipeline California 7/13/87 1 283. 27 CE-020 Company of Celeron Pipeline California 7/13/87 1 284. 28 CE-021 Company of Celeron Pipeline California 7/13/87 1 285. 29 CE-022 Company of California 7/13/87 1 286. 30 CE-023 Celeron Pipeline Company of Celeron Pipeline California 7/13/87 1 287. 31 CE-024 Company of Celeron Pipeline California 7/13/87 1 288. 32 CE-025 Company of Celeron Pipeline California 7/13/87 1 289. 33 CE-026 Company of California 7/13/87 1 Line 903 Elevation Profile Drawings 290. CA_PNTCA_EMDCA-RMS-001 American Plains All 6/10/15 Pipeline, L.P 1 291. CA_LSFCA_PNTCA-RMS-002 American Plains All Pipeline, L.P 6/10/15 1 292. CA_LSFCA_PNTCA-RMS-003 Plains All Pipeline, L.P American 6/10/15 1 293. CA_LSFCA_PNTCA-RMS-004 Plains All Pipeline, L.P American 6/10/15 1 DNV GL - OAPUS307KKRA (PP136049) December 4, 2015 B-17#

Page 414Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. Nº of Reference Number Document Name Issued by Dated Page 7. Public Documents 294. Letter Plains Pipeline Response to 6-5-15 Congressional American Plains All Pipeline, L.P 6/19/15 14 295. Excerpt from PAA's Anticipated 10Q Disclosure County of Santa Pipeline, L.P. American Plains All Regarding Line 901 8/7/15 1 296. sPipeline.asp http://www.sbcountyplanning.org/energy/projects/Plain Planning and Barbara, Energy Division Development- 297. https://www.plainsallamerican.com/about-us/company- American Plains All history Prepared Oral Testimony of Patrick Hodgins - June 26, Pipeline, L.P. 298. 2015 American Plains All /2552586/ http://www.plainsline901response.com/go/doc/7266 Pipeline, L.P. 6/26/15 8. Standards, Papers, etc. "Pipeline External Corrosion Direct Assessment NACE International Standard Practice SP0502-2010, International NACE 2010 60 Methodology" "Control of External Corrosion on Underground or NACE International Standard Practice SP0169-2007 International NACE 2007 NACE International Standard Practice SP0169-2013 Submerged Metallic Piping Systems". 36 "Control of External Corrosion on Underground or International NACE 2013 Submerged Metallic Piping". 60 302. pipeline buried in soil Cathodic protection criteria of thermally insulated Corrosion Science 43, 2001 303. NACE RP0198: Control of Corrosion Under Thermal Insulation and Fireproofing Materials - A Systems NACE 1998 Petrobras Transporte S.A. - Transpetro Solution for the Approach 304. Insulation Pipelines for Underground Heated Oil Minimization of External Corrosion of Thermal Proceedings of IPC 2004 2004 Transportation in Brazil. 305. pipelines On the cathodic protection of thermally insulated Failure Analysis Engineering (2009) 16 2047- 2053 306. Physical and Performance Properties of Coal Tar Urethanes - Pipe International NACE April 2- 6, 1984 8 Improvement In a Sour Crude Gathering and Under Deposit Corrosion Mitigation and ILI Accuracy 307. International NACE Chemistry Transportation System Accomplished Using Novel 2012 16 308. Leakage Signals in Inspection of Farside Metal-loss Influence of Corrosion Products on Magnetic Flux Japan Petroleum Journal of the January Defects in Oil Storage Tank Bottom Institute 2004 9 DNV GL - OAPUS307KKRA (PP136049) December 4, 2015 B-18#

Page 415Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis No of Reference Dated Page Number Document Name Issued by 309. Systems Qualification Standard Recommended Practice 1163 - In-Line Inspection Petroleum American August Institute 2005 50 310. Insulation and Fireproofing Recommended Practice 583 - Corrosion Under Petroleum American May Institute 2014 88 311. NACE International Publication 10A392 (2006 Edition) Insulated Underground Metallic Structures Effectiveness of Cathodic Protection on Thermally International NACE 2006 8 312. ASME B31G Manual for Determining the Remaining American Society Strength of Corroded Pipes of Mechanical Engineers 2012 60 313. Ordinances - Supplement 32 Update 2 (Ordinance No. Santa Barbara County, California - Code of Santa Barbara 4926) - Chapter 14 County Code 6/23/15 43 314. http://www.usclimatedata.com/climate/santa- US Climate Data for Santa Barbara County: barbara/california/united-states/usca1017/2000/1 US Climate Data 8/18/15 - 315. remaining strength of corroded pipe Citation: A modified criterion for determining the PRCI Contract PR-3-805 12/22/89 of Transportation U.S. Department 316. 49 CFR 195 - Transportation of Hazardous Liquids by and Hazardous (DOT) Pipeline Pipeline 70 Materials Safety Administration (PHMSA) December 4, 2015 DNV GL - OAPUS307KKRA (PP136049) B-19#

Page 416Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis APPENDIX C Supplemental Analyses from 2015 Digs DNV GL – OAPUS307KKRA (PP136049) December 4, 2015#

Page 417Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis SUPPLEMENTAL 2015 DIG ANALYSES 1.0 BACKGROUND Four priority digs, identified as Digs 1 – 4, were performed on Line 901 between May 29, 2015 and June 3, 2015, based on the preliminary findings of the 2015 ILI run. ‎Figure C-1 contains a topographical map and elevation plot of Line 901 showing the locations of Digs 1 - 4 relative to the failure location. All four digs were located D/S from the failure in relative low areas along the line; see ‎Figure C-2 and ‎Figure C-3. The locations of these digs were selected based on the maximum depths of external metal loss features on Line 901, as identified by the tool. ‎Table C-1 identifies the features associated with the four digs and summarizes the dimensional findings [Ref 181] for each feature. The maximum corrosion depths of the measured features were all lower than the depths identified by the tool (i.e. the features were over-called by the ILI tool). ‎Figure C-4 – ‎Figure C-7 contain field photographs from the four digs showing representative corrosion products associated with the external corrosion features. For all four digs, the corrosion was located at areas of disbonded coal tar urethane (CTU) coating, beneath an intact PU foam layer. The corrosion products were primarily dark brown in appearance with some areas that were rust-colored. In general, the products were dry, fairly rigid, and magnetic. Although portions of the corrosion products were removed as relatively thick, intact samples, the products were a bit more friable (i.e. crumbled) than the deposits removed near the failure location. All four samples exhibited evidence of a layered morphology; see ‎Figure C-8. DNV GL personnel were present during all four digs and collected various samples at each dig site for laboratory analysis. The collected samples included the following: (1) corrosion products associated with the external metal loss features, (2) swab samples for bacteria testing removed at and away from the features, (3) soil samples removed from the dig sites, and (4) insulation samples removed at the feature locations. The objectives of the analyses were to characterize the samples and to compare the results for the samples with the results obtained for samples removed near the failure location. 2.0 TECHNICAL APPROACH The procedures used in the analyses were in accordance with industry-accepted standards. Three of the general standards governing terminology and bacteria testing used are as follows:  NACE/ASTM G193 – 10a “Standard Terminology and Acronyms Relating to Corrosion.” DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 C-1#

Page 418Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis  NACE TM0106, “Detection, Testing, and Evaluation of Microbiologically Influenced Corrosion (MIC) on External Surfaces of Buried Pipelines.”  NACE TM0194, “Standard Test Method for Field Monitoring of Bacterial Growth in Oil and Gas Systems.” Corrosion products were collected during each dig for characterization. Analyses performed on these products included: (1) elemental analyses using energy dispersive spectroscopy (EDS) with a scanning electron microscope (SEM) and (2) compound identification using x- ray diffraction (XRD). Swab samples were also obtained for bacteria analyses, over a standard area of 1 cm2, at two locations per dig site (i.e. at an area of corrosion and an area where the coating was disbonded but there was negligible external corrosion). Separate swab samples were taken for serial dilution and microscopic analysis. Liquid culture media for acid-producing bacteria (APB), sulfate-reducing bacteria (SRB), nitrate-reducing bacteria (NRB), aerobic bacteria (AERO), anaerobic bacteria (ANA), and iron-related bacteria (IRB) was used for the serial dilutions to evaluate growth of various types of bacteria. A five vial serial dilution (1:10,000) was performed using each type of media. The swab obtained for the microscopic analysis was fixed in 1% glutaraldehyde. A five microliter specimen was removed from the fixed sample and prepared for examination by drying on a microscope slide and staining with 0.1% fluorescein isothiocyanate (FITC). The sample was examined using a CFI PLAN FLUOR 100X oil immersion objective on a Nikon Eclipse 50i epifluorescent microscope equipped with a FITC filter set to determine bacteria cell counts and morphology. Analyses were conducted on soil samples removed (in the field) from each dig site. The soils were tested for resistivity, moisture content, pH, total acidity, total alkalinity, concentration of soluble anions and cations, total dissolved solids, and linear polarization resistance; see ‎Table C-2 for a summary of the soil related procedures. Analyses were also performed on liquids extracted from insulation samples that were removed from the feature locations at each dig site. Due to the limited sample volumes, only two of the extracts were analyzed. The extracts were analyzed for only the following: soluble anions [Cl-, SO4 2-, NO2 - , NO3 -, CO3 2-, HCO3 -], total alkalinity, and total dissolved solids. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 C-2#

Page 419Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 3.0 RESULTS 3.1 Corrosion Product Analyses 3.1.1 X-ray Diffraction ‎Table C-3 shows the results of XRD analyses performed on the corrosion products from Digs 1 – 4. Compounds identified in all four samples were goethite (FeO(OH)) and magnetite (Fe3O4). Goethite is one of the most thermodynamically stable iron oxides under aerobic (high oxygen) conditions. Conversely, magnetite is a metastable phase formed under low oxygen conditions. In Dig 4, a third compound, akaganeite (Fe3+O(OH,Cl)) was also identified. Akaganeite is indicative of the presence of oxygen. 3.1.2 Energy Dispersive Spectroscopy The results of the EDS analyses performed on the corrosion products from Digs 1 through 4 are summarized in ‎Table C-4. The two primary constituents are iron (Fe) and oxygen (O), which are characteristic of iron oxides. Small quantities of chlorine (Cl) were identified, likely associated with chlorides. Small quantities of manganese (Mn) were identified, which is a common constituent of line pipe steels. A relatively high concentration of carbon (C) was identified in all scans, which may be from organics within the insulation, soil, and/or bicarbonate compounds found in ground water. 3.2 Microbiological Analyses The external surfaces of Joints 6550, 12420, 12460, and 14470 from Digs 1 – 4, respectively, were swabbed over a standard area of approximately 1 cm2 for bacterial analysis. For each pipe joint, the swabs were taken from a representative external corrosion pit and from an area away from the corrosion pit. Separate swab samples were taken from each location for the serial dilution and microscopic examination analyses. The results of the microbiological analyses are discussed below. 3.2.1 Serial Dilution – Liquid Culture Media ‎‎Table C-5 shows the results of the bacteria serial dilution testing for the swab samples collected from the pipe joints. The results reveal that the majority of the swab samples exhibited a positive indication for five types of bacteria (APB, AERO, ANA, IRB, and NRB). Only the swab samples taken at an area away from the corrosion feature for Digs 2 and 4 were positive for all six bacteria types (i.e. AERO, ANA, APB, SRB, IRB, and NRB). As seen in the table, the highest concentration of bacteria detected was 100,000 bacteria per cm2 , which is a relatively high value. There was no evidence to indicate that bacteria were preferentially flourishing at the corrosion pits. In many cases, higher concentrations of bacteria were found in the swabs taken from areas away from the corrosion features. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 C-3#

Page 420Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 3.2.2 Microscopic Examination for Total Bacteria The swabs collected from the four dig locations were fixed in 1% glutaraldehyde and examined using epifluorescent microscopy. The practical minimum detection limit for this method is approximately 103 cells/ml of fixed sample. The results of the analysis are provided in ‎Table C-6. As seen in the table, rod-shaped cells were detected for all the swab samples. The calculated concentration of cells for the swab samples ranged between 2.10 × 104 cells/mL and 2.8 × 104 cells/mL, which are high values. This type of microscopic examination does not differentiate between living and non-living organisms. 3.3 Soil Analyses ‎Table C-7 is a summary of the soil samples collected by DNV GL during the four priority digs performed in 2015. Information on the soil samples collected near the failure location are also provided in the table for comparison. The first column in the table identifies the location where the sample was obtained. Columns 2, 3, and 4 provide DNV GL’s designation for the soil, the associated Arcsset number ID, and a brief field description of the soil, respectively. Columns 5 and 6 provide the joint number where the soil was taken and whether the soil was analyzed. Six (6) soil samples were removed from the dig site near the failure location; see ‎Table C-7. Two samples were collected from under the pipe at each of three locations: 8 feet U/S of GW 5930 (IDs 10000151761 & 10000151762), 2 feet D/S of the failure location (IDs 10000151753 & 1000151759), and 12.5 feet D/S of GW 5940 (IDs 10000151754 & 10000151755). The only samples not contaminated with product, and thus representative of the soil prior to the failure, were the samples collected 8 feet U/S of GW 5930. One of these samples, ID 10000151761, was analyzed. ‎Figure C-9 is a photograph of Soil 10000151761 in the shipped bag. The soil consisted of clumps in a variety of sizes that were cream to tan colored in appearance. Five (5) additional soil samples were removed during the four priority digs performed following the release; see ‎Table C-7. The soil from Dig 1 (ID 10000151758) was removed below the pipe at a GW on Joint 6550, which was located west or D/S of the failure location. The soils from Dig 2 (ID 100151751) and Dig 3 (ID 1000195234) were removed at the pipe on Reference Joint 12420 and under the pipe at Reference Joint 12460, respectively. Both digs were located D/S of the failure location. The soil samples removed from Dig 4 (IDs 10000195233 and 10000195232) were collected from the top of the pipe from Reference Joint 14470, near a corrosion feature. Dig 4 was located D/S of the failure location. Only one of the samples from Dig 4 (ID 10000195233) was analyzed. ‎Figure C-10 contains photographs of the four soils that were analyzed from Digs 1 – 4. All four soils consisted of DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 C-4#

Page 421Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis clumps. The soils from Digs 1 and 3 consisted of equally sized larger rocks, while the soils from Digs 2 and 4 consisted of rocks of varying sizes. The soil from Dig 1 was black to charcoal in appearance, while the soils from Digs 2 – 4 were cream to tan in appearance. The following steps were performed for the soil analyses. The soil samples were collected, shipped, and handled in accordance with DNV GL’s standard operating procedure for soils. Analysis began with each soil sample pulverized into small pieces. The soils were then sifted through a #10 sieve (2.0 mm particle size) to remove gravel, leaving soil particles classified as sand, silt, and clay. The selected soils were tested for pH, moisture content, and resistivity. Testing was also performed to estimate the corrosion rate of carbon steel within the soil using linear polarization resistance (LPR), which is an electrochemical technique. Next, water soluble anions and cations were extracted from the soils, using a 5:1 water to soil ratio, to determine their relative concentrations. The extracts were also tested to determine the total acidity, total alkalinity, and total dissolved solids present in each extract. The procedures used in the analysis were in accordance with industry- accepted standards, which are summarized in ‎Table C-2. The results of the analyses are provided in ‎Table C-8 – ‎Table C-10. In general, the results of the analyses revealed that the soil removed near the failure location exhibited more corrosive properties, as received, than those soils removed from the priority dig locations. This conclusion is based on the following results for the as-received soil removed near the failure location: (1) the higher moisture content, (2) the lower resistivity, (3) the higher determined corrosion rate, and (4) the higher levels of sulfate (SO4 2-) anions. All five soil samples exhibited more corrosive properties in the saturated condition. In general, the soil removed near the failure location exhibited the most corrosive properties. This soil exhibited the lowest resistivity and the second highest corrosion rate in the saturated condition. Based on the findings, the corrosive properties of the soil are impacted by moisture content, which is expected. 3.4 Insulation Extract Analyses During Digs 1 – 4, DNV GL collected samples of the insulation that had been in contact with the pipe at each feature location. The samples were bagged and shipped to DNV GL’s laboratory in Columbus, OH, where the liquids within the samples were extracted. ‎Figure C- 11 is a photograph showing the liquids extracted from the insulation samples from each dig. ‎Table C-11 provides a summary and description of the four extracted samples. The volume of extracted liquids varied from approximately 20 to 120 mL. The extracts from the insulation samples from Digs 1 and 4 were relatively clear in appearance, while the extracts from the insulation samples from Digs 2 and 3 were rust-colored in appearance. Based on DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 C-5#

Page 422Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis the limited extract volumes, two representative samples were selected for chemical analysis. One sample (i.e. Dig 2 sample) was selected to represent a rust-colored extract and the second sample (i.e. Dig 4 sample) was selected to represent a clear extract. ‎Table C-12 is a summary of the chemical analyses performed on the Dig 2 and Dig 4 insulation extracts. Due to the limited sample volumes, these samples were analyzed for only the following: soluble anions [Cl-, SO4 2-, NO2 -, NO3 -, CO3 2-, HCO3 -], total alkalinity, and total dissolved solids. The concentrations of soluble anions were consistently higher for the Dig 2 Extracts compared to the Dig 4 Extracts. Both extract samples exhibited higher levels of chlorides (Cl-), nitrates (NO3 -), sulfates (SO4 -), and bicarbonates (HCO3 -) than the soil samples that were removed from these locations. These findings indicate that a higher concentration of corrosive species may have been in contact with the pipe at these locations. Furthermore, the insulation may facilitate the concentration process as the insulation experiences wet-dry cycling. 4.0 SUMMARY OF FINDINGS  The corrosion products ♦ Are primarily dark brown in appearance with some areas that were rust-colored. ♦ Are dry, rigid, and magnetic. ♦ Consist of a layered morphology comprised primarily of goethite and magnetite.  There is no strong evidence to indicate that MIC played a primary role in the observed external corrosion observed for Digs 1 – 4.  The results of analyses performed on soil samples, removed near the failure and dig locations, revealed that the soil removed near the failure location exhibited more corrosive properties.  Analyses of liquids extracted from insulation samples removed near the corrosion features from Digs 1 – 4 revealed higher concentrations of corrosive species (i.e. chlorides) than their respective soil samples. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 C-6#

Page 423Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Table C-1. Summary of features identified during Priority Digs 1 – 4 performed in 2015. Tool Calls Priority Dig Number Reference Joint Log Distance of Feature (ft) Tool Call Max Length (in) Max Depth (%) Max Depth Field (%) Dig 1 6550 23785.8 External metal loss 0.75 85 72 Dig 2 12420 44719.8 External metal loss 0.87 72 54 44874.43 External metal loss 1.98 53 49.3 Dig 3 12460 44877.52 External metal loss 0.98 83 74 51640.00 External metal loss 0.88 71 65 Dig 4 14470 51640.27 External metal loss 0.83 57 56.3 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 C-7#

Page 424Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. Table C-2. Summary of Soil Related Procedures. Test Parameter Methodology Standard Soil Handling Soil Permit Guidelines N/A Soil Extraction Water extraction of soluble anions and cations SW846-1311, 1312 (modified) pH 1:1 slurry ASTM D4972 Resistivity, As Received or Saturated 4pt Wenner method ASTM G57, AASHTO T288-91 Moisture Content weight loss technique ASTM D2216, AASHTO T265 Corrosion Rate of Soil by LPR Linear Polarization Resistance ASTM G5, G15, G59, G102, Linear Polarization Resistance Measurement Soluble Anions Nitrite, NO,- Nitrate Colorimetric, auto Analytical method: EPA 353.2 Chloride, Cr Sulfate, SO? lon chromatography Analytical method: EPA 300.0 Sulfide, S?- Colorimetric Analytical method: SM 4500-S2-D Carbonate, CO3* Bicarbonate, HCO3 Titrimetric Analytical method: SM2320B Total Alkalinity Total Acidity Titrimetric Analytical method: SM2310B Soluble Cations Calcium, Ca* Magnesium, Mg* Potassium, K* ICP Analytical method: EPA 6010, Preparation Method: EPA 3010 Sodium, Na* Total Dissolved Solids (TDS) Gravimetric residue Analytical method: SM2540C DNV GL - OAPUS307KKRA (PP136049) C-8 December 4, 2015#

Page 425Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Table C-3. Results of compound analyses, using X-ray diffraction, performed on corrosion products from Digs 1 – 4. Compound Dig 1 Dig 2 Dig 3 Dig 4 Goethite – FeO(OH) Present Present Present Present Magnetite – Fe3O4 Present Present Present Present Akaganeite – FeO(OH) – – – Present Table C-4. Results of elemental analyses, using EDS, performed on corrosion products from Digs 1 – 4 compared to ideal chemistry compositions of goethite and magnetite; values presented in mass percent (wt.%). Elements Dig 1 Dig 2 Dig 3 Dig 4 Goethite (FeOOH) Magnetite (Fe3O4) Carbon (C) 9.9 7.9 4.6 4.8 – – Oxygen (O) 34.2 33.2 34.7 34.4 36.01 27.64 Sodium (Na) – 0.6 – – – – Silicon (Si) – 0.2 – – – – Chlorine (Cl) 0.3 0.4 0.3 0.3 – – Manganese (Mn) 0.5 0.5 0.9 0.7 – – Iron (Fe) 55.1 57.2 59.5 59.8 62.85 72.36 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015#

Page 426Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Table C-5. Results of bacteria analyses performed on swabs taken, over an ~1 cm2 area, from the external surfaces of Joints 6550, 12420, 12460, and 14470 during Digs 1 – 4, respectively, at and away from corrosion features. Dig 1 (Joint 6550) Dig 2 (Joint 12420) Pit Area Away Pit Area Away Bacteria Type Test Result Number of Positive Vials Test Result Number of Positive Vials Test Result Number of Positive Vials Test Result Number of Positive Vials Aerobic (AERO) Positive 3 Positive 5 Positive 3 Positive 4 Anaerobic (ANA) Positive 2 Positive 5 Positive 3 Positive 2 Acid-Producing (APB) Positive 3 Positive 5 Positive 2 Positive 4 Sulfate-Reducing (SRB) Not detected – Not detected – Not detected – Positive 2 Iron-Related (IRB) Positive 1 Positive 5 Not detected – Positive 3 Nitrate-Reducing (NRB) Positive 3 Positive 5 Positive 5 Positive 4 Dig 3 (Joint 12460) Dig 4 (Joint 14470) Pit Area Away Pit Area Away Bascteria Type Test Result Number of Positive Vials Test Result Number of Positive Vials Test Result Number of Positive Vials Test Result Number of Positive Vials Aerobic (AERO) Positive 5 Positive 5 Positive 5 Positive 5 Anaerobic (ANA) Positive 5 Positive 5 Positive 5 Positive 5 Acid-Producing (APB) Positive 4 Positive 4 Positive 3 Positive 5 Sulfate-Reducing (SRB) Not Detected – Not Detected – Not Detected – Positive 1 Iron-Related (IRB) Positive 5 Positive 4 Positive 2 Positive 4 Nitrate-Reducing (NRB) Positive 5 Positive 5 Positive 4 Positive 4 Bacteria Concentration Key: 1 10 bacteria per cm2 2 100 bacteria per cm2 , 3 1,000 bacteria per cm2 , 4 10,000 bacteria per cm2 , 5 100,000 bacteria per cm2 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 C-10#

Page 427Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Table C-6. Results of optical microscopy examination for fixed swab samples taken, over an ~1 cm2 area, from the external surfaces of Joints 6550, 12420, 12460, and 14470 during Digs 1 – 4, respectively, at and away from corrosion features. Dig Number Sample Identification Aliquot Volume, uL Total Cells Observed Calculated № cells/mL Morphology Pit 5 >20 2.80 × 104 Rod 1 Area Away 5 >20 2.80 × 104 Rod Pit 5 >20 2.80 × 104 Rod 2 Area Away 5 15 2.10 × 104 Rod Pit 5 16 2.20 × 104 Rod 3 Area Away 5 >20 2.80 × 104 Rod Pit 5 >20 2.80 × 104 Rod 4 Area Away 5 18 2.50 × 104 Rod DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 C-11#

Page 428Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Table C-7. Summary of soil samples collected by DNV GL. DNV GL Designation Sample ID (ArcSSETT #) Field Description Near Failure 10000151761 @ 8 ft U/S of U/S GW 5930 below pipe – 10000151762 @ 8 ft U/S of U/S GW 5930 below pipe Soils Near Failure Location – 10000151753 2 ft D/S of leak location – 10000151759 2 ft D/S of leak location – 10000151754 12.5 ft D/S of GW 5940 – 10000151755 12.5 ft D/S of GW 5940 Dig 1 10000151758 Soil from dig West of leak location below pipe @ GW Dig 2 10000151751 Dig 2 @ pipe Priority Dig Soils Dig 3 10000195234 Dig 3 Soil under pipe – 10000195232 Dig 4 soil @ top of pipe near corrosion S/N 1 of 2 6/03/15 Dig 4 10000195233 Dig 4 soil @ top of pipe near corrosion S/N 2 of 2 6/03/15 1 - Duplicate sample 2 - Sample contaminated with crude oil DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 Reference Joint 5920 5920 5930 5930 5940 5940 6550 12420 12460 14470 14470 Analyses Performed Yes No 1 No 2 No 2 No 2 No 2 Yes Yes Yes No 1 Yes C-12#

Page 429Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Table C-8. Summary of various chemical and electrochemical properties for soil samples. DNV GL Designation pH Soil Moisture Content (%) 1 Resistivity (Ohm-cm) Corrosion Rate (mpy) As Received Saturated As Received Saturated Near Failure 7.95 27.59 3,800 400 2.517 2.718 Dig 1 8.53 18.31 2,500 810 0.359 1.933 Dig 2 8.21 11.57 29,000 580 0.160 2.244 Dig 4 8.52 12.77 78,000 12,000 0.094 2.328 Dig 3 7.56 6.80 14,000 690 0.410 4.405 1 – Percent moisture per AASHTO T265 & ASTM D2216 Table C-9. Summary of soluble cation and anion concentrations for soil samples removed near the failure location and from Digs 1 – 4. Soluble Cations mg/L Soluble Anions, mg/L DNV GL Designation Ca2+ Mg2+ Na+ K+ NO2 - NO3 - Cl- SO4 2- S2- CO3 2- HCO3 - Near Failure 898 320. 495 9.64 <2.1 114.84 117 3600 <0.67 <13.3 204 Dig 1 18.0 <6.10 218 <6.10 <2.0 12.68 29 49 <0.61 <12.2 744 Dig 2 53.0 38.7 413 8.88 10.472 40.33 78.6 338.8 <0.57 <11.4 529.5 Dig 3 9.57 <5.77 493 5.88 <1.9 21.18 108 200 <0.58 <11.5 524 Dig 4 60.0 24.8 95.2 <5.41 <1.8 9.09 26 206 <0.54 <10.8 146 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 C-13#

Page 430Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Table C-10. Summary of various chemical properties determined for soil samples removed near the failure location and from Digs 1 – 4. DNV GL Designation Total Alkalinity (mg CaCO3/L) Total Acidity (mg CaCO3/L) Total Dissolved Solids (TDS) (mg/L) Near Failure 204 < 66.5 6350 Dig 1 744 < 61.0 640 Dig 2 530 < 56.9 1550 Dig 3 524 < 57.7 1390 Dig 4 146 < 54.1 622 Table C-11. Summary of liquids extracted from insulation samples collected by DNV GL during Priority Digs 1 – 4 performed in 2015. Extract Identification Reference Joint Sample ID (ArcSSETT #) pH at Pipe/Insulation Interface (Field measurement) Estimated Extracted Volume (mL) Extract Appearance Chemical Analysis Performed Dig 1 Extract 6550 10000151104 6 – 7 20 Clear No Dig 2 Extract 12420 10000151105 7 – 8 120 Rust-colored Yes Dig 3 Extract 12460 10000151106 7 60 Rust-colored No Dig 4 Extract 14470 10000151107 6 40 Clear Yes Table C-12. Summary of results of chemical analyses performed on liquids extracted from the insulation removed during Priority Digs 2 and 4 performed in 2015. Sample ID NO2 - NO3 - Cl- Dig 2 Extract 48.0 Dig 4 Extract < 1.6 Soluble Anions (mg/L) 739.28 30.10 SO4 - CO3 2- HCO3 - Total Alkalinity As CaCO3 (mg/L) Total Dissolved Solids (mg/L) 1080 2000 < 2.0 397 397 7470 329 993 < 2.0 102 102 2020 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 C-14#

Page 431Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis CAVIOTA 006500 Refugio State Beach Line 901 Elevation Profile • Elevation [ft] - 2015 Failure location - Dig 1 Dig 2 - Dig 3 - Dig 4 1000 Gaviota Station FLOW Las Flores Station Dig 4 Dig 3 Dig 2 Dig 1 Failure 800 600 400 200 ... Distance Downstream from Las Flores Station (feet) Figure C-1. Topographical map and elevation plot showing the locations of Priority Digs 1 - 4 relative to the failure location. The white triangles on the map correspond to mile post markers. The green star or he map and the green line on the plot identify the location of the May 19, 2015 failure gust 28, 201 15#

Page 432Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis (a) Failure Location (b) Dig 1 Location Figure C-2. Plots showing close-ups of the elevation profile of Line 901 at: (a) the failure location and (b) the location of Dig 1. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 C-16#

Page 433Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis (a) Dig 2 and 3 Locations (b) Dig 4 Location Figure C-3. Plots showing close-ups of the elevation profile of Line 901 at: (a) the locations of Digs 2 and 3 and (b) the location of Dig 4. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 C-17#

Page 434Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Corrosion products Corrosion products PU foam 100 110 120 150 15 90 100 110 120 130 - 150 (24"') Corrosion products 20 30 40 50 60. 70 80 90 100 110 120 130 Figure C-4. Field photographs showing the corrosion products associated with the external metal loss feature from Dig 1. DNV GL - OAPUS307KKRA (PP136049) C-18 December 4, 2015#

Page 435Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. Corrosion products CTU coating Corrosion products PU foam CTU coating 40 90 100 -110 120 150 PU foam External metal loss features Figure C-5. Field photographs showing representative corrosion products associated with the external metal loss features from Dig 2. DNV GL - OAPUS307KKRA (PP136049) December 4, 2015 C-19#

Page 436Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Corrosion 20 30 40 50 80 90 100 - 110 120 13. products Corrosion products PU foam Corrosion products Corrosion products PU foam 50 60 10 20 50 40 50 B0 70 30 0Q 100110 Figure C-6. Field photographs showing representative corrosion products associated with the external metal loss features from Dig 3. DNV GL - OAPUS307KKRA (PP136049) C-20 December 4, 2015#

Page 437Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Corrosion products External metal loss features 1 3 4. 0E 152 132 2 9z L5/0302015 11:39г Flexib 05/03/2005 01141 Figure C-7. Field photographs showing representative corrosion products associated with the external metal loss features from Dig 4. DNV GL - OAPUS307KKRA (PP136049) C-21 December 4, 2015#

Page 438Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Dig 1 Dig 2 9 6p l or 2. 9 6 7 8 5 6 7 Dig 3 Dig 4 8Z = 07 g. * C L 9 82 2 02 .. = * 82 2 02 914 = MADE IN U.S.A TEMPERED 5R Figure C-8. Photographs showing the layered morphology of representative corrosion products associated with the external metal loss features from Digs 1 - 4. DNV GL - OAPUS307KKRA (PP136049) C-22 December 4, 2015#

Page 439Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Figure C-9. Photograph of Soil 10000151761, as-received, that was removed near the failure location. The scale pictured is in mm. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 C-23#

Page 440Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Dig 1 Dig 2 70 150 00 100 110 Dig 13 Dig 14 100 710 120 130 Figure C-10. Photograph of as-received soil for Priority Digs 1 - 4. The scales pictured are in mm. DNV GL - OAPUS307KKRA (PP136049) C-24 December 4, 2015#

Page 441Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. Dig 1 Dig 2 Dig 3 Dig 4 Figure C-11. Photographs of liquids extracted from insulation collected by DNV GL during 2015 Priority Digs 1 - 4. December 4, 2015 DNV GL - OAPUS307KKRA (PP136049) C-25#

Page 442Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis APPENDIX D Corrosion Products Supplemental Analyses Density Testing DNV GL – OAPUS307KKRA (PP136049) December 4, 2015#

Page 443Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis CORROSION PRODUCT SUPPLEMENTAL ANALYSES DENSITY TESTING 1.0 BACKGROUND DNV GL was requested by PHMSA to perform density and magnetic permeability testing on corrosion product samples removed near the 2015 failure location on Line 901. These tests were not part of the original scope of the metallurgical analysis performed by DNV GL and so were added to the root cause analysis (RCA). The results from the density testing are summarized in this appendix, while the results of the magnetic permeability testing are summarized in Appendix E. Density testing was performed on a representative corrosion product sample, identified as Corrosion Product Sample 10000195318, removed near the 2015 failure. The sample was collected along the 6:00 o’clock orientation, 17.8 – 19.5 feet from the upstream girth weld. The objectives of the analysis were to determine the approximate density of the sample and compare the results with the density of steel. 2.0 TECHNICAL APPROACH The density testing was performed using a Model XS 205 balance manufactured by Mettler Toledo and equipped with a density determination kit. This equipment was used to calculate the density of the corrosion product sample based upon Archimedes’ principle, which states that “any body immersed in a fluid becomes lighter by an amount equal to the weight of the fluid that has been displaced.” The testing involved weighing the corrosion product in air and then in an auxiliary fluid; deionized (DI) water. The density of the corrosion product was then calculated using the following two equations: 𝐴 𝐴−𝐵 (𝜌𝑜 − 𝜌𝐿) + 𝜌𝐿 With compensation for air density 𝜌 = 𝜌 = 𝐴· 𝜌𝑜 𝐴−𝐵 Without compensation for air density Where: ρ = density of the sample A = weight of the sample in air Ρo = density of the auxiliary liquid B = weight of the sample in the auxiliary liquid ΡL = density of air (0.0012 g/cm3) DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 D-1#

Page 444Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 3.0 RESULTS ‎Figure D-1 is a photograph showing the sample that was removed from Corrosion Product Sample 10000195318 for the density testing.1 The sample was irregular in shape with a maximum length, width, and thickness of approximately 2.33 inches, 1.28 inches, and 0.534 inches, respectively. Note that one edge of the sample was cut with a Dremel tool equipped with a cutting blade. The sample was rigid, non-friable, and easily handled. ‎Figure D-2 and ‎Figure D-3 contain photographs showing the test setup for measuring the weight of the sample in air and in water, respectively. Due to the presence of air within the sample, the weight of the sample in water was taken only after all large air bubbles escaped from the surface of the sample. This process took approximately 30 minutes. ‎Table D-1 provides a summary of the weights obtained for the sample in both air and water. The weights are provided in both milligrams, which was the value reported by the balance, and in grams. As expected, the weight of the sample in air was greater than the weight of the sample in water. ‎Table D-2 summarizes the density values calculated for the sample, with and without compensation for the density of air, based on the measurements in ‎Table D-1. The values are very similar, ranging from 3.533 to 3.537 g/cm3 with and without compensating for the density of air. These values were compared to the density for mild steel (i.e. 7.87 g/cm3). The densities obtained for the corrosion product samples were approximately 45% of the density of low carbon steel. 1 Note: Only a portion of Corrosion Product Sample 10000195318 was needed for the density testing (i.e. the entire sample was not consumed for this testing). DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 D-2#

Page 445Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Table D-1. Summary of weights measured for Corrosion Product Sample 10000195318 during the density testing. Testing Environment Density of Testing Environment (g/cm3) Weight (mg) Weight (g) Air 0.0012 (ΡL) 39794.36 39.79436 (A) DI water 0.99819 1 (Ρo) 28563.52 28.56352 (B) 1 – Density of water at 20.2 oC (i.e. temperature measured at time of testing) per Table 7.7 in the operating instructions manual for Excellence Balances, XS Models. Where: Ρo = density of the auxiliary liquid A = weight of the sample in air ΡL = density of air (0.0012 g/cm3) B = weight of the sample in the auxiliary liquid Table D-2. Summary of density values calculated for Corrosion Product Sample 10000195318 based on the data in ‎Table D-1. Density of Corrosion Product (g/cm3) Compensation for Air Density Density of Mild Steel (g/cm3) 3.533 Yes 1 7.87 3 3.537 No 2 1– ρ = A A−B (ρo − ρL) + ρL; see ‎Table D-1 for A, B, ρo, and ρL 2– 𝜌 = 𝐴· 𝜌𝑜 𝐴−𝐵 ; see ‎Table D-1 for A, B, ρo, and ρL 3 – Density for 0.06% C steel. Metals Handbook Desk Edition, Second Edition 1998 p. 64. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 D-3#

Page 446Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. 12 16 20 24 28 12 16 20 24 28 12 16 2 1 ENGINEERING 2 PRODUCTS 262-00 24 32 40 48 56 8 16 24 32 40 48 56 8 16 24 32 4 Figure D-1. Photograph showing the portion of Corrosion Product 10000195318 that was used for the density testing. Scale is in inches Corrosion product dry measurement Figure D-2. corrosion product. Photograph showing the test setup used to measure the dry weight of the December 4, 2015 DNV GL - OAPUS307KKRA (PP136049) D-4#

Page 447Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. Mettler Toledo Model: XS205DU S/N: B309083908 V Calibration: BCCA Calibration Due: 31 May 2016 • Installation Plan/Contract Bo Repair Technician: Tim Bucke Date: 28 May 2015 Corrosion product wet measurement Figure D-3. Photographs showing the test setup (Left) and close-up of the sample (Right) during the submerged testing. DNV GL - OAPUS307KKRA (PP136049) D-5 December 4, 2015#

Page 448Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis APPENDIX E Corrosion Products Supplemental Analyses Magnetic Permeability Testing DNV GL – OAPUS307KKRA (PP136049) December 4, 2015#

Page 449Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis CORROSION PRODUCT SUPPLEMENTAL ANALYSES MAGNETIC PERMEABILITY TESTING 1.0 BACKGROUND DNV GL was requested by PHMSA to perform density and magnetic permeability testing on corrosion product samples removed near the 2015 failure location on Line 901. These tests were not part of the original scope of the metallurgical analysis performed by DNV GL and so were added to the root cause analysis (RCA). The results from the magnetic permeability testing are summarized in this appendix, while the results of the density testing are summarized in Appendix D. Two representative corrosion product samples and a steel plate sample, all removed from the pipe joint that contained the failure, were selected for the magnetic permeability testing. Identifications and descriptions of the selected samples are provided in ‎Table E-1 and photographs of the samples are provided in ‎Figure E-1 and ‎Figure E-2, respectively. The corrosion product samples are identified as Sample 10000195331 (i.e. sample exposed to crude oil) and Sample 10000195318 (i.e. dry sample). The steel sample selected for the testing is identified as Sample 10000195363. Only portions of the corrosion products and the steel sample were used for the testing. Specifically, two specimens were removed from each sample type described above. The testing did not consume all of the product/material available for the three samples. The objectives of the testing were to measure and compare the magnetic properties of the corrosion product samples with those measured for the plate steel. 2.0 EXPERIMENTAL DETAILS 2.1 Test Technique The test method used for the magnetic property testing was in accordance with industry- accepted standard ASTM A773 / A773M, “Standard Test Method for Direct Current Magnetic Properties of Low Coercivity Magnetic Materials Using Hysteresigraphs.” This test method provides instructions on how to produce plots of magnetic induction (B, magnetic flux density) vs. magnetic field strength (H), from which basic magnetic properties for soft and semi-hard materials are determined. The curves were evaluated to determine the following parameters (see ‎Figure E-3):  Coercive force (Hc) in Oersteds (Oe)  Residual magnetization (Br) in Gauss (G) DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 E-1#

Page 450Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis  Maximum magnetic field strength (Hmax) in Oe  Maximum induction (Bmax) in G Based on the above determine values, the following magnetic permeability values, which are dimensionless, were calculated using the equations provided below.  Initial magnetic permeability (µin)  Maximum magnetic permeability (µr max)  Magnetic permeability amplitude (µr amp) 𝐵 = 𝐵𝑖 + 𝐻 Relative magnetic permeability of material 𝐵 µ𝑟 𝑎𝑚𝑝 = 𝐻 Amplitude magnetic permeability 𝑑𝐵 µ𝑟 𝑑𝑖𝑓𝑓 = 𝑑𝐻 Differential magnetic permeability where: H = magnetic field strength [Oe] B = normal induction in test specimen [G] Bi = intrinsic induction in test specimen [G] 2.2 Specimen Preparation Based on the test technique identified for this analysis, bar specimens were prepared from the corrosion products and steel samples selected. Two specimens per sample were prepared (i.e. six total specimens). The initial proposed dimensions for the test specimens were 3-inches in length by 0.5- inches in width by 0.25-inches in height. The width and height selected for the specimens were based on minimum allowances identified for the testing. During the course of the specimen preparation, the actual heights achieved for the corrosion product specimens were greater than the minimum 0.25 inches previously selected. So as not to significantly alter the nature of the deposits from their field condition, it was decided to maximize the heights used for the corrosion product samples. ‎Figure E-4 contains schematics showing the test specimen geometries and ‎Table E-2 summarizes the dimensions of each test specimen. Details regarding the specific preparation steps for the corrosion product specimens and steel specimens are discussed below by sample type. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 E-2#

Page 451Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 2.2.1 Corrosion Product Specimens The following steps were performed to prepare the corrosion product specimens. First, the deposits samples were laser scanned from both surfaces, using a FaroArm™, to produce 2D and 3D renderings of the samples. The renderings were used to map the thickness of the samples and to determine the optimal locations from which to take specimens. The approximate locations of two test specimens were marked on the flattest surface of each corrosion product sample; see ‎Figure E-5. These markings were made to provide guidance for the initial cuts once the samples were embedded. The markings were made by first preparing a template of the desired specimen geometry (i.e. length vs. width) on a sheet of transparency paper. The transparency paper was cut along the lines of the template, except at the corners. The template was then placed on top of the corrosion product samples and a yellow paint marker was used to trace the cut edges transferring the specimen geometry onto the corrosion products. Consideration to the thickness of the band saw blade (i.e. 0.02-inches) was given when using the template to mark the samples. Alphabetical reference points (i.e. A, B, C, etc.) were marked on the corrosion product samples to identify the approximate end points of each cut line; see ‎Figure E-5. For ease of discussion, these alphabetical markings will be referenced when identifying sectioning locations on the embedded samples. Next, the corrosion product samples were embedded in epoxy. Rectangular plastic containers, approximately 7-inches in length by 5-inches in width and 3-inches in height, were used as molds to embed the corrosion product samples in a clear, two-part epoxy. Prior to placing the samples in their molds, a mold release agent was sprayed on the internal surfaces of the containers to facilitate the release of the embedded samples once the epoxy had cured. The flattest surface of each corrosion product sample was placed on top of plastic spacers positioned within the container. The plastic spacers were positioned so that the surface of each sample was parallel to the bottom of their respective molds. ‎Figure E-6 contains photographs showing the spacers used beneath the samples and the samples once they were placed in their respective molds. Next, the two-part epoxy was mixed and poured slowly into the molds; see ‎Figure E-7a. The sample molds (i.e. mounts) were then placed in a vacuum chamber in order to remove any trapped air within the uncured epoxy of the mounts; see ‎Figure E-7b. The mounts were allowed to cure overnight in a fume hood. A small fan was positioned to blow on the mounts during the curing process to remove the exothermal heat from the chemical DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 E-3#

Page 452Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis reaction of the epoxy during curing. Once cured, the mounts were removed from their respective plastic containers; see ‎Figure E-7c and ‎Figure E-7d. The embedded samples were then cut using a diamond band saw that was lubricated with ethylene glycol; see ‎Figure E-8a. All cuts were made slightly outside of the final desired dimension to ensure that a sufficient amount of material remained for any grinding needed to achieve the final specimen dimension. The embedded samples were first cut along the outer end markings (i.e. along Lines G-H and I-J shown in ‎Figure E-7c and d) in order to facilitate handling of the samples during grinding. The corrosion product surfaces exposed by these cuts were then re-embedded in epoxy. Next, a cut was made along Line C-D on both samples (see ‎Figure E-8b and c). These cuts were the only cuts that were made directly along the line marked on the embedded samples. The corrosion product surfaces exposed by these cuts were then re-embedded in epoxy. The two remaining pieces of the original embedment were then cut along Lines A-B and E-F followed by re-embedment of the exposed corrosion product surfaces in epoxy (see ‎Figure E-8b and c). Each side of the embedded samples was individually ground by hand in order to achieve the desired over-all dimensions; see ‎Figure E-9a. Grinding was carried out using 600 grit silicon carbide paper strips attached to a flat granite block and ethylene glycol as a lubricant. Based on the integrity of the samples, a protective epoxy layer was not necessary for the final test specimens. ‎Figure E-9b and c are photographs showing the final test specimens for Corrosion Product Samples 10000195318 and 10000195331. 2.2.2 Steel Plate Specimens The following steps were performed to prepare the steel specimens. The approximate locations of the final test specimens were marked on the steel plate. Two specimens: one in the axial direction and one in the transverse direction were sectioned from the plate; see ‎Figure E-10. Consideration was given to account for the thickness of the saw blade (i.e. 0.02-inches) when marking up the steel plate samples. The samples were cut using a band saw. All cuts were made slightly outside of the final desired dimension to ensure that a sufficient amount of material was left to allow for any milling needed to achieve the final specimen dimensions. Care was taken to achieve straight cuts. The samples were then milled at slow speeds to the desired dimensions. ‎Figure E-10b and c are photographs showing the final longitudinal and transverse steel specimens. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 E-4#

Page 453Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 2.3 Test Procedure A soft magnetic hysteresigraph tester, Model SMT-700, that was computer automated and manufactured by KJS Associates, Inc., was used to measure the magnetic properties of the test specimens. ‎Figure E-11 contains photographs showing the magnetic tester and a representative test setup for the magnetic property testing. Prior to testing, the prepared specimens were wrapped with insulating tape that was approximately 0.009 inches thick. A 30 gauge magnetic wire was then wound around each specimen. This wire served as a secondary induction winding. During testing, each specimen was positioned using a pole piece adapter in a KJS Associates Model YOKE-100 electro-magnet and clamped into a closed magnetic test circuit. A calibrated Hall probe was placed at the surface of the coil in order to measure the applied magnetic field (H). The secondary winding was then connected to the system fluxmeter to determine the flux density in the sample. Prior to the start of each test, the test specimen was demagnetized. Once the specimen was demagnetized, the specimen was then magnetized to a maximum applied field of 1000 Oe in the yoke fixture. The full four- quadrant B vs. H curve was then measured at room temperature. Each test specimen was also measured at lower applied fields to account for the typical field strengths of the magnetic flux leakage (MFL) tool. The higher permeability specimens were tested at an induction level of 12 kG, while the lower permeability specimens were tested to an induction level of 1000 G. 3.0 RESULTS Several magnetic parameters were measured and/or calculated for this analysis. The parameters include Hmax, Hc, Bmax, Br, µin, and µmax. Some values are relevant to MFL tools and some are not. The parameters of interest, as related to the Line 901 failure, include the magnetic permeability (µ) and the magnetic field strength (H). ‎Figure E-12 contains composite plots showing the magnetic induction (B, magnetic flux density) vs. magnetic field strength (H) for the six specimens when tested at strong magnetic fields (up to 1000 Oe). The curves for the steel specimens are shown in red, while the curves for the corrosion product specimens are shown in green (195318) and blue (195331). The plot to the left in ‎Figure E-12 shows the data up to a maximum magnetic induction of 25,000 G, while the plot to the right in ‎Figure E-12 shows the data up to a maximum magnetic induction of 2,500 G. A clear difference is apparent between the steel and corrosion product specimens in both plots. As shown, the saturation magnetization of the corrosion product specimens (i.e. Bmax) is approximately 10% of the saturation magnetization measured for the steel specimens. These differences indicate that the steel DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 E-5#

Page 454Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis specimens are more magnetic than the corrosion product specimens. No significant differences were observed between the corrosion product specimens. The magnetic properties of the specimens were measured when exposed to both a strong magnetic field and a lower magnetic field. ‎Figure E-13 through ‎Figure E-18 contain the individual plots of magnetic induction vs. magnetic field strength for each of the six tested specimens when exposed to the two magnetic fields. The magnetic properties determined from these curves and the data extracted from these curves are summarized in ‎Table E-3. The first column identifies the specimen and the second column shows the sample type. Columns 3 - 6 contain values that were extrapolated from the curves shown in ‎Figure E-13 through ‎Figure E-18 and Columns 7 and 8 provide the initial and maximum differential magnetic permeability values. The values shown in Columns 7 and 8 are the parameters of interest for this analysis. As seen, the permeability values determined for the corrosion product specimens are similar and are much lower than the values determined for the steel specimens (i.e. less than 2% of the values determined for the steel specimens). Differences were observed between the longitudinal and transverse steel specimens, with the transverse specimen exhibiting higher magnetic permeability values. The magnetic permeability values shown in Table E-3 were obtained for a field strength that exceeds the strongest MFL tools. Thus, values were also determined within the typical field strengths of high-field MFL tools. The typical field strength of high-field MFL tools range from 140 Oe to 180 Oe, reaching a maximum of around 200 Oe.1 Based on this information, smoothing approximation curves were used in the range from 50 to 300 Oe to calculate the amplitude magnetic permeability. The curves were based on the measured B-H curves for the six test specimens, but were corrected for residual magnetization and non-zero initial field data. The results of these analyses are summarized in ‎Table E-4. The first column identifies the specimen and the second column identified the sample type. Columns three through seven list the amplitude magnetic permeability at the following magnetic field strengths: 50, 100, 150, 200, and 250 Oe. As shown, the amplitude magnetic permeability values for the corrosion product specimens generally increase with increasing magnetic field strength. In contrast, the amplitude magnetic permeability values for the steel specimens decrease with increasing magnetic field strength. Overall, the magnetic permeability values for the corrosion product specimens are much lower than the values measured for the steel samples (i.e. less than 3% of the values determined for the steel specimens). 1 Development of Dual Field Magnetic Flux Leakage (MFL) Inspection Technology to Detect Mechanical Damage, PRCI Report, 2013. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 E-6#

Page 455Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 4.0 SUMMARY OF FINDINGS A summary of the findings are provided below.  The corrosion product specimens were less magnetic than the steel specimens.  No significant differences were determined for the magnetic properties of the specimens removed from the two corrosion product samples.  There were differences between the magnetic properties of the steel specimen in the axial (longitudinal) direction and the magnetic properties of the steel specimen in the transverse (circumferential direction).  At the field strengths typically associated with MFL tools, the magnetic permeability values of the corrosion product specimens were significantly lower than the magnetic permeability values of the steel specimens. The values for the corrosion product specimens were less than 5% of the values determined for the steel specimens. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 E-7#

Page 456Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Table E-1. Summary of samples selected for the magnetic permeability testing. Sample ID (Arcsset #) Sample Type Description Pipe Joint Distance D/S from GW 5930 (ft) o’clock orientation 10000195318 Corrosion Product Corrosion product from Feature 2 5930 17.8 – 19.5 ~ 6:24 10000195331 Corrosion product adjacent to leak location (Feature 4) 5930 33.50 ~4:24 10000195363 Plate Steel Counter clockwise fracture surface; small plate 5930 Table E-2. Summary of the dimensions, as reported by Magnetic Instruments, for the magnetic permeability specimens. Specimen Identifications Sample Type Average Length (in) Average Width (in) Average Height (in) 195318 – 3A 2.251 0.514 0.327 195318 – 3B 2.251 0.514 0.295 Corrosion Product 195331 – 2A 2.251 0.518 0.483 195331 – 2B 2.251 0.510 0.463 195363 – 2 Longitudinal 2.251 0.505 0.250 Plate Steel 195363 – 3 Transverse 2.251 0.504 0.249 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 E-8#

Page 457Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Table E-3. Summary of magnetic properties for tested specimens. Values determined from plots Calculated values Maximum Applied Magnetic Initial Maximum Field Residual Differential Differential Strength, Coercive force, Magnetization, Magnetic Magnetic Specimen Sample Hc Bmax" Br Permeability Permeability Identifications Type (Oe) (G) (G) U in U max 195318 - 3A 1020.6 86.27 1680 278 2.229 2.890 195318 - 3B Corrosion 1009.9 92.31 2050 352 2.511 3.325 195331 - 2A Product 1008.0 95.16 1802 280 1.870 2.545 195331 - 2B 998.3 104.4 1987 426 2.475 3.598 195363 - 2 Longitudinal 1018.2 6.66 22193 12750 149.0 1467 195363 - 3 Plate Steel Transverse 1011.5 6.5 22337 14147 177.3 1863 1 Values calculated from corrected curves. DNV GL - OAPUS307KKRA (PP136049) E-9 December 4, 2015#

Page 458Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Table E-4. Summary of magnetic permeability amplitudes! for the tested specimens at typical magnetic field strengths of MFL tools. Magnetic Field Strength Specimen (Oe) Identification Sample Type 50 100 150 200 250 195318 - ЗА 2.292 2.508 2.632 2.678 2.675 195318 - 3B Corrosion 2.615 2.812 2.954 3.045 3.086 195331 - 2A Product 1.872 2.018 2.154 2.249 2.298 195331 - 2B 2.491 2.795 3.027 3.165 3.206 195363 - 2 Longitudinal 332.3 179.5 125.2 96.99 79.55 195363 - 3 Plate Steel Transverse 330.5 178.7 124.7 96.64 79.28 1 Smoothing approximation curves used in range from 50 to 300 Oe to calculate amplitude magnetic permeability. Curves were corrected for residual magnetization and non-zero initial field. 2 Values shaded in light gray are amplitude magnetic permeability values. DNV GL - OAPUS307KKRA(PP136049) E-10 December 4, 2015#

Page 459Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Sample 10000195318 Sample 10000195331 Figure E-1. Photographs of Corrosion Product Sample 10000195318 (Top) and Corrosion Product Sample 10000195331 (Bottom), which were used for the magnetic permeability testing: Surface that was in contact with the pipe surface (Left) and Surface that was in contact with the coating (Right). Top images are flipped about the horizontal axis and bottom images are flipped about the vertical axis. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 E-11#

Page 460Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Figure E-2. Photograph of Steel Plate Sample 10000195363, which was used for the magnetic permeability testing DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 E-12#

Page 461Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. 2000 1900 Major Hysteresis Loop 1000 Normal Magnetizing 500 Curve Normal and Intrinsic Magnetic induction, G 0 500 Coercive Force, He 3000 - Normal inductan 7 -1500 Residual magnetization, B, -2000 1500 -1000 -500 500 3000 1500 Magnetic Field Strength, Oe Figure E-3. Plot of normal and intrinsic magnetic induction vs. magnetic field strength for a representative sample, showing how coercive force (Hc) and residual magnetization (B,) are determined. 0.249 - 0.250 inches (steel) Specimen Height 0.295 - 0.327 inches 195318) Specimen 0.50 = 0.02 inches Width 0.463 - 0.483 inches 195331) 0.50 + 0.02 inches Length Width 2.251 inches (a) Cross-sectional view (b) Longitudinal view Figure E-4. Schematics showing the dimensions for the magnetic permeability specimens: (a) cross-sectional view and (b) longitudinal view. DNV GL - OAPUS307KKRA (PP136049) E-13 December 4, 2015#

Page 462Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. a 8 Marked and cut transparency b Transparency overlaid on sample 3 12143 67 8 -3 inches Josinches 2 15 20 za 20 = " 1"33 MASE IN VED Sample 10000195318 d Sample 10000195331 ERING 2 3. -006 MADE IN DEA • an ad Figure E-5. Photographs showing the labeling process for the corrosion product samples: (a) transparency template, (b) transparency overlaid on Sample 10000195331 for marking, (c) labeled Sample 10000195318, and (d) labeled Sample 10000195331. DNV GL - OAPUS307KKRA (PP136049) E-14 December 4, 2015#

Page 463Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. a Sample 10000195318 b Sample 10000195331 195331 195318 5 Plastic spacers Plastic spacers C d Sample 10000195318 Sample 10000195331 Figure E-7 Figure E-6. Photographs showing preparation of the corrosion product samples for embedment: (a) plastic rod used to elevate Sample 10000195318, (b) plastic rod used to elevate Sample 10000195331, (c) Sample 10000195318 in the mold, and (d) Sample 10000195331in the mold. December 4, 2015 DNV GL - OAPUS307KKRA (PP136049) E-15#

Page 464Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Molds b Vacuum chamber Epoxy Embedded sample Sample 10000195331 Sample 10000195318 Sample 10000195318 d Sample 10000195331 Figure E-7. Photographs showing the embedment process for the corrosion product samples: (a) embedment molds after samples were covered with epoxy, (b) mold samples in vacuum chamber to remove bubbles, (c) embedded Sample 10000195318, and (d) embedded Sample 10000195331. DNV GL - OAPUS307KKRA (PP136049) E-16 December 4, 2015#

Page 465Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis a b I G G I 9... ------ S_ E F E F H H Sample 10000195318 Sample 10000195331 Figure E-8. Photographs showing the cutting process for the corrosion product samples: (a) sample cutting on diamond saw, (b) Sample 10000195318 after cuts were made along Lines G-H, I-J, and C-D and (d) Sample 10000195331 after cuts were made along Lines G-H, I-J, C-D (yellow dashed line), A-B (green dashed line), and E-F (red dashed line). Refer to Figure E-7 for the line identifications. December 4, 2015 DNV GL - OAPUS307KKRA (PP136049) E-17#

Page 466Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. a b Sample 10000195318 Magnetic permeability samples C Sample 10000195331 Magnetic permeability samples 145331 195321 92re02g00 g * 3E +E 02 , 2 3 BZ TZ 0E + BE +2 02 62 +2.02 42 i= 02 BX +E 03 MADE NUSA TEMPERED 5R 9 g 1 24 40 31 90 5 • 1t за 23 10 19 08. 6 1 18 24 10 48 38 7 10 24 3°40 48.55 8 32 MADE IN USA TEMPERED Figure E-9. Photographs showing the finishing process for the corrosion product samples: (a) sample finishing with SiC paper, (b) Sample 10000195318 magnetic permeability samples, and (c) Sample 10000195331 magnetic permeability samples. DNV GL - OAPUS307KKRA (PP136049) E-18 December 4, 2015#

Page 467Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Plate Sample 10000195363 a Arcane b Longitudinal magnetic permeability sample Transverse magnetic permeability sample 52 72 0z 831202 3L 82 p2 02 9% 82p2. 02912 % 82 pz 0z 31 8 MADE IN U.S.A. TEMPERED MADE IN U,S A TEMPERED 4 • 16 24 22 30 48 00 5 8 10 24 22 40 49 56 6 в 18 24 32 40 48 56 4 8 16 24 32 40 48 5€ ฿ 10 24 32 6 40 40 68 в. 18 24 32 s0 48 56 в 16 24 32 д0 48 Бб 8 16 24 32 40 48 5 Figure E-10. Photographs showing the plate samples used for the magnetic permeability testing: (a) Plate Sample 1000095363 showing location where samples were removed, (b) Longitudinal magnetic permeability sample, and (c) Transverse magnetic permeability sample. DNV GL - OAPUS307KKRA (PP136049) E-19 December 4, 2015#

Page 468Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. SMT-700 System Closed Magnetic Circuit Specimen Coils Poles Figure E-11. Photographs showing the tester and a representative test setup for the magnetic property testing. December 4, 2015 DNV GL - OAPUS307KKRA (PP136049) E-20#

Page 469Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis ZOUU Steel specimens 20000 2000 15000 - Steel specimens 1500 10000 1000 5000 500 00 0 5000 Corrosion product specimens -500 Corrosion product -10000 -1000 specimens -15000 -1500 -20000 -2000 -2500 -1000 -750 -500 -250 H (Oe) 250 500 750 1000 -1000 -750 -500 -250 H (Oe) 250 500 750 1000 MS22802 363-2 LONG.csv MS22803 363-3 TRANS.csv MS22804 331-2 AB.csv MS22802 363-2 LONG.csv MS22803 363-3 TRANS.CSV MS22804 331-2A AB.csv MS22805 331-2B BD.csv MS22806 318-3A AB.csv MS22807 318-3B CD.csv MS22805 331-2B BD.csv MS22806 318-3A AB.csv MS22807 318-3BCD.csv Figure E-12. Composite plots of magnetic induction (B, magnetic flux density) vs. magnetic field strength (H) for the six specimens when tested at strong magnetic fields (up to 1000 Oe): Overall plot (Left) and close-up of overall plot with magnetic induction range of 2,500 G. DNV GL - OAPUS307KKRA (PP136049) E-21 December 4, 2015#

Page 470Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. Strong Magnetic Field Low Magnetic Field 1000 1500 750 1000 500 500 250 -500 250 -500 -1000 -750 -1500 -1000 -1000 H(de) 1000 -500 -300 -200 H (Ce) 500 Figure E-13. Plots of magnetic induction (B, magnetic flux density) vs. magnetic field strength (H) for Corrosion Product Specimen 195318 - 3A: Strong magnetic field of 1000 Oe (Left) and Low magnetic field of 460 Oe (Right). DNV GL - OAPUS307KKRA (PP136049) E-22 December 4, 2015#

Page 471Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. 2000 Strong Magnetic Field 1000 Low Magnetic Field 1:00 - 750 1000 500 500 250 -500 -250 -1000 500 -1500 - -750 -2000 -1000 -1000 -750 500 -250 (Ge) -300 -200 -100 H (0e) 10D 200 300 Figure E-14. Plots of magnetic induction (B, magnetic flux density) vs. magnetic field strength (H) for Corrosion Product Specimen 195318 - 3B: Strong magnetic field of 1000 Oe (Left) and Low magnetic field of 360 Oe (Right). DNV GL - OAPUS307KKRA (PP136049) E-23 December 4, 2015#

Page 472Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. 2000 Strong Magnetic Field 1000 Low Magnetic Field 1500 750 1000 500 500 250 0 +500 -250 -1000 500 750 -1500 -1000 -2000 -1000 -750 -500 -250 0 250 500 750 1000 -500 400 -300 -200 -100 100 200 300 400 500 Figure E-15. Plots of magnetic induction (B, magnetic flux density) vs. magnetic field strength (H) for Corrosion Product Specimen 195331 - 2A: Strong magnetic field of 1000 Oe (Left) and Low magnetic field of 460 Oe (Right). DNV GL - OAPUS307KKRA (PP136049) E-24 December 4, 2015#

Page 473Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. 2000 Strong Magnetic Field 1000 Low Magnetic Field 1500 750 1000 500 500 250 -900 -250 -1000 -500 - 1500 -750 -2000 -1000 -1000 -750 -500 -250 H (Ce) 250 500 750 1000 -300 -200 -100 H (Ge) 100 200 300 Figure E-16. Plots of magnetic induction (B, magnetic flux density) vs. magnetic field strength (H) for Corrosion Product Specimen 195331 - 2B: Strong magnetic field of 1000 Oe (Left) and Low magnetic field of 320 Oe (Right). DNV GL - OAPUS307KKRA (PP136049) E-25 December 4, 2015#

Page 474Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. Strong Magnetic Field Low Magnetic Field 20000 10000 15000 10000 5000 5000 B (G) g -5000 -10000 5000 -15000 10000 -20000 - -1000 HOei -15 -10 -5 15 Figure E-17. Plots of magnetic induction (B, magnetic flux density) vs. magnetic field strength (H) for Steel Specimen 195363-2 (Longitudinal): Strong magnetic field of 1000 Oe (Left) and Low magnetic field of 15 Oe (Right). DNV GL - OAPUS307KKRA (PP136049) E-26 December 4, 2015#

Page 475Line 901 Release (5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. Strong Magnetic Field Low Magnetic Field 20000 10000 — 150D0 50000 5000 5000 B (G) ฿ (G) -5000 -10000 5000 -15000 - -10000- 220000- -1000 -10 H(de) Figure E-18. Plots of magnetic induction (B, magnetic flux density) vs. magnetic field strength (H) for Steel Specimen 195363-3 (Transverse): Strong magnetic field of 1000 Oe (Left) and Low magnetic field of 13 Oe (Right). DNV GL - OAPUS307KKRA (PP136049) E-27 December 4, 2015#

Page 476Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis APPENDIX F Statistically Active Corrosion Assessment DNV GL – OAPUS307KKRA (PP136049) December 4, 2015#

Page 477Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis STATISTICALLY ACTIVE CORROSION ASSESSMENT 1.0 BACKGROUND This appendix provides a summary of the Statistically Active Corrosion (SAC) assessment completed on Line 901. The pipeline is comprised of 24-inch diameter by 0.344 inch wall thickness, API 5L Grade X65 line pipe steel that was manufactured by Nippon Steel and contains a high frequency (HF) electric resistance welded (ERW) longitudinal seam. It was installed in 1990 and is approximately 10.87 miles in length, spanning between Las Flores Station on the U/S end and Gaviota Station on the D/S end. The normal operating pressure and maximum discharge pressure (MDP) for the line are 616 psig and 1,025 psig, respectively. These pressures correspond to 33% and 55% of the specified minimum yield strength (SMYS), respectively. The pipeline was inspected by Rosen with a magnetic flux leakage (MFL) in-line inspection (ILI) tool in June 2007, July 2012, and May of 2015. 1.1 Objective The primary objective of the SAC assessment was to estimate the localized corrosion growth rates on Line 901 based on a comparison of the 2007 MFL and 2012 MFL ILI surveys. 1.2 Scope of Work In order to determine corrosion growth rates, DNV GL conducted its SAC assessment of changes in reported metal loss between the un-clustered1 metal loss reported in the 2007 MFL and 2012 MFL ILI surveys. Statistically (at a 95% confidence level) high growth areas were then reviewed in the ILI raw signal data sets to determine the actual hotspots of corrosion growth on the pipeline and the rates of that growth. 2.0 TECHNICAL APPROACH The following tasks were conducted within the SAC assessment:  Task 1: Data Alignment and Preparation of the Input Data  Task 2: Comparison of ILI-Reported and Field-Measured Depths  Task 3: Statistically Active Corrosion Assessment of the Inspection Data Sets  Task 4: Compilation and Review of the Statistical Screening Results  Task 5: Application of Corrosion Growth Rates 1 Clustering is defined as combining multiple indications within a specific distance. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 F-1#

Page 478Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 2.1 Task 1 – Data Alignment & Preparation of the Input Data DNV GL aligned the 2007 MFL and 2012 MFL un-clustered metal loss inspection data sets prior to performing the statistical analysis on individual pipe joints. The data sets were also matched in sensitivity to ensure that standard ILI survey instrument differences were considered during the screening process. The matching was conducted using unity plots based on pits reported in both inspections and by comparing raw signals (“boxed” data) from each ILI survey using the software provided by the ILI vendor. The unity plots were used to identify overall biases between the inspections. The box data were used to determine whether a sensitivity (depth) adjustment factor should be applied to either ILI data set. 2.2 Task 2 – Comparison of ILI-Reported and Field-Measured Depths To aid in determining whether any adjustments were warranted for the most recent ILI inspection, the field-measured depths were compared with the ILI-reported depths. Axial and circumferential location information, as reported by the ILI for a given feature, was used to define the search area for a corresponding anomaly within the provided excavation results. Unity plots were produced to graphically review the results, which were then used within the SAC assessment. 2.3 Task 3 – Statistically Active Corrosion Assessment DNV GL compared the two sets of ILI data (in this assessment, the 2007 MFL and 2012 MFL inspections) using its SAC assessment methodology. The SAC methodology identified pipeline locations for which the changes between the ILI data indicate a likelihood of active corrosion growth. Those locations that exceeded a desired level of confidence (95% confidence interval) were identified as statistically active locations. The SAC methodology is applied on a joint-by-joint basis. Internal and external features were grouped together for the SAC assessment. This is typically done when the ID/OD discrimination is suspect, especially for deeper (more significant) features. Potential locations of corrosion activity were identified from average depths, maximum depths, and metal loss anomaly frequency perspectives. If a joint exhibits a statistically significant increase in the average or maximum reported metal loss depth, it is identified as either a SAC Mean or SAC Max respectively: DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 F-2#

Page 479Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis  SAC Mean – identified locations and quantifies the corrosion growth where there is evidence of a statistically significant change between the average (mean) metal loss depths in each ILI survey.  SAC Max – identified locations and quantifies the corrosion growth where there is evidence of a statistically significant change between the deepest metal loss calls in each ILI survey. Estimated corrosion rates were calculated using the difference in the means or maximums and the time interval between inspections. To be conservative, DNV GL uses a default growth rate based on ILI tolerance, the nominal wall thickness, and the time frame between both inspections (see Equation (1)). Joints that were neither SAC Mean nor SAC Max were assigned this calculated minimum corrosion growth rate, CGRMin. 𝐶𝐺𝑅𝑀𝑖𝑛 = 0.5 × 𝐼𝐿𝐼𝑇𝑜𝑙𝑒𝑟𝑎𝑛𝑐𝑒 × 𝑛𝑜𝑚𝑖𝑛𝑎𝑙 𝑊𝑇 𝐷𝑎𝑡𝑒𝑅𝑒𝑐𝑒𝑛𝑡− 𝐷𝑎𝑡𝑒𝑃𝑟𝑒𝑣 (1) 2.4 Task 4 – Compilation and Review of Statistical Growth Results Following the statistical assessment, DNV GL performed a manual review of the signal data on selected pipe joints to:  Locate areas of growth that may not have been identified via the statistical analysis.  Confirm areas identified as containing statistically significant growth. Manually reviewed pipe joints were selected based on a number of characteristics determined from DNV GL’s experience from similar projects. Characteristics used to select joints for manual review include joints with:  The highest SAC Mean or SAC Max growth rates  Statistically significant differences in the number of SAC counts  The most unmatched metal loss features (Orphan and non-Orphan) in both 2012 and 2014  The largest maximum depth in 2014, both with and without a corresponding 2012 feature  The largest difference in maximum depths  The largest difference in depth between matched (one-to-one) pits DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 F-3#

Page 480Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Joints identified based on the characteristics above and the areas immediately upstream and downstream of these joints were reviewed for signs of growth by manually comparing the ILI signal data from each inspection. Each joint manually reviewed was classified per ‎Table F-1. Table F-1. Manual ILI Signal Review Classifications. Classification Description Probable Significant Growth The ILI signals appear to demonstrate a large difference between each tool survey for depth, length, or width. Possible Growth The ILI signals appear to demonstrate a difference between each tool survey, but this difference is not as pronounced as “Probable Significant Growth”. Unlikely Growth The ILI signals do not appear to demonstrate a difference between each tool survey. 2.5 Task 5 – Application of Corrosion Growth Rates The time to reach the scenarios, as defined below, was deterministically calculated for each metal loss indication from the 2012 ILI survey using the SAC growth rates. Internal and external indications were evaluated together in the SAC assessment and a single corrosion growth rate was calculated (see Section ‎2.3) for each pipe joint or the default growth rate was assigned. The estimated corrosion rate for each joint after the manual ILI signal review (i.e., after the estimated rate for joints identified as “Unlikely Growth” were adjusted to the determined minimum threshold rate) was applied to all metal loss indications reported within that joint. Metal loss indications that were reported to be repaired prior to the 2015 ILI were not included in the calculations. The following scenarios were evaluated:  Scenario 1 o The reported depth plus the stated tool tolerance exceeds 80% WT  Scenario 2 o The reported length and depth lead to a predicted failure pressure of 1.39 × MOP as calculated using modified (0.85 dL) B31G  (i.e. P0.85dL ≤ 1.39 × MOP) o The growth is assumed to occur only in depth (i.e., the length remains constant) DNV GL calculated a deterministic timeframe for each of the metal loss indications and identified the minimum predicted timeframe for each joint according to the two scenarios. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 F-4#

Page 481Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis The estimated timeframe for features located on joints that were classified as “Unlikely Growth” via the manual review process were used as-is. To be as consistent as possible with Plains’ re-assessment interval approach, the estimated timeframe for features on joints that were found to exhibit growth or are on joints that were not manually reviewed were multiplied by a factor of 0.72. Those features that were predicted to meet any of the scenarios within five years of the 2012 inspection were identified. 3.0 RESULTS The results from the assessment are presented in the following subsections. 3.1 Task 1 – Data Alignment and Preparation of Input Data The joint listings for the 2007 MFL and 2012 MFL ILI surveys were aligned, and the joints successfully matched using the reported joint lengths and odometer locations. Prior to the statistical review, the pit-to-pit matching algorithm was used to identify one-to- one matches between the 2007 and 2012 reported metal loss to aid in evaluating whether there is any bias in the ILI data. ‎Figure F-1 shows a plot of the matched metal loss3. The 95% confidence interval between the ratio of the two data sets is [1.09,1.21], which indicates the 2012 MFL as-reported data, on average, is deeper than the as-reported 2007 MFL data. DNV GL also compared raw signals from each ILI survey using the software provided by the ILI vendor in areas where the signal data did not show any evidence of change to identify any systematic differences between the sizing algorithms (sensitivities) used for each ILI. Results of the raw signal comparison are shown in ‎Figure F-2. No adjustment was made to either ILI data set based on either these comparisons. 2 Other factors of safety could also be employed to account for uncertainty. 3 A total of 167 one-to-one matches were identified. There were a total of 3618 un-clustered metal loss boxes reported by the 2007 MFL and 1705 un-clustered metal loss boxes reported by the 2012 MFL. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 F-5#

Page 482Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 100% 90% 80% 70% 60% 40% 30% 20% 10% 0% 0% 10% 100% Figure F-1. Comparison of 2007 and 2012 Internal and External Metal Loss Non-Clustered ILI Depths. December 4, 2015 DNV GL - OAPUS307KKRA (PP136049) F-6#

Page 483Plains All American Pipeline, L.P. Line 901 Release 5/19/15) Technical Root Cause Analysis 100% 90% 80% 70% 60% 50% 40% 30% • 20% 8000 • • 10% 0% 0% To 20 or i Repare Do i 006 00. 02 Figure F-2. Comparison of 2007 and 2012 Internal and External Metal Loss Non-Clustered ILI Depths in Areas of Unlikely Growth. 3.2 Task 2 - Comparison of ILI-Reported and Field-Measured Depths Excavation records from the 2007 MFL and the 2012 MFL response program were provided to DNV GL and were used to gauge tool performance and determine whether an additional sensitivity adjustment factor was warranted. Based on records provided to DNV GL, 15 field- measured depths were matched to metal loss reported in the 2007 MFL and 52 field- measured depths were matched to metal loss reported in the 2012 MFL (see Figure F-3 and Figure F-4, respectively). Further discussion of the comparison between the field and ILI are in the main report. No adjustment was made to either ILI data set based on these comparisons. DNV GL - OAPUS307KKRA (PP136049) December 4, 2015 F-7#

Page 484Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 100 ILI under reported 90 80 70 Field-Measured Depth, %WT 60 50 y = 0.1508x + 33.165 R² = 0.0475 40 30 20 33.3% within ±10% WT 66.7% outside ±10% WT 40.0% under reported by ILI 60.0% over reported by ILI 13.3% under reported & outside ±10% WT 53.3% over reported & outside ±10% WT 10 ILI over reported 0 0 10 20 30 40 50 60 70 80 90 100 2007 ILI-Reported Depth, %WT Figure F-3. Comparison of Field-Measured (following 2007 ILI) to 2007 ILI-Reported Depths. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 F-8#

Page 485Plains All American Pipeline, L.P. Line 901 Release 5/19/15) Technical Root Cause Analysis 100 ILI under reported 90 80 = 0.4988x ÷ 23.967 70 R: = 0.1716 60 50 0/ • • 30 • • 55 8% within +10% WT 20 44 2% outside +10% WT 55 8% under reported by ILl 36 5% over reported by lll 17 3% over reported & outside +10% WT 26.9% under reported & outside 10% WT 10 ILI over reported 10 90 100 Figure F-4. Comparison of Field-Measured (following 2012 ILI) to 2012 ILI-Reported Depths. 3.3 Task 3 - Statistically Active Corrosion Assessment As noted earlier, a corrosion growth rate was calculated for each pipe joint based on the results of the statistical assessment using either the differences in the ILI reported mean (average) or maximum depths between inspections. There were 11 identified joints that had SAC and the highest estimated corrosion growth rate prior to the manual signal review was 29.1 mpy. Calculated corrosion rates of joints were adjusted to the minimum threshold rate determined using Equation (1) if the calculated corrosion rates were less than the minimum rate. The nominal wall thicknesses taken into consideration for Equation (1) were 0.344- DNV GL - OAPUS307KKRA (PP136049) December 4, 2015 F-9#

Page 486Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis inch and 0.500-inch, resulting in minimum threshold rates of 3.5 and 5.0 mpy, respectively.4 3.4 Task 4 – Compilation and Review of Statistical Screening Results A total of 169 pipe joints were selected for manual ILI signal review based on the characteristics described previously. The 169 manually reviewed joints included a single SAC Max joint and ten SAC Mean joints. The other 158 manually reviewed joints where selected based on criteria listed in Section ‎2.4. Of the 169 joints manually reviewed, 87 joints (51%) were classified as “Unlikely Growth”, 53 joints (31%) were classified as “Possible Growth”, and 29 joints (17%) were classified as “Probable Significant Growth”. In general, the manual review confirmed that the screening process (including the statistical analysis and selection criteria) identified joints with the potential for growth, but it also identified joints where little change was evident. This is not uncommon as differences in analysis algorithms can lead to what appears to be growth based on reported depths where none is observed in the signal data. The complete manual review results are tabulated in Section 5.0 of this appendix. The results of the manual ILI signal review were superimposed on the calculated corrosion growth rates, which are displayed in ‎Figure F-5. Estimated rates for joints identified as “Unlikely Growth” were adjusted down to the determined minimum growth rate calculated using Equation (1) for the applicable WT. The estimated rates for “Possible Growth” joints were adjusted to the rate based on the differences of the mean depths. “Probable Significant Growth” joints were adjusted to the rate based on the difference of the means or maximums with the highest confidence level. 4 Minimum corrosion growth rates were calculated based on the ILI survey dates (June 1, 2007 and July 3, 2012) and were rounded up to the nearest 0.5 mpy. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 F-10#

Page 487Line 901 Release 5/19/15) Technical Root Cause Analysis Plains All American Pipeline, L.P. 35 • Estimated CGR • Probable Significant Growth (29 joints) 30 • Possible Growth (53 joints) X Unlikely Growth (83 joints) 25 20 Estimated Corrosion Growth Rate, mpy 15 10 10000 20000 ILl Odometer, ft 30000 40000 50000 60000 Figure F-5. Estimated Corrosion Growth Rate per Joint after Manual Review. After incorporating the manual signal review results, the highest estimated corrosion growth rate for joints occurs at a "Probable Significant Growth" joint and is 29.1 mpy. 3.5 Task 5 - Application of Corrosion Growth Rates There are eight joints with a predicted 70% minimum timeframe to scenario less than or equal to five years the maximum allowed by 49 CFR 195.452(j)(3)) using the approach described in Section 2.5 when metal loss indications repaired prior to the 2015 ILI are taken into account. Three of these joints have a 70% timeframe less than three years (the specified reassessment interval); one was excavated after the 2015 ILI. A list of all joints with minimum timeframes less than or equal to five years is included in (Table F-2. December 4, 2015 DNV GL - OAPUS307KKRA (PP136049) F-11#

Page 488Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Table F-2. Joints with 70% of the Predicted Timeframes Less Than or Equal to Five Years. Joint ID Odometer, ft Manual Review † Rate, mpy Max. Depth, % WT Scenario Min. Time to Scenario, yrs 70% Time, yrs 4220 15065.38 PS 12.2 47 80% WT 6.5 4.6 5930 21351.11 PS 14.9 45 80% WT 5.8 4.0 8280 30276.76 PS 24.3 46 80% WT 3.4 2.4 9430 34027.19 PS 16.9 35 80% WT 7.1 5.0 11060 39808.08 PS 16.9 36 80% WT 6.9 4.8 12850 46264.57 PS 17.6 37 80% WT 6.5 4.5 13210 47401.55 PS 19.6 49 80% WT 3.7 2.6 14470‡ 51618.37 PS 29.1 53 80% WT 2.0 1.4 † P = P, PS = Probable Significant Growth ‡ Features at 2015 ILI odometer 51640.00 and 51640.27 ft were repaired with a composite sleeve on June 4, 2015 [Ref 200]; the features on this joint in the 2012 ILI are between 51640.14 and 51642.68. The maximum depth in the field was measured at 65% WT. 4.0 SUMMARY REMARKS The statistically active corrosion (SAC) methodology was developed with the objective to identify pipeline locations for which ILI data indicate a likelihood of corrosion growth. For selected joints with the potential for significant growth, a manual review of the ILI signal data was performed to determine whether the likely growth is evident in the ILI signal or a result of ILI sensitivity differences. Based on the results of the corrosion growth screening and probabilistic assessment, DNV GL has developed the following conclusions:  There does not appear to be a systematic bias between the 2007 and 2012 ILI reported depths; no adjustments to reported depths were applied prior to the statistical analysis.  Based on the SAC analysis, when repairs prior to the 2015 ILI are accounted for, 11 joints out of 314 joints with metal loss indications (3.5% of joints with metal loss) were identified as potential growth locations. These are referred to as SAC joints.  One hundred and sixty nine pipe joints were subjected to manual ILI signal review. DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 F-12#

Page 489Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis o Of the 169 joints, 87 joints (51%) were classified as “Unlikely Growth”, 53 joints (31%) were classified as “Possible Growth”, and 29 joints (17%) were classified as “Probable Significant Growth”.  The highest estimated corrosion growth rate after adjusting the rates based on the manual signal review is 29.1 mpy. o This rate occurs on a joint repaired with a composite sleeve on June 4, 2015  There are eight joints with a predicted 70% minimum timeframe less than or equal to five years. o Features identified to have been repaired prior to the 2015 ILI were not included in the growth projections.  The joint that failed in 2015 (Joint 5930) is predicted to: o Have a SAC rate (15 mpy); a value between the rate used in the CGAR process (8 mpy) and the rate obtained via pit-to-pit matching (18 mpy) o Reach 80% WT in 5.8 years (70% of that time is 4.0 years)  The SAC process predicts a reassessment interval on the order of the reassessment interval utilized by Plains for Line 901 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 F-13#

Page 490Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis 5.0 TABULATED MANUAL REVIEW RESULTS Joint ID Odometer, ft Review Selection Criteria Manual Review † Review Comments 70.01 111.52 Estimated Corrosion Growth Rate U 3.5 80 128.88 Estimated Corrosion Growth Rate U 3.5 120 255.86 Estimated Corrosion Growth Rate 250 566.12 Estimated Corrosion Growth Rate 260 606.11 Estimated Corrosion Growth Rate 290 654.88 Deepest Features from 2012 Survey that were Matched Original Rate, mpy U Located near GW. Adjusted Rate, mpy 3.5 3.5 U 5.0 5.0 U 3.5 3.5 U 3.5 3.5 3.5 3.5 420 954.76 Deepest Features from 2012 Survey that were Matched U 3.5 3.5 480 1158.59 490 1198.74 500 1238.87 2012 Joints with Most Unmatched Metal Loss (non- Orphan) 2007 Joints with Most Unmatched Metal Loss (non- Orphan) Estimated Corrosion Growth Rate U 3.5 U 3.5 U 3.5 3.5 3.5 3.5 510 1279.00 520 1319.13 530 1359.26 540 1399.33 3.5 3.5 3.5 3.5 550 1439.41 560 1474.52 570 1514.60 580 1554.67 590 1594.69 2007 Joints with Most Unmatched Metal Loss (non- Orphan) U 3.5 2012 Joints with Most Unmatched Metal Loss (non- Orphan) U 3.5 2012 Joints with Most Unmatched Metal Loss (non- Orphan) U 3.5 2012 Joints with Most Unmatched Metal Loss (non- Orphan) U 3.5 2007 Joints with Most Unmatched Metal Loss (non- Orphan) U 3.5 3.5 2012 Joints with Most Unmatched Metal Loss (non- Orphan) U 3.5 2012 Joints with Most Unmatched Metal Loss (non- Orphan) U 3.5 2012 Joints with Most Unmatched Metal Loss (non- Orphan) U Located near GW. Deepest Features from 2012 Survey that were Matched P Located near GW. 3.5 3.5 3.5 3.5 3.5 3.5 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 F-14#

Page 491Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Joint ID Odometer, ft Review Selection Criteria Manual Review † Review Comments 600 1634.65 Original Rate, mpy 3.5 Adjusted Rate, mpy 4.5 610 1674.68 620 1714.66 2007 Joints with Most Unmatched Metal Loss (non- Orphan) P Located near GW. 2012 Joints with Most Unmatched Metal Loss (non- Orphan) U 3.5 3.5 2012 Joints with Most Unmatched Metal Loss (Orphan) U 3.5 3.5 650 1821.71 2012 Joints with Most Unmatched Metal Loss (Orphan) U 3.5 3.5 710 2060.81 2012 Orphan (no ML reported 2011) Joints with Largest Maximum Depth U 3.5 3.5 720 2100.91 730 2140.90 740 2180.93 750 2220.99 760 2260.83 Largest Difference Between Matched Pits (1:1 matches ONLY) Largest Difference Between Matched Pits (1:1 matches ONLY) Largest Difference Between Matched Pits (1:1 matches ONLY) Largest Difference Between Matched Pits (1:1 matches ONLY) Deepest Features from 2012 Survey that were Matched U 3.5 U 3.5 U 3.5 P Located near GW. U Located near GW. 3.5 3.5 3.5 3.5 3.5 3.5 3.5 770 2300.87 Deepest Features from 2012 Survey that were Matched U Located near GW. 3.5 3.5 970 2989.81 980 3029.58 2012 Joints with Most Unmatched Metal Loss (non- Orphan) 2012 Joints with Most Unmatched Metal Loss (non- Orphan) P 3.5 1050 3308.39 1070 3388.53 Largest Difference Between Matched Pits (1:1 matches ONLY) Deepest Features from 2012 Survey that were Matched U 3.5 New growth not visible in the previous inspection. U Located near GW. U 3.5 3.5 3.5 3.5 3.5 3.5 1350 4491.84 2012 Joints with Most Unmatched Metal Loss (Orphan) U Located near GW. 3.5 3.5 1360 4531.86 1370 4571.90 Largest Difference Between Matched Pits (1:1 matches ONLY) Deepest Features from 2012 Survey that were Matched 3.5 U Located near GW. U 3.5 3.5 3.5 1560 5305.93 2012 Orphan (no ML reported 2011) Joints with Largest Maximum Depth U 3.5 3.5 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 F-15#

Page 492Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Joint ID Odometer, ft Review Selection Criteria Manual Review † Review Comments 1570 5346.04 Deepest Features from 2012 Survey that were Matched 1700 5794.37 Deepest Features from 2012 Survey that were Matched P Located outside of repaired area. P Located near GW. Original Rate, mpy 3.5 3.5 Adjusted Rate, mpy 3.5 4.5 1990 6903.85 Deepest Features from 2012 Survey that were Matched U 3.5 3.5 2020 7016.64 2007 Joints with Most Unmatched Metal Loss (Orphan) U Located near GW. 3.5 3.5 2170 7617.42 Deepest Features from 2012 Survey that were Matched PS 2210 7777.78 2012 Orphan (no ML reported 2011) Joints with Largest Maximum Depth P 2640 9423.94 2012 Orphan (no ML reported 2011) Joints with Largest Maximum Depth P New growth not visible in the previous inspection. Located near GW. Feature appears to be growing wider. New growth not visible in the previous inspection. 3.5 3.5 3.5 5.1 3.5 5.9 2830 10080.65 2012 Joints with Most Unmatched Metal Loss (Orphan) P New growth not visible in the previous inspection. Located near GW. 3.5 3.7 2860 10174.60 2007 Joints with Most Unmatched Metal Loss (Orphan) U 3.5 3.5 2960 10556.14 Deepest Features from 2012 Survey that were Matched P 3.5 3.5 3420 12376.08 3810 13831.31 2012 Joints with Most Unmatched Metal Loss (non- Orphan) 2012 Joints with Most Unmatched Metal Loss (non- Orphan) U 3.5 U 3.5 3.5 3.5 4080 14741.05 Estimated Corrosion Growth Rate P 6.8 6.8 4150 14921.60 Deepest Features from 2012 Survey that were Matched P 3.5 4.6 4160. 01 14960.84 Deepest Features from 2012 Survey that were Matched U 5.0 5.0 4160. 02 14968.27 Deepest Features from 2012 Survey that were Matched U 5.0 5.0 4210 15025.35 Deepest Features from 2012 Survey that were Matched P 3.5 3.5 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 F-16#

Page 493Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Joint ID Odometer, ft Review Selection Criteria Manual Review † Review Comments 4220 15065.38 Deepest Features from 2012 Survey that were Matched PS Original Rate, mpy 3.5 Adjusted Rate, mpy 12.2 4240 15145.46 2012 Joints with Most Unmatched Metal Loss (Orphan) New growth not visible in the previous inspection. U Located near GW. 3.5 3.5 4270 15264.67 4430 15584.42 2012 Orphan (no ML reported 2011) Joints with Largest Maximum Depth Deepest Features from 2012 Survey that were Matched U Located on GW. U Located near GW. 3.5 3.5 3.5 3.5 4650 16459.45 4660 16499.44 2012 Orphan (no ML reported 2011) Joints with Largest Maximum Depth Deepest Features from 2012 Survey that were Matched PS PS 5100 18164.55 Deepest Features from 2012 Survey that were Matched Located near GW. Located near GW. Feature appears to be growing wider. U Located near GW. 3.5 3.5 14.9 3.5 3.5 3.5 5120 18212.77 5400 19284.44 5660 20324.54 Largest Difference Between Matched Pits (1:1 matches ONLY) Largest Difference Between Matched Pits (1:1 matches ONLY) Deepest Features from 2012 Survey that were Matched U Located near GW. U Located near GW. U Located near GW. 3.5 3.5 3.5 3.5 3.5 3.5 5680 20404.89 5840 21009.74 Largest Difference Between Matched Pits (1:1 matches ONLY) Deepest Features from 2012 Survey that were Matched U Located near GW. U Located near GW. 3.5 3.5 3.5 3.5 5930 21351.11 Deepest Features from 2012 Survey that were Matched PS 6060 21834.30 6070 21874.41 Largest Difference Between Matched Pits (1:1 matches ONLY) Estimated Corrosion Growth Rate 3.5 3.5 Located outside of repaired area. U Located near GW. U 3.5 14.9 3.5 3.5 6100 21994.61 2012 Orphan (no ML reported 2011) Joints with Largest Maximum Depth PS New growth not visible in the previous inspection. Located near GW. 3.5 16.2 6180 22315.00 2012 Joints with Most Unmatched Metal Loss (Orphan) U Located near GW. 3.5 3.5 6270 22652.08 2012 Joints with Most Unmatched Metal Loss (non- Orphan) U 3.5 3.5 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 F-17#

Page 494Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Joint ID Odometer, ft Review Selection Criteria Manual Review † Review Comments 6350 22972.10 2012 Joints with Most Unmatched Metal Loss (non- Orphan) P New growth not visible in the previous inspection. Located near GW. Original Rate, mpy 3.5 Adjusted Rate, mpy 3.5 6360 23012.16 Estimated Corrosion Growth Rate P 3.5 3.5 6370 23052.31 6520 23639.09 Largest Difference Between Matched Pits (1:1 matches ONLY) Deepest Features from 2012 Survey that were Matched P Located near GW. PS 3.5 3.5 3.5 3.5 6550 23746.71 Deepest Features from 2012 Survey that were Matched P Located near GW. 3.5 5.5 6590 23906.73 2012 Joints with Most Unmatched Metal Loss (Orphan) P New growth not visible in the previous inspection. Located near GW. 3.5 3.5 6600 23946.77 2012 Joints with Most Unmatched Metal Loss (Orphan) U Located near GW. 3.5 3.5 7120 25874.03 7400 26930.83 7420 26984.21 2012 Joints with Most Unmatched Metal Loss (non- Orphan) 2012 Joints with Most Unmatched Metal Loss (non- Orphan) 2012 Joints with Most Unmatched Metal Loss (non- Orphan) P 3.5 3.5 7490 27246.37 2012 Joints with Most Unmatched Metal Loss (Orphan) U 3.5 U Located near GW. New growth not visible in the previous inspection. U Located near GW. 3.5 3.5 3.5 3.5 3.5 7580 27595.59 2007 Joints with Most Unmatched Metal Loss (Orphan) U 3.5 3.5 7670 27956.49 Deepest Features from 2012 Survey that were Matched P 3.5 3.5 7690 28009.13 2012 Joints with Most Unmatched Metal Loss (Orphan) U Located near GW. 3.5 3.5 7990 29170.51 Deepest Features from 2012 Survey that were Matched P Located near GW. Feature appears to be growing wider. 3.5 3.5 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 F-18#

Page 495Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Joint ID Odometer, ft Review Selection Criteria Manual Review † Review Comments Original Rate, mpy 8010 29250.55 2012 Orphan (no ML reported 2011) Joints with Largest Maximum Depth PS New growth not visible in the previous inspection. Located outside of repaired area. 3.5 Adjusted Rate, mpy 10.1 8060 29451.25 Largest Difference Between Matched Pits (1:1 matches ONLY) PS 3.5 8140 29741.31 2012 Orphan (no ML reported 2011) Joints with Largest Maximum Depth Deepest Features from 2012 Survey that were Matched New growth not visible in the previous inspection. U 3.5 4.1 8280 30276.76 PS 3.5 3.5 24.3 8360 30596.93 8640 31550.78 2007 Joints with Most Unmatched Metal Loss (non- Orphan) Estimated Corrosion Growth Rate U 3.5 P Located near GW. 6.4 3.5 6.4 8660 31597.28 8680 31622.71 2012 Joints with Most Unmatched Metal Loss (non- Orphan) Deepest Features from 2012 Survey that were Matched U 3.5 U Located near GW. Previously repaired. 3.5 3.5 3.5 8690 31633.87 8980 32410.90 Largest Difference Between Matched Pits (1:1 matches ONLY) Estimated Corrosion Growth Rate 3.5 3.5 9060 32644.07 Largest Difference Between Matched Pits (1:1 matches ONLY) 3.5 9160 32962.17 Deepest Features from 2012 Survey that were Matched 9200 33122.06 2012 Joints with Most Unmatched Metal Loss (Orphan) 9250 33322.18 Deepest Features from 2012 Survey that were Matched 9260 33362.23 2012 Joints with Most Unmatched Metal Loss (Orphan) U Located near GW. P Located on GW. U Located near GW. U Located near GW. 3.5 3.5 3.5 3.5 P 3.5 3.5 P 3.5 9270 33401.66 Deepest Features from 2012 Survey that were Matched 9280 33441.67 2012 Joints with Most Unmatched Metal Loss (non- Orphan) P U Located near GW. 3.5 New growth not visible in the previous inspection. 3.5 3.5 3.5 P 3.5 3.5 3.5 3.5 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 F-19#

Page 496Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Joint ID Odometer, ft Review Selection Criteria Manual Review † Review Comments 9300 33521.75 Deepest Features from 2012 Survey that were Matched P Original Rate, mpy 3.5 Adjusted Rate, mpy 3.5 9310 33561.84 Deepest Features from 2012 Survey that were Matched New growth not visible in the previous inspection. P Located near GW. 3.5 3.5 9360 33761.78 2012 Orphan (no ML reported 2011) Joints with Largest Maximum Depth P 9390 33866.72 Deepest Features from 2012 Survey that were Matched PS New growth not visible in the previous inspection. Located near GW. 3.5 6.3 3.5 3.5 9420 33987.05 Deepest Features from 2012 Survey that were Matched PS 3.5 9430 34027.19 9450 34107.43 9650 34890.95 9860 35634.99 9880 35715.04 2012 Orphan (no ML reported 2011) Joints with Largest Maximum Depth 2012 Orphan (no ML reported 2011) Joints with Largest Maximum Depth Largest Difference Between Matched Pits (1:1 matches ONLY) Largest Difference Between Matched Pits (1:1 matches ONLY) 2012 Joints with Most Unmatched Metal Loss (Orphan) PS PS 3.5 3.5 New growth not visible in the previous inspection. Feature appears to be growing wider. Located outside of repaired area. U 3.5 P 3.5 U Located near GW. 6.5 3.5 16.9 14.9 3.5 12.8 3.5 9890 35755.13 9920 35875.25 2012 Orphan (no ML reported 2011) Joints with Largest Maximum Depth Deepest Features from 2012 Survey that were Matched PS 3.5 Located near GW. U 3.5 9.5 3.5 10540 38046.21 3.5 3.5 10950 39466.11 2012 Orphan (no ML reported 2011) Joints with Largest Maximum Depth U Located near GW. 2012 Joints with Most Unmatched Metal Loss (non- Orphan) U 3.5 3.5 10990 39592.06 Largest Difference Between Matched Pits (1:1 matches ONLY) P Additional pit near GW in both inspections not called. 3.5 6.8 11000 39614.43 Deepest Features from 2012 Survey that were Matched P 3.5 3.5 11030 39701.69 Deepest Features from 2012 Survey that were Matched U Located near GW. 3.5 3.5 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 F-20#

Page 497Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Joint ID Odometer, ft Review Selection Criteria Manual Review † Review Comments 11050 39768.01 11060 39808.08 11310 40693.57 Original Rate, mpy 2012 Orphan (no ML reported 2011) Joints with Largest Maximum Depth U 3.5 Largest Difference Between Matched Pits (1:1 matches ONLY) PS 3.5 16.9 Deepest Features from 2012 Survey that were Matched U Located near GW. Adjusted Rate, mpy 3.5 3.5 3.5 11330 40741.62 11470 41210.73 11540 41490.60 2012 Orphan (no ML reported 2011) Joints with Largest Maximum Depth 2007 Joints with Most Unmatched Metal Loss (non- Orphan) Estimated Corrosion Growth Rate 3.5 PS 4.8 11550 41530.68 2012 Joints with Most Unmatched Metal Loss (non- Orphan) P U Located on GW. U 3.5 New growth not visible in the previous inspection. New growth not visible in the previous inspection. Located near GW. 3.5 3.5 4.8 3.5 3.5 11570 41610.83 11590 41690.92 2007 Joints with Most Unmatched Metal Loss (non- Orphan) Deepest Features from 2012 Survey that were Matched PS 3.5 3.5 8.8 11600 41730.90 Deepest Features from 2012 Survey that were Matched PS 3.5 3.5 11610 41744.11 11650 41891.05 11990 43143.07 12160 43705.68 2012 Orphan (no ML reported 2011) Joints with Largest Maximum Depth 2012 Orphan (no ML reported 2011) Joints with Largest Maximum Depth 2012 Orphan (no ML reported 2011) Joints with Largest Maximum Depth Deepest Features from 2012 Survey that were Matched PS U 3.5 New growth not visible in the previous inspection. Located near GW. Feature appears to be growing wider. P Located near GW. 3.5 U Located near GW. 3.5 3.5 P 3.5 Feature appears to be growing wider and in length. 5.1 3.5 8.8 10.8 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 F-21#

Page 498Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Joint ID Odometer, ft Review Selection Criteria Manual Review † Review Comments 12170 43745.88 2012 Joints with Most Unmatched Metal Loss (non- Orphan) P Original Rate, mpy 3.5 Adjusted Rate, mpy 3.5 12230 43974.21 Deepest Features from 2012 Survey that were Matched There is growth on adjacent joint upstream. Feature appears to be growing wider. P Feature appears to be growing wider. 12240 44014.10 Estimated Corrosion Growth Rate PS 3.5 3.5 3.5 3.5 12270 44125.63 Deepest Features from 2012 Survey that were Matched U Located near GW. 3.5 3.5 12280 44165.64 Deepest Features from 2012 Survey that were Matched P 3.5 12300 44245.65 2007 Joints with Most Unmatched Metal Loss (Orphan) New growth not visible in the previous inspection. U 3.5 3.5 3.5 12410 44669.66 2012 Joints with Most Unmatched Metal Loss (Orphan) P New growth not visible in the previous inspection. 3.5 3.5 12420 44709.74 Deepest Features from 2012 Survey that were Matched PS 3.5 8.1 12430 44748.75 Largest Difference Between Matched Pits (1:1 matches ONLY) PS 3.5 7.3 12460 44868.95 12490 44988.84 2012 Orphan (no ML reported 2011) Joints with Largest Maximum Depth Deepest Features from 2012 Survey that were Matched P P 3.5 Located near GW. Feature appears to be growing wider. 3.5 6.2 3.5 12510 45069.09 12540 45182.34 2012 Joints with Most Unmatched Metal Loss (non- Orphan) P Feature appears to be growing wider. 3.5 3.5 2012 Orphan (no ML reported 2011) Joints with Largest Maximum Depth P Located outside of repaired area. 3.5 4.7 12550 45204.55 Largest Difference Between Matched Pits (1:1 matches ONLY) P 12590 45331.80 2012 Joints with Most Unmatched Metal Loss (Orphan) P New growth not visible in the previous inspection. New growth not visible in the previous inspection. 3.5 3.5 3.5 3.5 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 F-22#

Page 499Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Joint ID Odometer, ft Review Selection Criteria Manual Review † Review Comments 12710 45747.80 Deepest Features from 2012 Survey that were Matched P New growth not visible in the previous inspection. Located near GW. Original Rate, mpy 3.5 Adjusted Rate, mpy 3.5 12800 46063.76 Estimated Corrosion Growth Rate P 3.5 3.5 12820 46144.13 Deepest Features from 2012 Survey that were Matched P 12840 46224.40 Estimated Corrosion Growth Rate PS 12850 46264.57 Deepest Features from 2012 Survey that were Matched PS 12870 46344.80 Deepest Features from 2012 Survey that were Matched Located near GW. Feature appears to be growing wider. New growth not visible in the previous inspection. New growth not visible in the previous inspection. Located outside of repaired area. U Located near GW. 3.5 3.5 3.6 3.6 3.5 17.6 3.5 3.5 12880 46384.73 2012 Joints with Most Unmatched Metal Loss (non- Orphan) P Feature appears to be growing wider. 3.5 3.5 12900 46465.03 Deepest Features from 2012 Survey that were Matched U Located near GW. 3.5 3.5 13000 46781.15 13200 47361.45 2012 Joints with Most Unmatched Metal Loss (non- Orphan) Deepest Features from 2012 Survey that were Matched P 3.5 PS 3.5 3.5 4.0 13210 47401.55 2012 Joints with Most Unmatched Metal Loss (non- Orphan) PS 13260 47584.55 Deepest Features from 2012 Survey that were Matched New growth not visible in the previous inspection. U Located near GW. 3.5 19.6 3.5 3.5 13700 48881.37 Deepest Features from 2012 Survey that were Matched U Located near GW. 3.5 3.5 14060 50258.72 2012 Joints with Most Unmatched Metal Loss (non- Orphan) P New growth not visible in the previous inspection. Located outside of repaired area. 3.5 3.5 DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 F-23#

Page 500Plains All American Pipeline, L.P. Line 901 Release (5/19/15) Technical Root Cause Analysis Joint ID Odometer, ft Review Selection Criteria Manual Review † Review Comments 14470 51618.37 Estimated Corrosion Growth Rate PS Estimated Corrosion Growth Rate Joint. Original Rate, mpy 29.1 Adjusted Rate, mpy 29.1 15770 56459.87 Deepest Features from 2012 Survey that were Matched U 3.5 3.5 15900 56849.78 2012 Joints with Most Unmatched Metal Loss (Orphan) P 3.5 3.5 15910 56889.91 New growth not visible in the previous inspection. 2012 Orphan (no ML reported 2011) Joints with Largest Maximum Depth PS 3.5 13.5 † P = P, PS = Probable Significant Growth DNV GL – OAPUS307KKRA (PP136049) December 4, 2015 F-24#

Page 501ABOUT DNV GL Driven by our purpose of safeguarding life, property, and the environment, DNV GL enables organizations to advance the safety and sustainability of their business. We provide classification and technical assurance along with software and independent expert advisory services to the maritime, oil and gas, and energy industries. We also provide certification services to customers across a wide range of industries. Operating in more than 100 countries, our 16,000 professionals are dedicated to helping our customers make the world safer, smarter and greener.#

Page 502Appendix O NACE International: Effectiveness of Cathodic Protection on Thermally Insulated Underground Metallic Structures#

Page 503PSK-2 Item No. 24156 NACE International Publication 10A392 (2006 Edition) This Technical Committee Report has been prepared NACE By NACE International Specific Technology Group 35* on INTERNATIONAL Pipelines, Tanks, and Well Casings Effectiveness of Cathodic Protection on Thermally Insulated Underground Metallic Structures © September 2006, NACE International members who have reviewed this document, its scope, and provisions. Its acceptance does not in This NACE Interational technical committee report represents a consensus of those individual any respect rocesses, or procedures not included in this report. Nothing contained in this NACE report is to b preclude anyone from manufacturing, marketing urchasing, or using product. construed as granting any right, by implication or otherwise, to manufacture, sell, or use in connection anyone against liability for infringement of Letters Patent. with any method, apparatus, or product covered by Letters Patent, or as indemnifying or protecting as a restriction on the use of better procedures or materials not discussed herein. This report should in no way be interpreted the usefulness of this report in specific instances. port intended to apply in all cases relating to the subject. Unpredictable circumstances may nega interpretation or use of this report by other parties. ACE assumes no responsibility for t environmental, and regulatory documents and for determining their applicability in relation to this Users of this NACE report are responsible for reviewing appropriate health, report prior to its use. problems or environmental hazards associated with This NACE report may not necessarily address all potential health and safety operations detailed or referred to within this report. Users of this NACE report are also responsible use of materials, equipment, and/o applicable regulatory requirements prior to the use of this report. oite appropriate regurite auth tied sarnt talieve conces, i any elating reports are subject to periodic review, reports may receive current information on all NACE International publications by contacting the VACE FirstService Department, 1440 South Creek Drive, Houston, Texas 77084-4906 (telephone +1 Foreword external metallic surface corrosion control is the application establishing an effective level of of the effectiveness of CP. This report is intended for those of a barrier coating or adhesive on the metallic surface prior dealing with thermally insulated structures or pipelines. Experience as shown that there is generally a amited thermal insulating material. Although pipelines are the primary focus of this report, the beneficial effect from the application of cathodic protection principles discussed would be applicable when a thermal (CP) to a bare or ineffectively coated metallic surface under insulating material has been applied on or in the immediate thermal insulation. was originally prepared in 1992 by NACE Task Group (TG) proximity of an underground metallic surface. This report information guide for external corrosion control of thermally This NACE technical committee report was prepared as an Cathodic Protection and was reaffirmed with editorial T-10A-19, a component of Unit Committee T-10A on insulated underground metallic surfaces and considerations on Pipelines, Tanks, and Well Casings. It is published by changes in 2006 by Specific Technology Group (STG) 35 NACE under the auspices of STG 35. *Chair Paul R. Nichols, Shell Global Solutions, Houston, Texas. PAA_PHMSA_0000006#

Page 504NACE International NACE technical committee reports are intended to convey technical information or state-of-the-art knowledge regarding corrosion. In many cases, they discuss specific applications of corrosion mitigation technology, whether considered successful consideration when applying this or not. Statements used to convey this information are factual and are provided to the reader as input and guidance for recommendations for general application of this technology, and must not be construed as such. technology in the future. However, these statements are not intended to be BACKGROUND pipelines installed On most thermally insulated oil and prior to 1980 to 1981, a shop mold- gas transmission Thermally insulated pipelines have experienced relatively steel pipe, with an outer jacket applied to moisture-proof the formed thermal insulation was placed directly over the bare aggressive corrosion, with some failures occurring within three years of service, although acceptable industry were applied over the joint area to fit between the ends of system. At the field joint, preformed insulation half shells standards of CP had been applied and maintained shortly after line construction. The most predominant failures have the shop-applied insulation. After the insulation was fitted, a been those occurring at joints; however, moisture has insulation. When the integrity of the outer moisture barrier heat shrink sleeve or a tape wrap was applied over the migrated electrochemical corrosion cells remote from the field joint, along the pipeline steel surface to create edges of the preformed half shells and the shop-applied was compromised, the space, gap, or void between the of line. An article titled "Corrosion of Underground Insulated culminating in extensive replacements of substantial lengths steel beneath. insulation allowed oxygenated water to diffuse to the bare Pipelines"1 supports this committee's conclusions that also occurred remote from the joint, allowing oxygenated Damage to the outer moisture barrier has sufficient CP current from an external source may not reach the insulated metallic surface in sufficient quantity to ground water ingress. establish adequate corrosion control. BASIC CORROSION MECHANISM External failure of thermal insulated metallic surfaces has been primarily attributed to electrochemical corrosion cells conditions and annular spaces where oxygenated water has Figures 1a, 1b, and 1c detail various metallic surface have found generated from oxygenated ground waters, although some microbiologically and concluded that failures influenced corrosion (MIC). are due to migrated to a location remote or shielded from the external conventional CP is applied to a thermally insulated pipeline When wrap or sleeve, for some reason such as line movement, environment. Figure 1a shows a joint on which the joint where an annular void exists, protection along the length of has become disbonded from the exterior coating and allows the void often does not occur. In a paper titled "Cathodic water to ingress to the pipe surface. The oxygenated water at CORROSION/82, the authors submitted experimental Protection Levels Under Disbonded Coatings, "2 presented then migrates through the annulus, and active corrosion cells could be established if the foam or another barrier is data that suggested a distance limitation of effective not bonded to the pipe surface Sufficient CP current amount of bare metallic surface under the thermal insulation corrosion control by the use of externally applied CP. The generated externally cannot reach the metallic surface because of the shielding effect of the thermal insulation and coating) would (which can be equated to a severe condition of disbonded the natural phenomenon of electrochemical reactions that effectively protected by the externally applied CP. be the major factor limiting the area result in active corrosion cells. Joint wrap/sleeve Pipeline - Foam half shell water ingress Potential External coating- Shop-applied foam insulation FIGURE 1a: Typical Joint with Damaged Wrap or Sleeve 2 PAA_PHMSA_0000007#

Page 505NACE International discontinuity (holiday) in the exterior coating and a void in Figure 1b details a close-up of a metallic surface on which a Dye test experiments have indicated that impressed the thermal insulation has been created. applied CP current provides protection to the surface area The externally water migration patter once water has broken through the currents, such as those applied by CP systems, affect the at the void and for a limited distance beyond the void. insulation. These impressed currents in fact cause the of CP in which the insulating material affects or significantly Active corrosion cells result beyond the effective coverage water to migrate further from the point of entry than it may otherwise have done. inhibits current flow through the thermal material. been observed that when the rigid foam insulation adheres It has The conditions described are not necessarily the worst- or to the metallic surface, it forms an effective barrier and best-case situations. When the metallic surface except a discontinuity also exists in a coating that was corrosion usually does not occur. Figure 1c is similar to 1b temperature has been maintained so that moisture is not allowed surface contact, the presence of active corrosion applied directly to the pipe surface. elevated temperatures cannot be relied on as a corrosion cells by oxygenated water is usually eliminated; however, control method. - Path of water ingress through voids and insulation Path of CP current Serenal coating a That insan magna Limit of CP current - Foam bonded to surface No protection area - Bare surface area. Pipe . wall FIGURE 1b: Insulation Material Not Continuously Bonded to Metal 3 PAA_PHMSA_0000008#

Page 506NACE International -Path of water ingress through voids and insulation Path of CP current Thermal insulation material 88XX88 8XX coating barter Limit of CP current No protection area Steel surface wall Pipe FIGURE 1c: Pipe with Discontinuity in the Barrier Coating on the Pipe Surface TYPES OF THERMAL INSULATION industry has been a polyurethane foam. The most common thermal insulation utilized by the pipeline permeability of various types of thermal insulation. for thermal insulation and Table 1a reviews the water some of the properties of materials that have been utilized TABLE 1—Properties of Thermal Insulation Materials Type Typical Use Application Method Feasible Operating Temperature Heat Transfer Coeff. "K" Compression Strength kPa W/m=-K (psi) Polieihane Pipelines Shop Molding or Spray to 93°C (200°F) 0.12 207-414 (30-60) Isocyanurate Pipelines Shop Molding or Spray to 150°C (302°F) 0.18 (VERTICAL) 193(28) (PARALLEL) 138(20) Polystyrene Tank Bottoms Board Stock Laid in Sheet Form Cryogenic to 74°C (165°F) VERTICAL 41 (3: (PARALLEL) Fiberglass Pipe Half Shells to 316°C (600°F) 0.23 N/A Cellular Glass Pipe/Structures Board Stock/Half -268° to 538°C Shells (-450° to 1,000°F) 0.33 at 10°C (50°F) 689100) Calcium Silicate High Temperature Geothermal Pipe Half Shells to 593°C (1,100°F) 0.4 1379 (200)(A) Hot Water Lines 4 PAA_PHMSA_0000009#

Page 507• NACE International TABLE 1a—Water Permeability of Various Types of Thermal Insulation Type ASTM Methods Typical Value Rigid Polyurethane D28424 0.7 g/cm (0.05 lb//t) Isocyanurate D2842° 0.7 g/cm (0.05 lb/f) Polystyrene C27245 0.3% by volume Fiberglass N/A Less than 1% by volume Cellular Glass C240° 0.2% by volume Calcium Silicate Calcium silicate has a very high moisture absorption rate and may not be suitable for use on underground pipelines Cellular glass and calcium silicate insulations have been used underground in "pipe-within-a-pipe" systems. In these concentricity. Protective coatings supplemented with CP shells in the annular space, and metallic spacers provide systems, the thermal insulation is placed as half or quarter ensure a dry environment within the annulus. are provided for the outer casing This is necessary to APPLICATION PROCESSES buried service copied the technology used for above-grade Early methods of insulating steel line pipe intended for maximize this feature are selected. In circumstances in were placed over the pipe and held in place with ties or refinery piping. Preformed halves of polyurethane insulation which a coating or corrosion barrier is being applied to a bare steel pipe prior to application of insulation, a coating or bands of material and subsequently covered with an outer barrier is selected to withstand the application temperature wrap, such as polyethylene tape or polyethylene heat of the insulation. In other words, its effectiveness as a shrinkable sleeves of the type commonly utilized in pipeline corrosion barrier remains intact after the insulation has been From this technique evolved a process of placing steel line The principal change in the construction of thermally hot polyurethane was injected and allowed to form and set pipe, one joint at a time, in a "dunk" tank or mold into which insulated line from what was practiced prior to 1980 to 1981 is the addition of a corrosion barrier coating on the steel are available to provide the desired thickness of insulation. around the pipe. Mold tolerances of different dimensions pipe prior to applying the insulation materials. The typical to the use of rigid polyurethane insulation that requires both shop preparation and application process listed below refers process was developed for an external vapor/moisture barrier to maintain the thermal applying polyurethane insulation to a steel pipe that is integrity of the insulation system and a barrier at the steel rotating and travelling past a fixed point through a fixed surface to control corrosion: nozzle or jet. This method represents a recent development for the application of insulation to steel pipes intended for 1. Incoming pipe is inspected to ensure it is free of puried service, permitting greater control of such variables grease, oil, etc., that would impede proper coating as compressive strength and application temperature application. application, and physical conditions ariables in the polyurethane formulation, method of 2. Pipe is preheated to specified temperature. application determine the properties of the as-applied at the time o 3. Pipe is blast cleaned to the specified finish. are concerned with the integrity of the bond between the product. From a physical property point of view, end users 4. Corrosion coating is applied to specification with proper the insulation material, and maximum temperature at which nsulation and the steel pipe, the compressive strength of cutback. the properties of the insulation or its associated coating(s). he system can be operated without altering or damaging 5. The polyurethane insulation is applied to specified properties and thickness. Traditionally, emperature at which a pipeline can be operated. In orde pipeline coatings have determined the According to spectatonsure barier jacket is applied to realize maximum benefit and heat transfer efficiency from an insulated pipeline system, coating products that confirm conformance to specifications. The specified quality control tests are conducted to 5 PAA_PHMSA_0000010#

Page 508NACE International 8. Half shells of the required size are manufactured for shipment to the field for joint completion. 9. Heat shrink sleeves or polyethylene tape are provide jacket. for joint completion of the pipeline coating and the externa after the joint has been made and properly prepared Some companies prefer to inject insulation on site CONSTRUCTION PRACTICES Buried pipeline designs incorporate pipe-soil resistance in restraint calculations premise that the line pipe and any applied coatings or This characteristic works on the An alternative method of insulating field joints involves an injection molding at the job site followed by the application insulation are completely bonded and do not allow the pipe of shrink sleeves or tape wrap to move freelv inside the outer protective lavers. This relies on a good, long-lasting bond. the mign dr l dating banana note age, anticipated shipping, handling, and field bending employed Coatings and insulation that are able to withstand the auses a loss of thermal properties and may lead to pip during pipeline installation are chosen. A bending shoe or a orrosion. Some of the typical field construction practice hydraulic bender and an internal mandrel are usually used is used to resist pulverizing or breaking into numerous small Insulation with high compressive strength 1. Padded supports (typically sandbags) are used for pipe minimize damage to the insulation. pieces, and bending at low temperatures is avoided to handling and field stringing 2. Properly trained field personnel are employed for the used for the insulation of field joints at girthweld Preformed insulation half shells have been most commonly application and application is performed as specified and to adopt a completion of joints to ensure the the properly prepared and coated joint area and to fit These half shells are typically designed to be installed over specification that tests the joint integrity. insulation halves are in place, a heat shrink sleeve or tape is between the ends of the shop-applied insulation. crushing 3. Properly padded bending shoes are used to minimize compromising the external coating. of the thermal insulation and to avoid applied to provide an external seal and to hold the insulation in place (see Figure 1a). appropriate to avoid exteral coating perforation. 4. Proper ditch padding and select backfill are used when EVALUATION OF EXTERNALCORROSION MITIGATION locate/detect The use of internal pipeline corrosion inspection tools to been relatively successful in evaluating corrosion control. metal loss on external metallic surfaces has there is direct contact with the surrounding soil/water control under theral insulation. CP is only effective when judgement has also been used to locate external corrosion. Random excavation by personnel using sound engineering electrolyte. measurements with current applied, or when compersating pipe-to-soil (pipe electrolyte) When a metal loss area on a metallic surface is located, the for voltage drop, result in a reading between the reference location is excavated, and the corrosion status is evaluated electrode on the ground/earth surface and the nearest The line section is then recoated, repaired, or replaced conducting path to the metallic surface. The indicated reevaluate the condition on a regular basis is usually depending on the severity of the metal loss. A schedule to potential is therefore only representative of the nearest adopted. 2202 Conventional CP electrical evaluation techniques have not been able to accurately determine the status of corrosion 6 PAA_PHMSA_0000011#

Page 509NACE International Reference electrode KXXX Grade Damaged wrap/sleeve Extemal coating Thermal insulation material coating barier Steel surface wall Pipe FIGURE 2: Representative Circuit of Potential Measurement EXPERIENCES installed in North America, with virtually every owner substantial amount of thermally insulated pipe has beer A relatively documenting similar situations. TABLE 2—Experiences with Corrosion of Joints on Thermally Insulated Pipelines Incident Years in Service Corrosion Barrier Joint Coating Comments #1 10 No Shrink sleeve Poor shrink sleeve application #2 6 No Shrink sleeve tastrophic failu or shrink sleeve application; resulted #3 17 No Shrink sleeve Poor shrink sleeve application #4 17 No Shrink sleeve Poor shrink sleeve application #5 3 Yes Wraparound Poor application of undercoat tape; aiso shop- damaged barrier #6 6 No Wraparound specified but not installed Poor application; wrap too short; inner barrier tar epoxy coating of 400 um (16 mil), with 50 mm (2.0 in.) of Another experience involves a pipeline with a primary coal years, this 135-km (84.O-mile), 457-mm (18,0-in.) diameter polyurethane foam and a polyethylene jacket exterior line was checked with an instrumented and intelligent tool coated with 300 um (12 mil) of coal tar epoxy and 50mm coating of 4.1 mm (160 mil). Field joints were primed, top (smart pig) that indicated two locations with external corrosion. (2.0 in.) of foam, and finished with a shrink sleeve. After six 7 PAA_PHMSA_0000012#

Page 510NACE International Investigations revealed the corrosion was caused by water pipeline at the joint. migration through the shrink sleeves to the poorly coated many of these sites, close-interval surveys were taken where adequate potentials had been recorded Both locations were also at test whether the corrosion cell could be located in this manner. before the defect was excavated in order to determine during annual CP surveys. n every test section, this method proved ineffectiv reliminary evaluation of a 1989 log indicated that som Five years later (1987), the second intelligent tool survey corrosion was beginning to develop on sections of factory- at tield joints and areas where the nine was field-coated At showed accelerated external corrosion with most anomalies near field joints. CONCLUSIÓNS insulated metallic surfaces has been ineffective. 1. Generally, the application of external CP to thermally which causes corrosion at imperfections in the primary coating. thermally insulated metallic surfaces is the application of an 2. The principal or primary means of corrosion control of 4. When practical, the thermally insulated metallic effective coating on the metallic surface. surfaces need to be inspected at routine time intervals for metal loss (e.g., an internal pipeline inspection tool could be jacket and during pipe installation to minimize water ingress, 3. Care is typically taken in the application of the external used). REFERENCES 1. J.F. Delahunt, "Corrosion of Underground Insulated 1986): p. 36. ipelines." Journal of Protective Coatings and Linings 3, 1 5. ASTM C272 (latest revision), "Standard Test Method Sandwich Constructions" (West Conshohocken, PA for Water Absorption of Core Materials for Protection 2. R.R. Fessler, A.J. Markworth, R.N. Parkins, "Cathodic ProtecisioN, paper no. 118 Houston, a is 6. ASTM C240 (latest revision), "Standard Test Methods 1982). of Testing Conshohocken, PA: ASTM). Cellular Glass Insulation Block" West Measuring Compressive Properties of Thermal Insulations" ASTM C165 (latest revision), "Standard Method for • W.B. Holtsbaum, "Potential Measurement Pitfalls wit West Conshohocken, PA: ASTM). Eastern Conference, held November 18, 1991 (Houston hermally Insulated Pipes," NACE Canadian Regio for Water Absorption of Rigid Cellular Plastics" (West 4. ASTM D2842 (latest revision), "Standard Test Method Conshohocken, PA: ASTM). J.J. Baron, "Pipeline Field Joint Corrosion: Experienc paper no. 88-39-114 (Calgary, Alberta, Canada: id a Review of Materials," 39th Annual Technical Meeti Petroleum Society of CIM (" 1988) The (' The Petroleum Society of CIM (CIM), 500-5th Avenue SW Suite 720, Calgary, Alberta, Canada, T2P 3L5. 8 PAA_PHMSA_0000013#