Failure Analysis Associates

Integrity Analysis of the

Camisea Transportation
System,
Peru, S.A.,
June 2007
Integrity Analysis of the Camisea
Transportation System,
Peru, S.A.,
June 2007

Prepared for
Inter-American Development Bank
1300 New York Avenue N.W.
Washington, DC 20577

Prepared by

Exponent
320 Goddard, Suite 200
Irvine, CA 92618

June 8, 2007

Doc. No. SF36292.003 A0F0 0607 0806

 June 8, 2007

Contents

Page

List of Figures vi

List of Tables ix

Acronyms and Abbreviations x

Limitations xi

Executive Summary xii

Background and Scope of Work xii
Spill Incidents xv
Risk Identification xvi
Risk Evaluation xvii
Design-Related Risks xvii
Construction-Related Risks xviii
Geotechnical and Geology-Related Risks xx
Pipe Integrity–Related Risks xxiii
Seismic-Related Risks xxv
Scour-Related Risks xxvi
Future Activities xxvi

1 Introduction 1
1.1 Description of the System 1
1.2 Operational History 3
1.3 Report Organization 3

2 Exponent’s Investigation 5
2.1 Objective 5
2.2 Phases of Work 5
2.3 Scope 6

3 Review of NGL Pipeline Incidents 8

3.1 Background 8

SF36292.003 A0F0 0607 0806 ii

 June 8, 2007

3.2 First Incident 8

3.3 Second Incident 13
3.4 Third Incident 17
3.5 Fourth Incident 21
3.6 Fifth Incident 25
3.7 Sixth Incident 29
3.8 Summary and Conclusions 30

4 Design-Related Risks 32
4.1 Design Background 32
4.2 Hydraulic Design Risks 33
4.2.1 General Review 33
4.2.2 Hydraulic Design of NGL Pipeline 34
4.2.3 Hydraulic Design of NG Pipeline 35
4.3 Geotechnical Design Risks 37
4.4 Conclusions 38

5 Construction-Related Risks 40
5.1 Pipeline Construction Characteristics 40
5.2 Geotechnical Construction Characteristics 45
5.3 Clearing Related Risks 46
5.4 Trenching-Related Risks 47
5.5 Pipe Material–Related Risks 48
5.6 Pipeline Field Welding–Related Risks 50
5.7 Pipeline X-ray–Related Risks 51
5.8 Hydrostatic Testing–Related Risks 51
5.8.1 Background 51
5.8.2 Hydrostatic Testing Results for the Camisea System 52
5.8.3 Hydrogen-Induced Crack–Related Risks 53
5.9 Conclusions 54

6 Geotechnical and Geology-Related Risks 56

6.1 Geotechnical and Geologic Conditions 56

SF36292.003 A0F0 0607 0806 iii

 June 8, 2007

6.2 General Findings 57

6.3 Geotechnical Risk Assessment 60
6.3.1 Risk Assessment Methodology 60
6.3.2 Application of the Risk Assessment Method 62
6.4 Ongoing Geotechnical Risk Mitigation 69
6.5 Conclusions 70

7 Pipeline Integrity–Related Risks 73

7.1 Description of Potential Risks and Controls 73
7.2 Pipe Material and Damage Tolerance–Related Risks 74
7.3 Inline Pipe Inspection 75
7.3.1 Background 75
7.3.2 Inline Pipe Inspection Results 76
7.3.3 Circumferential Crack Detection 77
7.4 Conclusions 78

8 Seismic-Related Risks 80
8.1 Tectonic Overview 80
8.2 Seismic Hazards to Buried Pipelines 82
8.3 Seismic Risk Evaluation 84
8.3.1 Characterization of Seismic Demand 84
8.3.2 Permanent Ground Deformation (PGD) Hazards 85
8.3.3 Wave Propagation Hazards 86
8.4 Conclusions 87

9 Scour-Related Risks 88
9.1 River Crossings and Scour 88
9.2 Risks at Buried River Crossings 89
9.3 Conclusions 93

10 Summary and Conclusions 94

10.1 Pipeline Design–Related Risks 94
10.2 Pipeline Construction–Related Risks 94
10.3 Geotechnical and Geology-Related Risks 95

SF36292.003 A0F0 0607 0806 iv

 June 8, 2007

10.4 Pipeline Integrity–Related Risks 96

10.5 Seismic-Related Risks 96
10.6 Scour-Related Risks 97
10.7 Summary 97

11 Recommendations 98
11.1 Geotechnical and Geologic 98
11.2 Seismic 101
11.3 Scour 101
11.4 Pipe 101

SF36292.003 A0F0 0607 0806 v

 June 8, 2007

List of Figures

Page

Figure 1. Right-of-way of the Camisea Transportation System in Peru, with the

extraction and production (E&P) centered in Malvinas, a NGL fractionation
plant in Pisco, and a NG distribution point in Lurin. 2
Figure 2. General area of site of first spill incident, at KP 8+850, during ROW
stabilization work. Top: View downstream (looking downslope). Bottom:
View upstream (looking upslope). Plastic sheeting covers 14-inch NGL
pipeline. Photographs taken on June 12, 2006. 11
Figure 3. General area of first spill incident, at KP 8+850, looking downstream, after
stabilization work conducted in 2006. Photograph taken on September 11,
2006. 12
Figure 4. MCI photograph showing frontal view of fracture surface of the first rupture
at KP 8+850. 13
Figure 5. Location of second spill incident, at KP 222+500, during permanent pipeline
repair work. Top: View upstream (downslope) along excavated trench along
NGL pipeline. Bottom: View downstream (upslope) across trenches opened
for repairs. Photographs taken on June 14, 2006. 15
Figure 6. MCI photograph showing frontal view of fracture surface of the pipe from
the second spill incident of the NGL pipeline at KP 222+500. White arrows
indicate the weld’s root pass in the background to the fracture surface; red
arrows indicate the inner surface of the pipe where the hydrogen-induced
crack began; and yellow arrows indicate the narrow remaining ligament at
the outer surface of the pipe. 16
Figure 7. General area of the third spill incident, at KP 200+700, during stabilization
work. Top: View upstream showing large landslide above and below ROW.
Bottom: View downstream along ROW showing alignment of temporary
bypass pipeline (beneath stacked sacks) and new retaining wall at base of cut
slope. Photographs taken on June 14, 2006, and September 19, 2006. 19
Figure 8. MCI photograph of pipe from the third spill incident, at KP 200+700. 20
Figure 9. Site of fourth spill incident, at KP 50+900 showing river crossing and new
steel truss bridge carrying the NGL bypass pipeline. Top: View looking
downstream (uphill). Bottom: View looking upstream. Photographs taken
on June 13, 2006. 23

SF36292.003 A0F0 0607 0806 vi

 June 8, 2007

Figure 10. Observed groove (white arrows) and rupture (red arrows) on the outer surface
of the pipe, with the pipe being horizontally aligned and NGL product flow
being from the right to the left. Black material is the damaged protective
polyethylene cover. 24
Figure 11. General area of the fifth spill incident, at KP 125+950, during stabilization
work. Top: View upstream (downslope) along the ROW. Bottom: View
downstream (upslope) across excavation made to remove unstable soil.
Photographs taken on June 13, 2006. 26
Figure 12. MCI photograph showing fracture surface of NGL pipeline at the fifth spill
incident, at an amplification of 14×. Blue arrows demarcate three distinct
fracture zones. White arrows identify the narrow, slanted fracture associated
with crack nucleation. Each zone is numbered, with the outer surface of the
pipe being the top portion of the photograph. 28
Figure 13. Excavated NGL pipe section from the sixth spill incident, with protected
wrinkle from which NGL was reported to have been leaking. 29
Figure 14. Comparison of the pipeline elevation profile along the ROW, with the graph
origin being in Malvinas. 33
Figure 15. Ratio of the maximum operating pressure to the MAOP along the ROW of
the NGL pipeline. 34
Figure 16. Ratio of the maximum operating pressure divided by the design pressure
along the ROW of the NG pipeline. 36
Figure 17. Standard installation of NGL pipeline at KP 107. 42
Figure 18. Trenching and stockpiling of cuttings at KP 391. 42
Figure 19. Placement of selected fill as bedding at KP 358 using a machine that
separates larger stones from the remaining fill. 43
Figure 20. Laying of pipe over the Manugali River at KP 92. 44
Figure 21. Risk assessment results for May 2006. 64
Figure 22. Risk assessment results for September 2006. 65
Figure 23. Risk assessment results for October 2006, which includes the monitoring
program for the September 2006 results. 66
Figure 24. Change in risk from May to October 2006. 67
Figure 25. Risk by sectors for May 2006. 68
Figure 26. Risk by sectors for October 2006. 69
Figure 27. Cross-sectional view of the Peru-Chile Trench (after Worthey, Washington
State University website). 81

SF36292.003 A0F0 0607 0806 vii

 June 8, 2007

Figure 28. West coast of Peru showing source regions of great events of 1868 and 1877
and epicenters of notable 20th century earthquakes (after Dewey, Silva, and
Tavera 2003). 82
Figure 29. Flow chart for conducting scour analysis (from FHWA HEC-18). 90
Figure 30. Methodology utilized by Golder Associates for the scour analysis of the
Camisea Pipeline (taken from “EVALUACIÓN DE CRUCES Y
QUEBRADAS PROYECTO CAMISEA ‘RIVER CROSSING’” –
presentation by Golder Associates January 24 and 25, 2007, Lima, Peru. 92

SF36292.003 A0F0 0607 0806 viii

 June 8, 2007

List of Tables

Page

Table 1. Leaks identified during hydrostatic testing of the pipeline 49

Table 2. Risk category assessment chart 62

SF36292.003 A0F0 0607 0806 ix

 June 8, 2007

Acronyms and Abbreviations

ABSC ABS Consulting

API American Petroleum Institute
ASME American Society of Mechanical Engineers
BPD barrels per day
CFR Code of Federal Regulations
COGA Compania Operadora de Gas del Amazonas
DOT U.S. Department of Transportation
E&P extraction and production
Exponent Exponent® Failure Analysis Associates
FCAW Flux Arc Welding
FHWA HEC-18 Federal Highway Administration Hydraulic Engineering Circular No. 18
Golder Golder Associates
HAZ heat-affected zone
HDPE high-density polyethylene
HIC hydrogen-induced cracking
IDB Inter-American Development Bank
IGL Ingenieria y Geotecnia LTDA
LL Likelihood Levels
MAOP maximum allowable operating pressure
MCI Metallurgical Consultants Inc.
MFL Magnetic Flux Leakage
MMSCFD million standard cubic feet per day
MOP maximum operating pressure
MRA MR Associates
NG natural gas
NGL natural gas liquid
PCS pressure control station
PGD permanent ground deformation
PRS pressure reduction station
PS pump station
QWP Qualified Welding Procedures
RMP Risk Management Plan
ROW right-of-way
SCADA supervisory control and data acquisition
SEM scanning electron microscopy
SL Severity Levels
SMAW Shielded Metal Arc Welding
SR Safety Ratio
TgP Transportadora de Gas del Peru S.A.
the system Camisea Transportation System
TPS Tuboscope Pipeline Services
Vector Vector Peru S.A.C.

SF36292.003 A0F0 0607 0806 x

 June 8, 2007

Limitations

At the request of the Inter-American Development Bank, Exponent prepared this report to
summarize our evaluation of the integrity of the pipeline components in the Camisea
Transportation System, which experienced five spill incidents between December 2004 and
March of 2006, and a sixth spill incident in April 2007. This report also provides a site-specific
technical evaluation of the geotechnical and mechanical aspects of each incident, using the
currently available information to identify the most probable contributing factors. The scope of
services performed during this investigation may not adequately address the needs of other
interested parties, and any reuse of this report or the findings or conclusions presented herein is
at the sole risk of the user.

Our investigation included visual inspection of the pipeline right-of-way and adjacent areas,
document review, interviews of key personnel involved in the design and construction, limited
visual inspection of some failed pipe sections, and engineering analysis. Accordingly, Exponent
has no direct knowledge of, and offers no warranty regarding subsurface conditions, ground
stability, or the condition of concealed construction, beyond what was specifically revealed
during the site visits and our document review. Comments regarding concealed construction or
subsurface conditions are our professional opinion, based on engineering and geological
experience and judgment, and are derived in accordance with current standards of professional
practice. Changes in the conditions of the right-of-way may occur with time due to natural
processes or events, such as rains and landslides or human activities. In addition, Exponent has
used and relied upon certain information provided by sources that it believes to be reliable for the
purpose of this report. Accordingly, the findings of this report may be invalidated, wholly or in
part, by changes that are beyond our control.

SF36292.003 A0F0 0607 0806 xi

 June 8, 2007

Executive Summary

Background and Scope of Work

Exponent® Failure Analysis Associates (Exponent) was retained by the Inter-American

Development Bank (IDB) to perform a pipeline integrity analysis of the Camisea Transportation
System (the system). Exponent’s retention followed the occurrence of five spill incidents, during
the first 19 months of operation, each of which resulted in the release of hydrocarbons. The
intent of our investigation was to develop a risk profile for the two component pipelines and
identify the factors that contributed to the incidents. During our investigation of causal factors in
the five incidents and assessment of pipeline integrity, Exponent made recommendations to
Transportadora de Gas del Peru S.A. (TgP) in order to improve future pipeline integrity by
mitigating and controlling identified risks to the system. TgP has implemented many of these
interim recommendations and has undertaken other additional activities based on its experience
and knowledge. This report summarizes these efforts and provides a risk-based evaluation of the
system that incorporates extensive sources of information and field investigations by Exponent,
commencing in April 2006.

The Camisea Transportation System is owned and operated by TgP. TgP contracted with
Compania Operadora de Gas del Amazonas (COGA) for the operation and maintenance of the
pipeline. The system consists of two buried pipelines: 1) a natural gas (NG) pipeline, which
runs from the upstream facilities at Malvinas to a terminal station at Lurin, at the southern edge
of Lima; and 2) a natural gas liquid (NGL) pipeline, which transports the liquid condensates
from Malvinas to a fractionation plant near Pisco, on the coast of Peru south of Lima. The two
pipelines share a common right-of-way (ROW) that traverses the Peruvian jungle, climbs over
the Andes Mountains at an elevation of approximately 4,800 m, and descends steeply toward the
coast along the Pacific Ocean. The alignment of the ROW is shown on Figure 1. The NGL
pipeline is approximately 561 km long, and the NG pipeline is approximately 734 km long.1
Along this route, the NGL pipeline telescopes from a nominal pipe diameter of 14 to 10¾ inches,

1
True length of pipeline.

SF36292.003 A0F0 0607 0806 xii

 June 8, 2007

and the larger NG pipeline telescopes from a nominal pipe diameter of 32 to 24 to 18 inches.
Construction of the pipelines started in 2002, and commercial operation began in August 2004.

All of the spill incidents occurred in the first 222 km of the NGL pipeline—four of the spills
occurred in the selva sector, one in the transition zone between the selva and the sierra sectors,
and one in the sierra sector. In contrast, no leaks have occurred on the larger diameter NG
pipeline or on either pipeline in the costa sector. The locations and dates of the first five spill
incidents are shown on Figure 1. Specifics of the first five individual spill incidents are
presented later in this summary. Recently, TgP identified2 a sixth incident on April 2, 2007,
where a small amount of NGL was reported to have been released. TgP subsequently repaired
the NGL pipeline and is currently investigating the root cause of this incident. The location of
this spill incident is also shown on Figure 1.

Exponent’s pipeline integrity analysis was conducted in two integrated phases. The objective of
the first phase was to provide a forward-looking, overall assessment of the integrity of the
pipelines, primarily the NGL. The following components were included in our evaluation of risk
in the first phase: pipe material, pipeline design, pipeline construction, geologic and
geotechnical hazard mitigation, pipeline maintenance, and ongoing operation. Exponent
prioritized the identified hazards and evaluated the efficacy of currently used mitigation and
control measures. The first phase included an evaluation of the suitability of the seismic design,
the mechanical design, and the design of river crossings (scour analysis). As part of that study,
we established a baseline risk level for the system and performed a technical review of the five
spill incidents that occurred in the system between December 2004 and March 2006.

The objectives of the second phase were to further evaluate key risks and to evaluate the progress
made by TgP and COGA in reducing risks to the pipeline. The risk was evaluated following the
implementation of various mitigation measures constructed in 2006, which was compared to the
baseline risk established during the first phase of our investigation. Specifically, the second
phase included: 1) developing and assisting in the implementation of a hybrid risk-based system
to evaluate potential geotechnical and geologic hazards to the pipeline system, 2) evaluating the

SF36292.003 A0F0 0607 0806 xiii

 June 8, 2007

effectiveness of the geotechnical stabilization measures constructed in 2006 to mitigate external

soil pressures acting on the pipelines, and 3) evaluating the efficacy of the current pipe integrity
program.

In order to accomplish the pipeline integrity analysis, Exponent reviewed more than 400 sets of
documents related to the design, construction, operation, and maintenance of the Camisea
Transportation System. These documents included engineering specifications, construction
specifications, pipe material data, pipeline design drawings and calculations, geotechnical and
geological studies, seismic studies, hydrological studies, river-crossing studies, construction
progress surveys, various internal and external pipeline inspection reports, and operational data.
In some cases, Exponent performed its own engineering analysis to quantify certain risks to the
integrity of the pipeline.

In addition to the document review and engineering analysis activities, a multi-disciplinary team
of Exponent engineers and scientists performed inspections along the pipeline ROW in June and
September 2006. These inspections occurred at more than 50 sites along the ROW where
geotechnical stabilization measures were proposed, under construction, or had been completed.
Exponent personnel also interviewed key personnel involved with design, construction,
operations, and maintenance of the system. These interviews were supplemented with numerous
teleconference calls that included the designers, operators, service providers, and independent
consultants hired during the construction and maintenance of the pipeline by TgP. Finally,
Exponent participated in the metallurgical investigation of samples of pipe that were involved in
two of the incidents, and reviewed the metallurgical examination reports and evidence from the
five spill incidents that we investigated.

2
Exponent was informed that TgP detected this minor leak during planned activities of its pipeline integrity
program.

SF36292.003 A0F0 0607 0806 xiv

 June 8, 2007

Spill Incidents

Exponent reviewed information related to the first five NGL pipeline spill incidents as a means
of evaluating risk, and to consider the potential for systemic problems. Exponent was not
retained to perform a root-cause analysis of any of the spill incidents.3

Our evaluation identified similarities in fracture surfaces in the NGL pipe from the first and fifth
incidents, which occurred at KP 8+850 and KP 125+950, respectively. In both cases, Exponent
identified unstable geotechnical conditions as a significant contributor to the rupture of the pipe.
At both of these locations, progressive soil loading likely propagated an initial crack and induced
the rupture of the NGL pipeline. In the incident at KP 8+850, the crack resulted in a through-
wall leak of about 10 inches in extent. In the incident at KP 125+950, the crack resulted in
complete severance of the NGL pipe. The NG pipeline was not damaged at either location.

The second incident, located at KP 222+500, appears to have been primarily the result of a time-
delayed, hydrogen-induced crack in the weld. The hydrogen crack escaped detection by the
post-welding radiography because of the inherent incubation time associated with hydrogen-
induced crack initiation. It is currently Exponent’s opinion that the high toughness of the pipe
material allowed the pipe to pass subsequent hydrostatic testing (performed five months after the
welding), even though the crack had extended to approximately 90% of the wall thickness4 by
the time the hydrostatic test was performed. Our current understanding is that the combination
of hydrostatic load cycles and subsequent operational pressure fluctuations caused the hydrogen
crack to be further destabilized, ultimately rupturing the pipe and releasing the NGL at a very
slow rate.

The third incident occurred in an area that was well studied from a geologic perspective and was
known to be an area of very high risk of ground failure. While some measures were taken during
construction to mitigate this geologic risk, a sizable landslide ultimately overwhelmed these
measures, both undermining and overtopping the ROW and the road next to the ROW. Hence,

3
A root cause analysis is the integrated evaluation of all facts pertaining to the investigated failure, to uniquely
identify the cause or causes of failure.
4
This is a rather deep crack, and subsequent crack growth due to normal operational pressure fluctuations need be
only minimal to reach a critical crack depth that causes the remaining ligament ahead of the crack to fail.

SF36292.003 A0F0 0607 0806 xv

 June 8, 2007

the rupture of the NGL pipe at this location is attributed to overload caused by a substantial
landslide. The NG pipe at this location was not damaged.

For the fourth incident, located at KP 50+900, current information suggests that the rupture of
the NGL pipe was induced by mechanical damage to the exterior concrete coating and a dent in
the exterior wall of the NGL pipeline. Analysis performed to date indicates that the dent was not
made by a boulder washed downstream during the flash flooding that immediately preceded the
spill. Indeed, the pipe wall at this location was apparently capable of containing the NGL fluid
until some unknown external loading event caused the already weakened pipe wall to fail in
ductile overload at the damaged area. This triggering loading event could have been associated
with riverbed scouring caused by the flash flooding. Again, the NG pipe at this location was not
damaged.

Finally, it is important to note that the metallurgical testing confirmed that none of the five spill
incidents were related to pipe material quality.

Risk Identification

The pipeline integrity analysis evaluated the risk categories that influence the likelihood and
severity of potential pipeline failures. For purposes of this report, we define risk as the
likelihood that a given chain of events will occur and result in a consequence that has a defined
severity. The aggregate of the likelihood of failure and the severity of failure is risk. In this
report, risk was ranked to be minimal if the risk is currently not a concern and effectively
consistent with other pipelines.

The objective of risk management as part of a pipeline integrity management program is to

identify, eliminate if appropriate, and then prioritize the remaining sources of risk so they can be
mitigated, controlled, and/or monitored. In this context, during our study, we identified four

SF36292.003 A0F0 0607 0806 xvi

 June 8, 2007

primary categories of risk affecting the integrity of the pipeline: geotechnical and geological,5
mechanical pipe integrity, seismic events, and river scouring. Each of these areas of integrity
risk is discussed separately below. In summary, we found that the risk associated with
geotechnical and geological conditions is currently more significant than risks associated with
pipe integrity, seismic events, and scour, although pipe integrity–related risks typically increase
with the age of the pipeline. The higher risk level associated with geotechnical hazards is a
direct consequence of the steep topography, poor foundation (ground) conditions, and abundance
of water along the pipeline ROW in the selva sector. Exponent’s review determined that TgP
implemented various actions in 2006 to substantially reduce the risk of future incidents to the
pipeline. Furthermore, our communication with TgP to date indicates that they are committed to
continue identifying and reducing the geotechnical risks to the pipeline.

Risk Evaluation

The four primary categories of risk identified above are discussed in more detail in the
paragraphs that follow. However, we also recognize that the four risk categories identified
above depend, to varying degrees, on decisions made during the design and construction of the
pipeline. Due to their importance, we first provide a brief summary of our investigation into
potential systemic risks resulting from the design and construction of the system.

Design-Related Risks

The system was designed to comply with the engineering code requirements of the American
Society of Mechanical Engineers (ASME).6 Code compliance is established if the designer
demonstrates that all specific code requirements and all reasonably foreseeable load conditions
are addressed by the design. Foreseeable load conditions apply to internal pressures and to
external loads imposed by soil pressures or ground movement.

5
For the purposes of this report, geotechnical and geological hazards are defined as external pressures resulting
from ground movement. Geotechnical hazards are defined as movement in soil, typically when saturated,
whereas geological hazards are defined as movement in rock.

SF36292.003 A0F0 0607 0806 xvii

 June 8, 2007

Our review indicates that the pipe wall thickness is sufficient to contain the internal pressures of
the transported hydrocarbon products along the entire length of the pipeline (i.e., hydraulic risk).
We have independently verified that the computed design pressures are code compliant and in
good agreement with the measured operational pressures along the whole length of both
pipelines. Therefore, any risks associated with the internal pressure aspects of the design are
minimal and consistent with other pipelines.

Given the demanding route of these pipelines through the jungle and up the mountains, external
loads were an important element for the design and construction of the system. Our design
review revealed that the pipeline designers assumed that external soil loading would be entirely
mitigated by geotechnical mitigation measures implemented during construction at sites deemed
to pose a geotechnical or geological hazard.

Construction-Related Risks

The system was constructed simultaneously at several so-called “mini-spreads” along the ROW,
using more than 100,000 individual pipes, each of which is up to 12 meters long. A review of
the pipe book7 and the pipe manufacturing and coating records indicates that the pipe material
used was purpose built for the system at two pipe mills during 2002 and 2003. The pipe
manufacturer’s records indicate that these mills are located in Pindamonhangaba, SP, Brazil, and
Buenos Aires, Argentina. These electric-resistance-welded tubular pipes are manufactured per
the American Petroleum Institute’s API 5L standard, such that steel plates are rolled and
longitudinally welded at the mill. Subsequently, all pipes were coated with an outer high-density
polyethylene (HDPE) layer to protect them from external corrosion.

The pipes were transported to the individual construction sites, where the ROW itself was
commonly the only available route for transportation. This ROW had to be contained in a
3-km-wide, government mandated corridor, and deviations from this pre-approved corridor had

6
ASME B31.4 [Pipeline Transportation Systems for Liquid Hydrocarbons and Other Liquids] is the applicable
Code for the NGL pipeline, and ASME B31.8 [Gas Transmission and Distribution Piping Systems] is the
applicable code for the larger NG pipeline.
7
The pipe book lists relevant pipe data.

SF36292.003 A0F0 0607 0806 xviii

 June 8, 2007

to be granted by the Peruvian government. Within the 3-km-wide corridor, the cleared section of
the ROW was typically restricted to 25 meters.

To minimize environmental impact, the constructor decided to preferentially build the pipeline
along mountain ridges. The ROW was cleared and cut, and the work was inspected by outside
consultants. Exponent’s review indicates that this approach was appropriate and preferred, in
that placing the ROW near the base of the mountains, and along rivers and drainages, would
have likely resulted in substantially more construction-related damage to the environment.

Due to limitations imposed by the narrow lane of clearing, grading along the ROW consisted of
cutting into the hillside and placing the excavated material as “side-cast” or “spill” fill on the
downhill side of the clearing. Although these side-cast fills were generally placed outside the
limits of the pipe trenches, they ultimately became a source of concern related to the potential to
exert external soil pressures on the pipes. The ground conditions encountered during installation
of the pipe were reportedly assessed by geotechnical engineers, and some mitigation measures
were constructed at that time. Geotechnical engineers also supervised the geotechnical
mitigation measures to control surface water runoff and stabilize the ROW following installation
of the pipelines.

Next, the pipe was strung out, and girth welds were used to join individual pipes. A completed
pipe section that may be several hundred meters long was then lowered into the trench and
welded together to form an even longer pipe section. This standard construction methodology
was replaced by special construction methods in very steep terrain, at river crossings, and at
locations where the pipeline was laid along an existing road. Overall, the construction methods
are consistent with general pipeline construction practice, where potential risks to pipeline
integrity arise primarily from the girth welds, trench conditions, and potential damage to the pipe
exterior.

The more then 100,000 girth welds of both pipelines were to be welded per approved procedures
and to be x-rayed 24 hours later.8 This process was generally effective at minimizing and

8
Exponent’s scope of work did not include reviewing these x-rays; it is our understanding the government of Peru–
commissioned pipeline audit is performing this task.

SF36292.003 A0F0 0607 0806 xix

 June 8, 2007

detecting weld defects that needed to be repaired, where weld quality was to be evaluated per the
American Petroleum Institute’s API 1104 standard. To further reduce the likelihood of failure,
long individual sections of the pipeline were hydrostatically tested. Hydrostatic testing involves
filling each pipe section with water and pressurizing the water to a predefined level that exceeds
the maximum operating pressure. Next, the water-tightness of the pipeline was verified by
maintaining a constant water pressure for at least 24 hours. This hydrostatic test was performed
for both pipelines along the entire length of the system.

During the hydrostatic testing, eight leaks were identified, of which seven occurred in the NGL
pipeline and one in the NG pipeline. The causes of the leaks were determined to be a faulty girth
weld in three cases, a faulty longitudinal weld in two other cases, external damage during
construction in two cases, and a foreign object being introduced during rolling of the steel plate
used to manufacture a pipe in one case. All these failures were subsequently repaired, and the
pipeline section was successfully re-tested. Eight failures in more than 1,250 km of pipeline is a
low number, especially considering the challenging terrain. The test results are more an
indication of the test’s ability to detect preexisting faults. Despite the above-described actions,
some minimal risk exists because defects may be aligned or sized such that the hydrostatic test
would not rupture the pipe and the defect could go undetected, which may be similar to other
pipelines in the world. This situation arose with the second spill incident, in which a hydrogen-
induced crack survived the hydrostatic test, initiating the subsequent spill incident. This incident
is currently not considered to be indicative of any systemic problems for the more than 100,000
girth welds. Overall, the radiography of all girth welds and hydrostatic testing of the Camisea
Transportation System reduced the level of risk.

Geotechnical and Geology-Related Risks

The spill incidents and observed performance of the system as of early 2006 caused TgP to set
into motion an aggressive geotechnical remediation program that began in earnest in April 2006.
Geotechnical instability caused or substantially contributed to two of the five spill incidents (#1
and #5), and geologic instability caused one of the five spill incidents (#3). In this program,
more than 100 sites along the ROW were evaluated and mitigated in 2006 by implementing

SF36292.003 A0F0 0607 0806 xx

 June 8, 2007

geotechnical mitigation measures using more robust construction techniques. At the end of
2006, following extraordinary efforts to stabilize geotechnical and geologic conditions along the
ROW, TgP believed that geotechnical hazards due to soil movement were more effectively
documented, addressed, and controlled in this ongoing process.

Exponent observed over 50 sites, some on both occasions, during our field inspections in June
and September 2006. Based on our review, observations, and engineering experience, we
concluded that geotechnical and geologic conditions initially posed a substantial risk to the
pipeline. Construction of geotechnical mitigation measures in 2006 significantly reduced this
risk by improving the geotechnical and geological stability at specific sites of greatest hazard.
While the route of the system traverses a challenging and dynamic area, our initial inspection in
June 2006 indicated that some of the original stabilization measures implemented during
construction were not completely effective. As early as April 2006, TgP and COGA had begun
to implement a system of identifying those sites with the highest priority for mitigation based on
the perceived likelihood of failure and the potential consequence of failure. TgP and COGA had
also contracted additional external geotechnical and geological specialists to help assess the
hazards and evaluate the likelihood of failure.

Based on our second round of inspections in September 2006 and subsequent documentation
review, Exponent concluded that geotechnical stabilization measures constructed during 2006
are generally reliable and robust. Further, we determined that these measures were being applied
in a consistent and effective manner. As a result of these efforts, the risk of future failure of the
system from external geotechnical forces has been substantially reduced. In more critical areas,
instrumentation (e.g., strain gauge, piezometer, slope inclinometer, survey control, and rainfall
monitoring) is being used or is recommended to provide additional interpretation and warning of
ground instability, and data on movement characteristics, to permit more expeditious, reliable
repairs.

During the second phase of our project, Exponent worked collaboratively with TgP to develop a
hybrid risk matrix to adequately assess the likelihood of future failure resulting from
geotechnical and geologic conditions. This system was validated using information from our
field inspections. To date, the sites identified are listed in the hybrid risk matrix, representing

SF36292.003 A0F0 0607 0806 xxi

 June 8, 2007

locations that have exhibited manifestations of ground instability, and the vast majority of these
stations are located in the selva sector. The risk at each station was then evaluated at three
different points in time, reflecting the risk prior to and after construction of the new geotechnical
mitigation measures, and after implementation of the additional monitoring programs discussed
below.

Exponent also reviewed a monitoring program initiated by TgP and COGA to help reduce the
risk of future failure resulting from external geotechnical forces by detecting and quantifying
early signs of slope instability. This program allows for the early detection and correction of
potential problem areas. The core of this program involves regular visual inspections, including
during the rainy season, between KP 0 and KP 220. The multi-disciplinary team of inspectors is
trained by COGA’s technical consultants and is intimately familiar with the project and
conditions along the alignment. This program appears to be comprehensive and is integral to
reducing the risk of future failure resulting from geotechnical conditions.

The results of the geotechnical risk assessment are consistent with our field observations that
TgP and COGA have made substantial progress in reducing the overall risk. Thus, while we
initially ranked 45 of 94 sites as having “high” to “very high” risk (along the initial 455 km of
the ROW), the construction of the geotechnical mitigation measures in 2006 reduced this number
to 12 sites. The addition of monitoring provided a further reduction to 5 sites having a “high” to
“very high” risk, with solely the site at KP 108, adjacent to the second pump station, being now
ranked as “very high.” TgP has stated that construction of new geotechnical stabilization
measures is ongoing or completed at these sites. It is our understanding that TgP is updating the
hybrid risk matrix and is committed to implementing additional geotechnical measures in 2007.

We expect that the continuous, ongoing implementation of the risk matrix process will identify
additional sites that are not included in the current hybrid risk matrix due to the absence of
surface manifestations of ground movement. Some of these sites may even be ranked with a
high to very high risk and will need to be mitigated quickly. Therefore, we recommended that
TgP adopt a proactive approach of continually assessing geotechnical hazards along the ROW.
In this regard, TgP has committed to implementing the recommended Risk Management Plan
(RMP) that should govern the use of all risk assessment methods and guide TgP’s actions,

SF36292.003 A0F0 0607 0806 xxii

 June 8, 2007

decision process, and manner of execution. At this time, Exponent believes that a successful
implementation of the above, and the construction of additional geotechnical mitigation
measures in 2007, will identify additional sites and reduce the geotechnical-related risks further.

Pipe Integrity–Related Risks

Pipe integrity–related hazards are associated with pipe material, weld quality, and the quality of
protection the pipe is afforded against environmental conditions. As discussed above, the most
significant risks to the system arise from external loading caused by soil movement. Soil
movement imposes lateral loading upon the pipe, which can induce axial pipe stresses in
addition to those induced by the internal pressure of the transported hydrocarbons. These axial
stresses are known to affect the girth welds in the pipe. Exponent’s analysis of this loading
condition has shown that the as-designed NG pipeline has a significantly larger external load
capacity and flaw tolerance than the NGL pipeline, such that the NG pipeline generally has a low
risk of failure from external loads. Load capacity estimates for the NGL pipeline show that,
while sufficient for internal pressures, it is susceptible to this external failure mode, which is
consistent with four of the first five spill incidents.

Several approaches have been adopted by TgP to reduce this risk. First, during construction, all
the welds were x-rayed, and the entire pipeline was hydrostatically tested, reducing the potential
number of potential weld-related and pipeline material defects. Second, any growth of such
defects that would lead to the rupture of the pipe requires the presence of external loading.
Removal of the loading is a good way to further mitigate the risk, and TgP’s ongoing and prior
geotechnical construction program reduces the likelihood of soil movement. Thus, the residual
risk is now mostly confined to areas that may become geologically unstable and that may contain
potential weld anomalies.

To mitigate the residual risk, TgP performed an inline inspection of the NGL pipeline in 2006
using the Magnetic Flux Leakage (MFL) inspection tool and a geometric inline inspection tool.
The MFL inspection tool has the capability to detect metal loss and other potential anomalies,
and the geometric tool measures the pipe’s geometry along its length. TgP has reported that the
inspection of the NGL pipeline identified 30 reportable defects, per requirements by API 1160

SF36292.003 A0F0 0607 0806 xxiii

 June 8, 2007

and DOT9 195. TgP is currently excavating these sites to perform a more detailed evaluation and
initiate the appropriate repair measures, if required. In this regard, TgP has voluntarily identified
additional sites to further quantify the accuracy of the inline inspection tools. Results of the
MFL inspection tool, and evaluation of metal loss per ASME B31G, have shown that, currently,
no severe external or internal corrosion damage exists along the NGL pipeline. However, it is
Exponent’s opinion that some sections of the NGL pipeline may be subject to a potential long-
term risk that will need to be addressed in the long-term planning stages of TgP’s pipeline
integrity program.

In an effort to quantify the MFL tool’s ability to detect crack-like features, TgP and its
contractors conducted a research program to quantify the crack detection limit of the MFL tool.
In addition, Exponent performed a pipeline integrity study to determine the MFL inspection
tool’s utility in detecting circumferential cracks.10 Our analysis indicates that a potential
circumferential crack would need to be subjected to a significant external load to be detectable
with high certainty by the currently employed MFL inspection tool.

The MFL inspection tool has proven to be an excellent tool to detect internal and external
corrosion damage to the pipe in this project. Overall in 2006, TgP significantly reduced pipe
integrity–related risks and is currently engaged in additional efforts to reduce the risk profile
under technically and logistically difficult conditions. Specific resources have been committed
in 2007 to further improve TgP’s ability to detect potential circumferential cracks. A root cause
analysis of spill incidents 1 and 5 and the origin and nucleation of the potential circumferential
cracks will be performed by TgP, which will assist in assessing the implications of this concern
related to pipeline integrity.

At present, no pipeline inspection company is readily able to provide a commercially available

inspection tool to detect potential circumferential cracks, even though the technology to do so
appears to be available. Despite the fact that the use of in-line inspection tools to detect small
circumferential cracks is currently not a common practice among pipeline operators due to the

9
Department of Transportation of the United States of America.

SF36292.003 A0F0 0607 0806 xxiv

 June 8, 2007

relatively low risk to pipeline integrity posed by circumferential cracks under normal operating
loads, TgP will evaluate potential options.

Seismic-Related Risks

The system lies within regions that have the potential for very large and frequent earthquakes.
The earthquake hazard appears to be the greatest in the costa sector and least near the selva
sector. Recognizing the potential seismic hazard, the pipeline fabricator commissioned several
seismic hazard studies during the design process. The first study characterized the regional
seismic hazard for each of the three sectors (selva, sierra, and costa) and provided more specific
evaluations of ground motions for the pump stations, pressure control stations, and pressure
reducing stations. Another investigation identified active fault crossings along the pipeline and
evaluated the pipeline’s capacity for predicted fault displacements. Additional studies consisted
of evaluating the potential for wave propagation damage for straight sections of buried pipe and
12 surface facilities.

Although several seismic hazard studies were performed as part of the design of the system, our
review suggests that the potential for permanent ground displacements (e.g., landslides, slope
instability, liquefaction, and lateral spread) should be evaluated more comprehensively to reduce
uncertainty. Additionally, seismic risk management would benefit from an update of the design
ground motions with up-to-date scientific information. Exponent currently understands that TgP
is engaged in a review, as part of the pipeline integrity management plan of seismic risks, to
determine whether these potential seismic risks are acceptable for this system or whether
mitigation measures should be considered.

10
The service provider of the currently used MFL inspection tool has determined that only circumferential cracks
with a crack mouth opening of more than 0.1 mm can be detected with a probability of better than 90%. For
cracks with a smaller crack opening, the probability of detection decreases rapidly.

SF36292.003 A0F0 0607 0806 xxv

 June 8, 2007

Scour-Related Risks

Scour is defined as the erosion of streambed or bank material due to flowing water. In addition,
naturally occurring lateral migration of the mainstream channel within its floodplain may affect
the stability of the buried pipeline crossings. Flood-induced scour can occur over short periods
of time, with little warning and serious consequences. Therefore, pipeline crossings should be
designed and constructed to withstand floods of relatively extreme magnitude.

Exponent’s review of the scour analyses performed during the design of the project indicates that
some assessments were performed to determine the minimal depth at which to bury the pipeline.
These design studies included field investigations of stream crossings. In addition to the
mentioned studies, Exponent recommends evaluation of additional potential scour processes, as
described in the Federal Highway Administration’s Hydraulic Engineering Circular 18
(HEC-18), which is considered the industry standard to evaluate stream scour. However, HEC-
18 is not a design standard required under Peruvian law. Despite not being required, the HEC-18
scour design approach was partially utilized, along with other procedures that were deemed to be
appropriate by the designer. Given the potential uncertainty and the objective to minimize this
risk, additional scour studies and investigation of potential scour mitigation measures are
warranted and are currently being evaluated by TgP.

Future Activities

Exponent performed a pipeline integrity analysis of the pipeline components of the Camisea
Transportation System. The analysis identified four primary categories of risk to the integrity of
the pipeline: geotechnical, considered to be the most significant, and mechanical pipe integrity,
seismic events, and river scouring as secondary risks. The higher risk level associated with
geotechnical hazards is a direct consequence of the steep topography, poor foundation (soil)
conditions, and the abundance of water along the pipeline ROW in the selva sector. TgP has
implemented various actions to help reduce these risks, including various interim
recommendations made by Exponent during our investigation. However, in the opinion of

SF36292.003 A0F0 0607 0806 xxvi

 June 8, 2007

Exponent, additional actions are still recommended to continue the minimization of any existing
pipeline integrity risks.

In this context, a technical action plan has been established with TgP and the IDB. This plan
includes: 1) implementation of geotechnical mitigation and monitoring actions in 2007, along
with continued geotechnical risk assessment; 2) re-evaluation of potential scour risk at river
crossings and, based on the results, implementation of actions to reduce such risks; 3) re-
consideration of seismic risk, in particular wave propagation, permanent ground deformations,
excluding fault rupture, resulting from strong ground shaking; 4) experimental evaluation of
potential circumferential cracks and their impact on pipe integrity; 5) spill root cause analysis of
spill incidents #1, #4, and #5.

Exponent has been retained to provide continued technical assistance to the IDB related to the
Camisea Transportation System, which includes review of these actions and additional site visits
in 2007. Based on available information obtained during Exponent’s investigation and the
proposed actions, it appears to Exponent at this time that TgP is performing adequate pipeline
integrity actions, and that these actions have significantly reduced the risk to the system.
However, Exponent also notes that pipeline integrity management is a continuous process, and
thus, future information and risks need to be continually and properly evaluated. If and when
ongoing pipeline integrity management efforts identify additional issues, risk management
actions above and beyond those currently being taken may be required.

SF36292.003 A0F0 0607 0806 xxvii

 June 8, 2007

1 Introduction

1.1 Description of the System

The Camisea Transportation System (the system) is owned by Transportadora de Gas del Peru
S.A. (TgP). TgP contracted with Compania Operadora de Gas del Amazonas (COGA) for the
operation and maintenance of the pipeline. Construction of the system was performed by
Techint beginning in 2002. Commercial operation of the system began in August 2004.

The system consists of two buried pipelines: 1) a natural gas (NG) pipeline, which runs from the
upstream facilities at Malvinas to a terminal station at Lurin, at the southern edge of Lima; and
2) a natural gas liquid (NGL) pipeline, which transports the liquid condensates from Malvinas to
a fractionation plant near Pisco, on the coast of Peru south of Lima. The two pipelines share a
common right-of-way (ROW) that traverses the Peruvian jungle, climbs over the Andes
Mountains at a peak elevation of approximately 4,800 m, and descends steeply toward the
Pacific coast.. The alignment of the ROW, which is shown on Figure 1, follows a 3-km–wide,
government-mandated corridor. Deviations from this pre-approved corridor had to be granted by
the Peruvian government. Within this 3-km-wide corridor, the cleared section of the ROW was
typically restricted to a width of 25 meters.

The NGL pipeline is approximately 561 km long, and the NG pipeline is approximately 734 km
long, traversing the roughly 200-km-long selva sector, the ~300-km-long sierra sector, and the
200-km-long costa sector.11 Along this route, the NGL pipeline telescopes from a nominal pipe
diameter of 14 to 10¾ inches, and the larger NG pipeline telescopes from a nominal pipe
diameter of 32 to 24 to 18 inches. The wall thickness of the NGL pipeline ranges between 0.219
and 0.469 inch, and the wall thickness of the larger NG pipelines ranges between 0.406 and 1
inch, with heavier wall thickness typically used for sections with larger diameter. Both pipelines
are constructed of tubular high-strength steel in conformance with the American Petroleum
Institute (API) 5L standard and sleeved in a protective plastic layer.

11
True length of pipeline.

SF36292.003 A0F0 0607 0806

1
 June 8, 2007

The NGL pipeline is equipped with four pump stations (PS) on the east side of the Andes
Mountains and three pressure reduction stations (PRS) on the west side of the Andes Mountains.
The pump stations propel the liquid hydrocarbons up the Andes Mountains, and the pressure
reduction stations reduce the pressure as the liquid hydrocarbons flow downhill to the
fractionation plant located along the Peruvian coast. The NGL pipeline is also equipped with
nineteen block valves and ten check valves, mostly at major river crossings, to minimize the
amount of potential spillage. All block valves are connected to an automatic leak detection
system to quickly shut down the NGL pipeline in case of a pipe rupture. The NG pipeline
receives the processed gas products from Pluspetrol’s exploration and processing facility in
Malvinas at a sufficiently high pressure such that no further compression is currently required.
On the west side of the Andes Mountains, a pressure control station (PCS) reduces the pressure
of the natural gas. The PCS is equipped with several redundant safety features and a venting
capability to protect the pipeline from over-pressurization events.

Figure 1. Right-of-way of the Camisea Transportation System in Peru, with the extraction
and production (E&P) centered in Malvinas, a NGL fractionation plant in Pisco,
and a NG distribution point in Lurin.

SF36292.003 A0F0 0607 0806

2
 June 8, 2007

The system is centrally operated from Lurin using a state-of-the-art supervisory control and data
acquisition (SCADA) system. The SCADA system gathers all operational data and allows direct
remote control of all components. The system communicates primarily via its own fiber optic
cable that is installed along the pipelines and has two technologically different and independent
backup systems: a satellite and a radio frequency–based system.

1.2 Operational History

Since the Camisea Transportation System was placed into service in August 2004, the NGL
pipeline has experienced six incidents involving a release of NGL (see Figure 1). The NG
pipeline has not experienced a known gas release. The first (KP 8+850) and fifth (KP 125+950)
incidents were similar, in that external soil loading is currently considered to have been a
significant contributor to the subcritical crack growth in the pipe. An explosion occurred as a
consequence of the fifth incident. The second incident, located at KP 222+500, appears to have
been primarily the result of a time-delayed, hydrogen-induced crack in a girth weld. The third
incident occurred in an area that was well studied from a geologic perspective and known to be
an area of high risk of landsliding. The rupture of the pipe at this location is attributed to
overload caused by a landslide. The fourth incident, located at a river crossing at KP 50+900,
appears to have been induced by mechanical damage to the exterior concrete coating and a dent
in the exterior wall of the NGL pipeline and been triggered by riverbed scouring caused by a
flash flood. The sixth incident occurred upstream of the fifth incident in April 2007 and is
currently under investigation by TgP. A more detailed description of each incident is given in
Section 3 of this report.

1.3 Report Organization

This report is structured in three main parts: 1) the review of the six spill incidents of the NGL
pipeline; 2) our review of the design and construction methods and their risk assessment; and
3) the identification and evaluation of the geotechnical and geologic risk, seismic risk, scour risk,
and pipe-related risks of this system. In this regard, our work focused on the NGL pipeline.

SF36292.003 A0F0 0607 0806

3
 June 8, 2007

The spill incidents and their probable cause are reviewed in Section 3. Section 4 describes the
system’s mechanical and geotechnical design and evaluates whether any systematic risks were
introduced during the design process. Section 5 addresses construction and issues related to the
geotechnical mitigation measures that were constructed during this period. Issues related to pipe
material, pipe manufacture, and pipeline fabrication are also discussed in Section 5. Sections 6
through 9 describe both the hazards and means that were and will be used to mitigate pipeline-
related risks in the following four areas: geotechnical and geological risks, pipe integrity risks,
seismic risks, and scour risks at river crossings. Section 10 presents our overall conclusions and
recommendations for future activities.

SF36292.003 A0F0 0607 0806

4
 June 8, 2007

2 Exponent’s Investigation

2.1 Objective

Exponent® Failure Analysis Associates (Exponent) was retained by the Inter-American

Development Bank (IDB) to perform a pipeline integrity study of the Camisea Transportation
System. Exponent’s retention in April 2006 followed the occurrence of five spill incidents
during the first 19 months of operation, each of which resulted in the release of hydrocarbons.
The primary objectives of our investigation were to: 1) develop a risk profile for the two
component pipelines and identify the factors that contributed to the spill incidents, 2) investigate
the causal factors of the first five spill incidents, 3) review key information on the design and
construction of the system, 4) develop and assist TgP in implementing a qualitative risk
assessment method to evaluate geotechnical and geological hazards, 5) evaluate the effectiveness
of the geotechnical measures constructed in 2006 to stabilize the ROW, 6) evaluate the pipe
inspection program, 7) identify and evaluate seismic hazards, and 8) explore possible
contributions of scour to the risk of future pipeline failure at stream and river crossings.

During this investigation, Exponent made recommendations to TgP intended to improve future
pipeline integrity by mitigating and controlling identified risks to the system. Although TgP had
undertaken a substantial program to provide stabilization measures along the ROW beginning in
April 2006, TgP also implemented our interim recommendations. This report also summarizes
these efforts and provides a risk-based evaluation of the system that incorporates extensive
sources of information and field investigations by Exponent, commencing in April 2006.

2.2 Phases of Work

Exponent’s pipeline integrity analysis was conducted in two integrated phases. The objective of
the first phase was to provide a forward-looking, overall assessment of the integrity of the
pipelines, primarily the NGL. The following components were included in our evaluation of risk
in the first phase: pipe material, pipeline design, pipeline construction, geologic and

SF36292.003 A0F0 0607 0806

5
 June 8, 2007

geotechnical hazard mitigation, pipeline maintenance, and ongoing operation. Exponent

prioritized the identified hazards and evaluated the efficacy of currently used mitigation and
control measures. The first phase included an evaluation of the suitability of the seismic design,
mechanical design, and design of river crossings (scour analysis). As part of that study, we
established a baseline risk level for the system and performed a technical review of the five spill
incidents that occurred in the system between December 2004 and March 2006.

The objectives of the second phase were to further evaluate key risks and to evaluate the progress
made by TgP and COGA in reducing risks to the pipeline. The risk was evaluated following the
implementation of various mitigation measures constructed in 2006, which was compared to the
baseline risk established during the first phase of our investigation. Specifically, the second
phase included: 1) developing and assisting in the implementation of a qualitative risk
assessment method to evaluate potential geotechnical and geologic hazards to the pipeline
system, 2) evaluating the effectiveness of the geotechnical stabilization measures constructed in
2006 to mitigate external soil pressures acting on the pipelines, and 3) evaluating the efficacy of
the current pipe integrity program.

2.3 Scope

In order to accomplish the pipeline integrity analysis, Exponent reviewed more than 400 sets of
documents related to the design, construction, operation, and maintenance of the Camisea
Transportation System. These documents included engineering specifications, construction
specifications, pipe material data, pipeline design drawings and calculations, geotechnical and
geological studies, seismic studies, hydrological studies, river-crossing studies, construction
progress surveys, various internal and external pipeline inspection reports, and operational data.
In some cases, Exponent performed its own engineering analysis to quantify certain risks to the
integrity of the pipeline. Exponent did examine the radiographs of the spill incidents and some
additional of the more than 100,000 girth weld radiographs..

In addition to these activities, a multi-disciplinary team of Exponent engineers and scientists

performed inspections along the pipeline ROW in June and September 2006. These inspections

SF36292.003 A0F0 0607 0806

6
 June 8, 2007

occurred at more than 50 sites along the first 450 km of the ROW where geotechnical
stabilization measures were proposed, were under construction, or had been completed. Special
attention was given to the first 220 km of the ROW, where all spill incidents have occurred and
geotechnical as well as geologic risks are considered to be highest. Exponent personnel also
interviewed key personnel involved with design, construction, operation, and maintenance of the
system. These interviews were supplemented with numerous teleconference calls that included
the designers, operators, service providers, and independent consultants hired during the
construction and maintenance of the pipeline by TgP. Finally, Exponent participated in the
metallurgical investigation of samples of pipe that were involved in two of the incidents, and
reviewed the metallurgical examination reports and evidence from the first five spill incidents.

SF36292.003 A0F0 0607 0806

7
 June 8, 2007

3 Review of NGL Pipeline Incidents

3.1 Background

Since the Camisea Transportation System was placed into commercial service in August 2004, a
total of six spill incidents have occurred along the NGL pipeline. The locations and dates of
these six spill incidents are shown on Figure 1. Exponent reviewed information related to the
first five NGL pipeline spill incidents as a means of evaluating risk, and to consider the potential
for systemic problems. Exponent was not retained to perform a root-cause analysis of any of the
spill incidents.12 All of the spill incidents occurred in the first 222 km of the NGL pipeline—
four of the spills occurred in the selva sector, one in the transition zone between the selva and the
sierra sectors, and one in the sierra sector. In contrast, no leaks have occurred on the larger
diameter NG pipeline.

3.2 First Incident

The first incident occurred on December 22, 2004, in the jungle sector near KP 8+850. 13 The
NGL pipe has a diameter of 14 inches and a nominal wall thickness of 0.25 inch along this
section of the system. The topography at this site is characterized by undulating hills at a low
elevation (~400 m). The hills contain gentle to moderately steep slopes bisected by short ravines
that rise above the Urubamba River. At the point of rupture, the pipe was on an intermediate
bench14 and near a small stream, with a slope profile of approximately 15%.

This location had been inspected by GEOTEC15 in September 2004. In-place geotechnical
mitigation measures consisted of surface drainage channels constructed of soil-cement bags, and

12
Root cause analysis is the integrated evaluation of all facts pertaining to the investigated failure to uniquely
identify the cause or causes of failure.
13
Distance along the pipeline as measured from Malvinas and marked along the ROW at every kilometer point
(KP).
14
A relatively level strip of land bounded by steeper slopes above and below.
15
GEOTEC was the geotechnical engineering firm that supervised the implementation of geotechnical stabilization
measures during construction of the pipeline system.

SF36292.003 A0F0 0607 0806

8
 June 8, 2007

revegetation with grass. Cracks in the soil and displacement of some of the drainage channels
were interpreted as evidence of soil settlement. GEOTEC recognized that the cracks provided a
route for rainfall runoff to infiltrate the soil. They recommended sealing of cracks and repairing
channels and current breakers, but based on available information, they believed that there was
no need for significant remedial geotechnical stabilization work.

The pipeline rupture was detected in TgP’s control room in Lurin as an anomaly in the pressure
characteristics of the pipeline between PS #1 and PS #2. Seven minutes after noting the
anomaly, TgP stopped pumping hydrocarbons at PS #1, and by the ninth minute, the rupture site
was isolated between the block valve at the outlet of PS #1 and the block valve on the east bank
of the Urubamba River at KP 12. During their site investigation, TgP discovered that the pipe
was circumferentially cracked in the vicinity of a girth weld16 but was not completely severed.
The hydrodynamic forces of the escaping NGL had formed a sinkhole above the pipeline. TgP
estimated that 260 m3 of hydrocarbons were released.17 No fires or explosions occurred as a
consequence of the spillage.

Exponent inspected this site on June 12, 2006, and interviewed TgP personnel who were
knowledgeable about the history of construction, repairs, and stabilization measures. Both
pipelines were originally buried at a depth of up to 7 m. Remediation and geotechnical
stabilization work completed before our visit included repair of the pipeline, cleanup of
hydrocarbon-containing soil, addition of ditch breakers to control subsurface seepage along the
pipeline, and construction of retaining walls to mitigate lateral earth movement in the ROW. At
the time of our June 12, 2006, visit, the 14-inch NGL line was exposed, and additional
stabilization measures were being constructed to stabilize a shallow landslide that had occurred
in the vicinity of the rupture. Selected photographs of the area from that visit are shown in
Figure 2.

Exponent visited this site again on September 11, 2006, when additional geotechnical
stabilization measures were nearly complete. A substantial amount of overburden soil had been

16
Circumferential weld joining two pipe sections.
17
Stone and Webster Management Consultants (April 2006). “Analysis of Incident Report: Camisea Pipeline
Project.”

SF36292.003 A0F0 0607 0806

9
 June 8, 2007

removed from the ROW to improve the stability of the slope, and the depth of the NGL pipeline
at that time was reported to be about 1.0 to 1.5 m. Figure 3 provides an overview of the newly
constructed geotechnical stabilization measures at the site on September 11, 2006, looking
downstream. Stabilization measures included the construction of surface and subsurface
drainage features,18 as well as retaining walls and piles to stabilize the slope and prevent lateral
earth movement along the ROW. The extensive monitoring program being implemented for the
site consists of strain gauges to detect pipe movement, inclinometers to detect soil movement,
and piezometers to monitor the groundwater level.

18
Current breakers and surface channels were installed to control surface runoff, whereas ditchbreakers were
constructed to control subsurface water flow along the pipeline.

SF36292.003 A0F0 0607 0806

10
 June 8, 2007

Figure 2. General area of site of first spill incident, at KP 8+850, during ROW stabilization
work. Top: View downstream (looking downslope).
Bottom: View upstream (looking upslope). Plastic sheeting covers 14-inch NGL
pipeline. Photographs taken on June 12, 2006.

SF36292.003 A0F0 0607 0806

11
 June 8, 2007

Figure 3. General area of first spill incident, at KP 8+850, looking downstream, after
stabilization work conducted in 2006. Photograph taken on September 11, 2006.

At the request of TgP, Metallurgical Consultants Inc.19 (MCI) performed a metallurgical

investigation of the ruptured pipe section. MCI concluded that the fracture occurred between the
4 and 7 o’clock positions20 (about 10 inches long), with the crack beginning at the pipe’s outer
surface, as shown in Figure 4. MCI stated that this initial crack formed when offset fatigue
cracks coalesced, as evidenced by the presence of ratchet marks on the fracture surfaces.
Subsequent loading propagated this crack subcritically, with rapid incremental tearing in a few
load steps, until the crack became critical, and the wall of the pipe was breached (see Figure 4).

19
Metallurgical Consultants, Inc. is a Houston-based metallurgical laboratory.
20
The circumferential position on a pipe is referenced to a clock, with the 12 o’clock position being straight up and
the viewer looking downstream at the clock.

SF36292.003 A0F0 0607 0806

12
 June 8, 2007

Figure 4. MCI photograph showing frontal view of fracture surface of the first rupture at
KP 8+850.

Analyses done by GIE (GIE S.A.), IGL (Ingenieria y Geotecnia LTDA), and MCI, as well as our
own investigation, identified soil movement and the possible loss of pipe bedding21 due to water
infiltration as a significant contributor to the rupture of the NGL pipeline. Our current
understanding is that the circumferential crack grew subcritically as soil loading bent the pipe
until it ruptured. MCI’s metallurgical investigation of the failed pipe confirmed that the pipe
material and weld were in compliance with applicable codes but minor allowable anomalies per
API 1104 were identified.

3.3 Second Incident

The second incident was discovered on August 29, 2005, approximately nine months after the
first incident. The incident, referred to as the Pacobamba failure, occurred in the mountain sector
near KP 222+500. This spill incident occurred at an elevation of approximately 3,950 m on a
fairly broad topographic ridge crest at the head of a long valley. As shown in Figure 5, the
pipeline ascends an east-facing slope and is approximately 30 m away from the unimproved San

21
Soil placed beneath the pipe in the trench.

SF36292.003 A0F0 0607 0806

13
 June 8, 2007

Antonio-Pacobamba Road. At this location, the 14-inch NGL pipeline has a nominal wall
thickness of 0.219 inch, a slope profile of approximately 15%, and a burial depth of 1 to 1½ m.

A local resident discovered a small hydrocarbon surface stain caused by a minor leak in the NGL
pipeline. TgP’s control room received this information at 19:34 hours and commenced a leak
investigation of the NGL pipeline. Twenty minutes later, the pipeline was shut down, and at
20:04, the block valves upstream and downstream were closed to isolate the rupture site. On
September 1, 2005, at 15:00, the field investigation, NGL pipeline excavation, and temporary
repair of the NGL pipeline using a slip-on sleeve were completed.

On June 14, 2006, Exponent staff inspected this site and interviewed TgP personnel familiar with
the rupture. Remediation and stabilization work completed before the inspection included
temporary repair of the pipeline and cleanup of hydrocarbon-containing soil and rock. At the
time of our visit, the NGL line was exposed in a series of three trenches, and a bypass pipe was
being installed to enable removal of the fractured pipe section with the slip-on sleeve and
installation of a permanent repair. Selected photographs of the area from that visit are shown in
Figure 5.

MCI analyzed, with initial participation of Exponent, the removed fractured pipe section, which
verified the existence of a small (approximately 2- to 2.5-inch-long) circumferential crack along
a girth weld of the NGL pipeline. It is noteworthy that the normal operating pressure is very low
at this elevated location, just upstream of PS #4, which contributed to the low leakage rate
(significantly less than a fraction of 1% of the NGL pipeline’s product flow rate). MCI also
performed a detailed investigation of the fracture surface of the NGL pipeline. MCI identified
the leaking weld as being part of a tie-in section approximately 1.3 m long that connected two
much longer sections of piping.

SF36292.003 A0F0 0607 0806

14
 June 8, 2007

Road Ridge

Figure 5. Location of second spill incident, at KP 222+500, during permanent pipeline

repair work. Top: View upstream (downslope) along excavated trench along
NGL pipeline. Bottom: View downstream (upslope) across trenches opened for
repairs. Photographs taken on June 14, 2006.

SF36292.003 A0F0 0607 0806

15
 June 8, 2007

MCI’s investigation indicates that the leakage began as a time-delayed, hydrogen-induced crack
on the inside surface of the pipe and extended radially to a crack depth of more than 90% of the
wall thickness (see Figure 6). Mild steels used in pipeline construction are known to be
susceptible to hydrogen-induced cracking (HIC). Because hydrogen can be introduced
temporarily during welding, welded regions of the pipe are particularly susceptible. Typically,
these cracks can take between several hours and several days to develop after welding. While
the hydrogen-induced crack was evident in the post-failure radiograph, it was not visible in the
post-weld radiograph that was performed according to TgP one day after welding. This 90%
deep hydrogen-induced crack, though close to critical, was able to pass the subsequent
hydrostatic test, performed approximately 5 months after welding. Pressure cycling during
hydrostatic testing and the subsequent pressure fluctuations during operation of the pipeline
further destabilized the crack, which ultimately caused—at a much later time during normal
operation of the NGL pipeline—the formation of a tight through-wall crack from which the NGL
escaped very slowly. Exponent’s current understanding is that the hydrogen-induced crack
originated in the weld, and that the low normal operating pressure at this location likely delayed
the rupture of the weakened NGL pipeline.

Figure 6. MCI photograph showing frontal view of fracture surface of the pipe from the
second spill incident of the NGL pipeline at KP 222+500. White arrows indicate
the weld’s root pass in the background to the fracture surface; red arrows
indicate the inner surface of the pipe where the hydrogen-induced crack began;
and yellow arrows indicate the narrow remaining ligament at the outer surface of
the pipe.

SF36292.003 A0F0 0607 0806

16
 June 8, 2007

3.4 Third Incident

The third incident, referred to as the Toccate failure, occurred on September 16, 2005. This failure
occurred in the upper jungle sector near KP 200+700 on the Pacobamba variant, a portion of the
ROW that was rerouted before construction, primarily for environmental reasons. This rupture was
caused by a fairly substantial landslide that extended across the ROW and severed a girth weld that
joined a bend (wall thickness of 0.344 inch) to a straight pipe (wall thickness of 0.219 inch).

This section of the ROW is on a tall, steep, west-facing valley wall and follows a long,
unimproved road between San Antonio and Toccate. The third incident occurred approximately
1 km northeast of the town of Toccate at an elevation of approximately 2,350 m. Various studies
conducted prior to construction had identified this area as having many unstable geologic zones
that would require special engineering and construction methods. The location was described as
having a very high physical risk due to the potential for shallow landslides,22 rockslides,23 and
debris flows.24 Based on these findings, some geotechnical mitigation measures were installed to
stabilize the slope and protect the pipeline.

Mitigation measures implemented during construction included the establishment of a narrower

ROW to minimize the heights of cut slopes, and carefully controlled blasting to minimize
vibrations that could loosen rock masses above and below the roadway. GEOTEC’s October
2004 report, two months after the system was placed into operation, describes the area as the
most unstable site that they evaluated along the ROW, and the area of greatest risk for the
stability of the pipelines in case of ground movement. Evidence of ground movement in the
form of cracks and distorted retaining walls was noted in two zones, one above the pipelines and
one below. The upper zone of movement was adjacent to a slope failure that occurred after the
installation of the pipelines and forced their relocation. GEOTEC recommended long-term
monitoring and geotechnical stabilization measures. Exponent’s site visits along this section of

22
A mass of soil and/or detached bedrock that slides downslope. The moving mass is generally thinner than about
3 to 6 m. Movement can be slow or rapid – measured in meters per month to meters per second.
23
A mass of detached bedrock that slides downslope. The mass can be less than a meter to many tens of meters
thick. Movement is generally rapid – measured in meters per day to meters per second.
24
A mass of mud and rocks that flows downslope, generally during or shortly after heavy rainfall. The mass is
usually no more than a few meters thick. Movement is rapid – measured in meters per second.

SF36292.003 A0F0 0607 0806

17
 June 8, 2007

the ROW slopes showed evidence of incoherent, highly fractured rock and locally thick soil
along very steep slopes that extend a long distance above and below the road.

The rupture was detected in the control room due to a pressure drop at PS #3 on September 16,
2005. At 00:15 hours, the operating system reported normal operating conditions. A drop in
pressure and flow rate was registered at 00:16, which automatically activated the upstream and
downstream block valves to isolate the rupture site. One minute later, the pipe operator stopped
pumping NGL. The total amount of released NGL was estimated to be approximately 1,102 m3,
of which approximately 410 m3 remained after evaporation.25 No fires or explosions occurred as
a consequence of the rupture. TgP’s technical report on this rupture identifies several measures
that were deployed to capture the NGL liquid downslope of the rupture site, as well as along the
possible flow path of the Chunchubamba River, a tributary to the larger Apurimac River.

Exponent inspected this site on June 14 and September 19, 2006. Selected photographs of the
area from the first visit are shown in Figure 7. Remediation and stabilization work completed
before our June 2006 visit included construction of a bypass pipeline beneath the roadway, and
completion of various interim geotechnical works along approximately 500 m of the ROW. As a
result of ongoing stability problems at the site, a bypass tunnel, approximately 730 m long, was
under construction from Rio Corimayo northeast of the site to the next ravine southwest of the
site. Remediation works completed by September 19, 2006, just prior to completion of the
tunnel, included widening the road by cutting back the base of the upper slope, removing about
60,000 m3 of unstable material above the road, benching the slope above the road, building a
new retaining wall and subdrain at the base of the road cut, improving surface drainage, and
planting trees. Although the pipeline was to be rerouted through the tunnel, away from the
unstable area, stabilization work was performed to allow continued use of the San Antonio–
Toccate Road through the failure area.

25
Inter-American Development Bank (2006), “Camisea Project: Incident Report”.

SF36292.003 A0F0 0607 0806

18
 June 8, 2007

Landslide

Walls

Figure 7. General area of the third spill incident, at KP 200+700, during stabilization work.
Top: View upstream showing large landslide above and below ROW.
Bottom: View downstream along ROW showing alignment of temporary bypass
pipeline (beneath stacked sacks) and new retaining wall at base of cut slope.
Photographs taken on June 14, 2006, and September 19, 2006.

SF36292.003 A0F0 0607 0806

19
 June 8, 2007

MCI performed the metallurgical investigation of the pipe from this incident. Figure 8 shows the
ruptured NGL pipe section that was completely severed by the landslide in a ductile overload
tension failure. MCI determined that the fracture had no distinct fracture origin and occurred in
the base metal and heat-affected zone next to the circumferential weld. MCI conducted various
tests to quantify the material’s strength, hardness, and ductility and determined that it showed
satisfactory properties and compliance with the applicable codes. MCI confirmed that the weld
was in compliance with the applicable codes. Exponent reviewed MCI’s investigation and
visually inspected the failed pipe section.

In summary, Exponent concludes that the most significant contributing factor to the failure of the
pipe was external geotechnical forces induced by a large landslide that extended across the
ROW.

Figure 8. MCI photograph of pipe from the third spill incident, at KP 200+700.

SF36292.003 A0F0 0607 0806

20
 June 8, 2007

3.5 Fourth Incident

The fourth incident was detected on November 24, 2005, at the crossing of the Paratori River
(KP 50+900). The pipelines approach the river on relatively flat terrain, and ascend a steep slope
on the opposite side of the river. The river is less than 10 m wide, but the fast flow and the
presence of large boulders (approaching 3 m in maximum dimension) suggest a steep gradient.
Both pipelines were originally buried about 2.1 m below the stream bottom.26 At this location,
the NGL pipeline has a diameter of 14 inches and a nominal wall thickness of 0.219 inch. In
addition to the protective polyethylene cover that sheaths each pipeline, both the NGL and NG
pipelines were covered with a reinforced concrete layer at the river crossing, which is standard
construction practice. After installation of the pipelines, a cleaning pig with plates was passed
through the entire pipeline, including this river crossing section.27

TgP has indicated to Exponent that the sizing plates did not identify any significant deformation
to the pipe. Furthermore, the hydrostatic test of this section was performed and did not indicate
any leakage. The October 2004 GEOTEC report describes the river as carrying a high flow of
water, which can rise 1.0 to 2.0 m during the rainy season. A protective rockfill layer of
boulders 1.0 to 3.0 m in diameter had been installed along the river banks to protect the pipeline
from lateral erosion.

TgP operations personnel in Lurin identified this rupture by detecting a reduction in flow
downstream at PS #2 (KP 108). At approximately 04:55 in the morning of November 24, 2005,
the operational flow rate decreased by approximately 20% over a period of 5 minutes. At 05:00,
pumping was stopped by TgP. The pressure reduction was not sufficiently large to activate the
automatic rupture detection mechanism of the block valves upstream and downstream of the
rupture. Consequently, operations personnel closed the block valves 55 minutes after pumping
stopped and the leak analysis was completed.28 Estimates29 of total NGL volume leaked indicate
a loss of approximately 736 m3. No fires or explosions occurred as a consequence of the rupture.

26
The “as-built” drawing at this specific location identifies a minimum burring depth of 2.1 m for the NGL pipeline.
27
An inline pipe tool to clean the pipeline.
28
According to TGP procedures, the first action to perform after indication of a leak is to stop the pumping and start
an operating analysis to identify the segment of pipe involved in the leak.

SF36292.003 A0F0 0607 0806

21
 June 8, 2007

On June 13, 2006, Exponent made observations and interviewed TgP personnel who were
familiar with the construction and stabilization measures at this site. Selected photographs of the
area from this site visit are shown in Figure 9. A flash flood reportedly preceded the rupture.
The maximum water level was reported to be on the order of 2 to 3 m above the stream level
observed during our site visit (see Figure 9). Subsequent to the failure of the NGL pipeline, a
steel truss bridge was installed to support an aerial bypass pipeline. The original NGL pipeline
section that crossed the river was excavated in late May 2006.

Excavation revealed that the concrete coating of the NGL pipeline was missing over a length of a
few meters and that the slightly-higher-positioned NG pipeline’s concrete coating showed some
signs of erosion. TgP removed the section of the NGL pipeline containing the rupture and sent it
to MCI for metallurgical evaluation. Exponent participated in the initial metallurgical laboratory
investigation of this rupture in July 2006.

MCI’s external examination of this pipe section revealed a deep dent and an external groove that
bisected the approximately 7.5-inch-long rupture. No girth welds were present in the vicinity of
the rupture. Figure 10 shows that the rupture is inclined 50 degrees to the pipe axis below the
groove, and is nearly vertical above the groove. The groove itself is inclined at an angle of
55 degrees with respect to the pipe axis. The groove was more likely caused by a gradual
application of pressure due to construction equipment than by the scraping movement of rocks
carried by streamflow. Testing of the pipe material from this location revealed satisfactory
material characteristics.

29
Inter-American Development Bank (2006), “Camisea Project: Incident Report”.

SF36292.003 A0F0 0607 0806

22
 June 8, 2007

Approximate
Location of
Rupture

Approximate
High Water
Mark at
Rupture

Figure 9. Site of fourth spill incident, at KP 50+900 showing river crossing and new steel
truss bridge carrying the NGL bypass pipeline. Top: View looking downstream
(uphill). Bottom: View looking upstream. Photographs taken on June 13, 2006.

SF36292.003 A0F0 0607 0806

23
 June 8, 2007

Figure 10. Observed groove (white arrows) and rupture (red arrows) on the outer surface of
the pipe, with the pipe being horizontally aligned and NGL product flow being
from the right to the left. Black material is the damaged protective polyethylene
cover.

In summary, although the initial mechanical damage that caused the dent and groove initiated the
subsequent rupture, it was not sufficient to breach the wall of the pipe. Indeed, the pipe wall at
this location apparently was capable of containing the NGL fluid until some unknown external
loading event caused the already weakened pipe wall to fail in ductile overload at the damaged
area. This triggering loading event could have been associated with the flash floods that
immediately preceded detection of the leak. It is currently believed that the damage sustained to
both the NGL pipeline’s concrete coating and to the pipe itself was most likely associated with
external mechanical damage caused by human activity.

SF36292.003 A0F0 0607 0806

24
 June 8, 2007

3.6 Fifth Incident

The fifth incident occurred on March 4, 2006, in the jungle sector at KP 125+950. At this
location, the NGL pipeline has a diameter of 14 inches and a nominal wall thickness of
0.375 inch. This section of the pipeline is in a broad topographic swale,30 below a saddle,31 on a
ridge about 2 km southwest of Rio Cumpirusiato. The slopes at the site are gentle to moderate
and are bisected in some places by shallow ravines. The geological and geotechnical reports of
this sector indicate the potential for slope instability to be “high” to “very high,” with medium
potential for mudflows. Site soils are described as wet, but with good surface drainage and
moderately good subsurface drainage. At the point of rupture, the pipe was near the base of a
slope above an intermediate bench, as shown in Figure 11. The NGL pipeline at the rupture
location has a slope of about 15% to 20%, a burial depth of approximately 1 to 2 m, and an
elevation of approximately 1,000 m. The NG pipeline was buried at a slope and depth similar to
the NGL pipeline.

TgP’s control-room personnel in Lurin detected this rupture following a pressure-drop warning
from an upstream block valve at 15:27:00 in the afternoon. A few seconds later, this block valve
automatically started to close and was verified to be closed at 15:28:12. Pumping at the
upstream pump station (PS #2) was reported to have ceased 20 seconds after the pressure drop
was detected. The downstream block valve was closed at 15:32:12, approximately 5 minutes
after the pressure drop was detected.

Subsequent site inspection revealed that the NGL pipeline was completely severed at the rupture
point. The resulting hydrodynamic force of the escaping NGL formed a sinkhole. The rupture
of the NGL pipeline resulted in an explosion, which was apparently ignited by a source in a farm
dwelling located downhill from the rupture. Osinerg filings indicate that 661 m3 of NGL was
released during this rupture. No fatalities, but two injuries, were reported.32

30
A swale is a small valley, generally on a hillside.
31
A saddle is a low point on a ridgeline.
32
Inter-American Development Bank (2006), “Camisea Project: Incident Report”.

SF36292.003 A0F0 0607 0806

25
 June 8, 2007

New
Stabilization
Measures

Figure 11. General area of the fifth spill incident, at KP 125+950, during stabilization work.
Top: View upstream (downslope) along the ROW.
Bottom: View downstream (upslope) across excavation made to remove
unstable soil. Photographs taken on June 13, 2006.

SF36292.003 A0F0 0607 0806

26
 June 8, 2007

On June 13 and September 20, 2006, Exponent inspected this site and interviewed TgP personnel
who are familiar with the construction and stabilization measures. Selected photographs of the
area from that visit are shown in Figure 11. The rupture was believed to have been caused by a
shallow landslide that was part of a complex, interconnected soil movement of at least three
landslides that occurred at about the same time. The first, lowest, landslide was located in the
area that has since been excavated; the second was at the point of failure; and the third,
uppermost, was upslope along the ROW.

Remediation and stabilization work completed before our visit included repair of the pipeline,
cleanup of hydrocarbon-containing soil, installation of subdrains along and across the ROW,
construction of a buried retaining wall parallel to and to the left of the NGL pipeline with
stabilizing ties to the pipeline, excavation of unstable soil to the left of the point of rupture and
extending left of the ROW, construction of the foundation and first above-ground layer of a
retaining wall in the excavated area, and construction of most of the length of a concrete-lined
channel along the right edge of the ROW to replace the original berm-type drainage channel
lined with sacks of soil cement.

MCI was retained to perform the metallurgical investigation of this rupture at KP 125+950. MCI
performed a detailed investigation of both the fracture surface of the NGL pipeline and the
material properties of the pipe. MCI concluded that crack nucleation occurred at the bottom
portion of the pipe in the heat-affected zone of the weld. This nucleation zone was characterized
by mostly ductile narrow slant fractures that joined to form a single large crack with a depth of
approximately 0.055 inch (see Figure 12). Subsequent loading propagated this crack, with rapid
incremental tearing, to form a flat fracture that followed a 10.5-inch-long arc.

Using scanning electron microscopy (SEM), it was determined that the nucleated crack grew
through about half the thickness of the pipe wall, after which its rate of growth increased to
approximately 0.01 inch per load step. The crack then became critical, and the pipe wall was
breached and severed around its complete circumference. As with the preceding four failures,
MCI determined that the ductility, hardness, strength of the pipe materials and weld materials were
in compliance with the required codes.

SF36292.003 A0F0 0607 0806

27
 June 8, 2007

Figure 12. MCI photograph showing fracture surface of NGL pipeline at the fifth spill
incident, at an amplification of 14×. Blue arrows demarcate three distinct fracture
zones. White arrows identify the narrow, slanted fracture associated with crack
nucleation. Each zone is numbered, with the outer surface of the pipe being the
top portion of the photograph.

Recent geotechnical studies by IGL,33 as well as Exponent’s field inspection of the rupture site,
indicate that both pipelines pass through a shallow active landslide at this location, and this
shallow slide overlies an older, larger landslide. The cause of movement of the shallow landslide
was likely the infiltration of surface water into the ground due, in part, to the truncation of a
natural drainage crossing the ROW and direct precipitation and surface runoff. We also
understand that a local inhabitant had altered some of the surface drainage in this area prior to
the rupture. We therefore conclude that external forces produced by ground movement, likely
preceded or accompanied by washout of the pipeline’s bedding, were likely a substantial factor
in the rupture of the NGL pipeline at this location. We also believe that the geotechnical
stabilization measures implemented after the rupture have significantly reduced the future risk of
landslides at this location.

33
Ingeneria y Geotecnia LTDA is a geotechnical firm in Columbia that has extensive experience in jungle areas.

SF36292.003 A0F0 0607 0806

28
 June 8, 2007

3.7 Sixth Incident

Recently, TgP identified as part of its pipeline integrity analysis a sixth incident of the 14-inch-
diameter NGL pipeline on April 2, 2007, at KP 125+500. TgP reported that only a small amount
(approximately 0.3 m3)34 of NGL was released. TgP informed us that the NGL pipeline was
repaired by mid-April 2007, and this leak was discovered during planned activities of TgP’s
pipeline integrity management program. The location of this spill incident is shown on Figure 1.
Figure 13 depicts the excavated pipe section with the localized wrinkle away from the girth
welds. Typically, external loading is responsible for the formation of such wrinkles if no thermal
stresses are induced; however, TgP is currently investigating the root cause of this incident. To
date, Exponent has not conducted a complete investigation, and this analysis is based on the
material submitted by TgP to Exponent, including the reports TgP submitted to OSINERG.

Figure 13. Excavated NGL pipe section from the sixth spill incident, with protected wrinkle
from which NGL was reported to have been leaking.

34
Coga, “Estamacion Del Volumen De Perdida De NGL En KP 125+500,” April 16th, 2007.

SF36292.003 A0F0 0607 0806

29
 June 8, 2007

3.8 Summary and Conclusions

Exponent’s review of the first five NGL pipeline spill incidents identified similarities in the
fracture surfaces of the first and fifth incidents, which occurred at KP 8+850 and KP 125+950,
respectively. In both cases, Exponent’s investigation identified unstable geologic conditions as a
significant contributor to the ruptures. In both cases, progressive soil loading is the likely driver
that propagated an initial crack and induced the rupture of the NGL pipeline. In the case of the
first incident, at KP 8+850, the crack resulted in a through-wall leak of about 10 inches in extent;
in the fifth incident, at KP 125+950, the crack resulted in complete severance of the pipe.

The second incident, at KP 222+500, appears to have been primarily the result of a time-delayed,
hydrogen-induced crack in the weld. The hydrogen crack escaped detection by the post-welding
radiography because of the inherent time delay of hydrogen cracking. The toughness of the pipe
material allowed the pipe to pass subsequent hydrostatic testing, even though the crack extended
to approximately 90% of the wall thickness.35 Our current understanding is that the combination
of hydrostatic load cycles and subsequent operational pressure fluctuations caused the initial
hydrogen-induced crack to be further destabilized, ultimately perforating the pipe wall and
releasing the NGL at a very slow rate. The pressure fluctuations at this location, though
frequent, are relatively small, which helped delay the propagation of the crack through the
remaining wall thickness.

The third incident occurred in an area that was well studied from a geologic perspective and was
known to be an area of very high risk of ground failure. While some measures to mitigate this
geologic risk were taken during construction, a sizable landslide ultimately overcame these
measures and both undermined and overtopped the road and ROW. Hence, the rupture of the
pipe at this location is attributed to overload caused by a substantial landslide.

For the fourth incident, at KP 50+900, current information suggests that the rupture of the NGL
pipe was induced by mechanical damage to the exterior concrete coating and a dent in the
exterior wall of the NGL pipeline. Analysis performed to date indicates that this dent was not

35
This is a rather deep crack, and subsequent crack growth due to normal operational pressure fluctuations need be
only minimal to reach a critical crack depth that causes the remaining ligament ahead of the crack to fail.

SF36292.003 A0F0 0607 0806

30
 June 8, 2007

caused by contact with a boulder being washed downstream. Indeed, the pipe wall at this
location was apparently capable of containing the NGL fluid until some unknown external
loading event caused the already-weakened pipe wall to fail in ductile overload in the damaged
area. This triggering loading event could have been associated with the flash floods that
immediately preceded detection of the leak.

The sixth incident is still under investigation, and no final conclusions concerning its origin and
relation to the other incidents have been reached to date.

SF36292.003 A0F0 0607 0806

31
 June 8, 2007

4 Design-Related Risks

4.1 Design Background

Several different engineering disciplines are involved in designing pipelines. The mechanical
engineer is ultimately responsible for the overall pipeline design, but reliable input is sought
from geotechnical engineers and geologists to evaluate the manner in which the pipe will interact
with soils and other environmental components. Commonly, the first step in the mechanical
design process is the hydraulic analysis, where the sizing of pumps, valves, and pipes is
performed. This information is carried over to the next step, a pipe stress analysis, wherein the
designer considers load cases other than the internal design pressure in the pipe. These design
load requirements vary slightly from code to code, but are fundamentally intended to ensure
pipeline integrity and safe operation. This typically leads to an iterative design process between
the engineers performing the hydraulic and the pipe stress analyses, with guidance from the
geotechnical engineer or geologist.

The Camisea Transportation System was designed to the requirements of standards issued by the
American Society of Mechanical Engineers (ASME) that are national standards of the United
States. Specifically, the NGL pipeline was required to conform to ASME B31.4 (Pipeline
Transportation Systems for Liquid Hydrocarbons and Other Liquids), and the NG pipeline was
required to conform to ASME B31.8 (Gas Transmission and Distribution Piping Systems) as
well as the Code of Federal Regulations (CFR), Title 49, Part 192, Transportation of Natural and
Other Gas by Pipeline. The primary purpose of these codes is to establish requirements for
design, construction, inspection, testing, operation and maintenance of pipeline systems.
However, these codes do not provide requirements for all conditions, whether usual or unusual,
that the designer may encounter. Rather, the designer needs to exercise due care and
consideration for all reasonable and expected load cases. In this section, we review the hydraulic
analysis and the geotechnical design considerations for the system during the design phase.
Seismic hazards and scour at river crossings are evaluated in Sections 8 and 9, respectively.

SF36292.003 A0F0 0607 0806

32
 June 8, 2007

4.2 Hydraulic Design Risks

4.2.1 General Review

Exponent reviewed the design hydraulic analysis and operational data from TgP’s SCADA
system. The hydraulic modeling of the system requires as input a well-defined pipeline profile
and the hydraulic performance requirements of the pipeline. Using this information, the
hydraulic analysis determines the internal pressures along the pipeline route. This internal
pressure is then used to compute a minimum allowable pipe wall thickness that adheres to the
code design requirements for the internal pressure. Figure 14 provides a comparison of the
elevation profile of the pipeline for the first 550 km between the initial design and the as-built
condition. The profiles show good agreement, indicating that only minor deviations should be
expected between the design calculations and the as-built alignment.

Figure 14. Comparison of the pipeline elevation profile along the ROW, with the graph
origin being in Malvinas.

SF36292.003 A0F0 0607 0806

33
 June 8, 2007

4.2.2 Hydraulic Design of NGL Pipeline

The hydraulic analysis can be divided into two parts: the steady-state analysis and the transient
analysis. For the NGL pipeline, the steady-state analysis was carried out for flow rates of 10,000
to 70,000 BPD (barrels per day). Our review indicates that the hydraulic analysis relied on
typical modeling techniques to compute the internal pressures along the NGL pipeline. Based on
this internal pressure, a minimum allowable wall thickness was determined. Figure 15 provides
a graphical depiction of the ratio of the maximum internal operating pressure to the maximum
allowable operating pressure (MAOP) per ASME B31.4, where this ratio is to be smaller than 1
at all times, with the smallest margins typically existing downstream of the four pump stations,
at KP 0, 108, 209, and 226.

1.2

1.1PS #1 PS #2 PS #3 PS #4

1
Maximum Operating Pressure/MAOP

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0
0 100 200 300 400 500 600

Distance (Km)

Figure 15. Ratio of the maximum operating pressure to the MAOP along the ROW of the
NGL pipeline.

SF36292.003 A0F0 0607 0806

34
 June 8, 2007

In addition to this static analysis, a transient study was performed using the Stoner36 pipe
simulation software. This analysis determined transient pressures along the NGL pipeline due to
valve closure, pump shutdown and pump startup, or any combination of these events. The
largest transient operating pressures are found along the first 50 km of the NGL pipeline and
downstream of PS #2 at KP 108. Our review indicates that the considered transient events are at
all times smaller than 1.1 times the allowable MAOP, and are in compliance with ASME B31.4.

In summary, our review indicates that the hydraulic analysis of the NGL pipeline has been
properly executed in conformance with code requirements. Furthermore, the actual internal
pressure, flow rates, pumping capacity, and pipeline control in the operating pipeline appear to
be consistent with the hydraulic analysis. The requirements of ASME B31.4 concerning the
internal pressure design have been met, because the hydraulic analysis shows that the Camisea
NGL pipeline does not exceed either the static or transient MAOP. Accordingly, the mechanical
design was carried out in compliance with applicable codes and engineering practice, and no
undue hydraulic risks were introduced.

4.2.3 Hydraulic Design of NG Pipeline

Techint’s hydraulic analysis of the NG pipeline included a steady-state analysis and a blocked-in
load condition providing the largest pressures. The steady-state analysis was carried out for flow
rates ranging from 205 to 215 MMSCFD (million standard cubic feet per day). This static
analysis determines the internal operating pressures along the ROW. In addition, an analysis was
performed that included a compressor station at KP 208 to increase the flow rate.

The hydraulic analysis relied on typical modeling techniques to compute the pipe friction
coefficient and associated pressure drops. The maximum delivery pressure of the NG pipeline at
Malvinas is approximately 147 barg, and the minimum specified delivery pressure at Lurin is
40 barg. The maximum operating pressure is the blocked-in pressure if the block valve at the end
of the pipeline were to be closed. Based on this internal pressure, a minimum allowable wall

36
Stoner Pipeline Simulator (SPS) is widely used for the transient flow simulation of natural gas and
liquid transmission systems.

SF36292.003 A0F0 0607 0806

35
 June 8, 2007

thickness was determined per ASME B31.8. Figure 16 clearly illustrates that the design pressure
is, at all times, larger than the largest expected operating pressure, since the ratio of them is at all
times smaller than 1. The largest margins are typically found at river crossings where the pipe has
been thickened.

1.1

1
Max. Operating Pressure/Design Pressure

0.9

0.8

0.7

0.6
0 50 100 150 200 250 300 350 400 450 500 550 600 650 700 750
Distance (km)

Figure 16. Ratio of the maximum operating pressure divided by the design pressure along
the ROW of the NG pipeline.

In summary, the hydraulic analysis of the NG pipeline is adequate. The operational internal
pressure and flow rates appear to be consistent with the results of the hydraulic analysis. The
requirements of ASME B31.8 concerning the internal pressure design appear to have been met.
The mechanical design was carried out in compliance with codes and general engineering
practice, and no undue hydraulic risks were introduced.

SF36292.003 A0F0 0607 0806

36
 June 8, 2007

4.3 Geotechnical Design Risks

ASME codes require the designer to demonstrate that all reasonable foreseeable load conditions,
which include external loads, imposed by soil pressures or ground movements, be considered
during the design process. External soil loads can be addressed by either designing the pipeline
to withstand external soil loads, or adequately mitigating the geotechnical hazards. Given the
challenging geotechnical conditions and diverse terrain present along the pipeline, particularly in
the selva sector, the system designers opted to construct geotechnical stabilization measures to
mitigate potential external soil loads.

Prior to construction, Golder Associates (Golder) conducted a preliminary geologic and

geotechnical investigation in the selva sector, with the stated objective of determining the
geologic and geotechnical feasibility of the proposed alignment. 37 Golder reviewed aerial
photographs and satellite images, and conducted field inspections along the proposed pipe route.
The Golder study identified the five main geotechnical and geologic hazards in the selva sector
as being landslides, rock falls, debris flows, hillside erosion, and flooding/river erosion. In
addition, Golder developed an algorithm to evaluate the risk at sites along the route and identify
zones of concern. Orlando Felix38 and Dario Verastegui39 performed more detailed geologic and
geotechnical evaluation at thirteen critical sites in the selva sector and made preliminary
recommendations to mitigate hazards. Both the Felix and Verastegui reports recommended
further geological and geotechnical studies during construction, to identify additional unstable
zones and recommend stabilization measures.

MR Associates40 (MRA) conducted a preliminary geologic and geotechnical investigation in the

sierra and costa sectors, with the stated objective of determining the geologic and geotechnical
feasibility of the proposed alignment. MRA reviewed aerial photographs, satellite images,
topographic information, and geologic maps, and conducted field inspections along the proposed

37
Golder Associates, “Final Report – Geologic and Geotechnical Reconnaissance, Natural Gas and Liquid Natural
Gas Pipelines, Route in Sector Selva – Project Camisea,” December 2001.
38
Orlando Felix Salvador, “Pipeline –Sector Selva – Geologic Inspection of the Route – Critical Zones and
Pumping Stations 2 and 3,” February 2002.
39
Dario Verastegui, “Geotechnical Inspection of Route – Camisea Pipeline – Sector Selva,” March 2002.
40
MR Associates, “Final Report – Geologic and Geotechncial Reconnaissance, Route of Camisea Pipeline, Sierra
and Costa Sections,” March 21, 2003.

SF36292.003 A0F0 0607 0806

37
 June 8, 2007

pipe route. MRA recommended implementing surface and subsurface drainage facilities to
prevent erosion and soil creep, and evaluation of alternative route alignments where active slides
or marshy lands were identified; increasing the depth of pipe embedment at sites susceptible to
erosion, debris flows, or scour; identifying locations where excavation cuts would reduce the
stability of the slopes near the pipes; and recommending corrosion prevention near the costa
sections.

The preliminary geotechnical and geologic reports and more specific geotechnical studies
performed during the design phase indicated that the geotechnical and geologic conditions would
be complex and challenging along the alignment.

4.4 Conclusions

The system was designed to comply with ASME engineering code requirements.41 Code
compliance is established if the designer demonstrates that all specific code requirements and all
reasonably foreseeable load conditions are addressed by the design. Foreseeable load conditions
apply to internal pressures and to external loads imposed by soil pressures or ground movement.

Our review indicates that the pipe wall thickness is sufficient to contain the internal pressures of
the transported hydrocarbon products along the entire length of the pipeline (i.e., hydraulic risk).
We have independently verified that the computed design pressures are code compliant and in
good agreement with the measured operational pressures along the whole length of both
pipelines. Therefore, any risks associated with the internal pressure aspects of the design are
minimal and consistent with other pipelines.

Given the demanding route of these pipelines through the jungle and over the mountains,
external loads were expected to be an important consideration in designing and constructing the
system, as evidenced by the geotechnical and geologic studies. Our review revealed that the
pipeline was designed such that external soil loads would be mitigated by geotechnical

SF36292.003 A0F0 0607 0806

38
 June 8, 2007

stabilization measures to be constructed at sites deemed to pose a geotechnical or geological

hazard.

41
ASME B31.4 [Pipeline Transportation Systems for Liquid Hydrocarbons and Other Liquids] is the applicable
Code for the NGL pipeline, and ASME B31.8 [Gas Transmission and Distribution Piping Systems] is the
applicable code for the larger NG pipeline.

SF36292.003 A0F0 0607 0806

39
 June 8, 2007

5 Construction-Related Risks

5.1 Pipeline Construction Characteristics

Techint commenced construction of support infrastructure (i.e., access roads, fuel depots, camps,
pipe staging sites, and loading docks) in 2001. At the end of the 2001–2002 rainy season,42
Techint finalized the pipeline alignment. The pipe route had to be contained within the 3-km-
wide, government-mandated corridor, and deviations from this pre-approved corridor had to be
granted by the Peruvian government. Within this 3-km-wide corridor, the cleared section of the
ROW was typically restricted to 25 meters or less.

In order to minimize the impact on the environment, TgP decided to preferentially build the
pipeline along or near topographic ridges of the government-mandated 3-km-wide corridor. The
first step was to clear and grade the ROW. Due to limitations imposed by the topography of the
ridges and the narrow lane of clearing (25 meters), grading along the ROW typically consisted of
cutting into the ridge or hillside and placing the excavated material as “side-cast” or “spill” fill
on the downhill side(s) of the clearing. Exponent interviewed key individuals involved in the
construction, who informed us that the ground conditions encountered during grading and
installation of the pipe were assessed by geotechnical engineers, and mitigation measures were
constructed at that time.

The pipeline was constructed simultaneously along the ROW between 2002 and early 2004.
Responsibility for constructing the system was divided into the selva, sierra, and costa sectors.
Each sector was supervised by a Project Manager and a Construction Manager, both of whom
were ultimately responsible for construction of the pipeline in their respective sectors. During
construction, Techint employed up to 10,000 people on 12 spreads, with each spread
constructing an assigned section of both pipelines.

42
The rainy season is a time period of elevated precipitation from late October until April.

SF36292.003 A0F0 0607 0806

40
 June 8, 2007

The pipes typically were transported to the individual construction sites along the ROW itself,
because it was the only available route for transportation. Techint used two general types of
construction methods to install the pipes, depending on the terrain: “standard” and “special.”
The “standard installation” method was used in flat to hilly terrain along the majority of the
pipeline ROW. The “special installation” method was reserved for very steep terrain with a
grade more than 35%. Another “special installation” method was also employed at river
crossings, and sites located along roads and narrow mountain ridges.

Techint’s standard installation method appears to have been flexible and adaptable to specific
site conditions. Techint preferred to string out the pipe prior to welding, resting each pipe on
two blocks. Afterward, the pipes were welded together into a continuous section measuring
several hundred meters in length. Each weld was x-rayed 24 hours after completion of the weld.
The pipe trench was then dug and prepared for the pipeline by placing several sand bags, or so-
called pillows, every 3 to 5 meters on the trench bottom, upon which the pipeline would initially
rest. Figure 17 shows an example of the standard construction method in hilly terrain. The
excavated material was placed next to the trench, and protective measures were reportedly taken
to prevent the soaking of the excavated material. The welded pipe sections were lowered into
the trench, and later joined to the already-laid pipeline. If the terrain was flat, the side-booms
moved continuously forward with roller-equipped slings. In some instances, trenching may have
occurred first, if the ROW was sufficiently wide, as depicted in Figure 18. The engineer
responsible for laying the pipe determined the spacing and number of side-booms and other
auxiliary equipment to be used.

SF36292.003 A0F0 0607 0806

41
 June 8, 2007

Figure 17. Standard installation of NGL pipeline at KP 107.

Figure 18. Trenching and stockpiling of cuttings at KP 391.

SF36292.003 A0F0 0607 0806

42
 June 8, 2007

Techint’s special installation procedure for steep hills started with trenching. Once the trench was
dug, a backhoe carrying the pipe was winched from the top of the hill to the end of the pipeline to
place the pipe in position. The pipe was clamped and welded to the pipeline, thereby constructing
the pipeline segment by segment, rather than in long strings as was done in more gently sloping
terrain. This construction method allowed advancement of only approximately two pipe segments
per day. Another specialized construction variation arose at ridges and roads, where the NG and
NGL pipeline share a single narrow trench that may be stepped to further reduce the trench width.
Backfilling of the trench had to satisfy Techint’s construction specification 2794-L-SP-0045,
which provided for the use of selected backfill to contain stones of sizes up to 1½ by 1½ inches.
When possible, the machine depicted in Figure 19 was used to backfill a portion of the trench.
Backfilling was finalized by forming a crown on top of the filled trench and driving an
approximately 4-ton vehicle over the trench to achieve some degree of compaction. In locations
where the pipeline followed a road, concrete plates were placed on top of the backfill to minimize
vehicle loads acting on the pipes.

Figure 19. Placement of selected fill as bedding at KP 358 using a machine that separates
larger stones from the remaining fill.

SF36292.003 A0F0 0607 0806

43
 June 8, 2007

Techint used two types of techniques to construct river crossings: micro-tunneling and
trenching. Micro-tunneling was used solely at the Urubamba River crossing. The most
commonly used technique relied on trenching the riverbed and laying the pipe into the trench. In
some cases, blasting was required to trench. Figure 20 shows the laying of the pipe over the
Manugali River at KP 92. Typically, the pipe was coated with a thick concrete layer to provide
ballast and protection (Construction Specification 2794-L-SP-0043: Concrete Casing of Pipes).
As discussed below, the construction methods employed standard engineering procedures.

Figure 20. Laying of pipe over the Manugali River at KP 92.

SF36292.003 A0F0 0607 0806

44
 June 8, 2007

5.2 Geotechnical Construction Characteristics

During construction of the pipelines, a consortium of geotechnical consultants named GEOTEC

was retained by TgP to develop and supervise the implementation of the geotechnical protection
work from KP 0 to KP 520. Supervision of construction required an evaluation of future risks of
erosion, landslides, rockfalls, and river erosion. GEOTEC was also responsible for evaluating
the geotechnical and geologic risks after installation of the system and commencement of system
operation in August 2004. In October 2004, two months after the system was put into operation,
TgP requested that GEOTEC complete an extensive risk study to characterize the post-
construction stability conditions of the ROW, and recommend measures to prevent, reduce, and
control potential damages from natural phenomena to the system, people, and the environment.43
The report concentrated on critical areas of the selva sector and included descriptions of specific
stations along the ROW, detailing examples of areas of erosion and ground movement, as well as
areas of successful geotechnical stabilization efforts. GEOTEC concluded that geotechnical
stabilization works had improved some of the sites during construction, but other sites already
showed signs of increased deterioration. Additionally, GEOTEC cautioned that maintenance in
the selva sector would be required for at least the first 5 years of operation.

GEOTEC documented numerous manifestations of instability in the selva sector, and to a lesser
extent in the sierra sector. They pointed out that other sites not described in the report could also
have stability problems in the future due to the dynamic nature of the terrain and significant
precipitation. GEOTEC concluded that most of the slope failures evaluated in the sierra sector
had occurred during construction and were believed to have reached equilibrium but to be at risk
of reactivation if subjected to an earthquake or a harsh winter. Sections of the pipelines
constructed along roadways were identified as being at higher risk of failure, whereas cut-slope
failures were judged to pose the greatest threat when the cut is in colluvium because of its origin
as a landslide deposit that could reactivate. The report also noted the potential long-term risk to
the pipelines arising from failures in side-cast fill downslope of the ROW. GEOTEC believed
that the landslide risk was highest between about KP 50 and KP 90, and between about KP 175
and KP 197. The report pointed out that the results were intended to prioritize areas for

SF36292.003 A0F0 0607 0806

45
 June 8, 2007

maintenance. Detailed stability analyses were performed for critical areas. When only limited
detailed data were available, conservative values were assumed.

During our inspections, Exponent noted sites being stabilized where the geotechnical stabilization
measures originally installed during or immediately after pipeline construction were not adequate
to stabilize the ROW.

5.3 Clearing Related Risks

As mentioned previously in Section 5.1, TgP decided to preferentially build the pipeline along or
near topographic ridges, where possible. Based on our field inspections, and discussions with
Techint employees who were directly involved in the construction of the pipeline, Exponent
generally concludes that this approach was appropriate for pipeline integrity purposes and helped
reduce environmental impacts. The route selection along mountain ridges, coupled with the
government mandate to limit the cleared section of the ROW to 25 meters, necessitated the
creation of side-cast fills in some particular areas where there was a limitation of the cleared
ROW section. Although these side-cast fills were generally placed outside the limits of the pipe
trenches, they ultimately became a source of concern related to the potential to exert external soil
pressures on the pipes.44, 45 In our opinion, not all of these side-cast fills were adequately
stabilized during construction, but TgP is currently addressing this issue through the stabilization
measures discussed in Section 6.

The October 2004 GEOTEC report describes numerous sites at which thick fill stockpiles had
experienced ground movement in the form of settlement, cracking, and shallow landsliding. In
many of these locations, soil saturation is mentioned as a key factor in the instability of stockpiles.
At some narrow ridges, excavated soil and rock materials had washed or slid down the steep ridge
flanks, necessitating geotechnical mitigation measures. During our site inspections, Exponent

43
Estudio de Riesgos por Fenomenos Naturales, Proyecto Camisea, Lineas de Conduccion de GN y LGN, October
2004.
44
Camisea Project, Variant Pisco, Field Reconnaissance on July 17, 2002,” memo to Lucio Costarrosa from Milos
Stepanek, no date.
45
“Camisea Project, Pacobamba Route Sector,” memo to Lucio Costarrosa from Milos Stepanek, no date.

SF36292.003 A0F0 0607 0806

46
 June 8, 2007

observed a few instances where this excessive material was apparently used to widen the ROW.
This is generally not a recommended practice, unless the slope is sufficiently stabilized.

During our field inspection, we noted that TgP constructed many geotechnical stabilization
measures in 2006 to address the issues described above. A detailed description of the stabilization
measures undertaken by TgP in 2006 is given in Section 6.

5.4 Trenching-Related Risks

Based on our discussions with Techint construction personnel, the trench was backfilled using
material processed from excavated material stockpiled along the ROW. Except in special
circumstances, the material surrounding the pipe was dumped via a mechanized conveyor. The
backfill around the pipe reportedly was lightly to moderately compacted, using the tracks and/or
wheels of the construction vehicles. The materials used for the pipe bedding and embedment
were derived from excavated native materials, which were known to vary substantially along the
alignment. The moisture conditions in the ground varied from dry (in the coastal sector) to wet
(in the jungle sector) to saturated (at river crossings). Therefore, soil conditions in the trench are
expected to have varied significantly along the alignment.

The ability of material dumped into the trench to flow around and beneath the pipe is dependent
on the size of the gap on the sides and beneath the pipe, and the material type and wetness of the
backfill. Dry sand, for example, would likely flow freely around and beneath the pipe. Clods of
clay, on the other hand, will not flow, but rather will fall and stack on the sides of the pipe,
potentially creating a large, continuous void beneath the pipe and interspersed voids among the
clods. Exponent reviewed representative photographs taken during construction of the project in
the three sectors and at river crossings (e.g., Figures 17 to 20), which provide a glimpse of the
range of ground conditions that existed during construction of the pipe zones and possible
current conditions. The conditions in the selva sector, specifically in natural drainages, likely
represent some of the worst conditions for producing material in a suitable form for backfilling
the pipe, primarily because of the wet ground conditions and heavy precipitation that wets the
stockpiles.

SF36292.003 A0F0 0607 0806

47
 June 8, 2007

Potential geotechnical risks from trench conditions relate primarily to internal soil erosion
(“piping”). Without proper use of flow barriers to reduce the flow of water traveling through the
backfill (“ditch breakers”), the loss of backfill from erosion could result in external soil pressures
acting on the pipe. We understand that TgP is in the process of installing piezometers in the
ROW and inspecting the performance of the subterranean drainage systems as a means to assess
the potential for internal soil erosion.

5.5 Pipe Material–Related Risks

The system was constructed using more than 100,000 individual pipe segments, each
approximately 12 m in length. A review of the pipe book46 and the pipe manufacturing and
coating records indicates that the entire stock of pipe segments used for this system was
fabricated at one of two pipe mills during 2002 and 2003. The pipe manufacturers’ records
indicate that these two pipe manufacturers are located in Pindamonhangaba, SP, Brazil, and
Buenos Aires, Argentina. These electric-resistance welded tubular products were manufactured
per API 5L standard. Specifically, flat steel plates were rolled into a tubular cross-section and
longitudinally welded into a pipe segment. Each segment was then coated with an outer high-
density polyethylene (HDPE) layer to protect the pipe’s exterior from the environment.

API 5L provides provisions to minimize potential risks to pipe integrity that can arise from the
fabrication of the steel plate and the longitudinal electric-resistance weld by requiring minimum
material performance, as well as material sampling and quality assurance procedures. Adherence
to DIN 30670, as described in Specification 2794-P-SP-00005, is intended to ensure the integrity
of the outer HDPE coating. Documents and evidence reviewed showed that the pipe was
specifically built for this system and consistent with API 5L requirements. Available data
(i.e., the hydrostatic testing and the metallurgical analysis of the five spill incidents) indicate that
the pipe material was in compliance with the required codes. Additionally, hydrostatic testing
revealed that none of the 59,906 installed pipes of the NG pipeline and only three of the 49,193
installed pipes of the NGL pipelines had pipe material defects. One of these three defects was
related to the rolled plate material, and two were defective longitudinal welds. All three pipes

SF36292.003 A0F0 0607 0806

48
 June 8, 2007

were replaced and successfully passed a subsequent hydrostatic test. For details on these three
leaks, see leaks 3, 4, and 5 in Table 1.

Table 1. Leaks identified during hydrostatic testing of the pipeline

ID Pipeline Location Date Failure Type Failure Cause Description

Partial circumferential Incorrect assembly of weld joint during
1 NGL KP 9+906 August 8, 2003 rupture of girth weld construction
External mechanical deformation and
2 NGL KP 1+726 July 27, 2003 Longitudinal fracture gauges
Localized material defect of
Partial circumferential manufactured steel plate used for pipe
3 NGL KP 171 October 4, 2003 rupture manufacture
Short longitudinal fracture Localized defect along the factory
4 NGL KP 31+494 February 16, 2004 along the longitudinal weld manufactured longitudinal weld
Short longitudinal fracture Localized defect along the factory
5 NGL KP 48+830 March 23, 2004 along the longitudinal weld manufactured longitudinal weld
Partial circumferential
6 NGL KP 170 October 4, 2003 rupture of girth weld Failure of a repair girth weld
External damage of pipe due to
unauthorized use of construction
7 NGL KP 210 January 9, 2004 External damage equipment near pipeline
Partial circumferential
8 NG KP 388 September 16, 2003 rupture of girth weld Failure of a repair girth weld

Exponent currently considers the residual risk of future pipeline failures caused by pipe
manufacturing-related risks to be negligible, because hydrostatic testing provided an effective
means of identifying longitudinal welding defects. The three pipe material–related failures do
not indicate that the manufacturing process was inadequate; rather, they demonstrate that the
hydrostatic testing was successful in removing these defects before the pipeline was placed into
operation. This level of defect removal is not atypical, as can be seen from the more
comprehensive discussion of the hydrostatic testing given in Section 5.8.

The HDPE coating system is designed to mitigate external pipe corrosion. Pipeline integrity–
related risks associated with the manufacture of the coating are typically low; however, risks to
the integrity of the HDPE coating arise during the various stages of moving and handling the
coated pipes. The prevention of surface damage to the protective outer layer is an important step
in enhancing the protection against external corrosion. It needs to be recognized that, at each
circumferential pipe joint, a sleeve is placed, overlapping the adjacent pipe to provide continuous
surface protection. A failure of this protective coating system is significant to the pipeline’s

46
The pipe book lists relevant pipe data.

SF36292.003 A0F0 0607 0806

49
 June 8, 2007

long-term integrity and mitigated by operation of a cathodic protection system and detection of
material loss by inline inspection tools. In this regard, TgP operates a cathodic protection system
and, in 2006, performed an inline pipe inspection of the NGL pipeline to detect any external as
well as internal material loss. A more detailed description of the inline inspection’s results and a
discussion of our observations can be found in Section 7.3.

5.6 Pipeline Field Welding–Related Risks

The individual pipes are circumferentially joined in the field using the Shielded Metal Arc
Welding (SMAW) and Flux Arc Welding (FCAW) method in compliance with API 1104. API
1104 and ASME B31.4 (liquid) and ASME B31.8 (gas) standards establish the framework for
the many girth welds that join the individual pipes. These standards are used throughout the
world for the construction of pipelines.

The NGL pipeline was fabricated using the manual SMAW method, whereas for the NG
pipeline, automated methods were used. Qualified Welding Procedures (QWP) for girth welds
and their field repair are given in Construction Specifications 2794-L-SP-00012, 2794-L-SP-
00016, 2794-L-SP-00017, and 2794-L-SP-00031. Exponent reviewed these welding
specifications and found them to be acceptable and in conformance with API 1104. In addition,
a review of the welding records indicates that the issued welder certificates were consistent with
the requirements of API 1104 and ASME Sec. IX. Although four of the six incidents occurred at
girth welds, geotechnical loading conditions, and not weld quality, were the significant
contributing factors in three of these four incidents. The other incident was caused by a
hydrogen-induced crack in the girth weld that both escaped radiographic detection and passed
hydrostatic testing. The crack missed detection by both of these weld quality tests because the
crack appears to have developed after the weld was x-rayed, and because the pipe material
toughness exceeded the minimum required so that the crack was just below the failure threshold.
This hydrogen crack does not necessarily represent an unacceptable level of risk. It is not
unusual for a small number of preexisting line defects to cause small leaks that are detectible and

SF36292.003 A0F0 0607 0806

50
 June 8, 2007

easily repaired for a line that has been hydrostatically tested.47 It only becomes a significant risk
if such cracking is pervasive throughout the system. A more detailed discussion of the risks
associated with hydrogen-induced cracking of girth welds can be found in Section 5.8.3.

5.7 Pipeline X-ray–Related Risks

Techint specified that 24 hours after a weld was made, it was to be radiographed, and the
radiograph evaluated per API 1104 for any potential welding-induced defects. Any defects
determined not to be acceptable per API 1104 were to be removed and the weld repaired. The
inspectors who conducted and interpreted these radiographs were to be certified to Level II per
API 1104. A review of the certification records confirmed that the contracted inspection
personnel were certified to the required level.

Radiography is particularly effective in finding weld defects such as undercuts in the weld.
Because defects such as an undercut are more easily identified than a very tight hydrogen crack,
there is a very small possibility, as with all pipelines, that a tight hydrogen crack could escape
detection by the radiograph. Likewise, due to the time-delayed nature of a hydrogen crack, there
is also a finite risk that a hydrogen crack could develop just after the radiograph is taken.
Exponent reviewed a limited number of radiographs and associated reports that did not show any
API 1104 code deviations.

5.8 Hydrostatic Testing–Related Risks

5.8.1 Background

Hydrostatic testing of pipelines is an important tool to verify integrity and identify any leaks.
Hydrostatic testing of the system is mandated by the ASME codes. Hydrostatic testing involves
filling each pipe section with water and pressurizing the water to a predefined proof pressure that

47
G.H. Ewing, “The Development and Results of High Stress Hydrostatic Testing of Gas Transmission Lines in the
United States”, 1973, 12th World Gas Conference, page 10.

SF36292.003 A0F0 0607 0806

51
 June 8, 2007

exceeds the maximum operating pressure. Next, the water-tightness of the pipeline is verified by
maintaining a slightly smaller constant water pressure for at least 4 hours.48 This hydrostatic test
was performed for both pipelines along the entire length of the system. In the case of the NGL
pipeline, the specific code requirements are that the hydrostatic test’s proof pressure is 1.25 times
the internal design pressure, which is the expected maximum operating pressure (MOP).
A similar criterion applies to the NG pipeline and is defined per ASME B31.8. Per this criterion,
our review indicates that the hydrostatic tests were executed in conformance with the ASME
Codes. However, during hydrostatic testing, induced stresses were not uniform, such that the
effectiveness to detect defects varied along the length of the pipelines.

If a leak is detected during hydrostatic testing, the pipe is repaired and hydrostatic testing is
performed again to verify the pipe’s water tightness and strength. After the test, the individual
tested sections of the pipeline are welded back together using either pipe tie-in pieces or welding
the adjoining pipe sections together. These welds are then radiographed 24 hours after the
welding.

5.8.2 Hydrostatic Testing Results for the Camisea System

During the hydrostatic testing of the Camisea system, eight leaks were identified, of which seven
occurred in the NGL pipeline. Table 1 lists the specific information for each leak. Three leaks
were caused by faulty girth welds, two by faulty longitudinal welds, two by external damage that
had occurred during construction, and one by a foreign object being introduced during rolling of
the steel plate. All of these failures were subsequently repaired, and the pipeline section was
successfully re-tested. Eight failures during hydrostatic testing in more than 1,250 km of
pipeline is a low number, especially considering the challenging terrain. The test results are
more an indication of the test’s ability to detect preexisting faults. For example, the Texas
Eastern Transmission Corporation has in service over 5,000 miles of transmission pipelines that
have been hydrostatically tested. Hydrostatic testing removed 129 defects over a span of

48
TgP’s water tightness test was conservatively extended to be 24 hours long.

SF36292.003 A0F0 0607 0806

52
 June 8, 2007

3,424 miles of pipe.49 Of these 129 defects, 10 were plate defects, 16 were defects in the girth
welds, and the remainder were defects at the longitudinal welds. Hydrostatic testing plays an
important role in detecting flaws and cracks in pipelines in order to reduce operational risks and
the likelihood of subsequent pipeline incidents; however, it needs to be recognized that these
risks cannot be fully eliminated but need to be managed by the pipeline operator.

As in any pipeline, despite the precautions taken by hydrostatic testing, some minimal risk may
exist, because defects may be aligned or sized such that the hydrostatic test would not rupture the
pipe and the defect could go undetected. In the absence of significant external pipe loading, this
type of potential defect, if it results in a leak, would typically release only small amounts of gas
or liquid hydrocarbons. This situation arose with the second spill incident of the NGL pipeline,
where a hydrogen-induced crack was not detected by the radiograph, and the subsequent
hydrostatic test and initiated the subsequent spill incident.

5.8.3 Hydrogen-Induced Crack–Related Risks

Hydrogen-induced cracking (HIC) is a well-known phenomenon that affects welds in which

hydrogen can be temporarily introduced during welding of mild steels. This temporarily trapped
hydrogen embrittles the weld material and heat-affected zone (HAZ), but will not leave any trace
of its presence after it diffuses out of the weld. During the time the hydrogen is present, the weld
material and HAZ is susceptible to hydrogen-induced cracking. Typically, these cracks are
formed several hours to several days after welding. In the Camisea system and consistent with
general pipeline engineering practice throughout the world, this hazard was mitigated during
construction by radiographic inspection of all girth welds and hydrostatic testing. Despite these
actions, a potential residual risk is that partial penetration cracks can survive the hydrostatic test.
This situation arose with the second spill incident at KP 222+500, in which the tough pipe
material of the system did not rupture during hydrostatic testing. The multiple hydrostatic test
cycles and subsequent operational pressure cycles of the transported hydrocarbons further

49
G.H. Ewing, “The Development and Results of High Stress Hydrostatic Testing of Gas Transmission Lines in the
United States,” 1973, 12th World Gas Conference, page 10.

SF36292.003 A0F0 0607 0806

53
 June 8, 2007

propagated this initial hydrogen-induced crack, ultimately breaching the wall and causing a slow
leak in the NGL line at KP 222+500.

Exponent’s investigation currently indicates that the post-weld radiograph at KP 222+500

showed no signs of a hydrogen-induced crack, but the post-failure radiograph clearly showed a
hydrogen-induced crack. The hydrostatic test of this section was performed five months after the
fabrication of this joint and lasted for approximately a month. The reviewed test results showed
no indication of a leak for the final hydrostatic test. The five repetitions of this hydrostatic test
were necessitated by the pipe’s inability to hold pressure in the initial test. Exponent’s review
and analysis currently indicates that the most likely cause for the failure of the initial hydrostatic
tests was entrapped air. The incident at KP 222+500 is currently not considered to be indicative
of any systemic problems for the more than 100,000 girth welds. Overall, the radiography of all
girth welds and hydrostatic testing of the system provides a level of risk mitigation consistent
with general pipeline engineering practice. We currently believe that hydrogen-induced cracking
is not a systemic issue. The occurrence of this type of failure should be very low, and when it
does occur, risks are not as high as for a rupture situation, because absent soil movement, a
failure with a slow leak rate should be the more likely outcome.

5.9 Conclusions

Construction of the pipeline used methods that employed standard engineering procedures. On the
other hand, the constructed geotechnical stabilization measures proved to be insufficiently robust
that soil stability was initially problematic along the ROW. GEOTEC recognized this risk and
recommended additional geotechnical stabilization measures. However, our inspections in 2006
showed that a significant portion of these measures were not sufficiently effective and robust to
stabilize the ROW. The severity and challenges of the dynamic terrain led to the construction of
geotechnical remedial measures at more than 100 locations in 2006, which are discussed in more
detail in Section 6.

Overall, the review showed that (1) the pipe was specifically built for this system and consistent
with API 5L requirements, (2) the issued welder certificates and qualifications of the inspectors

SF36292.003 A0F0 0607 0806

54
 June 8, 2007

were in compliance with API 1104, and (3) hydrostatic testing of the system was in compliance
with the required ASME codes.

SF36292.003 A0F0 0607 0806

55
 June 8, 2007

6 Geotechnical and Geology-Related Risks

6.1 Geotechnical and Geologic Conditions

The Camisea Transportation System carries natural gas products through the jungle, over the
Andes Mountains, and along the Pacific coast. Each of these geographic regions possesses
distinctive terrain, amounts of precipitation, geology, and other conditions that pose a myriad of
geotechnical and geologic hazards50 and challenge the integrity and reliable operation of the
pipeline.

The jungle (selva) sector is characterized as having geotechnical and geologic conditions that
pose the highest risk to the integrity of the pipelines. The conditions include very steep and high
slopes, narrow ridges, relatively thick deposits of moderate to high-plasticity residual soils, high
groundwater levels, and heavy precipitation leading to underground seepage and numerous water
crossings. The rocks and soils in the selva sector are generally characterized as materials with
low durability, low resistance to weathering, low strength, and moderate to high susceptibility to
erosion, particularly when devoid of vegetation. The salient geotechnical and geologic hazards
for the selva sector include debris flows, wedge landslides, deep rotational landslides in colluvial
deposits, rock slides, rockfalls, and ridge instability, all of which are a direct consequence of the
steep topography and heavy precipitation.

The mountain (sierra) sector is characterized by natural conditions that are somewhat better than
those in the selva sector, but still pose significant risk to the pipelines. Specifically, the
conditions include very steep and high slopes, isolated areas of flatter slopes with high-plasticity
weathered soils, hard rock, freezing temperatures, isolated regions with infrequent occurrence of
high intensity rainfall, and high-gradient streams. Groundwater typically does not exist within
the pipe zone, but the water table can rise during periods of intense rainfall that occur

50
For purposes of this report, geotechnical and geologic hazards are external ground pressures resulting from
ground instability or movement. Geotechnical hazards are defined as movement in soil, typically when saturated,
whereas geologic hazards are defined as movement in rock.

SF36292.003 A0F0 0607 0806

56
 June 8, 2007

intermittently. The principal geotechnical and geologic hazards in the sierra sector include
debris flows, wedge landslides, rockslides, and rockfalls.

The coastal (costa) sector generally has the most favorable natural conditions with respect to the
static stability of soil and rock along the ROW. The costa sector is characterized by regions with
low topographic relief, very little rainfall, predominantly granular soils and gravel, and broad
river crossings. The primary non-seismic geotechnical hazards are mudflows and flash floods in
ravines during infrequent downpours, and lateral erosion of river terraces in the Pisco River
Valley.

These geotechnical and geologic conditions were known and evaluated by several geotechnical
and geologic consultants, including GEOTEC, the consortium hired to provide engineering
services during and immediately after construction of the system. In January 2005, five months
after the system was placed into operation, GEOTEC presented a plan of action, with
recommendations for remedial actions that included surface and subsurface drainage systems,
and stabilization systems such as buttress fills, slope armoring, reinforced concrete walls, crib
walls, gabion walls, and gravity walls, metal or concrete piles, rock bolts, and reinforced
shotcrete. These recommended measures were to be used in conjunction with regular
maintenance activities, particularly reconstruction of surface drainage works. This report also
recommended the following types of instrumentation: observation wells or piezometers, surface
surveying, and slope inclinometers, and also a hydro-meteorological (rainfall and river-stage)
alert network.

6.2 General Findings

As described previously in Chapter 4, the pipeline was designed such that external soil loading
would be mitigated by stabilization measures constructed at sites deemed to pose a geotechnical
or geologic hazard. For a challenging alignment such as this system, this design approach
requires the construction of effective stabilization measures.

SF36292.003 A0F0 0607 0806

57
 June 8, 2007

Also described in Chapter 4 are the preliminary geologic and geotechnical investigations that
were conducted by Golder and MRA for design of the system with the stated objective of
determining the geologic and geotechnical feasibility of the proposed alignment. These
preliminary studies evaluated the geologic (and geotechnical) risk along the proposed route,
identified critical zones in the selva sector, and made preliminary recommendations. These
preliminary studies recommended further geologic and geotechnical studies during construction
to identify additional unstable zones and recommend stabilization measures. As described in
Chapter 4, some site-specific evaluations of geotechnical and geologic hazards were performed
prior to or during construction by various consultants (Felix, Verastegui, and Stepanek). The
most comprehensive study of geotechnical risks was performed by GEOTEC after installation of
the pipeline was complete.

The geotechnical risk assessment requested by TgP from GEOTEC identified areas (slopes and
ravine/river crossings) that required additional geotechnical stabilization. In addition, GEOTEC
made several other recommendations, including implementation of subsurface exploration,
topographic surveys, and routine annual geotechnical maintenance. Based on this study, TgP
developed a master plan to address geotechnical and geologic recommendations.

Following our June 2006 inspections, Exponent concluded that some of the stabilization
measures implemented during, or immediately following, construction were not completely
effective in mitigating external soil loads acting on the pipes. In some instances, we observed
evidence of substantial ground movement, which indicated that the stabilization measures were
not adequately mitigating geotechnical or geologic hazards at the time of our inspections. For
example, we observed gabion walls that had displaced laterally with significant tilting, and
displaced or disrupted drainage facilities.

The spill incidents and the observed performance of the system as of early 2006 caused TgP to
set into motion an aggressive geotechnical stabilization program that began in earnest in April
2006. More than 100 sites along the ROW were evaluated and remediated by implementing
geotechnical stabilization measures using designs and construction techniques that were more
robust than those used previously. The primary components of the 2006 geotechnical
stabilization measures included: (1) gabion walls that were typically founded in weathered rock

SF36292.003 A0F0 0607 0806

58
 June 8, 2007

below the landslide plane; (2) in some limited cases, rows of small-diameter pipe or timber “pin”
piles that were placed parallel to the ROW; (3) reduction in thickness of soil overburden above
the unstable ground or landslide; (4) subsurface drains (“filters”) consisting of perforated PVC
pipe wrapped in geo-fabric and connected to solid pipe discharging into a lined drainage
channel; (5) subsurface “ditch breakers” to collect water flowing through trench backfill in steep
terrain and route it to underground drain pipes; (6) surface “current breakers” to collect runoff
water and convey it in a controlled, lined ditch to a main surface collector system; and (7) lined
drainage channels to collect water from current breakers and ditch breaker drains.

Based on our second set of site inspections, in September 2006, and subsequent review of
documentation of individual site stabilization design approaches, Exponent concluded that the
remedial geotechnical stabilization measures at most sites are sufficiently robust and appear to
represent a reasonable and practical approach to reduce the risk of failure to the pipeline.
Further, we determined that these measures were being applied in a consistent and effective
manner. As a result of these efforts, we believe that the risk of future failure of the system
resulting from external geotechnical forces has been substantially reduced at locations where
stabilization measures were implemented in 2006, with the exception of one site at PS #2, where
evidence of soil movement was observed above the gabion wall installed at the toe of the
landslide. Based on our observations at PS #2, Exponent recommended that a detailed
geotechnical investigation, including soil sampling in borings and test pits, be performed at PS
#2 and that appropriate geotechnical stabilization measures be implemented as soon as possible.
We understand that TgP has followed our recommendations, and a significant portion of the
stabilization measures have been completed at PS #2.

Exponent also noted that effective and continuous monitoring of the performance of surface and
subsurface drainage systems is necessary, because high groundwater conditions have been a
major factor in the initiation of slope movements. Hence, in September 2006, Exponent
recommended the deployment of instrumentation and monitoring equipment (e.g., rain gauges,
piezometers, strain gauges, slope inclinometers, and survey control) at sites with the greatest
residual risk to assess surface and subsurface conditions. This instrumentation will provide
earlier warnings of ground instability and additional data on ground movement locations, depths,

SF36292.003 A0F0 0607 0806

59
 June 8, 2007

directions, and rates of movement will be available to permit more expeditious and reliable
repairs.

In late 2006, Exponent also reviewed a proposed inspection program (marcha vigilante) initiated
by TgP and COGA to help reduce the risk of future failure resulting from external geotechnical
forces by detecting and quantifying early signs of slope instability. The core of this program
involves regular ROW visual inspections within the selva and sierra sectors by multi-disciplinary
teams trained by COGA’s technical consultants. The marcha vigilante inspection teams
document relevant geotechnical conditions at each site on a weekly basis, and this information is
used to update the geotechnical risk assessment method described below.

6.3 Geotechnical Risk Assessment

6.3.1 Risk Assessment Methodology

Exponent and TgP collaboratively developed a qualitative method to assess the likelihood and
severity of future failure resulting from geotechnical and geologic conditions at a given site
along the ROW. This method was developed to assess geotechnical and geologic risk in a
project setting that is diverse, dynamic, complex, and sensitive as a result of both natural and
man-made (e.g., development, deforestation, and remediation) causes. In this method, risk is
assessed using traditional geotechnical failure modes, with a separate evaluation of the
likelihood of failure versus the severity of failure. In our opinion, this approach enhances the
characterization of overall risk, providing a stronger basis to make fundamental decisions on
acceptable and unacceptable levels of risk, and develops the most effective means to reduce risk
and to prioritize remedial and monitoring efforts.

The geotechnical risk assessment method formulated by Exponent and TgP consists of the
following four major steps: (1) evaluation of geo-integrity parameters, (2) evaluation of the
likelihood of failure using a screening analysis and the geo-integrity parameters,
(3) characterization of the severity of failure should one occur, and (4) qualitative evaluation of
the overall risk of failure.

SF36292.003 A0F0 0607 0806

60
 June 8, 2007

Geo-integrity parameters are typically evaluated during the inspection program (e.g., marcha
vigilante) to characterize the likelihood of a potential pipeline failure using readily available or
measurable information from a given site. These parameters were chosen based on our
experience and knowledge of the conditions along the ROW, including geologic and
geotechnical conditions that likely led to three of the first five pipeline ruptures.

The initial step in the likelihood evaluation involves computing a Safety Ratio (SR) for the site
using geo-integrity parameters for each failure mode. The Safety Ratio relationships developed
for this method are intended to be an interim surrogate measure of safety factors51 in the absence
of more rigorous engineering analysis. The following geotechnical failure modes are currently
considered in this risk method and capture most of the observed landslides for the system:
(a) deep rotational landslide, (b) wedge landslide, (c) translational landslide, and (d) ridge
instability.

The computed Safety Ratios are then used to assign Likelihood Levels (LL) for each failure
mode. Likelihood Levels consist of four numerical categories that correspond to our
interpretation of the probability of occurrence of the failure mode at that site. The higher the
Likelihood Level, the greater the susceptibility. Severity Levels (SL) are characterized using four
rating levels, with higher numerical levels also corresponding to greater, more severe,
consequences. The Severity Level considers the following four different consequence categories:
environmental, property, health, and safety. The final step in the risk assessment method
involves assigning the final Risk Category for each site by combining the computed Likelihood
and Severity Levels as shown in Table 2. The Risk Categories are:

Risk Category 1 – Low risk

Risk Category 2 – Moderately low risk that is acceptable

Risk Category 3 – Medium risk that should be evaluated for mitigation measures

51
The ratio between the forces resisting ground movement and the forces driving ground movement. A safety
factor of 1 indicates that the forces are equal, and a slope is at the point of marginal stability or imminent failure.

SF36292.003 A0F0 0607 0806

61
 June 8, 2007

Risk Category 4 – High risk that should be evaluated for mitigation measures with
high priority

Risk Category 5 – Very high risk that requires immediate evaluation of the need for
mitigation measures.

The final Risk Category is evaluated using the most critical combination of the Likelihood Level,
considering all four failure modes, and Severity Level as shown in Table 2. Exponent has
recommended to TgP that all sites in Risk Categories 3, 4, and 5 be subjected to a formal
engineering stability analysis to determine whether mitigation measures are necessary.

Table 2. Risk category assessment chart

Likelihood
1 2 3 4

1 1 2 2
1
1 2 2 3
2
Severity

2 2 3 4
3

4 2 3 4 5

6.3.2 Application of the Risk Assessment Method

The geotechnical risk assessment method was applied to 95 different sites52 along the system to
create a risk matrix. The sites were selected by COGA based on their interpretation of the
geotechnical and geologic hazards present along the ROW. Of the 95 sites in the risk matrix, 67
are in the selva sector, 28 are in the sierra sector, and none are in the costa sector.

52
Although more than 100 sites were mitigated as part of the 2006 geotechnical stabilization program conducted by
TgP and COGA, the geo-integrity parameters necessary to apply the geotechnical risk assessment method were
characterized for only 95 sites as of October 2006.

SF36292.003 A0F0 0607 0806

62
 June 8, 2007

In general, sites are entered into the risk matrix if manifestations of earth instability (e.g., tension
cracks, surface depression, slope bulging, soil movement, pipe movement, excessive fiber optic
attenuation) are observed. The risk matrix is based on the geotechnical expert evaluation of the
first 450 km of the ROW. However, the risk matrix does not necessarily include all potential
risk sites along the entire pipeline alignment. For example, even though narrow ridge instability
is a substantial hazard on some sections of the ROW in the selva sector, and ridge instability is
included in the risk matrix, there currently are no sites in the matrix involving this potential
hazard, because there were no known manifestations of ground movement in these same sections
as of 2006. In this regard, it is our understanding that TgP will transition in 2007 toward a
proactive implementation to address this potential geotechnical risk.

Exponent inspected more than 50 sites, some on both occasions, during our field inspections in
June and September 2006. Thirty-five of these sites were in the COGA risk matrix. The geo-
integrity parameters for these 35 sites were evaluated initially by COGA, and subsequently by
Exponent for the sites that we visited. Exponent did not independently evaluate the geo-integrity
parameters for the remaining 61 sites but did collaborate with COGA in their evaluation.

The overall risk at each site was first evaluated assuming conditions that existed prior to the
construction of the new geotechnical stabilization measures (May 2006), in order to provide a
baseline assessment of risk. The risk at each site was then re-evaluated following the completion
of the stabilization measures (September 2006). Finally, the risk was re-evaluated to incorporate
the implementation of the marcha vigilante inspection program (October 2006). The results of
the geotechnical risk assessment for May, September, and October 2006 are presented in
Figures 21 to 23, respectively. For each time period, the maximum combination of likelihood-
severity pairings of each individual site is presented. The 35 sites in the risk matrix that were
inspected by Exponent are shown in blue. In addition, the results were compiled in pie charts to
summarize the overall risk level of the 95 sites in the COGA risk matrix (Figures 21 to 23).

The results of the geotechnical risk assessment and our field observations indicate that TgP made
substantial progress in 2006 to diminish the overall risk of future failure resulting from external
soil or rock pressures. The risk matrix initially ranked 48% of the sites (45 sites, along the initial
455 km of the ROW) as having “high” to “very high” risk (Risk Categories 4 and 5) in May

SF36292.003 A0F0 0607 0806

63
 June 8, 2007

2006. By comparison, the percentage of sites ranked “high” to “very high” risk was 13% (12
sites) in September 2006. Further reductions to the risk level were observed in the October 2006
risk matrix results wherein only 5% of the sites (5 sites) were characterized as having a “high” to
“very high” risk, and 74% of sites were ranked as having “moderately low” risk (Risk
Category 2) by October 2006. By October 2006, only one site adjacent to the second pump
station was ranked as “very high.” As mentioned before, TgP has recently been implementing
stabilization works at this location to reduce the geotechnical risk.

94 Likelihood
1 2 3 4

1 0 0 0 0
Severity

2 0 4 1 7

3 3 17 13 24

4 2 2 9 12

DISTRIBUTION for MAY 2006

1 2 3 4 5

35%

23%

0%
29% 13%

Figure 21. Risk assessment results for May 2006.

SF36292.003 A0F0 0607 0806

64
 June 8, 2007

95 Likelihood
1 2 3 4

1 0 0 0 0
Severity 0 5 6 2
2

3 9 22 20 6

4 3 16 4 2

DISTRIBUTION for SEPTEMBER 2006

1 2 3 4 5
40%

47%
0% 11%

2%

Figure 22. Risk assessment results for September 2006.

SF36292.003 A0F0 0607 0806

65
 June 8, 2007

95 Likelihood
1 2 3 4

1 0 0 0 0

Severity 2 0 8 4 0

3 16 33 6 3

4 9 14 1 1

DISTRIBUTION for OCTOBER 2006

1 2 3 4 5

74%

21%

0%
4%
1%

Figure 23. Risk assessment results for October 2006, which includes the monitoring
program for the September 2006 results.

SF36292.003 A0F0 0607 0806

66
 June 8, 2007

The substantial reduction in overall risk from May to October 2006 is directly attributable to
changes in Likelihood Levels, because Severity Levels were likely unchanged. The change in
Likelihood Levels from May to September coincides with COGA’s progress in completing robust
geotechnical stabilization measures, which began in earnest in April 2006. The construction of
geotechnical stabilization measures is reflected in the risk matrix through re-evaluation of geo-
integrity parameters used to characterize various site conditions (i.e., slope inclination, surface
drainage control, groundwater depth, and presence of gabion walls). The further reduction of
Likelihood Levels in the October risk matrix corresponds to TgP’s commencement of its
comprehensive inspection program.

The rankings in the May and October risk matrices were compared to evaluate the total progress
made as of October 2006 in mitigating geotechnical and geologic hazards along the ROW
through geotechnical stabilization, inspections, and monitoring. Figure 24 summarizes the
computed reduction in Risk Category versus the baseline risk (i.e., the Risk Category as of May
2006). Figure 24 shows that the most significant reductions in risk were achieved at sites
possessing “high” to “very high” risk (Risk Category ≥ 4), whereas marginal effects were
observed at “moderately low” to “medium” risk sites.

3 1
Reduction in Risk

2 21 9

1 21 9 1

0 27 1 3 1

2 3 4 5
May 2006 Risk Category

Figure 24. Change in risk from May to October 2006.

SF36292.003 A0F0 0607 0806

67
 June 8, 2007

The results of the risk assessment for the 95 sites were separated by sector, to geographically
delineate variations in the potential for geotechnical risk along the system. However,
representation across the three sectors was not possible, because 67 of the 95 sites are located in
the selva sector—the location of the vast majority of sites that have exhibited manifestations of
ground instability—and the remaining sites are in the sierra sector. Figures 25 and 26 show the
risk by sectors for the May and October risk matrix rankings, respectively. The May 2006
rankings show that the majority of the selva sector sites evaluated fall into Risk Categories 4 and
5, whereas the majority of the sierra sector sites fall into Risk Category 2. In contrast, the
October 2006 results show a substantial reduction in the risk for the selva sector, wherein most
of the selva sites fall into Risk Categories 2 and 3. Hence, comparison of Figures 25 and 26
suggests that the majority of the geotechnical stabilization works and monitoring programs are
appropriately focused on the selva sector.

While TgP had been using a risk assessment method for some time, the current geotechnical risk
assessment method was developed in 2006. Results of this method appear to work reasonably
well in providing a qualitative assessment of the geotechnical and geologic hazards present along
the ROW. However, engineering judgment should be exercised at all times when applying the
risk assessment method, recognizing the qualitative nature of the approach.

50
Selva (May 2006)
45
Sierra (May 2006)
40
35
Frequency

30
25
20
15
10
5
0
1 2 3 4 5
Risk Category

Figure 25. Risk by sectors for May 2006.

SF36292.003 A0F0 0607 0806

68
 June 8, 2007

50
Selva (October 2006)
45
Sierra (October 2006)
40
35
Frequency

30
25
20
15
10
5
0
1 2 3 4 5
Risk Category

Figure 26. Risk by sectors for October 2006.

6.4 Ongoing Geotechnical Risk Mitigation

We expect that the continuous, ongoing implementation of the risk matrix process will identify
additional sites that are not included in the current risk matrix as new manifestations of ground
movement are observed or detected using the installed instrumentation (e.g., strain gauges and
inclinometers). Some of these potential new sites may even be ranked with a high to very high
risk, requiring immediate measures to sufficiently mitigate the hazard. Therefore, we
recommend that TgP adopt a proactive approach of continually assessing geotechnical and
geologic hazards along the ROW where manifestations of instability are not present. Initially,
this approach will likely require an assessment of geologic and topographic maps to identify the
most susceptible sites. In addition, Exponent recommends that TgP develop a Risk Management
Plan (RMP) that governs the use of all geotechnical risk assessment methods and guide TgP’s
actions, decision processes and means of execution. This RMP should state at least the
following:

SF36292.003 A0F0 0607 0806

69
 June 8, 2007

• All sites with a risk ranking 4 and 5 should be mitigated immediately.

• All sites with a likelihood ranking of 4 should be mitigated immediately.

• Clear organizational structures and the level of responsibility and authority

need to be assigned.

• Mandated actions ranging from authorizing investigations to shutting down

the pipeline need to be provided.

6.5 Conclusions

The geotechnical and geologic conditions along the pipe alignment are diverse, dynamic,
complex, and sensitive. During our inspections in 2006, we observed evidence that some
geotechnical stabilization measures implemented during or immediately following construction
proved insufficient to mitigate external soil pressures acting on the pipes. Based on Exponent’s
review, observations, and engineering experience, we conclude that geotechnical and geologic
conditions posed the most significant risk to the integrity and reliable operation of the system.

The observed performance of the pipeline system, wherein three of the spill incidents were
attributed to geotechnical and geologic instabilities, caused TgP to implement aggressive
geotechnical stabilization measures. This remedial program began in earnest in April 2006 and
employed robust construction techniques that were applied in a consistent and effective manner.
Extraordinary efforts were made to complete the geotechnical stabilization measures along the
ROW before the start of the 2006–2007 rainy season.

Exponent visited more than 50 sites during our September 2006 inspection, when stabilization
measures were completed or significantly underway, and concluded that the geotechnical risk
has been reduced substantially at the remediated sites, with the exception of a site adjacent to the
second pump station. Exponent recommended that a detailed geotechnical investigation,
including soil sampling in borings and test pits, be performed at PS #2 as soon as possible, and
TgP has communicated that such efforts have been substantially completed. Furthermore,
Exponent recommended the deployment of instrumentation and monitoring equipment at critical

SF36292.003 A0F0 0607 0806

70
 June 8, 2007

sites to provide earlier warnings of ground instability, and to provide additional data on ground
movement locations, depths, directions, and rates to permit more expeditious and reliable repairs.

Exponent also reviewed the inspection program initiated by TgP to help reduce the risk of future
failure from external geotechnical forces by detecting and quantifying early signs of slope
instability. This program, which involves weekly visual inspections of critical sections of the
system during the rainy season, appears to be comprehensive and aggressive.

During the second phase of our project, Exponent worked collaboratively with TgP to develop a
geotechnical risk assessment method to evaluate the occurrence of geotechnical hazards along
the ROW that could ultimately affect the stability of the system. This risk assessment method
was validated using information from our field inspections and applied to 95 sites along the
system to create a risk matrix.

The risk at each site was evaluated at three different points in time, reflecting the risk prior to
construction of the new geotechnical remedial measures (May 2006), after the construction of
those measures (September 2006), and after implementation of the additional monitoring
programs (October 2006). In general, the results of the geotechnical risk assessment are consistent
with our field observations that TgP has made substantial progress in diminishing the overall risk.
The reduction in risk in 2006 was achieved primarily through the construction of geotechnical
stabilization measures and implementation of monitoring programs to decrease the likelihood of
future failures. Furthermore, the results suggest that the geotechnical stabilization measures and
monitoring programs have appropriately targeted sites that formerly had high to very high risk,
which are located primarily in the selva sector.

We expect that the continuous, ongoing implementation of the risk matrix process will identify
additional sites that are not included in the current risk matrix as new manifestations of ground
movement are observed or detected using the installed instrumentation (e.g., strain gauges and
inclinometers). Some of these sites may even be ranked with a high to very high risk and will
need to be mitigated quickly. Therefore, we recommended that TgP adopt a proactive approach of
continually assessing geotechnical hazards along the ROW, such that sites possessing higher risk
profiles that have not exhibited instability manifestations will be included in the risk matrix. In

SF36292.003 A0F0 0607 0806

71
 June 8, 2007

this regard, TgP has committed to implementing a RMP, recommended by Exponent, that will
govern the use of all risk assessment methods and guide TgP’s actions, decision process, and
manner of execution. At this time, Exponent believes that a successful implementation of the
above, and the construction of additional geotechnical mitigation measures in 2007, will further
reduce the geotechnical-related risks to the pipeline system.

SF36292.003 A0F0 0607 0806

72
 June 8, 2007

7 Pipeline Integrity–Related Risks

7.1 Description of Potential Risks and Controls

In any operating pipeline, pipe integrity–related risks are associated with pipe loading conditions,
pipe material, weld quality, and the quality of protection the pipeline is afforded against
environmental conditions. Pipe material and pipe manufacture–related risks have been discussed
in prior sections and been deemed to be at a typical and acceptable level for pipelines. Risk
mitigation during the construction of the pipeline relied foremost on hydrostatic testing and
radiography of girth welds, which is in compliance with the requirements of the applicable
codes.

During operation, inline inspections and external pipe inspections are the most effective means
of controlling long-term hazards to pipeline integrity, most commonly manifested as internal and
external corrosion. As with the Camisea system, cathodic protection and the exterior HDPE
coating are common methods of protecting the pipeline from external environmental attack.
Inline inspection tools, particularly the Magnetic Flux Leakage (MFL) tool, in conjunction with
pipe excavation at areas identified as problematic, are common means of identifying internal
metal loss and other potential anomalies. These inspections are typically conducted in
compliance with API 1160 and federal regulations by the U.S. Department of Transportation
(DOT) 49 CFR 195.452. In this regard, TgP has implemented an inline inspection program in
compliance with these requirements. This section summarizes the results of these pipe
inspections performed by TgP in 2006 on the NGL pipeline and discusses the key findings and
future needs to mitigate potential residual risks.

In the dynamic landscape of the Camisea selva sector, soil movement has been identified to be
the most significant hazard to the system, because soil movement imposes a lateral loading upon
the pipe. Lateral loading from external soil pressures can induce axial stresses in the pipe, in
addition to the stresses induced by the internal pressure of the transported hydrocarbons. These
axial stresses affect the girth welds in the pipe, by growing circumferential defects if they exist,

SF36292.003 A0F0 0607 0806

73
 June 8, 2007

and this could also significantly affect the integrity of a pipeline. Thus, an increased residual risk
due to subcritical crack propagation may exist at sites where geotechnical mitigation measures
have not been constructed. Identification of these sites and construction of effective geotechnical
measures mitigates this risk most effectively. Elimination of potential defects in the pipe or weld
reduces the potential risk even further.

7.2 Pipe Material and Damage Tolerance–Related Risks

The tubular pipe material used for the system was required to conform to API 5L. This standard
requires minimum yield strength of 70,000 psi and a minimum ultimate strength of 82,000 psi
for an X70 grade pipe. In addition to the material testing required by API 5L during pipe
manufacturing, TgP’s outside consultant, MCI, tested samples of pipe from the first five spill
incidents. These tests independently substantiated the adequacy of pipe strength.

Exponent’s analysis of the system’s ability to withstand lateral soil movement has shown that the
as-designed NG pipeline has a significantly larger external load capacity and flaw tolerance than
the NGL pipeline, such that the NG pipeline generally has a low risk of failure from external
loads. This basic residual strength advantage of the NG pipeline versus the NGL pipeline is
approximately a factor of two in tension and a factor of six in bending. Load capacity estimates
for the NGL pipeline show that, while sufficient for internal pressures, the pipeline is less able to
resist external soil loads, because the NGL pipeline only requires a smaller diameter and thus
thinner walls than the NG pipeline per the hydraulic design considerations of the ASME codes.
Therefore, these external soil loads should be mitigated by geotechnical stabilization measures.
The results of Exponent’s analysis are consistent with the operational experience, where soil
movement was identified to be a significant contributor in three of the five spill incidents.53 In
particular, for the first and fifth spill incidents, axial loading due to internal pressure alone cannot
account for the observed fast54 crack growth rates and final overload failure. Significant
progressive lateral soil loading is the likely force that propagated these cracks and induced the

53
The sixth spill incident is currently being investigated and is not included in this comparison.
54
We refer to fast crack growth in the context of geologic time scales and not dynamic fast fracture due to impact
loading.

SF36292.003 A0F0 0607 0806

74
 June 8, 2007

overload failure of the NGL pipeline. This failure evolution occurred with very few load cycles
until the crack reached critical length. Furthermore, this soil loading is mostly load controlled
and therefore is more detrimental to the NGL pipeline than the NG pipeline, since the NG
pipeline has a larger diameter such that smaller stresses are induced by the same soil load.

Under normal operating pressures, any potential defects that may have survived the hydrostatic
test are unlikely to cause rupture. However, geologic hazards pose a risk to the NGL pipeline,
especially if circumferential cracks are deeper than 40% of the wall thickness,55 because soil
movement can induce the growth of such cracks, potentially leading to pipe failure. However,
repeated hydrostatic testing will not reduce this risk, because the pipe material is sufficiently
tough to pass the hydrostatic test.

7.3 Inline Pipe Inspection

7.3.1 Background

TgP contracted Tuboscope Pipeline Services (TPS) in 2006 to perform an inline inspection of the
entire NGL pipeline, using the MFL inspection tool and a geometric inline inspection tool. The
MFL inspection tool has the capability to detect metal loss and other potential pipe wall
anomalies. The geometric tool continuously measures the pipe’s geometry along its length.
These tools are inserted at a launch point at each pump station and record data from their sensors
for later analysis.

The goal of the pipe inspections is to identify pipe wall anomalies, metal loss, pipe gauge, and
dents. DOT regulates inspections per 49 CFR 195.452, which lays out the classification and
reporting of these conditions. API 1160 is a standard that is used to implement these rules,
providing guidance to the pipeline industry in maintaining the integrity of pipelines that transport
hazardous liquids. ANSI B31G is a manual that is used and referenced in DOT 49 CFR 195.452,
to determine the remaining strength of corroded pipes. Our review indicates that this set of

55
The hydrogen-induced crack that imitated the second spill incident was significantly deeper but relatively short.
Our analysis indicates that this crack was at the limit of detection for the hydrostatic test performed at this
location.

SF36292.003 A0F0 0607 0806

75
 June 8, 2007

regulations and standards are being used by TgP to conduct their pipeline integrity review of the
recently performed inline pipe inspection. In this regard, the geometric inspection tool is
typically used to identify dents, and the more sophisticated MFL inspection tool is used to
identify other reportable wall thinning conditions encountered in the pipeline industry. Our
discussion of the results of the inline inspection is divided into two parts. First, we discuss the
results of the inline inspection with regard to the requirements of 49 CFR 195.452. Then we
discuss the MFL inspection tool’s ability to detect potential circumferential cracks and its impact
on the NGL pipeline’s risk.

7.3.2 Inline Pipe Inspection Results

TgP used the combined data from the MFL inspection and geometric inspection, as well as
external excavations, to determine which conditions were reportable per DOT 49 CFR 195.452.
The inline inspection results of the entire NGL pipeline (48,188 pipes) only identified 30 locations
with reportable conditions per the requirement. Among the 30 locations, 14 were related to metal
loss, 13 to dents, 2 to girth welds and 1 to dent and metal loss.

The approach for each of these sites is to do a field investigation at each site, which includes
excavation, a series of detailed non-destructive testing (i.e., visual inspection, measure of pipe
diameter and axial pipe misalignment, ultrasound wall thickness measurement, radiographic
examination of girth weld, external ultrasound inspection of girth weld, external magnetic particle
inspection of girth weld, micro hardness measurement), and based on the results, implementation
of any necessary remedial actions (e.g., patching, re-weld, etc.). By the end of March, 12 sites
have been investigated and necessary repair actions completed. In all cases, the existing pipe was
in sufficient condition to reinforce, and it was not necessary to replace the pipe or repair a girth
weld. Good correlation was found between the inline inspection results and the field verifications.
The investigation of the remaining 18 anomalies is ongoing.

Beyond what is required by applicable norms (i.e., in terms of required follow-up

action/investigation), TGP is also investigating additional locations where the internal inspection
tool reported some indications (e.g., potential anomalies). As of March 2007, numerous sites have

SF36292.003 A0F0 0607 0806

76
 June 8, 2007

been verified in the field without having encountered any significant defect. A verification of
other sites is ongoing.

Overall, results of the MFL inspection tool and evaluation of metal loss per ASME B31G has
shown that, currently, no severe external and internal corrosion damage exists along the NGL
pipeline (i.e., the burst pressure of the pipeline is at all times larger than the maximum operating
pressure). Specifically, the data showed that the vast majority of metal loss occurrences had
depths of less than 15% with respect to the pipe’s wall thickness. The deepest metal loss was
reported to have occurred in the selva sector, with a depth of 49%, which is still less than the
mitigation threshold of 80% as required per DOT 49 CFR 195.452. All other metal loss reported
in the selva sector is less than 25% deep, and nearly 80% of the detected damage is less than
10% deep. Based on these findings, it also appears that there is a relatively elevated frequency
of metal loss occurrences within the first 50 km of the NGL pipeline. Therefore, this portion of
the NGL pipeline may be subject to this potential risk over the long term, and we recommend
that this should be evaluated as part of TgP’s ongoing pipeline integrity program.

TgP employed the geometric inspection tool to identify dents with a depth of more than 2%, per
DOT 49 CFR 195.452. Specifically, TgP requested that TPS use a more stringent criterion of
identifying any diametrical change larger than 2%. This inspection identified a total of 90
locations along the first 452 km of inspected NGL pipeline with deviations. However, more than
95% of these were deviations in diameter of less than 6%.

7.3.3 Circumferential Crack Detection

In an effort to quantify the MFL tool’s ability to detect circumferential cracks, TgP and its
contractors conducted a research program to quantify the crack detection limit of the MFL tool.
In addition, Exponent performed a pipeline integrity study to determine the MFL inspection
tool’s utility in detecting circumferential cracks.56 Our analysis indicates that a potential
circumferential crack would need to be subjected to a significant external load to be detectable

56
The service provider of the currently used MFL inspection tool has determined that only circumferential cracks
with a crack mouth opening of more than 0.1 mm can be detected with a probability of better than 90%. For
cracks with a smaller opening, the probability of detection decreases rapidly.

SF36292.003 A0F0 0607 0806

77
 June 8, 2007

with a high degree of certainty using the currently employed MFL inspection tool. In this regard,
the most effective means at this time to contain this potential risk is to identify sites with
geologic and geotechnical instability and construct the geotechnical mitigation measures needed
to eliminate any potential external soil loadings that could cause these circumferential cracks to
grow. TgP has also committed resources in 2007 to further assess the ability to detect potential
circumferential cracks. Unfortunately, at present, no pipeline inspection company is able to
provide a commercially viable inspection tool that can detect potential circumferential cracks,
even though the technology appears to be readily available. Despite the fact that the use of in-
line inspection tools to detect small circumferential cracks is currently not a common practice
among pipeline operators due to the relatively low risk to pipeline integrity posed by
circumferential cracks under normal operating loads, TgP has committed to evaluating potential
options.

7.4 Conclusions

The NG pipeline is at least twice as strong as the NGL pipeline under a variety of loading
conditions. Under bending loads that are typically induced by external soil loads, the NG
pipeline’s strength advantage is even larger. Our fracture mechanics study showed that the flaw
tolerance of the NG pipeline is significantly better than that of the NGL pipeline, because the NGL
pipeline only requires a smaller diameter and thus thinner wall than the NG pipeline, per the
hydraulic design considerations of the ASME code. This is especially true if the pipe is loaded in
bending, the loading condition that commonly results from external soil loads. Specifically, in
bending, the NG pipeline with a nearly 60% deep circumferential crack will typically have a
residual strength that will surpass that of a perfectly good NGL pipeline, whereas, in pure axial
loading, only a 40% deep circumferential crack will typically have a residual strength greater than
the NGL pipeline with no flaws.

Exponent expects that the most commonly encountered geologic and geotechnical hazards will
induce bending moments that may be accompanied by tension loading. Clearly, the third failure
where tension loads were larger was an exception to this loading assumption, but it was caused
by a specific site condition. Therefore, based on this study, the NG pipeline is deemed to be of

SF36292.003 A0F0 0607 0806

78
 June 8, 2007

significantly lower risk than the NGL pipeline. Geological and geotechnical hazards are the
most significant hazards for the NGL pipeline, and are even more of a concern when
circumferential cracks deeper than 40% preexist. Specific data from the fifth failure suggest that
rapid crack growth due to soil movement is possible, and only a few cycles were required to
propagate the crack to a critical length.

TgP used the combined data from the MFL inspection and geometric inspection, as well as
external excavations, to determine which conditions were reportable per DOT 49 CFR 195.452.
This review only identified 30 reportable locations in the entire NGL pipeline per this
requirement. These 30 locations are being investigated and repaired as needed.

Overall, results of the MFL inspection tool and evaluation of metal loss per ASME B31G has
shown that, currently, no severe external and internal corrosion damage exists along the NGL
pipeline. However, it appears that there is a relatively elevated frequency of metal loss
occurrences within the first 50 km of the NGL pipeline. Therefore, this portion of the NGL
pipeline may be subject to this potential risk over the long term, and we recommend that this be
evaluated as part of TgP’s ongoing pipeline integrity program.

The in-line inspections and external excavations performed at many sites have shown that the
MFL inspection tool is an excellent tool to detect internal and external metal loss in this system.
However, its ability to identify circumferential cracks is dependent on external loading
conditions, which reduces the probability of detecting potential circumferential cracks within an
actionable time frame. In addition, hydrostatic testing will also not likely identify potential
circumferential cracks, because identifiable critical crack sizes need to be at least 60% deep and
very long, given that the pipe material has such good toughness properties. We understand that
TgP has committed specific resources in 2007 to further assess the ability to detect potential
circumferential cracks. A root-cause analysis of the first and fifth spill incidents and the origin
and nucleation of the potential circumferential cracks, to be performed by TgP in 2007, will
assist in assessing the implications of this concern related to pipeline integrity.

Overall, in 2006, TgP significantly reduced pipe integrity–related risks and is currently engaged
in additional efforts to further reduce the risk profile.

SF36292.003 A0F0 0607 0806

79
 June 8, 2007

8 Seismic-Related Risks

8.1 Tectonic Overview

Peru represents one of the most seismically active regions in the world. This distinction is the
consequence of its location along a segment of the circum-Pacific seismic belt where the
continental block of South America is drifting westward, overriding and forcing down the denser
Pacific Ocean (Nazca) plate along a major offshore fault zone known as a subduction zone. This
deformation of the earth’s crust causes elastic strain energy57 to accumulate until a breaking
point is reached—an earthquake. At least nine large-magnitude earthquakes58 have been
recorded in or near Peru during the last 60 years.

The intensity of ground shaking and significant effects of a particular earthquake depend largely
on the magnitude, faulting mechanism, distance to the origin of rupture, and local site and
subsoil conditions. The Nazca and South American plates are “slipping” at a rate of 78 to
84 mm per year.59 By comparison, the famous San Andreas fault between the Pacific and North
American plates is slipping at a rate of approximately 50 mm per year. Because of the position
and orientation of the subduction zone, earthquakes in Peru generally occur at increasingly
greater depths toward the east, as illustrated in Figure 27. Earthquakes affecting Peru have three
distinct mechanisms:60

• Shallow, offshore inter-slab thrust events wherein failure occurs at the

boundary of the subducting Nazca and overriding South American plates.

57
Elastic strain energy is potential energy stored in a volume of the earth’s crust that has been deformed but not yet
ruptured.
58
A large-magnitude earthquake is defined as having a magnitude of 7 or greater.
59
DeMets, C., Gordon, R.G., Stein, S. and Argus, D.F. (1990). “Current plate motions,” Geophysical Journal
International, 101, pp. 425-478.
60
Yeats, R.S., Sieh, K. and Allen, C.R. (1997). “The Geology of Earthquakes,” Oxford University Press, New
York.

SF36292.003 A0F0 0607 0806

80
 June 8, 2007

• Shallow continental in-slab events where some of the relative motion of the
Nazca and South American plates is accompanied by deformation within the
overriding South American plate.

• Finally, deep onshore in-slab events caused by internal deformation of the

subducting Nazca slab at depths of 40 to 700 km.

As shown in Figure 28, the Peru-Chile segment of the circum-Pacific-seismic belt has
experienced two great61 earthquake events in historical time. Recent notable earthquakes
(e.g., the 2001 southern Peru event) have significantly reduced the elastic strain energy along a
300-km-long segment of the circum-Pacific seismic belt. However, the approximately 700-km-
long plate interface between Chala and Limahas has been quiescent since the great earthquakes
of 1868 and 1877, and thus has been accumulating elastic strain energy since the late 19th
century.62 Consequently, elastic strain energy with the potential to produce large-magnitude
earthquakes in the upcoming decades exists along the part of the plate boundary closest to the
Camisea ROW.

Figure 27. Cross-sectional view of the Peru-Chile Trench (after Worthey, Washington State
University website).

61
A “great” earthquake has an approximate Richter magnitude of at least 8.0.
62
Dewey, Silva, and Tavera (2003). “Seismicity and Tectonics” in Southern Peru Earthquake 23 June 2001
Reconnaissance Report, EERI Supplement A to Volume 19.

SF36292.003 A0F0 0607 0806

81
 June 8, 2007

Figure 28. West coast of Peru showing source regions of great events of 1868 and 1877
and epicenters of notable 20th century earthquakes (after Dewey, Silva, and
Tavera 2003).

8.2 Seismic Hazards to Buried Pipelines

Incidents of seismically induced pipeline damage are typically characterized as arising from one
of two earthquake effects:

1. Permanent ground deformation: Strong ground shaking during earthquakes

can induce permanent ground deformations (PGD) in regions where
underlying soil materials or bedrock are susceptible to ground failure
mechanisms. Excessive movement associated with ground failure can, in
turn, lead to pipeline damage, and this mechanism is considered an indirect
consequence of earthquake ground shaking.

SF36292.003 A0F0 0607 0806

82
 June 8, 2007

2. Wave propagation: Seismic waves traveling through the earth (i.e., ground
shaking) induce transient strains63 in buried pipelines and the surrounding soil.
When these transient strains become sufficiently large, damage to buried
pipelines may result. Wave propagation–induced damage can be thought of as
a direct consequence of earthquake ground shaking.

Studies of the performance of pipelines during past earthquakes clearly indicate that the most
serious pipeline damage during earthquakes is caused by PGD, resulting in breaks throughout the
entire portion or large portions of the pipeline system within the PGD zone.64 However, PGD-
induced damage is typically confined to a local geographic area susceptible to ground failure
(e.g., faults, unstable slopes, liquefiable soils, or poorly compacted fills). Ground failure
mechanisms capable of inducing PGD damage to pipelines include surface faulting,65
landslides,66 slope instability liquefaction,67 lateral spread,68 and seismic compression.69

Wave propagation typically induces smaller strains in the pipe than PGD but covers a much
longer extent of the pipeline. Therefore, it follows that wave propagation tends to affect weaker
pipeline components. Earthquakes generate different types of seismic waves. Wave propagation
damage can occur due to the propagation of body waves (i.e., compression or shear) or surface
waves (e.g., Rayleigh or Love). Recent studies have shown that pipes are typically more
susceptible to wave propagation damage when subjected to surface waves, as opposed to body
waves, for similar levels of earthquake motion.70

63
Transient strain is short-term deformation.
64
O’Rourke (1998). “An overview of geotechnical and lifeline earthquake engineering,” ASCE Geotechnical
Special Publication No. 75, Reston, VA, Vol. 2, 1392-1426.
65
Surface faulting is the rupture and displacement of the ground where the earthquake fault intersects the ground
surface.
66
Landsliding occurs during earthquakes when inertial forces introduced during strong shaking cause shear stresses
to exceed the shear strength of the slope materials.
67
Liquefaction refers to a phenomenon wherein saturated soils temporarily lose their strength and behave like a
viscous fluid. This phenomenon generally occurs in loose, cohesionless soils subjected to rapid loadings
(i.e., earthquakes), which cause the development of excess pore-water pressures.
68
Lateral movement of ground, typically with little to no perceptible slope, toward an open face such as a stream
channel. The mechanism of lateral spread requires the occurrence of soil liquefaction.
69
Seismic compression refers to the settlement of unsaturated soils due to strong earthquake shaking.
70
O’Rourke and Deyoe (2004). “Seismic damage to segmented buried pipe,” Earthquake Spectra, EERI, Vol.
20(4), 1167-1183.

SF36292.003 A0F0 0607 0806

83
 June 8, 2007

8.3 Seismic Risk Evaluation

8.3.1 Characterization of Seismic Demand

Vector Peru S.A.C. (Vector)71 characterized the regional seismic demand72 for each of the three
sectors (selva, sierra, and costa) and provided more specific evaluations of ground motions for
the pump stations, pressure control stations, and pressure reducing stations. Vector evaluated the
earthquake loading for two different scenario earthquake events: (i) an “extreme” event based
on a deterministic assessment corresponding to the greatest earthquake that a seismogenic zone
can produce under well-defined tectonic conditions; and (ii) a “design” event based on a
probabilistic assessment assuming an event with a 475-year recurrence interval. For a
subduction-zone earthquake, Vector concluded that the costa sector would experience the
strongest shaking, with estimated peak ground accelerations of 0.7g73 for the extreme event and
0.5g for the design event. They also identified the potential for continental seismic activity
occurring along the Razuwilcas fault system near Ayacucho in the sierra sector.

Exponent reviewed the Vector study and concluded that engineering models used to evaluate the
seismic demand may under-predict the actual ground-shaking hazard. Several studies have
significantly advanced the scientific understanding of ground motions generated by subduction-
zone earthquakes. Specifically, there are substantial differences in the ground motion
characteristics between the various faulting mechanisms (i.e., inter-slab versus in-slab events).
These findings may have important implications as to which earthquake scenarios may affect the
hazard along various segments of the Camisea ROW. Consequently, additional studies may be
warranted to re-evaluate the seismic demand.

71
Vector Peru S.A.C. (2001). “Estudio de Peligro Sismico Traza de los Ductors de Gas Y Condensados Proyecto
Camisea.”
72
Seismic demand is the earthquake loading imposed on the system as a result of strong ground shaking, and is
typically expressed in acceleration and/or velocity.
73
Ground acceleration is measured as a percent of the acceleration of gravity, which is 1.0g. Therefore, 0.5g
represents half the acceleration of gravity.

SF36292.003 A0F0 0607 0806

84
 June 8, 2007

8.3.2 Permanent Ground Deformation (PGD) Hazards

ABS Consulting74 (ABSC) was retained by Techint to investigate surface fault rupture hazards
posed by active faults along the system. Their investigation consisted of identifying active faults
that cross the pipelines, characterizing potential fault displacements, and evaluating the
pipeline’s capacity for such displacements. ABSC identified only one active fault crossing, at
the intended alignment, referred to as the Los Libertadores fault. Empirical correlations were
used to develop design displacements for the fault. The Los Libertadores fault was characterized
as a predominately left-lateral strike-slip fault, with expected mean horizontal displacements of
0.52 m. APA Consulting was subsequently subcontracted by ABSC to evaluate the pipe
performance for such fault displacements and concluded that both the NG and NGL pipes are
expected to withstand the design displacements at the Los Libertadores fault.

ABSC also reviewed aerial photographs to identify regions susceptible to liquefaction, and did
not observe surface evidence of any past liquefaction episodes. For the sierra and selva sectors,
ABSC postulated that soil deposits are less susceptible to liquefaction, because the depositional
environment in the Andes (i.e., rapid uplift and erosion) creates poorly sorted gravels, sands, and
silts with high fines and clay content. They also concluded that large sections of the costa sector
are covered by windblown deposits (e.g., sand dunes), which could mask any evidence of past
liquefaction episodes.

Based on our review of documents provided to Exponent, we conclude that PGD hazards arising
from fault rupture pose a low risk to the pipeline. However, it appears that other modes of PGD
(i.e., landsliding, slope instability, liquefaction, lateral spread, and seismic compression) should
be evaluated on a site-specific basis outside limited areas that have been improved with
geotechnical stabilization measures. Based on our understanding of the regional geology and
descriptions of ground failures during the 2001 Southern Peru Earthquake, seismically induced
landslides pose a substantial risk to the pipeline, particularly for slopes in the selva and sierra
sectors that are marginally stable under non-seismic conditions, and mountainous areas
exhibiting recent or ancient deep-seated landslides. Seismically induced slope instability poses a

74
ABS Consulting (2002). “Seismic hazard investigations of active faults for the Camisea Pipeline, Peru.”

SF36292.003 A0F0 0607 0806

85
 June 8, 2007

substantial risk in the costa sector. Furthermore, sections of the pipeline near river deposits and
in low coastal regions may be susceptible to liquefaction and lateral spread.

8.3.3 Wave Propagation Hazards

ABSC75 evaluated the potential for wave propagation damage to occur along the system. They
performed the analysis for a straight section of buried pipe using the earthquake loadings for the
475-year return period design event developed by Vector. ABSC evaluated the strain levels
associated with these ground motions using conservative soil properties, and concluded that both
the NG and NGL pipes can adequately resist the passage of seismic waves during the design
earthquake.

Techint evaluated the effects of ground shaking on 12 surface facilities using CAESAR II, a
numerical pipe stress analysis software program. Each pipeline model encompassed the
respective surface facility and several hundred meters of buried pipeline upstream and
downstream of the facility. Based on the results of the seismic stress analysis using earthquake
loadings per Vector’s recommendations, Techint found that the surface facilities were code
compliant per ASME B31.4. However, the directional dependency of the seismic loading should
be investigated in more detail.

Based on Exponent’s review of wave propagation studies, we conclude that straight sections of
buried pipe are unlikely to be damaged by seismic wave passage during a major earthquake.
However, segments of above-ground pipe with rigid connections to surface facilities, or with
major curves or bends, may be vulnerable to wave propagation damage. The seismic stress
analysis performed by Techint assumed that all supports move in unison with the defined ground
motion, and hence does not capture the wave-propagation effect. Consequently, an analysis
should be performed to evaluate the performance of above-ground pipe sections with rigid

75
ABS Consulting (2002). “Seismic Verification of Camisea Pipelines.”

SF36292.003 A0F0 0607 0806

86
 June 8, 2007

connections, which are most susceptible to wave propagation damage due to the high
impedance76 contrast.

8.4 Conclusions

The Camisea system lies within regions that are prone to very large and frequent earthquakes.
Consequently, strong ground shaking generated by large earthquakes poses a substantial risk of
damaging the pipeline and disrupting operations. Although seismic hazard studies were
performed as part of the system design, our review suggests that the potential for permanent
ground displacements should be evaluated more comprehensively at the most susceptible sites to
reduce the uncertainty in the risk. Additionally, seismic risk management would benefit from an
update of the design ground motions with up-to-date scientific information. The seismic hazards
arising from fault rupture and wave propagation along straight sections of the buried pipeline are
considered to pose a low risk to the pipeline. However, the effects of wave propagation on the
pipeline at rigid connections and major curves and bends should be evaluated further. Exponent
currently understands that TgP is engaged in a review of seismic risks, as part of their pipeline
integrity management plan, to determine whether these potential seismic risks are acceptable for
this system or whether mitigation measures should be considered.

76
Impedance contrast provides a measure of the stiffness of one material relative to another.

SF36292.003 A0F0 0607 0806

87
 June 8, 2007

9 Scour-Related Risks

9.1 River Crossings and Scour

Scour is defined as the erosion of streambed or bank material due to flowing water. Erosion of
the streambed is considered to have three principal components: (1) long-term scour, (2) general
scour, and (3) local scour. Long-term scour refers to aggradation and degradation of the
streambed due to natural or anthropogenic causes. General scour is the lowering of a streambed
due to the passing of a flood. Local scour is the removal of material due to acceleration of flow
around submerged obstructions.

In addition to the types of scour mentioned above, naturally occurring lateral migration of the
mainstream channel within its flood plain may affect the stability of buried crossings. Factors
that affect lateral stream migration are the geomorphology of the stream, location of the pipe
crossing, characteristics of the flood, and characteristics of the bed and bank materials. Lateral
migration of a streambed can occur gradually over decadal time periods or episodically during
very large flood events.

Pipelines buried in streams that are perennial (year-round), intermittent (flowing reaches
interspersed with dry reaches), and ephemeral (flowing only after rainfall) can become exposed
or undermined as a result of scour. Once exposed or undermined, pipelines can fail or be
damaged due to debris impact or from the spanned weight of the pipe. Once a pipeline is
exposed or undermined, it can potentially become a source of further erosion as water accelerates
around the pipe.

To prevent breakage and subsequent accidental contamination of runoff during the lifetime of the
pipeline, all stream crossings should have burial depths designed and constructed according to
site-specific conditions.

SF36292.003 A0F0 0607 0806

88
 June 8, 2007

9.2 Risks at Buried River Crossings

A rigorous and robust scour analysis of buried river crossings is outlined in Figure 29, which is
taken from the Federal Highway Administration Hydraulic Engineering Circular No. 18 (FHWA
HEC-18). FHWA HEC-18, which provides guidance on the hydrologic and hydraulic aspects of
a scour analysis, is symbiotic with FHWA HEC-20 and -23. HEC-20 provides guidance on the
evaluation of the long-term stability and geomorphology of the stream, and HEC-23 provides
guidance on the design and construction of scour countermeasures.

This methodology outlined in FHWA HEC-18, -20, and -23 considers the factors likely to play a
role in the scour behavior during the lifetime of a pipeline. Although these documents were
developed in the United States to estimate bridge scour depths, they are widely used for various
types of scour studies internationally. Furthermore, aspects of the FHWA approach were utilized
in the original design of the Camisea system. However, the FHWA documents are guidance
documents and, to our knowledge, do not constitute design standards required by Peruvian law.

Exponent reviewed documentation and interviewed the parties involved in the design of the 62
identified river crossings along the ROW. Exponent’s review indicates that specialist firms were
retained to support the general contractor, Techint, in determining the pipelines’ design burial
depth at each river crossing. Techint retained Hydroconsult and GMI Consulting Engineers to
complete hydrology and hydraulic studies on separate areas of the pipeline ROW.77, 78 This
hydrologic information was used by Golder to aid in their calculations of the burial depth of the
NG and NGL pipeline at each river crossing. Techint then used Golder’s studies and design
recommendations to construct each river crossing.79, 80, 81

77
Hydroconsult (2003), “Camisea Project Peru Rio Pisco Preliminary Report on Instream Allignments at Prog
477.3 and 479.”
78
GMI S.A. Ingenieros Consultores, “Estudio Hidrologico General Gasoducto Camisea.” – date unknown.
79
Golder Associates (2002), “Cruce De Ríos Sector Costa Poliducto de Gas y Condensados Proyecto Camisea.”
80
Golder Associates (2002), “Cruce De Ríos Sectores Selva y Sierra Poliducto de Gas y Condensados Proyecto
Camisea.”
81
Golder Associates (2003), “Cruce De Ríos Sectores Costa, Sierra Y Selva Poliducto De Gas y Condensados
Proyecto Camisea.”

SF36292.003 A0F0 0607 0806

89
 June 8, 2007

Figure 29. Flow chart for conducting scour analysis (from FHWA HEC-18).

Exponent’s review of the scour studies indicates that a portion of the guidance in the FHWA
publications was used by Techint to establish the design burial depth. As seen in Figure 30, the
design team incorporated the following elements of the FHWA methodology in their
recommended burial depth, including:

• Hydroconsult and GMI Consulting Engineers completed hydrology and

hydraulic studies for the river crossings. The main objective of these studies
was to quantify the magnitude of the floods in the rivers crossed by the
pipeline. The design return period was 200 years (0.5% chance of occurrence

SF36292.003 A0F0 0607 0806

90
 June 8, 2007

during any year) for most crossings and was 500 years (0.2% chance of
occurrence during any year) for crossings deemed critical.

• Data from 19 rain gauges and 5 stream gauges were relied upon for the GMI
hydrologic study, but in some cases, the data were incomplete or had been
recorded only for short periods of time. Incomplete or short records were
supplemented using the U.S. Army Corps of Engineers’ HEC-4 model, a
stochastic stream flow generation program.

• The river crossing locations were surveyed, and site-specific conditions were
documented during field investigations.

• Data from the hydrology and hydraulic study and the field investigations were
used to calculate the potential scour depth at each river crossing. Different
methodologies for calculating the potential scour depth were evaluated. The
clear-water contraction scour equation (FHWA HEC-18 equation 5.4) was
deemed the most conservative and was used for many of the crossings. The
details of the other evaluated methodologies were not given in the Scour
Analysis Summary Reports published by Golder in 2002 and 2003.

• If the calculated scour depth for rock material was less than 1 meter, it was
increased to 1 meter. If the calculated scour depth for unconsolidated
alluvium was less than 2 meters, it was increased to 2 meters at each crossing.
This provides an additional safety factor above the adopted equations.

• Scour countermeasures, such as reinforcement of the riverbed, stabilization of

embankment with gabion walls, and energy dissipators, were designed and
built in selected locations.

• At each river crossing, the system is reinforced with a concrete cover,

providing a certain amount of protection against abrasion and impact from
small boulders.

SF36292.003 A0F0 0607 0806

91
 June 8, 2007

Figure 30. Methodology utilized by Golder Associates for the scour analysis of the Camisea
Pipeline (taken from “EVALUACIÓN DE CRUCES Y QUEBRADAS PROYECTO
CAMISEA ‘RIVER CROSSING’” – presentation by Golder Associates January 24
and 25, 2007, Lima, Peru.

All of the above-listed factors, when combined, result in a reasonable overall design basis for
pipeline burial depth. However, our general practice and experience in scour prediction, and our
review of the design process shown in Figure 30 and the documentation provided to us, indicate
that certain phenomena could play an important role in the prediction of scour risk for the
Camisea system. These include:

• Stream classification and evaluation of long-term stream stability

• Evaluation of stream bank migration
• Identification of possible headcutting or knickpoint migration
• Evaluation of the potential effects of debris flows.

SF36292.003 A0F0 0607 0806

92
 June 8, 2007

Our review of the overall scour design concludes that there is some uncertainty as to whether
each stream crossing is protected from scour over the long term and during extreme flood events.
The primary sources of uncertainty arise from sparse hydrologic data and the additional
phenomena mentioned above. Given the uncertainty and recognizing that the objective of all
parties involved is to minimize risk, TgP has committed resources in 2007 to investigate the risk
of scour damage at each river crossing, and to design and construct additional mitigation
measures where necessary. During our field inspections in 2006, TgP fortified several river
crossings and smaller streams to mitigate scour risk. The more comprehensive identification of
potential scour risks in 2007 will be helpful to better identify potential hazards and reduce the
current uncertainty.

9.3 Conclusions

The design team’s approach of utilizing their local knowledge and site-specific investigations in
their scour analysis appears to be sound. These methodologies are incorporated into the design
procedure set forth in FHWA HEC-18, -20, and -23, which are widely used in the design of
scour protection for key infrastructure. Our review of the overall scour design concludes that
there is some uncertainty as to whether each stream crossing is protected from scour over the
long term and during extreme flood events, primarily due to the limitations of the available
hydrologic data and the use of the clear-water contraction scour equation. To mitigate this
potential residual risk, TgP has committed to study all river crossings in 2007, to identify which
crossings may require further risk mitigation measures, such as the construction of additional
fortifications and flow control measures. These measures would help to mitigate potential risks
from large-scale flood events.

SF36292.003 A0F0 0607 0806

93
 June 8, 2007

10 Summary and Conclusions

Exponent was retained to provide continued technical assistance to the IDB, related to the
Camisea Transportation System. Overall, our investigation indicates that the primary risks to the
pipeline are geological and geotechnical, and that TgP has taken actions in 2006 to significantly
reduce these risks to the system. Furthermore, TgP continues to address pipeline integrity and
risk issues on an ongoing basis. As these efforts identify risks to the integrity of the pipeline,
risk management actions above and beyond those underway and completed to date may be
necessary.

10.1 Pipeline Design–Related Risks

Our review indicates that the system was designed to comply with the engineering code
requirements of the American Society of Mechanical Engineers (ASME)82. First, the pipe wall
thickness is sufficient to contain the internal pressures of the transported hydrocarbon products
along the entire length of the pipeline, such that hydraulic risks of the design are minimal and
consistent with other pipelines. Second, the system was designed such that external soil loads
would be mitigated by geotechnical stabilization measures to be constructed at sites deemed to
pose a geotechnical or geological hazard.

10.2 Pipeline Construction–Related Risks

Overall, the review showed that (1) the more than 100,000 individual pipe segments were
specifically built for this system and consistent with API 5L requirements, (2) all pipe segments
were subsequently coated with an outer high-density polyethylene (HDPE) layer to protect the
exterior from the environment, (3) the issued welder certificates and qualifications of the
inspectors were in compliance with API 1104, (4) radiography of all girth welds was performed
per API 1104, (5) hydrostatic testing of the system was in compliance with the required ASME

SF36292.003 A0F0 0607 0806

94
 June 8, 2007

codes and performed along the entire length of both pipelines, and (6) construction of the
pipeline used methods that employed standard engineering procedures.

10.3 Geotechnical and Geology-Related Risks

The geotechnical and geologic conditions along the pipe alignment are diverse, dynamic,
complex, and sensitive. During our inspections in 2006, we observed evidence that some
geotechnical stabilization measures implemented during or immediately following construction
proved to be insufficient to adequately mitigate external soil pressures acting on portions of the
NGL pipeline. Based on Exponent’s review, observations, and engineering experience, we
conclude that geotechnical and geologic conditions posed the most significant risk to the
integrity and reliable operation of the system. Geotechnical instability caused or substantially
contributed to two of the five spill incidents (#1 and #5), and geologic instability caused one of
the five spill incidents (#3).

The spill incidents and observed performance of the system as of early 2006 caused TgP to set
into motion an aggressive geotechnical remediation program that began in earnest in April 2006.
Ultimately, the severity and challenges of the dynamic terrain led to the construction of
additional robust geotechnical remedial measures at more than 100 sites in 2006.

Exponent also reviewed a monitoring program initiated by TgP and COGA to help reduce the
risk of future failure resulting from external geotechnical forces. This program allows for the
early detection and subsequent correction of potential problem areas. We expect that the
continuous, ongoing implementation of the risk matrix process and monitoring program may
identify additional sites that are not included in the current risk matrix as new manifestations of
ground movement are observed or detected. However, Exponent believes that the successful
implementation of the above, and the construction of additional geotechnical mitigation
measures in 2007, will further reduce the geotechnical-related risks to the pipeline system.

82
ASME B31.4 [Pipeline Transportation Systems for Liquid Hydrocarbons and Other Liquids] is the applicable
Code for the NGL pipeline, and ASME B31.8 [Gas Transmission and Distribution Piping Systems] is the
applicable code for the larger NG pipeline.

SF36292.003 A0F0 0607 0806

95
 June 8, 2007

10.4 Pipeline Integrity–Related Risks

Several approaches have been adopted by TgP to reduce pipeline integrity-related risks. First,
during construction, all the welds were x-rayed, and the entire pipeline was hydrostatically
tested, reducing the potential number of weld-related and pipeline material defects. Second, any
growth of such defects that would lead to the rupture of the pipe requires the presence of external
loading. Removal of the loading is a good way to further mitigate the risk, and TgP’s ongoing
and prior geotechnical construction program reduces the likelihood of soil movement.

TgP performed an inline inspection of the NGL pipeline in 2006 using the Magnetic Flux
Leakage (MFL) inspection tool and a geometric inline inspection tool. TgP has reported that the
inspection of the NGL pipeline identified 30 reportable defects per requirements of DOT 49
CFR 195.452. TgP is currently excavating these sites to perform a more detailed evaluation and
initiate the appropriate repair measures if required. Overall, results of the MFL inspection tool
and evaluation of metal loss per ASME B31G have shown that, currently, no severe external and
internal corrosion damage exists along the NGL pipeline. However, it appears that there is a
relatively elevated frequency of metal loss occurrences within the first 50 km of the NGL
pipeline. Therefore, this portion of the NGL pipeline may be subject to this potential risk over
the long term, and we recommend that this should be evaluated as part of TgP’s ongoing pipeline
integrity program (see Section 11).

10.5 Seismic-Related Risks

The Camisea system lies within regions that are prone to very large and frequent earthquakes.
Consequently, strong ground shaking generated by large earthquakes poses a substantial risk of
damaging the pipeline and disrupting operations. Although seismic hazard studies were
performed as part of the system design, our review suggests that (1) the potential for permanent
ground displacements should be re-evaluated more comprehensively at the most susceptible
sites, (2) seismic risk management would benefit from an update of the design ground motions,
and (3) the effects of wave propagation on the pipeline at rigid connections and major curves and
bends should be re-evaluated further. It is our understanding that TgP is currently engaged in a
review of seismic risks, as part of their pipeline integrity management plan, to determine whether

SF36292.003 A0F0 0607 0806

96
 June 8, 2007

these potential seismic risks are acceptable for this system or whether mitigation measures
should be considered.

10.6 Scour-Related Risks

The design team’s approach of utilizing their local knowledge and site-specific investigations in
their scour analysis appears to be sound. However, our review of the overall scour design
concludes that there is some uncertainty as to whether each stream crossing is fully protected
from scour over the long term during extreme flood events, primarily due to the limitations of the
available hydrologic data. To mitigate this potential residual risk, we recommended and TgP has
committed to study all river crossings in 2007, to identify which crossings may require further
risk mitigation measures, such as the construction of additional fortifications and flow control
measures.

10.7 Summary

TgP has agreed with the IDB to implement the recommendations listed in Section 11, and
Exponent has been retained to provide continued technical assistance to the IDB on this matter,
which includes review of these actions and additional site visits in 2007. Based on available
information obtained during Exponent’s investigation and the proposed actions, it appears to
Exponent at this time that TgP is performing adequate pipeline integrity management actions,
and that these actions have significantly reduced the risk to the system. However, Exponent also
notes that pipeline integrity management is a continuous process, and thus, future information
and risks need to be continually and properly evaluated. If and when ongoing pipeline integrity
management efforts identify additional issues, risk management actions above and beyond those
currently being taken may be required.

SF36292.003 A0F0 0607 0806

97
 June 8, 2007

11 Recommendations

Exponent performed a pipeline integrity analysis of the pipeline components of the Camisea
Transportation System. Overall, our investigation indicates that the primary risks to the pipeline
are geological and geotechnical, with mechanical pipe integrity, seismic events, and river
scouring as secondary risks. TgP has implemented various actions to help reduce these risks,
including various interim recommendations made by Exponent during our investigation, such as
the excavation of specific sites to verify data from the inline MFL inspection, construction of
additional geotechnical stabilization measures in 2006, performance of rigorous site-specific
stability analysis, and installation of monitoring equipment at specific sites. Exponent has
recently made several recommendations to further reduce the risk, based on the results and
conclusions of this investigation, from which TgP and the IDB have developed a technical action
plan for 2007. These recommendations, which TgP has committed to implement, are described
in the following sections.

11.1 Geotechnical and Geologic

The following are geotechnical and geological recommendations that apply to the pipeline
alignment:

1. Exponent recommends that TgP develop a comprehensive risk management

plan (RMP) that establishes a framework for pipeline integrity risk
management. At a minimum, the plan should clearly define the framework
and approach for making management decisions in terms of what geotechnical
remedial measures should be implemented, and will identify means and
responsibilities for execution of such remedial measures.

2. TgP should continue to apply the risk assessment method as described herein
in a proactive manner to other sections of the pipeline.

SF36292.003 A0F0 0607 0806

98
 June 8, 2007

3. TgP should continue the implementation of the ROW inspection program as

described in Section 6.

4. Prior to our site inspections, TgP installed strain gauges on the pipelines at
seven locations. The strain gauge program should provide the type of data
that will allow TgP to evaluate movement in the pipelines and should allow a
preemptive remedial program to be implemented at a specific location, should
the data indicate that stabilization is either necessary or desirable. We
recommend that the results and findings from this analysis be transmitted
from engineers at Lurin to the COGA field personnel on a regular and
frequent basis.

5. Exponent recommends that attention and care be given to proper soil

conditioning and compaction of backfill behind and in front of retaining walls
and other remedial structures. Proper compaction will reduce forces acting on
the walls and reduce potential ground cracks that could form in poorly
compacted fill behind walls. The reduction of potential ground cracks will
simplify the geotechnical investigation at these locations.

6. Exponent recommends that the project purchase a small drilling and sampling
rig that is portable using helicopters.

7. During the past site inspections, Exponent recommended some basic

geotechnical testing equipment to perform index property testing of soil
samples taken from the borings made on the project. We understand that
some equipment has been purchased since then and is being used at the
COGA office at Kiteni.

In addition to the above general recommendations, Exponent recommends that the following
geotechnical technologies be employed at specific sites as appropriate:

8. Piezometers to monitor water levels at nearly all sites of geotechnical

stabilization. At most of the locations, piezometers can be constructed by
hand by driving 2.5-cm-diameter perforated metal pipe into the ground using

SF36292.003 A0F0 0607 0806

99
 June 8, 2007

simple hand drive hammers. Pipe should be relatively heavy (schedule 80)
pipe with threaded couplings.

9. Slope inclinometers to monitor subsurface movements in soil and rock.

Installation of these will require the use of portable drilling equipment, as
discussed with TgP, and qualified personnel to install, monitor, analyze, and
interpret results.

10. Detailed geotechnical investigations, including soil sampling in borings and

test pits, as necessary, to be performed at PS #2 as soon as possible. This
work will require the use of the portable drilling and sampling equipment, as
discussed with TgP. Following completion of this work, Exponent
recommends that a detailed geotechnical analysis be performed to identify
subsurface conditions and construct cross sections for this site, to evaluate soil
strength and groundwater levels, and to perform conventional stability
analyses of existing slopes at the site. It is our understanding that TgP has
performed this geotechnical study, and construction of remedial measures is
currently underway.

11. Specific observations and physical measurements to be made at individual

sites. Examples of observations and measurements are as follows:

a. Alignment of concrete-lined surface water channels and soil-cement

bag current breakers and lined channels.

b. Nature and amount of the occurrence of surface water flows,

especially noting new occurrences, presence of discoloration, increases
or decreases in the rate of flow, and other such observations.

c. Nature and size of ground cracking or other indications of

displacement in the slopes. This may involve placement of simple
staking to assist in making precise quantitative measurements of
movement.

SF36292.003 A0F0 0607 0806

100
 June 8, 2007

d. Alignment of the gabion walls at each location.

e. Formation of erosion gullies, retreat of stream banks, damage to check

dams, or shifting of rock protection on the floors of streambeds.

11.2 Seismic

TgP should perform a complete and comprehensive upgrade to the seismic hazard assessment for
the system, specifically related to the potential for permanent ground displacements, design
ground motions, and effects of wave propagation on the pipeline at rigid connections and major
curves and bends. This information shall be used to update, if necessary, the seismic risk
management plan and perform any required upgrades.

11.3 Scour

TgP should evaluate the scour potential and existing countermeasures at each river crossing and
should perform a scour risk analysis that classifies the scour risk at each river crossing. This
study shall consider factors that are very likely to play a part in some of the stream crossings
where more complex processes are active, such as long-term degradation, lateral shifting of the
stream banks, and possible floods of relatively extreme magnitude. Based on the analysis, TgP
should implement, if necessary, appropriate remedial measures.

11.4 Pipe

To reduce pipe integrity–related risks, TgP has committed to perform a root-cause analysis of the
first, fourth, and fifth spill incidents, including an experimental evaluation of the nucleation of
circumferential cracks at pipe girth welds that are loaded externally (i.e., soil movement). Later,
TgP will investigate the feasibility of identifying circumferential cracks along the system and
will continue with its external excavation program to make further advances in the interpretation
of the recently performed MFL inline inspection.

SF36292.003 A0F0 0607 0806

101