journal of pipeline engineering - pipemag.com - home page - sample... · pipeline engineering...

March, 2009 Vol.8, No.1

ScientificSurveys Ltd, UK

ClarionTechnical Publishers, USA

Journal ofPipeline Engineering

incorporatingThe Journal of Pipeline Integrity

SAMPLE C

OPY

Journal of Pipeline Engineering

Editorial Board - 2009

Obiechina Akpachiogu, Cost Engineering Coordinator, Addax PetroleumDevelopment Nigeria, Lagos, Nigeria

Mohd Nazmi Ali Napiah, Pipeline Engineer, Petronas Gas, Segamat, MalaysiaDr Michael Beller, NDT Systems & Services AG, Stutensee, Germany

Jorge Bonnetto, Operations Vice President, TGS, Buenos Aires, ArgentinaMauricio Chequer, Tuboscope Pipeline Services, Mexico City, Mexico

Dr Andrew Cosham, Atkins Boreas, Newcastle upon Tyne, UKProf. Rudi Denys, Universiteit Gent – Laboratory Soete, Gent, Belgium

Leigh Fletcher, MIAB Technology Pty Ltd, Bright, AustraliaRoger Gomez Boland, Sub-Gerente Control, Transierra SA,

Santa Cruz de la Sierra, BoliviaDaniel Hamburger, Pipeline Maintenance Manager, El Paso Eastern Pipelines,

Birmingham, AL, USAProf. Phil Hopkins, Executive Director, Penspen Ltd, Newcastle upon Tyne, UK

Michael Istre, Engineering Supervisor, Project Consulting Services,Houston, TX, USA

Dr Shawn Kenny, Memorial University of Newfoundland – Faculty of Engineeringand Applied Science, St John’s, Canada

Dr Gerhard Knauf, Mannesmann Forschungsinstitut GmbH, Duisburg, GermanyLino Moreira, General Manager – Development and Technology Innovation,

Petrobras Transporte SA, Rio de Janeiro, BrazilProf. Andrew Palmer, Dept of Civil Engineering – National University of Singapore,

SingaporeProf. Dimitri Pavlou, Professor of Mechanical Engineering,

Technological Institute of Halkida , Halkida, GreeceDr Julia Race, School of Marine Sciences – University of Newcastle,

Newcastle upon Tyne, UKDr John Smart, John Smart & Associates, Houston, TX, USA

Jan Spiekhout, NV Nederlandse Gasunie, Groningen, NetherlandsDr Nobuhisa Suzuki, JFE R&D Corporation, Kawasaki, Japan

Prof. Sviatoslav Timashev, Russian Academy of Sciences – Science& Engineering Centre, Ekaterinburg, Russia

Patrick Vieth, Senior Vice President – Integrity & Materials,CC Technologies, Dublin, OH, USA

Dr Joe Zhou, Technology Leader, TransCanada PipeLines Ltd, Calgary, CanadaDr Xian-Kui Zhu, Senior Research Scientist, Battelle Pipeline Technology Center,

Columbus, OH, USA

❖ ❖ ❖

SAMPLE C

OPY

1st Quarter, 2009 1

The Journal ofPipeline EngineeringincorporatingThe Journal of Pipeline Integrity

Volume 8, No 1 • First Quarter, 2009

Contents

Peter Tuft .................................................................................................................................................................... 5The Australian approach to pipeline safety management

Dr Érika S M Nicoletti and Ricardo Dias de Souza ............................................................................................... 19A practical approach in pipeline corrosion modelling: Part 1 – Long-term integrity forecasting

Dr John Beavers, Patrick Vieth, and Dr Narasi Sridhar ....................................................................................... 29Ethanol transportation: status of research, and integrity management

Dr Chris Alexander .................................................................................................................................................. 35Evaluating damage to on- and offshore pipelines using data acquired using ILI

Professor Andrew Palmer and Dr Yue Qianjin ..................................................................................................... 49Rethinking laybarge pipelaying

H S Costa-Mattos, J M L Reis, R F Sampaio, and V A Perrut ............................................................................... 53Rehabilitation of corroded steel pipelines with epoxy repair systems

Assadollah Maleknejad ............................................................................................................................................. 63Technical and commercial challenges in procurement and implementation of major international pipeline projects

❖ ❖ ❖

As part of an American Petroleum Institute study, experimental efforts were undertaken to assess the effects ofwrinkle bends on the fatigue life of pipelines, and three 36-in x 0.281-in pipes were fitted with wrinkle bends havingnominal depths of 2%, 4%, and 6% (wrinkle depth percentage calculated by dividing wrinkle depth by the nominal

diameter of the pipe). OUR COVER PICTURE shows the pipe sample with 2% wrinkles, and details of this researchare included in the paper by Dr Alexander on pages 35-47.

SAMPLE C

OPY

The Journal of Pipeline Engineering2

1. Disclaimer: While every effort is made to check theaccuracy of the contributions published in The Journal ofPipeline Engineering, Scientific Surveys Ltd and ClarionTechnical Publishers do not accept responsibility for theviews expressed which, although made in good faith, arethose of the authors alone.

2. Copyright and photocopying: © 2009 Scientific SurveysLtd and Clarion Technical Publishers. All rights reserved.No part of this publication may be reproduced, stored ortransmitted in any form or by any means without the priorpermission in writing from the copyright holder.Authorization to photocopy items for internal and personaluse is granted by the copyright holder for libraries andother users registered with their local reproduction rightsorganization. This consent does not extend to other kindsof copying such as copying for general distribution, foradvertising and promotional purposes, for creating newcollective works, or for resale. Special requests should beaddressed to Scientific Surveys Ltd, PO Box 21, BeaconsfieldHP9 1NS, UK, email: [email protected].

3. Information for subscribers: The Journal of PipelineEngineering (incorporating the Journal of Pipeline Integrity)is published four times each year. The subscription pricefor 2009 is US$350 per year (inc. airmail postage). Membersof the Professional Institute of Pipeline Engineers cansubscribe for the special rate of US$175/year (inc. airmailpostage). Subscribers receive free on-line access to all issuesof the Journal during the period of their subscription.

4. Back issues: Single issues from current and past volumes(and recent issues of the Journal of Pipeline Integrity) areavailable for US$87.50 per copy.

5. Publisher: The Journal of Pipeline Engineering ispublished by Scientific Surveys Ltd (UK) and ClarionTechnical Publishers (USA):

Scientific Surveys Ltd, PO Box 21, BeaconsfieldHP9 1NS, UKtel: +44 (0)1494 675139fax: +44 (0)1494 670155email: [email protected]: www.j-pipe-eng.com

www.pipemag.com

Editor and publisher: John Tiratsooemail: [email protected]

Clarion Technical Publishers, 3401 Louisiana,Suite 255, Houston TX 77002, USAtel: +1 713 521 5929fax: +1 713 521 9255web: www.clarion.org

Associate publisher: BJ Loweemail: [email protected]

6. ISSN 1753 2116

THE Journal of Pipeline Engineering (incorporating the Journal of Pipeline Integrity) is an independent, international,quarterly journal, devoted to the subject of promoting the science of pipeline engineering – and maintaining and

improving pipeline integrity – for oil, gas, and products pipelines. The editorial content is original papers on all aspectsof the subject. Papers sent to the Journal should not be submitted elsewhere while under editorial consideration.

Authors wishing to submit papers should send them to the Editor, The Journal of Pipeline Engineering, PO Box 21,Beaconsfield, HP9 1NS, UK or to Clarion Technical Publishers, 3401 Louisiana, Suite 255, Houston, TX 77002, USA.

Instructions for authors are available on request: please contact the Editor at the address given below. All contributionswill be reviewed for technical content and general presentation.

The Journal of Pipeline Engineering aims to publish papers of quality within six months of manuscript acceptance.

Notes

v v v

www.j-pipe-eng.comwent live on 1 September 2008

SAMPLE C

OPY

1st Quarter, 2009 3

Editorial

THE NEWLY-published report from Chatham Houseentitled Transit troubles – pipelines as a source of conflict*

raises a number of interesting and important issues, and isworth studying in detail, and can be downloaded from thereference below. We are pleased to have the agreement ofthe report’s author, Professor Paul Stevens, to publish thereport’s summary; a brief biography for him appears at theend of this article, from which it will be seen that he iseminently well-placed to be a commentator in this area ofthe pipeline industry, where engineering and politics eithermeet or clash, depending on the viewpoint.

“RECENT EVENTS between Russia and Ukraine at the start of 2009, and Russia and Georgia in2008, have brought transit pipelines back into the mediaspotlight. Any reading of the history of transit oil and gaspipelines suggests a tendency to produce conflict anddisagreement, often resulting in the cessation of throughput,sometimes for a short period and sometimes for longer. Itis tempting to attribute this to bad political relationsbetween neighbours. This is certainly part of the story, butalso important is the nature of the ‘transit terms’ – tariffsand offtake terms – whereby transit countries are rewardedfor allowing transit. Put simply, the trouble with transitpipelines has a significant economic basis. The reportaddresses three questions:

• Why will oil and gas transit pipelines become moreimportant to global energy markets in the future?

• Why has the history of such pipelines been litteredwith conflict between the various parties?

• What might be done to improve this record in thefuture and make transit pipelines less troublesome?

Chapter 1 defines transit pipelines as lines which crossanother’s ‘sovereign’ territory to get the oil or gas to market.Such lines have a number of relevant, commoncharacteristics which tend to generate conflict. Differentparties are involved, each with different interests andmotivations. This invites disagreement between the partiesbecause of the benefits to be shared and the fact thatmechanisms exist toï: encourage one or other party to seek

a greater share. Even though this would apply to anycommercial transaction, the key difference with transitpipelines is that there is no overarching jurisdiction. Moretransit pipelines will be needed in the future, since oil andgas reserves close to market are being depleted, and there isgrowing demand for natural gas in the world’s primaryenergy mix. In recent years, there has been a noticeablefragmentation of legal jurisdictions as the Soviet Unionand former Yugoslavia both collapsed. Many of the newtransit pipeline projects being discussed are essentially theresult of gaming strategies between the various players andwill fail to materialize.

Chapter 2 starts with a brief history of the many transitpipelines which have been associated with very negativeexperiences. In the past, they included those operating inthe Middle East; more recently, attention has been focusedon those in the former Soviet Union. The chapter thendescribes lines which can be viewed either as success storiesor as having too recent a history for the outcome to bedetermined. This history helps in identifying whichcharacteristics make for ‘good’ and ‘bad’ transit countries.These include:

• the importance of foreign direct investment in thetransit country’s development strategy;

• the importance of the transit fee in the country’smacro economy;

• the dependence upon offtake from the line;• the availability of alternative routes;• whether the transit country is also an oil or gas

exporter in its own right.

Chapter 3 seeks explanations for poor performance interms of politics but with the main discussion focusing onthe underlying economics which generate conflict. Oneobvious source of political disputes is a history of badrelations between neighbouring countries. As for theeconomics, the key explanation is that there is no reasonable,objective basis for determining ‘transit terms’. The onlysensible reason for the existence of a transit fee is to allowthe transit country to share in the benefits of the project.This share will reflect the relative bargaining power of theparties to the negotiations. Over time this changes and thusthere are always pressures to change the transit terms. This

*Transit troubles: pipelines as a source of conflict. Prof. Paul Stevens,2009. A Chatham House Report – see www.chathamhouse.org.uk.

Pipelines as a source of conflict

SAMPLE C

OPY


trend is greatly encouraged by the existence of the‘obsolescing bargain’, the structure of pipeline costs, andthe growing volatility of oil and gas prices.

Chapter 4 considers possible solutions to help reduceconflict and supply disruptions. These include:

• a military solution;• encouraging the transit country into the global

economy to make it dependent upon foreign directinvestment;

• making the transit country dependent upon its owngas and oil supplies from the pipeline, although thiscan be a double-edged sword;

• considering alternatives to the transit country notonly in terms of geographic routes but (for gas) theactual means of transport including, for example,the use of liquefied natural gas (LNG);

• encouraging multilateral jurisdictional solutionssuch as the Energy Charter Treaty;

• developing mutual dependence between the transitcountry and the producer/consumer country.

Finally, the report considers a new solution: basing the‘transit terms’ on a progressive fiscal arrangement similarto the sort of systems which govern upstream oil agreements.The report concludes that there will be an increasing needfor and dependence upon oil and gas transit pipelines butsuch pipelines are inherently unstable because of politicaldisputes and also, of equal importance, as a result ofcommercial disputes over the transit terms. Thesecommercial disputes arise because there is no objective,reasonable or fair way of setting the transit terms. Many ofthe apparent solutions to this problem are, on closerexamination, at best ineffective, at least in currentcircumstances. More generally, history suggests that a goodexperience with transit pipelines requires certain best-practice conditions to be met. These include:

• a clear definition and acceptance of the rules;• projects driven by commercial considerations;• credible threats to deter the ‘obsolescing bargain’;• mechanisms to create a balance of interest.

However, it is difficult to turn this ‘wish list’ into a practicalagenda. The only practical, realistic solution in the nearterm is to introduce ‘progressive’ transit terms to existingand new agreements. However, ultimately both consumersand producers must diversify as far as is economicallypractical.

Professor Paul Stevens is Senior Research Fellow for Energy atChatham House, Emeritus Professor at Dundee University andConsulting Professor at Stanford University. He was educated asan economist and as a specialist on the Middle East at CambridgeUniversity and the School of Oriental and African Studies,London. He taught at the American University of Beirut inLebanon (1973–79), interspersed with two years as an oilconsultant; at the University of Surrey as lecturer and senior

lecturer in economics (1979–93); and as Professor of PetroleumPolicy and Economics at the Centre for Energy, Petroleum andMineral Law and Policy, University of Dundee (1993–2008) –a chair created by BP.

US companies explore ethanolpipeline through US Midwest

TWO MAJOR US pipeline companies have announcedtheir plans to assess the feasibility of constructing an

ethanol pipeline through the Midwest. If built, the pipelinewould the first one totally dedicated to transporting ethanolin the US. Oklahoma-based Magellan Midstream Partnersand Pennsylvania-based Buckeye Partners have partneredto explore creating the 2720-km long pipeline to transportethanol from plants in Illinois, Iowa, Minnesota, andSouth Dakota to major cities including Pittsburgh,Philadelphia, and New York. The project is estimated tocost more than $3bn.

The American Coalition for Ethanol’s 2007 report listsIllinois as the second largest producer of ethanol in the US,at 317m gall/yr, and corn grown in Illinois is used toproduce 40% of the ethanol consumed in the US, accordingto the Illinois Corn Growers Association (ICGA). Nearlyone-third of all gasoline in the US already contains lowlevels of ethanol – usually between 5.7% and 10%, and theICGA reports that 95% of the gasoline sold in the Chicagoarea contains 10% ethanol. However, high levels of ethanolcannot be piped through existing gasoline lines withoutdamaging them. Once ethanol has been transported throughexisting pipelines, they can’t be shared with other refinedproducts. “In pipelines today, you can ship differentmaterials through in batches, with plugs that separate theshipments. However, ethanol – because it absorbs water,and is a corrosive agent – is really difficult to use in a non-dedicated pipeline,” John Urbanchuk, the director ofexpert-resources firm LECG, said.

Magellan and Buckeye may be years away from construction,simply because not much is known about transportingethanol through pipelines. Studies on the technical issuesand economic impact of creating an ethanol pipeline arecontinuing, as highlighted in Dr John Beaver et al.’s paperon pages 29-34; no ethanol pipelines exist in the US,though Brazil is in the process of constructing one andHouston-based Kinder Morgan is understood to haveannounced plans to test an ethanol pipeline in Florida thisyear.

Changing the way ethanol is transported may have more ofan effect on consumer costs than adopting alternate fuelsor even falling oil prices. “If you looked at something inIllinois or maybe Iowa, sending it to the East Coast byfreight is anywhere between 16 and 18 cents a gallon. If you

”

concluded on p61

SAMPLE C

OPY

1st Quarter, 2009 5

The Australian pipeline industryTransmission pipelines in Australia are often long but ofrelatively-small diameter (maximum 34in, typically 12-18in).They traverse vast lengths of remote and sparsely-populatedcountry, but there are also substantial lengths in semi-ruralareas and urban outskirts, and some within urban areas.There is a growing problem of urban encroachment onpipelines originally constructed in rural locations. MostAustralian pipelines are relatively young (80% built since1975) and therefore in reasonably good condition as aresult of being designed and built to modern practices andwith modern coatings, as well as having had limited time todeteriorate.

The Australian pipeline industry is relatively small byglobal standards. The total length of high pressuretransmission pipelines is just under 30,000km, and thereare only a handful of major pipeline operating companies.Nevertheless the industry is quite large enough to bevigorous and to support a healthy population of specialistpipeline engineers. Some of the larger Australian pipelineconstruction and engineering service companies havesuccessful export businesses with projects in diverse locationsaround the world.

The Australian Pipeline Industry Association (APIA)sponsors an active research programme and has a co-operative research agreement with PRCI in North Americaand EPRG in Europe; the most recent tripartite JointTechnical Meeting was held in Canberra in 2007.Also well supported by APIA is Standards AustraliaCommittee ME38, responsible for AS 2885. This committeehas been active in developing standards for pipeline design/construction, welding, operation, and pressure testing.The committee and its working groups includerepresentatives from all sectors of the industry as well thetechnical regulators from each state, and has been responsive

Author’s contact details:tel: +61 2 9983 1511email: [email protected]

This paper was presented as part of the proceedings of the 7th InternationalPipeline Conference – IPC 2008 – held in Calgary on 29 September –3 October, 2008, and organized by the ASME’s Pipeline SystemsDivision. It is published here by kind permission of ASME.

The Australian approach topipeline safety management

by Peter TuftPeter Tuft & Associates, West Pymble, NSW, Australia

THE AUSTRALIAN APPROACH to management of pipeline safety and risk differs from that used in mostother parts of the world: there is a strong focus on identifying causes of failure and designing against

them using a cause/control model of risk management, and little use of quantitative risk assessment.

Oil and gas pipelines in Australia are designed, constructed, and operated in accordance with AS 2885. Sincea major revision in 1997, this has been a risk-based standard. While it does contain numerous design rules,their application is flexible and to some extent dependent on the outcomes of a mandatory safetymanagement study. Key elements of the standard include separation of wall thickness selection frompressure design factor, mandatory protection against external interference, special requirements for high-consequence areas, and a safety management study process including qualitative assessment of residualrisks.

The AS 2885 process has been shown to be workable and effective :. It results in a design which is optimizedfor safety at every point along the pipeline while not incurring costs for features that do not reduce risk.The process is oriented principally to design of new pipelines, but is equally applicable to management ofolder pipelines which are suffering degradation or subject to changed conditions such as urban encroachment.

SAMPLE C

OPY


in meeting the needs of both groups. The industry andregulatory representatives have a very co-operative approachand opinions diverge only on peripheral issues. The APIAresearch programme includes a number of projects initiatedin response to the needs of the ME38 committee and theresearch outcomes are incorporated in new revisions ofstandards.

The considerable distances and small loads in Australiacreate economic pressure to minimize pipeline costs, whichprovides a driver for technical innovation. The unifiedapproach of the APIA and the ME38 committee providesthe means by which innovation can be relatively quicklyincorporated into standards and applied to new and existingpipelines.

Basis for AS 2885Prior to 1997, AS 2885 and its precedent standards hadbeen developed from the ASME/ANSI B31.4 and B31.8codes, although considerable differences from those codeshad evolved over time. In preparation for the 1997 revision,the code committee recognized that, despite the bestintentions, a rigid rule-based code would often producedesigns that were less than optimum in terms of safety,economics, or both. There was particular concern aboutanomalies that arose from the rule-based approach atboundaries between different location classes (reflectingpopulation density, often defined in a very arbitrary way),and also with the way that the rules handled changes inpopulation density as a result of urban growth.

AS 2885, of course, still includes many rules. However,they are more flexible than previously, and the overridingrequirement is to assess risks and ensure that they are

satisfactorily controlled by any means that are appropriaterather than by application of a narrow set of fixed rules.

A fundamental aspect of the standard is the safetymanagement study (SMS), described in more detail later inthis paper. Virtually all aspects of the design must bereviewed through the SMS. While this may appear onerous,it is the route to flexibility in the application of rules so thatthe design can be optimized for safety.

The SMS includes a qualitative risk review process, with theobjective of identifying threats which may cause failure andensuring that they are managed so that the residual risk istolerable. The intention is that safety and risk managementshould be done by the engineers responsible for the designand operation of the pipeline, rather than being outsourcedto risk specialists. Pipeline engineering and risk managementshould be integrated, and a corollary of this is thatengineering and risk management form an iterative process;the design and operating procedures affect the risk profile,and treatment of risk feeds back to the design andprocedures. Since pipeline design and operation aregenerally not complex processes, it is eminently sensiblethat this loop be contained entirely within the small teamresponsible for pipeline engineering.

Risk specialists have an occasional role in providing technicalanalysis of the consequences of a pipeline failure (forunusual cases where the standardised approach is notapplicable), and also for those few cases where quantitativerisk assessment may be required.

There is little use of quantitative risk assessment (QRA) inthe analysis of Australian pipelines. Attempts have beenmade to use statistically-based QRA, but for such methodsto produce realistic results they must be based on defensible

Fig.1. The Australian pipeline network.

SAMPLE C

OPY

1st Quarter, 2009 7

failure rates. The Australian pipeline incident record is toosparse for meaningful data to be extracted. There are lessthan 200 recorded pipeline incidents in the APIA database,and only a small fraction of these involve loss of containment.This is far too little from which to develop average failurerates which account for the range of parameters that canaffect pipeline failure (wall thickness, depth of cover, andlocation class, to nominate just three that are most critical).Some QRA studies have been done based on UK and/orEuropean statistical failure-rate data, notwithstanding thatthat the pipeline may be in the remote outback, and theresulting predicted failure rates have been one or twoorders of magnitude higher than the overall averageAustralian failure rate (which includes the higher rate ofincidents from more populated areas).

Such misuse of QRA methods has done much to damagetheir credibility in the Australian pipeline industry, whichis unfortunate because there are applications of quantitativemethods that are valid and useful. In particular, modernreliability-based quantitative methods have considerablepotential but have not yet been adopted. AS 2885acknowledges that QRA potentially has a role in assistingthe evaluation of risk-treatment alternatives, to permitcomparison of the risk-reduction benefits of various options.Statistical QRA may also have a role in assessing the risksassociated with pipeline facilities which comprise standardprocess plant components and can therefore call on theextensive process plant failure data.

A tacit feature of the AS 2885 principles is that whilepipeline safety is the overriding priority and cannot becompromised, there is also flexibility to avoid incurringcosts that do not add any safety benefit. The optimizationof both safety and cost is a recurring theme in this paper.

External interference protectionDamage by external forces is a major contributor to pipelineincidents worldwide, but is particularly dominant inAustralia where it accounts for at least 80% of all incidents.

(This is not because the external damage rate is unusuallyhigh, but because the relatively-young age of Australianpipelines means that to date they have experienced only afew corrosion-related failures.) Also, Australian pipelinestend to be thin-walled because of the relatively-smalldiameters and high-grade steels used, and this makes themmore vulnerable to loss of containment should seriousexternal damage occur.

For these reasons AS 2885 places considerable emphasis onexternal interference protection (EIP). There are mandatoryrequirements for both physical and procedural protectivemeasures, and these must be appropriate to the level ofthreat that is identified:

“A pipeline shall be designed so that multipleindependent physical controls and proceduralcontrols are implemented to prevent failure fromexternal interference by identified threats.

“The purpose of physical controls is to preventfailure resulting from an identified externalinterference event by either physically preventingcontact with the pipe, or by providing adequateresistance to penetration in the pipe itself.

“The purpose of procedural controls is to minimisethe likelihood of external interference activity, withpotential to damage a pipeline, occurring withoutthe knowledge of the pipeline operator, and tomaximise the likelihood of people undertakingsuch activity being aware both of the presence of thepipeline and the possible consequences of damagingit.” (Clause 5.5.1)

The standard requires that all practicable controls beapplied, with a minimum of one physical measure in rurallocations and two in urban locations, and always a minimumof two procedural measures (see Table 1). There isconsiderable detail in the standard on the minimumrequirements for each type of control to be consideredeffective. The overall effectiveness of the external

0.13

0.015

0.22

0.3

0.12

0.6

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Australia, AllIncidents

Australia,LOC

Europe GasLOC

Europe OilLOC

USA Gas LOC USA Oil LOC

Inci

dent

rate

, per

100

0 km

-yr

LOC = Loss of Containment

Australian data for 86- 03Europe data for '97-'01USA data for '98-'02

Europe & US data fromW. Guijt, O&GJ, Jan 26 '04

Fig.2. Comparison of Australian andnon-Australian data.

SAMPLE C

OPY


interference protection design must be reviewed as part ofthe safety management study.

Penetration resistanceThe last line of defence against loss of containment causedby external damage is the resistance of the pipe itself topenetration, and for this reason AS 2885 gives considerableattention to penetration resistance as a physical protectionmeasure. Emphasis to date has been on resistance toexcavators, given both their ubiquity and the state ofknowledge, but it is recognized that other equipment –such as boring rigs – can also pose a significant threat.

APIA sponsored research to determine the relationshipsbetween penetration force, pipe properties (grade, wallthickness), and excavator parameters (tooth dimensions,bucket force, excavator mass). (Previous work had beendone by others, particularly in Europe, but was not directlyapplicable to typical thin-walled Australian pipelines.) Itwas found that for a given pipe and tool dimensions thereis excellent agreement between experimental and finite-element results for the force required to penetrate, but ofcourse some variability enters the relationship betweenmachine size and bucket force capability. Nevertheless the

resulting relationships are quite adequate for estimatingthe maximum size of excavator capable of penetrating anygiven pipe, and these formulae have been incorporated inAS 2885.

Because there is huge uncertainty about the actual impactconditions the equations include an empirical parameterbased on limited full-scale field trials. Adjustment of theparameter permits calculation of an upper-bound value(penetration quite likely) and lower-bound value(penetration not credible) for the size of excavator that mayresult in puncture.

AS 2885 mandates penetration resistance as a physicalprotection measure in the higher location classes. It isoptional in rural areas, but the standard expects that thepenetration resistance calculations will always be done inorder to provide reference data that can be used in the SMSfor assessing failure mode and consequences.

A key feature of the design for penetration resistance is thatit is divorced from the pressure design factor. Pipeline wallthickness has traditionally been based on a pressure designfactor of 0.72 in remote areas, with progressively-lowerdesign factors as population density increases. However fora pipeline of small diameter and low pressure rating even a

Table 1. Physical and proceduralcontrols (from AS 2885).

Fig.3. An illustration ofa situation where

penetration resistanceis the governing

influence.

slortnoclacisyhP

noitarapeS

lairuB)gnicnef.g.e(noisulcxE

dnuorg-evobanarofreirrabhsarcelcihev.g.e(reirraB)noitcesepip

noitartenepotecnatsiseR ssenkcihtllaW)tnemesacnerobalsetercnoc.g.e(reirrabnoitarteneP

slortnoclarudecorP

ssenerawaenilepiP

lacol(seitrapdrihtdnasrenwodnalhtiwnosiaiL)cte,seitilitu,tnemnrevog

margorpssenerawaytinummoC)"giDuoYerofeBlaiD"(ecivresllac-enO

dnasngisgninraw(gnikramenilepiP)epatrekramdeirub

srodirrocderahsfosresurehtohtiwstnemeergA

noitcetedecnerefretnilanretxEgnillortaP

senoznoitacifitongninnalPnoitcetednoisurtnietomeR

Allowances (corrosion etc)

Penet

ration

resi

stan

ce

"No

ruptu

re"

Oth

ers

as p

er

Cla

use

5.4

.2

Hyd

rost

atic

te

stin

g

Const

ruct

abili

ty

Req

uir

ed

WT

Manufacturing tolerance (seamless pipe only)

Inte

rnal

pre

ssure

No

min

al W

T

SAMPLE C

OPY

1st Quarter, 2009 9

design factor as low as 0.4 will yield a wall thickness that canbe penetrated by a backhoe. Conversely, at the maximumdesign factor of 0.72 a pipeline of large diameter and highpressure rating will have a wall thickness that cannot bepenetrated by a machine of any size, so imposing a lowdesign factor adds very considerable cost without anysignificant improvement in the risk of failure due toexternal interference. By separating the design for internalpressure from the design for penetration resistance, theoverall pipeline can be optimized for both safety and cost.

Wall thicknessAS 2885 explicitly de-couples wall thickness determinationfrom the traditional location class/design factor formula. Itspecifies that the required in-service wall thickness at eachlocation along the pipeline shall be the greatest of thethicknesses required by whichever of the following factorsare applicable at that location:

• pressure containment• penetration resistance• “no rupture” (discussed later)• other stress and strain criteria• control of fast-running fracture• special construction (such as bridges)• vehicle loads at road and rail crossings• mitigation of stress-corrosion cracking• fatigue life• external pressure

The design factor for pressure containment is independentof the location classification. Failure modes other than

overpressure are addressed explicitly through considerationof the other factors influencing wall thickness. Hence inprinciple it is acceptable to operate a pipeline at 72% or80% SMYS in an urban area if the wall thickness can meetall the other requirements without any increase above thatfor pressure containment. The SMS provides a thoroughreview of these issues before a design is finalized

Figure 3 (simplified from the Standard) illustrates oneexample of how this approach is applied in a case wherepenetration resistance happens to be the governinginfluence. The intent of this approach is again the principlethat the design can be optimized for both safety and cost ateach point along the pipeline route.

Location classificationVirtually all pipeline codes use some concept of locationclassification to identify areas where the risks both to andfrom a pipeline are increased by higher population density.AS 2885 is no different, but has refined the concept in twoways and also applies it quite differently.

Firstly, location classification is based on the area thatwould be seriously affected by an ignited full-bore rupture.Location classes are determined from the land use (as aproxy for population density) within a radiation contour of4.7kW/m2 (1500BTU/hr.ft2). This is the generally-acceptedradiation level at which an unprotected person will suffersecond degree burns after 30s exposure. It will vary with thediameter and MAOP of the pipeline and, in principle,could be calculated for each pipeline on the basis of releaserate and flame radiation correlations. However for most gas

1R sgnillewddetalosiylnohtiwdnaldepolevednurolarur-larurdaorB

2R ah5-1yllacipytaerafoskcolbecnediser-elgnis-laitnediserlaruR

1T erutcurtsarfnidetaicossadnanabrubus-laitnediseRsertnecgnippohsllamssahcus

2T fosrebmunegralroyerots-itlumyltnanimoderp-ytisnedhgiH)sertnecliaterrojam.g.e(tneserpelpoep

S cte,sliaj,slatipsoh,sloohcs-evitisneS 2TottnelaviuqE

I selasrac,yrtsudnithgil-lairtsudnIcte,sdray 1TottnelaviuqE

IH royvaeh-lairtsudnIyvaeHseirtsudnisuodrazah

firettal,2Tot2RmorfseiraVetalacseyamtnedicnienilepip

CIC

-rodirroCerutcurtsarfnInommoCsahcussecivresrehtohtiwderahs

daehrevorodeirubdnatropsnartseitilitu

dnanosiailotnoitnettaralucitraPniseitraprehtohtiwstnemeerga

rodirroc

S doolfdnasgnissorcretaw-degrembuSsnialp ngisedlaicepS

Table 2 (above). Primary location classes (from AS 2885).

Table 3 (below). Secondary location classes (from AS 2885).

SAMPLE C

OPY


pipelines the calculation can be standardized to the extentthat it can be simply read from graphs provided in thestandard. (In fact for gas pipelines with the commonMAOP of 10.2MPa there is an even-simpler rule of thumb:the 4.7kW/m2 radiation contour in metres is equal to thepipe diameter in millimetres – 300m for a DN 300 pipeline.)

Basing location classification on the actual worst-caseradiation damage permits the pipeline design to berealistically optimized for both safety and cost. The use ofthe full-bore rupture radiation distance still applies evenfor a “no rupture” design (see below), because it is thepossibility of serious impacts on surrounding people thatgives rise to the “no rupture” requirement in the first place.

AS 2885 defines four primary location classes, summarizedin Table 2 (using very abbreviated definitions). There is aninevitable element of subjectivity in the allocation of locationclass in borderline areas. However this matters little, giventhe nature of the SMS process outlined later and theflexible approach to achieving an adequate level of safety.

A second refinement of the location classification system isthe addition of five secondary location classes to highlightspecial features that may not be adequately identified by theprimary location classification, as shown in Table 3.

As has already been made clear, AS 2885 does not directlylink wall thickness to location class. Location class isinstead used to adjust certain requirements of the safety-management system. In particular there are higher demandsfor external interference protection in higher locationclasses, and special requirements for high-consequenceareas.

High-consequence areasAS 2885 uses the concept of high-consequence areas,although the definition and approach differ from those inthe USA. A high-consequence area is formally defined as “alocation where pipeline failure can be expected to result inmultiple fatalities or significant environmental damage”.In practical terms this includes (but is not necessarilylimited to) residential, high-density, sensitive, and industriallocation classes.

For a new pipeline there are two requirements that must bemet in order to limit the consequences of any failure(Clause 4.7):

• No rupture – “The pipeline shall be designed suchthat rupture is not a credible failure mode.”

• Maximum discharge rate – “... the maximumdischarge rate shall not exceed 10GJ/s in residential,industrial and sensitive locations, or 1GJ/s in high-density locations.” (This brief extract omits otherqualifying requirements.)

For existing pipelines there are other requirements to beapplied when the location class changes as a result of urbandevelopment.

No rupture

The no-rupture requirement can be achieved by either oftwo means. Firstly, the hoop stress may be limited to lessthan 30% SMYS (the approximate level at which there isinsufficient elastic energy in the pipe for any defect topropagate); for some pipelines this may lead touneconomically-large wall thickness. Alternatively, throughthe SMS, the largest credible threat to the pipeline must beidentified, the resulting maximum defect length determined,and the linepipe selected so that the critical defect length(above which the pipe will rupture) is at least 150% of thismaximum hole size. Detailed guidance is provided for thecalculations.

For example, a pipeline of DN 450 (18in NB) and 6.8mm(0.268in) wall thickness in X70 steel operating at 10.2MPa(1480psi) has a critical defect length of 64mm (hoop stress72% SMYS). In a suburban area it is plausible to expect thatthe largest excavation machinery would not exceed 30t,and such a machine fitted with sharply-pointed penetrationteeth is capable of penetrating this pipe. The resulting holefrom the penetration tooth would be around 70mm long,which exceeds the critical defect length, and thus rupturewould be possible. Hence this pipe is not acceptable in ahigh-consequence area. If the wall thickness is increased to9.8mm, the same machine can no longer penetrate at all, so“no rupture” is achieved although the hoop stress is stillwell above 30% SMYS. Clearly this design process dependson the threats that apply to the particular pipeline and theconclusion from this example is not generally applicable(for example, there has been no consideration here of thethreats posed by boring machines).

Discharge rate limit

The limitations on discharge rate were derived from genericQRA studies for suburban and high-density areas. Thesestudies determined the magnitude of the largest ignited gasrelease that would fall within tolerable criteria for societalrisk.

The rate of discharge from a punctured pipeline dependsmainly on the size of hole and the operating pressure, sothese are the only parameters that the design engineer canadjust in order to comply with the limits. Even the scope foradjustment of wall thickness is quite constrained becausepenetration by excavators is largely a binary outcome – agiven machine will either penetrate or it won’t, and if itdoes penetrate then, to a first approximation, the hole sizeis unaffected by the wall thickness. So the options are toeither:

• increase wall thickness until penetration by thelargest identified threat is not possible, or

SAMPLE C

OPY

1st Quarter, 2009 11

• reduce the maximum allowable operating pressureuntil the discharge rate from the largest crediblehole is within the specified limit.

Table 4 shows, as an indication, the hole sizes correspondingto the specified discharge rates for two illustrative operatingpressures.

Change of location class

Clearly the “no-rupture” and discharge-rate limitationscannot easily be applied retrospectively to existing pipelines,particularly those which are affected by urbanencroachment. However in order to maintain consistencyand integrity in the approach to pipeline safety management,the committee revising AS 2885 felt it was necessary tointroduce a requirement that goes as far as possible towardsachieving equivalent results.

These requirements for changed location class are bestsummarized by quoting almost in full:

“Where land use ... changes along the route ofexisting pipelines to permit ... [high consequenceareas] in areas where these uses were previouslyprohibited, ... [it] shall be demonstrated that therisk from a loss of containment involving rupture isALARP [As Low As Reasonably Practicable].

“This assessment shall include analysis of at least thealternatives of the following:

(a) MAOP reduction (to a level where ruptureis non-credible).

(b) Pipe replacement (with no rupture pipe).(c) Pipeline relocation (to a location where the

consequence is eliminated).(d) Modification of land use (to separate the

people from the pipeline).(e) Implementing physical and procedural

protection measures that are effective incontrolling threats capable of causing ruptureof the pipeline.

“For the selected solution, the assessment shalldemonstrate that the cost of the risk reductionmeasures provided by alternative solutions is grossly

disproportionate to the benefit gained from thereduced risk that could result from implementingany of the alternatives.” (Clause 4.7.4, emphasisadded.)

The list of alternatives to be considered indicates that theassessment of risk level and the determination of ALARPis to be taken very seriously. This is one situation whereQRA studies may be of some value in helping withcomparison of the alternatives; even if the absolute valuesof the quantitative risk predictions are questionable, theremay be much use in their comparative rankings, to beassessed alongside the cost of each alternative.

Safety managementstudy process

As noted previously, a formal safety management study is afundamental requirement for any pipeline designed to AS2885. Overall pipeline safety review is essentially a two-stepprocess:

• Design review: identify every potential threat to theintegrity of the pipeline, and if possible apply controlsso that “failure as a result of that threat has beenremoved for all practical purposes”.

• Risk assessment: rank any remaining threats thatare not fully mitigated, and ensure that the residualrisk is reduced to a tolerable level.

The intention, and general experience, is that the vastmajority of threats are eliminated by application of controlsat the design review stage and only a small number progressto risk assessment.

While the SMS process is defined in terms of pipelinedesign, it is equally applicable (and mandated) for regularreview of existing pipelines.

Threat identification

A threat is “any activity or condition that can adverselyaffect the pipeline if not adequately controlled”. Identifyingthreats is conceptually similar to a HAZOP, although not

aPM2.01)isp0541(

aPM7.2)isp004(

s/JG1 mm54 mm09

s/JG01 mm541 *mm082

senilepipynamroferutpurerob-llufylevitceffE*Table 4. Hole sizes corresponding to discharge rates fortwo operating pressures

SAMPLE C

OPY


as formally structured. It helps to have very experiencedpeople involved, and it also helps to use checklists. Mostimportantly, it helps to think laterally about anything thatmight go wrong. Even threats that are believed to be alreadymitigated should be included in the documentation, partlybecause it may lead to identification of weaknesses in theexisting mitigation, and partly because it forms a soundbasis for the safety and operating plan (which is mandatedby AS 2855).

A checklist of potential threat types may include over 100items, ranging from all sorts of external interference events,through a wide range of defects in design, materials, andconstruction, through to diverse mishaps involvingcorrosion, natural events, and operations and maintenance.

It is fundamental that a threat exists at a location; sometimesthat “location” may be the entire pipeline and the threat isconsidered to be non-location-specific (such as corrosion,some design defects, etc). However, the great majority ofthreats are associated with activities or events that occur ata particular location along the pipeline route. This may bea single point, such as threats associated with roadmaintenance at a road crossing, or may be more extended,such as threats associated with logging activities in a forest.

Identifying external interference threats requires particularattention, including real data from the field. Field personnelinvolved in landowner liaison and pipeline patrol areinvaluable aids, to the extent that an SMS that does notinclude their input is seriously devalued. Such people canprovide details of the type and size of excavation machinerylikely to be used at every point along the route (forconsideration in the context of penetration resistance),and can often provide background information on otherthreat types as well.

Gathering this information may appear onerous, but withadequate planning and support (such as a brief “land usersurvey” form) good field personnel can acquire it in thecourse of their ordinary duties. An additional benefit ofgathering the real data is that the maximum equipmentused is not uncommonly found to be rather smaller than

office-based engineers may have guessed (although justsometimes the opposite occurs, which is an equally-compelling reason for gathering the data).

For a new pipeline it is also vital to consider the design forfuture land use, and hence liaison with the local governmentor other planning authority is necessary.

Threat control

Control of external interference threats has already beendiscussed. For other threats, appropriate controls must beput in place, and these may range from standard corrosion-control measures to quality-assurance procedures for design,manufacturing, and construction.

In all cases, the key question to be asked is “are the controlssufficient to prevent failure as a result of the identifiedthreat?” This may appear to be subjective, but there isusually a definitive answer if there is a clear understandingof the identified threat and the controls. (Of course,another threat that may require consideration is failure ofthe controls, but that can be addressed as a separate threatin its own right.) Once sufficient controls are in place, thethreat is accepted and requires no further consideration,other than ensuring that the controls are documented andimplemented.

Threats which cannot be controlled by the application ofexternal interference protection and other design measuresbecome hazardous events which required risk assessment.

Risk assessment

The risk-assessment phase involves qualitative estimationof the likelihood and consequences of failure leading to aranking of risk on a scale of extreme, high, intermediate,low, or negligible. Extreme and high risks are intolerableand must be reduced (but they are also very uncommon ifthe pipeline is well designed in the first place). Intermediaterisks are acceptable only if formally shown to be ALARP(discussed below).

RISKASSESSMENT

DESIGN REVIEWReview Controls

Failure?

LowHigh

Consequences & LikelihoodYes

Accepted Design

No

Yes

Intermediate

ALARP?No

Initial DesignIdentify Threats

Fig.4. The SMS process.

SAMPLE C

OPY


Risks are ranked according to a standard frequency/severityrisk matrix (see Appendix 1), which includes considerableguidance on the ratings for likelihood and severity.

Severity of a failure is most commonly assessed in terms ofthermal radiation effects on people, assuming that any lossof containment will ignite. Given the basis for the locationclassification described previously it is usuallystraightforward to make this type of qualitative assessment.Consideration is also required for the effects of a failure onthe environment and continuity of supply.

As low as reasonably practicable (ALARP)

The concept of ALARP is the basis for determining whethera risk ranked intermediate can be tolerated. “ALARPmeans the cost of further risk reduction measures is grosslydisproportionate to the benefit gained from the reducedrisk that would result.” (definition, Clause 1.5.3). In practicalterms ALARP can be assessed by asking:

• what else can we do to reduce risk? (adjust the route,for instance)

• why haven’t we done it?

ALARP is achieved when either the answer to the firstquestion is “nothing” or the answer to the second is“because the cost is grossly disproportionate”.

SMS implementationThe phases of an SMS are:

• initial design (for new pipelines only)• data gathering, discussed under threat identification

(above)• desktop design review (pre-analysis for workshop)• validation workshop(s) involving all stakeholders

The workshop is mandated by AS 2885; for a major project,the workshop may take a week. A workshop on a singlemajor encroachment problem for an existing pipeline mayoccupy a full day.

The great value of a workshop is that it generates synergiesfrom the interaction of a diverse group of stakeholders,identifying both threats and solutions that would not beapparent to an individual working alone. It also providesconsensus and buy-in from all participants. The value of aworkshop is demonstrated by the observation that nomatter how well the engineering and risk team think theyhave prepared, the workshop will always produce newissues. There is a clear analogy with a HAZOP meeting.Stakeholders who should attend the workshop includedesign engineers, operations’ management, field personnel(land agents, patrol officers, etc.), construction management,relevant technical specialists (involved in corrosion,materials, etc., and possibly part-time), relevant outside

parties (including major landowners, developers, alsopossibly part-time), owner’s representatives, and thetechnical regulator or other government representatives.

The SMS process is defined in Part 1 of AS 2885 (Designand Construction) but Part 3 (Operation) mandates that itbe reviewed every five years, or more frequently if there isa change in circumstances surrounding the pipeline (suchas a proposed development nearby). This means that forpractical purposes the SMS remains “live” for the life of thepipeline. It is of course mandatory that all SMS deliberationsbe recorded in full, and because it is “live” it is desirable forthe documentation to be readily updated.

A database is the preferred means of recording the threats,controls, risk evaluation, and risk treatments. Anappropriately structured database can also record andclose-out various corrective actions that arise during theprocess. Attempts are sometimes made to use a basicspreadsheet but, except in the simplest cases, this rapidlybecomes unwieldy because of the large quantity ofinformation and explanatory comment that must berecorded.

The SMS documentation forms part of the Safety andOperating Plan that is mandated by AS 2885 Part 3, as isentirely appropriate since risk management tacitly orexplicitly underlies almost all pipeline operations andprocedures, other than those involved in scheduling andcommercial metering of the pipeline contents.

Past and future trendsIn Australia over the past 20 years there has been slow butdeliberate movement away from “design by rules” towardsrisk-based “design by thinking”. The first edition of AS2885 in 1987 recognized in principle a need to move awayfrom rules based on location class and design factor (derivedfrom US codes), but it did not achieve a practical change inthe way pipelines were designed. The 1997 revision of AS2885 was a substantial rewrite which introduced the conceptof risk assessment as an underlying principle of pipelinedesign, and was a bold move that required significantchange to the pipeline design process. While the principleswere correct, the wording in the standard did not fullydefine all the requirements necessary to implement iteffectively: conscientious players could do it well, but somejust paid lip service to the new concepts. Nevertheless, theindustry as a whole embraced the idea and graduallydeveloped a broadly-agreed set of good practices. Thesehave now been codified in the 2007 revision of AS 2885.

In AS 2885.1-2007 the principles have not changed but therequirements are specified more explicitly with the aim ofminimizing loopholes. The most substantial change is theaddition of the high-consequence area requirements asdescribed previously. This revision of the standard waspublished little more than a year ago so there has been only

SAMPLE C

OPY


limited opportunity to assess how well the newerrequirements are being accepted and implemented by theindustry. However there was extensive public consultationand issue of drafts over a period of two years, so the industrygenerally has been improving risk-management practicesand there is no evidence to date that the changes arecausing major difficulties.

Future trends in risk management of Australian pipelinesare difficult to assess, especially since the latest version ofthe standard is still so new. The need for further changesmay only become apparent after the new standard has beenin use for a considerable time.

A key issue may be the extent to which quantitative methodsare used. As discussed previously, the Australian pipelineindustry in general does not believe that quantitative risk-analysis methods add value to routine design or riskassessment, and such methods are not currently usedexcept where mandated by local regulations. However, ascities continue to expand around pipelines built to rurallocation class standards, it is perhaps increasingly likely thatsome intractable questions of risk versus cost may benefitfrom comparative numerical risk estimates. Any trend inthis direction will be encouraged by the new AS 2885requirement that the risks due to a pipeline subject tourban encroachment must be rigorously demonstrated tobe ALARP, including comparison with alternatives such asreconstruction or relocation of the line.

In the very long term, as the Australian pipeline networkcontinues to age, quantitative methods may also findincreasing application in prioritising deteriorating pipelinesfor repair.

Reliability-based quantitative risk methods have to datebeen barely recognized in Australia. There is clearly greatpotential in such methods, but it seems likely that they willbe adopted only when driven by risk issues which aredifficult to resolve in any other way.

DiscussionImplicit in the AS 2885 approach to pipeline safety is acause/control model of pipeline incidents: they haveidentifiable causes, and those causes can be controlledthrough design and operation so that the possibility ofpipeline failure is either eliminated or reduced to a tolerablelevel. An alternative view is that incidents have randomcauses and can never be totally prevented; this underliessome QRA approaches which use statistical failure rates.

In fact, all incidents do have causes, but there is uncertaintyin the knowledge of those causes, so the cause/control andrandom views of risk management are really at oppositeends of a spectrum of knowledge. Nevertheless, AS 2885 isbiased strongly towards the identification and managementof specific factors that might lead to failure.

The cause/control model was adopted when the Australianpipeline industry became concerned that inappropriatestatistical QRA methods may have been imposed on it bysafety regulators who were comfortable with this approachin the management of hazardous industries, but who failedto appreciate its shortcomings when applied to pipelines ingeneral and, particularly, pipelines in Australia. As a pre-emptive defence against any such moves, the industrysought to establish a safety-management strategy which itconsidered to be more appropriate, and which wouldprovide genuine improvements in safety while not incurringcosts that did not achieve practical reductions in risk. Onthe whole the AS 2885 approach has been well accepted byregulators, with only limited areas where QRA is imposed(and almost invariably done badly, using inappropriatedata and methods, as noted previously).

In comparison with other pipeline codes, one feature of AS2885 that may appear distinctive is the relatively-broaddiscretion permitted, bordering on the subjective. This is adeliberate strategy, and to date there are no indications thatthere has been any abuse of the flexibility that is permitted.

In fact, to the contrary, it appears that most pipelineengineers are quite conservative people who like to haverules and who are keen to be seen to be complying with theStandard. Hence the Standard is generally being appliedconservatively. The review through the SMS workshop, thestate technical regulators and (in some states) independentdesign validation go a long way to ensuring that thepermitted discretion is properly applied.

There are some minor concerns that the requirements ofthe SMS process are not always well understood, but this islikely to fade as time passes and the industry becomesincreasingly familiar with the new requirements.

Having said all that, the new requirements for high-consequence areas have not yet been seriously tested incases where there has been very extensive urbanencroachment over pipelines built for rural conditions(minimum wall thickness, minimum cover, but now withhouses within metres of the pipeline for many kilometres).At least one such SMS review is imminent at the time ofwriting, and the outcomes will be observed with interest.

Overall, the Australian pipeline industry appears to besatisfied with the AS 2885 approach to pipeline safety andrisk management. It works well for us in allowing both thesafety and costs of pipelines to be optimized.

References1. AS 2885.1-2007 Pipelines - Gas and liquid petroleum. Part

1: Design & Construction. Standards Australia, 20072. AS 2885.3-2001 Pipelines - Gas and liquid petroleum. Part

3: Operation and maintenance. Standards Australia, 2001

SAMPLE C

OPY


Appendix 1: AS 2885.1-2007 risk matrix (adapted).

C O N S E Q U E N C E S

sessalc

ytireves

laci

py

Tci

hp

ortsat

aC

roj

aM

ereve

Sr

oni

Mel

bigil

geN

elp

oeP

seitilatafelpitl

uM

.tluser

ro,seitilatafwe

Felpoep

lareves-efil

htiw

gni

netaerht

seiruj

ni

ssenlli

royr

ujnI

latipsoh

gniri

uqert

nemtaert

seiruj

nItsrif

gniri

uqert

nemtaert

dia

tcapmi

lami

niM

ytefas&

htlaeh

no

ylp

pu

Smret

gno

Lfo

noitpurret

niylpp

us

degnolor

Pro

noitpurret

nimret-g

nolfo

noitcirtserylpp

us

mrettro

hS

ronoitp

urretni

degnolorp

fonoitcirtserylpp

us

mrettro

hS

ronoitp

urretni

fonoitcirtser

tub

ylppus

tem

llaftrohs

rehto

morfsecr

uos

on;tcap

mio

Nfo

noitcirtserylpp

use

nilepip

tne

mn

oriv

nE

:E

TO

Nt

nacifingi

Slat

nem

norivne

yam

secne

uqesnoc

snoitacol

nir

uccoera

hcih

wlla

msylevitaler

.detalosid

na

stceffE

,daerpsediw

foytilibaiv

ros

metsysoce,detceffa

seicepsroja

mt

nena

mrepseg

nahc

etis-fforoja

M-g

nolro

tcapmi

erevesmret

rostceffe

noitacifitcertl

uciffid

)ah

1<(

desilacoL

mret-trohs

&,stceffe

)ry2

<(deifitcer

ylisae

yrevtceff

Edesilacol

dna

)ah

1.0<(

mrettro

hsyrev

lami

nim,)skee

w(noitacifitcer

F R E Q U E N C Y

tne

uqer

Fr

uccoot

detcepxE

roraey

repec

noero

me

mertx

Ee

mertx

Eh

giH

etai

demret

nIw

oL

la

nois

accO

rucco

yaM

yllanoisacco

emert

xE

hgi

Het

aide

mretnI

wo

Lw

oL

ylekil

nU

rucco

otylekil

nU

elbissopt

ubh

giH

hgi

Het

aide

mretnI

wo

Lel

bigil

geN

eto

meR

detapicitna

toN

enilepip

siht

rofnoitacol

siht

tah

giH

etai

demret

nIw

oL

elbi

gilge

Nel

bigil

geN

lacite

hto

py

H

yllaciteroeh

Treve

nt

ubelbissop

ralimis

ano

enilepip

etai

demret

nIw

oL

elbi

gilge

Nel

bigil

geN

elbi

gilge

N

SAMPLE C

OPY


Appendix 1 (continued): AS 2885.1-2007 risk matrix (adapted).:s

noitc

at

neme

ga

na

mksi

R

:emert

xE

.rewol

roetaide

mretnI

otdec

udersi

ssalcksir

eht

taht

erus

neot

secne

uqesnoc

eht

royc

neuqerf

eht,taer

hte

htyfido

M.yletaide

mmi

decuder

ebts

um

ksire

hte

nilepipecivres-

nina

roF

:h

giH

.rewol

roetaide

mretnI

otdec

udersi

ssalcksir

eht

taht

erus

neot

secne

uqesnoc

eht

royc

neuqerf

eht,taer

hte

htyfido

Mero

mto

nfo

elacs-emit

ani

htiw

yllacipyt,elbissopsa

noossa

decuder

ebts

um

ksire

hte

nilepipecivres-

nina

roF

.skeew

wefa

naht

:etai

demret

nI

;noita

mitseksir

eht

yfitna

uq,elbissopere

hw,d

nayfirev

otsessecorp

noitaulave

ksird

nanoitacifit

neditaer

httaepe

R,etaide

mretnI

ebot

demrif

nocsi

ssalcksir

eht

ereh

W.

noitamitse

eht

foyt

niatrecn

ud

naycar

uccae

hte

nimreted

.elbigilgeN

rowo

Lot

decuder

sissalc

ksire

hter

usne

otec

neuqes

noce

htro

ycne

uqerfe

ht,taerht

eht

yfidom

ecuder

ro,staerht

evomer

otnekat

eblla

hsnoitca

elbigilgeN

rowo

Lot

decuder

ebto

nnac

ssalcksir

eht

ereh

We

hte

nilepipecivres-

nina

roF

.P

RA

LA

otksir

eht

ecuder

otsa

ossec

neuqes

nocfo

ytirevesec

uderro

seicne

uqerfto

nfo

elacs-emit

ani

htiw

yllacipyt,elbissopsa

noossa

detelpmoc

ebts

um

PR

AL

Aro

elbigilgeN,

woL

otnoitc

uder.s

htno

mwef

ana

htero

m

:w

oL

eht

tceffadl

uochci

hw

segna

hcroti

nom

otd

naec

nerrucco

tneverp

ottaer

hte

htrof

nalpt

nemega

nam

eht

eni

mreteD

.noitacifissalc

SAMPLE C

OPY


THIS NEW INTERNATIONAL CONFERENCE and its accompanying courses and exhibition will cover a wide range of issues concerning pipeline rehabilitation, ranging from the initial stages of evaluation of a pipeline’s condition to the steps required to undertake rehabilitation of the structure to ensure its continued fitness-for purpose and prolong its economic lifetime. The event is being planned to discuss the latest developments in the industry, to showcase some of the industry’s latest achievements, and to provide an unmatched opportunity for both networking and learning.

ORGANIZING COMMITTEEDr Michael Beller,

BJ Lowe,

Sid Taylor

John Tiratsoo,

CALL FOR PAPERS

ORGANIZED BY

ABSTRACT SUBMISSION

SCHEDULE

SUPPORTED BY

rehabConf-a4.indd 1 3/6/2009 9:33:12 AM

SAMPLE C

OPY


SAMPLE C

OPY


Nomenclature

Dt1: pipeline service life under original operatingconditions [years]

Dt2: pipeline service life under posterior operatingconditions [years]

Dtc: coating degradation lag [years]

Dtf: forecasting lag [years]

Dts: pipeline service life [years]

sdi: forecast dimension standard deviation [mm]

sRi

: local depth corrosion rate standard deviation [mm]

d: maximum metal-loss depth [mm]D

f: maximum metal-loss forecast dimension [mm]

Dj: pig-reported dimension (depth, width and length)

[mm]d

j: pig-reported defect depth [mm]

Epig

: tool measurement error [mm] (at 80% confidencelevel)

Fh: service conditions linearization factor

H: defect odometer [m]l: maximum defect length [mm]lf: defect forecast length [mm]

LSEGi

: local segment length [m]N: total number of active corrosion sitesn: vicinity parameterR

i: individual defect dimension corrosion rate [mm/

year]R

Di: local defect dimension corrosion rate [mm/year]

RDij

: individual corrosion rate at a nearby defect [mm/year]

s: scoring factor for service condition changesw: maximum defect width [mm]w

f: defect forecast width [mm]

FOR OVER ONE hundred years pipelines have beenused to transport hydrocarbons from their distant

location to refineries and onwards to consumers. Manymajor world markets nowadays depend upon thisincreasingly-ageing pipeline infrastructure to supply mostof their energy demands. It is unfortunate that ageingadversely affects a pipeline’s integrity, and it can suffer from

A practical approach in pipelinecorrosion modelling: Part 1 –Long-term integrity forecasting

by Dr Érika S M Nicoletti* and Ricardo Dias de SouzaPetrobras Transporte SA, Rio de Janeiro, RJ, Brazil

NOWADAYS, MANY MAJOR MARKETS worldwide depend upon an increasingly-ageing pipelineinfrastructure to supply most energy demands. As corrosion damage accumulation is usually

expected under typical pipeline service conditions, forecast metal-loss growth over time is a key elementin their integrity management; but there is little industrial guidance on this issue. The current work has beenundertaken aiming to provide a corrosion rate model by means of straightforward stochastic treatmentof metal-loss ILI data. This first part will present a model framework regarding long-term scenarios andremaining-life predictions, based on a cost-effective pecuniary threshold for the system’s future remedialactions. The concept of local activity breaks new ground by merging two traditional approaches: theindividual defect and the pipeline segment corrosion growth rates. The model’s underlying assumptions aredetailed, together with its mathematical framework; an empirical balance has been established betweenover- and under-conservative premises, and the accuracy of the results has been considered suitable forforecasting intervals of up to 30 years. The technique provides powerful information with no need to carryout any further expensive and/or laborious analyses: the whole algorithm could be easily put into practiceusing commercial mathematical packages. In order to illustrate the model’s applicability, four case studieswill be presented.

*Author’s contact details:tel: +55 21 3211 7264email: [email protected]

SAMPLE C

OPY


many types of damage under typical service conditions.

Corrosion has historically been the greatest time-dependentthreat to pipeline integrity. The process itself reduces thelocal metal cross-section, affecting the remaining strengthand, consequently, reducing the pipeline’s containmentcapacity in the area of the damage.

Operators can quantify corrosion in their systems throughperiodic metal-loss in-line inspections (ILI). Subsequently,the system’s fitness-for-purpose can be assessed at eachpoint by carrying out damage tolerance analysis using, asinput, ILI data and required service conditions [1, 2].

However, given that pig inspections show only the state ofstatic damage at the time of the inspection, integrityforecasting must take into account corrosion growthestimates. Furthermore, although the phenomenon ofcorrosion is widely known, a plethora of factors impact theprocess kinetics along a pipeline’s length, and the inherentrandomness generally associated with real field conditions,makes its mechanistic modelling a complex task. Thisrequires highly-skilled work, the difficulties of which areoften compounded by an inconvenient lack of historicaldata concerning many of the process control parameters.Thus, it has become common practice to adopt an empiricalapproach, mostly based on worst-case scenarios(recommended practices and/or historical data) [3, 4].However, such procedures usually give rise to highly-inaccurate forecasts, particularly when dealing with long-term scenarios.

Indeed, the US’ Office of Pipeline Safety estimated that theability to accurately forecast corrosion rates could saveAmerican pipeline companies more than US$ 100 millionper year through reduced maintenance costs and accidentavoidance [5].

Fortunately, ILI metal-loss mapping reflects either unknownservice condition variances and/or local electrochemicalmechanism abnormalities, providing a good background,insofar as processing past behaviour is concerned, forcorrosion-rate inferences.

The current work aims to develop a simplified methodologyto allow reasonably-accurate pipeline-integrity forecasts,chiefly by using ILI data. Two basic algorithms have beenconstructed: the first, which is presented in this paper, hasbeen directed to long-term scenarios. The second part ofthe methodology has targets short-term predictions, andwill be published in the second art of the paper [in the June,2009, issue of the Journal of Pipeline Engineering].

The main differences between the algorithms result fromtheir diverse application expectations. Short-term scenariosare usually applied in order to define reinspection intervalsand rehabilitation scopes. As operational pipeline safetyand reliability often depends on the result, conservativeapproaches have always been preferable. On the other

hand, long-term scenario predictions are typically associatedwith the system’s economic viability forecasting (andestimating its remaining life); such analyses are usuallybetter served with accurate modelling.

Despite both algorithms being developed with the aim ofincorporating them into a company’s proprietary defect-assessment software [6], they can also be easily implementedusing any common commercial mathematical package.Accordingly, both are presented in only their most simplisticinterpretations. The underlying assumptions of the models,and the descriptive formulations, will be described anddiscussed. Real cases studies will then be presented toillustrate the methodology anticipated results and overallperformance, before final conclusions are given.

Theoretical backgroundThe corrosion process is irreversible: once it takes place,metal-loss damage at a particular site can only either grow,or remain the same over time, the latter being a sign of siteinactivity (and possible repassivation).

As time goes by, it is expected that new active sites will arise,while some of the existing ones will cease growing.Additionally, time-dependent defect enlargements usuallyslow down over time although, as a general rule,deterministic approaches treat those processes as linear [7,8]. Valor et al. [9] suggest the following should be taken intoaccount:

• the slowing down effect: 0-10% of reduction in pastcorrosion rates

• the cessation of growth effect: 0-20% of the numberof sites nucleated per year

• new defect nucleation: 0-65% of the number of sitesnucleated per year

Conversely, probabilistic models often use the followingdistributions in order to represent:

• nucleation time – exponential and Weibull [9, 10]• number of sites nucleated over time: Poisson [9]• growth rate: gamma, log-normal, or extreme-value

distributions [6, 10, 11]

The Bayesian approach and the Markovian process havealso been widely used in probabilistic framework modelling[12-15].

Given that metal-loss measurements reflect operationalcondition variations and the overall randomness of thecorrosion process itself, the proper treatment of ILI datacan lead to reasonable estimates of past behaviour [16-18].

In order to determine the corrosion rate, damage must bequantified at two different points in time. However, inorder to avoid the usual laborious defect-matching

SAMPLE C

OPY


procedures [5, 19], the current model has been adapted touse a single ILI data set. Also, the model has been constructedunder the premise of a linear relationship existing with theprocess’s past behaviour. Its breakthrough came with theassumption that only adjacent metal loss represents thecorrosion activity at each point, which will be describedfurther below.

The local corrosion activity principle

The authors consider it a rational assumption that allmetal-loss located in close vicinity and on the same side ofthe pipe wall (external/internal) is under similar conditionsof corrosion attack. If, regarding the variations in corrosionactivity along a pipeline’s length, it is expected that futureservice conditions remain similar, then future corrosiongrowth can be predicted based on the metal-loss anomalypopulation located in the defect neighbourhood.

To define the range of each defect’s environment, a vicinityparameter must be empirically determined, using therelationship expressed in Equn 1. Each defect will also haveits associated characteristic length, as defined by Equn 21.

2n+1 > = 7 (1)

L H HSi i n i n= −+ − (2)

Figure 1 illustrates the principle in a pipe section from casestudy 4. All the anomalies displayed are internal metal-loss,and channelling can be clearly noted. Two anomalies havebeen arbitrarily chosen to exemplify the neighbourhood’sdelimitation mechanism: for the purpose of illustration,the lower recommended value for the vicinity parameterhas been used in the figure (n = 3). Note that only axialproximity is taken into account: n anomalies immediatelyup- or downstream are considered as belonging to eachdefect’s local population.

A number of additional simplistic assumptions have beenmade, and a general outline of them will be given in thefollowing paragraphs.

• Process characterization: irreversible, evolving at aconstant rate, and at discrete time intervals.

• Defect population: ILI reported metal-loss anomaliestrimmed, based on the empirical criterion definedin Equn 3:

DE

jt≥ 2

1 28.(3)

where Et represents tool the measurement error and

– eventually – measurement bias, with a confidencelevel of 80%. The mathematical framework to bepresented is independently applied to the anomalypopulations located on the external and internalsurfaces. New defect generation, as well as the rateof cessation of defect growth, are considered to benegligible.

• Nucleation time: defect populations are assumed tobe instantaneously nucleated at the first exposure tocorrosive conditions.

• Defect growth: determined based on the pastbehaviour of local corrosion activity. The details ofthis premise regarding external and internal surfacecorrosion are described below.

• Coating protection effectiveness: the coatingcondition is considered to be perfect at the time ofpipeline commissioning. All pipeline coatingholidays are considered to be instantaneouslygenerated after a specific coating degradation lag.The protection effectiveness is assumed to be 0% atall active sites of external corrosion, and 100% inholiday-free regions. Water and air permeationtime dependency is not taken into consideration.

• Coating degradation lag: must be empirically definedbased on coating data history and engineering bestjudgment.

• Cathodic protection: is assumed to remain in asteady-state condition throughout the entire servicelife of the pipeline.

• Probability density functions (PDFs): defect depthdimensions and corrosion rates are described byGaussian PDFs. It is worth noting that, given thateach defect’s corrosion rate is represented in termsof a local average, there is a normalizing effect on theoverall depth corrosion rate data set2.

Mathematical framework

Corrosion rates PDFs

The probability density functions should be individuallydefined, taking account of the damage accumulated in eachdefect neighbourhood, according to the previously-outlinedprinciple of local corrosion activity3. If a significant changein the system’s operating conditions takes place after any

1. The parameter n should be adjusted in order to obtain an averagesegment length not exceeding 1-2 km.

2. Pipeline geometry, material features, and the axial and circumferentialcorrosion rates, have only been considered deterministically, as will bethe allowable damage as a consequence.

3. Clustering criteria should preferably be applied after a futuremorphology forecast, not before.

SAMPLE C

OPY


particular event, then a factor Fh is defined accordingly,

using Equn 4; otherwise, Fh is assumed as 14.

Ft s t

ths

= +Δ ΔΔ

1 2(4)

Internal defects

Individual defect growths (radial, axial, and circumferential)are determined by means of Equn 5a. The subsequentapplication of the local corrosion activity principle leads tothe determination of the corrosion rate average for thedefect population located in the adjoined region by usingEqun 5b, while the dispersion is obtained from Equn 5c.Furthermore, the characteristic length associated with eachdefect neighbourhood (L

seg) can also be defined, as previously

discussed.

Hence, each flaw on a pipe’s inside surface will have onesingle PDF representing its depth corrosion growth rate,while axial and circumferential rates, as well as itsneighbourhood characteristic length, are deterministicallydefined.

RD

tii

s

=Δ (5a)

R F

R

nLi h

jj i n

j i n

=+

= −

= +

∑( )2 1

(5b)

σLi

Li jj i n

j i n

R R

n=

−( )= −

= +

∑ 2

2(5c)

External defects

It is proposed that pipeline coating holidays are consideredstationary. Thus, circumferential and axial growth rates areassumed as zero at all active sites located on the pipeline’sexternal surface. Equation 5d represents the depth growthrate, considering the lag in coating degradation.

Rd

t tdij

s c

=−Δ Δ (5d)

Equations 5b and 5c must therefore also be applied inorder to characterize the defect’s depth corrosion rate PDF,

while Equn 2 should be used to define its neighbourhoodcharacteristic length.

Future defect morphology

The average dimensions of future defects can be calculatedfrom Equn 6a, and Equn 6b is used to determine theassociated dispersion.

D D R tf i Li f= . .Δ (6a)

σ σDf f Litt

E= ( ) + ⎛⎝⎜

⎞⎠⎟

Δ 22

1 28.(6b)

Damage tolerance

There are a number of metal-loss assessment criteria thatcan be used to determine damage tolerance. The mostsimplistic and widely known is ASME B31.G, which onlytakes into account axially-oriented corrosion defectssubmitted to internal pressure loading. Depending on theparticular system’s damage characteristics (which can includecircumferential- or even helically-oriented defects), or theexistence of axial loads (such as those geotechnically orthermally induced), an appropriate criterion should bechosen to deterministically find out the maximum allowabledefect depth as a function of its forecast width and length,according to Equn 7.

d f l wa f f= ( , ) (7)

Probability of exceedance

The future defect depth (df) shall not exceed its allowable

depth (da) [22, 23], as represented by the limit-state function

in Equn8:

d df a− < 0 (8)

In the current approach, df is characterized by a normal

distribution, while da is deterministic. This means that the

probability of a pipeline exceeding the limit-state conditionat each defect can be determined as the area on the right-hand side of the allowable depth under the d

f PDF (see

Fig.2)5.

Economic remediation rate

The economic remediation rate which provides cost-effectiveoperation must be ascertained by a pipeline’s own operator,considering each case individually. It is outside the scope of

4. The scoring factor for changes in service conditions (s) should bedetermined based on historical data (coupons/probes, comparison ofmultiple ILI data or computation simulations) and engineering bestjudgment [20]. 5. Most commercial packages have standard functions to perform this.

SAMPLE C

OPY


this work to accomplish a full perspective into problem, butsome of the factors that must be taken into account in suchan analysis include:

• technical and economic viability of alternativepipeline systems, or other modes of transportation

• the ratio between the cost of a new pipeline and theestimated maintenance costs of the existing one

• the impact of a possible delivery shortage on thelocal economy

• current, and possible future, economic scenarios

Restriction on the model’s applicability

As the whole model is based on averaging the behaviour ofthe local corrosion process, its application is notrecommended to systems where hot-spot mechanisms (suchas stray current, under-coat corrosion, etc.) are significantfeatures.

Case studiesIn order to illustrate the model’s application, four casestudies have been chosen, the input data for which issummarized in Table 1. A brief introduction is given foreach, before the model results and overall performance arediscussed.

• Pipeline 1: an onshore pipeline carrying dry gassince its operation began. Accumulated corrosiondamage was slight on both the external and internalpipeline surfaces.

• Pipeline 2: an onshore gas pipeline that has beenused to transport both wet and sour products.Accumulated internal corrosion is severe although,on the other hand, almost no external metal-lossindications have been reported as a result of thedryness of the of region crossed by this pipeline, inthe NE of Brazil.

• Pipeline 3: a trunk line responsible for transportingall of one refinery’s crude oil supply. During itsoperational life, it endured production waterpumped through recurrently, together with somehigh-BSW content product. Long shut-down periodswere also a regular occurrence. Internal corrosiondamage is quite severe and channelling damage isgeneral. In order to meet an increase in demand, anincrease in flow capacity was required. The resultantnew service conditions were simulated by the worst-case hydraulic scenarios, and the maximumoperational pressure profile was defined accordingly.

• Pipeline 4: an onshore line which has been used totransport naphtha and crude oil, the latter usually

1enilepiP 2enilepiP 3enilepiP 4enilepiP

)ni(retemaiD 61 41 22 61

muminiM)mm(ssenkciht 7.8 2.8 3.6 9.7

lairetamepiP 06X 56X 64/04X 53X

)mk(htgneL 481 822 89 89

)ry(efilecivreS 62 63 23 14

)mcqs/gk(POAM 001 79 *65-12 *14-13

.egnardetalumisciluardyhoiranecsesactsrow*Table 1. Constructionand operational data.

Fig.1. Local corrosion activity.

SAMPLE C

OPY


with a high BSW content. Again, production watertransportation was a frequent occurrence togetherwith extensive shutdown periods. The whole pipelinehas bad channelling damage, as shown in Fig.1.

Results and discussion

The model’s output data from the four case studies aresummarized in Table 2, while Fig.3 shows the overallnormalized local corrosion activity. Figures 4 and 5 presentthe expected probability of exceedance for safe operations(at the required levels) without repair to the 200 mostcritical metal-loss areas in each case study, for the next 20and 30 years, respectively. Pipeline 2 was not analysed forexternal corrosion, due the lack of significant indicationson its external surface. In view of a desirable operational

POE threshold range of 10-4-10-5, and the economicremediation rate specified for each case, it can be concludedthat pipeline 3 could be safely operated for almost 30 years,while pipeline 4 would be cost-effectively operational forapproximately 20 years at most.

Conversely, with the exception of pipeline 1, Figs 6 and 7show that internal corrosion developing over 30 yearswould be a direct threat. Pipeline 4 is not expected tomaintain its present use for long, while the operationalreliability of pipelines 2 and 3 will not be cost-effective formore than 20 and 10 years, respectively.

As a result of applying these forecasts, the company’s boardof directors has undertaken the following:

Fig.2. The probabilistic limit-statefunction.

1enilepiP 2enilepiP 3enilepiP 4enilepiP

EXTERNAL

deretlif-noitalupoP 312 - 768 222

)n(retemarapytiniciV 5 - 5 5

naissuaGsetarnoisorroclacoLsretemarapnoitubirtsid

]raey/mm[600.0-80.0 - 400.0-350.0 600.0-08.0

etarnoitaidemeryrainucePsriapergnitaocevitceffe-tsoC

rebmun05 - 001 001

rednudetcepxEefilgniniameR]sraey[snoitidnoccirotsih 03> - 03 02

INTERNAL

deretlif-noitalupoP 733 073,01 523,05 04232

)n(retemarapytiniciV 5 01 02 51

naissuaGsetarnoisorroclacoLsretemarapnoitubirtsid 800.0-560.0 600.0-080.0 300.0-540.0 700.0-180.0

etarnoitaidemeryrainucePsriapergnitaocevitceffe-tsoC

rebmun04 08 08 05

rednudesaBefilgniniameR]sraey[snoitidnoccirotsih 03 02-51 01 5

Table 2. Modellingparameters and output.

SAMPLE C

OPY


�

��

��

��

��

��

��

mm/yearLocal IndividualFig.3. Internal corrosion rate

histogram for pipeline 3.

Fig.4. 20-year POE forecast of theworst external metal-loss anomalies.

Fig.5. 30-year POE forecast ofthe worst external metal-lossanomalies.

SAMPLE C

OPY


• pipeline 4 was converted to specified dieseltransportation;

• a major rehabilitation project is being carried outon pipeline 3, together with several mitigating actions(including a new strategy regarding productionwater);

• a brand new pipeline is under construction toreplace pipeline 2 (mainly in order to supply thelocal market’s forecast rising demand) while analternative use for pipeline 2 is being studied.

ConclusionsNowadays, new onshore pipeline systems must be plannedwell in advance. Sometimes almost a decade can pass

between the conceptual design and commissioning stages,mostly as consequence of the complexities concerning thelegal agreements with landowners through whose land thepipeline will be routed, together with the tougher regulationsregarding environmental and operational issues.

Pipeline operators therefore need to forecast their systems’remaining lives with reasonable long-term accuracy. Despitecorrosion being the major time-dependent threat to ageingpipeline systems, there is little available guidance concerningcorrosion modelling for real pipeline service conditions,and the subject remains controversial.

The current work has been developed to support theoperator’s long-term strategic planning, by providing astraightforward stochastic model to forecast the remaining

Fig.6. 10-year POE forecast of theworst internal metal-loss anomalies.

Fig.7. 20-year POE forecast of theworst internal metal-loss anomalies.

SAMPLE C

OPY


life of corroded pipelines. As input data, the newly-developedmodel requires pipeline geometry and material properties,the worst-case scenario for operational pressure, a good-quality set of metal-loss ILI data, and also the economicthreshold for the system’s future remediation. Thepioneering concept of local corrosion activity wasintroduced, and the underlying simplistic assumptions aredetailed together with the entire mathematical framework.The definitions and roles of the empirical parameters havealso been described.

A balance has been established between over- and under-conservative assumptions, and the model had beenconsidered suitable for forecasting periods of up to 30years. Its algorithm is set out in detail and it can easily beimplemented using standard commercial mathematicalpackages.

The technique provides powerful information with noneed for further expensive or laborious analyses. The use ofthe model has already proved to be particularly relevant toforecasting critical problems long before they present anyreal threat. The model has also been used to give rise toactive mitigation planning, such as a review of inhibitorstrategy and definition of the scope of coating rehabilitationprojects.

Additionally, if more-sophisticated mathematical packagesare available, the model could be easily adapted toincorporate further refinements, incuding:

• non-Gaussian behaviour (for which an automaticbest-fitting-distribution tool is required)

• full limit-state approach: pipeline geometry andmaterial properties could also be consideredprobabilistically (if convolution integrals can beeasily solved) [2, 10, 24]

• any specifics of a system’s history could be takeninto consideration by making the necessaryadjustments to the model’s premises andassumptions.

AcknowledgmentsThe authors thank Petrobras Transporte SA for permissionto publish this paper, and their colleagues Dr SérgioCunha, Carlos Alexandre Martins, and João Hipólito deLima Oliver for many enlightening discussions andcontributions.

References1. B.Gu, R.Kania, and M.Gao, 2004. Probabilistic based

corrosion assessment for pipeline integrity. Corrosion 2004,NACE International, New Orleans.

2. R.Bea et al., 2003. Reliability based fitness-for-serviceassessment of corrosion defects using different burst pressurepredictors and different inspection techniques. 22ndInternational Conference on Onshore Mechanics and ArcticEngineering, June 8-13, Cancun.

3. NACE RP-0775. Preparation, installation, analysis andinterpretation of corrosion coupons in oilfield operations.

4. NACE SP0502, 2008. Pipeline external corrosion directassessment methodology.

5. J.M.Race, S.J.Dawson, L.Stanley, and S.Kariyawasam, 2006.Predicting corrosion rates for onshore oil and gas pipelines.International Pipeline Conference, Calgary.

6. S.B.Cunha, A.P.F.Souza, E.S.M.Nicoletti, and L.D.Aguiar,2006. A risk-based inspection methodology to optimize in-line inspection programs. The Journal of Pipeline Integrity,pp133-144.

7. M.Ahammed, 1998. Probabilistic estimation of remaininglife of a pipeline in the presence of active corrosion defects.International Journal of Pressure Vessels and Piping, 75, pp321-329

8. S.L.Fenyvesi, H.Lu, and T.R.Jack, 2004. Prediction ofcorrosion defect growth on operating pipeline. Proc.International Pipeline Conference, October 4 - 8, Calgary,Canada.

9. A.Valor, F.Caleyo, L.Alfonso, D.Rivas, and J.M.Hallen, 2007.Stochastic modeling of pitting corrosion: a new model forinitiation and growth of multiple corrosion pits. CorrosionScience, 49, pp559–579.

10. A.Ainouche, 2006. Future integrity management strategy ofa gas pipeline using Bayesian risk analysis. 23rd World GasConference, Amsterdam.

11. P.J.Laycock and P.A.Scarf. Exceedances, extremes,extrapolation and order statistics for pits, pitting and otherlocalized corrosion phenomena. Corrosion Science, 35. no 1-4,pp135-145, 193.

12. J.L.Alamilla and E.Sosa, 2008. Stochastic modelling ofcorrosion damage propagation in active sites from fieldinspection data. Corrosion Science, 50, pp1811–1819.

13. J.L.Alamilla, D.De Leon, and O.Flores, 2005. Reliabilitybased integrity assessment of steel pipelines under corrosion.Corrosion Engineering, Science and Technology, 40, 1.

14. S.A.Timashev, 2003. Updating pipeline remaining lifethrough in-line inspection. International Pipeline PiggingConference, Houston.

15. S.A.Timashev et al., 2008. Markov description of corrosiondefect growth and its application to reliability based inspectionand maintenance of pipelines. Proc. 7th International PipelineConference, Calgary.

16. G.Desjardins, 2002. Optimized pipeline repair and inspectionplanning using in-line inspection data. Pipeline Pigging,Integrity Assessment, and Repair Conference, Houston.

17. B.Gu, R.Kania, S.Sharma, and M.Gao, 2002. Approach toassessment of corrosion growth in pipelines. 4th InternationalPipeline Conference, Calgary.

18. G.Desjardins, 2001. Predicting corrosion rates and futurecorrosion severity from in-line inspection data. MaterialsPerformance, August, 40,8.

19. J.M.Race et al., 2007. Development of a predictive model forpipeline external corrosion rates. Journal of Pipeline Engineering,6, pp15-29.

20. R.B.Eckert and B.Cookingham, 2002. Advanced proceduresfor analysis of coupons used for evaluating and monitoringinternal corrosion. CC Technolgies, Doublin, OH, USA.

21. ASME B 31G. Manual for determining the remaining strengthof corroded pipelines.

SAMPLE C

OPY


22. H.Plummer and J.M.Race, 2003. Determining pipelinecorrosion growth rates. Corrosion Management, April.

23. F.Caleyo et al., 2002. A study on the reliability assessmentmethodology for pipelines with active corrosion defects.International Journal of Pressure Vessels and Piping, 79, pp77-86.

24. G.Pognonec, 2008. Predictive assessment of externalcorrosion on transmission pipelines. 7th InternationalPipeline Conference, Calgary.

TITLES OF

INTEREST

FROM

ASME PRESS

Technical Writing A–Z: A CommonsenseGuide to Engineering Reports and Theses,British English Editionby Trevor M. Young

Topics include: format and content of reports and theses;copyright and plagiarism; print and Internet reference cita-tion; abbreviations; units and conversion factors; significantfigures; mathematical notation and equations; writing stylesand conventions; frequently confused words; grammaticalerrors and punctuation; commonsense advice on issuessuch as getting started and holding the reader’s attention.

2005 256 pp. Softcover ISBN: 0-7918-0237-XOrder No. 80237X $29 (list)/$23 (ASME member)Order sets of 10 copies at a special price. Order No. 80236S $199

American Edition:2005 256 pp. Softcover ISBN: 0-7918-0236-1Order No. 802361 $29 (list)/$23 (ASME member)Order sets of 10 copies at a special price. Order No. 80236S $199

Pipeline Operation and Maintenance: A Practical Approachby M. Mohitpour, J. Szabo, and T. Van Hardeveld

Covering pipeline metering, pumping, and compression, thebook covers day-to-day concerns of the operators andmaintainers of the vast network of pipelines and associatedequipment and facilities that deliver hydrocarbons andother products. It is a useful reference for veterans and atraining tool for novices.

2004 600 pp. Hardcover ISBN: 0-7918-0232-9Order No. 802329 $125 (list)/$99 (ASME member)

Mister Mech Mentor, Volume I:Hydraulics, Pipe Flow, Industrial HVAC &Utility Systemsby James A. Wingate

Gain practical knowledge from frank, colorful cases andlearn to solve mechanical problems related to hydraulics,pipe flow, and industrial HVAC and utility systems withthese organized solutions to the problems involving: waterand steam hammer phenomena; gravity flow of liquids inpipes; siphon seals and water legs; regulating steam pres-sure drop; industrial risk insurers’ fuel gas burner pipingvalve train; controlling differential air pressure of a roomwith respect to its surroundings; water chiller decoupledprimary-secondary loops; pressure drop calculations ofincompressible fluid flow in piping and ducts; water chillersin turndown; hydraulic loops; radiation heat transfer; andthermal insulation.

2005 160 pp. Softcover ISBN: 0-7918-0235-3Order No. 802353 $45 (list)/$36 (ASME member)

Pipeline Design and Construction:A Practical Approach, Second Editionby M. Mohitpour, H. Golshan and A. Murray

This second edition includes updated codes and standardsinformation, solutions to technical problems, additional ref-erences, and clarifications to the text. It offers straightfor-ward, practical techniques for pipeline design and con-struction, making it an ideal professional reference, trainingtool, or comprehensive text.

2003 700 pp. Hardcover ISBN: 0-7918-0202-7Order No. 802027 $110 (list)/$88 (ASME member)

Order Now! North America: www.asme.org • Europe: www.ihsatp.com

SAMPLE C

OPY


THE US Energy Policy Act of 2005 (amended in 2007)established a nationwide renewable fuels standard

starting from 15 billion litres (4 billion gallons) of allbiofuels in 2006 to 136 billion litres (36 billion gallons) in2022. Ethanol will constitute almost 90% of this renewablefuel. As the price of gasoline has oscillated, the economicviability of alternative fuels also is undergoing re-evaluation.However, the long-term need for biofuels, both for reducinggasoline dependence and carbon footprint, is undeniable.Biofuels can be broadly classified into several generationsdepending on their status of commercial productionreadiness. The ethanol produced from sugars (such as sugarcane, beets, and grapes) and starch (corn, wheat) isconsidered to be a first-generation biofuel. The productiontechniques for these feedstocks are well established, andworld-wide commercial production of ethanol from thesesources was approximately 50 billion litres (13.2 billiongallons) in 2008.

The ethanol produced from grasses (such as switch grass),

agricultural/food processing wastes, and other cellulosicmaterials requires enzyme treatments that will requiresignificant additional process development. The ethanol(or alcohol) from these sources is referred to as second-generation ethanol (and also as cellulosic ethanol). Afurther development in alcohol production will be the useof transgenic materials (low-lignin trees), which will requireadvanced enzymatic treatments, and the resulting alcoholswill be the third-generation cellulosic fuel. A somewhatsimilar categorization exists for biodiesel production.

The ethanol supply chain is illustrated in Fig.1. Ethanol isproduced in bio-refineries and must be transported toterminals, where it is blended with gasoline to produce themost commonly used blends E-10 (10% ethanol) and E-85(85% ethanol). As shown in Fig.1, rail and truck arecurrently the predominant means of transporting ethanolin North America. Brazil has a history of transportingethanol via pipelines and ships. In terms of the volumestransportable by the different modes shown in Fig.1, onebarge load is roughly equivalent to 15 to 20 rail cars or 60-80 truck loads. In comparison, a 16-in pipeline can transportan equivalent of 15 barges on a daily basis. The number ofnew rail cars constructed has to rise substantially, and newterminals that can accommodate unit train shipments haveto be constructed, to allow rail shipment of the futureanticipated fuel volumes. Barge transport can benefitsubstantially by a pipeline delivery system, while increasing

Author’s contact information:tel: +1 614 761 1214email: [email protected]

This paper was presented at the 21st Pipeline Pigging and IntegrityManagement conference held in Houston on 10-12 February, 2009,organized by Clarion Technical Conferences, Houston, and ScientificSurveys Ltd, Beaconsfield, UK.

Ethanol transportation: status ofresearch, and integritymanagement

by Dr John Beavers1, Patrick Vieth1, and Dr Narasi Sridhar2

1CC Technologies, Inc (a DNV Company), Dublin, OH, USA2 DNV Research and Innovation, Dublin, OH, USA

THE PIPELINE INDUSTRY is undertaking considerable research to determine the best approach tomanage the potential for internal stress-corrosion cracking (SCC) to occur while transporting fuel-

grade ethanol (FGE). Based on the results to date, it appears that FGE meeting the ASTM D 4806specification can cause SCC of carbon steel. The parameters that affect the potential for SCC (oxygen,water, etc.) are understood, and the research is now focused on methods to reduce the likelihood of SCC.The current state of the research is discussed.

SAMPLE C

OPY


the truck transportation poses significant logisticalproblems, including training a much larger number ofdrivers than anticipated to be available in the future. Thus,pipeline transportation is the most cost-effective mode oftransporting large volumes of ethanol.

Figure 2 shows that most of the bio-refineries in the US arelocated close to the middle of the continent, whereas thepopulation centres are along the coasts. Most of the currenthydrocarbon pipelines move products from the Gulf ofMexico region to the east and west coast and the midwest.Thus, new pipelines will be required to transport the fuel-grade ethanol (FGE) from the bio-refineries to thepopulation centres.

BackgroundA survey of published literature and service experience withSCC in FGE was published by the American PetroleumInstitute (API) in 2003 [1]. Documented SCC failures ofequipment in users’ storage and transportation facilitieshave dated back to the early 1990s: the majority of thecracking has been found at locations near welds where theprimary stresses leading to SCC have been residual weldingstresses. No cases of SCC were reported in ethanolmanufacturer facilities, tanker trucks, railroad tanker cars,or barges, or following blending the FGE with gasoline. Alloccurrences of SCC were at the first major hold point (theFGE distribution terminal) or in the subsequent end-usergasoline-blending and -distribution terminals.

An example of SCC in terminal piping is shown in Fig.3:note that the leak is near a girth weld adjacent to a pipingtee.

The API survey did not pinpoint what causes ethanol SCC,but the failure history suggests that the SCC may be relatedto changes in the FGE as it moves through the distributionchain over a period of days, weeks, or months. Theseobservations led to an industry-sponsored researchprogramme to identify the causative factors. A‘Roadmapping Workshop’ held in October, 2007, identifieda number of research gaps in safely transporting FGE viapipelines [2], which were divided into four areas:

(i) ethanol sources and quality,(ii) pipeline operations,(iii)standards, guidelines, and training, and(iv) pipeline integrity.

Ethanol SCC potentially impacts all four of these areas andresearch is continuing or planned to address these gaps. Anumber of factors were identified as contributors to SCC,and SCC-mitigation strategies are being developed. Thispaper summarizes the current state of the research.

Environmental factorsaffecting SCC in FGE

The results of research on chemistry effects on SCC havedemonstrated that FGE that meets applicable API standards(Table 1) is a potent cracking agent in the presence ofoxygen [3, 4]�. Several research programmes have examinedthe effects of the contaminants (such as chloride) or thedenaturant in FGE on SCC behaviour in aerated ethanolsolutions. Studies by Sridhar et al. [5] and Beavers et al. [6]showed that chloride significantly increased thesusceptibility of carbon steels to SCC in ethanol.

Oil Refinery

Blending Terminal

Ethanol plant

Rail (30%)

Truck (67%)

Barge (2%)

Pipeline

Blend

Fig.1. Ethanol supply chain for North America.

SAMPLE C

OPY


Furthermore, the fracture mode changed frompredominantly intergranular to predominantlytransgranular as the chloride concentration increased from0 to 40ppm. Methanol also appeared to increase SCCsusceptibility [5]. The water content of the FGE has alsobeen shown to affect the SCC behaviour: anhydrous ethanolwill not promote SCC [4] and water contents above about4.5% by volume completely inhibit SCC [7]. Between theselimits, water does not appear to have a significant effect onSCC [5, 7].

Factors that have not been shown to have a significantinfluence on SCC in FGE include acidity, one commongeneral corrosion inhibitor, and the denaturant [4, 5]. SCCwas observed in SSR tests over a wide range of pHE andacetic acid concentration. Sridhar et al. [5] showed that onecommon corrosion inhibitor added to FGE to protectagainst automotive corrosion (Octel DCI-11) did not haveany effect on SCC of steel.

A statistically-designed study by Sridhar et al. [5] showedthat dissolved oxygen is the most important factor affectingSCC in FGE: no SCC occurred under any circumstanceswithout the presence of dissolved oxygen. Based on theoxygen concentration, a “critical” potential regime wasidentified for SCC [8], Fig.4, which depends on chlorideconcentration. In the presence of chlorides, SCC extendsto lower corrosion potentials.

At high corrosion potentials in some ethanols, SCC wasnot observed, and the reason for this behaviour is stillunclear. Beavers et al. [6] showed that removing oxygen bychemical, mechanical, or electrochemical methods allresulted in suppression of SCC in slow-strain-rate (SSR)tests in a simulated FGE. Oxygen removal also caused anegative potential shift in the free corrosion potential, asshown in Fig.5. It is well known that dissolved oxygenincreases SCC susceptibility of steel in other non-aqueousenvironments, such as ammonia and methanol. Therefore,

tnemeriuqeR

stimilMTSA)6084DMTSA(

muminiM mumixaM

%.lov,lonahtE 1.29 -

%.lov,lonahteM - 5.0

lm001/gm,mugdehsaw-tnevloS - 0.5

%.lov,tnetnocretaW - 0.1

%.lov,tnetnoctnarutaneD 69.1 67.4

)L/gm(mpp,edirolhccinagronI - )23(04

gk/gm,reppoC - 1.0

)L/gm(%ssam,)HOOC3HCdicacitecasa(ytidicA - )65(700.0

eHp 5.6 0.9Table 1. Specification offuel-grade ethanol.

Fig.2. Ethanol productionlocations in the US.

SAMPLE C

OPY


it is not surprising that dissolved oxygen is a majorcontributor to SCC in FGE.

The earlier research, funded by API, demonstrated thatFGE as well as E-85 (85% ethanol – 15% gasoline) promotedSCC [4]. More recently, Beavers et al. [9] evaluated theeffect of ethanol-gasoline blend ratio on SCC in a researchproject funded by PRCI. The study was performed withnotched SSR specimens of an X-46 linepipe steel in asimulated FGE. No SCC was observed in gasoline or E-10(10% ethanol blend), but SCC susceptibility increasedrapidly with increasing ethanol concentration for E-20 andhigher blends. Surprisingly, E-30 was nearly as susceptibleto SCC as FGE, as shown in Fig.6.

Studies [8, 9] have shown that no two ethanols are createdequally in terms of SCC tendency. Some ethanols do notcause SCC even at high dissolved oxygen levels; otherscause significant SCC. Aging of ethanol samples appears toalter their SCC tendency significantly. The variations inthe ethanol chemistry and their impact on SCC behaviourare the subject of ongoing studies.

Metallurgical factorsaffecting ethanol SCC

Field experience and laboratory testing indicate that severestraining is required for ethanol SCC to occur. SCC ofethanol storage tanks has been observed only in severely-strained areas associated with non-post weld heat-treatedwelds and/or in tanks with design/installation issues [1].For example, floor areas that were not adequately supportedexperienced SCC as a consequence of cyclic loading fromfilling and withdrawal of ethanol. Some of the earliestlaboratory studies of SCC in ethanol were conducted usingU-bend specimens. SCC was not observed in these tests

unless a “bad” welding bead perpendicular to the stressingdirection and an extremely severe bending mode wereincluded. In SSR tests with un-notched specimens, SCCwas observed near the necked region of the specimen [6];notched-SSR tests exhibited SCC at the notch root [9]. Allthese observations suggest that severe plastic deformationand the presence of dynamic plastic strain are necessary forSCC to occur. In more-recent crack-growth tests usingcompact tension specimens, the presence of a cyclic loadingcomponent has been shown to exacerbate SCC [7].

Recent studies reported by Beavers et al. [9] have shown thatthe extent of SCC was not dependent on steel graderanging from X-42 high frequency electric-resistance weldedpipe material to cast steel for pumps. For one grade, theweld area of a double submerged arc weld appeared to beslightly more resistant to SCC than the base metal.

Mitigation of ethanol SCCThe field experience and research results specificallyaddressing ethanol SCC, as well as broader experience withother forms of SCC in the pipeline and other industries,point to potentially-effective methods for mitigation ofethanol SCC. The research on oxygen effects on ethanolSCC clearly demonstrates that, regardless of how theoxygen is removed, SCC can be mitigated. Both mechanicaldeaeration and one chemical oxygen scavenger were shownto be effective. It is probable that the oxygen is absorbed inthe ethanol during the transportation process and it mightbe possible to minimize oxygen contamination, as opposedto removing it once it is already in the ethanol.

True SCC inhibitors also potentially could be effective.Research by Beavers et al. [6] showed that some film-forming amines have an inhibiting effect on ethanol SCC.The identification of the best possible inhibitors or inhibitor

LeakLeak

Fig.3. SCC observed in terminal piping system containing FGE.

SAMPLE C

OPY


-400 0 400 800Corrosion Potential, mV vs. Ag/AgCl EtOH

360

400

440

480

520M

axim

um L

oad,

Kg

Only ClOnly MeOHNo Cl or MeOHLow H2OEtOH-10%GasolineEtOH-15% GasolineWet Milling EtOHDry Milling EtOHReagent EtOH+airE85 Sample 1 DeaeratedE85 Sample 1 AeratedHigh Potential EtOH AeratedReagent EtOH Still AirE85 Sample 1 Still AirE85 Sample 2 Deaerated

SCC

No SCC

With Cl Without Cl

0.0E+00

5.0E-07

1.0E-06

1.5E-06

2.0E-06

-400 -300 -200 -100 0 100 200Average Corrosion Potential in Test, Ag/AgCl EtOH

Cra

ck G

row

th R

ate,

mm

/s

Steel Wool Mechanical Deaeration

No Deaeration

NitrogenDeaeration

Hydrazine

Fig.4. SCC vs corrosion potential indicating a critical potential (dependenton chloride level) below which SCC was not observed [8].

Fig.5. Crack growth rate as a function of average potential forSSR tests in simulated FGE with various deaeration methods [6].

packages, taking into consideration diverse issues such astoxicity, compatibility with combustion engines, cost etc.,requires further research.

The association of SCC in the terminals with residualstresses provides another mitigation avenue for new ethanol

pipelines. Post-weld heat treatment of all welds couldminimize SCC, although the hoop stresses from the internalpressure in transmission pipeline might play a bigger rolein the SCC process. Grit blasting prior to coating has beenshown to play a role in the mitigation of external SCC ofgas transmission pipelines [10]. The compressive residual

SAMPLE C

OPY


stress imparted by the grit blasting process has been shownto effectively overcome the effects of residual stresses andthe tensile hoop stress from internal pressurization. Asimilar process might be effective for the mitigation ofethanol SCC.

SummaryLarge increases in production and transportation volumesof FGE are expected as a result of new energy policiesdictating substantially-higher usage of alternate fuels.Ethanol has been shown to cause SCC of steel in thepresence of dissolved oxygen and chloride. Significantadvances have been made in understanding the variousparameters that can affect SCC and in identifying mitigationstrategies. Going forward, this information will be neededto manage the integrity of ethanol pipelines.

References1. R.D.Kane and J.G.Maldonado, 2003. Stress corrosion

cracking of carbon steel in fuel grade ethanol: review andsurvey. API Technical Report 939-D, American PetroleumInstitute, Washington, DC, September.

2. Energetics, Inc., 2007. Safe and reliable ethanol transportationand storage technology roadmapping workshop, October 25-26, Dublin, Ohio.

-1.0E-06

0.0E+00

1.0E-06

2.0E-06

3.0E-06

4.0E-06

5.0E-06

10 20 30 50 95Ethanol Concentration, %

SCC

Cra

ck G

row

th R

ate,

mm

/s

Fig.6. Crack growth rate as a function of ethanol concentration for X-46 linepipe steel specimens tested in simulated FGE-gasoline blends [9].

3. R.D.Kane, N.Sridhar, M.Brongers, J.A.Beavers, A.K.Agrawal,and L.Klein, 2005. Materials Performance, 44, 12.

4. R.D.Kane, D.Eden, N.Sridhar, J.Maldonado,M.P.H.Brongers, A.K.Agrawal, and J.A.Beavers, 2007. Stresscorrosion cracking of carbon steel in fuel grade ethanol:review, experience survey, field monitoring, and laboratorytesting. API Technical Report 939-D, 2nd Edn, AmericanPetroleum Institute, Washington, DC, May.

5. N.Sridhar, K.Price, J.Buckingham, and J.Dante, 2006.Corrosion, 62, 8, pp687-702.

6. J.A.Beavers, M.P.Brongers, A.K.Agrawal, and F.A.Tallarida,2008. Prevention of internal SCC in ethanol pipelines.NACE, Corrosion 2008 Conference, New Orleans, LA,March, Paper 08153.

7. J.A.Beavers and N.Sridhar, 2008. Unpublished results, PRCISCC Program.

8. J.G.Maldonado and N.Sridhar, 2007. SCC of carbon steel infuel ethanol service: effect of corrosion potential and ethanolprocessing source. Corrosion, Paper 07574, Houston, TX,NACE International.

9. J.A.Beavers, N.Sridhar, and C.Zamarin, 2009. Effects of steelmicrostructure and ethanol-gasoline blend ratio on SCC ofethanol pipelines. NACE, Corrosion 2009 Conference,Atlanta GA, March, Paper 095465.

10. J.A.Beavers, N.G.Thompson, and K.E.W.Coulson, 1993.Effects of surface preparation and coatings on SCCsusceptibility of line pipe: Phase 1 – laboratory studies.Corrosion, NACE Paper 597, New Orleans, LA, March.

SAMPLE C

OPY


DENTS GENERATED IN onshore pipeline are typicallythe result of third-party damage, although rock dents

are certainly a contributor for bottom-side defects. Damageto subsea pipelines typically occurs as the result of impactwith an anchor. After the subsea incident occurs, ROVs(remotely-operated vehicles) are then deployed to surveythe damage, followed by survey efforts to determine if thepipeline has been moved or laterally displaced. If it isbelieved that localized damage has been inflicted, it isessential that the profile of the dented region be determined,and in-line inspection is ideally-suited for collecting thisdata. From a geometry standpoint, the data collectedincludes points measuring radius, circumferentialorientation, and longitudinal position (i.e. R-q-Zcoordinates).

Presented in this paper is a background section that discusses

how to evaluate dents considering previous research effortsand experience, following which is a discussion on how rawILI geometry data are converted into the mesh for evaluationusing the finite-element (FEA) method. FEA is used tocalculate the alternating stresses in the dented region; oncethe stresses due to cyclic pressure are calculated, a fatiguecurve is used to estimate the remaining life for the givendents. Results are presented from previous research onfatigue testing of pipes having plain dents, and the finalsection of the paper provides recommendations for industryin using ILI data to estimate the remaining life of damagedpipelines, and integrating previous test data whereappropriate for validation purposes.

BackgroundIn the 1990s, a significant body of work on evaluatingdented pipelines was performed under the direction of thePipeline Research Council International, while other workwas also performed on plain dents and related defects forthe American Petroleum Institute. For the most part, thiswork focused on damage to pipelines involving plain dentsand dents with gouges. Full-scale testing involving pipelinessubjected to static and cyclic pressures were used to evaluate

Author’s contact information:tel: +1 281 955 2900email: [email protected]

This paper was presented at the 21st Pipeline Pigging and IntegrityManagement conference held in Houston on 10-12 February, 2009,organized by Clarion Technical Conferences, Houston, and ScientificSurveys Ltd, Beaconsfield, UK.

Evaluating damage to on- andoffshore pipelines using dataacquired using ILI

by Dr Chris AlexanderStress Engineering Services, Inc, Houston, TX, USA

EVALUATING THE INTEGRITY of pipelines often involves assessing data acquired from an in-lineinspection (ILI) run. ILI generates a range of data types, one of which is geometric data from a caliper

tool. Once the data are collected, engineers are required to evaluate the relative severity of any indicationsthat might have been found. With recent advances in storage capacity and instrumentation, the resolutionof the acquired data is of sufficient magnitude to make relatively accurate assessments of the potentialdamage that might exist within a given pipeline system.

In this paper a case study is provided that used data collected during an in-line inspection run of a damagedpipeline. The assessments included the development of finite-element models constructed using thegeometric ILI data. Integral to the assessments were integration of actual pressure history data that, whenused in conjunction with a cumulative damage assessment model, determined the remaining life of theselected anomaly. Additionally, the assessment used prior full-scale experimental data to confirm theaccuracy of the models. A systematic approach for evaluating damaged pipelines using ILI caliper tool datais described.

SAMPLE C

OPY


the effects of dents having varying degrees of severity on theintegrity of pipelines, and interested readers are encouragedto consult the reference documents provided in this paper.The predominant conclusion from these research efforts isthat to properly assess a defect’s severity, one mustappropriately categorize the defect. The major defectclassifications that typically arise when assessing pipelinedamage are:

• plain dents• constrained dents• gouges• mechanical damage• wrinkles

The sections that follow discuss in detail experimentaltesting that has been conducted to address the severalclasses of dent listed previously by different researchprogrammes around the world. Detailed in each section arethe appropriate references, critical variables associatedwith the defect in question, and the effects of loading (staticor cyclic) on failure behaviour.

Plain dents

Plain dents are defined as dents having no injurious defects– such as a gouge – and possessing a smooth profile (theyare often classified as smooth dents). The critical variablesrelating to plain dents are:

• dent depth (depth after rerounding due to pressure)• pipe geometry (relationship between diameter and

wall thickness)• profile curvature of the dent profile• pressure at installation• applied cyclic pressure range.

While the effects of certain variables are not clearlyunderstood, it is apparent that the denting process plays acritical role in determining the future behaviour of thedent. Early research recognized that dent depth was one of,if not the most important, variable of interest. The dent

created initially changes as a function of applied pressure(statically or cyclically).

The following equation was developed Maxey [1] andcorrelates the relationship between initial dent depth andthe residual dent depth as a function of applied pressureand yield strength.

D D

-0.5066 log 10,000

oR

y

=

×+

⎛

⎝⎜⎜

⎞

⎠⎟⎟

⎡

⎣⎢⎢

⎤

⎦⎥⎥

σσ

(1)

where:

s = hoop stress at instant of damage (psi)s

y = yield strength of pipe (psi)

Do = dent depth at instant of damage (in)

DR = residual dent depth after removal of damaging tool(in)

A review of the preceding equation by Hopkins [2] revealedsome levels of unconservatism because the aboveformulation is lower-bound and ignores the elastic spring-back of the dent at zero internal pressure. Later work byRosenfeld [3] indicates that some degree of progressivererounding occurs with pressure cycles. It is these changesin dent depth, and associated changes in dent profile, thatdetermine the eventual long-term behaviour of the dent.When considering pipes with relatively-high diameter towall thickness ratios, a significant level of reroundingoccurs on pressurisation. Work conducted for the AmericanPetroleum Institute (API) [4] showed that for 12.75-in x0.188-in X52 pipes, it was not possible to achieve dentdepths greater than 3% of the pipe diameter when the pipewas pressurised to the maximum allowable operatingpressure, even though initial dent depths as great as 18%were initially established. As will be discussed later in thispaper, this rerounding reduces the severity of the dent.

The behaviour of plain dents in static and cyclic pressureenvironments differ. The sections that follow provideinsights on these differences.

elpmaSrebmun yrtemoegepiP

dnaedarGdleiy

htgnerts

yprahCtcapmi)sbl-tf(

laniFtnedhtped

)%,D/d(

eruliaFsserts)isk(

setoN

2UPB ni-45.0xni-63 06X)isk5.76( 2.44 4.3 2.76 SYMS%211tadeliaF

2ONB ni-05.0xni-63 06X)isk8.06( 1.91 5.4 7.76 SYMS%311tadeliaF

1EIB ni-83.0xni-42 25X)isk1.35( 0.41 4.5 9.42 SYMS%84tadeliaF

1VLB)1( ni-13.0xni-03 25X

)isk7.25( 2.91 5.3 1.17 )2etonees(

1YUE ni-66.0xni-63 56X)isk9.86( 7.13 8.4 5.62 SYMS%14tadeliaF

)1(1BJF ni-84.0xni-03 25X)isk8.85( 8.22 2.3 9.81 SYMS%63tadeliaF

)1(1BJF ni-84.0xni-03 25X)isk8.85( 8.22 9.4 6.21 SYMS%42tadeliaF

Table 1. Burst pressures for plaindents.

Note: (1) cracks detected on insideseam weld of the sample;

(2) sample yielded but did not break.

SAMPLE C

OPY


Response of plain dents to static pressure loading

The response of plain dents to static pressure loads dealsprimarily with the effects of the damage on the burststrength of the pipe. In addition to concerns relating todent depth and profile, the mechanical properties of thedamaged pipe material are also important. Work wasreported in the 1980s that correlates burst pressure withdent depth and material properties for pipes with differentgeometries and grades [5]. The tests involved pipe ringsamples that were dented prior to pressure testing; Table 1provides a summary of the test results.

The definitive conclusion based on all available research isthat plain dents do not pose a threat to the structuralintegrity of a pipeline other than the potential for reducedcollapse/buckling capacity associated with the inducedovality. A discussion on the subject matter will follow in alater section of this paper. However, the classification of aplain dent assumes that no cracks, gouges, or materialimperfections are present in the vicinity of the dent.Interaction of plain dents with weld seams, especially girthwelds and submerged arc welds (SAW), can significantlyreduce the burst strength of the damaged pipeline [4]. Theprimary cause of the reduction is crack development at thetoe of the welds during pressurizing the pipe and associatedrerounding of the dent.

Response of plain dents to cyclic pressure loading

While plain dents do not pose a threat to pipeline integrityin a static environment, cyclic pressure applications canreduce the life of a pipeline. A survey of several gas andliquid transmission companies revealed the number ofapplied pressure cycles that can be expected for the respectivefuel types [6]. A gas transmission line can be expected to see60 cycles per year with a pressure differential of 200psi;however, the same pressure differential can occur over1,800 times on a liquid pipeline in the course of a year. Forthis reason, liquid pipeline operators are considerablymore concerned with fatigue than gas pipeline operators.

The impact that a plain dent has on the fatigue life of apipeline is directly related to two factors, the first of whichconcerns the dent geometry in terms of shape and depth.Dents that are deeper and possess greater levels of local

curvature reduce fatigue lives of pipes more than dents thatare shallow with relatively-smooth contours. Workconducted for the American Gas Association [6], AmericanPetroleum Institute [2], and by EPRG [5] all validate thisposition. The second factor determining the severity ofplain dents is the range of applied pressures. In general, afourth-order relationship can be assumed between theapplied stress range and fatigue life: in other words, adented pipeline subjected to a pressure differential of200psi will have a fatigue life that is 16 times greater thanif a pressure differential of 400psi were applied. Barring theeffects of rerounding (which change the local stress in thedent), the fatigue lives of plain dents are reduced to agreater degree when increased pressure differentials areassumed.

Table 2 provides several data points extracted from the APIresearch programme showing the effects of dent depth onfatigue life. As noted in the data, the 6% dent never failedand had a fatigue life that exceeded the fatigue life for the18% dent by one order of magnitude.

In assessing the overall impact that plain dents have onpipelines subjected to cyclic service, one must considerboth the applied pressure range and geometry of the dent.A given dent may not be serious in gas service, but couldpose a detriment to fatigue life when considering theservice requirements of liquid transmission pipelines.

Dents with gouges

While plain dents may be regarded as rather benign interms of their impact on structural integrity, dents withgouges are a major concern for pipeline companies. Theleading cause of pipeline failures is mechanical damage,which often occurs during excavation of pipelines, and theUnited States Department of Transportation (US DOT)has specific criteria for reporting outside incidents. Therate of reportable incidents for gas pipelines from 1970 toJune 1984 was 3.1 x 10-4/km-yr, while the rate wasapproximately 6.8 x 10-5/km-yr for the period from July1984 to 1992. A more-conservative estimate assumes thatthe actual incident rate may be as high as 10-3/km-yr due tounreported incidences. Regardless of the assumed incidentrate, world-wide efforts have focused on the need formechanical damage research. In the United States, most of

rebmunelpmaS edargdnayrtemoegepiP

laitinItnedhtped

)%,D/d(

laniFtnedhtped

)%,D/d(

otselcyCeruliaf

2-A6SU 25XedarG,ni-881.0xni-57.21 6 3.1 322,703,1

3-A21DU 25XedarG,ni-881.0xni-57.21 21 5.2 309,486

82-'A81DU 25XedarG,ni-881.0xni-57.21 81 7.0 650,101

Table 2. Cyclic pressure tests on plaindents.

Note: (1) no pressure in pipe sampleduring indentation.(2) residual dent measured with nopressure in pipe after sample waspressurised to a 65% SMYS stresslevel.(3) cycles to failure listed based uponMiner’s Rule in combining results from two applied pressure ranges (36% and 72% SMYS).(4) sample did not fail. Testing terminated due to excessive number of applied pressure cycles.

SAMPLE C

OPY


the experimental work has been conducted by BattelleMemorial Institute and Stress Engineering Services, Inc,and has been funded by the American Gas Association andthe American Petroleum Institute. In Europe, testing hasbeen conducted primarily by British Gas and Gaz de Francewith funding from the European Pipeline Research Group.

The severity of mechanical damage is rooted in the presenceof microcracks that develop at the base of the gouge duringthe process of dent rerounding due to pressure (and tosome extent elastic rebound). As with plain dents, dentswith gouges respond differently to static and cyclic pressureloading. The discussions that follow provide greater detailsregarding the associated responses.

Response of dents with gouges to static pressure loading

Unlike plain dents that do not severely affect the pressure-carrying capacity of pipelines, the deleterious nature ofdents with gouges requires careful investigation. The failurepatterns of dents with gouges that are subjected to staticpressure overload involve the outward movement of thedent region, while development and propagation ofmicrocracks at the base of the gouge occur with increasingpressure levels. Hopkins et al. [5] conducted numerous ringtests to address the failure pattern of dents combined withgouges and concluded that the failure mechanism wasductile tearing within an unstable structure.

Testing was conducted by Kiefner & Associates, Inc/StressEngineering Services, Inc [4] for determining the burstpressure of dents containing gouges. All testing wasconducted using 12-in NPS X52 pipes. Machined V-notcheswere made at various depths in the pipe samples, whichwere pressurised to 920psi (60% SMYS) and then dentedwith a 1-in wide bar. Table 3 lists six of the test samples and

the pressures at which they failed. As noted in the table,dent and gouge combinations that exceed 10% of the pipediameter and wall thicknesses (respectively) are likely tohave burst pressures that are less than the pressurecorresponding to SMYS. The pipes used in testing hadrelatively-good ductility and toughness (32% elongationand Charpy V-notch impact energy of 51ft-lbs at roomtemperature); however, pipes without such materialqualifications will fail at lower pressures. Work conductedby the Snowy Mountains Engineering Corporation inAustralia) validates the importance of having sufficientductility and toughness in reducing the potential for lowfailure pressures.

Based upon a review of the data and the experience of theauthor in experimental testing, it is difficult to envision aclosed-form solution for predicting the failure pressure dueto static overload of dents containing gouges. Althoughattempts have been made to do so, a paper written by Eiberand Leis [7] shows that the current models (developed forthe PRCI and EPRG) do not satisfactorily predict burstpressures. Several of the primary reasons for the complexitiesin predicting burst pressure of dents with gouges are:

• material properties (especially ductility andtoughness)

• sharpness and depth of gouge• pressures at indentation and during rerounding• dent profile and depth as well as resulting plastic

deformation of pipe• local work-hardening and variations in through-

wall properties due to denting

The key to future experimental testing is only to addressone variable while holding all others constant; the abovelist represents a satisfactory starting point.

rebmunelpmaS htpedeguoG)%,t/a(

tnedlaitinI)%,D/d(htped

erusserptsruB)isp(

SYMStnecrePP( tsrub )SYMS/

N1-1B 5 5 561,2 141

N3-1B 01 5 589,1 021

N6-1B 01 01 974,1 69

N7-1B 51 51 028 35

N8-1B 01 21 715,1 99

N11-1B 5 51 577 15

Table 3. Burst tests for dents withgouges.

Note: (1) dents installed with aninternal pressure of 920 psi. Dents

permitted to reround afterpressurisation.

(2) material properties: 53.6ksi yieldstrength; 72.1ksi UTS; 51 ft-lbs CVN.

htpedtnedlaudiseR)retemaidepip%(

htpedeguoG)ssenkcihtllawepip%( efileugitaF

enoN %02 selcyc005,541nahtretaerG

%4 enoN selcyc039,6nahtsseL

)dlewepipni(%4 enoN selcyc987nahtsseL

%4 %02 selcyc991nahtsseLTable 4. Fatigue life for gouges,

plain dents, and dents with gouges

SAMPLE C

OPY


Response of dents with gouges to cyclic pressure loading

Initial efforts in the pipeline research community focusedon static burst testing of mechanical damage, but once abasic level of understanding of the fracture mechanismswere developed, efforts focused on fatigue testing. Cyclicpressure tests have been conducted on pipe specimens witha variety of defect combinations [4, 5, 6]; the researchefforts conducted for the EPRG, AGA, and PRCI indicatethat if the fatigue life for plain dents is on the order of 105cycles, then the presence of gouges (in dents) reduces thisvalue to be of the order of 103. Table 4 summarises datafrom the research conducted for the EPRG on ring-testspecimens for relating plain dents and dents with gougessubjected to cyclic pressure service [5]. As noted, thepresence of a gouge significantly reduces the fatigue life ofa plain dent, although a gouge by itself is non-threatening(an observation validated by Fowler et al., [6]).

Response of dents inwelds to cyclic pressure loading

In addition to considering interaction of dents with gouges,efforts to assess the interaction of welds with dents havebeen conducted. Testing on submerged and double-submerged arc welds indicated that the dents in seam weldscould significantly reduce the burst pressures and fatiguelives of the affected pipelines. The recommendation byHopkins et al. [5] is that these defects should be treated withextreme caution and immediate repair considered.

Research efforts funded by AGA and API indicate thatwhen dents are installed in ERW seams the fatigue resistanceis on the same order as plain dents [6, 8]. This assumes thatgood-quality seam welds are present in the pipe material.The presence of girth welds was shown to reduce the fatiguelife of dents to a level less than ERW seams, but more thanSAW seams. As an example, consider that the researchprogramme for API tested a dent in a SAW weld seam thatfailed after 21,603 cycles, while the same dent in a girthweld failed after 108,164 cycles [8].

Experimental study ofstrains in dented pipes

While numerous studies have addressed the failure patternsof plain dents and dents with gouges, less effort has beenmade to evaluate the strains in dented pipes. Obviously, thecomplex nature of dent mechanics is a contributing factor;also, the use of finite-element analysis (FEA) permitsengineers to accurately understand the stress/straindistribution in dents as will be discussed later in this paper.

Lancaster et al. has conducted numerous tests directed atdeveloping an understanding of strains caused bypressurization of pipes with dents, employing the use ofboth strain gages and photoelastic coatings. His workprovides several useful findings,

• During the process of rerounding the dents withinternal pressure, approximately 60% of the dent

Fig.1. 36-in diameter pipe with 2% wrinkles.

SAMPLE C

OPY


had been recovered at a pressure equal to 70% ofthe yield pressure. There was evidence of creep atpressures above yield.

• The locations having the highest strains are on therim of the dent. Interestingly, this location wasconsistent with the failure location for unconstraineddome dents in the API research programme thatresulted in longitudinally-oriented cracks thatdeveloped on the exterior of the pipe [8].

• The highest strain measured on the rim of the dentwas 7000me, and the maximum hoop stressconcentration (SCF) was calculated to be 10.0. Incomparing this SCF with those generated by finite-element methods (FEM) for the API researchprogramme, the maximum FEM SCF was calculatedto be 7.2 for an unconstrained dome dent having aresidual dent depth of 10% [8].

In addition to the work conducted by Lancaster, Rosenfeld[3] developed a theoretical model that describes the structuralbehaviour of plain dents under pressure. His efforts alsoinvolved dent rerounding tests for validation purposes.

Wrinkle bends

Wrinkle bends are associated with the bending of pipe thatresults in creating local indentations that may be regularlyor irregularly spaced, along the length of the affected area.Wrinkle bends are not considered favourably by the pipelinecodes and most operators. As a point of reference, ASMEB31.8 841.231(g) states that wrinkle bends are permittedonly on systems that operating at hoop stress levels less than30% of the specified minimum yield strength.

As part of the American Petroleum Institute study [8],experimental efforts were undertaken to assess the effectsof wrinkle bends on the fatigue life of pipelines. Three 36-in x 0.281-in pipes were fitted with wrinkle bends havingnominal depths of 2%, 4%, and 6% (wrinkle depthpercentage calculated by dividing wrinkle depth by the

nominal diameter of the pipe). Figure 1 shows the pipesample with 2% wrinkles, while Fig.2 shows thecorresponding profiles for the three wrinkles that weretested.

Pressure-cycle testing was performed where the sampleswere pressure cycled to 100% of the operating pressure.The following fatigue results were obtained:

2% wrinkle – no failure after 44,541 cycles4% wrinkle – failure after 2,791 cycles6% wrinkle – failure after 1,086 cycles

The above results were a significant find for the APIresearch programme. The critical observations is thatalthough depth of damage is important (wrinkle or dent),the more important factor is the profile shape of thedamage. The change in radius of curvature along the lengthof the line is directly related to bending strains. As noted inthe fatigue data, a wrinkle having a depth of 6% poses asignificant threat to the integrity of the pipeline. Althoughintentional wrinkle bends are unlikely to occur offshore,the authors observed several anchor impact zones thatclearly resembled the damage profile associated with wrinklebends. For this reason, any damage in an onshore oroffshore pipeline that resembles a wrinkle bend (i.e. defecthaving a sharp curvature, as in a kink) should be removedas soon as is prudent.

Summary of experimental work

The information presented in this paper indicates that asignificant level of research has been conducted world-widein an effort to characterize and assess the severity of plaindents and dents with gouges. It can be concluded that acertain hierarchy exists in terms of defect severity, althoughunquestionable scatter is present in both the static andfatigue data. Empirical models and semi-empirical modelshave been able to predict with some success the failurepressure for dents with gouges; however, the large numberof variables has so far precluded the development of ageneral model that can accurately forecast the burst and

0 5 10 15 20 25Longitudinal Position (inches)

0.00

0.50

1.00Dept

h af

ter c

yclin

g(in

ches

)Sample Configuration

2 percent buckle

4 percent buckle

6 percent buckle

Fig.2. Wrinkle profile for the three test samples.

SAMPLE C

OPY


fatigue behaviour of all possible types of mechanical damage.Any evaluation involving numerical modelling based onILI geometry data should be validated by referencingprevious experimental work.

Dent analysisThe primary focus of this paper is to specifically address theuse of ILI data in evaluating dent severity, and the approachpresented can be used for both on- and offshore pipelines.The presentation includes a discussion on converting rawILI data into a format useful for generating a finite-elementmesh, actually performing the analysis using FEM, andinterpreting the data in terms of estimating futureperformance.Converting raw ILI data

The ILI data that are typically measured by an in-lineinspection tool is presented in cylindrical coordinates (i.e.R-q-Z). Figure 3 provides a portion of an example data settaken from an ILI tool run: as can be seen, radial coordinatesare provided as functions of circumferential and axialpositions. In this particular data set the circumferentialpositions are provided every 12o, or approximately every1.75in for the given pipe diameter. To generate accurateanalysis results, this spacing is too large, and therefore analgorithm was developed to increase the mesh density andgenerate a more-refined mesh for the FEM based on a fast

Fourier transform (FFT) routine. As a point of reference,where the raw data had 30 points circumferentially resultingin nodal spacing of 1.75in, the FFT-modified procedureproduces 177 points circumferentially spaced atapproximately 0.50in. The number of data points in theaxial direction is adjusted to match the circumferentialspacing so that the two are approximately equal (an elementaspect ratio of 1:1).

Finite-element analysis

Once the required level of mesh refinement has beenmade, the finite element model is generated, for which theR-q-Z coordinates serve as the nodes. A Fortran code wasdeveloped to read the reduced data and generate an Abaqusinput file. The coordinates for each node were developedusing the relationships shown below.

X r t

Y r t

Z Z

= +

= +

=

( ) * sin

( ) * cos

2

2

θ

θ

In these relationships, r is the inside radius from the ILIdata; q, is the circumferential position relative to the pipeaxis measured clockwise from the top of the pipe. Thethickness of the pipe, t, is taken based on the pipe’snominal wall thickness. The axis of the pipe was taken asthe global z-axis. Figure 4 shows an overall view of a dentmodel, while Fig.5 shows an enlarged view of the region

391300.1426 196.1399 199.0881 199.5950 201.6023 204.6078 207.6003 209.8907 211.0764391300.1459 196.1445 199.0792 199.5974 201.6105 204.6081 207.5990 209.8935 211.0715391300.1491 196.1445 199.0838 199.6020 201.6066 204.6069 207.6073 209.8920 211.0837391300.1524 196.1445 199.0884 199.6066 201.6026 204.6057 207.6155 209.8904 211.0959391300.1557 196.1469 199.0876 199.6119 201.6034 204.6089 207.6123 209.8949 211.0951391300.1590 196.1494 199.0868 199.6172 201.6042 204.6122 207.6090 209.8994 211.0943391300.1623 196.1518 199.0859 199.6225 201.6050 204.6154 207.6057 209.9039 211.0935391300.1655 196.1494 199.0920 199.6131 201.6026 204.6171 207.6025 209.9008 211.0968391300.1688 196.1469 199.0981 199.6037 201.6001 204.6187 207.5992 209.8978 211.1000391300.1721 196.1445 199.1042 199.5944 201.5977 204.6203 207.5960 209.8947 211.1033391300.1754 196.1506 199.1183 199.5910 201.6014 204.6298 207.5948 209.8981 211.1057391300.1787 196.1567 199.1323 199.5877 201.6050 204.6393 207.5935 209.9014 211.1082391300.1819 196.1606 199.1311 199.5983 201.6018 204.6338 207.5974 209.8984 211.1114391300.1852 196.1644 199.1299 199.6089 201.5985 204.6283 207.6013 209.8953 211.1147391300.1885 196.1683 199.1287 199.6194 201.5953 204.6228 207.6051 209.8923 211.1179391300.1918 196.1753 199.1271 199.6203 201.6001 204.6252 207.6039 209.8959 211.1130391300.1951 196.1823 199.1256 199.6213 201.6050 204.6277 207.6027 209.8996 211.1082391300.1984 196.1772 199.1266 199.6162 201.6050 204.6283 207.5994 209.8927 211.1171391300.2016 196.1721 199.1277 199.6111 201.6050 204.6289 207.5962 209.8858 211.1261391300.2049 196.1671 199.1287 199.6060 201.6050 204.6295 207.5929 209.8788 211.1350391300.2082 196.1720 199.1381 199.6142 201.6148 204.6307 207.5978 209.8779 211.1262391300.2115 196.1768 199.1476 199.6225 201.6246 204.6319 207.6027 209.8770 211.1173391300.2148 196.1768 199.1541 199.6225 201.6205 204.6328 207.5954 209.8823 211.1167

Circumferential position (every 12 degrees)

Axial position Data shown (other than first column) are radial coordinates.

Fig.3. Raw in-line inspection data in cylindrical coordinates.

SAMPLE C

OPY


where the mesh density can be seen. The “S4” type shellelements were specified in Abaqus, and symmetry boundaryconditions were specified at each end of the pipe model.For each analysis, a linear elastic analysis was performedwhere the internal pressure was the yield pressure of thepipe using the specified minimum yield strength (SMYS) ofthe respective pipe grade (for example, X52 has an SMYS of52,000psi). A typical finite-element model has of the orderof 25,000 elements.

Once the model pre-processing was completed, stresseswere calculated based on the internal pressure loading.Although plastic strains are induced in any dented pipeline,experience has shown that after several pressure cycles ashakedown to elastic action occurs and the alternatingstresses are typically within the elastic regime. Therefore, itis appropriate to elastically model cyclic stresses in dents.From the finite-element model, the principal stresses in thedented region of the model are calculated. From this stressstate a stress concentration factor (SCF) is calculated bydividing the maximum principal stress by the nominalhoop stress. Figure 6 provides a contour plot showing themaximum principal stresses in a dent that resulted in a

maximum SCF of 3.58. It is noted in this figure that themaximum stress occurred on the outside surface of themodel: these results are consistent with previous findingsfrom experimental studies where fractures in plain dentssubjected to cyclic pressures initiated on the outside surfaceof the pipe.

Data interpretation

Once the FEA model results are calculated and arepresentative SCF has been determined, the next stepinvolves estimating remaining life. It should be noted that,for this particular discussion, the focus is on plain dentswhere failure due to static pressure overload is unlikely. Ifplain dents do fail, they are most likely to do so in thepresence of cyclic pressures. Even if a large number of cyclicpressures is not likely, the process of calculating SCFsprovides operators with a means for evaluating the relativeseverity among competing dents.

From the author’s experience, the API X’ fatigue curvefrom API RP2A, Planning, designing, and constructing fixedoffshore platforms, reasonably predicts the fatigue behaviour

Fig.4. Global view of dentin finite-element model.

Fig.5. Close-up view ofdent in finite-element model.

SAMPLE C

OPY


of plain dents subjected to cyclic pressure conditions.Provided below is the equation for the API X’ curve whereDs represents the stress range in units of psi.

N = 2.978 x 1021 Ds-3.74 (2)

As an example, consider the previously-presented dentanalysis with the SCF of 3.58 (cf. Fig.6). If one assumes acyclic pressure range of 36% SMYS for an X52 pipe, thenominal hoop stress range is 18,720psi. Including the SCF,the corresponding stress range in the dented region is67,000psi. Using the API X’ curve, the resulting fatigue lifeis 2,657 cycles.

While the above presentation is certainly useful, for mostoperators an important unanswered question remains:how many years of useful service remain? In the absence ofactual historical operating data, the 2,657 cycle number isnot entirely useful. Therefore, to complete the analysis onemust consider actual operating history. Listed below arethe steps involved in evaluating the remaining life of adented pipeline considering the ILI-based stressconcentration factor used in conjunction with actualoperating pressure cycle data.

1. Obtain pressure history plot similar to one shownin Fig.7.

Resulting SCF of 3.58 on outside surface of dented region.

Units in psi

Fig.6. Maximum principal stresses onoutside surface of FEA model.

Fig.7. Historical pressure cycle datafrom an operating pipeline (pressurein psi).

SAMPLE C

OPY


2. Use rainflow counting to develop a pressure cyclehistogram similar to one shown in Fig.8.

3. Use histogram to determine a single equivalentcycle count such as 100 cycles at DP = 36% SMYS.

4. Divide the calculated fatigue life by the annual cyclecount to determine the remaining life in years.

Referring once again to the previous example, we determinedthat for a stress range of 36% SMYS the fatigue life was2,657 cycles. If a given pipeline annually experienced 100cycles at DP = 36% SMYS, the remaining life in years wouldbe 26.5 years.

DiscussionThe integrity of dents is related to not only the severity ofthe dent itself, but also to the possibility that the dent caninteract with other features such as seam and girth welds.The author’s company was the principal investigator in astudy conducted for the American Petroleum Institute toevaluate the severity of plain and constrained rock dents.Included in this study were evaluating the effects of seamand girth welds that interacted with dents.

Listed below are the major dent groupings extracted fromthe dataset from this API study, and related data for thesetest samples are included in Table 5. Within these samplesare groups based on a number of common characteristics.

These groups are important as they serve as the basis forsome of the assumptions regarding dent performance. Asan example, the test results associated with girth welds indents provides information regarding the expectedperformance of plain dents versus those dents containinggirth welds. Unless noted, all dents are unconstrained.

• plain dents – samples 1, 3, and 28• constrained dents – samples 15, 26, and 27• dents with welds – samples 16 and 20• dents with welds subjected to hydrotest – samples

30 and 31• double dents – sample 32

As noted in Equn 2 for the API X’ RP2A S-N curve, thereis a numerical relationship of 3.74 between design cyclesand applied stress range. This exponent will be used indeveloping empirical stress concentration factors for specificpipeline imperfections.

One of the objectives of this study was to evaluate how thefatigue life of plain dents is reduced when consideringfeatures such as girth welds, seam welds, and double dents.The data presented in Table 5 is used to provide numericalcorrelation among these dents, as presented below.

Stress concentration factor for dents interacting with ERWseam welds

• Sample 16 (unconstrained dent with ERW)– 22,375 cycles

Pressure Cycle Histogram for Yellow Creek Suction Location From 11/1/06 - 11/1/07 (283 Days)

159

101

7060

5140

4840 39

19 20 271612 10 10 9 6 5 5 5 3 1 6 2 0 2 0 0 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

286

0

50

100

150

200

250

300

350

60 120

180

240

300

360

420

480

540

600

660

720

780

840

900

960

1020

1080

1140

1200

1260

1320

1380

1440

1500

Peak-to-Peak Cycle Magnitude (psi)

Cyc

le C

ount

Fig.8. Pressure cycle histogram showing stress range cycle count.

SAMPLE C

OPY


• Sample 3 (unconstrained dent)– 684,903 cycles

A stress concentration factor is calculated using the abovecycles to failure using a 3.74 order relationship betweenstress and cycle life.

SCF22,375 cycles

684,903 cycles2.49

13.74

= ⎛⎝⎜

⎞⎠⎟ =

−

(3)

Stress concentration factor for dents interacting with girthwelds

• Sample 16 (unconstrained dent with ERW) – 20,220 cycles



SCF20,220 cycles

684,903 cycles2.56

13.74

= ⎛⎝⎜

⎞⎠⎟ =

−

(4)

Stress concentration factor for double dents

• Sample 16 (unconstrained dent with ERW)– 217,976 cycles



SCF217,976 cycles

684,903 cycles 1.36

13.74

= ⎛⎝⎜

⎞⎠⎟ =

−

(5)

Using the calculated stress concentration factors, it ispossible to develop a fatigue reduction factor, FRF, for eachrespective imperfection type. This value can then be usedto estimate the effect that a particular anomaly has on thefatigue life of a plain dent. Several example calculations areprovided. The FRF is calculated using the followingequation, with results for the three anomalies tabulated inTable 6.

FRF = (SCF)-3.74 (6)

Sample DescriptionInitial Dent(% pipe OD)

Rebound Dent(% pipe OD)

Final Dent(% pipe OD)

N (DDP=50% MAOP)

1 Plain dent, unconstrained 6 4.9 2.7 1,307,2233 Plain dent, unconstrained 12 6.8 2.5 684,903

15 Constrained dent 12 N/A N/A 426,58516 ERW, Plain dent, unconstrained 12 7.7 1.4 22,37520 GW, dent, unconstrained 12 7.6 1.4 2,02021 GW 2" offset from dent, unconstrained 12 6.8 1.5 38,97226 Constrained dent 24 N/A N/A 98,48327 Constrained dent 18 N/A N/A 235,00828 Plain dent, unconstrained 18 11.3 0.7 101,05630 ERW, Plain dent, unconstrained, hydrotest 12 5.9 0.7 277,39631 GW, dent, unconstrained, hydrotest 12 6.0 1.0 213,876

22 Double dent unconstrained (dents 3.5 inches apart) 125.25.6

0.81.2

217,976

69 Plain dent, unconstrained (4-inch dome indenter) 6 3.3 0.7 359,35070 Plain dent, unconstrained (4-inch dome indenter) 12 7.1 2.3 263,91071 Plain dent, unconstrained (4-inch dome indenter) 18 15.8 4.9 204,24672 Plain dent, unconstrained (4-inch dome indenter) 24 15.9 5.0 234,934

Table 5. Test results for dents subjected to cyclic pressure fatigue testing.

Notes:(1) sample 1 (unconstrained 6% plain dent) and sample 15 (constrained 12% plain dent) did not fail even after

extensive pressure cycling.(2) the final dent depth was measured after all phases of testing were completed.(3) observed failure pattern for unconstrained dents was an OD-initiated longitudinal flaw.(4) observed failure pattern for constrained dents was an ID-initiated circumferential flaw.(5) the tested cycles to failure, N, presented above assumed an applied pressure range of 50% MAOP (36% SMYS). For

the 12.75-in x 0.188-in X52 pipe used in the testing the 50% MAOP value corresponds to 550psi.

epytegamaD FCS FRF

maesdlewWREhtiwtneD 94.2 330.0

dlewhtrightiwtneD 65.2 030.0

tnedelbuoD 63.1 813.0Table 6. Fatiguelife reduction factors.

SAMPLE C

OPY


A final comment concerns two factors that were notconsidered in the analysis efforts discussed here. The firstconcerns the presence of corrosion: if corrosion is expected,one can assume that the remaining life of the dent will bereduced relative to the non-corroded case. Secondly, noconsideration of tool tolerance was included in the geometryof the finite-element models. On this second issue, readersare encouraged to interface with tool vendors regardingtolerances and what, if any, effect they would have on theresulting dent geometry.

ConclusionsThis paper has discussed methods for using ILI data toevaluate the severity of dents in pipeline systems. The mostpowerful feature of this technique is the ability for anoperator to compare the relative severity of multiple dent-like defects in an effort to make decisions regarding whichones require immediate attention. In a world of unlimitedresources, operators could evaluate and repair all defects;however, in the real world such options do not exist, andoperators must prioritize their responses based on the bestavailable sources of information.

From the author’s perspective there is no standardizedmethod for evaluating the severity of dents. It is hoped thatthe methods presented here can serve as a means foropening lines of communication between ILI companies,pipeline operators, and industry experts in formalizing amore systematic approach for evaluating dents. There iscertainly ample evidence to suggest that a reasonableunderstanding of dent behaviour exists among subjectmatter experts. When this knowledge is coupled with astandardized analysis approach, the pipeline community atlarge will be well-served.

References and bibliography1. W.A.Maxey, 1986. Outside force defect behaviour. NG-18

Report 162, AGA Catalog no. L51518.2. P.Hopkins, 1991. The significance of mechanical damage in

gas transmission pipelines. Paper 25, EPRG/PRC 8th BiennialJoint Technical Meeting on Line Pipe Research, Paris, May14-17.

3. M.J.Rosenfeld, 1998. Investigations of dent reroundingbehaviour. Proc. Int. Pipeline Conference, 1, pp299-307,Calgary, Canada.

4. C.R.Alexander, J.F.Kiefner, and J. R. Fowler, 1997. Repairof dents combined with gouges considering cyclic pressureloading. 8th Annual International Energy Week Conferenceand Exhibition, Houston, Texas, January.

5. P.Hopkins, D.G.Jones, and A.J.Clyne, 1989. Significance ofdents and defects in transmission pipelines. Proc. Int. Conf.on Pipework Engineering and Operations, London, February21-22.

6. J.R.Fowler, C. R.Alexander, P.J.Kovach, and L.M.Connelly,1994. Cyclic pressure fatigue life of pipelines with plaindents, dents with gouges, and dents with welds. Prepared byStress Engineering Services for the Offshore and OnshoreApplications Supervisory Committee of the Pipeline ResearchCommittee, PR-201-9324, June.

7. R.J.Eiber and B. N.Leis, 1995. Line pipe resistance to outsideforce. Paper 14, EPRG/PRC 10th Biennial Joint TechnicalMeeting on Line Pipe Research, Cambridge, April 18-21.

8. C.R.Alexander and J.F.Kiefner, 1997, 1999. Effects of smoothand rock dents on liquid petroleum pipelines, Phases 1 and2. API Publication 1156, May, and October, respectively.

Bibliography

C.R. Alexander, 2006. Assessing the effects of external damageon subsea pipelines. Paper No. IOPF2006-014, Proceedingsof the ASME International Offshore Pipeline Forum, October24-25, 2006, Houston, Texas.

C.R.Alexander, J.R.Fowler, and K. Leewis, 1997. Analysis ofcomposite repair methods for pipeline mechanical damagesubjected to cyclic pressure loads. 8th Annual InternationalEnergy Week Conference and Exhibition, Houston, Texas,January.

C.R.Alexander and L. M.Connelly, 1998. Analytical recreationof a dent profile considering varied soil, operating andboundary conditions. Energy Sources Technology Conference& Exhibition, Sheraton Astrodome Hotel, Houston, Texas,February 2-4.

C.R.Alexander, 1999. Analysis of dented pipeline consideringconstrained and unconstrained dent configurations. EnergySources Technology Conference & Exhibition, SheratonAstrodome Hotel, Houston, Texas, February 1-3.

American Society of Mechanical Engineers, 1991. Manual fordetermining the remaining strength of corroded pipelines.ASME B31G-1991, New York.

American Society of Mechanical Engineers, 1992. Liquidtransportation system for hydrocarbons, liquid petroleumgas, anhydrous ammonia and alcohols. ASME B31.4, NewYork.

American Society of Mechanical Engineers, 1995. Gastransmission and distribution piping systems. ASME B31.8,New York.

I.Corder and P. Corbin, 1991. The resistance of buried pressurisedpipelines to outside force damage. Paper24, EPRG/PRC 8th

Biennial Joint Technical Meeting on Line Pipe Research,Paris, France, May 14-17.

D.G.Jones and P. Hopkins, 1983. Influence of mechanicaldamage on transmission pipeline integrity. Proc. Int. GasResearch Conf., London, June 13-16.

P.B.Keating and R. L.Hoffman, 1997. Fatigue behavior of dentedpetroleum pipelines (Task 4), Office to the Office of PipelineSafety, US Department of Transportation, Texas A&MUniversity, May.

J.F.Kiefner, W.A.Bruce, and D.R.Stephens, 1994. Pipeline repairmanual. Prepared for the Line Pipe Research SupervisoryCommittee of the Pipeline Research Committee.

J.F.Kiefner, C.R.Alexander, and J.R.Fowler, 1996. Repair ofdents containing minor scratches. Proc. 9th Symposium onPipeline Research, Houston, Texas, October.

SAMPLE C

OPY


SAMPLE C

OPY


The Pigging Productsand Services Association

An international trade associationserving the pipeline industry

Our aims are to promote the knowledge of pigging and its related products andservices by providing a channel of communication between the membersthemselves, and with users and other interested parties.

Services include:Free technical information service available to all

Complimentary Buyers Guide and Directory of Members

Sourcing of pigs and pigging services

Pigging seminars – next one19th November 2008 AberdeenPPSA newsletter, “Pigging Industry News”

PPSA’s book “An Introduction to Pipeline Pigging”

Training courses

PPSA web site – www.ppsa-online.com

Want to join?Full members - pigging manufacturers and service providersAssociate members - pipeline operators, suppliers and allied industriesIndividual members - anyone with an interest in piggingTo find out more visit our web site www.ppsa-online.comor contact the Secretary at [email protected]

Pigging Products and Services AssociationP O Box 2, Stroud, Glos., GL6 8YB, UK

Telephone: +44 (0) 1285 760597Facsimile: +44 (0) 1285 760470Email: [email protected]

www.ppsa-online.com

SAMPLE C

OPY


THIRTY YEARS AGO, pipelaying productivity wasextremely low by present-day standards. The Forties 1

pipeline (32-in, 170km, maximum depth 125m) took twobarges two seasons. The Frigg pipeline system (32-in, 730km,maximum depth 140m) had an average productivitysomewhere around 0.5km/day: precise figures are notavailable. The first writer unwisely opined that the laybargesystem was so slow and inefficient that it would have to bereplaced by a different system. He was obviously wrong,which once more confirms how foolish it is to try to forecastthe future.

Productivity has much improved. The second Fortiespipeline was laid 15 years later by one barge in one season,and had a much higher average layrate of 1.9km/day, andrates near the average were achieved after only a few days[1]. The average layrate had become a much higher fractionof the peak layrate. A BP press release for a 22-in, 105-km,line to the Shetlands from the North Atlantic, laid by theSolitaire in 2003, reported an average layrate of 6.9km/dayand a peak of 7.8km/day. Layrates of the order of 5km/dayare nowadays almost routine.

An improvement from 0.5km/day to 5km/day over the 28years from 1976 to 2003 corresponds to less than 9% a year.Without for a moment wishing to downplay the courageousfinancial and engineering investments that made thatimprovement possible, it is clearly modest by comparisonwith other industries such as computing and electronics,where the progress over the same period has been severalorders of magnitude.

Creative dissatisfaction ought to encourage us to look forfurther progress, perhaps through radical change. Oneinspiration to every pipeline engineer is the PLUTO project,more than 60 years ago. The military recognized that the

armies that would invade the European mainland fromEngland would consume enormous quantities of gasoline(petrol), and sought the advice of Anglo-Iranian (theforerunner of BP) on how to transport the fuel across theEnglish Channel. Two ideas were put forward: the first wasa hollow submarine cable, laid from a cable ship; the otherwas a 3-in steel ERW pipe, which would be wound onto afloating reel and unwound as tugs towed the reel across theChannel, some 120km from Shanklin on the Isle of Wightto Cherbourg on the Cotentin Peninsula in France. Thepipeline had no cathodic protection anodes and no coating,there was no tensioner, and the positioning of the pipemust have been very imprecise. The girth welds were madeby flash-butt welding. The project had many impressiveaspects that might make us wonder how much progress hasbeen made in the intervening years. The first trial is said tohave been carried out exactly one week after the firstmeeting, something no oil company could accomplishtoday: the papers would still be in the in-tray of thecontracts’ lawyers. In 1944 a pipeline could be laid over120km in 10hrs, which is not possible in 2008, presumablybecause we know so much more about it.

Many aspects of the project are instructive. Everythingpossible was done to minimize work at sea. The pipelinewas welded together onshore, tested, and wound onto reelsin great lengths. No connections were made at sea. The onlymarine operation was to lay the pipeline, and that could bedone in one night. There were of course very good reasonsin the special context of the project: it was possible that atsea the operation could still be attacked.

Alternativeconstruction methods

The reeling method is of course still widely applied,principally for small-diameter lines, though existing

Author’s contact details:email: [email protected]

Rethinking laybarge pipelayingby Professor Andrew Palmer*1 and Dr Yue Qianjin2

1 Centre for Offshore Research and Engineering, Department of Civil Engineering,National University of Singapore, Singapore

2 State Key Laboratory for Structural Analysis of Industrial Equipment, Department ofEngineering Mechanics, Dalian University of Technology, Dalian, China

SAMPLE C

OPY


equipment can lay up to 18-in pipe. A limiting factor is thata large-diameter pipeline on a small-diameter reel has largebending strains, and may buckle unless the wall thicknessis large. Another limiting factor used to be that conventionalconcrete weight coating cannot be reeled, but to resolvethis problem a rubber-like flexible concrete has beendeveloped. It is more expensive than conventional concretebut much cheaper than adding steel to gain additionalsubmerged weight. Tests have demonstrated that flexibleconcrete can be applied to pipelines, and that it protects theanti-corrosion coating.

Reeling is attractive because the bulk of the work requiredis carried out at the shore make-up site. This is particularlyattractive if the application of internal lining is expected toslow down welding. Once the pipeline has been loaded itcan be laid very rapidly, typically at 1m/s. However, nobodyhas yet been prepared to make the serious investmentneeded for a reelship or reelbarge that could lay 36-inpipelines, for example.

Another option is tow, but it has turned out primarily to beattractive for relatively-short lengths, up to about 8km,generally as bundles. It is of course possible to constructgreater lengths by towing a number of long sections andthen joining them together, but that has only occasionallybeen done and is generally thought not to be economicallyattractive.

Alternative laybarge schemesThe conventional sequence includes a number ofoperations, carried out by different organizations and oftenat widely-separated locations. The sequential operationsare fragmented and not vertically integrated. One companymanufactures the pipes; other companies apply the anti-corrosion coating and the concrete weight coating, andinstall anodes and buckle arresters. Lengths of pipe aretransported to a laybarge, and there welded together inpairs in double joints, or sometimes into longer multiplejoints. The girth welds are all made offshore. The laybargeneeds to be large and expensive, and a great number ofpeople are engaged. Each of them needs support, so that foreach person working on the pipeline there are severalothers, controlling and positioning the barge, cooking,cleaning, managing, inspecting, transporting people to and

fro, doing laundry, organizing exercise rooms and puttingon videos, and so on.

An alternative – potentially better and cheaper – concept isto integrate the process, and to adopt the PLUTO principleof doing as much as possible onshore and as little asnecessary at sea. Imagine an integrated onshore processthat manufactures long lengths of pipeline, say 125m (10joints) but perhaps longer still. The conceptual model is thehigh-technology factory, not the construction site or theoffshore barge. The factory workers travel to the factory foreach shift, and go home afterwards: they themselves lookafter and pay for all the accommodation, food,transportation, and recreation issues that would have to beprovided by an offshore constructor. They work full timeand are paid accordingly: it does not occur to them toexpect to work 15 days on and 15 days off.

The lengths that the factory produces are complete, in thesense that when they leave the factory the only work thatstill has to be done on them is to connect them to thelengths on either side. Any necessary inspection, such asgirth-weld gamma ray or UT, has already been carried out.The lengths include girth-weld anti-corrosion protectionand infill, anodes, barcode internal and externalidentification, pre-installed connections for CP monitoring,and buckle arresters and internal coating (if required).Figure 1 illustrates a length: the lengths can be factory-hydrotested, and quality control in a factory environmentis under much less time pressure than it is offshore.

The lengths now have to the transported to the barge thatwill connect them into the pipeline and lower them to theseabed. The factory is at the shore, and the lengths arerolled from the factory onto specialized transportationvessels. The vessels sail to the laybarge, and unload theminto a storage area.

Figure 2 illustrates an S-lay version of the laybarge. It makesone weld at one station, gives the weld anti-corrosioncoating and if necessary girth-weld infill, and lays the pipeover a stinger in the conventional way.

How to make the girth connections is a separate question.Conventional ‘automatic’ welding is only one of manypossible techniques [2]. There are many welding alternatives,and some of them are known to be much faster, cheaper,

anode bucklearrester

10 joints or more; 125 m+ ?

concretebarcode

Fig.1. Factory-fabricated length of pipeline.

SAMPLE C

OPY


less dependent on skilled personnel, and more able to welddifficult materials such as some kinds of corrosion-resistantalloy. Among the alternatives are:

• friction-stir welding• flash-butt welding• homopolar welding• electron-beam welding• laser welding• explosive welding

It is sometimes argued that these are newfangled innovationsthat are still in the early development stage. That is notcorrect. Flash-butt welding, for instance, is widely appliedto rails, and in the former Soviet Union was used toconstruct some 30,000km of large-diameter pipeline; arecent Russian book says that it is not in use a present,though no reasons are given. McDermott bought the rightsto the process from the Paton Welding Institute in Kiev,and reportedly spend $10 million on development forlaybarge pipelaying in the West. McDermott was at onetime enthusiastic [3], but was unable to find an operatorwilling to be the first to apply flash-butt, and appears nowto deploy its enthusiasm elsewhere. Much effort has goneinto the development of friction welding and electron-beam welding for laybarges. Homopolar welding wasdeveloped by the Center for Electromechanics at theUniversity of Texas Austin: it makes a girth weld in a 12-inpipeline in 3sec by passing through the end butt a massivepulse of electricity, briefly reaching 15MW. The energy iskinetic energy stored in a flywheel brought back up to speedbetween one weld and the next, and so the power sourcedoes not need a continuous high-power service. At onetime, commercialization of the homopolar system was inprospect, but it has gone quiet.

Regrettably, the adherents of conventional welding areresistant to the welding systems that are successful in otherengineering fields, and devote their energy to findingreasons not to examine alternatives. However, the perceivedchallenge of alternative welding have at least spurred theimprovement of conventional techniques.

×rampconnection

stinger

Fig.2. Pipelaying schematic

There is of course no reason why a pipeline should necessarybe welded. High-quality premium threaded connectionshave been demonstrated to be reliable and able to withstandthe moments and torques applied by pipelaying. Under thescheme proposed here, there would only be one threadedconnection every 125m. The cost of a premium connectionwould become less significant, and would be paid for by theacceleration of pipelaying that it would make possible. Athreaded connection can be made up in a matter ofseconds.

ConclusionMarine pipelaying can be made cheaper and faster byabandoning the traditional process of making up thepipeline from short sections on a laybarge, and insteadmaking up much longer lengths onshore and joining themon a barge dedicated to that purpose, preferably applying arapid process for making the girth connections between thelengths.

AcknowledgementAn earlier version of this paper was presented at theDeepwater Operations Symposium in Singapore inNovember 2008.

References1. C.J.London, 1991. Forties export pipeline project. Proc.

Offshore Pipeline Technology Seminar, Copenhagen.2. A.C.Palmer, J.Hammond, and R.A.King, 2008. Reducing

the cost of offshore pipelines. Proc. Marine OperationsSpecialty Symposium, Singapore, paper MOSS-11, 275-284.

3. Flash-butt welding, video, McDermott, 1990..

SAMPLE C

OPY


SAMPLE C

OPY


THE REHABILITATION OF corroded pipelines withepoxy repair systems is becoming a well-accepted

engineering practice and an interesting alternative to theclassic repair methods for metallic pipes, mainly in the oilindustry, saving time and allowing safer operation [2].Since offshore platforms are hydrocarbon atmospheres,any repair method that uses equipment that produces heatand sparkling is forbidden: type B sleeves, leak clamps, andhot tapping are therefore excluded from the list of allowablerepair methods. According to Ref.2, only Bolt-On Clampswith seals are allowed for leak repairs on offshore platforms.

Corroded pipelines can be repaired or reinforced with acomposite sleeve system, in which a pipe segment isreinforced by wrapping it with concentric coils of compositematerial after the application of epoxy filler at the corrosiondefect. Generally, the composite sleeve is not only used asrepair system itself (mainly to avoid or to restrain the

propagation of internal flaws), but also as a complementaryprocedure to enhance the reliability of weldments,eliminating the necessity of heat treatment (in the weldingoperation there is always a possibility of metallurgicalchanges in the parent metal in the vicinity of the weld).Technical specification ISO 24817 [1] gives requirementsand recommendations for the qualification and design,installation, testing, and inspection of the externalapplication of composite repairs to corroded or damagedpipework. Nevertheless, so far, composite repair systemsare not effective for through-thickness corrosion defectsbecause generally they cannot avoid leaking.

The present paper presents a very simple and systematicmethodology for repairing leaking corrosion defects inmetallic pipelines with epoxy resins. The focus is to ensurethat the pipe will not leak after a repair, and such aprocedure can be associated with a composite sleeve thatwill further ensure a satisfactory level of structural integrity.The study is focused on what ISO 24817/TS defines as adefect type B – where the substrate requires structuralreinforcement and sealing of through-wall defects (leaks) –and all three classes of repair, although mainly Class 3

Author’s contact information:tel: +55-21-2629-5565email: [email protected]

Rehabilitation of corroded steelpipelines with epoxy repairsystems

by H S Costa-Mattos1, J M L Reis*1, R F Sampaio1, and V A Perrut2

1Programa de Pós-Graduação em Engenharia Mecânica, Laboratório de Mecânica Teórica eAplicada, Universidade Federal Fluminense, Niterói, Brazil

2 Centro de Pesquisas e Desenvolvimento da Petrobrás – CENPES, Ilha do Fundão, Rio deJaneiro, Brazil

THE REHABILITATION OF corroded pipelines using epoxy repair systems is becoming a well-acceptedengineering practice and an interesting alternative to the classic repair methods for metallic pipes in

the oil industry, both saving time and allowing safer operation. In these repair systems, a pipe segment isreinforced by wrapping it with concentric coils of composite material after the application of epoxy fillerat the corrosion defect. The technical specification ISO 24817 [1] gives requirements and recommendationsfor the qualification and design, installation, testing, and inspection for the external application of compositerepairs to corroded or damaged pipework. Nevertheless, so far, composite repair systems are not totallyeffective for through-thickness corrosion defects because generally they cannot avoid leaking. The presentpaper presents a simple and systematic methodology for repairing leaking corrosion defects in metallicpipelines with epoxy resins. The focus is to ensure an adequate application of the epoxy filler such that thepipe will not leak after the repair. Such a procedure can be associated with a composite sleeve that willensure a satisfactory level of structural integrity. Examples of repair systems in different damage situationsare presented and analysed, showing the practical use of the proposed methodology.

SAMPLE C

OPY


which is appropriate for systems transporting producedfluids.

The main motivation for the study presented on this paperis leaking defects found in the produced water pipelinesused in offshore oil platforms. The damages derived fromcorrosion process in produced water pipelines in platformscause very important economic losses because the operationmust be stopped while the repair is being performed(Fig.1). Although the operation pressure of these pipelinesis not very high, the water temperature is between 60oC and90oC, which can be a major shortcoming if polymericmaterials are used as repair systems.

The objective is to ensure the pipe will not leak under theoperational pressure and temperature after the repair. Themaximum time allowed between the beginning of therepair and the return to operation is 75mins. Hydrostatictests were carried out with water at room temperature andat 80oC to validate the epoxy repair systems that are appliedin offshore produced water pipelines, and the experimentaltests were aimed at analysing the performance of differentepoxy resins in real offshore platform repair situations.Examples of repair systems in different damage situationsare presented and analysed, illustrating the possibilities ofpractical use of the proposed methodology.

Epoxy resinsTwo different commercial fast-curing epoxy resins wereanalysed: both are two-component systems consisting of abase and solidifier. The first one (System A) is designed forleak repairs on tanks and pipes, as well as for otheremergency applications, and is based on a silicon steel alloy

blended within high molecular weight polymers andoligomers. It is partly cured (machining and/or light loading)after 35mins at 25oC and is fully cured after 1hr at thistemperature. Further technical data for System A includes:

• flexural strength: 59.3MPa• tensile shear on steel: 17.2MPa• compressive strength: 55.8MPa• heat-distortion temperature: 51oC

The second system (System B) is also a polymer-basedsystem specially developed for repairs, and consisting of amixture of epoxy resin and aluminium powder. It is partlycured (machining and/or light loading) after 18mins at25oC and is fully cured after 40mins at this temperature.Further technical data for System B includes:

• flexural strength: 67MPa• tensile shear on steel: 19MPa• compressive strength: 104MPa• heat-distortion temperature: 120oC

Since the heat-distortion temperature for System A is verylow (51oC) it was only tested at room temperature. Thehydrostatic tests with pipes repaired with System 2 wereperformed at two different temperatures: room temperatureand 80oC.

Methodology forthe epoxy repair system

Since epoxy repair systems do not necessarily avoid leakage,even if a composite sleeve is used, the following methodology

Fig.1. Corrosion damage inproduced-water pipelines.

SAMPLE C

OPY


was created to improve the effectiveness of such repairsystems in the produced-water pipelines used on offshoreoil platforms. The experimental set-up in the laboratorywas designed to approximate a real repair operation, wherethe resin has to be applied in field conditions (which affectthe quality of the resulting epoxy repair). To optimize theprocess, avoiding stopping production for a long period, amaximum repair time of 75mins is suggested from thebeginning of the repair procedure to the return to operation.

In a repair of a pipeline with through-thickness defects withepoxy resins, two mechanisms of brutal failure can occurwhen pressure is applied, see Fig.2. The experimentalprocedure was designed to minimize the possibility of suchfailure modes.

Defect sizing

Defect sizing is important in order to define the limits foran effective use of the repair procedure. The dimensions ofthe defect should be determined by the smallest ellipse,with one axis parallel to the axis of the pipe, which fullycontains the area of the flaw (see Fig.3). The maximumallowable defect size for the proposed repair procedure is

Fig.2. Types of failure.

Fig.3. Defect sizing.

Through-thickness defect

Fig.4. Surface preparation.

SAMPLE C

OPY


defined by the semi-major axis of the ellipse, a, which isgiven by:

aR

tmax max ,≤ ⎧⎨⎩

⎫⎬⎭10 (1)

where R is the inner radius of the pipe and t is the wallthickness. This means that the maximum allowabledimension for the semi-major axis a is the greatest value ofeither the wall thickness t or 1/10 of the inner radius R.

Proposed repair procedure

The repair methodology can be described as follows:

Surface preparation

Surface treatment often involves chemical reactions whichproduce surface modifications on adherends, or mechanicalprocedures, which improve adhesion by increasingmechanical interlocking of the adhesive to the adherend.In this way, the primary objective of a surface treatment isto increase the surface energy of the adherend as much aspossible, and/or improve the contact between the adhesiveand the adherend by increasing the contact area. Increasingroughness, or an increase in surface area, has been shownto give good results in improving adhesion. Subsequently,a relationship exists between good adhesion and bonddurability.

In order to obtain these properties, sanding with 120 or150 sandpaper was used to achieve a white metal appearanceand to remove some of the existing oxide layer in the

substrate. A final rinse with solvent was made to provide asurface free of oil, grease, and dirt surface. After this, theadhesive was mixed according to the manufacture’sprocedure, and applied to the pipe. It is important to pointout that, in a real situation, the pipe may be so corrodedthat sandpaper should be used with extreme care (seeFig.4). Also, since offshore platforms are hydrocarbonatmospheres, any method of mechanically roughening thesurface that may produce heat or sparking (such assandblasting, cutting, grinding), is unacceptable.

Introduction of an internal rubber cap to avoid spillage ofepoxy resin

An elliptically-shaped rubber cap must be used to avoidresin spillage inside the pipe. Since the rubber is verydeformable, it is easy to introduce the cap into the pipe, andit is maintained in position using a simple system of nylonstrings.

The cap should allow formation of and internal layer ofadhesive with approximately the same thickness as the pipewall, and with average dimension twice the size of the defect(see Fig.5). For through-thickness defects with the semi-major axis less than or equal to 5mm, it may be difficult tointroduce the rubber cap, and a metallic wedge should beused instead (Fig.6). The following steps in the repairprocedure are exactly the same if either the wedge or the capis used.

Application of the first external layer of epoxy adhesive

The epoxy adhesive layer applied externally should cover anarea approximately five times that of the ellipse (Fig.7), and

Fig.5. Rubber cap toavoid adhesive spillage.

Fig.6. Metallic wedgefor smaller defects.

SAMPLE C

OPY


the thickness of this first layer must be at least equal to thethickness of the pipe; the layer should also have a smoothboundary for improved performance. After application, aninitial epoxy polymerization time is allowed according tothe manufacturer’s instructions (the maximum desirablebeing 20mins).

Application of the second layer of epoxy adhesive

A second layer of adhesive must be applied without sanding.The repair procedure is considered adequate when:

Non-Smooth Finishing

Smooth Finishing

Fig.7. External epoxy adhesive layer.

Fig.8. Equivalent system..

Pipeline with through-thickness defect

Rubber Cap to avoid adhesive spilling

Rubber Band Metallic Clamp

Epoxy Adhesive

Fig.9. Complete repair system.

SAMPLE C

OPY


1 2+⎛⎝⎜

⎞⎠⎟⎛⎝⎜

⎞⎠⎟

≥a

b

PR

t yσ (2)

where a and b are, respectively, the semi-major and the semi-minor axis of the ellipse, R is the inner radius of the pipe,t the wall thickness and s

y the yield stress of the pipe

material.

The stress distribution in a general through-thicknesscorrosion defect is very complex, but if the size of the defectis limited, a rough estimate of the magnitude of thepermanent deformation close to the defect can beperformed. The term on the left-hand side of Eqn 2 is themaximum stress in a thin-walled infinite plate with anelliptical defect with semi axes a and b subjected to tractionof a uniform force per unit area S = PR/t (see Fig.8). Thestress concentration factor in this case isKt

ab

= +( )1 2 . The criterion in Ref.2 states that apermanent deformation close to the defect in a pipe can beneglected when K

tS is less than the yield stress s

y. For

closed-ended pipes, the yield stress should be adjusted by afactor of 1.115 [3].

If this condition is verified, immediately after the applicationof the second epoxy layer a rubber sheet should be appliedover the repair around the perimeter and a simple metallicclamp, similar to those used for garden hoses, can beattached (Fig.9). The clamp is not used to improve the

structural integrity of the pipe, but to prevent the twopossible major failure mechanisms of the adhesive repairshown in Fig.2, mainly at the beginning of operation whenthe resin may not be fully cured.

Under these circumstances, the proposed procedure iseffective as a repair system by itself. Nevertheless, thisprocedure is intended to be used together with a compositesleeve (which is normalized, for instance, by Ref.1). Themain objective is to ensure that composite repairs ofleaking defects when qualified, designed, installed, andinspected using ISO/TS 24817 and the proposed procedure,will meet the specified performance requirements. Thesuggestion is to apply the epoxy resin as described in thispaper and then apply a composite material sleeve, of anormalized thickness, to restrain the plastic strain and toassure a satisfactory level of structural integrity.

An alternative method for defining the necessary thicknessof composite material to ensure both the safety of repairsunder operational conditions and the lifetime extensionunder operational conditions, can be found in Ref.4. Thismethod, although simple, is acceptable for different failuremechanisms, including plasticity, fatigue, and fracture.The method meets the most widely-used criteria for theassessment of corrosion defects under internal pressureloading – a family of criteria described in [5] as the effective-area methods. These include the ASME B31G criterionand the RSTRENG 0.85 criterion (also known as the

Fig.10. Test apparatus.

Fig.11 – Detailed temperaturecontrol system: 1 – the pressured

water machine connection; 2 – thetemperature control thermostat; 3 –

the electrical resistance.

SAMPLE C

OPY


modified B31G criterion). Nevertheless, this study is mainlyfocused on metal loss due to corrosion treated as a part-through-wall defect in the pipe, and not on through-thickness defects.

Results and discussionAn experimental set-up was designed to examine theeffectiveness of the methodology, approximating to a realrepair operation as far as possible. Five different specimensof API 5L grade B steel pipes, normally used in offshoreplatform for produced water, were used as the specimensfor hydrostatic tests:

• specimen 1: 2-in diameter Schedule 80 pipe with a3-mm diameter circular hole

• specimen 2: 2-in diameter Schedule 80 pipe with a10-mm diameter circular hole

• specimen 3: 12-in diameter Schedule 20 pipe,1300mm long, with a 10-mm diameter circular hole

• specimen 4: 12-in diameter Schedule 20 pipe,1300mm long, with a 30-mm diameter circular hole

• specimen 5: 3.5-in diameter Schedule 20 pipe,1000mm long, taken from the field with realcorrosion defects (see Fig.4).

Initially, all the repaired specimens (no composites sleeveswere used, only the clamp) with the two systems weresubmitted to a classical hydrostatic test at room temperatureto evaluate its strength and effectiveness. The maximumallowable time for each repair was 60mins, and all testsbegan exactly 75mins after the start of the repair process. Inthe tests, the pipe pressure was raised to 30kg/cm2 andmaintained at this level for 60mins. After five cycles, if therepair did not fail, the specimen was unloaded and inspected

to check any eventual small leaks or reinforcementdisbonding.

As a second step, the specimens were repaired with systemB (no sleeves were used, only the clamp) and submitted tofive pressure cycles (60mins at 30 kg/cm2) with the watertemperature inside the specimen at 80ºC, increased whilethe water was at atmospheric pressure. The internal pressurewas not increased until after the temperature had stabilized.After each pressure cycle, the specimen was cooled to roomtemperature, and each specimen was therefore alsosubmitted to five temperature cycles during testing.

Once again, the maximum allowable time for each repairwas 60mins, and all tests began exactly 75mins after thebeginning of the repair. The temperature level of 80ºC waschosen in order to simulate average offshore fluidconditions.

The system to control water temperature inside thespecimens was designed specially for this procedure, andthe whole system (including the electrical resistance) wasinstalled at one end of the specimen, as can be seen in Figs10 and 11.

All the repairs performed with Systems A and B using theabove methodology withstood the five pressure cycles withwater at room temperature. The repairs also resisted thehigh-pressure tests; it was not possible to obtain a failurepressure since the pipe end caps were not designed for bursttesting and they deformed plastically and failed before therepair failed, as can be seen in Fig.13.

If the proposed procedure is not adopted, however, therepair may not be able to resist the loading. Table 1 showsthe failure pressure obtained for specimen 2 – the 2-in

.

Fig.12. 12-in SCH-0 steel pipe with a 10-mm repairedhole, before and after testing.

Fig.13. Deformed end cap after testingat 60kg/cm2 and 80ºC.

SAMPLE C

OPY


diameter Schedule 80 pipe with a 10-mm diameter circularhole – repaired using system A (no cap and no clamp).

All the pipes repaired with System B at 80ºC resisted for thefive cycles. In order to decide whether a given epoxy systemcan be used at higher temperatures, it is suggested that thesame conditions are used as presented in Ref.1 for compositesleeves: “For a design temperature greater than 40oC therepair system shall not be used at a temperature higher thanthe glass transition temperature (T

g) less 30oC. For repair

systems where Tg cannot be measured, the repair system

shall not be used above the heat-distortion temperature less20oC. For repair systems which do not exhibit a cleartransition point, i.e. a significant reduction in mechanicalproperties at elevated temperatures, then an uppertemperature limit, T

m, shall be defined (or quoted) by the

repair supplier.”

As an example, the failure pressures observed in hydrostatictests performed with specimen 4 (which has heat-distortiontemperature of 51oC) repaired using system A at 80oC arepresented in Table 2.

It is interesting to note that the adhesive System A behavedsurprisingly well when the proposed repair procedure wasadopted, even at temperatures above the heat-distortiontemperature. All the repairs resisted to five cycles at 80oCin tests performed on specimens 1, 2, and 3.

ConclusionsThe present work is a first step towards the definition of

safer and more-reliable procedure for applying epoxy repairsystems to through-thickness flaws caused by corrosion inmetallic pipelines. This procedure is designed to be usedtogether with a composite-sleeve repair system (which isnormalized, for instance, by the ISO technical specification24817). The proposal is to apply the epoxy resin as describedin this paper and then to apply a composite material sleeve,with a normalized thickness, to restrain the plastic strainand to ensure a satisfactory level of structural integrity. Themain objective is to ensure that composite repairs toleaking defects when qualified, designed, installed, andinspected using ISO/TS 24817, and also the proposedcomplementary procedure, will meet the specifiedperformance requirements.

The main requirements for epoxy resins to be used as repairsystems are: fast curing, high heat-distortion temperature,and a thermal expansion coefficient similar to that of thematerial of the pipe. The full validation of this simplifiedrepair methodology still requires an extensive programmeof experimental investigation, mainly concerning fatigue,creep, ageing, and resistance to UV degradation andweathering.

References1. ISO Technical Specification 24817, 2006. Petroleum,

petrochemical and natural gas industries - composite repairsfor pipework - qualification and design, installation, testingand inspection.

2. C.A.Jaske, B.O.Hart, and W.A.Bruce, 2006. Pipeline repairmanual. Pipeline Research Council International, Inc.Virginia.

tseT mc/gk(erusserperuliaF 2)

1 29.8

2 46.71

3 71.61

4 53.81

5 72.41

egarevA 70.51

tseT mc/gk(erusserperuliaF 2)

1 )elcyctsrif(81.02

2 )elcycdnoces(29.4

3 )nim01retfa-elcyctsrif(00.03

4 )elcyctsrif(29.31

5 )elcyctsrif(48.9

Table 1. Failure pressure forspecimen 2 if the repair procedure is

not adopted.

Table 2. Failure pressure forspecimen 4 at 80oC.

SAMPLE C

OPY


3. A.T.de Mello dos Santos, 2006. Simplified analysis of thecaps influence in elasto-plastic pipe burst tests. MSc Thesis,Universidade Federal Fluminense, January.

4. H.Costa Mattos, R.F.Sampaio, J.M.L.Reis, and V.A.Perrut,2007. Rehabilitation of corroded steel pipelines with epoxyrepair systems. In: Solid mechanics in Brazil 2007, EdsM.Alves and H.S.da Costa Mattos, Brazilian Society of

Mechanical Sciences and Engineering, ISBN 978-85-85769-30-7, pp485 – 496.

5. D.R.Stephens and R.B.Francini, 2000. A review andevaluation of remaining strength criteria for corrosion defectsin transmission pipelines. ETCE2000/OGPT-10255,Proceedings of ETCE/OMAE2000 Joint Conference, Energyfor the New Millenium, New Orleans, USA, 2000.

take into consideration a refined product travelling thesame distance, it would probably be under a nickel. So ifyou could get ethanol from Illinois to the East Coast for 12cents a gallon cheaper than you can today, obviously a lotwould change in the world, and the interest in ethanolwould increase,” Robert White, director of operations forthe Omaha, Nebraska-based Ethanol Promotion andInformation Council, said. John Urbanchuk agreed, “Thecost of shipping ethanol would be about the same as it is toship gasoline through a pipeline.”

In the US, ethanol is primarily moved by rail and roadtankers, which are costly and time-consuming methods oftransportation. While a pipeline should help in the cost ofdistributing ethanol (and presumably in the cost ofconsuming it), it could also take away jobs from the rail andtrucking industry, particularly in the Midwest. Most expertsadmit there’s something of a chicken-and-the-egg effect ascompanies consider shipping ethanol. The production ofethanol isn’t high enough today to create a desperate needfor pipelines, but without a pipeline infrastructure in place,companies are hesitant to produce more ethanol. One

motivator is the Energy Independence and Security Act of2007, which President Bush signed in December. The lawrequires that American fuel producers use 36bn gallons ofrenewable fuels by 2022, with is more than five times whatis currently used.

But the biggest challenge to Magellan and Buckeye nowmay not be moving products through an ethanol pipeline,but moving funding through the federal governmentpipeline. “They have reached out to Congress and said,‘Some of this has merit, but we need some support and weneed to know how much that support’s going to be, becausewe need to make business decisions on our end,’” MrWhite said. “[This $3-bn project is] a substantial investmentthat you have to recapitalize somehow.” The companies’press release repeatedly stresses the importance ofgovernment support: “Congressional support and assistanceis necessary for a project of this nature given the changingfederal policies associated with renewable fuels.” The EnergyAct included a provision requiring that the US governmentundertakes its own feasibility study on ethanol-dedicatedpipelines, and this study is due to be released in 2010.

Editorial (continued form page 4)

SAMPLE C

OPY


SAMPLE C

OPY


IN STRUCTURING a finance package for an oil or gaspipeline project, it is essential that the project risks be

identified, and how they will be allocated to the partiesinvolved who include shareholders, banks, contractors,and governments. The ultimate allocation of these risks isa matter of negotiation, which can only be successfullyconcluded when all parties have agreed to what extent theyare willing to commit funds and bear certain risks.

The key to successfully financing an oil or gas pipelineproject is the equable allocation of risks among the projectparticipants. Risks must be allocated to the party that is inthe best position to manage them, and a rigorousmethodology must be followed including:

• risk identification• risk evaluation (sensitivity analysis)

Technical and commercialchallenges in procurement andimplementation of majorinternational pipeline projects

by Assadollah MaleknejadVice President – Finance and Economic Affairs, Pars Oil Co, Tehran, Iran

OIL AND GAS pipelines have significant potential for enhancing stability and improving living standardsin the host countries. Pipeline projects require significant investment to diversify upstream energy

supply, downstream security of demand, and enhanced midstream transport infrastructure to increasemarket access and interconnectivity for both. Nevertheless, international pipeline projects face considerabletechnical and commercial challenges in their procurement and implementation. There are considerablerisks attached to pipeline projects, pertaining to the techno-economic viability of the projects. However,technical risks can be seen as challenges that have to be faced to make the projects successful.

The technical challenges include:

• the possibility of construction cost overruns• the possibility of construction delays• the possibility of operating cost overruns• can the project be completed within acceptable performance levels?• the possibility that the project will not operate continuously• the possibility that the project will not operate according to environmental requirements

There are also other risks in respect of economic and security aspects as well as the investment climatein the region. However, these risks can also be seen as challenges that have to be faced to make the projectssuccessful. These include:

• will the project fail due to economic pressuresor contract failure?• is there a possibility of interest increases?• is there a possibility of revenue reductions?• is there a possibility of interest increases?• will the project be affected by political events, such as legal changes, exchange rates, etc.?• will the project operate under inconvertibility and limited transferability of currency?• will the project operate under the risk that the government may unilaterally change the economic or

fiscal conditions on which it is based?

This paper presents a risk analysis and risk mitigation methodology, and addresses the concerns of thestakeholders in relation to cross-border pipeline projects.

SAMPLE C

OPY


• risk mitigation• risk allocation (commercial negotiation), and• contractual documentation

The risks discussed below are described from a financialpoint of view and have been separated into the three phasesof the project: pre-development, construction, andoperation.

Pre-developmentThis is the phase in which relatively small amounts ofmoney have been, or are being, spent. The oil or gaspipeline project may fail for a number of reasons, but thefinancial risks for the shareholders are limited.

During this phase, providers of finance will form preliminaryconclusions about whether the project is financeable. Dueto the various uncertainties in this stage, the providers offinance will ascertain whether the project is based on solidtechnical, economic, and legal foundations. Once theprimary parameters have been agreed, it will be necessary toestablish the chances of survival of the project undervarious pessimistic conditions, including higher capitalcosts, delay of start-up, higher operating costs, higherfinancing costs, or combinations of any of these.

Of particular importance to the project during the pre-development stage is how to deal with the perceived countryrisk exposure, and the ability of providers of finance to bearthis risk. Providers of finance will generally not wish tocommit money, time, or effort if the political risks of theproject are not fully mitigated. Due to importance of thisparameter, the political risks and required mitigations willbe discussed further later in this paper.

Construction phaseDuring the construction phase, the main risk is failure tocomplete the project within acceptable performance levelsand an acceptable timeframe and budget. In the case of oiland gas projects, one of the key risks is the ability of thecontractor to construct and lay the pipeline according toacceptable standards: API certification and OA duringconstruction, as well as independent technical due diligence,are normally sufficient mitigation for financiers.Furthermore, recognized experienced project managementwill be a requirement. Risks with respect to earthquakesand ground movement, as well as accessibility of theterrain, will be addressed prior to construction, and shouldbe mitigated by technical solutions in line with theindependent technical consultant’s recommendations.

Construction risk falls on the project and its sponsors who,in turn, may be able to hedge their risk by purchasingvarious forms of insurance and obtaining guarantees fromcontractors regarding costs, completion schedules, and

operational performance. The construction risk isconsidered by the providers of finance to be relatively high,and potentially involves significant losses. It is therefore amost important financial risk; for example, should theproject fail during the construction phase, the security overthe assets of the project would be of little value. Thus,providers of finance do not want to take the constructionrisk, and normally ask for recourse to the sponsors’ otherresources until the project is completed and tested.

Operation phaseOnce the pipeline is in operation, the main concern is thatit may not operate on a continuing basis within acceptableeconomic, technical, and environmental parameters. Suchoperational risks are numerous, and are borne by theproject and its limited-recourse providers of finance.However, the project can hedge against the risks throughcontractual and guarantee arrangements that in effecttransfer some of these to other parties. The following areexamples:

• guarantees from suppliers of equipment for technicalperformance

• supply guarantee by means of “supply or pay contract”from the supplier of the oil or gas

• off-take guarantee by means of “take or pay contract”from the off-taker of the oil or gas

• guarantees against political events (see below)• recruiting qualified operator• establishment of regional support and cooperation

between countries in case of cross-border pipelines

Environmental issuesIt is important that the project should operate according tothe latest environmental requirements, and providers offinance require an independently-performed environmentalimpact audit. This serves two purposes: firstly, it establishesa baseline environmental status, and secondly, it providesreassurance that (it is assumed) state-of-the-art standardsare being implemented, meeting – as a minimum – bothWorld Bank and local standards. With respect to a baselinereport, this would mitigate exposure to damages resultingfrom existing environmental disturbances.

Political risksThroughout the project phases, political risk is of primeconcern. Examples of political risk include (but are notlimited to):

• Inconvertibility and limited transferability ofcurrency: the risk that sufficient hard-currency fundsare not available to meet foreign currency obligationstowards the project.

SAMPLE C

OPY

• Expropriation or nationalization by the government:the risk that the government unilaterally takescontrol of a project or changes the economic orfiscal conditions (creeping expropriation) underwhich a project operates.

• War, revolution, and civil war.

• Breach of contract by the host or transit governmentto the material rights of the project, such as throughtaxation terms, approvals, export rights, andproduction rights; restrictions on import or exportof equipment; prohibition of the entry of personnel.

Providers of finance will therefore, prior to committing tofinance, evaluate the political risks compared to the exposurethey are willing to take in combination with politically-insured funding from multilateral agencies (MLA) such asthe World Bank and/or export credit guarantee agencies(ECGA).

As already discussed in this paper, the sponsors of theproject will generally not be able to commit money, time,and effort on any international pipeline if the financialinstitutions (MAL and/or ECGA) are unwilling to establishexposure to political risks. Even though the sponsors maybelieve there will be opportunities for MAL and/or ECGAto participate in funding arrangements in the future, theymay be unable at this moment to give a positive indicationto proceed with the project.

Once MLA and ECGA funding is available in principle,more interest can be expected from commercial banks andcapital markets to participate into the project. They wouldfind reassurance in the political circumstances whenever

MLA and ECGA financing is available. At the same time,the MLA will require the host countries to provideguarantees against political events, legal changes, taxationchanges, exchange rate fluctuations, and investmentrecovery by means of political support. If legislation for thisis seen as unsatisfactory in the host countries, the guaranteeswill be based on offshore receivables, reserves, anddisbursement accounts.

Risk-analysis proceduresAn analysis of the risks from the perspective of the sponsorswill consist of understanding the risks as identified by anumber of independent experts and determining theultimate effect on the viability of the project to service debt.The services provided by independent experts will include:

• technical advice with respect to supply• technical advice with respect to procurement,

construction, costs, timing, and environment• insurance advice• market advice with respect to offtake• legal opinions with respect to local jurisdiction,

security aspects, taxation, recovery of foreigncurrency

• legal opinions of the sponsors

Assuming a satisfactory outcome of this advice, a financialanalysis will be completed, and a base case will be generatedand the effect of possible construction cost overruns,construction delays, operating cost overruns, revenuereductions, and interest increases will be modelled for thepurpose of selecting ratios in which the project may proceedfor construction.

SAMPLE C

OPY

SAMPLE C

OPY

journal of pipeline engineering - pipemag.com - home page - sample... · pipeline engineering...

Documents