Journal of Pipeline Engineering distribution - … - Dec 13...headline refers to the rate of...

December, 2013 Vol.12, No.4

Great Southern Press Clarion Technical Publishers

Journal of Pipeline Engineering

incorporating The Journal of Pipeline Integrity

Sample

copy

- not

for di

stribu

tion

Journal of Pipeline Engineering

Editorial Board - 2013

Dr Husain Al-Muslim, Pipeline Engineer, Consulting Services Department, Saudi Aramco, Dhahran, Saudi Arabia

Mohd Nazmi Ali Napiah, Pipeline Engineer, Petronas Gas, Segamat, MalaysiaDr-Ing Michael Beller, Rosen Engineering, Karlsruhe, Germany

Jorge Bonnetto, Operations Director TGS (retired), TGS, Buenos Aires, ArgentinaDr Andrew Cosham, Atkins, Newcastle upon Tyne, UK

Dr Sreekanta Das, Associate Professor, Department of Civil and Environmental Engineering, University of Windsor, ON, Canada

Leigh Fletcher, Welding and Pipeline Integrity, Bright, AustraliaDaniel Hamburger, Pipeline Maintenance Manager, Kinder Morgan, Birmingham, AL, USA

Dr Stijn Hertele, Universiteit Gent – Laboratory Soete, Gent, BelgiumProf. Phil Hopkins, Executive Director, Penspen Ltd, Newcastle upon Tyne, UK

Michael Istre, Chief Engineer, Project Consulting Services, Houston, TX, USA

Dr Shawn Kenny, Department of Civil and Environmental Engineering, Faculty of Engineering and Design, Carleton University, Ottawa, ON, Canada

Dr Gerhard Knauf, Salzgitter Mannesmann Forschung GmbH, Duisburg, GermanyProf. Andrew Palmer, Dept of Civil Engineering – National University of Singapore, Singapore

Prof. Dimitri Pavlou, Professor of Mechanical Engineering, Technological Institute of Halkida, Halkida, Greece

Dr Julia Race, School of Marine Sciences – University of Newcastle, Newcastle upon Tyne, UK

Dr John Smart, John Smart & Associates, Houston, TX, USAJan Spiekhout, DNV Kema, Groningen, Netherlands

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

Patrick Vieth, President, Dynamic Risk, The Woodlands, TX, USADr Joe Zhou, Technology Leader, TransCanada PipeLines Ltd, Calgary, Canada

Dr Xian-Kui Zhu, Principal Engineer, Edison Welding Institute, Columbus, OH, USA

� � �

Sample

copy

- not

for di

stribu

tion

4th Quarter, 2013 273

The Journal of Pipeline Engineeringincorporating

The Journal of Pipeline IntegrityVolume 12, No 4 • Fourth Quarter, 2013

ContentsDr Robert M Andrews, Harry Kamping, Henk de Haan, Otto Jan Huising, and Neil A Millwood ......................277

Guidance for mechanized GMAW of onshore pipelinesDr Adilson C Benjamin .....................................................................................................................................301

Prediction of the failure pressure of corroded pipelines subjected to a longitudinal compressive force superimposed on the pressure loading

Peter Tuft and Sergio Cunha .............................................................................................................................313Comparing international pipeline failure rates

Daniel Sandana, Mike Dale, Dr E A Charles, and Dr Julia Race .........................................................................321Internal stress-corrosion cracking in anthropogenic CO2 pipelines: is it possible?

Indranil Guha, Beau Whitney, Raúl Flores-Berrones, Aditya Barsainya, and Gaurav Arya ................................. 335Earthquakes and the Indian pipeline industry

❖ ❖ ❖

OUR COVER PHOTO shows pipeline welding in the field: photo provided, with thanks, by Lincoln Electric, Cleveland, OH, USA.

The Journal of Pipeline Engineering has been accepted by the Scopus Content Selection & Advisory Board (CSAB) to be part of the SciVerse Scopus database and index.

Sample

copy

- not

for di

stribu

tion

The Journal of Pipeline Engineering274

1. Disclaimer: While every effort is made to check the accuracy of the contributions published in The Journal of Pipeline Engineering, Great Southern Press Ltd and Clarion Technical Publishers do not accept responsibility for the views expressed which, although made in good faith, are those of the authors alone.

2. Copyright and photocopying: © 2013 Great Southern Press Ltd and Clarion Technical Publishers. All rights reserved. No part of this publication may be reproduced, stored or transmitted in any form or by any means without the prior permission in writing from the copyright holder. Authorization to photocopy items for internal and personal use is granted by the copyright holder for libraries and other users registered with their local reproduction rights organization. This consent does not extend to other kinds of copying such as copying for general distribution, for advertising and promotional purposes, for creating new collective works, or for resale. Special requests should be addressed to Great Southern Press Ltd, PO Box 21, Beaconsfield HP9 1NS, UK, or to the editor.

3. Information for subscribers: The Journal of Pipeline Engineering (incorporating the Journal of Pipeline Integrity) is published four times each year. The subscription price for 2013 is US$350 per year (inc. airmail postage). Members of the Professional Institute of Pipeline Engineers can subscribe for the special rate of US$175/year (inc. airmail postage). Subscribers receive free on-line access to all issues of the Journal during the period of their subscription.

4. Back issues: Single issues from current and past volumes are available for US$87.50 per copy.

5. Publisher: The Journal of Pipeline Engineering is published by Great Southern Press Ltd (UK and Australia) and Clarion Technical Publishers (USA):

Great Southern Press, PO Box 21, Beaconsfield HP9 1NS, UK• tel: +44 (0)1494 675139• fax: +44 (0)1494 670155• email: [email protected]• web: www.j-pipe-eng.com• www.pipelinesinternational.com

Editor: John Tiratsoo• email: [email protected]

Clarion Technical Publishers, 3401 Louisiana, Suite 110, Houston TX 77002, USA• tel: +1 713 521 5929• fax: +1 713 521 9255• web: www.clarion.org

Associate publisher: BJ Lowe• email: [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 aspects of the subject. Papers sent to the Journal should not be submitted elsewhere while under editorial consideration.

Authors wishing to submit papers should do so online at www.j-pipeng.com. The Journal of Pipeline Engineering now uses the Aires Editorial Manager manuscript management system for accepting and processing manuscripts, peer-reviewing, and informing authors of comments and manuscript acceptance. Please follow the link shown on the Journal’s site to submit your paper into this system: the necessary instructions can be found on the User Tutorials page where there is an Author's Quick Start Guide. Manuscript files can be uploaded in text or PDF format, with graphics either embedded or separate. Please contact the editor (see below) if you require any assistance.

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.comis available for subscribers

Sample

copy

- not

for di

stribu

tion


Editorial

TO START with, let us make clear that the above headline refers to the rate of pipeline failure in

Australia vs that in the rest of the world. The Journal certainly would not dream of making comparisons with any other aspect of life or engineering between that stimulating country (where it is headquartered) and any other community anywhere else!

But in terms of their failure rates, pipelines in Australia are happily different from other pipelines around the world, and the reasons for this are by no means clear. The rates of pipeline failure in Australia are substantially lower than in the Americas and Europe, at only 10-20% of the international mean for failures in onshore transmission pipelines.

At the recent Joint Technical Meeting held in Sydney between the Australian Pipeline Industry Association (APIA), the European Pipeline Research Group (EPRG), and Pipeline Research Council International (PRCI), this issue was examined in depth in a paper from Peter Tuft and Sergio Cunha which is published in this issue of the Journal (on pages 313-320). The authors looked at the validity of the Australian data and then explored reasons for the difference. Some reasons are obvious, such as the relative youth of Australian pipelines which results in a negligible rate of corrosion failure. However, there is no obvious explanation for the markedly lower rate of failure due to third-party damage. It is hypothesized that Australian practices for managing third-party damage may differ in some way and the authors suggest that, given the high social and economic cost of pipeline failures, there should be a comparative study to identify any beneficial differences between the third-party damage protection practices in Australia and those elsewhere.

The background to this paper is the significant fact that that the Australian pipeline industry has been collecting incident data since about 1965. The lines for which data are collected are gas and liquid transmission pipelines that comply with the over-arching Australian pipeline industry standard AS 2885, and with a maximum allowable operating pressure above 1050 kPa. The resulting database contains about 100 fields to record data about the pipe itself, the events causing the incident, details of any damage and repairs, and operating practices (particularly relating to external interference protection). Reporting is voluntary, and data are collected by the APIA from members of its Pipeline Operators Group (POG). Messrs Tuft and

Cunha point out that POG members represent 94% of the total transmission pipeline length in Australia; each year POG member companies submit a signed declaration that they have either reported all incidents or have had none.

In their paper, the authors point out that the mostly likely explanation for the low rate of third-party damage failures in Australia is a potentially different approach to managing such interference. Exactly what that difference might be is not clear and there has been no study that casts light on this area. They go on to speculate on the Australian practices which might contribute to low third-party failure rates, although they are only very tentative in their explanations for the different failure rates pending a time when there is better information on whether practices elsewhere do in fact differ significantly. Despite these reservations, the authors suggest several factors that may be of relevance:

• Australian pipelines in populated areas tend to be patrolled frequently, usually on the ground but sometimes by air. In and around at least two major cities the transmission pipelines are patrolled every day – or every weekday – and in other cities they are patrolled weekly. The incident database contains about 20 near-miss events where patrollers have caught third parties in the act of digging or preparing to dig on the right-of-way (RoW), and there are many more instances where work near the pipeline was forestalled by patrollers before encroaching on the RoW (such off-RoW activity is not reportable, but there is ample anecdotal evidence).

• The ‘One-Call’ or ‘Dial-Before-You-Dig’ system is well used in Australia by third parties with only rare lapses.

• Pipeline marker signs tends to be frequent, conspicuous, and explicit in their warning message.

• Since 1997, AS 2885 has required a safety-management study to be undertaken during design and then reviewed every five years or whenever the environment around the pipeline changes (such as for new urban development). This study is a fine-grained analysis, often on a metre-by-metre basis, of all possible causes of pipeline failure. Threats to pipeline integrity are explicitly identified and mitigated, with great emphasis on protection against third-party damage.

Why is Australia different from the rest?

Sample

copy

- not

for di

stribu

tion


the Japan Roads Association, and in 1984 the American Society of Civil Engineers (ASCE) first published formal guidelines for seismic design of pipeline systems; however, until 2007, there was no specific standard for seismic design of pipeline systems. In that year the Gujarat State Disaster Management Authority (GSDMA) published a standard for Indian application, as the State was the worst one affected during the 2001 Bhuj Earthquake, in which over 20,000 people died; numerous public- and privately-owned oil and gas companies have pipeline networks within the State. The first draft of the standard resulted from a project at the Indian Institute of Technology in Kanpur, sponsored by GSDMA. An analysis of the effect of earthquakes on a continuous pipeline in the state of Gujarat in India was subsequently presented based on the GSDMA report, in which the authors also discuss the design and construction methodology to minimize the effect of loading on the pipeline during ground movement.

In summary, the paper by Guha et al. provides an overview of past performance of buried pipelines during earthquake events, including fault rupture, liquefaction, and seismic shaking, and goes on to illustrate the nascent understanding of earthquake hazards both around the world and in India. Identification of seismic sources and geohazard-prone areas during the early phases of a project allows these data to be incorporated during the design phase. Appropriate site-specific seismic design during the engineering stage can then reduce the risks posed by earthquake hazards on buried pipeline structures.

The September, 2014, issue of the Journal will explore these issues at greater depth: under the guest editorship of Prof. Shawn Kenny – the Wood Group Chair in Arctic and Harsh Environments Engineering in the Faculty of Engineering and Applied Science at the Memorial University of Newfoundland, and a member of the Journal’s Editorial Board – a special issue is being planned around the topic of engineering analysis and design for seismic ground movement. Prof. Kenny is soliciting contributions for this issue, and further details can be obtained from him [email protected] or the Journal’s editor (see page 274).

• While the safety-management study is an engineering process, it may have a cultural side effect: because it is integral to Australian pipeline design and operation it may help keep safety matters – and particularly the consequences of pipeline failure – in the forefront of pipeline engineers’ thinking at all times.

Messrs Tuft and Cunha reiterate that all or none of these factors may be responsible for the low rates of Australian failures due to third-party damage, although they at least suggest possible initial directions for investigation. As they go on to conclude, the absence of a simple explanation for the low Australian failure rate is not a clear-cut and satisfying conclusion. However the objective of their paper is to initiate discussion of whether others might benefit from research into the differences between Australian and international failure rates. The authors believe it provides a convincing case that Australian pipeline failure rates are indeed substantially lower than elsewhere, and hence that investigation of that difference has potential to provide benefits in reducing failure rates in other parts of the world.

How do earthquakes affect pipelines?

Seismic design and engineering of pipelines has advanced significantly in recent decades, although little has been accomplished to address the vulnerability of buried pipelines to seismic hazards. In their paper on pages 335-344, Indranil Guha and his co-authors examine this proposition from the viewpoint of the Indian sub-continent which has a record of earthquakes of varying magnitude and devastation. Since 2003, for example, there have been 18 major earthquakes in the country, averaging around 5 on the Richter Scale, but peaking at over 9. As this paper explains, with new and proposed cross-country pipelines in India, it is becoming more important to understand the effects on buried pipelines of seismic hazards (such as shaking, liquefaction, and fault surface rupture).

India, of course, is not alone in suffering from such tectonic issues. In 1974 the first seismic design code Technical standard for oil pipelines was developed by

Sample

copy

- not

for di

stribu

tion


INDUSTRY STANDARDS FOR pipeline welding generally had their origins in cellulosic welding of onshore pipelines, and this is still the dominant process in some regions. Over the years standards have been adapted

to include new material grades, new processes, and more-demanding applications. Even though mechanized gas-metal-arc welding (GMAW) is now the dominant process for offshore pipelines, and is widely used in some areas of the world for large-diameter long-distance cross-country pipelines, the industry standards still do not fully reflect the subtleties of this process. This results in owners and operators having to issue amending company specifications. Moreover, there have been continual technical developments in equipment and control technology which makes mechanized GMAW ever more sophisticated.

The European Pipeline Research Group identified a need to develop guidelines focused on mechanized GMAW. This paper summarizes a review document produced to form the basis of such guidelines. The document has reviewed the main industry standards and also had input from company specifications. It covers typical equipment and consumables, procedure and welder qualification, typical equipment including ancillaries, production welding, inspection and testing, acceptance criteria, and repair options.

It is hoped that this work will identify best practice across the industry. Based on the initial work it is intended to develop a guidance document and input into national and international standards and others working on pipeline welding requirements.

This paper was presented at the Joint Technical Meeting held between the APIA, EPRG, and PRCI in Australia in April, 2013, and is reproduced by kind permission of the meeting’s organizers.

* Corresponding author’s contact details:tel: +44 191 215 4010email: [email protected]

1 MACAW Engineering, Newcastle upon Tyne, UK2 Gasunie, Groningen, Netherlands3 5G Orbital, Loughborough, UK

Guidance for mechanized GMAW of onshore pipelines

by Dr Robert M Andrews*1, Harry Kamping2, Henk de Haan2, Otto Jan Huising2, and Neil A Millwood3

THE RATE AT WHICH a pipeline can be constructed is largely dictated by the time

taken to produce the first pass (i.e. the root pass). Traditionally, ‘5G’ positional welding of pipeline girth welds has been accomplished using shielded-metal-arc welding (SMAW) with cellulosic electrodes, operated vertically down. This ‘stovepiping’ technique is well suited to pipeline welding, mostly due to its versatility. However, with increasing usage of large-diameter pipelines and higher grades, the effectiveness of the cellulosic stovepiping method is diminished. For L555 grades, most specifications only permit cellulosic consumables to be used for the root and hot pass, with low-hydrogen consumables required for the fill

and cap passes. For grade L690, cellulosic consumables are not permitted, since the risk of hydrogen-assisted cold cracking is too high, and achieving the required overall strength overmatching becomes even more problematic. The most widely used process for girth welding of large-diameter, high-strength steel pipelines is mechanized gas-metal-arc welding (GMAW). This process offers high productivity combined with good mechanical properties at the required design temperatures. Several systems are available in the market, but in essence they all rely on the following features

• precision bevel (narrow gap)• internal clamping• one or more welding heads, or ‘bugs’, mounted

on a band• gas-shielded solid wire

The basic systems use a short-circuit ‘dip’ transfer mode, with either 100% CO2 shielding gas or an argon-CO2 mixture. The process is continually being

Sample

copy

- not

for di

stribu

tion


• (single) tandem wire configurations: two electrically insulated contact tips in one torch with both wires feeding into the same weld pool

• dual tandem torch/wire configurations: two single-tandem torches mounted on one bug producing two weld pools.

A mechanized welding system is defined as one which moves the welding torch along the joint and controls the welding parameters, but allows the welder to have some (a few %) control over speed, position of the torch, and the welding parameters. As systems become more sophisticated, it is possible to control more of the parameters and adapt to variations in joint geometry, and this has resulted in welders having less freedom to override the set parameters. Some would argue that

developed to increase welding speeds and reliability without sacrificing weld-metal properties. In recent years, some of the developments have included:

• dual torch configurations: two torches each with a single wire mounted on one bug

• pulsed GMAW: globular or spray transfer• controlled short-circuit transfer, or ‘surface

tension’ transfer (STT): particularly suited to open-gap root-pass welding

• FCAW (flux-cored arc welding) with a narrow cap bevel (7°)

• seam tracking: electrical or vision based• inverter power sources and control systems:

precise control of power waveforms and welding parameters around the pipe circumference

Fig.1. Some examples of external welding heads for mechanized GMAW/FCAW.

(a.) CRC

(c.) Vermaat

(e.) Saturnax dual torch (f.) Saturne 8 (Photo from Serimax.)

(d.) FCAW

(b.) PWT Mk-I

Sample

copy

- not

for di

stribu

tion


Equipment

Most welding standards might say something about the type of bevelling machines and line-up clamps to be used, but they say very little about the equipment other than that it ‘shall be of a size and type suitable for the work and shall be maintained in a condition that ensures acceptable welds, continuity of operation and safety of personnel’.

By and large, the onus is placed on the contractor to supply equipment which is suitable for the intended task. It is appropriate that the contractor should be responsible for using the right equipment and ensuring that it is maintained in good working order. However, the client should retain the right of inspection and verification of calibration of the welding equipment prior to start of production.

With mechanized welding there is less opportunity for the welder to overcome any shortcomings of the equipment, so that the whole system is important – including technician support. Only items of equipment which are different from conventional pipeline construction are discussed in the following sections:

Pipe-facing machines

It is normal practice for steel pipes to be supplied from the mill with a standard 30° bevel. However, for narrow-gap mechanized GMAW welding it is necessary to apply a bespoke precision bevel designed to suit the particular welding system. This should be machined immediately prior to field welding to avoid mechanical damage in transit or corrosion during long storage periods.

There are many different types of machine for bevelling pipe ends; however, the type which has been found to be most suited to pipeline construction (both on- and offshore) is a hydraulically operated machine which uses floating-head cutters mounted on a rotating end plate (Fig.2). The non-rotating body locates firmly in the pipe

mechanized GMAW/FCAW is becoming more like an automatic or robotic process, with less need for skilled welders to operate the welding heads. There is a trend in some quarters to employ welding-machine operators rather than skilled pipeline welders: whilst this may reduce costs, it is not clear if this is a welcome development. Mechanized GMAW/FCAW is best suited to ‘mainline’ pipe-to-pipe welding where the joint geometry is uniform and an internal clamp can be used. A selection of photographs of mechanized welding heads is shown in Fig.1. Mechanized GMAW/FCAW can be used on pipe-to-flange welds if it is possible to machine the appropriate bevel onto the flange weld-neck.

The EPRG project compared and contrasted the main standards (API 1104 [1], BS 4515 [2], DNV OS-F101 [3], CSA Z662 [4], and AS 2885.2 [5]). Member companies provided copies of their amending company specifications and valuable input was provided by several welding contractors:

Quality and personnel requirements

Although the obvious aspects of pipeline girth welding are the equipment, consumables, and procedures, the successful application of mechanized GMAW is also dependent on the implementation of an adequate quality management system and employing competent personnel to cover all aspects of the welding process. In addition to compliance with any general project-wide QA system, it is recommended that the quality requirements of ISO 3834-2 [6] should be implemented and the key welding personnel should be competent. Pipeline construction by conventional welding (manual welding) is in essence no different from pipeline construction using mechanized welding equipment, with regard to personnel, except for two key ‘players:

• Welding technician – responsible for setting-up and maintaining welding and ancillary equipment. The equipment manufacturer’s training requirements should be followed.

• Welder – qualification of welders will be discussed later in more detail.

Professional welding contractors invest a lot of time and money training welders and technicians in their workshops prior to construction. The time spent perfecting a weld procedure beforehand can save a lot of difficulties later in the field. It is also essential that the ultimate owner or operator of the pipeline has an effective quality management system and personnel in place. Due to the high productivity of these systems it is important that the quality management system responds rapidly to any developing problems to minimize the number of defective welds and the consequent re-work.

Fig.2. Example of pipe facing machine.

Sample

copy

- not

for di

stribu

tion


Some systems combine pipe alignment with internal root pass welding capability, which obviates the need for copper backing shoes (Fig.5). The internal welding machine is fitted with either four, six, or eight welding heads, each welding a portion of the root in the vertical-down direction. Clamping and welding is controlled by an operator at the free end of the pipe. The power, shielding gas, and control information are passed along a reach-rod or umbilical to the clamp.

Welding bands

The welding band is simply a means of transporting the welding head around the girth weld. One band per weld is set on one pipe end, usually prior to ‘stabbing-on’ the new pipe. Once the band is set in place it is usually left there until completion of that weld joint. The band is placed at a set distance from bevelled pipe end; it needs to be rigid when fixed to the pipe and robust enough to cope with being ‘knocked about’, while needing to be lightweight for routine manual handling. Feet, or pads, provide a ‘stand-off’ from the external pipe surface. The geared edge which engages with the drive of the welding head is usually placed on the opposite side away from the weld to minimize damage from spatter.

In principle the welding bands should be suitable for the automatic ultrasonic testing (AUT) scanner head as well, but in practice each system tends to have its own design and there is limited compatibility.

Welding head or ‘bug’

The main function of the welding head, or bug, is to carry the welding torch(es) along (or around) the weld joint. The head incorporates motors for providing travel, lateral movement, oscillation, and contact-tip-to-work

end ensuring good end squareness. Carbide cutting tips are set into the floating heads which track the internal surface of the pipe as the end plate is rotated (Fig.3). Most contractors fit a cup wire brush onto the floating head in order clean any loose dirt or millscale from the internal surface of the pipe prior to welding.

Internal line-up clamps

An internal line-up clamp is an essential item of equipment for mechanized pipeline welding. Generally, they are used for joining components of equal thickness, i.e. pipe-to-pipe welds, and are not suitable for pipe-to-fitting welds which involve a change in internal diameter across the joint. Internal line-up clamps are usually capable of passing through a cold field bend, but not a pulled bend or induction bend. Internal line-up clamps tend to be pneumatically operated instead of using hydraulics, thereby avoiding the risk of oil contamination in the pipe near the joint preparation. The larger clamps have a reasonable capability to re-round the pipe ends (albeit elastically) thus aiding joint fit-up. Most clamps have motorized wheels to propel themselves inside the pipe.

Most mechanized GMAW/FCAW welds where all the passes are deposited from the outside of the joint rely on internal line-up clamps having retractable copper backing shoes to provide adequate support for the molten weld pool of the root pass. Some recently developed systems are able to weld a closed root without backing shoes. The radius of curvature of the shoes is chosen to match that of the internal surface of the pipe (Fig.4). The merits and potential issues of copper shoes are discussed later in this paper.

Fig.3. Pipe-facing machine showing rotating plate, floating-head cutters, cutting tips, and cup wire brush.

Fig.4. Copper backing shoes in use on a DN1200 internal line-up clamp.

Sample

copy

- not

for di

stribu

tion


Power source

Early mechanized GMAW systems used constant-voltage power supplies with dip-transfer mode. Nowadays, most systems use inverter power supplies which provide a more stable power characteristic and are more suited to pulsed GMAW.

Welding machines need to be properly grounded to avoid the occurrence of stray arcs. Traditionally, the earth return for onshore SMAW pipeline construction was a saddle with spike which was prodded into the girth weld bevel. This method has been somewhat refined for mechanized GMAW, but the quality of the earth return path is often poor. This can be problematic for high-performance pulsed-GMAW systems which use measurements of arc voltage to maintain contact-tip-to-work distance and seam tracking. Some contractors use bespoke earth-return clamps which provide a better contact path.

The welding return cables need to have sufficient cross-sectional area to cope with the maximum current and anticipated duty cycle. It is also important to ensure the length of cable is minimized and kept constant so that the voltage drop is a fixed variable. Weld-procedure qualification testing should simulate the cable lengths that are to be used in the production system.

Consumables

Filler wires

Welding consumables need to be selected to provide the specified mechanical properties and adequate resistance to degradation from pipeline contents and intended operating conditions. This requirement applies

distance adjustment. Power, shielding gas, filler wire, and control information are carried via umbilical cables to the welding head. Most modern mechanized GMAW systems have a pendant console which allows the welder to control the run sequence, start/stop commands, lateral position, and travel speed. There are many variants of welding head, which can be broadly classified as:

• single torch• dual torch• tandem wire• hybrid laser GMAW

Wire-feed unit

The wire feed can be either on the welding bug, or a separate unit mounted on the inside wall of the welding shack. The main advantage of having the wire feed motor on the bug is to minimize the distance between wire feeder and arc, resulting in better wire speed control and hence arc stability. However, this does limit the user to small spools or ‘bobbins’ of wire, which tend to be more expensive and have to be used with a wire straightener due to the tighter radius of curvature or ‘cast’.

Most modern systems use wall-mounted wire-feed units which means the wire has to be fed through an umbilical cable. The advantages of this arrangement are that it reduces the weight of the welding head and allows larger, standard, off-the-shelf spools of wire to be used. For onshore pipeline construction it is normal to use 15-kg spools, whereas on some laybarges much larger packs of wire are convenient. The larger spools mean less downtime due to changing of spools.

The smaller ‘bobbins’ tend to make use of 0.9-mm or 1.0-mm diameter wire, whereas for remote wire feed units, which rely on having to ‘push’ the wire for a longer distance, it has been found that 1.2-mm diameter wires are more successful.

Controller

Modern microprocessor systems enable rapid communication between power source, wire-feed unit, and welding head. They can be programmable in the field, but access needs to be limited to the welding engineer and/or nominated technicians. Due to the increasingly large amount of information contained in each procedure programme, it is normal practice for the signed, approved, weld procedure to include the name and unique version number of the control software.

Measurement of welding parameters can be inbuilt or independent. Generally, for third-party verification, it is preferred to have independent measurement and monitoring of parameters.

Fig.5. CRC internal-welding machine (DN1000).

Sample

copy

- not

for di

stribu

tion


having BS EN 10204:2004 type 3.1 certification where each batch is tested by the manufacturer. For demanding applications it is also common practice to insist that production welding is performed using the same batch(es) that were used for welding-procedure qualification testing (WPQT): whilst this may be a laudable requirement, in practice it is not always easy to achieve.

One of the companies surveyed as part of this project commented on the batch-to-batch variations observed with flux-cored wires. Its policy is to perform the equivalent of a weld-procedure qualification test for each batch of flux-cored wire.

Handling and storage

It is important to observe the manufacturer’s recommendations for handling and storage of consumables. For filler wires this principally involves keeping them in a humidity-controlled room or container to prevent moisture pick-up and rusting. Packing and re-packing boxes of consumables should be avoided to reduce the risk of mechanical damage. For systems which make use of small reels (or bobbins) it is important to ensure reeling is done by the consumable manufacturer. Transfer in the field from large reels to small reels should be prohibited due to concerns over process control.

Shielding gas

100% carbon dioxide, or a mixture of argon and carbon dioxide, are usually used for mechanized GMAW welding of C-Mn steels. Carbon dioxide is an active gas and tends to produce a hotter arc: this is good for ensuring root penetration and sidewall fusion, but more spatter is formed and weld pool control is not so good for the cap pass. Pulsed GMAW normally uses an 80%-Ar 20%-CO2 mixture. In addition to effects on weld-pool fluidity and amount of spatter, several studies have shown that the mechanical properties are affected by changes in shielding-gas composition.

Principally for safety reasons, gases must be stored in the containers in which they were supplied, and no inter-mixing is permitted in the containers. Pre-mixed gases are more commonly used for onshore pipeline construction, and although more expensive, they provide good reliability with minimal effort. Offshore pipeline construction barges or vessels tend to mix the gas on demand, probably because it is easier to permanently install the pipework and mixer units.

During the WPQT and pre-production phase it is important to ascertain what operational procedures are in place to control the storage, handling, and usage of shielding gases. Similarly, it is important to assess the competency of the welding technicians with regard to the shielding-gas systems. Regular checks should be

for any welding process, manual or mechanized, but there are additional issues for automated GMAW pipeline welding.

The standard consumable classification groups tend to be quite broad allowing considerable variation in chemistry. For demanding mechanized GMAW applications it is accepted that small changes in the concentration of alloying elements and/or residual elements can have a significant effect on technological properties, hot cracking susceptibility, bead profile, and arc stability. Some contractors apply their own additional requirements – defining steel de-oxidation practice, placing limits on alloy additions, and ‘tramp element’ concentrations and setting guaranteed mechanical properties.

Surface quality, coating thickness, cast (the diameter of a loose turn), and helix of the filler wire are important factors which can affect the ‘feedability’ and contact resistance. Surface quality can be degraded if the storage and handling procedures are not adhered to. Most filler wires are coated with a very thin layer of copper. The coating thickness is difficult to measure and if it is not thick enough it will affect the electrical contact between the wire and the contact tip. The spool size will affect the amount of straightening required and hence the amount of physical resistance to the wire-feed motor.

It is widely agreed that strength overmatching is beneficial and, in fact, it is a premise for adoption of the EPRG and most other alternative weld-defect-acceptance criteria, as discussed below. It is important to base the overmatching requirement on the narrow gap weld and not the consumable batch test; this is because the measured narrow gap all-weld yield strength is usually much higher (up to 150 N/mm2). The level of overmatching varies from standard to standard, but most require 5% to 10% above the SMYS of the parent material. One company standard requires overmatching of the actual strength (of all the pipes), which essentially means SMYS plus 150 N/mm2. Overmatching is not really an issue for low grades, but for high-strength steels (≥ L555) this can pose significant challenges.

Batch testing

Batch testing is usually the responsibility of the consumable manufacturer. The testing is performed according to a procedure which minimizes the effects of parent metal dilution, i.e. in a very wide groove or a built-up ‘pad’ of weld metal. Whilst these tests are simple and repeatable, they bear little relation to the properties achievable in a narrow-gap weld. Guidance on the number and type of verification tests is given in AWS A5.01 and EN 14532-1.

There are different levels of batch testing and certification depending on the demands of the application. Most projects involving mechanized GMAW use consumables

Sample

copy

- not

for di

stribu

tion


• arc characteristics – spray arc, globular arc, pulsating arc, or short-circuit arc;

• contact-tip-to-work distance;• contact tip – type, and size/dimensions;• torch angle(s);• separation distance between wires (for dual-torch

and tandem systems).

This could mean that, for current systems, the WPD would consist of a program identification, a sketch of the layer build-up, and the related pre-choices the welder has to make on his remote. That would be sufficient for the welder in the field. But all standards require a minimum set of parameters to be addressed in the WPD. Some of these are relevant, some less interesting, but some important values are not required, as is discussed in the following section.

Essential variables

A list of the welding parameters seen as essential by the authors of this paper for mechanized GMAW/FCAW welding, is given in the Appendix. Also included in the Appendix are the parameter requirements as addressed in the codes BS 4515-1:2009, API 1104 20th edition, and ISO 15614-1/ISO 15609-1. In addition to the usual variables for a girth-welding procedure such as preheating, interpass temperatures, thickness, and chemical composition, the list includes others specific to the mechanized process. These additional variables include the information listed in the preceding paragraph. For multiple-wire (dual-torch and/or tandem-wire) systems, the number of wires and their spacing should also be treated as essential variables.

Heat input is controlled, since it influences the weld profile and incidence of defects such as lack of fusion. The welding technician is more interested in arc energy since that is something that can be controlled. Arc energy is given by voltage multiplied by current divided by travel speed. The amount of arc energy ‘absorbed’ by the metal will depend on various factors, but generally for GMAW it is assumed that 80% of the arc energy is available as heat input; the rest is dissipated by radiation or convection. Although other factors also influence the effective heat input, wire-feed rate also has a large influence on the heat-affected zone profile and on the risk of burn-through or lack-of-fusion defects.

Heat input is a derived parameter, so whilst it is interesting to limit its fluctuation, it is more important to control the primary parameters which contribute to arc energy. The correct way to calculate arc energy is to use the instantaneous values of arc voltage, current, and travel speed. The minimum, maximum, and average values of arc energy can then be reported for each pass. Minimum heat input has in some cases been

performed during production to confirm the actual flow rate of shielding gas exiting the torch(es). A portable measuring device, plugged onto the end of the torch-gas shroud, can be used. It is also important to clean the torch-gas shroud regularly as the build-up of spatter can have an adverse effect on the shielding-gas flow pattern.

Procedures and procedure qualification

Procedures for mechanized GMAW

The weld procedure defines the parameters that control the production of the weld. The question of how much information to state on the weld procedure document (WPD) is one which causes discussion amongst welding engineers from time to time. The simple answer is that it should contain all the information necessary for the welder to be able to make welds which are demonstrably the same as those which were subject to non-destructive and destructive testing.

Most modern mechanized-welding systems now have segment control which means parameters can be changed depending on the angular position around the joint. Systems like this require a large set of pre-programmed parameters which are not easy to present on a printed WPD.

Another complication is the increased use of pulsed GMAW, which requires several parameters to define the waveform, but in most cases the average (mean) parameters are only reported on the WPD. This information is almost meaningless. So, without making the WPD over-complicated, the most sensible approach is to state the name and version number of the control software.

The following information specific to mechanized GMAW welding should be stated:

• control software – programme and/or software version;

• list of pre-set welding parameters that cannot be adjusted by the welder;

• list of pre-set welding parameters that can be adjusted by the welder (i.e. ‘hot key’ limits);

• wire-feed speed for each pass;• oscillation width, frequency, and dwell time for

each pass;• calculated arc energy for each pass using the

recorded values of current, voltage, and travel speed without the addition of a percentage. For multiple, electrically isolated, tandem-electrode welding processes the effective arc energy per run must be calculated as the sum of the individual energies for each electrode. For pulsed welding, the effective arc energy shall be calculated on the basis of RMS (root mean square) values;

Sample

copy

- not

for di

stribu

tion


This should also apply to any welds which are made at a later date for additional testing, for example for corrosion or full-scale tests.

Inspection and testing

There is often a question whether inspection methods and techniques during qualification should be same as in production or whether they should be more onerous. For example, during WPQT one usually has access to the internal surface (root) which would enable magnetic-particle inspection (MPI) to be carried out. Similarly there is more freedom to be able to grind off the root bead to aid MPI, or to grind off the cap bead to aid ultrasonic testing (UT).

If automated ultrasonic testing (AUT) is being performed during WPQT it may be appropriate also to carry out radiographic testing (RT) in order to provide comparison inspection data. Test-ring sizes are more manageable, which means radiography can be performed in a purpose-made bunker, or the welds could even be subjected to immersion UT.

Welder qualification

The purpose of the welder-qualification test is to demonstrate that the welder is able to make sound welds using the particular procedure and equipment that is to be used in production. Although manual control of the weld pool is not performed by the welder, it is important to remember the welder will need to have an insight into weld pool behaviour and the influence of the movement of the torch on the weld pool. So either a trained manual welder will be able to be qualified within a limited period of time, or a considerable time will be needed to train the operator to get to know the weld-pool behaviour. The welder-qualification test should be performed in the required principal position (5G, etc.) on full pipe joints or pup pieces. API 1104 [1] 20th edition allows segments or coupons of pipe to be used. However, this should not be permitted for qualification of welders on mechanized GMAW/FCAW systems, as it is not representative of production conditions. For positional welding, each welder must complete at least 50% of the pipe circumference. The ends must be sealed until at least the root bead and hot pass have been completed around the full circumference in order to prevent draughts interfering with gas shield and possibly also affecting the weld-cooling rate.

The welder-qualification test should be performed using the same, or equivalent, equipment as will be used for production welding. The welder should also be able to demonstrate competency in the correct use of the equipment, for example how to set the band on the pipe, change wire spools, change contact tips, clean the gas shroud, check wire-feed speed, and use of grinder

erroneously calculated by taking minimum voltage for the whole pass, multiplying it by minimum current for the whole pass, and dividing it by maximum travel speed for the whole pass. Similarly, maximum heat input has been calculated by taking the maximum voltage and current and dividing by minimum travel speed. Clearly, this is physically incorrect and can give rise to unqualified and impractical extremes of heat input, and should be avoided.

In addition to the essential variables, it is important to consider whether to impose a time limit on the validity of a WPD to account for technological advances in equipment, pipe materials, and consumables which might render the WPD invalid. Specifically such a time limit is recommended when welding materials with a yield strength of 450 N/mm2 and above, where any changes in the welding consumables might have a negative influence on the as-welded properties of the weld metal.

Procedure qualification

The purpose of qualifying a weld procedure is to demonstrate that production welds made in accordance with that procedure will have the required mechanical properties and be of sound quality. There are various approaches which can be taken with regard to qualification of weld procedures depending on the specific project application. Australian Standard AS 2885.2 describes four methods, namely: (i) qualification by testing, (ii) qualification by documentation of previous testing and approval, (iii) qualification by pre-qualification without testing, and (iv) qualification by the use of supervision. This is also a possibility with the various BS EN ISO standards (i.e. BS EN ISO 15607, 15609, 51613, and 15614).

For new pipeline projects, the default requirement is option (i), qualification by testing. Option (ii) may also be adopted for short pipeline projects, such as diversions, subject to consideration of essential variable restrictions. It may also be possible on occasion to consider the other alternatives.

Most standards stipulate that a qualified WPD of a particular contractor is valid for welding only in workshops or sites under the operational technical and quality control of that contractor. In other words, a transfer from one contractor or sub-contractor to another is not permitted without re-qualification. This is a good requirement which ensures the in-house knowledge and expertise which is implicit in a WPD is maintained. In reality, for mechanized welding systems, it is not realistic for third parties to use someone else’s pre-qualified procedure as it is unlikely they would have all the correct equipment and know-how.

Sample

copy

- not

for di

stribu

tion


Good pipe-end tolerances are likely to give better fit-up which should result in better productivity, lower repair rates, and greater longevity of copper backing shoes on the internal line-up clamp. Typically, there is a ± 1.6-mm tolerance on diameter for the pipe sizes of interest to mechanized GMAW. Usually, the tolerance on diameter and out-of-roundness is greater for the pipe body than for the factory-supplied ends. This can cause problems when a cut end is used for production welding.

Mechanized GMAW systems rely on very specific, precise, narrow-gap bevels which need to be applied just prior to welding in the field. It is not a good idea to bevel the pipe ends too far in advance especially if there is a possibility of the machined surfaces rusting. Carbide cutting tips can produce very reliable radii in the preparation, but it is important to check for wear, especially in higher-grade steels which are harder. If it is not possible to machine the bevel, for example at a tie-in, it may not be possible to use mechanized GMAW welding, and other processes such as FCAW are often used. Mechanized FCAW processes can be used either in the ‘traditional’ V-bevel configuration or in an narrow gap bevel as a fill and cap. In these cases, a mechanized or manual solid-wire process or SMAW is utilized to weld the root pass.

Some methods of overcoming residual magnetism in the pipe ends are not practical for GMAW, and it may be necessary to use a dedicated demagnetizing unit and coil. Stray arcs tend not to be an issue with mechanized GMAW/FCAW systems since the supply of arc current to the electrode is controlled by the machine. There are various interlocks to provide protection when cleaning the torch or changing a contact tip. If an arc burn does occur despite these precautions then remedial action is required according to an approved procedure.

Weld bands should ideally be placed on bare pipe surfaces although, for various reasons beyond the control of the welding contractor, this is often not possible. It is often the case that the ‘feet’ of the bands are either both on the coated pipe surface, or half on and half off. This can cause problems with alignment and stability of the welding band, as well as damage to the external anti-corrosion coating, usually in the form of indents into the softened coating. In extreme cases, this can result in additional repairs having to be made at the field-joint coating stage.

When using an internal line-up clamp with copper backing shoes it is also normal practice to heat up the shoes with a flame torch at the start of the day, or after any prolonged delay, in order to eliminate condensation or moisture and prevent rapid cooling of the root pass on start up.

and power brush to clean and/or dress the weld. Other functions such as checking the earth clamp, gas settings, power source settings, and connections are arguably the responsibility of the technician.

Comprehensive training prior to testing and ‘tool-box’ talks during production should be used to convey the importance of good welding practice and the specification requirements applicable to the project.

There is a trend in some quarters to employ welding-machine operators rather than skilled pipeline welders. However, it can be argued that this downskilling is not appropriate. All ‘mechanized’ welders should also be capable of producing good manual welds. Manual welders control the melting pool by adjusting welding parameters and have the ability to look into the pool and see if something wrong. Operators don’t have these abilities. So, prior to qualification welding, the welders search for the optimal welding parameters in order to produce good-quality welds without defects and with good mechanical properties and a high productivity. In case of quality issues in production (due to weather, machine or consumable problems), the welder is the key person to solve the problems. Operators who just push the buttons are not capable of doing this.

Production welding

The clearance between the bottom of the horizontal pipe and the ground, or floor of the welding shack, should be sufficient to allow unimpeded access for the welding head, as well as physical access for the welder to see the weld pool and to carry out any remedial grinding. After completion of welding there should also be sufficient access for visual inspection and other non-destructive testing methods which rely on equipment travelling around the outside of the pipe. Generally, it is accepted that a minimum clearance of 400 mm is sufficient.

Another aspect to consider is the clearance between the welding and inspection equipment and any external pipe coating, or concrete weight coating. The ‘cut-back’ distances, with suitable tolerances, need to be stated in the linepipe coating specification. It is generally accepted that a cut-back distance of 115 ± 15 mm for the anti-corrosion coating will provide sufficient access for welding and subsequent phased-array AUT inspection. For projects involving heavy-wall pipe, the cut-back distance may need to be increased. The cut-back distance for the concrete weight coating for offshore pipelines tends to be around 350 mm. Again, for very thick concrete weight coatings, it is possible that the cut-back distance may need to be increased to avoid interference with the welding and/or inspection heads.

Sample

copy

- not

for di

stribu

tion


be a cause of cold laps and may also mask missed-edge defects from being seen during visual inspection. For this reason, it is good practice to power brush the weld just prior to making the cap pass. When using the FACW process, interpass cleaning is done by power brush to remove the small amount of slag from this process.

After completion of the weld, it is also necessary to remove all slag and weld spatter from the weld and adjacent pipe surfaces. In addition to being a sign of good workmanship, it is important to have a smooth surface for the AUT probes to pass over unhindered, and for good field-joint-coating integrity. Cleaning is usually accomplished using a power brush, although other tools may be required. Excessive grinding of the pipe surfaces should be avoided, as this may also impede the effectiveness of the AUT inspection.

Most, if not all, mechanized GMAW/FCAW systems have the capability to measure and record the welding parameters relative to the weld pass and angular position. The information can be interrogated in real time, or afterwards, to confirm whether the weld was made in accordance with the approved WPD, or not. With advances in pulsed welding, segment control, spatter-free ignition, seam tracking, and through-arc sensing, a huge amount of data is generated. This cannot possibly be assimilated by an inspector. However, it is feasible automatically to evaluate the data and present the data as either a ‘pass’ or ‘fail’.

Real-time data and event logging can also be performed to record other activities such as manual override commands for seam tracking and travel speed, delays between passes, stop/starts, short circuits, and tip change-outs. It is conceivable that in the future this information, together with a vision sensor to measure weld bead profile and ‘hi-lo’, could be used in place of non-destructive testing.

Repair of defective welds

When a discontinuity or indication is deemed to be outside the acceptance criteria, the contractor is faced with the choice of either making a weld repair or a ‘cut-out’. Remedial grinding of surface-breaking discontinuities is not usually counted as a repair. The selection of repair method and consumable should take into consideration the metallurgical effect of the repair on the original mechanized GMAW weld deposit.

Methods

Repair-welding processes need to be flexible to cope with a variety of geometrical configurations and non-ideal site conditions, and for these reasons, the mechanized systems are generally not suitable for repair welding. The most commonly used processes for repair

One company standard states that a minimum of two welders is required for pipe sizes greater than or equal to 457 mm OD. However, most – if not all – mechanized GMAW welding with ‘bug-on-a-band’ systems use two or more welders welding on opposite sides of the pipe to maintain a balanced deposition sequence, particularly for the root and hot passes. The fill and cap passes could be welded with one welder, but this would not be economic. Internal welding machines make use of four, six, or eight welding arcs to make the root pass, but are controlled by one operator at the end of the pipe. Machine-mounted external welding systems, such as the Saturne-8 system, control multiple welding heads simultaneously, avoiding interactions between welders.

Some contractors allow both welders to start at top dead centre (TDC) for vertical-down progression. Other contractors adopt the practice of one welder starting at TDC, whilst the other welder starts at 3 o’clock, welding to 6 o’clock, returning to TDC and finishing the root pass at 3 o’clock. This is more effective in terms of maintaining a balanced sequence, but it does mean that there is always a stop/start at the 3 o’clock position. Some standards require a root stop/start to be qualified during WPQT. Some contractors use more than two welders on each side (simultaneously) on pipe sizes over DN500. This can be achieved, but requires careful organization.

The time limit between the start of the root and the start of the hot pass is a throw-back to the days of cellulosic welding where there was a real risk of cold cracking. For mechanized GMAW welding it is normal to specify a maximum time limit between the completion of the root pass and the start of the hot pass. But even the necessity of this variable depends on the actual thickness of the root pass compared to the pipe thickness. If the root has sufficient thickness, i.e. 25% of the wall, there is no need put a restriction on the time between passes

It is particularly important for GMAW and FCAW welding to remove clusters of surface porosity, stop/start overlap, high points in the bead and other visible defects before deposition of the next pass. This is because GMAW is a low heat input process and only has limited capacity to ‘burn out’ defects from previous passes. Vertical-up flux-cored welding with a rutile wire is more tolerant, but even so, it is better to grind-out known defects. For narrow-gap GMAW or FCAW, grinding needs to be done very carefully so as not to damage or alter the bevel uniformity which in itself could cause more weld defects.

GMAW is accepted as a non-slag forming process; however, in heavy-wall, multi-pass, welds it is common to see a build-up of a thin silicate ‘glassy’ slag. Generally, the effects are benign, but this fine layer of slag can

Sample

copy

- not

for di

stribu

tion


input may produce an unacceptable microstructure in the heat-affected zone. Most pipeline standards also require qualification testing of partial-penetration repair welds and other options, such as back weld repairs.

The position selected for qualification of repair welds should simulate the most onerous conditions. For example, one end of a partial/full-penetration weld should be at the 6 o’clock position on a fixed-position 5G weld, and a vertical-down single-pass cap repair should be made at the 3 o’clock position. For qualification of partial-penetration repair welds, it is normal practice to leave some of the original (mechanized GMAW) weld so that the fusion boundary between the repair and original weld can be tested (usually using Charpy testing).

The qualification of repair welds should be subject to at least the same levels of inspection and mechanical testing as the original girth welds. In this context it is worthy of note that AS 2885.2 [5] does not currently require any mechanical testing of repair welds, instead relying only on macro examination and hardness testing. In the authors’ opinion this is not acceptable for mechanized GMAW welds in high-strength linepipe.

Inspection of welds

Visual inspection

A challenge with most mechanized GMAW systems is to keep the cap height within specification limits (typically 1.6 mm at the 5-7 o’clock position on 5G positional welds). The main reason for limiting the cap height is to avoid problems with the integrity of the field-joint coating. Welding systems according to the current developments, like segment control and pulsed weld parameters, are more able to control cap bead height at the bottom of 5G welds. Making use of a ‘split cap’ reduces the width of oscillation and helps to control the size of the weld pool. This split cap is, depending on the weld preparation and wall thickness, often used in practice by twin-wire and dual-torch systems.

Typically, visual inspection on production welds is limited to the external cap profile. However, whenever possible, the root should be inspected as well. In some instances and depending on safety rules, it may be possible to climb inside the pipe, but nowadays several camera-based vision systems are available, which allow rapid and reliable visual inspection of the root-pass profile. These systems are vital for CRA pipelines where it is not possible to make full-penetration repairs. Internal visual inspection is performed after the root or hot pass has been completed, and if a defective profile is found, then the whole weld is cut out and re-made. On a laybarge this saves the time and effort of having to back-up the barge just to re-make one defective weld.

welding are (i) basic coated low hydrogen SMAW, (ii) gas-shielded semi-automatic FCAW, and (iii) self-shielded semi-automatic FCAW. Cellulosic electrodes are generally not permitted for repair welding. Some contractors use semi-automatic STT for the root pass of full-penetration repair welds.

Removal or excavation of the defective weld material can be by grinding and/or arc air gouging. The latter method is preferred since it is very quick and often it is possible to see the defect as the hot metal is ‘scooped’ out. For a full-penetration (or ‘root’) repair it is normal practice to gouge to within 2 or 3 mm of the root and to grind-out the remaining metal. It is recommended that the repair welder performs the gouging and grinding, since it allows him to visualize the extent of the repair and mentally prepare for the subsequent welding.

The repair excavation length is usually restricted to 20% of the pipe circumference for full-penetration repairs, and 30% for partial penetration repairs. However, the designer or welding engineer should assess each case to determine whether the stresses acting on the pipe may place tighter restrictions on the excavation length. This is particularly likely to be the case for deepwater offshore installations where considerable lengths of pipe may be hanging from the laybarge. Engineering critical assessment methods have been used to determine allowable excavation lengths, taking advantage of the blunt shape of the groove to disregard fracture and consider only plastic collapse.

Most standards also place limits on the proportion of a girth weld which may be repaired and how many attempts may be made at a repair. For example, BS 4515-1 restricts the cumulative repair lengths to 20% and 30% for partial-penetration and full-penetration welds, respectively. Only one attempt at a full-penetration repair is permitted, whilst a second attempt at a partial-penetration weld repair may be permitted.

Qualification of repairs

Qualification of repair welds is normally performed as part of the WPQT programme for mainline mechanized GMAW welds. The number and type of repair-weld configurations should be established beforehand. Retrospective qualification of repair welds should not be permitted, as this is considered bad practice due to the risk of repair-weld-procedure qualification failure which places unnecessary pressures on the project.

As a minimum, it is necessary to qualify a full-penetration repair and a single-pass cap repair where this is to be permitted. Qualification testing of a single-pass cap repair is more important if a vertical-down technique is being used, since the low heat

Sample

copy

- not

for di

stribu

tion


over several years. The orientation of lack-of-fusion defects on the sidewall of narrow-gap welds is generally quite favourable for detection by this technique, but other planar defects, such as cracks, are difficult to detect. The probability of detection (POD) of radiography related to planar defects is rather low. Other consequences of radiography are safety (radiation), environment (chemicals), and efficiency (examination time and wait for inspection results). Since the image on the film is a two-dimensional representation of a 3-D object, it is not possible to reliably ascertain the depth and height of indications in the weld. For this reason, conventional film radiography is not ideally suited to alternative acceptance criteria which rely on depth and height information as well as length. The acceptance criteria of radiography are based on good workmanship originally.

In the past decade, radiographic inspection of pipeline girth welds has largely been overtaken by AUT (automated ultrasonic testing), which is claimed as having better reliability (i.e. POD) with a focus on critical defects, being more suited to alternative acceptance criteria, providing a permanent record and to be safer as it does not use ionizing radiation. However, in recent years, advances in digital radiography, with lower radiation intensity due to the sensitivity of digital detectors, have led to a resurgence in interest in radiography as the primary means of girth-weld inspection, especially suited for austenitic materials and CRA pipelines.

Automated ultrasonic inspection

AUT systems using a combination of pulse-echo probes (fixed-array or phased-array) and time-of-flight diffraction techniques are favoured for inspection of pipelines. These AUT systems can inspect a girth weld very quickly. Before use on a project, the proposed AUT system should be qualified in a manner analogous to weld-procedure qualification.

At this moment there is ongoing discussion about the application of AUT, such as: identifying critical variables and how they affect AUT inspection; understanding the technical capabilities of AUT systems regarding POD; and sizing accuracy and evaluation of the physical limits of AUT inspection, such as minimum detectable f law size and inspection frequency.

Pipeline owners and operators need a better understanding and reliability of the application of AUT systems focused on their specific pipeline project. A guideline for specification of AUT systems has to be developed for cross-country and offshore pipelines. Standards like API 1104, DNV-OS-F101, ASTM-E-1961-11, and selected company specifications,

Additional visual inspection related to GMAW welding

It is not just the completed welds which need to be inspected. Copper components (contact tips and shoes from internal line-up clamps) should be inspected regularly for signs of damage or wear during production. Poor fit-up and non-optimized bevel or welding parameters can cause excessive penetration of the root bead onto the copper shoes. This reduces the useful life of the shoes, and evidence of this can be seen on shoes with pits on the surface and blackened surfaces. In extreme cases, the shoes can stick to the root bead preventing them from being retracted by the pneumatic actuator or even tearing out part of the root bead.

Contamination from the contact tip can occur either from direct contact between the tip and the bevel wall, or from ‘burn-back’ when the arc is switched off. If, for some reason, copper contamination does occur from the contact tip, it will inevitably result in ‘copper cracks’ being formed (Fig.6). All welding and NDT specifications stipulate that if cracks are found during inspection, the weld shall be cut-out and remade. However, with regard to copper cracking, some pipeline specifications permit this defect to be repaired. Copper cracking can be difficult to detect and the possible presence of an adjacent embrittled but uncracked copper-containing area must be considered if copper cracks are to be repaired.

Radiography

Panoramic X-radiography has been used successfully on many thousands of mechanized GMAW girth welds

Fig.6. Example of copper cracking due to contamination from the contact tip.

Sample

copy

- not

for di

stribu

tion


to the applied loads and the material properties. For new pipeline construction, these alternative criteria are usually derived during the design stage using generic assumptions for the loads and material properties. The assumed material properties would then be an input to the specifications for linepipe procurement and the development of welding procedures. It is also possible to use alternative acceptance criteria to assess a specific defect taking account of the actual loads and defect position both along the pipeline route and around the pipe circumference. This may occur after a post-construction audit, or if the service conditions change during the life of the pipeline.

In principle either workmanship criteria or fitness-for-purpose approaches could be used to set acceptance criteria for mechanized GMAW girth welds. Generally, fitness-for-purpose criteria are used with mechanized welds as they are able to take advantage of the through-wall height information provided by the AUT inspection systems which are generally used with mechanized GMAW. An example of the integration of AUT and defect-acceptance criteria is given in [9] which describes the determination of AUT sizing errors and the use of wide-plate testing to determine defect-size limits.

Pipeline-specific alternative criteria are given in CSA Z662 (Annex K for general use and Annex J for assessing a specific defect), AS 2885.2, and Appendix A of API 1104. The European Pipeline Research Group has developed a set of guidelines for weld-defect acceptance [10]. Originally aligned with workmanship criteria, recent work [11] has extended the Tier-2 requirements to make them more usable with mechanized welding. A common feature of these criteria is that the weld metal should overmatch the parent pipe strength and have a minimum level of toughness. The intention is to ‘shield’ any defects in the weld from high stresses as the lower-strength pipe will yield before the weld and so limit the stress on the defect. The practical application of this apparently simple requirement is difficult due to factors such as variations in both pipe and weld metal actual strengths. For mechanized GMAW, the narrow gap complicates testing, as it is difficult to ensure that an all-weld-metal tensile specimen does not contain some parent metal. It is also arguable whether a small-round-bar tensile specimen fully reflects the performance of the weldment.

The pipeline-specific assessment criteria usually have limitations in their range of application, due to the underlying test data used to calibrate them. For example, Tier 2 of the EPRG Guidelines is currently limited to C-Mn steels up to grade L485, 0.5% total axial applied strain, and require a weld metal Charpy impact energy of 40 J. If the project conditions or

have to be analysed as a reference for this guideline. A technical justification is required as part of the development of the guideline.

Another important issue is the interpretation of the AUT system output (signals). This should be carried out by personnel who have a thorough knowledge of the AUT system and an understanding of the welding system being used. They should hold a Level II qualification in ultrasonic testing and have successfully completed a full training course. The guideline shall specify the minimum requirements for these personnel.

Inspection and acceptance of repair welds

Production repair welds should be inspected by the same NDT methods as were used to inspect the original welds, as well as any methods which may be more suited to the repair weld configuration. Generally, this means that the repair welds are inspected by a combination of AUT – to check that the original defects have in fact been removed – and manual UT and/or ToFD. The manual inspection is required because the AUT system will not be optimized to examine the variable position of the repair weld fusion line. Even though the repair weld may be inspected by AUT, the acceptance criteria should either be based on workmanship standards or an ECA. If using workmanship criteria, the repair weld must achieve mechanical properties, particularly for toughness, that are consistent with those that would be required when workmanship standards are used. An ECA approach must use fracture-toughness and yield-strength data obtained for the repair weld; it cannot be assumed that data from the original GMAW testing can be used. Guidance on mechanical testing of repairs is given in standards such as OS-F101, Appendix B [3].

Defect-acceptance criteria

The ubiquitous workmanship criteria have served the pipeline construction industry well over the years. However, they are empirical and take little account of (i) the material properties of the parent pipe and the weldment, (ii) the actual service conditions, (iii) the height of the defect, or (iv) whether the defect is embedded or surface-breaking. Due to their empirical nature, it is not always clear how conservative they are for new high-strength materials or for arduous service conditions such as high strains beyond yield or severe fatigue loading. Workmanship criteria also do not fully complement inspection methods such as AUT.

With the development of fracture mechanics and fitness-for-purpose methods it has become possible to develop alternative acceptance criteria which can analyse the severity of defects in a quantitative manner. Thus it is possible to relate the allowable defect size

Sample

copy

- not

for di

stribu

tion


the international standards to descriptive documents will not lead to better welds; knowledge and QA/QC in the field will.

Acknowledgments

This project was carried out with the support of the EPRG Materials Technical Committee, with Dr Chris Thornton of BP acting as project manager. This is gratefully acknowledged. The authors also acknowledge the support of the EPRG member companies and welding contractors who provided examples of their specifications and procedures. The opinions expressed in this paper do not imply endorsement of any particular system or equipment by the authors, their employers or EPRG. An earlier version of this paper was published in 3R International; the agreement of Vulkan Verlag to allow the publication of the present paper is gratefully acknowledged.

References

1. API 1104, 2005. 20th Edition: Welding of pipelines and related facilities. American Petroleum Institute.

2. BS 4515-1, 2009. Specification for welding of steel pipelines on land and offshore. Part 1: Carbon and carbon manganese steel pipelines. British Standards Institution.

3. Det Norske Veritas, 2010. OS-F101: Submarine pipeline systems.

4. Canadian Standards Association, 2007. CSA Z662-07: Oil and Gas Pipeline Systems.

5. Standards Australia, 2007. AS 2885.2-2007: Pipelines – Gas and liquid petroleum – Welding.

6. International Standards Organization, 2005. ISO 3834-2: Quality requirements for fusion welding of metallic materials – Part 2: Comprehensive quality requirements.

7. British Standards Institution, 1999. BS 4515-2: Specification for welding of steel pipelines on land and offshore. Part 2: Duplex stainless steel pipelines.

8. Det Norske Veritas,2010. RP-F118: Pipe girth weld AUT System qualification and project specific procedure validation.

9. R.M.Andrews and L.L.Morgan, 2004. Integration of automated ultrasonic testing and engineering critical assessment for pipeline girth weld defect acceptance. In: International Conference on Pipeline Technology, May 2004. Scientific Surveys, Beaconsfield, Vol 2, pp 655-667.

10. G.Knauf and P.Hopkins, 1996. EPRG guidelines on the assessment of defects in transmission pipeline girth welds. 3R International, 35, 10/11, pp 620-624.

11. R.Denys, R.M.Andrews, M.Zarea, and G.Knauf, 2010. EPRG Tier-2 guidelines for the assessment of defects in transmission pipeline girth welds IPC2010-31640.

materials are outside the code limits, then generic methods such as BS 7910 or R6 can be used. These offer greater flexibility by the correct choice of inputs to match the pipeline geometry.

Experience has shown that these generic assessment methods often produce excessively conservative results when applied to pipeline girth welds. Much of this conservatism is due to these methods using a lower-bound material toughness measured using the standard single-edge-notch bend (SENB) specimen. This uses a deep notch (50% of the thickness) loaded in bending, conditions which are designed to give a high crack tip constraint and hence a lower bound to the toughness. In contrast, most pipeline defects are much less than 50% of the thickness in through-wall height and the dominant loading is tension. These conditions reduce the constraint and increase the effective toughness.

Attempts have been made to quantify constraint effects and reduce the conservatism using constraint-based fracture mechanics’ methods, but these have not been widely accepted in the pipeline industry. DNV has proposed an approach, RP-F108[14], [15] based on testing a single-edge-notch tension (SENT) geometry, particularly for situations involving cyclic plastic strain such as installation by reeling. To date, the SENT test specimen has not been standardized, which is a major limitation on the wider adoption of this test.

Pipeline owners and operators need guidelines for a better understanding of these considerations, and for interpretation of the application of defect-acceptance criteria. It is intended that the guidelines to be developed will address the differentiation of defect-acceptance criteria and the level of AUT system qualification required depending on the application.

Concluding remarks

There is no consensus in different standards regarding the essential welding parameter requirements for mechanized GMAW girth welding. The industry would benefit from the development of a guidance document covering best practices and the technical background for the essential parameters, as the existing standards do not give that guidance at the present time.

The guidance document can also have some positive influence in discussions between the pipeline owner and the construction company regarding the expectations for high-quality welds.

It is not the purpose of the EPRG to rewrite all international standards in order to diminish quality and integrity problems during construction. Adjusting

Sample

copy

- not

for di

stribu

tion


Appendix: Essential variablesThis table is an amalgamation, or summary, of the essential variables of relevance to girth butt welding. In the last three columns it is indicated whether BS EN ISO 15614-1, BS 4515-1 and/or API 1104 address these variables, either as essential (E), non-essential (NE), or not addressed (NA).

In the authors’ opinion, there is a danger of making the essential variables so onerous that weld-procedure qualification testing on project-specific pipe has to be done each time.

International Pipeline Conference, Calgary, Canada. ASME, New York.

12. British Standards Institution, 2005. BS 7910:2005 Incorporating amendment 1: Guide to methods for assessing the acceptability of flaws in metallic structures.

13. British Energy Generation Ltd, 2001. R6. Assessment of the integrity of structures containing defects; R6 Revision 4. Barnwood.

14. Det Norske Veritas, 2006. RP-F108. Fracture control

for pipeline installation methods introducing cyclic plastic strain.

15. A.Cosham and K.MacDonald, 2008. Fracture control in pipelines under high plastic strains – a critique of DNV-RP-F108 IPC2008-64348. International Pipeline Conference, Calgary, Canada. ASME, New York.

16. R.M.Andrews, N.A.Millwood, and P.Roovers, 2012. An update on mechanized gas metal arc welding (GMAW) of pipelines. 3R International, Special edition, 2, 18-29.

Sample

copy

- not

for di

stribu

tion


Var

iabl

eC

hang

es r

equi

ring

re

-qua

lifica

tion

Gui

danc

e no

tes

BS

EN I

SO

1561

4-1:

200

4B

S 45

15-1

: 20

09A

PI

1104

: 20

08

1W

eldi

ng c

ontr

acto

rA

ny c

hang

e in

resp

onsib

ility

for o

pera

tiona

l, te

chni

cal a

nd q

ualit

y co

ntro

lT

his h

as th

e ef

fect

of l

imiti

ng th

e us

e of

pr

e-qu

alifi

ed w

eld

proc

edur

es.

EE

NA

2W

eldi

ng p

roce

ss

The

pro

cess

(es)

use

dA

ny c

hang

eE

EE

The

ord

er o

f pro

cess

es u

sed

Any

cha

nge

whe

n m

ultip

le p

roce

sses

are

us

ed.

EE

E

Man

ual,

part

ly m

echa

nize

d,

mec

hani

zed,

or a

utom

atic

wel

ding

Any

cha

nge

betw

een

man

ual,

part

ly

mec

hani

zed,

mec

hani

zed,

and

aut

omat

ic

wel

ding

EE

E

3W

eldi

ng e

quip

men

t

Wel

ding

Any

cha

nge

in m

ake,

type

, and

mod

el fo

r pa

rtly

mec

hani

zed,

mec

hani

zed,

or a

utom

atic

w

eldi

ng

Thi

s lim

its th

e pr

oced

ure

to a

spec

ific

syst

em, w

hich

is se

nsib

le. H

owev

er, t

he

syst

em sh

ould

incl

ude

the

pow

er so

urce

, an

d co

ntro

ller.

NA

NA

E

Soft

war

e pr

ogra

mA

ny c

hang

eW

PD sh

ould

stat

e re

visio

n nu

mbe

r of

soft

war

e/pr

ogra

m.

NA

NA

NA

Hot

key

lim

itsA

ny in

crea

seO

ver-

ride

func

tion

NA

NA

NA

Num

ber o

f wire

sA

ny c

hang

e fr

om si

ngle

wire

to m

ultip

le w

ire

syst

ems a

nd v

ice

vers

a.T

his i

s im

port

ant f

or m

oder

n sy

stem

s w

hich

may

use

a d

ual t

orch

, or t

ande

m

wire

arr

ange

men

t.

EE

NA

4B

ase

mat

eria

ls

Mat

eria

l gra

deA

cha

nge

from

a lo

wer

- to

a hi

gher

-str

engt

h gr

ade,

but

not

vic

e ve

rsa.

Gro

upin

gs (

as p

er A

PI 1

104)

:(a

) SM

YS

≤ 2

90 M

Pa(b

) 29

0 <

SM

YS

< 4

48 M

Pa(c

) SM

YS

≥ 4

48 M

Pa (

each

gra

de sh

all b

e se

para

tely

qua

lified

.)

The

re a

re se

vera

l way

s of g

roup

ing

mat

eria

l gra

de. M

ost,

if no

t all,

sp

ecifi

catio

ns a

dopt

the

prin

cipl

e of

re

stric

ting

the

max

imum

stre

ngth

to th

at

wel

ded

in th

e pr

oced

ure

test

.T

he g

roup

ings

sugg

este

d by

API

110

4 se

em se

nsib

le si

nce

it is

not g

ood

prac

tice

to o

verm

atch

by

too

high

a m

argi

n.

EE gr

oupi

ng o

n ch

emic

al

anal

ysis

E

Supp

ly c

ondi

tion

A c

hang

e in

the

supp

ly c

ondi

tion

(TM

CP,

Q

T, o

r nor

mal

ized

).E

EE

Sample

copy

- not

for di

stribu

tion


Var

iabl

eC

hang

es r

equi

ring

re

-qua

lifica

tion

Gui

danc

e no

tes

BS

EN I

SO

1561

4-1:

200

4B

S 45

15-1

: 20

09A

PI

1104

: 20

08

Stee

l sup

plie

rFo

r SM

YS

> 4

50 M

Pa: a

cha

nge

in b

ase

mat

eria

l orig

in (

stee

l mill

).O

ne sh

ould

be

care

ful t

o di

ffere

ntia

te

betw

een

plat

e m

ill a

nd p

ipe

mill

, i.e

. a

pipe

mill

can

obt

ain

plat

e fr

om m

ore

than

one

pla

te m

ill, a

nd si

mila

rly a

pla

te

mill

can

supp

ly p

late

to m

ore

than

one

pi

pe m

ill.

NA

NA

NA

Che

mic

al c

ompo

sitio

nA

n in

crea

se in

:(i

) Pc

m o

f mor

e th

an 0

.020

,(i

i) C

E of

mor

e th

an 0

.030

, and

(iii)

Car

bon

cont

ent o

f mor

e th

an 0

.02%

for

C-M

n an

d lo

w a

lloy

stee

l.

BS 4

515

has a

muc

h m

ore

com

preh

ensiv

e se

t of l

imits

.G

roup

ing

as p

er

CR

-ISO

156

08

incl

udes

che

mic

al

limits

.

EN

A

Man

ufac

turin

g pr

oces

sA

cha

nge

in m

anuf

actu

ring

proc

ess (

rolle

d,

seam

less

, for

ged,

cas

t).

NA

NA

NA

UN

S nu

mbe

rsA

cha

nge

in th

e U

NS

num

ber f

or C

RA

s.E

EE

5M

ater

ial t

hick

ness

and

dia

met

er

Mat

eria

l thi

ckne

ss o

f tes

t joi

nt

(whe

re t

is th

e no

min

al th

ickn

ess)

.Fo

r non

-sou

r ser

vice

:t <

25

mm

: cha

nge

beyo

nd 0

.75

t to

1.5

tt >

25

mm

: cha

nge

beyo

nd 0

.75

t to

1.25

For s

our s

ervi

ce:

a ch

ange

out

side

the

thic

knes

s ran

ge 0

.75

t to

1.2

5 t

The

mat

eria

l thi

ckne

ss w

ill h

ave

an

effe

ct o

n th

e t8

/5 c

oolin

g ra

te a

nd h

ence

th

e H

AZ

hard

ness

val

ues.

So, i

t is n

ot

alw

ays w

ise to

hav

e to

o w

ide

a ra

nge

of

wal

l thi

ckne

ss c

over

ed b

y on

e PQ

R.

EE

E, b

ut n

o cl

ear l

imits

de

fined

.

Nom

inal

OD

of p

ipe

Cha

nge

in n

omin

al o

utsid

e di

amet

er b

eyon

d th

e gr

oups

qua

lified

as f

ollo

ws:

D <

60.

3 m

m60

.3 m

m <

D <

323

.9 m

mD

> 3

23.9

mm

Alte

rnat

ivel

y, w

here

ther

e is

a ch

ange

in

diam

eter

from

a q

ualifi

ed p

roce

dure

of m

ore

than

50%

of t

he n

omin

al O

D.

Som

e sp

ecifi

catio

ns re

fer t

o th

e pi

pe

ID, b

ut fo

r sim

plic

ity th

e au

thor

pre

fers

no

min

al O

D.

The

min

imum

effe

ctiv

e siz

e fo

r m

echa

nize

d G

MA

W/F

CA

W is

6 in

.

EE

E, b

ut n

o cl

ear l

imits

de

fined

.

Sample

copy

- not

for di

stribu

tion


Var

iabl

eC

hang

es r

equi

ring

re

-qua

lifica

tion

Gui

danc

e no

tes

BS

EN I

SO

1561

4-1:

200

4B

S 45

15-1

: 20

09A

PI

1104

: 20

08

6Jo

int c

onfig

urat

ion

Join

t des

ign/

confi

gura

tion

Any

cha

nge

in jo

int d

imen

sions

out

side

the

tole

ranc

es sp

ecifi

ed in

the

agre

ed W

PS.

Sugg

este

d to

lera

nces

are

as f

ollo

ws:

Beve

l ang

les (

±1°

)Si

ze o

f roo

t fac

e (±

50%

)W

idth

of r

oot g

ap (

+0.

5 m

m).

An

incr

ease

in th

e pe

rmitt

ed le

vel o

f hi

gh-lo

w b

eyon

d th

at q

ualifi

ed sh

ould

al

so b

e co

nsid

ered

as a

n es

sent

ial

varia

ble.

NA

E, fo

r mai

n co

nfigu

ratio

n,

tole

ranc

es

are

to b

e ap

prov

ed b

y op

erat

or

E, J,

to V

or

vic

e ve

rsa.

Back

ing

and

back

ing

mat

eria

lA

dditi

on o

r del

etio

n of

bac

king

, or c

hang

e of

ba

ckin

g m

ater

ial.

For e

xam

ple,

cop

per s

hoes

.E

EN

A

7A

lignm

ent a

nd ta

ck w

eldi

ng

Tac

k w

eldi

ngA

redu

ctio

n in

the

num

ber o

r siz

e of

tack

w

elds

or b

oth.

Any

cha

nge

in re

mov

al o

f tac

k w

elds

or

inte

grat

ion

of ta

ck w

elds

in th

e [fi

nal]

wel

d.

Tac

k w

elds

are

gen

eral

ly n

ot u

sed

for

mec

hani

zed

GM

AW

, but

may

be

used

fo

r sem

i-aut

omat

ic S

TT

root

pas

s.

NA

NA

NE

Line

-up

clam

pO

miss

ion

of a

line

-up

clam

p an

d a

chan

ge

betw

een

exte

rnal

and

inte

rnal

line

-up

clam

p.Ex

tern

al c

lam

ping

app

lies t

o SM

AW

or

sem

i-aut

omat

ic G

MA

W/F

CA

W ro

ot

pass

.M

echa

nize

d G

MA

W sh

ould

be

perfo

rmed

to

geth

er w

ith a

n in

tern

al li

ne-u

p cl

amp.

NA

NA

NE

Rem

oval

of l

ine-

up c

lam

pFo

r mec

hani

zed

GM

AW

: Any

redu

ctio

n in

th

e nu

mbe

r of r

uns.

For m

anua

l or s

emi-a

utom

atic

wel

ding

: Any

re

duct

ion

in th

e pe

rcen

tage

of r

oot p

ass

com

plet

ed.

For m

echa

nize

d G

MA

W 1

00%

of

the

root

pas

s mus

t be

com

plet

ed, a

s a

min

imum

, bef

ore

rem

oval

of t

he li

ne-u

p cl

amp.

> 5

0% o

f roo

t pas

s sha

ll be

com

plet

ed

befo

re li

ne-u

p cl

amp

is re

leas

ed. I

f cla

mp

is re

leas

ed b

efor

e ro

ot p

ass c

ompl

eted

th

en a

t lea

st 8

0% o

f bot

h to

p an

d bo

ttom

qu

adra

nts s

hall

be c

ompl

eted

.

NA

NA

NA

Low

erin

g-of

f (on

land

),

Barg

e m

ove-

up (

offs

hore

).A

ny re

duct

ion

in th

e nu

mbe

r of r

uns.

Stru

ctur

al a

naly

sis sh

ould

be

carr

ied

out t

o co

nfirm

min

imum

requ

ired

wel

d lig

amen

t.

NA

NA

NA

Inte

rnal

misa

lignm

ent

Any

incr

ease

for c

lad

and

lined

pip

eN

AN

AN

A

Sample

copy

- not

for di

stribu

tion


Var

iabl

eC

hang

es r

equi

ring

re

-qua

lifica

tion

Gui

danc

e no

tes

BS

EN I

SO

1561

4-1:

200

4B

S 45

15-1

: 20

09A

PI

1104

: 20

08

8W

eldi

ng c

onsu

mab

les

Elec

trod

e or

fille

r met

alA

ny c

hang

e of

dia

met

er o

r cro

ss-s

ectio

nal

area

For m

anua

l wel

ding

it is

nor

mal

to

spec

ify:

- any

cha

nge

for t

he c

appi

ng la

yer o

r the

fir

st tw

o la

yers

, or

- any

incr

ease

for o

ther

runs

.

EE

E

Any

cha

nge

of ty

pe, c

lass

ifica

tion

or b

rand

.Fo

r low

er m

ater

ial g

rade

s, ch

ange

of

bran

d ne

ed n

ot b

e an

ess

entia

l var

iabl

e.W

here

ther

e is

a sig

nific

ant c

hang

e in

th

e pr

opor

tion

of th

e th

ickn

ess w

elde

d w

ith d

iffer

ent e

lect

rode

cla

ssifi

catio

ns

ther

e m

ay b

e so

me

mer

it in

rest

rictin

g th

e ra

nge

of q

ualifi

catio

n.

E, w

ith C

harp

y re

quire

men

tsE

E

Any

use

of a

n un

test

ed b

atch

whe

n ba

tch

test

ing

is re

quire

d.Ba

tch

test

ing

is re

quire

d fo

r mos

t of

fsho

re p

ipel

ine

proj

ects

and

hig

h-st

reng

th o

nsho

re p

roje

cts.

NA

NA

NA

Any

use

of a

wel

ding

con

sum

able

bat

ch w

ith

a re

duct

ion

in te

nsile

or i

mpa

ct p

rope

rtie

s of

mor

e th

an 1

0% fr

om th

e ba

tch

used

for

WPQ

T w

hen

batc

h te

stin

g is

not r

equi

red.

NA

NA

NA

Wire

feed

spee

dA

ny c

hang

e be

yond

rang

e qu

alifi

ed.

NA

EN

A

Elec

trod

e/w

ire sp

acin

gA

ny c

hang

eFo

r dua

l tor

ch o

r tan

dem

wel

ding

EE

NA

Elec

trod

e/w

ire a

ngle

Any

cha

nge

beyo

nd ra

nge

qual

ified

NA

NA

NA

9Sh

ield

ing,

bac

king

and

pla

sma

gase

s

Gas

es a

ccor

ding

to E

N 4

39.

Any

cha

nge

in d

esig

natio

n, c

lass

ifica

tion

and

purit

y ac

cord

ing

to E

N 4

39.

EE

E

Oth

er g

ases

and

mix

ture

s.A

ny c

hang

e of

the

nom

inal

com

posit

ion

(±10

%),

pur

ity a

nd d

ew p

oint

.E

EE

Oxy

gen

cont

ent o

f bac

king

gas

Any

incr

ease

EE

E

Sample

copy

- not

for di

stribu

tion


Var

iabl

eC

hang

es r

equi

ring

re

-qua

lifica

tion

Gui

danc

e no

tes

BS

EN I

SO

1561

4-1:

200

4B

S 45

15-1

: 20

09A

PI

1104

: 20

08

Shie

ldin

g ga

s flow

rate

.Fo

r pro

cess

es 1

31, 1

35, 1

36, 1

37 &

141

: any

ch

ange

in fl

ow ra

te b

eyon

d ±

10%

.So

me

spec

s are

inte

rest

ed in

the

decr

ease

in

shie

ldin

g ga

s flow

rate

and

the

decr

ease

in th

e no

zzle

or c

up si

ze; r

athe

r th

an a

ny in

crea

ses.

NA

EE

Shro

ud d

iam

eter

Any

dec

reas

eFo

r tan

dem

GM

AW

NA

NA

NA

10El

ectr

ical

cha

ract

eris

tics

and

puls

ing

data

Pola

rity

Any

cha

nge

in p

olar

ityE

EE

AC

, DC

or p

ulse

d cu

rren

t.A

ny c

hang

e in

type

of c

urre

nt a

nd a

cha

nge

from

nor

mal

to p

ulse

d cu

rren

t and

vic

e ve

rsa.

Som

e sp

ecs o

nly

refe

r to

a ch

ange

of

elec

tric

al c

urre

nt b

etw

een

AC

and

DC

, bu

t with

mod

ern

syst

ems i

t is i

mpo

rtan

t to

con

sider

any

cha

nges

bet

wee

n no

rmal

an

d pu

lsed

curr

ent.

EE

E

Pulse

freq

uenc

y ra

nge

in p

ulse

d m

anua

l wel

ding

.A

ny c

hang

e in

pul

se fr

eque

ncy

for

back

grou

nd a

nd p

eak

curr

ent e

xcee

ding

±

10%

and

pul

se d

urat

ion

rang

e ex

ceed

ing

±10

%.

E vi

a H

eat i

nput

E, b

ut

tole

ranc

e to

be

defi

ned

by

oper

ator

E

Con

tact

tip

to w

ork

dist

ance

(ele

ctric

al st

ick-

out)

.A

ny c

hang

e in

ele

ctric

al st

ick-

out e

xcee

ding

±

5 m

m (

diffe

rent

val

ues m

ay b

e us

ed fo

r ea

ch ru

n).

Som

e sp

ecifi

catio

ns d

efine

a p

erce

ntag

e ra

nge

(± 1

0%).

NE

EN

A

Para

met

er ra

nges

Any

cha

nge

in a

rc v

olta

ge e

xcee

ding

± 1

0%A

ny c

hang

e in

wire

feed

spee

d or

wel

ding

cu

rren

t exc

eedi

ng ±

10%

Any

cha

nge

in tr

avel

spee

d ex

ceed

ing

± 1

0%N

o se

para

te re

stric

tion

for c

alcu

late

d va

lue

of

arc

ener

gy.

Tol

eran

ce sh

ould

be

base

d on

the

mea

n va

lue;

not

add

ed to

the

rang

e!N

otic

e th

e su

btle

diff

eren

ce b

etw

een

‘hea

t inp

ut’ a

nd ‘a

rc e

nerg

y’.

E, v

ia h

eat i

nput

EE

Arc

ene

rgy

rang

e fo

r eac

h pa

ss.

For C

-Mn

and

low

allo

y st

eels

with

SM

YS

< 4

50 M

Pa in

non

-sou

r ser

vice

: any

cha

nge

exce

edin

g ±

15%

.Fo

r C-M

n an

d lo

w a

lloy

stee

ls w

ith S

MY

S >

45

0 M

Pa in

non

-sou

r ser

vice

: Any

cha

nge

exce

edin

g ±

10%

.Fo

r CR

As:

Any

cha

nge

exce

edin

g ±

10%

.

E, v

ia h

eat i

nput

EE

Sample

copy

- not

for di

stribu

tion


Var

iabl

eC

hang

es r

equi

ring

re

-qua

lifica

tion

Gui

danc

e no

tes

BS

EN I

SO

1561

4-1:

200

4B

S 45

15-1

: 20

09A

PI

1104

: 20

08

Mod

e of

met

al tr

ansf

erG

MA

W: a

cha

nge

from

spra

y ar

c, g

lobu

lar

arc,

or p

ulsa

ting

arc

to sh

ort-

circ

uitin

g ar

c,

and

vice

ver

sa.

FCA

W: a

cha

nge

from

shor

t-ci

rcui

ting

tran

sfer

to sp

ray

or g

lobu

lar t

rans

fer.

Qua

lifica

tion

with

spra

y or

glo

bula

r tra

nsfe

r qu

alifi

es b

oth

spra

y or

glo

bula

r tra

nsfe

r.

For h

igh-

perf

orm

ance

syst

ems i

t wou

ld

be re

ason

able

to sp

ecify

the

type

of

pow

er so

urce

as a

n es

sent

ial v

aria

ble.

EN

AE

11W

eldi

ng te

chni

ques

Ang

le o

f pip

e ax

is to

the

horiz

onta

l.A

cha

nge

of m

ore

than

±15

° fro

m th

e po

sitio

n w

elde

d.T

he L

045

posit

ion

qual

ifies

for a

ll po

sitio

ns

prov

ided

all

othe

r ess

entia

l var

iabl

es a

re

fulfi

lled.

Som

e sp

ec st

ate

a ±

5° to

lera

nce

from

no

min

al p

ositi

on, w

hilst

oth

ers a

re m

ore

gene

rous

in a

llow

ing

±25

°.

EE

NE

Wel

ding

dire

ctio

nA

cha

nge

from

upw

ards

to d

ownw

ards

and

vi

ce v

ersa

.H

oriz

onta

l (2G

) w

eldi

ng w

ould

not

qu

alify

fixe

d 5G

pos

ition

al w

eldi

ng.

EE

E

Strin

ger/

wea

veA

cha

nge

in a

mpl

itude

, fre

quen

cy, o

r dw

ell

time

of a

ny m

echa

nize

d w

eave

.T

he m

axim

um a

mpl

itude

of a

ny

mec

hani

zed

wea

ve is

to b

e ag

reed

.T

he fr

eque

ncy

of a

ny m

echa

nize

d w

eave

is

to b

e ag

reed

.T

he d

wel

l tim

e at

the

side

of a

ny

mec

hani

zed

wea

ve is

to b

e ag

reed

.

E, v

ia h

eat i

nput

E, b

ut

tole

ranc

e to

be

defi

ned

by

oper

ator

.

NA

Sequ

ence

of d

epos

ition

of d

iffer

ent

cons

umab

les

Any

cha

nge

in th

e se

quen

ce.

EE

E

Sequ

ence

of s

ides

wel

ded

first

and

la

st (

doub

le-s

ided

wel

ds)

Any

cha

nge

in th

e se

quen

ce.

NA

EN

A

Num

ber o

f pas

ses

Not

lim

ited

unle

ss it

is a

redu

ctio

n to

less

th

an fo

ur p

asse

s.A

ny c

hang

e in

the

num

ber o

f cap

pas

ses.

Dua

l-tor

ch sy

stem

s som

etim

es m

ake

use

of a

split

cap

tech

niqu

e (t

wo

pass

es, o

ne

run)

.

NA

NA

NA

Pass

es w

elde

d fr

om e

ach

side.

Cha

nge

from

sing

le to

mul

ti-pa

ss w

eldi

ng a

nd

vice

ver

sa.

EE

E

Num

ber o

f wel

ders

Any

dec

reas

e in

the

num

ber o

f wel

ders

for

wel

ding

of r

oot a

nd h

ot p

ass.

Nor

mal

pra

ctic

e is

to u

se tw

o w

elde

rs

per b

utt.

NA

EN

A

Sample

copy

- not

for di

stribu

tion


Var

iabl

eC

hang

es r

equi

ring

re

-qua

lifica

tion

Gui

danc

e no

tes

BS

EN I

SO

1561

4-1:

200

4B

S 45

15-1

: 20

09A

PI

1104

: 20

08

Tim

e la

pse

betw

een

com

plet

ion

of

root

pas

s and

star

t of h

ot p

ass.

Any

incr

ease

in th

e tim

e la

pse

beyo

nd th

e ra

nge

qual

ified

.Fo

r cel

lulo

sic e

lect

rode

s it i

s the

tim

e la

pse

betw

een

the

star

t of t

he ro

ot ru

n an

d th

e st

art o

f the

seco

nd ru

n w

hich

is

impo

rtan

t.H

owev

er, f

or h

ydro

gen-

cont

rolle

d el

ectr

odes

it is

usu

ally

the

time

laps

e be

twee

n co

mpl

etio

n of

root

pas

s and

st

art o

f hot

pas

s whi

ch is

mea

sure

d.

NA

Add

ress

ed

but o

nly

esse

ntia

l w

hen

cellu

losic

w

eldi

ng is

ap

plie

d.

E

Wel

d co

mpl

etio

nA

ny re

duct

ion

in th

e nu

mbe

r of p

asse

s co

mpl

eted

bef

ore

cool

ing

to b

elow

pre

heat

te

mpe

ratu

re.

NA

EN

A

Acc

eler

ated

wel

d co

olin

gA

ny c

hang

e in

met

hod

and

med

ium

and

any

in

crea

se in

max

imum

tem

pera

ture

of t

he

wel

d at

star

t of c

oolin

g.

Thi

s is f

or w

elds

subj

ect t

o A

UT

in

spec

tion.

If fa

st in

spec

tion

is re

quire

d.

Usu

ally

in fi

eld

set u

p, th

is is

not r

equi

red.

NA

NA

NA

12P

rehe

atin

g

Preh

eat t

empe

ratu

reA

ny re

duct

ion

Som

e sp

ecs a

llow

a p

lus t

oler

ance

(50

°C

or 7

5 °C

) on

pre

heat

, but

real

ly it

is th

e m

inim

um p

rehe

at th

at is

of i

nter

est.

The

te

st w

eld

mus

t sim

ulat

e th

e w

orst

cas

e to

be

enc

ount

ered

dur

ing

prod

uctio

n.

EE

NE

Met

hod

of a

pplic

atio

nA

cha

nge

from

flam

e he

atin

g to

ele

ctric

al

heat

ing,

but

not

vic

e ve

rsa.

It is

gen

eral

ly a

ccep

ted

that

ele

ctric

al

met

hods

(re

sista

nce

and

indu

ctio

n he

atin

g) p

rovi

de a

mor

e ev

en,

cont

rolle

d, te

mpe

ratu

re p

rofil

e th

an

flam

e he

atin

g.

NA

EN

A

Met

hod

of c

ontr

ollin

g te

mpe

ratu

reN

/AM

etho

d (i

.e. t

empe

ratu

re in

dica

ting

cray

on, c

onta

ct p

robe

) sh

ould

be

stat

ed

on W

PD. I

t is t

he a

utho

rs’ o

pini

on

that

non

-con

tact

ther

mom

eter

s (i.e

. in

frar

ed p

yrom

eter

s) sh

ould

not

be

used

fo

r mea

surin

g pr

ehea

t and

inte

rpas

s te

mpe

ratu

re o

n w

elds

.

NA

EN

A

Initi

al te

mpe

ratu

re w

hen

preh

eat i

s no

t use

d.A

ny re

duct

ion

belo

w 1

0°C

.N

AE

NA

Sample

copy

- not

for di

stribu

tion


Var

iabl

eC

hang

es r

equi

ring

re

-qua

lifica

tion

Gui

danc

e no

tes

BS

EN I

SO

1561

4-1:

200

4B

S 45

15-1

: 20

09A

PI

1104

: 20

08

13In

terp

ass

tem

pera

ture

Max

imum

and

min

imum

inte

r-pa

ss

tem

pera

ture

.A

ny in

crea

se a

bove

25°

C fo

r C-M

n an

d lo

w

allo

y st

eel.

Any

incr

ease

for C

RA

s. A

ny

redu

ctio

n be

low

the

preh

eat t

empe

ratu

re.

Usu

ally

mea

sure

min

imum

pre

heat

and

m

axim

um in

terp

ass t

empe

ratu

re. F

or

C-M

n st

eels

limit

inte

rpas

s to

250

°C

max

.

EE

E

14C

lean

ing

of b

evel

and

wel

dN

/AT

here

may

be

cert

ain

inst

ance

s whe

re

the

clea

ning

of t

he b

evel

and

wel

d w

ill

affe

ct th

e pe

rfor

man

ce o

f the

wel

d, b

ut

this

is no

t rea

lly a

n es

sent

ial v

aria

ble.

C

lean

ing

shou

ld b

e by

han

d to

ols,

pow

er

brus

h or

grin

der.

NE

NA

NA

15P

ost-

wel

d he

at tr

eatm

ent

Post

hea

ting:

hyd

roge

n re

leas

eA

ny re

duct

ion

in ti

me

and

tem

pera

ture

; de

letio

n (b

ut n

ot a

dditi

on)

of p

ost h

eatin

g.E

EE

Post

-wel

d he

at tr

eatm

ent (

stre

ss

relie

f).

Add

ition

or d

elet

ion

of p

ost-

wel

d he

at

trea

tmen

t.A

ny c

hang

e in

met

hod

of a

pply

ing

heat

.A

ny c

hang

e in

hol

ding

tem

pera

ture

ex

ceed

ing

±20

°C.

Any

cha

nge

in h

oldi

ng ti

me

and

any

chan

ge

in h

eatin

g an

d co

olin

g ra

tes o

utsid

e ±

5%.

Gen

eral

ly d

o no

t per

form

pos

t-w

eld

heat

tr

eatm

ent o

n T

MC

P lin

epip

e.13

Cr m

arte

nsiti

c st

ainl

ess s

teel

wel

ds a

re

usua

lly su

bjec

t to

a sh

ort h

eat-

trea

tmen

t cy

cle

(for

exa

mpl

e, 5

min

utes

at 6

50°C

).

EE

E

Sample

copy

- not

for di

stribu

tion

September 30 – October 2, 2014Calgary TELUS Convention Centre

Calgary, Alberta, Canada

EXHIBIT NOW

internationalpipelineexposition.com

Ensure your place at the world’s largest pipeline gathering

Held alongside International Pipeline Conference

@petroleumshow #IPE14

Sample

copy

- not

for di

stribu

tion


SEVERAL METHODS FOR the assessment of corroded pipelines subjected to internal pressure loading are currently available: for example, the ASME B31G method, the RSTRENG 0.85dL method, and the

DNV RP-F101 method for single defects (Part B). These methods do not take into account the effect of a longitudinal compressive force on the corroded pipeline failure pressure.

When a corroded pipeline is subjected to a longitudinal compressive force superimposed on the pressure loading, its failure pressure may be reduced. For a buried pipeline, the most common source of a longitudinal compressive force is temperature loading; ground movement may also produce a longitudinal compressive force in the pipeline.

The DNV RP-F101 method for combined loadings and the RPA-PLLC method are two of the currently available assessment methods which can take into account the effect of a longitudinal compressive force on the corroded pipeline failure pressure.

In this paper a comparison between the DNV RP-F101 method for combined loadings and the RPA-PLLC method is presented.

Prediction of the failure pressure of corroded pipelines subjected to a longitudinal compressive force superimposed on the pressure loading

by Dr Adilson C BenjaminPetrobras R&D Centre, Rio de Janeiro, Brazil

Nomenclature

A – longitudinal area of metal loss Asteel – area of the cross section of the pipe De – outside diameter of the pipeline d – maximum depth of the corrosion defect E – Young’s modulus (fR)h – reduction factor in the hoop direction (fR)L – reduction factor in the longitudinal direction I – moment of inertia of the cross sectional area of the pipeL – length of the corrosion defect M – Folias bulging factor Mb – external applied bending moment

n – ratio of the stress (σL)p to the stress σh Nc – external applied longitudinal compressive force p – pipeline internal pressure pd – design pressure (pa)comb – allowable pressure of a corroded pipeline subjected to pressure loading plus longitudinal compression (pa)press – allowable pressure of a corroded pipeline subjected to pressure loading only (pf)comb – failure pressure of a corroded pipeline subjected to pressure loading plus a longitudinal compression (pf)press – failure pressure of a corroded pipeline subjected to pressure loading only (pf)test – failure pressure measured in the burst test t – wall thickness of the pipeline Tinst – pipeline installation temperature w – width of the corrosion defect α – thermal expansion coefficientα area – factor that depends on the geometric shape adopted to represent the longitudinal area of metal loss A

This is an updated version of the author’s paper that was first published at IPC, Calgary, in 2008.

Contact details:tel: +55 21 2162 4782email: [email protected]

Sample

copy

- not

for di

stribu

tion


The software performs shell finite-element analyses of pipeline corrosion defects under combined internal pressure and longitudinal loads.

In a paper published in 1998 [18], Roberts and Picks describe research work on corroded pipelines subjected to a longitudinal compressive force or a bending moment performed at the University of Waterloo. Based on finite-element analyses and burst tests, the authors developed an expression to calculate a longitudinal stress factor to correct B31G or RSTRENG predictions of burst pressure.

In a paper published in 2008 [19], Benjamin presented the RPA-PLLC method. This method is a modified version of the RPA method [20] and can take into account the effect of a longitudinal compressive force on the corroded pipeline failure pressure. The RPA method is itself a modified version of the RSTRENG 0.85dL method [21].

Note that RPA is the acronym for rectangular parabolic area; PLLC is the acronym for pressure loading plus longitudinal compression.

In this paper a comparison between the DNV RP-F101 method for combined loadings and the RPA-PLLC method is presented.

Stresses in a pipeline subjected to the pressure loading combined with a longitudinal compressive force

Assuming that the pipeline is a thin shell the radial stress σR at any point of the pipe wall is negligibly small. Consequently there are only two stresses at any point of

THE PUBLISHED LITERATURE about the assessment of corrosion defects in pipelines subjected

to a longitudinal compressive force is not extensive. Several papers found in the published literature [1-9] are related to a research project carried out by the Southwest Research Institute (SwRI) in the 1990s: this project was funded by the Alyeska Pipeline Service Co which operates and maintains the Trans Alaska Pipeline System. The purpose of this project was to investigate the structural behaviour of corroded pipelines subjected to longitudinal compression due to temperature loading and bending moment due to the freeze-thaw action of arctic permafrost. The results of this project include full-scale tests, finite-element analyses, and a computer program named shell analysis failure envelope (SAFE).

Other papers found in the published literature [10-15] are related to the joint-industry project (JIP) on Reliability of corroded pipes carried out by DNV in the 1990s. One of the objectives of this JIP was to investigate the pressure strength of pipelines subjected to internal pressure combined with a longitudinal compressive force or a bending moment. Full-scale tests and finite-element analyses were performed. Based on these results, a method for the calculation of the pipeline failure pressure taking into account the effect of a longitudinal compressive force was developed. This method, called the DNV RP-F101 method for combined loadings (DNV CL method), was first published in the Section 7.3 of the DNV RP-F101 [16] released in 1999.

In a report published in 1995 [17], Stephens, Bubenik, and Francini presented the background of the development of the PCORR computer program. This research project was performed at Battelle under the sponsorship of the Pipeline Research Committee.

ΔT – temperature rise above the temperature Tinst γd – design factor, for example the safety factor applied to the yield stress to establish the allowable circumferential stress used to calculate the pipe wall thickness (according to the ASME B31.4 code, γd is equal to 0.72)γeqv – safety factor applied to yield stress to establish a limit to the equivalent stress (von Mises stress or Tresca stress) acting at any point of the pipe wall (according to the ASME B31.4 code γeqv is equal to 0.90)γm – modelling factor adopted by the DNV RP-F101 method for combined loading in the calculation of the allowable pressure besides the usual design factor (γm = 0.9) ν – Poisson’s ratio Σ(σL)add i – summation of the longitudinal stresses generated by the additional loadings (loadings other than the pressure loading). Herein it is supposed that the stress Σ (σL)add i is negative (compressive)σ1, σ2 – principal stresses

σallow – allowable stress (σallow)h – allowable stress in the hoop direction (σallow)L – allowable stress in the longitudinal direction σfail – failure stress of the material (σfail)h – failure stress in the hoop direction (σfail)L – failure stress in the longitudinal direction σflow – flow stress of the material σh – hoop stress σL – longitudinal stress (σL)p – longitudinal stress generated by the internal pressure (σL) ΔT – longitudinal stress generated by the temperature rise ΔT (σL)Mb – longitudinal stress due to an external applied bending moment Mb (σL)Nc – longitudinal stress due to the force Nc

σR – radial stress σult – ultimate tensile stress of the material σyield – yield stress of the material

Sample

copy

- not

for di

stribu

tion


longitudinal stress generated by the bending moment will be a tensile stress; otherwise it will be a compressive stress. The longitudinal stress due to an external applied bending moment is given by the following equation:

σ L Mbb eM

ID( ) = ±2 (7)

where

I D De i= −π64

4 4( ) (8)

The resultant longitudinal stress is equal to the longitudinal tensile stress plus the stress Σ(σL)add i which is the summation of the longitudinal stresses generated by the additional loadings (loadings other than the pressure loading).

σ σ σL L p L add i= ( ) + ( )∑ (9)

It is assumed here that the stress Σ(σL)add i is negative (compressive). The resultant longitudinal stress is positive (tensile) or negative (compressive) depending on the value of the stress Σ(σL)add i.

Tresca yield criterion

The equations of the Tresca yield criterion for a biaxial state of stress are as follows [23]:

σ σ σ1 2− = yield if σ1 σ2 ≤ 0 (10)

σ σ σ σ1 2= =yield yieldor σ1 σ2 ≥ 0 (11)

Figure 1 presents a graphical representation of the Tresca yield criterion for a biaxial state of stress.

the pipe wall, the hoop stress σh and the longitudinal stress σL. The assumption that the pipeline is a thin shell is valid provided that the ratio of the pipeline outside diameter De to the pipeline wall thickness t is greater than or equal to 20 (De/t ≥ 20).

Besides the hoop tensile stress σh the internal pressure also generates a longitudinal tensile stress (σL)p in the pipe wall. The hoop tensile stress and the longitudinal tensile stress are related to the internal pressure p by the following equations:

σ hep Dt

=2 (1)

σ σL p hn( ) = (2)

where

n = 0.5 for longitudinally unrestrained pipe (3a)n = ν = 0.3 for longitudinally restrained pipe (3b)

For a buried pipeline the most common source of a longitudinal compressive force is a temperature rise ∆T. It is assumed here that even if the pipeline is carrying a hot product, the steel properties at ambient temperature can be used in the calculations. According to the ASME B31.4 code [22], this assumption is valid provided that the pipeline temperature (Tinst + ΔT) is less than or equal to 120oC.

The longitudinal compressive stress generated by the temperature rise in a straight pipe which is fully restrained in the longitudinal direction is given by the Equn 4:

σ αL TE t( ) = −

∆∆ (4)

A ground movement, such as a landslide, may also produce a longitudinal compressive force in the pipeline.

The longitudinal compressive stress due to an external applied longitudinal compressive force is given by Equn 5:

σ L Ncc

steel

NA

( ) = (5)

where

A D Dsteel e i= −π4

2 2( ) (6)

A ground movement, such as a landslide or a differential settlement, may also produce a bending moment Mb. In this case there is a contribution of the bending moment to the resultant longitudinal stress acting on the corrosion defect. If the corrosion defect is situated above the neutral axis of the pipe cross section, the

Fig.1. Tresca yield criterion for a biaxial state of stress.

Sample

copy

- not

for di

stribu

tion


If the longitudinal stress is positive (tensile), the equations of the failure criterion are:

σ σL fail L= ( )

(16)or

σ σh fail h= ( ) (17)

Equations 16 and 17 are Equn 11 re-written such that the longitudinal stress σL is the principal stress σ1, the hoop stress σh is the principal stress σ2, and the failure stress of the material is (σfail)L or (σfail)h.

Figure 2 presents a graphical representation of the failure criterion based upon the Tresca yield criterion.

RPA-PLLC method

Level-1 assessment methods are the methods which represent the longitudinal area of metal loss A on the basis of the maximum defect depth d and defect length L.

The RPA method is a Level-1 method which was developed by Benjamin and Andrade [20], and is a modified version of the RSTRENG 0.85dL method [21]. The method uses two different equations to predict the failure pressure of a corrosion defect. The equation for short defects is the same equation as that used by the original 0.85dL method in the assessment of short defects. However, the equation for long defects is different from the equation used by the original 0.85dL method in the assessment of long defects, and gives conservative results for long uniform-depth defects.

Several methods for the assessment of corroded pipelines subjected to internal pressure loading are currently available; for example, the ASME B31G method [25], the RSTRENG 0.85dL method [21], the DNV RP-F101 method for single defects [26], and the RPA method [20]. These methods do not take into account the effect of a longitudinal compressive force on the corroded pipeline failure pressure.

When the corroded pipeline is subjected to a longitudinal compressive force superimposed onto the pressure loading, its failure pressure may be reduced.

The RPA-PLLC method is a modified version of the RPA method which can take account of the effect of a longitudinal compressive load on the corroded pipeline failure pressure. This method can only be applied for the assessment of pipelines made of steel.

The RPA-PLLC method was developed in 2004 and was included in the first edition of the Petrobras

Failure criterion

The failure criterion adopted is based upon the Tresca yield criterion. If the longitudinal stress is negative (compressive), the start point to derive one of the equations of the failure criterion is Equn 12 below.

σ σ σL h fail− = (12)

Equation 12 is Equn 10 re-written such that that the longitudinal stress σL is the principal stress σ1, the hoop stress σh is the principal stress σ2, and the failure stress of the material is σfail.

Considering that the longitudinal stress is negative (compressive) and the hoop stress is positive (tensile), Equn 12 can be re-written as:

σ σ σh L fail− = (13)

or

σσ

σσ

h

fail

L

fail

− =1 (14)

Considering that, due to presence of the corrosion defect, the failure stress in the hoop direction is different from the failure stress in the longitudinal direction, Equn 14 can be re-written as:

σσ

σσ

h

fail h

L

fail L( ) − ( ) =1

(15)

Equation 15 is a general format equation that can be used to derive different assessment methods [24] depending on the expressions adopted to calculate the failure stresses (σfail)h and (σfail)L.

h

Fig.2. Failure criterion based upon the Tresca yield criterion.

Sample

copy

- not

for di

stribu

tion


The failure stress in the longitudinal direction is given by:

σ σfail L flow R Lf( ) = ( ) (26)

The design pressure is given by:

p tDd d yield

e

= γ σ2

(27)

Failure pressure

The failure pressure of the corroded pipeline subjected to the pressure loading only is given by:

p tD

d td t Mf press

eflow

area

area( ) =

−− −

2 11 1σ

αα

( / )( / )

(28)

The failure pressure of the corroded pipeline subjected to the pressure loading plus a longitudinal compression is given by the following equations:

p tD

Hf combe

fail h f( ) = ( )2σ

(29a)

p p if p pf comb f press f comb f press( ) = ( ) ( ) > ( ) (29b)

where

H nf

L add i

fail L

fail h

fail L

=+

−

1

1

Σ ( )( )( )

( )

σσσσ (30)

The derivation of Equn 29a was performed using Equns 15, 9, 2, and 1.

Equation 29b is a consistency check to guarantee that the failure criterion in the circumferential direction is not violated (Equn 17)).

Equation 29a is valid provided that the following inequalities hold:

σ σfail L fail hn( ) > ( ) (31)

H f > 0 (32)

In order to be consistent with the physical behaviour, the internal pressure must be positive ((pf)comb > 0). Consequently the parameter Hf in Equn 29a must

standard N-2786 [27], released in 2005. The method was developed based on a failure criterion which checks the pipeline plastic collapse. It is assumed that the pipeline’s global buckling is prevented, and the local buckling in the corrosion defect region will be checked in parallel by another method.

Basic equations

The flow stress of the material is given by:

σ σflow yield MPa= + 69 (18)

The reduction factor in the hoop direction is given by:

f d td t MR h

area

area

( ) =−

− −

11 1

αα

( / )( / ) (19)

where

αareae

if LD t

= ≤0 85 202

. (20)

αarea

e

eLD t

if LD t

= −×

>1 0 15 64 10 206

2 6

2

.

(21)

M LD t

LD t

if LD te e e

= + −

≤1 0 6275 0 003375 202 2 2 1 2

2

. .

/

(22)

M LD t

if LD te e

= + >2 1 0 07 202 2

. . (23)

Equations 18 to 23 are identical to the ones adopted by the RPA method [20].

The failure stress in the hoop direction is given by:

σ σfail h flow R hf( ) = ( ) (24)

The reduction factor in the longitudinal direction is given by:

f dt

wDR L

e

( ) = −

1

π (25)

The reduction factor in the longitudinal direction (Equn 25) is the ratio of the cross-section area of the corroded pipe (π De t – w d) to the cross section area of the uncorroded pipe (π De t).

Sample

copy

- not

for di

stribu

tion


p p if p pa comb a press a comb a press( ) = ( ) ( ) > ( ) (39b)

where

H na

L add i

allow L

allow h

allow L

=+

−

1

1

Σ ( )( )( )

( )

σσσσ

(40)

The derivation of Equn 39a was performed using Equns 35, 9, 2, and 1. The equation is valid provided that the following inequalities hold:

σ σallow L allow hn( ) > ( )

(41)

Ha > 0 (42)

In order to be consistent with the physical behaviour, the internal pressure (pa)comb must be positive ((pa)comb > 0). Consequently the parameter Ha in Equn 39a must be positive (Ha > 0). If the parameter Ha is negative or null (Ha ≤ 0) there is no room for the pressure loading and the additional loadings must be reduced.

If the condition expressed by the inequality (41) is not fulfilled, it is necessary to use Equn 36 (if σL is negative) or Equns 37 and 38 (if σL is positive) to obtain a set of (pa)comb and additional loadings which comply with the allowable stress criterion.

DNV RP-F101 method for combined loadings

The DNV RP-F101 method for combined loadings (the DNV CL method) was developed in the 1990s during the Reliability of corroded pipes JIP carried out by DNV, and was first published in the Section 7.3 of the DNV RP-F101 [16] released in 1999. Further details about the method were given by Bjornoy, Sigurdsson, and Marley [15]. The DNV CL method is recommended in the Section 8.3 of the current edition of DNV RP-F101 [26], released in 2004.

Basic equations

The flow stress of the material is given by:

σ σflow ult= (43)

The reduction factor in the hoop direction is given by:

f d td t MR h

area

area

( ) =−

− −

11 1

αα

( / )( / ) (44)

be positive (Hf > 0). If the parameter Hf is negative or null (Hf ≤ 0), there is no room for the pressure loading and the additional loadings must be reduced.

If the condition expressed by the inequality (31) is not fulfilled, it is necessary to use Equn 15 (if σL is negative) or Equns 16 and 17 (if σL is positive) to obtain a set of (pf)comb and additional loadings which comply with the failure stress criterion.

Allowable pressure

The allowable pressure (pa)press of the corroded pipeline subjected to pressure loading only is given by:

p pa press d f press( ) = ( )γ (33a)

p p if p pa press d a press d( ) = ( ) > (33b)

Under service loadings, the state of stress at any point of the pipe wall is limited by the following equation:

σ σ σh L allow− = (34)

or

σσ

σσ

h

allow

L

allow

− =1 (35)

Due to presence of the corrosion defect, the allowable stress in the hoop direction (σallow)h is different from the allowable stress in the longitudinal direction (σallow)L.

σσ

σσ

h

allow h

L

allow L( )−( )

=1 (36)

The allowable stress in the hoop direction is given by:

σ γ σallow h eqn flow R hf( ) = ( ) (37)

The allowable stress in the longitudinal direction is given by:

σ γ σallow L eqn flow R Lf( ) = ( ) (38)

The allowable pressure (pa)comb of the corroded pipeline subjected to the pressure loading plus a longitudinal compression is given by the following equations:

p tD

Hf combe

allow h a( ) = ( )2σ

(39a)

Sample

copy

- not

for di

stribu

tion


σ σfail L fail h( ) > ( )0 5. (53)

H f > 0 (54)

It is proposed that if the condition expressed by the inequality (53) is not fulfilled, it is necessary to use Equn 15 (if σL is negative) or Equns 16 and 17 (if σL is positive) to obtain a set of (pf)comb and additional loadings which comply with the failure stress criterion.

Allowable pressure

The allowable pressure (pa)comb of the corroded pipeline subjected to the pressure loading plus a longitudinal compression is given by the following equation:

p pa comb m d f comb( ) = ( )γ γ

(55)

Comments about the RPA-PLLC and the DNV RP-F101 method for combined loadings

Even though there are many similarities between the RPA-PLLC method and the DNV RP-F101 method for combined loadings, there still remain many differences.

The similarities are the failure criterion based upon the Tresca yield criterion (Equn 15), the general format of the reduction factor in the hoop direction (Equns 19 and 44), and the reduction factor in the longitudinal direction (Equns 25 and 48).

The main differences are the pipeline restraint considered in the longitudinal direction (the difference is clearly detected by comparing Equns 30 and 52), the flow-stress expression (Equns 18 and 43), the αarea parameter expression (Equns 20, 21, and 45), the bulging factor expression (Equns 22, 23, and 46), and the allowable pressure expression (Equns 39a and 55).

DNV laboratory test results

Full-scale tests were performed as part of the DNV JIP, and a detailed description of these tests and their results were published by Bjornoy et al. [13]. Only a brief description will be presented here.

Twelve burst tests were performed, of which nine were with external longitudinal corrosion and three with external circumferential corrosion defects. The tubular specimens were loaded with internal pressure plus an external load, except for two specimens which were loaded with internal pressure only. The external load considered was a longitudinal compressive force or a bending moment.

where

αarea =1 0. (45)

M LD te

= +

1 0 31

2 1 2

./

(46)

The failure stress in the hoop direction is given by:

σ σfail h flow R hf( ) = ( ) (47)

The reduction factor in the longitudinal direction is given by:

f dt

wDR L

e

( ) = −

1

π (48)

The failure stress in the longitudinal direction is given by:

σ σfail L flow R Lf( ) = ( ) (49)

Failure pressure

The failure pressure considering the corroded pipeline subjected to the pressure loading only is given by:

p tD t

d td t Mf press

eult( ) =

−−

− −

2 11 1( )

( / )( / )

σ (50)

The failure pressure (pf)comb of the corroded pipeline subjected to the pressure loading plus a longitudinal compression is given by the following equations:

p tD

Hf combe

allow h f( ) = ( )2σ (51a)

p p if p pf comb f press f comb f press( ) = ( ) ( ) > ( )

(51b)

where

H f

L add i

fail L

fail h

fail L

=+

−

1

10 5

Σ ( )( )

. ( )( )

σσσ

σ (52)

Although in the DNV RP-F101 [16, 26] there is no comment about the range of validity of Equn 51a, this equation is valid only if the following inequalities hold:

Sample

copy

- not

for di

stribu

tion


API 5L X-52 steel (SMTS = 455.1 MPa). The ratio of the mean ultimate tensile strength to the mean yield strength is equal to 1.35. This value is greater than the ratio SMTS / SMYS ((SMTS / SMYS) = 1.27).

Table 1 presents the dimensions of the tubular specimens and the applied failure loadings.

Table 2 presents the values of the flow stress σflow adopted by the two assessment methods, calculated using Equns 18 and 43. The ratio of the flow stress (σflow) DNV adopted by the DNV RP-F101 method for combined loadings to the flow stress (σflow)RPA adopted by the RPA-PLLC method is also presented in this table. The most conservative expression is the one adopted by the RPA-PLLC method, and the less conservative is the one adopted by the DNV RP-F101 method for combined loadings.

For the specimens in the DNV tests, the flow stress adopted by the DNV RP-F101 method for combined loadings is 1.14 times the flow stress adopted by the RPA-PLLC method.

The specimens were cut from seamless tubes made of API 5L X-52 steel with a nominal outside diameter of 324 mm (12.75 in) and a nominal wall thickness of 10.3 mm (0.406 in).

The corrosion defects were longitudinal uniform-depth defects generated using spark erosion, and mechanically machined circumferential uniform-depth defects.

The tensile specimens were tested to determine the material properties. The average yield strength and the average ultimate tensile strength determined were 380 MPa and 514 MPa, respectively. The mean yield strength is 6.0% greater than the SMYS of API 5L X-52 steel (SMYS = 358.5 MPa). The mean ultimate strength is 12.9% greater than the SMTS of

Table 1. Dimensions of the tubular specimens and the applied failure loadings. Note: In the specimens subjected to bending moment, the corrosion defect is located on the compressed side of the specimen’s cross section.

Specimend

(mm)L

(mm)w

(mm)d–t

L2

–––Det

w–––π De

applied failure loadings

pf (MPa)

Nc (N)

Mb (N.mm)

1 5.15 243.0 154.5 0.50 17.7 0.15 23.20 - -

2 5.15 243.0 154.5 0.50 17.7 0.15 21.90 - -1.290E+08

3 5.15 243.0 154.5 0.50 17.7 0.15 19.50 - -2.120E+08

4 3.09 162.0 30.9 0.30 7.9 0.03 29.00 - -7.300E+07

5 3.09 162.0 30.9 0.30 7.9 0.03 28.60 -2.563E+06 -

6 3.09 162.0 30.9 0.30 7.9 0.03 28.70 -2.943E+06 -

7 5.15 243.0 30.9 0.50 17.7 0.03 18.60 -3.000E+06 -

8 5.15 243.0 30.9 0.50 17.7 0.03 22.00 - -

9 7.21 243.0 30.9 0.70 17.7 0.03 12.30 -2.070E+06 -

10 5.15 12.0 1017.9 0.50 0.04 1.00 32.00 -2.289E+06 -

11 5.15 12.0 1017.9 0.50 0.04 1.00 33.50 -2.343E+06 -

12 7.21 12.0 1017.9 0.70 0.04 1.00 32.10 -2.399E+06 -

De = 324 mm

t = 10.3 mm

σyield = 380 MPa

σult = 514 MPa

Table 2. Flow stress adopted by the assessment methods.

(σflow)DNV (MPa)

(σflow)RPA (MPa)

(σflow)DNV /(σflow)RPA

514 449 1.14

Sample

copy

- not

for di

stribu

tion


(pf)comb. For this reason, the longitudinal force Nc and bending moment Mb predicted by the DNV CL method presented in Table 2 are 100% accurate.

For the RPA-PLLC method, the external loading (longitudinal force Nc or bending moment Mb) applied in the test was input to the procedure applied by the assessment method to calculate the failure pressure (pf)comb except for the specimens 10, 11, and 12. For these specimens it was necessary to reduce the absolute value of the longitudinal force Nc in order to obtain a non null value of the failure pressure (pf)comb. These calculations were performed using Equn 15 and imposing the condition

Comparison between test results and assessment method results

Table 3 presents the failure loadings applied in the laboratory tests of the 12 tubular specimens along with those predicted by the DNV RP-F101 method for combined loadings (DNV CL method) and the RPA-PLLC method.

For the DNV CL method the external loading (longitudinal force Nc or bending moment Mb) applied in the test was the input to the procedure applied by the assessment method to calculate the failure pressure

Table 3: Actual and predicted failure loadings.

Specimen Test DNV CL RPA-PLLC

(pf)test (MPa)

(Nc)test

(N)(Mb)test

(N.mm)(pf)comb

(MPa)(Nc)method

(N)(Mb)method

(N.mm)(pf)comb

(MPa)(Nc)method

(N)(Mb)method

(N.mm)

1 23.20 0.0 0.0 21.00 0.0 0.0 18.82 0.0 0.0

2 21.90 0.0 -1.290E+08 20.51 0.0 -1.290E+08 17.47 0.0 -1.290E+08

3 19.50 0.0 -2.120E+08 13.34 0.0 -2.120E+08 9.88 0.0 -2.120E+08

4 29.00 0.0 -7.300E+07 28.19 0.0 -7.300E+07 23.80 0.0 -7.300E+07

5 28.60 -2.563E+06 0.0 24.57 -2.563E+06 0.0 17.77 -2.563E+06 0.0

6 28.70 -2.943E+06 0.0 20.99 -2.943E+06 0.0 14.32 -2.943E+06 0.0

7 18.60 -3.000E+06 0.0 12.77 -3.000E+06 0.0 9.38 -3.000E+06 0.0

8 22.00 0.0 0.0 21.00 0.0 0.0 18.82 0.0 0.0

9 12.30 -2.070E+06 0.0 10.53 -2.070E+06 0.0 10.09 -2.070E+06 0.0

10 32.00 -2.289E+06 0.0 33.53 -2.289E+06 0.0 28.27 -2.257E+06 0.0

11 33.50 -2.343E+06 0.0 33.53 -2.343E+06 0.0 28.27 -2.257E+06 0.0

12 32.10 -2.399E+06 0.0 27.61 -2.399E+06 0.0 28.00 -2.235E+06 0.0

Table 4. Ratios of the predicted to the actual failure loadings.

Specimen DNV CL RPA-PLLC

(pf)test / (pf)comb (Nc)test / (Nc)method (Mb)test / (Mb)method (pf)test / (pf)comb (Nc)test / (Nc)method (Mb)test / (Mb)method

1 0.91 - - 0.81 - -

2 0.94 - 1.00 0.80 - 1.00

3 0.68 - 1.00 0.51 - 1.00

4 0.97 - 1.00 0.82 - 1.00

5 0.86 1.00 - 0.62 1.00 -

6 0.73 1.00 - 0.50 1.00 -

7 0.69 1.00 - 0.50 1.00 -

8 0.95 - - 0.86 - -

9 0.86 1.00 - 0.82 1.00 -

10 1.05 1.00 - 0.88 0.99 -

11 1.00 1.00 - 0.84 0.96 -

12 0.86 1.00 - 0.87 0.93 -

Sample

copy

- not

for di

stribu

tion


Table 4 presents the ratios of the predicted to the actual failure loadings for the 12 tubular specimens, and Fig.3 present the ratios of the predicted to the actual failure pressures. The maximum accuracy that one method could achieve would be a ratio (pf)comb / (pf)test equal to the unity.

A majority of the predicted failure pressures are conservative, i.e. the ratios (pf)comb / (pf)test presented in Table 4 are smaller than or equal to the unity. Among the failure pressures predicted by the DNV CL method there was only one unconservative ((pf)comb / (pf)test > 1). Among the failure pressures predicted by the RPA-PLLC method none was unconservative.

Table 5 presents the errors of the failure pressure predictions.

The DNV CL method was the method that predicted the failure pressures closest to the actual failure pressures. The DNV CL method presented a mean error equal to 13.4% and a standard deviation equal to 10.6% while the RPA-PLLC method presented a mean error equal to 26.4% and a standard deviation equal to 14.9%.

The DNV CL method predicted five failure pressures to be adequately conservatives (0% ≤ |error| ≤ 10%), four failure pressures as moderately conservative (10% < |error| ≤ 30%), two failure pressures as overly conservative (|error| > 30%), and one failure pressure as unconservative (error > 0%).

The RPA-PLLC method predicted eight failure pressures to be moderately conservative (10% < |error| ≤ 30%) and four failure pressures to be overly conservative (|error| > 30%).

In relation to the DNV CL method, it is important to mention that in the determination of the allowable pressure according to this method, besides the usual

that the longitudinal stress σL is null (σL = (σL)p – (σL)Nc = 0). Actually the condition that the longitudinal stress σL is null was the loading condition under which the burst tests of specimens 10, 11, and 12 were carried out [28]. For this reason the longitudinal force Nc predicted by the RPA-PLLC method for the specimens 10, 11, and 12 (see Table 2) are, respectively, 1.41%, 3.68%, and 6.82% smaller than those applied in the test.

In the input data for the RPA-PLLC method, the tubular specimens were considered longitudinally unrestrained (n = 0.5) because they were end-capped and longitudinally unrestrained.

Specimen DNV CL RPA-PLLC

1 -9.48 -18.87

2 -6.33 -20.23

3 -31.56 -49.31

4 -2.80 -17.91

5 -14.10 -37.85

6 -26.88 -50.12

7 -31.32 -49.56

8 -4.55 -14.45

9 -14.40 -17.96

10 4.79 -11.66

11 0.09 -15.61

12 -13.99 -12.77

mean 13.4 26.4

standard deviation 10.6 14.9

Table 5. Error between the failure pressure predictions (%). Note 1: error (%) = ((predicted – experimental) / experimental) x 100 Note 2: mean = ∑ |error|/12

Fig.3. Ratios of the predicted to the actual failure pressure.

Sample

copy

- not

for di

stribu

tion


the DNV CL method, there was only one that was unconservative. Among the failure pressures predicted by the RPA-PLLC method, none were unconservative.

With the exception of specimen 12, all the failure pressures predicted by the DNV CL method are greater than the ones predicted by the RPA-PLLC method.

The DNV CL method predicted the failure pressures closest to the actual failure pressures. The DNV CL method presented a mean error equal to 13.4% and a standard deviation equal to 10.6%, while the RPA-PLLC method presented a mean error equal to 26.4% and a standard deviation equal to 14.9%.

In relation to the DNV CL method, it is important to mention that in the determination of the allowable pressure according to this method, besides the usual design factor, a modelling factor of 0.9 is always applied. This additional safety factor may be interpreted as a de-rated factor that should be applied directly to the flow stress established in Equn 43. According to this approach, the value of the flow stress adopted by the DNV CL method would be 90% of the ultimate tensile stress (σflow = 0.90 σult) instead of 100% of the ultimate tensile stress (σflow = σult). If this de-rated flow stress had been adopted the results of the DNV CL method presented in Table 3 would be similar to the ones predicted by the RPA-PLLC method (mean error equal to 29.1% and standard deviation equal to 19.2%).

It is also worth remembering that in any set of experimental results there is always some scatter.

design factor, a modelling factor of 0.9 is always applied. This additional safety factor may be interpreted as a de-rated factor that should be applied directly on the flow stress established in Equn 43. According to this approach, the value of the flow stress adopted by the DNV CL method would be 90% of the ultimate tensile stress (σflow = 0.90 σult) instead of 100% of the ultimate tensile stress (σflow = σult). If this de-rated flow stress had been adopted, the results of the DNV CL method presented in Table 3 would be similar to the ones predicted by the RPA-PLLC method (mean error equal to 29.1% and standard deviation equal to 19.2%).

It is also worth remembering that in any set of experimental results there is always some scatter.

Table 6 presents the failure pressure ((pf)comb)DNV predicted by the DNV CL method and the failure pressure ((pf)comb)RPA predicted by the RPA-PLLC method. The ratio ((pf)comb)DNV / ((pf)comb)RPA is also presented in this table. Except for specimen 12, all the failure pressures predicted by the DNV CL method are greater than the ones predicted by the RPA-PLLC method. Apart from specimen 12, the ratios ((pf)comb)DNV / ((pf)comb)RPA are between 1.04 and 1.47.

Conclusions

In this paper, a comparison between the DNV RP-F101 method for combined loadings and the RPA-PLLC method was presented. Even though there are many similarities between the RPA-PLLC method and the DNV RP-F101 method for combined loadings (DNV CL method), there still remain many differences.

The similarities are the failure criterion based upon the Tresca yield criterion, the general format of the reduction factor in the hoop direction, and the reduction factor in the longitudinal direction.

The main differences are the pipeline restraint considered in the longitudinal direction, the flow stress expression, the αarea parameter expression, the bulging factor expression and the allowable pressure expression.

The two methods were developed based on a failure criterion which checks the pipeline’s plastic collapse. It is assumed that the pipeline’s global buckling is prevented and the local buckling in the corrosion defect region will be checked in parallel by another method.

The results of the full scale tests performed as part of the Reliability of corroded pipes JIP, carried out by DNV in the 1990s, were used to evaluate the performances of the DNV CL and RPA-PLLC methods.

A majority of the predicted failure pressures are conservative. Among the failure pressures predicted by

Table 6. Predicted failure pressures.

Specimen ((pf)comb)DNV (MPa)

((pf)comb)RPA

(MPa)((pf)comb)DNV ((pf)comb)RPA

1 21.00 18.82 1.12

2 20.51 17.47 1.17

3 13.34 9.88 1.35

4 28.19 23.80 1.18

5 24.57 17.77 1.38

6 20.99 14.32 1.47

7 12.77 9.38 1.36

8 21.00 18.82 1.12

9 10.53 10.09 1.04

10 33.53 28.27 1.19

11 33.53 28.27 1.19

12 27.61 28.00 0.99

Sample

copy

- not

for di

stribu

tion


14. O.H.Bjornoy and M.J.Marley, 2001. Assessment of corroded pipelines: past, present and future. 11th International Offshore and Polar Engineering Conference, ISOPE 2001.

15. O.H.Bjornoy, G.Sigurdsson, and M.J.Marley, 2001. Background and development of DNV-RP-F101 corroded pipelines. 11th International Offshore and Polar Engineering Conference, ISOPE 2001.

16. Anon, 1999. DNV Recommended Practice – DNV-RP-F101 – Corroded pipelines. Det Norske Veritas, Norway.

17. D.R.Stephens, T.A.Bubenik, and R.B.Francini, 1995. Residual strength of pipeline corrosion defects under combined pressure and axial loads. Final Report to Line Pipe Research Supervisory Committee of the Pipeline Research Committee of the American Gas Association, NG-18 Report No. 216, A.G.A. Catalog No. L51722, Battelle Memorial Institute, February.

18. K.A.Roberts and R.J.Pick, 1998. Correction for longitudinally stress in the assessment of corroded line pipe. 2nd International Pipeline Conference, 1, pp 553-561.

19. A.C.Benjamin, 2008. Prediction of the failure pressure of corroded pipelines subjected to a longitudinal compressive force superimposed on the pressure loading. International Pipeline Conference, September.

20. A.C.Benjamin and E.Q.Andrade, 2003. Modified method for the assessment of the remaining strength of corroded pipelines. Rio Pipeline Conference, October.

21. J.F.Kiefner and P.H.Vieth, 1989. A modified criterion for evaluating the remaining strength of corroded pipe. Final Report on Project PR 3-805, Pipeline Research Committee, American Gas Association.

22. Anon, 2006. ASME B31.4-2006 Pipeline transportation systems for liquid hydrocarbons and other liquids. The American Society of Mechanical Engineers, October.

23. L.M.Kachanov, 1971. Foundations of the theory of plasticity. North-Holland Publishing Co, London.

24. A.Cosham, 2002. Assessment methods for corrosion in pipelines. A report to the Pipeline Defect Assessment Manual (PDAM) Joint Industry Project, Report NR99012/4238.1.72, Revision 3, September.

25. Anon, 1991. ASME-B31G – Manual for determining the remaining strength of corroded pipelines – a supplement to ANSI/ASME B31 Code for pressure piping. The American Society of Mechanical Engineers, New York.

26. Anon, 2004. DNV Recommended Practice – DNV-RP-F101 – Corroded pipelines. Det Norske Veritas, Norway.

27. Anon, 2005. Petrobras Standard N-2786 – Assessment of defects and failure modes of land and submarine steel pipelines. Petrobras, NORTEC, July.

28. O.H.Bjornoy, G.Sigurdsson, and E.H.Cramer, 1997. Laboratory burst tests. DNV JIP Reliability of corroded pipes. Report 96-3393, Revision 2, 7 May 1997.

References

1. S.Roy, M.Q.Smith, S.C.Grigory, H.R.Couque, and M.F.Kanninen, 1995. The development of methodologies for determining the residual strength of corroded line pipe under combined loading. 50th NACE Annual Conference and Corrosion Show (CORROSION’95), pp 22/1,22/12.

2. H.R.Couque, M.Q.Smith, S.C.Grigory, and M.F.Kanninen, 1996. The development of methodology for evaluating the integrity of corroded pipelines under combined loading - Part 1: Experimental testing and numerical simulations. Pipelines, Terminals & Storage Conference, API-ASME Pennwell Publishing Co Energy Week, pp 58-66.

3. Ibid. Part 2: Engineering model and PC program development, pp 67-76.

4. S.C.Grigory and M.Q.Smith, 1996. Residual strength of 48-inch diameter corroded pipe determined by full scale combined loading experiments. 1st International Pipeline Conference, 1, pp 377-386.

5. M.Q.Smith and S.C.Grigory, 1996. New procedures for the residual strength assessment of corroded pipe subjected to combined loads. 1st International Pipeline Conference, 1, pp 387-400.

6. S.Roy, S.C.Grigory, M.Q.Smith, M.F.Kanninen, and M.Anderson, 1997. Numerical simulations of full scale corroded pipe tests with combined loading. J.Press.Vessel Technology, 119, pp 457-466, November.

7. M.Q.Smith, D.P.Nicollela, and C.J.Waldhart, 1998. Full-scale wrinkling tests and analyses of large diameter corroded pipes. 2nd International Pipeline Conference, 1, pp 543-551.

8. W.Wang, M.Q.Smith, C.H.Popelar, and J.A.Maple, 1998. A new rupture prediction model for corroded pipelines under combined loadings. 2nd International Pipeline Conference, 1, pp 563-572.

9. M.Q.Smith and C.J.Waldhart, 2000. Combined loading tests of large diameter corroded pipelines. 3rd International Pipeline Conference, 2, pp 769-779.

10. O.H.Bjornoy, E.H.Cramer, and G.Sigurdsson, 1997. Probabilistic calibrated design equation for burst strength assessment of corroded pipes. 7th Offshore and Polar Engineering Conference (ISOPE’97), 4, pp 160-166.

11. O.H.Bjornoy, G.Sigurdsson, E.H.Cramer, B.Fu, and D.Ritchie. 1999. Introduction to DNV-RP-F101. 19th International Conference on Offshore Mechanics and Arctic Engineering, OMAE 99.

12. G.Sigurdsson, E.H.Cramer, O.H.Bjornoy, B.Fu, and D.Ritchie, 1999. Background to DNV-RP-F101 corroded pipelines. 19th International Conference on Offshore Mechanics and Arctic Engineering, OMAE 99.

13. O.H.Bjornoy, G.Sigurdsson, and E.H.Cramer, 2000. Residual strength of corroded pipelines, DNV test results. 10th International Offshore and Polar Engineering Conference, ISOPE 2000.

Sample

copy

- not

for di

stribu

tion


RATES OF PIPELINE FAILURE in Australia seem to be substantially lower than in the Americas and Europe, at only 10-20% of the international mean for failures in onshore transmission pipelines. This

paper examines the validity of the Australian data and then explores reasons for the difference. Some reasons are obvious, such as the relative youth of Australian pipelines which results in a negligible rate of corrosion failures. However there is no obvious explanation for the markedly lower rate of failures due to third-party damage. It is hypothesized that Australian practices for managing third-party damage may differ in some way. Given the high social and economic cost of pipeline failures, there should be a comparative study to identify any beneficial differences between third-party damage protection practices in Australia and elsewhere.

This paper was presented at the Joint Technical Meeting held between the APIA, EPRG, and PRCI in Australia in April, 2013, and is reproduced by kind permission of the APIA Pipeline Operators Group and the meeting’s organizers.

* Corresponding author’s contact details:tel: +61 2 9983 1511email: [email protected]

1 Peter Tuft & Associates, Sydney, Australia2 Petrobras Transporte – Transpetro, Rio de Janeiro, Brazil

Comparing international pipeline failure rates

by Peter Tuft* and Sergio Cunha

AT FACE VALUE the failure rates of onshore transmission pipelines in Australia appear to be

much lower than in the Americas and Europe. The difference is large (note that the failure rates are loss-of-containment events per 1000 km-yr for the most recent five-year period. Derivation of these values is explained later in the paper):

International Australia Ratio

Gas pipelines 0.15 0.032 21%

Oil pipelines 0.28 0.032 11%

This striking difference demands explanation and raises a number of questions that will be addressed in this paper:

• Are the Australian data valid?• If so, is there something inherent in Australian

pipelines or their environment that minimizes the likelihood of pipeline failures?

• Or is there something different about Australian pipeline management that reduces the number of failures?

The objective of this paper is to draw attention to what appear to be real differences in some aspects of pipeline failure rates in order to encourage further investigation into the causes of those differences so that they might be applied, if possible, for a wider benefit.

(Throughout this paper ‘failure’ refers to loss of containment, while ‘incident’ is more general and can include damage without loss of containment, and near misses.)

Failure rates in the Americas and Europe

A comprehensive overview of loss-of-containment rates (excluding Australia) was presented by Cunha [1], and Tables 1 and 2 from that paper are reproduced here. More complete references to the data sources can be found in the original paper. Note the variation in reporting criteria from different regions. For gas pipelines, most regions except the US include all events involving loss of containment. Because the US database contains only events that exceed thresholds for severity1, it is possible that US rates are under-reported relative to other regions.

For the purpose of this paper we will focus mainly on the international mean values (highlighted in the tables) for comparison with Australian data.

The Australian pipeline-incident database

The Australian pipeline industry has been collecting incident data since about 1965; at least, the earliest recorded incidents occurred in 1965 and the oldest transmission pipelines in Australian were only constructed around 1960.

1 Significant incidents involve any of the following: a. fatality or injury requiring in-patient hospitalization. b. $50,000 or more in total costs, measured in 1984 dollars (equivalent

to about $115,000 at the end of 2012).

Sample

copy

- not

for di

stribu

tion


types is an attempt to learn from the more-frequent events involving minor damage (about 130 items in the database) and the much-more-frequent events involving near misses (about 460 items).

Since 2007 the scope of the database has also included incidents from New Zealand, but the NZ data are not included in this analysis.

The database contains about 100 fields to record data about the pipe itself, the events causing the incident, details of any damage and repairs, and operating practices (particularly relating to external interference protection).

Reporting is voluntary and data are collected by the Australian Pipeline Industry Association from members of its Pipeline Operators Group (POG). Practices for reporting and recording incidents have developed over time and there was a period when it appears that reporting was less rigorous. However we are confident that the data collected since about 2002 are reasonably

Pipelines for which incident data are collected are gas and liquid transmission pipelines that are required to comply with AS 2885 [2]. Maximum allowable operating pressure must be above 1050 kPa.

The Australian incident database collects three broad categories of incident:

• loss-of-containment events (LoC)• pipeline damage that required repair but did

not result in leak or rupture (gouges, dents, deformation, etc., but excluding routine corrosion repairs)

• near misses, comprising unauthorized third-party excavation activity on the pipeline easement.

There is on average only about one LoC event each year, although the number fluctuates widely and in recent years has ranged from zero to three. Limiting the data collection to these rare LoC events would provide the industry with very limited scope for learning. The collection of a broader range of incident

Region PeriodExposure (km-year)

Frequency of failure (/103 km-yr)Reporting criteria

HistoricFive-year average

Source

Europe 1970 – 2010 3.55 x 106 0.35 0.16 EGIG No lower limit.

Canada 2000 –2008 1.91 x 105 0.10 NA NEB Pipelines at 15 bar or more.

UK 1962 – 2010 7.73 x 105 0.23 0.093 UKOPA No lower limit.

USA 1985 – 1997 5.96 x 106 0.11 NA DOT–PRCIDeath, injury,

cost > US$ 50,000

Brazil 1978 – 2010 8.23 x 103 0.36 NA Transpetro No lower limit.

Average(1) 0.23 0.13

Mean(2) 0.20 0.15

Region PeriodExposure (km-year)

Frequency of failure (/103 km-yr) Reporting criteria

HistoricFive-year average

Source

Europe 1971 - 2010 1.01 x 106 0.55 0.28 CONCAWE 1 m3 release.

Canada 2000 - 2008 2.41 x 105 0.10 NA NEB 1.5 m3 release.

Brazil 1978 - 2010 6.13 x 104 0.70 0.23 Transpetro No lower limit.

Average 0.46 0.25

Mean 0.48 0.28

Table 2. Frequency of failure for liquid pipelines.

Table 1. Frequency of failure for gas pipelines. Note: (1) arithmetic mean of the frequency of failures; (2) total number of failures / total exposure.

Sample

copy

- not

for di

stribu

tion


the voluntary incident-reporting system lapsed among many operators during this period.

Before the 1990s it seems that the largely government-owned pipeline authorities reported LoC and damage incidents at a rate which appears plausibly realistic. Full reporting would have been consistent with what is known of the organizational cultures of the time. However there is no hard evidence that reporting of failures was complete.

Because there is some doubt about the completeness of early data, the analysis presented here de-emphasizes failure rates prior to 2002 although it is sometimes necessary to use the full data set because the small number of failures since 2002 is not always sufficient for meaningful analysis.

The most recent publicly available analysis of Australian incidents was presented at the Australian Pipeline Industry Association convention in 2009 [3]. The data presented in the current paper are based on more recent analyses, including failures since 2009.

Australian failure rates

The historic mean Australian failure rate, based on all recorded failures, is 0.063 failures per 1000 km-yr, based on a total of 43 LoC events2. The total exposure underlying this failure rate is 684,000 km-yr, which is in the middle of the range of the exposures for other regions listed in Tables 1 and 2.

The failure rate in recent years is more relevant than the overall historical rate for two important reasons. Firstly, more-recent Australian data have greater validity as discussed above. Secondly, if we are seeking lessons from the data about possible improvements to pipeline operation then it is important to look at data that

2 The 43 LoC events include 15 corrosion failures, mostly in the 1960s and 1970s, on a single pipeline that was particularly badly managed (it started leaking within a year or so of commissioning and is no longer in service). When reporting within Australia we usually omit these failures since they reflect practices no longer used and are highly unrepresentative. However for the purpose of comparing overall historical failure rates they are included here because the data for other regions may include failures on similarly poorly maintained lines.

complete, and that any omissions are insufficient to affect the overall conclusions drawn in this paper. That confidence comes mainly from two observations:

• POG members represent 94% of the total transmission pipeline length in Australia.

• Each year POG member companies submit a signed declaration that they have either reported all incidents or have had none.

In practice, we find that each year a few POG members fail to submit their declarations and/or incident reports by the deadline. At face value that has given an impression of poor reporting but we also find that the problem is mainly one of timing rather than non-reporting, since the missing data are almost always provided before the closure of the next reporting period.

The total length of pipelines considered in this analysis is 32,020 km, representing all pipelines operated by POG member companies that have reported consistently for the last few years. There is a further 1100 km operated by POG members that was omitted because recent incident reporting has been somewhat patchy (probably no incidents, but we are not completely certain). Non-members of POG operate another 2400 km of pipeline for which there are no recent incident data.

Gas pipelines comprise about 83% of the total Australian pipeline length and the remaining 17% carry diverse fluids including various forms of oil as well as LPG, ethane, etc. The incident database is capable of distinguishing gas pipelines from others, but the distinction has not been crucial to this analysis. In particular, because the overall failure rate is low, it is not very meaningful to present results on a small subset of the pipeline system for which the number of failures per year may be zero for many years in a row.

Data prior to 2002 should be treated with caution. A sharp drop in reported incidents during the 1990s corresponds to a transition of pipeline ownership from government authorities to various private operators. This was not a clean transition and many pipelines changed hands (or at least management) several times as the newly privatized industry slowly settled into a new regime that was commercially viable. It appears that

Table 3. International and Australian comparison.

Period International Australia Ratio

Historical(all data)

Gas pipelines 0.200.063

32%

Oil pipelines 0.48 13%

Five-year(most recent period)

Gas pipelines 0.150.032

21%

Oil pipelines 0.28 11%

Sample

copy

- not

for di

stribu

tion


The single event in 2006 was an SCC defect revealed by in-line inspection and found to be almost imperceptibly weeping when exposed for repair.

• It is also notable (and surprising) that there have been no LoC events caused by third-party damage since 2005. Third-party damage incidents do occur: there have been 31 pipeline strikes since 2002 but most have resulted in only minor harm to the pipe or coating.

• Of the five failures due to natural events, four were due to lightning plus one due to earth movement. There are three additional non-LoC lightning damage events (largely found through in-line inspection) plus one more leak in 2001 just prior to period covered by the Table 4. The fact that lightning damage is the single largest cause of LoC was unexpected and is likely to be the subject of further investigation within Australia.

Are the Australian data valid?

The answer to this question is implicit in the earlier description of the Australian incident-reporting system. While some questions may be raised about the overall historical rate derived from all failures since 1965, we are confident in the data collected since 2002.

It has been necessary in this analysis to make some assumptions or approximations. In every case (in both preceding and following sections of this paper)

reflects current practice and avoids past practices that are no longer applicable. The rolling five-year average failure rate was the basis for comparison of regions in Tables 1 and 2 and will be adopted here as well.

The current Australian five-year average is 0.032 failures per 1000 km-yr, but due to the very small number of failures each year, it fluctuates considerably and for 2011 was only 0.013 per 1000 km-yr. The rolling 10-year average hovers at around 0.028 per 1000 km-yr. Nevertheless, for consistency, we will use here the somewhat higher five-year value.

Table 3 compares international and Australian failure rates for both all data and the most-recent five-year period. The Australian values are markedly lower than the international mean and also significantly below any individual region reported in Tables 1 and 2.

Table 4 shows a breakdown of Australian LoC events since 2002. The last row of the table shows the average failure rate for the period since 2002, but given the erratic occurrence of failures due to any single cause, these figures are at best only indicative and are not discussed further. However it is possible to make some general comments and observations:

• The paucity of corrosion failures is notable. Further, apart from the badly managed pipeline mentioned in Footnote 2, there are no failures due to metal-loss corrosion in the entire database.

Year Total Third party CorrosionMaterial/

constructionNatural events

2002 3 2 1

2003 0

2004 0

2005 1 1

2006 2 1 1

2007 0

2008 0

2009 1 1

2010 0

2011 1 1

2012 3 1 2

Total 11 3 1 2 5

Average rate(1) 0.034 0.009 0.003 0.006 0.016

Table 4. Australian loss-of-containment events. Note: (1) average rate in failures per 1000 km-yr based on total exposure of 320,000 km-yr since 2002.

Sample

copy

- not

for di

stribu

tion


been only three third-party damage failures since 2002, one in a remote rural location and two in suburban areas. Hence breakdown against population density must be based on the whole data set back to 1965.

Table 5 summarizes the available data: in both urban and rural areas the Australian failure rate is a fraction of the international mean failure rate, and hence the low Australian rate cannot be attributed to a low level of third-party activity in vast expanses of unpopulated land.The international mean values in Table 5 are based on data from Europe, the UK, the US, and Brazil for the first row (all third-party failures), but only on UK data for the last two rows (see Tables 9 and 10 in [1]). Data collected from other regions do not allow analysis based on location class or population density. This difference in source data explains why the ratio of international to Australian failure rates for all third-party failures is not a weighted average of the rates for urban or rural areas but is lower than either. Because the overall UK third-party failure rate is little more than half of the international mean (see [1]), the figures in the ‘ratio’ column are probably inflated (perhaps roughly double) relative to the values they would take if a true international mean was available for comparison.

The basis for assigning failures to urban or rural areas in Table 5 also deserves explanation. AS 2885 nominates four location classes: R1 (broad rural), R2 (rural residential), T1 (suburban), and T2 (high density), and these roughly correspond to location classes 1 to 4 in US pipeline codes. In practice, R2 locations tend to have quite a lot of roads, underground services, farming activities, etc., so the level of third-party activity is much higher than in the more-remote R1 locations. However the comparison is largely with UK data, and Australian R2 areas are probably roughly comparable to UK rural areas. Hence for this analysis, Australian failures in location classes R1 and R2 are grouped in Table 5 as ‘Rural areas’, while T1 and T2 are grouped as ‘Urban areas’.

As it happens, there was only one failure in an R2 location, but grouping R1 and R2, and T1 and T2, also minimizes the exposure of pipeline in urban areas which has the effect of increasing the calculated failure rate. This approach is conservative in the sense that

we have taken the approach least favourable to the Australian data, and yet Australian failure rates still compare very favourably.

Further, the difference between recent Australian and international failure rates is so large that it far outweighs any possible under-reporting of Australian failures. It is not credible that under-reporting of Australian failures could be so extreme that the true Australian failure rate could be even doubled, let alone comparable to the international mean rates.

Are Australian circumstances different?

Population density: influence on third-party damage

A natural reaction on learning of the low Australian failure rate is to assume that Australian pipelines are in unpopulated desert areas and hence largely unaffected by third-party activities. While it is true that Australia has thousands of kilometres of pipelines in remote locations it also has a significant proportion in more populated areas including semi-rural regions, city outskirts, and within cities. It is helpful to break down failure rates based on location class.

As a factor influencing failures rates, location classification is relevant only to failures caused by third-party damage. All other things being equal, the likelihood of third-party damage depends mainly on the level of third-party activity around a pipeline, which depends on the population density that is in turn reflected in the location class. In general it seems reasonable to expect that failures due to corrosion, natural events, and defects in material or construction should be largely independent of the land use and population density around the pipeline. At most, failures in these categories may be influenced by location class only insofar as the pipeline operator may take additional precautions (for example, more intensive corrosion monitoring) in locations where the consequences of failure are higher, but that is a secondary effect that can be ignored for the present purposes.

Because third-party damage failures have been so infrequent in Australia it is not meaningful to analyse population effects in the most recent period; there have

Table 5. Third-party failure rates. Note: (1) For these rows the values in the ‘international’ column are for UK only.

Location International or UK Australia RatioNo. of Aust.

failuresAust. pipeline

length, km

All third-party failures 0.09 0.022 24% 15 31,200

Urban areas (1) 0.18 0.066 37% 9 2,100

Rural areas (1) 0.049 0.016 33% 6 29,100

Sample

copy

- not

for di

stribu

tion


the single poorly protected pipeline that suffered 15 external corrosion leaks mostly in the 1960s and 1970s. There have been no other metal-loss corrosion failures.

There have been only two stress-corrosion-cracking failures, both on the same pipeline. SCC occurs too erratically and is too pipeline-specific for any broad conclusions to be possible, and so is not discussed further here.

There have been four failures due to material or construction defects, all pinhole leaks in girth welds (3) or a seam weld (1). Two occurred in 1983 (on the same poorly managed pipeline as the multiple corrosion leaks), the others in 2002 and 2012. With such sparse failures it is not very meaningful to calculate average failure rates.

The relative youth of Australian pipelines is the most obvious explanation for these low rates of failure due to corrosion or defects. It is possible that there may be other factors contributing to the low corrosion failure rate but deeper investigation would be necessary to uncover them.

Are Australian circumstances different for failures due to corrosion or defects in material or construction? Perhaps, in that most Australian pipelines are relatively young and well protected.

Natural causes

In terms of overall failure rate due to natural causes, the Australian average is not very different from the international mean. Historical and five-year rates are 0.010 and 0.026 respectively (but the latter is highly variable), compared with international means of 0.018 and 0.014 respectively. There is no obvious conclusion to be drawn here.

However there is a difference in the distribution of failures due to natural causes. There have been only seven Australian failures due to natural causes, two (29%) due to earth movement and five (71%) due to lightning. In contrast, in the Americas and Europe earth movement accounts for 50 - 100% of natural-event failures and lightning causes only 0 - 10%.

The low rate of geologically caused failures is not surprising given the stable terrain in most of Australia. However the high rate of lightning failures is unexpected and perhaps reflects a high-lightning environment, but that is only speculation and more investigation is necessary.

It seems that the Australian environment is rather different for pipeline failures due to natural events, although this is reflected in the distribution of failure causes rather than the overall failure rate.

any resulting bias will be in the direction of making the Australian failure rate in populated areas higher rather than lower.

Another possible explanation for the low failure rate is that Australian pipelines are better protected against external interference through greater wall thickness or deeper burial. We believe that neither is the case. Most Australian pipelines are relatively small diameter and hence thin-walled. There are few pipelines larger than DN 500 (20 in) (the maximum existing size is DN 750 (30 in) but DN 1050 (42 in) currently under construction). Wall thicknesses are typically under 10 mm, often around 6 mm or down to 4.8 mm on small-diameter lines. In populated areas, pipelines built since the late 1990s have increased thickness but there are many older pipelines in urban areas that are thin. Standard burial depth in rural areas is 750 mm although some pipelines are at 900 mm cover while in urban areas the minimum cover is 1200 mm. These thicknesses and depths do not appear to provide unusually robust protection.

In answer to the question “Are Australian circumstances different?”, we would argue that the answer is “No” but the failure rate is low nevertheless.

Pipeline age: influence on construction defects and corrosion failures

Australian pipelines are relatively young, which offers several benefits, including the use of modern standards for materials and construction, the use of modern practices for corrosion protection, and less time to accumulate deterioration.

Roughly 80-85% of Australian pipelines were built after 1980 when factory-applied coatings on grit-blasted pipe became standard practice and replaced over-the-ditch coatings on poorly prepared pipe surface. Good cathodic-protection systems have been applied since at least the 1970s. Also, since the 1970s, there have been generally high standards of specification and quality assurance for materials and construction. Government ownership of many pipelines built between about 1970 and 1990 may have been a contributing factor to generally high standards, as the culture of the government authorities tended to be risk-averse and also the commercial pressures to minimize costs may have been a little lower for such organizations.

It seems that this fortunate history is reflected in very low failure rates of Australian pipelines due to corrosion and defects in materials or construction.

The rate of metal-loss corrosion failures on Australian pipelines has been zero for over 20 years. Arguably, it has been zero for all ‘modern’ pipelines, omitting

Sample

copy

- not

for di

stribu

tion


design and then reviewed every five years or whenever the environment around the pipeline changes (such as for new urban development). This study is a fine-grained analysis, often on a metre-by-metre basis, of all possible causes of pipeline failure. Threats to pipeline integrity are explicitly identified and mitigated, with great emphasis on protection against third-party damage.

• The safety-management study is an engineering process but may have a cultural side effect: because it is integral to Australian pipeline design and operation it may help keep safety matters – and particularly the consequences of pipeline failure – in the forefront of pipeline engineers’ thinking at all times.

We reiterate that all or none of these factors may be responsible for the low rates of Australian failures due to third-party damage, although they at least suggest possible initial directions for investigation.

The absence of a simple explanation for the low Australian failure rate is not a clear-cut and satisfying conclusion. However the objective of this paper is to initiate discussion of whether others might benefit from research into the differences between Australian and international failure rates. We believe it provides a convincing case that Australian pipeline failure rates are indeed substantially lower than elsewhere and hence that investigation of that difference has potential to provide benefits in reducing failure rates in other parts of the world.

References

1. S.B.Cunha, 2012. Comparison and analysis of pipeline failure statistics. International Pipeline Conference, Calgary, Canada, September, paper IPC2012-90186.

2. Standards Australia, 2012. AS 2885.1-2012: Pipelines – Gas and liquid petroleum, Part 1: Design and construction.

3. P.Tuft and C.Bonar, 2009. Experience with the Australian pipeline incident database. Australian Pipeline Industry Association convention.

Summary of differences

Do these differences between Australian and international conditions for pipelines explain the markedly lower Australian failure rate? Only partly. They may explain the virtual absence of corrosion failures in Australia, and they may also explain the different pattern of failures due to natural events. However they do not explain the much lower rate of failures due to third-party damage.

Conclusion

The mostly likely explanation for the low rate of third-party damage failures in Australia is some difference in approach to managing third-party interference. Exactly what that difference might be is not clear and there has been no study that might cast light in this area.

Following are some speculative comments on Australian practices that might contribute to low third-party failure rates, although they are only very tentative explanations for the different failure rates until there is better information on whether practices elsewhere do in fact differ significantly.

• Australian pipelines in populated areas tend to be patrolled frequently, usually on the ground but sometimes by air. In and around at least two major cities the transmission pipelines are patrolled every day or every weekday, and in other cities they are patrolled weekly. The incident database contains about 20 near-miss events where patrollers have caught third parties in the act of digging or preparing to dig on the easement, and there are many more instances where work near the pipeline was forestalled by patrollers before encroaching on the easement (such off-easement activity is not reportable but there is ample anecdotal evidence).

• The ‘One-Call’ or ‘Dial-Before-You-Dig’ system is well used by third parties with only rare lapses.

• Pipeline marker signs tends to be frequent, conspicuous, and explicit in their warning message.

• Since 1997, AS 2885 has required a safety management study to be undertaken during

Sample

copy

- not

for di

stribu

tion

.

www.clarion.org

ORGANIZED BY

CALL FOR PAPERS NOW OPEN

20 – 23 October 2014 Berlin, Germany

ABSTRACT SUBMISSIONProposals should include an abstract of no more than 200 words,

accompanied by the author’s complete contact information and affiliation, and should be sent to:

John Tiratsoo, Director, Tiratsoo Technical e-mail: [email protected]

SCHEDULEAbstracts deadline: 1 July, 2014 | Final papers: 8 September, 2014

FPP_2014_FP_CallforPapers_A4.indd 1 4/12/13 3:38 PM

Sample

copy

- not

for di

stribu

tion


TRANSPORTING ANTHROPOGENIC CO2 in pipelines, either in dense phase or gaseous phase, is an essential component in the practical realisation of carbon capture and storage (CCS). Whichever phase is

considered, the likelihood and severity of internal degradation mechanisms arising from CO2 transportation under normal operating conditions and under process upsets needs to be assessed.

Whilst internal corrosion has been a focus of research in this area, the risk of stress-corrosion cracking (SCC) has not been extensively investigated. This paper explores the level of risk posed by SCC in CO2 pipelines, and the gaps in current knowledge, together with a presentation of test results that investigate the presence of SCC in simulated CO2 environments in the presence of impurities.

This paper was presented at the 4th International Forum on the Transportation of CO2 by Pipeline held in June, 2013, in Newcastle upon Tyne, UK, and organized by the University of Newcastle, Tiratsoo Technical, and Clarion Technical Conferences.

*Corresponding author’s contact detailstel: +44 191 215 4010email: [email protected]

1 MACAW Engineering, Newcastle upon Tyne, UK2 School of Chemical Engineering and Advanced Materials, Newcastle University, Newcastle

upon Tyne, UK3 School of Marine Science and Technology, Newcastle University, Newcastle upon Tyne, UK

Internal stress-corrosion cracking in anthropogenic CO2 pipelines: is it possible?

by Daniel Sandana*1, Mike Dale1, Dr E A Charles2, and Dr Julia Race3

THE RELEASE OF GREENHOUSE gases – such as carbon dioxide (CO2), methane, and nitrous

oxides – into the atmosphere due to human activities has been associated with global warming and climate change [1]. Carbon dioxide (CO2) represents a significant component (~77%) of the total greenhouse emissions.

To significantly cut the sustained increase of global atmospheric CO2 emissions, one promising option that has attracted interest from governments and industry has been to capture the CO2 produced from fuel use at major point sources and prevent it from reaching the atmosphere by storing it. This has been referred to as carbon capture and storage (CCS).

Transporting anthropogenic CO2 in pipelines is an essential component in the realization and implementation of CCS. Transportation of dense CO2 has generally been the preferred economic solution, but projects in the UK have also considered transportation of gaseous CO2. Whichever option is selected, provision

will need to be made to mitigate or prevent internal corrosion. Over the past few years the development of such CCS technology has generated increased research and development activities to evaluate the integrity risks related to the transport of anthropogenic CO2 in pipelines, i.e. in the presence of impurities such as oxides of sulphur (SOx) or nitrogen (NOx).

So far, whilst the corrosion research in this domain has focused on identifying plausible corrosion rates which may occur in these environments, the risk of SCC has not been extensively investigated [2-7]. This needs to be assessed [8] in order to implement the correct conditioning plant process and materials design, and to set up suitable integrity-management strategies for the infrastructures to ensure they fulfil their required operating life [9].

CO2 transport and the presence of impurities

There are currently over 3,500 km of operational, long-distance, high-pressure CO2 pipelines. These pipelines are mainly located in the USA and are mostly transporting CO2 from natural sources for enhanced oil recovery. CO2 from natural sources is relatively pure, and this means that defining a specification for such product is, in many respects, not too complex a problem [10].

Sample

copy

- not

for di

stribu

tion


The Inter-governmental Panel for Climate Change (IPCC)1 considers that the main impurities in flue gases generated from a post-combustion process by coal combustion will include not only nitrogen (N2), oxygen (O2), and water, but also SOx and NOx, hydrochloric and hydrofluoric acids (HCl and HF), and mercury (Hg) [11]. In comparison, flue gases from natural gas combustion processes typically contain low levels of SOx and NOx and higher concentrations of O2; HF can, however, also be present. Desulphurization plant is generally necessary to prevent sulphur-poisoning of the solvent in the CO2-absorption process.

In the pre-combustion process, the captured CO2 may contain very small levels of impurities such as N2, O2, hydrogen (H2), methane (CH4), carbon monoxide (CO), and sulphur compounds such as H2S. The levels of SOx and NOx present in captured CO2 from pre-combustion processes are insignificant. CO2 from pre-combustion physical solvent scrubbing processes typically contains about 1-2% H2 and CO, and traces of H2S and other sulphur compounds.

In the oxy-fuel process, the CO2-rich stream commonly contains O2, N2, argon (Ar), sulphur (S), and NOx.

The presence of impurities will have a significant impact on the pipeline design and operation [10, 12, 13] but also on the pipeline integrity in terms of internal corrosion risks [2, 9].

1 IPCC Secretariat, c/o World Meteorological Organization, 7 bis, Avenue de la Paix, CP 2300, CH-1211 Geneva 2, Switzerland.

In contrast, the capture and transport of CO2 from power plants means that different types and levels of impurities are present in the CO2 stream, and the purity of the CO2 becomes significantly affected. The composition of such product via pipelines to its destination is dependent on many parameters, essentially:

• the nature and chemical composition of the fuel-gas source;

• the carbon-capture technology: currently there are three main options – pre- or post-combustion, or oxyfuel;

• the conditioning / treatment process: for example, sulphur- or nitrogen-scrubbing plants to limit hydrogen sulphide (H2S), SOx, and NOx;

• the use and type of drying plant to restrict water content;

• the process used to separate CO2 from the rest of the flue gases, including chemical or physical absorption.

Additional local regulatory requirements and safety considerations will bring further complication into the definition of a universal specification for the transport of anthropogenic CO2 [5, 10]. The types and levels of impurities as a function of the fuel source and the carbon-capture technology source used are provided in various publications. Table 1 gives a comparison of published CO2 compositions from different capture technologies [10].

Table 1. CO2 % vol. compositions from different capture technologies [10].

Comp (vol %) CO2 CH4 H2S C2+ CO O2 Ar N2 NOx SOx H2 HCN COS NH3 CH2OH

IPCC Coal >99.97 0.01 <0.01 <0.01

IPCC Gas >99.97 0.01 <0.01 <0.01

ENCAP 99.8 0.003 0.001 0.003 0.021 0.021 0.002 0.001

Oosterkamp et al <99 0.01 Trace 0.01 0.001 0.01 Trace 0.17 <0.005 <0.001 Trace

IEAGHG – Comp1 99.93 0.001 0.015 0.045 0.002 0.001

IEAGHG – Comp2 99.92 0.001 0.015 0.045 0.002 0.001

IEAGHG – Comp3 99.81 0.002 0.03 0.045 0.002 0.001

IPCC Coal >96.39 0.01 0.01-0.6 0.03-0.4 0.03-0.6 0.8-2.0

IPCC Gas >95.65 2 >0.01 0.04 1.3 1

ENCAP – CO2/H2S 97.8 0.035 0.01 Unknown 0.17 Unknown 0.05 0.03 Unknown 1.7 <0.0005 <0.0005 0.003

ENCAP – CO2+H2S 95.6 0.035 23 Unknown 0.17 Unknown 0.049 0.03 Unknown 1.7 <0.0005 <0.0005 0.003

Oosterkamp et al >95.6 >0.035 <3.4 <0.01 <0.4 Trace <0.05 <0.6 <0.05 Unknown

IEAGHG – Selexol 97.95 0.01 0.01 0.04 0.03 0.9 1 With H2S

IEAGHG – Rectisol 99.7 0.01 0.01 0.04 0.15 0.21 0.002 With H2S 0.02

IPCC Coal >95.79 3.7 0.01 0.5

IPCC Gas >95.88 4.1 <0.01 <0.01

ENCAP – CO2/H2S 91 Unknown 1.6 5.7 0.61 0.25 0.076 Trace Trace

ENCAP – CO2+H2S 90 Unknown 1.6 5.6 0.6 0.24 1.5 Trace Trace

Oosterkamp et al >90 Trace Trace <3 <5 <7 <0.25 <2.5 Trace

IEAGHG – Comp1 85 4.7 4.47 5.8 0.01 0.007

IEAGHG – Comp2 98 0.67 0.59 0.71 0.01 0.007

IEAGHG – Comp3 99.94 0.01 0.01 0.01 0.01 0.007

Post

Com

bust

ion

Pre

Com

bust

ion

Oxy

fuel

Sample

copy

- not

for di

stribu

tion


The presence of impurities and internal corrosion considerations

It is generally accepted as a simplified rule that if no aqueous phase condenses out from the CO2 stream, the risk of internal corrosion in the transporting facilities will remain low. The occurrence of aqueous-phase condensation is dependent on the water solubility in the CO2 fluid which is a function of the operating pressure, operating temperature, and the types and levels of impurities present. In pure CO2, the condensation of water will occur when the water content is above its solubility limit for the considered pressure and temperature. The additional presence of impurities such as SOx and NOx will, however, drive the precipitation of acid aqueous phases at water contents much lower than the water solubility limit in otherwise-pure CO2 streams. This is related to the formation of acids (whose nature is dependent on the type of oxide compounds) at a temperature much higher than the water dewpoint, which is commonly referred to as the acid dewpoint[14].

Studies have investigated the chemistries of the aqueous phase in the presence of pure CO2 in the view of assessing corrosion of steel materials in high-pressure CO2 environments. The levels of carbonic acid, bicarbonates, and carbonates in deionized water as function of the CO2 partial pressure and temperature are identified in Figs 1-3 [4]. The concentrations increase with increasing pressure whilst decreasing with temperature. The pH value also decreases with higher CO2 pressures, and will be in the range of 3 to 3.4 above 50 bar. At high CO2 pressures, the pH values are so low that the formation of a surface protective film may become difficult (i.e. scale-free CO2 corrosion) due to the high solubility of iron carbonate under these conditions. However, the expected increased dissolution of iron due to higher carbonic acid concentration and lower pH with pressure may provide an increased source of ferrous ions which might lead, over time, to iron carbonate precipitation even at low pH, although probably not as a protective layer. The chemistry of the aqueous phase in rich CO2 environments will be further modified by the presence of impurities in the CO2 stream, and the magnitude of the change will be dependent on the type and concentration of the impurity, but also – and as important – on the partitioning behaviour of these impurities between the different CO2 and water phases. This partitioning will be a function of the CO2 physical state, temperature, pressure, and on the synergy effects between the different mixture compounds. The presence of impurities of SOx and NOx in the CO2 stream will further decrease the pH of any aqueous phase formed, due to the formation of acids such as sulphuric/sulphurous acids or nitric/nitrous acids, which will aggravate further general and localized corrosion in these environments and render any protective film formation more difficult. Ayello et

Fig.1. Predicted carbonic acid concentrations in water vs CO2 partial pressure at various temperatures [4].

Fig.2. Predicted bicarbonate concentrations in water vs CO2 partial pressure at various temperatures [4].

Fig.3. Predicted carbonate concentrations in water vs CO2 partial pressure at various temperatures [4].

Sample

copy

- not

for di

stribu

tion


However, in the case of a delay in response to process upsets or in the case of a shut-down (pipeline shut-in or total depressurization), situations could exist where free water may be present over long periods of time in the pipeline. Indeed, during long shut-downs, water-removal operations will be required and experience from the oil and gas industry has indicated that this may take weeks.

The following sections will aim at understanding whether SCC should be considered as a potential internal corrosion risk under such extreme, but realistic, operating scenarios, when water and high levels of impurities may be present in the pipeline during its operating life. As part of the authors’ experience in other systems related to the oil and gas industry, the review will consider the following impurities: CO, H2S, NOx [8].

The presence of impurities in CO2: why SCC should be considered

Stress-corrosion cracking leads to the sudden and catastrophic failure of alloys as a result of materials cracking in a corrosive environment. Three key parameters are essential for the initiation and sustained propagation of SCC:

• a material susceptible to SCC;• an environment in which this material is susceptible

to SCC;• a tensile stress (residual or applied) whose magnitude

is sufficient to initiate and propagate cracking.

Should one of these parameters not be present, SCC will not be produced.

al. suggested that sulphur dioxide (SO2) can drop the pH by a magnitude of 1 unit [7]. General and localized corrosion rates in the presence of free water in rich CO2 environments and in the presence of impurities are expected to be significant; the importance of water content and the effect of impurities such as SOx and NOx on general and localized (pitting) corrosion risks in rich CO2 streams have been well discussed elsewhere, and are beyond the scope of this paper [2, 4, 5, 6, 7, 9, 15]. A mitigation strategy to prevent the realization of internal corrosion in CO2 transport pipelines would be to identify (from corrosion testing) the levels of water and impurities acceptable in terms of pipeline internal integrity prior to the pipeline commissioning, and then to ensure suitable physical and chemical processes have been put in place upstream of operating CO2 pipelines such that the conveyed product is in accordance to specification. However, it should be recognized that despite all the precautions taken at the process design stage to guarantee acceptable water and impurity contents in CO2, upset conditions in the CO2 dehydration and conditioning process are still a possibility in complex pipeline systems, especially over decades of operating life. Under such scenario, free water is expected at the 6 o’clock position of the pipeline transportation system in presence of significant levels of impurities, typically as high as concentrations in flue gases. Hence the resulting risk of internal corrosion from such excursions should not be excluded. One may argue that if a continuous flow of dry CO2 is present following an event of water incursion, the water may rapidly dissolve in the stream minimizing any internal damage to the pipeline [6, 9].

Stress to cause cracking, MPa

Crack Growth Rate at 448 MPa

Carbon monoxide partial pressure, bar

Str

ess

to c

ause

cra

ckin

g, M

Pa

Cra

ck g

row

th r

ate,

mm

/sec

x 1

0-7

470

420

370

320

270

220

170

24

20

16

12

8

4

00 0.2 0.4 0.6 0.8 1 1.2 1.4 2.6

Fig.4. Effect of carbon monoxide partial pressure on minimum stress to initiate cracking and on crack-growth rate (total pressure 7.9 bar (0.79 MPa), +100 mV from FCP) [18].

Sample

copy

- not

for di

stribu

tion


• The additional presence of oxygen will increase the susceptibility to SCC in this system.

• Crack-growth rates of 10-6 mm/s were reported.• The mechanism of SCC for the CO2-CO-H2O

system can be classed under the ‘strain-generated active path’ model. This is the result of the formation of a mono-molecular CO film on the surface of the carbon steel and its rupture under stress.

• Most of the experimental data are limited to low partial pressures of CO2 (<20 bar (<2 MPa)).

Under upset dehydration conditions, there is a potential risk of CO2-CO-H2O SCC in pipelines transporting CO2 from pre-combustion capture processes. There is a fundamental requirement to investigate the susceptibility of pipeline steels to CO2-CO-H2O SCC at the high CO2 partial pressures typical of gaseous and dense CO2 pipelines.

H2S and sour cracking

Hydrogen sulphide is a significant impurity in CO2 captured from pre-combustion processes, but can be also present in streams generated from post-combustion and oxy-fuel technologies.

The presence of H2S can lead to different types of sour cracking, mainly sulphide-stress-corrosion cracking (SSCC) and hydrogen-induced cracking (HIC). These threats and their respective mitigation requirements have been well documented in the oil and gas industry. In particular, the standard NACE2 MR0175/ISO3 15156 has been used to mitigate the risk of sour cracking [20]. This standard is based on oil and gas industry experience and testing for hydrocarbon systems (i.e. CO2 is present as an impurity). Although this document is a starting point to potentially decrease the susceptibility of sour cracking in CO2 pipelines, there is a fundamental requirement to obtain data at the high partial pressures of CO2 in presence of H2S to understand in which conditions of H2S partial pressure, pH, temperature, SSCC, or HIC can be realized in CO2 pipelines.

If a CO2 pipeline is expected to see significant levels of H2S during service, sour-resistant steels will need to be considered to prevent catastrophic failure of the pipeline which could occur within days, or even hours, under the most favourable conditions. Compliance to maximum hardness will have to be specified for the parent and weld materials.

2 NACE, 1440 South Creek Drive, Houston, TX 77084-4906, USA.3 ISO Central Secretariat, 1, ch. de la Voie-Creuse, CP 56, CH-1211 Geneva 20, Switzerland.

It is generally considered that the presence of CO2 alone (in the presence or not of free water) is not sufficient to drive initiation and propagation of SCC. Hudgins et al. [17] nevertheless indicate that cracking may be generated on high-strength carbon steel in high-pressure CO2 environments under extreme stress conditions in relatively long exposure times. At 20 bar CO2, failures were produced within exposures as low as 22 hrs on steel materials with hardnesses of 34Rc and deformation levels of 115%; the production of cracks was associated with the potential leaching of sulphur from the steel materials. Scenarios under which such high stresses may arise, such as from geological ground motion or at localized corrosion features, should be considered.The additional presence of other impurities in the CCS stream may further increase the likelihood of SCC in CO2-H2O environments.

Carbon monoxide SCC in the CO2-CO-H2O system

Carbon monoxide is a significant impurity in CO2 captured from pre-combustion processes [1]. In the 1970s cracking of carbon steels was observed in environments constituted of wet mixtures of CO2 and CO gases, such as those present in coal-processing plants, and town-gas manufacture, transport, and storage systems.

Microscopic examination of the failures indicated fine transgranular cracks initiated from the internal surface of the vessel containing the gas mixture. The investigations also showed that cracking initiated at sites subject to tensile stress typically generated from the high pressure of the contained gas. The occurrence of such cracking in practical engineering situations worldwide, particularly in town-gas high-pressure pipelines, meant that great interest was generated and various research studies were conducted, mainly by Brown et al. and Kowaka and Nagata [18, 19]. Kowaka and Nagata were the first to indicate that transgranular SCC of steel is possible in the CO2-CO-H2O system. The current understanding for the occurrence of CO2-CO-H2O SCC is summarised [18, 19]:

• The presence of water is critical for the incidence of cracking.

• The presence of CO in CO2-H2O systems is critical for the occurrence of transgranular cracking in carbon steels.

• An increase of the CO activity in CO2-H2O systems increases the susceptibility to cracking (see Fig.4 [18]), i.e. the crack growth rate is greater, and the minimum initial stress to be applied for SCC occurrence is lower. At high CO activity, fine branched cracks are formed during crack propagation, whilst at low CO activities voids are created below the metal surface. It is possible to generate cracking under freely corroding conditions at high CO partial pressures.

Sample

copy

- not

for di

stribu

tion


100-mV shift criterion is used, where the CP is monitored by ‘on’ potentials, or where there is some degree of CP shielding, for example in areas of disbonded coating or from a porous coating.

• In the bicarbonate-carbonate system, the susceptibility to cracking decreases with decreasing concentration of bicarbonate. This is illustrated in Fig.5 [21].

• The susceptibility to cracking is maximum at temperatures of 75-80ºC, and diminishes with decreasing temperature. The risk of cracking is usually considered to be low at temperatures lower than 40ºC, but it is still possible at ambient temperatures, as illustrated in Fig.6 [21].

The risk of bicarbonate/carbonate SCC in CO2 pipelines in the presence of free liquid water may in the first instance be discarded due to:

• Non-existence of surface electrochemical polarization to drive the internal pipeline surface steel potential in the critical range for SCC initiation.

• A low pH (< 4) which will mitigate the formation of protective iron carbonate.

• Relatively low concentrations of bicarbonates (< 0.001M for CO2 partial pressures < 100 bar (<10 MPa) – see Fig.2) and carbonates in free water.

Bicarbonate and carbonate SCC, and the effect of impurities

The excessive presence of water in rich CO2 environments will drive the formation of an acid aqueous phase in which dissolved CO2, bicarbonates, and carbonates will be in equilibrium. The concentrations of these species as a function of the partial pressure are illustrated in Figs 1-3 [4].

Aqueous concentrated carbonate-bicarbonate solution environments are known to result in intergranular SCC of low-alloyed steel pipeline materials. This mechanism is generally referred to as high-pH SCC since it readily occurs in solutions with a pH of 9-10, conditions under which iron carbonate films will precipitate on steel. The occurrence of such cracking has only occurred on operational pipelines as an external mechanism due to the importance of potential polarization to generate SCC initiation and growth. The critical potential for SCC is related to the steel surface active-passive transition as a result of iron carbonate formation. This potential is dependent on the actual pH environment and temperature of the pipe.

This mechanism has been extensively studied:

• The critical potential for SCC is in the range of -650 to -750 mV (CSE); this potential can be present on operating pipelines where the

Fig.5. The effect of bicarbonate levels on the average crack velocity for low-carbon steel in 1 M Na2CO3 solution [21].

Sample

copy

- not

for di

stribu

tion


Nitrate SCC

Nitrates can cause SCC of carbon steel materials on their own. The susceptibility to cracking increases with the concentration of nitrates in solution, and with the temperature. The occurrence of SCC becomes significant at temperatures above 70ºC due to the rapid formation of a magnetite film.

It is plausible that occurrences of concentrated solutions of nitrates are present over the pipeline operating life as a result of low water condensation or water pool evaporation following process upset. Considering the spectrum of operating temperatures of 40ºC at the pipeline inlet (downstream of compression after-cooler unit) to ambient temperature (< 20ºC), the risk of nitrate SCC is considered to be low. However, the nitrate interaction with bicarbonate-carbonate environments may increase the susceptibility to cracking at relatively low temperatures, < 40ºC, as discussed above, in naturally corroding conditions.

Nitrate SCC should be considered upstream of the pipeline inlet.

Experimental testing

SCC in environments containing concentrated levels of bicarbonates and carbonates as a result of the evaporation of a pool of water present in a CO2 pipeline following an upstream process upset.

However, other operational considerations should be given to the impact of events/conditions in generating an SCC mechanism in CO2 pipelines, whose characteristics can be similar to bicarbonate/carbonate SCC:

• Following significant dehydration upsets, impact of water evaporation due to re-establishment of dry-gas operations in concentrating the levels of bicarbonates/carbonates, and impurities, such as nitrates and sulphates in existing water pools.

• The solution has initially a low pH due to the presence of acids; over time as the solution becomes saturated with bicarbonates/carbonates and corrosion products, the pH of the solution may increase to become alkaline.

• Impact of chemicals (such as amine, sulphite) and by-products (such as ammonia) carry-over from the capture and conditioning process into the pipeline to generate more alkaline conditions in the aqueous phases present in a CO2 pipeline.

• Impact of impurities such as nitrates and sulphates to shift the free corrosion potential into the critical range for SCC and increase the susceptibility to SCC under freely corroding conditions.

• The presence of scales such as iron sulphide may shift the potential in the critical range of potentials for SCC.

Fig.6. The effect of temperature on the average crack velocity for low-carbon steel in 1 M NaHCO3 + 0.75 M Na2CO3 solution [21].

Sample

copy

- not

for di

stribu

tion


concentration ratio of sodium bicarbonate to sodium carbonate was maintained constant at 2. Addition of 10%wt sodium nitrate (NaNO3) or sodium sulphite (Na2SO3) to the carbonate/bicarbonate solutions was made to assess effect of impurity. Each solution was tested at 23ºC, 40ºC, and 75ºC to simulate the range of temperatures possible in service.

All the solutions had pH values between 8.5 and 9.5, depending on the composition and the temperature. The addition of nitrate or sulphite to bicarbonate/carbonate solutions resulted in a slight acidification of the solution; the most significant impact on pH was from the sulphite addition

Slow-strain-rate testing

The X-80 tensile test specimens were produced from the pipe section. The specimens were polished over their gauge length to a 1200 grit-finish. After polishing, the specimens were cleaned with methanol, swept with cotton wool, and air dried.

A potential was applied to the specimen via a potentiostat; the reference electrode was a saturated calomel electrode coupled to the test solution via a salt bridge, and the counter electrode was a platinum wire directly in contact with the test environment. The temperature

The experimental work was conducted using pipeline steel of API4 5L X-80 grade. The materials composition (%wt) of the steel is given in Table 2.

Electrochemical testing

Potentiodynamic testing of the X-80 steel materials was carried out to identify:

• If this system is potentially susceptible to SCC, i.e. if the materials shows passivation at anodic potentials.

• Within which range of electrode potentials, SCC can be produced. The critical region of potentials for SCC can be determined from any significant difference in current flow between anodic potentiodynamic polarization curves plotted at a slow and a fast sweep rate.

The electrochemical measurements were conducted in a three-electrode cell using an automated potentiostat. Potentiodynamic testing was carried out at two potential sweep rates, of 0.2 mV/s and 10 mV/s. Each test was conducted in a fresh test solution of approximately 600 ml.The test solutions are summarized in Table 3. The molar

4 API, 1220 L Street, NW, Washington, DC 20005-4070, USA.

Elements C Si Mn P S Cr Mo Ni Al Cu Nb+Ti+V

%wt 0.064 0.309 1.96 0.010 <0.003 0.217 <0.001 0.005 0.029 0.009 0.132

Bicarbonate/Carbonate solutions Sodium Nitrate or Sulphite addition (%wt) Temperature (ºC)

0.25M NaHCO3+0.125M Na2CO3 0, 10 23, 40, 75

0.5M NaHCO3+0.25M Na2CO3 0, 10 23, 40, 75

0.70M NaHCO3+ 0.35M Na2CO3 0, 10 23, 40, 75

1M NaHCO3+0.5M Na2CO3 0, 10 23, 40, 75

Bicarbonate/Carbonate solutions Potentials (mV SCE)

0.25M NaHCO3+0.125M Na2CO3 -755, -740, -725, -710, -695

0.25M NaHCO3+0.125M Na2CO3 + 10 %wt NaNO3 -720, -710, -690, -650

0.25M NaHCO3 + 0.125M Na2CO3 + 10% Na2SO3 -820, -790, -770, -750, -730

0.70M NaHCO3 + 0.35M Na2CO3 -710, -685, -675, -655

0.70M NaHCO3 + 0.35M Na2CO3 + 10% NaNO3 -700, -670, -650, -628, -590

0.70M NaHCO3 + 0.35M Na2CO3 + 10% Na2SO3 -730. -720, -710, -690

Table 4. Test environments for SSRT, 75°C.

Table 3. Test environments for electrochemistry.

Table 2. Materials chemistry of tested X-80 pipeline steel.

Sample

copy

- not

for di

stribu

tion


then prepared and hot mounted, and the cross-section was observed under an optical microscope.

Results

Electrochemical data

Potentiodynamic curves

The polarization curves at slow (0.2 mV/s) and fast sweep rates (10 mV/s) for 0.70 M NaHCO3 + 0.35M Na2CO3 + 10%wt NaNO3, and for 0.25 M NaHCO3

of the solution was controlled via a thermostat. The strain rate during testing was 1.4 x 10-6 sec-1. The test ended when fracture of the specimen occurred. Table 4 shows the conditions at 75°C at SSRT was conducted. Following SSRT, the reduction of area of the specimen in the necking region was measured.

The specimen fracture surfaces were cleaned in Clarks Solution. The specimen surface at proximity of the fracture was then observed using a scanning-electron microscope (SEM) to identify the presence of any stress-corrosion cracks. A cross-section of the specimen tip was

Fig.7. Polarization curves at slow and fast sweep rates for 0.70M NaHCO3 + 0.35M Na2CO3 + 10%wt NaNO3 at various temperatures.

Fig.8. Polarization curves at slow and fast sweep rates for 0.25M NaHCO3 + 0.125M Na2CO3 + 10%wt NaNO3 at various temperatures.

Sample

copy

- not

for di

stribu

tion


SCC may also be generated in lower bicarbonate/carbonate systems with nitrates (Fig.8).

An active-transition region was also observed with the addition of sulphites, as illustrated in Figs 9 and 10.

Free-corrosion potential, Ecorr

The effect of nitrate and sulphite in various bicarbonate/carbonate solutions on the free-corrosion potential response (Ecorr) is illustrated at 23˚C, 40˚C, and 75˚C in

+ 0.125M Na2CO3 + 10%wt NaNO3, are illustrated in Figs 7 and 8, respectively. The anodic curve shows an active-passive transition at all temperatures, suggesting the formation of a protective surface film in carbonate/bicarbonate environments in the presence of nitrates. The anodic current difference between the active and the passive state is lower as the temperature is decreased; this suggests that the surface film is less protective with the drop of temperature. The presence of an active-transition region is often indicative of the possible occurrence of SCC in low-alloy steels, and

Po

tent

ial,

mV

SC

E

Fig.9. Polarization curves at slow and fast sweep rates for 0.75M NaHCO3 + 0.35M Na2CO3 + 10%wt Na2SO3 at various temperatures.

Fig.10. Polarization curves at slow and fast sweep rates for 0.25M NaHCO3 + 0.125M Na2CO3 + 10%wt Na2SO3 at various temperatures.

Sample

copy

- not

for di

stribu

tion


in the presence of nitrates under naturally occurring conditions (i.e. at Ecorr) or under very low polarization (such as resulting from the presence of semi-conductive scales cathodic to the steel) may be increased.

The addition of sulphites to bicarbonate or carbonate environments shifts the value of Ecorr to more cathodic potentials for all the test temperatures used, suggesting the system becomes more corrosive with the addition of sulphites. This is probably related to the surface films being less stable or protective in the presence of sulphur compounds.

Figs 11-13, respectively. The addition of nitrate to bicarbonate or carbonate environments shifts the value of Ecorr to more anodic potentials for all the test temperatures used, suggesting the system becomes nobler with the addition of nitrates.

This also suggests that in the presence of nitrates the electrode potential will shift towards the active-passive steel transition in systems containing concentrated bicarbonates and carbonates. This may indicate that the susceptibility to cracking in concentrated bicarbonates and carbonates

Fig.11. Effect of nitrate and sulphite on the free corrosion potential Ecorr in bicarbonate and carbonate solutions at 23˚C.


Sample

copy

- not

for di

stribu

tion


lower concentrated bicarbonate and carbonate environments in the presence of nitrates, i.e. 0.50M NaHCO3 + 0.25M Na2CO3 + 10%wt NaNO3, -670 mV SCE, at 75°C; this is illustrated in Fig.16.

Intergranular stress-corrosion cracking was also produced in the presence of sulphites (see Fig.17).

Initial results associated with reduction of area measurements for tests at 75°C on failed specimens (Fig.18) suggests that nitrates do not significantly affect the maximum cracking susceptibility other than that obtained in a pure

It is noted that the free-corrosion potentials are taken from the potentiodynamic curves, and there will be an over-potential related to the dynamic nature of the test. The free-corrosion potential values obtained under long steady exposure conditions will be slightly lower.

SSRT

Multiple intergranular cracking was produced along the gauge of X-80 in 0.70M NaHCO3 + 0.35M Na2CO3 + 10%wt NaNO3, -650 mV SCE, at 75°C, as illustrated in Figs 14 and 15. Significant cracking was still produced in


Fig.15. Intergranular SCC of X-80 in 0.70M NaHCO3+0.35M Na2CO3+10%wt NaNO3, 75˚C, -650 mV SCE, x100.

Fig.14. Intergranular SCC on X-80 gauge surface: 0.70M NaHCO3+0.35M Na2CO3+10%wt NaNO3, 75˚C, -650 mV SCE.

Sample

copy

- not

for di

stribu

tion


• Consideration should be given to the impact of extreme operational events/conditions (i.e. upsets) which can generate a SCC mechanism in CO2 pipelines, whose characteristics can be similar to bicarbonate/carbonate SCC due to the presence of impurities such as nitrates. » Intergranular SCC was produced in bicarbonate and carbonate environments in the presence of nitrates and in the presence of sulphites.

» In the presence of nitrates, the Ecorr shifts towards the active-passive steel transition in systems containing concentrated bicarbonates and carbonates. The susceptibility to cracking in concentrated bicarbonates and carbonates in the presence of nitrates under naturally occurring conditions or under very low polarization (for example resulting from the presence of semi-conductive scales cathodic to the steel) may be increased.

» The presence of nitrates does not appear to affect significantly the cracking susceptibility compared with a pure carbonate or bicarbonate system.

» The presence of sulphites appears to decrease the cracking susceptibility compared with a pure carbonate and bicarbonate system. This may be associated with sulphur compounds hindering the formation of protective surface films.

carbonate or bicarbonate system. Sulphites, however, appear to decrease the susceptibility to stress-corrosion cracking, especially in the higher concentrated bicarbonate and carbonate systems. This may be associated with sulphites hindering the formation of protective surface films on the steel and hence enhancing general corrosion or the formation of shallow pits.

Conclusions

• Internal stress-corrosion mechanisms should be considered at the process design stage, i.e. CO2 specification definition, and during the operating lifecycle of the CO2 transporting pipeline system.

• Stress-corrosion cracking in the CO2-H2O system may occur at high partial pressures of CO2, typically above 20 bar, under severe stress conditions. The presence of sulphur in the steel may increase susceptibility. Scenarios under which high stresses can arise, such as from geological ground motion or localized corrosion features, should be considered.

• CO and H2S can result in internal SCC in CO2 pipelines depending on the fuel source and the CO2-capture technology used. Further experimental work is required to quantitatively assess the crack susceptibility at high CO2 partial pressures (>20 bar).

Fig.17. Intergranular SCC on X-80 gauge surface, 0.70M NaHCO3+0.35M Na2CO3+10%wt Na2SO3, 75˚C, -650 mV SCE.

Fig.16. Intergranular SCC of X-80 in 0.50M NaHCO3+0.25M Na2CO3+10%wt NaNO3, 75˚C, -670 mV SCE, x400.

Sample

copy

- not

for di

stribu

tion


11. IPCC, 2005. Carbon dioxide capture and storage. Special Report, Intergovernmental Panel on Climate Change (IPCC).

12. P.Seevam et al., 2008. Transporting the new generation of CO2 for carbon capture and storage: the impact of impurities on supercritical CO2pipelines. IPC, paper 64063.

13. P.Seevam et al., 2010. Capturing carbon dioxide: the feasibility of re-using existing pipeline infrastructure to transport anthropogenic CO2. IPC, paper 2010-31564.

14. Kear. Low-temperature corrosion by flue gas condensates Part 1.

15. Farelas et al., 2012. Effects of CO2 phase change, SO2 content and flow on the corrosion of CO2 transmission pipeline steel. Corrosion/2012, paper C2012-01322, NACE International.

16. R.N.Parkins, 1972. Stress corrosion spectrum. British Corrosion Journal, 7, pp 15–28.

17. Hudgins et al., 1966. Hydrogen sulfide cracking of carbon and alloy steels. Corrosion, 22, pp 238-251, 56.

18. Brown et al., 1973. Electrochemical investigation of SCC of plain carbon steel in carbon dioxide-carbon monoxide-water system. Corrosion-NACE, 5, pp 686-695.

19. M.Kowaka et al., 1976. SCC of mild and low steels in CO2-CO-H2O environments. Corrosion-NACE, 32, pp 395-401, October.

20. NACE, 2009. MR0175/ISO 15156: Petroleum and natural gas industries – materials for use in H2S environments in oil and gas production. NACE Standard.

21. R.N.Parkins and S.Zhou, 1997. SCC of C-Mn steel in CO2/HCO3-/CO32, I: Stress corrosion data. Corrosion Science, 39, 1, pp 159-173.

References

1. IPCC, 2007. Climate change 2007: Synthesis report. Intergovernmental Panel on Climate Change, Adopted by the IPCC Plenary XXVII, November.

2. A.Dugstad et al., 2012. Internal corrosion in dense phase CO2 transport pipelines - state of the art and the need for further R&D. Corrosion/2012, paper C2012-01452, NACE International.

3. Seiersten, 2001. Materials selection for separation, transportation and disposal of CO2. Corrosion/2001, paper 01042, Houston, TX, NACE International.

4. Choi et al., 2010. Determining the corrosive potential of CO2 transport pipeline in high pCO2–water environments. Int. J. of Greenhouse Gas Control.

5. A.Dugstad et al., 2011. Corrosion of transport pipelines for CO2 – effect of water ingress. Energy Procedia, 4, 3063–3070.

6. Zhang et al., 2011. Water effect on steel corrosion under supercritical CO2. Corrosion/2011, paper 11378, NACE International.

7. Ayello et al., 2010. Effect of impurities on corrosion of carbon steel in supercritical CO2. Corrosion/2010, paper 10193, NACE International.

8. D. Sandana et al., 2013. Transport of gaseous and dense carbon dioxide in pipelines: is there an internal stress corrosion cracking risk? Corrosion/2013, paper C2013-02516, NACE International.

9. D.Sandana et al., 2012. Transport of gaseous and dense carbon dioxide in pipelines: is there an internal corrosion risk? J. of Pipeline Engineering, 11, 3, pp 229-238.

10. J.Race, 2012. Towards a CO2 pipeline specification: defining tolerance limits for impurities. J. of Pipeline Engineering, 11, 3, pp 173-190.

Fig. 18. Reduction of area of X-80 steel in different bicarbonate and carbonate environments in the presence of nitrates and sulphites at 75°C.

Sample

copy

- not

for di

stribu

tion


PIPELINES ARE THE SAFEST, most reliable, affordable, and efficient means for the transportation of water and other commercial fluids such as oil and gas. In last few decades the importance of the pipeline

transportation system has increased due to its extensive use in the oil and gas industry. Pipelines pass through myriad geologic environments and some are subjected to earthquake hazards. Historically, the most catastrophic damages are the once resulting from faulting, seismic shaking, and liquefaction. Among others, the San Francisco (1906); Meckering, Australia (1968); Mexico City (1985); Loma Prieta, California (1989); Northridge, California (1994); Kobe, Japan (1995); Bhuj, India (2001); Denali, Alaska (2003); and Sumatra (2004) earthquakes all triggered damages to critical pipeline routes. Ruptures or severe distortions of the pipeline are often associated with landslides, liquefaction, loss of support, fault surface rupture, or differential motion at abrupt interfaces between rock and soil.

The performance of buried and above-ground pipeline structures subjected to seismic hazards has become an important subject of study. Approximately 2,000,000 km of pipelines has been laid worldwide. In India there are already more than 23,000 km of oil and gas pipelines that have been laid through many geographical areas and geologic conditions. More pipelines are slated to be laid in the future. There are plans for cross-border pipelines from Afghanistan; the route is in areas of high seismicity. There is also plan to construct an offshore pipeline over 1300 km of highly variable geologic conditions between the Middle East and India across the Indus River delta. The performance of buried and above-ground pipeline structures subjected to ground-surface rupture, soil liquefaction, and other seismic hazards is critical for engineers to understand in the Indian context.

Seismic design and engineering of pipelines has advanced significantly in last few decades; still, little has been accomplished to address the vulnerability of buried pipelines to seismic hazards. With new and proposed cross-country pipelines in India, it is becoming more important to understand the effects of seismic hazards (such as shaking, liquefaction, fault-surface rupture) on buried pipelines.

*Corresponding author’s contact details:email: [email protected]

1 Centre for Offshore Foundation Systems, School of Civil and Resource Engineering, University of Western Australia, Crawley, WA, Australia

2 Mexican Institute of Water Technology, Jiutepec, Mexico 3 Samit Spectrum Eit Pvt Ltd, Gurgaon, India4 Siti Energy Ltd, Moradabad, UP, India

Earthquakes and the Indian pipeline industry

by Indranil Guha*1, Beau Whitney1, Raúl Flores-Berrones2, Aditya Barsainya3, and Gaurav Arya4

PIPELINES ARE THE SAFEST, most economic and efficient means for the transportation of

water and other commercial fluids such as oil and gas and, nowadays, minerals in the form of slurry [1]. The designation of pipeline systems as ‘lifelines’ signifies that their operation is essential in maintaining public safety and critical services (for example, power, water, and sewer) at all times, even immediately after natural disasters. It is a linear system which traverses a large geographical area, and soil conditions thus, is

susceptible to a wide variety of geohazards. Ruptures or severe distortions of the pipeline are most often associated with relative motion arising from earthquake movements, landslides, liquefaction, or differential motion at abrupt interfaces between rock and sediment. Historically the most catastrophic damage results from ground rupture, seismic shaking, and liquefaction. The performance of buried and above-ground pipeline structures subjected to seismic hazards has become an important subject of study. This paper will give an over view of pipeline damage along with its failure modes during past earthquakes, and also the seismic vulnerability of Indian pipeline industry.

Sample

copy

- not

for di

stribu

tion


History of pipeline damage during earthquake events

In 1971 the San Fernando Earthquake resulted in direct losses to pipeline systems by damaging a 1.24-m diameter water pipeline at nine bends and welded joints. The ductile steel pipelines were able to withstand ground shaking but could not withstand ground deformation associated with fault rupture and lateral spreading [4, 5]. The 1983 Coalinga Earthquake caused a number of breaks in a natural gas line, and several pipeline failures occurred in oil drilling and processing facilities. In 1987, after the Whittier Narrows Earthquake, the Southern California Gas Co reported 1411 gas leaks that were directly caused by the earthquake. In 1989 the Loma Prieta Earthquake, with a magnitude of 6.9 Mw (Richter scale, moment magnitude), also caused failure of numerous water pipelines, and broken water pipelines occurred at the Ford automobile plant as a result of liquefaction and excessive soil pressures. During the Tennent Creek earthquakes of 1988 in Australia, three powerful earthquakes ranging from 6.3 to 6.7 ML (Richter scale, local magnitude) shook the region. The main infrastructure damage was the severe warping of a major natural gas pipeline as ground ruptures occurred along a 35-km long fault scarp (with up to 2 m of vertical displacement) due to the ductile failure of mild steel pipeline materials [6] as shown in Fig.3b. On 28 December, 1989, an earthquake of magnitude 5.6 in the Newcastle region of Australia caused 22 water-main breaks in 150-mm diameter and smaller pipes, mainly due to the seismic shaking. A few cast iron pipes were found to have circumferential cracking,

Indian seismic hazards

The Indian subcontinent is one of the most seismically active zones in the world. A deterministic seismic hazard map is shown in Fig.1a [2]; the Indian subcontinent has undergone catastrophic earthquakes claiming huge loss of life and property, and socio-economic losses, and a few of the major earthquakes locations are shown in Fig.1b [3], and these have happened due to active seismic/tectonic activities in Himalayan-Hindukush region spread over Afghanistan in the west to Bangladesh in the east across Pakistan, India, Nepal, and Bhutan. This is one of the earth’s youngest mountain range and is still isostatically imbalanced and geodynamically active due to the movement of the Eurasian Plate towards the north and the Indian Plate towards north-east. Due to tectonic movement of both the Plates, stresses are generated; when the stresses reach a critical stage, they are released in the form of earthquakes.

Consequently, the Indian subcontinental region is always at threat of earthquakes in present or future scenarios. The Indian landform has been classified into five different zones from Zone I (lowest) to Zone V (highest) in relation to seismic activity, and geographical statistics clearly show that 54% of the land is vulnerable to earthquakes. Figure 2 is a map showing that the north east of the country and a portion of Gujarat lie under Zone V, and major proportion of the country lies in Zone III. Consequently, any major or minor earthquake may create havoc, with huge losses of life and property due to high population densities and developed infrastructure.

Fig.1. (a) Generalized tectonic map of India and surrounding countries [2]; (b) Major earthquakes in India [3].

(a) (b)

Sample

copy

- not

for di

stribu

tion


deformation (PGD) received particular attention from the engineering community after the 1999 Kocaeli and Chi-Chi earthquakes [9), following faults displacing the ground surface by 2 and 9 m, respectively, and causing damage to water and gas pipelines. These kinds of events are likely to cause permanent damage to pipelines that cross the fault line. The response of buried pipelines to earthquake-induced PGD has received attention by the pipeline engineering community in the past decade (for example, Ref.10).

and also areas of localised corrosion appeared to have been shaken by the shockwaves from the earthquake and pieces blown out of the pipes [7].

The 1994 Northridge Earthquake caused about 1400 pipeline breaks in the San Fernando Valley area in the USA, and leaking gas ignited at several locations. Some broken water and gas lines were found to have experienced between 152.4 and 304.8 mm of separation (Fig.4). The area experienced widespread ground cracking and differential settlements. In the 1999 Kocaeli, Turkey, Earthquake (of magnitude 7.4 Mw) substantial water supply damage occurred in many cities [8]. For example, the entire water distribution system of Adapazari was damaged. A 2.4-m diameter steel water pipe was damaged at Kullar due to a right-lateral strike-slip motion along the fault, and a 2.2-m diameter butt-welded steel raw water pipeline crossing the Sapanca Segment of the North Anatolian fault and was damaged at the fault crossing. Damage was observed at three locations where a small surface leak was observed in the pipe at a point near the fault crossing; a significant leak occurred at yet another point; and a minor leak had happened at a bend in the pipe. In 1999, the Chi Chi Earthquake in Taiwan damaged many buried water and gas pipelines at many sites. It was reported that buried gas pipelines underwent bending deformation due to ground displacement along a reverse fault near the Wushi Bridge about 10 km south of Taichung. The risk to pipelines from permanent ground

Fig.3. (a) Brittle failure of water pipe during the Meckering Earthquake (Source: AEES Gallery); (b) Damage to the natural gas pipeline during the Tennant Creek Earthquake [6].

Fig.2. Seismic zone map of India [22].

(b)(a)

Sample

copy

- not

for di

stribu

tion


friction between the soil and the pipe, the depth of burial, and the pipe properties. Governing strains induced in the pipeline may be tensile or compressive depending on the type of motion along the underlying fault (or seismically triggered failure surface) and the angle at which the pipeline crosses. Seismic hazards have been classified as being either permanent ground deformation (PGD) or seismic-wave-propagation hazards [12]: PGD may be due to

Failure modes of pipelines during earthquakes

Approximately 3% of natural gas pipelines failures in the USA are due to ground movement during seismic events [11]. Soil movement can induce stresses and strains in the pipe walls, the magnitude of which depends on the magnitude and direction of soil movement, the soil shear strength, the

Fig.4. Pipeline damage during the Northridge Earthquake.

Fig.5. Schematics of fault-pipeline interaction: (a) strike-slip fault; (b) normal slip fault [13].

(a) (b)

Sample

copy

- not

for di

stribu

tion


of the pipe wall, and this is very common failure mode for steel pipes. Figure 6 illustrates the local buckling of a 77-in welded steel pipe during the 1994 Northridge Earthquake [13]. For a shallow buried pipeline, during movement along a fault in the vertical plane, global buckling is predominant over local buckling: the uplift resistance is much less than the downward bearing capacity. Typically, the relative

a discrete movement along a geologic fault (Fig.5), or triggered from seismic shaking such as landsliding Fig.6 and liquefaction.

Compressive failure of a continuous buried pipeline occurs due to fault rupture, landslide, liquefaction, or relative ground motions. Local buckling or wrinkling in the buried pipelines is due to local instability

Fig.7. Mexico City Earthquake, 1985 [17]: (a, b) Damage to steel pipelines; (c) close-up of a joint failure; (d) joint failure of a concrete pipeline.

Fig.6. Schematics of landslide-pipeline interaction: (a) longitudinal; (b) lateral [13].

(a) (b)

(a)

(c)

(b)

(d)

Sample

copy

- not

for di

stribu

tion


Circumferential flexural failure and joint rotation

When a segmented pipe is subjected to bending due to permanent ground movement or shaking due to seismic events, the ground curvature is accommodated by a combination of rotation at the joints and flexure on that pipe segment, and these are the two major failure modes for cast iron or asbestos-cement pipes.

Crushing of bell-and-spigot joints occurs due to compressive strain and was observed following the Mexico City Earthquake in 1985 [17], Fig.7, and following the Bhuj earthquake of 2001 [16] where many of the cast iron pipes were damaged due to this failure mechanism.

Seismic vulnerability of existing and proposed Indian pipelines

In India there are already more than 33,000 km of oil and gas pipelines which have been laid through many geographical areas and geologic conditions, and these pipelines contribute 32% of total transportation modes of hydrocarbons in India, as shown in Fig.8 [18]. More pipelines are planned to be laid in the future.

The previous history of pipeline damage in India due to earthquake was summarised by Das and Jain in Ref.13. In the 6.6-M Bihar Earthquake of 1988, some minor damage to facilities at the IOCL refinery was reported. In the 6.8-M Chamoli Earthquake, water supply lines in Chamoli and Gopeshwar was disrupted due to landslides. In the 7.7-M Gujarat Earthquake in 1991, while a little damage occurred at the joints of pipelines to pump station equipment, no major damage was reported. Similarly, in the 9.0-M Sumatra

movement is distributed over a considerable length, and hence the compressive strains in the pipeline are not too large and the potential of tearing of the pipe wall is reduced.

In tension, a corrosion-free steel pipe with arc-welded butt joints is ductile and capable of mobilizing large strains, associated with significant tensile yielding, before failure. However, older steel pipe with gas-welded joints often cannot withstand large tensile strains before rupture. The strain associated with tensile rupture is generally well above about 4% [14]; for analysis and design, an ultimate tensile value of approximately 4% [15] is often used, beyond which the pipeline is considered to have failed in tension.

Welded slip joint

The material strength of the pipe governs the failure of the arc-welded butt joints. However, for pipelines on steel slopes with slip joints and gas-welded joints, the failure criterion is different and is governed by the strength of the joints which is less than the strength of the pipe material. During the San Fernando Earthquake in 1971, the 48-in diameter Granada Trunk Line was an example of welded slip joint failure during an earthquake [5].

Axial pull-outs

The shear strength of the joint material is much less than that of the pipeline material. During seismic activity in areas of tensile ground strain, this kind of failure mechanism dominates especially for water transport pipelines [16].

Fig.8. Indian hydrocarbon transportation modes (% share) (after Mathur, 2010 [18]).

Sample

copy

- not

for di

stribu

tion


superimposing this route map with the tectonic map of Fig.1 shows that the pipelines entering India from the north west border are particularly vulnerable to seismic activity. Even the eastern side of the country falls under seismic threat, and the pipeline routes from Bangladesh and Burma also cross high earthquake zones as shown in Fig.2.

There is also plan to construct an offshore pipeline of over 1300 km length [19] across highly variable geologic conditions between the Middle East and India across the Indus River delta. In the Indian context, it is critical for engineers to understand the performance of on- and offshore buried and above-ground pipeline structures subjected to ground surface rupture, soil liquefaction, and other seismic hazards. Seismic-hazard analysis for the proposed Oman-India subsea pipeline was carried out few years ago [19]. The

Earthquake, most of the water pipelines were damaged in the Andaman and Nicobar Island areas, Fig.9b.

In addition to various pipeline projects being undertaken within India, there are plans to import piped gas from gas-rich neighbouring countries, such as Iran and Turkmenistan in Central Asia, Qatar and Oman in West Asia, and also from Burma and Bangladesh. As part of these schemes, the following international pipelines are being proposed:

• Iran-Pakistan-India (IPI) gas pipeline• Turkmenistan-Afghanistan-Pakistan-India (TAPI)

gas pipeline• Bangladesh-India onshore gas pipeline

For the pipelines from Afghanistan and Iran, the routes are in areas of high seismicity as shown in Fig.11. The route map of the IPI pipeline is shown in Fig.10;

Fig.9. (a) Axial pull-out at the joint of a water supply pipeline during the Tangshan Earthquake, 1976 [23]; (b) Leaking at a bell-and-spigot joint of a water supply pipeline due to bending during the Sumatra Earthquake, 2004 [13]; (c) Local buckling/wrinkling of a product pipeline in compression in the California Earthquake [24]; (d) Beam buckling of a water pipeline [13].

(a)

(c) (d)

(b)

Sample

copy

- not

for di

stribu

tion


the effect of earthquakes on a continuous pipeline in the state of Gujarat in India was presented in Ref. 21 based on the GSDMA report, in which the authors also discuss the design and construction methodology to minimize the effect of loading on the pipeline during ground movement.

In summary, this paper has given an overview of past performance of buried pipelines during earthquake events including fault rupture, liquefaction, seismic shaking, and illustrates the nascent understanding of earthquake hazards around the world and in India. Identification of seismic sources and geohazard-prone areas during the early phases of a project allows these data to be incorporated during the design phase. Appropriate site-specific seismic design during the engineering stage can then reduce the risks posed by earthquake hazards on buried pipeline structures.

References

1. I.Guha and B.Whitney, 2012. Seismic vulnerability of Australian pipeline industry – an overview. 11th Australia New Zealand Conference on Geomechanics. Melbourne, Australia.

2. I.A.Parvez, F.Vaccari, and F.P.Giuliano, 2003. A deterministic seismic hazard map of India and adjacent areas. Geophys. J. Int., 155, pp 489-508.

3. V.Kamalakar and R.P.Kumar, 2009. Damage based life of historical constructions in seismic environment. European J. of Scientific Research, 35, 2, pp 254-273.

ground-shaking hazard along the route of the pipeline and in the Indus delta was found out to be relatively high. Due to this, and other potential geohazards such as liquefaction, submarine landslides can be triggered. Figure 11 shows the route of the pipeline and seismic activities in nearby areas.

Conclusion

Seismic design and engineering of pipelines has advanced significantly in last few decades, although little has been accomplished to address the vulnerability of buried pipelines to seismic hazards. With new and proposed cross-country pipelines in India, it is becoming more important to understand the effects of seismic hazards (such as shaking, liquefaction, fault surface rupture) on buried pipelines. In 1974 the first seismic design code Technical standard for oil pipelines [20] was developed by the Japan Roads Association [13]. Thereafter, in 1984, ASCE first published formal guidelines [15] for seismic design of pipeline systems, although until 2007 there was no specific standard for seismic design of pipeline systems. However, the Gujarat State Disaster Management Authority (GSDMA) felt the urge to develop such a standard for Indian application as the state was the worst affected during the 2004 Bhuj Earthquake, and numerous public and privately owned oil and gas companies have laid their pipeline networks across that state. GSDMA has sponsored a project at the Indian Institute of Technology in Kanpur, and in 2007 the first draft was published [13]. An analysis of

Fig.10. Route map of the IPI pipeline [18].

Sample

copy

- not

for di

stribu

tion


10. M.J.O’Rourke, 2003. Buried pipelines. Earthquake Engineering Handbook, Eds. W-F.Chen and C. Seawthorn, CRC Press.

11. PRCI, 2003. Next generation design and assessment methods for onshore oil and gas pipelines. Pipeline Research Council International Inc. R & D Forum, Vancouver, Canada.

12. M.J.O’Rourke and X.Liu, 1999. Response of buried pipelines subjected to earthquake effects. Multidisciplinary Centre for Earthquake Engineering Research (MCEER), Monograph 3.

13. S.R.Das and S.K.Jain, 2007. Guidelines for seismic design of buried pipelines: provision with commentary and explanatory examples. IITK-GSDMA, National Information Centre of Earthquake Engineering, Kanpur, India, November, p. 88, ISBN 81-904190-7-2 (http://www.iitk.ac.in/nicee/IITK-GSDMA/EQ28.pdf).

14. N.M.Newmark and W.J.Hall, 1975. Pipeline design to resist large fault displacement. Proc. US National Conference on Earthquake Engineering, Ann Arbor, MI, pp 416–425.

15. American Society of Civil Engineers, 1984. Guidelines for the seismic design of oil and gas pipeline systems. Committee on Gas and Liquid Fuel Lifelines. Technical Council on Lifeline Earthquake Engineering, ASCE, New York.

4. T.L.Youd and D.M.Perkins, 1978. Mapping liquefaction-induced ground failure potential. J. Geotech. Eng. Division, ASCE, 104, pp 433-446.

5. T.O’Rourke and M.Twafik, 1983. Effects of lateral spreading on buried pipelines during the 1971 San Fernando earthquake. Earthquake behaviour and safety of oil and gas storage facilities buried pipelines and equipment, PVP-Vol 77, ASME, New York, USA.

6. K.McCue, 1994. The rise to lifelines from earthquakes in Australia and in Canberra in particular. In: Proc. Survival of Lifelines in Earthquake. Seminar held by the Australian Earthquake Engineering Society, Canberra, Australia.

7. H.L.Raja Sekar, 1991. Seismic design consideration for buried pipelines. 14th Federal Convention of Australian Water & Wastewater Association (AWWA). Perth, WA, Australia.

8. A.Kurtulus, 2011. Pipeline vulnerability of Adapazari during the 1999 Kocaeli, Turkey earthquake. Earthquake Spectra, 27, 1, pp 45-66.

9. M.Miyajima and T.Hashimoto, 2001. Damage to water supply system and surface rupture due to fault movement during the 1999 Ji-Ji earthquake in Taiwan. Proc. 4th Int. Conf. on Recent Advances in Geotechnical Earthquake Engineering and Soil Dynamics, University of Missouri-Rolla, paper 10.45.

Fig.11. Seismic vulnerability of the Oman-India subsea pipeline [19].

Sample

copy

- not

for di

stribu

tion


21. I.Guha and R.Flores-Berrones, 2008. Earthquake effect analysis of buried pipelines. Proc. 12th International Association for Computer Methods and Advances in Geomechanics, (IACMAG), Goa, India, pp 3957-3967.

22. W.K.Mohanty and M.Y.Walling, 2008. Seismic hazard in mega city Kolkata, India. Nat Hazards, 47, pp 39-54.

23. EERL, 2004. The Great Tangshan Earthquake of 1976 (CD), Vol. 4, Earthquake Engineering Research Laboratory, California Institute of Technology, Pasadena, California.

24. D.Ballantyne, 2008. The shakeout scenario: oil and gas pipelines. Prepared for United States Geological Survey and California Geological Survey. USGS open file report 2008-1150.

16. S.R.Dash and S.K.Jain, 2005. Seismic design of buried pipelines in Indian context. Proc. National Symposium on Structural Dynamics. Random Vibrations and Earthquake Engineering (C.S.Manohar and D.Roy, Eds), Bangalore, India, pp 25-32.

17. R.Flores-Berrones and X.L.Liu, 2003. Seismic vulnerability of buried pipelines. Geofίsica International, 42, 2, pp 237-246.

18. S.Mathur, 2010. Pipeline perspective on India – 2010. Petromin Pipeliner, Oct-Dec, pp 22-35.

19. K.W.Campbell, P.C.Thenhaus, J.E.Mullee, and R.Preston, 1996. Seismic hazard evaluation of the Oman India pipeline. Proc. Offshore Technology Conference, Houston, USA. OTC 8135.

20. JRA, 1974. Technical standard for oil pipelines. Japan Road Association.

Sample

copy

- not

for di

stribu

tion

What if you could have the acclaimed Pipeline Engineering courses from Clarion Technical Conferences and Tiratsoo Technical

right on your desktop?

NOW YOU CAN!In partnership with Penspen Integrity, Clarion and Tiratsoo Technical are pleased to announce the immediate availability of the popular course via online distance-learning:> Pipeline Integrity Assessment (advanced level): a self-paced learning and reference resource.Purchase the entire course or targeted groups of modules of most interest. All complete course purchases receive a 20% discount. COURSE ALUMNI: If you attended the public course offered by Clarion and Tiratsoo Technical, you are eligible for a 10% discount on any of the four targeted groups of modules available, or 25% off the complete course.

For more information and to purchase the online courses, visit www.tiratsootechnical.com

OnlineLearningCourses_2013_FP.indd 1 6/12/13 11:19 AM

Sample

copy

- not

for di

stribu

tion

Sample

copy

- not

for di

stribu

tion

Journal of Pipeline Engineering distribution - … - Dec 13...headline refers to the rate of...

Documents

Transcript of Journal of Pipeline Engineering distribution - … - Dec 13...headline refers to the rate of...