September, 2012 Vol.11, No.

Journal of
Pipeline Engineering
incorporating
The Journal of Pipeline Integrity

n
io
st y
ut
di op
rib
or c
t f ple
no m
Sa

Great Southern Press Clarion Technical Publishers

 Journal of Pipeline Engineering

Editorial Board - 2012

Obiechina Akpachiogu, Cost Engineering Coordinator, Addax Petroleum Development Nigeria, Lagos,
Nigeria
Dr Husain Al-Muslim, Pipeline Engineer, Consulting Services Department, Saudi Aramco, Dhahran,
Saudi Arabia
Mohd Nazmi Ali Napiah, Pipeline Engineer, Petronas Gas, Segamat, Malaysia
Dr-Ing Michael Beller, Landolt Steuer & Unternehmensberatung AG, Luzern, Switzerland
Jorge Bonnetto, Operations Director TGS (retired), TGS, Buenos Aires, Argentina
Dr Andrew Cosham, Atkins Boreas, Newcastle upon Tyne, UK
Dr Sreekanta Das, Associate Professor, Department of Civil and Environmental Engineering, University
of Windsor, ON, Canada
Prof. Rudi Denys, Universiteit Gent – Laboratory Soete, Gent, Belgium
Leigh Fletcher, Welding and Pipeline Integrity, Bright, Australia
Roger Gomez Boland, Sub-Gerente Control, Transierra SA,
Santa Cruz de la Sierra, Bolivia
Daniel Hamburger, Pipeline Maintenance Manager, El Paso Eastern Pipelines, Birmingham, AL, USA

n
Prof. Phil Hopkins, Executive Director, Penspen Ltd, Newcastle upon Tyne, UK

io
Michael Istre, Engineering Supervisor, Project Consulting Services,
st y
ut Houston, TX, USA
di op

Dr Shawn Kenny, Memorial University of Newfoundland – Faculty of Engineering and Applied

Science, St John’s, Canada
rib
Dr Gerhard Knauf, Salzgitter Mannesmann Forschung GmbH, Duisburg, Germany
or c

Prof. Andrew Palmer, Dept of Civil Engineering – National University of Singapore, Singapore
Prof. Dimitri Pavlou, Professor of Mechanical Engineering,
t f ple

Technological Institute of Halkida , Halkida, Greece

Dr Julia Race, School of Marine Sciences – University of Newcastle,
Newcastle upon Tyne, UK
no m

Dr John Smart, John Smart & Associates, Houston, TX, USA

Jan Spiekhout, Kema Gas Consulting & Services, Groningen, Netherlands
Sa

Dr Nobuhisa Suzuki, JFE R&D Corporation, Kawasaki, Japan

Prof. Sviatoslav Timashev, Russian Academy of Sciences – Science
& Engineering Centre, Ekaterinburg, Russia
Patrick Vieth, Senior Pipeline Engineer - Pipelines & Civil Engineering, BP America, Houston, TX,
USA
Dr Joe Zhou, Technology Leader, TransCanada PipeLines Ltd, Calgary, Canada
Dr Xian-Kui Zhu, Senior Research Scientist, Battelle Pipeline Technology Center, Columbus, OH,
USA

���
3rd Quarter, 2012 141

The Journal of
Pipeline Engineering
incorporating
The Journal of Pipeline Integrity

Volume 11, No 3 • Third Quarter, 2012

Contents
Dr Jan Hayes and Prof. Andrew Hopkins.................................................................................................................145

n
Deepwater Horizon – lessons for the pipeline industry

io
Russell Cooper........................................................................................................................................................... 155
st y

ut
National Grid’s COOLTRANS research programme
di op
rib
Dr Julia M Race, Dr Ben Wetenhall, Dr Patricia N Seevam, and Prof. Martin J Downie....................................173
or c

Towards a CO2 pipeline specification: defining tolerance limits for impurities

t f ple

Dr Phil Cleaver and Harry Hopkins.................................................................................................................................. 191

The application of individual and societal risk assessment to CO2 pipelines
no m

Dr Simon Gant......................................................................................................................................................................201
Framework for validation of pipeline release and dispersion models for the COOLTRANS research programme
Sa

Dr Andrew Cosham.............................................................................................................................................................. 213

The saturation pressure and the design of dense-phase CO2 pipelines

Daniel Sandana, Matthew Hadden, Dr Julia Race, and EA Charles...................................................................... 229

Transport of gaseous and dense carbon dioxide in pipelines: is there an internal corrosion risk?

Dr Janice A Lake, M D Steven, K Smith, and B H Lomax..................................................................................... 239

Environmental impact of a hypothetical catastrophic leakage of CO2 onto the ground surface

❖❖❖

OUR COVER PHOTO shows a pipeline booster station at Jackson Dome on Texas-
based Denbury Resources’ 293-km long, 20-inch diameter, NEJD CO2 pipeline The Journal of Pipeline Engineering
that runs from Jackson Dome to near Donaldsonville, Louisiana, as part of the the has been accepted by the Scopus
Content Selection & Advisory Board
company’s 2001 acquisition of the Jackson Dome source field. The photo is reproduced (CSAB) to be part of the SciVerse
by courtesy of Denbury Resources Inc. Scopus database and index.
142 The Journal of Pipeline Engineering

T HE 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 ScholarOne 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
1. Disclaimer: While every effort is made to check the 4. Back issues: Single issues from current and past volumes
accuracy of the contributions published in The Journal of are available for US$87.50 per copy.
Pipeline Engineering, Great Southern Press Ltd and Clarion

n
Technical Publishers do not accept responsibility for the 5. Publisher: The Journal of Pipeline Engineering is
views expressed which, although made in good faith, are published by Great Southern Press Ltd (UK and Australia)

io
those of the authors alone. and Clarion Technical Publishers (USA):
st y
ut
di op

2. Copyright and photocopying: © 2012 Great Southern Great Southern Press, PO Box 21, Beaconsfield
rib
Press Ltd and Clarion Technical Publishers. All rights HP9 1NS, UK
or c

reserved. No part of this publication may be reproduced, • tel: +44 (0)1494 675139
stored or transmitted in any form or by any means without • fax: +44 (0)1494 670155
t f ple

the prior permission in writing from the copyright holder. • email: jtiratsoo@gs-press.com
Authorization to photocopy items for internal and personal • web: www.j-pipe-eng.com
use is granted by the copyright holder for libraries and • www.pipelinesinternational.com
no m

other users registered with their local reproduction rights

organization. This consent does not extend to other kinds Editor: John Tiratsoo
Sa

of copying such as copying for general distribution, for • email: jtiratsoo@gs-press.com

advertising and promotional purposes, for creating new
collective works, or for resale. Special requests should Clarion Technical Publishers, 3401 Louisiana,
be addressed to Great Southern Press Ltd, PO Box 21, Suite 255, Houston TX 77002, USA
Beaconsfield HP9 1NS, UK, or to the editor. • tel: +1 713 521 5929
• fax: +1 713 521 9255
3. Information for subscribers: The Journal of Pipeline • web: www.clarion.org
Engineering (incorporating the Journal of Pipeline Integrity)
is published four times each year. The subscription price for Associate publisher: BJ Lowe
2012 is US$350 per year (inc. airmail postage). Members of • email: bjlowe@clarion.org
the Professional Institute of Pipeline Engineers can subscribe
for the special rate of US$175/year (inc. airmail postage). 6. ISSN 1753 2116
Subscribers receive free on-line access to all issues of the
Journal during the period of their subscription.

v v v

www.j-pipe-eng.com
is available for subscribers
3rd Quarter, 2012
Editorial
Guest editorial: CO2 pipelines
T his special issue of the Journal of Pipeline Engineering is
dedicated to the topic of carbon dioxide transportation
by pipeline, and is the second issue of the Journal that has
been devoted to this theme. The first edition in 2010 (Vol.
9, No. 4) published key papers from the First International
Forum on the Transportation of CO2 by Pipeline held
in Newcastle upon Tyne, UK, and organized by Tiratsoo
Technical and Clarion Technical Conferences in association
with Newcastle University. Since this first Forum, two
further Fora have been run, and this edition of the Journal
of Pipeline Engineering publishes a selection of papers from
the 2012 Forum, held in June in Newcastle.
n
The aims of the first Forum were to highlight the key issues
io
relating to CO2 pipelines and bring together leading academics
st y
ut
and industry experts to discuss current and future Carbon
di op
Capture and Storage (CCS) ventures and the international
rib
research activities being undertaken to support these projects.
or c
So how has this very important research area moved on in
the intervening two years? Although the aims of the Forum
t f ple
have ostensibly remained the same, it was noticeable at
the recent event that, rather than describing proposals for
research programmes, as had been the case in 2010, presenters
no m
were now publishing the initial results from these research
programmes or describing construction projects for CO2
Sa
pipelines, specifically for CCS schemes. This change in
focus of the papers being presented at the Forum reflects
the commitment of international governments to move from
pilot to demonstration plant schemes for CCS1, a move that
must be welcomed if CCS is going to deliver the required
impact for climate change mitigation.
One of the research programmes that has been initiated in the
last two years, and which is beginning to impact the design and
construction of high pressure anthropogenic CO2 pipelines,
is the COOLTRANS (CO2 Liquid Pipeline Transportation)
project funded by National Grid. In the first paper in this
issue, Russell Cooper describes the approach that is being
applied in the COOLTRANS research programme to allow
the development of a comprehensive safety justification for
the onshore pipeline transportation of dense phase CO2
in the UK. According to Mr Cooper, one of the critical
gaps that was identified by the project team was the fact
that current UK pipeline codes were not applicable to the
transportation of CO2 without additional development.
1. UKCCS Research Centre. Demonstration and deployment of CCS. Available
from http://www.ukccsrc.ac.uk/uk-ccs-community/what-ccs/demonstration-and-
deployment-ccs Accessed on 14.08.12.
143
His paper describes how a strategy was developed to fill
this gap through the specification of a research programme
that combines theoretical studies with an underpinning
experimental test programme. The paper describes the
initial results from the theoretical studies and experimental
tests and, more significantly, indicates how these results are
being rapidly integrated into the design process.
One of the conclusions from Russell Cooper’s paper is
that the composition of the CO2 into the pipeline is a key
pipeline design parameter. This conclusion is expanded in the
second paper in this issue which describes work that has been
conducted by Newcastle University as part of the MATTRAN
(Materials for Next Generation CO2 Transport Systems) project
to investigate the definition of a CO2 pipeline specification.
The CO2 specification from a capture plant is dependent
on the technology that is being used to capture the CO2
and the type of fuel that is being burnt (in the case of power
plant capture). The approach that is taken in this paper is to
analyse the effects that the impurities in the CO2 could have
on pipeline design (including hydraulic efficiency, pipeline
sizing and pressure drop), fracture control, material degradation
and health and safety with the aim of specifying maximum
impurity levels for entry into the pipeline system. However, the
main conclusion of the paper is that the specification is system
specific and that, in defining a CO2 purity, a balance has to be
struck between techno-economic considerations and limiting
material properties specifically related to fracture control and
corrosion and cracking mechanisms. These themes are taken
up in following papers by Dr Andrew Cosham and by Daniel
Sandana and co-authors.
The paper by Andrew Cosham focuses on the constraints
placed on the CO2 specification by the requirement to
prevent long-running ductile fractures. This paper is
significant in that, up to this point, the effects of composition
on crack propagation and therefore ultimately on the
required toughness of the pipeline had not been considered.
Dr Cosham presents a methodology, based on the Battelle
Two-Curve Method, which relates the composition of the
CO2 stream to its decompression behaviour and then to the
toughness and wall thickness required for the prevention
of ductile fracture propagation.
In his paper Transport of gaseous and dense carbon dioxide
in pipelines: is there a corrosion risk?, Daniel Sandana and
his co-authors presents a review of relevant literature related
to the exposure of carbon steel to CO2 mixtures containing
144
water. The main conclusion from this review is worthy of
emphasis which is that corrosion rates in pure CO2 and CO2
containing impurities are negligible as long as the product
stream remains dry. Nevertheless, the paper presents three
operational scenarios in which upsets in dehydration could
result in water being present in the pipeline. The authors
then explore the corrosion effects that might be expected
in each scenario based on the experimental data available.
However, one of the main conclusions that they draw is that
a larger database of experimental data is required, particularly
for CO2 containing impurities.
For the final three papers in this special issue, the focus turns
to risk assessment, dispersion prediction, and environmental
impact of CO2 pipelines. The first of these papers by Dr
Phil Cleaver and Harry Hopkins presents a methodology for
routeing a dense-phase CO2 pipeline. What is novel in this
approach is that the authors propose a methodology based
on societal risk assessment rather than the “traditional”
natural gas approach of area classification and reduced
design factors.
n
Russell Cooper described the modelling predictions and
experimental validation that is being conducted as part of
io
the COOLTRANS project in the area of dispersion of CO2
st y
ut
from different failure scenarios in the pipeline. This area is
di op
expanded by Dr Simon Gant in his paper Framework for
rib
validation of pipeline release and dispersion models for
or c
the COOLTRANS project. Dr Gant describes a framework
that can be used to validate dispersion models for use in
t f ple
no m
P receding the papers on CO2 Transportation, the first
Sa
paper of this issue examines the Deepwater Horizon
incident in the Gulf of Mexico, and the lessons that can
be learned from this by the pipeline industry. Dr Jan
Hayes and Prof. Andrew Hopkins are from the School
of Sociology of the Australian National University, and
they therefore are able to examine the problem with a
refreshing and interesting perspective. In their paper,
the authors highlight the Swiss cheese model as a useful
way of introducing the concept of what others might
call Murphy’s Law, paraphrased as ‘if things can go
wrong, they will’. Dr Hayes and Prof. Hopkins point out
that social science research shows that high-performing
organizations seek opportunities to learn from failures,
and Deepwater Horizon provides such an opportunity
for the pipeline industry. Whilst the technical details of
well-control systems may not be directly relevant, there are
broader organizational factors that can provide significant
lessons to any organization that designs, maintains, and/
or operates complex and potentially hazardous technology.
Accident investigations, including Deepwater Horizon,
continue to highlight common organizational failures
that can have catastrophic consequences, and it is these
lessons that are the subject of this paper. The key question
The Journal of Pipeline Engineering
quantitative risk assessments for CO2 pipelines that is based
on an independent assessment of their performance.
The final paper in this issue is by Dr Janice Lake and co-
authors from the University of Nottingham. Dr Lake has
been conducting experimental studies on the effect of a
simulated leak from a CO2 pipeline on plant biochemistry.
Although in this study it was concluded that there was no
lasting effect on the flora for the leak scenario simulated,
further studies in this area are being conducted to simulate
different leak and rupture scenarios.
As you read the papers in this issue I hope that you are
encouraged by the breadth of research work in CO2 pipeline
transportation that is now starting to be published. However,
many of the papers conclude that there is still further work to be
done and a limited time in which to do it to meet the challenges
of climate change mitigation afforded by CCS. It is therefore
critical that researchers continue to work collaboratively and
it is hoped that as much information as possible is published
in the public domain to allow the rapid dissemination and
implementation of research results.
Dr Julia Race
Senior Lecturer in Pipeline Engineering
School of Marine Science and Technology,
Newcastle University
Newcastle upon Tyne, NE1 7RU
UK
Email: julia.race@ncl.ac.uk
is why accidents continue to occur despite the level of
engineering and management system controls that are
apparently in place.
The authors quote some of the considerable volume of
published material available regarding the causes of the
Deepwater Horizon blowout, and their paper is based
on analysis of that material to emphasize lessons that are
relevant to management of pipeline companies, including:
managing the cost vs. safety tradeoff, the down side of
good news, and understanding the difference between
personal safety and system integrity. The paper makes
some very important points but also asks some far-reaching
questions of those responsible for pipeline systems and
networks and is worthy of study by all with responsibilities
for operating pipelines. Among these questions are: How
does your organization maintain a focus on safety in the
face of apparent success? How would you know if norms
were changing? In the face of cost pressures, how do you
judge if sufficient safety redundancy still exists in your
systems? What key tests are done of your systems? Do
systems ever fail the tests and, if not, are you sure the
test is effective? We hope the answers do not keep you
awake at night.
3rd Quarter, 2012 145

Deepwater Horizon – lessons for

the pipeline industry
by Dr Jan Hayes* and Prof. Andrew Hopkins

The Australian National University, Canberra, ACT, Australia

T he Deepwater Horizon incident has impacted significantly on the offshore drilling industry in the US and
globally. The final environmental, political, and financial consequences will not be known for decades,
but it is already clear that the incident has changed, and will continue to change, the way business is done
for all operating companies, drilling contractors, regulators, and industry associations that function in the
offshore oil and gas industry.

Social science research shows that high-performing organizations seek opportunities to learn from failures,
and Deepwater Horizon provides such an opportunity for the pipeline industry. Whilst the technical
details of well-control systems may not be directly relevant, there are broader organizational factors

n
that can provide significant lessons to any organization that designs, maintains, and/or operates complex
and potentially hazardous technology. Accident investigations, including Deepwater Horizon, continue to

io
highlight common organizational failures that can have catastrophic consequences, and it is these lessons
st y
ut
that are the subject of this paper. The key question is why accidents continue to occur despite the level
di op

of engineering and management system controls that are apparently in place.

rib
or c

There is an enormous volume of published material available regarding the causes of the Deepwater
Horizon blowout. This paper is based on analysis of that material to draw out lessons that are relevant to
t f ple

management of gas pipeline companies including:

• managing the cost / safety tradeoff

no m

• the down side of good news

• maintaining a focus on the need for multiple barriers
Sa

• the value of professional judgment and experience

• understanding the difference between personal safety and system integrity
• contractor relationships, and
• linking incentives for senior management to system safety performance.

A key business objective for the pipeline

industry is to effectively control risk to the public from
hydrocarbon pipelines and associated facilities. Strategies
for how to achieve this can be drawn from studies of high
reliability organizations (or HROs) [1-3]. This social-science
work on high-performing organizations that manage complex
and hazardous technologies has emphasized the benefit of
seeking to learn from a wide range of failures. Using trial-
and-error learning, i.e. waiting for a significant failure to
occur within one’s own organization and then adjusting
company systems to prevent a recurrence, is not a sufficient
strategy in organizations where the consequences of failure
are potentially severe. In addition to the potentially terrible
*Corresponding author’s contact details:
tel: +61 2 6125 2140
email: jan.hayes@anu.edu.au
physical consequences for those people directly involved in
a serious industrial incident, such events can be catastrophic
in business and political terms for the organizations involved
due to the response of regulators, governments, investors,
and other industry participants.
For this reason, the pipeline industry should be very
interested in the lessons that can be learned from recent major
disasters in the offshore oil and gas sector, in particular the
loss-of-well-control incident that occurred when a well was
being drilled by the Deepwater Horizon rig in the Macondo
field in the Gulf of Mexico.
Loss-of-well-control incidents, known commonly as blowouts,
are a well-known hazard in the offshore oil and gas industry.
There were 39 blowouts on the US Outer Continental
146
Shelf in the period 1992 to 2006 [4]. Eight months prior
to the events on the Deepwater Horizon, a blowout off the
north coast of Australia (known as the Montara incident)
provided to this global industry yet another reminder of the
need for comprehensive well control [5, 6]. As a result of the
potential for disaster, and because of the large sums of money
involved, well-construction activities are tightly controlled,
both within operating companies and by regulation. How
is it then, that in 2009/10, the offshore oil and gas industry
can experience such incidents? This paper provides some
answers to that question about the Deepwater Horizon
blowout in particular. As will be described, those responsible
for the technical integrity of the hazardous work involved
in drilling and completing the well demonstrated a poor
understanding of both the hazards inherent in the work
they were doing and the measures that should have been
in place to manage those hazards. The prevailing attitude
was apparently one of forging ahead, regardless: this seems
to have been the case both for those on the facility and for
those in onshore technical and management positions.
This paper looks critically at why these choices were made
n
and reflects on some questions that organizations in the
pipeline industry should perhaps ask themselves to reduce
io
their vulnerability to similar failures, also with potentially
st y
disastrous consequences.
ut
di op
Organizational accidents
rib
or c
Social-science research has shown that disasters and potential
t f ple
disasters can most usefully be thought of as organizational
accidents [7, 8]. Organizational accidents are events that
occur within complex modern technologies such as nuclear
no m
power stations, commercial aviation, and oil and gas facilities.
They have multiple causes involving many people operating in
Sa
different areas and at different levels. Analysing an incident
in this framework involves a search for technical causes, but
also for causes related to systems of work and the actions of
people throughout the organization.
The most widely recognized model of organizational
accidents is James Reason’s Swiss cheese model shown in
Fig.1 [8]. In this way of thinking about accidents, there is in
place a range of defences that are functionally designed to
prevent any given hazard from leading to a loss of some kind
(such as an accident). In practice, these defences are imperfect
(like holes in Swiss cheese), but the multiple hardware and
procedural measures that should be in place ensure that
failure of any individual measure is not catastrophic. An
accident occurs only when the holes in the cheese line up
and provide an accident trajectory through all the defences.
In this model, the ‘holes’ in the cheese have two interesting
features. Firstly, they may be due to active failures, for
example an operating error of some kind, or they may be
latent failures. Latent failures are weaknesses in the system
that do not, of themselves, initiate an accident and may lie
undetected for months or years but they fail to prevent an
The Journal of Pipeline Engineering
Fig.1. Swiss cheese model [10].
accident when an active failure calls them into play on a
given day. Problems arise when latent failures in the system
accumulate – maintenance is not carried out, records are
not kept, audits are incomplete. The consequence of a small
active failure can then be catastrophic as the protective
systems fail to function as expected.
The second quality of the holes in the Swiss cheese is that
they are a function of the organization itself. In this model
of accident causation, operator actions in the field are
linked to workplace factors such as competency, rostering,
control-room design, task design, etc., and these issues
are linked to organizational factors such as budgets, safety
priorities, management styles, etc. In this way of thinking
about safety defences, the performance of each component
in the system is interlinked. Organizational failures are seen
as similar to what reliability engineers would call common
mode failures, i.e. a serious organizational failure can cause
multiple, simultaneous failures across a range of apparently
independent technical barriers.
The San Bruno pipeline failure, which led to the fire in
the suburbs of San Francisco in September, 2010, is a good
illustration of how inadequacies in both the operating
organization and the regulatory bodies allowed a seriously
defective pipe to remain in service and undetected until
it failed catastrophically. The report of the National
Transportation Safety Board included explicit findings
that the failure was the result of an organizational accident,
that there was a systemic problem within the operating
organization, and that ineffective regulation allowed these
organizational failures to continue over many years [9, see
Findings 21, 22, and 27].
The concept that accidents occur within an organizational
context explains why accidents from other industries may
have useful lessons for the pipeline industry, even when
the technical circumstances of a particular accident may
not be particularly relevant to pipelines. One difference
between failures in pipelines and other process industries
is that pipeline incidents are often not the direct result of
some active operation on the part of pipeline personnel.
Rather, pipeline incidents may more often be initiated by
3rd Quarter, 2012
third parties or result from corrosion, construction defects,
ground movements, etc. Nevertheless it is the latent failures
within a pipeline organization’s systems that can allow a
relatively minor incident to escalate to a catastrophe.
With these issues in mind, we now turn to the sequence of
events that led to the blowout.
Deepwater Horizon
The Deepwater Horizon was a semi-submersible drilling
rig owned and operated by Transocean. In April, 2010, the
rig was working on contract to BP undertaking exploration
drilling in the Macondo field in the Gulf of Mexico. The water
depth at Macondo was approximately one mile, introducing
some major technical challenges to the drilling programme
(and to the response after the blowout had occurred).
Work on the Macondo well began in October, 2009, using
a different drilling rig. Drilling proceeded for 34 days before
the work was interrupted by a hurricane and the rig moved
off station. The Deepwater Horizon was moved into place
n
on 31 January, 2010, to recommence the work. By this time
the drilling programme was six weeks behind schedule and
io
$58 million over the budget of $96.2 million. Despite the
st y
ut
weather delay and a series of technical problems, the well
di op
was seen as a success overall. Complex testing of the well
rib
once the target depth had been reached assessed that it had
or c
significant production potential. The decision was made to
set production casing in the well before the rig departed so
t f ple
that it would be ready for recovery of oil and gas at a later
date. The final stage of the work before departure was setting
the production casing in the well and cementing it in place.
no m
The detailed sequence of events that led to the blowout
Sa
started with cementing. The goal of the cementing operation
is simple to understand. Fluid is pumped down the well
through the casing and back up the annular space between
the outside of the casing and the wellbore. A slug of cement
slurry is included and measurements of pressure and pumped
volume are used to try to ensure that the cement is located
correctly deep underground to secure the well and prevent
hydrocarbons from flowing up the inside of the casing or
up the annulus. Once the cement has been allowed to set,
various tests can be conducted to ascertain the integrity of
the arrangement. Whilst this sounds simple, in practice
the cement volume is located literally miles away from the
rig floor and the pumps. There are many ways in which
the integrity of the cement can be compromised and these
can only be inferred from the operational data available
at the time.
In the case of Macondo, there had been some earlier
problems with the well (fracturing of the reservoir) caused
by the high pressures in the drilling fluids necessary for such
a deep well. In order to minimize the potential for further
fracturing, the engineers made several decisions regarding
how the cement was to be put in place that minimized the
147
operating pressure, at the expense of increasing the potential
to compromise the integrity of the cement left in place.
Because of some preliminary indications of success, a
decision was also made not to conduct a specialist cement
integrity test and the contractors were sent home (saving
time and direct cost). Instead, a negative pressure test would
be conducted: the pressure in the well would be reduced
to ensure there was no inflow or increase in pressure
which would indicate leakage from the reservoir, through
the cement, into the casing itself. The well failed this test
three times: in other words, three times the pressure was
bled down but then rose again within the casing. The test
was then repeated, with pressure measured via a different
connection (but one that should have been hydraulically
connected to the other measurement point). When the
pressure was not seen to rise, this was interpreted as a
successful test (although there was other evidence that this
pressure measurement was flawed and that the test should
still be seen as a failure).
The final stage in the activities on the Deepwater Horizon
to temporarily abandon the well was for drilling mud to
be displaced out of the wellbore and casing by seawater,
a fluid with a lower density. The volume of mud being
returned (and the pressure at various points in the system)
can be monitored during this operation to confirm that
no hydrocarbons are entering the well as the static head of
the heavy drilling mud is replaced by seawater. In drilling
jargon, the well is ‘overbalanced’ with the heavier mud in
the well. Data available after the event show that there were
indications of mounting trouble during the last hour before
the blowout. It appears that those with responsibility for
monitoring these parameters may have been distracted by
other activities going on at the time (in preparation for the
departure of the rig from the field). This is discussed more
fully in a forthcoming book [10].
During this final stage of the work, control of the well was
lost and firstly drilling mud and then large volumes of oil
and gas flowed from the well in an uncontrolled manner,
spraying widely across the facility. The gas cloud ignited
almost immediately and the people on the drill floor died
in the subsequent fire. Ultimately, 11 people were killed and
the remaining crew abandoned the facility which sank the
following day. Oil continued to leak from the well at the sea
floor for almost three months before attempts to cap the
well and stop the hydrocarbon flow were at last successful.
The final device that should have been able to prevent
complete loss of well control at the time of the initial event
was the blowout preventer (BOP). This device includes several
systems designed to seal the well. In this case, closure of the
BOP was not initiated as well-control problems developed,
but only after the blowout was in progress. It is likely that if it
had been activated it may not have been functional because
of a lack of preventative maintenance which had left parts of
the actuation system in an unserviceable condition. There
148
Barrier Testing/monitoring
Cemented casing string Negative pressure test mistakenly seen to be successful. Not effective
Overbalanced well fluids Mud volume monitoring (and other parameters).
Blowout preventer (BOP) Preventative maintenance / function testing not done.
Table 1. Barrier status.
was a further significant design issue with the BOP. Even if
functioning correctly, it needed to be initiated BEFORE
full flow from the well was established. In the operating
conditions seen at Macondo, the BOP was not designed
to be able to stop completely uncontrolled flow from the
well once it was established.
In summary, the devices and systems in place to prevent
flow from the well (and their status) are shown in Table 1.
The Report of the US National Commission [11, p.90]
n
states that “the well blew out because a number of separate
risk factors, oversights, and outright mistakes combined
io
to overwhelm the safeguards meant to prevent just such
st y
ut
an event from happening. But most of the mistakes and
di op
oversights at Macondo can be traced back to a single,
rib
overarching failure – a failure of management. Better
or c
management by BP, Halliburton, and Transocean would
almost certainly have prevented the blowout by improving
t f ple
the ability of individuals involved to identity the risks
they faced, and to properly evaluate, communicate, and
address them. A blowout in deepwater was not a statistical
no m
inevitability.”
Sa
The technological failures related to cementing and well
control are clearly not directly relevant to the pipeline
industry, but using the organizational-failures’ model
described earlier, there are broader lessons to be learned
which may be just as relevant to us. Some of these are
described in the following sections.
Cost versus safety
Rig time is expensive. Any competent company would be
concerned about the time taken to drill a well but, on the
Deepwater Horizon, cost considerations appear to have
taken precedence over the technical integrity of the work
done and hence ultimately the safety of personnel. There
are many examples of this in the way drilling decisions were
made at BP prior to the blowout, but two are described here.
One of the key decisions made in the period immediately
prior to the cementing work was about the use of
centralizers. These devices are used to keep the casing in
the centre of the wellbore in order to facilitate good flow
of cement up the annular space between the outside of the
casing and the wellbore itself, and thus help to ensure a
The Journal of Pipeline Engineering
Status
Crew apparently distracted by
other activities.
good seal. On the other hand, centralizers are also a possible
source of operational problems during installation if the
protrusions hang-up on other equipment as the casing
with the centralizers on the outside is pushed literally
miles through concentric pipes. Effective design of the
centralizers themselves and their placement along the
length of the entire string of piping is therefore a key well
integrity and operational issue. BP’s initial well design called
for at least 16 centralizers, but in the event only six were
installed. This was the subject of much discussion amongst
the offshore crew who saw their options as constrained by
the equipment immediately available to them. There were
only six centralizers available from the supplier of the type
favoured by the BP Wells Team Leader. More devices of a
different design (but with the same function) were available,
but he was concerned that they were of a design that was
more difficult and time-consuming to install and had a
greater chance of hanging up during installation. The pros
and cons of the various options were discussed amongst
the BP team and its specialist contractor (Halliburton).
This decision was not formally documented and in the
end, as one BP engineer wrote, “But, who cares, it’s done,
end of story, [we] will probably be fine and we’ll get a good
cement job” [11, p.116].
One of the BP engineers gave some indication of the
thinking behind this choice when he wrote in an email
that using only the six centralizers of the preferred type
was “right on the risk/reward equation” [11, p.97]. This
risk/reward tradeoff could have been simply related to
cost. Installing extra centralizers of the less-preferred type
would incur the cost of the equipment items themselves,
the cost of the rig time to install them, and the possible
additional rig time in the event of an installation problem.
Not incurring these costs is the reward. The potential risk of
going ahead with fewer centralizers would be an ineffective
cement job due to channelling, which would then require
significant time (and hence expense) to fix. Of course this
assumes that the integrity of the cement job would be
thoroughly tested and then fixed if necessary. Without
this step, the risk/reward equation becomes a direct cost/
safety tradeoff – saving money at the expense of making
a blowout more likely. This is, in fact, what happened.
A second key decision where cost and safety were in direct
conflict occurred after the cementing work itself had been
completed. The BP cementing plan for the well included
3rd Quarter, 2012
the option of conducting a test called a cement bond log
once the cement had been pumped into the well. This test
required the services of a specialist contractor team (from
Schlumberger) and the team had been standing-by on the
rig for at least one day waiting to see if its services would be
required. Aspects of the actual cementing activity had not
gone smoothly. Adding this experience to the situation with
the centralizers described above could be seen in hindsight
to provide ample justification for conducting an additional
integrity test, such as the cement bond log. Instead, the
decision was made to send the specialists back onshore,
saving the $128,000 fee and the rig time required for the test.
The pipeline industry has structured management systems
in place to ensure that the tension between cost and safety
is managed when large decisions that have an obvious
potential safety impact are made. In Australia, pipeline design
decisions are guided by the safety management study process
detailed in AS2885 which is intended to ensure that safety
is properly considered at all stages of a pipeline’s life cycle
from preliminary planning through to decommissioning.
The Deepwater Horizon incident reminds us, however, that
n
the tension between cost and safety is ever-present when
dealing with hazardous operations in the field, and that
io
high-stakes’ changes are sometimes made on the run. Whilst
st y
ut
two examples have been described here, the Presidential
di op
Commission report lists five other examples of decisions
rib
made offshore in the days before the blowout that were
or c
driven by cost and increased risk of failure [11, p.125].
t f ple
The safety implications of the decisions described above
were presumably seen as negligible at the time, but they
contributed to disastrous consequences. It is easy for
no m
awareness of this tension to be lost in run-of-the-mill daily
decision-making by field personnel, leading to drift in
Sa
standards regarding what is acceptable. Consider how your
people in the field make those choices when conducting
maintenance or when putting new facilities into service (for
example). Are all safety versus cost/schedule tradeoffs made
with an explicit consideration of alternatives and the risks
involved? And, critically, in your organization, how would
you know if they were not?
Expecting good news
We have already described how the decision regarding
centralizer installation made the cement job more vulnerable
to channelling and hence of compromised integrity. We have
also pointed out the link to the later decision not to conduct
a cement bond log because it was seen to be unnecessary.
Even without the cement bond log, there was another
opportunity for the problems with the integrity of the
cementing to be revealed. This relates to the test that was
conducted on the well after cementing, the negative-pressure
test. For this test, the pressure in the well is reduced to
simulate the case where the rig has departed. The pressure
at the top of the shut-in well is then monitored to see
149
if any build-up occurs. Any increase in pressure means
that hydrocarbons are leaking into the wellbore from the
reservoir and that the cement job is compromised. Whilst
the test is somewhat complicated by the need to take into
account the static head of the various fluids that can be in
the wellbore, it is, in essence, this simple. On the Deepwater
Horizon, the pressure was bled down but instead of achieving
zero psi, the minimum pressure that could be achieved was
266psi. When the bleed line was shut in, the pressure rose
to 1262psi. Some adjustments were made to the equipment
and the test was repeated twice. Whilst the adjustments
made it possible to get the pressure down to zero, each time
significant pressure quickly built back up in the system.
At this stage, the only credible explanation should have
been that the well had failed the test and that the cement
was leaking. There is no other source of pressure. Instead,
the test was repeated for a fourth time using a different
measuring point. This point was on a different line, but
in effect should have been hydraulically connected to the
original sensing location, i.e. the pressure readings from the
two locations should have been consistent. On this fourth
occasion, the pressure at the new sensing point remained at
zero and the team declared the test a success and the integrity
of the cement to be proven. No credible explanation was
offered or found for the three failed tests (or the continuing
high pressure at the original measurement location during
the fourth test).
Before the negative pressure tests began, various celebratory
and congratulatory emails were sent out (from the
Halliburton crew to its onshore manager, and from
one BP person to the rest of the BP team). The general
message of the emails was that the cement job had been
successfully completed – a job well done by everyone. In
fact, the work was not complete because it had not been
tested, but in the minds of all participants the work was
over. The negative pressure test is a powerful example
of what psychologists call confirmation bias, i.e. bias
towards seeing evidence that confirms a particular view,
in this case that the cement job was successful and the
well was secure. Despite overwhelming evidence to the
contrary, confirmation bias leads the team to hold this
view to be true. The only evidence that is seen as valid is
the fourth test and the evidence of the three earlier tests
is dismissed without needing any explanation because it
fails to confirm the view that is believed to be factual.
This is a very different view of safety barriers to that put
forward by the Swiss cheese model described earlier. In
that view, all defences are seen as imperfect no matter
how hard we try to make them perfect, and so evidence
of faults and flaws in the system is very valuable.
It must be said that requirements for multiple barriers were
also built-in to many of the standards and procedures that
applied to drilling operations on Deepwater Horizon and
yet those involved did not behave as if such requirements
were necessary. The answer may rest with how experienced
150
people learn. We learn by trial and error. Consciously and
subconsciously we learn from our experience – what worked
last time and what did not. This is very dangerous when
it comes to high-consequence, low-frequency, accidents
where the feedback most of us get over our professional lives
luckily does not include exposure to catastrophe. If we cut
corners with no apparent downside, then our expectation
of what is normal quickly shifts. Diane Vaughan’s analysis
of the Challenger disaster [12] tells us that this problem
is not limited to the oil and gas industry. The problem is
that the Swiss cheese acquires more and more holes with
no apparent penalty until disaster strikes.
This is why high-performing organizations go to great
lengths to learn from small failures. Recording, analyzing,
and sharing information about near misses and sharing
war stories of past failures as we are doing today is such
an important part of ongoing professional development
for all of us. Whilst the drilling team on Deepwater
Horizon on that day apparently could not conceptualize
that the cement job might have failed, there were plenty of
broader examples available in industry as to the credibility
n
of such a failure, the potential for catastrophe, and the
need for vigilance.
io
st y
ut
How does your organization maintain a focus on safety in
di op
the face of apparent success? How would you know if norms
rib
were changing? In the face of cost pressures, how do you judge
or c
if sufficient safety redundancy still exists in your systems?
What key tests are done of your systems? Do systems ever
t f ple
fail the tests and, if not, are you sure the test is effective?
Professionalism and judgment
no m
The preceding sections have described various technical
Sa
details of barriers to flow from the wells that should have
been in place and the ways in which these barriers were
compromised. Unfortunately, the people in control of
the work made – on the run – what were, in hindsight,
significant changes to drilling plans. This work was done
without effective consideration of the safety implications,
and those involved argued away flaws and faults in specific
devices, procedures, or actions that were designed to reduce
risk. Apparently, no-one had the necessary understanding of
the cumulative implications of all the judgments that were
being made in regards to well integrity (or if they did, they
failed to act on that understanding). HRO (high reliability
organization) theory emphasizes the value to organizations
of those experienced professional staff who have a broad
system knowledge – the people with a strong ‘gut feel’ for
when things are just not right [1]. Researchers in the field
of naturalistic decision making [13, 14] have found that
experienced people under time pressure make high-stakes’
decisions based on (often unconscious) pattern recognition.
This is in contrast to the classical decision-making process
which is often seen by technical organizations as desirable –
identifying and analyzing options and then making a rational
choice based on fixed criteria. Naturalistic decision making
The Journal of Pipeline Engineering
suggests that supporting good decision making in the field
involves a focus on increasing the store of experience that
people can draw on in any given situation. This includes
not only things that people have experienced personally, but
having the ability to draw on broader stores of knowledge
from within their profession.
Modern organizations place a great emphasis on management
and the need to control the actions of employees to be
in accordance with approved procedures consistent with
organizational goals and objectives. As we would expect
from resources companies in the 21st century, BP’s activities
were generally undertaken under company-wide safety-
management systems. There was an expectation that work
would generally be done in accordance with documented
procedures. In fact, even with such management systems in
place, complex technical organizations also rely critically on
the professionalism of their technical staff. On Deepwater
Horizon, it is clear in hindsight that personnel offshore
were routinely required to make decisions with significant
and potentially serious consequences in a way that relied
on their professional skills and judgment. Research suggests
this is typical in complex industrial situations, but often
unrecognized. One key example of this is that BP had no
internal procedure for running or interpreting negative-
pressure tests, despite the critical role that such a test
played. There was no Macondo-specific procedure prepared
for conducting the negative pressure test, nor was there a
procedure or policy in place regarding how confusing test
results should be treated. These issues were simply left to
the team offshore to deal with as best it saw fit. In practice,
as described above, poor choices were made. Undoubtedly,
additional written procedures could have been in place,
but it is inevitable that decisions will be made in the field
by drilling professionals. Given their critical nature, such
decisions must be supported.
Professionals can be seen as having the following features [15]:
• technical and theoretical expertise with the authority
and status flowing from such expertise;
• the establishment and the exercise of trust as a basis
for professional relationships (with clients, the public,
and between professionals);
• adherence to particular standards and professional
ethics;
• independence, autonomy, and discretion; and
• specific attitudes towards work, including dedication,
reliability, flexibility, and creativity.
It is not difficult to see that some of the actions taken
in the lead-up to the blowout were lacking some of these
features. The overarching driver appears to have been project
considerations of cost and schedule, rather than issues of the
technical integrity of the work. Qualities such as trust and
dedication seem to have been exercised by those involved
primarily in regards to cost issues, but not in relation to system
integrity. It should be noted that fostering a professional
3rd Quarter, 2012
attitude to system integrity is not simply an issue of ensuring
that sufficient technical expertise exists in an organization.
It is also about ensuring that this expertise is valued and has
an appropriate organizational voice so that concerns can be
acted upon. Given the right organizational focus, research in
high-performing organizations has emphasized the strength
and value of professional identity and the extent to which
professionals will go to ensure that functional impairment
of any barriers is minimized [16].
The issues associated with ensuring the continued availability
of the necessary professional expertise within the pipeline
industry should not be underestimated. Incidents such as
Deepwater Horizon serve to remind the pipeline industry of
the importance of acquiring and retaining the correct levels
of technical, theoretical, and professional expertise, and the
need to do this in a changing environment. The pipeline
industry may currently have access to well-trained technical
and theoretical expertise; however, much of this expertise will
be unavailable within the next ten years due to retirement of
older workers. We are aware that the continued availability of
technical and theoretical expertise for the pipeline industry
n
is regularly discussed within the industry. One initiative
from the Australian Pipeline Industry Association has been
io
to form the Young Pipeliners Forum which, among other
st y
ut
things, has provided a forum for the more senior experts to
di op
mentor younger industry participants. The general issue
rib
of ensuring a ‘talent pipeline’ of appropriate professional
or c
expertise into the future has also been identified as a
research opportunity during the Joint Technical Meeting
t f ple
of Pipeline Research Council International (PRCI), the
European Pipeline Research Group (EPRG), and the
Research and Standards Committee of the Australian
no m
Pipeline Industry Association (APIA RSC) in San
Francisco in May, 2011.
Sa
Contractor relationships
Offshore drilling operations are controlled by the lessee,
in this case BP, but typically nearly all the operations are
carried out by contractors. Two contractors in particular
played a part in the Macondo accident: Halliburton,
the company responsible for the cement used in sealing
the well, and Transocean, the company that owned and
operated the drilling rig. These are global companies
that BP relied on to follow appropriate procedures. The
Presidential Commission concluded that “the immediate
causes of the Macondo well blowout can be traced to a
series of identifiable mistakes made by BP, Halliburton,
and Transocean that reveal such systematic failures in
risk management that they place in doubt the safety
culture of the entire industry” [11, p.vii]. Leaving aside
BP’s mistakes for the moment, there is clearly a question
of contractor management and communications here.
One of the lessons that BP drew from the events was
that there was a need for much greater scrutiny of the
competence of its contractors. As was said on more than
one occasion, it is not enough for principals to trust that
151
their subcontractors behave professionally; they must
also verify that this is the case. This applies equally to
global companies like Transocean and Halliburton but
also to the host of smaller contractors used for other
specialist services.
There are also important questions here about the nature
of the contract between the two parties involved in any
contract, especially responsibilities. In essence, the party
paying for the goods or services holds significant power
over the provider, but the provider may have a higher
level of expertise and hence be the appropriate decision
maker over some issues, even when the consequences may
incur costs for the primary party, not the provider. Issues
discussed earlier about the tendency to prefer good news
over bad and the need to value professional judgment
become even more critical when the individuals involved
may have different employers.
There is clearly a lesson here for the pipeline industry.
Pipeline owners cannot and should not rely blindly on
the competence of their subcontractors. They should also
ensure that concerns of contractor professional staff can be
heard and not swamped by consideration of commercial
issues between the parties.
Lack of effective management control
Four senior managers (two from BP and two from Transocean)
were on the rig at the time of the blowout. They had arrived
on the facility about five hours earlier as part of a management
visibility tour. They witnessed some of the discussions on
the rig floor about how to deal with the conflicting negative-
pressure test results. They saw this activity as a cue to remove
themselves from the worksite so as not to interfere, rather
than as an opportunity to ask challenging questions.
As described above, people offshore sent emails with glowing
reports of their work to onshore management BEFORE the
work was actually completed – in the sense that cementing
had been completed but no integrity testing had been done.
In response to this, BP’s drilling operations managers replied
to say “great job guys” [11, p.4]. Ironically, this manager was
offshore only 12 hours later when the blowout occurred.
The pipeline-industry analogy to the offshore/onshore divide
is the office/field divide. With the exception of some of the
more remote operations bases, the barriers to crossing the
offshore/onshore divide are generally much greater than
those for the office/field divide within the pipeline industry.
In principle, it should therefore be logistically easier for
supervision and management to go to site to oversee what
is happening. The big issue is what to do on such visits.
Managers often think that the purpose of such visits is to
make themselves ‘visible’, to show that they care, and even to
tell workers about how important safety is. The management-
visibility tour of the Deepwater Horizon just hours before
the blowout graphically demonstrates the weakness of this
152
approach. Instead, managers should go with one or both the
following aims:
1. To investigate for themselves how well the company’s
hazard-management systems are working, in particular,
the major hazard-management systems. This means that
they will need to be briefed before they go into the field
on just how the relevant hazards are being managed,
so that they can ask intelligent questions. Some ideas
about appropriate questions are developed in [17].
2. A second purpose of such walk-arounds is to draw on
the knowledge of front-line workers to identify early
warning signs about where things may be going wrong.
Accident investigations nearly always discover that
there was information available to front-line workers
which, had it been acted upon, would have averted
the accident. One of the aims of senior management
walk-arounds should be to identify this early-warning
information. Sometimes the communication from
the front line may be difficult to decipher, consisting
of dark mutterings about management hypocrisy. But
such perceptions are worth investigating for the kernel
n
of truth that they usually contain.
io
Focus on personal safety
st y
ut
di op
Like all oil and gas companies, BP put a lot of time and
rib
effort into safety and yet the Deepwater Horizon was an
or c
appalling outcome. The details of the various decisions
made as described so far seem to indicate a disregard for
t f ple
safety. It is worth considering how BP as an organization
thought about safety before this incident for some clues
as to why this was the case.
no m
In one sense, those responsible for work on the Deepwater
Sa
Horizon were very safety conscious. The managers on the
rig at the time of the blowout were there partly to present
an award to the crew for a very creditable seven years’
operation without a lost-time injury. The rig was a standout
performer within Transocean and BP. More broadly, BP
overall claims a serious commitment to safety and before
the accident reported that “since 1999, injury rates and
spoils have reduced by 75%” [11, p.218]. But there is an
important distinction to be made between personal and
process safety (or system integrity). Process safety is about
keeping the process under control or, more colloquially,
keeping the hazardous substances in the pipes, or otherwise
contained. Failure to do so can lead to catastrophic
consequences. Personal safety, on the other hand, concerns
hazards that are not directly related to the process itself,
but which can injure or even kill people, such as falling
from a height. BP had an admirable focus on personal
hazards but an entirely inadequate appreciation of process
hazards. This problem was identified in the inquiries into
the BP Texas City Refinery disaster in 2005. Five years
later, BP still had not properly addressed this issue. For
example, the management walk-around that occurred on
the Deepwater Horizon in the hours before the explosion
The Journal of Pipeline Engineering
ignored process hazards entirely and focused exclusively
on personal safety.
Many companies in hazardous industries continue to fall
into this trap, and companies in the pipeline industry
need to exercise great care lest they too make the mistake
that BP made.
Remuneration systems
Remuneration systems, as embodied in the performance
agreements of top managers, serve powerfully to focus
everyone’s attention on key drivers, typically cost
minimization and personal safety. If there is nothing in
the system of pay to focus attention on the management
of process safety, then this is likely to be done very poorly,
if at all. Macondo engineers facing performance reviews
went to great lengths to demonstrate to their superiors
how much money they had saved, but there was no
corresponding pressure to demonstrate the contribution
they had made to managing system integrity.
Reports following the Texas City Refinery disaster also
drew attention to this issue. The Baker panel concluded
that: “a significant proportion of total compensation
of refining line managers and supervisors [should be]
contingent on satisfactorily meeting process safety
performance goals…” [18, p.251]. As one observer
remarked: “Managers did not act to prevent Texas City
because every incentive and potential penalty they faced
told them not to” [19, p.85].
All major-hazard industries, including the pipeline industry,
should examine their remuneration systems carefully to
determine if they are providing people with incentives to
focus on cost at the expense of management of system
integrity.
Conclusions
Unfortunately, the Deepwater Horizon disaster was not
the first serious safety incident in BP’s recent history. In
2005, a fire and explosion at their Texas City Refinery killed
15 people [20]. Regrettably, some of the organizational
weaknesses seen at Texas City in 2005 were to appear
again in the Gulf of Mexico in 2010. For this reason, it is
tempting to conclude that BP is a poor operator and by
implication that other large organizations have little to learn
from BP’s dreadful recent history, but this is to miss the
fundamental lesson of each of these incidents. Complex
accidents occur for complex reasons – combinations of
minor problems and failures that have accumulated. Many
of the minor failures occur for the most banal reasons –
warning bells can be very difficult to hear.
This paper has highlighted several areas of organizational
failure that were seen in the circumstances surrounding
the Deepwater Horizon blowout. These provide some key
3rd Quarter, 2012
lessons for places to look for organizational improvements.
High-performing organizations:
• are good at identifying cost/safety tradeoffs and
carefully consider the right choice in each case;
• are sceptical of good news, preferring ‘bad’ news for
the opportunities it brings to improve;
• value the professionalism of technical staff and
support development of expertise;
• have effective relationships with their contractors,
exerting appropriate control and listening to and
acting on contractor expertise, where appropriate;
• have senior managers who can exert effective control
on technical integrity issues;
• understand the difference between personal safety
and technical integrity, and manage both;
• have remuneration systems that support these
organizational priorities, including system integrity.
These organizational vulnerabilities are all things that
the pipeline industry is aware of to some extent, but the
dynamic nature of our industry and the environment in
n
which we operate, plus events such as the San Bruno disaster,
emphasize the need for a continuing focus on improvement
io
in these areas.
st y
ut
di op
In Australia, Research Programme 4 of the Energy Pipelines
rib
Cooperative Research Centre regarding public safety
or c
and security of supply is taking up several of the issues
mentioned above. There are many opportunities available
t f ple
for organizations to participate in this work as detailed on
the EPCRC website (see https://www.epcrc.com.au) and
we welcome further discussion in these important areas.
no m
Acknowledgements
Sa
This work is partially funded by the Energy Pipelines
CRC, supported through the Australian Government’s
Cooperative Research Centres Programme.
The cash and in-kind support from the APIA RSC is
gratefully acknowledged.
References
1. M.Bourrier, 2011. The legacy of the high reliability
organization project. Journal of Contingencies and
Crisis Management, 19, 1.
2. E.Roe and P.R.Schulman, 2008. High reliability
management: operating on the edge. Stanford
University Press, Stanford.
153
3. K.W.Weick and K.M.Sutcliffe, 2001. Managing the
unexpected: assuring high performance in an Age of
complexity. Jossey-Bass, San Francisco.
4. D.Izon, E.P.Danenberger, and M.Mayes, 2007. Absence
of fatalities in blowouts encouraging in MMS study of
OCS incidents 1992-2006, Drilling Contractor.
5. D.Borthwick, 2010. Report of the Montara Commission
of Inquiry.
6. J.Hayes, 2012. Operator competence and capacity –
lessons from the Montara blowout. Safety Science,
50, pp563–574.
7. J.Reason, 1997. Managing the risks of organizational
accidents. Ashgate, Aldershot.
8. Idem, 2000. Human error: models and management.
British Medical Journal, 320, pp768-770.
9. NTSB, 2011. Pipeline accident report: Pacific Gas and
Electric Company natural gas transmission pipeline
rupture and fire, San Bruno, CA, September 9, 2010.
National Transportation Safety Board, Washington.
10. A.Hopkins, 2012. Disastrous decisions: the human and
organisational causes of the Gulf of Mexico blowout.
CCH, Sydney.
11. National Commission on the BP Deepwater Horizon
Oil Spill and Offshore Drilling, 2011. Deep water: the
Gulf oil disaster and the future of offshore drilling.
Report to the President.
12. D.Vaughan, 1996. The Challenger launch decision:
risky technology, culture and deviance at NASA.
University of Chicago Press, Chicago.
13. G.Klein, 1998. Sources of power: how people make
decisions. MIT Press, Cambridge, Massachusetts.
14. Idem, 2003. The power of intuition: how to use your
gut feelings to make better decisions at work. Currency/
Doubleday, New York.
15. R.Middlehurst and T.Kennie, 1997. Leading
professionals: towards new concepts of professionalism.
In: The end of the professions? The restructuring of
professional work. Eds: J.Broadbent, M.Dietrich, and
J.Roberts. Routledge, London.
16. J.Hayes, 2012. Use of safety barriers in operational
safety decision making. Safety Science, 50, pp424–432.
17. A.Hopkins, 2010. Management walk-arounds: lessons
from the Gulf of Mexico oil well blowout. National
Research Centre for OHS Regulation, Canberra.
18. J.Baker, 2007. The report of the BP US refineries
independent safety review panel.
19. T.Bergin, 2011. Spills and spin: the inside story of BP.
Random House, London.
20. A.Hopkins, 2008. Failure to learn: the BP Texas City
refinery disaster. CCH, Sydney.
INTERNATIONAL PIPELINE CONFERENCE & EXPOSITION
September 24 – 28, 2012 • Calgary, Alberta, Canada

MEETING THE CHALLENGES

The international pipeline industry will come together for the

International Pipeline Conference & Exposition 2012 to share
leading edge information which provides the industry with
the tools to meet the evolving challenges of pipelining in

n
today’s world.

io
st y
ut
di op
rib
or c
t f ple

THE CONFERENCE THE EXPOSITION

September 24 - 28, 2012 September 25 - 27, 2012
Hyatt Regency Hotel & Calgary TELUS
no m

TELUS Convention Centre Convention Centre

• 13 Technical Tracks with over • Over 150 exhibiting companies
Sa

300 quality papers • Displays of innovations in

• Highly Informative Tutorials
• International Presenters and
Keynote Speakers
• Panel Sessions/Poster Sessions
• Student Paper Competition and
Best Paper Award
• Networking Opportunity
Luncheons & Receptions
pipeline technology
• Over 3,500 visitors from
40 countries
• IPE Industry Reception
• Complimentary entry for
Conference delegates and
industry professionals

Media Sponsor Major Exhibition Sponsors

For more information and updates visit our websites:

Conference: www.InternationalPipelineConference.com @IPC_Calgary
Exposition: www.internationalpipelineexposition.com @petroleumshow
3rd Quarter, 2012 155

National Grid’s COOLTRANS

research programme
by Russell Cooper

National Grid, Warwick, UK

N ational Grid is progressing the £8-million COOLTRANS (CO2Liquid pipelineTRANSportation) research

programme to address and resolve the key issues relating to the safe routeing, design and construction
of onshore pipelines for the transportation of anthropogenic, high-pressure, dense-phase, carbon dioxide
(CO2) from power stations and other industrial emitters to offshore locations for underground storage.

In constructing the COOLTRANS research programme, National Grid has brought together major technical
capability in UK academic and industry organizations, and is co-ordinating research studies carried out by
these organizations in order to provide a fast track for the identification and application of key learning
to pipeline projects. An overview of the COOLTRANS research programme was given at the 2011 CO2

n
Transportation Forum [1], which explained the integrated analysis strategy combining state-of-the art
numerical modelling of pipeline decompression, near- and far-field-dispersion studies being conducted by

io
three university groups, and the use of full-scale experimental tests carried out at GL Noble Denton’s
st y
Spadeadam test site.
ut
di op
rib
The COOLTRANS programme is over 50% complete, and has generated significant knowledge from the
or c

results of the advanced analyses and the large-scale experiments.

t f ple

This paper provides a detailed description of the research programme, the strategy, organization, programme,
and progress. The extensive experimental programme is detailed. The programme of case studies, which is
designed to confirm that the independently conducted theoretical studies can be efficiently integrated to
no m

allow validation by experimental data, is described. The key learning obtained to date from the research
programme and the application of this to pipeline project studies is described.The key differences between
Sa

transporting dense-phase CO2 and other products are highlighted. The challenges posed by the transport
of dense-phase CO2 by pipeline compared with the conventional design of natural and rich gas, and liquid,
pipelines are detailed.

The COOLTRANS The COOLTRANS programme is now over 50% complete,

research programme
The purpose of the £8-million COOLTRANS research
programme is to establish and demonstrate the requirements
for the safe design, construction, operation, and maintenance
of dense-phase CO 2 pipelines. This will allow the
development of a comprehensive safety justification for
the onshore pipeline transportation of dense-phase CO2.
An overview of the COOLTRANS workstreams is given in
Fig.1; the workstreams are interlinked and are underpinned
by large-scale experimental studies.
This paper was presented at the Third International Forum on the Transportation of
CO2 by Pipeline, organized by Newcastle University, Tiratsoo Technical, and Clarion
Technical Conferences, and held in Newcastle in June, 2012.
Author’s contact details:
email: russell.d.cooper@uk.ngrid.com
and it is now appropriate to summarize the history of the
programme, review progress, and consider the learning to
date and how this is being applied to project studies.
Background
In 2008, National Grid initiated a programme of work to
investigate the business potential for the transportation of
CO2 in the UK. In order to meet Government targets for
the capture, transportation, and storage of CO2 (CCS),
a £4-billion investment in CO2 pipelines will be required
by 2030 [2]. The pipelines will collect CO2 from emitters
(power stations, etc.) and transport it to a location (such
as saline aquifers, and depleted gas or oil fields) where it
can be safely stored.
156 The Journal of Pipeline Engineering

n
io
st y
ut
di op

Fig.1. COOLTRANS research programme workstreams.

rib
or c

Gap Technical work required

t f ple

Hazard category for gaseous and dense-phase Assess the thermodynamic properties of dense-phase CO2
anthropogenic CO2. and the consequences of a release.
no m

Area classification and design factor requirements for Develop risk-assessment methodology, risk-based routeing
gaseous and dense-phase CO2 pipelines guidelines.
Sa

Routeing guidelines Determine the requirements for area classification and

Assessment of block valve spacing
CO2 specification, including dryness
Flow assurance design
Fracture control requirements
Performance requirements for non-metallic materials
Management of internal corrosion
Table 1. CO2 pipelines: identified code gaps.
design factor for (thick-wall) dense-phase CO2 pipelines.
Determine the effect of impurities (including water) on
safety, pipe integrity, and hydraulic performance.
Assess the impact of topology on pipeline maximum
allowable operating pressure (MAOP).
Assess potential for phase change.
Assess impact of impurities on hydraulic performance.
Assess the decompression behaviour of dense-phase CO2
and CO2 with impurities, and determine the material
properties required to control the behaviour of damage
when subjected to fluid-decompression forces.
Assess the impact of warm pre-stressing and the effect of
Joule Thompson cooling on brittle initiation on defect
behaviour.
Assessment and specification of elastomer composition to
avoid loss of elasticity and explosive decomposition.
Establish corrosion rates in wet CO2 environments
required dryness specification to reduce the risk of
internal corrosion.
3rd Quarter, 2012
The current pipeline codes used in the UK are not directly
applicable to the transportation of CO2 without additional
development. In addition, the UK is a densely populated
country in which land use requires careful consideration
and management. The CO2 to be transported will contain
impurities due to combustion or chemical processing which
will affect the safety, integrity, and hydraulic performance
of pipelines used for CO2 transportation.
Existing world-wide operating experience does not directly
apply. The North American experience of safe CO2 transport
is mainly associated with enhanced-oil recovery (EOR) and
typically applies to pure, naturally extracted CO2. The US
experience also relates to the routeing of pipelines through
remote areas. Consequently the experience, while relevant,
cannot be directly applied in the UK without further
development.
Furthermore, the existing National Grid experience and
culture relates to the safe operation of natural gas pipelines
where the consequences of an unplanned release are ignition
of a flammable cloud and the resulting thermal radiation.
n
For CO2 the consequences are asphyxiation and toxicity,
so it is essential the effect of these different consequences
io
on public safety are fully understood.
st y
ut
di op
National Grid is currently pursuing plans to transport CO2
rib
through pipelines from the proposed Don Valley power
or c
station near Doncaster to a saline aquifer off the Yorkshire
coast. There is also an aspiration to develop the first pipeline
t f ple
into a network configuration that links-up multiple CO2
emitters in the Yorkshire and Humberside areas. The
planned developments are supported by European Union
no m
grants which have been used to partly fund this research.
Sa
The need for a comprehensive research programme to
support these planned developments was therefore identified
and initiated. The research strategy developed by National
Grid addresses technical gaps associated with gaseous phase
and dense-phase CO2 pipelines.
Strategy
To address the technical gaps, two research and development
(R&D) programmes (gaseous and dense-phase CO2) were
specified. At the heart of the R&D is the knowledge acquired
in the 1970s and 1980s for the safe operation of natural- and
rich-gas pipelines.
The existing knowledge is being extended to CO2 by multi-
disciplinary programmes involving leading UK universities
with expertise in the relevant areas and by expert industrial
companies with specialized knowledge and/or proven track
record and/or proven full-scale test facilities and experience.
Due cognisance is taken of all other relevant CO2 research
and studies being conducted world-wide, and National
Grid is supporting/contributing to the programmes
which complement its research. The National Grid R&D
157
programmes are directly managed and there is a strong
team ethos to ensure delivery within the timescales critical
to National Grid’s interests.
In developing the requirements for the R&D programmes
for CO2 pipelines, National Grid reviewed the work carried
out in the 1970s and 1980s to establish the code rules and
safety requirements for natural-gas and rich-gas mixtures,
and extracted the appropriate learning and relevant data
from this work. In addition, the requirements of the pipeline
codes were reviewed to identify gaps in relation to CO2 [3].
The two primary UK pipeline standards, PD 8010-1 [4] and
IGEM/TD/1 [5], were reviewed in detail, and a number of
gaps were identified: these are summarized in Table 1. The
assessment of these technical gaps in the pipeline codes
for CO2 pipelines was used as a checklist in defining the
requirements for the R&D.
An R&D programme to investigate the requirements for
the transportation of gaseous-phase CO2 has already been
completed by National Grid to facilitate the change of use
of existing pipelines which are no longer required for the
transportation of natural gas. Existing knowledge and tools,
including knowledge gained from the gaseous-phase R&D
programme, have been applied and extended to dense-phase
CO2 through the use of advanced modelling and analysis
– supported by experimental work where necessary –
to dense-phase CO2.
Links with other CO2 research programmes
In developing the strategy for R&D into the transportation
of CO2 by pipelines, National Grid has taken full account
of current and planned programmes relating to CO2
transportation. National Grid is on the Advisory Boards for
the MATTRAN research programme1 funded by EPSRC2 and
E.ON, and the CO2PIPEHAZ research programme3, funded
under the European 7th Framework. National Grid is a
member of the joint-funded research programmes COSHER4
(a Statoil/Gasunie initiative to establish a collaboration of
European stakeholders to carry out large-scale CO2 release
experiments and measurement programmes to obtain data
that can be used by the partners to improve and validate
safety models for CO2 pipelines), and SARCO25 (a research
programme being progressed by Centro Sviluppo Materiali
(CSM) on behalf of the European Pipeline Research Group
(EPRG), using the European Research Fund for Coal and
Steel and support by industry partners). National Grid has
also supported a Pipeline Research Council International
(PRCI) project to conduct shock-tube tests on dense-phase
1. Material for Next Generation CO2 Pipeline Transport Systems, involving Newcastle
University, Nottingham University, University College London (UCL), and Leeds
and Cranfield Universities.
2. Engineering and Physical Sciences Research Council.
3. Quantitative Failure Consequence Hazard Assessment for Next Generation CO2
Pipelines, involving UCL, Leeds University, the UK Health and Safety Executive
(HSE), National Research Centre for Physical Sciences (Greece), Dalian University
(China), INERIS (France), and GEXCON (Norway).
4. Carbon Dioxide, Safety, Health, Environment and Risks.
5. Requirements for Safe And Reliable CO2 Transportation by Pipeline.
158 The Journal of Pipeline Engineering

n
io
st y
Fig.2. COOLTRANS management structure.
ut
di op
rib
or c
t f ple
no m
Sa

Fig.3. COOLTRANS experiments at the GLND Spadeadam test site.

3rd Quarter, 2012
CO2 at the TransCanada’s gas dynamic test facility in
Didsbury, Alberta, Canada, and is supporting projects
considering leak prevention in CO2 pipeline valves and pig
trap launchers by correct seal material selection and the
technology challenges for liquid CO2 pumping stations.
In addition to support of the PRCI project which addresses
USA interests, enquiries were made regarding any other
research studies being undertaken in the USA. However
at the time of enquiries (2009), no opportunities to share
interests in technical studies were identified. National
Grid has investigated USA operational experience relating
to the transportation of CO2, and confirmed that this
experience predominantly relates to the transport of
dense-phase CO2 in relatively small-diameter pipe at
high pressures (typically 250barg) for EOR, for which
the pipelines are mainly routed through remote areas.
There is currently limited experience in the USA relating
to the transportation of anthropogenic CO2 containing
the impurities resulting from combustion or chemical
processing for carbon capture.
n
Management and organization of COOLTRANS
io
The large, multi-disciplinary programme required to meet
st y
ut
National Grid’s needs and timescale involves the definition
di op
of a number of interrelated work packages. These work
rib
packages are being undertaken by universities which have
or c
or are developing expertise in the relevant areas, and expert
industrial companies which have specialist knowledge and/
t f ple
or a recognized track records in theoretical and empirical
research. Management of this programme to ensure
delivery and exchange of appropriate information between
no m
the various expert organizations and delivery of research
findings that can be quickly applied in pipeline projects
Sa
is critical in ensuring that National Grid’s programme
for the development of dense-phase CO2 pipelines is met.
The organization established by National Grid to direct
and manage the COOLTRANS research programme
is shown in Fig.2. This organization brings together
National Grid’s design expertise with leading experts
comprising the National Grid CO2 Technical Group, which
directs and progresses the research work and technical
studies, confirms work completion, and recommends its
application to pipeline projects. The Technical Group is
multi-disciplinary, and the experience and capability of this
group includes knowledge, expertise, and involvement in:
• the major research programme undertaken
by British Gas in the 1970s and 1980s for the
development of the UK natural gas pipeline
network;
• development of pipeline code rules;
• academic research into carbon capture and
transportation of CO2;
• pipeline design, construction, and operation;
• fracture control;
159
• dispersion analysis and measurement, and
consequence analysis;
• experimental design and execution;
• quantified risk assessment (QRA) and integrity
analysis;
• extensive national and international pipeline
integrity experience.
In addition to the above, the Technical Group supports
National Grid in interpreting the requirements of the
UK legislative safety framework.
The capability of the Technical Group and its close
working relationship with National Grid has allowed the
focusing of research and technical studies and the early
application of research results as these become available
in pre-FEED pipeline studies.
The COOLTRANS workstream and the research
organizations and lead researchers involved in the delivery
of the work are summarized in Table 2.
Scope of the COOLTRANS research programme
As shown in Fig.1, the COOLTRANS research programme
comprises six workstreams, which are described briefly
below.
1. Thermodynamic and flow characteristics of dense-phase CO2
This workstream involves the investigation, modelling,
and analysis of the thermodynamic and flow behaviour of
dense-phase CO2 including the effect of impurities. This
is of particular significance in understanding the phase
and dispersion behaviour of the fluid in the event of a
controlled or uncontrolled release, and understanding
how the energy released by the decompressing fluid
will provide a driving force which could cause defects
to propagate. As the fluid is toxic and an asphyxiant in
large concentrations, it presents a hazard to people over
the range that it disperses. The aspects which require
research are the phase behaviour of CO2 and CO2 with
impurities, decompression and outflow, and near- and
far-field dispersion.
2. Fracture control
Fracture control is an integral part of the design of a
pipeline and is required to minimize the likelihood of
failures (fracture-initiation control) and to prevent long-
running brittle or ductile fractures from occurring in a
pipeline (fracture-propagation control).
Fracture control requires a fundamental understanding of
the material properties and the behaviour of defects which
are subject to the forces generated by the decompression
of the fluid being released from the pipeline. In simple
terms, the code requirements for fracture control are to
160 The Journal of Pipeline Engineering

Experimental Organization and leading

Work stream Theoretical investigations
investigations researcher
Measurement of the
phase behaviour
of CO2 and CO2 + Nottingham University,
impurities Dr T Drage
Assessment of equations of state

Outflow and decompression UCL,

modelling
1.
Thermodynamic
Near-field source modelling
characteristics of
dense-phase CO2
Far-field dispersion modelling
Computational fluid dynamics
(CFD) and pragmatic modelling
of CO2 releases and the
assessment of model performance
Pragmatic modelling of CO2
dispersion
Prof H Mahgerefteh
Measurement of
Leeds University,
droplet and particle
Prof M Fairweather
sizes in CO2 release
Kingston University,
Prof J Wen
HSL,
Dr S Gant
GLND,
Dr P Cleaver, Ms A Halford

n
io
Source model including release GLND,
st y
-crater geometry
ut Dr P Cleaver, Ms A Halford
di op

Puncture and rupture GLND (Spadeadam),

tests K Armstrong
rib
Requirements for pipeline QRA,
or c

including failure frequency

modelling, simple dispersion
t f ple

2. Quantified risk
analysis modelling and harm criteria Newcastle University,
Assessment of simple dispersion Dr J Race
models and harm criteria
no m

Requirements for QRA

Sa

GLND,
Demonstration QRA
Dr P Cleaver, Ms A Halford
Applicability of existing
Newcastle University,
fracture-mechanics’ models to
Dr J Race
thick-wall pipes
GLND (Spadeadam),
Shock-tube tests
K Armstrong
Assessment of pipe toughness
PIE and Atkins,
requirements and determination
D G Jones and Dr A Cosham
of full-scale test strategy
Instrumented
3. Fracture control burst/’West Jefferson’
(vessel) type tests GLND (Spadeadam),
K Armstrong
Full-scale crack-
propagation tests
Atkins, PIE and Penspen,
Warm pre-stressing and brittle
Dr A Cosham, D G Jones, Prof P
initiation studies
Hopkins
GLND (Spadeadam),
Cold gas-release tests
K Armstrong
3rd Quarter, 2012 161

Experimental Organization and leading

Work stream Theoretical investigations
investigations researcher
Newcastle University,
Hydraulic studies
Dr J Race
GLND,
Venting analysis
Dr P Cleaver, Ms A Halford
GLND (Spadeadam),
Venting tests
K Armstrong
UCL,
ESDV studies
4. Design and Prof H Mahgerefteh
integrity GLND,
Block-valve spacing studies
Dr P Cleaver, Ms A Halford
Assessment of welding MACAW Engineering,
requirements Dr K Prosser, Dr P Boothby
Welding thermal GLND (Spadeadam),
decay tests K Armstrong
Internal corrosion PIE and Newcastle University,
testing G Senior, Dr J Race
Investigation of the
5. Environmental Nottingham University,

n
effects of leaks on
and social studies Dr B Lomax, Dr J Lake

io
plants
st y
ut Tyndall Centre at Manchester
Public perception of risk studies University,
di op

Dr C Gough
rib
Table 2. Summary of the COOLTRANS workstreams and deliverables.
or c
t f ple

ensure the material toughness is adequate to prevent are used to carry out the risk assessments required to
initiation from small flaws resulting from material manage operations and reach agreement with regulatory
manufacturing process and fabrication, and to arrest a authorities. National Grid has developed an equivalent
no m

propagating ductile fracture. strategy for gaseous-phase CO2 pipelines and is currently
developing a methodology for dense-phase CO2 pipelines.
Sa

The models used to predict the initiation and speed of crack
propagation and fluid decompression are semi-empirical,
so their application to dense-phase CO2 pipelines needs
to be confirmed and high-quality empirical data relating
to the operating conditions are required. In addition, the
effect of warm pre-stressing on material performance and
the potential for brittle initiation due to Joule-Thompson
cooling of the decompressing fluid must be considered.
A number of large-scale experimental tests are therefore
needed to provide the relevant information.
3. Quantified risk assessment (QRA)
QRA is required to establish a routeing corridor based on
individual risk, in accordance with the British Standards
Institute (BSI) code of practice, PD 8010-1:2004 [4], and to
confirm that the risks associated with a CO2 pipeline are
‘as low as reasonably practicable’ (ALARP) in accordance
with the Pipelines Safety Regulations 1996 [6].
National Grid has extensive experience in the application
of pipeline QRAs for the safe management of natural gas
pipelines. Models and methods based on best practice
This workstream carried out under COOLTRANS will
produce a public domain description of a methodology
for dense-phase CO2 pipelines based on the results from
this overall programme. The work involves the assessment
of failure-frequency modelling, consequence modelling,
and harm criteria on the individual and societal risks
posed by CO2 pipelines.
4. Pipeline design and integrity
This workstream involves the development of the technical
basis for the standards, specifications, and operational
requirements essential to the safe design and operation and
integrity management of CO2 pipelines. The workstream is
therefore extensive, but the primary issues are the requirements
for material specifications (specifically, toughness requirements),
pipeline routeing requirements, design wall thickness
requirements for class 1 and class 2 locations6, management and
control of internal corrosion, welding requirements (including
under-pressure welding on live pipelines), block-valve spacing,
and the requirements for safe venting.
6. Class 1 locations have a population density of ≤ 2.5 persons per hectare, class 2
locations have a population density > 2.5 persons per hectare.
162 The Journal of Pipeline Engineering

Test Purpose Number of tests Status

To obtain decompression data for pure CO2

Shock-tube tests and CO2 containing impurities and to obtain 33 Complete
dispersion data where practicable.
To provide dispersion data for venting through
Vent tests 13-mm, 25-mm, 50-mm, and 75-mm diameter 14 13 complete 1 scheduled
vents at heights 3m and 5m above ground level.
To provide detailed visual evidence of the
behaviour of realistic releases from a buried
pipeline by simulating external interference and
Puncture tests 8 Complete
corrosion.
To obtain measurements of outflow and
dispersion where practicable.
To measure local pipe-wall cooling around leaks
of different geometries (holes, slot, crack).
Cold gas release 3 complete
To validate computational fluid dynamics (CFD) 4
tests 1 scheduled
studies and provide input for the demonstration
test.
To investigate the potential to generate a long-
Instrumented
running shear fracture in the full-scale crack- 3 Complete
burst tests

n
propagation test.

io
To demonstrate crack arrest in 914-mm outside
st y
diameter, 25.4-mm wall thickness, L450 grade
Full-scale crack- ut 1 complete
di op

pipe with a dense-phase CO2 mixture. 2

propagation test 1 scheduled
To validate the methodology for estimating
rib
crack-arrest toughness.
or c

To carry out temperature decay tests during

underpressure welding operations on pipe
t f ple

Underpressure
containing circulating dense-phase CO2. 28 Scheduled
welding trials
To carry out a full-scale under-pressure
connection welding trial.
no m

To provide visual evidence of the behaviour

of realistic releases (ruptures) from a buried
Sa

Rupture tests pipeline. 4 Scheduled

To obtain measurements of the outflow and
dispersion where practicable.
To demonstrate whether rupture (continuous
1
brittle initiation) can occur from a leak due to
Demonstration localized cooling.
Scheduled
test To conduct three associated leak/rupture
experiments to allow correct interpretation of
3
the results of the demonstration test.

Table 3. Summary of the COOLTRANS experimental programme.

5. Environmental and social impact studies
Releases from dense-phase CO2 pipelines may have an
effect on the environment local to the leak, which needs
to be understood and evaluated both in terms of the
physical effect on the soil caused by the pressure and
temperature of the fluid, and the effect on local flora
and fauna. A release of CO 2 may occur as a leak from
a small hole in a buried pipeline or as a major release
due to a pipeline rupture.
A programme of work to investigate the effects of CO2 leaks on
plants is being undertaken by the University of Nottingham
at its ASGARD (artificial soil gassing and response detection)
facility. This facility was established in 2006 to simulate
the effects of elevated soil CO2 concentrations caused by
a leak of CO2 from an underground source. This work is
investigating the effects of (i) small, long-term leaks through
the soil on crop growth, (ii) large releases leading to local
flooding of the ground surface with CO2, and (iii) different
soil types on the responses observed in (i) and (ii).
3rd Quarter, 2012
The ASGARD facility is currently being used to carry
out research for the EU Framework 7 study, RISCS
(Research into Impacts and Safety of CO2 Storage). The
experiments planned for RISCS and COOLTRANS
programmes are complementary, and are being carried
out on a collaborative basis, in which the facilities and
data will be shared. The RISCS study provides the
experimental framework and involves the measurements
of soil parameters, including gas concentrations, plant
growth, root development, and a range of potential stress
responses. Additional investigations under COOLTRANS
include investigation of the stress response of plants to
elevated soil CO2 levels.
Planning applications for pipeline projects require
extensive public consultation. In order to ensure public
awareness and address any concerns, the public perception
of the hazards and risks posed by CO2 pipelines must be
understood so that accurate and relevant information can
be provided. Studies are being undertaken by the Tyndall
Centre at Manchester University to investigate these
factors. The research includes case studies to assess the
n
social impacts of existing and planned pipeline projects,
followed by a detailed assessment of the public perception
io
of CO2 transportation in order to inform the provision of
st y
ut
information to the public. This work will use and extend
di op
existing work on the public perception of CCS.
rib
or c
6. Application of research findings
t f ple
This workstream is currently being defined. The results
of the research work obtained in workstreams 1- 5 will be
used to develop technical specifications and procedures
no m
for application by National Grid for the design,
construction, operation, and maintenance of dense-phase
Sa
CO2 pipelines, and the development of a comprehensive
suite of documents based on those in place for natural
gas pipelines is planned. Work has commenced on the
development of the requirements for design to address
the technical gaps in the current pipeline codes, and these
are being developed as supplementary requirements to
PD 8010-1:2004 [4] for use by National Grid.
Summary of progress and implementation
COOLTRANS is a three-year R&D programme which
commenced in January, 2011, and will complete in
December, 2013, at a total cost of approximately
£8 million.
The COOLTRANS programme is over 50% complete,
and has already delivered results that challenge the
conventional pipeline design approach, and a summary
of the key learning to date is given later in this paper.
In progressing its interests in the development of a CO2
pipeline network, National Grid has used the results as they
become available to carry out pre-front end engineering
and design (FEED) studies for several ongoing projects.
163
Experimental test programme
A comprehensive programme of large-scale experimental
tests is being carried under COOLTRANS, in order to
provide validation and demonstration data, and its status
is summarized in Table 3. The test programme (Figs 3 and
4) is being carried out by GL Nobel Denton (GLND) at its
Spadeadam test site in the UK.
Use of case studies to integrate
theoretical and experimental
dispersion studies
One of the primary deliverables from the COOLTRANS
research programme is the development of a robust and
practical methodology for pipeline QRA. QRA is required
as part of the approach for the design and routeing of
dense-phase CO2 pipelines and can be used to help address
any gaps in existing codes. The decompression, outflow,
near-field expansion, and far-field dispersion of CO2 is
complex, and new models are required to accurately describe
the behaviour.
The theoretical studies by UCL, the University of Leeds,
and Kingston University are being carried out to provide
state-of-the-art models for the outflow, near-field dispersion
behaviour, and far-field dispersion behaviour, associated
with releases from below-ground CO2 pipelines that are
ruptured or punctured. This will involve the universities
using their specialist software and different modelling
approaches in the different stages of the release. In order
to obtain a comprehensive, state-of-the-art solution for the
dispersion of a release of dense-phase CO2 from a pipeline
puncture or rupture, the individual university studies
addressing outflow, near-field, and far-field dispersion
must be integrated.
The integration of the three stages of analysis is required
in order to understand how the predictions of each stage
relate to the subsequent and previous stage, and how
assumptions and input data used in one stage may influence
the predictions from the next stage. The approach adopted
in COOLTRANS is to use a series of case studies designed
to confirm that the independently conducted theoretical
studies can be efficiently integrated to allow validation
by experimental data. This approach allows some direct
comparison of the models and also allows the effect of the
differences in factors such as the equation-of-state (EOS),
input assumptions, and solution methods to be assessed.
This level of integration of the stages of analysis required in
the COOLTRANS programme is innovative, and requires
careful management and full collaboration between the
parties involved.
Programme of case studies
A programme of case studies of increasing complexity is
being carried out during the COOLTRANS programme.
164 The Journal of Pipeline Engineering

n
io
st y
ut
di op
rib
or c
t f ple
no m
Sa

Fig.4. COOLTRANS experiments at GLND Spadeadam.

3rd Quarter, 2012 165

Case study Description Purpose

Vertical vent, steady-

state 25-mm diameter and To predict the conditions at the vent exit and flow rates to compare
1
variable flow rate 50-mm with measured values.
diameter
Horizontal discharge from a
2 Predict the time dependent flow rate.
shock-tube test rig
Below-ground pipeline
3 Predict flow rates and compare with measured values.
puncture

5 Below-ground rupture As above

6 Below-ground rupture As above

Table 4. COOLTRANS integrated dispersion case studies.

The case studies are defined on completed experiments independent review of the capabilities and limitations
so that data will be available for model comparison. In of the various models used for predicting discharge and
this way, the universities will gain information throughout dispersion of CO2 releases from punctures and ruptures

n
the programme that may prove useful in developing the resulting from pipeline failure. The results will be of
models. By proposing case studies in order of increasing benefit to pipeline operators in selecting appropriate

io
complexity, the intention is to minimize the risks inherent consequence-model tools, and to regulatory authorities
st y
ut
in the parallel developments planned for the models. The in assessing pipeline QRAs.
di op

programme of case studies is summarized in Table 4.

CO2 research: key learning
rib
or c

Case study 1 [1] has been completed and reviewed, case

study 2 is in its final stages, and work is currently progressing As stated above, the COOLTRANS programme is over 50%
t f ple

on case study 3. complete. The key learning to date is summarized below.

Model-evaluation protocol CO2 entry specification

no m

The Health and Safety Laboratory (HSL), participating A pipeline transporting CO2 must be designed to take
Sa

in the COOLTRANS research programme on behalf of
the Health and Safety Executive (HSE), is developing a
model-evaluation protocol (MEP) for CO2 discharge and
dispersion models for application to the models developed
and used in COOLTRANS. The purpose of the MEP,
described by Gant in [7]7, is to provide a comprehensive
evaluation methodology for determining the suitability of
models for simulating releases of dense-phase CO2 from
pipelines.
account of the phase behaviour of the fluid, and the effect
of any impurities in the CO2 stream resulting from the
capture process. The pressure-temperature phase diagram
for CO2 is shown in Fig.5: CO2 can exist as a vapour (gas)
or liquid within the typical operating temperature range of
a pipeline. CO2 can be transported at high pressures as a
liquid or a ‘dense-phase’ fluid, or at low pressures as a gas.
In contrast, natural gas exists as a gas within the typical
operating temperature range of a pipeline.

The MEP is designed to consider release types, fluid type
(gas, liquid, supercritical), and situations/environments,
with the objective of confirming the accuracy of predictions
for hazardous clouds. The process will include a review of
the capability of the models, and comparisons with data
based on the case studies will provide an independent
assessment of model performance against experimental
data. In this respect, modelling of a CO2 release from
vessels, short pipe, and pipeline depressurization will be
considered, to assess model sensitivity and parameter trend
predictions. The application of the MEP will provide an
7. Published in this issue on pages 201–211.
The addition of other components (‘impurities’) changes
the thermodynamic properties of the fluid. Figure 5
also shows the pressure-temperature phase diagram for
a binary mixture of CO2 together with hydrogen (H2),
hydrogen sulphide (H2S), and nitrogen dioxide (NO2).
The introduction of these impurities significantly extends
the two-phase region.
A pipeline can be designed to transport a wide range of
different CO2-rich mixtures. There may, however, be a cost
penalty associated with transporting higher concentrations
of impurities, because the impurities have an effect on the
design and operation of the pipeline.
166 The Journal of Pipeline Engineering

100

90

80

70

60
Pressure /bar

50

40

30 Pure Carbon Dioxide

5% Hydrogen
20
5% Hydrogen Sulphide
10 5% Nitrogen Dioxide

0
0 5 10 15 20 25 30 35 40 45 50
Temperature /degC

n
io
Fig.5.The pressure-temperature phase diagram for CO2 and a binary mixture.
st y
ut
di op
rib
pi = 100 barg, Ti = 40°C
or c
t f ple
no m
saturation pressure, barg

Sa

80
0.5
1 mol.%

75

H2 N2 Ar CH4

Fig.6.The effect of composition on the saturation pressure.

3rd Quarter, 2012 167

22 25
610 mm, L450
21 24
20 23
f = 0.72
wall thickness, mm

wall thickness, mm
19
22
18 ps
21
17 = 85 barg
20
16 = 80 barg
19
15
18 = 75 barg
14
13 17 = 70 barg

12 16
50 60 70 80 90 100 150 200 250
saturation pressure, barg design pressure, barg

Fig.7.The relationship between saturation pressure and wall Fig.8.The relationship between design pressure, saturation
thickness (indicative). pressure, and wall thickness (indicative).
of dense-phase CO2 are such that the nominal wall thickness
600 25
610 mm, L450 may, in fact, be determined by the requirements for fracture

n
control. Consequently, thicker pipe may be required for
500
dense-phase CO2 pipelines, as detailed below.

io
20
st y
defect length, mm

defect depth, mm

400 ut Pipelines transporting gaseous fluids, two-phase fluids,

di op

dense-phase fluids, or liquids with a high vapour pressure

300 15
rib
are susceptible to long-running (propagating) ductile (shear)
or c

fractures. CO2 in the liquid or dense-phase is a high vapour

200 pressure liquid. Pipelines designed to transport CO2 in
t f ple

10 the dense-phase are susceptible to running-ductile fractures

100 [9-11]. The prevention of running fractures (fracture
control) is a design requirement in the pipeline code PD
no m

0 5 8010-1:2004 [4].
10 15 20 25 30
wall thickness, mm
Sa

The toughness required to arrest a running-ductile fracture

Fig.9.The size of part-wall defect that is predicted to fail as
a rupture in 610-mm (24-in) diameter pipe.
A specification needs to be used to define the quality
requirements to limit the range of compositions that may
be transported in a pipeline, and this must take account
of safety (toxicity), impact on pipe integrity (corrosion
and corrosion induced cracking), hydraulic efficiency,
and saturation pressure. Care is needed when using other
published specifications; a number do not take account of
all these aspects. This is discussed in [8] by Race8.
Effect of saturation pressure on wall thickness
The wall thickness of a typical onshore pipeline is
determined by the limit on the design factor (for example,
0.72), and the population density along the proposed route
of the pipeline. However, the decompression characteristics
8. Published in this issue on pages 173–190.
in a pipeline transporting CO2 in the dense-phase depends
upon the geometry and grade of the pipeline material, and
the ‘saturation pressure’ of the CO2-rich mixture [9, 10].
The implication is that the geometry, material grade, and
toughness of a pipeline define a limit on the saturation
pressure. CO2-rich mixtures with a higher saturation
pressure cannot therefore be transported in the pipeline.
The saturation pressure is defined by the composition of
the mixture and the initial pressure and temperature of
the fluid. The addition of other components increases the
saturation pressure. Different components have different
effects; for example, hydrogen has a greater effect than
methane. Consequently, saturation pressure cannot be
defined in terms of a maximum level (for instance, 2
mol.%) of impurities, and this is illustrated in Fig.6: a
mixture that contains 2 mol.% hydrogen (H2) has a much
higher saturation pressure than a mixture that contains 2
mol.% methane (CH4).
Semi-empirical models, such as the Two-Curve model,
developed by the Battelle Memorial Institute in the 1970s,
168 The Journal of Pipeline Engineering

1.0E-05
Individual Risk (per year)

1.0E-06

1.0E-07

1.0E-08
-1000 -800 -600 -400 -200 0 200 400 600 800 1000

n
Distance from Pipeline (m)

io
st y
ut NG Pipeline CO2 Pipeline
di op
rib
Fig.10. Individual risk transect for natural gas (NG) pipeline and a CO2 pipeline (ruptures only).
or c
t f ple
no m
Sa

Fig.11. Environmental testing at the ASGARD facility at Nottingham University.

3rd Quarter, 2012
are available to define the toughness required to arrest a
running-ductile fracture [9-11]. However, these models have
not been validated for CO2. National Grid has (i) conducted
a series of shock-tube tests to validate the decompression
models used to calculate the saturation pressure [12,13]; and
(ii) is also conducting full-scale fracture-propagation tests
to validate the Two-Curve model for CO2-rich mixtures.
Modern, high-quality, linepipe is very tough; full-size
Charpy V-notch impact energies of 250 Joules (J) are
achievable without a cost-penalty. Figure 7 presents an
indicative relationship between the saturation pressure
of the mixture and the wall thickness of 250-J pipe [14].
As discussed above, the saturation pressure is defined by
the composition of the mixture and the initial pressure
and temperature. Thus, the composition that can be
transported in a pipeline is limited by the wall thickness.
If a more-severe composition (in terms of saturation
pressure) is transported, then an increased wall thickness
is required [14].
Figure 8 presents a relationship between the design pressure
n
and the wall thickness of the pipe. The inclined line shows
the wall-thickness requirement based on the (0.72) design
io
factor. Each of the horizontal lines corresponds to the wall
st y
ut
thickness requirement based on the indicated saturation
di op
pressure (fracture-control requirement). As an example,
rib
we consider a pipeline transporting a mixture with a
or c
saturation pressure of 80barg. For design pressures greater
than 190barg, the wall thickness is based on design-pressure
t f ple
requirements (inclined line); for design pressures less than
190barg, the wall thickness is based on the more-onerous
fracture-control requirement (horizontal line), i.e. thicker
no m
pipe is required. Also the (fracture-control) wall thickness
requirements increase as the saturation pressure increases.
Sa
This is discussed in [14] by Cosham9.
Essentially, a pipeline designed to transport a relatively high
proportion of impurities will cost more than one designed
to transport a relatively low proportion, because thicker
pipe will be required. Conversely, the capture plant will
cost less. Techno-economic studies would be required to
optimize the design of the CCS system.
Quantified risk analysis: pipeline individual and
societal risk levels
The understanding and application of pipeline risk analysis
covered in the UK pipeline codes [4, 5, 15] are based on
thermal hazards. The work completed under COOLTRANS
has shown that the hazards arising from a drifting toxic
cloud produce a different shape to the individual risk
curves when compared with curves for a flammable gas,
such as natural gas. This is illustrated in Fig.10, where the
individual risk values are plotted against the distance along
a line perpendicular to the pipeline (the risk transect)
9. Published in this issue on pages 213–228.
169
for a typical high-pressure natural gas pipeline and for a
large-diameter, thick-walled, dense-phase CO2 pipeline.
The differences arise as the thermal hazard distances from
a natural gas pipeline are less sensitive to variations in the
atmospheric conditions, wind speed, and topography than
the dispersion distances associated with a drifting cloud
of CO2. The sensitivity produces a long tail to the risk
transect for the CO2 pipeline. Further, the lower failure
frequencies associated with the large-diameter, thick-walled,
CO2 pipelines mean that the individual risk levels are
relatively low. As a result, the balance between individual
and societal risk is different for CO2 pipelines. This means
that the conventional approach that is used for assessing the
consequences and risks of transporting substances such as
natural gas has to be reviewed and, if necessary, modified.
Further, the greater distance and longer time period
associated with the CO2 hazard mean that a number of
factors may be more significant in modelling the risk, such
as the influence of the source conditions and subsequent
dispersion from the crater; and population factors, including
the density of people at larger distances from the pipeline,
the time people spend out of doors, escape and shelter
assumptions, and the influence of isolation valves along
the pipeline.
The more-fundamental CFD-based modelling studies in
the COOLTRANS R&D programme are providing the
theoretical understanding that is required to address these
issues. In parallel, the more-practical risk-assessment models
are being applied to investigate the sensitivity of the results
to the above factors and to advise on any modifications
required for CO2 pipelines. For example, the requirements
for individual and societal risk evaluation for dense-phase
CO2 pipelines and a possible screening method for societal
risk-based routeing are discussed in [16] by Cleaver and
Hopkins10.
Environmental effects of CO2 releases
Work completed to date by Nottingham University, reported
in [16]11 and shown in Fig.11, has indicated that major
short-term (8-hr) surface leaks at ambient temperature
have shown no long-term detrimental effects on the species
studied, which include Agrostis stolonlifera – creeping bent
grass, Lolium perenne – perennial rye grass, and Taraxacum
officinalis – dandelion. Monitoring of plant responses to
long-term (seasonal) low-level sub-surface gaseous CO2
leaks indicates that with the tests undertaken so far there
is unlikely to be extensive damage. National Grid considers
that as the likelihood of both leaks and ruptures in thick-wall
dense-phase CO2 pipelines is very low, the environmental
impact of CO2 pipelines is no different to that of other
hazardous pipelines.
10. Published in this issue on pages 191–200.
11. Published in this issue on pages 239–248.
170 The Journal of Pipeline Engineering

n
io
st y
ut
di op
rib
or c
t f ple
no m
Sa

Fig.12. Dense-phase CO2 pipeline design process.

3rd Quarter, 2012
Summary
The impact of the key learning described above and discussed
in Refs 8, 14, 16, 17 in terms of pipeline design can be
summarised as follows:
a. The CO2 entry specification impacts on the
requirements for integrity and fracture control of
the pipeline, and may result in (i) wall thicknesses
greater than required for a design factor of 0.72, and
(ii) the need for use of sour-grade materials.
b. Historically, pipe toughness levels to ensure the
arrest of long-running ductile cracks are based on the
most onerous conditions of maximum pressure and
lowest operating temperature (and composition). For
dense-phase CO2 pipelines, more-onerous toughness
levels are required for lower operating pressures and
(particularly) the highest operating temperature.
c. Requirements for increased wall thickness and use
of sour-grade materials result in significant increases
in pipeline costs and introduce the need for techno-
economic assessment as part of the design process.
n
d. The approach to pipeline routeing and QRA specified
in the pipeline codes is based on thermal hazards:
io
this must be modified for the toxic hazards posed
st y
by dense-phase CO2.
ut
di op
e. The environmental impact of CO2 pipelines is
rib
equivalent to that for pipelines transporting other
or c
fluids, and is limited mainly to the construction
phase.
t f ple
Application to pipeline design
no m
National Grid has developed a pipeline-design process which
applies the learning described in the preceding section, and
Sa
which is shown in Fig.12. The process has been developed
to ensure that safety is established, the interlinks between
the parameters affecting the pipeline design are properly
assessed, and a techno-economic assessment is applied to
challenge technical requirements and to evaluate project
feasibility.
Implications of research findings
The research undertaken under the COOLTRANS
programme has demonstrated the linkages between key
parameters which influence the design of the pipeline.
The CO2 quality specification influences corrosivity and
saturation pressure:
• The potential for corrosion (due to the presence of
water and other highly reactive compounds such
as hydrogen sulphide (H2S), NOx, and SOx – the
collective terms for oxides of nitrogen and sulphur,
respectively) requires corrosion control, which in turn
requires (i) a corrosion allowance and/or sour-grade
material, or (ii) reliable drying of the CO2 inlet stream.
171
• The potential for increased saturation pressure
(due to the presence of non-condensable substances
such as hydrogen (H2), nitrogen (N2), argon (Ar),
and methane (CH4)) changes the decompression
characteristics of dense-phase CO2 and this means
that a higher material toughness and wall thickness
may be required for fracture arrest.
The wall thickness of a typical pipeline is determined by
the limit on the design factor (for example, 0.72). However,
the corrosivity and decompression characteristics of dense-
phase CO2 are such that the nominal wall thickness may,
in fact, be determined by the more-onerous (thicker-walled)
requirements for integrity and fracture control. Furthermore,
dense-phase CO2 pipelines will be required to transport
impurities. The wall thickness required for both corrosion
and fracture control is influenced by the level of impurities.
Transporting higher concentrations of impurities requires
increased wall thickness which has an associated cost increase
on the transportation elements of CCS projects.
The cost of increasing the wall thickness of a pipeline is
significant. For example, for a 100-km long pipeline, the
costs in terms of steel for various diameters (i.e. excluding
the increase in other construction costs such as welding) of
a 1-mm increase in wall thickness is (based on May, 2012,
prices) are:
• for 914-mm diameter pipeline: 3.96 million Euro
• for 610-mm diameter pipeline: 2.53 million Euro
• for 457-mm diameter pipeline: 1.98 million Euro
A pipeline for the transportation of dense-phase CO2 will
have a higher cost if the CO2 stream contains higher levels
of impurities (i.e. thicker pipe will be needed). However, the
need for purer CO2 will increase the costs of the capture
plant. This means that project costs must be assessed and
evaluated across all parts of the CCS chain in order to select
the best option. The assessment should consider for example,
whether the pipeline should be constructed of the thinnest
pipe for the proposed levels of impurities, or of thicker pipe
to allow for higher levels of impurities in the future as new
emitters add to the transportation requirements.
The above requirements demonstrate that the pipe wall
thickness for a dense-phase CO2 pipeline will be greater
than that required for the maximum design factor of 0.72
allowed by the current pipeline codes. In order to ensure
the additional costs are optimized, a techno-economic
assessment is required to identify the best project option.
In order to address the current code gaps affecting CO2,
QRA is required to establish the required separation
distance between the pipeline and occupied buildings
(i.e. the minimum distance to occupied buildings, or the
‘building-proximity’ distance), the width of the route corridor
for population-density assessment, and to establish design
features such as the spacing of block valves. The high wall
172
thicknesses required for dense-phase CO2 pipelines results
in reduced failure frequencies and this, together with the
difference in failure consequences for a toxic CO2 pipeline,
means that a robust QRA methodology which addresses
these factors is required. The development of the QRA
methodology under the COOLTRANS R&D programme
has allowed this changed methodology to be established and
applied to pipeline routeing and design studies.
Concluding remarks
1. The £8 million COOLTRANS research programme is
now 50% complete and is delivering results which are being
applied to ongoing project studies.
2. The COOLTRANS research programme includes a
comprehensive programme of large-scale experimental
tests, some of which are the first of their type in the world.
3. The results are demonstrating complex interactions
between key design parameters which require new design
rules.
4. National Grid has developed a pipeline design process
which uses and applies the key learning obtained from the
n
COOLTRANS R&D programme.
5. The design process relies upon a CO2 specification, which
io
allows the assessment of corrosion- and fracture-control
st y
ut
requirements to be accounted for in pipeline design.
di op
6. National Grid is developing and applying methodologies
rib
which enable the fast and practical assessment of the impact
or c
of CO2 stream composition on pipeline design. These include
the use of a model developed from parametric studies for
t f ple
the rapid calculation of saturation pressure.
7. The conventional pipeline-code approach for deriving
wall thickness on a maximum design factor of 0.72 requires
no m
modification for CO2 pipelines. The wall thickness needs to
be based on the more-onerous (thicker-wall) requirements
Sa
for corrosion and fracture control. There are significant cost
implications which require consideration.
8. The conventional approach used for assessing the
consequences and risks of transporting hazardous substances
such as natural gas requires modification for CO2 pipelines.
The reason is that the hazard from a drifting toxic cloud is
directional and dependent on topology, and is time-based (i.e.
it extends with time), whereas the maximum hazard due to
thermal radiation from a fireball due to flammable gas cloud
is less directional, is not dependent upon topography, and
is short lived (i.e. it is either immediate or does not occur).
The COOLTRANS research has enabled the differences
to be quantified.
9. The environmental impact of CO2 pipelines is considered
to be no different from that of any other pipeline, and
is limited mainly to the environmental impact due to
construction.
10. The £8-million investment in the COOLTRANS R&D
programme has provided National Grid with the knowledge
to deliver CO2 pipelines which are viable, safe, and cost
effective.
The Journal of Pipeline Engineering
References
1. R.Cooper, 2011. Integrated CO2 dispersion studies. National
Grid. Second International Forum on Transportation of
CO2 by Pipeline, 2011.
2. Department of Energy and Climate Change, 2012. CCS
Roadmap supporting deployment of carbon capture and
storage in the UK. April.
3. J.M.Race, et al., 2009. Development requirements for a CO2
transport specification. Task 4: implications for pipeline
design to existing codes. Newcastle University Contract
Research Report for National Grid.
4. BSI, 2004. Code of practice for pipelines – Part 1 Steel
pipelines on land. PD 8010-1:2004.
5. IGEM, 2008. Steel pipelines and associated installations
for high pressure gas transmission. IGEM/TD/1 Edition
5, Communication 1735.
6. HSE, 1996. A guide to the Pipeline Safety Regulations
1996. L82 HSE Books ISBN 07176 1182 5.
7. S.Gant, 2012. Framework for validation of pipeline release
and dispersion models for the COOLTRANS project.
Health and Safety Laboratory. Third International Forum
on Transportation of CO2 by Pipeline.
8. J.M.Race et al., 2012. Towards a CO2 pipeline specification:
defining tolerance levels for impurities. Third International
Forum on Transportation of CO2 by Pipeline.
9. A.Cosham and R.J.Eiber, 2008. Fracture control in
carbon dioxide pipelines – the effect of impurities. Paper
No.: IPC2008-64346, Proc.7th International Pipeline
Conference, Calgary, Alberta, Canada.
10. A.Cosham, R.J.Eiber, and E.B.Clark, 2010. GASDECOM:
carbon dioxide and other components. Paper No.:
IPC20010 31572, Proc.8th International Pipeline
Conference, Calgary, Alberta, Canada.
11. M. Mohitpour, P.Seevam, K.K.Botros, B.Rothwell, B., and
C.Ennis, 2011. Pipeline transportation of carbon dioxide
containing impurities. ASME Press, New York, USA.
12. J.Barnett and D.G.Jones, 2011. The Longannet CCS
project shock tube tests. Second International Forum
on the Transportation of CO2 by Pipeline.
13. A. Cosham, D.G.Jones, K.Armstrong, D.Allason, and
J.Barnett, 2011, The decompression behaviour of carbon
dioxide in the gaseous phase. Ibid.
14. A.Cosham, 2012. The saturation pressure and the design
of dense-phase CO2 pipelines. Third International Forum
on the Transportation of CO2 by Pipeline.
15. BSI, 2009. PD 8010-3:2009 Code of practice for pipelines
– Part 3 Steel pipelines on land – Guide to the application
of pipeline risk assessment to proposed developments
in the vicinity of major accident pipelines containing
flammables – Supplement to PD 8010-1:2004.
16. P.Cleaver and H.Hopkins, 2012. The routeing of dense-
phase CO2 pipelines. Third International Forum on
Transportation of CO2 by Pipeline.
17. J.A.Lake, M.Steven, K.Smith, and B.Lomax, 2012.
COOLTRANS – CO2 and the environment. Thirds
International Forum on Transportation of CO2 by
Pipeline.
3rd Quarter, 2012 173

Towards a CO2 pipeline

specification: defining tolerance
limits for impurities
by Dr Julia M Race*, Dr Ben Wetenhall, Dr Patricia N Seevam, and Prof.
Martin J Downie

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

A s more carbon capture and storage (CCS) projects are proposed, it is becoming clear that the
specification of the CO2 in the pipeline is an under-researched area. Research has been conducted
into the effect of the impurities on the different aspects of pipeline hydraulic design, pipeline integrity, and
public safety; however, the inter-relationship of these differing, and sometimes competing, requirements
has not been investigated in detail.

n
This paper reviews the current pipeline specifications for CO2 pipelines and then discusses the effects that

io
different impurities have on key aspects of pipeline design, operation, integrity, and health and safety and
st y
ut
the requirements that need to be considered when specifying the maximum levels of these impurities for
di op

entry into the pipeline system.

rib
or c

W orldwide, it is estimated that there are over increased cost associated with the removal of impurities
t f ple

3500km of operational, long-distance, high-pressure to low levels), legislative and regulatory requirements,
CO2 pipelines (Table 1). The majority of these pipelines are specifications, and safety considerations.
located in the USA, and are transporting mainly natural sources
no m

of CO2 for onshore enhanced-oil recovery (EOR). There is The definition of a CO2 pipeline specification from carbon-
only one high-pressure pipeline currently transporting CO2 capture plants can therefore be undertaken by analysing
Sa

for storage purposes, the offshore Snøhvit pipeline operated
by Statoil in Norway. None of the pipelines introduced in
Table 1 transports pure CO2 and, whether the source is natural
or anthropogenic, the CO2 stream will contain some level of
other components. There is therefore a requirement to define
a pipeline entry specification for the CO2.
In many respects the definition of a universal specification
is less complicated when the CO2 is derived from naturally
occurring sources, which are relatively pure and fairly
consistent in composition. As a result, in the USA, and
particularly for EOR applications, the quality specification
is driven by the requirements of EOR as well as pipeline
integrity. However, for pipeline transportation for CCS
schemes, particularly from power plant capture, the purity of
the CO2 is affected, not only by the various types of capture
technology and processes, but also by economics (i.e. the
This paper was presented at the Third International Forum on the Transportation of
CO2 by Pipeline, organized by Newcastle University, Tiratsoo Technical, and Clarion
Technical Conferences, and held in Newcastle in June, 2012.
*Corresponding author’s contact details:
tel: +44 (0)191 222 6964
email: j.m.race@ncl.ac.uk
the types and levels of impurity that could be present from
each capture technology and then by specifying the levels of
impurities that could be safely transported from a health,
environmental, and technical point of view.
This paper will review the current pipeline specifications
for CO2 pipelines and then discuss the effects that different
impurities have on key aspects of pipeline design, operation,
integrity, and health and safety, and the requirements that
need to be considered when specifying the maximum levels
of these impurities for entry into the pipeline system. It is
accepted that this approach only considers the effect on the
pipeline section of the CCS chain and it is appreciated that
the requirements of the storage site will also influence the
CO2 specification.
Review of CO2 specifications for
pipeline transportation
Before analysing proposed CO2 specifications for CCS, it
is first informative to consider the quality specifications
of currently operating CO2 pipelines in order to put the
discussion into context.
174 The Journal of Pipeline Engineering

CO2
Length MAOP
Pipeline Location capacity Source Sink Year
(km) (bar)
(Mt/y)

Canyon Reef Carriers USA 5.2 225 140 Anthropogenic EOR 1972

Bati Raman Turkey 1.1 90 170 Natural EOR 1983

Cortez USA 19.3 808 186 Natural EOR 1984

Bravo USA 7.3 350 165 Natural EOR 1984

Central Basin Pipeline USA 20 278 170 Natural EOR 1985

Bairoil USA 8.3 180 - Anthropogenic EOR 1986

Val Verde USA 2.5 130 140 Anthropogenic EOR 1998

Weyburn USA/Canada 5 328 186 & 204 Anthropogenic EOR 2000

n
Snøhvit Statoil 0.7 153 150 Anthropogenic Storage 2008

io
st y
ut
Table 1.Typical operating conditions for long-distance, high-pressure CO2 pipelines
di op

(Source data: Gale and Davison [43], Seevam, Race and Downie [15]).
rib
or c

Weyburn
t f ple

KM CO2 CRC Pipeline KMCO2 Cortez Pipeline [2]

Pipeline
Pipeline specification
[1] Specification Actual Actual
Specification Actual
[3] [21] 44
no m

Source Anthropogenic Natural Anthropogenic

Sa

CO2 >95% >95% 95% >95% 98.35% 96%

N2 4% 4% 0.5% 4% 1.514 300ppm

0.7% (CH4)
Hydrocarbons 5% 5% 5% 1-5% 0.136%
2.3% (C2+)
30 lbs/
H2O 0.48 g/m3 0.11g/m3 30 lbs/MMscf Not reported 20ppm
MMscf

O2 10 ppm wt 10 ppm wt -- -- 50ppm

H2S 10-200 ppm 1500 ppm wt -- 0.002% -- 0.9%

0.3 gal/
Glycol 4x10-5 L/m3
MMcf

CO -- -- -- -- -- 0.1%

Table 2: Examples of CO2 pipeline quality specifications and compositions (all percentages are vol%).
3rd Quarter, 2012
Currently operating pipelines
CO2 quality specifications
As mentioned previously, for transmission pipelines in the
USA, the quality requirements for the CO2 are primarily
dictated by the effects of impurities on the EOR process
rather than economic, safety, or hydraulic considerations.
For EOR, impurities can affect the minimum miscibility
pressure (MMP) of CO2 with crude oil. Impurities such
as nitrogen (N2), hydrogen (H2), and methane (CH4)
increase the MMP and therefore are undesirable; however,
hydrogen sulphide (H2S) decreases the miscibility of CO2
and therefore it could be considered to be a desirable
impurity, ignoring all other effects of H2S.
Table 2 provides a comparison between the required
CO2 quality specifications and the actual compositions
achieved in currently operational CO2 pipelines. Most
natural sources of CO2, which are transported by pipeline,
contain impurities such as CH4, higher hydrocarbons, and
N2, as shown in the Kinder Morgan (KMCO2) CO2 quality
specification [1] and the Cortez pipeline specification [2].
n
Anthropogenic sources of CO2 can also include carbon
monoxide (CO), oxygen (O2), and hydrogen sulphide
io
(H2S). A typical quality specification for the anthropogenic
st y
ut
Canyon Reef Carriers (CRC) CO2 pipeline, as defined
di op
in IPCC [3], is also presented in Table 2.
rib
or c
It is interesting to note from Table 2 that the Cortez
pipeline, operated by KMCO2, is transporting CO2
t f ple
at concentrations well above those required by the
specification and that the Weyburn pipeline is transporting
significantly more H2S than the other pipelines. As
no m
discussed previously, the presence of H2S is desirable for
EOR applications. In addition, as CO2 is odourless, the
Sa
presence of H2S acts as an odorant which can alert the
public to the presence of a leak in the pipeline.
Projected CO2 purity specifications
from different capture technologies
The focus for CCS projects is to capture CO2 predominantly
from power plants. The amount and type of impurities in
the CO2 stream captured from a power plant are dependent
on the capture process, the capture technology, the fuel
source, regulatory constraints, and also – as mentioned
previously – economic considerations. Studies of the
levels and types of impurity that could be present in the
captured CO2 from different capture technologies and
fuel sources have been published by IPCC [3], the EU
Framework 6 ENCAP (European Enhanced Capture of
CO2) project[4], Oosterkamp and Ramsen [5], and the
International Energy Agency Greenhouse Gas R&D
Programme [6, 7]. A review of some of these specifications,
as they relate to pipeline transportation, is provided in
Mohitpour et al. [8] and a summary is presented in Table 3.
Comparison of Table 2 with Table 3 indicates that there
are some types of impurity which are not currently being
175
transported in operational CO2 pipelines including,
for example, NOx, SOx1, Ar, and H2. It is therefore
important to understand the impact of these impurities, in
particular, on the specification. In terms of defining a CO2
specification, it should also be noted from Table 4 that,
although some impurities are not produced by some of the
capture technologies (for example, H2S is only produced
in pre-combustion processes), co-mingling of CO2 from
different capture technologies in a single transportation
pipeline requires a specification to be ‘universal’ and to
consider all impurities that could be present from every
capture technology. Another challenge for a pipeline
specification is illustrated by the difference in the levels
of some impurities between different technologies; for
example, in post-combustion capture the combined levels
of Ar and N2 could be as low as 0.01vol%, while for some
types of oxyfuel capture technology, the combined levels
could be over 10vol%. This range of requirements for
a pipeline specification makes it difficult to specify the
CO2 composition from the point of view of the capture
processes and the components that could be present
in the captured CO2 stream. An alternative position
has therefore been to define the composition from the
point-of-view of the requirements of the pipeline. This
is the basis of the Dynamis and Ecofys studies described
in the next section.
Projected CO2 pipeline
specifications for CCS transport
The Dynamis transport specification [9] presented in Table
4 is based on the ENCAP specification [4] for pre- and
post-combustion capture (Table 3). It is highlighted that,
as other capture technologies were not considered in the
analysis (such as oxyfuel), the application of the Dynamis
specification to pipelines carrying CO2 from processes
other than pre- and post-combustion technologies should
be considered with care.
For the Dynamis specification, the ENCAP specification has
been modified to take account of safety and toxicity limits,
in the event of a release from the pipeline; infrastructure
durability, in terms of the need to avoid free water
formation to prevent hydrate formation and corrosion;
and transport efficiency [9]. These considerations have
decreased the allowable levels of H2S, CO, SOx, and NOx
from the ENCAP limits, using the short-term exposure
limits (STELs) for these compounds to set a maximum
concentration in the CO2 stream on the basis of toxicity
effects. The allowable water content has also been increased
from the ENCAP specification. This increase has been
justified on the basis that the specified level is below the
water-saturation limit at the pipeline-operating conditions
considered, allowing for the effects of other impurities
on the solubility of water in CO2, particularly H2S and
CH4 and the risk of hydrate formation.
1. Collective terms for oxides of nitrogen and sulphur, respectively.
176 The Journal of Pipeline Engineering

Comp (vol %) CO2 CH4 H2S C2+ CO O2 Ar N2

IPCC Coal >99.97 0.01

IPCC Gas >99.97 0.01

ENCAP 99.8 0.003 0.001 0.003 0.021 0.021

Oosterkamp et al >99 0.01 Trace 0.01 0.001 0.01 Trace 0.17

IEAGHG - Comp1 99.93 0.001 0.015 0.045

IEAGHG - Comp2 99.92 0.001 0.015 0.045

IEAGHG - Comp3 99.81 0.002 0.03 0.045

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

IPCC Gas >95.65 2 0.01 0.04 1.3

n
io
ENCAP - CO2/H2S 97.8 0.035 0.01 Unknown 0.17 Unknown 0.05 0.03
st y
ut
di op

ENCAP - CO2+H2S 95.6 0.035 2.3 Unknown 0.17 Unknown 0.049 0.03
rib
or c

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

t f ple

IEAGHG - Selexol 97.95 0.01 0.01 0.04 0.03 0.9

no m

IEAGHG - Rectisol 99.7 0.01 0.01 0.04 0.15 0.21

Sa

IPCC Coal >95.79 3.7

IPCC Gas >95.88 4.1

ENCAP - CO2/SO2 91 Unknown 1.6 5.7 0.61

ENCAP - CO2+SO2 90 Unknown 1.6 5.6 0.6

Oosterkamp et al >90 Trace Trace 3 5 7

IEAGHG - Comp1 85 4.7 4.47 5.8

IEAGHG - Comp2 98 0.67 0.59 0.71

IEAGHG - Comp3 99.94 0.01 0.01 0.01

Table 3. Comparison of published CO2 compositions from different capture technologies (all percentages are vol%).
3rd Quarter, 2012 177

Comp (vol %) NOx SOx H2 HCN COS NH3 CH3OH

IPCC Coal 0.01 0.01

IPCC Gas 0.01 0.01

ENCAP 0.002 0.01

Oosterkamp et al 0.005 0.001 Trace

IEAGHG - Comp1 0.002 0.001

IEAGHG - Comp2 0.002 0.001 0.005

IEAGHG - Comp3 0.002 0.001

IPCC Coal 0.8-2.0

IPCC Gas 1

n
io
ENCAP - CO2/H2S Unknown 1.7 0.0005 0.0005 0.003
st y
ut
di op

ENCAP - CO2+H2S Unknown 1.7 0.0005 0.0005 0.003

rib
or c

Oosterkamp et al 0.05 Unknown

t f ple

IEAGHG - Selexol 1 With H2S

no m

IEAGHG - Rectisol 0.002 With H2S 0.02

Sa

IPCC Coal 0.01 0.5

IPCC Gas 0.01 0.01

ENCAP - CO2/SO2 0.25 0.076 Trace Trace

ENCAP - CO2+SO2 0.24 1.5 Trace Trace

Oosterkamp et al 0.25 2.5 Trace

IEAGHG - Comp1 0.01 0.007

IEAGHG - Comp2 0.01 0.007

IEAGHG - Comp3 0.01 0.007

178 The Journal of Pipeline Engineering

DYNAMIS
ECOFYS
Storage EOR

CO2 >95% >95% >95%

H2O 500 ppm 500 ppm 500 ppm

SOx 100 ppm 100 ppm Not critical

NOx 100 ppm 100 ppm Not critical

H2S 200 ppm 200 ppm 200 ppm

CO 2000 ppm 2000 ppm 2000 ppm

H2
Total
Ar

n
non-condensable
gases

io
Total Total
st y
N2 ut non-condensable non-condensable
gases gases
di op

O2 100 ppm
rib
or c

CH4 100 ppm

t f ple

Table 4. Pipeline specifications proposed by the Dynamis and Ecofys projects [9, 10].
no m

150
Sa

Pressure /bar

140
130
Dense Phase CO2

Supercritical CO2

120
110
100
90
80
70 Critical Point
Solid CO2 60
50
40
Liquid CO2 30 VLE Curve
20 Vapour CO2
Triple Point 10
0
-100 -90 -80 -70 -60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60
Temperature/ºC

Fig.1. Phase diagram for pure CO2.

3rd Quarter, 2012
The Ecofys study took a similar approach to the Dynamis
project, except that the case study was based on potential
impurities from coal fired power plants [10]. In contrast to
the Dynamis specification, the Ecofys specification does not
set a limit for SOx and NOx as these are not considered to
be critical impurities if no free water is present (Table 4).
In both the Dynamis and Ecofys studies, the total amount of
the non-condensable components such as N2, O2, H2, CH4,
and Ar, is limited to less than 4%. This limit is primarily
based on the assumption that these gases will reduce
the pipelines’ transport capacity and therefore increase
the pipeline investment and compression energy costs.
Although no cost-benefit analysis is presented in either the
Dynamis or Ecofys studies for this limit, Yan et al. [11] have
studied the techno-economic impact of non-condensables
at different levels (13%, 4%, and 1% by volume) on the
transportation of CO2 from oxyfuel capture. They conclude
that the limit on non-condensable components of less than
4% is a reasonable purification limit in terms of the cost
balance of the CCS chain. However, they indicate that, for
short distances and where the storage conditions permit,
n
the level of non-condensables could be raised to 10%vol.
io
Effect of impurities
st y
on pipeline design ut
di op
rib
Having reviewed the literature relating to CO2 specifications
or c
and discussed the impurities that could be present in a
pipeline transporting CO2 for CCS schemes, the remainder
t f ple
of this paper will discuss the impact of these impurities
on pipeline design, operation, integrity, and health and
safety aspects. Each section will discuss the effect of the
no m
CO2 specification on hydraulic efficiency, fracture control,
material degradation (in terms of corrosion and cracking),
Sa
and health and safety. In each section, the effect of the
individual impurities will be discussed in isolation without
reference to any other constraints; for example, the effect
of NOx on hydraulics is not coupled to the effect of NOx
on health and safety. In addition, the concentrations
of some impurities are relatively high in this analysis
to demonstrate an effect, although it is recognized that
some levels of impurity are not feasible from the capture
processes discussed previously.
CO2 specification for hydraulic efficiency
In order to understand the issues relating to the hydraulic
efficiency and, in particular, the composition of the
CO2 in the pipeline, it is first necessary to consider the
phase in which the fluid is being transported. Reference
to the phase diagram for pure CO2 in Fig.1 defines a
critical point in the phase diagram at 74bar and 31°C. At
pressures and temperatures above the critical point, CO2
no longer exists in distinct gaseous and liquid phases.
In this paper, two regions are defined above the critical
pressure: the ‘supercritical’ phase, which lies above the
critical temperature, and the ‘dense’ phase which lies below
179
the critical temperature. In moving from the supercritical
phase to the dense phase there is no distinct phase change
although the density of the fluid increases with decreasing
temperature. However, below the critical pressure there is
an abrupt change in phase and physical properties when
crossing the vapour-liquid equilibrium (VLE) curve. At
any point along the VLE curve both gas and liquid phases
can co-exist at the same conditions of temperature and
pressure resulting in a two-phase condition.
In pipelines it is most efficient and economic to transport
the CO2 as a supercritical or dense-phase fluid, as under
these conditions the fluid has the density of a liquid but
the viscosity of a gas. It is also important to prevent the
pipeline from operating near the VLE conditions, and
consequently all of the high-pressure pipelines referred to
in Table 1 are operating above the critical pressure. The
outlet temperature from the compressor may be as high as
40-50°C [12], and therefore at the start of these pipelines
the CO2 will be in the supercritical phase. However,
further along the pipeline, the temperature will drop
due to heat transfer from the pipeline, and although the
CO2 in the pipeline will start in the supercritical phase, it
could move into the dense phase with increasing distance
along the pipeline.
Although the supercritical phase is the most efficient phase
in which to transport large volumes of CO2, gaseous-phase
pipelines are also being considered for CCS schemes
and could be viable for short pipeline sections or for
transport through more densely populated areas. In these
regions, high-pressure pipelines may not receive regulatory
approval and it may be desirable to make use of existing
infrastructure that has been designed for lower pressures
[13]. For example, the Lacq pipeline operated by Total
in France is transporting CO2 in the gaseous phase from
an oxyfuel capture plant through a 27-km pipeline at a
pressure of 30bar to an onshore storage site [14].
Impact of impurities on pipeline-operating
pressure
The phase behaviour of CO2 changes when impurities are
introduced into the system. This change is dictated by the
type, amount, and combination of impurities present, as
the impurities interact both with CO2 and each other (as
in the case of water). The general effect of any impurities
when added into the CO2 stream is to raise the critical
pressure and open-out a two-phase (liquid-gaseous) area
in the phase diagram. Components having a critical
temperature and pressure higher than CO2 form a phase
envelope that expands below the VLE of pure CO2, whereas
those with a lower critical temperature and pressure than
CO2 expand above the VLE for pure CO2. The critical
temperatures and pressures of the major components from
power-plant capture (as presented in Table 3) relative to
CO2 are illustrated in Fig.2, and the effect on the phase
diagram for binary combinations of CO2-NO2 and CO2-H2
180 The Journal of Pipeline Engineering

6
Relative Critical Temperature (ºC)
Relative Critical Pressure (bara)
4

2
Relative value to carbon dioxide

0
Sulphur Hydrogen Nitrogen Hydrogen Oxygen Nitrogen Carbon Argon Methane Carbon
dioxide sulphide dioxide monoxide dioxide
-2

-4

n
-6

io
st y
-8
ut
di op
rib
or c

Fig.2. Relative critical pressures and temperatures.

t f ple

100
no m

90
Sa

80

70
Pressure /bara

60

50

40

30 Pure Carbon Dioxide

20 5% Hydrogen

10 5% Nitrogen Dioxide

0
0 5 10 15 20 25 30 35 40 45 50
Temperature/ºC

Fig.3. Pressure-temperature diagram for pure CO2, CO2-5mol%H2, and CO2-5mol%NO2.

3rd Quarter, 2012 181

1000

900

800

700
Pure Carbon Dioxide
Density /kg/m

600 5% Argon
5% Carbon Monoxide
500 5% Hydrogen
5% Hydrogen Sulphide
400 5% Methane
5% Nitrogen
300 5% Nitrogen Dioxide
5% Oxygen
200 5% Sulphur Dioxide

100

n
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190
Pressure /bar

io
st y
ut
di op

Fig.4.Variation of density with pressure and temperature for pure CO2, and CO2 with 5mol% impurity.
rib
Impact of impurities on pipeline sizing
or c

is illustrated in Fig.3. The implications of the size of the

two-phase area and changes in the critical pressure lie in
t f ple

the operating flexibility of the pipeline system. One of the first stages in the design of a pipeline is to size the
pipeline, in terms of internal diameter, for the anticipated
For dense-phase pipelines, a high critical pressure and large flow rate. The capacity of the pipeline is dependent on the
no m

two-phase area will require the pipeline to be operated required pressure and temperature as well as on the fluid’s
at higher pressures to reduce the risk of two-phase flow physical properties, in particular compressibility and density.
Sa

during upset conditions. Operating at a higher pressure
will have economic implications in that, in order to comply
with the maximum-allowable stress in the pipeline, the
wall thickness will have to be increased, or the diameter
of the pipeline will have to be reduced, or a higher-
strength steel specified for the pipeline. However, for
a gaseous-phase CO2 pipeline, the formation of a large
two-phase area below the pure CO2 VLE line reduces
the maximum-allowable operating pressure (MAOP) of
the pipeline to avoid two-phase flow, particularly at low
ground temperatures. Reducing the operating pressure
level limits the throughput and operational flexibility of
the pipeline, and therefore has economic implications.
Therefore, it can be concluded that, in order to increase
the operational flexibility of the pipeline, the components
with the highest relative critical pressures and temperatures
(for example SOx and NOx) should be limited when
specifying the quality of the CO2 to be transported in the
gaseous phase, and the components with the lowest relative
critical pressures and temperatures (such as H2 and N2) in
the CO2 mixture should be limited when specifying the
quality of the CO2 to be transported in the dense phase.
The way in which the density of CO2 changes with pressure
and temperature, and the sharp discontinuity it exhibits close
to the VLE curve, is significantly influenced by the level of
impurity present. This behaviour, and the effect on CO2
pipeline transportation, has been discussed in Seevam et al.
[15] and the relationship between density, pressure and level
of impurity is illustrated in Fig.4. It is highlighted that around
the discontinuity in Fig.4, small changes in pressure can have
large influences on the density. The addition of impurities will
move the location of the discontinuity to higher pressures (for
components with lower critical temperatures and pressures
than CO2) and to lower pressures (for components with higher
critical temperatures and pressures than CO2). A further key
point that was concluded by Seevam et al. [15] relates to the
effect of temperature on density. Reducing the temperature
increases the density of the CO2 fluid and therefore, for
dense-phase pipeline transportation in particular, lowering
the inlet temperature will increase pipeline capacity.
In order to illustrate the effect of the specification of the
pressure, temperature, and the level of impurities on pipeline
diameter, a series of hydraulic simulations are presented for
a case-study pipeline transporting pure CO2 and a mixture
182 The Journal of Pipeline Engineering

50
150bar;40°C
45
Difference in internal diameter compared with pure

150bar;30°C
110bar;40°C
40
110bar;30°C
35

30
CO2 (mm)

25

20

15

10

n
0

io
st y
400 ut 800 1200
Flow Rate (kg/s)
di op
rib
or c

Fig.5. Difference in diameter between pure CO2 and a CO2-5mol%H2 mixture with changing temperature and pressure.
t f ple

of CO2-5mol%H2. The study was conducted for a pipeline rate, thereby reducing the capital cost of pipeline projects.
length of 100km with a 0.0002bar/m pressure drop along However, this cost has to be balanced against the operational
the route; the ground temperature was specified as 5°C, costs associated with achieving these conditions.
no m

and the results are presented in graphical form in Fig.5.

As would be expected from the density changes described Impact of impurities on pressure
and temperature drop
Sa

previously, the study showed that increasing the temperature

and reducing the pressure results in larger diameters having
to be specified for any given fluid at any given flow rate.
However, Fig.5 illustrates that, for the CO2-5%H2 mixture,
the effect of the addition of H2 is to increase the required
diameters over pure CO2 and that the increase (in terms
of additional steel requirements) is greater at higher flow
rates and lower pressures. It should be noted that this
relationship will not hold for all impurities and is dependent
on the density of the CO2 mixture at the inlet temperature
and pressure relative to pure CO2. For example, reference
to Fig.4 indicates that, at some conditions of temperature
and pressure, a CO2-5%NO2 mixture is denser than pure
CO2 and at some conditions it is less dense, and therefore
the diameter required to transport this mixture will either
be greater or less than that required to transport pure CO2
depending on the operating conditions.
For dense-phase pipeline transportation, it can therefore
be concluded that lower inlet temperatures, higher inlet
pressures, and the minimization of low-density impurities
(such as H2) will maximize the CO2 throughput. This allows
smaller-diameter pipelines to be used for the same flow
As mentioned previously, for pipeline transport in the dense
phase, the pressure has to be maintained above the critical
pressure to keep the CO2 in the dense phase. If, due to
frictional losses or gravity-based losses, the pressure drops
below the critical point, intermediate pumping stations would
be required to raise the pressure back into the dense-phase
region. Alternatively, the diameter of the pipeline would have
to be increased or the initial pressure raised to ensure that
the pressure drop did not result in two-phase flow within
the pipeline length. All of these mitigation methods have
associated economic impacts and in some cases are mutually
exclusive. For example, increasing the diameter to avoid two-
phase flow will increase the stress in the pipeline and, if wall
thickness and steel grade remain constant, would require a
reduction in operating pressure to maintain the allowable
operating stress of the pipeline.
The impact of impurities on pressure drop along a pipeline
in the dense phase has been reported previously in Seevam,
Race, et al. [16]. This work concluded that, for binary
combinations of impurities, the addition of H2 resulted in
3rd Quarter, 2012
n
io
Fig.6. Battelle TCM for CO2 illustrating the decompression characteristics [22].
st y
ut
the largest pressure and temperature drops along the pipeline
di op
for the impurities studied. However, all of the impurities
rib
with lower critical temperatures and pressures than CO2 also
or c
showed this effect of increasing the pressure and temperature
drops relative to pure CO2. It can be concluded that the
t f ple
levels of H2 in particular should be minimized for dense-
phase transportation. Conversely, it was observed that the
addition of NO2, SO2, and H2S resulted in a lower pressure
no m
drop than observed for pure CO2 and therefore could be
seen to be beneficial in terms of the hydraulic characteristics
Sa
of dense-phase pipelines.
It is recognized that, due to the high density of the CO2
fluid, the effect of static head on the pressure drop can be
significant. Therefore, in some cases, gravitational effects
due to the changes in elevation can counteract any frictional
losses and the effects of impurities.
For transportation in the gaseous phase, the pressure
profile is not as significant as the temperature profile and,
in particular, the ground temperature. In this case it is
important that the temperature does not drop to a level such
that the fluid will enter the two-phase region. As mentioned
previously the addition of components such as NOx and
SOx should be limited in this respect. However, reducing
the inlet pressure will have a greater effect on increasing the
operating temperature range of the pipeline, but has impacts
for operating efficiency.
Impact of impurities on compression and pumping
In addition to the effects of impurities on the design of
the pipeline in terms of inlet temperature and pressure,
183
diameter, wall thickness, and steel grade, the level of non-
condensable impurities in the CO2 stream will also affect the
energy required for pumping and compression. Although
there is not much data available on cost comparisons for
compression, preliminary work has indicated that the
compression and pumping costs increase as the level of
impurities increases. In addition, as has been shown in the
preceding sections, if a higher inlet pressure is required for
certain stream compositions, then this will also increase
the power and the number of compression stages required,
escalating project and energy costs. These costs should also
therefore be included in any techno-economic analysis.
CO2 specification for fracture control
The problem of ductile fracture propagation was recognized
in the gas industry over 40 years ago, and fractures of in-
service natural gas pipelines have propagated in this mode
for distances up to 300m. Many authors have indicated
that ductile-fracture propagation may be an issue for CO2
pipelines [17-22], and the requirement to consider fracture
propagation in CO2 pipelines is included in the federal
regulations in the USA [23].
The concept and modelling of fracture propagation in CO2
pipelines has been described in detail in Seevam et al. [24]
and is reviewed here. On initiation of a fracture the fluid
starts to decompress, and a decompression wave propagates
in both directions from the fracture point at a velocity which
is dependent on the fluid properties. Whether the fracture
will propagate once initiated is dependent on whether there
is sufficient driving force for propagation, i.e. whether
the initial pressure is high enough to sustain a fracture. If
184
3
Argon
Carbon Monoxide
2.5 Nitrogen
Sulphur Dioxide
Relative Saturation Pressure (to pure CO2)
1.5
0.5
n
1 2
io
st y
ut
di op
Fig.7. Relative saturation pressures (compared with pure CO2) for binary CO2-%X mixture when decompressing from 100bar, 5°C.
rib
or c
fracture can be sustained, then a crack will propagate along
the pipeline at a velocity which is dependent on the strength
t f ple
and toughness of the pipe steel and also on the geometry
of the pipeline (i.e. the diameter and wall thickness). If the
crack-propagation velocity is slower than the decompression
no m
velocity then there is no driving force for propagation and
the crack will arrest.
Sa
Extensive world-wide research has led to the establishment of
a number of models, which describe the fracture-propagation
behaviour of gas pipeline systems. These models have been
very successful in defining toughness requirements for pipe
material for natural gas and rich-gas mixtures, which ensure
fracture arrest. The most widely used model is the Battelle
Two-Curve model (BTCM) [25]. Although, the BTCM has
not yet been validated for ductile-fracture propagation in CO2
pipelines, it can be used to study the effects of impurities
on the decompressing fluid and therefore the likelihood of
being able to arrest the crack once it has started to propagate.
An illustration of the BTCM for pure CO2 is provided
in Fig.6 [22]. The ‘two curves’ in the model are the fluid-
decompression curve (shown in brown), and the fracture-
velocity curve (shown in black). If the curves intersect
or are tangent, then there exists a pressure at which the
decompression wave and the crack are travelling at the same
speed, and the crack can propagate indefinitely under those
conditions. In order to prevent crack propagation, therefore,
the toughness of the steel is increased so that the fracture-
velocity curve lies above the fluid-decompression curve. By
The Journal of Pipeline Engineering
Methane
Hydrogen
Oxygen
3 4 5
Mole Percentage Impurity
observation of Fig.6, it was concluded by Maxey [20] and
Cosham and Eiber [22] that an estimate of the toughness
requirement for arresting a ductile fracture in a CO2 pipeline
could be obtained from the calculation of the arrest pressure2
and the saturation pressure3; in other words, in order to
arrest a ductile fracture, the arrest pressure must be greater
than the saturation pressure and either the arrest pressure
must be raised or the saturation pressure must be lowered.
Maxey [20] indicates that the saturation pressure can
be lowered by lowering the operating temperature or by
removing impurities with lower critical temperatures than
CO2. Figure 2 illustrates the impurities with lower critical
temperatures than CO2: the effect of initial temperature,
pressure, and impurity levels is currently being investigated
by the authors using the BTCM decompression model [25].
This work is ongoing, but a sample calculation at the starting
decompression conditions of 100bara and 5°C is presented
in Fig.7. This preliminary work confirms the result of other
researchers that H2 in particular has the most potent effect,
in terms of mol% addition, in raising the saturation pressure
above pure CO2 and therefore the largest detrimental effect
on fracture propagation [17]. At the pressure and temperature
condition presented in Fig.7 it can, however, been seen that
the SO2 would have a beneficial effect.
2. The arrest pressure is defined as the pressure below which a propagating ductile
fracture cannot be sustained.
3. The pressure at which the fluid crosses the two-phase boundary, i.e. in the case of
pure CO2 decompressing from the dense phase, the pressure at which fluid changes
to the gaseous phase.
3rd Quarter, 2012
Conversely, it has been shown that the arrest pressure can
be raised by increasing the wall thickness, increasing the
toughness, decreasing the pipe diameter, or increasing
the pipe material yield strength [18]. All of these measures
have economic implications and must also be balanced
against the hydraulic requirements mentioned previously
(i.e. decreasing the diameter of the pipeline will increase
the pressure drop and reduce the hydraulic efficiency). In
addition, wall thickness, yield strength, and toughness all
have practical upper limits in terms of the capabilities of steel
and pipe manufacturers. The balance therefore has to be
made between the lowering of the saturation pressure (and
the resultant costs on the CO2 purification) and the raising
of the arrest pressure (and the resultant costs on pipeline
materials). However, one conclusion that can be drawn is
that, for certain pipe diameters and wall thicknesses, the
CO2 may need to be specified to ensure fracture arrest, as
the required levels of toughness and yield strength cannot
be achieved. If this specification cannot be attained then
mechanical crack arrestors would need to be installed along
the pipeline, as is the practice on some pipelines in the USA
[21, 26] at additional cost of construction for the pipeline.
n
CO2 specification to prevent
io
material degradation
st y
ut
di op
Material degradation in terms of internal corrosion or
rib
cracking mechanisms will only occur in the presence of water.
or c
Therefore, prior to a discussion on the effect of impurities
on corrosion and cracking mechanisms, it is first necessary
t f ple
to discuss the specification of the maximum water content
allowable in the CO2 stream to avoid water formation in
the pipeline.
no m
Water specification
Sa
The solubility of water in pure CO2 has been studied
extensively as a function of temperature and pressure and is
discussed in the context of CO2 pipeline transportation in
Mohitpour et al. [8]. However, the effect of impurities on the
solubility of water in CO2 containing impurities is less well
researched. Seiersten and Kongshaug [27], Heggum et al. [28],
and Austegaard et al. [29] have studied the CO2-H2O-CH4
system and their experimental results and models indicate
that the solubility of water in the system decreases with
increasing CH4 content, i.e. adding CH4 to the system will
require the water content to be more stringently specified
to avoid free water formation. Conversely, on the basis of
calculations alone, for the CO2-H2O-H2S system at 4°C and
100bar, de Visser and Hendriks [9] found that the addition
of H2S would have the opposite effect and would increase
the solubility of water, although the effect was found to
be negligible at the levels of H2S proposed in the Dynamis
specification.
The specification of water content in currently operating
pipelines has been reviewed by Seevam et al. [24] and ranges
between 640ppmv (parts per million by volume) in the
185
KMCO2 pipelines and 20ppmv in the Weyburn pipeline.
Both the Dynamis and Ecofys projects recommend a water
content of 500ppmv to ensure that no free water is present in
the pipeline and therefore to minimize the risk of corrosion
and hydrate formation [9,10]. It is noted, however, that the
Dynamis project concluded that this water level should be
reviewed if the operating temperature and pressure of the
pipeline could result in operation below the solubility limit
for water in CO2 or if other impurities, not considered in
the Dynamis report, were present in the CO2 stream.
Despite the lack of data, the conclusion that can be drawn
from these observations is that the drying specification for
CO2 containing impurities is dependent on the type and
quantity of impurities present and that the effect of different
impurities on the saturation limit needs to be understood
in order to be able to ensure dry conditions in the pipeline.
Hydrate formation
Apart from potential material degradation, another
consideration in the specification of the water content in the
CO2 stream is the risk of hydrate formation. The published
literature in this area becomes confusing as some authors
indicate that free water is required for hydrates to form [30],
whereas others indicate that hydrates can form with dissolved
water [10, 31]. It has been stated that, under CO2 pipeline
operating pressures, it would be possible for hydrates to form
at around 10-11°C [30, 32]. However, it was considered by
the Dynamis project that the amount of hydrate that could be
formed with the levels of water required to prevent corrosion
will not be sufficient to cause pipeline operational problems
[9]. Chapoy et al. [33], from their work on hydrate formation
and two-phase flow, have concluded that in pure CO2 systems,
it is unlikely that hydrate will form in the temperature range
-2°C to 30°C and for pressures up to 200bara provided that
the water content is less than 250ppm.
There has been little work on the effect of impurities on
the CO2-H2O phase behaviour although work by Chapoy
et al. [33] on the CO2-H2 and CO2-H2-N2-CO systems in
the presence of water indicates that higher dehydration
requirements might be necessary in the presence of H2 to
prevent hydrate formation.
Corrosion
Pure, dry CO2 is not corrosive to carbon or stainless steel.
However, in the presence of water, CO2 can dissolve and
form a concentrated carbonic acid (H2CO3) solution which is
highly corrosive to carbon steel. This form of CO2 corrosion
is termed ‘sweet’ corrosion and is well understood in the oil
and gas industry where the combination of CO2 in the product
and free water can cause internal pipeline corrosion. As a
result there is an extensive volume of literature published on
the mechanism, influencing parameters, and prediction of
sweet corrosion rates under the conditions of temperature,
pressure, CO2 concentration, and flow rates relevant to the
186
300
250
200
Concentration of H2S (ppmv)
150
100
50
n
0 20 40 60 80
io
Pipeline pressure (bar)
st y
ut
di op
Fig.8. Concentration of H2S required to prevent sour cracking.
rib
or c
oil and gas industry. Less work has been published on sweet
corrosion in CO2 pipelines, particularly in the presence of
t f ple
impurities. In CO2 pipelines there are two scenarios whereby
corrosion could occur; due to the presence of water in a
water-rich phase, and due to the presence of water dissolved
no m
in a CO2-rich phase.
Sa
If the level of water is above the saturation limit at the
operating temperature and pressure, then water can drop-out
of solution as free water. Water could also enter the pipeline
due to an upset in the dehydration equipment. Recent work
on CO2 corrosion relevant to the transportation of CO2
for CCS schemes indicates that high rates of corrosion (up
to 20mm/yr) could be observed in the water-rich phase in
equilibrium with pure CO2 at high pressure and temperature
[34]. In the CO2-rich phase, water is dissolved in the CO2 and
the corrosion rates reported were much lower and typically
around 0.2mm/year. However, work conducted by McGrail,
et al. [35] suggests that corrosion can occur in the CO2-rich
phase at water levels below the water-solubility limit at the test
temperature and pressure, suggesting that a separate water
phase is not necessary for corrosion to occur. From their
work it can be concluded that a threshold level did appear
to exist for the onset of corrosion, which has implications for
the specification of the water content to prevent corrosion.
There has been little experimental work conducted on the
effect of impurities on corrosion rates; however, the principle
impurities that might be considered to have an effect are
H2S, O2, SO2, and NO2.
The Journal of Pipeline Engineering
100 120 140 160 180 200
From work conducted for the hydrocarbons industry, the
influence of small amounts of H2S (i.e. less than 0.0035
bar partial pressure H2S) on CO2 corrosion is dependent
on the concentration of H2S and whether iron sulphide
can precipitate as a protective surface scale. As the H2S
concentration increases, the formation of the iron sulphide
surface film becomes more favourable and can reduce the
rate of corrosion. McGrail et al. [35] indicate that one of
the effects of H2S was to promote corrosion in the CO2-
rich phase at lower water levels than seen with pure CO2.
Another potential issue with H2S and the formation of iron
sulphide films is reported by Wong [36] in compression
equipment. If the iron sulphide film becomes dislodged
it can coat the stainless steel aerial coolers resulting in a
decrease in compressor efficiency.
There is limited experimental evidence to indicate that the
addition of O2, SO2, and NO2 in ternary combinations
with CO2 and H2O or mixtures of O2-SO2-CO2-H2O
increase the corrosion rate above those observed with pure
CO2 in the CO2- rich phase under the same experimental
conditions [37-39]. Although none of these works indicate
acceptable levels of these components from a corrosion
point of view, Xiang et al. [38] recommend that the
specification of the water content should be reviewed
when SO2 is present. However, Ayello [39] indicates
that, in terms of corrosion, NO2 could be more effective
than SO2 in increasing corrosion rates. Consequently,
the influence of NO2 on the water specification could
also be crucial.
3rd Quarter, 2012
The levels of SOx and NOx in the CO2 stream have recently
been reviewed by Santos and Yan [40] as it has been reported
that, during the compression process, SO2 and NO can
react with water and O2 respectively to form sulphuric and
nitric acid, causing corrosion problems in the compressor.
Work reported by these authors indicates that this reaction
could occur at very low concentrations, i.e. below those
specified by Dynamis; however, no limits are specified.
CO2 specification to prevent cracking
H2S (sour) cracking
In CO2 environments containing partial pressures of H2S
greater than 0.003bar, as found in the oil and gas industry,
the mechanism of corrosion described in the previous
section changes from one of general or pitting corrosion
to cracking, and the product is termed ‘sour’. This limit
has been defined in the standard BS EN ISO 15156-2
[41] to prevent sour cracking in H2S environments for oil
and gas production. The methodology outlined in that
standard has been applied in the current study to CO2
n
pipelines and the effect on the specification of H2S levels
at different pipeline pressures is illustrated in Fig.8. This
io
figure indicates that the specification of H2S is dependent
st y
ut
on system pressure, and that raising the pressure reduces
di op
the allowable levels of H2S. This is significant when
rib
considered against the other implications discussed in this
or c
paper regarding the raising of system pressure on hydraulic
efficiency and fracture propagation.
t f ple
If a pipeline is going to be operated in sour service (i.e. above
the levels in Fig.8), sour-resistant steel has to be selected
no m
to prevent rapid failure, as failure times as short as days or
hours have been observed for sour-corrosion mechanisms
Sa
under test conditions in oil and gas environments. High-
strength steels are more susceptible to cracking in a sour
environment, and consequently steels have to be selected
to comply with maximum hardness levels to avoid cracking
in both the parent and the weld material. As well as the
additional costs associated with specifying sour-resistant
steel, there are additional requirements on welding and
inspection when operating a sour pipeline which will
increase pipeline construction and maintenance costs and
need to be taken into consideration in the cost-benefit
analysis associated with reducing levels of H2S and/or
ensuring that the CO2 stream is completely dry.
CO cracking
Another contaminant that may pose a threat to pipeline
steels in terms of cracking is CO. This type of cracking has
been identified as a potential risk in CO-CO2 environments
in carbon steels [42]. There is very little published research
in this area and therefore the level of impurity required
to cause CO2-CO cracking under pipeline operating
conditions is not yet known.
187
CO2 specification for health and safety
As mentioned previously, the Dynamis specification sets the
levels of H2S, CO, NOx, and SOx using an approach based
on the STELs, which defines the maximum concentrations
for H2S, CO, NOx, and SOx in the CO2 stream based on
the maximum amount of these components to which a
person can be exposed for a period of 15 minutes without
adverse effects. The approach outlined by de Visser and
Hendriks [9] assumes that there is no additional toxic effect
on people due to the combination of the components,
which could increase the toxicity of the resultant stream
and may require further investigation.
Defining CO2 specifications
for pipeline transportation
Although the approach taken in this paper has been to
consider each aspect relating to the pipeline specification
individually, one of the principal conclusions that is drawn
from this analysis is that the specification of CO2 must be
considered as a whole, as any changes to the specification
to improve one aspect of pipeline design and operation
will affect another aspect. For example, increasing NOx
levels to improve operational flexibility for dense-phase
pipelines will adversely affect corrosion in the event of
water being present in the pipeline, and there are also
safety concerns in the event of a release.
In addition, the results presented have indicated that the
pressure and temperature of the system is also important in
defining a pipeline specification. For example, increasing
the pressure of the system to accommodate mixtures with
higher critical pressures will have an impact on the cost
of the pipeline system and also on the compression costs.
In addition, at higher system pressures, the specification
of the maximum-allowable levels of H2S to prevent sour
cracking would also have to be reduced, adding additional
costs at the capture plant.
Therefore in defining a specification, for some impurities,
the specification will be driven by a cost-benefit analysis
to balance the additional costs required to attain required
inlet temperature and pressure conditions and to specify
larger-diameter or thicker wall thickness pipe against the
costs of improving the purity of the CO2 stream. However,
for other impurities, the specification will be driven by the
requirement to maintain the integrity of the pipeline and
prevent a release of CO2.
This concept is illustrated in the summary in Table 5. From
this table, it is important to highlight that some of the
drivers for the specification presented here (such as fracture
control and sour cracking) have not been considered in
previous pipeline specifications [9]. However, the analysis
presented here indicates that these aspects may drive the
specification to lower limits than have been previously
specified.
188 The Journal of Pipeline Engineering

Hydraulics Hydraulics
Fracture control Water solubility
(dense phase) (gas phase)

H2O Not studied

Reduces saturation
SOx
pressure Decreases water
Increases critical Reduces operating solubility
NOx Effect unknown
pressure pressure
Increases water
H2S Effect unknown
solubility

CO
Increases critical
H2 pressure
Reduces
Ar pipeline capacity Effect unknown
Increases Increases operating Increases saturation
pressure drop pressure pressure
N2
Increases power
O2 requirements

n
io
Decreases water
CH4
st y
ut solubility
di op
rib
Corrosion Cracking Hydrate formation Health and safety
or c

Promotes hydrate
H2O Promotes cracking
t f ple

formation

SOx Promotes corrosion

no m

Effect unknown
NOx
Sa

Toxic in event of
Effect unknown
release
H2S Effect unknown
Promotes cracking
CO

Promotes hydrate
H2 Effect unknown
formation
Not studied
Ar Effect unknown

Promotes hydrate
N2 Effect unknown
formation
Not studied
O2 Promotes corrosion
Effect unknown
CH4 Not studied

Table 5. Summary of effects of different impurities.

3rd Quarter, 2012
From Table 5, it is considered that the specification of
impurity levels for hydraulic analysis and fracture control
is primarily driven by a cost-benefit analysis. However, as
highlighted in this paper, the specification for fracture
control does become limited by available pipe dimensions
and material properties. In this respect, H2 and N2 have the
greatest effect on saturation pressure and should be limited
in this situation.
It is also considered that the specification of the water level
is critical in driving the specification of impurities such as
NOx, SOx, H2S, O2, and CO. The mechanisms of corrosion
and cracking discussed in this paper will only occur in the
presence of water, whether that is dissolved in the CO2 or as
a separate liquid phase, and therefore in normal operation it
is critical to prevent corrosion and cracking by limiting the
water content. In the event of an upset condition developing,
where water could enter the pipeline, it is important to limit
the extent of corrosion (through limiting the quantities
of NOx, SOx, and O2) and prevent cracking by limiting
the concentrations of H2S and CO to below the cracking
thresholds. A cautious approach to the specification of water
n
would therefore seem prudent in the early stages of CCS
transportation until further work has been conducted which
io
would allow a relaxation of these limits.
st y
ut
di op
Conclusion
rib
or c
The effect of the impurities which may be present in the
CO2 stream on pipeline design, operation, integrity, and
t f ple
health and safety has been considered in this paper. The
work has demonstrated that, in order to define a pipeline
specification, these requirements have to be considered
no m
as a whole. In particular, it has been shown that fracture
control and cracking mechanisms will play a key role in
Sa
limiting the allowable concentrations of certain components.
These aspects have not been considered previously in the
published literature.
The review conducted in this paper has also indicated that
the majority of the work conducted to date on issues related
to the CO2 specification has only considered the effects
of individual impurities. There has been little research
published on the combined effects from different impurities
in the CO2 stream. In conclusion, there remain a number
of uncertainties which require further investigation in order
to define a CO2 specification which maintains safety and
integrity whilst ensuring cost-efficient design.
Acknowledgements
This work has been conducted under the auspices of the
MATTRAN project (Materials for Next Generation CO2
Transport Systems) and the authors gratefully acknowledge
the financial support of EPSRC and E.ON for this research
(E.ON-EPSRC Grant Reference EP/G061955/1). Additional
work has also been conducted under the National Grid
COOLTRANS research programme.
189
References
1. Kinder Morgan, 2006. CO2 transportation. World
Resources Institute. 8th February. www.wri.org/
powerpoints/ccs_martin.pdf (accessed May 27, 2012).
2. G.W.Marsden and D.Wolter, 1986. Mass custody-transfer
measurement technique. Oil and Gas Journal, November,
pp74-78.
3. IPCC, 2005. IPCC special report on carbon dioxide capture
and storage. Working Group III of the Intergovernmental
Panel on Climate Change, Eds: B.Metz, O.Davidson,
H.C.de Coninck, M.Loos, and L.A.Meyer, Cambridge
University Press.
4. M.Anheden, A.Andersson, C.Bernstone, S.Eriksson,
S.Liljemark, and C.Wall, 2005. CO2 quality requirement
for a system with CO2 capture, transport and storage. Proc.
7th Greenhouse Gas Technologies Conference GHGT-7,
Vancouver, pp2259-2563.
5. T.Oosterkamp and J.Ramsen, 2008. State of the art review
of CO2 pipeline transportation with relevance to offshore
pipelines. Norway: Report No. UCD-ITS-RR-06-14,
Polytec.
6. IEAGHG, 2011. Effects of impurities on geological storage
of CO2. 2011/04.
7. A.Kather and S.Kownatzki, 2011. Assessment of the
different parameters affecting the CO2 purity from coal
fired oxyfuel process. Int. J. of Greenhouse Gas Control, 55,
S204-S209.
8. M.Mohitpour, P.N.Seevam, K.K.Botros, B.Rothwell, and
C.Ennis, 2012. Pipeline transportation of carbon dioxide
containing impurities. American Society of Mechanical
Engineers, New York.
9. E.de Visser and C.Hendriks, 2007. Towards hydrogen and
electricity production with carbon capture and storage -
DYNAMIS quality recommendations. DYNAMIS.
10. C.Hendriks, S.Hagedoorn, and H.Warmenhoven, 2007.
Transportation of carbon dioxide and organisational
iIssues of CCS in the Netherlands. Report prepared by
Ecofys for EnergieNed, the Ministry of Economic Affairs
and the Ministry of Housing, Spatial Planning and the
Environment.
11. J.Yan, M.Anheden, C.Liljemark, S.Bernstone, H.Pettersson,
D.Simonsoon, and H.Li, 2008. Impacts on non-
condensable components on CCS. IEA CO2 Specification
Working Group Meeting. Stockholm, October.
12. C.Farris, 1983. Unusual design factors for supercritical
CO2 pipelines. Energy Progress 3, no. 3, pp150-158.
13. P.N.Seevam, J.M.Race, M.J.Downie, J.Barnett, and
R.Cooper, 2010. Capturing carbon dioxide: the feasibility of
re-using existing infrastructure to transport anthropogenic
CO2. Proc. 8th Int. Pipeline Conf., Calgary, Alberta,
Canada: American Society of Mechanical Engineers.
14. Total, 2007. Project information dossier: Lacq CO2 capture
and geological storage pilot project. October, www.total.
com (accessed May 25, 2012).
15. P.N.Seevam, J.M.Race, and M.J.Downie. 2007. Carbon
dioxide pipelines for sequestration in the UK - engineering
gap analysis. Conf. on transmission of CO2, H2, and biogas:
190
exploring new uses for natural gas pipelines. Clarion and
Scientific Surveys, Amsterdam, pp141-162.
16. P.N.Seevam, J.M.Race, M.J.Downie, and P.Hopkins, 2008.
Transporting the next generation of CO2 for carbon capture
and storage: the impact of impurities on supercritical CO2
pipelines. Proc. 9th Int. Pipeline Conf., Calgary, Alberta:
American Society of Mechanical Engineers.
17. G.G.King, 1982. Here are key design considerations for
CO2 pipelines. Oil and Gas Journal, 80, 39, pp219-222.
18. Idem, 1982. CO2 pipeline design - fractures, pipe sizing
probed for high operating pressure CO2 pipelines. Ibid.,
80, 40, pp100-102.
19. L.C.Decker, J. R.Checak, W.H.Heath, and R.N.Warren,
1985. CO2 pipeline design, construction and operation.
Facilities, pipelines, and measurements: a workbook for
engineers. Proc. 41st Petroleum Mechanical Engineering
Workshop and Conference, Kansas City, MO, USA.
20. W.A.Maxey, 1986. Long shear fractures in CO2 lines
controlled by regulating saturation arrest pressures. Oil
and Gas Journal, 84, 31, pp44-46.
21. D.L.Marsili and G.R.Stevick. Reducing the risk of ductile
failure on the Canyon Reef Carriers CO2 pipeline. 65th
n
Annual Tech. Conf. of the Society of Petroleum Engineers,
New Orleans, SPE, pp311-319.
io
22. A.Cosham and R.J.Eiber, 2007. Fracture control in CO2
st y
ut
pipelines. Proc. Conf. on Transmission of CO2, H2, and
di op
biogas: exploring new uses for natural gas pipelines. Clarion
rib
and Scientific Surveys, Amsterdam.
or c
23. 49CFR195, 2008. Title 49 Code of Federal Regulation
Part 195: Transportation of hazardous liquid and carbon
t f ple
dioxide by pipeline. http://ecfr.gpoaccess.gov.
24. P.N.Seevam, J.M.Race, and M.J.Downie, 2010.
Infrastructure and pipeline technology for carbon dioxide
no m
(CO2) transport. In: Developments and innovation in
carbon dioxide (CO2) capture and storage technology, Ed.
Sa
M.M.Maroto-Valer, 408-434. Woodhead Publishing Ltd,
Cambidge.
25. W.A.Maxey, 1974. Fracture initiation, propagation and
arrest. Paper J, Proc. 5th Symposium on Line Pipe Research.
Houston, Texas, USA. Pipeline Research Committee of
the American Gas Association.
26. D.E.McCollough, 1986. The Central Basin pipeline: a
CO2 system in West Texas. Energy Progress, 6, 4, December,
pp230-234.
27. M.Seiersten and K.O.Kongshaug, 2005. Materials selection
for capture, compression, transport and injection of CO2.
Vol.2, In: Carbon dioxide capture for storage in deep
geological formations, Eds: D.C.Thomas and S.M.Benson,
pp937-953. Elsevier.
28. G.Heggum, T.Weydahl, R.Mo, M.Molnvik, and
A.Austegaard, 2005. CO2 conditioning and transportation.
Vol. 2. In: Carbon dioxide capture for storage in deep
geological formations, Eds: D.Thomas and S.M.Benson,
pp925-936. Elsevier Ltd.
29. A.Austegaard, E.Solbraa, G.de Koeijer, and M.J.Molnvik,
2006. Thermodynamic models for calculating mutual
solubilities in H2O-CO2-CH4 mixtures. Chemical Engineering
Research, 84, no. A9 (September): pp781-794.
The Journal of Pipeline Engineering
30. C.B.Wallace, 1986. Drying supercritical CO2 demands
care. Oil and Gas Journal, 6, 3, p155.
31. J.J.Carroll, 2008. An introduction to gas hydrates.
FlowPhase Inc. http://www.telusplanet.net/public/
jcarroll/HYDR.HTM (accessed March 19, 2011).
32. A.Fradet, S.Saysset, P.Odru, P.Ruer, J.Broutin, and
M.Bonnissel, 2008. Technical and economic assessment
of CO2 transportation for CCS purposes. J. Pipeline Eng.,
6, 3, pp173-180.
33. A.Chapoy, R.Burgass, B.Tohidi, J.M. ustell, and C.
Eickhoff, 2009. Effect of common impurities on the phase
behaviour of carbon dioxide rich systems: minimizing
the risk of hydrate formation and two phase flow. SPE
Offshore Europe Oil and Gas Conf., 8-11 September,
Society of Petroleum Engineers.
34. Y.-S.Choi and S.Nesic, 2010. Determining the corrosive
potential of CO2 transport pipeline in high pCO2-water
environments. Int. J. of Greenhouse Gas Control, 5, 788-797.
35. B.P.McGrail, H.T.Schaef, V.-A.Glezakou, L.X.Dang,
and A.T.Owen, 2009. Water reactivity in the liquid and
supercritical CO2 phase: has half the story been neglected?
Energy Procedia, 1, pp3415-3419.
36. S.Wong, 2005. Module 4 – CO2 compression &
transportation to storage reservoir. Building capacity for
CO2 capture and storage project in the APEC region:
a training manual for policymakers and practitioners.
APEC Reference #205-RE-01.3. www.delphi.ca/apec.
37. Y.-S.Choi, S.Nesic, and D.Young, 2010. Effect of
impurities on the corrosion behavior of CO2 transmission
pipeline steel in supercritcal CO2-water environments.
Environmental Science Technology, 44, pp9233-9238.
38. Y.Xiang, Z.Wang, X.Xu, C.Zhou, Z.Li, and W.Ni, 2011.
Impact of SO2 concentration on the corrosion rate
of X70 steel and iron in water-saturated supercritical
CO2 mixed with SO2. Journal of Supercritical Fluids, 58,
pp286-294.
39. T.Ayello, 2009. Materials performance in supercritical CO2
environments. Corrosion/2009. NACE International.
40. S.Santos and J.Yan, 2009. CO2 processing unit; challenges
in meeting the required CO2 quality. Oxyfuel Combustion
Network - 2nd Working Group on CO2 quality and other
relevant issues, 7th September, Cottbus, Germany.
41. BS EN ISO 15156-2, 2009. Petroleum and natural
gas industries - Materials for use in H2S-containing
environments in oil and gas production Part 2: Cracking-
resistant carbon and low-alloy steels, and the use of cast
irons, ISO 15156-2:2009, British Standards Institute.
42. W.E.Berry and J.H.Payer, 1979. Internal stress corrosion
cracking by aqueous solutions of CO and CO2. Proc.
Annual Symposium - Society of Flight Test Engineers,
Z1-Z19.
43. J.Gale and J.Davison, 2004. Transmission of CO2 - safety
and economic considerations. Energy, 29, 9-10, pp1319-
1328.
44. APGTF, 2003. Carbon dioxide capture and storage:
Report of DTI international technology service mission
to the USA and Canada. Advanced Power Generation
Task Force.
3rd Quarter, 2012 191

The application of individual

and societal risk assessment
to CO2 pipelines
by Dr Phil Cleaver*1 and Harry Hopkins2
1 GL Noble Denton, Loughborough, UK
2 HH Risk, Solihull, UK

T his paper starts by giving a summary of the historical background and development of guidance for
the control of risks of possible releases from pipelines transporting flammable substances in the UK.
The application of this guidance to CO2 pipelines is then considered. It is noted that the application of
the accepted individual-risk approach for a natural gas pipeline is likely to be cautious when applied to a
gaseous-phase CO2 pipeline, but does not pose any significant problems in its application. In contrast, it
is shown that because dense-phase CO2 pipelines are likely to be thick-walled, the individual risk levels
around these pipelines are likely to be low, but because of the great variability of atmospheric dispersion

n
of dense gas clouds, the risk levels are likely to decay very slowly beyond about 200m from the pipeline.
This poses something of a problem, as an approach based on individual risk would suggest that only small

io
separation distances between the pipeline and occupied buildings are required and yet the maximum hazard
st y
ut
distances if an event occurred could be considerably larger. Such a difference does not arise for natural
di op

gas pipelines, as the risk transects decay more rapidly with distance for a flammable hazard. It is suggested
rib
that societal risk evaluation using the criterion curve in PD8010: Part 3 offers a way of resolving this issue.
or c

Because such an approach is time-consuming and may not be warranted in every case, a screening method
is suggested to use to help in routeing and design studies. In order to take into account the nature of the
t f ple

risk transect for CO2, the screening approach provides different initial separation distances from cities,
towns and isolated villages.
no m

P ipeline routeing for transporting numbers and location of any population groups in the
Sa

flammable substances has the aim of locating the
most cost-effective route option while satisfying the safety
requirements to protect both the general public and the
pipeline, and minimizing the impact on the environment and
on owners, occupiers, and authorities who will be required
to accommodate the pipeline.
Once a need for a pipeline has been determined, the route-
selection process is based on choosing the most suitable
corridor, taking into account areas which must be avoided,
such as historic sites, environmentally or ecologically sensitive
sites, difficult terrain, and centres of population; engineering
issues related to crossings or unstable ground; and the location
of installations, such as intermediate block valve installations.
When the route corridor has been selected, the location
of the pipeline within the corridor takes account of the
vicinity of the pipeline with the aim of minimizing both
the number of people within the hazard zone and, for
the pipeline design, the need to use thick-walled pipe.
This aspect of the detailed routeing has been embodied
in UK pipeline codes and standards, primarily applied
to natural gas, for more than 30 years. Guidance on
managing the risk, both individual and societal, is based
on a consideration of the separation distance from the
pipeline to the nearest occupied building, to provide
protection from more-credible events such as a leak,
and a reduction in risk level from larger, less-credible,
events; and by limiting the design factor to reduce the
likelihood of the more-serious failure modes where there
are numbers of people at risk. These requirements are
implemented in straightforward rules which are applied
in the pipeline design.

The purpose of this paper is to summarize and review

This paper was presented at the Third International Forum on the Transportation of
CO2 by Pipeline, organized by Newcastle University, Tiratsoo Technical, and Clarion
the guidance for flammable substances and to consider
Technical Conferences, and held in Newcastle in June, 2012. the application of this information to CO2 pipelines. To
this end, the background to the current regulations and
*Corresponding author’s contact details:
tel: +44 (0)1509 282426 guidance is discussed initially. This covers the historical
email: phil.cleaver@gl-group.com
192
development of risk guidance and its application to the
transportation of natural gas and other fluids by pipelines.
Next, the application of this approach for dense-phase
CO2 pipelines is discussed, and the use of risk assessment
is illustrated by means of an example. The results are
discussed subsequently, and an approach to a screening
assessment to address the societal risk impact of the
pipeline is put forward thereafter. This is proposed in a
form that can be used in the early stages of routeing to
ensure that the risks are controlled1. The final section
of the paper discusses the findings and conclusions that
have been drawn from this work.
Historical background
Natural gas
Early guidance on high-pressure natural gas pipeline
design and construction in the UK was based on the US
Code ASME B31.8 [1]. Detailed guidance on routeing,
taking account of UK conditions and in the form still
used today, was first published in the Institution of
n
Gas Engineers’ requirements for high pressure natural
gas pipelines, IGE/TD/1 in 1970 [2]. Since then, this
io
guidance has been progressively developed, based on
st y
ut
experimental work and operational data on the causes
di op
of pipeline failure and on the consequences of failure,
rib
taking advantage of the availability of thicker-walled pipe
or c
of adequate quality. These developments allowed higher
operating pressures in populated areas on the basis of
t f ple
increasing the pipe wall thickness to limit the design factor
to 0.3. This control of design factor at higher pressures in
populated (Suburban or type-S) areas enabled operation
no m
at these higher pressures in Rural (type-R) areas at up to
0.72 design factor.
Sa
The separation (or proximity) distance, later referred to
as building-proximity distance (or BPD), from occupied
buildings in type-R areas was intended to protect the
population from the hazard of more-credible failure modes
(leaks or punctures) and provide a degree of control of
the risk from less-credible events (breaks or ruptures).
For type-S areas, defined at that time by the population
density in a corridor five times the type-R area separation
distance on each side, the design factor was controlled
such that only leaks would then need to be considered
when defining the separation distance. Together with a
procedure to determine the extent of any type-S area, a
set of simple workable rules were available to set out the
route for new pipelines, to determine any infringements
which may need attention, and to identify any planned or
anticipated land-use developments which would conflict
with the design code.
1. Note: the application of this approach is illustrated using one particular set of
consequence models. However, the approach is generic and is independent of the
particular software used, as long as it is able to meet certain minimum standards.
A formal protocol that can be used to evaluate the consequence models and assess
their adequacy is discussed further by Gant [31].
The Journal of Pipeline Engineering
Other fluids
Pipelines laid under the Pipelines Act 1962, as opposed to
natural gas pipelines laid under the various Gas Acts, were
covered by the British Standard Institute’s (BSI) code of
practice, BS CP 2010: Part 1: 1966 [3], and later BS 8010:
Part 1: 1989 and Part 2, Section 2.8 1992 [4]. The approach
was consistent with that in IGE/TD/1 and for methane the
requirements were identical with IGE/TD/1 Edition 2 [5].
The separation distance was based on an equation matching
IGE/TD/1 for methane and applying a substance factor
for other fluids. The population-density corridor was three
times the separation distance for category-C fluids (non-
flammable and including carbon dioxide), or four times
for category-D fluids (flammable or toxic). A risk analysis,
as part of a safety evaluation of the pipeline, could be used
to justify operation outside the prescriptive limits.
Land-use planning
The UK Health and Safety Commission (HSC) Advisory
Committee on Major Hazards (ACMH), set up after the
Flixborough incident in 1974, produced a series of three
reports. The first of these [6] identified the potential for
pipelines to be a major hazard; the second report [7] assessed
the hazard, and as a result certain pipelines were included
in the Notification of Installations Handling Hazardous
Substances (NIHHS) Regulations 1982 [8]. However
pipelines were not included in the Seveso Directive 1982
[9] and hence were not included in the subsequent Control
of Industrial Major Accident Hazard (CIMAH) Regulations
1984 [10]. The second report also recommended that
planning authorities be required by statute to consult on
planning applications for notifiable installations and for
any developments in their vicinity.
While pipelines were not classed as notifiable, there was a
requirement for the Health and Safety Executive (HSE) to
develop ‘general development control’ (land-use planning)
advice in the vicinity of high-pressure natural gas pipelines.
An assessment of the hazards was undertaken [11] which
related the predicted hazard range from a pipeline failure to
multiples of the IGE/TD/1 proximity distance BPD. For the
larger-diameter rural pipelines the hazard range extended to
around four times the BPD distance, and it was agreed that
this provided a good basis for planning advice. Guidance was
produced in the form of a matrix of development type, defined
in terms of sensitivity (type and numbers of occupants),
against multiples of BPD, used to define inner, middle,
and outer zones [12]. This subdivision of development
types was not taken into account when routeing a pipeline
to IGE/TD/1 although operators may separately account
for certain development types. The risk of failure was taken
note of but was not explicitly incorporated into the advice
policy and hence the policy, in line with the BPD approach,
was generally hazard based. As noted above, BS 8010: Part
2.8: 1992 used four times the separation distance for gases
other than methane, so following the hazard assessment, the
3rd Quarter, 2012
corridor width was reduced to 4BPD on either side of the
pipeline in the Third Edition of IGE/TD/1 in 1993 [13];
BS PD 8010: Part 1: 2004 [14] also reduced this distance to
four times the separation distance for methane.
Planning guidance for other pipeline fluids followed the
approach used for fixed installations, being largely based
on the hazard range and then progressively introducing
a more risk-based approach, particularly for toxic fluids
where the hazard range is large but the likelihood of the
release small [15].
Acceptable risk levels
Throughout the period covered here the development of a
risk-assessment methodology and approaches to acceptable
risk was progressed by industry, the HSE, and elsewhere,
including the Royal Society Study Group reports in 1983 [16]
and 1992 [17]. Outcomes included the use of a ‘three-band’
individual risk-based approach in which an individual risk
of one in a million per year (1 x 10-6/year) is regarded as a
generally accepted level. It was also a major topic addressed
in the Public Inquiries in 1982 into the Canvey Island
n
installations and in 1983-85 into the Sizewell B nuclear
power station. The Sizewell Inquiry recommended that the
io
HSE publish guidance on the acceptable risk from nuclear
st y
ut
power stations. This led to the publication in 1988 (and
di op
1992) of the Tolerability of risk from nuclear power stations
rib
document (TOR) [18] which introduced the three-band
or c
‘ALARP triangle’ approach to acceptable individual risk
of death, which was also considered to be applicable to
t f ple
non-nuclear activities. This was later extended to include
‘societal concern’ in Reducing risk, protecting people (R2P2)
in 1999 [19].
no m
Responding to the ACMH reports, the general development
Sa
of risk assessment and the publication of TOR, the HSE
published Risk criteria for land use planning in the vicinity of
major industrial hazards (RCD) in 1989 [20]. Taking note of
TOR, individual risk levels of 10, 1, and 0.3cpm (chances
per million per year or 10-6/year) were selected to define the
inner, middle, and outer zones within which the societal
risk to different types and numbers of people would be
controlled. ‘Dangerous dose or worse’ was used as the harm
criterion as a more-easily addressed predictive measure
than the historical risk of death used in TOR. The 10 and
0.3cpm risk levels were then included in PD 8010 Part1
(which replaced BS 8010) to define the separation distance
and population density corridor distances respectively if
the quantified risk assessment (QRA) option was chosen
for pipeline routeing.
The Health and Safety Commission’s Advisory Committee
on Dangerous Substances (ACDS) study into transport risks
[21] concluded that the use of individual risk as a surrogate
societal risk measure was not appropriate to transport routes,
and therefore used societal risk calculations presented as FN
curves to represent the risks assessed (that is, it produced
curves showing the cumulative frequency F with which N or
193
more casualties are produced). The UK natural gas industry
also selected societal risk and FN curves as the measure for
acceptable pipeline operation in the UK, and used this
approach to define an FN-based acceptability envelope based
on operation to IGE/TD/1. This was first widely published
in 1999 [22] then incorporated as an approach in Edition
4 of IGE/TD/1 [23] and subsequently in IGEM/TD/1:
Edition 5 [24] and IGEM/TD/2 [25], where the approach
for QRA of natural gas pipelines was set out.
For flammable fluids other than natural gas, a similar
approach to IGEM/TD/2 was followed and published as
PD 8010: Part 3: 2004 [26]. In this case, the acceptable
FN curve was developed based on the approach for fixed
installations, using the Canvey Island Inquiry acceptable
FN point, quoted in R2P2, and a slope of -1 [27] together
with an equivalence of pipeline length to a typical fixed
major-hazard installation.
In summary, therefore, the ability to use QRA for pipeline
routeing was recognized in both BS 8010 and Edition 3 of
IGE/TD/1, and by the time of Edition 4 of IGE/TD/1
and PD 8010 both methodology and criteria were available.
In practice, however, it was generally more straightforward
to route using the prescriptive rules in the codes and to use
QRA for the assessment of any non-standard locations or
to address issues arising once the pipeline was in service. In
part this was because the development of both code rules
and QRA were based on the same research and operating
experience. Using failure frequencies based on historical
failure and damage data from acceptable operation in
predictive models such as the current industry model
FFREQ, [28], produced results that, in general, aligned with
the acceptable risk levels defined in TOR and used in the
RCD. That is, both approaches would generally produce
a similar outcome for normal situations; however, the use
of the prescriptive code rules was a simpler more easily
understood approach and had well-defined acceptable limits.
Application to CO2
CO2 is categorized in PD 8010 as a category-C fluid, i.e.
a non-flammable fluid that is a non-toxic (noting that it
has a degree of toxicity) gas at ambient temperature and
atmospheric pressure conditions. It is currently not classified
as a ‘dangerous fluid’ under the Pipeline Safety Regulations
1996 [29]. At the time of publication of PD 8010 the large-
scale transportation of dense-phase CO2 was not envisaged.
Vapour phase
Transportation of CO2 in the vapour (or gaseous) phase
is limited in operating pressure to avoid the occurrence
of two-phase conditions. Because of the limited pressure
and its behaviour on release [30], the risk from CO2 in the
gaseous phase is lower than typical natural gas pipeline
operation. Hence the approach for natural gas produces
cautious results. Application of the PD 8010 rules for a
194
Individual Risk (per year)
-1000 -800 -600 -400
Distance from Pipeline (m)
n
io
NG Pipeline
st y
ut
di op
Fig.1. Individual risk transect for a natural gas (NG) pipeline and transect for a dense-phase CO2 pipeline (using a probit
rib
relationship between thermal or toxic dose and chance of fatality).
or c
914-mm diameter pipeline operating at 35barg and a design
t f ple
factor greater than 0.3 to CO2 with a substance factor of
0.3 would give a separation distance of 32m and a route
corridor (3 x separation distance) of 96m; the equivalent for
no m
methane, with a substance factor of 0.5, would be 50m and
200m (4 x separation distance). Using the QRA approach,
Sa
and assuming the water vapour content in the transported
CO2 is strictly controlled, that is, it is a ‘dry’ pipeline so that
the incidence of a significant failure due to internal corrosion
is similar to that for natural gas, there would not be a
10-cpm risk level and the 0.3-cpm distance would be less
than 20m due to the limited hazard range at ground level.
This illustrates the element of caution inherent in applying
the code rules in that both the individual and societal
risks would be acceptable if the code rules were applied.
The difference in the outcome of the two approaches is
manageable and, given the need to ensure access to the
pipeline, would not be expected to lead to any significant
issues in routeing or land-use planning.
Dense phase
It is not known at present which substance factor in PD
8010 would be applicable to the large-scale transportation
of CO2 in the dense phase or what fluid category would
be applied. The hazard-based equation for the separation
distance, and hence the population-density corridor, is not
applicable. Hence, there is a need to follow the QRA option
in routeing during design. That is, there is a need to consider
The Journal of Pipeline Engineering
1.0E-05
1.0E-06
1.0E-07
1.0E-08
-200 0 200 400 600 800 1000
CO2 Pipeline
the application of individual and societal risk assessments
to dense-phase CO2 pipelines in order to demonstrate that
the risks are being managed appropriately. This is discussed
in the following sections.
Risk assessment of CO2 pipelines
Methodology
The base-case analysis defined in this section involves an
evaluation of the risks arising from possible failure at the
mid-point of a 100-km length of buried 914-mm diameter
pipeline carrying dense-phase CO2 at an initial pressure of
150barg and temperature of 10°C. The specification for
the CO2 entering the pipeline and the reliability of the
dehydration system at the point of entry ensure that corrosion
can be managed to the extent that the contribution that
this makes to the failure frequency are no worse than for a
representative oil pipeline of the same wall thickness. The
possibility that the pipeline might be damaged by external
impact or by earth movement, man-made or natural, is
considered following the approach in PD 8010: Part 3. It is
assumed that the other causes of failure make a negligible
contribution to the failure frequency, compared with the
causes defined above that are included in the assessment.
It is assumed that the combination of the pipe wall thickness,
material grade and toughness, and the specification of the
composition of the CO2 mixture that can be transported have
3rd Quarter, 2012
been defined in such a way that the possibility of generating
a long-running fracture, i.e. several pipe lengths, has been
reduced to the point where such events can be ignored.
Further, for the purposes of the base-case assessment, the
possibility that a propagating failure would only cause a
long split has been ignored. That is, it has been assumed
that any sufficiently large through-wall defect will produce
a complete rupture of the pipe, leaving the two opposing
open ends of the pipeline separated by a fracture length
of 12m, as typically assumed for a high-pressure natural
gas pipeline. The consequences of punctures and ruptures
have been assessed at a failure frequency evaluated for the
pipeline specified above. The methodology includes the
following stages:
• Identify the failure causes and modes of the pipeline.
• Define failure frequencies for each of the modes of
failure that are being considered.
• Predict the rate of release from the pipeline for
each case.
• Evaluate the size of crater that will be produced by
the release from each failure.
n
• Predict the equivalent source conditions that apply
for this flow rate entering the specified below-ground
io
crater.
st y
ut
• Evaluate the time-dependent dispersion of the
di op
release, assuming the simplified equivalent source
rib
specified above for a range of wind speeds, directions
or c
and atmospheric stability categories.
• For each case, predict the time-dependent mean
t f ple
concentration that would be experienced at an array
of locations around the pipeline.
• Evaluate the dose experienced at each point,
no m
making allowance for the impact of concentration
fluctuations within the plume. The dose calculations
Sa
for a person outdoors initially is repeated, considering
a person who decides to attempt to escape and a
person who decides to remain stationary at each
location. The dose calculation for a person who is
indoors initially takes account of the CO2 ingress
and accumulation within a typical domestic property
and assumes that a person indoors does not attempt
to escape.
• By using an appropriate probit relationship or dose-
related casualty criterion, the location-specific risk at
each point can be evaluated for the length of pipeline
being considered.
A separate assessment is performed for day and night
conditions, as the distribution of atmospheric stability
categories are biased to reflect the tendency for stable
conditions to arise during the night time and unstable
conditions during the day time. The final values take a
weighted average of the two results. The location-specific
risk values can be used to provide a transect of individual
risk on a line perpendicular to the pipeline, assuming that
a person spends a certain percentage of time indoors and
outdoors.
195
Individual risk transects
The resulting risk transects are shown in Fig.1, for the base
case, where they are compared with the comparable risk
transect derived for an example 914-mm diameter natural
gas pipeline operating at 85barg.
A thermal- or toxic-dose related casualty criterion has been
used in producing these transects, for illustration, but the
shape of the curves for each substance remain the same if
a fixed casualty criterion had been applied. Figure 1 shows
that there are significant differences between the shape
and the level of the risk transect for a dense-phase CO2
pipeline and for a natural gas pipeline. The differences
in shape arise as the hazard range for a dense-phase CO2
pipeline is governed by the dispersion of a dense gas cloud,
compared with the thermal hazard ranges associated with
natural gas pipelines. The dispersion process is sensitive to
the weather conditions and this introduces a high degree
of variability into the results. The ratio of the ‘worst-case’
hazard distance to the average case value is large for a CO2
pipeline compared with the ratio deduced for the fire-based
hazard ranges for natural gas pipelines. The worst-case hazard
range extends a considerable distance and gives the long
‘tail’ to the risk transect. The difference in risk levels arise
as the failure frequencies for the dense-phase CO2 pipeline
are less than for the natural gas pipeline, as the CO2 pipeline
has a greater wall thickness.
The individual risk transect based on a criterion of receiving
a specified level of toxicity (SLOT), as defined by the HSE,
has been produced for the CO2 pipeline and it is found
that all of the individual risks are below a value of 1 x 10-5/
year and fall below 3 x 10-7/year at distances greater than
42m from the pipeline.
Societal risks
The approach recommended to societal risk evaluation in
PD 8010: Part 3 is to evaluate the risks to the surrounding
population from a fixed 1-km length of pipeline. The total
number of casualties, N, can be deduced for each event along
the pipeline if the population distribution at different times
of the day is specified. The cumulative frequency F with
which N or more casualties are produced can be evaluated
and a curve showing the variation of F with N can be plotted
for comparison with the criterion curve in PD 8010: Part 3.
This process is site specific and is a relatively time-consuming,
especially if full details of the variation in population with
time of day is taken into account. However, the results
obtained for the base-case pipeline indicate that it is possible
to construct a plausible example in which the FN curve for a
particular location on the pipeline approaches the criterion
curve given in PD8010 Part 3, despite having a distance of
only 42m to the 3 x 10-7/year individual risk level using a
SLOT casualty criterion. That is, the risks are moving into
the region where there is a requirement to think further
196
Uniform Population
Density
Separation
Distance Either
Side of Pipeline
Fig.2. Illustration of generic situation being assessed, with a uniform population density located beyond a certain distance
from the pipeline.
about risk-reduction measures in order to demonstrate
that the risks have been reduced to ‘as low as reasonably
practicable’ (ALARP), despite no individual being subject
to a level of individual risk that would cause concern. Using
individual risk, there would be no way in which this case
could be differentiated from a situation in which no-one
n
lived within ten times the individual risk based distance
of 42m of the pipeline, for example. Societal risk offers an
io
approach that could be used to rank such cases in a logical
st y
ut
order and a way of judging the acceptability of each case,
di op
by comparison with the criterion curve in PD 8010: Part 3.
rib
This suggests the need for further societal risk investigations
or c
in planning a route. The implications of this are discussed
further below.
t f ple
Discussion of results
no m
For natural gas, the relatively rapid decay in individual risk
makes the connection between the hazard range and the
Sa
individual risk levels easier to make. Further, experience has
shown that the rapid decrease in risk means that societal risks
are unlikely to cause a concern if the number of people at
higher risk than 3 x 10-7/year is controlled. This means that
for natural gas, the code-based approach and individual-risk
approach produce complementary results.
However, the wall thickness of the high-pressure dense-phase
CO2 pipes results in low predicted failure frequencies, and
hence distances to individual risk levels of 1 x 10-5/year and
3 x 10-7/year are relatively small. The results given above
show that, for the base case, the individual risk distances are
very much smaller than the hazard distance. This indicates
that it may not be conservative to take the individual risk
distances to define the separation distance for the pipeline
and a corridor width over which to assess population.
To address this it is recommended that a societal-risk
assessment is performed for the specified length of pipeline
and an FN curve produced for comparison with the criterion
curve in PD 8010: Part 3. However, there is a need for a
simplified screening approach that can be used to help
during the early stages of a project or in the event of a new
The Journal of Pipeline Engineering
Pipeline
proposed development near to an existing pipeline, as a
detailed site-specific societal-risk assessment would not be
warranted in every case. Such an approach is presented
below, based on the results from some generic societal risk
assessments.
Generic societal risk assessment
Generic studies have been carried out to investigate how
societal risk varies with population density and distance
from the pipeline. The generic situation that is assessed is
sketched in Fig.2.
The analysis is carried out for different separation distances
and population densities on either one side or both sides of
the pipeline. The predicted number of casualties is evaluated
and FN curves are produced.
The results are analysed to present the expectation value
of the potential number of casualties per year (casualties
per year) plotted against the separation distance assumed
(m), shown in Fig.3, for a representative natural gas (NG)
pipeline operating at 85bar and the base-case dense-phase
CO2 pipeline. The curves are plotted in the figure for three
different representative values of population density beyond
the separation distance.
If the separation distance of people from the pipeline was set
at 200m, the expectation value for the case of a population
density of 2.5 / hectare beyond 200m would be approximately
7 x 10-6 casualties per year for the CO2 pipeline (red curve),
and approximately 2 x 10-5 casualties per year for the natural
gas pipeline (orange curve). However, if there was a separation
distance of 400m, the expectation value would be less than
10-8 casualties per year for the natural gas pipeline but just
over 5 x 10-6 casualties per year for the CO2 pipeline.
This example shows that the shape of the above curves follows
from the shape of the risk transects. It is possible to get quite
a significant value of the expectation value for quite large
separation distances with a CO2 pipeline. Therefore, there
is a need to consider the societal risk beyond the individual
3rd Quarter, 2012 197

1.0E-02

Expected Number of Fatalities (per year)

1.0E-03

1.0E-04

1.0E-05

1.0E-06

1.0E-07

1.0E-08
0 200 400 600 800 1000
Distance from Pipeline (m)

NG - 2.5 /ha - both sides of the pipeline NG - 14 /ha - both sides of the pipeline
NG - 42 /ha - both sides of the pipeline CO2 - 2.5 /ha - both sides of the pipeline
CO2 - 14 /ha - both sides of the pipeline CO2 - 42 /ha - both sides of the pipeline

n
io
st y
Fig.3.Variation in expected value of the number of casualties per year with the pipeline-separation distance to people for
ut
different population densities beyond the separation distance.
di op
rib
risk distance of 3 x 10-7/year for CO2 (and equally, there is interest, to define the population densities required to
or c

no requirement for this for a natural gas pipeline). exceed an expectation value of 3 x 10-5 casualties per year,
in a stepped manner as follows:
t f ple

An expectation value of greater than 1 x 10-4 casualties per

year can be used to gauge whether there is potential for an • Step 1: Consider routeing required to avoid cities
FN curve for the pipeline to go beyond the assumed ALARP Use the curve for a population density of 100
no m

curve shown in PD 8010: Part 3 for societal risk. Curves persons per hectare to find a separation distance
are provided in Fig.4 to show the value of the uniform of people from the pipeline to avoid giving an
Sa

population density that would be a cause of concern if sited expectation value in excess of 3 x 10-5 casualties
uniformly, either on one side or both sides of a pipeline per year, using the SLOD or probit criterion.
beyond a given separation distance. Make sure there are no cities within this distance
of the route.
Based on the above results, the following procedure is • Step 2: Consider routeing required to avoid suburbs
recommended in screening. Firstly, define three target Use the curve for a population density of 40 persons
types of population: per hectare to find a separation distance of people
from the pipeline to avoid giving an expectation
• Scattered villages – population density up to value in excess of 3 x 10-5 casualties per year, using
10 persons per hectare averaged over the region of the SLOD or probit criterion. Make sure there are
interest. no suburban areas within this distance of the route.
• Suburban development – population density • Step 3: Consider routeing required to avoid villages
approximately 40 persons per hectare. Use the curve for a population density of 10 persons
• Intense inner city/urban areas of terraced housing/ per hectare to find a separation distance of people
high rise flats – population density approximately from the pipeline to avoid giving an expectation
100 persons per hectare. value in excess of 3 x 10-5 casualties per year, using
the SLOD or probit criterion. Make sure there are
Assume that a contribution to the expectation value no villages within this distance of the route.
of 3 x 10-5 casualties per year will be allowed from each • Step 4: Individual risk stage
development type. Use the original risk transect for individual risk to
check for an inner exclusion zone, with no people
Use similar curves to the ones shown in Fig.4 for other within a distance with an individual risk of greater
population densities and for the particular pipeline of than 1 x 10-5 / year.
198
1000
Population Density (people per ha) to give
Expectation Value of 10 (per year)
100
-4
10
1
0 100
st y
ut
di op
Fig.4. Population density to give an expectation value of 1 x 10-4 casualties per year for a given separation distance.
rib
or c
General discussion and conclusions
t f ple
The historical background and development of guidance that
is applied to control the risks to the surrounding population
no m
arising from pipelines transporting flammable substances
have been reviewed. It has been seen that using a hazard-based
Sa
approach or a risk-based approach produces results that are
similar for natural gas. This arises as the difference between
the greatest and smallest hazard range for an ignited release
are not very large, because of the positive buoyancy of the
natural gas fires and its dominance over the effects of the
wind. This allows screening for route selection to be carried
out in a quick and efficient manner for these pipelines.
However, it is concluded from the worked example that the
individual risk transect for a thick-walled CO2 pipeline, and
hence the risk profile, is very different from that for natural
gas. The absolute values of the individual risk are lower but
the rate of decay with distance from the pipeline is much
slower beyond about 200m. This latter behaviour arises in
part because of the significant influence of the atmospheric
conditions on the dispersion of a dense CO2 cloud that has
slumped back to the ground. There is no suitable release
factor that has been defined for dense-phase CO2 that takes
this variation and dependency on atmospheric conditions
into account. It is concluded that there is a potential conflict
between using the relatively large worst-case hazard range
and the relatively small distance to an individual risk of
3 x 10-7/year to control dense-phase CO2 pipelines.
The Journal of Pipeline Engineering
200 300 400 500 600 700 800 900 1000
Distance from Pipeline (m)
n
NG - one side of the pipeline only NG - both sides of the pipeline
io
CO2 - one side of the pipeline only CO2 - both sides of the pipeline
In some respects this is similar to the situation that arises
around large liquefied natural gas (LNG) storage tanks,
where the failure frequency predicted for the modern
full-containment tanks implies that the individual risks
are low, yet the hazard distances could be large and would
vary with atmospheric conditions. The results from the
generic societal risk assessments presented imply that it
would be possible to define a plausible-looking example
in which there is no-one living within the 3 x 10-7/year
individual risk contour around the pipeline, yet the FN
curve could still approach the PD8010 Part 3 criterion
curve. The traditional approaches adopted for natural
gas pipelines are therefore not sufficient.
Societal risk assessment offers a way to differentiate
between such cases in a logical and consistent manner
by taking into account the full range of possible release
behaviour and measuring its impact on all the surrounding
people. However, it is time-consuming to carry out a
full site-specific assessment and it may be unwarranted,
especially at the early stages of a project where numerous
options may be being considered. Therefore, a screening
method, based on the expectation value of the potential
number of casualties per year, has been proposed. This
method considers three types of population around a
pipeline; cities, other urban areas, and isolated villages.
A separation distance is defined for each type to ensure
that overall the societal FN curve will not go above the
PD 8010:Part 3 criterion.
3rd Quarter, 2012
The application of this screening approach to natural gas
pipelines is not required as, in practice, the separation
distance required for each type would be similar, as the
individual risk falls off so rapidly with distance. The multi-
distance approach is useful for CO2 pipelines, however.
It avoids the excessive caution that would be imposed by
having to apply the overly cautious city–type values to every
case. These values approach the maximum hazard distances
and so would represent a considerable penalty both at the
design stage and in controlling future developments around
an existing pipeline. Equally, it avoids going to the other
extreme of being excessively optimistic by applying the
smaller individual risk-like values, appropriate for isolated
villages, regardless of any significant conurbations just
beyond this distance.
References
1. ASME. Gas transmission and distribution piping systems,
B31.8, periodic (industry specific) editions from 1955.
2. IGE, 1970. Steel pipelines for high pressure gas
transmission. TD/1, Communication 674A.
n
3. BSI, 1966. Pipelines: installation of pipelines in land.
CP 2010 Part1: 1966.
io
4. BSI, 1992. Code of practice for pipelines, Part 2. Pipelines
st y
ut
on land: Design, construction and installations, Section
di op
2.8 Steel for oil and gas. BS 8010, Section 2.8: 1992.
rib
5. IGE, 1984. Steel pipelines for high pressure gas
or c
transmission. TD/1 Edition 2, Communication 1234.
6. HSC, 1976. Advisory Committee on Major Hazards, First
t f ple
Report, The Identification of Major Hazards, (ACMH1).
7. Idem, 1979. Second Report, The Assessment of Major
Hazards, (ACMH2).
no m
8. The Notification of Installations Handling Hazardous
Substances Regulations, 1982 (NIHHS), SI 1982/1357.
Sa
9. European Community Directive on the Major Accident
Hazards of Certain Industrial Activities. 82/501/EEC,
1982 (Seveso Directive).
10. Control of Industrial Major Accident Hazards
Regulations 1984 (CIMAH), SI 1984/1902.
11. D.A.Jones and G.D.Fearnehough, 1985. Natural gas
transmission by pipeline: a reassessment of the hazards
and risks and their influence on UK legislative safety
procedures. Fifth International Symposium on Safety
and Loss Prevention in the Process Industries, Cannes.
12. D.A.Carter, 1989. Developments in the vicinity of
pipelines operated by British Gas: advice to local
planning authorities.
13. IGE, 1993. Steel pipelines for high pressure gas
transmission. TD/1 Edition 3, Communication 1530.
199
14. BSI, 2004. Code of Practice for Pipelines- Part1: Steel
pipelines on land, PD 8010-1: 2004.
15. D.A.Jones and D.A.Carter, 1990. Pipeline safety
evaluations and their relevance to land-use planning
decisions. Pipeline Management ’90.
16. Royal Society, 1983. Risk assessment: a study group report.
17. Royal Society, 1992. Risk: analysis, perception,
management.
18. HSE, 1988. The tolerability of risks from nuclear power
stations 1988. Second Edition 1992 (TOR).
19. HSE, 2001. Reducing risk, protecting people. HSE’s
decision making process (R2P2).
20. HSE, 1989. Risk criteria for land use planning in the
vicinity of major industrial hazards, (RCD).
21. HSC, 1991. Advisory committee on dangerous substances
(ACDS): Major hazard aspects of the transport of
dangerous substances.
22. J.Haswell, 1999. The pipeline life cycle. Pipes and Pipelines
International, 44, 6. Scientific Surveys, Beaconsfield, UK.
23. IGE, 2001. Steel pipelines for high pressure gas
transmission. TD/1 Edition 4, Communication 1670.
24. IGEM, 2008. Steel pipelines for high pressure gas
transmission, TD/1 Edition 5, Communication 1735.
25. IGEM, 2008. Application of pipeline risk assessment to
proposed developments in the vicinity of high pressure
natural gas pipelines. TD/2, Communication 1737.
26. BSI, 2004. Code of practice for pipelines – Part3: Steel
pipelines on land- Guide to the application of pipeline
risk assessment to proposed developments in the
vicinity of major accident hazard pipelines containing
flammables. Supplement to PD 8010-1: 2004, PD
8010-3: 2009.
27. HSE. Guidance on ALARP decisions in COMAH. SPC/
Permissioning/37, www.hse.gov.uk/
28. I.Corder and G.D.Fearnehough, 1987. The prediction
of pipeline failure frequencies. Second International
Conference on Pipes, Pipelines and Pipeline Systems,
Utrecht.
29. The Pipelines Safety Regulations 1996 (PSR),
SI1996/825.
30. R.P.Cleaver, A.R.Halford, H.F.Hopkins, R.McConnell,
D.J.McCollum, and J.Barnett, 2011. Methods for
assessing the risks from transporting CO2 in the gaseous
phase by pipeline. Second International Forum on the
Transportation of CO2 by Pipeline, Newcastle upon
Tyne, UK.
31. S.Gant, 2012. Framework for validation of pipeline release
and dispersion models for the COOLTRANS project.
Third International Forum on the Transportation of
CO2 by Pipeline, Newcastle upon Tyne, UK.
 Sa
no m
t f ple
or c
di op
st y
rib
ut
io
n
3rd Quarter, 2012 201

Framework for validation of

pipeline release and dispersion
models for the COOLTRANS
research programme
by Dr Simon Gant*
Health and Safety Laboratory, Buxton, UK

T he framework of a model-evaluation protocol (MEP) is presented for CO2 discharge and dispersion
models. Its purpose is to provide a comprehensive evaluation methodology for determining the suitability
of models for simulating releases from CO2 transport infrastructure, i.e. pipelines and associated equipment.
The protocol has been developed by the Health and Safety Laboratory (HSL) in support of National Grid’s
COOLTRANS research programme, and it follows a similar structure to that developed previously by HSL

n
for liquefied natural gas (LNG).

io
The proposed MEP consists of five elements: context definition, scientific assessment, verification, validation,
st y
ut
and sensitivity analysis.The first of these involves the definition of the set of operating conditions and credible
di op

release scenarios that are to be simulated, together with identification of potential models. The second
rib
step of scientific assessment involves an examination of the underpinning physics of the models to discern
or c

whether they are suitable for the relevant release scenario. In the present work, verification is taken to mean
the process of ensuring that the model equations are implemented in the computer software correctly,
t f ple

which includes some consideration of code quality assurance. The validation step is a key component in
the MEP and involves the assessment of model performance against experimental data. In the present
work, four validation scenarios are considered: above-ground discharges from vents, transient discharges
no m

from a shock tube, steady releases from simulated punctures in a buried pipeline, and full-bore ruptures of
a buried pipeline. For each of these scenarios, experiments are being conducted at GLND Spadeadam as
Sa

part of the COOLTRANS research programme. The final element of the MEP involves sensitivity analysis,
where model-input parameters are varied in order to understand the effect of uncertainties in the physical
and model conditions. This final stage of the evaluation procedure also includes simulations of a ‘realistic’
full-scale release scenario that could, for example, be studied as part of a pipeline risk assessment.

The outcome of the MEP will be a review of the capabilities and limitations of various pipeline-discharge
and dispersion models, and an independent assessment of their performance.This will be of benefit to both
pipeline operators seeking to use appropriate consequence-modelling tools and regulatory authorities in
assessing pipeline quantified-risk assessments (QRAs).

A s part of the carbon capture and storage (CCS)

design and risk-assessment process, it is necessary
to understand the consequences of an intentional or
compressor stations and, in some cases, equipment on
offshore platforms.

accidental release from the CO2 transport infrastructure.
This infrastructure includes pipelines transporting the
CO2 from the emitter to the reservoir injection site,
This paper was presented at the Third International Forum on the Transportation of
CO2 by Pipeline, organized by Newcastle University, Tiratsoo Technical, and Clarion
Technical Conferences, and held in Newcastle in June, 2012. © Crown Copyright 2012.
*Author’s contact details:
tel: +44 (0)1298 218134
email: simon.gant@hsl.gov.uk
In many of the proposed CCS projects in the UK, the CO2
will be transported in pipelines as a dense-phase fluid (i.e.
in a liquid or supercritical state). When this material is
released into the atmosphere, it changes state into a two-
phase mixture of gas and solid CO2 particles (dry-ice). This
presents new challenges for consequence models, in terms
of predicting the complex depressurisation behaviour within
pipelines, the expansion of CO2 to atmospheric pressure
and the dispersion of the two-phase mixture.
202
Over the last five years, various calculation methods have
been developed to calculate the flow rate and atmospheric
dispersion from a dense-phase CO2 inventory [1-5]. These
have mainly been based on extending methods previously
developed for flashing-liquid releases (such as liquefied
petroleum gas). However, the accuracy of these models for
full-scale CO2 releases remains uncertain, and guidance on
the best practice use of these models in risk assessments
has yet to be produced. A recent study by TNO showed
that variations in modelling assumptions and risk criteria
could lead to significant variations in the predicted hazard
distances. As a consequence, the individual risk contour
was found to vary between 0m and 204m from the pipeline,
depending on assumptions made [6].
To address this issue, National Grid has commissioned
a comprehensive programme of experimental tests and
modelling exercises as part of the COOLTRANS research
programme. The partners in this project include: Pipeline
Integrity Engineers (PIE), Atkins, GL Noble Denton (GLND),
Nottingham University, University College London (UCL),
University of Leeds, Kingston University, and the Health and
n
Safety Laboratory (HSL). PIE is supporting National Grid
in defining the technical strategy and project management
io
of the programme. Atkins is involved principally on the
st y
ut
pipeline fracture propagation aspects rather than the release
di op
and dispersion-modelling analysis. The role of GLND is to
rib
conduct field-scale CO2-release experiments and to provide
or c
predictions using consequence models suitable for use in
risk assessments. Nottingham University is conducting
t f ple
laboratory experiments to develop an equation of state for
CO2 (with and without impurities), and is also conducting
separate field-scale experiments to examine the effect of
no m
fugitive CO2 emissions on vegetation, etc. UCL, and the
University of Leeds and Kingston University, are tasked
Sa
with modelling, respectively, the release rate, near-field, and
far-field dispersion behaviour of CO2 using sophisticated
computational fluid dynamics (CFD) models. Finally, the
role of HSL is to develop a model-evaluation protocol (MEP)
and to conduct some limited tests using the Det Norske
Veritas (DNV) consequence-modelling package, Phast. The
COOLTRANS project started in January, 2011, and is due
to be completed in December 2013. HSL’s involvement in
the project is financially supported solely by the UK Health
and Safety Executive (HSE).
The aim of the present paper is to describe the MEP that
is being developed by HSL for the COOLTRANS research
programme. At the present time, the programme of CO2-
release experiments is still ongoing and therefore details of
the MEP, such as the choice of quantitative model-acceptance
criteria, have yet to be finalized. Nevertheless, it is possible
to describe the general structure of the proposed MEP at
this stage and constructive feedback on the proposals is
welcomed.
The MEP proposed here is based on that developed
previously by HSL for assessing liquefied natural gas (LNG)
The Journal of Pipeline Engineering
dispersion models [7] which is currently used by the US
regulatory authorities for selecting models used in siting
LNG terminals1. This methodology was itself based on the
earlier work of the model-evaluation group (MEG), which is
described in various papers produced by the EU-sponsored
REDIPHEM and SMEDIS projects [8-11]. The proposed
model-evaluation protocol consists of the following five
elements:
• context definition
• scientific assessment
• verification
• validation
• sensitivity analysis
Each of these items are discussed in more detail in the
later sections, as applied to the case of CO2 discharge and
release modelling.
Context definition
A wide range of potential release scenarios needs to be
considered as part of the MEP. The transport of CO2 from
capture locations to sequestration reservoirs is likely to
involve gaseous-phase CO2 in the initial collection stage
from power stations and other emitters, supercritical CO2
following compression, and liquid-phase CO2 once the
supercritical fluid has cooled within the pipeline network,
either naturally from heat loss to the ground or through
forced cooling following compression. The locations of
potential CO2 releases include discharges from both above-
ground (for example, from compressor stations) and below-
ground, from the buried pipelines. Since the main focus of
pipeline-risk assessment is on the impact of hazardous CO2
clouds on populated areas, there is an interest in considering
typical populations and location topography. Finally, the
CO2 may contain various impurities depending upon the
emission source and pipeline specification.
To simulate this wide range of potential release scenarios,
three stages of model are required:
• discharge model
• near-field dispersion model
• far-field dispersion model
The primary role of the discharge model is to predict the
mass release rate of CO2 and the conditions at the release
orifice in terms of pressure, temperature, and CO2 quality
(i.e. the proportions of gas, liquid, solid, and supercritical
CO2). Different discharge models are needed in different
circumstances. For instance, a small leak from a large-
diameter pipeline may be treated similarly to a leak from a
vessel, in which case the conditions upstream of the orifice
are assumed to remain unchanged over time. Full-bore
pipeline ruptures, however, require a model that predicts the
1. NFPA 59A: Standard for the production, storage, and handling of liquefied natural
gas (LNG) – see: http://tinyurl.com/crd6525, accessed May 2012
3rd Quarter, 2012
time-varying depressurisation of the pipeline and accounts
for the changing conditions near the orifice. Various
discharge models have been developed for these scenarios
that incorporate different assumptions and varying levels
of complexity. For the pipeline-depressurization scenario,
three models are being tested in the COOLTRANS research
programme: the PipeTech model from UCL [12], the GLND
model [13], and the Pipebreak model in Phast [14].
Immediately downstream from the orifice there is a zone
in which the CO2 expands to atmospheric pressure. The
behaviour here is complex and may feature shocks and phase
transition from an initial high-pressure gas, supercritical
liquid, or two-phase, state into the final gas and solid
state at atmospheric pressure. Three different approaches
to model the expansion behaviour are being tested in the
COOLTRANS research programme: a compressible multi-
phase CFD model developed by University of Leeds and
two simpler integral models used by GLND and DNV (the
Phast software).
For the near-field models, one of the major challenges
n
is to simulate releases from buried pipelines, where the
high-speed jet of CO2 produces a crater around the release
io
location. This has a significant effect on the dispersion
st y
ut
behaviour, and hence the hazard range. The entrainment
di op
of air into the crater and the impact of the CO2 onto the
rib
crater walls (with the potential for CO2 solids to deposit)
or c
make this a very difficult scenario to model. There are also
uncertainties associated with the nature of the surrounding
t f ple
soil and the way in which the pipelines fracture (i.e. for
full-bore ruptures, whether the two pipeline ends become
misaligned). Furthermore, under low wind-speed conditions,
no m
the dense CO2 gas ejected from the crater may fall back and
be re-entrained into the crater, coupling the behaviour in
Sa
the near-field to the far-field dispersions situation.
In the COOLTRANS research programme, experiments
are being conducted to assess the shape and size of craters
formed by buried pipeline punctures and full-bore ruptures.
CFD simulations of the flow within the crater are being
conducted by the University of Leeds, whilst GLND is
developing a simple empirically-based crater-source model.
In the far-field, dispersion models need to account for the
sublimation of any solid CO2 particles present, the potential
for solids to ‘rain-out’, and the effects of ambient humidity,
which may lead to water vapour condensing into the cold
CO2 cloud. Previous dense-gas dispersion incidents, such
as the Lake Nyos disaster [15] and the Buncefield incident
[16], have shown that the far-field dispersion behaviour
can be strongly driven by topography and the presence
of obstacles, especially under low wind-speed conditions.
Ideally, far-field dispersion models therefore need to take
into account these effects. In the COOLTRANS research
programme, experiments are being conducted in which
the gas cloud disperses over either flat ground or obstacles
and slopes. Simulations of these experiments are being
203
performed by Kingston University using two different CFD
codes, and by GLND and HSL, using integral far-field
dispersion models.
Ultimately, the purpose of the discharge, near-field, and
far-field dispersion models is to estimate the hazard range.
For potential exposures to toxic substances, such as CO2, the
hazard and risk is estimated by HSE on the basis of the ‘toxic
load’. For CO2, this is calculated from the time-integral of
the gas concentration to the power eight2. As a consequence
of this highly non-linear dependence of the toxic load on
the concentration, turbulent concentration fluctuations
in the dispersing plume of CO2 have a significant effect
on the calculated hazard range [17]. The assessment of
far-field CO2 dispersion models therefore needs to take
into account the treatment of concentration fluctuations.
Scientific assessment
The scientific assessment involves a critical review of the
physical, mathematical, and numerical bass of the models
to ensure that they are fit for the purpose of simulating
CO2 releases. In the proposed MEP, the first stage of this
assessment will involve a questionnaire that is completed
by the model developers. The objective of this exercise will
be to ascertain information from the model developers in
a standardized format that will enable model features to be
compared. The information recorded in the questionnaire
will be examined by one or more independent experts.
Additional material may be considered in the assessment
process in the form of user-manuals, peer-reviewed
publications, and reports and the results of validation
exercises (including those made independently of the
COOLTRANS research programme).
The format of the questionnaire will be similar to that
previously developed by Ivings et al. [7]. A list of the likely
topics is given below for the far-field dispersion models.
Suitably modified questionnaires will be developed for the
discharge and near-field dispersion models.
1. General model description
a. model version number and release
b. short description
c. history of model
d. availability of model
e. level of expertise required to use model
f. hardware and software requirements
2. Scientific basis
a. model type (empirical, integral, CFD)
b. specification of the source
i. type of sources (single/two-phase jet, area source,
multiple sources)
ii. continuous or time-varying
iii. thermodynamic properties of source
2. http://www.hse.gov.uk/hid/haztox.htm, accessed May 2012.
204 The Journal of Pipeline Engineering

c. specification of the environment Verification

i. reference frame (2D/3D Cartesian, cylindrical
polar, spherical)
ii. atmosphere (wind profiles, stratification,
humidity, zero wind)
iii. turbulence parameterisation
iv. terrain
v. surface roughness characteristics
vi. obstacles
d. model physics and formulation
i. transport equations solved
ii. turbulence model
iii. initial and boundary conditions allowed
iv. dispersion treatment (advection, gravity
spreading, dilution, concentration fluctuations)
v. concentration profiles assumed
vi. treatment of aerosols
vii. thermodynamics
viii. transition to passive dispersion
e. solution technique
i. equation types
ii. analytical/numerical solution methods
In the present work, verification is taken to mean the
process of ensuring that the model equations are correctly
implemented in the computer software. This is distinct from
the process of model validation, which consists of ensuring
that the model provides an accurate representation of the
flow physics. More precise definitions can be found, for
example, in the book by Roache [18].
Information will be sought from the model developers on
verification as part of the scientific assessment exercise.
This will include consideration of code version control,
comparison of model predictions to known analytical
solutions, and use of the method of manufactured solutions.
It is not proposed to undertake specific demonstrations of
model verification as part of the MEP. The extent to which
a model has been verified will be assessed qualitatively by
an independent expert, based on the information provided
by the model developers.
Validation and model-sensitivity analysis

n
f. results output from model
i. concentration data output (centreline/centroid The objective of the model-validation stage of the MEP is to

io
values, pointwise, etc.) assess the accuracy of CO2 discharge and release models by
st y
ut
ii. further outputs (velocity, turbulence, etc.) comparing their predictions to experimental measurements
di op

g. source of model uncertainty for a range of representative scenarios.

rib
h. limits of applicability
or c

i. special features Atmospheric dispersion involves inherent randomness

j. planned developments due to turbulence and so, for the far-field dispersion
t f ple

modelling at least, the predictions can only be assessed

3. User-oriented aspects of model in terms of their statistical performance rather than their
a. documentation and help ability to predict directly the time-varying concentrations
no m

b. installation procedures in a particular experiment. Comparisons will therefore be

c. user interface made on the basis of the mean, variance, or frequencies of
Sa

d. internal databases occurrence of the temperatures or concentrations, rather

e. guidance in selecting model options than their instantaneous values at a particular position
f. assistance in data input and point in time.
g. error/warning messages and checks on use of model
beyond its scope In the previous LNG MEP [7], a dispersion model was
h. computational aspects considered to be ‘acceptable’ provided that the following
i. programming language statistical performance criteria were met:
j. user programming options
k. typical execution times • Mean bias of the predictions within ±50% of the
l. model output measured mean, corresponding to a mean-relative
m. facilities for post-processing results bias (MRB) of between -0.4 and 0.4, and a mean-
n. model suitability to users and usage geometric bias (MG) of between 0.67 and 1.5.
o. possible improvements • A scatter of a factor of three of the mean,
corresponding to a mean relative-square error
4. Verification (MRSE) of less than 2.3 and a geometric variance
a. verification undertaken (VG) less than 3.3.
b. quality assurance • The fraction of model predictions within a factor of
two of the measurements (FAC2) to be at least 50%.
5. Validation
a. previous validation studies Definitions of the statistical-performance measures: MRB,
MG, MRSE, VG, and FAC2 can be found in the report by
Ivings et al. [7]. These criteria were applied to maximum
arc-wise concentration and plume-width concentration
3rd Quarter, 2012
data in the LNG MEP. However, it was considered that
there was insufficient experience to apply the same criteria
to concentrations at particular point locations or to
temperatures.
The purpose in calculating these different measures is to help
understand the different facets of a model’s performance.
For example, one model may be ‘correct’ on average but
exhibit a high degree of scatter. This would show up in
the values of the MRB and the relatively large value of the
MRSE and VG. Another model may nearly always over-
predict the experiments by a small percentage. This would
show up in the sign and value of the MRSE and the value
of the MG. It is useful for a user or assessor to be aware of
the values such as these, to help in the interpretation of
the predictions of a particular model.
At present, it is unclear whether the same model-acceptance
criteria will be adopted in the CO2 MEP as for the LNG
MEP. This choice will depend upon the quality of the
measurement data from the COOLTRANS experiments.
Consideration also needs to be given to the fact that the
n
LNG MEP was devised to assess flammable gas dispersion,
whereas in the case of CO2 it is toxicity that is the concern.
io
The hazardous concentrations of CO2 and natural gas
st y
ut
are in fact reasonably close (in terms of a constant, non-
di op
time-varying, concentration). The lower explosive limit of
rib
methane is 4.4% vol/vol as compared to the immediately-
or c
dangerous to life and health (IDLH) concentration for CO2
of 4.0% vol/vol [19]. However, the HSE harm criteria in
t f ple
the case of CO2 is determined from the time integral of
concentration to the power eight, which means that peak
values of concentration in the plume have a significant
no m
effect. This is somewhat different from the criteria used to
assess flammable hazards, which are based on the predicted
Sa
concentration, rather than a time-integrated dose.
In the LNG MEP, data were provided for model validation
based on short and long time-averaged concentrations,
where in the former case averaging was typically over a period
of 1 second [20]. Preference was given to validation using
the short time-averaged data in view of the fact that the
application was for flammable releases. In the CO2 MEP,
different averaging times will be investigated. If possible,
experimental data for model validation will be presented
on the basis of the toxic load, which will be calculated by
integrating the time-averaged concentration to the power
eight over time, where the averaging period is similar to
duration of a human breath (i.e. a few seconds).
The dispersion models’ performance in the LNG MEP
was categorized into two groups, according to whether
dispersion took place over flat, open ground, or complex
terrain. A similar methodology to this will be adopted in
the proposed CO2 MEP.
One approach to model validation is to compare model
predictions to the experiments ‘blind’, which would involve
205
the modellers being given only the release geometry,
the reservoir state, and the weather conditions, without
any measurement data. This replicates the way in which
models tend to be used in risk assessments, when there
are no measurement data available. However, in most of
the scenarios simulated in the CO2 MEP, the discharge,
near-field, and far-field models are used in sequence, with
the outputs from one model being used as inputs to the
next model. This introduces difficulties, for example, in
independently assessing the performance of the far-field
dispersion model, since the input conditions provided to it
from the upstream discharge and near-field models could be
quite inaccurate. If the far-field model used a mass release
rate (from the discharge model) that was much lower than
the actual release rate in the experiment, its predictions
could not be expected to be the same as the measurements.
The alternative approach of allowing modellers free access to
the measurement data prior to the submission of results in
order to optimize their choice of model inputs (and obtain
the best agreement with the results) is also not entirely
acceptable. Different combinations of model inputs could
be used in each scenario, and the models would then no
longer be being used as predictive tools.
To address this issue, it is proposed in the CO2 MEP to
request two sets of predictions from the modellers:
• ‘Standard’ model predictions, using the default
input parameters for the particular model, without
any tuning; and
• ‘Optimzed’ model predictions using tuned input
parameters that achieve the best agreement with
the measurement data.
For example, in the shock-tube scenario (discussed in
more detail below), one of the model input parameters
to the pipeline discharge model is the orifice discharge
coefficient, Cd. A default value of Cd is used by integral
models such as Phast, but it is known that Cd is a function
of the particular orifice design and, moreover, the value
of Cd has a significant effect on the discharge rate. In the
proposed approach, ‘standard’ model predictions would
be produced with the default value of Cd. Better agreement
with the results may be achieved by changing Cd to some
other value. The results from both of these tests would be
submitted for the model-evaluation exercise. In addition
to providing two alternative model inputs to the near-field
and far-field models (one ‘default’ and the other more
accurate), the two sets of results will help to provide a
measure of the sensitivity of the results to uncertainties
in the input parameters.
A tree diagram is shown in Fig.1 that describes how it is
proposed to combine ‘standard’ and ‘optimized’ variants
of the discharge, near-field, and far-field dispersion models.
The combinations of model inputs and outputs have
been selected here in order to assess the accuracy of each
206 The Journal of Pipeline Engineering

Discharge Near-Field Far-Field

Dispersion Dispersion

Standard Standard Standard

Standard
Optimised
Standard
Optimised

Optimised

Fig.1. Proposed combinations of model variants for the discharge, near-and far-field dispersion models.

n
io
st y
ut
di op
rib
or c
t f ple
no m
Sa

Fig.2.Vertical vent release experiment.

3rd Quarter, 2012
model independently (based on accurate input values) and
separately to provide model predictions using the ‘standard’
version of all three models, to replicate the type of behaviour
that is expected if the models were used as they would be
in an industrial risk assessment. In some cases, the default
values may in fact be found to provide ‘optimum’ results,
or it may be difficult to determine tuned values (due to a
lack of validation data in the near-field, for example). If
this is the case, then the number of simulations required
will be reduced.
Whilst it is feasible to run multiple scenarios and optimise
model inputs using integral models such as Phast, the
situation is different for the CFD models used by UCL,
the University of Leeds, and Kingston University. Their
simulations are more time-consuming to run and it is
less clear what the ‘default’ values would be for these
models. Moreover, they are unlikely to be used directly for
consequence modelling as part of a pipeline-risk assessment.
Instead, their role in the COOLTRANS research programme
is primarily to provide reference solutions, using state-of-
the-art computational methods. Therefore, only one set
n
of model predictions from the university partners will
be considered for each of the validation cases, based on
io
the ‘optimum’ input conditions that have been found to
st y
ut
produce the best agreement with the measurements.
di op
rib
The following sections consider the modelling scenarios
or c
being considered as part of the COOLTRANS research
programme.
t f ple
Scenario 1: above-ground vent releases
no m
The first validation scenario involves above-ground releases
of CO2 from circular orifices. The results from this case
Sa
are relevant to leaks from large-diameter above-ground
pipelines or vessels, and full-bore ruptures of short lengths
of small-diameter pipes attached to vessels.
In total 16 release experiments were conducted at
Spadeadam that could potentially be used for model
validation of this scenario, comprising 11 vertical and
five horizontal releases, with orifice diameters of between
13mm (0.5in) and 51mm (2in). An example of one of the
vertical release tests is shown in Fig.2. Mass release rates
of CO2 were measured in most of the vertical releases but
not in the horizontal releases. Both gaseous and liquid-
phase CO2 was released in the vertical cases, with the
pressure maintained at an approximately constant value
using nitrogen padding gas on the input side to the CO2
reservoir. For the vertical liquid-phase CO2 releases, this
meant that conditions were maintained in the storage vessel
above the saturation conditions, at around 150barg and
ambient temperatures.
The horizontal releases were through either 13-mm (0.5-
in) or 25-mm (1-in) diameter orifices from the end of the
144-m long, 152-mm (6-in) diameter, shock-tube rig. After
207
an initial very fast transient, the pressure in shock tube
fell to the saturation conditions and then slowly decayed
to ambient pressure over a few tens of minutes. Tests were
conducted with the circular vent located in either the top
or the bottom of the shock tube’s end face, in order to
release either the gaseous or liquid-phase CO2.
It would be very onerous to simulate all 16 releases as part
of the validation exercise. Instead, it is proposed to select
four cases that cover a range of different conditions:
1a. Vertical gaseous-phase CO2 release at approximately
34barg from a 25-mm diameter orifice.
1b. Vertical liquid-phase CO2 release at approximately
150barg from a 25-mm diameter orifice.
1c. Vertical liquid-phase CO2 release at approximately
150barg from a 51-mm diameter orifice.
1d. Horizontal liquid-phase CO2 release from the
bottom of the shock-tube end face through a 25-mm
diameter orifice.
For the vertical tests (Cases 1a to 1c), a short section
of pipe leading to the orifice was instrumented to take
pressure and temperature measurements at either end of
the pipe. Therefore, it should be possible to validate the
discharge models using measured pressure, temperature,
and mass-release rate. Temperatures were also measured
on a horizontal array of points at two heights close to the
release orifice in these cases, as shown in Fig.2. These data
may be used for validating the near-field models, in terms
of the plume centreline temperatures and plume-widths.
In the far-field, for practical reasons, the measurements
were limited to probes close to ground level (rather than
in the elevated plume): there is therefore insufficient data
to derive the plume centreline concentrations, but the
measurements should still provide sufficient information
for the purposes of identifying qualitative differences in
cloud behaviour (i.e. occurrence of cloud touchdown).
For the horizontal release (Case 1d), a short section of 25-
mm vent pipe was attached to the shock-tube end face, which
was instrumented to record the pressure and temperature.
The mass-release rate was not measured directly. Only
partial validation of the discharge model was therefore
possible (since even if P and T were measured, the relative
amounts of gaseous and liquid-phase CO2 are unknown
and therefore the mass-release rate cannot be determined).
However, some indication of whether the mass-release rate
is predicted accurately can be determined from the time
taken for the shock tube to depressurize.
Scenario 2: transient shock-tube releases
The second validation scenario consists of full-bore
discharges from the end of a 144-m long, 152-mm
diameter, shock-tube test rig (Fig.3). The primary purpose
of these experiments was to record the highly transient
depressurization to saturation conditions that occurs within
208
n
io
st y
ut
di op
Fig.3. Photo of the shock-tube test rig.
rib
or c
the first second or so of a release, in order to validate
predictions of the decompression wave speed. The wave
t f ple
speed is an important parameter in determining the driving
force for pipeline fracture-propagation models.
no m
For the purposes of pipeline outflow model validation,
the only data available from these tests consist of the time-
Sa
varying pipe-wall temperatures along the length of the tube,
and the fluid pressure and temperature at a position close
to its closed end. No measurements are available for the
mass flow rate or the fluid pressure and temperature at
the open end of the shock tube, as the primary purpose
of the tests was to obtain depressurization data. The
amount of data is less than ideal for the assessment of
outflow and dispersion, but useful comparisons to model
predictions can still be made. The rate at which the
pressure and temperature decays over time provides some
indication of the release rate. Comparisons between model
predictions and measurements will therefore be made
initially by comparing time-varying profiles of pressure and
temperature. Other quantitative measures, such as the time
taken for the pressure to halve, may be considered later.
In the plume produced by the shock-tube releases,
measurements were made of the temperature, concentration,
and flow velocity, in order to study how such a highly
transient release behaves, i.e. whether it exhibits behaviour
similar to a steady plume or an instantaneous ‘puff’. The
effect of localized topographical features on the dispersion
behaviour was also studied by performing tests across either
The Journal of Pipeline Engineering
open, flat terrain, or across terrain that was sculpted into
mounds and/or surface depressions.
In total, 31 gaseous and liquid-phase CO2 shock-tube tests
were conducted which considered a range of initial fluid
pressures, temperatures, and compositions (including
mixtures of CO2 with various proportions of hydrogen,
nitrogen, oxygen, and methane), and also various types
of terrain. Three of these tests have been selected for the
model validation exercise:
2a. Pure liquid-phase CO2 at an initial pressure of
around 150bar, with flat terrain.
2b. Pure liquid-phase CO2 at an initial pressure of
around 150bar, with complex terrain featuring a
slope and a 5-m high mound.
2c. A CO2-rich mixture comprising 92% CO2 + 4% N2
+ 4% H2 at and initial pressure of around 140bar,
with complex terrain featuring a slope and a 5-m
high mound.
If the effect of impurities is found to have a significant
effect on the mass release rate and dispersion behaviour,
further validation cases may have to be considered.
Validation of the far-field dispersion models will be more
challenging for this scenario as compared to the previous
above-ground vent release, due to the highly transient nature
of the discharge and the uncertainty over the predicted
mass-release rate.
3rd Quarter, 2012
n
io
st y
ut
di op
rib
or c
Fig.4. Pre-formed crater prior to burial.
t f ple
Scenario 3: buried pipeline punctures
no m
The third model validation scenario consists of steady releases
through 25-mm and 51-mm diameter orifices from a buried
Sa
914-mm (36-in) diameter section of pipeline. The pressure
in the pipeline was maintained approximately constant
during each release using a long length of pipe filled with
CO2 that formed a charge line, which was connected to a
nitrogen padding system. Mass flow rates were measured
in the CO2 supply line using a Coriolis flowmeter. A series
of eight experiments were conducted with either gaseous
or liquid-phase CO2, using different release orientations
from the 914-mm pipeline (horizontal, vertically-up, and
vertically-down) and in different atmospheric conditions.
In most of the releases, the high-speed discharge from the
buried pipe was allowed to form its own crater. Two different
types of soil were studied: clay and sand. In a final test, a
pre-formed crater manufactured from sheet steel was used
instead to allow instrumentation to be placed closer to the
release location (Fig.4). The dimensions of this crater were
chosen to match the naturally-formed crater from one of
the previous tests.
In the experiments, the pressure and temperature close to
the release location within the pipeline was measured, in
addition to the mass-flow rate into the section of pipeline. For
the dispersion aspects, temperatures and concentrations were
209
measured on four arcs in the far-field and at other locations
up to 200-m from the release point. For the pre-formed crater
test, additional instrumentation was located immediately
above the crater to measure the plume temperatures, which
may be used to validate the near-field dispersion model.
The experiments for Scenario 3 have only recently been
conducted and the data have yet to be processed fully.
However, it is currently proposed to use results from four
tests for the purposes of model validation:
3a. Horizontal release from a 25-mm orifice into a pre-
formed crater in a low wind speed.
3b. Vertically down release from a 51-mm orifice into
sand in a high wind speed.
3c. Horizontal release from a 25-mm orifice into sand
in a low wind speed.
3d. Horizontal release from a 25-mm orifice into clay
in a medium wind speed.
Scenario 4: buried-pipeline ruptures
The final validation scenario consists of full-bore ruptures of
a buried section of 152-mm (6-in) diameter pipeline. Each
end of the ruptured pipeline will be fed from CO2 reservoirs
in order to produce a long-duration release. Fluid pressure
and temperature will be measured within the pipeline ends,
210
close to the release location, and the installation of Coriolis
meters in the two ends of the pipeline is currently being
considered. To validate the dispersion models, temperatures
and concentrations will be measured on arcs in the far-field
and at other locations up to 200-m from the release point.
Some velocity measurements will also be made. Releases
will take place into soil, forming natural craters, and at least
one test is planned to take place using a pre-formed crater.
For the pre-formed crater test, additional instrumentation
will be located close to the release point.
Details of the cases to be considered for model validation
will be decided upon when the experimental programme
is finalized, but they will probably consist of the following:
4a. Gaseous-phase CO2 release forming a natural crater.
4b. Liquid-phase CO2 release forming a natural crater.
4c. Liquid-phase CO2 release with a pre-formed crater.
Scenario 5: ‘realistic’ full-scale
pipeline-release scenario
n
The final scenario considered for the model evaluation
exercise consists of a ‘realistic’ full-scale pipeline release
io
that may be considered as part of a pipeline risk assessment.
st y
ut
The conditions modelled are currently planned to consist
di op
of a full-bore rupture of a 914-mm (36-in) diameter, 70-km
rib
pipeline with a mid-point break. Although no measurement
or c
data will be available for this case, useful information will
be obtained by comparing results from the different models.
t f ple
Conclusions
no m
A framework methodology has been proposed for evaluating
CO2 discharge and dispersion models as part of National
Sa
Grid’s COOLTRANS research programme. The aim of the
MEP is to produce an independent assessment of models,
in terms of their scientific basis, verification and validation.
One of the difficulties faced in validating CO2 discharge, near-
field and far-field dispersion models is the interdependence
of models, where errors in the output from one model
may propagate into the inputs of the next model. In the
present work, it has been proposed to address this issue by
performing multiple integral model simulations using both
default and optimized model input parameters. Additionally,
‘reference’ simulations will be obtained using state-of-the-
art CFD models.
An initial selection of validation case studies has been made,
based on four release scenarios, including: above-ground
discharges from vents, transient discharges from a shock
tube, steady releases from simulated punctures in a buried
pipeline, and full-bore ruptures of a buried pipeline. For
each of these scenarios, experiments are being conducted at
GLND Spadeadam as part of the COOLTRANS research
programme. A final stage of the evaluation procedure
includes simulations of a ‘realistic’ full-scale release scenario
The Journal of Pipeline Engineering
that could, for example, be studied as part of a pipeline-risk
assessment.
The range of scenarios will be used to validate each of the
models (discharge, near-field and far-field dispersion) at a
range of scales, from 25-mm diameter steady vent releases
to 152-mm diameter full-bore pipeline ruptures. Most of the
validation cases will involve releases of liquid-phase CO2 at
an initial pressure of around 150barg, but some tests will
consider gaseous-phase CO2 releases and, in one case, a
140-barg CO2 rich mixture with small amounts of nitrogen
and hydrogen. For the far-field dispersion modelling, most
of the validation cases will involve flat open terrain, but
in some of the shock-tube tests the CO2 disperses across
complex terrain featuring a slope and a mound.
One area where it could be argued that uncertainties will
remain following the proposed validation exercise is in
predicting the outflow from pipeline ruptures. It is important
to have confidence in the predicted mass-release rate from
the pipeline outflow model, since it has a strong effect on the
subsequent dispersion behaviour. Validation of the outflow
models is limited to comparing the time-varying pressure
and temperature at the closed end of the shock tube. No
mass-release-rate data or information from measurements
of temperature and pressure near the open end of the tube
are currently available. This limitation is to a large extent
understandable, given the severe technical challenges to be
overcome in measuring experimentally the mass-release rate
from full-bore pipeline ruptures. To partially address this
issue, it is proposed to compare in detail the predictions of
different pipeline depressurization models and study model
sensitivities. If, in the future, mass-release rate data become
available, it would be useful to incorporate this into the
validation exercise.
A second area where uncertainties remain is in scaling the
results from the 152-mm diameter rupture tests to full-scale
CO2 pipelines. In part, this issue should be addressed in
the review of the scientific basis of models, which will
examine whether the models appropriately account for
scaling issues. Model predictions will also be compared and
sensitivity tests undertaken for full-scale releases as part of
the final evaluation test case involving a simulated ‘realistic’
pipeline rupture. As part of the COOLTRANS research
programme, experiments are currently being conducted at
GLND Spadeadam to investigate crack propagation using
lengths of 914-mm diameter pipeline. These should help
to identify any significant qualitative changes in the release
and dispersion behaviour at full scale.
The outcome of the MEP will be a review of the
capabilities and limitations of various pipeline discharge
and dispersion models, and an independent assessment of
their performance. This will be of benefit to both pipeline
operators seeking to use appropriate consequence modelling
tools and regulatory authorities in assessing pipeline QRAs.
3rd Quarter, 2012
Acknowledgements
The author would like to thank National Grid for providing
access to the COOLTRANS experimental data, Phil Cleaver
and Ann Halford (GLND) for assistance in providing
information on the various experiments, and Jane Haswell
(PIE) and Mike Bilio (HSE) for providing constructive
comments on an earlier draft.
Disclaimer
This publication and the work it describes were funded by the
Health and Safety Executive (HSE). Its contents, including
any opinions and/or conclusions expressed, are those of
the author alone and do not necessarily reflect HSE policy.
References
1. C.Dixon and M.Hasson, 2007. Calculating the release
and dispersion of gaseous, liquid and supercritical
CO2. IMechE Seminar on Pressure Release, Fires and
Explosions, London, UK.
n
2. H.W.M.Witlox, M.Harper, and A.Oke, 2009. Modelling
of discharge and atmospheric dispersion for carbon
io
dioxide releases. J. Loss Prev. Process Ind., 22, pp795-802.
st y
ut
3. D.M.Webber, 2011. Generalising two-phase homogeneous
di op
equilibrium pipeline and jet models to the case of carbon
rib
dioxide. J. Loss Prev. Process Ind., 24, pp356-360.
or c
4. H.Mahgerefteh, S.Martynov, and S.Brown, 2011. Modelling
dry ice formation following rapid decompression of CO2
t f ple
pipelines. The Second International Forum on the
Transportation of CO2 by Pipeline, Newcastle, UK.
5. T.A.Hill, J.E.Fackrell, M.R.Dubal, and S.M.Stiff, 2011.
no m
Understanding the consequences of CO2 leakage
downstream of the capture plant. Energy Procedia, 4,
Sa
pp2230-2237.
6. J.Koornneef, M.Spruijt, M.Molag, A.Ramírez,
W.Turkenburg, and A.Faaij, 2010. Quantitative risk
assessment of CO2 transport by pipelines - A review
of uncertainties and their impacts. J. Haz. Mat., 177,
pp12-27.
7. M.J.Ivings, S.F.Jagger, C.J.Lea, and D.M.Webber, 2008.
Evaluating vapor dispersion models for safety analysis
of LNG facilities. Technical Report, Fire Protection
Research Foundation (http://www.nfpa.org/assets/
files//PDF/Research/LNGVaporDispersionModel.
pdf, accessed June 2008).
8. M.Nielsen and S.Ott, 1996. A collection of data from
dense gas experiments. Risø Report R-845(EN), Risø
National Laboratory, Roskilde, Denmark.
9. N.C.Daish, R.E.Britter, P.F.Linden, S.F.Jagger, and
B.Carissimo, 2000. SMEDIS: scientific model evaluation
of dense gas dispersion models. Int. J. Env. Poll., 14,
pp39-51.
211
10. B.Carissimo, S.F.Jagger, N.C.Daish, A.Halford, S.Selmer-
Olsen, K.Riikonen, J.M.Perroux, J.Wurtz, J.G.Bartzis,
N.J.Duijm, K.Ham, M.Schatzmann, and D.-R.Hall,
2001. The SMEDIS database and validation exercise.
Int. J. Env. Poll., 16, pp614-629.
11. N.J.Duijm and B.Carissimo, 2002. Evaluation
methodologies for dense gas dispersion models. In:
M.Fingas (Ed.) The handbook of hazardous materials
spills technology. McGraw-Hill.
12. A.Oke, H.Mahgerefteh, I.G.Economou, and Y.Rykov,
2003. A transient outflow model for pipeline puncture.
Chem. Eng. Sci., 58, pp4591-4604.
13. R.P.Cleaver, P.S.Cumber, and A.R.Halford, 2003.
Modelling outflow from a ruptured pipeline transporting
compressed volatile liquids. J. Loss Prev. Process Ind., 16,
pp533-543.
14. D.M.Webber, T.K.Fanneløp, and H.W.M.Witlox,
1999. Source terms for two-phase flow in long pipelines
following an accidental breach. International Conference
and Workshop on Modeling and Mitigating the
Consequences of Accidental Releases of Hazardous
Materials.
15. J.C.Stager, 1987. Silent death from Cameroon’s killer
lake. National Geographic, September.
16. S.E.Gant and G.T.Atkinson, 2011. Dispersion of the
vapour cloud in the Buncefield Incident. Process Safety
and Environmental Protection, 89 pp391-403.
17. S.E.Gant and A.Kelsey, 2012. Accounting for the effect
of concentration fluctuations on toxic load for gaseous
releases of carbon dioxide. J. Loss Prev. Process Ind., 25,
pp52-59.
18. P.J.Roache, 1998. Verification and validation in
computational science and engineering. Hermosa
Publishers.
18. P.J.Roache, 1998. Verification and validation in
computational science and engineering. Hermosa
Publishers.
19. NIOSH (1995) Documentation for immediately
dangerous to life or health concentrations (IDLH).
NIOSH chemical listing and documentation of revised
IDLH values (as of 3/1/95), NTIS Publication No. PB-
94-195047, National Institute for Occupational Safety
and Health (NIOSH).
20. S.Coldrick, C.J.Lea, and M.J.Ivings, 2010. Validation
database for evaluating vapor dispersion models for
safety analysis of LNG facilities: Guide to the LNG
model validation database. Fire Protection Research
Foundation, Quincy, MA, USA. (http://www.nfpa.
org/assets/files/pdf/research/lng_database_guide.pdf,
accessed May 2012).
 n
MaY 15–16 2013
io
st y
ut
di op

Houston Marriott WestcHase Hotel

rib
or c

Houston, tX, usa

t f ple
no m
Sa

Exhibition and sponsorship

options available, visit
www.clarion.org
3rd Quarter, 2012 213

The saturation pressure and the

design of dense-phase CO2 pipelines
by Dr Andrew Cosham
Atkins, Newcastle upon Tyne, UK

C arbon capture and storage (CCS) involves capturing CO2 from large industrial emitters, such as coal-
and gas-fired power stations, and transporting it to permanent storage sites, such as depleted oil or
gas reservoirs, or saline aquifers. Pipelines will be required to transport the CO2 from the emitter to the
storage site.The successful implementation of CCS depends upon avoiding unnecessary costs in either the
capture, transport or storage infrastructure.

A capture plant can be designed to produce a CO2-rich stream that contains a relatively high proportion
of other components (‘impurities’) or a relatively low proportion. Removing the ‘impurities’ from the
stream increases the cost of the capture plant. A pipeline can be designed to transport a wide range of
different CO2-rich mixtures. There may, however, be a cost-penalty associated with transporting higher

n
concentrations of ‘impurities’.

io
Pipelines must be designed to control running (propagating) fractures. Pipelines designed to transport
st y
ut
CO2 are susceptible to running-ductile fractures. The decompression behaviour of liquid or dense-phase
di op

CO2 and CO2-rich mixtures is very different from that of lean or rich natural gas. The different behaviour
rib
has significant implications for the design of pipelines to prevent running-ductile fractures. Some changes
or c

in thinking are required. Nevertheless, trends can be identified and simple relationships can be developed.
t f ple

The susceptibility to a running-ductile fracture can be characterized by the ‘saturation pressure’ of the
CO2-rich mixture: the susceptibility increases as the saturation pressure increases.The saturation pressure
depends upon the composition of the CO2-rich mixture and the operating conditions of the pipeline,
no m

and the saturation pressure increases as the concentration of ‘impurities’ increases. The geometry, grade,
and toughness of a pipeline define a limit on the saturation pressure of a CO2-rich mixture that can be
Sa

transported in the pipeline. In practical terms, the saturation pressure is limited by the wall thickness of
the linepipe. The wall thickness will increase as the concentration of ‘impurities’ increases.

Therefore, a pipeline designed to transport a relatively high proportion of ‘impurities’ will cost more than
one designed to transport a relatively low proportion. Conversely, the capture plant will cost less.Techno-
economic studies would be required to optimize the design of the system. A simple relationship between
the design pressure, the saturation pressure, and the required wall thickness is described. It is suitable for
use in techno-economic studies.

T he climate is changing [1]. The fourth report

from the Intergovernmental Panel on Climate Change
(IPCC) states that there is a “very high confidence that the
gases need to be reduced in order to limit the rise in average
global temperatures to a ‘safe’ level [1-3].

global average net effect of human activities since 1750
has been one of warming” [1]. Emissions of anthropogenic
(man-made) carbon dioxide (CO2) and other ‘greenhouse’
This paper was presented at the Third International Forum on the Transportation of
CO2 by Pipeline, organized by Newcastle University, Tiratsoo Technical, and Clarion
Technical Conferences, and held in Newcastle in June, 2012.
Author’s contact details:
tel: +44 (0)191 229 9522
email: andrew.cosham@atkinsglobal.com
Carbon capture and storage (CCS) is one of the technologies
that has been proposed as a means of reducing emissions
of anthropogenic carbon dioxide to the atmosphere [2-5].
CCS involves capturing carbon dioxide from large industrial
emitters, such as coal- and gas-fired power stations, and
transporting it to permanent storage sites, such as depleted
oil or gas reservoirs, or saline aquifers. CCS is an example
of a bridging technology in the transition to a sustainable,
low-carbon economy. The International Energy Agency has
214
estimated that the widespread adoption of CCS technologies
could contribute approximately 20% of the reduction in
emissions required to reduce projected global emissions in
2050 to their 2003 levels [6, 7]. One of the factors that will
determine whether or not CCS is indeed widely adopted is
the cost of implementation.
Pipelines can be expected to play a significant role in
the transportation infrastructure that is required for the
successful implementation of CCS [5].
The transportation of carbon dioxide by pipeline is an
established technology. The pipeline industry in the US
has 40 years experience of the design, construction, and
operation of CO2 pipelines [5, 8-10]. The Canyon Reef
Carriers pipeline system in West Texas, the first onshore
CO2 pipeline, was commissioned in 1972 [11], and there
are now over 6,611km (4,111 miles) of high-pressure CO2
pipelines in North America [10].
CCS introduces a number of differences, the most notable
being the routieng of pipelines through more densely
n
populated areas and ‘impurities’. The anthropogenic CO2-
rich mixture captured at an emitter may contain higher levels
io
of other components (sometimes referred to as ‘impurities’)
st y
ut
or different components, than is typical of the existing sources
di op
of naturally occurring or anthropogenic CO2 [5, 9, 12-14].
rib
The composition of the CO2-rich mixture produced by the
or c
capture plant depends upon the fuel and the technology
used to capture the CO2: post-combustion will produce the
t f ple
highest concentration of CO2; pre-combustion and oxy-fuel
will typically produce lower concentrations of CO2.
no m
The capture plant will produce a stream of nearly pure CO2,
and this stream must be suitable for transportation and
Sa
storage. Components that may be present in the CO2-rich
stream include: hydrogen, oxygen, argon, nitrogen, methane,
water, carbon monoxide, hydrogen sulphide, and oxides of
sulphur and nitrogen (SOx and NOx). Other components
may be present at near-trace levels. A simplistic approach
would be to say that the stream must be ‘very’ pure (say,
greater than 99.5 mol.% CO2). However, removing all of
the ‘impurities’ from the stream will increase the cost of the
capture plant. Therefore, it is necessary to understand the
effect of each component on the transportation and storage
infrastructure. This then allows a ‘quality specification’ to
be developed for CCS. In developing such a specification it
is important to ‘engineer out’ potential problems in order
to avoid adding excessive costs to the capture, transport,
or storage infrastructure. A simple example is water. A
pipeline can be design to transport a corrosive fluid but, if
the pipeline is a transmission line (i.e. long) it is better to
ensure that the fluid is dry prior to it entering the pipeline.
CCS will likely involve the development of pipeline networks
transporting CO2-rich mixtures from a number of different
emitters to one or more storage sites. Network solutions are
more efficient that point-to-point solutions. The quality
The Journal of Pipeline Engineering
specification then becomes the entry-specification for this
network.
A pipeline can be designed to transport a wide range of
different CO2-rich mixtures. There may, however, be a cost
penalty associated with designing a pipeline to transporting
higher concentrations of ‘impurities’. A simple example is
hydrogen sulphide. If the partial pressure of H2S is above
the limit specified in BS EN ISO 15156-2: 2009 [15], then
sour-resistant steels would need to be specified. Sour-resistant
steels are more expensive than conventional steels.
The operator of a pipeline is responsible for the safe
design, construction, installation, operation, maintenance,
and abandonment of the pipeline. The limits in the
entry-specification will follow from consideration of the
effect of the different impurities on issues such as safety
(some of the components are toxic at low concentrations),
hydraulic efficiency, water solubility, hydrate formation, and
corrosivity. Race et al. [16]1 review several entry-specifications
and discuss the effects of the various impurities on the
design, etc., of pipelines.
As an example of an entry-specification, the DYNAMIS CO2
quality recommendations [17] limit the total concentration
of the non-condensables (H2, O2, CO, Ar, N2, and CH4)
to 4 mol.%, and there are also individual limits on the
concentration of O2 and CO.
Avoiding unnecessary costs in the capture, transport,
or storage infrastructure implies that it is necessary to
understand what the costs are and what affects the costs.
A simple example is hydraulic efficiency. The higher the
concentration of impurities, the greater are the pumping
or compression requirements [9, 13, 14, 16].
Fracture control is a more complicated issue, but similarly,
it can be shown that the requirement that a pipeline be
designed to prevent running fractures means that there is a
relationship between the wall thickness and the composition
of the CO2-rich mixture that can be transported in the
pipeline. The higher the concentration of impurities, the
thicker the pipe that is required. Increasing the thickness
of a pipeline increases the cost of the pipeline. There is the
simple increase in the tonnage of the steel, but also increased
costs associated with construction (handling, welding, etc.).
The intermediate steps in the relationship between the
required wall thickness and the composition of the CO2-
rich mixture are summarized below, and are developed in
more detail through the paper:
• The susceptibility to a running-ductile fracture can
be characterized by the saturation pressure of the
CO2-rich mixture. The susceptibility increases as
the saturation pressure increases.
1. Published in this issue on pages 173–190.
3rd Quarter, 2012
• The saturation pressure depends upon the composition
of the CO 2-rich mixture and the operating
conditions of the pipeline. The saturation pressure
increases as the concentration of ‘impurities’
increases.
• The saturation pressure of the mixture must be less
than the arrest pressure of the pipe.
• The arrest pressure of the pipe depends upon the
geometry (diameter and wall thickness), grade and
toughness of the pipe.
• The required wall thickness of the pipe depends
upon the saturation pressure of the mixture.
• The required wall thickness of the pipe depends
upon the composition of the CO2-rich mixture and
the operating conditions of the pipeline.
Fracture control2
Pipelines transporting gaseous fluids, two-phase fluids,
dense-phase fluids, or liquids with a high vapour pressure
are susceptible to long-running (propagating) ductile
(shear) fractures. Running fractures are prevented by
n
specifying appropriate toughness requirements in the
pipe body or, if linepipe with the required toughness is
io
not available, by installing mechanical crack arrestors at
st y
ut
regular intervals. The prevention of running fractures is a
di op
design requirement – see PD 8010-1: 2004, Code of practice
rib
for pipelines – Part 1: Steel pipelines on land, Section 8.2.5
or c
[18]. ASME B31.4-2009, Pipeline transportation systems for
liquid hydrocarbons and other liquids, Section 402.5 Fracture
t f ple
propagation in carbon dioxide pipelines gives specific guidance
(see below) [19].
no m
Carbon dioxide in the liquid or dense-phase is a high vapour
pressure liquid. Pipelines designed to transport CO2 or
Sa
CO2-rich mixtures in the dense-phase are susceptible to
running shear fractures.
The toughness of linepipe steel is characterized using the
shear area measured using a drop-weight tear-test (DWTT)
specimen and the impact energy measured using a Charpy
V-notch specimen. Toughness requirements have been
incorporated into linepipe specifications, such as ANSI/
API Specification 5L [20]. The shear-area requirement
ensures that the linepipe steel is on the upper shelf, and
hence that brittle fracture propagation will not occur. The
impact energy requirement ensures that ductile fracture
propagation will not occur.
The shear -area requirement in codes and standards (i.e.
a minimum shear area of 85%) is applicable to CO2 (and
CO2-rich mixtures). However, the tabulated toughness
(impact energy) requirements are not applicable to CO2
(or CO2-rich mixtures).
2. Fracture control includes both fracture initiation control and fracture propagation
control. The subject here is fracture propagation control. The toughness requirements for
fracture-propagation control are more onerous than those for fracture-initiation control.
215
The North American experience of controlling running-ductile
fractures is informative. Mechanical crack arrestors are used
on most of the CO2 pipelines in the US to limit the length of
a running fracture. Crack arrestors are used on the Canyon
Reef Carriers pipeline (1972), the Cortez pipeline (1982-4), the
Sheep Mountain pipeline (1983), the Bravo pipeline (1984), the
McElmo Creek pipeline (1984-5), the Central Basin pipeline
(1985), and the North East Jackson Dome (formerly Choctaw)
pipeline (1986) [5, 12, 16, 21]3. Steel sleeves and composite
sleeves (ClockSpring) have been used as arrestors. The typical
spacing between each arrestor is 300-500m. In several cases (for
example, Canyon Reef Carriers), the crack arrestors were retro-
fitted to the lines [11, 22]. This was necessary because the fact
that CO2 pipelines were susceptible to running fractures was
not recognized at the time that these pipelines were constructed.
The Dakota Gasification (Souris Valley) pipeline, constructed
in 2000, does not use mechanical crack arrestors: the
toughness (Charpy V-notch impact energy) of linepipe steel
has significantly improved over the years, and is significantly
higher than that of 1970s or 80s vintage steel.
Design codes
The CO2 pipelines in the US have been designed to ASME
B31.4, a design code for pipelines transporting liquid
hydrocarbons and other liquids.
The pipeline industry in the US has 40 years experience of
designing CO2 pipelines. This is reflected in the design code,
ASME B31.4. It contains specific guidance on designing CO2
pipelines to prevent running fractures. However, perhaps
reflecting the fact that the pipelines are routed through largely
unpopulated areas, there is no other specific guidance in the code.
Other design codes give very little guidance, reflecting the lack
of experience outside of the US.
Carbon dioxide is quoted as an example of a ‘category C’
fluid in PD 8010-14. This classification is consistent with BS
EN 14161: 2003 (ISO 13623:2000 modified) [23] and DNV-
OS-F101, 2007 [24]. However, as CO2 is toxic, it would perhaps
be more appropriate to categorise it as a ‘category E’ fluid5.
Section 8.2.5 of PD 8010-1 states that “pipelines conveying
category D and category E fluids should meet the Charpy
energy values in BS EN 10208-1, BS EN 10208-2 and/or
BS ISO 3183-3.” The toughness requirements tabulated in
BS EN 10208-2 and BS ISO 3183-3 are based on the EPRG
Recommendation for crack arrest toughness for high-strength
linepipe steels [25-27], and are only applicable to pipelines
transporting lean, dry natural gas. It is highlighted in the
standards that the requirements for rich gas or two-phase
fluids may be different.
3. This is not an exhaustive list.
4. Category C fluids are: “Non-flammable fluids that are non-toxic gases at ambient
temperature and atmospheric pressure conditions.”
5. Category E fluids are: “Flammable and/or toxic fluids that are gases at ambient
temperature and atmospheric pressure conditions and are conveyed as gases and/
or liquids.”
216 The Journal of Pipeline Engineering

a. 100
CO2+N2
90
80
70
pressure, barg

60
50
40
10 mol.%
30
8
20 6
4
10 2
0
0
-40 -30 -20 -10 0 10 20 30 40
temperature, °C

b. 100 100
C1+C3

n
80 80

io
st y
pressure, barg

pressure, barg

60 ut 60
di op
rib
40 40
or c

10 mol.% 10 mol.%
8 8
t f ple

20 6 20 6
4 4
100 barg, 30°C

100 barg, 5°C

2 2
610x15.9 mm, L450 0 610x10.3 mm, L450 0
0 0
no m

0 100 200 300 400 0 100 200 300 400

velocity, m.s-1 velocity, m.s-1
Sa

c. 200 100
CO2+N2 C1+C3
180 t = 9.5 mm
90
t = 10.3 mm
160 80
1/1 CVN impact energy, J

t = 15.9 mm
1/1 CVN impact energy, J

140 70
120 60
100 50
80 40
60 30
40 20
20 10
610x15.9 mm, L450 610 mm, L450
0 0
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
concentration, mol.% concentration, mol.%

Fig.1.The effect of composition on: (a) the phase boundary, (b) the decompression behaviour, and (c) the toughness required
to arrest a running ductile fracture, illustrated for a binary mixture of carbon dioxide and nitrogen, and one of methane and
propane (a rich gas).
3rd Quarter, 2012
Similarly, the tabulated requirements in DNV-OS-F101,
Submarine pipeline systems [24], including those in
‘Supplementary requirement, fracture arrest properties
(F)’ are not applicable to CO2.
Annex G of ANSI/API Specification 5L, Specification for
Line Pipe [20], lists five different methods for determining
the pipe body toughness requirements to prevent running-
ductile fractures. Three are only applicable to lean natural
gas, while the other two, ‘G.9 Battelle Two-Curve method
Approach 3’ and ‘G.11 Full-scale burst testing Approach
5’, are applicable. However, it is important to note that the
Two-Curve model (see below) has not been validated for
application to CO2 (or CO2-rich mixtures).
DNV-RP-J202, Design and operation of CO2 pipelines, Section
5.5 Running ductile fracture control also gives specific guidance
[28]. The guidance is consistent with ASME B31.4 and
Annex G of ANSI/API Specification 5L. It also notes that
this is an area of ongoing research, and that “the latest
knowledge on this subject [should be employed] in the
development of … pipeline systems.”
n
ASME B31.4, Section 402.5 provides probably the
io
simplest and most concise summary of the issues: “The
st y
ut
possibility of brittle and ductile propagating fracture shall
di op
be considered in the design of carbon dioxide pipelines.
rib
… Ductile fracture propagation shall be minimized by the
or c
selection of a pipe steel with appropriate fracture toughness
and / or by the installation of suitable crack arrestors.
t f ple
Design considerations shall include pipe diameter, wall
thickness, fracture toughness, yield strength, operating
pressure, operating temperature, and the decompression
no m
characteristics of carbon dioxide and its associated
impurities.”
Sa
However, whilst the issues are identified and some guidance
is given, there are, as yet, no specific requirements for CO2
or CO2-rich mixtures that are analogous to the EPRG
Recommendation for crack arrest toughness for high strength line
pipe steels (for lean natural gas).
The decompression behaviour
of carbon dioxide
Carbon dioxide is a vapour at a pressure of 1bara and
a temperature of 15°C (i.e. standard pressure and
temperature). The critical point of carbon dioxide is
73.77bara (72.76barg) and 30.98°C.
The saturation pressure of CO2 at a temperature of 15°C
is 49.86barg. At higher pressures, CO2 is a liquid. CO2
can exist as a vapour (gas) or a liquid within the typical
operating temperature range of a pipeline. CO2 can be
transported at high pressures, as a liquid or a ‘dense-phase’
fluid, or at low pressures, as a gas. The typical CO2 pipeline
is designed to transport CO2 at high pressures, well above
the critical pressure.
217
In order to determine the toughness required to arrest a
running ductile fracture it is necessary to understanding the
decompression behaviour of the fluid in the pipeline.
The decompression behaviour of CO2 in the liquid or dense-
phase is different to that in the vapour phase, and different
from that of a lean or rich natural gas [9, 30-37]. Here, the
subject of interest is the decompression of CO2 and CO2-rich
mixtures in the liquid or dense-phase. The decompression
behaviour that is relevant to running-ductile fractures is that
which occurs immediately following a rupture.
A rupture initiates an expansion wave in the fluid that
propagates through the fluid in both directions away from the
rupture. The flow out of the opening is choked. The leading
edge of the decompression wave travels at the speed of sound
in the fluid at the initial pressure and initial temperature, less
the outflow velocity toward the rupture.
The decompression is rapid and it is therefore generally
considered to be an isentropic expansion process. An isentropic
process is one in which the entropy is constant.
The decompression behaviour of a fluid can be measured in a
shock-tube test. National Grid has undertaken a programme
of shock-tube tests to measure the decompression behaviour
of CO2 and CO2-rich mixtures in the liquid and dense
phase [29-37]. These experiments have confirmed that the
decompression behaviour can be conservatively modelled by
assuming an isentropic expansion process [36, 37].
The decompression behaviour of a fluid is characterized by
a decompression curve, a plot of pressure verses the velocity
of the decompression wave. The decompression behaviour
depends upon the composition of the fluid, and the initial
pressure and the initial temperature.
Figure 1 illustrates the decompression behaviour of CO2 and
a CO2-rich mixture, a binary mixture (two components) of
carbon dioxide and nitrogen6. Figure 1a shows the pressure-
temperature phase diagram for CO2 and the binary mixture.
A phase diagram describes the phases and stable equilibrium
states of a substance, and the phase boundaries mark the
dividing lines between the different phases. The liquid-
vapour phase boundary is shown in Fig.1a: the addition of
other components (‘impurities’) changes the thermodynamic
properties of the fluid, and the phase boundary changes. In
Fig.1a, increasing the concentration of nitrogen increases the
range of pressures and temperatures over which the liquid and
vapour phases can exist in equilibrium. The bubble-point curve,
the dividing line between the liquid phase and the two-phase
region, is at a higher pressure in the binary mixture than in
the pure substance. This is important, as described below.
6. The decompression behaviour of CO2 and CO2-rich mixtures has been calculated
using DECOM [34, 36, 37]. It adopts identical assumptions to the models developed
for natural gas, such as GASDECOM and DECAY, but uses ‘reference’ (i.e. the most
accurate) equations of state for CO2 and CO2-rich mixtures. The decompression
behaviour of the binary mixture of methane and propane has been calculated using
GASECOM [38].
218 The Journal of Pipeline Engineering

100 100
CO2+N2 CO2+N2

80 80
saturation pressure, barg

saturation pressure, barg

60 60

40 40
10 mol.% 10 mol.%
8 8
20 6 20 6
4 4
2 2
pi = 100 barg 0 pi = 150 barg 0
0 0
0 10 20 30 40 0 10 20 30 40
initial temperature, °C initial temperature, °C

Fig.2.The effect of composition, and the initial pressure and initial temperature on the saturation pressure of a binary mixture
of carbon dioxide and nitrogen.

n
Also shown in Fig.1a are the isentropic decompression The decompression behaviour of CO2 or a CO2-rich mixture
paths for an initial pressure of 100barg and an initial in the liquid (or dense) phase is characterized by a ‘long’

io
temperature of 30°C7, 8. The corresponding decompression plateau. This is important, as described below.
st y
curves are shown in Fig.1b: the decompression curves are
ut
di op

characterised by a ‘long’ plateau, and the pressure at this The toughness required to
arrest a running ductile fracture
rib
plateau or discontinuity is the ‘saturation pressure’, i.e.
or c

the pressure at which the decompression path intersects

the bubble-point curve. The pressure at which the The toughness required to arrest a running ductile fracture
t f ple

decompression path intersects the bubble-point curve depends upon the geometry and grade of the pipe, and the
increases as the concentration of nitrogen increases, see decompression behaviour of the fluid.
Fig.1a. Consequently, the pressure at the plateau in the
no m

decompression curve increases, see Fig.1b. The Two-Curve model, developed by the Battelle Memorial
Institute in the 1970s [38-40], is an example of a widely used
Sa

Figure 1b also shows decompression curves for methane
and a binary mixture of methane and propane. The initial
pressure is 100barg and the initial temperature is 5°C9.
The gas becomes ‘richer’ as the concentration of propane
increases. The decompression curves for the binary mixture
are characterized by a ‘short’ plateau: the pressure at the
plateau increases as the concentration of propane increases
(i.e. as the gas becomes richer).
The decompression behaviour of a lean gas is characterized
by no plateau or a ‘very short’ plateau.
The decompression behaviour of a rich gas is characterized
by a ‘short’ plateau.
semi-empirical model that defines the relationship between
geometry, grade and toughness, and the decompression
behaviour of the fluid. The model is an uncoupled model, and
assumes that the decompression behaviour and the dynamic
crack propagation behaviour are uncoupled processes, i.e. the
properties of the fluid can be considered separately from the
fracture resistance of the linepipe steel. The driving force for
a propagating fracture is represented by the decompression
curve, a relationship between pressure and the velocity of the
decompression wave. The resistance to a propagating fracture
is represented by the fracture-velocity curve, a relationship
between pressure and the velocity of the running fracture.
The minimum toughness required to arrest a running-
ductile fracture (the arrest toughness) is that at which the
decompression curve and the fracture-velocity curve are tangent.

7. The decompression behaviour of a CO2-rich mixture becomes more severe as The arrest pressure is the pressure below which a running
the initial temperature increases. Therefore, a high initial temperature is (in simple
terms) a more severe case. ductile fracture cannot be sustained, and corresponds to the
8. The isentropic decompression paths for the binary mixtures with 8 and 10 mol.% condition where there is insufficient energy in the system
nitrogen are based on initial temperatures of 28 and 22°C. The decompression
paths intersect the bubble-point curve at the cricondenbar (the maximum pressure
to drive a propagating fracture. The fracture-velocity curve
on the phase boundary). The resulting decompression curves are more severe than tends to the arrest pressure as the velocity tends to zero. This
those for an initial temperature of 30°C (as illustrated, in terms of the saturation is important, as described below. The arrest pressure is a
pressure, in Fig.2).
9. The decompression behaviour of a lean or rich gas becomes more severe as the initial function of the geometry, grade, and toughness.
temperature decreases. Therefore, a low initial temperature is a more severe case.
3rd Quarter, 2012
The Two-Curve model has been validated in full-scale fracture
propagation tests with lean and rich natural gas, but it has
not been validated in tests with CO2 or CO2-rich mixtures.
For this, National Grid is undertaking a programme of full-
scale tests to validate the model [29].
Consider a pipeline designed to transport a CO2-rich mixture
and a pipeline designed to transport a natural gas. Assume
that the design pressure of the pipeline transporting the CO2-
rich mixture is 150barg, and that of the pipeline transporting
natural gas is 100barg. Assume 610-mm (24-in) diameter and
Grade L450, and a design factor of 0.72. The wall thickness
of the pipeline transporting the CO2-rich mixture is then
15.9mm and that of the pipeline transporting natural gas
is 10.3mm (rounding-up to a ‘standard’ pipe size in both
cases). The toughness required to arrest a running ductile
fracture (the arrest toughness) depends on the decompression
behaviour of the fluid. Figure 1b shows the decompression
curves for these two pipelines10.
Also shown in Fig.1b are the fracture-velocity curves for the
two geometries, a 610 x 15.9-mm, Grade L450 pipe and a
n
610 x10.3-mm, Grade L450 pipe. Figure 1c is a plot of the
arrest toughness versus the composition of the mixtures: the
io
arrest toughness increases as the concentration of nitrogen
st y
ut
(in the CO2-rich mixture) or propane (in the rich gas)
di op
increases. The fracture-velocity curves in Fig.1b are plotted
rib
for the arrest toughnesses in Fig.1c.
or c
The plateau in the decompression curve for a binary mixture
t f ple
of methane and propane (a rich gas) is ‘short’. At low
concentrations of propane (a lean gas) the fracture-velocity
curves are tangent above the plateau in the curve, and
no m
the fluid is in the vapour phase. The level of the plateau
increases as the concentration of propane increases. At high
Sa
concentrations of propane, the fracture-velocity curves are
tangent below the plateau in the curve, and the fluid is two-
phase, a vapour with droplets of liquid. The arrest toughness
increases as the concentration of propane increases, see
Fig.1c. At first this increase is gradual, but there is then a
relatively large increase in toughness as the two curves move
from being tangent above the plateau to below the plateau.
The plateau in the decompression curve for a binary
mixture of carbon dioxide and nitrogen is ‘long’. At low
concentrations of nitrogen the fracture-velocity curves are,
effectively, tangent when the saturation pressure equals
the arrest pressure. The level of the plateau increases as
the concentration of nitrogen increases. The length of the
plateau also decreases. At high concentrations of nitrogen,
the fracture-velocity curves are tangent below the plateau in
the curve. The fluid is two-phase, a liquid with bubbles of
vapour. A conservative simplification has been adopted here,
10. The decompression behaviour of a CO2-rich mixture becomes more severe as
the initial pressure decreases. An initial pressure of 100barg is then a reasonable
assumption for a pipeline with a design pressure of 150barg. The decompression
behaviour of a lean or rich gas becomes more severe as the initial pressure increases.
An initial pressure of 100barg is then a reasonable assumption for a pipeline with
a design pressure of 100barg.
219
and it is assumed that the minimum toughness required
to arrest a running-ductile fracture (the arrest toughness)
is that at which the saturation pressure is equal to the
arrest pressure. This is a reasonable assumption given that
the Two-Curve model has not been validated in tests with
CO2 or CO2-rich mixtures. The arrest toughness increases
as the concentration of nitrogen increases, see Fig.1c. The
increase in toughness is non-linear, and is asymptotic. This
is important, as subsequently described below.
The toughness required to arrest a running-ductile fracture
in a pipeline designed to transport a CO2-rich mixture is
higher than in that designed to transport a natural gas,
because of the decompression behaviour of the CO2-rich
mixture. The plateau in the decompression curve for the
natural gas is ‘short’. The plateau in the decompression
curve for the CO2-rich mixture is ‘long’.
The saturation pressure
The decompression behaviour of CO2 or a CO2-rich mixture
in the liquid (or dense) phase is characterized by a ‘long’
plateau. The pressure at this plateau is the ‘saturation
pressure’. The toughness required to arrest a running-ductile
fracture corresponds to the toughness at which the ‘arrest
pressure’ is equal to the ‘saturation pressure’. The toughness
required to arrest a running ductile fracture increases as the
‘saturation pressure’ increases.
• The susceptibility to a running ductile fracture can
be characterised by the saturation pressure of the
CO2-rich mixture. The susceptibility increases as
the saturation pressure increases.
The saturation pressure
and composition
The saturation pressure is the pressure at which the
decompression path intersects the bubble-point curve. The
decompression path depends upon the composition of the
fluid, and the initial pressure and the initial temperature;
the bubble-point curve depends upon the composition of
the fluid.
Figs 1a and 1b illustrate these dependencies with reference
to a binary mixture of carbon dioxide and nitrogen, in
terms of the phase diagram and the decompression curve,
respectively.
Figure 2 shows the effect of composition, and the initial
pressure and initial temperature, on the saturation pressure,
again for a binary mixture of carbon dioxide and nitrogen,
in terms of a plot of the saturation pressure versus the
initial temperature. The saturation pressure is plotted for
an initial temperature from 0°C to 40°C, and an initial
pressure of 100barg and 150barg. The saturation pressure
increases as the concentration of nitrogen increases. The
saturation pressure increases to a maximum as the initial
220
100
CO2+H2
CO2+N2
90 CO2+CH4
saturation pressure, barg
80
70
60
pi = 100 barg, Ti = 30°C
50
0 1 2 3 4 5 6 7 8
concentration, mol.%
Fig.3.The effect of composition on the maximum saturation pressure, illustrated for binary mixtures of carbon dioxide and
hydrogen, carbon dioxide and nitrogen, and carbon dioxide and methane.
n
temperature increases, and then it decreases, because of
the effect of increasing the concentration of nitrogen on
io
the size and shape of the phase boundary (see Fig.1a). The
st y
maximum saturation pressure is equal to the cricondenbar
ut
di op
(the maximum pressure on the phase boundary). The
rib
saturation pressure for an initial pressure of 100barg is higher
or c
than that for a saturation pressure of 150barg.
t f ple
The saturation pressure increases (to a maximum) as the
initial temperature increases.
no m
The saturation pressure increases as the initial pressure
decreases.
Sa
The susceptibility to a running-ductile fracture increases as
the saturation pressure increases.
Therefore, a pipeline should be designed to the maximum
saturation pressure within its operating envelope. The effect
of the initial pressure on the saturation pressure indicates
that the relevant pressure is the minimum pressure within
the operating envelope. The implication of this is that
the saturation pressure is, effectively, independent of the
maximum operating pressure (the design pressure). The
effect of the initial temperature on the saturation pressure
indicates that the relevant temperature is the maximum
temperature within the operating envelope, or if it is lower,
the temperature at which the saturation pressure is equal
to the cricondenbar. A maximum saturation pressure can
be defined within the operating envelope of a pipeline.
The maximum saturation pressure depends on the
composition of the CO2-rich mixture. Figure 3 is a plot of
the maximum saturation pressure for three different binary
mixtures: carbon dioxide and hydrogen, carbon dioxide and
nitrogen, and carbon dioxide and methane. The maximum
The Journal of Pipeline Engineering
100
CO2+H2
CO2+N2
90 CO2+CH4
saturation pressure, barg
80
70
60
pi = 100 barg, Ti = 40°C
50
9 10 0 1 2 3 4 5 6 7 8 9 10
concentration, mol.%
saturation pressure is plotted for initial temperatures of
30°C and 40°C (i.e. the maximum temperature within the
operating envelope of the hypothetical pipeline). The initial
pressure is 100barg (i.e. the minimum pressure within the
operating envelope). The saturation pressure increases as the
concentration of hydrogen, nitrogen or methane increases.
As the concentration of the second component increases,
the saturation pressure tends to the cricondenbar for that
mixture. At high concentrations of hydrogen or nitrogen, the
maximum saturation pressure for an initial temperature of
30°C is identical to that for 40°C. At lower concentrations,
it is higher at an initial temperature of 40°C.
Figure 3 also shows that some components have more of an
effect on the saturation pressure than other components.
Hydrogen has more of an effect that nitrogen, which has
more of an effect than methane.
Although the discussion so far has been limited to binary
mixtures (two components), the observations are applicable
to higher-order mixtures (i.e. more than two components).
The maximum saturation pressure of a CO2-rich mixture
containing hydrogen and nitrogen (three components) is
bounded by that for binary mixtures of hydrogen and nitrogen
(as in Fig.3). Similarly, that for a mixture containing hydrogen,
nitrogen and methane (four components) is bounded by that
for binary mixtures of hydrogen and methane (as in Fig.3).
The effect of the composition of a CO2-rich mixture on
the maximum saturation pressure, and the implications
for the entry-specification for a hypothetical pipeline, can
be illustrated in simple terms:
Figure 4 illustrates, in simple graphical terms, the relationship
between the composition of a CO2-rich mixture and the
maximum saturation pressure. Figure 4 is a conservative
3rd Quarter, 2012 221

a.
pi = 100 barg, Ti = 30°C

80
saturation pressure, barg

75

0.5
1 mol.%
70

b.
H2 N2 Ar CH4 O2 CO
saturation pressure, barg

pi = 100 barg, Ti = 40°C

n
io
st y
80
0.5
ut
di op

1 mol.%
rib
or c

75
t f ple

H2 N2 Ar CH4 O2 CO
no m

Fig.4. A simple illustration of the effect of composition on the maximum saturation pressure.
Sa

simplification of Figure 3, but extended from binary to
higher order mixtures11. The coloured blocks represent the
different non-condensables (i.e. H2, CO, O2, N2, Ar, and
CH4). The height of each of the coloured blocks in Fig.4
represents the relative contribution of each component to
the saturation pressure of the mixture: the taller blocks have
a greater effect on the saturation pressure (and hence the
toughness requirements) than the shorter blocks. Hydrogen
has the most significant effect on the saturation pressure of
the mixture (a block representing 1 mol.% of hydrogen is taller
than the other blocks). The total height of a stack of blocks is
the saturation pressure of the mixture. The number of (half
or full size) blocks in a stack of blocks is the concentration
of the non-condensables in the mixture.
The initial temperature in Fig.4a is 30°C, whilst that in
Fig.4b is 40°C. The initial pressure is 100barg. The vertical
axes in the two figures are consistent (although the origins of
the axes are different). The saturation pressure of a mixture
11. Figure 4 is a simplification: the saturation pressure of a mixture is slightly less
than that implied in the figure. Nevertheless, Fig.4 correctly represents the trends
and implications of a limit on the saturation pressure.
at an initial temperature of 40°C is higher than that of an
otherwise identical mixture at an initial temperature of 30°C.
Consider a pipeline that is designed to transport CO2-rich
mixtures with a saturation pressure that does not exceed
80barg. Figure 4 is used to illustrate the implications of
this limit on the saturation pressure on the composition of
mixtures that can be transported in the pipeline:
The entry-specification for the pipeline will limit the
concentration of some of the individual non-condensables.
CO is toxic. The concentration might be limited to of the
order of 0.2 mol.%, or possibly lower (the DYNAMIS CO2
quality recommendations recommend a limit of 0.2 mol.%
[17]). The concentration of O2 might be limited to less than
100ppmv because of the effect of O2 on microbial activity in
a storage site [17, 41]. Therefore, only H2, N2, Ar, and CH4
need be considered in this illustration.
The entry-specification will also limit the total concentration of
non-condensables (because of the effect on the thermodynamic
and transport properties of the CO2-rich mixture); the
222
DYNAMIS CO2 quality recommendations recommend a limit
of 4.0 mol.%. Therefore, in terms of Fig.4, the height of a
stack of blocks is limited to that equivalent to a saturation
pressure of 80barg, and the number of (full-size) blocks is
limited to four (i.e. 4 mol.%). The following observations
can then be made (and assuming, unless otherwise indicated,
an initial temperature of 30°C):
• In a CO 2-rich mixture containing only H 2,
the concentration of H2 is limited to less than
2.5 mol.% (but, for Ti = 40°C, to less than 1.5 mol.%).
The concentration of non-condensable is therefore
limited to less than 4 mol.%.
• In a CO2-rich mixture containing 2 mol.% H2 and
one other component (i.e. a ternary mixture), the
concentration of Ar or N2 is limited to less than 1
mol.%, or the concentration of CH4 is limited to less
than 2 mol.%. The concentration of non-condensables
is therefore limited to less than 4 mol.%.
• In a CO2-rich mixture containing 1 mol.% H2 and
one other component (i.e. a ternary mixture), the
concentration of Ar is limited to approximately
n
2.5 mol.% (but, for Ti = 40°C, to less than 1 mol.%),
or the concentration of N2 is limited to less than
io
3 mol.% (but, for Ti = 40°C, to approximately
st y
ut
1 mol.%), or the concentration of CH4 is limited
di op
to 3 mol.% (but, for Ti = 40°C, to approximately
rib
2.5 mol.%). The concentration of non-condensable
or c
is therefore limited to less than 4 mol.% when the
third component is Ar or N2.
t f ple
• In a CO2-rich mixture containing 1 mol.% H2and
1 mol.% Ar and one other component (i.e. a
quadrenary mixture), the concentration of N2 is
no m
limited to less than 2 mol.%, or the concentration
of CH4 is limited to 2 mol.%. The concentration of
Sa
non-condensable is therefore limited to less than 4
mol.% when the fourth component is N2.
• In a CO2-rich mixture containing 1 mol.% H2,
1 mol.% Ar and 1 mol.% N2, and one other
component (i.e. a quinary mixture), the concentration
of CH4 is limited to 1 mol.%.
• In a CO2-rich mixture containing no hydrogen,
the concentration of non-condensable is limited to
4 mol.% (but, for Ti = 40°C, to less than 4 mol.%,
except when there is at least 2 mol.% CH4).
The composition of the CO2-rich mixture that can be
transported in the hypothetical pipeline (i.e. mixtures with
a saturation pressure that does not exceed 80barg) depends
on the maximum design (operating) temperature of the
pipeline (i.e. the initial temperature) and the composition
of the mixture. The latter follows from the fact that different
components have different effects. H2 has the most significant
effect on saturation pressure: a mixture that contains 2
mol.% H2 has a much higher saturation pressure than a
mixture that contains 2 mol.% CH4. The total concentration
of non-condensables in a mixture that contains hydrogen
may be limited to less than 4 mol.%.
The Journal of Pipeline Engineering
Increasing the maximum operating temperature (i.e. increasing
the initial temperature) decreases the concentration of non-
condensables that can be transported in a pipeline.
The implication of the limit on the saturation pressure is
that an entry-specification cannot simply be defined in terms
of the minimum concentration of CO2 or the maximum
concentration of non-condensables12. It is necessary to
calculate the saturation pressure of the mixture. Increasing
the concentration of non-condensables (i.e. removing fewer
impurities in the capture plant) will increase the saturation
pressure. Increasing the maximum operating temperature
(i.e. less after-cooling at the pump or compressor station) will
increase the saturation pressure.
A pipeline that is designed to a higher limit on the saturation
pressure would be able to transport a higher concentration
of non-condensables, subject to the total limit of 4 mol.%, or
an identical concentration at a higher maximum operating
temperature.
A pipeline can be design to a higher limit on the saturation
pressure by increasing the wall thickness. This is important,
as discussed below.
• The saturation pressure depends upon the composition
of the CO2-rich mixture and the operating conditions
of the pipeline. The saturation pressure increases as
the concentration of ‘impurities’ increases.
The saturation pressure
and arrest toughness
The toughness required to arrest a running-ductile fracture (the
arrest toughness) in a pipeline transporting a CO2-rich mixture
in the liquid or dense-phase depends upon the geometry
and grade of the pipeline, and the saturation pressure of the
mixture. The susceptibility to a running-ductile fracture can be
characterized by the saturation pressure of the CO2-rich mixture.
The requirement for a pipeline to have sufficient toughness to
arrest a running-ductile fracture limits the saturation pressure of
the CO2-rich mixtures that can be transported in the pipeline.
The saturation pressure is defined by the composition of the
mixture, and the initial pressure and temperature.
The arrest pressure is defined by the geometry, grade and
toughness of the pipe.
A pipeline has sufficient toughness to arrest a running-ductile
fracture if the arrest pressure is greater than the saturation
pressure. The limiting condition is when the arrest pressure
is equal to the saturation pressure. This is a conservative
simplification of the Two-Curve model.
12. The situation is analogous to the limit on the Wobble Number in the Gas Safety
(Management) Regulations 1996. The Regulations specify limits in terms of the
Wobble Number, the Incomplete Combustion Factor and the Soot Index, rather
than limits on the concentration of the individual components.
3rd Quarter, 2012 223

a. 350 350
610x15.9 mm, L450 355.6x9.525 mm, X70
300 300

1/1 CVN impact energy, J

1/1 CVN impact energy, J

Wilkowski et al. (1977, 2006) correction

250 250
Leis (2000), Eiber (2008) correction

200 200 Wilkowski et al. (1977, 2006) correction

Leis (2000), Eiber (2008) correction

150 150

100 100

50 50

0 0
50 60 70 80 90 50 60 70 80 90
saturation pressure, barg saturation pressure, barg

b. 350 350
610x15.9 mm, L450 355.6x9.525 mm, X70
300 300
1/1 CVN impact energy, J

1/1 CVN impact energy, J

Wilkowski et al. (1977, 2006) correction

250 250

n
Leis (2000), Eiber (2008) correction

io
200 200
st y
Wilkowski et al. (1977, 2006) correction
ut Leis (2000), Eiber (2008) correction
di op

150 150
rib
100 100
or c

50 50
t f ple

0 0
20 22 24 26 28 30 32 34 36 38 20 22 24 26 28 30 32 34 36 38
hoop stress, percent SMYS hoop stress, percent SMYS
no m
Sa

Fig.5.The relationship between the saturation (arrest) pressure, the hoop stress at the saturation pressure (the arrest stress)
and the arrest toughness.

The geometry, grade and toughness of the pipe and the
saturation pressure are related. The toughness required to
arrest a running-ductile fracture (the arrest toughness) can
be characterized in terms of the saturation pressure of the
CO2-rich mixture.
Figure 5a illustrates the relationship between the saturation
pressure (equal to the arrest pressure) and the arrest toughness for
610 x 15.9-mm, L-450 line pipe, and also for 355.6 x 9.525-mm,
X-70 line pipe (see below). The form of this relationship follows
from the conservative simplification of the Two-Curve model.
The relationship between the saturation pressure and the
arrest toughness is non-linear. The arrest toughness increases
asymptotically as the saturation pressure tends to a limiting value.
The limiting value of the saturation pressure depends on the
geometry and grade, and is independent of the toughness. It is
not possible to arrest a running-ductile fracture if the saturation
pressure is greater than this limiting value.
The relationship between the saturation pressure and the
arrest toughness can also be expressed in terms of the hoop
stress at the saturation pressure rather than in terms of the
saturation pressure. Figure 5b illustrates the relationship
between the hoop stress at the saturation pressure and the
arrest toughness: there is a limit to the hoop stress at the
saturation pressure, and this limiting value depends on
the grade of the pipe.
The arrest toughness in Fig.5 is expressed in terms of the
Charpy V-notch impact energy. The relationship between
Charpy V-notch impact energy and ductile-fracture
propagation resistance becomes increasingly non-linear
as the impact energy increases. This effect is illustrated
in Fig.5 by applying the Leis (2000) and Eiber (2008a,b)
[42-44] correction factor and the Wilkowski (1977, 2006)
[45, 46] correction factor. The correction factors do not
change the asymptote, but they do increase the toughness
as the saturation pressure tends to the limit.
224
The relationship between the toughness required to arrest
a running-ductile fracture and the saturation pressure is
dependent on the geometry and grade of the pipe, and is
independent of both the composition of the fluid and the
initial conditions; it is the saturation pressure that depends
on the composition of the fluid and the initial conditions.
• The saturation pressure of the mixture must be less
than the arrest pressure of the pipe.
• The arrest pressure of the pipe depends upon the
geometry (diameter and wall thickness), grade and
toughness of the pipe.
The saturation pressure
and wall thickness
The toughness required to arrest a running-ductile fracture
(the arrest toughness) in a pipeline transporting a CO2-rich
mixture in the liquid or dense-phase depends upon the
geometry and grade of the pipeline, and the ‘saturation
pressure’ of the mixture. The composition of the CO2-rich
mixture that can be transported in a pipeline depends on
n
the geometry, grade and toughness of the line pipe. The
relationship between the toughness and the saturation
io
pressure is non-linear.
st y
ut
di op
Modern, high-toughness linepipe steels are very tough.
rib
Full-size Charpy V-notch impact energies of over 250J are
or c
readily achievable, and at no additional cost.
t f ple
Consider again the 610 x 15.9-mm, Grade L-450 pipeline
designed to transport a CO2-rich mixture. The minimum
toughness of the line pipe steel will be at least 250J. A
no m
simple calculation would indicate that such a pipeline
could transport mixtures with a saturation pressure of 78.1-
Sa
80.7barg, depending on the correction factor (if any) that
is applied to the predicted toughness. The limiting value
of the saturation pressure for this grade and geometry is
81.2barg (34.6% SMYS). However, the non-linearity evident
in Fig.5 raises a number of questions. The applicability
of the Two-Curve model has not been demonstrated.
The asymptotic relationship might be real or it might
be an artefact of the simplifying assumption (that the
saturation pressure must be less than the arrest pressure).
In the absence of a full-scale fracture propagation test to
confirm that it is appropriate to operating a pipeline this
close to the asymptote, a bounding approach is warranted.
An upper bound (that might be non-conservative) to the
saturation pressure is given by the Two-Curve model (with
a correction factor), i.e. the curves in Fig.5.
A lower bound (that might be over conservative) can be
estimated from the experience in North America.
The Dakota Gasification (Souris Valley) pipeline was
constructed in 2000, and does not use mechanical crack
arrestors. The geometry and grade of the two sections of
The Journal of Pipeline Engineering
the pipeline are 355.6 x 9.525-mm (14.0 x 0.375in), Grade
X-70 and 323.96 x 9.525-mm (12.75-in), Grade X-70 [47].
The pipeline transports a mixture containing approximately
96 mol.% CO2, 0.7 mol.% methane, 2.3 mol.% ethane
and above, 0.9 mol.% hydrogen sulphide, and less than
0.1 mol.% nitrogen or oxygen, and no hydrogen [48,
49]. A conservative estimate of the maximum saturation
pressure of this mixture is approximately 72.8barg (the
critical pressure of CO2).
The toughness required to arrest a running-ductile
fracture in a 355.6 x 9.525-mm, Grade X-70 pipeline
transporting a mixture with a saturation pressure of
72.8barg is approximately 47J, see Fig.513. The hoop stress
at a pressure of 72.8barg is approximately 28% SMYS. The
limiting value of the saturation pressure for this grade and
geometry is 88.8barg (34.3% SMYS)14. The arrest toughness
is low relative to the toughness of modern, high-toughness
linepipe steels. The pipeline is not operating close to the
asymptote.
A reasonable lower bound to the saturation pressure is
then a hoop stress equal to 28% SMYS.
Therefore, it follows that a pipeline constructed from
modern, high-toughness linepipe steels (i.e. a minimum full-
size Charpy V-notch impact energy of 250J) can transport
mixtures with a saturation pressure that is somewhere
between the pressure corresponding to a hoop stress equal
to 28% SMYS (the lower bound) and the arrest pressure (the
upper bound). These are indicative limits. The lower bound is
based on a limit on the hoop stress at the saturation pressure.
The upper bound is based on the Two-Curve model (and
the assumed correction factor). The lower bound depends
upon the geometry and grade. The upper bound depends
upon the geometry, grade and toughness.
Consider a pipeline designed to transport a CO2-rich
mixture. Assume 610-mm (24-in) diameter and Grade L-450.
Assume that the full-size Charpy V-notch impact energy is
250J (i.e. consistent with modern, high-toughness linepipe
steel). The wall thickness and the saturation pressure are
related. If the wall thickness is assumed, then the saturation
pressure that can be transported in the pipeline can be
calculated. If the saturation pressure is assumed, then the
required wall thickness can be calculated.
Figure 6 is a plot of the wall thickness versus the saturation
pressure. The lower and upper bounds to the saturation
pressure are plotted. The lower bound is a hoop stress equal
to 28% SMYS, and the upper bound is based on the Two-
Curve model. The required wall thickness increases as the
saturation pressure increases.
13. In a 610 x 15.9-mm, Grade L-450 pipeline it is approximately 100J (and higher
if a correction factor is applied).
14. In PD 8010-1, the design factor is defined with respect to the specified minimum
wall thickness. The nominal wall thickness is quoted here. The wall thickness tolerances
for welded pipe given in BS EN 10208-2 have been applied.
3rd Quarter, 2012 225

22 25
610 mm, L450 610 mm, L450
21 24
20 23
f = 0.72
wall thickness, mm

wall thickness, mm
19 σθ = 0.28xSMYS
22
18 CVN = 250 J
21 ps
17 = 85 barg
20
16 = 80 barg
19
15 CVN = ∞ J
18 = 75 barg
14
13 17 = 70 barg

12 16
50 60 70 80 90 100 150 200 250
saturation pressure, barg design pressure, barg

Fig.6.The relationship between the wall thickness and the Fig.7.The relationship between the design pressure, the
saturation pressure, as implied by the Two-Curve model (with saturation pressure and the required wall thickness
and without corrections) and a limit on the hoop stress at (indicative).
the saturation pressure.

n
A 610 x 15.9-mm, Grade L-450 pipeline can transport mixtures fractures. In this case, the wall thickness of the pipeline is

io
with a saturation pressure somewhere between 65.7barg (the determined by the saturation pressure and not the design
st y
lower bound) and 78.1-80.6barg (the upper bound).
ut pressure. The required wall thickness is independent of
di op

the design pressure (up to design pressures equal to 195.4

rib
A mixture with a saturation pressure equal to 80barg can be barg). Increasing the saturation pressure would increase the
or c

transported in a pipeline with a wall thickness somewhere required wall thickness. Reducing the saturation pressure
between 19.4mm (the lower bound) and 15.8-16.3mm (the would decrease the required wall thickness.
t f ple

upper bound).
A pipeline has sufficient toughness to arrest a running-ductile
Assume that the design factor of the pipeline is 0.72 (i.e. the fracture if the arrest pressure is greater than the saturation
no m

hoop stress at the design pressure is 72% SMYS). The wall pressure. The geometry, grade, and toughness of the pipe
thickness and the design pressure are related. and the saturation pressure are related. The requirement for
Sa

a pipeline to have sufficient toughness to arrest a running-

Figure 7 is a plot of the wall thickness versus the design
pressure and the saturation pressure. The lower bound to
the saturation pressure is used. The required wall thickness
increases as the design pressure increases and as the saturation
pressure increases.
ductile fracture limits the saturation pressure of a mixture
that can be transported in the pipeline.
The saturation pressure of a mixture that can be transported
increases as the toughness increases.

Assume that the design pressure is 150barg. The wall thickness
required for a design pressure of 150barg is 14.9mm (the wall
thickness has not, here, been rounded up to a ‘standard’ pipe size).
A mixture with a saturation pressure equal to 80barg requires
a wall thickness of approximately 19.4mm.
Therefore, if the pipeline is designed to transport CO2-rich
mixtures with a saturation pressure equal to 80barg, the wall
thickness must be at least 19.4mm. This is greater than the
wall thickness required for a design pressure of 150barg (in
fact, the design pressure could be increased to 195.4barg
without the design factor exceeding 0.72).
The wall thickness of the pipe must be sufficient to meet the
limit on the design factor, and to prevent running ductile
The saturation pressure of a mixture that can be transported
increases as the wall thickness increases.
The saturation pressure of a mixture that can be transported
depends on the geometry, grade and toughness of the pipe.
Modern, high-toughness linepipe steel has a high toughness
(say, at least 250J). Therefore, in practical terms, the limit is
the wall thickness of the pipe, not the toughness. Increasing
the wall thickness of the pipe increases the saturation pressure
of mixtures that can be transported in the pipeline.
The saturation pressure of a CO2-rich mixture that can be
transported in a pipeline depends on the wall thickness of
the pipeline. A relationship can therefore be defined between
the saturation pressure and the wall thickness, given the
226
diameter, grade, and toughness. Figure 6 presents such a
relationship. The required nominal wall thickness increases as
the saturation pressure increases. The upper bound is based
on the Two-Curve model and assumes that the model is
applicable to pipelines transporting CO2 or CO2-rich mixtures
in the liquid or dense-phase. The lower bound is based on a
limit on the hoop stress at the saturation pressure.
A full-scale fracture-propagation test is required to validate
the application of the Two-Curve model to CO2 and CO2-
rich mixtures. In the absence of relevant full-scale fracture-
propagation test data, the lower bound (requiring a higher
wall thickness for a given saturation pressure) is the more
appropriate relationship at this time.
The nominal wall thickness depends on the diameter and
grade of the pipeline, the design pressure, and the design
factor. The required nominal wall thickness increases as the
design pressure increases.
The saturation pressure depends upon the minimum
operating pressure within the operating envelope of the
n
pipeline. It is effective, independent of the design pressure.
io
Therefore, a relationship can therefore be defined between
st y
ut
the nominal wall thickness and the design pressure and the
di op
saturation pressure, given the diameter, grade, and design
rib
factor. Figure 7 presents such a relationship.
or c
If the saturation pressure is high, or the design pressure is low,
t f ple
then the required wall thickness of the pipe is determined by
the saturation pressure; the wall thickness is independent of
the design pressure. If the saturation pressure is low, or the
no m
design pressure is high, then the required wall thickness of
the pipe is determined by the design pressure.
Sa
The relationship between the required wall thickness, the
design pressure, and the saturation pressure in Fig.7 is only
indicative. However, whilst the details might change as a data
from full-scale tests become available, the principles outlined
here, and summarized in Fig.7, apply irrespective of the
underlying fracture model. The wall thickness of a pipeline
designed to transport CO2 or CO2-rich mixtures in the liquid
or dense-phase might be determined by the saturation pressure
(in order to prevent running-ductile fractures), rather than
the design pressure.
• The required wall thickness of the pipe depends upon
the saturation pressure of the mixture.
• The required wall thickness of the pipe depends upon
the composition of the CO2-rich mixture (and the
operating conditions of the pipeline).
The implications for the cost of
carbon capture and storage
A pipeline must be designed to prevent long running fractures.
The toughness required to arrest a running-ductile fracture in
The Journal of Pipeline Engineering
a pipeline transporting a CO2-rich mixture in the dense-phase
depends upon the geometry and grade of the pipeline, and
the saturation pressure of the CO2-rich mixture. Modern,
high-toughness linepipe steel is very tough. Therefore, the
compositions that can be transported in a pipeline are, in
practical terms, limited by the wall thickness of the linepipe.
The geometry, grade and toughness of a pipeline define a
limit to the saturation pressure. Mixtures with a saturation
pressure higher than this limit cannot be transported in
the pipeline. The saturation pressure can be increased by
increasing the thickness of the pipeline. This increases the
cost of the pipeline.
A capture plant can be designed to produce a CO2-rich
stream that contains a relatively high proportion of other
components (‘impurities’) or a relatively low proportion.
Removing the impurities from the stream increases the cost
of the capture plant.
The higher the concentration of impurities, the higher the
saturation pressure of the CO2-rich mixture. The higher the
saturation pressure, the thicker the pipe. Therefore, a pipeline
designed to transport a relatively high proportion of impurities
will cost more than one designed to transport a relatively
low proportion. Conversely, the capture plant will cost less.
Techno-economic studies would be required to optimize the
design of the system, and the relationship between the wall
thickness and the composition of the CO2-rich mixture is
an important parameter in such studies. The relationship is
simplified by expressing it in terms of the saturation pressure
of the mixture. The following observations can be made:
• Higher design pressures might be more economic
than lower design pressures.
• Lowering the concentration of hydrogen in the
mixture might be more economic. The additional
cost of removing the hydrogen might be offset by
the reduction in the required wall thickness. Also,
hydrogen is valuable.
• Lower maximum operating temperatures might
be more economic than higher temperatures. The
additional cost of after-cooling at the pump or
compressor might be offset by the reduction in the
required wall thickness. A lower temperature also
increases hydraulic efficiency and thermal efficiency
(waste heat recovery).
Conclusions
A pipeline must be designed to prevent long-running fractures.
Pipelines designed to transport CO2 CO2-rich mixtures in
the liquid or dense-phase are susceptible to running-ductile
fractures. The decompression behaviour of liquid and dense-
phase CO2 and CO2-rich mixtures is very different to that of
lean or rich natural gas. The different behaviour has implications
for the design of pipelines to prevent running-ductile fractures.
3rd Quarter, 2012
The susceptibility of a pipeline transporting a CO2-rich
mixture in the liquid or dense-phase depends on the saturation
pressure of the mixture. The susceptibility increases as the
saturation pressure increases.
An entry-specification for a pipeline must include a limit on
the saturation pressure of the CO2-rich mixture.
The saturation pressure of a CO2-rich mixture depends upon
the composition of the mixture and the operating conditions
of the pipeline.
The geometry, grade, and toughness of a pipeline define a
limit on the saturation pressure of a CO2-rich mixture that
can be transported in the pipeline. In practical terms, the
saturation pressure is limited by the wall thickness of the
linepipe. Increasing the wall thickness of the pipe will increase
the saturation pressure of CO2-rich mixtures than can be
transported in the pipeline.
A simple relationship between the design pressure, the
saturation pressure, and the required wall thickness is
n
described. It is suitable for use in techno-economic studies
to optimize the design of the capture, transport, and storage
io
infrastructure that will be required for the successful
st y
ut
implementation of carbon capture and storage.
di op
Acknowledgements
rib
or c
The author acknowledges support and funding provided
t f ple
by National Grid through the COOLTRANS Project. The
author also acknowledges technical discussions with David
G Jones and Dr Jane Haswell (Pipeline Integrity Engineers)
no m
and Dr Julia Race (Newcastle University).
References
Sa
1. R.B.Alley, et al., 2007. Summary for policy makers,
contribution of Working Group I to the Fourth Assessment
Report of the Intergovernmental Panel on Climate Change,
IPCC WG I 4AR. Intergovernmental Panel on Climate
Change, April.
2. T.Barker, et al., 2007. Summary for policy makers, Working
Group III contribution to the Intergovernmental Panel
on Climate Change Fourth Assessment Report Climate
Change 2007: Mitigation of Climate Change, IPCC WG
III 4AR. Intergovernmental Panel on Climate Change, May.
3. N.Stern, 2007. The economics of climate change: the Stern
review. Cambridge University Press, Cambridge (also HM
Treasury, 2006 – www.hm-treasury.gov.uk).
4. S.Pacala and R. Socolow, 2004. Stabilization wedges: solving
the climate problem for the next 50 years with current
technologies. Science, 305, August, pp968-972.
5. Anon., 2005. IPCC special report on carbon dioxide capture
and storage. Eds B.Metz, O.Davidson H.de Coninck,
M.Loos, and L.Meyer. Intergovernmental Panel on Climate
Change, Cambridge University Press, Cambridge.
6. C.Philibert, 2007. Technology penetration and capital stock
227
turnover: lessons from IEA scenario analysis. International
Energy Agency, Organisation for Economic Co-operation
and Development, COM/ENV/EPOC/IEA/SLT(2007)4,
May.
7. IEA, 2009. Technology Roadmap, Carbon capture and
storage.
8. M.Mohitpour, H.Golshan, and A.Murray, 2007. Pipeline
design and construction, a practical approach. Third
Edition, ASME Press, New York.
9. M.Mohitpour, P.Seevam, K.K.Botros, B.Rothwell, and
C.Ennis, 2011. Pipeline transportation of carbon dioxide
containing impurities. ASME Press, New York.
10. Interstate Oil & Gas Compact Commission CCGS Task
Force, 2010. A policy, legal, and regulatory evaluation
of the feasibility of a national pipeline infrastructure for
the transport and storage of carbon dioxide. Pipeline
Transportation Task Force (PTTF) of the Interstate Oil
and Gas Compact Commission-Southern States Energy
Board (IOGCC-SSEB), Oklahoma City, Oklahoma,
September 10.
11. D.L.Marsili and G.R.Stevick, 1990. Reducing the risk of
ductile fracture on the Canyon Reef Carriers CO2pipeline.
SPE20646, 65th Annual Technical Conference and
Exhibition of the Society of Petroleum Engineers, New
Orlean.
12. A.Oosterkamp and J.Ramsen, 2008. State-of-the-art
overview of CO2 pipeline transport with relevance to
offshore pipelines. Report No. POL-O-2007-138-A, Polytec,
Norway.
13. P.N.Seevam, J.M.Race, M.J.Downie, and P.Hopkins, 2008.
Transporting the next generation of CO2 for carbon capture
and storage: the impact of impurities on supercritical
CO2 pipelines. Paper No. IPC2008-64063, Proc. 7th
International Pipeline Conference, IPC 2008, Calgary,
Alberta, Canada.
14. P.N.Seevam, J.M.Race, M.J.Downie, J.Barnett, and
R.Cooper, 2010. Capturing carbon dioxide: the feasibility
of re-using existing pipeline infrastructure to transport
anthropogenic CO2. Paper No. IPC2010-31564, Proc. 8th
International Pipeline Conference, IPC 2010, Calgary,
Alberta, Canada.
15. BSI, 2009. Petroleum and natural gas industries - Materials
for use in H2S-containing environments in oil and gas
production Part 2: Cracking-resistant carbon and low-alloy
steels, and the use of cast irons (ISO 15156-2 : 2009), BS EN
ISO 15156-2: 2009. British Standards Institute, London.
16. J.M.Race, B.Wetenhall, P.N.Seevam, and M.J.Downie,
2012. Towards a CO2 pipeline specification: defining
tolerance limits for impurities. Third International Forum
on the Transportation of CO2 by Pipeline, Newcastle upon
Tyne, UK.
17. E.de Visser, C.Hendriks, G.de Koeijer, S.Liljemark,
M.Barrio, A.Austegard, and A.Brown, 2007. DYNAMIS
CO2 quality recommendations, Project No. 019672,
Deliverable No. D 3.1.3, DYNAMIS Consortium, 21 June.
18. BSI, 2004. Code of practice for pipelines – Part 1: Steel
pipelines on land, PD 8010-1: 2004, British Standards
Institution, London.
228
19. ASME, 2009. Pipeline transportation systems for liquid
hydrocarbons and other liquids. ASME code for pressure
piping, B31, ASME B31.4, 2009 Edition, American Society
of Mechanical Engineers, New York.
20. API, 2007. Specification for line pipe, ANSI/API
Specification 5L, American Petroleum Institute, 44th
Edn, October 1.
21. Private communication with Robert J. Eiber, Robert J.
Eiber Consultant Inc.
22. G.King, O.Zarzour, T.Afachtal, S.Kumar, and
M.Balasubramanian, 2009. Design of dense-phase carbon
dioxide pipelines. 3rd International CO2 Forum, SOGAT
2009, Abu Dhabi, UAE, April 2.
23. BSI, 2003. Petroleum and natural gas industries – Pipeline
transportation systems, BS EN 14161: 2003, British
Standards Institution, London.
24. Det Norske Veritas, 2007. Submarine Pipeline Systems,
Offshore Standard DNV-OS-F101, October.
25. BSI, 2009. Steel pipes for pipelines for combustible
fluids – Technical delivery conditions Part 2: Pipes of
requirement class B, BS EN 10208-2: 2009, British
Standards Institution, London.
n
26. BSI, 1999. Petroleum and natural gas industries – Steel
pipe for pipelines – Technical delivery conditions – Part
io
3: Pipes of requirement class C, BS EN ISO 3183: 1999,
st y
ut
British Standards Institution.
di op
27. G.Re, V.Pistone, G.Vogt, G.Demofonti, and D.G.Jones,
rib
1995. EPRG Recommendation for crack arrest toughness
or c
for high strength line pipe steels. 3R International, 10, 11,
pp607-611.
t f ple
28. Det Norske Veritas, 2010. Design and operation of CO2
pipelines, Recommended Practice DNV-RP-J202, April.
29. R.Cooper, 2012. National Grid COOLTRANS
no m
research programme. Third International Forum on
the Transportation of CO2 by Pipeline, Newcastle upon
Sa
Tyne, UK.
30. W.A.Maxey, 1983. Gas expansion studies, final report to
the American Gas Association. NG-18 Report No. 133,
Battelle Memorial Institute.
31. Idem, 1986. Long shear fractures in CO2 lines controlled
by regulating saturation, arrest pressures. Oil and Gas
Journal, pp44-46.
32. A.Cosham and R.J.Eiber, 2008. Fracture control in
carbon dioxide pipelines – the effect of impurities. Paper
No.: IPC2008-64346, Proc. 7th International Pipeline
Conference, IPC 2008, Calgary, Alberta, Canada.
33. A.Cosham, 2009. CO2: it’s a gas Jim, but not as we know
it. International Conference on Pipeline Technology 2009,
Ed. R.Denys, Ostend, Belgium.
34. A.Cosham, R.J.Eiber, and E.B.Clark, 2010. GASDECOM:
carbon dioxide and other components. Proc. 8th
International Pipeline Conference, IPC 2010, Paper No.:
IPC2010-31572, Calgary, Alberta, Canada.
35. J.Barnett and D.G.Jones, 2011. The Longannet CCS
project shock tube tests. Second International Forum
on the Transportation of CO2 by Pipeline, Newcastle
upon Tyne, UK.
The Journal of Pipeline Engineering
36. A.Cosham, D.G.Jones, K.Armstrong, D.Allason, and
J.Barnett, 2011. The decompression behaviour of carbon
dioxide in the gaseous phase. Second International
Forum on the Transportation of CO2 by Pipeline,
Newcastle upon Tyne, UK.
37. A.Cosham, D.G.Jones, K.Armstrong, D.Allason, and
J.Barnett, 2012. The decompression behaviour of carbon
dioxide in the dense-phase. Paper No.: IPC2012-90461,
Proc. 9th International Pipeline Conference, IPC 2012,
Calgary, Alberta, Canada.
38. R.J.Eiber, T.A.Bubenik, and W.A.Maxey, 1993. Fracture
control technology for natural gas pipelines. NG-18
Report No. 208, Final Report to Line Pipe Research
Supervisory Committee of the Pipeline Research
Committee of the American Gas Association, Project
PR-3-9113, Catalog No. L51691.
39. W.A.Maxey, 1974. Fracture initiation, propagation
and arrest. Paper J, Proc. 5th Symposium on Line Pipe
Research, Pipeline Research Committee of the American
Gas Association, Houston, Texas.
40. R.J.Eiber and T.A.Bubenik, 1993. Fracture control
plan methodology. Paper 8, Eighth Symposium on
Line Pipe Research, Pipeline Research Committee of
the American Gas Association, Catalogue No. L51680,
Houston, Texas.
41. IEAGHG, 2011. Effects of impurities on geological
storage of CO2. Report: 2011/04, June.
42. B.N.Leis, 2000. Predicting fracture arrest based on
a relationship between Charpy Vee-notch toughness
and dynamic crack-propagation resistance. Vol. 1,
Pipeline Technology, Proc. 3rd International Pipeline
Technology Conference, Ed. R. Denys, Brugge, Belgium,
May, pp407-420.
43. R.J.Eiber, 2008. Fracture propagation – 1: Fracture-
arrest prediction requires correction factors. Oil & Gas
Journal, 106, 39, October 20.
44. Idem, 2008. Fracture propagation – Conclusion:
Prediction steel grade dependent. Ibid., 106, 40,
October 27.
45. G.M.Wilkowski, W.A.Maxey, and R.J.Eiber, 1977. Use
of a brittle notch DWTT specimen to predict fracture
characteristics of line pipe steels. Paper No. 77-Pet-21,
Energy Technology Conference, Houston, Texas.
46. G.Wilkowski, D.Rudland, H.Xu, and N.Sanderson,
2006. Effect of grade on ductile fracture arrest criteria
for gas pipelines. IPC2006-10350, Proc. International
Pipeline Conference, American Society of Mechanical
Engineers, Calgary, Alberta, Canada.
47. www.dakotagas.com/Gas_Pipeline/CO2_Pipeline/
index.html
48. www.dakotagas.com/About_Us/index.html
49. J.M.Race, P.N.Seevam, and M.J.Downie, 2007.
Challenges for offshore transport of anthropogenic
carbon dioxide, Paper No.: OMAE2007-29720,
Proc.26th International Conference on Offshore
Mechanics and Arctic Engineering (OMAE), San
Diego, California.
3rd Quarter, 2012 229

Transport of gaseous and dense

carbon dioxide in pipelines: is
there an internal corrosion risk?
by Daniel Sandana*1, Matthew Hadden1, Dr Julia Race2, and EA Charles2
1 Macaw Engineering, Newcastle upon Tyne, UK
2 Newcastle University, Newcastle upon Tyne, UK

T ransporting anthropogenic CO2 in pipelines is an essential component in the realization and implementation
of carbon capture and storage (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 may need to be made to mitigate or prevent internal corrosion. This will
require identifying and defining in the CO2 specification the maximum levels of water and impurities, such
as nitrogen oxides (NOx) and sulphur oxides (SOx), so that internal corrosion risks are maintained at an

n
acceptable level throughout the proposed service life of a pipeline. Equally, should there be a process upset
in the CO2 stream conditioning procedure (for example, a failure of a dehydration unit), then potential

io
internal-corrosion risks will need to be clearly defined in order to establish an effective mitigation strategy
st y
that maintains pipeline integrity.
ut
di op
rib
This paper reviews available corrosion data in the public domain for carbon steel exposed to gaseous and
or c

dense CO2 in the presence of impurities.The paper will describe different corrosion scenarios in pipelines
transporting anthropogenic CO2 and interpret associated corrosion risks from published data. Gaps in
t f ple

knowledge and the requirement for further work will be highlighted.

no m

T he release of greenhouse gases, including • Capturing CO2 from the gas streams emitted from
Sa

carbon dioxide, methane, and nitrous oxides, into the
atmosphere due to human activities have been associated
with global warming and climate change [1]. Carbon dioxide
(CO2) represents a significant component (approximately
77%) of the total greenhouse emissions. In 2004, the IEA
World Energy Outlook (WEO) estimated that in 2030 CO2
emissions will be approximately 90% higher than 1990
figures. Even using mitigation policies envisaged in 2004,
global CO2 emissions in 2030 were predicted to be 62%
higher than in 1990 [2].
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 is referred to as carbon capture and storage
(CCS), and involves three distinct stages:
This paper was presented at the Third International Forum on the Transportation of
CO2 by Pipeline, organized by Newcastle University, Tiratsoo Technical, and Clarion
Technical Conferences, and held in Newcastle in June, 2012.
*Corresponding author’s contact details:
tel: +44 (0)191 215 4010
email: daniel.sandana@macawengineering.com
power plants, industrial sources or natural gas wells.
• The CO2 is generally compressed into a supercritical
(dense) or liquid phase to allow easy and less costly
transport by pipelines to storage. However, CO2 may
also be transported in its gaseous state.
• Storing CO2 underground in deep saline aquifers,
depleted oil and gas reservoirs, or to re-inject it into
producing wells for enhanced oil recovery (EOR).
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. These
need to be assessed 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.
230
Fig.1. Solubility of water in pure CO2 as a function of pressure and temperature [7].
n
The importance of water content
specification and water solubility in
io
CO2
st y
ut
di op
Oil and gas industry: CO2 corrosion
and the effect of water
rib
or c
In the oil and gas industry, CO2 is present as an impurity in
t f ple
low hydrocarbons (as gaseous components such as methane),
and in more heavy hydrocarbons (as liquid components such
as hexane — C6H14), and in water from the underground
no m
production reservoirs. Considering the pressure and
temperature and the mixtures involved, CO2 naturally behaves
Sa
as a gas which can partially dissolve within the liquids and
aqueous media present, depending on the solubility limits.
CO2 can be a significant threat for carbon steel materials used
in pipeline design since it can, under particular circumstances,
cause corrosion – usually referred to as ‘sweet’ corrosion, and
this subject has been widely discussed in the literature [3-5].
It is understood in the oil and gas industry that sweet gas
corrosion will only be a risk if free liquid water is present.
Indeed if the water remains dissolved in the hydrocarbon
phase, whether it is in a gas or a liquid phase, the formation of
carbonic acid (due to the dissolution of CO2 in liquid water)
responsible for the sweet corrosion process is not possible. The
water-solubility limit is defined as the water content at which
water saturation is reached, and is dependent on pressure,
temperature, and fluid chemical composition.
Above the water-solubility limit, free liquid water will separate
out and form an aqueous phase where sweet corrosion can
take place.
Hence to prevent CO2 corrosion, oil and gas operators
undertake dehydration of the hydrocarbon fluids so that water
The Journal of Pipeline Engineering
condensation does not occur during pipeline transportation
resulting from changing pressure and temperature profiles
along the pipeline. For example in the natural gas industry,
pipeline internal integrity management is achieved by ensuring
the gas is transported above its water dewpoint temperature,
i.e. the temperature below which free water will condense.
CO2 pipelines: water solubility in pure CO2 streams
If the principles of dewpoint control are applied to CO2
transport, then CO2 corrosion should not be a risk, as long
as the water remains dissolved in the CO2 stream and no
aqueous phase precipitates. Experience of CO2 pipeline
operations in the US would support this reasoning. The water
level specification for the existing US pipelines is generally
strict, being below 50 parts per million weight (ppm wt), the
threshold which is usually considered as full dehydration
[6, 7]. Since the majority of CO2 pipelines are operated dry,
the number of reported failures of these pipelines has been
relatively low; between 1986 and 2006 the incident rate in
the US was 0.0003 per km year for CO2 pipelines. Crucially,
corrosion was not the major factor in those leaks that did
occur. The internal corrosion rate in a dry supercritical
CO2 pipeline was reported between 0.25-2.5 µm/year over
12 years of operation.
Figure 1 [7, 9-12] illustrates water solubility in pure CO2 as a
function of pressure and temperature; it can be seen that:
• A rise in temperature increases the solubility of water
(H2O) in gaseous and dense-phase CO2.
• Where CO2 is a gas, typically at low pressures, the
solubility of water decreases with increasing pressure.
However, during a phase transition of CO2 from a
gas to dense phase, the solubility of H2O in CO2
increases with pressure.
3rd Quarter, 2012
n
io
st y
ut
di op
rib
or c
Fig.2.Variation of sulphuric acid dewpoint for gases having different water vapour contents [16].
t f ple
An acceptable water specification for avoiding corrosion in
pipelines transporting pure CO2, at operating temperatures
no m
above 4°C, is generally considered to be in the range 300-
500ppm wt (0.4kg/m3 or 0.0003-0.0005ppm vol) [7, 13].
Sa
CO2 pipelines: water solubility in
CO2 and the effect of impurities
Depending on the type and concentration of impurities,
water solubility in CO2 may decrease. For example,
Austegard et al. [10] showed that CH4 will lower the
solubility of water in CO2, and this is also potentially the
case for other impurities [14], such as H2S, O2, and N2.
Similarly the presence of sulphur oxides (SOx) will lead
to a significant reduction in the water solubility in CO2.
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 as the acid dewpoint [15, 16]. Typically
the presence of SO2 or SO3, even at small concentrations,
will react with water vapour (or liquid) to form sulphurous
or sulphuric acids.
Although nitrogen oxides (such as NO and NO2) can also
react with water to form acids including nitrous (HNO2)
or nitric acids (HNO3), the effect of NOx on the dewpoint
231
temperature is negligible in comparison to the effect of
SO2 or SO3 [16, 17]. Hence, in the absence of SOx, the
dewpoint can be considered as being solely determined by
the water content in the gases. The variation of the acid
dewpoint for gases containing different concentrations of
water as a function of H2SO4 and SO3 is illustrated in Fig.2
[16] and Fig.3 [17] respectively. As the amount of water is
increased, the acid dewpoint temperature increases; this
is also the case when the concentration of SO3 in the gas
is increased.
The condensation of acids can cause severe acidic corrosion
by reducing the pH level. It has been reported [15, 16] that
CO2 with NOx and SOx contaminants present does not
corrode carbon-manganese steels as long as the relative
humidity is less than 60%. It is however likely that this
will be dependent on the type of impurities present and
the relative concentrations of these impurities.
Whilst the water content specification of 300-500ppm wt
(0.0003-0.0005ppm vol) for water needs to be reviewed to
consider the presence of impurities, a full dehydration to
50ppm wt (0.00005ppm vol) – or 60% relative humidity
– is often considered too stringent [6, 7]. It is likely that
water specification will be system specific, i.e. dependant
on the type and concentration of impurities present from
the conditioning and capture processes.
232 The Journal of Pipeline Engineering

n
io
Fig.3.Variation of sulphuric acid dewpoint for gases having different water vapour contents as a function of SO3 levels [17].
st y
Note that a gas with 6.5 v% H2O has a vapour pressure of 49.7mm Hg and a dewpoint of 38°C.
ut
di op
rib
Despite all the precautions taken at the process design • If there is a process upset in the stream conditioning
or c

stage to guarantee acceptable water contents in CO2, upset procedure (such as drying or desulphurization)
conditions in the CO2 dehydration process and water and the resulting stream composition goes out of
t f ple

breakouts are still a possibility in complex pipeline systems, specification, what is the required response time and
especially over decades of operating life. Hence the resulting what is the suitable mitigation strategy to minimize
risk of internal corrosion from such excursions should not corrosion potential?
no m

be excluded. One may argue that if a continuous flow of • Are there any other options to minimize internal
dry CO2 is present following a water-incursion event, the corrosion during transport operations instead
Sa

water may rapidly dissolve in the stream minimizing any of drying (such as the use of glycol or corrosion
internal damage to the pipeline [6]. However, in the case inhibitor)?
of a delay in response to process upsets or in the case of
a shutdown (pipeline shut-in or total depressurization), Corrosion tests vs operational
situations could exist where free water may be present scenarios: is there an internal
over longer times in the pipeline. Indeed, during long corrosion risk in CO2 pipelines?
shutdowns, water-removal operations will be required and
experience from the oil and gas industry has indicated Over the past decade the need to develop CCS infrastructures
that this may take weeks. For new pipelines, corrosion worldwide has generated considerable amounts of research
testing should take place at the design stage to establish work to identify plausible corrosion rates of pipeline
corrosion allowance requirements as well as a suitable carbon steel materials (typically API 5L X60-X65) in
mitigation strategy which will be specific to each system CO2 transportation system environments. The main test
[6, 19]. Similarly, corrosion testing should be undertaken parameters in these research programmes have been:
for operational pipelines to develop a suitable corrosion-
management strategy. • materials
• water content
Prior to commissioning a CO2 pipeline the following • CO2 pressure
issues need to be resolved: • type and partial pressure of impurities, mainly O2,
• Does the agreed CO2 specification mitigate internal SOx, and NOx
corrosion, i.e. what is the likely maximum corrosion • temperature
rate in the pipeline if the operating procedure • stagnant or dynamic flow
is correctly applied and the stream meets the • duration of exposure
specification?
3rd Quarter, 2012
The results of these corrosion laboratory tests need to be
translated into different operational production scenarios. For
the purposes of this paper, the following four scenarios during
transportation of anthropogenic CO2 have been investigated:
• A. The CO2 is dehydrated to specification and the
impurity levels are identified as negligible, (according
to CO2-plant conditioning specification).
• B. The CO2 is dehydrated to specification but upsets
in CO2 stream conditioning have occurred. Impurity
levels above agreed specification have entered the
pipeline transportation system.
• C. Upset in dehydration occurs such that a free
liquid water phase is present at the bottom (6-o’clock
position) of the pipeline transportation system. Levels
of impurities are considered negligible.
• D. Upset in dehydration occurs such that a free
liquid water phase is present at the 6-o’clock position
of the pipeline transportation system. Upset in the
CO2conditioning plant has also occurred with levels
of impurities above the specified levels.
n
Scenario A: water dissolved in pure CO2
io
The CO2 stream is considered free of impurities and has
st y
ut
been dried to low water levels below an agreed specification,
di op
for example less than 500ppm wt. Under such circumstances
rib
it is considered that water will remain dissolved in the CO2
or c
and hence gaseous or dense phase CO2 is considered as dry.
t f ple
It is suggested by research work [20-24], practical experience
from the natural gas industry, and the US’ CO2 pipeline
experience, that negligible corrosion rates are expected in
no m
this scenario. However it has been reported [24] that there
is a region below the water-solubility point where corrosion
Sa
may still occur in dense dry CO2. This may be associated with
the formation of carbonic acid due to CO2 hydration with
dissolved water. Nonetheless it is considered that – even in
such instances – the corrosion rate will be insignificant, due
to rapid saturation of precipitated water with iron products
and the formation of a protective film. The situation remains
the same under flowing conditions.
Scenario B: low water contents
in CO2 in presence of impurities
Although the water content in the CO2 stream is kept below
specification and at low levels, impurities are present in the
fluid which can lower the water solubility in the CO2 stream.
Water condensation and the occurrence and magnitude of any
corrosion will greatly depend on the type(s) and concentration
of impurities contained in the CO2 stream.
Under dissolved water conditions it is suggested [20, 21,
25, 26] that O2 and H2S do not significantly contribute to
carbon steel corrosion. However, the presence of SOx and
NOx can cause corrosion to occur even at very low water
contents – notably SO3 [20, 21, 26]. This has been associated
233
with the condensation of acids at temperatures much higher
than the water solubility point or water dewpoint, and is
usually referred as the ‘acid dewpoint’.
Dugstad [20, 21] reported that in dense-phase CO2, the
supplementary presence of SO2 (1000ppm wt) led to
corrosion (0.01 mm/year) at low water contents of 200ppm
wt. Farelas [27] showed that in dense CO2 conditions (80 bar,
50°C) and at water contents of 650ppm (mole), (typical of
the water specification in US CO2 pipelines used for EOR),
corrosion rates could be as high as 3-4mm/year when 1%
SO2 is present at pressure of 0.8bar. These rates decrease
to below 0.03mm/year at SO2 levels of less than 0.1% and
pressure less than 0.08 bar.
Over a wide spectrum of CO2 pressures and temperatures,
three fundamental questions remain to be resolved:
• What is the maximum level of water content in the
presence of SOx that will not give rise to significant
corrosion in dry CO2?
• What is the maximum level of water content in the
presence of NOx that will not give rise to significant
corrosion in dry CO2?
• What is the maximum level of water content in the
presence of SOx and NOx that will not give rise to
significant corrosion in dry CO2?
The resolution of these issues will necessitate the development
of a test matrix looking at the correlation between various
amounts of water, SOx, and NOx, over a range of CO2
pressures and temperatures. Such a programme of work will be
extensive, and initially it will be sensible to test for the SOx and
NOx levels that can typically be achieved from available CO2
conditioning technologies, (such as desulphurization plant).
A test programme should focus on impurity concentrations
at different levels of water and various CO2 pressures over
a range of typical pipeline operating conditions. Upsets in
conditioning will also need to be considered, and under this
scenario the concentrations of SOx and NOx to be tested will
be driven by maximum levels expected in flue-gas composition
from the combustion process.
Scenario C: aqueous water phase
in equilibrium with CO2, no impurities
In high-pressure CO2 environments it is clear that sweet
corrosion will be a risk for carbon steel exposed to free
liquid water. Data are available to identify the magnitude
of such corrosion rates as a function of CO2 pressure
and temperature. Although published sources agree that
significant and excessive corrosion could be expected in
such environments, it is clear there is a considerable scatter
of in the corrosion rates reported at various CO2 partial
pressures, as illustrated in Fig.4 [28].
Furthermore the trend of corrosion rates with, for example,
increasing pressure has not been consistent. Choi [11, 29]
234 The Journal of Pipeline Engineering

n
io
st y
ut
di op
rib
Fig.4.Variation of corrosion rate in an aqueous solution as a function of partial pressure of CO2 [28].
or c
t f ple
no m
Sa

Fig.5. Corrosion rates in CO2-rich and water-rich phases at 50°C as a function of pressure over 24-hr exposure: A (left) = 1%wt
Cl-; B (right) = free chlorides [11].

(Fig.5) suggests that increasing CO2 pressures from 40 to
80bar, which includes a phase transformation from gaseous
to dense-phase CO2 at 70bar, did not significantly increase
the corrosion rate at 50˚C over a 24-hr exposure. Conversely,
Seiersten [30] (Fig.6) clearly shows that corrosion rates
decrease with increasing CO2 pressures from 10 to 56bar
at 40˚C over a period of 1-2 weeks’ exposure.
The inconsistency of the reported corrosion rate data can
be associated with various test parameters:
• test temperatures
• test exposure time
• material surface to water volume ratio
• stagnant and flowing conditions
3rd Quarter, 2012
n
Fig.6. Corrosion rate of X-65 carbon steel and 0.5 Cr steel in CO2-saturated water phase vs pressure at 40°C [30].
io
st y
ut
All these parameters have an essential commonality insofar
di op
as they influence the presence or absence of protective iron
rib
carbonate film.
or c
An increase of temperature should increase the rate of corrosion.
t f ple
However, the precipitation rate of iron carbonate is higher at
increased temperatures and the formation of a protective film
over shorter exposure times is more likely – which will mitigate
no m
corrosion. Particularly under stagnant flow conditions that
could apply during shutdown of a pipeline, progressive iron
Sa
saturation of a water pool will lead to potential precipitation
of surface films over time. This will be dependent on the
water condensation rate which is associated with the water
saturation levels in CO2, as well as on the pipeline operating
temperature gradient. At temperatures below 50°C (which will
typically be the case in the pipeline transportation system) this
will require longer exposure times. At all temperatures and
under stagnant conditions the instantaneous corrosion rate is
expected to decrease due to the drop in water corrosivity arising
from corrosion product saturation. Consequently, average
corrosion rates should be lower with longer exposure times
under stagnant conditions. The internal exposure duration
of pipeline materials to a water phase, following dehydration
upsets, will be dependent on the actions undertaken to dry
the pipeline – for example by pigging.
A larger material surface to water volume ratio will generate
a greater amount of dissolved iron over shorter times, and
hence iron carbonate saturation will be reached quite rapidly.
In a pipeline, this will relate to the volume of water which has
condensed, which depends on the response time in correcting
the dehydration upset and pipeline temperature profile as
well as the exposed pipeline surface area. An additional factor
235
is that the pipeline vertical profile may contribute to water
accumulation at low points in the pipeline.
Flowing conditions will hinder iron saturation of the surface,
reducing the tendency of iron-carbonate precipitation. Should
a protective film be established, high flow velocities could lead
to film destabilisation or removal leading to severe localized
corrosion, such as MESA.
It has been suggested by Seiersten [30] that a drop in corrosion
rates can be expected with higher CO2 pressures (Fig.6).
Although one would think that this may be related to more
rapid iron carbonate super-saturation, due to higher levels
of dissolved iron in solution, it was stated that the lowest
corrosion rates at the highest pressures cannot be caused by iron-
carbonate formation. The cause may be an as yet unidentified
corrosion mechanism [30], but further work is required to test
this hypothesis. Other authors have, however, confirmed the
presence of iron carbonate at high CO2 pressures, depending
on temperature and exposure times.
It is difficult to draw clear conclusions from published data on
what corrosion rates will be over a typical operating temperature
range of 10-50°C, if water is present in a pipeline at high CO2
pressures (greater than 10bar). However, these rates are expected
to be very high for carbon steels but mitigation, in the form
of – for example – corrosion inhibitor injection [23, 31] could
prove to be beneficial when water is present in the pipeline
[31]. Corrosion mitigation by glycol at high CO2 pressures
has not proved to be entirely conclusive and further work
is required. At temperatures below 20°C, corrosion appears
to be manageable at high CO2 pressures in the presence of
50%wt MEG [31, 32].
236
The use of high-grade stainless steels such as Duplex and 904L
grades may be acceptable for service, even in the presence
of free liquid water at temperatures below 30°C. However,
the costs associated with these materials for long-distance
pipelines are likely to make this an uneconomic solution.
Scenario D: aqueous water phase in
equilibrium with CO2, presence of impurities
The impact of impurities on corrosion in pipelines in the
presence of an aqueous phase will not only be dependent
on the type of impurity and its concentration in the CO2
stream, but also on the partitioning behaviour of these
impurities between the different CO2 and water phases.
This partitioning will be altered according to whether CO2
is in a gaseous, liquid, or dense phase, and will depend on
any interactions between the different mixture compounds.
The CO2-phase behaviour will be greatly dependent on the
pipeline operating pressure and temperature as well as the
type and level of impurities in the CO2 stream [33, 34].
Dugstad [20, 21] suggests that during a drop in pressure the
amount of sulphuric acid that can collect in an aqueous phase
n
might be significant at 20°C. Details of such partitioning
data over a wide spectrum of pressures and temperatures
io
are currently very sparse in the public domain.
st y
ut
di op
Essentially the presence of O2, SOx, and NOx will aggravate
rib
the corrosion rates in CO2 environments should a water
or c
phase exist at the bottom of a pipeline. The effect of impurities
on the acceleration of corrosion is twofold:
t f ple
• Increase of the cathodic reduction:
• O2 directly attacks the steel as part of the O2 reduction
no m
in water or hydroxides (depending on fluid pH).
• In the presence of water, SOx and NOx produce acids
Sa
which provide an additional source for hydroniums
which will directly result in acid corrosion. Free liquid
water is not a requirement for corrosion to occur in
the presence of SOx or NOx since the condensation
of their acids will occur at temperatures higher than
the water dew point or solubility in pure CO2-H2O
system.
• Destabilization or increased dissolution of iron-
carbonate films leading to severe localized attack.
• O2 destabilizes the formation of protective iron
carbonate film due to the direct reduction of ferrous
cations with O2 to form iron oxides which are not
protective. The consumption of iron by O2 also
diminishes the iron-saturation level, which means
that iron carbonate is locally prone to breakdown,
leading to severe localized corrosion.
• SOx and NOx will lead to the establishment of very
low pH values (less than 3) in aqueous phase due
to the formation of acids. At such low pH values
the solubility of iron-carbonate films is increased
and their precipitation rate is decreased, rendering
the formation of protective iron-carbonate films
very unlikely.
The Journal of Pipeline Engineering
• SOx (sulphuric acid) can result in the formation of
iron sulphates but these films are not protective. In
the presence of O2 it was also discussed that iron
sulphate can be oxidized into hydro-oxide FeOOH
by liberating more sulphuric acid in the same time,
both of which will probably decrease film stability.
Reports of tests for corrosion in CO2 environments due to
the presence of O2, SOx, and NOx have to date been limited
and a wider spectrum of CO2 pressures and temperatures
need investigation. It will be beneficial to investigate CO2
impurity levels that result from available CO2 conditioning
technologies, such as desulphurization plant. Worst-case
scenarios will need to consider the maximum-allowable
impurity level in flue gases from the combustion process that
will not give rise to unacceptable corrosion levels.
Stress-corrosion cracking (SCC) of carbon steel or low-alloyed
steel materials is possible in the presence of H2S and CO.
• 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 industries. In particular, the standard NACE
MR0175/ISO 15156 [35] has been used to mitigate the
risk of sour cracking. This standard is based on oil and
gas industry experience and testing for hydrocarbon
systems (in which CO2 is present as an impurity).
Although it is a starting point to potentially reduce
the susceptibility of sour cracking in CO2 pipelines,
there is a fundamental requirement to obtain data at
the high partial pressures of CO2 in the presence of
H2S to understand in which conditions of H2S partial
pressure, pH, and temperature, SSCC or HIC can be
realized in CO2 pipelines.
• In the 1970s and the transport of ‘town gas’ in the
UK, CO was reported to cause – in the combined
presence of CO2 and water – the initiation and
propagation of transgranular stress-corrosion cracks
in carbon steel and low alloyed steel materials [36,
37]. Most of the experimental data are limited to
low partial pressures of CO2 (less than 20bar), and
there is a fundamental requirement to investigate the
susceptibility of pipeline steels to CO2-CO-H2O SCC
at high CO2 partial pressures typical of gaseous and
dense CO2 pipelines.
Conclusions
Negligible corrosion rates are expected in CO2 free of
impurities, as long as water remains dissolved in the CO2 fluid.
• In pure CO2, a water level of 300-500ppm wt (0.0003-
0.0005ppm vol) may be acceptable to prevent water
condensation, depending on pipeline operating
temperature.
3rd Quarter, 2012
In the presence of impurities, water solubility in CO2 can
be lowered and corrosion may occur, even at low water
contents (less than 500ppm wt).
• Water condensation, and the occurrence of corrosion
and its magnitude, will greatly depend on the types
and concentration of the impurities.
• The presence of SOx – notably SO3 – will lead to
significant corrosion occurring even at very low water
content (such as 200ppm wt) due to condensation
sulphuric acid at temperatures above the water
solubility point in pure CO2-H2O system.
In high-pressure CO2 environments, carbon steel materials
exposed to free liquid water will suffer high sweet corrosion
rates. However, there is a discrepancy between the different
corrosion rates reported and, equally significantly, the trend
of corrosion rates with CO2 pressure has not been consistent.
Further experimental work is needed to establish confidence
in the magnitude of likely corrosion rates generated by high
CO2 pressures in the presence of liquid water. Such test work
will need to clearly document procedures and parameters
n
and demonstrate repeatability.
io
• The inconsistency of the experimental data may be
st y
ut
associated with variations in test temperatures, test
di op
exposure time, material surface to water volume
rib
ratio, and stagnant or flowing conditions.
or c
• Iron-carbonate film may form over time, which will
mitigate corrosion over long exposures. However the
t f ple
likelihood of film formation at low temperatures and
under flowing conditions is decreased.
no m
Pipeline corrosion in the presence of an aqueous phase
will be dependent on the types of impurity and their
Sa
concentration, and on the partitioning behaviour of these
impurities between the different CO2 and water phases.
The presence of O2, SOx, and/or NOx will aggravate the
corrosion rates due to increased cathodic reduction and
increased iron carbonate film dissolution or destabilization.
Recommendations
A wider spectrum of CO2 pressures, temperatures and
impurity levels needs investigating to identify the maximum
water level below which no significant corrosion occurs in
the presence of SOx, NOx, and combinations of both.
Establish partitioning characteristics of impurities, typically
SOx, NOx, and SO2, between the different CO2 and water
phases at different CO2 pressures and temperatures.
Experimental investigation should be undertaken to confirm
the magnitude and trend of corrosion rates for carbon steel
exposed in the free liquid water phase in CO2 environments
over a wide spectrum of CO2 pressures (10-80bar) and
temperatures (10-50°C).
237
Further experimental work needs to be undertaken to
establish corrosion rates of carbon steel in the presence of
impurities (such as SOx and NOx) in free water in equilibrium
with CO2, for a range of typical pipeline operating conditions.
The risk of stress-corrosion cracking due to the presence
of H2S or CO needs to be investigated at the high partial
pressures of CO2 typical of gaseous or dense CO2 pipelines.
References
1. IPCC, 2007. Climate change 2007: synthesis report.
Intergovernmental Panel on Climate Change, Adopted
by the IPCC Plenary 27, November.
2. IEA, 2004. Prospects for carbon capture and storage.
Energy Technology Analysis, International Energy
Agency.
3. M.B.Kermani and L.M.Smith, 1997. CO2 corrosion
control in oil and gas production. European Federation
of Corrosion Publication 23, Institute of Materials,
London.
4. A.Dugstad, 2006. Fundamental aspects of CO2 corrosion,
Part 1: mechanism. Corrosion/2006, Paper 06111,
Houston, TX. NACE International.
5. M.B.Kermani and A.Morshed, 2003. Carbon dioxide
corrosion in oil and gas production – a compendium.
Corrosion, 59, 8, NACE.
6. A.Dugstad, B.Morland, and S.Clausen, 2011. Corrosion
of transport pipelines for CO2 – effect of water ingress.
Energy Procedia, 4, pp3063–3070.
7. E.De Visser, C.Hendriks, M.Barrio, M.J.Molnvick, G.De
Koeijer, S.Lijemark, and Y.Le Gallo, 2008. DYNAMIS
CO2 quality recommendations. Int. J. of Greenhouse Gas
Control, 2, 478-484.
8. P.W.Parfomak and P.Folger, 2007. Carbon dioxide
pipelines for carbon sequestration: emerging policy
issues. CRS Report for Congress, April.
9. K.Y.Song, 1988. Water content of a near-specification
CO2-rich gas mixture (CO2-5.31 mol% methane) along
the three-phase and supercritical conditions. SPE paper
18583, Houston, TX. SPE International.
10. A.Austegard, E.Solbraa, G.De Koeijer, and M.J.Molnvik,
2006. Thermodynamical models for calculating multual
solubilities in H 2O-CO 2-CH 4 mixtures. Chemical
Engineering and Design, 84, A9, 781-794.
11. Y.-S.Choi and S.Nesic, 2009. Corrosion behaviour of
carbon steel in supercritical CO2-water environments.
Corrosion/2009, NACE, paper 09256.
12. N.Spycher, K.Pruess, and J.Ennis-King, 2003. CO2-
H2O mixtures in the geological sequestration of CO2.
1: Assessment and calculation of mutual solubilities
from 12 to 100°C and up to 600 bar. Geochemica et
Cosmochimica Acta, 67, 16, pp 3015-3031.
13. M.Mohitpour, H.Golshan, and A.Murray, 2003. Pipeline
design and construction: a practical approach. American
Society of Mechanical Engineers, New York.
14. IPCC, 2005. Carbon dioxide capture and storage. Special
report, Intergovernmental Panel on Climate Change.
238
15. Rogers and Mayhew, 1980.
16. R.W.Kear. Low-temperature corrosion by flue gas
condensates, Part 1.
17. W.M.M.Huijbregts and R.Leferink, 2004. Latest advance
in the understanding of acid dewpoint corrosion:
corrosion and SCC in combustion gas condensates.
Anti-Corrosion Methods and Materials, 51, 3, pp173-188.
18. A.Fradet, S.Saysset, P.Odru, P.Broutin, J.Ruer, and
M.Bonnissel, 2007. Technical and economic assessment
of CO2 transportation for CCS purposes. J. Pipeline Eng.,
6, 3, September.
19. M.Seiersten, 2001. Materials selection for separation,
transportation and disposal of CO2. Corrosion/2001,
paper 01042, Houston, TX. NACE International.
20. A.Dugstad and B.Morland, 2010. Transport of dense
phase CO2: when is corrosion a threat? The First
International Forum of CO2 by pipeline proceedings,
Newcastle, July.
21. A.Dugstad, B.Morland, and S.Clausen, 2011. Corrosion
of transport pipelines for CO2 – effect of water ingress.
Energy Procedia, 4, 3063–3070.
22. W.A.Propp, T.E.Carleson, C.M.Wai, P.P.Taylor,
n
K.W.Daehling, H.Shaoping, and M.Abdel-Latif, 1996.
Corrosion in supercritical fluids. US Department of
io
Energy report DE96014006.
st y
ut
23. Y.Zhang, K.Gao, and G.Schmitt, 2011. Water effect on
di op
steel corrosion under supercritical CO2. Corrosion/2011,
rib
paper 11378, NACE International.
or c
24. B.P.McGrail, H.T.Schaef, V.-A.Glezakoub, L.X.Dang,
and A.T.Owen, 2009. Water reactivity in the liquid
t f ple
and supercritical CO2 phase: has half the story been
neglected? Energy Procedia, 1, 3415–3419.
25. W.Schremp and G.R.Roberson, 1973. Effect of
no m
supercritical carbon dioxide (CO2) on construction
materials. SPE paper 4667.
Sa
26. Ayello, Evans, Thodla, and Sridhar, 2010. Effect
of impurities on corrosion of carbon steel in
supercritical CO2. Corrosion/2010, paper 10193, NACE
International.
The Journal of Pipeline Engineering
27. Farelas, Choi, and Nesic, 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.
28. Materials issues in CO2 capture, transport, and storage
infrastructure. Corrosion/2012, NACE, paper C2012-
01259.
29. Y.-S.Choi and S.Nesic, 2010. Determining the corrosive
potential of CO2 transport pipeline in high pCO2–water
environments. Int. J. of Greenhouse Gas Control.
30. M.Seiersten, 2001. Materials selection for separation,
transportation and disposal of CO2. Corrosion/2001,
paper 01042, Houston, TX. NACE International.
31. M.Seiersten and A.Dugstad, 2002. Corrosion in
high pressure CO2 – equipment for corrosion studies
and water solubility measurements. IFE, IFE/KR/F-
2002/001.
32. K.Kongshaug and M.Seiersten, 2004. Baseline
experiments for the modelling of corrosion at high
CO2 pressure. Corrosion/2004, paper 04630. NACE
International.
33. P.N.Seevam, J.M.Race, P.Hopkins, and M.J.Downie,
2008. Transporting the new generation of CO2 for
carbon capture and storage: the impact of impurities
on supercritical CO2 pipelines. IPC 2008, paper 64063.
34. P.N.Seevam, J.M.Race, M.J.Downie, J.Barnett, and
R.Cooper, 2010. Capturing carbon dioxide: the feasibility
of re-using existing pipeline infrastructure to transport
anthropogenic CO2. IPC 2010, paper 31564.
35. NACE. MR0175/ISO 15156, Petroleum and natural
gas industries – materials for use in H2S environments
in oil and gas production.
36. A.Brown, J.T.Harrison, and R.Wilkins, 1973.
Electrochemical investigation of stress corrosion cracking
of plain carbon steel in carbon dioxide-carbon monoxide-
water system. Corrosion, 5, pp686-695.
37. M.Kowaka and S.Nagata, 1976. Stress corrosion cracking
of mild and low steels in CO2-CO-H2O environments.
Corrosion, 32, 10, October.
3rd Quarter, 2012 239

Environmental impact of a
hypothetical catastrophic leakage
of CO2 onto the ground surface
by Dr Janice A Lake1*, M D Steven2,3, K Smith2, and B H Lomax1

1 Agricultural and Environmental Sciences, School of Biosciences, University of Nottingham,

Sutton Bonington, UK
2 School of Geography, University of Nottingham, Nottingham, UK
3 School of Geography, University of Nottingham, Semenyih, Selangor, Malaysia

E nvironmental impact of CO2 leakage from transport pipelines forms part of the overall EIA associated
with development and progress of CCS and transport in the UK. Our work includes assessment of the
effect on terrestrial flora from a hypothetical catastrophic leak of CO2 into the environment.An experiment

n
was designed utilising replicated enclosed cloches to elevate the CO2 concentration of air above a grass
sward. In an overnight gassing period (the most likely scenario for CO2 pooling, prior to dispersion at

io
dawn), CO2 levels were raised to 42% (mean). Experiments were carried out in both October and June.
st y
ut
The effects on selected plants were monitored, both in the short term (biochemistry of leaves) and long
di op

term (% change in species composition up to 12 weeks). Results indicated that an overnight 8hr exposure
rib
to high CO2 concentration did not cause significant stress to plants and no significant change in species
or c

composition due to CO2 was recorded.

t f ple

C O2 is now recognized as being the major Plants, in general, respond positively to experimental
no m

contributor to anthropogenic-induced climate change increases in atmospheric CO2 up to ~1000ppm (0.1%),

(IPCC 2007). In response to global concerns the UK but responses vary from species to species and in terms
Sa

government has set reduction of emissions targets of 34%
of 1990s levels by 2020 and 80% by 2050. Carbon capture
and storage (CCS) was cited as having a critical role during
the transition period from fossil fuel-derived energy to
renewable and sustainable sources with an emphasis on
CCS as a low-carbon mitigation strategy [1].
Transportation of CO2 to offshore storage facilities requires
investment in infrastructure. Any large-scale change to
infrastructure in the UK requires an Environmental Impact
Assessment (EIA - EU CCS Directive 2009, [2]). Under a
collaborative research programme, COOLTRANS (Dense
Phase CO2 pipeline TRANSportation), forming part of the
EIA for proposed pipeline transport and storage of dense
phase CO2 by National Grid, investigated effects on surface
vegetation of a short-term catastrophic leak commensurate
with a hypothetical rupture in a transport pipeline.
This paper was presented at the Third International Forum on the Transportation of
CO2 by Pipeline, organized by Newcastle University, Tiratsoo Technical, and Clarion
Technical Conferences, and held in Newcastle in June, 2012.
*Corresponding author’s contact details:
email: janice.lake@nottingham.ac.uk
of magnitude. These studies are usually conducted
over the long-term (weeks to months to years) [3]. CO2
concentrations of the order of 20% or more have not
been carried out experimentally, or in terms of vegetation
responses and studies are limited to natural analogues
consisting of accessible volcanic springs and vents [4, 5].
Again these studies are conducted over the long-term and
CO2 concentrations fluctuate at these sites both spatially and
temporally. A replicated short-term investigation simulating
a catastrophic CO2 leak has not been undertaken before.
It is hypothesised that CO2 pooling on the ground surface
would most likely occur during the hours of darkness when
the atmosphere is stable. The gas would then rapidly dissipate
at dawn with the return of atmospheric turbulence [6].
This scenario would constitute a very short-term exposure
to CO2 of approx. 8 hours.
Plants as sessile (immobile) organisms respond to
perturbations in environmental conditions via a complex
set of biochemical signal pathways. For example, drought
conditions are detected in plant roots as soil drying occurs;
a signal from the roots to the shoots initiates changes in
the leaf chemistry and anatomy to bring about mechanisms
240
Level of control
n
io
1. Experiment – CO2 injection
st y
2. N2 injection
ut
di op
3. Chamber, no gas injection
rib
or c
4. Open ground
t f ple
Table 1. Experimental protocol.Three levels of control were necessary to discount potential effects arising from additional factors.
no m
which avoid leaf damage caused by drought and conserve
water within the plant [7]. Similarly, a leaf can detect changes
Sa
in atmospheric CO2 and signal to new leaves, resulting in
altered anatomy and greater efficiency [8]. Measurements of
increases and decreases in concentrations of a particular set
of biochemical compounds can indicate the levels of stress
a plant is under at any point in time. If plants are unable to
mitigate stress, then metabolic processes are diverted from
normal growth to fuel survival mechanisms. This diversion
comes at a cost to growth processes and can lead to a loss of
competitive ability against neighbouring plants. Prolonged
stress will inevitably result in leaf die-back and death. Here
we take an experimental approach to test the effects of high
atmospheric concentrations of CO2 through analysis of the
biochemistry and survival of three target species occurring
in a semi-natural grass sward.
Materials and methods
Field site
The field site was located adjacent to the Artificial Soil
Gassing and Response Detection (ASGARD) facility
situated at the University of Nottingham Sutton Bonington
Campus, UK (Fig.1). An open field of un-grazed grass sward
The Journal of Pipeline Engineering
Fig.1. ASGARD site showing position of
grass sward. Sutton Bonington Campus,
The University of Nottingham, UK. 52˚52'
N 1˚14'58 W.
Additional factor
CO2 pooling at ground level
O2 depletion
N2/CO2 injection
Chamber effects
was the chosen agricultural class within which four fully
replicated experimental sites were marked out.
Target species
Three species were chosen as target species for analyses.
The grass sward was heterogeneous with respect to species
diversity; therefore plots where all three species were present
dictated positioning of plots within the field. Species
studied were two moncotydenons – Agrostis stolonifera
(creeping bent grass) and Lolium perenne (perennial
rye grass) – and one dictoyledon – Taraxicum officinalis
(common dandelion).
Experimental protocol
It was not permissible to flood an area of land with CO2
to achieve the experiment, therefore a protocol compliant
with Health and Safety directives was devised. To contain
and maintain gas concentrations at high levels in contact
with the vegetation for 8 hours, a set of modified single-
skinned moulded chambers which could be sealed to the
ground were used. The chamber was placed inside lawn
edging which was inserted into the ground to a depth of
5cm. Foam sealant was then used to fix the chamber to
3rd Quarter, 2012 241

Fig. 2. Experimental chamber - a single-

moulded plastic cache. Dimensions are
1.04 x 0.67 x 0.4 m, providing an
approximate volume of 140 L, covering
a ground area of 0.69m2.

the edging. The chambers were modified to incorporate

n
inlet (for gas injection) and outlet (for gas concentration
measurement) pipes, both of which were plugged when

io
injection and measurement was complete. All other areas
st y
ut
were sealed (Fig.2). This prototype provided the best gas
di op

retention over time. As the chambers were not completely

rib
sealed to the soil surface, some gas was expected to escape
or c

over time via the soil and possible small leaks between the
chamber and edging. Decay rate of contained CO2 was
t f ple

measured as 0.3% CO2 per minute.

Each of the four replicate sites had four plots for experiment
no m

and controls. Three levels of control were necessary to

determine the specific effect of CO2 concentrations at
Sa

ground level. Atmospheric CO2 has profound effects

on plants as it constitutes the primary carbon source for
manufacturing sugars during photosynthesis. However,
plants are also aerobic organisms and require O2 for
respiration. As gases are competitive with respect to
volume such high levels of CO2 (>20%) cause a severe
depletion of O2 which may affect the ability of plants to
respire normally. To discount the effect of O2 depletion
and quantify the effect of CO2 only, the primary level of
control is to inject N2 to produce the same level of O2
depletion as CO2 injection. Gaseous N2 is not used by
plants and therefore does not impinge on plant metabolic
processes.
An additional factor which may affect normal growth
processes of plants include the presence of the chamber
itself, which is sealed for up to 12 hours before the
experiment during set up. Within the chamber, several
environmental factors to which plants respond are
affected. These include a reduced light level prior to the
experimental period, increased temperature (greenhouse
effect), and increased humidity as the chamber is sealed;
plants themselves will raised humidity through the
Fig. 3. Schematic of plot layout within each of the four
replicate sites.The area enclosed measure 0.69m2. An outside
strip (dark shade) of 10cm width was excluded from analysis
to discount edge effects.
process of transpiration (a loss of water through the leaves
concomitant with a gain of CO2 as a carbon source).
Therefore, secondary and tertiary levels of control using
a sealed chamber without gas injection and a plot without
a chamber were necessary to discount these effects.
Table 1 lists the experimental protocol with respect to
appropriate controls.
Fig. 3 shows a schematic of each site layout, incorporating
four plots. Each plot was randomly assigned to CO2
injection or a level of control. Ecological and agricultural
studies routinely exclude a strip surrounding an
experimental plot to discount edge effects [9]. These are
factors which may affect the measured parameters, but
242
n
io
st y
ut
di op
rib
or c
which are not due to the experimental manipulation itself
such as proximity to the chamber wall, which may affect
t f ple
moisture and light levels the vegetation receives prior to
gas injection and do not therefore represent the vegetation
plot as a whole. Figure 4 shows the chambers in situ just
no m
prior to the experiment.
Sa
CO2 and N2 gas injection was initiated at nightfall (22.00hrs)
and continued until dawn (06.00hrs). The chambers were
injected using industrial cylinders (BOC, UK).
The experiment was carried out at two different seasons:
autumn (October) and summer (June) to assess whether
there were seasonal differences. Plants in the autumn are
less active in terms of growth than in summer.
Analyses
Acidification of moisture inside chambers
Immediately following the experiment, samples of moisture
inside the chambers were taken to measure pH. Samples
were measured in the laboratory using a pH meter (Hanna
Instruments, RS, UK).
Biochemical changes in target species
After removal of the chamber a sample (~5g) of each
target species leaf material was cut and immediately
frozen in liquid N2 to arrest metabolism. Samples were
The Journal of Pipeline Engineering
Fig. 4. Chambers in
situ just prior to the
experiment showing
the four replicate sites
(October 2010).
stored in a super-cooled liquid N2 vessel (Harddig Inc.
USA), transported to the laboratory and stored at -80°C
until analysis.
Biochemical assays
Phenylalanine lyase (PAL) (After Olsen 2008).
50 mg of plant material is ground in a pestle and mortar in 2
ml 100mM Tris – HCl buffer with 12 mM mercaptoethanol
(Fisher, UK), transferred to an eppendorf and centrifuged
at 16,000 rpm for 5 mins. The sample supernatant is used
in the assay. 500-µl sample, 450µl 100mM Tris-HCl (pH
8.8) and 50µl 100-mM phenylalanine is placed in a water
bath for one hour at 37°C. The reaction this then stopped
by addition of 50µl 5-M HCl. Change in absorbance is
measured on a spectrophotometer (Cary 50 UV-Visible
Varian, Northstar Scientific, UK) at 290nm in 1-cm light
path cells against blanks containing 50µl 5-M HCl before
the addition of 50µl 100-mM phenylalanine. The amount
of PAL present is expressed as nmol trans-cinnamic acid
gram-1 plant tissue hour-1.
Starch content (kit T-TSTA, Megazyme, Ireland, www.megazyme.com).
100mg plant material is ground in a pestle and mortar
with 0.2-mL ethanol (80% v/v) and put in a test tube. The
tube is vortexed for 1 minute. Immediately 3 mL amylase
is added and incubated in a boiling water bath for 6 mins,
stirring every 2 mins. The tubes are transferred to a bath
3rd Quarter, 2012
n
io
st y
ut
di op
rib
or c
t f ple
at 50°C where 4mL sodium acetate buffer (200mM, pH
no m
4.5 with sodium azide) and 0.1mL amyloglucosidase are
added and vortexed. The volume is adjusted to 10mL.
Sa
This extraction is used in the assay. Duplicate aliquots
(0.1mL) of extraction are transferred into glass test tubes
with 3mL glucose/oxidase peroxidase (GOPOD reagent)
and incubated at 50°C for 20mins. Transferred to 1-cm
light path cells and read against blanks at 510nm on a
spectrophotometer (Cecil 1100, Camlab Ltd, Cambridge
UK). Blanks are D - Glucose control (0.1mL D-glucose
standard solution in 3-mL GOPOD and reagent blank
(0.1mL distilled H2O with 3mL GOPOD reagent). Starch
content is expressed on a % w/w basis.
Succinic acid (kit K-SUCC, Megazyme, Ireland, www.megazyme.com).
50mg plant material is ground in a pestle and mortar in
2mL distilled H2O. Samples are centrifuged at 3000rpm
for 10mins. The extract supernatant is used in the assay. In
1-cm light path cells, 2mL distilled water sample plus 0.2mL
buffer, 0.2mL nicotinamide- adenine dinucleotide (NADH),
0.2mL adenosine – 5’–triphosphate /phosphoenolpyruvate/
succinyl-Co-A (ATP/PEP/CoA) and 0.02mL pyruvate kinase/
L-lactate dehydrogenase (PK/L-LDH) are added to 0.1mL
sample. The suspension is mixed and absorbance read on
a spectrophotometer (Cecil 1100, Camlab Ltd, Cambridge
243
Fig. 5. Concentration of CO2 (a) and O2 (b)
maintained within the chambers for the duration
of the experiment October 2010. CO2 shows the
four experimental chambers for each replicate
sites, O2 shows the four experimental chambers
and the four N2 (O2 depletion) controls for each
replicate site. (1 to 4 = CO2 injected chambers,
4 to 8 = N2 injected chambers).
UK) at 340nm after 3mins against a blank (0.1mL distilled
water instead of sample). The reaction is started by addition of
0.02mL succinyl-Co-A synthase (SCS) and read again after 6
mins. Succinic acid content is expressed as µg/g fresh weight.
Plant resilience to CO2 pooling
Measurement of the percentage cover of each of the target
species was taken before gassing to monitor any effect
of the gas episode on each species’ ability to maintain
growth. Subsequent weekly measurements following the
experiment continued for four weeks in October 2010 and
12 weeks in June 2011 after the gassing event.
Results and discussion
Figure 5a shows the CO2 concentrations within the four
experimental chambers for the period of gassing during the
October, 2010, experiment. A mean of 42.4% CO2 (mean
minimum of 12.2% CO2, mean maximum of 58% CO2). The
June experiment had a mean CO2 concentration of 35.1%
(mean minimum of 11.8% CO2, mean maximum of 48.2%)
O2 levels measured in both CO2 and N2 injected chambers
showed that both gasses depleted oxygen levels to a mean of
10% O2 (±4.2%) in October 2010 (Fig.5b) with comparable
values in June 2011 (data not shown).
244
n
io
st y
Acidification of moisture inside chambers
ut
di op
rib
Acidity of the atmosphere surrounding plants can have
or c
detrimental effects on growth and metabolism, as in the case
of SO2 (acid rain). In both the October and June experiments,
t f ple
pH shows a trend for acidification in chambers that were
gassed compared to the non-gassed chambers, with CO2
injection producing the lowest mean pH (Fig.6). pH was also
no m
lower in October compared to June. The differences were
not statistically significant (p=.0.05, Student’s t-test, n = 4).
Sa
Biochemical changes in target species
Phenylalanine lyase (PAL) [10]
PAL is a plant compound that is indicative of stress in
plants. It is considered to be a major junction in the
biological chemical signalling pathways which organisms
use to respond to environmental stresses. Therefore, if
more of the compound is found in leaf material, it would
signify that the plant is under stress. Figure 7 shows PAL
activity in October (a) and June (b). PAL activity shows no
significant differences in response to CO2 injection, O2
depletion or chamber presence in all target species. Levels
of PAL in October are less than that in June for all target
species, indicating an additional stress placed on the plants
in spring. This is most likely due to drought experienced in
spring, 2011. Species’ difference is manifested as Taraxacum
officinalis exhibiting less PAL in the autumn. This is likely
due to a shut-down of metabolism in this plant, which as
a dicotyledonous weed, sheds above-ground organs during
winter, whereas grasses (monocotyledons) maintain leaf
material throughout the year.
The Journal of Pipeline Engineering
Fig. 6. pH of moisture
collected from
chambers following
overnight gas injection
of CO2 and no-gas
control chambers (n =
3, bar = SEmean).
Starch content
A reduction in starch content of leaves can be indicative of
increased respiration, which is also a sign of increased stress
levels. This is of interest as the O2 levels are severely depleted
during gassing and increased respiration may be anticipated
as a consequence. Figure 7 shows starch levels for October
2010 (c) and June 2011 (d). Starch content was not statistically
significantly affected by CO2 treatment in any of the target
species. There was also no effect of O2 depletion or presence
of the chamber. The target species differed very little, however
starch content was an order of magnitude higher in October than
June. This would be expected as plants accumulate starch over the
growing period for maintenance of underground organs (roots)
during the winter period. In June, carbon gain is used directly
for leaf growth in the spring/summer season rather than for
starch accumulation.
Succinic acid content
Succinic acid is a compound that forms part of the respiratory
(Tri-carboxylic acid - TCA) cycle in plants. Increased or
decreased succinate can inform on levels of respiration in
plants, an increase in respiration is indicative of stress. This
compound has also been found to increase under conditions
of elevated atmospheric CO2 (0.3%) in laboratory experiments
(pers. obs. JAL). Figure 7 shows succinic acid levels for target
species in October 2010 (e) and June 2011 (f).
Succinate concentrations are similar in both the autumn and
spring experiments, with no significant differences between
CO2 pooling and all three levels of control. Lolium perenne has
a significantly greater content than the other target species.
These results suggest that the TCA cycle is operating at a similar
 ARE YOU UP TO SPEED?
0 0 0 2 0 1 2

Training courses – late 2012

TRAINING
October 1–5 Practical Pigging Operations – Onshore (Tricht, Netherlands)
October 2–5 Subsea Production Systems Engineering (Aberdeen)
OCT 2012 October 15–16 Pipeline Repair Methods, Hot Tapping, and In-Service Welding (Berlin, Germany)
October 15–16 Performing Pipeline Rehabilitation (Berlin, Germany)
October 22–26 Practical Pigging Operations – Subsea (Bergen, Norway)

November 5–9 Onshore Pipeline Engineering (Houston)

n
November 12–13 DOT Pipeline Safety Regulations – Overview and Guidelines for Compliance (Houston)

io
November 12–14 Defect Assessment in Pipelines (Houston)
st y
ut
di op

November 12–16 Practical Pigging Operations (Houston)

November 12–16 Subsea Pipeline Engineering Course (Houston)
rib

2012
November 12–16 Pipeline Integrity Courses, Houston – November 2012 (Houston)
or c

NOV 2012 November 13–16 Subsea Production Systems Engineering (Houston)

t f ple

November 14–16 Defect Assessment Calculations Workshop (Houston)

November 14–16 Advanced Pipeline Risk Management (Houston)
November 19–20 Pipeline Repair Methods, Hot Tapping, and In-Service Welding (Kuala Lumpur)
no m

November 19–21 Defect Assessment in Pipelines (Kuala Lumpur)

Sa

November 21–23 Pipeline risk management and assessment (Kuala Lumpur)

November 21–23 Pipeline pigging and maintenance (Kuala Lumpur)
November 21–23 Inspection of offshore pipelines and structures (Kuala Lumpur)

December 3–4 Pigging & In-line Inspection (Calgary)

December 3–4 Defect Assessment Calculations Workshop (Calgary)
DEC 2012 December 5–7 Defect Assessment in Pipelines (Calgary)
December 5–7 Pipeline Integrity Management (Calgary)

Working with a faculty of 38 leading industry experts, Clarion and Tiratsoo Technical are privileged to provide some of the best avail-
able industry based technical training courses for those working in the oil and gas pipeline industry, both onshore and offshore.

Complete syllabus and registration details for each course are available at:
www.clarion.org
246 The Journal of Pipeline Engineering

n
io
st y
ut
di op

Fig. 7. PAL activity in October 2010 (a) and June 2011 (b);
rib
Starch content in October 2010 (c) and June 2011 (d);
or c

Succinic acid content in October 2010 (e) and June 2011 (f)
for the target species immediately following CO2/N2 injection
t f ple

and controls.T = Taraxacum officinalis (black), A = Agrostis

stolonifera (dark grey), P = Lolium perenne (light grey), Fig. 8. Change in % coverage of target species over
(n = 4 for each species, bar = SEmean). 4 weeks October – November 2010.
no m
Sa

rate in October and June and supports the suggestion that
increased starch content in autumn is for overwintering and
not to facilitate stress-related increases in respiration.
Plant resilience to CO2 pooling at night
Figure 8 shows changes in the % coverage of each species
for four weeks after the experiment in October 2010. CO2
injected plots are most similar to open control plots, showing
no effect of CO2. T. officinalis and L. perenne show opposite
trends indicative of plant competition as the herb dies
back for overwintering and the grass covers more ground
surface. The erratic cover changes of A. stolonifera in N2
injected and open plots are most likely due to grazing by
rabbits present on site. Changes following the experiment
in June show similar results (data not shown) with the lack
of a CO2-specific response.
Conclusion
In conclusion there are no significant effects on measured
plant biochemistry following a simulation of CO2 pooling
for 8 hrs overnight. Furthermore, there is no lasting effect
on target species’ ability to compete after monitoring over
the longer-term. While in scientific terms these results are
negative, they indicate that the effects on vegetation of an
overnight catastrophic leak of CO2 are likely to be small and
short term. However, the effects on fauna (not studied here)
may be severe and could affect the overall ecological response.
Acknowledgements
The authors acknowledge National Grid, UK for funding
via the COOLTRANS project.
References
1. SRREN 2011 published online http://srren.ipcc-wg3.
de/report/IPCC_SRREN_.
2. J.Koorneef, A.Ramirez, W.Turkenburg, and A.Faaij,
2012. The environmental impact and risk assessment
of CO2 capture, transport and storage – an evaluation
of the knowledge base. Progress in Energy and Combustion
Science, 38, 62-86.
 n
io
st y
ut
g
di op

ducinPPIM
IntroThe
rib
or c

YOUNG ACHIEVEMENT AWARD 2013

t f ple

Do you know a colleague or employee of a pipeline operating, service

or supply company who has demonstrated a valuable and original
contribution to the field of pipeline integrity?
no m

Nominate them now for the PPIM “Young Achievement Award”.

Sa

For more information on award criteria or to nominate a colleague visit

www.clarion.org

The international gathering of the global pigging industry!

Organized by
248 The Journal of Pipeline Engineering

3. A.D.B.Leakey, E.A.Ainsworth, C.J.Bernacchi, A.Rogers, 7. W.J.Davies, G.Kudoyarova, and W.Hartung, 2005.

S.P.Long, and D.R.Ort, 1999. Elevated CO2 effects on
plant carbon, nitrogen, and water relations: six important
lessons from FACE. Journal of Experimental Botany, 60,
2859-2876.
4. F.Miglietta and A.Raschi, 1993. Studying the effect
of elevated CO2 in the open in a naturally enriched
environment in Central Italy. Vegetatio 104/105, 391-400.
5. Y.Onada, T.Hirose, and K.Hikosaka, 2009. Does leaf
photosynthesis adapt to CO2-enriched environments?
Experiments on plants originating from three natural
CO2 springs. New Phytologist, 182, 698-709.
6. J.Rohatgi and G.Barnezier, 1999. Wind turbulence and
atmospheric stability – their effect on wind turbine
output. Renewable Energy, 16, 908-911.
Long distance ABA signalling and its relation to other
signalling pathways in the detection of soil drying and
the mediation of the plants response to drought. Journal
of Plant Growth Regulation, 24, 285-295.
8. J.A.Lake, W.P.Quick, D.J.Beerling, and F.I.Woodward,
2001. Plant development: Signals from mature to new
leaves. Nature, 411, 154-154.
9. Ries L, Sisk TD (2004) A predictive model of edge effects.
Ecology 85:2917-2926
10. K.M.Olsen, U.S.Lea, R.Slimestad, M.Verheul, and
C.Lillo, 2008. Differential expression of four Arabidopsis
PAL genes; PAL1 and PAL2 have functional specialization
in abiotic environmental-triggered flavonoid synthesis.
Journal of Plant Physiology, 165, 1491-1499.

n
Dr Mo Mohitpour

io
31 March, 1942 – 21 August, 2012
st y
ut
di op

It is with great sadness that we report the passing away of Dr Mo Mohitpour in Vancouver, as the Journal
rib
went to press. He passed away after a short illness from congestive heart failure, leaving behind his beloved wife
or c

Carol, his son Bijan, daughter Rachel, and the sunshine of his riper years – four grandchildren of whom he was
immensely proud. A memorial service and celebration will be held in Calgary, Alberta, in conjunction with the
t f ple

International Pipeline Conference, and a full obituary of this well-known and much-loved pipeline-industry expert
will be published in our December issue.
no m
Sa
 The new online information service
that unlocks the secrets of the global
pipeline industry

n
io
st y
ut
di op
rib
or c
t f ple

Pipelines International Premium is the international

oil and gas pipeline industry’s foremost in-depth
source of information, comprising a digest of
no m

high-quality papers covering the latest technology

Sa

and reviews of the pipeline industry worldwide, and

a comprehensive project database.

It is comprised of:
Pipelines International Digest which provides a month-
ly update of papers covering all areas of the
industry – from key projects, and engineering and
construction issues, to environmental, regulatory,
legal and financial issues.

Pipelines International Projects which allows subscribers

to access a searchable database of completed and
current projects.

Subscribe or get a free 14 day trial now at

www.pipelinesinternational.com/premium
 Sa
no m
t f ple
or c
di op
st y
rib
ut
io
n