4th Quarter, 2010 265

A dynamic boundary ductile-

fracture-propagation model for
CO2 pipelines
by Prof. Haroun Mahgrefteh*, Solomon Brown, and Peng Zhang

Department of Chemical Engineering, University College London, UK

T HE DEVELOPMENT and testing of a dynamic boundary ductile-fracture-propagation model for

pressurized CO2 pipelines is presented. The model accounts for all the important fluid-structure
interactions governing the fracture process.These include expansion-wave propagation, real fluid behaviour,
pipe/wall fiction, and heat transfer, as well as the rapidly diminishing dynamic loading effects as the crack tip
opens. The resistance to crack-tip propagation is determined based on the drop-weight tear test energy
approach. The performance of the fracture model is tested by comparison of its predictions of the crack-
propagation velocity versus crack length against real data. The latter include the High-Strength Line Pipe
Committee, ECSC X100 and Alliance full-scale burst tests conducted for pipes containing either air or rich
gas mixtures. In all cases good agreement is obtained between the model predictions and the real data.The
validated model is used to test the propensity of a hypothetical but realistic pressurised CO2 pipeline to
ductile fracture propagation failure. The simulations indicate the remarkably significant role of the starting
line temperature on fracture propagation in CO2 pipelines.

T HOUSANDS OF KILOMETRES of pressurized pipelines

are used to transport large amounts of hydrocarbons
across the world. Although this method of transportation
is generally considered to be safe, pipeline failures do occur
with some leading to catastrophic consequences (see for
example Refs 1, 2). In the US alone, despite having one of
the most stringent safety requirements across the globe, over
202 pipeline incidents were reported during 2005 – 2009 [3].
These resulted in an estimated $2 billion of damage leading
to 69 deaths and 254 serious injuries.
Propagating factures are considered as by far the most
catastrophic type of pipeline failure. Such failures
involve the rapid axial splitting or tearing of the pipeline,
sometimes running over distances of several hundred
meters resulting in massive loss of inventory in a very
short time. Deservedly, understanding and modelling of
the mechanisms responsible for such type of failure has
led to a large number of studies (see for example Refs 4,
5). Such interest has intensified recently [6-8] given the
prospect of using pressurised pipelines for transporting
captured CO2 from fossil plants for subsequent storage.
This paper is based on one presented at the First International Forum on Transportation
of CO2 by Pipeline, organized in Newcastle upon Tyne in July, 2010, by Tiratsoo
Technical and Clarion Technical Conferences, and with the support of the University
of Newcastle and the Carbon Capture and Storage Association.
*Author’s contact details
tel: +44 (0)20 7679 3835
email: h.mahgerefteh@ucl.ac.uk
Given that CO2 at concentration of >10% v/v is likely to
be instantly fatal [9], the rupture of a CO2 pipeline near
a populated area can lead to catastrophic consequences.
Fractures can initiate from defects introduced into the
pipe by outside forces such as mechanical damage, soil
movement, corrosion, material defects, or adverse operating
conditions. Fractures propagate when the stresses acting
on the defect overcome the fracture initiation tolerance
of the pipe, reaching a critical size based on the pipeline
material properties and operating condition. As such it is
highly desirable to design pipelines such that when a defect
reaches a critical size and fails, the result is a leak rather than
a long running facture.
The above requires a two-tiered design approach involving:
• providing sufficient fracture initiation resistance,
mainly via specifying the required pipe toughness,
wall thickness and operating conditions
• ensuring sufficient fracture propagation resistance
such that if a running fracture occurs its length is
limited to a short distance
Notwithstanding cost implications, fracture initiation can
be largely controlled a priori by specifying the required
fracture initiation toughness, minimum wall thickness and
the maximum stresses acting upon the defect.
266 The Journal of Pipeline Engineering

HLP
Parameter ECSC Alliance
A1 B1 C2
Rich Gas (see Rich Gas (see
Inventory Air Air Air
table 2) table 2)
Internal diameter (m) 1.182 1.182 1.182 1.4223 0.8856
Pipe thickness (m) 0.0183 0.0183 0.0183 0.0191 0.0142
Initial pressure (bara) 116 116 104 126 120.2
Initial temperature (oC) 12 6 -5 20 23.9
Ambient pressure (bara) 1.01 1.01 1.01 1.01 1.01
Ambient temperature (oC) 20 20 20 20 20
Pipe length (m) 35 35 35 35 100
Tensile stress (MPa) 505 505 505 807 505
Yield stress (MPa) 482 482 482 728 482
Pipe grade X70 X70 X70 X100 X70

Table 1. Pipeline characteristics and prevailing conditions used for the full-scale burst tests.

However, controlling fracture propagation once a leak has
formed is more complex, presenting a unique set of challenges.
As well as the fracture toughness of the steel and the backfill
conditions, the fracture-propagation velocity and arrest length
depend on the depressurization rate, the thermal stresses, and
the minimum pipe wall temperature relative to its ductile-
to-brittle transition temperature. To model the above and
hence develop methodologies for overcoming such a type of
failure, we need to understand the nature of the processes
taking place once a fracture has been initiated.
The onset of a leak in the pressurized pipeline results in a
series of expansion waves that propagate from the rupture
plane towards the intact end of the pipeline at the speed of
sound [10]. As the main driving force for crack propagation
is the crack tip pressure [11], the precise tracking of the
expansion waves, and their effect on the pressure profile
along the pipeline, is essential for the proper modelling of
fracture propagation.
Additionally, given the significant drop in the speed of
sound and hence the depressurization rate during the
transition from the gaseous to the two-phase region [12],
such analysis must also account for real fluid behaviour
through the use of an appropriate equation of state. Also,
non-isentropic effects such as the fluid/pipe wall friction
and heat transfer must be accounted for as these also directly
affect the depressurization rate. Finally the temperature
drop as a result of the Joule-Thomson expansion cooling
[13] of the fluid within the pipeline during discharge can be
significant. In the case of CO2, depending on the starting
conditions, such temperatures can reach as low as -70°C
resulting in very significant localised cooling of the pipe
wall in contact with the escaping fluid.
The minimum pipe wall temperature reached relative to
its ductile to brittle transition temperature will dictate
whether the pipeline will fail in the ductile or brittle
fracture manner. The modelling of brittle fractures in

Fig.1. Schematic representation of the

experimental setup used in the HLP
full-scale pipe burst tests [20].
4th Quarter, 2010
Fig.2.Variation of crack velocity with
crack length for test A1 south-running
crack.
Inventory:
air, initial pressure = 116bara, initial
temperature = 12°C.
Curve A: experimental data [20].
Curve B. DBFM prediction.
pressurized pipelines has been presented in the authors’
previous publication [13]. Ductile fractures are the focus
of our attention in this work.
The so called Battelle Two-Curve (BTC) approach by
Maxey [5] was the first used to express the criterion for the
propagation of a ductile fracture in terms of the relation
between the fluid decompression-wave velocity and the
crack-propagation velocity. If the fluid decompression-
wave velocity is larger than the crack velocity, the crack tip
stress will decrease, eventually dropping below the arrest
stress and causing the crack to arrest. Conversely, if the
decompression-wave velocity remains smaller than the crack
velocity, the crack tip pressure will remain constant resulting
in indefinite propagation.
Several studies have since been conducted for modelling
ductile fractures based on the BTC approach (see for
example Refs14, 15). Some employ sophisticated finite-
element methods for simulating material deformation
but use over-simplistic transient fluid flow models for
predicting the rupture plane pressure and hence the crack
driving force (see for example Refs 16, 17). Others, on
the other hand, although accounting for the transient
depressurization profile within the pipeline, do not
deal with the impact of pipe wall heat transfer and
friction effects on the fluid decompression behaviour
(see for example Refs 18, 19). Additionally a reliable
267
decompression model must also incorporate a suitable
equation of state. This is especially important in the
case of CO2 pipelines given the unique depressurization
thermodynamic trajectory of CO2 [8].
Crucially none of the studies reviewed simulate the dynamic
interaction between the rapidly changing crack tip opening
area and the pressure loading as the crack propagates.
In this paper, we report the development and validation of
a rigorous dynamic boundary ductile-fracture-propagation
model which takes into account all of the important transient
fluid/structure interactions governing the fracture process.
The performance of the model in terms of predicting the
crack-propagation velocity and arrest length is tested by
comparison against real data. These full-scale burst tests
conducted by the High-Strength Line Pipe Committee[20],
ECSC X100 [21] and Alliance[22] for pipes containing either
air or rich gas mixtures.
The validated model is used to test the propensity of a
hypothetical but realistic pressurized CO2 pipeline to ductile
fracture propagation failure, paying particulate attention
to the impact of the starting line temperature. The latter
investigation was prompted by the findings of Cosham
and Eiber [7] indicating the significance of the starting
temperature on the CO2 depressurization trajectory relative
to its phase transition boundary.
268
Fig.3.Variation of crack velocity with
crack length for test B1 south-running
crack.
Inventory:
air, initial pressure = 116bara, initial
temperature = 6°C.
Curve A: experimental data [20].
Curve B: DBFM prediction.
Theory
The full background theory of the fluid flow model employed
in this study to predict the fluid decompression velocity
and the crack tip pressure for a given opening area is given
elsewhere [23-26], and hence only a brief account of its
main features is given here. Based on the homogeneous flow
assumption, in the case of unsteady, one-dimensional flow
the mass, momentum and energy conservation equations
are respectively are given by:
where ρ, u, P and h are the density, velocity, pressure and
specific enthalpy of the homogeneous fluid as function of
time, t, and space, x; qh is the heat transferred through the
pipe wall to the fluid and βy is the friction force term given by:
The Journal of Pipeline Engineering
where, fw is the Fanning friction factor and D the pipeline
diameter.
Also,
where θ is the angle of inclination of the pipeline to the
horizontal.
Equations 1-3 are quasi-linear and must be solved numerically.
In this study, the Method of Characteristics (MOC) [27] is
used as the numerical solution method, as opposed to other
numerical techniques such as finite-element [28, 29] and
finite-difference methods [30-32] as both have difficulty in
handling the choking condition at the rupture plane. The
MOC handles the choked flow intrinsically via the Mach
line characteristics. Moreover, MOC is considered to be more
accurate than the finite-difference method as it is based on
the characteristics of wave propagation. Hence, numerical
diffusion associated with a finite-difference approximation
of partial derivatives is reduced.
The key step in the BTC method is the derivation of two sets
of curves: one set describing the crack velocity, and the other
the velocity of fluid decompression wave. The resistance to
crack propagation is indicated by the Charpy V-Notch (CVN)
4th Quarter, 2010 269

Fig.4.Variation of crack velocity with

crack length for test C2 south-running
crack.
Inventory:
rich gas (Table 2), initial pressure =
104bara, initial temperature = -5°C.
Curve A: experimental data [20].
Curve B: DBFM prediction.
energy [5]. However, in the full-scale pipe bust tests conducted
by the High-Strength Line Pipe Committee (HLP) [33], the Component HLP C2 Alliance Test 1
BTC theory is used in conjunction with the drop-weight CH4 89.57 80.665
tear test (DWTT) energy, as this is shown to provide a more
accurate indication of the pipeline resistance to fracture. C2H6 4.7 15.409
Consequently this is the model applied in this work. C3H8 3.47 3.090
iC4H10 0.24 0.232
The two-curve model for the crack propagation velocity, vc,
and crack arrest pressure, Pa, are respectively given by [33]: nC4H10 0.56 0.527
iC5H12 0.106 0.021
nC5H12 0.075 0.014
nC6H14 0.033 0.003
nC7H16 0.017 0
nC8H18 0.008 0
nC9H20 0.001 0
N2 0.5 0.039
CO2 0.72 0
Table 2. Rich gas feed compositions.
where σflow, Dp, and Ap are respectively the flow stress (the tip pressure Pt is taken to be the choked pressure at the
mean value of the tensile and yield stresses), pre-cracked pipeline release plane.
DWTT energy and ligament area of a pre-cracked DWTT
specimen. On the other hand Pt and tw are the crack tip For brevity, full details of the coupling of the fracture and the
pressure and pipe wall thickness respectively. The crack fluid decompression models will be reported in a separate
270
study. The calculation algorithm automatically corrects
for the effective pipeline length and hence the crack tip
pressure as crack opens. The required fluid decompression
velocity and the crack tip pressure are determined from
the numerical solution of the conservation equations (1-3)
using our CFD computational package, PipeTech [34]. The
Peng-Robinson [35] equation of state (PR EoS) is used for
the prediction of the pertinent fluid phase equilibrium data
for both air and rich gas mixtures. In the case of CO2, the
Modified Peng-Robinson [36] EoS is used. As compared
to the PR EoS, this equation has been shown to produce
better predictions of the phase equilibrium data during the
most part of the depressurization process [37].
Results and discussion
Validation
The following shows the results relating to the validation of the
dynamic boundary ductile-fracture model presented above,
hereby referred to as DBFM, by comparison of its predictions
against the following published experimental data:
• HLP full-scale burst test [20]
• ECSC X100 pipe full-scale burst test [21]
• Alliance full-scale burst tests [22]
The Journal of Pipeline Engineering
Fig.5.Variation of crack velocity with
crack length for test ECSC X100
south-running crack.
Inventory:
air, initial pressure = 126bara, initial
temperature = 20°C.
Curve A: experimental data [33].
Curve B: DBFM prediction.
Table 1 shows the pertinent conditions relating to each
test. Table 2 on the other hand shows the rich gas feed
compositions for HLP C2 and Alliance tests.
The full-burst-test pipelines used comprised several sections
of differing toughness for which the corresponding DWTT
energy may be calculated. In all simulations, the pipe wall
roughness and heat-transfer coefficient are taken as 0.05 mm
and 5 W/(m2 K) respectively. The latter correspond to the
uninsulated pipeline exposed to still air in all simulations.
An equidistant grid system comprising 100 nodal points is
employed for the fluid dynamic simulations using PipeTech.
The corresponding discretisation time element is determined
using 90% of the Courant, Friedrichs and Lewy value [38].
The HLP full-scale experiments involved three series of
burst tests, referred to as test series A, B, and C using X70
API grade pipelines containing air and a rich gas mixture.
Pipeline fracture was initiated using an explosive charge.
Figure 1 shows a schematic representation of the pipe setup.
Figures 2 to 4 show the variation of the crack velocity with
crack length for the south-running A1, B1 and C2 tests
respectively for the HLP full-scale experiments. Curves A
show the measured crack length; Curves B, on the other
hand are the simulation predictions. In all cases, the
4th Quarter, 2010
Fig.6.Variation of crack velocity with
crack length for test Alliance Test 1.
Inventory:
rich gas (Table 2), initial pressure
= 120.2bara, initial temperature =
23.9°C.
Curve A: experimental data [22].
Curve B: DBFM prediction.
corresponding Charpy Energy, Cv, for each pipe section is
given in the figures.
Figures 5 and 6 show the corresponding data for ECSC
X100 [21] and Alliance full-scale burst tests [22], respectively.
Returning to Figs 2-6, as it may be observed, the crack
velocity significantly decreases with increase in crack length.
This is due to the significant rapid decrease in the crack
tip pressure as the pipeline depressurizes. As an example,
such behaviour expressed in terms of the variation of the
cark tip pressure with time is shown in Fig.7 for the HLP
A1 south-running crack.
Also as expected, the crack velocity decreases as the crack
propagates into the pipeline section with the higher
toughness, eventually coming to rest in all cases. As expected,
the data in Fig.4 show the smallest crack length as compared
to the other tests due to the combination of the much higher
fracture toughness pipe material employed together with
the lowest initial pressure.
The initial rapid increase in the crack velocity observed in many
of the test data is due to the finite time taken for the initial
notch to fully develop into an open flap following detonation.
This time domain is ignored in the present simulations.
271
Returning to the simulation data (curves A), given the
experimental uncertainties, it is clear that in all cases the
DBFM predictions produce reasonably good agreement
with the test data.
CO2 pipeline ductile fracture investigation
The following describes the results of the application of
the validated DBFM to the rupture of a hypothetical CO2
pipeline. To ensure practical relevance, the respective
pipeline internal diameter and wall thickness of 590.7mm
and 9.45mm are employed in the simulations. Cosham and
Eiber [7] suggest that such dimensions are the most likely for
CO2 pipelines to be employed in CCS. The same authors
also propose that a Cv of 50 J would be sufficient to arrest
a fracture for typical operating conditions of 100barg and
10oC. The same pipeline operating conditions are chosen
in the proceeding simulations.
Figure 8, curve A, shows the predicted variation of the crack
velocity versus crack length based on the above conditions
for the CO2 pipeline. For the sake of comparison, the
analogous data for methane (curve B) and natural gas
(85% methane-15% ethane, curve C) inventories are also
presented.
272
As may be observed following an initial drop, the natural gas
pipeline (curve C) exhibits a relatively constant high-velocity
crack which propagates through almost the entire length of the
pipeline before coming to rest at approximately 89m. Similar
trends in behaviour but of a smaller magnitude is observed in
the case of the methane pipeline (curve B) where the fracture
comes to rest at a distance of approximately 18m. Of the
three cases examined, the CO2 pipeline (curve A) offers the
best resistance to ductile fracture. Here the fracture almost
instantaneously comes to rest at a distance of only 6m.
Impact of line temperature
Based on an analysis of the CO2 depressurization trajectory,
Cosham and Eiber [7] postulate that the initial temperature
of the CO2 pipeline may have a significant impact on its
resistance to ductile fracture failure.
Figure 9 shows impact of the line temperature on the
variation of fracture velocity versus fracture length for the
CO2 pipeline at four selected temperatures of 30°C (curve
A), 20°C (curve B), 10°C (curve C), and 0°C (curve D).
As it may be observed, in the range 0-20°C, an increase in
temperature results in a relatively modest increase in the
fracture velocity and fracture arrest length. The data at 30°C
(curve A) is an exception to this rule. Remarkably only a
The Journal of Pipeline Engineering
Fig.7. Simulated variation of crack tip
pressure versus time for test HLP A1
south-running crack.
10°C rise in the line temperature results in a fast-running
propagating fracture which covers the entire length of the
pipeline.
To explain the above, Fig.10 shows the variation of the
crack tip pressure with temperature for the starting line
temperatures of 10, 20, and 30°C relative to the CO2
saturation curve. The latter is generated using the Span
and Wagner [39] equation of state for CO2. The calculated
crack arrest pressure of 43bara is also indicated in the same
figure. In the case of the highest line temperature of 30°C
(curve A), soon after pipeline failure, the dense-phase CO2
inventory crosses the saturation curve at the maximum
pressure of ca. 60bara, some 17bara higher than the crack
arrest pressure of 43bara, thus resulting in a propagating
fracture. The fracture comes to rest once the crack tip pressure
is equal to the crack arrest pressure. By the time this occurs
in the case of the 30°C pipeline, the crack will have already
propagated through the entire length of the pipe.
Conclusion
The development and validation of a dynamic boundary
ductile-fracture-propagation model for pressurized pipelines
was presented. The model, based on the coupling of a
semi-empirical fracture model with the transient real fluid
4th Quarter, 2010
Fig.8.Variation of crack velocity with
crack length for a 100-m long pipe at
100barg and 10°C containing various
inventories.
flow simulator, PipeTech, takes into account all of the
important fluid/structure interactions governing the fracture
propagation and arrest process.
A particularly important feature is accounting for the change
in the effective pipeline length as the pipeline unzips and
its impact on the crack tip pressure. Following its successful
validation against real pipe burst data reported for air and
rich-gas mixtures, the model is used to test the propensity
of a hypothetical but realistic pressurized CO2 pipeline to
ductile-fracture-propagation failure. Such investigations are
particularly timely given the real prospects of using CO2
pipelines as part of the carbon capture and sequestration
(CCS) chain.
Simulations conducted using a 100-m long pipeline
containing methane, natural gas or CO2 at 100barg and
10°C revealed that whereas for the natural gas and methane
inventories the fracture propagated through most of the pipe
length, in the case CO2 pipe, the crack length was limited
to only a short distance.
The change in the temperature of CO2 however was found
to have a remarkable impact on the resistance to fracture
273
propagation. A relatively modest increase in the line
temperature from 20 to 30°C resulted in a running fracture
which propagated through the entire length of the 100m
pipe. Analysis of the data revealed that upon crack initiation,
the pipeline inventory rapidly transforms from the dense
phase into the saturated state, thereafter following a relatively
prolonged depressurization trajectory along the saturation
curve. The crack will propagate for as long as the crack tip
pressure remains higher than the crack arrest pressure. In
the case of the CO2 pipeline at 30°C this cross over will
not happen before the fracture has propagated through the
entire length of the pipe. Obviously such phenomenon will
have significant practical implications when transporting
CO2 at different ambient temperatures as part of the CCS.
The study assumes that the adopted rather simplistic but
nevertheless effective drop-weight tear test (DWTT) energy
approach validated for air and rich gas mixtures is also
applicable to CO2. This assumption is justified given the fact
that the only fluid parameter introduced into the DWTT
approach is the crack tip pressure.
Furthermore the fluid-flow model employed is based on
the plausible homogenous flow assumption in which the
274
constituent fluid phases remain at thermal and mechanical
equilibrium during the fracture-propagation process.
Addressing the non-equilibrium phase behaviour, together
with the impact of the typical impurities present within the
CO2 stream for the various capture technologies on the
ductile fracture behaviour of CO2 pipelines, is currently
being investigated by the authors.
References
1. A.Bartenev, 1996. Statistical analysis of accidents on the Middle
Asia-Centre gas pipelines. J. Haz. Mat., 46, 1, pp57-69.
2. G.Papadakis, 1999. Major hazard pipelines: a comparative
study of onshore transmission accidents. J. Los. Prev. Proc. Ind.,
12, 1, pp91-107.
3. PHMSA, 2010. Pipeline incidents and mileage reports. Retrieved
from primis.phmsa.dot.gov/comm/reports/safety/PSI.html.
4. B.N.Leis, X.Zhu, and T.Forte, 2005. Modelling running fracture
in pipelines: past, present and plausible future directions.
Proc. 11th Int. Conf. Frac., Italy.
5. W.A.Maxey, 1974. Fracture initiation, propagation and arrest.
In: Proc. 5th Symposium in Line Pressure Research. Houston.
6. H.Mahgerefteh, G.Denton, and Y.Rykov, 2008. CO2 pipeline
rupture. IChemE Symposium Series: HAZARDS XX Process
The Journal of Pipeline Engineering
Fig.9.Variation of crack velocity with
crack length for the 100-m long, 100-
barg CO2 pipe at different starting line
temperatures.
Safety and Environmental Protection, 154, pp869 - 879.
Manchester: IChemE.
7. A.Cosham and R.J.Eiber, 2008. Fracture propagation in CO2
pipelines. J. Pipe. Eng., 7, pp115-124.
8. M.Bilio, S.Brown, M.Fairwheather, and H.Mahgerefteh, 2009. CO2
pipelines material and safety considerations. IChemE Symposium
Series: HAZARDS XXI Process Safety and Environmental
Protection, 155, pp423–429. Manchester: IChemE.
9. H.Kruse and M.Tekiela, 1996. Calculating the consequences of
a CO2-pipeline rupture. En. Conv. Man., 37, 95, pp1013-1018.
10. D.J.Picard and P.R.Bishnoi, 1988. The importance of real-
fluid behaviour and non-isentropic effects in modeling
decompression characteristics of pipeline fluids for application
in ductile fracture propagation analysis. Cana. J Chem. Eng.,
66, 1, pp3-12.
11. G.Fearnehough, 1974. Fracture propagation control in gas
pipelines: a survey of relevant studies. Int. J. Pres. Ves. Piping,
2, 4, pp257-282.
12. Idem., 1987. Calculation of the thermodynamic sound velocity
in two-phase multicomponent fluids. Int. J. Multip. Flow, 13,
3, pp295–308.
13. H.Mahgerefteh and O.Atti, 2006. Modelling low-temperature-
induced failure of pressurized pipelines. AIChE J., 52, 3, pp1248-1256.
14. B.N.Leis, 1997. Relationship between apparent (total Charpy
V-Notch toughness) and the corresponding dynamic crack-
4th Quarter, 2010
propagation resistance. Proc. Int. Pipe. Conf., Calgary: ASME.
15. H.Makino, I.Takeuchi, and R.Higuchi, 2008. Fracture
propagation and arrest in high-pressure gas transmission
pipeline by ultra strength line pipes. In: 7th Int. Pipeline
Conf.. Calgary.
16. P.E.O’Donoghue, M.F.Kanninen, C.P.Leung, G.Demofonti,
and S.Venzi, 1997. The development and validation of a
dynamic fracture propagation model for gas transmission
pipelines. Int. J. of Pres. Ves. Piping, 70, 1, pp11-25.
17. Z.Zhuang, 1999. Analysis of dynamic fracture mechanisms in
gas pipelines. Eng. Frac. Mech., 64, 3, pp271-289.
18. H.Makino, T.Sugie, H.Watanabe, T.Kubo, T.Shiwaku, S.Endo,
et al., 2001. Natural gas decompression behaviour in high
pressure pipelines. ISIJ Int., 41, 4, pp389-395.
19. A.Terenzi, 2005. Influence of real-fluid properties in modelling
decompression wave interacting with ductile fracture
propagation. Oil Gas Sci. Tech., 60, 4, pp711–719.
20. T.Inoue, H.Makino, S.Endo, T.Kubo, and T.Matsumoto, 2003.
Simulation method for shear fracture propagation in natural
gas transmission pipelines. In: International Offshore and
Polar Engineering Conf., 5, pp121-128. Honolulu.
21. I.Takeuchi, H.Makino, S.Okaguchi, N.Takahashi, and
A.Yamamoto, 2006. Crack arrestability of high-pressure gas
pipelines by X100 or X120. In: 23rd World Gas Conference.
Amsterdam.
275
Fig.10.The variation of crack tip
pressure with temperature for the 100-
m long, 100-barg CO2 pipe at different
starting line temperatures.
22. D.M.Johnson, N.Horner, L.Carlson, and R.J.Eiber, 2000. Full
scale validation of the fracture control of a pipeline designed
to transport rich natural gas. Pipeline Technology, 1, 331.
23. H.Mahgerefteh, O.Atti, and G.Denton, 2007. An interpolation
technique for rapid CFD simulation of turbulent two-phase
flows. Pro. Safe. Env.
24. H.Mahgerefteh, A.Oke, and Y.Rykov, 2006. Efficient numerical
solution for highly transient flows. Chem. Eng. Sci., 61, 15,
pp5049-5056.
25. H.Mahgerefteh, P.Saha, and I.G.Economou, 1999. Fast
numerical simulation for full bore rupture of pressurized
pipelines. AIChE J., 45, 6, pp1191–1201.
26. A.Oke, H.Mahgerefteh, I.Economou, and Y.Rykov, 2003. A
transient outflow model for pipeline puncture. Chem. Eng.
Sci., 58, pp4591-4694.
27. M.J.Zucrow and J.D.Hoffman, 1975. Gas dynamics. Wiley,
New York.
28. C.Bisgaard and H.Sørensen, 1987. A finite element method
for transient compressible flow in pipelines. Int. J. Numer.
Methods Fluids, 7, pp291-303.
29. E.Lang, 1991. Gas flow in pipelines following a rupture
computed by a spectral method. J. App. Math. Phy., 42, March.
30. K.Bendiksen, D.Maines, and R.Moe, 1991. The dynamic two-
fluid model OLGA: theory and application. SPE Production,
6, 6, pp171–180. Society of Petroleum Engineers.
276
31. J.R.Chen, S.M.Richardson, and G.Saville, 1995. Modelling
of two-phase blowdown from pipelines – 1: a hyperbolic
model based on variational principles. Chem. Eng. Sci., 50, 4,
pp695–713.
32. idem, 1995. Modelling of two-phase blowdown from pipelines
– 2: a simplified numerical method for multi-component
mixtures.Ibid., 50, 13, pp2173-2187.
33. H.Makino, I.Takeuchi, M.Tsukamoto, and Y.Kawaguchi, 2001.
Study on the propagating shear fracture in high strength line
pipes by partial-gas burst test. ISIJ Int., 41, 7, pp788-794.
34. PipeTech. Pipeline rupture simulation software: www.
pipetechsoftware.com.
35. D.Peng and D.B.Robinson, 1976. A new two-constant equation
of state. Ind. Eng. Chem. Fund., 15, 1, pp59-64.
The Journal of Pipeline Engineering
36. D.Wu and S.Chen, 1997. A modified Peng-Robinson equation
of state. Chem. Eng. Comm., 156, 1, pp215-225.
37. H.Mahgerefteh, G.Denton, and Y.Rykov, 2008. A hybrid
multiphase flow model. AIChE J., 54, 9, pp2261–2268.
38. H.Mahgerefteh, Y.Rykov, and G.Denton, 2009. Courant,
Friedrichs and Lewy (CFL) impact on numerical convergence
of highly transient flows. Chem. Eng. Sci., 64, 23, pp4969-4975.
39. R.Span and W.Wagner, 1996. A new equation of state for
carbon dioxide covering the fluid region from the triple-point
temperature to 1100 K at pressures up to 800 MPa. J. Phys.
Chem. Ref. Data, 25, 6.