Scad 2011 2 2 307 PDF
Scad 2011 2 2 307 PDF
Scad 2011 2 2 307 PDF
Abstract
One of the major challenges in the design of ultra high grade (X100) gas pipelines is the identification of a
reliable crack propagation strategy. Recent research results have shown that the newly developed high
strength and large diameter gas pipelines, when operated at severe conditions, may not be able to arrest a
running ductile crack through pipe material properties. Hence, the use of crack arrestors is required in the
design of safe and reliable pipeline systems.
A conventional crack arrestor can be a high toughness pipe insert, or a local joint with higher wall thickness.
According to experimental results of full-scale burst tests, composite crack arrestors are one of the most
promising technologies. Such crack arrestors are made of fibre reinforced plastics which provide the pipe
with an additional hoop constraint. In this paper, numerical tools to simulate crack initiation, propagation
and arrest in composite crack arrestors are introduced.
First, the in-use behaviour of composite crack arrestors is evaluated by means of large scale tensile tests
and four point bending experiments. The ability of different stress based orthotropic failure measures to
predict the onset of material degradation is compared. Then, computational fracture mechanics is applied to
simulate ductile crack propagation in high pressure gas pipelines, and the corresponding crack growth in
the composite arrestor. The combination of numerical simulation and experimental research allows deriving
design guidelines for composite crack arrestors.
Keywords crack arrestors, toughness modelling, pipeline materials, fibre reinforced plastics, integrity
1
The occurrence of a longitudinal crack propagating along a gas pipeline is a catastrophic event, which
involves both economic losses and environmental damage. Hence, the fracture propagation control is an
essential strategy to ensure pipeline integrity. Fracture control is a tough task, since it requires knowledge
of the interaction between the dynamic forces driving crack growth, and the resistance forces opposing
fracture propagation.
While brittle fracture control is typically achieved by ensuring that the pipeline is operated well above the
Ductile to Brittle Transition Temperature, the ductile fracture propagation can be avoided (or at least limited)
by increasing the minimum specified toughness of the pipeline steel. However, for ultra high pipeline grades
( X100), the identification of a reliable crack propagation strategy is not longer straightforward. Indeed,
despite excellent toughness values at lab scale (Charpy upper shelf energy and Battelle shear fracture
area), one can no longer rely on pipe body arrest [1-3]. As a result, additional mechanical devices such as
crack arrestors have to be mounted on the pipeline in order to stop a running ductile crack.
According to experimental results of full-scale burst tests, it is argued [4] that the most promising materials
are
Steel sleeve arrestors, in particular tight sleeves, which are placed around the main linepipe with a
close fitting connection
Composite arrestors, made of fibre reinforced plastics which provide the pipe with an additional
hoop constraint.
307
E1
E2 = E3
" 12 = " 13
" 23
G 12 = G 13
G 23
44 200 MPa
11 400 MPa
0.356
0.343
4 420 MPa
5 027 MPa
XT
YT
XC
YC
700 MPa
7.2 MPa
588 MPa
42 MPa
30.1 MPa
30.1 MPa
When a pipeline is being installed, the crack arrestor can be subjected to tensile loads and bending
stresses. In addition, during the operational life of the pipeline system, the composite crack arrestor is
exposed to environmental loads, low temperatures, external damage, fatigue, ...
In order to evaluate the in-service behaviour of composite arrestors, an extensive testing program was
undertaken, including three and four point bending experiments and tensile tests at different temperatures.
In addition, low cycle fatigue and ageing were studied as well. For these experiments, the behaviour of a
unidirectional reinforced crack arrestor was compared with an arrestor with a winding pattern including
inclined fibres, contributing to the axial reinforcement. The winding patterns are compared on Figure 1.
308
Tensile tests
Tensile tests were performed on a medium scale S235 steel pipe (with length L = 4 m, outer diameter
D = 220 mm and wall thickness t = 3.2 mm). The length of the composite crack arrestor was L a = 660
mm, and the thickness of the windings was t a = 4.8 mm. The tensile test setup is shown on Figure 2.
For the quasi-static tensile tests at lower temperatures (-30C), a local cooling box is installed around
the crack arrestor.
Figure 2: Tensile test setup for room temperature (left) and lower temperatures (right)
In order to enable a proper introduction of the load to the pipe, the ends were furnished with flanges,
which were fasted to a clamping support, like shown on Figure 3. During the experiment, the applied
force and the piston displacement were measured. In addition, several strain gauges (shown on
Figure 3) were attached to the steel pipe and the composite crack arrestor to monitor local
deformations.
309
Figure 5: Comparison of axial strains on unidirectional crack arrestor and inclined winding pattern
Although the strain gauges clearly indicated composite material failure, no macroscopic damage
could be observed after the experiments. Dye penetrant testing revealed the exact location of the
cracks due to tensile loading. On Figure 6, the results of such a dye penetrant inspection are shown.
The tested specimen is shown before (left) and after (right) a developer has been applied to reveal
the penetrant, and hence the location of the surface-breaking cracks.
310
Figure 6: Dye penetrant testing without (left) and with (right) developer to reveal material damage
2.2
Bending experiments
Four point bending tests were performed to assess the response of the composite crack arrestors to
combined loading. The experimental setup is shown on Figure 7: the sample is subjected to four point
bending, and the load is applied adjacent to the crack arrestor. The applied load and the piston
displacement are measured, and strain gauges are applied to monitor local deformations.
Figure 7: Four point bending setup (left) and strain gauge arrangement (right)
Typical results for bending tests at ambient and lower (-30C) temperatures are shown on Figure 8,
where the applied load is shown as a function of the measured piston displacement.
311
Figure 10: Load time history (left) and strain gauge signals (right) for cyclic four point bending tests
The ends of the test specimens were fixed using bolts, and the load was applied through half-shell
shaped supports. The resulting load-deflection curve for a cyclic four point bending test is shown on
Figure 11, indicating that the maximum bending force is slightly higher than in the static bending
tests. For the unidirectional reinforced crack arrestors, failure of the epoxy resin similar to the failure
mode for the monotonic experiments- could be observed. The onset of composite material failure is
clearly captured by the strain gauges, like shown on Figure 10. Crack closure during the compressive
stress cycles contributes to an increased failure strain. During the cyclic bending tests, no separation
between the steel pipe and the composite crack arrestor could be observed, indicating that monotonic
load assumptions are justified to design composite crack arrestors to withstand ultra low cycle fatigue
loading.
312
Figure 11: Load-deflection curve for a cyclic four point bending test
3
Orthotropic (plane stress) failure measures are indications of composite material degradation, where a
sound material has an index I F = 0.0 and a failed material has an index I F = 1.0. The orthotropic linear
elastic behaviour can be extended with a failure envelope, according to different criteria. In [6], the ability of
different orthotropic failure measures was presented, and their ability to describe failure for composite crack
arrestors was evaluated. The criterion of Hashin [9] and Rotem [10] was identified as the best failure
measure for unidirectional reinforced composites.
To predict damage initiation, Hashin and Rotem propose four different criteria to distinguish between matrix
and fibre failure in tension and compression. For fibre rupture in tension (& 11 > 0), they suggest
2
(Eq. 01)
while
2
(Eq. 02)
is proposed for fibre buckling/kinking in compression (& 11 < 0). The initiation criterion for matrix cracking
under transverse tension and shearing (& 22 + & 33 > 0) reads
2
2
( & * & 33 ) ' 23
1 & 22 & 33
M + - 11
*
,1
.
T2
/ YT
0
T
(Eq. 03)
M +
C
2
& 22 * & 33 4 ( YC )
YC
(Eq. 04)
These failure measures are used to define the internal variables that characterize fibre damage
:; F T ; &11 0
df + < C
; &11 , 0
;= F
313
(Eq. 05)
:; F T ; &11 0
df + < C
; &11 , 0
;= F
(Eq. 06)
32
32
32
ds + 1 1 1 1 F T 1 1 F C 1 1 M T 1 1 M C
respectively, where the effective stress tensor
4 1
61 1 d
f
6
6
D+6 0
6
6
6 0
8
& + D &
0
1
1 1 dm
0
(Eq. 07)
5
0 7
7
7
0 7
7
1 7
7
1 1 ds 9
(Eq. 08)
The damage initiation criteria (Eq. 01) - (Eq. 04) according to Hashin are implemented in a finite element
model of the four point bending tests reported in the previous section. The pipe is modelled as a
deformable solid, with three elements through the thickness to accurately capture the bending stresses.
The unidirectional reinforced epoxy is modelled as an orthotropic composite material, with the elastic
properties of Table 1. The total problem size was 48 678 elements and 212 052 degrees of freedom. In
Figure 12, the four point bending simulation is compared with the experimental curve, indicating that the
T
tensile matrix cracking criterion M = 1 is capable of predicting the onset of damage in the composite crack
arrestor.
Figure 12: Four point bending simulation with the Hashin tensile matrix cracking criterion
4
For the simulation of ductile crack propagation in high pressure gas pipelines, the PICPRO (Pipe Crack
Propagation) code, developed by CSM and the University of Rome [11-12] was used. The model uses an
explicit integration algorithm based on a central difference scheme. As a result, it is able to take both
steady-state and transient fracture propagation conditions into account during the analysis, including abrupt
changes of constraint characteristics such as those which occur in the vicinity of crack arrestors.
The code also accounts for local strain rate effects [12], soil constraint effects [13] and decompression of
the gas flowing through the fracture breach according to the actual gas composition, pressure and
temperature. Material ductility is described by a Fracture Process Zone, which is explained in [14-15].
314
The combination of numerical simulation and experimental research allows deriving design guidelines for
composite crack arrestors. In this section, the Hashin damage model for the composite material is
combined with the PICPRO code to simulate ductile crack propagation, which enables to design fit for
purpose composite crack arrestors. First, the numerical tools are applied to calculate the optimum
dimensions (thickness and length) of a composite crack arrestor for a small-scale pipe. Then, the method is
extended to predict crack arrest in a full-scale burst test of a 36 natural gas pipeline.
5.1
315
Figure 14: Predicted crack speed diagrams for t a = 1.5 mm (left) and t a = 3.0 mm (right)
For a composite reinforcement with a thickness of 1.5 mm (i.e. equal to the pipe wall thickness), no
crack arrest is predicted. As shown in the crack speed diagram of Figure 14, the crack is initially
imposed to propagate at a speed of 250 m/s, for an internal pressure of 220 bar. When entering the
composite crack arrestor, the fracture speed is predicted to slow down to ca. 70 m/s, but propagates
further with an increasing speed. The composite reinforcement is predicted to be totally destroyed
and proves ineffective in arresting the fracture. When increasing the thickness of the crack arrestor to
3.0 mm, the simulated fracture is effectively slowed down and crack arrest is predicted.
5.2
Figure 15: Predicted crack speed diagrams for L a = 37.5 mm (left) and L a = 150 mm (right)
Although the crack speed is considerably decreased in the short arrestor, by virtue of the constraint
action exerted by the composite windings, the crack arrestor length is not sufficient to stop the
running fracture. When the length is increased to 150 mm, crack arrest is achieved within 35 mm.
The composite material damage predicted by the finite element analysis is compared in Figure 16.
Figure 16: Predicted composite material damage for L a = 37.5 mm (left) and L a = 150 mm (right)
316
[]
36
Wall thickness
[mm]
20.0
Burial depth
[mm]
1.5
MATERIAL
Grade
API 5L
X100
[%]
16.5
Yield Stress
[MPa]
760
Tensile strength
[MPa]
813
Elongation
GAS
Pressurizing medium
Burst pressure
Temperature
natural gas
[bar]
226
[C]
14
CRACK ARRESTOR
Length
[mm]
1 600
Thickness
[mm]
40.0
[MPa]
826.8
[-]
0.018
[m/s]
135
Ultimate strength
Ultimate strain
Crack speed
EXPERIMENTAL RESULT
arrest within 0.5 m
Figure 17: Demopipe composite crack arrestor before and after full-scale burst test [16]
317
Figure 18: Simulated crack speed diagram for the Demopipe full-scale burst test
CONCLUSIONS
Design considerations for crack arrestors used in ultra high grade gas transmission pipelines were
reviewed. In [5], unidirectional glass fibre reinforced epoxy was identified as the most promising material for
the manufacture of composite crack arrestors. An extensive experimental program was presented to
measure the elastic properties of the composite materials. The results from traditional mechanical
characterization and non destructive testing were compared. Micromechanical modelling of unidirectional
reinforced plastics revealed that the Hashin model is best fit to calculate the stiffness matrix, based on the
properties of the fibre reinforcement and the reson.
In this paper, the in-use behaviour of composite crack arrestors was evaluated by means of quasi-static
tensile tests and (both monotonic and cyclic) four point bending experiments. The Hashin damage model
was applied to predict the onset of composite material degradation. Finite element simulations confirmed
that the tensile matrix cracking criterion can accurately predict damage initiation.
In order to assess the ability of composite crack arrestors to stop a running fracture in a high pressure gas
pipeline, numerical simulations were performed. The combination of experimental data and finite element
analysis allows deriving design guidelines for composite crack arrestors. The design methodology was
validated by comparing numerical predictions with the results of a full-scale burst test.
ACKNOWLEDGEMENTS
The research results, presented in this paper, were obtained in the scope of the LINESPEC project on
Special Components and Strain Based Requirements for High Strength High Pressure Pipeline
Applications. This project is funded by the Research Fund for Coal and Steel (RFCS).
The authors gratefully acknowledge the support of the project partners BP, SZMF, Corus, ISQ, RWTH and
the Soete Lab (UGent).
318
REFERENCES
[1] Demofonti G. et. Al., Fracture Propagation Resistance Evaluation of X100 TMCP Steel Pipes for High
Pressure Gas Transportation Pipelines Using Full Scale Burst Tests, Proceedings of the Fourth
International Conference on Pipeline Technology, 9-13 May 2004, Ostend, Belgium
[2] Mannucci G., Di Biagio M., Demofonti G., Fonzo A., Salvini P. and Edwards A., Crack Arrestor Design
by Finite Element Analysis for X100 Gas Transportation Pipelines, Proceedings of the Fourth
International Conference on Pipeline Technology, 9-13 May 2004, Ostend, Belgium
[3] Mannucci G. and Demofonti G., Control of Ductile Fracture Propagation in X80 Gas Linepipe,
Proceedings of the Fifth International Conference on Pipeline Technology, 12-14 October 2009,
Ostend, Belgium
[4] Leis B.N., Clark E.B. and Eiber R.J., Crack Arrestor Review Concepts, Design, Testing and InService use, Battelle Institute, Report to BP Exploration and Production (2005)
[5] Van den Abeele F. and Di Biagio M., Design of Crack Arrestors for Ultra High Grade Gas Transmission
Pipelines Material Selection, Testing and Modelling, Conference on Sustainable Construction and
Design, Gent, Belgium, 16-17 February 2011
[6] Van den Abeele F. and Skocovsky T., Enhanced Failure Criteria for Composite Crack Arrestors,
Proceedings of the Fifth International Conference on Pipeline Technology, Ostend, Belgium, 12-14
October 2009
[7] Van den Abeele F., Amlung L., Di Biagio M. and Zimmermann S., Numerical Design of Composite
th
Crack Arrestors for High Pressure Gas Pipelines, Proceedings of the 8 International Pipeline
Conference, IPC2010-31191, Calgary, Canada, September 2010
[8] ECCS Technical Committee 1, Technical Working Group 1.3, Recommended Testing Procedure for
Assessing the Behaviour of Structural Steel Elements under Cyclic Loads, Brussels, Belgium, 1986
[9] Hashin Z., Failure Criteria for Unidirectional Fibre Composites, Journal of Applied Mechanics, vol. 47,
pp. 329-334, 1980
[10] Hashin Z. and Rotem A., A Fatigue Criterion for Fibre Reinforced Materials, Journal of Composite
Materials, vol. 7, pp. 448-464, 1973
[11] Berardo G., Salvini P., Mannucci G. and Demofonti G., On Longitudinal Propagation of a Ductile
Fracture in a Gas Line Pipe: Numerical and Experimental Analysis, Proceedings of the Third
International Pipeline Conference IPC2000, Calgary, Canada
[12] Fonzo A., Salvini P., Di Biagio M. and Mannucci G., Full History Burst Test Through Finite Element
Analysis, Proceedings of the Fourth International Pipeline Conference, IPC2002-27120, Calgary,
Canada
[13] Salvini P., Berardo G., Demofonti G. and Mannucci G., Numerical Model of the Backfill Constraint
Effect During Fracture Propagation on Buried Large Diameter Gas Pipelines, Proceedings of the Third
International Pipeline Conference IPC2000, Calgary, Canada
[14] Fonzo A., Meleddu A., Di Biagio M., Mannucci G., Demofonti G., Peterson C.W. and Biery N.E., Crack
Propagation Modelling and Crack Arrestor Design for X120, Proceedings of the Sixth International
Pipeline Conference, IPC2006-10319, Calgary, Canada
[15] Van den Abeele F., Di Biagio M. and Amlung L., Numerical Design of Composite Crack Arrestors for
th
High Pressure Gas Pipelines, Proceedings of the 8 International Pipeline Conference IPC2010-31191,
Calgary, Canada
[16] Demofonti G., Mannucci G., Di Vito L.F., Aristotile R., Di Biagio M., Malatesta G., Harris D; and
Harrison P., Ultra High Strength Pipeline Prototyping for Natural Gas Transmission, Demopipe, EUR
21440 EN, 2003
319