Documente Academic
Documente Profesional
Documente Cultură
Abstract
Glass reinforced ethylene (GRE) pipes are widely used in the oil and gas sector in Australia owing to their
economical/competitive life cycle costs. However, as with all pipelines, GRE pipes face significant risks of
external interference by earthmoving equipment where mechanical insult is the result of approximately 50% of
pipeline failures. Mechanical tests on un-pressurised pipes were conducted on thick and thin walled GRE pipe
samples rated to ANSI Class 600 at different loading rates to study the failure mechanisms. Localised failures
due to combined bending and shear with severe delamination under the indenter were observed in thick wall
GRE whereas thin walled GRE pipes failed due to buckling away from the point of impact. The finite element
analyses results show that GRE pipes fail due to a combination of bending and direct shear at the p oint of
impact. Further, the failures are functions of the angle of impact and the tip geometry. The magnitude of
puncture resistance as well as the location of failure and failure mechanisms were all found to be highly
dependent on the pipe wall thickness and applied internal pressure. GRE pipes exhibited 55 to 75% less
resistance to puncture as compared with steel pipes of same wall thickness. Hence, most commercially
available excavators have the potential to penetrate GRE pipes, so alternative methods must be employed to
reduce this risk.
Keywords:
GRE Pipes, Puncture Resistance, Penetration, Internal Pressure, Point Loading, finite element analysis
1 INTRODUCTION
Glass-Reinforced Epoxy (GRE) pipelines have been and Coal Mine Methane (CMM) gathering systems,
in use by the oil and gas industry in Australia for and one can see that fibreglass piping will be
over 20 years. There are now more than 1200 implemented more frequently in the near future
kilometres of high pressure fibreglass composite (Kirk-Burnnand, 2006).
pipelines in operation throughout the country. Among the many benefits that GRE brings to the
Fibreglass pipes are commonly implemented in pipeline industry, the most significant is its corrosion
corrosive environments due to performance resistance. Companies incur huge costs annually on
deficiencies of traditional materials such as steels. combating pipeline corrosion through the use of
For instance, in transmitting well production fluids, exterior coatings (i.e. paint), repairs to corroded
wet or sour gas, hydrocarbon production flowlines, sections of pipe work, and the substantial costs of
field gathering systems and waterflood applications. performing Non-Destructive Testing (NDT) on the
Add to this the growing list of potential applications, piping networks. Thus for the majority of fluid
such as dry gas and high pressure wet gas handling applications, GRE reduces the level of
transmission as well as Coal Seam Methane (CSM) resources necessary to perform day-to-day operations
41
J.Griffiths and P.Gudimetla
both efficiently and safely. Additionally, unlike An understanding of the puncture resistance of GRE
carbon steel pipelines, the pumping power is therefore vital given that this material is now being
requirements in GRE pipelines remain constant over implemented more frequently by industry. This will
time. This is because the surface roughness on the ensure that pipeline designers understand the risks
internal surface does not significantly change and as associated with their designs as well as the
such the Hazen- of life-cycle cost of fibreglass consequences of failure and help mitigate and even
Williams coefficient remains around 150. Hence, the avoid losses of life and/or severe harm to people,
whole is often substantially lower than that of property and the environment.
traditional piping materials.
2 BACKGROUND
Prior research into external interference of pipelines
has predominately focused on steel. Various studies
have been conducted into topics ranging from stress
configurations under loading, fracture mechanics,
denting characteristics and analysis of puncture
resistance of thick and thin-walled GRE pipes.
Recently, the research focus has shifted towards
(a) Underground (b) Free surface utilising finite element software packages to analyse
Figure 1: GRE Piping in Australia what would generally have been considered too
complex to study previously. Primarily two
researchers have recently performed detailed studies
Other benefits which fibreglass composites offer on the cause and effects of pipeline punctures in
include a high strength to weight ratio, light weight typical Australian pipelines. These studies were
material (which significantly reduces transportation conducted by Andrew Stewart and Daniel Brooker in
costs to remote sites during construction and conjunction with the Cooperative Research Centre
maintenance operations), dimensional stability and (CRC) at the University of Western Australia
moderate tooling costs during installation. between 2000 and 2004. As Brooker’s research was
Depending on the type of resin used to manufacture more results based, his work was used as the basis of
the pipes, GRE also has the potential to be resistant this investigation into the puncture resistance of
to chemical attack such as hydrolysis, oxidation, GRE.
pyrolysis or solution attack. Figures 1a and 1b show In contrast to the prior experiments performed by the
typical applications of GRE in the oil and gas European Pipeline Research Group (EPRG),
industry for both aboveground and underground fluid Brooker’s research focused primarily on thinner
transmission lines respectively. To date GRE has walled steel pipes and much smaller tooth penetration
been limited to relatively low pressure applications geometries more typically representative of
(typically up to 10 MPa). This is due to several Australian operating conditions. Brooker’s results
unresolved issues relating mainly to the weeping also indicated that the material’s failure point was
characteristics of composites such as GRE when significantly dependent on the bending stresses
placed under internal pressure as well as the risks acting beneath the corners of the penetrating tooth.
associated with external interference. According to This is in combination with membrane and/or shear
Cosham & Hopkins (2001), the major causes of stresses previously found in the pipe wall by Driver
pipeline failure around the world are external & Zimmerman (1998). Brooker’s derived equation
interference and corrosion, of which, corrosion has now been incorporated into AS2885 as a means
generally does not negatively affect fibreglass by which the puncture resistance (Rp) (see equation
pipelines. This research is backed up by other 1) of steel pipelines can be assessed. The puncture
sources who have calculated that mechanical load was found to be highly dependent on tooth size
interference has caused approximately 50% of
recorded failures in Europe (Bolt, 2001) and 53.5% (L & W), pipe wall thickness (t) and pipe material
in the United States (Jones, et al. 1986). Bolt (2001) grade (σu).
also found that earth moving operations, specifically ⎡ W ⎤
by excavator and/or backhoe style equipment, are the R p = 0.701(t )(σ u + 410)(L + 22.4)⎢ ⎥
(1)
⎣W + 3.14 ⎦
major cause of such external interference incidents.
42
Puncture Resistance of Pressurized & Unpressurized Glass Reinforced Epoxy (GRE) Pipes – A
Phenomenological Model of Failure Mechanisms
Puncture loads calculated using equation 1 were properties and pipe wall thickness have a significant
found to be consistent with the experimental data influence on puncture resistance. The pipe diameters
obtained from thick walled puncture tests performed chosen for the testing were typical of what is used in
by the EPRG and other tests at Batelle Columbus the oil and gas sector. Samples shorter than 10 times
Division in the mid-eighties. Thus, this equation has the pipe diameter have previously been found by
a wider range of applications (i.e. for various tooth Brooker and other researchers to give inaccurate
and pipe dimensions) than any other previously results. Consequently, pipe lengths were specified to
derived equation. be 2 metres for the steel and 1.83 metres for the GRE
It is important to note that the current Australian oil as this ensured the pipes were of a sufficient length to
and gas pipeline standard AS2885 does not provided negate end effects. This is due to deformation being
designers with any guidance on quantifying the present along the entire pipe length resulting in
puncture resistance of fibreglass. Hence, it is excessive ovalisation of the pipe and axial pull-in of
intended that this particular study will provide the ends.
designers with some insight into the risks and The aim of this testing was to experimentally resolve
potential consequences of implementing GRE pipes. several issues which were unearthed during the
Much of the research into the failure mechanisms of literature review and analysis of UWA’s data. A
filament wound composite pipes to date has focused small off-cut of GRE pipe manufactured by
on biaxial loading. Several papers have also Fiberglass Systems was sourced directly from the
analysed impact damage on thin walled fibreglass field. This specimen was cut from a 3.6mm thick
pipes but have not gone as far as analysing the actual pipe but tapered up to 12mm at one end to allow for a
puncture resistance of pipes commonly found in the machined thread to be cut for joining pipe lengths.
Australian pipeline industry (Li, et al. 2005, Curtis, et The pipe was cut into 5 sections with only two of the
al. 2000 and Christoforou & Swanson, 1991). Nor samples being suitable for comparison. These
has any prior research taken into account the tooth tip included two 175mm long samples of 12mm and
geometry of common earthmoving equipment 3.6mm wall thickness and an internal diameter of
coming into contact with a fibreglass pipe. This paper 164mm. The other three shorter (90, 113, 122mm
presents the mechanical testing of thick and thin length) specimens with thicknesses varying from
walled GRE pipes with a view to accurately develop 5.1mm to 12mm would have ideally been machined
a Phenomenological model that describes the to an equal thickness but unforeseen circumstances
deformation and failure mechanisms of GRE pipes. prevented this. Instead these three samples were
A preliminary finite element material model is utilised to gauge the effects of higher impact speeds
developed using available orthotropic properties and on the acquired puncture loads. Since no information
the numerical results are compared with experimental about the material properties of this pipe was
test data to establish trends and likely failure available, the results were only used to establish
mechanisms of various thickness GRE pipes. trends between the different data sets. Tooth
displacement was maintained at a quasi-static speed
3 EXPERIMENTAL PROCEDURE of 0.2 mm/s for all tests. Although earthmoving
Puncture testing with a 50mm spherical indenter on equipment operates at much higher rates, past
short lengths of GRE pipe procured from Fiberglass research had found that quasi-static loading gave
Systems was performed at QUT. It must be noted sufficiently accurate results of impact against steel
that the material properties used in this study were and GRE pipes (Curtis, et al. 2000). Loads were
specified by the manufacturers and are considered to applied by a 60 tonne Instron machine (Figure 5) and
be nominal values. Previous material testing were sustained until either a maximum displacement
performed by Brooker found that the material of 100mm was reached or a full puncture had formed
properties of steel were often considerably higher in the pipe. Measurements of the puncture load and
than the minimum guidelines set down in the relative tooth displacement were recorded via a
Australian Standards (AS4041-2006). Likewise, the digital data acquisition system and saved for
wall thickness of the pipes tested had a significant subsequent analysis. A 50mm spherical indenter was
variation amongst specimens with many of the GRE chosen as this was consistent with previous thin
samples having a measured wall thickness of up to 12 walled GRE impact studies (Li, et al. 2004) and
mm. It was therefore expected that the puncture testing was performed on unpressurised GRE pipes.
loads would be higher than expected since material Two impact speeds were used throughout the testing
43
J.Griffiths and P.Gudimetla
to validate the assumption that impact speed has little puncture load and tooth displacement for the
influence on the results. A speed of 0.2mm/s was pressurised tests ranged from 0.9 to 94.7% and -13.0
again used on the 3.6 and 12mm uniformly thick to 71.6%. Figure 4 shows the upper, lower and
sections as well as one of the tapered specimens (5.1 average loads required to puncture the GRE pipes. It
– 9.3mm). The other tapered specimen (9.3 – 12mm) can be seen that pressurising GRE pipes does
and threaded end (6.4 – 9.4mm) were subjected to an increase their puncture resistance but wide variations
impact speed arbitrarily chosen to be 4.0mm/s. It in the experimental results do not allow a more
was assumed that any influence speed had on the specific conclusion to be drawn.
results would be evident with the speed being 20 These large variations in obtained results were not
times higher. Testing was performed on an Instron expected since consistency in manufacturing quality
5500R universal testing machine (Figure 2) with the should have resulted in similar results for each of the
pipes being supported directly on the test bed. Force three repeated tests. Although prior research (Park &
and displacement curves were obtained from the Jang, 1998) has found that a slight change in the
Instron load cell via Bluehill computer software. adhesion between layers can have a significant effect
on a composite’s impact resistance.
r
Lo er
su r
in n
Tw tion
in n
ed
g
ne ng
ne ng
ed
P
P
P
ed
ge
es ige
Tw tio
Tw tio
n
G
G
G
g
ris
Lo
Lo
ad
Lo
ris
Ti
E n Ti
T
tra
tra
tra
su
in
P
P
P
ne
es
G
G
G
Pe
Pe
Pe
r
Pr
np
U
(below). 30
29.0 28.9
21.5
31.3
18.6
20 13.0
8.9
10
0.9
0
4 EXPERIMENTAL RESULTS & ANALYSIS
r
r
n
n
g
g
P
P
ed
ed
ge
ge
tio
tio
n
n
G
G
Lo
ad
Lo
ris
Ti
Ti
ra
ra
Lo
in
in
t
t
P
P
ne
ne
G
G
Tw
Tw
es
d
Pe
Pe
En
Pr
can be seen in Figure 3 that the puncture loads and Upper Limit (%) Average % reduction in resistance for GRE Lower Limit (%)
44
Puncture Resistance of Pressurized & Unpressurized Glass Reinforced Epoxy (GRE) Pipes – A
Phenomenological Model of Failure Mechanisms
20
15
10
5
0
(c) (d)
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75
Tooth Displacement [mm]
45
J.Griffiths and P.Gudimetla
It should be noted that ideally the images in Figure 6 the thin wall pipe had delaminated but still had not
would have been images of the pipe wall under the punctured through. Figure 8 below shows the
ends of the tooth. This is because bending stresses comparative load-indenter displacement curves for
are highest at these points and material under the the thick and thin walled GRE pipes.
mid-section of the tooth is at a significantly lower
state of stress. Hence, delamination should be more
visible in all images, especially 15b. These photos 50mm Spherical Indentation Tests
Load [kN]
20
15
10
6 THIN WALL AND HIGHER SPEED 5
IMPACT TESTING
0
46
Puncture Resistance of Pressurized & Unpressurized Glass Reinforced Epoxy (GRE) Pipes – A
Phenomenological Model of Failure Mechanisms
resulted in burst pressures suddenly dropping by Of particular interest is how this may relate to thick
60%. wall GRE pipes. The location of membrane stress or
shear flow circumferentially around the pipe wall
may account for the fracture and whitening observed
7 FINITE ELEMENT ANALYSIS
either side of the point of impact, following the
Efforts were made to model pipe puncture in the wound strands, during unpressurised testing (Figure
commercial finite element (FE) software ANSYS. 9c). This whitening was not present in the
This initially involved performing an implicit pressurised samples (Figure 10c) which compliments
analysis using idealised boundary conditions such as the fact that membrane and shear stresses were found
a ramped tooth displacement rather than an impact to be localised under the tooth within the FE models.
velocity and fixed nodes partially along the base of Further research efforts are required to confirm this
the pipe. These methods have been used in the past but it is not believed that the damage is due to
with a large degree of accuracy. shattering of the composite material. Instead, it is
Attempts to develop an explicit analysis using the thought that relatively slow fracture growth and
Tsai-Wu failure criterion and element death were delamination of the material brought on by excessive
futile. Hence, only trends in stress and strain deformation of the pipe wall are the causes. Note
distributions could be extracted from the FE that the ANSYS images included here are of quarter
simulations. The general trends show agreement with models with symmetry conditions which were
Brooker’s (2002) analysis since significant bending implemented in the FE analyses in order to minimise
stress was present on the inside wall of the steel pipes the required computational time and power.
directly under either end of each tooth. As with
Brooker’s and Driver & Zimmerman’s (1998)
analyses, bending combines with shear stress at this
point to cause failure. The FE models also show that
the von-Mises membrane stress pattern differs
between the unpressurised and pressurised thin
walled GRE pipes. It is still unclear if this is the case
for thick wall GRE pipes since an explicit analysis
was not possible and therefore the models do not
account for penetration of the tooth through the wall
thickness. Hence, accurate measurements of stress,
strain and deformation are not possible due to the
non-linear problem. Analysis of the thin walled GRE
however is more accurate because experimental tests
have shown that very little penetration actually
occurs. So there is some confidence that the (a)
developed orthotropic FE models for the thin wall
GRE pipes accurately depict what is actually taking
place. Figures 9b and 10b show that the maximum
membrane stress shifted from the side wall of the
unpressurised pipe to a location directly under the
tooth edge when the pipe was pressurised. Similarly,
it can be seen in Figures 9c and 10c that the locations
of the maximum compressive and tensile shear
stresses move from approximately 52 degrees around
the circumference of the unpressurised pipe to
directly under the tooth edge when it is pressurised. (b) (c)
This was expected as pipe deformation
Figure 9: FEA of thin-walled GRE Pipes
characteristics differ significantly between the two
(unpressurized)
loading scenarios due to induced bending and
buckling in the unpressurised pipe walls (also visible (a) Impact Puncture (b) Membrane stress
in Figures 9b & 9c). (c) Shear Stress
47
J.Griffiths and P.Gudimetla
48
Puncture Resistance of Pressurized & Unpressurized Glass Reinforced Epoxy (GRE) Pipes – A
Phenomenological Model of Failure Mechanisms
55 and 73%. Hence, most commercially available Buarque, E.N., & d’Almeida, J.R.M. (2006), The
excavators have the potential to penetrate GRE pipes, effect of cylindrical defects on the tensile strength
so alternative methods must be employed to reduce of glass fiber/vinyl-ester matrix reinforced
this risk (refer to procedures within AS2885). composite pipes. Composite Structures, Vol.79,
On a final note, it is important to remember that pp. 270-279.
although a broad approach to this project has been Christoforou, A.P. & Swanson, S.R. (1988), Strength
taken, this information relates specifically to the type loss in composite cylinder under impact. Trans
and size of pipes tested. Also, because the testing ASME, Vol. 110, 180-184.
was performed on pipe samples with significantly Christoforou, A.P. & Swanson, S.R. (1991), Analysis
different material properties and wall thicknesses of impact response in composite plates.
than the minimum manufacturing requirements, these International Journal of Solids and Structures,
tests may not give a true representation of the Vol. 27, 161-170.
puncture resistance of similar grades of pipe in the
Composites Institute. (1992), Fiberglass Pipe
field. It should therefore be accepted that these
Handbook. New York: Fiberglass Pipe Institute.
puncture loads represent the upper bounds and as
such design for external interference risks shall be Cosham, A., & Hopkins, P. (2001), A new industry
more conservative. document detailing best practices in pipeline
defect assessment. Fifth International Onshore
Pipeline Conference. Amsterdam.
9 REFERENCES
Curtis, J., Hinton, M.J., Li, S., Reid, S.R., & Soden,
AS2885. (2007), Pipelines – Gas and Liquid P.D. (2000). Damage, deformation and residual
Petroleum. Standards Australia. burst strength of filament-wound composite tubes
AS4041. (2006), Pressure Piping. Standards subjected to impact or quasi-static indentation.
Australia. Composites: Part B, Vol. 31, 419-433.
Alexander, C. & Connelly, L. (2001), Analytical Davies, G.O.A., & Zhang, X. (1994), Impact damage
recreation of a dent profile considering varied prediction in carbon composite structure.
soil, operating, and boundary conditions. Energy International Journal of Impact Engineering, Vol.
Sources Technology Conference and Exhibition. 16, pp. 149-170.
ASME. Driver, R., & Zimmerman, T. (1998), A limit states
Bolt, R. (2001), Database reflects recent trends. Oil approach to the design of pipelines for mechanical
and Gas Journal, pp. 48-54. damage. 17th International Conference on
Brooker, D. (2002), Resistance of Pipelines to Offshore Mechanical and Arctic Engineering.
External Interference. Centre for Oil and Gas Hopkins, P., Corder, I. & Corbin, P. (1992), The
Engineering, University of Western Australia. resistance of gas transmission pipelines to
Brooker, D. (2003), Numerical modelling of pipeline mechanical damage. CANMET International
puncture under excavator loading. Part I. Conference on Pipeline Reliability. Calgary.
Development and validation of a finite element Jones, D., Kramer, G., Gideon, D. & Eiber, R.
material failure model for puncture simulation. (1986), An analysis of reportable incidents for
International Journal of Pressure Vessels and natural gas transmission and gathering lines 1970
Piping, Vol. 80, pp. 715-726. through June 1984. Technical report 158:
Brooker, D. (2003), Numerical modelling of pipeline Battelle Columbus Division.
puncture under excavator loading. Part II. Kirk-Burnnand, D. (June 2006). GRP High Pressure
Parametric study. International Journal of Pipelines for Australia. Perth: APIA WA Chapter
Pressure Vessels and Piping, Vol. 80, pp.727- Seminar.
736. Li, S., Reid, S.R., Soden, P.D., & Hinton, M.J.
Brooker, D. (2005), Experimental puncture loads for (2005), Modelling transverse cracking damage in
external interference of pipelines by excavator thin, filament-wound tubes subjected to lateral
equipment. International Journal of Pressure indentation followed by internal pressure.
Vessels and Piping. Vol.82, pp. 825-832. International Journal of Mechanical Science,
Vol.47, pp. 621-646.
49
J.Griffiths and P.Gudimetla
50