International Journal of Fracture 79: 85-95, 1996.

85
(~) 1996 Kluwer Academic Publishers. Printed in the Netherlands.

Rapid crack propagation in plastic water pipes: measurement of

dynamic fracture resistance

C.J. GREENSHIELDS and RS. LEEVERS

Department of Mechanical Engineering, Imperial College of Science, Technology and Medicine, London, SW7
2BX, UK

Received 30 June 1995; accepted in revised form 7 May 1996

Abstract. A classical analysis of Irwin and Corten relates the minimum pressure to sustain crack propagation
along a pipeline to the dimensions, elastic modulus and fracture resistance of the pipe wall. Although the model is
inapplicable to gas pressurisation, it is here modified to become sufficiently accurate for water pressurised thick-
walled pipe to provide the basis of a dynamic fracture resistance test. Results for three commonly used water pipe
grade polymers agree well with the predictions of the 'Thermal Decohesion' model, and with other experimental
data where available. Rapid crack propagation behaviour, in both test and service pipeline configurations, is
dominated by the ratio of crack speed to fluid decompression wave speed. This ratio can be adjusted in a controlled
test environment.

1. Introduction

In 1984, the first rapid crack propagation (RCP) testing of water pressurised plastic pipe was
carried out by British Gas, using its full-scale test facility. RCP could only be sustained in
medium density polyethylene (PE-80) pipes which were partially filled with air [1 ]. Subsequent
testing of PE-80 pipe at Imperial College used a variant of the $4 small scale pipe test, which,
in its original air-pressurised form, had already shown satisfactory correlation with full-scale
tests [2, 3]. Results from water/air pressurised $4 tests also reflected their full-scale equivalents
[4]. In both cases, running cracks arrested promptly in 100 percent water-filled pipe at pressures
far exceeding the 'critical' pressure predicted from the Irwin-Corten analysis (see Section 2)
using available fracture resistance data. This analysis equates the dynamic fracture resistance
GD of the pipe material to the strain energy content of the pipe wall, but fails to account for
the energy released by the pressurising medium itself.
It was later found that RCP could be sustained in water filled pipe specimens by introducing
an axial internal notch along the bore [4]. Consequently, it was suggested that the stored strain
energy in the pressurised pipe wall was insufficient to overcome the enhanced resistance to
failure of a thin ductile zone along the pipe bore, and that the inclusion of a small amount of
air provided the extra driving energy required for RCP to occur.
Once crack propagation had been successfully achieved, the 100 percent water $4 test was
proposed as one means of measuring the dynamic fracture resistance GD of a pipe material,
using a modified Irwin-Corten analysis. Measurement of Go at high crack speed is otherwise
a difficult task involving exotic experimental methods [5] and dynamic analysis methods [6].
The $4 test, however, provides a straightforward method for sustaining steady RCP, whilst the
Irwin-Corten analysis leads to a simple expression for GD in closed form.
The aim of the research reported here was to use the $4 water test to measure the dynamic
fracture resistance of two more pipe grade plastics, namely unplasticised PVC (PVC-U) and
86 C.J. Greenshields and P.S. Leevers
a 'third generation' modified high density polyethylene (PE-100). During the course of the
work, however, observations were made which seemed to contradict the original conclusions
about RCP in PE-80 water pipe: the presence of the 'ductile layer' is only indirectly linked
to the resistance of 100 percent water pressurised pipe to RCP. It is now apparent that the
behaviour of the pressurising fluid during fracture is significant in the understanding of this
problem.

2. Fracture mechanics theory

The Irwin-Corten model for RCP [7] assumes that all, but only, the strain energy stored in
the pipe wall drives the crack; no energy is retained during fracture to be released as kinetic
energy in the wake of the crack. This may be valid until, at high crack speeds, the effect of
inertia becomes increasingly significant, as has been shown [8] using a finite strip dynamic
steady-state solution. The Irwin-Corten analysis is inappropriate for gas pressurisation, since
it does not account for the very high energy content within the pressurising medium itself.
The strain energy stored in water, however, is small, due to its relatively high bulk modulus
M. Neglecting the strain energy in the fluid, the Irwin-Corten crack driving force is

2 -
G = rrP°D(D* [ 1]
2) 2 1 - (1)
8E

where P0 is the initial internal pipe pressure, E is the dynamic tensile modulus of the pipe
material, D and t are the external diameter and thickness of the pipe and D* = D/t. Since
D* determines the ratio of mean circumferential stress to pressure

1, (2)

and is quoted (as the 'Standard Dimensional Ratio', SDR) for pipe classification.

2.1. STRAINENERGYCONTENTOF WATER

If all of the stored strain energy in the water were recovered during fracture, the crack driving
force would become

G= ~ D ( D * - 2 ) z EI+ (~-~--~-I) D---71, (3)

where

f dc'D* ]2
/3 = 1 - [D(D* - 2)J

is the water-filled fraction of the cross-sectional area of a pipe containing an incompressible

cylindrical core of diameter de. For water-filled pipelines in service or full-scale tests, obvi-
ously/3 = 1, but/3 is defined here to model the $4 test configuration in which such a solid
core locates the pipe over internal baffles and between its end caps (see Section 4).
 Rapid crack propagation in plastic water pipes 87

2.2. IRWIN-CORTEN CRACK DRIVING FORCE FOR 'THICK-WALLED' PIPE

In the Irwin-Corten analysis, the strain energy in the pipe wall is calculated from the solution
for circumferential strain energy, a0 in a thin-walled pipe (2). Many plastic pipelines in service
can be classed as 'thick-walled' (SDR < 20) and a correction to the Irwin-Corten analysis
is required to calculate G accurately. If longitudinal stress is assumed to be zero and radial
stress, C~r and lateral strains are accounted for, the strain energy per unit volume of pipe wall
Us released during fracture is
1 2
Us = ~ (~ro + o2r) - -~aoar, (4)

where v is the dynamic Poisson's ratio of the pipe material. The circumferential and radial
stresses can be assumed to vary with radius r as [9]

ao = _-5-Z-~p0
b2 1+ ~ (Sa)

and

err -- b2 _--~--~P0 1 - ~- (5b)

where r = a at the bore and b at the outer surface.

Hence for thick-walled pipe Us can be calculated from (4) and (5) by summing incremental
strain energies across a section of pipe wall:

Us - 2E (~r° + ar - "

-- E b - a b2~a 2 p2 ( 1 - v ) ( b - a ) + -- --dg-b . (6)

This can be expressed as the ratio, ~bof Us to the circumferential strain energy derived for the
thin-walled solution in (1) in terms of D* as

1 (_~ 2)-- 2[(l_v)+ ( 1I 6 v ) 6D*3-12D*2-I-8D*

-

Using the approximations:

6D .3 - 12D .2 + 8D* ~ 6D*(D* - 1) 2

and

D*
gives the solution:
1
qS=~ [(D.D~-~)(1 +v)+ (D_~. 2)(1-v)]. (7)
88 C.J. Greenshields and P.S. Leevers

Note that these approximations are extremely accurate for D* > 5, which is well within the
range for service pipelines: for D* > 10, the error is less than +0.03 percent. Finally from
(1), i.e. neglecting the strain energy stored in the fluid, the Irwin-Corten crack driving force
for thick-walled pipe becomes

2 ( D * - 1) ]
2
G = ~ - ~ D ( D * - 1)(D* - 2) 1 + E D, 2 (v -- 1 ) ] . (8)

For D* = 11 and v = 0.40, the thick-walled solution gives a value of G which is 10 percent
higher than that derived from the 'conventional' thin-walled solution. Most of this difference
can be attributed to the inclusion of lateral strain in the analysis.

2.3. CIRCUMFERENTIAL STRAIN ENERGY ASSUMPTION

If expansion of the pressurising fluid is ignored, the pipe diameter just behind the crack tip
remains equal to that in the intact section ahead of it. The fractured surfaces separate to
maintain this diameter as the wall contracts circumferentially, preventing the release of some
of the original circumferential strain energy. For a thin-walled pipe, the crack opening, w,
approximates to the increase in pipe circumference during pressurisation:

rrpo D (D* - 2)(D* - l)

w = -~- D*

The circumferential strain energy per unit length retained by fractured pipe with a crack
opening w is

1 E 11)2 __ 71"p2 D2 (D* - 2)

u~ -
67r (D* - 1) 3 D*~*-- 1)"

The overall stored strain energy per unit length in the pressurised section before fracture is

__rcp2D2(D* - 2)2(D * - 1)
Us 8E D .2

Therefore the fraction of the total strain energy retained in the pipe wall during fracture,
U r / U s , is

UF 1
(9)
Us 3 ( D * - 1) 2

Although the solution is strictly only applicable to thin-walled pipe, it must be reasonably
accurate when applied to thick-walled pipe. Since even for a relatively thick walled pipe
(D* = 11) only 0.33 percent of the strain energy is retained, it can be neglected in fracture
resistance calculations.

2.4. CRACK PATH

The Irwin-Corten analysis assumes that the crack extends in a straight, longitudinal direction
along the length of the pipe. It is well known that during RCP in air and air/water pressurised
 Rapid crack propagation in plastic water pipes 89
Table I. Pipe materials

Pipe Material
PVC-U PE-80 PE- 100

Rated Pressure(bar) 12 10 16
Min. Outside Diameter,D (mm) 114 125 125
Min. Wall Thickness, t (mm) 6.0 11.4 11.4
Max. InternalDiameter(mm) 102.0 102.2 102.2
SDR (D/t) 19 11 11
Mean Wall Thickness(mm) 6.15 11.90 I 1.7-12.2"
Actual (Measured)D* 18.5 10.5 10.2-10.6"
*Internally notchedPE-1O0

pipes, the crack follows a sinusoidal path along the length [1]. In the Irwin-Corten analysis, it
is possible to introduce a factor corresponding to the absolute crack length, to account for the
increased fractured area. As it transpires, the crack paths in water pressurised pipe specimens
tested so far have remained almost entirely straight and, therefore, no such scaling factor has
been necessary.

2.5. FLUIDMECHANICSTHEORY
In air-pressurised full scale pipe fracture tests, fluid decompression occurs primarily by escape
through the crack opening. Axial backflow of air from the uncracked region causes substantial
decompression even before the crack front arrives; this effect is suppressed in the $4 test to
accelerate the development of a steady state around the crack tip. During fracture of a 100
percent water pressurised pipe, on the other hand, expansion of the water results in rapid
decompression without significant fluid escape. It seems likely that decompression ahead of
the crack tip could also occur in a water filled pipe, provided that the decompression wave
speed exceeded the speed of the crack. Under these conditions, the strain energy within the
pipe wall would quickly be released and, with no driving force, the crack should arrest.
The analysis of this decompression wave is the direct inverse of the well documented
'waterhammer' effect [10]. The classical theory of waterhammer is believed to be valid when
the valve closure, or in this inverse case the pipe opening or fracture, is instantaneous and
the pipe is sufficiently rigidly anchored [11]. Thus the analysis can be applied directly to the
$4 test with one modification to account for the incompressible core of the rig. It is assumed
that the pipe wall can expand and contract longitudinally over frictionless and cap O-rings,
resulting in zero longitudinal stress. Neglecting any effects of inertia in the pipe wall, the
decompression wave speed cw is approximately

Cw (lO)
I +-~(D -l+2v)

where p is the density of water at atmospheric pressure.

This should provide a good approximation for a limiting minimum crack velocity, below
which RCP cannot, in theory, be sustained.
90 C.J. Greenshields and PS. Leevers

Figure 1. Crack paths in water (top) and air pressurisedPVC-U pipe.

3. Material specifications
The pipe materials used in the tests are described in Table 1. The manufacturer's specifications
for outside diameter were accurate to -4-0.5 percent, although there was greater variation of
wall thickness from the minimum specified value. Wall thicknesses of fractured material were
measured using a travelling microscope and the measured D / t ratio replaced the specified
SDR for fracture resistance calculations. Note that the PE-100 tests used several internally
notched pipe specimens: the crack path width was measured in each specimen and the crack
driving force calculated accordingly.

4. Experimental procedure and preliminary tests

The $4 testing procedure for air/water pressurisation was based upon the air-pressurised $4
method detailed in the draft International Standard [12] with some modifications appropriate
to water pressurisation as described elsewhere [13]. In the preliminary tests the water $4
configuration, in which the baffles are fitted with robber seals in direct contact with the pipe
wall, was used. Special care was taken to eliminate all possible air pockets through a bleed
valve in the test section end-cap.
Results from preliminary tests on PVC-U pipe showed, for the first time, that RCP could
be sustained in un-notched pipe specimens. The observed crack paths were almost entirely
straight, differing from the characteristic sinusoidal path exhibited in air and air/water pres-
surised tests [13]. Crack paths for air and water pressurised PVC-U pipe are compared in
Figure 1. The origin of the observed sinusoidal crack paths in rapid fracture of air pressurised
pipe is still uncertain. Since sinusoidal cracks seem to appear only in air pressurised pipe, it
may be that the cracks follow a weaving path as a result of the flaring of the pipe walls behind
the crack tip - a belief held by some even in the early days of research into RCP in pipes (e.g.
[14]).
Piezo-electric pressure transducers, mounted along the core of the $4 test rig, recorded the
change in pressure as the crack travelled along the pipe. During fracture, the onset of a drop
in measured pressure could be attributed either to:
(a) the crack passing over the transducer, leading to an almost instantaneous radial expansion;
or,
(b) a decompression wave travelling ahead of the crack tip passing across the transducer,
resulting in rapid axial expansion.
A control test was carried out monitoring both fluid pressure (using these transducers) and
crack length (using a gauge of 20 conductive silver timing lines painted at 28 m m intervals
 Rapid crack propagation in plastic water pipes 91

E
E
"~"
r-
.E
(3
700

600
i!}
O
•
Timing Une
Pressure Transducer

o
og
o
o
o •

O) o
o. 500 o
¢0 o

o" 400
o
o

< o
a) :300 0
(3 011
e- 0
0
2o0 0
O
.-~ 100 Crack Velocity,
~t= 540 m/$
• i . . . i . . . i . , . ! . . , t . i | I | , |
0
0 200 400 600 800 1000 1200 1400

Time from Initiation (~ts)

Figure 2. Measuring crack velocityusing 'timing lines' and 'pressure transducer' techniques for a PVC-U pipe
specimen at 3°C.

across the crack path and connected into a Wheatstone bridge, providing a step change in
voltage as each line was broken). The crack position and the point of initial decompression
are plotted against time in Figure 2, clearly showing that decompression begins only as the
crack passes over the transducer. Note that the decompression appears to start approximately
40 #s after the respective crack timing lines break. The transducers have a 3 #s rise time, but
were sampled at only 25 kHz, probably giving rise to the observed delay.
During a series of tests in which crack propagation occurred, crack speeds of 450-600 m s - 1
were measured using the pressure transducers. Equation (10) gives a value for decompression
wave speed of approximately 450 ms -1 for PVC-U pipe, corresponding to the observed
minimum speed at which a crack could propagate through a fully loaded wall.
For air pressurised tests on PE-80 pipe, typical crack speeds of 100-150 ms -1 have been
observed, compared with speeds of 200-300 ms -1 for PVC-U. This difference is significant,
since for 100 percent water pressurised tests on PE-80, it is likely that the decompression
wave speed (again, approximately 450 ms -1) exceeds any attainable crack speed, rendering
RCP impossible. Therefore, in order to sustain RCP and measure the fracture resistance of
unnotched PE-80 pipe, the decompression wave speed had to be reduced to less than any
expected crack speed.

5. Test results using an enlarged central core

The solution, deduced from the waterhammer wave speed in (10), was to reduce the decom-
pression wave speed to approximately 150 ms -1 by removing the baffles and increasing the
diameter of the central core in the $4 test to approximately 96 percent of the internal diameter
of the pipe. Using this method, RCP was indeed sustained in unnotched PE-80 pipe, and the
critical pressure was similar to that found from tests on notched specimens [4]. The crack paths
were almost straight, as seen in the previous tests on PVC-U, and the fracture surfaces were
not noticeably different from those from air and air/water pressurised PE-80 tests: smooth and
brittle with a thin, whitened ductile strip along the pipe bore.
Tests were repeated for PVC-U using the enlarged central core. The critical pressure was
significantly lower with the large core, reducing the calculated dynamic fracture resistance
92 C.J. Greenshields and P.S. Leevers

, . . . . o, . . . .
Od

¢3 O
O
d • O
,o. o O
Oo o
O
u. 4 O O O
r0"3
. . . . . . . . . •
°r-
a
o 3
0
O Propagation I
• Arrest
2 i i ! . . . . i . . . .

0 1. 2 3
Notch Depth (ram)

Figure 3. Response of the crack at various notch depths in solid core tests on 125 mm diameter SDR 11 PE-100
pipe at 3°C (+0 °, -3°).

by a factor of 2.5. Previous results have shown no indication that fracture resistance would
increase by a factor of 2.5 for an increase in crack speed from approximately 200 ms -1 to
450 ms -1 [5, 15]. It is, however, quite possible that at relatively high speeds, the amount of
strain energy retained during fracture as subsequent kinetic energy is significant, causing the
observed increase in critical pressure. This shows that there is a need to try to establish a better
model relating crack speed to the amount of stored strain energy which is able to contribute
to the crack driving force.
An additional advantage of using the large core is that the problem of assessing the
contribution of the fluid strain energy is clearly avoided. Consequently, values for fracture
resistances quoted in Sections 6 and 7 are calculated using (8).
The dynamic tensile moduli used in the calculations were measured using an ultrasonic
technique with a Fast Fourier Transform analysis, as described in [16]. In future, however, by
testing at low pressures to ensure crack arrest, and measuring the time between decompression
at adjacent transducers ahead of the arrest, it may be possible to establish the decompression
wave speed in water. Rearranging (10) and substituting this wave speed should provide an
alternative method for calculating the required dynamic tensile modulus.

5.1. TESTING OF PE- 100

Tests were carried out to measure the fracture resistance of PE-100. For consistency, the test
temperature was again 3°C, above the ductile-brittle transition temperature of between - 15°C
and - 5 ° C for 125 mm diameter PE-100 pipe [17].
To achieve RCP, the pipe specimens were pre-notched. Straight axial notches were cut
along the pipe bore using a knife blade mounted on a vehicle which travelled along an internal
steel rail, supported at each end of the specimen. The consistency of notch depth along the
pipe length was within + 10 percent. After each test the notch depth and pipe wall thickness
were measured at intervals across the fracture surface. The mean crack path width was then
established for each test and substituted for wall thickness t when calculating D*. G was then
computed from D* and P0 for each test. The results from all the tests are plotted as crack
driving force versus notch depth in Figure 3. The variation in pipe wall thickness along the
 Rapid crack propagation in plastic water pipes 93
Table 2. Thermal and physical properties for PVC-U, PE-80 and PE-100

Pipe Material
PVC-U PE-80 PE- 100

Melt Temperature, Tm ( ° C) 231 126 130

Mean Specific Heat, Cp (kJ/kg K) 1.05 2.22 2.07
Heat of Fusion, A H I (kJ/kg) 0 171 212
Weight Average Molecular Mass, 92250 173600 197050
3~t~o (kg/mol)
Polymer Repeat Unit Length (pm) 255 254 254
Chain Contour Length, g~o(#m) 0.38 1.57 1.79
Density (kg/m3) 1350 944 960

length of the pipe was approximately -t-0.2 mm. The overall uncertainty in the crack driving
force is -t-0.2 kJ m -2. The dividing line between crack propagation and arrest regimes defines
the dynamic fracture resistance, By this definition, indicated by the dashed line in Figure 3,
the plane-strain dynamic fracture resistance for PE-100 is 3.45 kJ m -2.

6. The thermal decohesion model

Values of dynamic fracture resistance for these materials obtained using alternative exper-
imental methods appear, at present, to be in short supply. The results are compared to the
minimum dynamic fracture resistance values for PE-80 and PE-100 predicted using the ther-
mal decohesion model for RCP in crystalline polymers [15, 18]. The model proposes that
RCP occurs by self-sustained melting of a layer of polymer material, one molecular chain
length thick, at each cohesive surface of a crack tip craze. The dynamic fracture resistance
GD is equivalent to the total enthalpy of both melt layers plus that of adjacent material heated
by conduction. The heat-sinking effect of the bulk polymer makes the heating process very
wasteful for low crack speeds at which RCP does not occur. As the crack speed increases GD
falls precipitously to a minimum of

a D m i n ~'~ 8wD[5Vp(Tra - T ) + 2AHf]. (11)

GD min is apparently dependent only on the temperature of the pipe material T thermal
properties, molecular chain length and density. Differential scanning calorimetry tests were
carried out on each material tested here to establish thermal properties: specific heat capacity
at constant pressure Cp averaged across the relevant temperature range; melt temperature Tin;
and the latent heat of fusion of the crystalline phase A H I. Since PVC-U is almost completely
amorphous, AHI was neglected and Tm was defined from the empirical approximation
Tg/Tm = 0.7, where Tg is the glass transition temperature, well defined in PVC-U by a
sudden increase in @. The weight average molecular weight 55/w was estimated for each
material using Gel Permeation Chromatography. The mean chain contour length gw was
calculated from the polymer repeat unit length and molecular weight accordingly. Finally, the
density p of a section of each pipe material was measured in accordance with ASTM Standard
D 792-66 [19]. Properties for the thermal decohesion model are given in Table 2.
94 C.J. Greenshields and P.S. Leevers
Table 3. Measured and predicted fracture resistances for PVC-U, PE-80 and PE-100

Pipe Material
PVC-U PE-80 PE- 100

Rated Pressure (bar) 12 10 16

100° Water Critical Pressure (bar) 7.48 12-14"
Large Core Critical Pressure (bar) 4.90 12.77 15-17"
Dynamic Tensile Modulus (GPa) 4.61 2.54 3.22
Dynamic Poisson's Ratio 0.32 0.40 0.38
Experimentally Measured Fracture 0.64 2.28 3.45
Resistance, GD (kJm -2)
Thermal Decohesion Model Min. 0.61 2.54 2.98
Fracture Resistance, GD min(kJm-2)

*Results using notched pipe specimens

Table 3 presents a comparison between values of dynamic fracture resistance measured

using pipe tests and those predicted using the thermal decohesion model. The material prop-
erties used to calculate GD from the water pipe test are given at the top of Table 3; those used
to calculate GDmin from (11) are all given in Table 2.
The experimental and predicted values of fracture resistance, at the bottom of Table 3,
show fairly good agreement in all three materials.

7. Conclusions and practical considerations

Results in Table 3 show a higher resistance to RCP in PE-100 and PE-80 than in PVC-U. The
100 percent water pressurised critical pressure falls well below the rated pressure in PVC-U
and above the rated pressure in PE-80 and PE-100. Note that the critical pressures correspond
directly to the full-scale case due to a full-scale/S4 test conversion factor of unity [13].
Previously, RCP had only occurred in internally notched PE-80 pipe specimens. Internal
notching enables the crack to travel at much higher velocities; the increased crack speeds may
then have exceeded the decompression wave speed, allowing RCP to occur. The presence of
the ductile layer is the most probable cause of the relatively low crack velocities seen in PE-80
pipe tests [11]. The dramatic reduction in RCP resistance seen in air pressurised PE-80 and
PE-100 pipe after the introduction of 0.35-0.50 mm axial internal scores [2] may be as a result
of such a 'crack velocity effect' and is a subject for further research.
It is important to reiterate that RCP was only made possible in PE-80 by the introduction
of the large central core to reduce the decompression wave speed. If material properties and
pipe dimensions are such as to increase the decompression wave speed above any maximum
attainable crack speed, then it should be impossible to sustain long RCP cracks in service water
pipes. Although it is difficult to estimate the maximum crack speed for a particular material,
experimental data from transducer measurements give 600 ms-1 for PVC-U and 300 ms-1
for PE-80 at 3°C just above the critical pressure. Substituting dynamic modulus data into (5)
for a full-scale test (/3 = 1), these wave speed values correspond to a minimum SDR 13 for
PVC-U and SDR 29 for PE-80. Although such high SDRs are rarely seen in PE-80, almost all
PVC-U pipe is 'thin-walled' (at least higher than SDR 13) and as such is capable of sustaining
RCP in 100 percent water pressurised pipe.
 Rapid crack propagation in plastic water pipes 95

Acknowledgements
Chris Greenshields was in receipt of the WRc/IC Postgraduate Training Partnership award.
The authors wish to thank Richard Tyrrell for the work on PVC-U pipe.

References
1. J.M. Greig, Proceedings 7th International Conference Plastic Pipes, Bath, UK (1988) 12.1-12.7.
2. J.M. Greig, ES. Leevers and P. Yayla, Engineering Fracture Mechanics 42 (1992) 4 663-673.
3. P. Yayla and P.S. Leevers, Engineering Fracture Mechanics 42 (1992) 4 675-682.
4. C.J. Greenshields, S.J.K. Ritchie and P.S. Leevers, Proceedings 8th International Conference Deformation,
Yield and Fracture of Polymers, Cambridge, UK (1994) 36/1-36/4.
5. M.A. Wheel and P.S. Leevers, International Journal of Fracture 61 (1993) 349-359.
6. A. Ivankovic, I. Demirdzic, J.G. Williams and ES. Leevers, International Journal of Fracture 66 (1994)
357-371.
7. G.R. Irwin and H.T. Corten, Evaluating the feasibility of basing pipeline operating pressure on in-place
hydrostatic test pressure. Report to Northern Natural Gas Company and E1 Paso Natural Gas Company
(1968).
8. E Nilsson, International Journal of Fracture Mechanics 8:4 (1972) 403-411.
9. S.E Timoshenko and J.N. Goodier, Theory of Elasticity, 3rd edn., McGraw-Hill Book Co., London, UK
(1970) 68-71.
10. B.S. Massey, Mechanics of Fluids, 2nd edn., Van Nostrand Reinhold Co. Ltd., London, UK (1970) 412-427.
1I. A.G.T.J. Heinsbroek and A.S. Tijsseling, Proceedings 2nd International Conference WaterPipeline Systems,
Edinburgh, UK (1994) 17-30.
12. International Standards Organisation, Draft International Standard prEN 13478, Thermoplastics pipes for
the conveyance of fluids - Determination of the resistance to Rapid Crack Propagation (RCP)-Small-scale
steady-state test ($4).
13. C.J. Greenshields and P.S. Leevers, Plastics, Rubbers and Composites Processing and Applications 24 (1995)
7-12.
14. R.W.E. Shannon and A.A. Wells, International Journal of Fracture 10:4 (1974) 471-486.
15. P.S. Leevers, International Journal of Fracture 65 (1995) R53-R57.
16. N.N. Dioh, High strain rate behaviour of polymers at various temperatures, PhD thesis, University of London
(1993) 190-205.
17. R.E. Morgan, Ductile-brittle transitions in pipe grade polyethylene, PhD thesis, University of London (1994)
50-78.
18. P.S. Leevers, International Journal of Fracture 73 (1995) 109-127.
19. ASTM Standard D 792-66, Specific gravity and density of plastics by displacement, Method A-1 (1975).