7/31/2019 A Fracture Mechanics Approach for the Prediction of the Failure
1/12
A fracture mechanics approach for the prediction of the failure
time of polybutene pipes
L. Andena a,*, M. Rink a, R. Frassine a, R. Corrieri b
a Dipartimento di Chimica, Materiali e Ingegneria Chimica G. Natta, Politecnico di Milano, Piazza Leonardo da Vinci 32, 20133 Milano, Italyb Basell Poliolefine Italia, PT&C ARC, G. Natta R&D, P.le P.to Donegani 12, 44100 Ferrara, Italy
a r t i c l e i n f o
Article history:
Received 31 October 2008
Received in revised form 11 May 2009
Accepted 13 October 2009
Available online 17 October 2009
Keywords:
Polybutene
Fracture mechanics
Timetemperature superposition
Pipes
a b s t r a c t
In this work two grades of Isotactic polybutene-1 with a different degree of isotacticity
have been investigated; fracture tests have been performed at various temperatures and
testing speeds on DCB and SENB samples. Optical methods have been used to record crack
advancement.
Results of the tests have been interpreted using the fracture mechanics framework; a
timetemperature superposition scheme has been adopted to describe crack propagation
behaviour over several decades of time-scale. An analytical model has been applied to pre-
dict the lifetime of pressurised pipes from experimental fracture data. There is good agree-
ment between model predictions and experimental data obtained from full-scale tests on
real pipes.
2009 Elsevier Ltd. All rights reserved.
1. Introduction
There are several areas in which isotactic polybutene-1 (i-PB1) finds application thanks to its good thermal and mechan-
ical properties: the packaging industry, hot-melt adhesives, tanks for various domestic appliances. In Europe and Asia i-PB1
also became during the past years one of the preferred materials to be used for the manufacturing of hot and cold water
plumbing and heating piping systems. i-PB1 offers many advantages in terms of easy, fast installation with a reduced num-
ber of joints and connectors compared to much stiffer conventional plumbing materials (such as metals). i-PB1shares with
more traditional polyolefins good resistance to chemicals and environmental stress cracking in addition to its excellent creep
properties even at high temperatures.
In the literature there are many works concerning i-PB1s crystallization behaviour (e.g. [1]) and the subsequent transi-
tion which occurs between its two crystalline forms (I and II) [2,3]. Fewer works involve its mechanical properties, with
widely different approaches. For example, AFM investigation has been used recently to study crazing at the micrometricand nanometric scales [4]. Cohesive zone modelling (CZM), a phenomenological approach which proved to be a powerful
method to describe fracture of adhesives and tough polymers [5,6], has been adopted to describe mode I fracture of i-PB1
[7] and different methods have been used to identify cohesive zone parameters. It was shown, however, that i-PB1exhibits
a complex fracture behaviour, previously unreported in the literature, with partial instability arising during crack propaga-
tion and this limited the effectiveness of CZM in reproducing crack initiation and propagation. Although yielding of i-PB1 has
not been extensively studiedper se, a better understanding of the damage mechanisms preceding crack initiation could sup-
port the investigation of the fracture behaviour of i-PB1.
0013-7944/$ - see front matter 2009 Elsevier Ltd. All rights reserved.doi:10.1016/j.engfracmech.2009.10.002
* Corresponding author. Tel.: +39 0223993207; fax: +39 0270638173.
E-mail address: [email protected] (L. Andena).
Engineering Fracture Mechanics 76 (2009) 26662677
Contents lists available at ScienceDirect
Engineering Fracture Mechanics
j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / e n g f r a c m e c h
http://dx.doi.org/10.1016/j.engfracmech.2009.10.002mailto:[email protected]://www.sciencedirect.com/science/journal/00137944http://www.elsevier.com/locate/engfracmechhttp://www.elsevier.com/locate/engfracmechhttp://www.sciencedirect.com/science/journal/00137944mailto:[email protected]://dx.doi.org/10.1016/j.engfracmech.2009.10.0027/31/2019 A Fracture Mechanics Approach for the Prediction of the Failure
2/12
A key issue in the use of i-PB1 for pressurised pipe applications is the evaluation of their lifetime as related to creep crack
growth. This phenomenon consists in the initiation and subsequent slow growth of a crack originating from a surface flaw.
The materials resistance to this type of fracture is hard to characterise, as the lifetime of pressurised pipes in typical oper-
ating conditions may exceed 100 years. Therefore it is common practice to perform accelerated tests at high temperature andextrapolate data to predict pipes lifetime [8]. The schematic diagram shown in Fig. 1 illustrates the outcome of a typical full-
scale test on i-PB1 [9]; two distinct regions can usually be recognised. For high hoop stress values (region A), failure occurs
due to ductile yielding of the material when the stress in the pipe wall exceeds the yield stress of the material. The term
ductile failure is used as large deformations can generally be observed when the pipe cross-section yields before fracture;
however, this is not the case for i-PB1 pipes which fail without exhibiting ballooning phenomena, which are quite common
in this regime for other polyolefins. At lower values of the applied hoop stress (region B) creep crack growth occurs and fail-
ures in this region are termed as brittle. This field is more interesting from the application point of view as pipe failures typ-
ically take place under this regime. The main drawback of this kind of test is their long duration (i.e. 12 years) and high cost.
Fracture mechanics (FM) can provide an alternative, useful approach. With FM it is possible to characterise fracture prop-
erties of a given material from laboratory tests and use them to predict the lifetime of any manufactured article. In [10] FM
has been used to study fracture of two grades of i-PB1performing creep tests at high temperature on SENB specimens. The
tests lasted for several weeks, thus granting a significant time saving when compared with full-scale tests on pipes. The
authors also developed an analytical model able to predict pipes lifetime and a promising comparison with the referencecurves shown in [9] was made.
A similar approach has been followed in the present work, performing fracture tests on laboratory specimens. Yet in this
study a constant displacement rate rather than a constant load has been applied. This allows a further, significant reduction
in testing times which in this study ranged between a few seconds and a couple of hours these times are much shorter than
those required by creep tests, not to speak of full-scale tests on pipes. In addition to that, tests have been carried out with
varying speed and temperature in order to ascertain the influence of these variables on the general fracture behaviour ofi-
PB1 and especially on crack stability.
Finally, pipe predictions have been obtained using the model developed in [10] and they have been validated against data
obtained from full-scale tests.
2. Theoretical background
Fracture mechanics data was analysed in terms of the stress intensity factor at the crack tip Kfor any given crack size a. Inthe present work only mode I (opening) conditions were considered.
Several authors, including Williams [11] and Schapery [12], suggested possible approaches to extend linear elastic FM to
viscoelastic materials. Under certain simplifying assumptions, Williams derived the following relationships between the
stress intensity factor K, the initiation time ti and the crack speed _a:
ti B Kp 1_a A Kq 2
in which A, B, p and q are material properties which generally depend on external conditions, such as the temperature.
Following Schapery [13] it is recognised that _a depends on the current value of K but not on its past values and for this
reason Eq. (2) applies for any loading history. It becomes thus possible to determine crack propagation parameters from any
convenient loading history in a laboratory test and use the obtained data to predict the behaviour of any manufactured item.
By combining Eqs. (1) and (2) a prediction of the lifetimetf
of a pipe under constant pressure with wall thicknesss
can be
obtained:
h
Log tf
ductile failure
Log
brittle failureA
B
Fig. 1. Schematic diagram of hoop stress vs. lifetime for a polymeric pressurised pipe.
L. Andena et al. / Engineering Fracture Mechanics 76 (2009) 26662677 2667
http://-/?-http://-/?-7/31/2019 A Fracture Mechanics Approach for the Prediction of the Failure
3/12
tf ti tp B Kp0 Zsa0
da
A Kq 3
where tp represents the time required for a crack of initial size a0 to propagate across the wall thickness after initiation oc-
curs. Before crack initiation, as the crack size remains constant and equal to a0, the stress intensity factor is also constant and
equal to K0.
In the present work Eq. (3) has been used, neglecting the initiation time ti in the evaluation of tf. This leads to a conser-
vative prediction of the pipe lifetime.
3. Experimental details
The materials investigated are two pipe grades of i-PB1 kindly supplied in the form of pellets by Basell Polyolefins. The
two grades will be called PB1 and PB2, with PB2 having a higher degree of isotacticity and consequently crystallinity. Full-
scale testing run by the producer on pipes made from both materials showed that PB2 offers a better resistance to creep
crack growth. The greater degree of crystallinity of PB2 has also been reported to increase elastic modulus and yield stress
[10]. The tensile behaviour of i-PB1 is characterised by the absence of strain localisation and necking.
The pellets were compression moulded into 170 120 10 mm plates. After cooling from the melt, i-PB1 crystallizes in
form II, which is characterised by tetragonal symmetry. This form is unstable at room temperature and spontaneously
evolves into form I, which has an hexagonal lattice. To allow for completion of the transition, specimens were cut and ma-
chined at least 15 days after moulding [10], and then tested.
Fracture experiments under pure mode I conditions have been run on double cantilever beam (DCB) and single edgenotch bending (SENB) samples, shown in Fig. 2. Relevant dimensions are listed in Table 1. SENB configuration was used only
on preliminary tests on PB2, before moving onto DCB which grants a more stable crack propagation. Also, the longer liga-
ment of DCB specimens grants the acquisition of more data and extended fracture surfaces (see Table 1).
Notches in the case of SENB were made by means of razor sliding. The same apparatus could not be used for DCB, due to
the larger dimensions: the samples were first cut using a saw and then a razor blade was pushed into the material. On both
configurations the final root radius of the notches was about 13lm. The use of two different notching techniques can induce
a different degree of damage in the area surrounding the notch tip. This may in turn lead to a different behaviour at crack
initiation; however this is not very important for this study in which only crack propagation has been considered.
2h
Bn
P
P
W
aB
W
P
2h
a
Bn B
Fig. 2. DCB (above) and SENB (below) samples used for the fracture tests.
2668 L. Andena et al. / Engineering Fracture Mechanics 76 (2009) 26662677
7/31/2019 A Fracture Mechanics Approach for the Prediction of the Failure
4/12
Side grooves were also introduced in order to guide crack advancement along the notch plane. Great care was taken dur-
ing specimen preparation in order to ensure proper alignment of notch and grooves.Tests were performed on an Instron 1185R5800 screw-driven electro-mechanical dynamometer fitted with an environ-
mental chamber. Constant crosshead speeds of 1, 10 and 100 mm/min were used for the tests run at temperatures of 23 C,
50 C, 70 C and 90 C For every testing condition (sample geometry, speed, temperature) at least two specimens were tested.
Crack advancement was monitored using a photo-camera (at 1 mm/min) or a video-camera (at 10 and 100 mm/min) with a
calibration gauge applied on the specimens. ImageJ software was used to process the captured images.
4. Evaluation of the stress intensity factors
The stress intensity factor Khas been evaluated for both testing configurations from the measurements of load and crack
length recorded during the tests. In the case of SENB specimens the widely known formula:
K
f
a=W
P
B W0:5
4
has been used, in which f(a/W) is a non-dimensional shape factor [14].
For the DCB configuration the formula:
K 2ffiffiffi3
p P aB h1:5
5
is generally used. However, Eq. (5) works well only for a/h > 70, which is not the case for the samples tested in the present
study. This has been discussed in [15] where an alternative formula by Kanninen is proposed, in which the accuracy of the
simple beam theory is improved using EulerBernoulli beam theory together with a Winkler foundation. The resulting equa-
tion applies for a/h > 2:
K 2ffiffiffi3
p P aB h1:5
1 0:64ha
6
These expressions are derived according to linear elastic fracture mechanics (LEFM). To ensure validity of LEFM, smallscale yielding and plane strain conditions should be fulfilled. This can be guaranteed if the specimens meet appropriate size
criteria: the size of the plastic zone around the crack tip shall be significantly smaller than the specimen dimensions, i.e. the
thickness B, the crack length a and the ligament length (Wa). The characteristic length of the plastic zone, rp, can be esti-
mated from the following equation [15]:
rp KCrY
27
in which KC and rY are the material fracture toughness and yield stress respectively. For both i-PB1 grades rp is approxi-
mately 8 mm [7], a value which is comparable with the specimen thickness. However, the influence of thickness on the frac-
ture properties of PB1 and PB2 has already been investigated on SENB samples in [16] and no effect has been reported in the
range between 5 and 20 mm. An effect of the ligament width has been observed instead, with a decrease of the toughness for
small values of (Wa
); this was already reported by Hashemi and Williams in [17] and it can be explained considering the
constraint that such a small ligament size exerts on the plastic zone which, as a consequence, is not free to fully develop.
Table 1
Nominal dimensions of DCB and SENB samples.
DCB SENB
2 h 45 mm 2 h 80 mm
W 150mm W 20mm
a 4575 mm a 10mm
B 10 mm B 10mm
Bn 6 mm Bn 8 mm
a 60 a 60U 8 mm
Table 2
Parameters of the pipe model.
s 2 mm
R0 22mm
a0 50lm
e 1
L. Andena et al. / Engineering Fracture Mechanics 76 (2009) 26662677 2669
7/31/2019 A Fracture Mechanics Approach for the Prediction of the Failure
5/12
7/31/2019 A Fracture Mechanics Approach for the Prediction of the Failure
6/12
However the presence of frequent and/or large crack jumps limits the usefulness of this approach. Yet even when instabil-
ities occur there are large load drops but K values remain almost constant because a increases (see Fig. 7). Moreover, the
combination of stable and unstable crack propagation gives rise to an average crack speed which can be determined by a
linear fit of crack length vs. time data. The average K and da/dtvalues are reproducible within tests performed in the same
conditions (temperature and speed) and they have been used in the following analysis, as they are believed to truly represent
the materials behaviour. However, this approach has the obvious drawback of generating only a single data point for each
test.
Figs. 8 and 9 show Kvs. da/dtdata at the various temperatures for PB1 and PB2 respectively. According to Eq. (2), Kvs. da/
dtcurves are expected to be straight lines on a bilogarithmic scale. It is hard to detect a single slope of the data for all tem-
peratures. Moreover, data at 50 C do not fall on straight line.
Fig. 4. Fracture surfaces of PB1 and PB2 samples tested at 1, 10 and 100 mm/min and 23 C.
Fig. 5. Fracture surfaces of PB1 and PB2 samples tested at 1, 10 and 100 mm/min and 50 C.
L. Andena et al. / Engineering Fracture Mechanics 76 (2009) 26662677 2671
http://-/?-http://-/?-7/31/2019 A Fracture Mechanics Approach for the Prediction of the Failure
7/12
Data at different temperatures can be interpreted using timetemperature superposition, a reduction scheme which is
widely accepted in the literature concerning polymers. An example of its application to Kvs. da/dtfracture data can be found
in [18,19]. Basically one temperature is selected as a reference temperature; data points belonging to another temperature
are shifted along the crack speed axis until they superimpose with the reference curve. The process is repeated for the next
temperature and so on, until all data merge on a single master curve at the reference factor. The time shift factor a23C
T T
required for each temperature is usually reported on an Arrhenius plot as a function on the reciprocal of temperature and
the slope of this plot can be related to the activation energy of the mechanical process involved.
This scheme was applied to PB1 and PB2 data in Figs. 8 and 9 and a Kvs. da/dtmaster curve at 23 C was obtained for both
materials, as shown in Fig. 10. The shift factor a23C
T T was found to be the same: this quantity seems to be independent of
the materials crystallinity. A similar result was found in [10] for the shift factor related to relaxation modulus and yield
stress. Values of a23C
T T are reported on an Arrhenius plot in Fig. 11, where a linear dependence on the reciprocal of tem-
perature can be observed. In Fig. 10 a knee is clearly visible between crack speeds of about 103 to 102 mm/s, indicating a
transition between two regions with a different slope in the Kvs. da/dtcurve. Data analysis reveals that the two regions are
Fig. 6. Fracture surfaces of PB1 and PB2 samples tested at 1, 10 and 100 mm/min and 70 C.
0 600 1200 1800 2400 3000 3600 4200 4800 54000.0
0.5
1.0
1.5
2.0
2.5
3.0
avg da/dt
K
Crack length
Time (s)
K(MPa*
m1/2)
avg K
40
50
60
70
80
90
100
110
120
130
140PB2 23C 1mm/min
Crack
length(mm)
Fig. 7. Typical K and crack length vs. time curves for two DCB samples having different initial crack lengths of 45 and 75 mm; dashed lines indicate the
average values of K and da/dt determined during the analysis.
2672 L. Andena et al. / Engineering Fracture Mechanics 76 (2009) 26662677
7/31/2019 A Fracture Mechanics Approach for the Prediction of the Failure
8/12
-5 -4 -3 -2 -1 0 1
-0.05
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
PB1
logK(MPam
)
log da/dt (mm/s)
stable
crack
propagation
partially
unstable
crack
propagation
PB2
Fig. 10. K vs. da/dt master curves for PB1 and PB2 at 23 C. Dashed lines represent the slopes for the stable and partially unstable regimes.
-5 -4 -3 -2 -1 0 1-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
DCB23CDCB50CDCB70C
PB1
logK(M
Pam
)
log da/dt (mm/s)
Fig. 8. K vs. da/dt data at 1, 10 and 100 mm/min for PB1.
-5 -4 -3 -2 -1 0 1-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40PB2
DCB23CDCB50CDCB70CSENB50CSENB70CSENB90C
logK(MPam
)
log da/dt (mm/s)
Fig. 9. K vs. da/dt data at 1, 10 and 100 mm/min for PB2.
L. Andena et al. / Engineering Fracture Mechanics 76 (2009) 26662677 2673
7/31/2019 A Fracture Mechanics Approach for the Prediction of the Failure
9/12
characterised by a distinct behaviour: for K values below the knee stable crack propagation is observed while the values
above are associated to partially unstable propagation. In the latter region the crack speed is much more sensitive to the
applied K, i.e. an increase of K will cause an increase of da/dtwhich is larger than in the case of purely stable propagation.It is found then that the simple power-law model detailed in Section 2 can be used to describe propagation data for a given
mechanism but the transition between the two mechanisms needs to be correctly taken into account according to the effec-
tive K range for a given application.
Direct comparison of the two master curves clearly shows that crack propagation on PB1 is faster than PB2 for any level of
applied K. This result is in good agreement with results of full-scale testing on pipes, as mentioned earlier at the beginning of
Section 3. The present fracture mechanics approach can be used to rank fracture resistance of different materials with an
enormously reduced effort.
6. Prediction of pipes lifetime
Eq. (3) can be used as the basis of a simple analytical model able to predict lifetime of polymeric pipes; in the present
study the contribution of initiation time to the total failure time was neglected, as previously done in [10]. This is only a first
approximation giving conservative predictions that were compared against experimental data obtained from full-scale tests
on pipes.
In order to apply the model one needs to properly define the geometry of both the pipe and the initial defect, and use a
suitable shape factor; relevant material parameters (namely A and q) need to be known as well.
A pipe geometry analogous to that sketched in Fig. 12 was considered: a semi-circular flaw was assumed to be situated at
the pipe inner surface, lying on a radial plane. Physical dimensions were chosen according to the actual dimensions of the
pipes used by Basell for full-scale tests on i-PB; they are listed in Table 2. Experimental observations indicate that fracture
always initiates at the inner surface and quality controls performed prior to pipe testing excluded the presence of defects
larger than 50 lm: the location and size of the initial defect were chosen accordingly.
When internal pressure is applied to a pipe pure mode I conditions are generated at the crack tip of a radial defect. The
latter was assumed to propagate keeping a semi-circular shape: in this way K could be calculated as a function of the hoop
stress and the defect size by using the same shape factor (taken from [20]) throughout all the analysis.
Pipe lifetimes were evaluated for different levels of applied hoop stress (up to 20 MPa) by integrating Eq. (3) for a crack
growing from the initial defect size a0 to the wall thickness s.
The choice of which material parameters to use is not straightforward, since both materials exhibit a transition in the Kvs.
da/dtmaster curve. However, even for the highest level of applied stress (20 MPa) Kvalues ranged from 100.66 (0.2 MPa m,
for aa0) up to 100.48 ((3.0 MPa m, when as). By looking at Fig. 10 it is obvious that in these conditions K values lie in the
stable propagation region for most of the pipe life. Therefore, predictions were made considering A and q as obtained from
the stable part of each of the two master curves at 23 C, thus extrapolating the stable behaviour to the whole K range.
Experimental data from full-scale tests on pipes performed by Basell were available at 23, 70 and 95 C and the same tem-
peratures were considered in the model. The shift factor shown in Fig. 11 was used to obtain the K vs. da/dtcurve at 70 C;
the curve at 95 C was generated by applying a shift factor obtained by linear extrapolation.
A comparison between model predictions and full-scale experiments is shown in Fig. 13. The FM model predictions are
shown as the lines on the right and they should represent the region of brittle failure for the two materials. However, in the
time-scale considered most of the experimental pipe failures were reported as being ductile. Actually a fair estimate of the
failure time in this region can be obtained by simply considering time to yield data for each stress level and temperature.
Time to yield was calculated from yield stress vs. time curves reported in [10].
Fig. 11. Shift factor of K vs. da/dt curves as a function of temperature for both materials studied.
2674 L. Andena et al. / Engineering Fracture Mechanics 76 (2009) 26662677
http://-/?-http://-/?-http://-/?-http://-/?-7/31/2019 A Fracture Mechanics Approach for the Prediction of the Failure
10/12
Only a few pipes corresponding to data points at the highest failure times (greater than 10 4 h) presented a failure mech-
anism which was reported as mixed, indicating a transition towards the brittle region; the experimental curves show a
hint of a knee for these data. This is where the curves describing the brittle region are expected to intersect those for ductile
Fig. 12. Cracked pipe model considered for the prediction of pipe lifetimes.
1
10
100
101
102
103
104
105
106
107
108
109
experiment PB2
experiment PB1
95C
70C
23C
model PB2
model PB1
time (h)
stress(MPa)
Fig. 13. Comparison between model predictions and experimental data from full-scale tests on pipes at 23, 70 and 95 C.
L. Andena et al. / Engineering Fracture Mechanics 76 (2009) 26662677 2675
7/31/2019 A Fracture Mechanics Approach for the Prediction of the Failure
11/12
failure and indeed this is what happens for model predictions at 23 and 70 C, with remarkably good agreement. At 95 C
predicted curves seem to slightly overestimate lifetimes for both materials, nevertheless the overall agreement is substan-
tially good and shows that this approach can be used to obtain reliable estimates of pipe lifetimes. Moreover, the model cor-
rectly reproduces the different behaviour of PB1 and PB2, with curves of the latter slightly above those of the less resistant
material as it is observed on full-scale tests.
It may be surprising that such a good agreement has been obtained despite neglecting initiation times: this should lead to
conservative estimates of pipe lifetimes. However the model predictions strongly depend on the initial flaw size. A different
value of this parameter would cause an horizontal shift of the predicted curves, as discussed in [10].
This research is continuing with the aim of studying crack initiation for these materials and future developments of the
model will take it into account as well.
7. Conclusions
Fracture properties of two pipe grades of polybutene have been studied performing experiments on DCB and SENB con-
figurations. The existence of two mechanisms of stable and partially unstable crack propagation has been observed, as pre-
viously reported in [7]. The effect of testing speed and temperature on crack stability has been investigated and a transition
from stable to partially unstable crack propagation has been detected on both materials. It has been found that higher testing
speeds and lower temperatures promote the occurrence of instabilities.
The combined effect of testing speed and temperature fits well into a timetemperature superposition scheme and crack
propagation master curves could be obtained for both materials. The two curves are characterised by a bilinear trend with aknee separating the two regions of stable and partially unstable crack propagation.
The analysis performed using the fracture mechanics approach gave two main results:
1. Direct comparison of the crack propagation master curves can in most cases (unless they intersect) give a ranking of dif-
ferent materials with respect to their creep crack growth resistance. In the present case the more crystalline grades curve
lies above the other materials, thus indicating a slower crack speed for any value of the applied stress.
2. Quantitative predictions of manufactured items lifetime may be obtained by using crack propagation data in
conjunction with simple models based on FM and the different performance of the materials investigated has been
evaluated.
The analysis and the predictions agree well with experimental data obtained by the materials supplier from full-scale test
on pipes. Fracture mechanics therefore can be used to perform accelerated testing and proves itself to be a quick, inexpensive
and reliable method to evaluate the long-term performance of different materials.
Acknowledgements
The authors wish to thanking Evaristo Odinolfi for his precious support in performing DCB tests and analysing the data
and Mr. Oscar Bressan for the specimen preparation.
References
[1] Yamashita M, Kato M. Lamellar crystal thickness transition of melt-crystallized isotactic polybutene-1 observed by small-angle X-ray scattering. J Appl
Crystall 2007;40:s6505.
[2] Chatterjee AM. Butene polymers. In: Encyclopaedia of polymer science and engineering; 1985. p. 590.
[3] Azzurri F, Flores A, Alfonso GC, Balt Calleja FJ. Polymorphism of isotactic poly(1-butene) as revealed by microindentation hardness. 1. Kinetics of the
transformation. Macromolecules 2002;35:9069.[4] Thomas C, Ferreiro V, Coulon G, Seguela R. In situ AFM investigation of crazing in polybutene spherulites under tensile drawing. Polymer
2007;48:60418.
[5] Bianchi S, Corigliano A, Frassine R, Rink M. Modelling of interlaminar fracture processes in composites using interface elements. Compos Sci Technol
2006;66:25563.
[6] Andena L. Rink M. Fracture of rubber-toughened poly(methyl methacrylate): measurement and study of cohesive zone parameters. In: Proceedings of
ICF XI Turin; 2005.
[7] Andena L, Rink M, Williams JG. Cohesive zone modelling of fracture in polybutene. Engng Fract Mech 2006;73:247685.
[8] Plastics piping and ducting systems determination of the long-term hydrostatic strength of thermoplastics materials in pipe form by extrapolation.
ISO9080; 2003 (E).
[9] Polybutene (PB) pipes effect of time and temperature on the expected strength. ISO12230; 1996 (E).
[10] Passoni P. Frassine R. Pavan A. Small scale accelerated tests to evaluate the creep crack growth resistance of polybutene pipes under internal pressure.
In: Proceedings of plastics pipes XII Milan; 2004.
[11] Williams JG. The use of fracture mechanics in design with polymers. Plasticon. Engineering design with plastics: principles and practice, vol.
81. University of Warwick; 1981.
[12] Schapery RA. A theory of crack initiation and growth in viscoelastic media. I. Theoretical development. Int J Fract 1975;11:14159.
[13] Schapery RA. A theory of crack initiation and growth in viscoelastic media. III. Analysis of continuous growth. Int J Fract 1975;11:54962.
[14] Plastics determination of fracture toughness (GIC and KIC) linear elastic fracture mechanics (LEFM) approach. ISO13586; 2000 (E).[15] Stam G. The stress intensity factor for grooved DCB specimens loaded by splitting forces. Int J Fract 1995;76(44):34154.
2676 L. Andena et al. / Engineering Fracture Mechanics 76 (2009) 26662677
7/31/2019 A Fracture Mechanics Approach for the Prediction of the Failure
12/12
[16] Andena L. Frassine R. Rink M. Roncelli M. Thickness effect on fracture behaviour of polybutene. In: Proceedings of 4th ESIS TC4 conference Les
Diablerets; 2005.
[17] Hashemi S, Williams JG. Size and loading mode effects in fracture toughness testing of polymers. J Mater Sci 1984;19:374659.
[18] Frassine R, Rink M, Leggio A, Pavan A. Experimental analysis of viscoelastic criteria for crack initiation and growth in polymers. Int J Fract
1996;81:5575.
[19] Viscoelasticity of rubber-toughened poly(methyl methacrylate). Part II: fracture behavior. Polym Engng Sci 1996;36(22):275864.
[20] Murakami Y. Stress intensity factors handbook, vol. 2. Oxford: Pergamon Press; 1987.
L. Andena et al. / Engineering Fracture Mechanics 76 (2009) 26662677 2677