Engineering Fracture Mechanics 76 (2009) 2780–2787

Contents lists available at ScienceDirect

Engineering Fracture Mechanics

journal homepage: www.elsevier.com/locate/engfracmech

A fracture mechanics concept for the accelerated characterization

of creep crack growth in PE-HD pipe grades
Andreas Frank a,*, Werner Freimann a, Gerald Pinter b, Reinhold W. Lang b
a
Polymer Competence Center Leoben GmbH, Roseggerstrasse 12, A-8700 Leoben, Austria
b
Institute of Materials Science and Testing of Plastics, University of Leoben, Franz-Josef-Strasse 18, 8700 Leoben, Austria

a r t i c l e i n f o a b s t r a c t

Article history: For the lifetime prediction of pressurized polyethylene (PE) pipes based on methods of the
Received 25 September 2008 linear elastic fracture mechanics the knowledge of the crack resistance and the kinetics of
Received in revised form 8 June 2009 creep crack growth (CCG) is essential. In the present work a rather brittle nonpipe material
Accepted 15 June 2009
was used to develop a methodology for an accelerated measurement of crack kinetics in
Available online 24 June 2009
fatigue tests on cracked round bar (CRB) specimens at ambient temperatures of 23 °C. A
material and specimen specific compliance calibration curve was generated to detect the
Keywords:
crack kinetics with only one single CRB test. Based on an already proposed concept the
Polymers
Pipe
kinetics at different R-ratios (minimum/maximum load) was measured and extrapolated
Polyethylene to the case of CCG. To demonstrate the transferability of this concept to pipe materials a
Fatigue PE 80 pipe grade was used. Although the necessary testing time increased considerably
Slow crack growth the concept still has the potential to reduce the overall testing time for new pipe materials
to be certified significantly. With the presented procedure a characterization of CCG in
modern PE pipe grades at room temperature and without the use of stress cracking liquids
is possible within a few months.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction

It is generally accepted, that for long-term applications crack initiation and slow crack growth (SCG) are the critical failure
mechanisms of pressurized polyethylene (PE) pipes. The knowledge of the long-term failure behavior – especially SCG – of
the material is of essential interest for lifetime and safety assessment [1–7]. The standard method for the determination of
the long-term behavior and the estimation of lifetimes is based on internal pressure tests on pipe specimens and the extrap-
olation method by EN ISO 9080:2003 [8]. A number of test methods using fracture mechanics considerations were developed
for an accelerated characterization of the resistance against SCG, like the Notched Pipe Test, the Pennsylvania Notch Test or
the Full Notch Creep Test [6,9]. A decrease of the failure times in these tests was achieved by increasing temperature or by
the influence of stress cracking liquids.
Essential improvements in the raw materials have led to pipe materials with increased crack resistance. Especially the
controlled implementation of short chain branches and a bimodal molecular mass distribution result in materials with a
minimum required strengths (MRS) of 10 MPa and above that are classified in PE 100 grades [10,11]. However, these modern
materials represent new challenges to usual test methods, as their higher resistance against crack initiation and SCG also
extends the testing times into unpractical time frames.

* Corresponding author. Tel.: +43 (0) 3842 42962 26; fax: +43 (0) 3842 42962 6.
E-mail addresses: Andreas.Frank@pccl.at (A. Frank), Werner.Freimann@pccl.at (W. Freimann), Gerald.Pinter@mu-leoben.at (G. Pinter), Reinhold.
Lang@unileoben.ac.at (R.W. Lang).

0013-7944/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engfracmech.2009.06.009
 A. Frank et al. / Engineering Fracture Mechanics 76 (2009) 2780–2787 2781

One possibility to reduce testing times are fatigue tests under cyclic loads and using fracture mechanics concepts for data
evaluation [4,12]. Moreover, the choice of suitable specimen geometries influences the testing time and especially cracked
round bar (CRB) specimens show very promising results even at temperatures of 23 °C, which are very close to the applica-
tion temperatures of real pipes, and without any stress cracking liquids [13,14].
An extrapolation concept how to transfer cyclic fatigue tests into the case of static loading conditions to describe SCG has
already been proposed [15,16]. One essential step in this concept is the determination of the crack kinetics at several R-ratios
(ratio of minimum to maximum loading) [15–18]. Hence, the main objective of this work was the development of a technol-
ogy for a direct measurement of crack kinetics on CRB specimens, which was not previously possible. For this purpose a
rather brittle nonpipe PE was used to evaluate the reliability of the measurement of the crack kinetics based on crack length
dependent material compliance. Afterwards a PE 80 pipe grade material was used to apply this technology and to demon-
strate the transferability of the generated crack kinetics into the proposed extrapolation concept.

2. Background

The long-term failure behavior of pressurized PE pipes is well investigated and can be separated into three characteristic
regions (Fig. 1) [1]. In region A at relative high stress levels the failure is dominated by ductile deformation with large plastic
zones. At lower loadings the failure mechanism passes a transfer knee and change to quasi-brittle failure region B where the
longer failure times of this region can be separated into crack initiation and SCG. The nearly load independent third region C
is reached after very long times and is caused by ageing processes, polymer degradation and stress corrosion cracking [19].
The failure mechanism in region B (crack initiation and SCG) can be described with linear elastic fracture mechanics (LEFM)
concepts and therefore several laboratory tests have been developed. Basic requirements for the applicability of the concepts
of LEFM in plastics materials are that global loadings are within the linear viscoelastic deformation range and the size of plas-
tic zones at the crack tip is only small. The stress distribution in the vicinity of the crack tip is described by the stress intensity
factor (SIF) KI (index I stands for a crack opening load normal to the crack plane) that is a function of the crack length a, the
applied stress r and a geometric factor Y that is known for several specimens and component shapes (Eq. (1)) [20,21].
pffiffiffi
KI ¼ r  aY ð1Þ
The crack growth kinetics da/dt under static loading is plotted in a double logarithmic diagram as a function of the SIF KI
(Fig. 2) [17]. In the region of quasi-brittle crack growth usually a distinct linear correlation is found that can be described
by the power law of Paris and Erdogan (Eq. (2)), where the constants A and m depend on the material as well as on test con-
ditions like temperature. The linear region of the crack kinetics can be used to compare different materials concerning crack
growth resistance. An improved material shifts the kinetics to higher SIF and also a lower slope can be a result of higher crack
resistance. The material parameters A and m are of high importance for a fracture mechanics lifetime prediction.
da
¼ A  Km
I ð2Þ
dt
One possibility to reduce testing times are fatigue tests under cyclic loading, where the loading conditions are specified by
the R-ratio, the ratio of minimum to maximum loading (Fig. 3). Several studies show, that independent of the loading
conditions (cyclic vs. static) the mechanism of slow crack growth is very similar and material rankings based on fatigue tests

Fig. 1. Schematic illustration of the long-term failure behavior of pressurized PE pipes [1].
2782 A. Frank et al. / Engineering Fracture Mechanics 76 (2009) 2780–2787

Fig. 2. Crack growth kinetics under static loads [15].

Fig. 3. Crack tip stress field for static and cyclic loads at various R-ratios [15].

are comparable to those of static tests [4,12,22–24]. In the case of cyclic loading fatigue crack growth (FCG) is usually de-
picted in a double logarithmic diagram, where da/dN is a function of the range of the SIF DKI = KI,max  KI,min. A ratio of
R = 1.0 describes static loading (KI,max = KI,min). By decreasing R testing times decrease too and every single R-ratio will result
in a specific kinetics curve. According to Fig. 3 the approach from cyclic to static tests in this work was done at an equal max-
imum SIF, so that all illustrations of this work are related to KI,max.
Another possibility for a reduction of testing times is the choice of a suitable specimen geometry. The use of cracked round
bar (CRB) specimens provides near plain strain conditions which reduce the formation of the plastic zone size at the crack tip
to a minimum [25–28]. The high constraint in this specimen ensures a relatively quick formation of craze zones and an accel-
erated crack initiation. It was shown that even at ambient temperatures of 23 °C and without any stress cracking liquids a
material ranking with fatigue tests on CRB specimens give the same results as time consuming tests like Full Notch Creep Tests
[29]. Moreover, not only a differentiation of different PE-grades (PE 80, PE 100) can be done, but also a distinction within one
grade up to process related batch variations is possible within a few days. Due to the very simple geometry it is possible to
manufacture the CRB specimens either from moulded plates or even from extruded pipes. Therefore these fatigue tests on
CRB specimens open an interesting potential for product ranking, quality assurance and material development [9,13–16].
For LEFM lifetime prediction the knowledge of creep crack growth (CCG) is important. Based on an already proposed
extrapolation methodology [15,16] the FCG curves at different R-ratios are transformed into a diagram of constant crack
growth rates, where KI,max is a function of R (Fig. 4). After the extrapolation of the crack growth rates with appropriate math-
ematical methods to R = 1.0 (static loading) the generated SIF’s are transformed back into the kinetics diagram to generate a
‘‘synthetic” CCG curve. From this curve the material constants A and m can be determined and furthermore used for a frac-
ture mechanics based lifetime calculation of components like pressurized pipes.
Due to the specimen geometry and the tendency of eccentric crack growth a reliable direct measurement of the crack
kinetics da/dN on a CRB specimen is difficult [18]. Nevertheless a determination of the crack kinetics is possible indirectly
by using specimen compliance [30]. The compliance C is defined as the ratio of the crack opening displacement COD and
the applied load F (Eq. (3)). As displayed in Fig. 5 the specimen compliance also depends on the crack length a, which means,
that with increasing crack length C increases, respectively.
COD
C¼ ð3Þ
F
 A. Frank et al. / Engineering Fracture Mechanics 76 (2009) 2780–2787 2783

Fig. 4. Methodology to generate ‘‘synthetic” creep crack growth curves for static loading (R = 1.0) based on cyclic experiments with CRB specimens [15,16].

Fig. 5. Left: Schematic illustration of crack opening displacement COD. Right: Dependence of compliance C from crack length.

3. Experimental details

All investigations were performed on commercially available PE materials. First the development of the method for direct
measurement of the crack kinetics on CRB specimens was done with a rather brittle blow-moulding type PE-BF at R-ratios of
0.1, 0.3, 0.5 and 0.7. The extrapolation concept for generating CCG was applied to this material. In the following step further
investigations were done with a PE 80 pipe grade.
For specimen preparation compression moulded plates with a thickness of 15 mm were produced. The CRB specimens
with a diameter of 13.8 mm were manufactured out of these plates and the circumferential initial crack with the depth
of 1.5 mm was inserted with a razorblade. All fatigue tests were executed on a servo-hydraulic closed-loop testing system
MTS Table Top (MTS Systems GmbH, GER) with a sinusoidal loading at 23 °C and a frequency of 10 Hz.
For the indirect determination of the crack kinetics a crack length depending compliance calibration curve was developed.
Therefore the COD of CRB specimens with different initial crack lengths from 1.0 to 3.0 mm was measured with three extens-
ometers (Type 632.13–20, MTS Systems GmbH, GER) that were positioned at equal intervals of 120° around the crack. Impor-
tant was to guarantee that during these tests no crack initiation took place and the loads were chosen in the linear
viscoelastic range. Because the loads F in fatigue tests change between a minimum and a maximum value, the crack opening
displacement was measured as the difference DCOD of minimum and maximum COD. Analog the specimen compliance was
calculated to DC = DCOD/DF. Those compliance calibration curves could then be used to calculate crack lengths in the cyclic
CRB tests and to generate crack kinetic curves.

4. Results and discussion

Fig. 6 shows the compliance calibration curve for PE-BF and PE 80 at 23 °C where the correlation of DC with increasing
crack lengths a is displayed. For equal temperature the compliance is only depending on the material and the specimen
geometry. This is proven by the results for material PE-BF where the measured calibration curves for the R-ratios of 0.1
2784 A. Frank et al. / Engineering Fracture Mechanics 76 (2009) 2780–2787

Fig. 6. Compliance calibration curve DC at 23 °C as a function of the crack length a for CRB specimens of PE-BF (R = 0.1 and 0.3) and PE 80 (R = 0.1).

Fig. 7. Extensometer signals DCOD and corresponding specimen compliance DC as a function of cycles N. The first step in the curves indicates crack
initiation.

and 0.3 overlap indicating load independency. The values for DC were recorded after a few thousand cycles, so that starting
effects (e.g. stabilization of extensometer assembling) are eliminated, and before crack initiation. A curve fitting was done
with a simple polynomial function. By converting this material and specimen specific function into a quadratic equation
it was possible, to calculate the crack length in a cyclic CRB test from any corresponding measured DC.
The results in Fig. 6 show that the compliance curve for PE-BF is clearly lower than for PE 80 what is caused by the com-
paratively brittle material behavior of the blow-moulding type. A typical characteristic of modern pipe grade materials is the
improved toughness against crack initiation and crack growth what lead to a higher compliance of the PE 80 material.
Fig. 7 shows the typical signals of the extensometers DCOD and the associated DC values of a single fatigue test. After
starting effects at very low cycles the values of DCOD and DC are constant. The deviation of the single signals for the indi-
vidual extensometers can be assigned to tolerances in specimen preparation and clamping tools that led to diverse crack
opening at the surface. The slow crack growth is initiated anywhere around the circumferential initial crack and is clearly
detectable at the first step in the extensometer signals. After the initiation the crack grows around the entire notch and with
continuing test an increase of DCOD and DC indicates slow crack growth and the very typical stepwise characteristic of qua-
si-brittle crack growth in PE can be noticed.
With continuing crack growth and increasing crack length the SIF at the crack tip raises and the failure mechanism
changes from slow crack growth into the ductile failure mode, which is reflected by the distinctive raise of the signals.
 A. Frank et al. / Engineering Fracture Mechanics 76 (2009) 2780–2787 2785

Fig. 8. Cycles until crack initiation Nini and failure Nf at different R-ratios as a function of the maximum stress intensity factor KI,max.

Fig. 9. Crack kinetics da/dN of PE-BF at R = 0.1 from tests started at different initial stress intensity factors KI,max,ini. Comparison of CRB and CT specimens.

The fact, that with ongoing test an eccentric crack growth appears causes additional bending in the specimen, which causes a
decrease of the signal at 240°. However, the acquisition of the crack kinetic is only permitted within the quasi-brittle failure
mode where this effect of eccentric crack growth can be neglected. At the end of the test the specimen approaches its final
failure and the signals turn into infinite.
The fatigue tests were varied in the maximum loads for each R-ratio, so that the initial maximum SIF differ between 0.60
and 0.80 MPa m0.5. Fig. 8 shows the crack initiation and failure cycle number of PE-BF at different R-ratios as a function of the
maximum SIF KI,max. The crack initiation cycle number Nini shows the same linear correlation as the failure cycle number Nf,
however, at significantly lower testing times. As expected, Nini and Nf increase with rising R-ratios. Several studies prove that
fatigue tests on CRB specimens at R = 0.1 are highly reliable as a method for a quick and easy material ranking [9,13–16]. A
characterization by crack initiation cycles may have the potential for a further reduction of testing times.
Using the compliance calibration curve in Fig. 6 the crack kinetics can be calculated for the individual fatigue tests. Fig. 9
displays the crack kinetics of PE-BF at R = 0.1 from tests started at different initial SIF’s. The results show a very good repro-
ducibility and the characteristic of the crack kinetics is independent of the applied initial load. As a reference the data are
correlated to crack growth kinetics measured with compact tension (CT) specimens of the same material. However, this test
was performed in a previous project [23]. The CT specimens show slightly faster crack growth rates, what may be a result of
marginal changes in the raw material (different batches). Nevertheless, the functional characteristics of the curves confirm
the reliability and the potential of the compliance calibration method with CRB specimens for a fast measurement of crack
kinetics. Whereas the considerable higher constraint in CRB specimen induces a quick crack initiation, the geometry is
responsible for a fast increase in the SIF, what lead to shorter testing times compared to CT specimens.
The crack kinetics for PE-BF at R-ratios of 0.1, 0.3, 0.5 and 0.7 as a function of the maximum SIF KI,max are displayed in the
left chart of Fig. 10. With increased R-ratio the kinetic lines are shifted to higher SIF what denotes the decrease in the crack
2786 A. Frank et al. / Engineering Fracture Mechanics 76 (2009) 2780–2787

Fig. 10. Left: Fatigue crack kinetics da/dt of PE-BF at different R-ratios and extrapolated ‘‘synthetic” creep crack growth curve (R = 1.0). Right: Extrapolation
of R-ratio depending SIF to R = 1.0 (static loading).

Fig. 11. Crack kinetics da/dN of PE 80 at different R-ratios and extrapolated ‘‘synthetic” creep crack growth curve (R = 1.0).

growth rate. Based on the methodology described above (background) for each R-ratio the SIF’s at several constant crack
growth rates were defined and transformed into a diagram, where KI,max is a function of R (Fig. 10, right). For curve fitting
between R = 0.1 to 0.7 different simple calculations like polynomial or logarithmic functions were tested empirical. These
functions were used to extrapolate the constant crack growth rate lines to R = 1.0. Finally the associated SIF’s at R = 1.0 were
transformed back into the kinetics diagram to provide a ‘‘synthetic” CCG curve. The logarithmic function shows the best
agreement with the existing test data and furthermore this function results in the most conservative approach (which means
highest crack growth rate) in the extrapolated crack growth kinetics and is also shown in Fig. 10, left. With this ‘‘synthetic”
CCG curve it is possible to calculate the important material parameters A and m for static loading conditions and continue
with lifetime prediction. In the case of PE-BF this parameters are A = 4.65  105 and m = 7.32.
Fig. 11 shows the measured crack kinetics for the PE 80 pipe grade at the R-ratios of 0.1, 0.3 and 0.5. Data from a previous
project measured with CT specimens of the same material are included as a reference [23]. In contrast to the PE-BF material,
the compliance calibration based kinetics at R = 0.1 shows an excellent correlation to the CT data. However, at R = 0.3 and 0.5
the crack growth rates in the CRB specimens is higher than in CT specimens. Remembering the low constraint and high plas-
tic zones this could be a reason for retarded crack growth in CT specimens. With increasing R-ratio the material shows slower
crack growth rates. The calculation of the ‘‘synthetic” CCG curve was again done by extrapolation of SIF with constant crack
growth rates to R = 1.0 with logarithmic functions. The material parameters were calculated to A = 1.40  106 and m = 6.33.

5. Conclusions

In the present work a new technique for a direct measurement of crack growth kinetics of polyethylene (PE) on
cracked round bar (CRB) specimens in fatigue tests was demonstrated. Fatigue tests under sinusoidal load, a frequency of
10 Hz and at ambient temperatures of 23 °C were performed with two different PE pipe materials at different R-ratios
 A. Frank et al. / Engineering Fracture Mechanics 76 (2009) 2780–2787 2787

(minimum/maximum loading). Three extensometers that were positioned at equal intervals of 120° around the crack were
used to measure the crack opening displacement (COD) at the specimen surface and to develop crack length depending com-
pliance calibration curves. Those compliance calibration curves were then used to calculate crack lengths in the cyclic CRB
tests from COD data and to generate crack kinetic curves. The comparison of the generated kinetics with results form com-
pact tension (CT) tests shows a good comparability.
An extrapolation concept for the estimation of creep crack growth (CCG) was applied for two different PE’s. The kinetics of
fatigue tests were extrapolated to the R-ratio of 1.0 what represents static loading. With this ‘‘synthetic” CCG curves lifetime
prediction based on methods of the linear elastic fracture mechanics (LEFM) is possible. As already confirmed fatigue tests on
CRB specimens are a reliable method for a quick ranking of PE pipe grade materials. The high sensitivity in the detection of
crack initiation by three extensometers opens an interesting option for further acceleration of material comparison.

Acknowledgements

The research work of this paper was performed at the Polymer Competence Center Leoben GmbH (PCCL, Austria) within
the framework of the Kplus-program of the Austrian Ministry of Traffic, Innovation and Technology with contributions by the
University of Leoben, AGRU Kunststofftechnik GmbH (Austria), Borealis Polyolefine GmbH (Austria), OMV Exploration & Pro-
duction GmbH (Austria), Österreichische Vereinigung für das Gas und Wasserfach (Austria) and SABIC Europe (Netherlands).
The PCCL is funded by the Austrian Government and the State Governments of Styria and Upper Austria.

References

[1] Lang RW, Stern A, Doerner G. Die Angewandte Makromolekulare Chemie 1997;247:131.
[2] Gaube E, Gebler H, Müller W, Gondro C. Kunststoffe 1985;75(7):412–5.
[3] Ifwarson M. Kunststoffe 1989;79(6):525–9.
[4] Barker MB, Bowman JA, Bevis M. J Mater Sci 1983;18:1095–118.
[5] Brömstrup H. Essen, Deutschland: Vulkan Verlag; 2004.
[6] Brown N, Lu X. In: 12th Plastic fuel gas pipe symposium, Boston, Massachusetts, USA; 1991.
[7] Brown N, Lu X. In: 13th Plastic fuel gas pipe symposium, San Antonio, Texas, USA; 1993.
[8] Janson LE. Borealis, Axelsson, S. AB/Fäldts Grafiska AB, Stockholm, Schweden; 1999.
[9] Haager M. Institute of materials science and testing of plastics, Austria: University of Leoben; 2006.
[10] Brown N, Lu X, et al. Plast Rubb Compos Process Appl 1992;17(4):255–8.
[11] Pinter G, Lang RW. The application of fracture mechanics to polymers, adhesives and composites. In: Moore DR, editor. vol. 33, Oxford (England):
Elsevier Science Ltd. and ESIS. ESIS Publication; 2004. p. 47–54.
[12] Parsons M, Stepanov EV, Hiltner A, Baer E. J Mater Sci 2000;35:2659–74.
[13] Pinter G, Haager M, Lang RW. ANTEC 2006. Charlotte (North Carolina, USA): Society of Plastics Engineers; 2006.
[14] Pinter G, Haager M, Balika W, Lang RW. Polym Test 2007;26(2):180–8.
[15] Lang RW, Pinter G, Balika W. 3R International. 2005; 44(1–2): 33–41.
[16] Lang RW, Pinter G, Balika W, Haager M. Plastics pipes XIII, Washington DC, USA; 2006.
[17] Pinter G, Lang RW, Haager M. Chem Monthly 2007;138:347–55.
[18] Frank A, Pinter G, Lang RW. ANTEC 2008. Milwaukee (Wisconsin, USA): Society of Plastics Engineers; 2008.
[19] Choi B-H, Chudnovsky A, Paradkar R, Michie W, Zhou Z, Cham P-M. Polym Degrad Stab 2009;94(5):859–67.
[20] Kinloch AJ, Young RJ. Fracture behaviour of polymers. London (England): Applied Science Publ; 1983.
[21] Anderson TL. Fracture mechanics – fundamentals and application. Boca Raton (Florida, USA): CRC Press Inc.; 1991.
[22] Haager M, Zhou W, Pinter G, Chudnovsky A. ANTEC 2005. Boston (Massachusetts, USA): Society of Plastics Engineers; 2005.
[23] Pinter G, Balika W, Lang RW. Temperature–fatigue interaction. In: Remy L, Petit J, editors. vol. 29, Amsterdam: Elsevier Science Ltd. and ESIS. ESIS
Publication; 2002. p. 267–75.
[24] Lang RW, Balika W, Pinter G. The application of fracture mechanics to polymers, adhesives and composites. In: Moore DR, editors. vol. 33, Oxford
(England): Elsevier Science Ltd. and ESIS. ESIS Publication; 2004. p. 83–92.
[25] Duan D-M, Williams JG. J Mater Sci 1998;33:638–52.
[26] Ting SKM, Williams JG, Ivankovic A. Polym Eng Sci 2006;46:763–77.
[27] Ting SKM, Williams JG, Ivankovic A. Polym Eng Sci 2006;46:778–91.
[28] Ting SKM, Williams JG, Ivankovic A. Polym Eng Sci 2006;46:792–8.
[29] Haager M, Pinter G, Lang RW. ANTEC 2006. Charlotte (North Carolina, USA): Society of Plastics Engineers; 2006.
[30] Saxena A, Hudak SJ. Int J Fract 1978;14(5):453–68.