Engineering Failure Analysis 17 (2010) 701711

Contents lists available at ScienceDirect

Engineering Failure Analysis

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

Characterization of the fracture behavior of NBR and FKM grade elastomers

for oileld applications
Z. Major a,c,*, R.W. Lang b,c
a

Institute for Polymer Product Engineering, Johannes Kepler University Linz, Austria
Institute for Polymeric Materials and Testing, Johannes Kepler University Linz, A, Former Institute of Materials Science and Testing of Plastics,
University of Leoben, Leoben, Austria
c
Polymer Competence Center Leoben GmbH, Leoben, Austria
b

a r t i c l e

i n f o

Article history:
Received 5 January 2009
Received in revised form 6 August 2009
Accepted 17 August 2009
Available online 21 August 2009
Keywords:
Rapid gas decompression failure
Crack initiation and propagation
Fracture toughness of rubber
Tearing energy
Continuous and discontinuous crack growth

a b s t r a c t
The monotonic fracture behaviour of two elastomer compounds (HNBR and FKM) was
characterized in this study. These elastomeric materials are frequently used in many oileld applications (e.g., seals and hoses) and are exposed to a complex combination of
mechanical, thermal and environmental loads. Furthermore, rapid gas decompression failure of these elastomeric materials was observed under real service conditions. A phenomenon termed rapid gas decompression (RGD) damage occurs when elastomer seals exposed
to high gas pressure fail in a brittle manner upon the sudden release of the gas pressure.
Hence, to support both material development and design efforts, it is of prime theoretical
and practical importance to characterize the fracture behaviour of these materials. Fracture
mechanics tests using a faint waist pure shear specimen (FWPS) conguration were performed over a wide loading rate range. First, global force and displacement values were
measured and single parameter fracture mechanics values in terms of peak tearing energy,
Tp values were calculated. Furthermore, the crack growth process and the local crack tip
deformation were characterized by non-contact optical devices. The digital image correlation technique used allows for the determination of displacement and full-eld strain up to
a high strain and crack growth rate. Based on above measurements, crack growth resistance curves in terms of tearing energy (T  Dc) and crack tip opening angle (CTOA  Dc)
functions were derived. While a continuous crack growth was observed for HNBR at all
loading rates, a discontinuous crack extension was observed for FKM. Finally a phenomenological model was deduced for describing the rapid monotonic crack extension process
for these elastomeric compounds.
2009 Elsevier Ltd. All rights reserved.

1. Introduction and objectives

A phenomenon termed rapid gas decompression (RGD) damage occurs if elastomer seals exposed to high gas pressure fail
upon the sudden release of the gas pressure in a brittle manner. The rapid gas decompression failure of elastomer seals is an
important issue for the oil exploration industry and since the mid of the seventies is the objective of an intensive theoretical
and practical research. The process is relative complex and includes the gas permeation, solution in the bulk rubber, the
extension of the cavitations in the rubber matrix and the occurrence of crack initiation and crack growth under non-isother-

* Corresponding author. Address: Institute for Polymer Product Engineering, Johannes Kepler University Linz, Polymer Competence Center Leoben GmbH,
Altenberger STr 69, Linz, Austria. Tel.: +43 732 2468 1651.
E-mail addresses: zoltan.major@jku.at (Z. Major), reinhold.lang@jku.at (R.W. Lang).
1350-6307/$ - see front matter 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engfailanal.2009.08.004

702

Z. Major, R.W. Lang / Engineering Failure Analysis 17 (2010) 701711

mal and non-isobar test conditions. The overall RGD process is described by Briscoe et al. [1], Derham and Thompson [2],
Stewenson and Morgan [3] and Seregely and Schunck [4] for elastomers and by Lorge et al. [5] for a thermoplastic polymer
with many different aspects. More specically, the gas permeability and diffusivity problems are described and discussed by
Klopfer et al. and Filus et al., [6,7] and the cavitation process analyzed along with a simple solution among others by Chao
and Gent [8]. Recently, a novel model was proposed by Volokh [9] and simulations were performed on cavities in hydrostatic
tension applying the softening hyperelasticity theory and a coupled approach for modeling thermo-mechanical loading in
gaseous environment was developed by Rambert and Grandidier [10]. Furthermore, a novel instrumented test was recently
developed by the authors group and several elastomer grades were characterized regarding to their rapid gas decompression
failure behavior [11,12]. Moreover, it was early recognized that fracture mechanics concepts and methods can provide an
effective tool for characterizing the rapid gas decompression failure process as described by Briscoe and Savvas [1], Derham
and Thompson [2], Lakes [13], and Medri and Strozzi [14]. In spite of these signicant efforts, hardly any accurate description
of the fast crack growth process was found and no material parameters in terms of fracture toughness (Gc, Jc or CTOD) are
available for both material development and for design efforts. Hence, the objective of this paper is to characterize the crack
initiation and crack growth behaviour of two selected RGD relevant elastomer grades over a wide deformation rate range
using pure shear fracture specimen and to derive appropriate fracture toughness parameters.

2. Experimental
2.1. Materials and test specimens
Two elastomer types, a hydrogenated nitrile rubber (HNBR) and a uorelastomer type (FKM) were selected for these
investigations. The compounds were produced with a nominal hardness of 85 5 SHA by the company partner (SKF Economos GmbH, Judenburg, Austria) and provided for the investigations as cylindrical specimens with a nominal diameter of
10 mm and height of 25 mm and as faint waist pure shear (FWPS) test specimens. Furthermore, the nominal stressstrain
diagrams of both materials both for an unloaded case and after pressurization are shown in Fig. 1a and b. Moreover, scanning
electron microscopy images of both materials are shown in Fig. 1c and d. It is clearly seen that in spite of the exactly same
nominal hardness values, both the microstructure and the macroscopic deformation behavior of these compounds are signicantly different. More controversially, both materials revealed a very similar rapid gas decompression failure performance in the rst experiments [11,12].

2.2. Test methods and test set-ups

In the rst part of the study pressurization/depressurization experiments were performed in a high pressure autoclave (SITEC, Sieber Engineering AG, Zrich, Switzerland) using cylindrical specimens with 10 mm nominal diameter. In these experiments the pressure and temperature change of the chamber and the volume expansion of the test specimens were
continuously measured and recorded. The specimen expansion was measured by a CCD camera, the images were stored
and the expansion was determined using a software tool developed for a video extensometer. The various stages of deformation during the pressurization/depressurization phase are seen in Fig. 2. In the pressurization phase a moderate increase
of material volume was observed up to the saturation (compare Fig. 2a and b). After the pressure release a sudden increase of
the volume up to very high volumetric strain was observed in a short time (compare Fig. 2b and c). Finally, an instability of
the material and crack appearance on the specimen surface was recognized (see Fig. 2d). More detailed information about
the experimental set-up, data recording and reduction procedure and the material performance can be fund in [11,12].
In the second part of the study fracture tests were run on a high rate servohydraulic test system (MTS 831.59 Polymer Test
System, Berlin, Germany). A single edge notched pure shear specimen conguration with a faint waist in the mid-section
(SEN-FWPS specimen) with a nominal width, W of 200 mm and with a nominal thickness, B of 2 mm was used in this study.
The FWPS specimen conguration along with the test set-up used are shown in Fig. 3. The force was measured by the load
cell of the testing machine and the global displacement was measured by the LVDT integrated in the actuator. For more details to the specimen geometry and the test procedure refers to Feichter [15]. Single parameter critical tearing energy values,
Tc. were calculated based on the loaddisplacement or nominal stressstrain curves of FWPS specimens as follows.

R smax

T Wh0

Fds
0
BW  a

where W is the strain energy density, B is the specimen thickness and (W  a) is the remaining ligament length. The critical
tearing energy, Tc was dened as the peak load tearing energy, Tp and it was determined at the peak load value. The loading
rate was dened as the velocity of the actuator and a nominal strain rate (far eld) was calculated using this loading rate
divided by the nominal gage length of the specimen (h0 = 16 mm).
In addition to the conventional global forcedisplacement measurement, optical, non-contact devices (Aramis, GOM,
Braunschweig, Germany) and an IR-thermography testing system (CEDIP Infrared Systems, Paris, France) were also applied
for characterizing the crack tip and the ligament deformation behaviour. The complete test system (testing machine, Aramis

703

Z. Major, R.W. Lang / Engineering Failure Analysis 17 (2010) 701711

40

H-ECORUBBER - HNBR
T=23 C

nominal stress , N/mm2

nominal stress , N/mm2

O-ring Tensile Test

30

virgin O-ring

20

pressurized O-rings

10

40

O-ringTensile Test
ECORUBBER 2 - FKM
T=23 C
30

20

virgin O-ring

10

pressurized O-ring
0

0
0

50

100

150

200

250

300

350

nominal strain , %

50

100

150

200

250

300

350

nominal strain , %

Fig. 1. Structureproperty relationship for the two elastomer grades investigated; Nominal stressstrain curves of virgin (unloaded) and pressurized/
depressurized O-rings, (a) HNBR and (b) FKM; SEM images with a magnication of 5000; (c) HNBR and (d) FKM.

and thermo-camera system) is shown in Fig. 4. The Aramis system uses the digital image correlation technique and both the
local deformations can be measured and 2 or 3-dimensional full-eld strain analysis can also be performed. While an excellent overview of the image correlation technique is given by Sutton and co-workers [16] the details of the procedure used in
our experiments are described by Feichter et al. [17] and by Jerabek et al. [18]. As the digital image correlation system applied here has a maximal image acquisition rate of 8000 frames/s with reduced resolution (512  256) and about 500
frames/s with full (1024  1024) resolution, the rapid crack propagation can be followed and recorded with sufcient quality
for further analysis.
Furthermore, multiple parameter crack growth resistance curves (T  Dc) for various elastomer types under monotonic
test conditions were determined over a wide loading rate range. The crack length, c was measured by the Aramis system
from the edge of the specimen to the crack tip in the crack plane and the actual crack growth rate, dc/dt was also calculated.
Due to the continuous measurement of the crack length the actual cross section can be calculated. Hence, the tearing energy
was calculated based on true (actual cross-section) stress values. Moreover, the crack tip opening angle; CTOA was dened at
1 mm distance from the actual crack tip to the specimen edge between two secant lines on the crack contour. The results of
the fracture experiments are also analysed based on the crack growth kinetics (dc/dt  c curves) and crack tip opening angle
vs. crack extension (CTOA  Dc) curves.
Finally, to gain more insight into the real local material behavior under complex loading conditions full-eld strain and
temperature measurements were performed both in the near crack tip and in the far eld in the specimen ligament and the
strain distribution is depicted in Fig. 5a. Although, the full-eld strain analysis did not always work perfectly on the contour
edge of the crack, the maximal crack tip strain, ect
max was realistically estimated. In addition to the tearing energy derived from
global forcedisplacement data, the knowledge of local strain makes the determination of local strain or local strain energy
based fracture parameters possible [17].

704

Z. Major, R.W. Lang / Engineering Failure Analysis 17 (2010) 701711

Fig. 2. Volume change of round bar test (RBT) specimen and the appearance of the crack in autoclave; (a) original state, (b) compression phase (saturated),
(c) decompression phase (maximal volume strain) and (d) crack appearance.

Fig. 3. Specimen conguration and test set-up for bulk elastomer fatigue tests; (a) test system, (b) xture and (c) FWPS test specimen with initial notch.

The temperature and temperature distribution across the specimen was measured by the thermo camera. The crack tip
temperature of FWPS specimen is shown at the moment of maximum load and at about 1 s1 nominal strain rate in Fig. 5b
for HNBR. As expected, high strain and temperature gradient was observed in the vicinity of the crack tip. These strain and
temperature gradients depend on the crack length and reveal also pronounced loading rate dependence. In addition to the
gradient a higher temperature crack layer was also observed. Further details to this phenomenon and about the possible
characterization methods can be found at Chudnovsky and Moet [19] and at Persson and Brener [20] and will also be described in a future work by the authors.
3. Results and discussion
The change of the test parameters (pressure, p and temperature, T) and the volumetric deformation, DV behaviour of a
cylindrical test specimen is shown in Fig. 6 during an instrumented RGD test. While the overall process is shown in
Fig. 6a, the pressurization process is shown in Fig. 6b and the depressurization in Fig. 6c in an enlarged time scale. In the
pressurization phase a moderate increase of the material volume was observed up to the saturation (up to about 40% volume

Z. Major, R.W. Lang / Engineering Failure Analysis 17 (2010) 701711

705

Fig. 4. Test set-up with full-eld strain and temperature analysis systems.

Fig. 5. Strain and temperature distribution of a FWPS specimen at the crack tip and in the ligament; (a) the vertical strain, y component and (b) the
temperature prole is depicted.

expansion) until 250300 min. After the pressure release a sudden increase of the specimen volume up to very high volumetric strain was observed in a short time (from 40% to 500% in 30 s, compare Figs. 1c and 6c). Furthermore, the due to
the crack initiation the volume increase was stop and a sudden decrease of the volume was detected (compare Figs. 1d

706

Z. Major, R.W. Lang / Engineering Failure Analysis 17 (2010) 701711

600

300

250

500

200

400

150

300

100

50

200

100

volume expansion V, %

pressure p bar
temperature T, C

0
0

200

400

600

800

1000

1200

1400

time t, min

90

compression

80
70

250

60
200

150

50
40
30

100

20
10

50

350

600

300

volume expansion V, %

FKM, 100C, 260bar, CO2

500

explosion

250
200

FKM, 100C, 260bar, CO2

150

decompression

50

100

150

time t, min

200

250

300
200

100
100
50

0
-10
300

400

volume expansion V, %

300

pressure p bar
temperature T, C

100

350

pressure p bar
temperature T, C

0
0
1359,0

1359,5

1360,0

1360,5

1361,0

time t, min

Fig. 6. The change of the test parameters and the size of the round bar test specimen during an instrumented RGD test; (a) the complete test sequence, (b)
pressurization phase and (c) decompression phase.

and 6c after 1360 min). Moreover, the chamber temperature was not constant but rst decreased with the pressure and than
increased after the crack initiation. This is associated with the adiabatic heating during the fast deformation process.
Based on these diagrams and images depicted above (see Fig. 1) the entire process can be analyzed in detail. Due to the
rapid pressure decrease the gas solved in rubber expands and tries to move from the bulk to the free surface of the rubber.
During this process a very high degree of volumetric deformation of the elastomer was observed and simultaneously
decreasing chamber temperature was recorded. The possible locations of the crack initiation are; (i) rubber matrix/ller
interface and (ii) the space between particles where the matrix is highly constrained. A detailed comparison of RGD behaviour of the elastomers along with scanning electron microscopy images of the fracture surfaces after RGD test is provided by
Schwarz et al. [12].
As the autoclave experiments are very time consuming and expensive, efforts were made to characterize the elastomers
by fracture mechanics methods on the laboratory specimen scale.
3.1. Determination of single fracture parameters using loaddisplacement data (global measurements)
Global loaddisplacement curves at two testing rates are shown in Fig. 7a for the HNBR compound. First, single parameter
fracture toughness values in terms of tearing energy were determined at the peak load, Tp. Both compounds revealed distinct
rate dependence, with increasing loading rate continuously increasing peak tearing energy values, Tp were observed for both
elastomer grades (see Fig. 7b) with moderate date scatter (about 15%). Furthermore, as it is seen in Fig. 7b no signicant difference between HNBR and FKM was obtained. This is may be due to the fact that at about 100% strain similar stress values
were observed in the nominal stressstrain diagrams for both compounds (compare Fig. 1a and b). Hence, further experiments were performed and local parameters related to the crack tip deformation and crack growth were derived and discussed below.

707

Z. Major, R.W. Lang / Engineering Failure Analysis 17 (2010) 701711

3.2. Determination of single fracture parameters using local measurements

The nominal strain rate dependence of the maximal strain at the vicinity of the crack tip, ect
max is shown in Fig. 8a. A material dependent decrease of the ect
max values was observed with increasing loading rate and FKM revealed signicantly higher
strain rate dependence. Contrary to the ect
max values a signicant increase of the maximum crack tip temperature, Tmax was
observed with increasing loading rate and with similar manner for both elastomers as it is seen in Fig. 8b. In good agreement
with the nearly same stiffness and the similar hysteretic behaviour, the crack tip temperatures are in the same range for both
elastomers with somewhat lower values for FKM.

3.3. Determination of multiple fracture parameters using local data

In addition to the crack tip strain and temperature distribution and to the peak strain and peak temperature values at the
peak load, the crack growth was also measured. Crack resistance (R curves) in terms of tearing energy and crack extension,
T  Dc were constructed and plotted in Fig. 9a and b for both materials. To gain more information about the crack growth
kinetics and to compare with single point values the locations of peak load values, Tp are plotted in the diagrams. As expected, highly non-linear R curves were observed for both materials for all loading rates. However, some minor differences
in terms of the non-linear slope were recognized. While HNBR revealed a continuous increase of the slope, a small but signicant rate dependent decrease of the slope was observed for FKM. Furthermore, the location of Tp values uniformly increases with the crack length (from 10 to 30 mm) for increasing loading rate for HNBR. In contrary, Tp values were

2500

HNBR

100 mm/s

2000

load F, N

peak load tearing energy TP, N/mm

1500

10 mm/s

1000

500

60

50

HNBR
FKM

40

30

20

10

0
-10

10

20

10-1

30

100

101

102

103

104

loading rate v, mms-1

displacement s, mm

Fig. 7. Single parameter fracture tests; (a) loaddisplacement curves and (b) loading rate dependence of peak load tearing energy, Tp for HNBR and FKM.

200

b
ct ,

max. crack tip temperature T p C

max. strain at crack tip max, %

150

100

50

FKM
HNBR
0
10 -2

10 -1

10 0

10 1

nominal strain rate d/dt, s-1

10 2

10 3

50

45

40

35

30

25

20
10-2

FKM
HNBR
10-1

100

101

nominal strain rate d/dt, s

102
-1

Fig. 8. Nominal strain rate dependence of (a) the maximal strain at the crack tip, max ct and (b) the max crack tip temperature, Tp. ct.

103

708

Z. Major, R.W. Lang / Engineering Failure Analysis 17 (2010) 701711

observed at signicantly lower crack length values (from 5 to 15 mm) for FKM. This difference is assumed to be associated
with the different crack tip deformation and crack growth initiation process for the two elastomer grades.
Moreover, the kinetics of the crack growth was also analysed and the crack growth rate, dc/dt was calculated based on the
video sequences. The crack length and loading rate dependence of the crack growth rate is shown in Fig. 10a for HNBR and in
Fig. 10b for FKM. While a constant increase of the crack growth rate with increasing crack length with similar slope at all
loading rates was observed for HNBR, signicant changes are seen in the same curves for FKM. It is assumed that the crack
growth is continuous in the pure shear specimen for HNBR and contrary to this, it is discontinuous for FKM with crack deceleration and acceleration phases. Latter phenomenon (crack jumping) is associated with the crack tip blunting and sharpening process. Considering the real RGD loading situation where the gas expansion induces a stress controlled loading with
high loading rates, this could be a disadvantageous property for the FKM grade.
Furthermore, the crack tip blunting was also analyzed during the crack growth. A crack tip opening angle value, CTOA was
dened and CTOA values were determined for all test conditions. The loading rate dependence of CTOA values is shown in
Fig. 11a for HNBR and in Fig. 11b for FKM. While the CTOA value remains nearly constant during the crack growth process
and reveals only small rate dependence for HNBR, signicant and with increasing loading rate increasing changes of the CTOA
values were observed for FKM (see Fig. 11b). This higher strain rate dependence of FKM is in good agreement with the observation shown in Fig. 8a.
The nominal strain rate dependence of the CTOAmax values at the maximal crack length are plotted in Fig. 12 for both elastomer grades. While the CTOAmax values revealed a small decrease with increasing strain rate for HNBR (about 20%) a signicant decrease was observed for FKM (more than 100%). Furthermore, the rate dependence of CTOAmax is opposite as it
was observed for peak load tearing energy, Tp values (compare Fig. 12 with Fig. 7b). It was assumed, however, that this tendency corresponds to and describe more realistic the crack initiation process observed in the autoclave experiments.

60

50

40

30

20

HNBR, v= 1 mm/s
HNBR, v= 10 mm/s
HNBR, v= 100 mm/s
HNBR, v= 500 mm/s

10

peak load
Tearing Energy

0
0

10

20

30

40

tearing energy T, N/mm

tearing energy T, N/mm

50

60

40
30
20
FKM, v= 1 mm/s
FKM, v= 10 mm/s
FKM, v= 100 mm/s
FKM, v= 500 mm/s

10
peak load
Tearing Energy

0
0

50

10

crack extension c, mm

20

30

40

50

crack extension c, mm

Fig. 9. Loading rate dependence of crack growth resistance curves and the location of peak load tearing energy, Tp values for; (a) HNBR and (b) FKM.

10 6

HNBR (v= 1 mm/s)

HNBR (v= 10 mm/s)
HNBR (v= 100 mm/s)
HNBR (v= 500 mm/s)

10 5

crack growth rate c/dt, mm/s

crack growth rate dc/dt, mm/s

10 4
10 3
10 2
10 1
10 0

106
FKM
FKM
FKM
FKM

105

(v=
(v=
(v=
(v=

1 mm/s)
10 mm/s)
100 mm/s)
500 mm/s)

104
103
102
101
100

10 -1
0.1

10

crack length c, mm

100

1000

0.1

10

crack length c, mm

Fig. 10. Crack length and loading rate dependence of crack growth rate, dc/dt for; (a) HNBR and (b) FKM.

100

1000

709

Z. Major, R.W. Lang / Engineering Failure Analysis 17 (2010) 701711

140
120

b
crack tip opening angle CTOA,

HNBR
100
80
60
40
v= 1 mm/s
v= 10 mm/s
v= 100 mm/s
v= 500 mm/s

20
0
0

10

20

30

40

140

FKM

increasing loading rate

120
100

CTOAmax

a
crack tip opening angle CTOA,

These crack tip deformation processes are schematically summarized in Fig. 13 for both elastomers investigated. The
crack tip blunting is uniform during the entire crack growth process and remains the same over the loading rate range inves-

80
60
40
20

v=
v=
v=
v=

50

10

20

30

40

1 mm/s
10 mm/s
100 mm/s
500 mm/s

50

crack length c, mm

crack length c, mm

crack tip opening angle CTOAmax,

Fig. 11. Loading rate dependence of the crack tip opening angle, CTOA for; (a) HNBR and (b) FKM.

140
120
100
80
60
40
20

FKM
HNBR

0
0.1

10

100

nominal strain rate d/dt, s-1

Fig. 12. Nominal strain rate dependence of the crack tip opening angle, CTOA at maximal load for both elastomer grades (FKM and HNBR) investigated.

Fig. 13. Schematic representation of the crack tip deformation process with increasing crack length for both elastomer grades (FKM and HNBR)
investigated.

710

Z. Major, R.W. Lang / Engineering Failure Analysis 17 (2010) 701711

tigated for HNBR elastomer. This results in a continuous crack growth and in a uniform crack growth rate prole. In contrary
to this, the degree of the crack tip blunting changes during the crack growth process for FKM grade elastomer. The variation
of blunted and sharpened crack tip results in continuous and discontinuous crack growth for FKM. Finally, the kinetics of this
continuous and discontinuous process highly depends on the global loading rate of the specimen.

4. Summary and conclusions

A comprehensive characterization of the rapid gas decompression behavior of these two elastomer grades was performed
using both standardized tests and using the novel method by our group [11] and by the researchers of the company partner
[12]. It was found in both test series that in spite of the different microstructure (polymer type, cross-link density, ller type
and size, processing) and macroscopic deformation behavior (stressstrain curves) and the different volume change in both
the pressurization (5% HNBR and 15% FKM) and depressurization phase (100% HNBR and 300% FKM) hardly any difference
was observed between the two materials. Microscopic observations revealed, however, some differences regarding the crack
initiation site and the internal crack surface arisen during the crack growth process. Furthermore, it was stated in [12] that
no clear relationship between the basic mechanical properties and the RGD failure behavior was found. It was also speculated, that the quality of the polymer network and the processing quality (dispersion of the ller particles) may signicantly
affect the RGD behavior.
Due to the high volume deformation induced by the expansion of cavitations inside the materials, crack initiation and
crack growth during the RGD process may occur in the specimen. These cracks frequently initiate at the vicinity existing
imperfections (e.g., particle agglomerate, a single large particle or void). It was proposed by many authors [15] that this
process can adequately be characterized by fracture mechanics concepts and methods. To support material development efforts by comparison of materials or by the denition a target toughness value in terms of a relevant fracture parameter, reliable fracture mechanics parameters should be determined. Furthermore, the appropriate modeling and simulation of the
RGD process necessitates the determination and use of relevant material models along with proper material parameters.
While novel extended hyperelastic material models [7] may successfully be used for modeling the cavitation and expansion
process, computational fracture mechanics tools and models are needed to characterize the crack growth process. The prerequisite of the applicability of these models is the existence of adequate fracture mechanics parameters (single geometry
independent fracture toughness values) or parameter functions (i.e., cohesive zone model, crack resistance curve approach).
Hence, single and multiple fracture toughness values were determined in this paper and the ndings of these experiments
are summarized below:
To characterize the deformation and failure behavior of elastomers under complex loading conditions, experiments
were performed both on component level under near service conditions and on laboratory test specimen level over a
wide loading rate range using two model elastomer grades selected (HNBR and FKM). Crack initiation and rapid crack
propagation was observed during the instrumented autoclave tests in the cylindrical specimens during the rapid gas
decompression phase. It was assumed that the crack growths from the bulk (from an imperfection located in the inside)
to the surface. Furthermore, fracture experiments on laboratory specimen level applying FWPS specimens have also been
carried out for both elastomer grades. First, conventional fracture analysis was conducted and peak tearing energy values
determined based on global loaddisplacement data. Moreover, to gain more insight into the material response, novel
non-contact, full-eld strain and temperature analysis methods and techniques were also applied. The kinetics of the
loading process and the material response was measured, recorded and analyzed in terms of directly measured quantities
as local crack tip strain, crack length, crack tip opening angle and crack tip temperature. Multiple fracture mechanics
parameters were derived in terms of crack resistance curves, T  Dc and crack growth rate, dc/dt  c curves.
Material grade and loading rate dependent distinct crack tip blunting was observed in the FWPS test specimens. A uniform blunting process goes hand in hand with a continuous crack growth and in contrary to this, for a non-uniform blunting
process a discontinuous crack growth was observed. While an increase of global peak tearing energy, Tp values with increasing loading rate was observed, the local deformation based fracture parameters, maximal crack tip strain, ect
max and the crack
tip opening angle, CTOA were found to clearly decrease with increasing loading rates for both elastomer grades investigated.
It is expected, however, that due to the particularly high loading rates in the RGD experiments, the material resistance
against crack initiation and crack growth signicantly decreases. The tearing energy values based on global forcedisplacement curves do not reect this reduced material resistance against crack growth.
Moreover, high material temperatures were measured in the specimens during the fracture process and with increasing loading rate increasing maximum crack tip temperatures were observed. It is speculated that the same material temperature increase would be observed in the cylindrical specimen in the extension phase. On the other hand a decreasing
temperature was measured in the pressure chamber during the decompression. It is assumed that the resultant of these
two opposite tendencies is a highly non-uniform temperature prole in the bulk material during the decompression
process.
As overall conclusion it is stated that simple fracture experiments on laboratory specimen scale does not sufciently reect the very complex thermo-mechanical loading situation in rapid gas decompression and hence, can not substitute the
component tests in autoclave. However, these laboratory fracture experiments provide useful information both about the
fracture behavior as well as provide input data for further modeling and simulation work.

Z. Major, R.W. Lang / Engineering Failure Analysis 17 (2010) 701711

711

Acknowledgement
The research work for this paper was performed at the Polymer Competence Center Leoben GmbH (PCCL, Austria) within
the framework of the Kplus-programs of the Austrian Ministry of Trafc, Innovation and Technology with the contributions of
the University of Leoben as scientic partner and the SKF Economos (Judenburg, Austria) as company partner. The PCCL is
founded be the Austrian Government and the State Governments of Styria and Upper Austria. Special appreciation goes
to Mr. Klaus Lederer for performing the majority of these tests and data reduction in his M.Sc. Thesis and for efciently supporting us during the project work. The additional technical support from Dr. T. Schwarz, M. Moitzi and M. Mitterhuber (SKF
Economos) is also gratefully acknowledged.
References
[1] Briscoe BJ, Savvas T, Kelly CT. Explosive decompression failure of rubber. Rubber Chem Technol 1994;67:384416.
[2] Derham C.J., Thompson B. Mechanisms of explosive (rapid gas) decompression and related phenomena in elastomers (rubbers) and thermoplastics, 4th
oileld engineering with polymers, MERL, 34 November, London; 2003. p. 728.
[3] Stewenson A, Morgan G. Rubber Chem Technol 1995;68:197211.
[4] Seregely ZI, Schunck P. Plast Rubber Compos Process Appl 1994;22:917.
[5] Lorge O, Briscoe BJ, Dang P. Polymer 1999;40:298191.
[6] Flaconnche B, Klopfer MH, Taravel Condat C. 5th oileld engineering with polymers, MERL, 2930 March, London; 2006. Paper 7.
[7] Filus Z et al. Polym Test 2007;26:60613.
[8] Cho K, Gent AN. J Mat Sci 1988;23:1414.
[9] Volokh KY. Int J Solids Struct 2007;44:504355.
[10] Rambert G, Grandidier JC. Eur J Mech A/Solids 2005;24:15168.
[11] Major Z et al. Development of a test and failure analysis methodology for elastomeric seals exposed to explosive decompression, 5th oileld
engineering with polymers, MERL, 2930 March, London; 2006. Paper 10.
[12] Schwarz T, Moitz M, Mitterhuber M. Failure mechanisms of elastomeric sealing materials exposed to explosive decompression, 15th international
sealing conference, Stuttgart, 78 October; 2008.
[13] Lake GJ. Fracture mechanics and its application to failure of rubber articles. Rubber Chem Technol 2003;76:56791.
[14] Medri G, Strozzi A. Rubber Chem Technol 1987;59:70921.
[15] Feichter C. Ph.D. Thesis, University of Leoben; 2006.
[16] Sutton MA, Orteu JJ, Schreier HW. Image correlation for shape, motion and deformation measurements. Springer; 2009.
[17] Feichter C, Major Z, Lang RW. Strain 2006;42(4): 299304.
[18] Jerabek M. Ph.D. Thesis, University of Leoben; 2009 [chapter 2].
[19] Chudnovsky A, Moet A. A theory for crack layer propagation in rubber. J Elastom Plast 1986;18(1): 5055.
[20] Persson BJH, Brener EA. Phys Rev 2005;E 71: 036123.