BASF Corporation

PLASTICS PART DESIGN: LOW CYCLE FATIGUE STRENGTH

OF GLASS-FIBER REINFORCED POLYETHYLENE
TEREPHTHALATE (PET)

Abstract The various topics of fracture and fatigue of glass-

fiber reinforced polyamide (PA) and PET based plastics
This paper summarizes our extensive investigation have been discussed in [1-8]. Largest portion of these
on the low cycle (up to Nf = 5x104, where Nf is the number investigations was oriented on fatigue performance of
of cycles to failure) fatigue behavior of short glass-fiber various polyamides (PA 6, PA 66, PA 46, etc.) and sig-
reinforced poly(ethylene, terephthalate), or PET, thermo- nificantly less study were oriented on fatigue of PET [2-3,
plastic1. The modes of fatigue test include tension- 7-8]. The number of published article on low cycle fa-
tension, compression-compression, four-point bending tigue of plastics is very limited. In our previous report to
(flexural) -- all at frequency f = 1-3 Hz, and flexural fa- ANTEC we discussed the influence of time-temperature
tigue at f = 30 Hz (ASTM D-671). All tests were stress- [4], and fiber loading [5] on fatigue performance of short
controlled with stress ratio R = Smin/Smax = 0.1, except for glass-fiber reinforced polyamide 6.
flexural fatigue at f = 30 Hz where stress ratio R = -1. Resistance to low cycle fatigue has some specific [1,
The fracture surfaces of tested specimens were analyzed 8] related to method of testing, test frequency, geometry
using scanning electronic microscopy (SEM). of the specimens, loading methods (tensile, compression,
The results from this investigation provide compre- flexural, torsion, etc.), influence of hysteretic heating,
hensive, up-to-date information and recommendations microstructure, etc. The purposes of this paper is to in-
concerning methods for fatigue testing of injection vestigate low cycle fatigue characteristics of injection-
molded specimens and models, prediction and optimiza- molded PET with the influence of:
tion of low cycle fatigue properties that play a key role in
determining a highly stressed plastic parts life and end- • Loading modes (types) such as four-point bending
use performance, pre-selection of PET based plastic for (flexural) and flexural (ASTM D 671-93), tension-
various industrial applications. tension and compression-compression.
• Geometry and sizes of used specimens and models.
Introduction
Low cycle fatigue resistance of short glass-fiber rein-
Much of the recent growth in fiber-glass reinforced forced PET plastic is analyzed and compared. Important
PET has been found in various industrial applications comprehensive information was provided for plastic pre-
such as automotive, appliances, furniture and so on, selection, optimized design, product development, and
where PET made parts and structures are gradually re- technical support.
placing steels, light alloys, and in some case expensive
plastics thermoplastics and thermosets. One of the critical
Specific of test procedures and generated data
factors in structural design of highly loaded plastic parts
is the aspect of the part’s fatigue endurance (at various The typical flexural fatigue tests for plastics per
numbers Nf of cycles to failure from 102 to 108 and ASTM D 671 are performed using well-known testers
above). Very often the fatigue endurance of thermoplas- (SATEC SF-2U). The standard specimen is a cantilever
tics must be tested and analyzed under various design beam with 3.2 mm in thickness and the width increasing
versions, stress-strain and loads modes, frequency f, time- linearly toward the beam root) is used for the test. The
temperature (t –T), and moisture (Figure 1). beam is subject to a symmetrical (stress ration R =
Smin/Smax = -1, where Smin and Smax are the minimum and
maximum stress levels respectively) cyclic loading under
controlled stress amplitude” and at frequency f = 30 Hz.
1
PET – thermoplastic polyester for injection molding. This test and used equipment are inexpensive and pro-
* Author to whom correspondence should be addressed.
vides a basic data for comparison of different plastic ma- 112) was developed for various applications where in-
terials. With some assumptions, developed fatigue data (S creased strength, stiffness, engineering property and di-
– Nf curves) is used for design of various plastic parts. mensional stability performance is required. Typical me-
chanical properties are shown in Table 1. This data rep-
There are obvious limitations in the flexural fatigue resent basic mechanical properties generated by ISO pro-
data obtained using the ASTM D-671 procedure above, cedures for well-organized glass-fibers orientation in
among them are: multi-purpose specimen only.
• The data is obtained in a single flexural mode with
Test samples were molded into the following three
stress ratio R = -1, whereas the real applications often
types:
encounter fatigue in tension-tension or compression-
1. ISO multi-purpose tensile bars (ISO-3167) with 4
compression mode, or flexural mode with stress ratio
mm thickness that can be easily adapted for low cy-
R other than –1;
cle fatigue tests in three modes: tension, compres-
• The dimensions of thickness of actual plastic compo- sion, and 4-point bending (flexural).
nents, and thickness in particular, are often different 2. The “boss” (hollow cylinder, Figure 2).
from those of test specimens. As a result the differ- 3. Bars with cross-section 6.25 mm x 6.25 mm (Figure
ent molding conditions experienced by the plastic 3).
parts and specimens may invoke different resistance
to cyclic loading due to specifics in orientation and
Mechanical Tests
distribution of reinforcements, skin-core effects, etc.;
• The flexural loading mode in applications, as charac- The focus of our investigation was in the low-cycle
terized by stress ratio R, and mean stress Sm = 0, is fatigue Nf range from (2-3)x102 to (2-5)x104 cycles to
very different from the one in the laboratory flexural failure. The low cycle fatigue tests were conducted using
fatigue tests (ASTM D-671). The real stress ratio R Instron-1361, a servo-hydraulic system. All tests were
is ~ 0 (or above) where in the flexural test R = -1. conducted in stress-controlled mode with a constant Smax,
The mean stress Sm > 0 for tensile-tensile mode and Smin, and stress ratio R. A frequency f of 3 Hz was se-
Sm < 0 for compression-compression mode. lected for most of the low cycle fatigue tests except for
• The cyclic stress frequencies f in real applications are four-point bending where f = 1 Hz was used. The stress
often substantially lower than the 30 Hz from the ratio R = 0.1 was used for all low cycle fatigue tests, ex-
flexural ASTM tests; cept the classical flexural fatigue (ASTM D671), where R
= -1 at a frequency f = 30 Hz, as discussed before. Addi-
• The shape of the cyclic loading function can also be
tionally we conducted short-term tensile (compressive)
very different from the sinusoidal used in the ASTM
and flexural tests for strength data for every type of used
flexural fatigue test.
specimens. Fatigue data (Figures 4-5) and short-term
• The differences in fatigue cracks initiation and
properties (Table 1, Figure 6) were developed at con-
growth under variable loading modes (flexural, ten-
trolled laboratory conditions at 23 °C, air and 50% RH).
sion or compression, and multi-axial).
Specific of Loading Modes, Specimens and Models
Although for some types of materials – metals in par-
ticular - the shape and frequency of the cyclic function The following four cyclic loading modes were used:
may not have significant impact on the low cycle fatigue 1. Tension-tension fatigue, conducted on injection
of the materials, this is not the case for thermoplastics. molded ISO-3167 multi-purpose specimens with a
Due to the intrinsic viscoelastic behavior of thermoplas- thickness of 4 mm.
tics, internal friction can easily cause heating in material 2. Compression-compression fatigue, conduced on three
samples at high frequency. This behavior results in lower types of injection-molded specimens:
fatigue resistance, especially when the often unmonitored • Rectangular specimens modified from the ISO-3167
and uncontrolled temperature goes above Tg, the glass used for tension-tension fatigue. A 20-mm long
transition temperature of the material [1, 4-5, 8]. prism was obtained from the center section where the
two ends were carefully machined to make sure that
they are even and parallel to each other.
Experimental
• The “boss” (hollow cylinder) models were used as in
Material, Test Specimens and Models applications (Figure 2 a, b).
• The same boss specimen with the 3-mm removed
45 wt.% short-glass-fiber reinforced PET thermo-
from the top. The top layer was suspected to have a
plastic colored in black (carbon black) was used for this
molding imperfection (knit line) and fibers re-
study. This injection molding grade (Petra®2 140 BK- orientation that could be responsible for the reduced
2
fatigue strength in many cases (Figures 2 and 7).
Petra® is a registered trademark for BASF Corporation
PET plastic products.
3. Four-point bend (flexural) fatigue test at a frequency We already discussed the anticipated reduction of
f = 1 Hz, performed on 50 mm long specimens cut flexural fatigue strength on a larger specimen. The fre-
from the center section of ISO-3167. The bottom quency f effect, if any, should lead to the opposite direc-
span was 38 mm, and the top span was 12.7 mm. tion. The loading method should not lead to some sub-
stantial differences, because in both cases (the central
Results and Discussion section of the four-point flexural and ASTM test) the max
stresses are constant through comparable areas of the sur-
Figure 4 shows developed comprehensive data (Sa - face. So what we can see from Figure 4 is that compres-
Nf curves) for the stress amplitude Sa vs. cycles to failure sive fatigue strength is higher than both flexural and ten-
Nf for the fatigue tests under various stress modes. In Fig- sile fatigue strength. The Sa – Nf curves are parallel within
ure 5, stress amplitude Sa under each loading condition is the same loading mode, but the slopes of the curves are
normalized with the corresponding static strength Su (for steeper in the flexural and the tension modes, where they
used specimen, Figure 6). On the most basic level, all are parallel again within each mode. At the same time,
these diagrams may be anticipated to be the same. Some Figure 5 shows that the differences within the compres-
static strength of material theories, such as the Tresca- sion mode become statistically negligible when the rela-
Guest (maximum shear stress) criterion or the specific tive (with respect to the compressive static strength) val-
energy of volume distortion are not sensitive to the sign ues are considered (Table 1 and Figure 6). The differ-
(tension or compression) of the main (normal) stresses. ences between the tensile and the four-point flexural
Therefore, the tension and the compression could repre- modes also disappear when looked at in the relative
sent the same level of fatigue tolerance as soon as the terms. These observations confirm the hypothesis that the
maximum shear stress is the same. The same can be ap- major factor for the differences between the modes is the
plied to flexural where the volumes subjected to the same sign of the normal stress component in the max shear
absolute level of maximum tension and compression are stress plane.
present.
What is important is the observed substantial differ-
On a somewhat higher level and considering experi- ence between the flexural modes of our four-point, where
ence of measuring static strength of materials and com- R = 0.1, and the flexural per ASTM D-671, where R = -1.
paring for tension and flexural one can anticipate a higher If only the value of stress ratio R is considered, one would
fatigue strength obtained in flexural test. Indeed, the fol- have expected a quite opposite picture to the one in Fig-
lowing probability considerations are usually used to ex- ures 4-5, namely, the stress amplitudes for R = 0.1 should
plain the phenomenon: have been higher then the ones for R = 0.1. Indeed, the
• The failure is correlated with the largest defect, and maximum stress achieved during the cycle is higher for R
one can assume that the defects are uniformly dis- = 0.1. This controversy can be explained by examining
tributed in the tested volume (the specimen). In flex- the effect of loading frequency f. Apparently, as men-
ural test, only the outer surface adjacent layers (vol- tioned before, the relatively high frequency f = 30Hz used
umes) of material are exposed to the maximum stress. in ASTM D-671 (as opposite to the f = 1 ~ 3 Hz used in
The probability to find the largest defect in these ar- other tests), can easily cause internal heating in the PET
eas is therefore lower then in tension where the due to the viscoelastic nature of the material. The heat
whole volume is under maximum stress. The experi- generated in the cyclic loading process cannot be dissi-
ence of observing higher static flexural strength then pated quickly enough, causing rising temperature and
tensile one is rather consistent on different materials, softening of the material. The specimen, therefore, has
and this theory seems reasonable. been effectively tested at higher temperature when high
frequency is applied. The reduced fatigue strength, as can
• The same probability considerations can be used in be seen in the analysis, is caused more by the induced
explaining differences observed between larger and heat than the frequency itself. The flexural fatigue by
smaller samples (Figures 4-6). Indeed for the same ASTM D-671, therefore, will produce a Sa – Nf curve
reason the strength, especially, fatigue strength of a lower than other types of curves generated at lower fre-
smaller specimen should be higher. quencies f when internal heating is not a problem.
Figures 4-5 show also the classical flexural fatigue
data (ASTM D-671). The difference between the condi- Summary and Conclusions
tions of the two flexural tests is: The results from this investigation will provide com-
• The specimen configuration, sizes (thickness); prehensive, uniform, and up-to-date information and rec-
ommendations for design against fatigue in various
• The loading frequency f; molded parts, pre-selection of short glass-fiber reinforced
• The loading method. PET thermoplastic for design, mechanical performance
prediction and optimization.
 Acknowledgment
Boss
The authors wish to express their appreciation to
John Macur, Roberto Sanchez, for help in preparing this
paper for publishing. A special thanks is going to Cath-
erine Ruiz for constant support of this investigation and
helpful discussions.

References
1. Hertzberg, R. W., Manson, J., A., “Fatigue of Engi- Ribs
neering Plastics”, 295 pages, Academic Press,
(1980).
2. Karger-Kocsis, J. and Friedrich, K., “Fracture and
Figure 1. An example of injection molded load bearing
Fatigue of Unfilled and Reinforced Polyamides and
thermoplastic part with various modes of local cyclic
Polyesters”, Chapter 5, in Solid State Behavior of
loading (boss – compressive, ribs – flexural).
Linear Polyesters and Polyamides (editors J. Schultz
and S. Fakirov), pp. 249-322, Prentice Hall, (1990).
3. Malzahn, J., C. and Schultz, J., M., “Tension-Tension
a b
and Compression-Compression Fatigue Behavior of
Injection-molded Short-glass-fiber/Poly(ethylene
terephthalate) Composite”, Composite Science and
Technology, Vol. 27, pp. 253-289, Elsevier Applied
Science Publishers Ltd., England (1986).
4. Jia, N. and Kagan, V., “Effects of Time and Tem-
perature on Tension-Tension Fatigue Behavior of
Short Fiber Reinforced Polyamides”, ANTEC Pro-
ceedings (1997).
5. V., Kagan, R., McPherson, and J., Chung, “An Ad-
vanced High Modulus (HMG) Short Glass-Fiber Re-
Figure 2. Fatigue failure mode for the boss at
inforced Nylon6: Part II – Mechanical Perform-
compression-compression loading. Agenda: a - shear
ance”, ANTEC Proceedings (2001).
cleavage along max shear stress plane; b - advanced stage
6. Srivastava, V., K. and Kawada, H., “Fatigue Behav-
of failure with the break along the knit line.
ior of Alumina-Fiber-Reinforced Epoxy Resin Com-
posite Pipes Under Tensile and Compressive Load-
ing Conditions”, Composites Science and Technol-
ogy, Vol. 61, pp. 2393-2403, Elsevier Science Ltd.,
(2001).
7. Chen, W., Cheng, F., Lu., “Tension and Compres-
sion Test of Two Polymers Under Quasi-Static and
Dynamic Loading”, Polymer Testing, Vol. 21, pp.
113-121, Elsevier Science Publishers Ltd., (2002).
8. Krohn, J., A., Novak, G., E., and Wyzgoski, M., G.,
“Reinforced Plastics Design: Tensile Versus Flex-
ural Fatigue”, ”, ANTEC Proceedings (2001).

Property (in MPa) Strength Modulus

Tensile 190 12,700
Flexural 278 17,100 Figure 3. Fatigue failure mode for rectangular specimen
Compressive 191 8,087 at compression-compression loading (shear cleavage
along max shear stress plane).
Table 1. Typical mechanical properties of 45 wt.% GF
PET colored in black at 23°C (multi-purpose test speci-
men ISO-3167).
 90 300
278

80
250
70

Strength, MPa
200 191 190
60 179
Sa, MPa

50 150 140
Compression (ISO Bar)
40 Compression (Boss)
Compression (Shaved Boss) 100
30 4-pt Bending (Flexural)
Tension -Tension
20 50
Flexural @ 30Hz

10
100 1000 10000 100000 0
Boss- PET (ISO Flex Ten.
Boss
Cycles to Failure (N) Shaved Bar) Strength Strength
Mean 139.9 178.6 190.9 278.0 190.1

Figure 4. Comparison of Fatigue Properties of 45 wt.%

short-glass-fiber reinforced PET. Agenda: Sa – is stress
amplitude and Nf – number of cycles to failure. Figure 6. Mechanical Properties of 45 wt.% short-glass-
fiber reinforced PET plastic.
0.45

0.40
Compressive Load Re-orientation of Fibers
0.35

0.30
Sa/Su

0.25

0.20
Compression (ISO Bar)
0.15 Compression, Boss
Compression, Shaved Boss
0.10
4-pt Bending (Flexural)

0.05 Tension-Tension
Flexual @ 30Hz a b
0.00
100 1000 10000 100000
Figure 7. Specific of glass-fiber orientation for hollow
Cycles to Failure (N) boss (top layer content fibers oriented perpendicular to
Figure 5. Master Sa /Su – Nf Curve for All Fatigue Tests. stress direction). Agenda: a – morphology of hollow boss;
Agenda: where Sa is stress amplitude; Su is static strength b – glass-fiber orientation and distribution at top layer,
of plastic/used specimen, and Nf is number of cycles to which was removed).
failure.
Keywords
Poly(ethylene terephthalate), PET, fiberglass, re-
inforced, orientation, low cycle, fatigue, strength, me-
chanical performance, tensile, flexural, compressive.
Petra is a registered trademark of BASF Corporation.
Copyright BASF Corporation 2003.

Copyright BASF Corporation 2003

This information is provided for your guidance only. We urge you to make all tests you deem appropriate prior to use. No warranties, ei-
ther expressed or implied, including warranties of merchantability or fitness for a particular purpose, are made regarding products de-
scribed or information set forth, or that such products or information may be used without infringing patents of others.

BASF Corporation
3000 Continental Drive - North
Mount Olive, New Jersey 07828-1234

www.basf.com/usa
www.plasticsportal.com

©Copyright BASF Corporation 2003

HELPING MAKE PRODUCTS BETTER™