Previous Paper Back to Programme Back to Topic
DETERMINATION OF TIME-TEMPERATURE SHIFT
FACTOR FOR LONG-TERM LIFE PREDICTION OF
POLYMER COMPOSITES
K. Fukushima*, H. Cai**, M. Nakada*** and Y. Miyano***
* Graduate School, Kanazawa Institute of Technology
7-1 Ohgigaoka, Nonoichi, Ishikawa 921-8501, Japan
[email protected] ** School of Materials Science and Engineering, Xi'an Jiaotong University
28 Xianningxi Road, Xi’an, Shaanxi,710049, China
*** Materials System Research Laboratory, Kanazawa Institute of Technology
3-1 Yatsukaho, Hakusan, Ishikawa 924-0838, Japan
SUMMARY
The methodology for accurately determining the time-temperature shift factor used in
the accelerated testing for the long-term life prediction of polymer composites is
proposed in this paper.
Keywords: Polymer composites, Life prediction, Viscoelasticity, Time-temperature
superposition principle
INTRODUCTION
In recent years materials that possess high specific strength and specific modulus were
developed to fulfill the need for advanced lightweight structures. Fiber reinforced
plastics (FRP) are materials that have these properties, and are being used in structures
as primary as well as secondary load carrying members. Therefore, the development of
a testing procedure to evaluate the lifetimes of materials in extreme service
environments is becoming a high priority.
The stress-strain relation of polymer resin exhibits time and temperature dependence,
called viscoelastic behavior, not only above the glass-transition temperature Tg but also
below Tg. Thus, it can be presumed that the mechanical behavior of FRP also depends
on time and temperature even below Tg which is within the normal operating-
temperature range. These examples have been shown by Aboudi et al. [1], Gates [2],
Rotem et al [3].
With service lifetimes measured in years, it is almost unthinkable to do real time testing
under a variety of conditions. Accelerated testing methodologies for metals have been
studied for some time. One of the most popular tools used to predict the fatigue life of
metals is the S-N curve, which is based on the assumption that fatigue life depends on
cycles, but not on time. Therefore, the cyclic loads can be applied at much higher
frequencies than the actual loading to accelerate the fatigue test. Unlike metals,
polymer composites are viscoelastic and their properties exhibit strong time and
temperature dependencies as mentioned above. Since time plays an important role in
the fatigue and creep of polymer composites, simply applying the S-N curve to polymer
composites will not provide accurate prediction of the fatigue life.
We proposed the accelerated testing methodology (ATM) based on the time-
temperature superposition principle (TTSP) for polymeric materials. This principle was
originally developed for non-destructive material properties, but our recent studies have
shown that it can also be applied to failure properties of composite materials [4-6]. In
this case, elevated temperature states are used to accelerate the mechanical degradation,
which occurs under loads over long period of time at lower temperature. We predicted
the fatigue strengths of various kinds of FRP and structures under various types of
loading methods using our methodology to discuss their applicability [7].
In this paper, the methodology for accurately determining the time-temperature shift
factor used in ATM for the long-term life prediction of polymer composites is proposed
and experimentally demonstrated.
ACCELERATED TESTING METHODOLOGY
Our ATM which rests on the three hypotheses, (A) same TTSP for all strengths, (B)
linear cumulative damage law for monotonic loading, and (C) linear dependence of
fatigue strength upon stress ratio. When these hypotheses are met, the fatigue strength
under an arbitrary combination of frequency, stress ratio, and temperature can be
determined based on the following test results: (a) master curve of strength under
constant strain-rate (CSR) loading and (b) master curve of fatigue strength for zero
stress ratio as shown in Fig. 1. The master curve of CSR strength is constructed from
the test results at single strain-rate and various temperatures. On the other hand, the
master curve of fatigue strength at zero stress ratio can be constructed from the test
results at single frequency for various temperatures based on the hypothesis (A). The
details of ATM can be obtained from our published paper [7].
Figure 1 Accelerated testing methodology (ATM)
DETERMINATION OF TIME-TEMPERATURE SHIFT FACTOR
Inside of the frame of Fig. 2 shows the storage modulus E' versus time t (inverse of
frequency) at various temperatures T (T1~T3) for matrix resin. The master curve of E'
versus the reduced time t’ was constructed by shifting E' at various constant
temperatures along the log scale of t and the log scale of E'. The horizontal time-
temperature shift factor aTo(T) and the vertical temperature shift factor bTo(T) at a
reference temperature T0 plotted in Fig. 3 are respectively defined by following equation.
log aT 0 (T ) = log t − log t ' (1)
log bT 0 (T ) = log DC (t , T ) − log DC (t ' , T0 ) (2)
Figure 2 Master curve of storage modulus
Figure 3 Time-temperature shift factor aTo(T) and temperature shift factor bTo(T)
To determine accurately the time-temperature shift factor in TTSP, we propose the
following steps:
(1) To evaluate the influence of physical aging on the viscoelastic modulus of matrix
resin during the accelerated testing,
(2) To determine the post curing conditions of specimen based on (1),
(3) To measure the viscoelastic modulus at various times and temperatures for
specimens which are cured at conditions obtained in (2),
(4) To construct the master curve of viscoelastic modulus using the measured data in (3)
without personal errors.
At the time of master curve creation, the automatic shift program was used. For
example shown in Fig. 4, the auto-shift will shift the creep compliance curve measured
at T2 horizontally and vertically to that at T1 with aTo and bTo, respectively, to minimize
the sum of the deviations (Σ∆yi) from the fitted curve y = a2’x2+a1’x+a0’. The fitting
parameters a2, a1, and a0 are finally determined after shifting.
Figure 4 Principle of automatic shift
EXPERIMENTS
Preparation of Specimens
The epoxy resin used in this study consists of deglycidyl ether of bis-phenol A (jER828),
metylnadic anhydride, and 2-ethyl-4-methyl-imidazole as shown in Table 1. Epoxy
resin was cured by casting with a cure schedule of 70°C for 12 hours, 150°C for 4 hours,
190°C for 2 hours followed by a slow cooling at 0.5°C per minute. By these heat
treatments, the specimens were stabilized chemically during the accelerated testing.
The specific volume of polymer resins consists of the occupied volume and the free
volume [8]. The occupied volume is the volume occupied by molecular chains and the
free volume is the space which permits movement of molecular chains. The free
volume decreases moderately and the thermodynamic equilibrium state is approached
with the passage of time. This behavior is called physical aging.
If physical aging of resin progresses during the accelerated testing, the viscoelastic
property of resin will be influenced. In this study, specimens were put into the constant
temperature chamber under 120°C for 50 hours after curing to conduct physical aging
before the accelerated testing. We confirmed that the creep compliance of epoxy resin
without aging treatment (120°C for 50 hours) which measured in 2nd times decreases
remarkably during the creep test at 110 oC as shown in Fig. 5 (a). On the other hand, the
creep compliance of epoxy resin with aging treatment which measured in 2nd times
decreases scarcely during the creep test at 110 oC as shown in Fig. 5 (b).
Table 1 Molding condition for epoxy resin
(a) Measured by using specimen without aging treatment
(b) Measured by using specimen with aging treatment
Figure 5 Creep compliance of epoxy resin
Viscoelastic Test
The creep compliance and storage modulus of epoxy resin were obtained as the
viscoelastic modulus. The creep compliance was obtained by the creep test machine
which can measure the viscoelastic deformation precisely. Three point bending creep
tests for epoxy resin were carried out under various temperatures using an creep testing
machine with temperature chamber. The creep compliance Dc was calculated from the
deflection δ at the center of specimen using the following equation.
4bh 3δ
DC = (3)
P0 L3
where P0 is dead load which is decided as the strain at the center of the specimen less
than 1%. L is the span (50mm), and b and h are the width (25mm) and the thickness
(3.0mm) of specimen, respectively.
The storage modulus was obtained by the dynamic mechanical analyzer (DMA) which
can evaluate easily the storage modulus in wide ranges of temperature and frequency.
The strain amplitude of 0.06% by the sinusoidal wave with frequency range of 0.01-
10Hz was applied to specimen. The width, thickness and length of specimen are 6.4mm,
1.6mm, and 44.5mm, respectively. The span is 38mm.
RESULTS AND DISCUSSION
The left sides of Figs. 6 and 7 show the creep compliance Dc and the storage modulus
E’ versus time t at various temperatures T of epoxy resin. The master curves of Dc and
E’ versus the reduced time t’ were constructed by shifting Dc and E’ at various constant
temperatures along the log scale of t and the log scale of Dc and E’ as shown in the right
sides of Figs. 6 and 7. Since Dc and E’ at various temperatures can be superimposed
smoothly, TTSP is applicable for Dc and E’.
Figure 6 Master curve of creep compliance of epoxy resin
Figure 7 Master curve of storage modulus of epoxy resin
Figure 8 shows the time-temperature shift factor aTo and the temperature shift factor bTo
those obtained from the master curves of Dc and E’, respectively. The aTo obtained from
DMA and creep test agree well with each other, while the bTo do not agree with each
other. The difference in the bTo from DMA and creep test was caused by the loading
method.
(a) Time-temperature shift factor aTo(T) (b) Temperature shift factor bTo(T)
Figure 8 Shift factors obtained from master curves of Dc and E’.
The master curves of compliance obtained from DMA and the creep test are shown in
Fig. 9, where the compliance from DMA is the inverse of E’. The master curves from
DMA and creep test do not agree with each other, however those shapes are almost
same. From these results, it can be considered that the time-temperature shift factor is
obtained accurately and easily by using DMA.
Figure 9 Dc of creep test and DMA
CONCLUSION
This paper is concerned with the accurate determination of the time-temperature shift
factor used in the accelerated testing for the long-term life prediction of polymer
composites. Aging treatment is performed to the specimen before the accelerated
testing, and the specimen is stabilized physically. The dynamic viscoelastic test and
creep test were performed as the accelerated testing to measure the storage modulus and
the creep compliance, respectively. The master curves of the storage modulus and the
creep compliance were constructed by the automatic shift program and the time-
temperature shift factor is determined without personal errors. From the comparisons of
both test results, it can be considered that the time-temperature shift factor is obtained
accurately and easily by the dynamic viscoelastic test.
REFERENCES
1. Aboudi, J. and Cederbaum, G., Analysis of Viscoelastic Laminated Composite
Plates, Composite Structures, 12 (1989) 243-256.Sullivan, J., Creep and Physical
Aging of Composites, Composite Science and Technology, 39 (1990) 207-232.
2. Gates, T., Experimental Characterization of Nonlinear, Rate Dependent Behavior in
Advanced Polymer Matrix Composites, Experimental Mechanics, 32 (1992) 68-73.
3. Rotem, A. and Nelson, H. G., Fatigue Behavior of Graphite-Epoxy Laminates at
Elevated Temperatures, Fatigue of Fibrous Composite Materials, ASTM STP 723
(1981) 152-173.
4. Miyano, Y., Kanemitsu, M., Kunio, T. and Kuhn, H., Role of Matrix Resin on
Fracture Strengths of Unidirectional CFRP, Journal of Composite Materials, 20
(1986) 520-538.
5. Miyano, Y., McMurray, M. K., Kitade, N., Nakada, M. and Mohri, M., Role of
Matrix Resin on the Flexural Static Behavior of Unidirectional Pitch-based Carbon
Fiber Laminates, Advanced Composite Materials, 4 (1994) 87-99.
6. Miyano, Y., McMurray, M. K., Enyama, J. and Nakada, M., Loading Rate and
Temperature Dependence on Flexural Fatigue Behavior of a Satin Woven CFRP
Laminate, Journal of Composite Materials, 28 (1994) 1250-1260.
7. Miyano, Y., and Nakada, M., Cai, H., “Formulation of long-term creep and fatigue
strengths of polymer composites based on accelerated testing methodology”,
Journal of Composite Materials, Vol.42, pp.1897-1919 (2008).
8. J. D. Ferry, "Viscoelastic Properties of Polymers", 218, John Wiley & Sons, Inc.,
1961.
