12 Fatigue Life Estimation of Polymers

R.P.M. Janssen, L.E. Govaert, H.E.H. Meijer

Eindhoven University of Technology, Department of Mechanical Engineering

Introduction: Static Fatigue Using the plastic strain evolution in the model, this can be
Life time-estimation is imperative for reliable design of poly- solved analytically for a sawtooth signal.
mer components. Previous research showed for static fa- 3
10
tigue: cyclic stress signal
analytical solution
numerical predictions

2 Quantitative life time prediction possible, taking ther-

impact factor [−]

stress [MPa]
2
10

mal history into account. mean stress signal fails 20x faster

2 Failure governed by strain softening triggered by accu- 1

10

mulation of plastic strain. σ

ampl
= 6 MPa

80 cycle time
0
10
120 annealed
time [s]
0 2
σampl
4
[MPa]6 8 10

70
applied stress [Mpa]
true stress [MPa]

80 annealed 60 quenched Figure 3 Left: Impact of sawtooth signal on mean stress.

50
Right: Impact factor az for a sawtooth signal.
quenched
40
40
This approach yields the following impact factor for a saw-
0
0 0.25 0.5 0.75 1
30 0
10 10
2
10
4
10
time-to-failure [s]
6 8
10 tooth wave shape:
true strain [−]

Figure 1 Left: Intrinsic deformation of quenched and annealed PC. sinh(σampl /σ0 )
Right: Life time prediction of static fatigue on PC.
az =
(σampl /σ0 )

Remarkably, for this wave-shape only the stress amplitude

Impact of the Dynamic Component σampl affects the acceleration factor az . From this analytical
From cyclic fatigue experiments and simulations having a approach, it appears that frequency and mean stress do not
sawtooth-shape, it appears that (see figure 2):
play a role at all.
2 Again quantitative life time prediction is possible.
2 The dynamic component shifts the life time to lower Fatigue simulations, which are performed to validate the
values on the log(t)-axis with increasing amplitude. analytical result show that the impact factor derived from
65 65
these simulations closely matches the analytical result. Next
creep (s =0)
to a sawtooth-wave, also simulations of a sine-wave and a
d
Mean stress [MPa]

square-wave match the derived impact factor, see figure 4a.

mean stress [MPa]

60 60

constant shift
cyclic (s =4)
d

55 55
applied stress [MPa]

σ
ampl
= 0 MPa
σ (t)
σampl = 4 MPa
dyn
circles/squares = experimental σampl = 6 MPa
line = numerical prediction
50 1 50 2
10 10
2
10
3
10
4
10
5
10
3
10
4
10 σ
Time to failure [s] time to failure [s] mean

Figure 2 Left: Life time prediction of cyclic fatigue on PC. Right:

Numerical simulations indicate a shift in life time due cycle time

to the dynamic component time [s]

Based on the results from the simulation, an impact factor Figure 4 Left: Impact factor az for a sawtooth, square and sine
aimpact is defined. This relates the dynamic life time to the signal. Right: Any periodic stress signal has impact factor.
life time of the static mean stress, as follows:
tf ail,static Discussion
aimpact = It is shown that cyclic fatigue life time can be estimated from
tf ail,dynamic
the static life time. Based on the development of plastic
It is proposed that such a impact factor can also be derived strain during a cycle an impact factor was defined. Using this
from the plastic strain γ pl developed during a cycle, as de- impact factor, failure due to any periodic dynamic stress sig-
picted in figure 3a. Since numerical simulations of cyclic fa- nal (figure 4b) can be estimated from the plastic strain devel-
tigue are time-consuming, such an impact factor could save oped during a cycle. This means that time-consuming simu-
simulation time. lations can be avoided.
γpl,dynamic (tcycle )
aimpact = References:
γpl,static (tcycle )
[1] K LOMPEN , E.T.J., P H D-T HESIS , TU/ E , 2005

/department of mechanical engineering PO Box 513, 5600 MB Eindhoven, the Netherlands