Experimental and Numerical Analysis of Creep

Behavior of Expanded Polystyrene (EPS)

S. D. Mhaske1, U. H. Yadav2, S. V. Jagtap3

Abstract— Expanded Polystyrene (EPS) is an extremely light As a consequence, modeling creep phenomenon in polymers
weight material. It has features like good thermal insulation, becomes an important issue for predicting service lifetime of
moisture resistance, durability and low thermal conductivity. It products. Creep data is used to determine the viscoelastic
has been extensively used for the purpose of packaging. The material coefficients. These coefficients are used to derive an
inventory of refrigerators in a warehouse is done in the form of empirical relationship between time dependent strain, stress
stacks. On an average four refrigerators are stacked one-above-
the-other for a period of 30 to 45 days. The EPS material at the
and density [3]. Test data is used to model the creep behavior
base is under immense compressive load and is strained for long of the EPS in Ansys. Non linear regression analysis is done in
durations. EPS is vulnerable to time-dependent creep Ansys to determine the curve fitting constants to model creep
deformation when a constant magnitude of load is applied even behavior of EPS.
at room temperatures. Thus study of creep behavior of the EPS
becomes essential to find deformations due to creep. In design of
EPS packaging material creep deformation is the primary II. LITERATURE REVIEW
concern rather than shear failure. The creep behavior of EPS is
mostly influenced by its sample size, loading and density. The
Y. Z. Beju and J. N. Mandal [1] studied the effect of low
tests have been carried out on EPS with sample size of 50 mm
cube and density 24 kg/m3 under the applied compressive loads
density and sample size of EPS on the creep behavior. Small
of 41 and 55 kPa. A mathematical equation is proposed to predict samples overestimated the creep deformations. Most creep
the time dependent strains. It is then compared with the deformation tends to develop in the few weeks after loading.
experimental results. A good correlation between these values is High density EPS develop less creep deformation at same
observed. Also the material behavior of EPS is modeled in Ansys loading. S. Srirajan et al [2] studied the effect of sample size
to carry out creep simulations. The simulation results are in and load for different densities of EPS. Results showed that
agreement with both experimental values and mathematical the compressive strengths are not significantly affected by the
values. sample size. Density affects the creep behavior. John S.
Horvath [3] developed a deterministic approach to model the
Keywords- Expanded Polystyrene (EPS), packaging, creep,
correlation, material model, Ansys, creep simulation. time-dependent component of strain using power law
equation. Five material parameters were evaluated using
compressive creep tests. Martin J. Dropik et al [4] modeled
I. INTRODUCTION primary creep behavior for polypropylene using Ansys.
Maxwell model is used to represent the creep behavior in
polypropylene. A good correlation is observed between Ansys
Creep behavior is an important design consideration for EPS and experimental results for low values of stress but a high
[1-5]. Creep strains cause changes in the dimensions of a error is encountered for high value of stresses. Saulius Vaitkus
product with time and also cause reduction in strength that [5] studied exponential model for the relationship between the
could affect the ability of products to resist design loads [6]. creep deformation and time. D. L. May and A. P. Gordon [6]
EPS being a polymer is particularly susceptible to creep even calculated creep constants by using Norton- Bailey law and
at room temperature [4]. Since design applications require that utilizing bivariate power-law stastical regression. J. T.
a part be loaded for long-time duration, this loading will lead Maximov et al [7] developed constitutive models for modeling
to creep which may induce unacceptable permanent strain hardening and creep behavior of 2024T3 aluminum
deformations or loss of assembly preload [1]. EPS material alloy at room and high temperatures in order to be used in
shows a time dependent behavior arisen from viscoelastic/ finite element simulations of cold hole working process.
viscoplastic nature of polymers [7]. Experimental study of Roham Rafiee and Behzad Mazhari [8] developed a
creep requires a time consuming and costly experimental setup comprehensive model to evaluate creep phenomena for
[8]. Moreover, the whole experimental study should be different laminate configurations and geometries. Wensu Chen
repeated for any variation in materials and/or lay-up et al [9] studied static and dynamic properties of EPS foams
configurations. using static and dynamic compressive and tensile tests. Gnip
1
et al [10] showed that stress relaxation depends on the initial
S. D. Mhaske is Post Graduate Student in Dept. of Mechanical Engineering
at Sinhgad college of Engineering, Pune-41
value of the stress. Higher the initial stress greater is the
equilibrium stress that remains. The attenuation factor is used
2
U. H. Yadav is Assistant Professor in Department of Mechanical Engineering as a criterion to calculate the relaxation resistance of EPS. S.
at Sinhgad college of Engineering, Pune-41. Vaiktus et al [11] predicted the creep strain of expanded
3
S.V. Jagtap is Global Senior Engineer at Whirlpool of India, GTEC, Pune-14
polystyrene in long term compression.
Many researchers have carried out studies to study the effect
of density and sample size on the creep behavior of the EPS.
Also several analytical methods are proposed to predict the
long term creep strains. However very less amount of work is
done in modeling the creep behavior of EPS for simulation
purpose. The objective of this study is to model creep behavior
of EPS in Ansys and compare the results obtained from creep
simulations with experimental and analytical results.

III. EXPERIMENTAL SET UP AND RESULTS

A. TEST DESCRIPTION

Tests were done at BHTC lab in Michigan by authorized

Whirlpool lab technicians. The experimental results provided Fig.2. Creep test results: Total strain vs Time
by BHTC lab is used for correlation with numerical
simulation. The creep-test result for a set of two specimens of
density 24 kg/m3 were selected. Each specimen was a 50 mm
cube. Specimens of this shape and size are worldwide standard
for testing EPS block [1]. Figure 1 represents set up for creep
test. Load of 55 kPa and 41 kPa was applied on the two
specimens respectively. This load was applied considering the
stacking of refrigerators. The specimens were loaded for
duration of 30 days. Since the study of effect of variation of
temperature on creep behavior of EPS is not in the purview of
research, the tests were performed at room temperature.

Fig.3. Creep strain vs Time

IV. EMPIRICAL MODELS

A. Modified Findley Equation

Findley equation is one of the equations used to describe the

creep behavior of EPS material. It is given as:
Fig.1- Experimental setup

(1)
B. EXPERIMENTAL RESULTS
Where:
The initial axial strain for the EPS blocks at loads 55 kPa and = a dimensionless Findley material parameter,
41 kPa are found to be 0.66% and 0.26% respectively as
= a dimensionless Findley material parameter,
shown in figure 2. As the load applied is less than 25% of the
compressive strength (200-210 kPa) the initial strain is = a dimensionless Findley material parameter,
expected to be less than 1% [1]. Figure 3 shows that most of = a Findley material parameter with dimensions of stress,
the creep deformation is developed in the first couple of weeks = a Findley material parameter with dimensions of stress,
after the loading. About 70% of the total creep strain was σ = applied load,
observed in the first two weeks. The creep strain increases t = time in hours.
with increase in applied load.
Considering the rheological behavior of EPS under
compression test the hyperbolic functions are used. All the
Findley material parameters ( , , , , ) are
assumed to be dependent on material and temperature and are
independent on stress, strain and time.
Findley assumed specific relationships for initial strain and
m as follows:

(2)

(3)

All the Findley material parameters can be determined by

using the test data. is the average slope of creep strain vs
Fig.4. Plot of Log creep vs Log time
time plot for the two applied loads. The average of slope in
figure 3 is 0.55. Therefore the value of is 0.55. The initial In a similar fashion taking log of equation (6) holding time
strain values ɛ0 for the applied loads 55kPa and 41 kPa are constant yields:
0.66% and 0.26 respectively. Putting these values in equation
(2) we have two equations with two unknowns. Solving these (8)
equations the values for and are 0.0357 and 15.23 kPa
respectively. Again equation (8) represents equation of line with slope m.
Figure 5 shows log creep plotted vs log stress. The slope of
For each applied load, parameter m is the magnitude of creep above plot m is equal to 0.64.
strain at a time 1 hour. Again putting these values in equation
(3) we have two equations with two unknowns. Solving these
equations the values for and are 0.00205 and 16.79
kPa respectively.
Thus the Findley equation for total strain is given by:

(4)

Creep strain can be obtained by subtracting the initial strain

from the total strain.

B. Norton- Bailey Equation

Norton- Bailey Equation for calculating creep strain is given

by:
Fig.5. Plot of log creep vs log stress
(5)
The coefficient ‘a’ is determined by putting the value of m and
Where, m and n are the constants representing exponents of n into equation (6) along with creep strain data from
the stress and time variables. ‘t’ represents time in hours. Q experiment. The average value of ‘a’ is 0.0019.
represents activation energy, R represents Boltzmann’s Putting all constants in equation (6) the Norton equation for
constant and T represents the absolute temperature. As the creep strain is:
temperature is constant for this study Q/RT is zero, making
exp (0) equal to 1. The equation (5) thus reduces to: (9)

(6)

Taking log of equation (6) while holding stress constant

yields:

(7)
Equation (7) represents equation of line with slope ‘n’. Figure
4 shows log creep plotted vs log time. The average slope n
obtained from figure 4 is 0.52.
 V. FINITE ELEMENT ANALYSIS MODEL

ANSYS provides thirteen model types for creep curve fitting.

The model you choose will determine the experimental data
required for the curve fitting process. Modified time hardening
type curve fitting require creep strain, time, stress and
temperature values as input data. All other type of curve fitting
require creep strain rate. We don’t have creep strain rate value.
Modified time hardening formula to calculate creep strain is
given by:

(10)

Where, C1, C2, C3 and C4 are the creep constants. Creep

constants are determined using curve fitting technique in
Ansys APDL. Creep strain vs time data is given as input in
Fig.7. Comparison of experimental, analytical and simulation
APDL to obtain creep constants. These constants are used to
results for 55 kPa loading
model the creep behavior of EPS material in Ansys. To
simulate test conditions in Ansys the EPS block is kept
For 41 kPa loading the experimentation data, simulation
between two plates as shown in figure (6). Load is applied on
results and analytical results are compared in figure 8. Poor
the top plate and the EPS block rests on the bottom plate.
correlation is obtained between experimental, simulation and
Frictional contact is given between block and plates. The
analytical results for the initial values. The correlation
bottom plate is fixed. Implicit model is used. Analysis is a non
improves with time. Average error of around 20% is obtained
linear analysis involving material non linearity. Solid 185
between experimental and simulation results. A very poor
element type is used.
correlation is observed between Modified Findley equation
and the experimental results. But average error of around 16%
is observed between Norton equation and experimental results.
Percentage errors obtained from different methods of
calculating creep are show in table 1.

VI. RESULTS AND DISCUSSION

For 55 kPa loading the experimentation data, simulation

Fig.8. Comparison of experimental, analytical and simulation results for 41
results and analytical results are compared in figure 7. A good kPa loading
correlation is observed between experimental, simulation and
analytical results as time increases. Poor correlation is
observed for the initial values. The simulation results and the
analytical results show good correlation throughout. Average
error of less than 15% is obtained between experimental and
simulation results. And average error of less than 13% is
obtained for experimental and analytical results.
Table 1 Comparison of % Error obtained from different methods of [4] M. J. Dropik, D. H. Johnson, D. E. Roth. Developing
calculating creep strain
an ANSYS Creep Model for Polypropylene from
Experimental Data, Penn State-Erie, Erie, PA, USA.
% Error
[5] I. Y. Gnip , S. Vaitkus, V. Kersulis, S. Vejelis
55 kPa 41 kPa
(2011).Analytical description of the creep of
Loading Loading
expanded polystyrene (EPS) under long-term
Test & Simulation 14.7 20
compressive loading. Polymer Testing. 493-500.
Test & Modified Findley [6] D. L. May, A. P. Gordon, D.S. Segletes (2013). The
Equation 12.8 46 Application of the Norton-Bailey Law for Creep
Test & Norton Equation 13.8 16 Prediction through Power Law Regression. Presented
Modified Findley & Norton at Turbine Technical conference and Exposition.
Equation 2.1 73 [7] J. T. Maximov, G. V. Duncheva, A. P. Anchev, M.
Norton Equation & Simulation 9.6 8.7 D. Ichvoka (2014). Modeling of strain hardening and
creep behavior of 2024T3 aluminium alloy at room
temperature and high temperatures. Computational
VII. CONCLUSIONS Materials Science, 381-393
[8] R. Rafiee, B. Mazhari (2015). Modeling creep in
1. Maximum creep strain is developed in first couple of polymeric composites: Developing a general
weeks after loading. integrated procedure. International journal of
2. Creep strains are observed to be dependent on Mechanical Sciences, 112-120
loading. Creep strains decreases with decrease in [9] W. Chen, H. Hao, D. Hughes, Y. Shi, J. Cui, Z. Li
load. (2015). Static and dynamic mechanical properties of
3. The simulation results vary from experimental values expanded polystyrene. Materials and Design, 170-
for initial duration of time but a good correlation is 180.
observed as time increases. [10] S. Vaitkus, V. Granev, I. Gnip, S. Vejelis, A. Kairyte
4. Norton equation is more reliable than Modified (2013). Stress Relaxation in Expanded Polystyrene
Findley for predicting creep for both the load cases. (EPS) Under Uniaxial Loading conditions. Presented
5. Initial errors are result of end effect and seating error. at 11th International Conference on Modern Building
To avoid this size of specimen can be increased. Materials, Structures and Techniques.
[11] S. Vaitkus, I. Gnip, V. Kersulis, S. Vejelis (2007).
Prediction of Creep Strain of the Expanded
ACKNOWLEDGMENT Polystyrene (EPS) in long-term Compression.
Materials Science. 13 (4).

Authors would like to thank the Whirlpool Corporation for

their financial support and providing opportunity to access
Ansys software. Furthermore, the authors would like to
express their gratitude to lab technicians of BHTC lab in
Michigan for providing experimental results as per
requirement of research.

REFERENCES

[1] Y. Z. Beju and J. N. Mandal (2016). Compression

creep test on expanded polystyrene (EPS) geofoam,
International Journal of Geotechnical Engineering.
[2] S. Srirajan, D. Negussey and N. Anasthas (2006).
Creep behavior of EPS geofoam, Geofoam Research
Center, Syracuse University, Syracuse, New York,
USA.
[3] J. S. Horvath, “Mathematical Modeling of the Stress-
Strain-Time Behavior of Geosynthetics Using the
Findley Equation: General Theory and Application of
EPS-Block, Manhattan College Research Report No.
CE/GE-98-3, Bronx NY, USA, 1-35.