Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

MECHANICAL PROPERTIES OF POLYURETHANE-

FOAM IMPACT LIMITERS

By A. K. Maji,· Member, ASCE, H. L. Schreyer,2 Member,


ASCE, S. Donald,3 Q. ZUO,4 and D. Satpathi5

ABSTRACT: Rigid, closed-cell, polyurethane foams possess excellent energy-absorbing


properties and are used as impact limiters for various packaging applications. They can
absorb large quantities of energy as they transmit a stress equal to their crushing strcngth.
Recently, their mechanical properties have come under increased scrutiny due to appli-
cations in nuclear-waste shipment packages. In such an application. the matcrial is sub-
Downloaded from ascelibrary.org by Istanbul Teknik Universitesi on 08/17/23. Copyright ASCE. For personal use only; all rights reserved.

jected to multiaxial state of stress. Strain-rate effects due to the dynamic naturc of the
loading in an hypothetical accident situation is also likely to be encountered. This papcr
describes a number of tests performed on polyurethane foams of various densities along
several loading paths. These include uniaxial, hydrostatic and triaxial compression. Strain-
rate effects were also studied under compressive loading. Since the material is significantly
anisotropic, the experimental data are used to infer elasticity parameters and the shape
of the anisotropic yield surface. Some strain-rate effects are also examincd.

INTRODUCTION
Cellular materials are widely used in applications ranging from coffee mugs to the fect of
Apollo II, for which they were used as shock absorbers. Due to their excellent thermal, acoustic.
buoyancy, and structural properties, their use has increased immensely, particularly in civil
engineering applications.
Rigid, closed-cell polyurethane foams are frequently used as impact limiters in padding nu-
clear-waste shipment casks. Fig. I(a) shows a micrograph demonstrating the closed-cell structure
of the foam. This picture was taken with a backscatter scanning electron microscope, under a
magnification of ISO. Polyurethane foams absorb impact energy that may result from accidents
during transportation and handling. Although the expected stresses resulting from an accidental
impact are primarily compressive, some areas of the package are expected to experience tensile
and hydrostatic stresses due to the confinement of the foam between the inner and outer shells
of transportation casks.
The NuPac 140B plutonium shipping cask (from Nuclear Packaging, Federal Way, Wash.)
uses the foam as an excellent impact energy absorber and as insulation in case of a hypothetical
fire accident condition ("Safety" 1989). Fig. l(b) shows the cask and the foam placement along
with the other structural components of the cask. The stainless-steel/foam sandwich construction
of the outer containment assembly provides a tough, puncture-resistant, impact-absorbing shell
that insulates and cushions the contents of the vessel, A large number of casks are still to be
designed for future transportation needs of the U.S. Department of Energy and equivalent
international agencies.
Although these materials are used in important applications, test data supporting theorctical
formulations that describe the general behavior of these materials are limited. Theoretical
formulations developed by Gent and Thomas (1959) and Lederman (1971) were useful to some
extent, but their formulations assumed that the cell walls bear only the axial load. Many other
theories (Matonis 1964; Ko 1965; Chan and Nakamura 1969) were set forth to predict the
mechanical properties of foams. Later, Patel and Finnie (1970) provided an excellent description
of the behavior of cellular materials by including the effects of anisotropy and the packing of
the cell units. These early researchers sought to understand the relationship between the elastic
properties of the foams (such as modulus and yield strength) as a function of their geometry
(primarily the density). Although their analyses were successful in modeling the uniaxial response
for a limited set of materials, the empirical parameters in the equations they developed preclude
their generalization to the present-day materials.
Gibson et al. (1982) studied the mechanical properties of honeycombs in a morc general
manner by studying the properties in terms of bending, elastic buckling, and plastic collapse of
'Asst. Prof., Dept. of Civ. Engrg., Univ. of New Mexico, Albuquerque, NM R7131-1351.
'Prof.. Dept. of Mech. Engrg .. Univ. of New Mexico. Albuquerque, NM.
'Res. Asst., Dept. of Civ. Engrg .. Univ. of New Mexico, Albuquerque. NM.
Asst.. Dept. of Mech. Engrg .. Univ. of New Mexico, Albuquerque, NM.
'Res. Asst., Dept. of Civ. Engrg., Univ. of New Mexico, Albuquerque, NM.
Note. Associate Editor: Jean-Lou A. Chameau. Discussion open until September \,1995. To extend the closing
date one month. a written request must be filed with the ASCE Manager of Journals. The manuscript for this
paper was submitted for review and possible publication on October 13, 1992. This paper is part of the Journal
of Engineering Mechanics. Vol. 121. No.4. April. 1995. ©ASCE, ISSN 0733-9399195/0004-052X-05401$2.00 +
$.25 per page. Paper No. 4937.

528 JOURNAL OF ENGINEERING MECHANICS

J. Eng. Mech., 1995, 121(4): 528-540


Downloaded from ascelibrary.org by Istanbul Teknik Universitesi on 08/17/23. Copyright ASCE. For personal use only; all rights reserved.

FIG. 1(a). Closed-Cell Microstructure of Rigid Polyurethane Foams Tested

POLYURETHANE
IMPACT LIMITER

FIG. 1(b). Cross Section of TRUPACT-II Waste Transportation Cask

cell walls. Gibson and Ashby (19112) treated the mechanical properties of foams as functions of
cell-wall properties and cell geometry. and used an electron microscope to compare their theories
with physical observations of cell-wall bending of polyurethane foams under uniaxial compres-
sIon.
All these theories were used to predict the behavior of cellular materials subjected to uniaxial
stresses only. However. in most applications the cellular materials are subjected to multiaxial
stresses. and it is necessary to examine results from multiaxial tests. Shaw and Sata (1966)
studied the failure of polystyrene foams subjected to compressive, tensile. and hydrostatic forces.
and predicted that the foams yield according to a maximum principal stress criterion. Triantafillou
et al. (19119) and Triantafillou and Gibson (1990b), performed experimental studies to predict
the properties of the materials subjected to multi axial forces. Gibson et al. (19119) have developed
failure criteria for both isotropic and anisotropic foams subjected to general multi axial stresses.
To compare the results with theoretical modeling Triantafillou et al. (19119) tested different
types offoams: an open-cell polyurethane, a closed-cell polyethylene. a closed-cell polyurethane.
and an open-cell aluminum foam subjected to uniaxial compression. biaxial compression. axi-
symmetric compression, and hydrostatic compression. These more recent investigations have
been summarized by Gibson (19119). Neilsen et al. (19117) developed a phenomenological con-
stitutive model with combined volumetric and deviatoric plasticity for some low-density foams.
Their study covered a small range of foam densities, and recommended a study of strain-rate
effects. Slow strain-rate conditions were studied by Huang and Gibson (1991) on rigid poly-
urethane foams under constant shear. Triantafillou and Gibson (1990b) have developed a con-
stitutive matrix for open-cell isotropic foams.
This paper presents the of an extensive and unique series of experiments performed
on rigid closed-cell polyurethane foams. and provides a constitutive model that supports these
experimental data.
EXPERIMENTAL PROGRAM
A series of tests was performed on various low- and high-density polyurethane foams:
0.0411-. (J.(lll-, 0.16-. and 0.32-g/cm' (3-, 5-. 10-, and 20-pcf) foams were tested under uniaxial.

JOURNAL OF ENGINEERING MECHANICS 529

J. Eng. Mech., 1995, 121(4): 528-540


triaxial, and hydrostatic compression. Table 1 shows the various tests that will be discussed in
this paper.
The polyurethane foams were supplied by General Plastics Manufacturing Co., Tacoma,
Wash., under the trade name LAST-A-FOAM FR-3703, 3705, 3710, 3720. The last two digits
of the trade name represent the density of the material. The primary chemicals used to produce
the foams are methylene di-isocyanate and polymethylene polyphenylisocyanate reacted with
polyether polyol. Foams of 0.048 and 0.08 g/cm-' (3 and 5 pet) were made by a gas-blown process
using Freon as the gaseous media. The 0.16- and O.32-g/cm-' (10- and 20-pef) foams were made
by a water-blown process. All foams tested were closed-cell. rigid thermosets. The foams under
investigation were anisotropic as will be evident from the experimental data reported later. The
"rise" direction of the foam is the direction in which the material expands during production.
This will also be referred to as the out-of-plane direction, or the one-direction. Directions 2
and 3 refer to the two in-plane directions. The foams arc isotropic in the two-three plane, and
have different properties in the one direction.
Downloaded from ascelibrary.org by Istanbul Teknik Universitesi on 08/17/23. Copyright ASCE. For personal use only; all rights reserved.

The military standard MIL-STD-40IA ("Sandwich" 1967) and ASTM guidelines ("Standard"
1<J<JO) have been used as a guide for conducting the tests and designing the fixtures required for
testing. An Instron model 1323 servo-hydraulic testing machine was used for the ksts. This
machine has a uniaxial loading capacity of 445 kN (100 kips) and a torsional loading capacity
of 5.6 kN-m (50 kip-in.). Axial and torsioal loading can be applied simultaneously, although
that feature was not used in our tests. A data acquisition system consisting of an Isaac hardware
system and a personal computer were used for the various tests. The data collected in the binary
form were converted into an ASCII format by using a program called SCONV. As the loading
frame of Instron was not supplied with the loading platens, a number of test fixtures were
fabricated locally.
A large number of tests were performed along all the mentioned loading paths. It is not
possible to present all the data in the limited scope of a research paper. A more elaborate
description of these tests and the results have been presented in two reports to the Sandia
National Laboratories, Albuquerque, N. M. (Maji et al. I<J<JO; Glass et al. I<J<JO).

INVESTIGATION OF COMPRESSION BEHAVIOR


Uniaxial Compression

Two cold-formed E4130 steel loading platens were fabricated. The loading platens were
precisely machined to assure parallelism of the faces for an even distribution of the load. The
top loading platen was connected to the loading frame of the Instron by a 1.11-cm (0.43X-in.)
pin and the bottom loading platen was threaded on to the actuator.
The specimens used for compression testing were 10.16-cm (4-in.) cubes as prescribed by
Sandia to facilitate comparison of the results with other types of impact limiter materials such
as aluminum honeycombs. The specimens were cut into shape by a band saw. Care was taken
to ensure that the loaded ends of the test specimen were parallel to each other. Specimens used
for testing were randomly selected from all parts of a large slab (approximately I m by I m by
10 em). Load was applied by a constant displacement of O.<J5 mm/s (0.0375 in./sec). Figs. 2(a)
and 2(b) show the typical compression test results of the foams of various densities.
Additional compression tests were performed to facilitate comparison between the uniaxial
compression and the confined compression tests for the O.OX- and O.16-g/cm' (5- and IO-pef)
foams. Lateral expansion was measured with a set of callipers at different stages of loading.
These tests were identical to those mentioned above, except for the size of the specimens, which
were 7.1 cm (2.R in.) in diameter, as the specimens reported in the following. The results from
these tests were identical to those from the 10.16-cm (4-in.) cubes.
End effects due to the friction between the loading platen and the specimens were studied
in compression tests performed on the O.08-g/cm.1 (5 pef) foam. The specimens used were more
slender, as is shown in Fig. 3. The three sections were saw-cut and put back together, to reduce
the end effects in the central portion of the specimen. Lateral expansion of the specimen was
measured at points A, S, and C with a set of callipers, at different stages of loading until the
plateau stress was reached (at about 2.67 kN [600 lb/). The results are shown in Fig. 4. Until
TABLE 1 Schedule of Static Tests Reported

Triaxial Compression-Confinement Pressure


(MPa)
Density Uniaxial Uniaxial Hydrostatic
(g/cm 3 ) compression tension compression 0.35 0.52 0.69 1.04
(1 ) (2) (3) (4) (5) (6) (7) (8)
O.O"X 6 6
.4 .... -
...
- -

(!.OX 6
.
6 - -
0.16
0.32
6
6
6
6
.... -
- - -
-
-

530 JOURNAL OF ENGINEERING MECHANICS

J. Eng. Mech., 1995, 121(4): 528-540


// 0.32 //'
,.-
,_ ....- i..-_// Ij
J

0.32 ,;/

7.0 ,-r-
....
!!
J' -;;;
7.0
!
Ii
:' i
1 .
;
'"
!; ,
I

'"'" ,
'"'" 3.5
'"'"..,
...'" 3.5
. 0.16
...'"'" 0.16 '"
'" .048
0.08

0.0 0.\ 0.2 0.:5 0.4 0.5 O.li 0.7 0.8 0.9 1.0 0.0 0.1 0.2 0.) 0.4 0.5 O.li 0.7 0.8 0.9 1.0

STRAIN
STRAIN
Downloaded from ascelibrary.org by Istanbul Teknik Universitesi on 08/17/23. Copyright ASCE. For personal use only; all rights reserved.

FIG. 2. (a) Axial Response from Uniaxial Compression of All Densities of Foams in Parallel Directions; (b) Axial Response from
Uniaxial Compression of All Densities of Foams in Perpendicular Directions

4000 r - - - - - - - - - - - - - - - - ,
3500
d=7.6em I
3000

5.1 em
A 2500

--&- poiIlt A
.§ 1500 "0-- polDt B
10.2 em
B 1000 - . - poiIlt C

500

0.0 0.4 0.8 1.2 1.6


5.1 em
C Lateral displacement (mm)
FIG. 4. Load-versus-Lateral-Deflection Plot for End·
FIG. 3. End-Effect Specimen Effect Tests (0.08 g/cm 3 )

the plateau stress is reached, the lateral expansion of the different parts of the specimen, including
point A near the contact with the platen were reasonably close. Thus, specimen end effects
were minimal during these tests.

Hydrostatic and Triaxial Compression


A triaxial test system (T-1000 series) obtained from Soil-Test Corp., Chicago, was used to
test the foams under hydrostatic and triaxial loads. The T-IOOO series testing system consisted
of a T-Il)]5 triaxial cell, a T-lOOO control panel, and three pressure interface devices (PIDs).
The triaxial cell consisted of a transparent plexiglass cylinder, a metal head plate, and a base
plate. The connections between these individual components is shown in Fig. 5 and Table 2.
The Instron (model 1323) was used for applying axial loads. A maximum confinement pressure
of 2 MPa (300 psi) could be applied with this setup.
Since the apparatus was originally designed for soil testing, it had to be redesigned through
experience to suit the foams. A metal rod 21.6 cm long and 1.9 cm in diameter was used as a
piston for loading the specimen, and was fabricated along with an aluminum base plate 22.9 cm
in diameter, with grooves spaced according to the distance between the legs of the triaxial cell.
This plate was used to provide stability to the cell while load was being applied and to align the
cell on the loading platen of Instron. As the volumetric changes of the specimens were very
large and exceeded the capacity of the burette used for volume-change measurements, a PID
was converted into performing the functions of a burette. The accuracy of the volume measured
with this setup was 8 mL.
All the specimens tested for triaxial and hydrostatic compression were 7.1 cm (2.8 in.) in
diameter and 7.1 cm (2.8 in.) high. The specimen dimensions were dependent on the capability
of the apparatus used. Because of their low density, the specimens would float out of position
during testing, resulting in eccentric loading. Thus, to avoid buckling, the height of the specimens
were restricted to 7.1 cm (2.8 in.). The diameter was based on the limited capacity of the triaxial
cell and the loading piston, designed for a maximum load of 4.45 kN (10 kips). Accordingly.
for maintaining the uniformity of the test results, specimens tested for hydrostatic compression
also had the same dimensions.

JOURNAL OF ENGINEERING MECHANICS 531

J. Eng. Mech., 1995, 121(4): 528-540


2

H--Bj-------{ 4
9
IJ---------{ 5
Downloaded from ascelibrary.org by Istanbul Teknik Universitesi on 08/17/23. Copyright ASCE. For personal use only; all rights reserved.

FIG. 5. Test Equipment Used for Testing Specimens in Hydrostatic and Triaxial Compression

TABLE 2. Test Equipment Used in Fig. 5


Part Name Material
(1 ) (2) (3)
I Piston Steel
2 Head plate Aluminum
J Linear ball bushings Stainless stecl
4 Cylinder Transparent plastic
5 Tic rod Stainless steel
6 Specimcn Foam
7 Specimen basc Steel
X Base plate Aluminum
<) Specimen cap Plastic

A core cutter with an internal diameter of 7.1 cm in.) was used to prepare cylindrical
specimens. The porous specimen was then made waterproof by placing it in a triaxial membrane
whose one end was clamped on to the lower platen and the other end on to the upper plexiglass
platen. Care was taken to avoid the eccentricity of loading applied through the 1.9-cm- (0.75-
in.-) diameter piston.
All the densities of foams were tested in duplicate in each direction. The pressure applied
and the volume changes of the specimen were monitored manually using the pressure gages and
burettes.

Hydrostatic Compression
Hydrostatic compression was achieved by applying a uniform pressure at the rate of 0.035
MPa/min (5 psi/min) for and O.08-g/cm 3 (3- and 5-pcf) foams, and at the rate of 0.07
MPa/min (10 psi/min) for 0.16- and O.32-g/cm' (10- and 20-pcf) foams. The pressure was applied
by compressed air which pushed water through a pressure interface device into the triaxial cell.
The rate at which the pressure was applied was determined by a trial-and-error process of testing
the lowest- and highest-density foams and observing the drop in water level in the burette used
for measuring volume changes. Due to the viscoelastic behavior of the material, the water level
dropped gradually and continuously. Therefore, a certain amount of time had to be allowed for
the water level to stabilize. One minute was optimum for the water level to stabilize.
The specimen to be tested was placed on the base plate, then covered by a rubber membrane.
A plexiglass platen was used on the top to keep the specimen in place. The rubber membrane
had to be fastened to the base plate and the platen had to be fastened with gaskets. But. due
to insufficient water rightness provided by the manufacturer's rubber gaskets at high pressures.
aluminum clamps along with silicon grease were used to clamp the rubber membrane to the
upper and lower plates. Silicon grease was applied to the interiors of the membranes where
they were to be clamped. After the specimen was rendered watertight, the triaxial cell was
sealed and the specimen was held in place with a piston.
The cell was pressurized with water at the rate given earlier. Volume changes were monitored
at each increment of pressure. Pressure was increased until the foam specimen locked up (the
cells collapsed and came in contact with each other, leading to increasing strength). The lockup

532 JOURNAL OF ENGINEERING MECHANICS

J. Eng. Mech., 1995, 121(4): 528-540


pressure could not be achieved for the 0.32-g/cm J (20-pef) foam due to limitations on the pressure
capacity of the cell. Results of the hydrostatic tests have been presented in Table 3. A typical
pressure vs. volumetric strain plot has been shown in Fig. 6.

Triaxial Compression
The explained procedure was also used for the application of triaxial load. However, instead
of increasing the pressure gradually, a predefined confinement pressure was applied at the
beginning and then the specimen was loaded in uniaxial compression. The predefined pressure
applied was 172 kPa (25 psi) for (J.()48-g/cm J (3-pef) foam, 345 kPa (50 psi) and 517 kPa (75
psi) for the 0.08-g/cm' (5-pef) foam, 690 kPa (100 psi), and 1,034 kPa (150 psi) for the 0.16-
g/cm' (IO-pcf) foam, and 1,034 kPa (150 psi) for the O.32-g/cm' (20-pef) foam. After the ap-
plication of pressure additional uniaxial load was applied using the Instron loading machine,
the rate of head movement being 0.076 mm/s (0.003 in./sec). Data acquisition and the analysis
Downloaded from ascelibrary.org by Istanbul Teknik Universitesi on 08/17/23. Copyright ASCE. For personal use only; all rights reserved.

of the acquired data was then done in the same manner as described in the compression and
tensile test procedures. Table 4 summarizes the triaxial test results. Note that the axial stresses
tabulated there are the deviatoric stresses.

ANALYSIS OF EXPERIMENTAL DATA


Uniaxial Compression
The experimental results were analyzed with formulae reported by Gibson and Ashby (1988)
based on the mechanics of deformable bodies. The following equation describes the variation
of the elastic modulus with density, specifically for isotropic closed-cell foams:

E = <p 2 + (l _ <p) + PII(l - 211) (I)


E, p, p, E, (1 _
p,

TABLE 3. Comparison between Maximum and Minimum Peak Hydrostatic Pressure Obtained from
Testing Two Polyurethane Foam Specimens in Each Direction

Minimum peak Maximum peak Average peak


Density pressure pressure Difference pressure
(g/cm 3 ) (MPa) (MPa) (%) (MPa)
(1) (2) (3) (4) (5)
0.04R' 0.276 0.276 0.0 0.276
0.04X" 0.311 0.311 0.0 0.311
O.OX' 0.725 0.794 X.6 0759
O.OX" 0.725 0.725 0.0 0.725
0.16' I.X29 l.R29 0.0 l.X29
0.16" 1.794 1.932 7.1 I.R63
"PerpendICular direction.
"Parallel direction.

2.0 . ... -.-'- "


: .'

:
j.
J-:
III
:.
""
:>:
1.0
:;
::
..
'"
en
J.
;) ... ---------
,
'"
""'"
a 4 Jf
oz f--
OL.----'-_..L--..JL..--L_-'---'_-'-_-'---'_--l
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
VOLUMETRIC STRAIN
FIG. 6. Volumetric Response from Hydrostatic Compression of All Densities of Foams

JOURNAL OF ENGINEERING MECHANICS 533

J. Eng. Mech., 1995, 121(4): 528-540


TABLE 4. Comparison between Maximum and Minimum Peak Stress Obtained from Triaxial Test of
Two Foam Specimens in Each Direction
Confinement Minimum peak Maximum peak Average peak
Density pressure stress stress Difference stress
(g/cm 3 ) (MPa) (MPa) (MPa) (%) (MPa)
(1 ) (2) (3) (4) (5) (6)
(l.04S' o.ln 0.191 0.207 7.h 0.19')
O.04X" o.ln OAm OAm (UI 0.403
O.OX' (U45 O.XI2 O.X71 6.X O.X43
O.OX" 0.345 OAI4 OAI4 0.0 (1.414
(l.OX" O.5IX 0629 0.69 X.X O.hh
(l.OX" O.5IX 0.254 0.264 3.7 0.259
O. I (y' 0.69 IAX9 1.501 O.X 1.495
O.U)" O.h9 UX 1.4h I 5.5 1.421
Downloaded from ascelibrary.org by Istanbul Teknik Universitesi on 08/17/23. Copyright ASCE. For personal use only; all rights reserved.

O.th' 1.04 1.299 UX 5.9 1.339


0.16" 1.04 1.15 1.21X 5.5 1.IX4
"PerpendIcular directIon.
"Parallel direction.

0.1
y = 0.975x2 + 0.057x - 0.000

0.D75 0.D75

0.05 0.05
u.J w

0.025 0.025

0
o 0 N
V1
o
o
V1
0
0 0 0
0
V1
N
0
""
0

p·tps
FIG. 7. (a) Correlation between E and Density in Parallel Direction; (b) Correlation between E and Density
in Perpendicular Direction

where p and p, = the densities of the foam and the solid polyurethane; E, = the elastic modulus
of the solid polyurethane; Po = the atmospheric pressure; v = the Poisson's ratio of the foam;
and <!J = a parameter describing the proportion of the foam's constituent material that is located
in the cell edges. The three terms in (1) account for the contributions of the cell edges, the cell
walls, and the internal fluid pressure, respectively. As we shall show later, the contribution of
the internal air pressure can be neglected in our foams. Experimental data for E, both parallel
and perpendicular to the rise direction, are plotted in Figs. 7(a) and 7(b). The second-order
interpolation curves with the excellent correlation coefficients have also been shown in these
figures. Gibson and Ashby (1988) had suggested a value of <\> = 0.6-0.8 for closed-cell foams.
Since our material is anisotropic, the <!J values obtained from the correlation curves (<\>2 = 0.438
and 0.975 for perpendicular and parallel loading, respectively) are different in the two directions,
<!J = 0.66 and <!J = 0.99).
Note that the high value <\> = 0.99 (parallel to the rise direction) indicates that the relationship
is identical to that observed in open-cell foams, where

(2)

and there is no material in the cell faces in that direction, i.e., (<!J 1). A smaller value of <!J
perpendicular to the rise direction indicates more cell-wall material along those directions. This
observation is in keeping with the anisotropy of the material, and why specimens were stronger
in the perpendicular (two or three) direction, in comparison to the out-of-plane (one) direction.
The variation of plateau stress (Tpl with density p, was given for closed-cell isotropic foams by
the semiempirical equation as:

(TId

(II,
= c, (£.) 'c + c, (£.) +
p\
P" - P",
(T\\
(3)

where (Tn and p, = the yield strength and the density of the solid polyurethane, and were taken
as 127 MN/m 2 and 1.2 g/cm" respectively. The last two terms in (3) arc unique to closed-cell

534 JOURNAL OF ENGINEERING MECHANICS

J. Eng. Mech., 1995, 121(4): 528-540


foams, and only the first term is needed for open-cell foams. The second term incorporates the
membrane stresses on the cell-walls. The last term accounts for the entrapped air (PI) and Po,
are the initial internal pressure and the atmospheric pressure respectively). We shall demonstrate
later in the section on analysis of dynamic test data, that the effect of entrapped air is negligible
for our foams, so the last term is O. Fig. 8 shows the curve-fitting of the experimental data. A
constant C 1 = 0.4 provides a very high level of correlation (r 1 = 0.997) to the experimental
data. The second term of (3) is nonessentiaL i.e., the relationship is analogous to that of open-
cell foams. In Fig. 8, the value of (f", plotted is the average a", for the two directions of anisotropy.
The linear relationship for the densification strain En versus density p is given as:

En = 1 - 1.4 (4)

Fig. 9 demonstrates that such a linear fit may not be appropriate, and a higher-order equation
will fit the experimental data better.
Downloaded from ascelibrary.org by Istanbul Teknik Universitesi on 08/17/23. Copyright ASCE. For personal use only; all rights reserved.

Traditionally, the lockup strain is defined as the strain at which the stress-strain diagram is
almost vertical. Our lockup strain En (Fig. 9) is defined quite differently, as the strain at which
the stress level is 1.5 times the original plateau stress level. It was not experimentally possible
to get a vertical stress-strain curve and absolute lockup in our materials due to the limitations
of the testing machine's capacity. Furthermore, the engineering motivation of this project was
to assist in the design of nuclear-waste transportation casks, and to limit the stresses transmitted
into the cask in the case of a hypothetical accident. Therefore, a safety factor of 1.5 determined
the definition of En, rather than the conventional definition of absolute lockup where the stresses
transmitted are too large and therefore catastrophic to the cask containment.
The reported analysis demonstrates that the various semiempirical equations available in the
literature can qualitatively describe the uniaxial properties as a function of the density. However,
more accurate expressions can be developed specific to individual type of foams. This observation
also provides justification for determining the properties relevant to design from experiments.

Multiaxial Loading
The polyurethane foams that we tested under various loading paths showed either a negative
dependency or no dependency of the yield stress on the confining pressures. Mohr's circle plots
shown in Fig. 10 demonstrate the typical response of 0.08-g/cm.1 (5-pcf) foams under varying
confining pressures. The peak stress is quite independent of the confinement pressure, a phe-
nomenon not observed in conventional civil engineering materials such as concrete or soils. This
is because lateral pressure facilitates buckling of the cell walls, thereby reducing the axial load
necessary to cause failure (plateau stress). Gibson et al. (1989) have demonstrated that under
multiaxial loading elastic buckling precedes plastic collapse, defining portions of the failure
surface. This observation can also explain the phenomenon discussed above. The hydrostatic
test results shown in Fig. 6 demonstrate significant reduction in volume with increasing hydro-
static load. Therefore, conventional failure theories (Von Mises or Tresca) based solely on the
deviatoric stresses, which are appropriate for other materials, are not appropriate for foams.
Triantafillou and Gibson (1990a) have developed a plasticity based failure criterion for iso-

0.75.-----------------,
y = -0.779x + 0.687. = 0.778
0.06 r - - - - - - - - - - - - - - - - - ,
1 = 0.4 XI.5
0.7

0.05 .:a • O.lIll7
rn 0.04
w
o
"-
Q. 0.03

0.02

0.01

o.00
0.00 0.05 0.10 0.15 0.20 0.25 0.30
pips (P*/Ps)
FIG. 8. Correlation between (J'p' and p FIG. 9. Lockup Strain as Function of Density

JOURNAL OF ENGINEERING MECHANICS 535

J. Eng. Mech., 1995, 121(4): 528-540


.8

.6,..------------=----

•5
•..................... .............
....... poroIIel

.................
i/

.2 IV ;/V "'--;

( ( \ \ .1
Downloaded from ascelibrary.org by Istanbul Teknik Universitesi on 08/17/23. Copyright ASCE. For personal use only; all rights reserved.

o ,
o
o .2 .4 .6 ·8 1-0 ·5 1-0 1.5 2D 2.5
PriDcipII streII (1II'a) 11 (MPa)
FIG. 10. Mohr Circles for 5-pcf Foam (a 2 -Perpendicular Direction) FIG. 11. Plot of J 2 versus " for O.OS-g/cm 3 Foam

tropic open-cell foams. The theoretical yield function used by them involved a linear relationship
between v'l'2 and II of the form:

f = \/3Y1, + 0.09 -
---- (P)(f)
--'- - I = 0 (5)
(T,d p, (1/){

where II = the first stress invariant (II = (TIl + IT'2'2 + IT,_,). (TI " = the plateau stress level of
the foam; and 1 2 the second invariant of the deviatoric stresses given by
1
12 = 2: Si,Si' (6a)

where Si, the deviatoric stress tensor. Si; = (Ti, - (T"Oi;' An alternative form for 1 2 is

Fig. II once again demonstrates the axisymmetric anisotropy of the data obtained from our
experiments plotted in the V'll and II space. The data with (TIl in the out-of plane direction
approximate a linear relationship similar to that discussed earlier. However. in the in-plane
direction. a parabolic curve-fitting is more appropriate. This approach is similar to the cap model
in the deviatoric stress space used to capture this behavior of soils and geomaterials under high
confinement pressure (Desai and Siriwardane 1984) _Gibson et al. (1989) and Triantafillou et al.
(1989) have developed multi axial failure criteria for anisotropic foams. The purpose of our
analysis was to demonstrate the feasibility of using a simple coordinate transformation process.
and to obtain the failure surfaces in terms of lode parameters to be described next.

Anisotropy
It is evident that all the foams tested were anisotropic. Therefore isotropic elasticity and a
failure criterion based on stress invariants is not appropriate (LeMaitre and Chaboche 19(0).
An anisotropic yield criterion such as the Hill criterion does not take into account the Bauschinger
effect. i.e .. the yield surface is different in tension from that in compression. This is particularly
true for foams that are plastic or fail by clastic buckling under compression but are brittle under
tension _
The foam is isotropic in the two-three plane. In direction I. which is the rise direction. it is
anisotropic. For this class of materials. the flexibility matrix is as follows:

(7)

where £1 = E,: VI::' = VI3: V21/£::, = v,)£,: V"I£3 = vn/£,: and v,"IE, = vl,l£l- In addition.
there are two independent shear modulii G 23 and G I2 = GI." The total number of independent
elasticity parameters. is therefore 6.
Since stress invariants are not appropriate. a coordinate transformation in the principal stress
space was used to analyze the experimental data (Chen and Han 1(87):
IT, - IT, u, + u, - 21T.
CI, = Y2 : CI, =
\/6
: Cf, = P = (Xa-c)

536 JOURNAL OF ENGINEERING MECHANICS

J. Eng. Mech., 1995, 121(4): 528-540


where q, and q2 designate coordinates in the deviatoric plane; and q, the hydrostatic com-
ponent. The Lode angle e and a radius q are defined as follows:
q, = q cos !l and q, = q sin !l (9)
where e = the direction of loading in the deviatoric plane; and q the magnitude. The data
obtained from the uniaxial, hydrostatic, and triaxial tests are shown in Table 5. These stresses
designate the load combination at which the material fails (yielding or elastic buckling) under
different load paths. Fig. 12 shows a plot of q, versus q2 for the 0.08-g/cm·' (S-pef) foam. All
the experiments performed with IT, > (J2 = (J, fall on one straight line, with a lode angle of e
= 300. All the experiments with IT2 > ITI = IT, fall on another straight line, with a lode angle
fl = - 90°. These lines should be 120° apart. A specific lode angle, therefore, corresponds to
a certain loading path. Once the lode angle is established, a failure envelope in the q versus p
space can be used to describe when the material reaches its yield plateau. Fig. 13 shows the
yield surfaces corresponding to the two different lode angles discussed earlier. Similar analyses
Downloaded from ascelibrary.org by Istanbul Teknik Universitesi on 08/17/23. Copyright ASCE. For personal use only; all rights reserved.

were performed for the 0.16-g/cm (Ill-pcf) foam, and are shown in Figs. 14 (lode angles) and
15 (failure envelopes). The available experimental data were fit with a second-order polynomial
to obtain the failure envelopes. A more extensive experimental program could be used in the
future to refine these equations.

INVESTIGATION OF RATE DEPENDENCY


Dynamic Tests
Slow-rate dynamic testing (strain rate of lis) was done on the Instron using the same specimens
and instrumentation as in the static compression tests. The constant strain rate (based on en-
gineering strain) was achieved through the displacement control mode of the test machine.
TABLE 5. Summary of Static Test Results

If 11 U 22 U 33

Test (MPa) (MPa) (MPa)


(1) (2) (3) (4)
(1I) O.OH g/em'
I O.73H O.73H O.73H
2 0.776 0.51H 0.71H
3 0.75') 0.345 0.345
4 0.71H 0.0 0.0
5 0.51H 1.17 O.51H
6 0.345 1.19 0.345
7 0.0 0.991 0.0
(b) 0.16 g/cm'
I.H49 I.H49 I.H49
2 2.219 1.035 1.035
3 2.11 0.69 0.69
-1 265 0.0 0.0
5 I.C)J5 2.374 1.035
6 0.69 2.IH5 0.69
7 0.0 2.262 0.0
Note: 1T IS in parallel direction: IT" and IT.,; are in perpendicular direction.
"

1000
400

200 800 - 6

0
1 •
800
• .::>
g
-200 ... 400

-400 •'3 _ 2
5 200
-600
6
-600
0
7
100 200 300 400 500 600
o
o
,
100 200 300 400 500 600 700- f
800
ql (kPa) p(kPa)
FIG. 12. q, versus q2 Plot for O.08-g/cm 3 Foam FIG. 13. q versus p Plot for O.08-g/cm 3 Foam

JOURNAL OF ENGINEERING MECHANICS 537

J. Eng. Mech., 1995, 121(4): 528-540


1000 ,------,-----,-----,-----,------,----,------, 2000

3 ·7
500
1500
1
0 4 E

IIJ I... 1000 ..•• .5


_.
-1000
6 500 •

-1500

7
1
Downloaded from ascelibrary.org by Istanbul Teknik Universitesi on 08/17/23. Copyright ASCE. For personal use only; all rights reserved.

-2000 I i
0 200 400 600 800 1000 1200 1400 250 500 750 1000 1250 1500 1750 2000
ql (kPa) P (kPa)
FIG. 14. q, versus q2 Plot for O.16-g/cm 3 Foam FIG. 15. q versus p plot for O.16-g/cm 3 Foam

TABLE 6. Comparison of Static and Dynamic Results of Polyurethane Foams in Uniaxial Compression

Dynamic Dynamic
Static peak Static plateau Dynamic (1.0/s) Dynamic (1.0/s) (100.0/s) (100.0/s)
Density stress stress peak stress plateau stress peak stress plateau stress
(g/cm 3 ) (MPa) (MPa) (MPa) (MPa) (MPa) (MPa)
(1 ) (2) (3) (4) (5) (6) (7)
O.04N·' 0.435 (U73 0.569 O.4N3 0.69 05N7
O.04N" 0.302 0.279 0.345 0.276 No pe,lks O.42N
(l.OR' O.99[ O.N56 1.173 0.966 ISi3 1.20N
O.ON" O.7IN O.66N O.N56 O.N3N I. 00 I o<n2
O.I(y' 2.26 2.102 No peaks 2.346 4.092 345
0.16" No peaks I.N63 No peaks 2.20N No peaks 2.174
0.32' No peaks 6.969 No peaks 9.074 11.351 11.126
(U2" No peaks 7.776 No peaks 9.591 IO.N6N 10095
Note: Stress implies deviatorie stress.
"Perpendieular direetion.
"Parallel direetion.

Higher strain-rate testing (rate of about 100/s) was performed on an instrumented Charpy
impact tester using 2.54-cm (I-in.) cylindrical specimens. It is not possible to maintain a constant
strain rate during such tests, since the hammer is slowed down after initial impact. The size of
the specimen and the hammers drop height dictate the initial stain rate. The reported rate is
only the average rate during the loading history, obtained from the rotational velocity data
described in the next paragraph. The load signals for the impact processes were measured by
a load cell (Kistler 9041 A). The positioning of the load cell is not critical in such soft materials
since the mass and inertia effects of the foam can be considered negligible (Hinckley and Yang
1(75).
The displacements were inferred from a rotational variable differential transformer (RVOT)
(Schaevitz 30A) that measured the rotational movement of the impact hammer. Linear variable
differential transformers (LVOl's) used in the early part of the project were unsuccessful. Since
the specimens undergo large displacements, miniature LVOl's could not be used. and larger
LVOl's had excessive inertial effect. A Nicolet (4094-C) digital oscilloscope was used for data
acquisition. The mass of the loading hammer was adjusted for the various densities of materials
tested. This was necessary due to the large variation in the energy absorption capacity of the
four different densities tested. For lower densities it was necessary to prevent damage to the
load cell; for higher densities, it was necessary to ensure that the impact mass was large enough
to deform the specimens beyond the lockup strain.

Analysis of Dynamic Test Results


Table 6 shows a comparison between the peak and plateau stress levels observed in the foams.
The peak stresses reported here are the arithmatic mean of three specimens tested in each
direction for each density. A significant rate effect was observed on all the different densities
of foams in both anisotropy directions. No particular trend could be observed on the strains at
which the foams lockup and the stress rises again. Rate effect in cellular materials is often
attributed to the escape of air from the cellular voids. However. this rate effect is not typically
expected in closed-cell foams (like the ones reported here) since the walls preclude
escape of air. which is responsible for the rate effect (Gibson

538 JOURNAL OF ENGINEERING MECHANICS

J. Eng. Mech., 1995, 121(4): 528-540


TABLE 7. Effect of Gas Confinement on Plateau Stress
Density Plateau initiation p' (v = 0) p' (v = 0.3)
(g/cm 3 ) strain (MPa) (MPa)
(1 ) (2) (3) (4)
0.041''' 0.09 0.010 0.0031'9
0.041''' 0.10 (Ul116 0.00434
lUll''' O.D7 0.001'11 0.00309
(l.Og" (l.OR 11.00940 0.00355
0.16" 0.07 0.001'79 0.00330
0.16" IUl9 (Ull160 O.OO·B4
0.32" 0.065 0.00973 0.()()36g
Note: p" = 0.1 MPa; p, = 1200 kg/m'.
"Perpendicular direction.
"Parallel direction.
Downloaded from ascelibrary.org by Istanbul Teknik Universitesi on 08/17/23. Copyright ASCE. For personal use only; all rights reserved.

An estimate of the increase in pressure due to the decrease in volume of a cellular material
can be obtained from the following equation (Gibson and Ashby 1999):

P , = P - p" = p" --v:- - 1)


(V,'' = p" [ 1- ,(I - 2,") _ .] ( 10)
E(l - 2v")

where pi = the increase in internal gas pressure due to volumetric compression; p" = 0.1 MPA
= the atmospheric pressure; E = the applied strain; p, = solid density of the foam = 1,200
kg/m''; and v'" and p* = the Poisson's ratio and density of the foam, respectively.
The increasing stress carried by the air during compaction was evaluated. This analysis assumes
that the air is entrapped in the specimens during impact load and is free to escape during static
loading. The actual value of the Poisson's ratio during elastic loading was determined to be
about 0.2g from static tests. and used to evaluate the possible effect of air entrapment. The
expected increase in compressive stress from this analysis are tabulated in Table 7. These numbers
pertain to the situation when the materials first reach the plateau stress level.
As evident from Table 7, the expected rise in stress level due to gas entrapment under such
low strains is almost negligible. Thus. air entrapment cannot be a cause for the rate effect
observed in the foams. Huang and Gibson (1991) demonstrated the viscoelastic behavior of the
foam material in creep tests performed under shear loading. The same inherent rate dependency
of the cell-wall material seems to be responsible for the increase in plateau stresses that we
observed under high-strain-rate tests. The rate dependency of our closed-cell polyurethane foams
is therefore due to the viscoelastic nature of the cell-wall material. rather than due to air
entrapment within the closed cells.

SUMMARY OF OBSERVATIONS
Results of an extensive test program on polyurethane foams of various densities was discussed.
These test data can be used by designers of packaging and structural components. Data from
uniaxial compression tests were evaluated and analyzed with equations developed by other
researchers, based on a strength-of-materials approach.
The response under compression is either insensitive or negatively sensitive to the confinement
pressure. This phenomenon is distinct from what is observed in typical engineering materials.
The deviatoric and mean pressure coordinates can be used to illustrate the effects of anisotropy
and pressure sensitivity on materials such as the polyurethane foams.
The plateau stresses responsible for the impact energy absorption is rate-dependent. This rate
dependency cannot be attributed to the entrapment of air within the cell walls.

ACKNOWLEDGMENTS
This research was funded hy the Department of Energy's Waste Management Education and Research Con-
sortium and by the Sandia National Lahoratories. This paper was presented in part at the ASCE Engineering
Mechanics Conference. held at Texas A&M University in 1992.

APPENDIX I. REFERENCES
Carpenteri. A. (191'6). Mechanical damage and crack grow/h in concye/e. Martinus Nijhoff Publishers. the Neth-
erlands.
Chan. R .. and Nakamura. M. (1969). "Mechanical properties of plastic foams." I Cellular Maler.. Mar./Apr ..
112-111'.
Chen. W. F .. and Han. D. J. (191'7). Plas/icily J()r s/mc/ural mgineering. Springer-Verlag. 110.
Desai. C. S .. and Siriwardane. H. J. (191'4). Cons/ilu/ive laws .f(Jr mgineering m{lIerials wi/h emphasis 01/ geologic
malerials. Prentice-Hall Inc., Englewood Cliffs. N J.

JOURNAL OF ENGINEERING MECHANICS 539

J. Eng. Mech., 1995, 121(4): 528-540


Gent. A. N., and Thomas, A. G. (1959). "The deformation of foamed elastic materials." 1. Appl. !'oll'llwr Sci.,
15, 107-113.
Gibson, L. J. (19S9), "Modelling the mechanical behavior of cellular materials." Maler. Sci. alld ElIgrg" 1\ 110,
1-36.
Gibson, L. J .. and Ashby, M. F. (19SS). Cellular solids. slruclure alld properlies. Pcrgamon Prcss, Ncw York,
N.Y.
Gibson, L. J .. Ashby. M. F., Schagcr, G. S .. and Robertson, C. I. (19S2). "The mcchanics of two-dimcnsional
cellular materials." Proc., Royal Society London, England, A3S2, 25-42.
Gibson. L. J .. Ashby. M. F .. Zhang. J .. and Triantafillou, T. C. (19S9). "Failure surfaccs for cellular matcrials
under multiaxial loads-I. modelling," 1111. 1. Mech. Sci., 3 L 635-663.
Ghlss. R. E .. Neilsen. M. K., Donald, S .. and Maji, A. K. (1990). Sialic leslillg or aillmillllm IWllevcolllhs alld
polVlIrelhalle .f{Jllms. Sandia National Laboratories, AlbuqueHJue, N.M.
Hinckley, W. M .. and Yang, J. C. S. (1975). "Analysis of rigid polyurcthane foam as a shock mitigator." Exper.
Mee'!z., May. In-IR3.
Huang, .I. S., and Gibson. L. J. (1991). "Creep of polymer foams," 1. Maler. Sci., 26, 637 -647.
Ko. W. L. (1965). "Deformation of foamed elastomers." 1. Cell. Plaslics . .Ian., 45-50.
Downloaded from ascelibrary.org by Istanbul Teknik Universitesi on 08/17/23. Copyright ASCE. For personal use only; all rights reserved.

Lnlerman . .I. M. (1971). "The prediction of the tensile properties of flexible foams." .I. Appl. PolvlI/er Sci .. 15.
693-703.
LeMaitre, J., and Chaboche, J. (1990). Mechallics olsolid malerials. Cambridge Univ. Press. Cambridge. U.K.,
IS3-IS5.
Maji, A. K., Neilsen, M. K., Glass. R. E., and Satpathi. D. (1990). /)I'II(//ilic leslillg or ill/paCllimiler 1IU1laials.
Sandia Nat. Lab., Albuquerque, N.M.
Matonis. V. (1964). "Elastic behavior of low density. rigid foams in structural applications." SPE J.. Sep., 1024-
1030.
Ncilscn. M. K.. Morgan. H. S.. and Krieg. R. D. (19S7). "A phcnomcnological constitutivc modcl for low density
polyurethane foams." SAND8n-2927, Sandia Nat. Lab., Albuqucrquc, N.M.
Patel, M. R., and Finnic. I. (1970). "Structural fcatures and mechanical propertics of rigid ccllular plastics."./.
Maler.. 909-932.
"Safcty analysis report for thc TRUPACT-II Shipping Package." (19S9). NRC Dockct No. 71-92\ S. Nuclcar
Packaging Inc., Federal Way. Wash.
"Sandwich construction and core materials, general test methods." (1967). MIL-STD-401 B.
Shaw, M. C., and Sata. T. (1966). "The plastic behavior of ccllular matcrials," 1111. 1. Mech. Sci., S. 469-47S.
"Standard tcst mcthod for compressive propcrties of rigid cellular plastics." (1990). AlIlIlIal hook or ASTM
\/(Illdards, Vol. OR.02, Plastics (II): ASTM. Philadelphia, Pa., D 1601- D3()<)9.
Triantafillou, T. c., Zhang, 1.. Shercliff. T. L., Gibson. L. 1.. and Ashby. M. F. (19R9). "Failure surfaccs for
ccllular materials undcr multiaxial loads- II. Comparison of modcls with cxpcrimcnt." 1111. J. Mee'!z. Sci.. 31.
665-67S.
Triantafillou. T. C., and Gibson. L. J. (1990a). "Constitutive modeling of clastic-plastic open-ccll foams." ./.
Ellgrg. Mech., ASCE. 116( 12). 2722-2nR.
Tri'lIltafilloll. T. C., and Gibson. L. J. (1990b). "Multiaxial failure criteria for brittle foams." 1111 . ./. Mnh. Sci.,
32. 479-496.

APPENDIX II. NOTATION


The following symbols are lIsed in this paper:

E elastic modulus;
E, elastic modulus of solid polyurethane;
I, first stress invariant;
12 second invariant of deviatoric stresses;
p" initial internal pressure in foam;
p, atmospheric pressure;
qi transformed stress vectors;
Si, deviatoric stress tensor;
V, volume of pores;
V,,, original volume of pores;
F strain;
f /} densification strain;
e lode angle;
v'" poisson's ratio of foam;
p density of foam;
p' density of foam;
p, density of solid polyurethane;
u" stress tensor;
(T", plateau stress level of foams;
(T" yield strength of solid polyurethane; and
<I> parameter defining cell geometTY·

540 JOURNAL OF ENGINEERING MECHANICS

J. Eng. Mech., 1995, 121(4): 528-540

You might also like