ASM Handbook, Volume 8: Mechanical Testing and Evaluation Copyright © 2000 ASM International®

H. Kuhn, D. Medlin, editors, p143–151 All rights reserved.

DOI: 10.31399/asm.hb.v08.a0003265 www.asminternational.org

Uniaxial Compression Testing

Howard A. Kuhn, Concurrent Technologies Corporation

COMPRESSION LOADS occur in a wide materials, particularly along the direction of re- forms, as shown in Fig. 1(c), smaller L/D ratios
variety of material applications, such as steel inforcement, is much different in compression lead to a single barrel, as in Fig. 1(d). Barreling
building structures and concrete bridge sup- than in tension. indicates that the deformation is nonuniform
ports, as well as in material processing, such as In this article, the characteristics of deforma- (i.e., the stress and strain vary throughout the
during the rolling and forging of a billet. Char- tion during axial compression testing are de- test specimen), and such tests are not valid for
acterizing the material response to these loads scribed, including the deformation modes, measurement of the bulk elastic and plastic
requires tests that measure the compressive be- compressive properties, and compression-test properties of a material. Barreling, however,
havior of the materials. Results of these tests deformation mechanics. Procedures are de- can be beneficial for the measurement of the lo-
provide accurate input parameters for product- scribed for the use of compression testing for calized fracture properties of a material, as de-
or process-design computations. measurement of the deformation properties and scribed in the section “Instability in Compres-
Under certain circumstances, compression fracture properties of materials. sion” of this article.
testing may also have advantages over other If the compression test can be carried out
testing methods. Tension testing is by far the without friction between the specimen and com-
most extensively developed and widely used pression platens, barreling does not occur, as
test for material behavior, and it can be used to Deformation Modes in shown in Fig. 1(e), and the deformation is uni-
determine all aspects of the mechanical behav- Axial Compression form (homogenous). For measurement of the
ior of a material under tensile loads, including bulk deformation properties of materials in com-
its elastic, yield, and plastic deformation and its pression, this configuration must be achieved.
Compression tests can provide considerable A final form of irregular deformation in axial
fracture properties. However, the extent of de-
useful information on plastic deformation and compression is an instability that is the antithe-
formation in tension testing is limited by neck-
failure, but certain precautions must be taken to sis of necking in tension. In this case, the insta-
ing. To understand the behavior of materials
assure a valid test of material behavior. Figure
under the large plastic strains during deforma-
1 illustrates the modes of deformation that can
tion processing, measurements must be made occur in compression testing. The buckling
beyond the tensile necking limit. Compression mode shown in Fig. 1(a) occurs when the
tests and torsion tests are alternative ap- length-to-width ratio of the test specimen is
proaches that overcome this limitation. very large, and can be treated by classical anal-
Furthermore, compression-test specimens are yses of elastic and plastic buckling (Ref 4).
simpler in shape, do not require threads or en- These analyses predict that cylindrical speci-
larged ends for gripping, and use less material mens having length-to-diameter ratios, L/D,
than tension-test specimens. Therefore, com- less than 5.0 are safe from buckling and can be
pression tests are often useful for subscale test- used for compression testing of brittle and duc-
ing and for component testing where ten- tile materials. Practical experience with ductile
sion-test specimens would be difficult to materials, on the other hand, shows that even
produce. Examples of these applications in- L/D ratios as low as 2.5 lead to unsatisfactory
clude through-thickness property measure- deformation responses. For these geometries,
ments in plates and forgings (Ref 1), weld even slightly eccentric loading or nonparallel (a) (b) (c)
heat-affected zones, and precious metals (Ref compression plates will lead to shear distortion,
2) where small amounts of material are avail- as shown in Fig. 1(b). Therefore, L/D ratios less
able. than 2.0 are normally used to avoid buckling
In addition, characterizing the mechanical and provide accurate measurements of the plas-
behavior of anisotropic materials often requires tic deformation behavior of materials in com-
compression testing. For isotropic polycrys- pression.
talline materials, compressive behavior is cor- (d) (e) (f)
Friction is another source of anomalous de-
rectly assumed to be identical to tensile behav- formation in compression testing of ductile ma- Modes of deformation in compression. (a)
ior in terms of elastic and plastic deformation.
Fig. 1 Buckling, when L/D > 5. (b) Shearing, when
terials. Friction between the ends of the test
However, in highly textured materials that de- specimen and the compression platens con- L/D > 2.5. (c) Double barreling, when L/D > 2.0 and fric-
tion is present at the contact surfaces. (d) Barreling, when
form by twinning, as opposed to dislocation strains lateral flow at the contact surfaces, L/D < 2.0 and friction is present at the contact surfaces.
slip, compressive and tensile deformation char- which leads to barreling or bulging of the cylin- (e) Homogenous compression, when L/D < 2.0 and no
acteristics differ widely (Ref 3). Likewise, the drical surface. Under these circumstances, for friction is present at the contact surfaces. (f) Compressive
failure of unidirectionally reinforced composite L/D ratios on the order of 2.0, a double barrel instability due to work-softening material

Downloaded from http://dl.asminternational.org/handbooks/book/chapter-pdf/479841/a0003265.pdf

by Lawrence Livermore Natl Labs user
 144 / Tension, Compression, Bend, and Shear Testing

bility occurs due to work softening of the mate- from voids that initiate at the center of the neck the material, as described in the section “Insta-
rial and takes the form of rapid, localized region. The fracture strength of a brittle mate- bility in Compression” in this article.
expansion, as shown in Fig. 1(f). rial in tension, on the other hand, is limited by
its cleavage stress.
In compression of a ductile material, necking
does not occur, so the void generation and Plasticity Mechanics
Compressive Properties growth mechanism that leads to complete sepa-
ration in the tension test does not terminate the Further understanding of the axial compres-
The bulk elastic and plastic deformation compression test. Ductile fractures can form, sion test can be obtained by examining the in-
characteristics of polycrystalline materials are however, on the barreled surface of a compres- teractions between the plastic flow and forces
generally the same in compression and tension. sion specimen with friction. These fractures acting during the test. The essential features
As a result, the elastic-modulus, yield-strength, generally grow slowly and do not lead to com- of this interaction can be developed by consi-
and work-hardening curves will be the same plete separation of the specimen, so the dering a thin, vertical slab of material in a
in compression and tension tests. Fracture load-carrying capacity of the material is not compression-test specimen (Fig. 2a). Pressure,
strength, ultimate strength, and ductility, on the limited. As a result, there is no definition of P, from the compression platens acts on the top
other hand, depend on localized mechanisms of fracture strength in compression of ductile ma- and bottom of the slab. Because this slab is to
deformation and fracture, and are generally terials. Surface cracks that may form on the the right of the centerline, the slab moves to the
different in tension and compression testing. barreled surface of compression tests with fric- right as the compression test progresses. Mo-
Anisotropic materials, such as composite mat- tion depend not only on the material, but also tion of the slab to the right, coupled with the
erials and highly textured polycrystalline ma- on the amount of friction and the L/D ratios of pressure from the platens, causes friction, f, on
terials, also exhibit considerable differences the specimen, as described in the section the top and bottom surfaces of the slab. The di-
between tensile and compressive behaviors be- “Compression Testing for Ductile Fracture” in rection of friction on the slab is to the left, op-
yond initial elastic response. this article. posing the motion of the slab.
Measurements of bulk elastic modulus and In compression of a brittle or low-ductility Extracting the slab from the compression
yield strength require accurate measurements material, however, fracture occurs catastrophi- test, shown in Fig. 2(b), it is clear that the fric-
of the axial strain of the material under com- cally by shear. The failure either occurs along tion forces on the top and bottom of the slab
pression testing. This is accomplished by at- one large shear plane, leading to complete sep- cause an imbalance of forces in the horizontal
taching to the specimen an extensometer, aration, or at several sites around the specimen, direction. This implies that there must be inter-
which uses a differential transformer or strain leading to crushing of the material. In either nal horizontal forces acting on the vertical
gages to provide an electronic signal that is case, the load-carrying capacity of the material faces of the slab to maintain force equilibrium
proportional to the displacement of gage marks comes to an abrupt halt, and the fracture (forces due to acceleration are negligible). As
on the specimen. Extensometers are most easily strength of the material is easily defined as the shown in Fig. 2(b), the resulting horizontal
used in tension testing because tension test load at that point divided by the cross-sectional pressure, q, acting on opposite sides of the slab
specimens are long and provide ample space area. must differ by some amount, dq, to achieve
for attachment of the extensometer clips. Due The ultimate strength of a material in tension equilibrium.
to the limitations noted in the previous section is easily defined as the maximum load-bearing Applying the principle of equilibrium to the
(Fig. 1a and b), compression-test specimens are capacity. In a ductile material, this occurs at the slab in the horizontal direction gives a sim-
considerably smaller in length and make attach- initiation of necking. In a brittle material, it oc- ple differential equation for the horizontal pres-
ment of the extensometer clips difficult. Alter- curs at fracture. Because necking does not oc- sure q:
natively, a differential transformer can be used cur in compression testing, there is no ultimate
to measure the displacement between the com- compressive strength in ductile materials, and
in brittle materials the ultimate compressive dq/dr = –2f / L (Eq 1)
pression platen surfaces. Because the measure-
ment is not made directly on the specimen, strength occurs at fracture. The only exception
however, elastic distortion and slight rotations to this is in materials that exhibit severe work where L is the thickness of the compression-
of the platens during testing will give false dis- softening, in which case, plastic instability test specimen. At the outside edge of the test
placement readings. (Fig. 1f) leads to an upper limit in load-carrying specimen (r = D/2), the horizontal pressure
Measurement of the work-hardening, or capacity, which defines the ultimate strength of must be zero (free surface); therefore, Eq 1
plastic-flow, curve of a material is best carried
out by compression testing, particularly if the P
application, such as bulk metalworking, re- f
quires knowledge of the flow behavior at large
plastic strains beyond the necking limit in ten-
sion testing. In this case, the strains are many F
orders of magnitude larger than the elastic
strains, and indirect measurement of the axial q q + dq
strain by monitoring the motion of the com- L
pression platens is sufficiently accurate. Any
systematic errors caused by elastic deformation
F
of the platens or test equipment are insignificant
compared to the large plastic displacements of D
the compression specimen. f
The fracture strength of a material is much P
different in tension and compression. In ten-
sion, the fracture strength of a ductile material (a) (b)
is determined by its necking behavior, which
Fig. 2 Interactions between plastic flow and forces acting during compression testing. (a) Schematic of a compres-
concentrates the plastic deformation in a small sion test showing applied force F, radial expansion away from the centerline, and a slab element of material
region, generates a triaxial stress state in the in a compression test. (b) Forces acting on the slab. P, pressure from the compression platens; f, friction at the contact
neck region, and propagates ductile fracture surfaces, acting opposite to motion of the slab; q, internal radial pressure in the test specimen

Downloaded from http://dl.asminternational.org/handbooks/book/chapter-pdf/479841/a0003265.pdf

by Lawrence Livermore Natl Labs user
 Uniaxial Compression Testing / 145

shows that q increases from zero at the edge to zones (Ref 6): (a) nearly undeformed wedges at contact surfaces form a low-friction layer be-
positive values inside the test specimen. Fur- the top and bottom (referred to as dead-metal tween these surfaces. However, during compres-
thermore, Eq 1 shows that the rate of increase zones), (b) crisscrossing regions of intense sion of high-strength materials, the interface
of q toward the centerline is larger for high val- shear deformation, and (c) moderately de- pressure between the test specimen and die
ues of friction and low values of specimen formed regions near the barrel surfaces. The se- platens becomes extremely high, and the lubri-
thickness. If f is constant, the internal pressure verity of barreling and the differences in degree cant squeezes out, leaving metal-on-metal con-
distribution is: of deformation between the three regions in- tact, resulting in high friction.
crease as friction at the contact surfaces in- One approach to retaining lubricants at the
q = f (D/L)(1 – 2r/D) (Eq 2) creases. contact surface involves machining concentric
circular grooves into the end faces of the test
specimen (Fig. 5a) (Ref 7). Another approach
which has a peak value at r = 0. was pioneered by Rastegaev and refined by
Finally, the vertical pressure, P, is related to Homogenous Compression for
Plastic Deformation Behavior Herbertz and Wiegels (Ref 8), in which the en-
the internal pressure, q, by the yield criterion tire end face is machined away except for a
for plastic deformation: small rim, as shown in Fig. 5(b). This traps a
Under homogenous-compression conditions small volume of lubricant in the cavity, form-
P = q + σ0 (Eq 3) (frictionless compression), height reduction ing a hydrostatic cushion with nearly zero fric-
and the resulting radial and circumferential ex- tion. This approach was modified by machin-
where σ0 is the yield strength of the material. pansion are uniform throughout the test speci- ing a tapered recess, as shown in Fig. 5(c),
Therefore, P has the same distribution as the ra- men. Furthermore, under these conditions, ra- which reduces the amount of material re-
dial stress, q, plus the material yield strength. dial and circumferential stresses are zero, and moved and diminishes the strain measurement
Integrating this pressure distribution over the the only stress acting is the uniform compres- error. Furthermore this lubricant recess pro-
contact area gives the total force, F. sive stress in the axial direction, as described in vides greater lubrication at the rim where mate-
Schematic plots of the pressure distribution, the previous section. rial movement is greatest. During compression
P, in axial compression are given in Fig. 3. Note Homogenous compression is accomplished testing, radial displacement of the test material
that even though the deformation is uniform at by eliminating friction at the contact surfaces, is zero at the center and increases linearly to the
every point, the compressive stress is not uni- which obviously requires the use of lubricants. outer rim. Evaluations of lubrication practice for
form, but reaches peak values at the centerline. Polishing the ends of the compression-test high- temperature testing have shown that the
The values of this peak pressure increase as specimens as well as the die platens provides tapered lubricant reservoir shown in Fig. 5(c)
friction increases and as the test specimen as- smooth surfaces, and lubricants applied to the leads to the greatest reproducibility (Ref 9).
pect ratio, L/D, decreases. More importantly, if
friction is zero, Eq 2 shows that internal pres-
sure, q, is zero throughout the test specimen.
Then, from Eq 3, P is uniform and equal to σ0.
Frictionless conditions, therefore, must be used
to measure the plastic deformation response of
a material, as described in the next section.
More detailed analysis of the plasticity me-
chanics of axial compression are given in Ref 5.
The analysis given above is strictly valid only
for specimens having very low aspect ratios.
However, the essential roles of friction and ge-
ometry are valid qualitatively for test speci-
mens having large aspect ratios; for these test
specimens, the deformation patterns are very
complex and vary in the thickness direction, as
well as in the lateral direction. A macrograph of (b)
(a)
a compression test cross section, shown in Fig.
4(a), reveals the nonuniformity of internal de- Fig. 4 Internal deformation in compression testing. (a) Macrograph of the internal deformation in a compres-
formation patterns due to friction at the contact sion-test specimen with high-contact surface friction. Source: Ref 5. (b) Schematic representation of the inter-
nal deformation into three zones. I, nearly undeformed wedges at the contact surfaces (dead-metal zones); II, criss-
surfaces. In general, the internal deformation crossing regions of intense shear deformation; III, moderately deformed regions near the bulge surface. Source: Ref 6
depicted in Fig. 4(b) can be described as three

0.003 in. 0.007 in.

P 0.01 in. 0.1 in.
0.010 in.

Angle = 2°
f(D/L) increasing

σ0 f=0
r

Fig. 3 Schematic of pressure distributions, P, in a

compression test. When friction is zero, P is
uniform and equal to the material flow stress, σ0, but in- (a) (b) (c)
creasing friction and decreasing L/D with friction lead to
increasingly nonuniform pressure distributions with peak Compression-test end profiles for lubricant entrapment. (a) Concentric grooves. Source: Ref 7. (b) Rastegaev
values at the centerline.
Fig. 5 reservoir. Source: Ref 8. (c) Modified Rastegaev reservoir. Source: Ref 9

Downloaded from http://dl.asminternational.org/handbooks/book/chapter-pdf/479841/a0003265.pdf

by Lawrence Livermore Natl Labs user
 146 / Tension, Compression, Bend, and Shear Testing

Several high-pressure lubricants are available and dimensional changes of the specimen. In tarded from moving outward while the material
for room-temperature compression tests, in- the latter measurement, it is necessary to re- at the midplane is not constrained. As a result,
cluding mineral oil, palm oil, stearates, and mo- move systematic errors by first carrying out the barreling occurs, as shown in Fig. 1(c) and (d).
lybdenum disulfide. Teflon (E.I. DuPont de compression test with no test specimen in Under these conditions, for a given axial com-
Nemours & Co., Inc., Wilmington, DE) in the place. This provides a load-stroke curve for the pressive strain, the bulge profile provides cir-
form of spray or sheet is also widely used at test-machine load train and measures the com- cumferential strain at the equator that is greater
room temperature and can be used at tempera- pliance of the various elements in the loaded than the strain that occurs during homogenous
tures up to 500 °C (930 °F). For high-tempera- column. Subtracting this compliance from the compression. At the same time, due to the
ture testing of steels, titanium, and superalloys, measured crosshead stroke during a compres- bulge profile, the local compressive strain at
one can use emulsions of graphite, molybde- sion test then provides a more accurate mea- the equator is less than the strain that would
num disulfide, and various glasses. It is impor- surement of the specimen deformation. In any have occurred during homogenous compres-
tant to match the grade of glass and resulting event, if constancy of volume can be assumed sion for the same overall height strain. These
viscosity with the test temperature. for the material being tested, then the cross-sec- surface strain deviations from homogenous
In homogenous compression tests, the plastic tional area can be readily calculated from the compression increase as bulging increases; the
stress-strain curve can be easily calculated by specimen height at any point throughout the severity of the bulge, in turn, is controlled by
measurement of the load, cross-sectional area, test. the magnitude of friction and the L/D ratio of
and height of the specimen throughout the the specimen. Figure 6 illustrates the progres-
test. The test can be conducted incrementally sive change in strain at the bulge surface for
at room temperature wherein the specimen Compression Testing different lubrication and L/D ratios (Ref 10).
height and lateral dimensions are measured for Ductile Fracture These strain combinations lead to tensile
after each increment of deformation. For high- stress around the circumference and reduced
temperature deformation or continuous testing, compressive stress at the bulge equator. There-
the test-equipment load cell and crosshead dis- When friction exists at the die contact sur- fore, compression tests with friction, and con-
placement can be used to determine the load faces, material at the contact surfaces is re- sequent bulging, can be used as tests for frac-
ture. Figure 7 shows compression-test speci-
mens with and without friction. Note that the
compression test with the bulge surface, that is,
with friction at the contact surfaces, has a crack
caused by the tensile stress in the circumfer-
ential direction at the bulge surface. The ho-
mogenous compression specimen, even after
greater height compression, has not bulged;
therefore, there is no tensile stress in the cir-
cumferential direction, and the specimen has
not cracked.
The stress and strain environment at the
bulge surface of upset cylinders suggests that
axial compression tests can be used for work-
ability measurements by carrying out the tests
under a variety of conditions regarding inter-
face friction and L/D ratios. By plotting the sur-
face strains at fracture for each condition, a
fracture strain locus can be generated repre-
senting the workability of the material. Figure 8
illustrates such a fracture locus. Modifications
of the cylindrical compression-test specimen
geometry have been used to enhance the range
of strains over which fracture can be measured
Fig. 6 Progressive change in strain at the bulge surface in compression testing. (a) Strains at the bulge surface of a (Ref 11).
compression test. (b) Variation of the strains during a compression test without friction (homogenous com-
pression) and with progressively higher levels of friction and decreasing aspect ratio L/D (shown as h/d )

Fig. 8 Locus of fracture strains (workability) deter-

Fig. 7 Compression tests on 2024-T35 aluminum alloy. Left, undeformed specimen; center, compression with fric- mined from compression tests with friction.
tion (cracked); right, compression without friction (no cracks) Source: Ref 10

Downloaded from http://dl.asminternational.org/handbooks/book/chapter-pdf/479841/a0003265.pdf

by Lawrence Livermore Natl Labs user
 Uniaxial Compression Testing / 147

Instability in Compression dσ/de = σ/(1 + e) (Eq 4) conducted by techniques described in test stan-
dards, such as:

In tension testing, the onset of necking indi- where σ is true stress, ε is true strain, e is engi- • ASTM E 9, “Compression Testing of Metal-
cates unstable flow, characterized by a rapid neering strain, F is force, and A is area. lic Materials at Room Temperature”
decrease in diameter localized to the neck re- Equation 4 indicates that instability occurs • DIN 50106, “Compression Test, Testing of
gion. Up to this point, as the test specimen when the slope of the true stress–engineering Metallic Materials”
elongates, work hardening of the material com- strain curve equals the ratio of true stress to one • ASTM E 209, “Compression Tests of Metal-
pensates for the decrease in cross-sectional plus the engineering strain. This leads to the lic Materials at Elevated Temperatures with
area; therefore, the material is able to carry an Considére construction for instability (Fig. 9). Conventional or Rapid Heating Rates and
increasing load. However, as the work-harden- The upper part of Fig. 9 shows the Considére Strain Rates”
ing rate decreases, the flow stress acting across construction for a tension test. When the work
the decreasing cross-sectional area is no longer hardening stress-strain curve reaches point C, This section briefly reviews the factors that in-
able to support the applied axial load. At this necking begins and unstable deformation con- fluence the generation of valid test data for tests
point, necking begins and the rate of de- tinues through to complete separation or frac- conducted in accordance with ASTM E 9 and
crease of cross-sectional area exceeds the rate ture. This defines the ultimate strength of the the capabilities of conventional universal test-
of increase of work hardening, leading to insta- material in tension. In the lower part of Fig. 9, ing machines (UTMs) for compression testing.
bility and a rapidly falling tensile load as the the Considére construction for the compression
neck progresses toward fracture. test shows that, for a work softening material, Specimen Buckling
In compression testing, a similar phenome- unstable flow commences at point C ′ , leading
non occurs when work softening is prevalent to a configuration as shown in Fig. 1(f). Thus,
(Ref 12). That is, during compression, the the ultimate strength of the material in com- As previously noted, errors in compressive
cross-sectional area of the specimen increases, pression in this case can be defined as the stress stress-strain data can occur by the nonuniform
which increases the load-carrying capability of at this point. stress and strain distributions from specimen
the material. However, if work softening oc- Materials that undergo severe work softening buckling and barreling. Buckling can be pre-
curs, its load-carrying capability is decreased. are prone to compressive instabilities. While vented by avoiding the use of specimens with
When the rate of decrease in strength of the ma- useful in itself, this precludes measurement of large length-to-diameter ratios, L/D. In addi-
terial due to work softening exceeds the rate of the bulk plastic deformation behavior of the tion, the risk of specimen buckling can be re-
increase in the area of the specimen, an unsta- material, just as the necking instability in ten- duced by careful attention to alignment of the
ble mode of deformation occurs in which the sion testing prevents measurement of plastic loading train and by careful manufacture of the
material rapidly spreads in a localized region, deformation behavior at large strains. Several specimen according to the specifications of
as shown in Fig. 1(f). metallurgical conditions can lead to such work flatness, parallelism, and perpendicularity given
Instability in tension and compression can be softening. These include dynamic recovery and in ASTM E 9. However, even with well-made
described through the Considére construction. dynamic recrystallization where substructure specimens tested in a carefully aligned loading
Instability occurs when the slope of the rearrangements and dislocation reductions lead train, buckling may still occur. Conditions that
load-elongation curve becomes zero, that is: to a rapid decrease in flow stress. Morphologi- typically induce buckling are discussed in the
cal changes in second phases, such as the rapid following sections.
dF = d(σA) = σdA + Adσ = 0 spheroidization of pearlite at elevated tempera- Alignment. The loading train, including the
tures, the coarsening of small spherical precipi- loading faces, must maintain initial alignment
tates, and the coarsening of martensitic sub- throughout the entire loading process. Align-
or ment, parallelism, and perpendicularity tests
structures, are another source of work softening.
Further examples of work softening include in- should be conducted at maximum load condi-
dσ/σ = –dA/A = dε = de/(1 + e) cipient melting of eutectic phases and localized tions of the testing apparatus.
shear-band formation, seen commonly in tita- Specimen Tolerances. The tolerances given
nium alloys. in ASTM E 9 for specimen end-flatness, end-
and
parallelism, and end-perpendicularity should be
considered as upper limits. This is also true for
concentricity of outer surfaces in cylindrical
Test Methods specimens and uniformity of dimensions in
σ
rectangular sheet specimens. If tolerances are
reduced from these values, the risk of prema-
C Axial compression testing is a useful proce- ture buckling is also reduced.
dure for measuring the plastic flow behavior Inelastic Buckling. Only elastic buckling is
Tension and ductile fracture limits of a material. Mea- discussed in ASTM E 9. This may be somewhat
suring the plastic flow behavior requires unrealistic, because for the most slender speci-
frictionless (homogenous compression) test men recommended, the calculated elastic buck-
e conditions, while measuring ductile fracture ling stresses are higher than can be achieved in
limits takes advantage of the barrel formation a test. This specimen has a length-to-diameter
and controlled stress and strain conditions at ratio of 10. An approximate calculation using
the equator of the barreled surface when com- the elastic Euler equation for a steel specimen
Compression
pression is carried out with friction. Axial com- with flat ends on a flat surface (assumed value
C′ pression testing is also useful for measurement of end-fixity coefficient is 3.5) yields a buck-
of elastic and compressive fracture properties ling stress in excess of 4100 MPa (600 ksi); the
1 of brittle materials or low-ductility materials. In comparable value for an aluminum specimen
any case, the use of specimens having large would be 1380 MPa (200 ksi). These values,
L/D ratios should be avoided to prevent buck- however, are not realistic.
Fig. 9 Considére construction showing instability
conditions in tension testing (due to decreasing
ling and shearing modes of deformation. Buckling stress in the above example should
work-hardening rate) and in compression testing (due to Axial compression tests for determining the not be calculated by an elastic formula but
work softening) stress-strain behavior of metallic materials are by an inelastic buckling relation. In terms of

Downloaded from http://dl.asminternational.org/handbooks/book/chapter-pdf/479841/a0003265.pdf

by Lawrence Livermore Natl Labs user
 148 / Tension, Compression, Bend, and Shear Testing

inelastic buckling it has been concluded that throughout the test (Ref 14). A typical compres- lubricant at the loading surface of the specimen
the following relation appropriately calculates sion jig and contact-point compressometer are reduces friction. One such material commonly
inelastic buckling stresses (Ref 13): shown in Fig. 11(a) and (b) respectively. used is 0.1 mm (0.004 in.) thick Teflon sheet.
The action of the lubricant may be enhanced if
 Et  Barreling of Cylindrical Specimens the bearing surfaces that apply the load are hard
S cr = Cπ 2  (Eq 5)
2  and highly polished. The use of tungsten-car-
 ( L / ρ)  When a cylindrical specimen is compressed, bide bearing blocks is recommended for all ma-
Poisson expansion occurs. If this expansion is terials undergoing compression testing. Other
where Scr is the buckling stress in MPa (ksi); C techniques have been used to reduce non-
is the end-fixity coefficient; Et is the tangent restrained by friction at the loading faces of the
specimen, nonuniform states of stress and strain uniformity of stress and strain distributions
modulus of the stress-strain curve in MPa (ksi); along the gage length (Ref 15, 16).
L is the specimen length in mm (in.); and r is occur as the specimen acquires a barreled shape
(Fig. 12). The effect on the stress and strain dis- The contact area between the lateral faces of
the radius of gyration of specimen cross section the specimen and the lateral support guides of
in mm (in.). Equation 1 reduces to the Euler tributions is of consequence only when the de-
formations are on the order of 10% or more. the testing jig must be well lubricated. Person-
equation if E, the modulus of elasticity, is sub- nel engaged in sheet compression testing
stituted for Et. Friction on the loading face causes rollover.
As shown in Fig. 12, points originally on the should become familiar with the literature on
Rearranging Eq 5 to combine the stress- the subject. A selected bibliography on this
related factors results in: sides of the specimen are ultimately located on
the specimen end face. Use of a high-pressure subject is given in ASTM E 9.
2
 1   L =
Et
 2   (Eq 6)
 Cπ   ρ S cr

Note that the value of the right side of Eq 6 de-

creases as stress increases in a stress-strain
curve. In a material with an elastic-pure-plastic
response, the right side of Eq 6 vanishes, be-
cause Et becomes zero, and buckling will al-
ways occur at the yield stress. When the mate-
rial exhibits strain hardening, calculations
using Eq 6 will yield the appropriate specimen
dimensions to resist buckling for given values
of stress.
Side Slip. Figure 10 illustrates one form of
buckling of cylindrical specimens that can re-
sult from misalignment of the loading train un-
der load or from loose tolerances on specimen
dimensions. The ends of the specimen undergo
sideslip, resulting in a sigmoidal central axis.
This form of buckling could be described by Eq
5 and 6, provided an appropriate value of the
end-fixity coefficient can be assigned.
Thin-Sheet Specimens. In testing thin sheet
in a compression jig, approximately 2% of the
specimen length protrudes from the jig. Buckling
of this unsupported length can occur if there is
misalignment of the loading train such that it (a) (b)
does not remain coaxial with the specimen
Fig. 11 Compression testing of thin-sheet specimens. (a) Sheet compression jig suitable for room-temperature or el-
evated-temperature testing. (b) Contact-point compressometer installed on specimen removed from jig.
Contact points fit in predrilled shallow holes in the edge of the specimen.

Fig. 10 Schematic diagram of side-slip buckling. The

original position of the specimen centerline is Fig. 12 Barreling during a test when the friction coefficient is 1.00 at the specimen loading face. Note that as the
indicated by the dashed line. deformation increases, points A, B, and C originally on the specimen sides, move to the loading face.

Downloaded from http://dl.asminternational.org/handbooks/book/chapter-pdf/479841/a0003265.pdf

by Lawrence Livermore Natl Labs user
 Uniaxial Compression Testing / 149

Testing Machine Capacity tain parallel faces during deformation. A typi- measurement to much lower strain rates. When
cal grip assembly is shown in Fig. 14, in which a constant-strain-rate test is desired, deforma-
In a compression test performed to large a compression specimen (5.1 mm, or 0.2 in., tion must be initiated by an impact due to the
strains (e.g., to obtain fracture data), a large long by 5.1 mm, or 0.2 in., diam) is in place and acceleration time required by the ram. This im-
load capacity may be required. For example, ready for testing. The ram is shown in position pact can excite the natural vibrational mode of
consider four medium-length cylindrical speci- and is separated from the subpress by approxi- the load cell, which will produce oscillations in
mens suggested in ASTM E 9, where speci- mately 20 mm (0.8 in.). This gap allows time the output signal that can mask the actual load
mens are specified with diameters that range (approximately 2 ms at the highest ram veloc- measurement. Unless the impact is dampened
from 12.7 to 28.4 mm (0.50 to 1.12 in.) and ity) for the ram to accelerate to the specified ve- by some means, load measurement at strain
with length-to-diameter ratios of 3. Using these locity. In this test, the stroke of the ram must be rates greater than about 1 s–1 can be subject to
specimen sizes, consider the testing of a mate- set accurately to ensure the desired deformation. load-cell ringing.
rial with a yield stress of 1380 MPa (200 ksi) Measurement of Load and Displacement. Ringing of the load cell can be minimized by
and a compression strain-hardening exponent As the strain rate increases, the measurement selecting a load cell with a high vibrational fre-
of 0.05. Figure 13 illustrates the load-capacity of load and displacement becomes increasingly quency. If the natural frequency is sufficiently
requirements to reach a height reduction of more difficult. The requirement for adequate high, the vibrational mode may not be excited
60% for each of the four cylinders recom- frequency response in the signal conditioners by the impact, or if excited, it may be possible
mended in ASTM E 9. The maximum required and the problems associated with load-cell ring- to remove it from the signal with a low pass fil-
load is approximately 3.5 times the load at ing were discussed in the introduction to this ter. Another method to reduce ringing is to
yield. The required capacity for testing the article. In this section, the measurement of load dampen the impact that initiates deformation
same specimens to failure at 60% strain in and displacement at medium strain rates is de- within the specimen. Often, a thin layer of de-
tension would be no more than 1.5 times the scribed in more detail. formable material placed between the impacting
yield loads. Measurement of Load. A typical load cell de- surfaces is sufficient to remove the higher fre-
termines load by measuring displacement in an quencies generated by the impact that can ex-
elastic member, such as a diaphragm or cylin- cite the natural frequency of the load cell. For
Medium-Strain-Rate Testing der. The displacements are measured with bonded example, in the configuration shown in Fig. 14,
strain gages; this gives the load cell sufficient a single loop, approximately 50 mm (2 in.) in
Medium-strain-rate compression testing with intrinsic frequency response for testing at me- diameter, of 0.51 mm (0.02 in.) diameter
conventional load frames is very similar to dium strain rates. However, a problem often lead-tin solder wire placed on the impacting
low-strain-rate compression testing. For me- arises due to ringing in the load cell. The load face of the hydraulic ram was found to be effec-
dium-rate testing, the load frames require the cell has a natural frequency of vibration deter- tive in minimizing load-cell ringing. Such lay-
capability to generate higher crosshead or ram mined by geometry and physical properties, ers, however, may complicate measurement of
velocities. An important consideration is the such as density and elastic modulus. Typical displacement within the specimen.
stiffness of the machine, as discussed in more load cells have a natural frequency in the range At strain rates close to 100 s–1, the standard
detail in the article “Testing Machines and of 500 to 5000 Hz. In effect, the natural fre- load cell either may not possess the necessary
Strain Sensors” in this Volume. For tests at a quency of vibration sets the bandwidth of the frequency response, or it may ring excessively.
uniform strain rate, a high machine stiffness is load-measuring system. These characteristics can make the load cell in-
desired; techniques to increase the stiffness of a By this criterion alone, load cells should be adequate for load measurement. Under these
hydraulic machine are described in Ref 17. sufficient for compression testing at strain rates conditions, a quartz piezoelectric device, such
This section describes some of the techniques as high as 100 s–1. However, the transient re- as a load washer (Fig. 14), is useful. The load
used to obtain medium strain rates with con- sponse of the load cell in practice limits the washer is convenient because it is easily adapted
ventional test frames and additional experimen- to a compression test; it also has excellent in-
tal factors for measurement of load and strain at trinsic frequency response and a high funda-
medium rates. mental vibrational frequency. However, these
Grip design for compression testing at me- devices require special signal conditioning and
dium strain rates requires the same consider- low-capacitance cables.
ations that apply to grip design for low strain Measurement of Strain. The direct measure-
rates. The compression specimen typically is ment of strain at medium strain rates presents a
sandwiched between two hard, polished platens challenge. Many of the devices typically used
that are placed in a subpress designed to main- for low-strain-rate testing are inappropriate at
medium strain rates. Extensometers, for exam-
ple, may have the necessary response charac-
3 teristics for medium-strain-rate testing. How-
600
Load, kip (1000 lbf)

ever, it is difficult to ensure that the rapid and

500 large displacement in small compression speci-
Load, MN

2
A 400 mens will not damage the fragile extensometer.
300
Many hydraulic test frames use a linear vari-
1 B able differential transformer (LVDT) to control
200
D
the motion of the hydraulic ram. This LVDT
C 100 signal is comprised of displacements within the
0 specimen as well as elastic displacements
0 0.1 0.2 0.3 0.4 0.5 0.6 throughout the test frame. To relate this signal
Height reduction ratio (∆L/L) to displacements within the specimen, the latter
Load requirements for compressing speci- contribution must be subtracted; this problem
Fig. 13 mens of various diameters made of a material also is encountered at low strain rates. If a de-
Fig. 14 Subpress assembly for medium strain-rate
with a yield stress of 1380 MPa (200 ksi) and a testing with conventional load frame. The formable material is placed between the impact
strain-hardening exponent of 0.05. Diameters: A = 28.4 specimen, which is 5.1 mm (0.2 in.) diam by 5.1 mm (0.2 surfaces to dampen the impact, the displace-
mm (1.12 in.), B = 25.4 mm (1.00 in.), C = 20.3 mm (0.80 in.) long, is sandwiched between two highly polished
in.), D = 12.7 mm (0.50 in.). Length-to-diameter ratio platens. A quartz load washer is shown positioned above ments within this layer also must be subtracted
(L/D) = 3 the subpress assembly. from the LVDT signal.

Downloaded from http://dl.asminternational.org/handbooks/book/chapter-pdf/479841/a0003265.pdf

by Lawrence Livermore Natl Labs user
 150 / Tension, Compression, Bend, and Shear Testing

Fig. 15 Two views of a 72% compressed specimen of aluminum alloy 7075-T6 displaying orange peel effect. The loading axis is vertical. Extensive macrocracking is evident in
the severely wrinkled surface. Microscopic examination of the surface revealed extensive microcracking in the valleys of the wrinkles. Source: Ref 16

Microcrack to Macrocrack Coalescence.

In some tungsten alloys, the first visible evi-
dence of fracture is a shear macrocrack that ap-
pears at the equator of the specimen after 45 to
50% compressive deformation. However, using
fluorescent-dye penetrant methods, microcrack
initiation was detected at 25% deformation
(Ref 18). For this material, if crack initiation is
the criterion of failure, it is necessary to state
the method of crack detection with the selected
Shear cracks in a 72% compressed specimen of AISI-SAE 4340 steel. The cracks initiated one at a time, parameter for strength.
Fig. 16 starting when the deformation was 30%. Source: Ref 16

ACKNOWLEDGMENTS
A common practice is to mount the LVDT at compressed. The cracks generally initiate on
an off-axis position adjacent to the specimen. the outer surface of the compressed specimen. Portions of this article were adapted from R.
The benefit of this configuration is that a dis- As the specimen is further deformed, the initi- Papirno, Axial Compression Testing (p 55–58)
placement measurement is possible between ated cracks propagate, and new cracks form. and P.S. Follansbee and P.E. Armstrong, Com-
two points that are quite close to the specimen; Some different modes of compression fracture pression Testing by Conventional Load Frames
this measurement includes less of the elastic are described in Ref 18 and some examples are at Medium Strain Rates (p 192–193) in Me-
deformation in the load frame. When a mea- described in the following sections. chanical Testing, Vol 8, ASM Handbook, ASM
surement is made at an off-axis position, it is Orange Peel Cracking. In many materials, International, 1985.
important to verify that the measurement truly roughening or wrinkling of the surface (orange
represents displacements within the sample. peel effect) occurs prior to compressive crack-
Often, two LVDT units are mounted at diamet- ing. This effect is particularly prominent in REFERENCES
rically opposite positions, and their outputs are some aluminum alloys. An extreme example is
processed to eliminate the effects of nonplanar illustrated for an aluminum alloy 7075-T6
1. T. Erturk, W.L. Otto, and H.A. Kuhn, Ani-
motion. The LVDT suffers from an intrinsic specimen in Fig. 15. The specimen is shown af-
sotropy of Ductile Fracture—An Applica-
frequency-response limitation determined by ter 72% deformation. Wrinkling first appeared tion of the Upset Test, Metall. Trans., Vol
the excitation frequency. Standard excitation at 10 to 15% compressive deformation, and 5, 1974, p 1883
frequencies are in the range of 1 to 5 kHz, macrocracking occurred after 50 to 60% defor- 2. W.A. Kawahara, Tensile and Compressive
which limits the frequency response to around mation. Microscopic examination revealed Materials Testing with Sub-Sized Speci-
100 to 500 Hz. many microcracks in the valleys of the wrin- mens, Exp. Tech., Nov/Dec, 1990, p 27–29
Velocity transducers, which have good in- kles, with greatest concentration in the equato- 3. W.A. Backofen, Deformation Processing,
trinsic frequency response, have been used to rial region of the specimen. Defining a com- Addison-Wesley, Reading, MA, p 53
measure the motion of the specimen and grip pression strength or a strain criterion of fracture 4. J.H. Faupel and F.E. Fisher, Engineering
assembly (Ref 17). Their output can be inte- would be difficult for this material. Design, John Wiley & Sons, 1981, p
grated electronically or by computer to obtain Macrocracks in Steel. A case in which 566–592
the displacement. Generally, these also require macrocracks form without apparent precursor 5. G.E. Dieter, Mechanical Metallurgy, 2nd
mounting at off-axis locations. microcracks is shown in Fig. 16. The material ed., McGraw-Hill, 1976, p 561–565
Strain measurement by noncontact methods is AISI-SAE 4340 steel tempered at 204 °C 6. G.E. Dieter, Evaluation of Workability: In-
is becoming more common with optical exten- (400 °F), yielding a hardness of 52 HRC. The troduction, Forming and Forging, Vol 14,
someters or laser interferometers. Laser inter- cracks initiated one at a time and extended ASM Handbook, ASM International, 1988,
ferometers, which are capable of operating at across the surface of the specimen almost in- p 365
high sampling rates, can be used to measure stantaneously. The first cracks appeared when 7. J.E. Hockett, The Cam Plastometer, in Me-
strain at strain rates exceeding 103 s–1. the compressive deformation reached 30%, and chanical Testing, Vol 8, ASM Handbook,
other cracks continued to initiate until the test ASM International, 1985, p 197
was concluded at 72% deformation, which is 8. R. Herbertz and H. Wiegels, Ein Verfahren
Types of Compressive Fracture the condition shown in Fig. 16. The specimen Zur Verwirklichurg des Reibungsfreien
was still intact, and subsequent sectioning re- Zylinderstanch versuchs für die Ermittlung
For all but the most ductile materials, cylin- vealed that the cracks penetrated inward a dis- von Fliesscurven, Stahl Eisen, Vol 101,
drical specimens develop cracks when they are tance of diameter. 1981, p 89–92

Downloaded from http://dl.asminternational.org/handbooks/book/chapter-pdf/479841/a0003265.pdf

by Lawrence Livermore Natl Labs user
 Uniaxial Compression Testing / 151

9. K. Lintermanns Fander, “The Flow and at Elevated Temperatures, Advances in De- 16. R. Chait and C.H. Curll, “Evaluating Engi-
Fracture of Al-High Mg-Mn Alloys at High formation Processing, J.J. Burke and V. neering Alloys in Compression,” Recent
Temperatures and Strain Rates,” Ph.D. dis- Weiss, Ed., Plenum, 1978, p 238 Developments in Mechanical Testing, STP
sertation, University of Pittsburgh, 1984, p 258 13. G. Gerard, Introduction to Structural Sta- 608, ASTM, 1976, p 3–19
10. H.A. Kuhn, P.W. Lee, and T. Erturk, A bility Theory, McGraw-Hill, 1962, p 19–29 17. R.H. Cooper and J.D. Campbell, Testing of
Fracture Criterion for Cold Forging, J. Eng. 14. R. Papirno and G. Gerard, “Compression Materials at Medium Rates of Strain, J.
Mater. Technol. (Trans. ASME), Vol 95, Testing of Sheet Materials at Elevated Mech. Eng. Sci., Vol 9, 1967, p 278
1973, p 213–218 Temperatures,” Elevated Compression 18. R. Papirno, J.F. Mescall, and A.M. Hansen,
11. H.A. Kuhn, Workability Theory and Appli- Testing of Sheet Materials, STP 303, “Fracture in Axial Compression of Cylin-
cation in Bulk Forming Processes, Forming ASTM, 1962, p 12–31 ders,” Compression Testing of Homoge-
and Forging, Vol 14, ASM Handbook, 15. T.C. Hsu, A Study of the Compression Test neous Materials and Composites, R. Chait
ASM International, 1988, p 389–391 for Ductile Materials, Mater. Res. Stand., and R. Papirno, Ed., STP 808, ASTM,
12. J.J. Jonas and M.J. Luton, Flow Softening Vol 9 (No. 12), Dec 1969, p 20 1983, p 40–63

Downloaded from http://dl.asminternational.org/handbooks/book/chapter-pdf/479841/a0003265.pdf

by Lawrence Livermore Natl Labs user