See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/268278141

Behavior of High-Strength Concrete under Biaxial Stresses

Article  in  Aci Structural Journal · January 2000

CITATIONS READS
37 501

2 authors, including:

Hesham Marzouk
Ryerson University
157 PUBLICATIONS   1,486 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

High-Strength Concrete Offshore Design View project

High Strength Concrete Applications View project

All content following this page was uploaded by Hesham Marzouk on 13 September 2016.

The user has requested enhancement of the downloaded file.

 ACI MATERIALS JOURNAL
Title no. 97-M4
Behavior of High-Strength Concrete under Biaxial Stresses
by A. Hussein and H. Marzouk
Four different types of high-strength concrete plate specimens
(150 x 150 x 40 mm) were tested under different biaxial load
combinations. The specimens were subjected to biaxial combi-
nations covering the three regions of compression-compres-
sion, compression-tension, and tension-tension. The principal
deformations in the specimens were recorded, and the crack
patterns and failure modes were examined. Based on the
strength data, the failure envelopes were developed for each
type of concrete.
The test results revealed that the failure envelopes of concrete
depend on the concrete strength and on the type of aggregates.
The biaxial compression-tension region reflected a significant
difference in the behavior between high-strength and normal
strength concrete. The compression capacity was radically
reduced for high-strength concrete when a small tension is
introduced in the other direction. A second pronounced differ-
ence was found between the high-strength lightweight and the
high-strength normalweight concrete. High-strength light-
weight concrete indicated higher strength gain under biaxial
load at all stress ratios compared to normal and high-stength
concrete. The deformation characteristics indicated that high-
strength concrete shows a linear behavior up to a higher stress
than normal strength concrete. It also has higher discontinuity
limits. The observed failure modes showed that there is no fun-
damental difference in the crack patterns and failure modes due
to the increase in the compressive strength of the concrete or
due to the use of lightweight aggregates under different biaxial
loading combinations.
Keywords: biaxial loads; compressive strength; deformation; failure
mechanisms; high-strength concretes; lightweight aggregates; stress-strain
relationships; tensile strength.
INTRODUCTION
In the past two decades, considerable attention has been given
to the use of silica fume as a partial replacement for cement in
the production of high-strength concrete. Concretes with
strengths exceeding 70 MPa are produced commercially using
conventional methods and materials and are not unusual in con-
struction today. High-strength concrete has been used for off-
shore platforms, marine structures, tall buildings, and long-span
bridges.1,2
The finite element method has been used directly for the anal-
ysis and design of complex structures, such as offshore oil plat-
forms and nuclear containment structures. These types of
structures cannot be treated properly by the more approximate
methods. The finite element method, however, requires a good
understanding of the actual material behavior under different
load combinations to yield accurate and realistic results. For
normal strength concrete, reasonable amounts of data are avail-
able. This is not the case for high-strength concrete. In fact, very
few experiments have been carried out to ascertain the behavior
of high-strength concrete under biaxial and triaxial states of
stress.
Several researchers 3-6 have studied the behavior of normal
strength concrete subjected to biaxial stresses. The experimental
programs covered the whole range of biaxial loading (compres-
ACIMaterialsJournal/January-February2000
TECHNICAL PAPER
sion-compression, compression-tension, and tension-tension).
On the other hand, previous biaxial tests conducted on high-
strength concrete7-9 were limited to the compression-compres-
sion loading region only. In addition, the maximum compressive
strength of the concrete specimen was around 70 MPa.
Few experimental investigations of the behavior of normal
strength lightweight aggregate concrete under biaxial stress
were performed.10-12 To the best of the authors’ knowledge,
there is no published data in the literature concerning the behav-
ior of high-strength lightweight concrete under biaxial state of
stress.
In this paper, the behavior of high-strength concrete under
mechanical biaxial loading is investigated. The experiments
cover the entire range of stress combinations from biaxial com-
pression to biaxial tension. The investigation was carried out on
four different types of concrete: a normal strength concrete mix-
ture (42 MPa), two high-strength concrete mixtures made with
normalweight aggregates (74 and 96 MPa), and a high-strength
lightweight concrete mixture (66 MPa). The strength data were
collected and analyzed. The failure envelopes were then devel-
oped for each type of concrete. The principal stresses and strains
at failure were recorded and presented as functions of the stress
ratios. Finally, the failure modes and crack patterns were ob-
served and examined.
RESEARCH SIGNIFICANCE
Many structural elements such as slabs, thin shells, and shear
walls are essentially in a biaxial state of stress. The behavior of
concrete, under a biaxial state of stress, has to be examined care-
fully before developing any analytical idealization. Although
considerable data are available regarding the behavior of normal
strength concrete under biaxial loading, relatively few studies
have been conducted on high-strength concrete with compres-
sive strength in excess of 70 MPa. In addition, it appears that
there is no data available on the behavior of high-strength light-
weight aggregate concrete. This paper provides the strength and
deformational characteristics of high-strength concrete under
the whole range of biaxial loading. The study includes high-
strength concrete that contains normalweight aggregates as well
as lightweight aggregates. The experimental results should form
an essential part in calibrating any constitutive model for the fi-
nite element analysis of high-strength concrete.
EXPERIMENTAL STUDY
Test specimens
The concrete plate specimens used in this study were 150 x
150 x 40 mm in dimension. The specimens were subjected to bi-
axial stress combinations that cover the regions of compres-
sion-compression, compression-tension, and tension-tension.
Four different types of concrete were tested. Within the regions
of compression-compression and compression-tension, four dif-
ACI Materials Journal , V. 97, No. 1, January-February 2000.
Received September 23, 1998, and reviewed under Institute publication policies.
Copyright  2000, American Concrete Institute. All rights reserved, including the
making of copies unless permission is obtained from the copyright proprietors. Perti-
nent discussion will be published in the November-December 2000 ACI Materials
Journal if received by August 1, 2000.
27
A. Hussein is a structural engineer with Newfoundland Design Associates Ltd., St.
Table 1—Mixture proportions of 0.1 m3 of concrete
John’s, Newfoundland, Canada. He received his BSc (Eng) from Ain Shams Univer- High-
sity, Cairo, Egypt, in 1984, and his MEng and PhD from Memorial University of New- Normal High- Ultra-high- strength
foundland, in 1990 and 1998, respectively. His research interests include mechanical strength strength strength con- lightweight
and structural behavior of high-strength concrete and nonlinear finite element analy- concrete concrete crete concrete
sis of concrete structures. Constituent Units (NSC) (HSC) (UHSC) (HSLWC)
ACI member H. Marzouk is a professor of civil engineering at Memorial University Cement (c) kg 35 40 40 40
of Newfoundland. He has also served as a full-time structural consultant in North Silica fume
America for 10 years. He received his MSc and PhD from the University of kg — 3.4 3.4 3.4
(s f )
Saskatchewan, Saskatoon, Canada. He is a member of ACI Committees 209, Creep
Cement and
kg 35 43.4 43.4 43.4
silica fume
ferent stress ratios were selected. Only three stress ratios were
Fine aggregate kg 65 60 53 61
chosen for the tension-tension region. Three specimens were
Coarse
tested for each ratio. Nonetheless, only two specimens were aggregate
kg 110 100 85 67
tested for the biaxial-compression tests.
Retarder mL — 75 100 75
Four different types of concrete were tested: 1) one normal
Superplasti-
weight concrete mixture (NSC) with a target strength of 40 MPa; mL — 1000 1300 1000
cizer
2) two high-strength concrete mixtures with a target strength of w/(c + s f ) — 0.47 0.28 0.23 0.33
70 MPa (the first mixture is made with normalweight aggre-
Mixture
gates [HSC]; the second is almost identical in composition ex- density kg/m 3 2350 2410 2385 1950
cept that lightweight aggregates were used [HSLWC]. The
purpose of these two mixtures is to study the effect of aggregate
type on the behavior of high-strength concrete under biaxial
loading conditions); and 3) an ultra high-strength concrete mix-
ture (UHSC) with a target strength of 100 MPa.
The concrete test specimens were made of ordinary portland
cement (Type 10) CSA3-AS5 with a modified C3A content of
about 6%, as produced in Newfoundland. The silica fume was
supplied from a Canadian source in Quebec. The normalweight
coarse aggregate was mostly crushed granite with a maximum
nominal size of 6 mm. The fine aggregate had a composition
identical to the coarse aggregate, with a fineness modulus (FM)
of 3.02. Both types of aggregates were locally available. For the
lightweight concrete mixture, a slate-based lightweight aggre-
gate was used as coarse aggregate. The lightweight aggregate
was brought from the construction site for the Hibernia offshore
oil platform project. This aggregate had a maximum nominal
size of 19 mm. Thus, it had to be separated and recombined to a
desirable grading. A specified maximum nominal size of 9 mm
was achieved.
Due to the low water-cement ratios of the concrete mixtures,
a superplasticizer of sulfonated naphthalene formaldehyde base,
conforming to ASTM C 494 Type F, was employed. In addition,
a nonchloride water-reducing agent of polyhydroxycarboxylic
base, conforming to ASTM C 494 Type B and D, was used. The
retarder was also useful in delaying the setting time so that the
same consistency was achieved for all specimens prepared from Fig. 1—Test setup.
each batch. The materials used made a 0.1 m3 batch of fresh con-
crete for each type. The mixture proportions for the four batches Test facilities
are listed in Table 1.
Loading frame—The specimens were tested in a specially de-
signed biaxial test machine, as shown in Fig. 1. The biaxial load-
Specimen fabrication ing on the specimens was provided by means of two
Special plexiglass molds were manufactured for casting the servo-controlled MTS compression-tension actuators with a max-
test specimens. Each mold was used to cast four specimens. imum capacity of 710 kN (160 kips). A closed-loop system in dis-
During casting, compaction was achieved by placing the mold placement control was utilized. The displacement control was
on a vibrating table and vibrating it at a frequency of 40 Hz for provided by means of an external ultra-precision AC linear vari-
60 s. The casting surfaces were finished, and the specimens able displacement transducer (LVDT) that was mounted on the
were placed under a plastic cover to avoid drying out. Two days platens. The closed-loop control of the actuators was executed
after casting, the specimens were demolded and placed in a through an MTS 406 controller.
fresh water basin. At 56 days, the specimens were transported to Loading platens—To eliminate the friction between the spec-
a humidity room with a 99% relative humidity and a constant imen and the loading platens, brush-type loading platens were
temperature of 20 C. The specimens were taken out of the hu- used. The platens were mounted on the MTS hydraulic actuators
midity room 2 days before testing, and were then ground and by means of a spherical seating (swivel) and were used through-
prepared for testing. The four sides of the specimens were out the experimental program. The brushes were manufactured
ground to ensure that the specimen had flat edges and right-an- using the electric discharge wire cutting (EDWC). It is com-
gle corners. The grinder employed a wet-type grinding method monly called the wire EDM, and is considered a special form of
using a diamond grinding wheel. The final ground surface finish electric discharge machining that uses a small-diameter wire as
was within ±0.01 mm. the electrode to cut a narrow workpiece.

28 ACIMaterialsJournal/January-February2000
 Table 2—Strength data for different mixtures
Splitting tensile Compressive
Compressive strength of cylindrical strength of cylindrical strength of plate
specimen, MPa specimen, MPa specimen σ co, MPa
Mixture 28 days 56 days 91 days 180 days 91 days 91 days
Normal strength concrete
38.1 42.2 44.3 45.2 3.60 42.7
(NSC)
High-strength concrete 71.1 74.7 76.6 78.2 5.28 73.7
(HSC)
High-strength lightweight 96.0 98.1 100.1 102.9 6.31 96.5
concrete (HSLWC)
Ultra-high-strength
61.8 65.3 68.7 69.8 3.71 66.7
concrete (UHSC)

Each brush was manufactured from a solid steel block 150 mm

thick, 40 mm wide, and 150 mm high. The filaments of the brush
(5 x 5 mm in cross section) were formed by cutting slots, with a
0.2 mm wire, in the solid block using the wire EDM method. The
height of each filament was 75 mm. Numerical control was spec-
ified to control the workpiece motions during cutting to achieve
the desired cutting path. The fastness of the ends of the rods were
within ±0.01 mm. The steel used for manufacturing the brush was
SAE 4140 steel. Annealed condition heat treatment was used.
The ultimate tensile strength of the steel was 700 MPa with a
yield strength (at 0.2% offset) of 485 MPa with a Brinell hardness
number (BHN) of 207. Figure 2 shows a photograph of the man-
ufactured brush loading platens.
Control scheme—The control scheme for the biaxial testing
program was composed of two closed control loops. The first
control loop was in displacement control in the major loaded di-
rection. The command signal was supplied using the digital
function generator and fed into the MTS controller. The AC
LVDT was mounted near the specimen supplied the feedback
signal. Through the controller, both signals were compared, and
if necessary, an error signal was created. The correcting signal
was then given to the servo-valve at the hydraulic actuator. The
output from the load cell, in the major direction, was used as a
command signal for the other actuator that operated in a load
control closed loop. The proportionality of the two loads was
maintained through the MTS controller. A block diagram high-
lighting the details of the used closed-loop test scheme is shown
in Fig. 3.
Deformation measurements—The surface strains were mea-
sured at the center of the specimen in the two loading directions
by means of two extensometers. The extensometer was a canti-
lever type with two strain gages mounted on each leg. The strain Fig. 2—Brush loading platens.
gages were wired in such a way that only flexural strains (in
each leg) could be detected. The extensometers’ legs were glued
trol cylinders. The specimens were tested using a 2670 kN (600
to the specimen with epoxy. The gage length for each extensom-
kips) compression testing machine. For each mixture, three cylin-
eter was 100 mm. The lateral displacement was measured via
ders were tested in accordance with ASTM C 39 at 7, 28, 91, and
two short stroke LVDTs mounted on a special frame and touch-
180 days. Table 2 provides the compressive strength values for
ing both sides of the specimen.
the different concrete mixtures at different ages. All normal
The analog output signals from the different measuring devic- strength concrete specimens formed a cup-and-cone type of fail-
es were connected to different input channels of the data acqui- ure due to the bond failure mechanism. For high-strength con-
sition system. Menu-driven software was used to drive the crete specimens, however, the failure was mainly due to weak
different modules of the data acquisition system. In each mea- link in the aggregate, bond, or mortar chain. The splitting tensile
suring step, all the connected devices, including the elapsed strength for the different mixtures at the age of 91 days is given
time (in seconds), were scanned and the data was stored on the in Table 2. Only two 100 x 200 mm cylinders were tested. The
hard disk of the PC. In addition, different x-y plots were dis- tests were carried out in accordance with ASTM testing proce-
played on the screen of the computer for visual control of load dures.
versus deformation. Details of the test facilities and the experi-
To ensure that there was no significant gain in the compres-
mental procedures are given in Reference 13.
sive strength, it was decided that the biaxial test program on the
concrete plates should start at 91 days from the casting of the
TEST RESULTS specimen. The uniaxial compressive strength, obtained from
Ultimate strength testing the concrete cylinders, is slightly different than the
Uniaxial strength—The ultimate uniaxial compressive strength uniaxial strength obtained from testing the plate specimen. Such
for each mixture was determined from testing 100 x 200 mm con- a difference can be attributed to the difference in sizes between

ACIMaterialsJournal/January-February2000 29
Fig. 3—Block diagram highlighting details of closed-loop test scheme.
Fig. 4—Biaxial strength envelopes for four different types of
concrete under combined tension and compression, biaxial ten-
sion, and biaxial compression.
30
the two specimens and to the effect of end conditions in the two
different tests.
The experimental program was carried out using steel brush-
es. Accordingly, the same end conditions used for testing re-
main the same. Thus, it seems rational to use the uniaxial
compressive strength of the concrete plates as the representative
uniaxial compressive strength for the different types of concrete
specimens throughout this chapter. It is referred to as σ co.
Biaxial strength data—The plate specimens were tested un-
der different combinations of biaxial loading. The tests were
performed under constant stress ratio. Note that notations for the
principal stresses are σ 1 > σ 2 > σ3, σi negative in compression.
In the presentation of the data and the subsequent discussion, all
strength data are normalized with respect to the unconfined
uniaxial compressive strength σco obtained from the plate spec-
imen.
Figure 4 shows the relationship between the normalized (with
regard to σ co) principal stresses at failure for the different con-
crete mixtures. The graphical representation of this relationship
is also referred to as the strength envelopes.
In general, the ultimate strength of concrete under biaxial
compression is higher than under uniaxial compression due to
the increased confinement from biaxial compression. The
strength increase under biaxial compression is dependent on the
biaxial stress ratio. The maximum biaxial strength occurs at a
biaxial stress ratio of 0.5 for all specimens tested. At this stress
ratio, a strength increase of about 31% for the NSC specimen,
and 32, 35, and 38% for the HSC, UHSC, and HSLWC specic-
mens, respectively, was observed.
At equal biaxial compression (σ 2/σ 3 = 1.0), the relative
strength increase becomes smaller as the compressive strength
is increased for all the normalweight mixtures. The strength in-
crease is 19, 14, and 9% for the NSC, HSC, and UHSC mix-
tures, respectively.
ACIMaterialsJournal/January-February2000
Fig. 5—Stress-strain relationships for high-strength concrete Fig. 6—Stress-strain relationships for high-strength concrete
mixture (UHSC) under biaxial compression. mixture (UHSC) under combined tension and compression.

The effect of aggregate type on the biaxial behavior of con-

crete can be demonstrated by comparing the lightweight con-
crete mixture HSLWC with the corresponding HSC mixture,
which is relatively similar in compressive strength with dif-
ferent types of aggregates. The HSLWC indicated a higher
strength gain, at all stress ratios, than the HSC (Fig.4 (b)). This
contradicts the findings by Yin et al. 14 for normalweight aggre-
gates.
The results of that study suggested that the biaxial strength of
concrete increases as the strength of the aggregates is increased.
The discrepancy can be attributed to the nature of lightweight
aggregate concrete. When lightweight aggregate concrete is
subjected to a uniaxial compressive load, the aggregate is sub-
jected to a lateral confinement of the surrounding matrix.15
Thus, the lightweight aggregates benefit from the confinement
effects that occur in multiaxial stress conditions. Moreover,
lightweight aggregates improved bond characteristics (with ce-
ment mortar) better than normalweight aggregates. Accordingly,
these factors can lead to the enhancement of the behavior of high-
strength lightweight concrete over the high-strength normal-
weight concrete under biaxial compression.
The uniaxial tensile strength f ′t , found to be 3.30, 4.15, 4.82,
and 3.38 MPa for the NSC, HSC, UHSC, and HSLWC speci-
mens, respectively, resulted in a ratio of f ′t /f ′c equal to 0.078,
0.056, 0.050, and 0.051. The f ′t /f ′c ratio decreases as the com-
pressive strength increases. It is noted herein that the uniaxial ten-
sile strength is in good agreement with previous research work on
high-strength concrete.16,17 The tensile strength of the HSLWC is
less than the corresponding HSC mixture. The different combina- Fig. 7—Stress-strain relationships for high-strength concrete
tions of the biaxial tension-tension test resulted in a strength that mixture (UHSC) under biaxial tension.
was almost equal to the uniaxial tensile strength.
In previous studies on normal strength concrete, and in the bi- concrete tensile failure strength with f ′t is not significant
axial tension region, Tasuji et al.6 noticed a slight increase in the enough to necessitate a more accurate verification in this region.
tensile strength when the two principal tensile stresses are equal, It should be noted herein that the values of f ′t /f ′c that were report-
whereas Kupfer et al. 3 observed that there is no difference. It ed by Kupfer et. al.3 for the 59 MPa concrete appear to be slightly
should be noted, however, that in both cases the differences of high compared with the results obtained in the current study.

ACIMaterialsJournal/January-February2000 31
Fig. 8—Stress-strain relationships for lightweight concrete
mixture (HSLWC) under biaxial compression.
Fig. 9—Stress-strain relationships for lightweight concrete
mixture (HSLWC) under combined tension and compression.
In the biaxial compression-tension region, there is a signifi-
cant difference in the behavior between high-strength concrete
and normal strength concrete. Introducing a small amount of
tension decreases the compressive capacity more radically for
high-strength concrete than for normal strength concrete. This is
32
Fig. 10—Stress-strain relationships for lightweight concrete
mixture (HSLWC) under biaxial tension.
clear from the inspection of Fig.4(a). The same trend was re-
ported by Kupfer et al.3
Comparing the values of the uniaxial tensile strength and the
splitting tensile strength in this study indicates that the tensile
splitting strength was always found to be higher than the uniax-
ial tensile strength for high-strength concrete.
Typical stress-strain curves
The loads and deformations in the three principal directions
were measured for all the tests using the devices mentioned pre-
viously. The stress-strain curves, presented as follows, show the
relationships of normalized stress to actual strain. The conven-
tions for the principal strains are such that ε1 > ε 2 > ε3, with ten-
sile stress being positive.
Figure 5 through 10 show typical stress-strain curves under
different load combinations for Mixtures UHSC and LWAC.
Details of the stress-strain curves for all the mixtures are given
in Reference 13. In uniaxial compression, the average value of
the principal compressive strain at the ultimate stress ( ε3) in-
creases as the compressive strength was increased (for the nor-
malweight aggregate concretes). The strain at ultimate stress was
2076, and 2600 microstrain for the NSC and the high-strength
concrete UHSC. The lightweight concrete mixture (HSLWC)
showed a higher value of ε3 (3080 microstrain) than the corre-
sponding high-strength concrete normalweight mixture HSC
(2200 microstrain). An average value of the modulus of elastic-
ity E was found to be 27.7, 39, 42, and 25.3 GPa for the NSC,
HSC, UHS, and UHSC specimens, respectively.
In uniaxial tension, the principal strain ε 1 at failure was found
to be 156, 126, 141, and 172 microstrain for NSC, HSC, UHSC,
and HSLWC, respectively. In uniaxial and biaxial tension, the
variation of the average value of the principal tensile strain at ul-
timate stress (ε 1) was not significant for each individual mixture
under different load combinations. Note that HSLWC had high-
er values than those of the corresponding HSC mixture.
ACIMaterialsJournal/January-February2000
 Fig. 11—Poisson’s ratio versus applied stress for different types of concrete in uniaxial com-
It was noted that the material constants obtained in the exper-
imental study were somewhat different in uniaxial compression
and uniaxial tension for all the mixtures. The modulus of elas-
ticity in tension was found to be 30, 42.6, 46, and 28.5 GPa, re-
spectively. Those values are slightly higher than those in
compression. This is in good agreement with previous research
on high-strength concrete.17
The proportional limit values can be defined as the point at
which the stress-strain curve deviates from linearity. In uniaxial
compression, the high-strength concrete specimens, HSC and
UHSC, showed a linear behavior up to a higher stress than normal
strength concrete. Also, the HSC specimen showed a more linear
behavior than the corresponding HSLWC.
In biaxial compression, as the minor principal stress σ 2 is in-
creased, the proportional limit is also increased. A major cause for
the nonlinearity of the stress-strain curve for concrete is internal
microcracking. The microcracks begin as bond at the aggregate-
mortar interface (transition zone), and propagate through the mor-
tar matrix to cause failure. A stress σ2 preventing these microc-
racks from propagating results in a stiffer, more linear stress-
strain response in the major principal direction.
A manifestation of this explanation can be seen from the
HSLWC specimen. It seems that the lightweight aggregates
benefit from the confinement effects that occur in multiaxial
stress conditions. The stress-strain curves were more linear over
a higher range as the confining stress was increased. The light-
weight aggregates were more compressable than harder granite.
Therefore, stress concentration formed in the transition zones
around the angular granite at higher stress levels.
The behavior of the concrete specimen was considered at the
discontinuity level, as well as at the ultimate strength. The dis-
continuity level represents the onset of major microcracking of
concrete as defined by Newman18 and used by Tasuji et al. 6 in
interpreting the biaxial behavior of normal strength concrete. It
is apparently related to the beginning of extensive mortar crack-
ing.
For the uniaxial and biaxial compression tests, discontinuity
was defined as the point at which the ratio of principal tensile
strain to the principal compressive strain began to increase. For
uniaxial compression, this is equivalent to the point at which
Poisson’s ratio started to increase. For biaxial tests involving di-
ACIMaterialsJournal/January-February2000
rect tension, discontinuity was defined as the point at which the
tensile strain began to deviate from linearity. Tasuji et al.6 found
that the discontinuity occurred at approximately 70% of the ul-
timate load in uniaxial and biaxial compression tests, and at ap-
proximately 60% of the ultimate load in tests involving direct
tension for normal strength concrete.
Figure 11 shows a plot of Poisson’s ratio versus applied stress
for the different mixtures under uniaxial loading. From this fig-
ure, the discontinuity limits can be estimated to be 70, 85, 90,
and 76% for the NSC, HSC, UHSC, and HSLWC specimens,
respectively. Close values were also obtained under different bi-
axial compression loading combinations, and thus, those values
can be used as discontinuity levels for the different types of con-
crete.
It can also be seen from Fig. 11 that the Poisson’s ratio curves
were not almost vertical to a certain level. Thus, a definitive val-
ue of Poisson’s ratio could not be obtained. Rather, an average
value can be used. The scatter of the ratio of longitudinal to lat-
eral stress was very large at the beginning of the loading, and up
to 10% of the ultimate load.
Failure modes
The crack patterns and failure modes were observed for all
the test specimens. In general, the failure modes observed in the
high-strength concrete (HSC and UHSC) tests were similar to
those obtained for normal strength concrete specimens. There
was no fundamental difference in the crack patterns and failure
modes due to the increase in the compressive strength of the
concrete or due to the use of lightweight aggregates under differ-
ent biaxial loading combinations.
The crack patterns observed in the normal strength concrete
were in good agreement with those obtained in previous inves-
tigations.3,4,6 Under uniaxial compression, fracture of the
speci mens occurred by the formation of cracks that were in-
clined at an angle of 20 to 30 degrees to the direction of the ap-
plied load, and perpendicular to the larger unloaded surface of
the specimen. This type of failure can be referred to as faulting
failure14 as opposed to the splitting type of failure where the
splitting occurs along the direction of loading.
Figure 12(a) shows photographs of typical cracking patterns
of an NSC specimen under uniaxial loading. The high-strength
33
concrete specimen HSC and UHSC, as well as the high-
strength concrete specimen HSLWC showed a splitting type
of failure. The cracks were formed in a direction parallel to the
applied load. Figure 12(b), (c), and (d) show a typical specimen,
at failure, under uniaxial compression for the different high-
strength concrete types.
Examination of the failure surfaces of normal strength NSC
specimens showed that there were no cracks that passed directly
Fig. 12—Failure modes of specimens subjected to uniaxial com-
Fig. 13—Failure modes of specimens subjected to biaxial compression stresses σ 2/ σ3
34
through the coarse aggregate; the cracks passed mainly through
the mortar. The failure surfaces of the high-strength specimens
(HSC and UHSC) passed through some of the aggregates, with
a larger portion of the failure surface passing through the coarse
particles for the UHSC type. The failure surfaces of HSLWC
specimens revealed that the cracks passed through the coarse
aggregate and the mortar. It should be noted herein that mixture
UHSC had a uniaxial compressive strength (f ′c ) of 96 MPa.
This value is close to the fracture strength of the coarse aggre-
gate (crushed granite). Thus, a better compatibility of the
strength between the mortar and the coarse aggregate is
achieved. This, in turn, leads to a more homogeneous type of be-
havior and influences the cracks to pass through the aggregate
as well as the mortar.
Under biaxial compression, and at the low stress ratio of 0.2,
the specimen showed formation of microcracks parallel to the
free surface of the specimen. Failure occurred, however, by the
formation of a major crack that had an angle of 18 to 25 degrees
to the free surfaces of the specimen. This failure mode was ob-
served for the different types of concrete (Fig. 13).
For specimens subjected to biaxial compression ratios of 0.5
and 1, failure progressed due to tensile splitting cracks that oc-
curred along the direction of loading and in a plane parallel to
the free surfaces of the specimen (σ 1 − σ 2 plane). It was also ob-
served that Specimen HSLWC developed several more microc-
racks along the loaded surfaces and parallel to the loaded
surfaces of the specimen than in the case of the normalweight
specimen. Figure 14 shows the failure modes for the different
types of concrete under equal biaxial compression.
The tests under combined compression and tension revealed
that only one continuous crack normal to the principal tensile
stress was formed. For the smaller ratio of tension-to-compres-
sion (σ 1/σ 3 = − 0.05), however, several cracks were observed in
the direction of compressive loading before failure for the NSC
specimen. The HSC specimen showed a somewhat different be-
havior. Failure always occurred with one crack normal to the
principal tensile stress at such a small stress ratio of σ 1/ σ3 = −
0.05.
Under uniaxial tension, fracture of the specimen and failure
occurred by the formation of a single crack perpendicular to the
direction of loading and normal to the plane of the specimen.
Examination of the failure surfaces of the test specimens for the
concrete type NSC showed there were no cracks that passed di-
ACIMaterialsJournal/January-February2000
 Fig. 14—Failure modes of specimens subjected to biaxial compression stresses σ 2/ σ3 = 1.0.

rectly through the coarse aggregate. The fracture surface of the

HSC and UHSC specimen showed that failure surface pro-
gressed through some coarse aggregates. A crude comparison
between the failure surface and that of uniaxial compression in-
dicated that the percentage of broken aggregates was higher in
tension than in compression. Inspection of the failure surfaces
of the HSLWC specimens revealed that the cracks passed
through the coarse aggregate as well as the mortar. Figure 15
shows the failure modes for some of the concrete specimens un-
der uniaxial tension.
Under biaxial tension loading, a single crack was formed in a
direction normal to the unloaded surface of the specimen and
perpendicular to the maximum principal stresses. In the case of
equal biaxial tension, there was no preferred direction for the
fracture surface, and the cracks were always normal to the un-
loaded surface.

CONCLUSIONS
The following conclusions may be reached from the present
investigation concerning the behavior of high-strength concrete
under biaxial stress:
1. The ultimate strength of concrete under biaxial compres-
sion is higher than under uniaxial compression. The maximum
biaxial strength occured at a biaxial stress ratio of 0.5 for all
specimens tested. At this stress ratio, a strength increase of
about 31% for the NSC specimen, 32, 35, and 38% for the HSC,
UHSC, and HSLWC specimens, respectively, was observed;
2. At equal biaxial compression (σ 2 / σ 3 = 1.0), the relative
strength increase becomes smaller as the compressive strength
is increased for all the normalweight mixtures. The strength in-
crease is 19, 14, and 9% for the normal strength, high-strength,
and ultra-high-strength concrete, respectively;
3. HSLWC has a higher strength gain at all compression Fig. 15— Failure modes of specimens subjected to uniaxial ten-
stress ratios than the corresponding high-strength normalweight
concrete; mal strength concrete. Introducing a small amount of tension
4. Test results confirmed that high-strength concrete has a low- decreases the compressive capacity more radically for high-
er tensile-to-compressive strength ratio than normal strength con- strength concrete than for normal strength concrete;
crete. The uniaxial tensile strength f ′t was found to be 3.30, 4.15, 6. Under different biaxial compression loading combinations,
4.82, and 3.38 MPa for the NSC, HSC, UHSC, and HSLWC the results show that the introduction of a second principal stress
specimens, respectively, resulting in a ratio of f ′t /f ′c equal to significantly affects the effective elastic modulus of concrete
0.078, 0.056, 0.050, and 0.051; specimen in the direction of the first principal stress;
5. In the biaxial compression-tension region, there is a signif- 7. The material constants obtained in the experimental study
icant difference in the behavior between high-strength and nor- are somewhat different in uniaxial compression and uniaxial

ACIMaterialsJournal/January-February2000 35
 tension for all the mixtures. The modulus of elasticity in tension
was found slightly higher than that in compression;
8. High-strength concrete showed a linear behavior up to a
higher stress than normal strength concrete;
9. The discontinuity limits were found to be 70, 85, 90, and 76%
for the NSC, HSC, UHSC, and HSLWC specimens, respectively.
Thus, as the compressive strength of concrete is increased, the dis-
continuity levels become higher. The discontinuity level for the
high-strength lightweight concrete is lower than the corresponding
high-strength concrete; and
10. There is no fundamental difference in the crack patterns
and failure modes due to the increase in the compressive
strength of the concrete or due to the use of lightweight aggre-
gates under different biaxial loading combinations.
ACKNOWLEDGMENTS
The authors are grateful to the Natural Sciences and Engineering Re-
search Council of Canada (NSERC) for providing the funds for the project.
Sincere thanks are due to the Technical Staff of the Structural Engineering
Laboratory at Memorial University for their assistance during testing, espe-
cially A. Bursey, C. Ward, and R. O’Driscoll. The authors express their
warm appreciation to the members of the Technical Services of Memorial
University for their assistance in manufacturing the test setup.
Our sincere thanks are extended to North Star Cement Ltd., Newfound-
land, SKW Canada Inc., Quebec, Strenson Ltd., Ontario, for providing the
cement, silica fume, and all the chemical admixtures, respectively.
CONVERSION FACTORS
1 mm = 0.0394 in.
1 m3 = 35.3 ft3
1L = 61 in.3
1 mL/m 3 = 0.026 oz./yd 3
1 MPa = 145 psi
1 GPa = 145 × 103 psi
REFERENCES
1. Randall, V., and Foot, K., “High-Strength Concrete for Pacific First
Center,” Concrete International , V. 11, No. 4, Apr. 1989, pp. 14-16.
2. Haug, A. K., and Sandvik, M., “Mix Design and Strength Data for
Concrete Platforms in the North Sea,” Concrete in Marine Environment,
SP-109, V. M. Malhotra, ed., American Concrete Institute, Farmington
Hills, Mich., 1988, pp. 896-906.
3. Kupfer, H.; Hilsdorf, H.; and Rusch, H., “Behavior of Concrete under
Biaxial Stresses,” ACI JOU RNA L, Proceedings V. 66, No. 8, Aug. 1969, pp.
36
View publication stats
656-666.
4. Nelissen, L. J., “Biaxial Testing of Normal Concrete,” Heron (Delft),
V. 18, No. 1, 1972, pp. 603-611.
5. Liu, T.; Nilson, A.; and Slate, F., “Stress-Strain Response and Fracture
of Concrete in Uniaxial and Biaxial Compression,” ACI J OURNAL, Pro-
ceedings V. 69, No. 5, May 1972, pp. 291-295.
6. Tasuji, I.; Slate, F.; and Nilson, A., “Stress-Strain Response and Frac-
ture of Concrete in Biaxial Loading,” ACI JOURNAL , Proceedings V. 75,
No. 7, July 1978, pp. 306-312.
7. Nojiri, Y.; Kotani, K.; and Abe, Y., “Failure Envelope of Concrete
Subjected to Multiaxial Compressive Stresses,” Proceedings of Interna-
tional Conference on Concrete under Multiaxial Conditions , Toulouse,
RILEM, 1984, pp. 141-148.
8. Herrin, J. C., “Behavior of High-Strength Concrete Model Subjected
to Biaxial Compression,” MS thesis, University of Texas at Austin, May
1983, 148 pp.
9. Chen, R. L., “Behavior of High-Strength Concrete in Biaxial Compres-
sion,” PhD dissertation, University of Texas at Austin, Dec. 1984, 298 pp.
10. Niwa, Y.; Kobayashi S.; and Koyanagi, W., “Failure Criterion of
Lightweight Aggregate Concrete Subjected to Biaxial Compression,”
Memoirs, Faculty of Engineering, Kyoto University, V. 29, Part 2, Apr.
1967, pp. 119-131.
11. Taylor, M. A.; Jain, A. K.; and Ramey, M. R., “Path Dependent Biax-
ial Compressive Testing of an All-Lightweight Aggregate Concrete,” ACI
JOURNAL , Proceedings V. 69, No. 12, Dec. 1972, pp. 758-764.
12. Atan, Y., and Slate, F., “Structural Lightweight Concrete under Biax-
ial Compression,” ACI JOURNAL, Proceedings V. 70, No. 3, May 1973, pp.
182-186.
13. Hussein, A., “Behavior of High-Strength Concrete under Biaxial
Loading Conditions,” PhD thesis, Memorial University of Newfoundland,
St. John’s, Newfoundland, Canada, Apr. 1998, 245pp.
14. Yin, W.; Su, E.; Mansour, M.; and Hsu, T., “Biaxial Tests of Plain
and Fiber Concrete,” ACI Materials Journal , V. 86, No. 3, May-June 1989,
pp. 236-243.
15. Zhang, M., and Gjørv, O., “Mechanical Properties of High-Strength
Lightweight Concrete,” ACI Materials Journal, V. 88, No. 3, May-June
1991, pp. 240-246.
16. Daerga, P.; Pettersson, M.; and Pontinen, D., “Fracture Properties in
Tension of a High-Performance Concrete,” High-Strength Concrete 1993,
Proceedings, Lillehammer, Norway, June 1993, pp. 1193-1200.
17. Zhang, M.; Rønning, T.; and Gjørv, O., “Mechanical Properties of
High-Strength Concrete,” High-Strength Concrete 1993, Proceedings,
Lillehammer, Norway, June 1993, pp. 1271-1279.
18. Newman, K., “Criteria for the Behavior of Plain Concrete under
Complex State of Stress,’’ International Conference on the Structure of
Concrete, Proceedings, Cement and Concrete Association , London, 1968,
pp. 255-274.
ACIMaterialsJournal/January-February2000