ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 98-S38

Stress-Strain Behavior of High-Strength Concrete

Confined by Ultra-High- and Normal-Strength Transverse
Reinforcements
by Li Bing, R. Park, and H. Tanaka

When high-strength concrete is used for reinforced concrete mem- reinforcement can effectively increase the ductility of rein-
bers subjected to seismic loading, it is more difficult to achieve forced concrete columns. Ultra-high-strength transverse re-
ductile behavior of such members than when normal-strength con- inforcement is especially used for columns with high-
crete is used. In this paper, an experimental study of a number of strength concrete.
quasi-static axial loading tests on high-strength concrete speci-
mens confined by various amounts of transverse reinforcement is
described. The main parameters were concrete strengths ranging RESEARCH SIGNIFICANCE
from 35.2 to 82.5 MPa and yield strength of Grade 430 and 1300 The most fundamental issue in predicting the behavior of
transverse reinforcement. A stress-strain relationship for confined reinforced concrete members is the stress-strain behavior of
high-strength concrete is proposed that is found to give reasonably the constituent materials. Concrete is used to resist compres-
good prediction of the experimental behavior of circular and sion and its behavior in compression is important to the de-
square specimens with high-strength concrete confined by either signer. If the behavior of concrete subjected to uniaxial
normal- or ultra-high-yield-strength with various configurations. compression is known, the flexural behavior of reinforced
An empirical formula for the ultimate longitudinal strain of con- concrete can be estimated. The confinement steel require-
fined high-strength concrete corresponding to the first hoop or spi-
ral fracture is also proposed.
ments for normal-strength concrete are reasonably well es-
tablished in current building codes. In recent years, the
possible use of high-strength concrete for buildings con-
Keywords: column; confinement reinforcement; high-strength concrete;
strains; stresses; tests.
structed using reinforced concrete has been considered. Re-
search findings for high-strength concrete, however, are
relatively scarce in the literature.
INTRODUCTION
Reinforced concrete buildings are generally designed to
SUMMARY OF MODELS
behave in a ductile manner under the action of a severe earth-
A number of stress-strain models have been proposed in
quake. To achieve such ductile behavior, structural members
the past 15 years. A detailed review of existing models was
of the buildings should be carefully detailed. In the case of
presented elsewhere.1 The following section provides an
buildings with moment-resisting frames, detailing of trans-
overview of these analytical models that cover high-strength
verse reinforcement in the potential plastic hinge regions of
concrete.
the columns is a major consideration. For many years re-
searchers have been investigating a method for detailing the
transverse reinforcement to increase strength and ductility Martinez, Nilson, and Slate (1982)
of reinforced concrete columns. It has been demonstrated Martinez, Nilson, and Slate2 conducted experiments on
that adequate confinement of the core concrete and tying of several small diameter cylindrical specimens of high-
longitudinal reinforcement using transverse reinforcement strength concrete confined by spiral reinforcement without
can improve column ductility most effectively. Past and re- concrete cover and longitudinal reinforcement. The concrete
cent earthquakes have proved the validity of this philosophy. strength of the specimens ranged from 21 to 69 MPa and the
confining pressure ranged from 1.7 to 2.1 MPa. The yield
The gradual development of concrete technology has pro- strength of the lateral reinforcement was approximately 414
moted the use of high-strength concrete in the construction MPa. They proposed a theoretical model for the complete
industry. Concrete technology has developed to an extent stress-strain curve of spirally confined high-strength con-
where concrete compressive strengths up to 100 MPa and high- crete columns based on their own test results.
er can be reached without difficulties. There are, however, only
a limited number of studies relating to the confining effects
on high-strength concrete, even though the use of high- Fafitis and Shah (1985)
Fafitis and Shah3 tested a large number of small concrete
strength concrete has been increasing in recent years. Mean-
cylinders with practically no cover to the spiral steel and had
while, the strength of reinforcing steel has also been im-
no longitudinal reinforcement. The spiral wire was 3.2 mm
proved markedly. Normally the yield strength of
reinforcement is approximately 300 to 500 MPa. Ultra-high-
strength reinforcement with yield strength over 1000 MPa, ACI Structural Journal, V. 98, No. 3, May-June 2001.
MS No. 00-135 received June 6, 2000, and reviewed under Institute publication
however, has been recently used for transverse reinforce- policies. Copyright © 2001, American Concrete Institute. All rights reserved, includ-
ment of concrete columns in Japan. Some research work has ing the making of copies unless permission is obtained from the copyright propri-
etors. Pertinent discussion will be published in the March-April 2002 ACI Structural
shown that the use of ultra-high-strength steel for transverse Journal if received by November 1, 2001.

ACI Structural Journal/May-June 2001 395

Li Bing is an assistant professor in the School of Civil and Structural Engineering at
the Nanyang Technological University, Singapore. He received his BE from Tong Ji
University, China, and his PhD from the University of Canterbury, New Zealand. His
research interests include reinforced concrete, precast concrete structures, and
designing for earthquake resistance and blast loading.

R. Park, FACI, is an Emeritus Professor of civil engineering at the University of Can-

terbury, New Zealand. He is a corecipient of ACI’s 1984 and 1989 Raymond C. Reese
Research Award. His research interests include reinforced and prestressed concrete
structures, and designing for earthquake resistance.

H. Tanaka is a professor in the Disaster Prevention Research Institute, Kyoto Univer-

sity, Japan. He received his BE and ME from Kyoto University, Japan, and his PhD in
civil engineering from the University of Canterbury. His research interests include
reinforced concrete and high-strength concrete structures, particularly designing for
earthquake resistance.

Fig. 2—Test units set up in 10 MN capacity hydraulic uni-

versal testing machine.
Fig. 1—Short column with axial compressive loading: prin-
cipal dimensions; test setup; instrumentation; and position
of strain gages. to 87.5 MPa. The yield strength of hoops ranged from 191 to
1397 MPa. Based on their experimental work, they proposed
a three-part stress-strain curve.
in diameter and had a yield strength of 413 MPa. Variables
investigated in the experiment included the concrete com- Cusson and Paultre (1995)
pressive strength, the spiral spacing, and the strain rate effect Cusson and Paultre7 have developed a stress-strain model
(slow and fast). Based on their test results and the results of and calibrated it against their experimental results from high-
Martinez, Nilson, and Slate,2 Fafitis and Shah proposed a the- strength concrete tied columns. All the tested specimens were
oretical model for the complete stress-strain curve of circular 235 x 235 x 1400 mm. The concrete strength considered in
and square confined high-strength concrete. the experimental program ranged from 60 to 120 MPa and the
yield strength of confinement steel ranged between 400 to
Yong, Nour, and Nawy (1989) 800 MPa.
Yong, Nour, and Nawy4 tested 24 square prisms (152 x
152 x 457 mm) that were made of high-strength concrete Razvi and Saatcioglu (1996)
with compressive strengths ranging from 83.6 to 93.5 MPa, Razvi and Saatcioglu8 developed a stress-strain model for
confined with square hoops with a yield strength of 496 confined high-strength concrete based on their tests conducted
MPa. Based on their test results, Yong, Nour, and Nawy pro- on a large number of near-full size circular and square column
posed a three-part stress-strain curve for rectilinear confined specimens. The concrete strength considered in the experimen-
concrete. tal program ranged from 60 to 124 MPa and the yield strength
of confinement steel ranged between 400 to 1000 MPa.
Bjerkeli, Tomaszewicz, and Jansen (1990) Most of the models show very similar trends that indicates
Bjerkeli, Tomaszewicz, and Jansen5 tested a large number any one of them could be used to generally predict the behavior
of plain and confined high-strength concrete columns (150 of high-strength concrete, if properly calibrated.
mm diameter x 500 mm high cylinders, 150 x 150 x 500 mm
prisms, and 300 x 500 x 2000 mm prisms) confined with var- EXPERIMENTAL PROGRAM
ious volumetric ratios of spiral reinforcement. The concrete A total of 40 reinforced concrete columns 720 mm in length
compressive strengths ranged from 65 to 115 MPa. The test were tested. They were either 240 mm square or 240 mm diam-
specimens contained longitudinal steel but no concrete cov- eter circular sections. The arrangement of longitudinal bars,
er. They proposed a three-part stress-strain curve for con- transverse hoops, square helices, and spiral reinforcement used
fined high-strength concrete based on their test results. are given in Fig. 1. The reinforcement details are listed in Table
1 and 2. Four different concrete compressive strengths were
Muguruma and Watanabe (1992) used, ranging from 35.2 to 82.5 MPa. Two grades of transverse
Muguruma and Watanabe6 tested small square specimens reinforcement were used in the test units, namely Grade 430
confined laterally by square helix hoops of different yield steel (fyh = 445 MPa) and ultra-high Grade 1300 steel (fyh =
strengths and with various volumetric ratios. The specimens 1318 MPa). The hoop bars were anchored by a 135-degree bend
contained no longitudinal reinforcement and had no cover. around longitudinal bars. They were extended beyond the bend
The concrete strength of the specimen was varied from 31.5 of at least eight hoop bar diameters in the concrete core. The

396 ACI Structural Journal/May-June 2001

 Table 1—Specimen properties for square test units
Longitudinal reinforcement Transverse reinforcement
Diameter, Diameter, Spacing,
Unit no. ′ , MPa No. of bars
fco mm fy, MPa mm mm fyh, MPa ρs , %
1A 60.0 4 12 443 6.0 20 445 2.63
1B 72.3 4 12 443 6.0 20 445 2.63
2A 60.0 8 12 443 6.0 20 445 4.48
2B 72.3 8 12 443 6.0 20 445 4.48
4A 60.0 4 12 443 6.0 35 445 1.50
4B 72.3 4 12 443 6.0 35 445 1.50
5A 60.0 8 12 443 6.0 35 445 2.56
5B 72.3 8 12 443 6.0 35 445 2.56
7A 60.0 4 12 443 6.0 50 445 1.05
7B 72.3 4 12 443 6.0 50 445 1.05
8A 60.0 8 12 443 6.0 50 445 1.79
8B 72.3 8 12 443 6.0 50 445 1.79
10A 60.0 4 12 443 6.0 65 445 0.80
10B 72.3 4 12 443 6.0 65 445 0.80
11A 60.0 8 12 443 6.0 65 445 1.38
11B 72.3 8 12 443 6.0 65 445 1.38
1HA 35.2 8 12 443 6.4 35 1318 2.86
1HB 52.0 8 12 443 6.4 20 1318 5.00
1HC1 82.5 8 12 443 6.4 20 1318 5.00
3HA 35.5 8 12 443 6.4 53 1318 1.89
3HB1 52.0 8 12 443 6.4 35 1318 2.86
3HB3 52.0 8 12 443 6.4 35 1318 2.86
3HC1 82.5 8 12 443 6.4 35 1318 2.86
3HC3 82.5 8 12 443 6.4 35 1318 2.86
5HA 35.5 8 12 443 6.4 70 1318 1.43
5HB 52.0 8 12 443 6.4 50 1318 2.00
5HC 82.5 8 12 443 6.4 50 1318 2.00

Table 2—Specimen properties for circular test units

Longitudinal reinforcement Transverse reinforcement
Diameter, Diameter, Spacing,
Unit no. ′ , MPa No. of bars
fco mm fy, MPa mm mm fyh, MPa ρs, %
3A 63.0 6 12 443 6.0 20.0 445 1.53
3B 72.3 6 12 443 6.0 20.0 445 1.53
6A 63.0 6 12 443 6.0 35.5 445 0.82
6B 72.3 6 12 443 6.0 35.5 445 0.82
9A 63.0 6 12 443 6.0 50.0 445 1.68
9B 72.3 6 12 443 6.0 50.0 445 1.68
12A 63.0 6 12 443 6.0 65.0 445 2.94
12B 72.3 6 12 443 6.0 65.0 445 2.94
2HA 35.2 6 12 443 6.4 35.0 1318 1.68
2HB 52.0 6 12 443 6.4 20.0 1318 2.94
2HC1 82.5 6 12 443 6.4 20.0 1318 2.94
4HA 35.2 6 12 443 6.4 53.0 1318 1.10
4HB1 52.0 6 12 443 6.4 35.0 1318 1.67
4HC 82.5 6 12 443 6.4 35.0 1318 1.67
6HA 35.2 6 12 443 6.4 70.0 1318 0.84
6HB 52.0 6 12 443 6.4 50.0 1318 1.17
6HC 82.5 6 12 443 6.4 50.0 1318 1.17

spacing of transverse hoops and helices was reduced to 25 mm trolled by load, displacement, or strain. The steel reaction col-
within the region of 100 mm from each end of the test units to umns of the machine are stiff enough to permit the machine to
provide extra confinement and to ensure that failure occurs in measure the descending-branch of the load-deformation curve
the central region. of the test specimens. Two different methods of recording axial
Concentric vertical load was provided by a 10 MN capacity deformations were used during testing of the specimens. The
hydraulic universal testing machine. The machine can be con- first method was to measure the overall shortening of the spec-

ACI Structural Journal/May-June 2001 397

Table 3—Mixture proportions for different
compressive strengths of concrete
Weights, m3
Contents fc′ = 35 MPa fc′ = 52 MPa fc′ = 75 MPa fc′ = 82.5 MPa
13 mm aggregate 1210 kg 1150 kg 1150 kg 1188 kg
Kaiapoi sand 370 kg 600 kg 600 kg 495 kg
Yaldhurst sand 150 kg 150 kg 160 kg 124 kg
Ordinary portland
cement 308 kg 400 kg 410 kg 400 kg

High-range water-
1.5 L 2.0 L 2.5 kg 2.5 L
reducing admixture
Water 195 L 160 L 160 L 138 L
Silica fume solid — — — 40 kg
Fig. 3—Typical stress-strain curves for normal-yield-
Water-cement ratio 0.630 0.400 0.390 0.345
strength steel and high-yield-strength steel.

Fig. 4—Comparison of experimental and analytical curves for square cross sections confined by normal-yield-strength steel.

imen between the platen and the crosshead. The second method MATERIALS
used four linear potentiometers of two different gage lengths to A local ready-mix plant supplied the high-strength concrete
measure the axial strain of the specimen. Two 50 mm travel lin- used for the test units. The mixture proportions are shown in Ta-
ear potentiometers were used to measure strains over the antic- ble 3. The measured slumps of the concrete were all approxi-
ipated failure regions (central part of the specimen), that is, gage mately 120 mm. A typical stress-strain curve of the
reinforcement used is shown in Fig. 3.
length of approximately 300 mm. Two other linear potentiome-
ters were used to measure strains over the specimen’s length
GENERAL DESCRIPTION OF TEST BEHAVIOR
from 40 mm from the top to 40 mm from the bottom of the spec- The measured stress-strain curves of confined concrete are
imen (a 640 mm gage [Fig. 2]). If a diagonal shear plane shown in Fig. 4 to 7. The previous theoretical curves determined
formed, this larger gage length could be used instead of the gage by Fafitis and Shah,3 Yong, Nour, and Nawy,4 Bjerkeli, To-
length over the central part of the specimen. maszewicz, and Jansen,5 Muguruma and Watanabe,6 Cusson

398 ACI Structural Journal/May-June 2001

Fig. 5—Comparison of experimental and analytical curves
for square cross section confined by ultra-high-yield- Fig. 6—Comparison of experimental and analytical curves for
strength (fyh = 1318 MPa) steel. circular cross section confined by normal-yield-strength steel.

and Paultre,7 and Razvi and Saatcioglu8 were compared with ultimate concrete longitudinal strain was defined as the
some experimentally derived stress-strain curves (Fig. 8). strain at first hoop or spiral fracture.
Stress-strain curves predicted by models of Fafitis and The effect of a change in compressive strength of plain
Shah, Yong et al., Bjerkeli et al., Muguruma et al., Cusson concrete fco′ on the degree of ductility of columns is signifi-
and Paultre, and Razvi and Saatcioglu are denoted by F&S, cant, particularly for fco′ exceeding 60 MPa. Regardless of
Yong, BJ, MU, C&P, and R&S, respectively. A comparative the concrete compressive strength, an increase in the con-
study showed that most of the empirical models are effective finement ratio increases the peak stress attained, increases
only in interpreting their own test results or selected data. In the ultimate strain at first hoop or spiral fracture, and de-
comparing these models with experimental results obtained creases the slope of the descending branch. An increasing
from this study, it was found that the models could not pre- spacing of transverse reinforcement tends to reduce the effi-
dict accurately the descending branch of the stress-strain ciency of the confinement. The failure of the normal- and
curve of confined high-strength concrete. In seismic design, high-strength concrete specimens confined by transverse re-
knowledge of the descending branch is very important to en- inforcement with high yield strength steel was sudden, vio-
sure proper behavior of columns, which undergo large defor- lent, and explosive. This is because the transverse
mations during an earthquake. The level of strength and reinforcement fractured when the longitudinal bars buckled
ductility depends very much on the slope of the descending at very high strain, and the core concrete was crushed explo-
branch. Effects of concrete compressive strength, transverse sively due to a lack of confinement (Fig. 9). Consequently
reinforcement strength, and the amount of confining reinforce- the specimen lost all of its load capacity. In contrast, the fail-
ment on confined high-strength concrete are investigated. ure observed for specimens confined by normal yield
The most significant parameters affecting the shape of the strength steel was usually gradual and quite gentle after the
stress-strain curve of confined high-strength concrete for all first transverse bar fractured (Fig. 10). In axial loading tests
section shapes are the volumetric ratio and the yield strength of normal- and high-strength concrete cylinders and prisms
of the confining reinforcement. As the yield strength of the confined by spirals or square helices with two different yield
confining reinforcement is increased, both the strength and strengths of transverse reinforcement, namely, Grade 430
ultimate longitudinal strain of the confined concrete in- and 1300 steel, the strength and ductility of the confined con-
creased, while the slope of the falling branch decreased. The crete were enhanced when high yield steel was used.

ACI Structural Journal/May-June 2001 399

Fig. 7—Comparison of experimental and analytical curves
for circular cross section confined by ultra-high-yield-
strength (fyh = 1318 MPa) steel.

DISCUSSION OF TEST RESULTS

Effect of concrete compressive strength
Concrete compressive strength is a significant factor on
the behavior of confined concrete, as can been seen in the
test results. For example, comparing 8A and 8B in Fig. 4,
which have the same volumetric ratio of confining reinforce-
ment but with different concrete compressive strengths, the
test results indicated that the confinement effectiveness in
term of axial strains was less with a higher concrete com-
pressive strength. This is because the lateral dilation of high-
strength concrete was smaller than that of normal-strength
concrete.
Referring to the studies of Mander, Priestley, and Park9 for
normal-strength concrete confined by normal yield strength
transverse steel, the equation for design purpose can be ap- Fig. 8—Theoretical normalized stress-strain curves for con-
proximated as fined concrete using different models compared with specimens.

f cc′ = f co′ + 5.5f l ′ (1) Circular confinement— For normal-strength concrete

In this study, for high-strength concrete confined by recti- f cc′ = f co′ + 4.6f l ′ (3)
linear or circular normal yield strength confining reinforce-
ment, the equation for design purpose can be approximated For high-strength concrete
as
f cc′ = f co′ + 2.7f l ′ (4)
f cc′ = f co′ + 4.0f l ′ (2)

Rectilinear confinement— For normal-strength concrete

For concrete with typical level of rectilinear or circular ultra
high yield strength confining reinforcement, the following
equations for design purposes can be approximated as: f cc′ = f co′ + 2.1f l ′ (5)

400 ACI Structural Journal/May-June 2001

Fig. 9—Specimens with ultra-high-strength (fyh = 1318 MPa)
confining reinforcement after testing.

For high-strength concrete

f cc′ = f co′ + 1.9f l ′ (6)

Fig. 10—Specimens with normal yield strength confining

Effect of yield strength of transverse reinforcement after testing.
reinforcement
In this study, almost all gages on the Grade 430 steel spirals
and hoops indicated that the spirals and hoops had yielded at the tribution of the longitudinal steel play an important role in
peak of the load-longitudinal strain curves or shortly after. the confinement of concrete.
This finding is independent of concrete compressive
strength. Hence, it can be concluded that it is reasonable to Effect of confining reinforcement spacing
take the nominal yield strength of the transverse steel when To prevent early loss of strength of high-strength con-
calculating the confining stress. Most of the strain gages on crete columns, caused by both buckling of reinforcing bars
the high yield steel helices or spirals, however, indicated that and excessively deep arching of the confined concrete be-
the helices and spirals had not yielded at the peak of the load- tween the spirals or hoops, all columns should have suffi-
longitudinal strain curves. This contradiction may be ex- ciently close spacing of transverse reinforcement. In this
plained as follows: the effectively confined area of core con- study, spacing of hoop or spirals varied from 1.66db to
crete was significantly reduced due to incipient or serious 5.83db. Comparing the behavior of Specimen 2B (s = 1.66db)
buckling of longitudinal reinforcement at or near such a and Specimen 11A (s = 5.14db), the stress-strain curves of
large longitudinal strain at peak stress where the ultra-high- 2A and 11A (Fig. 4) show that as the hoop spacing was in-
strength transverse steel can yield. Also due to passive con- creased, the behavior became less ductile, the enhanced
finement, the damage of core concrete may have already strength became smaller, and the longitudinal strain at the
been significant when the lateral pressure reached the max- first hoop fractured decreased. Based on the test results of
imum due to yielding of transverse reinforcement. On the this study, the maximum pitch should not exceed 4db for nor-
other hand, in the case of Grade 430 reinforcement, the re- mal yield strength steel if satisfactory behavior is to be
inforcement can yield at a significantly smaller longitudinal achieved. A pitch of 6db, however, will result in reasonable
strain of concrete and hence the effectively confined area will ductility. The same trend was observed for the high-strength
be at least as large as that estimated by the Mander, Priestley, concrete column confined by ultra high yield strength steel.
and Park’s method.9 Also, additional confinement due to lon- It is recommended, therefore, that the maximum pitch should
gitudinal reinforcement may also be expected unless the ax- not be more than approximately 5db if satisfactory behavior
ial strain of longitudinal reinforcement greatly exceeds the is to be achieved.
yield strain. Therefore, the full yield strength of ultra-high
steel cannot be used when calculating the confining stress. Effect of volumetric ratio of confining
reinforcement
Effect of confinement reinforcement configuration From Fig. 4 to 7, the effect of the volumetric ratio of confin-
The hoop configuration and the resulting distribution of ing reinforcement on the behavior of the specimen is demon-
the longitudinal steel play an important role in the confine- strated. As expected, the larger the volumetric ratio of
ment of concrete. In this study, it was possible to observe the confining reinforcement, the more ductile the behavior of the
effect of varying the hoop configuration. Comparing the be- specimen. For the confined specimen using high yield strength
havior of Specimen 4A, constructed using single hoops, and of transverse steel, the strength and ductility of the concrete
Specimen 11A, confined using double hoop configuration, it were remarkably enhanced. This can been seen by comparing
was observed that Specimen 4A behaved in less ductile fash- the test results of three specimens (1HA, 3HB1, and 3HC1)
ion with a sharp dropping of load. A similar comparison with the same volumetric ratio of confining reinforcement but
was made between Specimen 4B and 11B. In this case, the with different concrete compressive strengths; the peak stress
effect was even more pronounced, with Specimen 4B enhancements due to confinement were 2.04, 1.57, and 1.28,
showing a pronounced reduction in load capacity after respectively. When the volumetric ratio was reduced to 1%
maximum load had been reached. These comparisons show level, the behavior of a high-strength concrete specimen was
that there was an improvement in the strength and ductility similar to that of an unconfined specimen. The test results in-
of the specimens as a result of better hoop configurations, dicted that a limitation on the minimum volumetric ratio of
indicating that the hoop configuration and the resulting dis- confining reinforcement is necessary to ensure ductile behav-

ACI Structural Journal/May-June 2001 401

 herein. The reason can be explained as follows: the confining
effects are negligible in the initial part of the ascending
branch of the stress-strain curve due to passive confinement
of transverse reinforcement and small transverse strain. This
means that the model curve needs to be changed little in the
initial part of ascending branch but significantly altered near
and after peak stress regions depending on the magnitude of
confinement. It is hard to find a single polynomial function
to satisfy such a condition.
In the case of a model that consists of several branches de-
fined by different functions, it is easier to adjust the model
curve to the experimental curve because the characteristics
of ascending and descending branches can be controlled in-
dependently. Hence, the authors decided to establish a model
Fig. 11—Stress-strain relationship proposed for confined by modifying the model proposed by Muguruma and Wa-
high-strength concrete. tanabe,6 as shown as follows. The model consists of three
branches expressed by Eq. (7) to (9) and a tail with a constant
ior of confined high-strength concrete, as was evident by the stress of 0.4fcc′ .
behavior of Specimens 10A, 10B, and so on.
When 0 ≤ εc ≤ εco
Effect of amount of longitudinal steel
The confinement effectiveness is more dependent on the ( f co′ – E c ε co ) 2
spacing of confining reinforcement than on the amount of f c = E c ε c + -------------------------------
- εc (7)
2
ε co
longitudinal reinforcement. Thus, while doubling of the
amount of longitudinal reinforcement improves the descend-
ing part of the curve, the improvement is not as significant as when εco ≤ εc ≤ εcc
that obtained by decreasing the hoop spacing and volumetric
ratio of confining reinforcement or changing the hoop con-
( f cc′ – f co′ ) 2
figuration. Thus, it is apparent that the distribution of the lon- f c = f cc′ – ------------------------------ × ( ε c – ε cc ) (8)
2
gitudinal reinforcement in the column is more important for ( εcc – ε c co )
confinement effectiveness for a given amount of reinforce-
ment. Also, there seems to be a limit on the amount of longi-
tudinal reinforcement necessary for ductile behavior of when ε > εcc
confined high-strength concrete. There must not be less than
eight longitudinal bars distributed along the perimeter of the f cc′
column. f c = f cc – β -------
- × ( ε c – εcc ) ≥ 0.4f cc′ (9)
εcc

MODELING OF STRESS-STRAIN RELATIONSHIP

OF CONFINED CONCRETE A typical stress-strain curve of confined concrete deter-
The modeling of material stress-strain relationships is a mined by the model is schematically shown in Fig. 11. To
basic requirement for the prediction of the behavior of struc- draw the stress-strain curve of confined concrete using the
tural elements. Theoretical moment-curvature analysis for previous equations, the maximum strength of confined con-
reinforced concrete structural elements, indicating the avail- crete fcc′ , the axial strain at maximum strength εcc, and the β
able flexural strength and ductility can be constructed pro- value, which controls the slope of the descending branch,
viding the stress-strain relations for the concrete and steel is need to be determined. The stress-strain curve is terminated
known. The formulation of mathematical stress-strain rela- at the ultimate compressive strain εcu where the first hoop
tionships involves a large number of factors, particularly for fractures due to serious buckling of longitudinal bars. These
high-strength concrete. variables are determined as follows.
The principal factors considered in modeling the stress- Maximum strength of confined concrete fcc′ —In the last
strain relationships are as follows: several decades, failure criteria associated with a maximum
1. Type and strength of concrete; stress surface have been established for concrete under triax-
2. Amount, pitch, and configuration of transverse rein- ial states of stress. Several numerical models, based on test
forcement; results, have been proposed to represent this failure surface.
3. Amount and distribution of longitudinal reinforcement; In this study, the method used by Mander, Priestley, and
4. Mechanical properties of transverse and longitudinal re- Park9 was employed. The five-parameters multiaxial failure
inforcement; criteria of William and Warnke10 were used to describe the
5. Ratio of confined area to gross area; theoretical ultimate strength surface in this study. The tensile
6. Monotonic and cyclic loading; and and compressive meridians are expressed as follows
7. Size and shape of confined concrete.
τ oct σ oct σ oct 2
-------- = a 0 + a 1 --------- + a 2  --------- (10)
Monotonic loading curve f co′ f co′ f co′ 
In the case of monotonic loading, it was not easy to find or
derive a single polynomial function that fits well with all
types of the stress-strain curves observed in tests described at θ = 0 degrees (tensile meridian)

402 ACI Structural Journal/May-June 2001

 τoct σoct σ oct 2
-------- = b 0 + b 1 --------- + b 2  --------- (11)
f co′ f co′  f co′ 

at θ = 60 degrees (compressive meridian)

Because attention is paid only to the compressive meridian,

the equation of compressive meridian can be transformed as
follows

σ1 + σ2 + σ3 fcc′ 2f l ′
σ oct = ------------------------------ = – ----------
- – ----------
- (12)
3f co′ 3f co′ 3f co′

Fig. 12—Ultimate strength surfaces on octahedral plane.

fl ′ 
τoct = ----------- ( f cc′ – f l ′ ) = – 2  σ oct + -------
2

- (13)
3f co′ f co′  For the properties of concrete confined by high yield
strength steel, it should be noted that fl′ defined by Mander,
For triaxial case σ1 = σ2 = fl′, then Priestley, and Park9 was based on the yield strength of the
confining strength of the confining steel. The comparison
between Mander, Priestley, and Park’s model and the test re-
2
τ oct = b 0 + b 1 ( σoct ) + b 2 ( σ oct ) (14) sults, however, show that the former model generally over-
estimated fcc′ for concrete confined by high yield strength
steel. This is because the ultra high yield strength of steel is
Substituting Eq. (12) into Eq. (14) not developed until at high transverse strains, hence there is a
delayed confining effect on the concrete relative to that pro-
 3 ( b1 + 2 vided by normal yield strength steel.
f cc′ = f co′  ------------------------
-+ (15) In this study, a regression analysis was conducted using
 2b 2 the test results to modify the fl′/fco′ value used to calculate
fcc′ . Because the test results cover both confined normal- and
high-strength concrete, Eq. (18) proposed by Mander, Priest-
2
 3 ( b 1 + 2 ) 9b 0 9 2 f l ′ fl ′ ley, and Park9 was adopted for modification. The modifica-
 --------------------------- – -------- – ---------- -------- + – 2 --------  tion factor was found to be
 2b 2  b2 b 2 f co ′ f co′

fl ′ fl ′
Mander, Priestley, and Park9 determined fcc′ using the test f cc′ = f co′ – 1.254 + 2.254 1 + 7.94 α s -------
- – 2 α s -------
- (18)
results of Schickert and Winkle11 (b0 = 0.12229, b1 = –1.15, f co′ f co′
and b2 = –0.315)
fl ′
When fco′ ≤ 52 MPa α s = ( 21.2 – 0.35f co′ ) -------
- (19)
fl ′ fl ′ f co′
f cc′ = f co′ – 1.254 + 2.254 1 + 7.94 -------
- – 2 -------
- (16)
f co′ f co′
fl ′
When fco′ > 52 MPa α s = 3.1 -------
- (20)
In this study, using the test results of Khaloo and Ahmad 12 for f co′
high-strength concrete, b0 = 0.113, b1 = –1.26, and b3 = –0.559,
and fcc′ for high-strength concrete was proposed as follows where fl′ is the effective lateral confining pressure, calcu-
lated using the equations proposed by Mander, Priestley,
and Park, given as follows:
fl ′ fl ′
f cc′ = f co′ – 0.413 + 1.413 1 + 11.4 -------
- – 2 -------
- (17) For circular confined section
f co′ f co′
f l ′ = 0.5K e ρ s f yh (21)
In this study, fcc′ was established using the results of active
confinement. The peak stresses of concrete under passive For circular hoops
confinement with different stiffness, however, were always
around the failure envelope of active confinement and the 2
differences were found to be insignificant.13 Therefore, it ( 1 – 0.5s ′ ⁄ d s )
K e = ------------------------------------
- (22)
seems reasonable to consider that the difference between the 1 – ρ cc
active and passive confinement fracture envelope is negligible.
In effect, the failure envelope could be considered to be For spirals
stress path independent. Comparison between the previous
equations is shown in Fig. 12. It is evident that the octahedral
( 1 – 0.5s ′ ⁄ d s )
shear stress of high-strength concrete is smaller than that of K e = ---------------------------------- (23)
normal-strength concrete at a given octahedral normal stress. 1 – ρ cc

ACI Structural Journal/May-June 2001 403

 For rectangular confined section: β = 0.07 when fyh > 1200 MPa and fco
′ ≤ 80 MPa (37)
The effective lateral pressure is given with good accuracy by
β = 0.1 when fyh > 1200 MPa and fco′ > 80 MPa (38)
f l ′ = 0.5K e ( ρ x + ρ y ) f yh (24)
For rectilinear confinement with ordinary yield strength
where (fyh ≤ 550 MPa), it was not possible to determine a uniform
coefficient to fit all experimental curves. Therefore, the β
n 2
value for each experimental curve was found using a trial-
Ci s′ s′ and-error method. The β values previously recommended do
1– ∑ 6b c d c
- × 1 – 0.5 ----- × 1 – 0.5 -----
-------------
bc dc not cover all ranges of material strength with continuity, and
i=1 hence, improvement in future research is expected.
K e = ---------------------------------------------------------------------------------------------------------
- (25)
1 – ρ cc Maximum concrete strain εcu—The maximum concrete
strain εcu has often been used in ductility calculations.
Axial strain at maximum strength εcc—For circular con- Scott, Park, and Priestley14 observed that it is reasonably
finement with ordinary-strength steel conservative to define the limit of useful concrete compres-
sive strain as the strain at which fracture of a hoop first oc-
ε cc fl ′ 2 curs. In this study, the following empirical equations were
- = 1.0 + 384 -------
------ - (26) proposed to estimate the ultimate compressive strain εcu,
ε co f co′ where the first hoop fractures occurs due to serious buckling
of longitudinal bars.
For rectilinear confinement with ordinary-strength steel For circular confinement with normal-strength steel

ε cc fl ′ 0.7 ε cu fl ′
------
- = 1.0 + 11.3 -------
- (27) ------- = 2 + ( 143.5 – 1.48f co′ ) ------- ′ < 80 MPa (39)
- when fco
εco f co′ ε co f co′

For circular confinement with ultra-high-strength steel ε cu fl ′

------- = 2 + ( 89.8 – 0.74f co′ ) ------- ′ ≥ 80 MPa (40)
- when fco
ε co f co′
ε cc fl ′ 
- = 1.0 + ( 120 – 1.554f co′ )  -------
------ - when fco ≤ 50 MPa (28)
ε co  f co′ 
For rectilinear confinement with normal-strength steel

ε cc fl ′ 
- = 1.0 + ( 71.4 – 0.623f co′ )  -------
------ ′ > 50 MPa (29)
- when fco ε cu fl ′
------- = 2 + ( 122.5 – 0.92f co′ ) ------- ′ < 80 MPa (41)
εco  f co′  - when fco
ε co f co′

For rectilinear confinement with ultra-high-strength steel

εcu fl ′
------- = 2 + ( 82.75 – 0.37f co′ ) ------- ′ ≥ 80 MPa (42)
- when fco
εco f co′
ε cu fl ′
- when fco ≤ 50 MPa (30)
------- = 2.0 + ( 87 – 1.06f co′ ) -------
ε co f co′
For circular confinement with ultra-high-strength steel

εcu fl′ εcu fl′

------- = 2.0 + ( 53.4 – 0.42f co′ ) ------- ′ > 50 MPa (31)
- when fco ------- = 2.0 + ( 86.7 – 1.06f co′ ) ------- ′ ≤ 50 MPa (43)
- when fco
εco f co′ εco f co′

Factor to control the slope of the descending branch β—

ε cu fl ′
For concrete confined by circular confinement ------- = 2.0 + ( 53.5 – 0.42f co′ ) ------- ′ > 50 MPa (44)
- when fco
ε co f co′
β = 0.2 when fyh ≤ 550 MPa and fco
′ ≤ 80 MPa (32)
For rectilinear confinement with ultra-high-strength steel
β = 0.08 when fyh > 1200 MPa and fco′ ≤ 80 MPa (33)
εcu fl ′
- when fco ≤ 50 MPa (45)
------- = 2.0 + ( 70.0 – 0.6f co′ ) -------
β = 0.2 when fyh > 1200 MPa and fco
′ > 80 MPa (34) εco f co′

For concrete confined by rectilinear confinement

ε cu fl ′
------- = 2.0 + ( 49.0 – 0.2f co′ ) ------- ′ > 50 MPa (46)
- when fco
ε co f co′
fl ′  1 ⁄ 3
β = ( 0.048f co′ – 2.14 ) – ( 0.098f co′ – 4.57 )  -------
- (35)
f co′ 
Figure 4 to 7 show comparisons between the experimental
curves and theoretical curves determined using the previously
When fyh ≤ 550 MPa and fco
′ > 75 MPa (36) described method. Overall the stress-strain model of con-

404 ACI Structural Journal/May-June 2001

Fig. 13—Stress-strain curves of specimens under cyclic and
monotonic loading conditions.

fined high-strength concrete proposed herein is quite close to

the tested data, especially in the descending branch of the
stress-strain curve.
It was also evident that none of the available high-strength
concrete models are able to predict satisfactorily the measured
stress-strain behavior of high-strength concrete confined by
circular or rectangular transverse reinforcement with different
yield strength. All the models predict the ascending branch
of the stress-strain relationship fairly well, whereas the pre-
dicted descending branch of the curve is not consistent. All Fig. 14—Comparison of experimental and analytical cyclic
the models underestimated the maximum concrete strength, stress-strain curves.
thus giving a very conservative stress-strain prediction for
concrete confined by ultra high yield strength steel. More
and monotonic loading conditions. Comparison between the
attention must be given to the following aspects, especially
experimental results and those predicted using the cyclic stress-
for the case of high-strength concrete. strain model proposed by Mander, Priestley, and Park’s9 mod-
1. It is notable that the magnitude of the compressive con- el modified by Dodd and Cooke15 showed very good agree-
crete strain measured in the postpeak stress region depends ment (Fig. 14).
significantly on the gage length and location of strain mea-
surement. This aspect does not relate directly to the confin- CONCLUSIONS
ing effect but influences their assessment. Because the gage Based on the results of this investigation, the following
length and the location of strain measurement are different conclusions can be drawn.
among researchers, it is difficult to compare data from differ- 1. The most significant parameters affecting the shape of
ent sources; the stress-strain curve of confined high-strength concrete
2. In the past tests on confined high-strength concrete, for all section shapes are the volumetric ratio and the yield
small scale models with less than 150 mm square section strength of the confining reinforcement. As the yield
were used due to the limit of loading capacity of the testing strength of the confining reinforcement increases, the
machine. The difference in specimen size may also lead to strength of the confined concrete also increases. The ulti-
difficulty in comparing test results from different sources; mate longitudinal strain was defined as the strain at first
and hoop or spiral fractured;
3. The relative stiffness of the testing machine and the con- 2. The influence of concrete compressive strength on the
crete specimen is one of the key points, particularly for high- degree of column ductility is significant, particularly for
strength concrete. A flexible testing machine used close to its concrete compressive strengths exceeding 60 MPa. Re-
upper capacity will tend to snap back once peak load is reached, gardless of the concrete compressive strength, however, an
giving nonrepresentative results. This factor will make compar- increase in the confinement ratio increases the peak stress
ing the test results from different sources difficult. attained, increases the ultimate strain at first hoop or spiral
fracture, and decreases the slope of the descending branch-
Cyclic loading curves ing branch. An increase in the spacing of transverse rein-
For ordinary strength concrete confined by ordinary forcement tends to reduce the efficiency of the
strength steel, it is normally assumed that the monotonic confinement;
loading curve corresponds to the envelope curve of the 3. A stress-strain model for high-strength concrete confined
stress-strain curves under cyclic loading. In this study, it using different types of normal-strength and high-strength
was found that the assumption that the monotonic loading confining reinforcement is proposed. The model is found to
curve represents the skeleton curve of the stress-strain give reasonably good predictions of experimental behavior of
curves under cyclic loading is still valid, regardless of the circular and square specimens with high-strength concrete
concrete compressive strength and the yield strength of confined by either normal or high yield strength confining re-
transverse reinforcement. Figure 13 shows the results for inforcement with various configurations;
two specimens with the same confinement and nearly the 4. In axial loading tests on normal- and high-strength con-
same compressive strength of concrete tested under cyclic crete cylinders and prisms confined by spirals or helices with

ACI Structural Journal/May-June 2001 405

two ultra high and normal strength transverse reinforcement, εcc = strain at maximum confined strength of concrete fcc′
the strength and ductility of the confined concrete was sig- εco = compressive strain at maximum in-place unconfined concrete
strength fco′
nificantly enhanced when the ultra-high yield strength of
εcu = ultimate concrete compressive strain
steel was used. The expected general improvement of the be- εy = yield strain of steel
havior of confined high-strength concrete with confining re- ρcc = volumetric ratio of longitudinal reinforcement in confined core
inforcement was observed in the tests; concrete
5. To prevent a relatively early loss of strength of high- ρs = volumetric ratio of confining reinforcement to core concrete
strength concrete caused by both buckling of reinforcing ρx = lateral confining steel parallel to x-axis
bars and excessively deep arching of the confined concrete ρy = lateral confining steel parallel to y-axis
in between the spirals and hoops, all specimens should have
sufficiently close spacing of transverse reinforcement. A REFERENCES
maximum pitch of 4db for Grade 430 transverse reinforce- 1. Bing, L.; Park, R.; and Tanaka, H., “Strength and Ductility of Rein-
ment or 5db for ultra-high-strength steel is recommended. forced Concrete Members and Frames Constructed using High-Strength
Concrete,” Research Report 94-5, University of Canterbury, New Zealand,
For rectilinear confinement in high-strength concrete, the May 1994, 389 pp.
number of longitudinal bars must not be less than eight and 2. Martinez, S.; Nilson, A. H.; and Slate, F. O., “Spirally Reinforced
those bars should be distributed along the perimeter of the High-Strength Concrete Columns,” ACI JOURNAL, Proceedings V. 81, No.
column; 5, Sept.-Oct. 1984, pp. 431-442.
6. The analytical stress-strain model proposed in this paper 3. Fafitis, A., and Shah, S. P., “Lateral Reinforcement for High-Strength
Concrete Columns,” SP-87, American Concrete Institute, Farmington
was found to give reasonably good predictions of the exper- Hills, Mich., 1985, pp. 213-232.
imental behavior of circular and square specimens with high- 4. Yong, Y. Y.; Nour, M.; and Nawy, E. G., “Behavior of Lateral Con-
strength concrete confined by either normal or high yield fined High-Strength Concrete under Axial Loads,” Journal of Structural
strength confining reinforcement with various configura- Engineering, ASCE, V. 114, No. 2, 1988, pp. 332-350.
tions; and 5. Bjerkeli, L.; Tomaszewicz, A.; and Jansen, J. J., “Deformation Proper-
7. When cyclic and monotonic loading were applied to ties and Ductility of High-Strength Concrete,” Proceeding of the Second
International Symposium on Utilization of High-Strength Concrete, Uni-
prisms and cylinders confined by transverse reinforcement versity of California, Berkeley, Calif., May 1990.
with yield strength less than 500 MPa, the envelope skeleton 6. Muguruma, H., and Watanabe, F., “Ductility Improvement of High-
of stress-strain curves coincided closely with the stress- Strength Concrete Column With Lateral Confinement,” Proceedings of the
strain curve obtained from a monotonic loading test, even for Second International Symposium on Utilization of High-Strength Concrete,
high-strength concrete specimens. Berkeley, Calif., 1990.
7. Cusson, D., and Paultre, P., “Stress-Strain Models for Confined High-
Strength Concrete,” ASCE, Journal of Structural Engineering, V. 121, No. 3,
ACKNOWLEDGMENTS Mar. 1995.
Financial support from the Cement and Concrete Association of New 8. Razvi, S., and Saatcioglu, M., “Confinement Model for High-Strength
Zealand, the New Zealand Concrete Society, Firth Certified Concrete, Pacif- Concrete,” ASCE, Journal of Structural Engineering, V. 125, No. 3, Mar.
ic Steel Ltd., Koshuha-Netsuran Co. of Japan, Taisei Corp. of Japan, WG 1999.
Grade (NZ) Ltd., and the New Zealand Ministry of External Relations and 9. Mander, J. B.; Priestley, M. J. N.; and Park, R., “Seismic Design of
Trade is gratefully acknowledged. Bridge Piers,” Research Report 84-2, Department of Civil Engineering,
University of Canterbury, New Zealand, 1984, 444 pp.
NOTATION 10. William, K. L., and Warnke, E. P., “Constitutive Model for the Triax-
db = nominal diameter of longitudinal reinforcing bar ial Behavior of Concrete,” International Association for Bridge and Struc-
dc = diameter of confined concrete core of circular column section, tural Engineering, Proceedings, V. 19, 1975.
measured to center-line of spiral or circular hoop 11. Schickert, G., and Winkle, H., “Results of Tests Concerning Strength
ds = effective core diameter between circular hoop or spiral bar centers and Strain of Concrete Subjected to Multiaxial Compressive Stresses,”
Ec = Young’s modulus of elasticity for concrete Deutscher Ausschuss Fur Stahlbeton, Heft, 277, Berlin, Germany, 1977.
Es = Young’s modulus of elasticity for steel 12. Khaloo, A. R., and Ahmad, S. H., “Behavior of High-Strength Con-
fc = concrete stress crete under Torsional Triaxial Compression,” ACI Materials Journal, V. 86,
fc′ = specified concrete compressive strength No. 6., Nov.-Dec. 1989, pp. 550-558.
fcc′ = confined concrete compressive strength 13. Sun, X.; Tan, T. H.; and Irawan, P., “Effect of Stress-Path on the Fail-
fco′ = in-place unconfined concrete compressive strength ure of Concrete under Triaxial Stress,” Proceedings of the EASEC-7 Con-
fl = transverse confining stress ference, Koichi, Japan, 1999.
f l′ = effective transverse confining stress 14. Scott, B. D.; Park, R.; and Priestley, M. J. N., “Stress-Strain Behavior
fy = yield strength of steel in tension of Concrete Confined by Overlapping Hoops at Low and High Strain
fyh = yield strength of transverse reinforcing steel Rates,” ACI JOURNAL, Proceedings V. 79, No. 1, Jan.-Feb. 1982, pp. 13-27.
Ke = confinement effectiveness coefficient, based on area ratio 15. Dodd, L. L, and Cooke, N., “The Dynamic Behavior of Reinforced
s′ = clear spacing between circular hoops or spirals Concrete Bridge Pier Subjected to New Zealand Seismicity,” Research
sh = center-to-center spacing of spiral or hoop sets Report 92-6, Department of Civil Engineering, University of Canterbury,
εc = concrete compressive strain New Zealand, 1992, 460 pp.

406 ACI Structural Journal/May-June 2001