ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 94-S5

Modification of the ACI Rectangular Stress Block

for High-Strength Concrete

by Hisham H. H. Ibrahim and J. G. MacGregor

Most concrete codes do not explicitly cover concrete with strengths to 100 MPa, [14,500 psi]; and UHSC over 100 MPa) are
above 50-60 MPa (7000 to 9000 psi). Compression stress blocks reviewed and new stress block equations are suggested. The
from three current codes that do allow design for high-strength study includes compression zones with rectangular and
concrete (HSC) sections are presented. Comparisons of test data triangular cross-sectional areas.
from the literature to strengths calculated using the current ACI
rectangular stress block indicate it is not conservative for high
strength concrete column sections failing in compression. New RESEARCH SIGNIFICANCE
equations for the parameters that define the ACI rectangular stress This paper compares the results from different tests
block are suggested and compared to tests. conducted on concentrically and eccentrically loaded normal
strength concrete (NSC), HSC, and UHSC columns to inter-
Keywords: columns; design; flexural strength; high-strength concrete; column action diagrams based on the ACI rectangular stress block.
interaction diagrams; reinforced concrete; stress-strain relationships. New equations for the parameters that define the ACI rectan-
gular stress block are suggested.
INTRODUCTION
In North America the design of reinforced concrete for REVIEW OF CURRENT DESIGN CODES
flexure is based on a rectangular compression stress block. ACI 318-M896 and Canadian Code CAN3-A23.3-M847
An analytical study by Garcia and Nilson,1 based on stress- The ACI code and the Canadian code have identical stress
strain curves of concentrically loaded cylinders with a peak block provisions. Both codes specify the concrete compres-
stress equal to 0.85fc′ suggests that the ACI rectangular sion strength using 152 x 304 mm (6 x12 in.) cylinders. In
stress block is unconservative by up to 12 percent for the the ACI and CSA codes, rectangular, trapezoidal, parabolic,
design of high strength concrete (HSC) rectangular sections or other stress blocks may be assumed, provided the relation-
subjected to axial loads with small eccentricities. The ACI ship between the concrete compressive stress distribution
code equations for the flexural stress block for HSC were and the resulting concrete strain is in agreement with test
based on tests by Kaar, et al.,2 and Nedderman,3 where the data. Figure 1(c) shows the stress block parameters obtained
sections were subjected to axial forces with small eccentric- from tests: k1, k2, and k3. The rectangular stress block is
ities. Leslie, et al.,4 and Garcia and Nilson have proposed defined by the parameters α1 and β1 as shown in Fig. 1(d)
non-rectangular stress blocks for design of HSC sections. (k1, k2, k3, α1 and β1 are defined in the Notation). A distinc-
Swartz, et al.,5 and Leslie, et al., suggest that the code value tion has been made between the k1, k2, and k3 obtained from
of 0.003 for the ultimate compressive strain is not conserva- tests and the parameters α1 and β1 used to define the rectan-
tive for design of HSC sections. gular stress block because the rectangular stress block uses
two parameters to approximate the effects of the three
In order to design a concrete member that is subjected to
measured in tests.
different cases of loading, an interaction diagram is gener-
In the ACI and CSA codes the parameter α1 is assumed to
ated for the cross-section. Using different stress-strain equa-
have a constant value of 0.85. The parameter β1 is equal to 0.85
tions for concrete in compression can lead to substantial
for concrete strengths, fc′ up to 30 MPa and is reduced contin-
differences in the resulting interaction diagrams.
uously at a rate of 0.08 for each 10 MPa of strength in excess
The current Norwegian code has a stress block specifically of 30 MPa. The parameter β1 is not taken less than 0.65. The
derived for HSC. It is conservative compared to the ACI limiting compressive strain εuis assumed to have a constant
code. The difference between interaction diagrams gener- value of 0.003. The parameter k2 is taken equal to 0.5β1.
ated using these two codes for a HSC section is significant
for the part of the diagram where the axial load is high (when
the section is subjected to axial forces with small eccentrici-
ties). For the pure moment case the effect of using a different ACI Structural Journal, V. 94, No. 1, January-February 1997.
stress block is negligible. Received February 8, 1995, and reviewed under Institute publication policies. Copy-
right © 1997, American Concrete Institute. All rights reserved, including the making of copies
In this paper the ACI design procedures for high-strength unless permission is obtained from the copyright proprietors. Pertinent discussion including
author’s closure, if any, will be published in the November-December 1997 ACI Struc-
concrete and ultra high-strength concrete sections (HSC up tural Journal if the discussion is received by July 1, 1997.

40 ACI Structural Journal/January-February 1997

Hisham H. H. Ibrahim is employed by Buckland and Taylor Ltd., Consultant and
Bridge Design Engineers, in Vancouver, Canada. He received his BSc and MSc from
Cairo University and his PhD from the University of Alberta. His research interests
are reinforced and prestressed structures and the applications of high-strength con-
crete in concrete structures.

James G. MacGregor is an ACI Fellow as well as University Professor Emeritus at

the University of Alberta. He is a member of ACI Committees 318, 441, and 445, and
is a past president of the Institute.

Norwegian Code NS 3473-19898

The highest grade for normal density concrete based on
tests of 100x100x100 mm (4 in.) cubes is 105 MPa (C
105) (15,225 psi). The conversion factor to a 152/304 mm
cylinder is given as 0.8 up to grade C55 and the cube
strength minus 11 MPa (1595 psi) for higher grades. The
Norwegian code assumes an effective k3 value (ratio between
structural strength to cylinder strength, see Fig. 1(c)) that
decreases with increasing strength, going from 0.84 for
cylinder strength of 20 MPa (2900 psi) to 0.66 for cylinder
strength of 94 MPa (13,630 psi). Fig. 2(a) illustrates the
recommended stress strain relationship for different concrete
grades. The strain values at the peak stress, εco , change from
0.00197 to 0.00215 and the strain values at ultimate strain,
εcu, change from 0.00384 to 0.00268 with a change in the
cylinder strength from 20 to 94 MPa. Equations that define
the parameters of the stress-strain relationship are given in
the Norwegian code. The symbol fcn in the figure represents Fig. 1—Stress block parameters for rectangular sections.
the strength of the concrete in the structure (fcn /fc′ is equal to
k3). The moment and the normal force capacity of rectan- Table 1—Stress block parameters of Norwegian
gular sections can be calculated by direct use of the magni-
tude of the resultant force (R), which corresponds to k1, the Code NS 3473-19898
strength of concrete in the structure, fcn, and the position of Strength in
the resultant (C), which corresponds to k2. Table 1 (from Cube strength Cylinder structure
fcube , MPa strength fc′, MPa fcn = k3 fc′, MPa R = k1 C = k2
Thorenfeldt, et al.9) gives values of fcn, R and C for different
25 20 16.8 0.877 0.445
concrete strengths.
35 28 22.4 0.829 0.424
45 36 28.0 0.804 0.413
Finnish Code Rak MK4 198910
55 44 33.6 0.778 0.403
The highest grade concrete defined using 150 x 150 x 150 mm
65 54 39.2 0.751 0.392
(6 in.) cubes is 100 MPa (K100) (14,500 psi). The conver-
sion factor to a 152x304 mm cylinder is taken as the cube 75 64 44.8 0.722 0.382
strength minus 8 MPa (1160 psi). The Finnish code assumes 85 74 50.4 0.66 0.361
a constant value of k3 equal to 0.70. 95 84 56.0 * *

Figure 2(b) shows the stress-strain relationship recom- 105 94 61.6 * *

*
mended by the code. The symbol fck in the figure represents Depends on modulus of elasticity measured in tests of concrete to be used.
Note: 1 MPa = 145 psi.
the concrete strength in the structure.

CEB/FIP Model MC9011 DERIVATION OF STRESS BLOCK PARAMETERS

The highest grade concrete defined using the uniaxial FOR USE IN THE ACI RECTANGULAR STRESS
BLOCK
compression strength of 150x300 mm cylinders is 80 MPa
Structure to cylinder compression strength
(C80) (11,600 psi) for normal density concrete. Stress-strain ratio, k3
diagrams of the form shown schematically in Figure 2(c) are The parameter k3 is the ratio between the strength of the
presented for analytical purposes. The value of εc1 is set at concrete in columns compared with concrete of the same
0.0022. Values of Ec, Ec1, εcu vary with changes in the mix in standard compression test cylinders (see Figure 1(c)).
concrete grade. This parameter can be determined from tests on concentri-
For design, two stress blocks are described, the first a cally loaded columns by dividing the part of the maximum
parabola-rectangle that consists of a second degree parabola load carried by the concrete section by the gross cross-
with apex at a stress of 0.85fc′ and a strain of 0.002, followed section area multiplied by the cylinder compression strength.
by a yield plateau with a uniform stress of 0.85fc′ for strains It can also be determined from tests on eccentrically loaded
between 0.002 and 0.0035, and the second a rectangular columns, as described in Reference 2, and those reported in
stress block with α1 a function of the concrete strength, two companion papers, References 12 and 13. In this case
decreasing from 0.782 for fc′ = 20 MPa (2900 psi) to 0.578 the parameter k3 cannot be obtained from the equilibrium of
for f'c = 80 MPa (11,600 psi), and β1= 1.0. the external and internal loads since the column strength is

ACI Structural Journal/January-February 1997 41

 Fig. 3—Values of k3 from tests of eccentrically-loaded
specimens.

Fig. 4—Values of k3 from tests of concentrically-loaded

columns.

Schade19 and Ibrahim and MacGregor12 included both plain

and reinforced concrete specimens. For the latter two series,
the net loads and moments carried by the longitudinal rein-
forcement at failure were subtracted out, based on the
reported stress strain curves, before computing k3. The figure
Fig. 2—Stress-strain diagrams for concrete in compression. indicates that despite the scatter in the test results, the ACI
value of 0.85 is conservative with respect to the values of k3
related to the cylinder strength by the parameters k1k3 as obtained from eccentrically loaded tests. This conclusion
illustrated in Figure 1(c) and it is necessary to compute the does not mean that the ACI code is conservative in
stress-strain relationship of the cross-section with certain predicting the applied loads on eccentrically loaded sections
assumptions to calculate k3. since the force in the cross-section is related to the cylinder
The report of ACI Committee 36314 on HSC recommends strength by the product of the parameters k1k3 and the
the use of a constant value of k3 equal to 0.85. This recom- moment arm factor k2, as discussed before. Although not
mendation was based, in part, on the results of an experi- evident in the figure, there were no significant differences
mental program on concentrically loaded plain concrete between the k3 values of the plain concrete specimens and
high-strength cylinders of different sizes, conducted by the reinforced concrete specimens.
Martinez, et al.15 The CEB/FIP16 state-of-the-art report on Figure 4 shows the values of k3 obtained from 90 tests of
HSC included results from two test series on concentrically concentrically loaded columns. The columns had a wide
loaded reinforced HSC columns with strengths ranging from range of amounts of lateral reinforcement ranging from
75 to 100 MPa (10,900 to 14,500 psi), conducted by Høiseth columns without any lateral reinforcement to columns that
and Jensen17 and Bjerkeli, et al.18 The average value of k3 had more reinforcement than required by the ACI code for
was 0.82 in the first test series and ranged between 0.94 to seismic regions. The loads carried by the longitudinal rein-
0.96 in the second test series. forcement at failure were subtracted out, based on the
Figure 3 shows k3 values obtained from 49 tests of eccen- reported stress strain curves, before computing k3. The
trically loaded columns with concrete strengths greater than values of k3 obtained from the test series conducted by
40 MPa (5800 psi). All of the specimens were tested under a Sheikh and Uzumeri20 were calculated using the gross cross-
triangular strain distribution with zero strain on one side of sectional area and the peak load that occurred after spalling
the column. The test series by Kaar, et al.,2 and Swartz, et of the cover. This peak load was higher than the maximum
al.,5 were on plain concrete specimens. The test series by load before spalling because of the confinement effects. For

42 ACI Structural Journal/January-February 1997

the other tests the k3 value was calculated at the first peak
that occurred at or before spalling of the cover.
Figure 4 shows that the ACI value of 0.85 is not conserva-
tive. All the test results of Cusson and Paultre21 were lower
than the code value. The figure also shows more scatter
between the results of different HSC column test series than
different NSC column test series, and a trend for k3 to
become lower as the concrete strength increases. Some test
series of HSC columns showed high values of k3, on average,
and others showed quite low values of k3. High values of k3
were also reported in the FIP/CEB16 report on HSC, as
discussed before. Other tests conducted at the University of
Toronto (not plotted in Fig. 3 and 4) showed values of k3 as
low as 70 percent of the value given by the ACI code. These
tests included eccentrically loaded columns made with 60
MPa (8700 psi) concrete tested by Sundararaj22 and concen-
trically loaded columns made with 66 MPa concrete tested
by Polat.23 Tests on concentrically loaded spirally reinforced
cylinders made with 63 MPa (9100 psi) concrete conducted
by Martinez24 (also not plotted in Fig. 4) showed an average
value of k3 equal to 0.8. The lower strength concrete speci-
mens of the same test series showed an average value of k3
equal to 0.98.
The lower values of k3 reported in the excluded HSC
column tests were explained by the researchers in different
ways. Martinez claimed that the quality of the concrete in the
protective cover of the spiral columns was inferior to that of
the concrete inside their cores. This difference in quality is
believed to have occurred because of inadequate compaction
of the concrete in the cover caused by the low workability of Fig. 5—Normalized ACI interaction diagram for rectangular
the HSC mixes. The low failure loads of University of plain concrete sections.
Toronto tests were explained by the early spalling of the
concrete cover, possibly caused by the formation of spliting stress block parameters k1, k2, k3 and the relative position of
cracks in the plane of the longitudinal reinforcing bars. the neutral axis in the test, ζ twhere ζ t = c/h, which is equal to
Cusson and Paultre explained the low capacity of their 1.0 for the strain distribution in Figure 1(b). For each test these
columns by the existence of longitudinal planes of weakness coordinates were determined using the following equations:
between the cover and the core due to the high density of
steel reinforcement.
Normal force = C = k 1 k 3 bcf c ′ = ζ t k 1 k 3 bhf c ′ (1)
More tests on concentrically loaded HSC columns are
required to provide better understanding of the behavior of
these columns. The design codes should provide a conserva- Bending moment = M = C ( 0.5h – k 2 c ) = Ch ( 0.5 – ζ t k 2 ) (2)
tive value of k3 compared with the available test data.
C
Parameters defining the rectangular stress block n t = ------------ = ζ t k 1 k 3 (3)
bhf c ′
The stress block parameters k1k3 and k2 obtained from
previous tests and two companion papers12,13 are used in the
following analysis to check the ACI design procedures. The M
- = ζ t k 1 k 3 × ( 0.5 – ζ t k 2 )
m t = ------------- (4)
experimental data are given in Refs. 2, 3, 5, 12, 25, and 29. 2
bh f c ′
These tests had a wide range of variables including rate of
loading, size of specimen, concrete mix design, type of
aggregates, amount of reinforcement, and test procedures. To generate the normalized interaction diagram corre-
The ACI rectangular stress block parameters were used to sponding to the code values of α1 and β1, a series of points
generate interaction diagrams for the plain concrete sections in that diagram are calculated. Each point in the diagram has
as shown in Figure 5. The interaction diagram was normal- the coordinates n and m defined by the following equations:
ized with respect to f'c and the cross-section dimensions. For
reinforced concrete sections, this normalized interaction n = ζα 1 β 1 (3a)
diagram represents the contribution of the concrete section to
the loads and the moments carried by the section after
excluding the part of the loads and the moments that were m = 0.5ζα 1 β 1 × ( 1 – ζβ 1 ) (4a)
carried by the reinforcement. In the normalized interaction
diagram each test can be represented by one point Rt, as illus- These equations are identical to Equations (3) and (4) for
trated in Figure 5. This point represents the maximum the nt and mt of the test point except that the code stress block
capacity of the specimen obtained during the test. The coor- parameters α1 and β1 are used instead of the stress block
dinates of this point, nt and mt, are a function of the reported parameters obtained from the test. By changing the value of

ACI Structural Journal/January-February 1997 43

Fig. 6—Tests of eccentrically loaded columns compared Fig. 8—Equations for β1 /2 compared to tests of eccentri-
with ACI 318-89. cally-loaded columns.
plain concrete specimens except those by Ibrahim12 and
Schade.19 Before making the comparisons for the reinforced
specimens, the net axial load and moment carried by the rein-
forcement was subtracted out, based on the strains in the
various steel layers and the stress strain curves for the steel.
Figure 6 indicates that the ACI code is not conservative for
most concrete strengths, especially for HSC and UHSC
sections. Most of the test points were between ±20 percent
from the code prediction. About 55 percent of the test points
had lower strengths than the code prediction. The mean value
of δ for all the tests was -0.1 percent (test/calculated capacity
of concrete = 0.999) with a standard deviation of 9.91 percent
(coefficient of variation C.O.V of 0.0992). Figure 7 shows the
mean value of δ in each 10 MPa increment of concrete
strength. The figure shows that the mean value of δ decreased
from about +12 percent for NSC test points to about –12
percent for UHSC test points. Figures 6 and 7 suggest that the
ACI rectangular stress block parameters should be modified.
Fig. 7—Mean values of δ versus the concrete strength for
ACI 318-89. Proposed parameters for the rectangular
stress block
ζ, different points are calculated and the normalized interac- The first step in deriving stress block parameters was to
tion diagram shown in Figure 5 can be generated. find a more accurate equation to represent the position of the
Figure 5 also shows that any data point can be defined by resultant force. The position of the resultant force is repre-
the radial distance from the origin (0-Rt) and the slope θ. The sented by the parameter k2 obtained from test results, or k2 =
relative position of the neutral axis depth ζ for the point in β1/2 as suggested by the ACI code. Figure 8 compares the
the interaction diagram that has the same slope θ, Rc, can be experimental data for k2 and the current ACI equation for β1/2
calculated from the equation of tan θ, as follows: plotted versus the compression strength. The ACI value of
β1/2 falls below the data showing that it is too small. If β1/2
1
1 - = ------------------------------ is too small, the internal lever arm is too big and the moment
tan θ = ---------------------- - (5) capacity will be overestimated. The proposed equation
0.5 – ζ t k 2 0.5 – ζ c β 1 ⁄ 2
plotted in Figure 8 has been chosen to pass through the center
of the data and is conservative compared to the current ACI
From the value of ζ the radial distance (0-Rc) can be calcu- equation for any concrete strength. The expression for β1 is
lated. The percentage ratio δ, defined in Figure 5 indicates represented by the equation:
how far the test point is from the ACI interaction diagram. A
positive value of δ for a particular test point means that the
code is conservative while a negative value of δ means that fc ′
β 1 = 0.95 – --------
- ≥ 0.70 f c ′ in MPa (6a)
the code is not conservative in predicting the maximum 400
capacity of the tested specimen. The ratio δ is equivalent to
(test/calculated - 1.0) expressed as a percentage. fc ′
The percentage ratio δ was calculated for a total of 94 tests of β 1 = 0.95 – -----------------
- ≥ 0.70 f c ′ in psi (6b)
eccentrically loaded columns. All of the specimens, except for 58, 000
those by Rüsch,25 were of C-shaped (in elevation) speci-
mens. Figure 6 shows the percentage ratio δ of these tests The second step of the analysis was to choose the parameter
plotted versus the concrete strength. All the tests were on α1. This parameter should provide a conservative lower

44 ACI Structural Journal/January-February 1997

Fig. 9—Equation for α1 compared to tests of concentrically- Fig. 10—Equations for α1β1 compared to tests of eccen-
loaded columns. trically-loaded columns.

bound for the experimental data of k3 obtained from concen-

trically loaded columns. It also should provide conservative
design for eccentric sections when combined with the param-
eter β1. Several trials were made to choose a suitable equation
for α1. For each trial, figures similar to Figures 6 and 7 were
used to check the validity of this equation for design. These
trials showed that the use of a constant value for α1 could
provide a safe design for HSC and UHSC sections but the
resulting value of α1 would be much lower than 0.85 and
would give very conservative design for NSC sections. The
parameter α1 is best represented by an equation that decreases
with an increase in the concrete strength. The following equa-
tion was chosen to represent the parameter α1:

fc ′
α 1 = 0.85 – --------
- ≥ 0.725 f c ′ in MPa (7a)
800 Fig. 11—Tests of eccentrically-loaded columns compared
with proposed equations.
fc ′
α 1 = 0.85 – --------------------
- ≥ 0.725 f c ′ in psi (7b)
116, 000 mean value of δ is greater than zero for all concrete strengths.
The mean value of δ for all of the 94 tests was 10.8 percent
For concrete strength greater than 100 MPa (14,500 psi), (test/calculated capacity of concrete = 1.108) with a standard
α1 and β1 are both taken constant and are equal to 0.725 and deviation of 9.93 percent (C.O.V of 0.0896).
0.70, respectively. The shape of the actual stress-strain curve Figures 13 and 14 show the interaction diagrams generated
for fc′ greater than 100 MPa is very close to triangular. The using the ACI code, the Norwegian code, and the proposed
stress block parameters α1β1 and β1/2 for a triangular stress- parameters for two eccentrically loaded specimens tested by
strain curve with the maximum stress fc = fc′, are 0.5 and Ibrahim and MacGregor. The ACI interaction diagrams over-
0.33, respectively. The same stress block parameters using estimate the capacity of these specimens. The interaction
the proposed equations are 0.5075 and 0.35, respectively. diagrams of the Norwegian code provide a very conservative
Figure 9 compares the experimental data for k3 obtained estimate to the capacity of these specimens. The proposed
from concentrically loaded columns and the k3 or α1 rela- parameters provide a safe design.
tionships from different codes. The figure shows that the
ACI value of 0.85 is not conservative, although the Norwegian Design of triangular compression zones
code is very conservative while the proposed equation A triangular stress block with the extreme compression
provides a lower bound for most of the data points. Figure 10 fiber at the apex is a particularly severe test of the applica-
shows the experimental data, the current and proposed design bility of a rectangular stress block because the tip reaches the
parameters α plotted versus the concrete strength. The figure limiting compressive strain before the bulk of the stress
shows that the proposed parameters give higher resultant block reaches an inelastic state. The ACI design parameters
forces for LSC and HSC sections than the ACI equations and the proposed parameters were used to compute the
while they give lower resultant forces for UHSC sections. maximum strength of the six triangular specimens tested by
Figure 11 shows the percentage ratio δ for the same 94 Ibrahim and MacGregor. The computed strengths were
tests of eccentrically loaded columns calculated with respect compared with the maximum strength measured during each
to the normalized interaction diagram based on the proposed test. This analysis followed the same procedures described
design parameters. Only 16 percent of the experimental data earlier (Eq. (1) to (5)) for rectangular specimens illustrated
points had lower capacity than the predicted capacity, four of in Figure 5(b). Figure 15 shows the strains and stresses in a
which were more than 5 percent low. Figure 12 shows that the triangular cross-section. The test specimens had b = 2h. The

ACI Structural Journal/January-February 1997 45

Fig. 12—Mean values of δ versus the concrete strength-pro-
posed equations, Fig. 14—Test of specimen fc′ = 72.5 MPa.

Fig. 15—Strains and stresses in a triangular section.

2 2
n = ζ β1 α1 (10a)

m = 2--- ζ β 1 α 1 × ( 1 – ζβ 1 )
2 2
(11a)
3

1 1
tan θ = ------------------------- = ----------------------------------- (12)
Fig. 13—Test of specimen fc′ = 129.3 MPa. 2 ⁄ 3 – ζt k2 2 ⁄ 3 ( 1 – ζc β1 )

Figure 16 shows the results from the analysis of the six

coordinates of each data point (nt, mt) were calculated as triangular specimens. All six specimens had lower capacity
follows: than that predicted by the ACI code as shown by the square
points. For the triangular specimens the mean value of δ
2 2 2 using the ACI design procedures was –11.8 percent (test/
Normal force = C = k 1 k 3 bcf c ′ ⁄ 2 = ζ k 1 k 3 bhf c ′ ⁄ 2 (8) calculated capacity of concrete = 0.882) with a standard
deviation of 6.29 percent (C.O.V of 0.071). Only one spec-
imen had lower capacity than that predicted by the proposed
Bending moment = M = C × ⎛⎝ 2--- h – k 2 c⎞⎠ = Ch × ⎛⎝ 2--- – ζk 2⎞⎠ (9) parameters as shown by the crosses. The mean value of δ
3 3
using the proposed equations was 2.0 percent (test/calculated
capacity of concrete =1.02) and a standard deviation of 6.21
2C 2 2 percent (C.O.V of 0.0609).
n t = ------------ = ζ t k 1 k 3 (10)
bhf c ′
Limiting strain
Values of the maximum strain εu, obtained from speci-
2M = 2 k 2 k × ⎛ 2
m t = -------------
- ζt 1 3 --- – ζ t k 2⎞ (11) mens of Ibrahim and MacGregor (0.0033 to 0.0046 for HSC
2 ⎝3 ⎠
bh f c ′ and 0.0039 to 0.0043 for UHSC rectangular columns) were
considerably higher than the limiting strain value of 0.003.
These specimens showed higher strain values at the peak
To generate the ACI interaction diagram, a series of points stress than the maximum strain values from previous test
in that diagram was calculated. Each point had coordinates series of HSC columns. Despite that, based on the reported
similar to those of Eq. (10) and (11) based on the ACI design values of maximum strain in previous tests of C-shaped
parameters β1, 2/3β1 and α1 instead of k1, k2 and k3. The specimens, the value of 0.003 for the ultimate strain speci-
resulting equations and the expression for tanθ of the test fied by the ACI code seems appropriate as a conservative
point were as follows: lower bound of experimental data.

46 ACI Structural Journal/January-February 1997

 REFERENCES
1. Garcia, D. T., and Nilson, A. H., “A Comparative Study of Eccentri-
cally Loaded High Strength Concrete Columns,” Research Report No. 90-2,
Department of Civil Engineering, Cornell University, Ithaca, Jan. 1990,
206 pp.
2. Kaar, P. H., Hanson, N. W., and Capell, H. T., “Stress-Strain Charac-
teristics of High Strength Concrete,” Douglas McHenry International
Symposium on Concrete and Concrete Structures, SP-55, American
Concrete Institute, Farmington Hills, MI, 1978, pp. 161-185.
3. Nedderman, H., “Flexural Stress Distribution in Very High Strength
Concrete,” M.A.Sc. thesis, Department of Civil Engineering, University of
Texas, Arlington, 1973.
4. Leslie, K. E., Rajagopalan, K. S., and Everard, N. J., “Flexural
Behavior of High Strength Concrete Beams,” ACI JOURNAL, Proceedings
V. 73, No. 9, Sept. 1976, pp. 517-521.
5. Swartz, S. E.; Nikaeen, A.; Narayan Babu, H. D.; Periyakaruppan, N.;
and Refai, T. M. E., “Structural Bending Properties of Higher Strength
Concrete,” High-Strength Concrete, SP-87, H. G. Russell, ed., American
Concrete Institute, Farmington Hills, MI, 1985, pp. 147-178.
6. ACI Committee 318, “Building Code Requirements for Reinforced
Fig. 16—Values of δ for the triangular specimens of Ibrahim Concrete (ACI 318-89) and Commentary (318R-99),” American Concrete
and MacGregor.12 Institute, Farmington Hills, MI, 1989, 353 pp.
7. CAN3-A23.3 M84, “Design of Concrete Structures for Buildings,”
Canadian Standard Association, Rexdale, Ontario, Canada, Dec. 1984, 280 pp.
8. NS 3473 1989, “Norwegian Standard for Design of Concrete Struc-
CONCLUSIONS AND DESIGN tures,” The Norwegian Council for Building Standardisation, N.B.R., Oslo,
RECOMMENDATIONS Norway, 1989.
The ACI design procedures for HSC and UHSC sections 9. Thorenfeldt, E.; Tomaszewicz, A.; and Jensen, J. J., “Mechanical
were investigated. The results of tests conducted on 90 tests Properties of High Strength Concrete and Application in Design,” Utiliza-
of concentrically and 94 tests of eccentrically loaded tion of High Strength Concrete, Proceedings, First International Sympo-
sium in Stavanger, Norway, 1987, pp. 149-159.
columns made from NSC, HSC, and UHSC were used in this 10. Rak MK4 1989, National Building Code of Finland, Concrete Struc-
investigation. Based on the analysis presented in the paper tures, Concrete Association of Finland, 1989.
the following conclusions and design recommendations are 11. CEB/FIP Model MC9O, Committee Euro-International du Beton,
drawn: Bulletin d’Information Nos. 195 and 196, Lausanne, Mar. 1990, 348 pp.
1. More tests are required to investigate the reasons for the 12. Ibrahim, H. H. H., and MacGregor, J. G., “Tests of Eccentrically
Loaded High-Strength Concrete Columns,” ACI Structural Journal, V. 93,
early and sudden spalling of the concrete cover reported in No. 5, Sept.-Oct. 1996, pp. 585-594.
some of the test series for HSC and UHSC columns. 13. Ibrahim, H. H. H., and MacGregor, J. G., “Flexural Behavior of
2. The current ACI rectangular stress block parameters were Laterally Reinforced High Strength Concrete Sections,” ACI Structural
found to overestimate the moment capacity of HSC and UHSC Journal, V. 93, No. 6, Nov.-Dec. 1996, pp. 674-684.
columns failing in compression. A rectangular stress block can 14. ACI Committee 363, “State of the Art Report on High-Strength
Concrete,” ACI JOURNAL, Proceedings V. 81, No. 4, July-Aug. 1984,
be used for design, with a limiting strain value of 0.003 and pp. 364-411.
modified stress block parameters α1 and β1 as follows: 15. Martinez, S.; Nilson, A. H.; and Slate, F. O., “Short-Term Mechanical
Properties of High-Strength Light-Weight Concrete,” Department Report
fc ′ No. 82-9, Structural Engineering Department, Cornell University, Ithaca,
α 1 = 0.85 – --------
- ≥ 0.725 f c ′ in MPa (7) Aug. 1982, 98 pp.
800 16. CEB/FIP, “High Strength Concrete State-of-the-Art Report,” CEB
Bulletin d’Information No. 197, 1990, 61 pp.
17. Høiseth; Hoff; and Jensen, J. J., “Høyfast Betong,” Delrapport 2.
fc ′ Søyler Under Sentrisk Last., SINTEF Report STF65 A83049, 1983.
β 1 = 0.95 – --------
- ≥ 0.70 f c ′ in MPa (6)
400 18. Bjerkeli, L.; Tomaszewicz, A.; and Jensen, J. J., “Deformation Prop-
erties and Ductility of High Strength Concrete,” Utilization of High
Strength Concrete, Proceedings, Second International Symposium in
ACKNOWLEDGMENTS Berkeley, CA, 1990, pp. 215-238.
Funding for this research was provided by the Canadian Network of 19. Schade, J. E., “Flexural Concrete Stress in High Strength Concrete
Centres of Excellence on High Performance Concrete, now known as Columns,” M.A.Sc thesis, Department of Civil Engineering, University of
Concrete Canada. Calgary, Calgary, AB, Canada, Sept. 1992, 156 pp.
20. Sheikh, S. A., and Uzumeri, S. M., “Strength and Ductility of Tied
NOTATION Concrete Columns,” Journal of the Structural Division, ASCE, V. 106,
b = width of compression zone STS, May 1980, pp. 1079-1102.
c = neutral axis depth 21. Cusson, D., and Paultre, P., “Behavior of High-Strength Concrete
fc′ = cylinder compressive strength of concrete Columns Confined by Rectangular Ties under Concentric Loading,”
h = overall height of the cross section Department of Civil Engineering, University of Sherbrooke, Report SMS-
k1 = ratio of the average compressive stress to the maximum com- 92/02, Aug. 1992, 39 pp.
pressive stress 22. Sundararaj, P., “High Strength Concrete Columns Under Eccentric
k2 = ratio of the distance between the extreme compressive fiber and Load,” M.A.Sc thesis, Department of Civil Engineering, University of
the resultant of the compressive stress to the distance between Toronto, Toronto, ON, Canada, 1991, 261 pp.
the extreme fiber and the neutral axis 23. Polat, M. B., “Behavior of Normal and High Strength Concrete
k3 = ratio of the maximum compressive stress in a stress block in a under Axial Compression,” M.A.Sc. thesis, Department of Civil Engi-
structure to the cylinder strength neering, University of Toronto, Toronto, ON, Canada, 1992, 175 pp.
α1 = ratio of the stress in the rectangular stress block to the cylinder 24. Martinez, S., “Spirally Reinforced High Strength Concrete
strength Columns,” PhD Thesis, Cornell University, Ithaca, Jan. 1983.
β1 = ratio of the depth of the rectangular stress block to c 25. Rüsch, H., “Tests on the Strength of the Flexural Compression
εu = ultimate compressive strain of concrete Zone,” Bulletin N9. 120, Deutscher Ausschuss Für Stahlbeton, Berlin,
ζ = ratio of neutral axis depth to the overall depth of the section 1955, 94 pp.

ACI Structural Journal/January-February 1997 47

 26. Richart, F. E.; Brandtzaeg, A.; and Brown, R. L., “The Failure of Plain Other Variables in Reinforced Concrete Columns,” University of Illinois
and Spirally Reinforced Concrete in Compression,” University of lllinois Engi- Engineering Experiment Station, Bulletin No. 368, Urbana, Nov. 1947, 128 pp.
neering Experiment Station, Bulletin No. 190, Urbana, Apr. 1929, 73 pp. 29. Hognestad, E.; Hanson, N. W.; and McHenry, D., “Concrete Stress
27. Richart, F. E., and Brown, R. L., “An Investigation of Reinforced Distribution in Ultimate Strength Design,” ACI JOURNAL, Proceedings V. 52,
Concrete Columns,” University of Illinois Engineering Experiment Station, Dec. 1955, pp. 455A79.
Bulletin No. 267, Urbana, June 1934, 91 pp. 30. Yong, Y. K.; Nour, M. G.; and Nawy, E. G., “Behavior of Laterally
28. Richart, F. E.; Draffin, J. O.; Olson, T. A.; and Heitman, R. H., “The Confined High-Strength Concrete under Axial Loads,” Journal of the
Effect of Eccentric Loading, Protective Shells, Slenderness Ratio, and Structural Division, ASCE, V. 114, No. 2, Feb. 1988, pp. 332-351.

48 ACI Structural Journal/January-February 1997