ACI JOURNAL TECHNICAL PAPER

Title no. 82-72

Stress-Strain Relationship for Plain Concrete in Compression

by Domingo J. Carreira and Kuang-Han Chu

A general form of the serpentine curve is proposed to represent the relationship is strongly affected by the strain rate, the
complete stress-strain relationship of plain concrete in compression. quality of the cement matrix, and the aggregate char-
The parameters that define the relationship are physically significant
and can be estimated from empirical relationships or determined ex- acteristics and their content.
perimentally. Proposed equations fit a wide range of testing condi-
tions and concretes for both the ascending and descending branches
of the stress-strain diagram in compression. The conditions that
should meet with an equation representing the stress-strain relation-
CONDITIONS FOR THE STRESS-STRAIN
ship and the effects of the testing conditions on the relationship are RELATIONSHIP EQUATION
also discussed. An equation representing the stress-strain relation-
ship of concrete should meet the following conditions:
I. The equation should compare favorably with ex-
Keywords: compression; compressive strength; initial tangent m?dulus; modu-
lus of elasticity; plain concrete; secant modulus; stress-stram d1agram; stress- perimental data from carefully conducted experiments.
strain relationships. 2. Ascending and descending branches should be
shown.
3. The equation should be based on physically sig-
A general equation to represent the complete stress-
nificant parameters that can be experimentally deter-
strain relationship of plain concrete in compression is
mined. At point of origin, d{f)/ dE = Eu, where f is the
proposed. The shape of the uniaxial stress-strain dia-
concrete stress, t: is the concrete strain, and E;1 is the
gram is strongly affected by the following conditions.
initial tangential modulus.
For testing, the conditions include stiffness of the test-
4. At point of maximum stress, d{f)/ dE = 0.
ing machine, size and shape of the specimen, specimen
The desirable conditions are:
versus machine stiffness, strain rate, type of strain
I. The mathematical form should be simple.
gage, gage length, and type of loading (preloading, cy-
2. There should be only one equation for both the
cling, etc.). For concrete characteristics, the conditions
ascending and descending branches.
include water-cement ratio, cement characteristics and
3. Integrals should be easily evaluated.
content, concrete unit weight, aggregate characteristics
The most common parameters with physical signifi-
and content, and type of curing and age when tested.
cance used to define the stress-strain relationship in-
Although a number of empirical expressions for the
clude the following:
stress-strain diagram have been proposed, most cannot
1. f; is the maximum stress, usually considered as
completely represent concrete behavior for the full
the concrete strength and determined in accordance
spectrum of concretes and testing conditions. The main
with ASTM C 39, "Standard Test Method for Com-
limitation of these expressions is that the effects of the
pressive Strength of Cylindrical Specimens."
forementioned variables cannot easily be accounted for
2. E; is the strain corresponding with the maximum
in the parameters given as constants.
stress J:.
The problem of finding a general stress-strain rela-
3. Ec is the modulus of elasticity.
tionship is further complicated because the parameters
4. Eu is the slope at the origin or initial tangent mod-
defining the relationship are not intrinsic properties of
ulus.
concrete, i.e., they depend on the testing conditions.
The nonlinearity and the existence of the descending
branch of the stress-strain diagram are usually ex-
plained by specimens versus testing machine stiffness Received June 7, 1984, and reviewed under Institute publicationpolicies.
Copyright 1985, American Co~crete Institute. All rights reserved, mcludmll
and by microcracking at the aggregate and hydrated the makin~ of copies unless perm1sS10n 1s obtamed from the copynght propn-
etors. Pertment discussion will be published in the September-October 1986 ACI
cement matrix interfaces. This is why the stress-strain JOURNAL if received by June I, 1986.

ACI JOURNAL I November-December 1985 797

 Table 2 - Fitted parameters in Eq. (1) for
ACI member Domingo J. Carreira is an engineering specialist in the Structural
Project Division of Sargent & Lundy in Chicago, Ill. He is involved in the de-
lightweight aggregate concrete test data
sign and construction of nuclear and fossil power plants. Dr. Carreira is a Specimen Refer-
member and past chairman of ACI Commillee 209, Creep and Shrinkage in Item size and shape, Age, J: ence
Concrete; a member of ACI Commil/ee 301, Specifications for Structural Con- No. in. (mm) days ksi MPa (3 t/ number
crete; and also a member of the Subcommillee of Nuclear Accreditation of the I from 3.4 23.44 6.0 0.0026
ACI 3591ASME Boiler and Pressure Vessel Code. 2 3X 6 28 4.4 30.34 7.5 0.00305
(76x 152) 0.0032 10
3 to 5.65 38.96 9.5
ACI member Kuan-Han Chu is Professor Emeritus of Civil Engineering a/the 4 cylinders 31 7.9 54.47 9.9 0.0037
Illinois Institute of Technology (liT) in Chicago. He was employed with the
firms of Ammann & Whitney and D. B. Steinman in New York, N.Y., before
5 - 4.8 33.10 2.6* 0.00225
6 5 X 6 X 16 - 5.6 38.61 2.6 0.0029
joining liT in 1956. Dr. Chu has published numerous articles in various tech- 7 - 8.1 55.85 4.4* 0.0028
nical journals. (127 X 152 X 406)
8 reinforced - 9.3 64.12 25.0* 0.00285 9
9 flexural - 3.6 24.82 2.7 0.0023
10 prisms - 6.0 41.37 7.0* 0.0028
II - 7.7 53.09 15.0* 0.0027
12 - 11.5 79.29 4.5* 0.00355
Table 1 - Fitted parameters in Eq. (1) for normal *No significant descending branch shown in test data.
weight concrete test data
Specimen Refer- PROPOSED EQUATION FOR THE STRESS
Item size and shape, Age, J: ence
No. in. (mm) days ksi MPa (3 : number STRAIN RELATIONSHIP
I 10 1.30 8.96 1.45 0.0019 A general form of the serpentine curve, as given by
2 40 3.0 20.68 1.8 0.0018
5
Eq. (1) and (2), is proposed to represent the complete
3 40 4.42 30.48 2.4* 0.0018
4 40 7.18 49.51 3.6* 0.00195 strain-stress relationship for unconfined concrete. These
6 X 12
5 (152x305) 28 1.55 10.69 1.7 0.0015 equations satisfy all the conditions previously stated.
6 cylinders 28 2.90 19.99 2.0 0.0019
7 28 5.05 34.82 3.0* 0.0022 7
8 28 6.80 46.89 5.2* 0.0021
9 28 7.6 52.40 6.5* 0.00195 (1)
10 91 7 4.65 32.06 2.9 0.0033 3
II 3 1.11 7.65 1.9 0.0015
12 15 2.20 15.17 2.85 0.0022 and
13 22 2.60 17.93 2.5 0.00195 6
14 42 3.56 24.55 2.2 0.00195
15 3x6 84 3.95 27.23 2.1 0.0020
(76 X 152) (3 (2)
16 3.0 20.68 2.8 0.0029
17 cylinders 2 5.82 40.13 5.3 0.0029
to
18 125 7.30 50.33 5.2 0.00295 10
19 10.75 74.12 6.9 0.00345
20 56 10.2 70.33 8.5 0.0032
21 - 10.6 73.09 7.0* 0.00295 for (3 -~ 1.0 and E ~ Eu
22 cylinders - 6.0 41.37 3.55 0.00392 8
23 Core' 210 10.2 70.33 8.3 0.00360 10 where (3 is a material parameter that depends on the
24 2% X 5Y, 1.17 6.24 43.02 3.6 0.0020 shape of the stress-strain diagram.
(69 X 140) to 12
25 cores 1.37 20.30 139.97 7.0* 0.0040 Eq. (1) and (2) are defined in terms of three param-
26 28 1.90 13.10 1.7 0.0015 eters only: J:, E;, and (3 or Eit. These parameters can
27 28 3.52 24.27 2.3 0.00195 be determined from compression tests in which the
29 28 5.20 35.85 3.8 0.00185 7
29 28 6.25 43.09 5.5 0.00215 strain rate is controlled. For design purposes, an ulti-
30 28 7.50 51.71 7.6 0.0020 mate strain Eu is specified to limit the degree of failure
31 5 X 6 X 16 - 6.20 42.75 2.2* 0.0020 allowed in the concrete.
32 (127 X 152 X 406) - 7.60 52.40 4.0* 0.0019
33 reinforced - 9.70 66.88 5.0* 0.00235 Eq. (1) and (2) have the limits of the perfectly elastic
34 flexural - 12.80 88.26 5.2* 0.00282 material when (3 ->oo (or J: IE; ::::: E,J and of the per-
35 prisms - 15.0 103.42 5.0* 0.0031
36 - 6.5 44.82 2.6 0.0027
fectly plastic material when (3 = 1.0. The equations
37 - 7.7 53.09 3.3* 0.0023 9 thus provide a nonlinear transition for these two ex-
38 - 9.5 65.50 4.0* 0.0028 treme cases.
39 - 11.5 79.29 4.8* 0.00265
40 - 14.3 98.60 6.0* 0.0032 The equation proposed by Desayi and Krishnan 1 and
41 - 10.6 73.09 3.8* 0.0023 TodeschinP is a particular case of Eq. (1) and (2) when
42 - 12.4 85.50 5.1* 0.00265
(3 = 2. Another particular case was proposed by Tulin
43 - 13.4 92.39 6.0* 0.00265
*No significant descending branch shown in test data.
and Gerstle 3 for (3 = 3. Similar equations were used by
'Steam cured. Popovics 4 for the ascending branch.

FITTING THE PROPOSED EQUATION TO

5. Eu is the ultimate strain or strain at which failure is EXPERIMENTAL DATA
defined. Compressive stress-strain data from References 3 and
These parameters are strongly dependent on test 5 through 11 were fitted by Eq. (1) using a minimiza-
conditions, although they are not necessarily indepen- tion of the square of the error technique. Fitted param-
dent of each other, as will be discussed later in this pa- eters are given in Tables 1 and 2; examples of the fitted
per. curves are shown in Fig. 1, 2, and 3.
798 ACI JOURNAL I November-December 1985
 50
10

40
60

30

20
40

10

Strain

(a)

3 x 6 in.
molded cylinders

50
3
(a)
40

30 10

60
20

10
..
0..
::ii
40 ~
13 8.5
Strain
1: =10.2 ksi u;
(b) E~; 0.0032

Fig. 1- Test data in Reference 7 and fitted stress-strain 20

diagrams: (a) flexural tests, reinforced prismatic speci-
mens, 5 x 6 x 16 in. (125 x 150 x 400 mm) and (b) 3 x 6 in.
compression tests, molded cylindrical specimens, 6 x 12 molded cylinders
in. (150 x 300 mm) Test data

The close agreement between the fitted curves and

experimental data indicates that Eq. (I) can be used to
(b)
represent the complete stress-strain relationship in
compression for a wide range of concrete characteris- Fig. 2- Test data in Reference 10 and fitted stress-
tics and testing conditions. strain diagrams (normal weight concrete): (a) concrete
tested from 2 to 125 days and (b) concrete from three
different sources
EFFECTS OF THE TESTING CONDITIONS ON
THE STRESS-STRAIN RELATIONSHIP 60
Lightweight aggregate concrete
8.0
As given by Eq. (1}, the shape of the curve, particu- 3 x 6-in.
molded cylinders
larly its descending branch, is defined by parameter (3.
Parameters f; and (3 from data in References 3, 5
through 7, 9, and 10 are shown in Fig. 4. (3 has a strong 6.0 ~:~:.,; h 40

dependence on f: and is affected by the shape and the fc 0.0037 \

size of the test specimens.

The value of (3 may be given by the equation
20

(3)
..
where f~ has the same units as f:, and ~ and f3c are di-
mensionless constants. Strain

Fig. 4 shows the fitted curves for (3 in terms off: Fig. 3-Test data in Reference 10 and fitted stress-
with the corresponding values off~, ~. and f3c strain diagrams (lightweight aggregate concrete)
ACI JOURNAL I November-December 1985 799
 MPa
Table 3 - The Eit e;lf; ratio from Eq. (2), (4), and
(5)
,. ]' 0 I :/{~ [LJ' 2.19
8 ~- + 1.55
[47
'
6 x 12-in. C)'linders
(150 x 300mm)
" oj I 0 /
/ /'/
,/
'/
/

vv 4.44

3 X 6-in. cylinders
(75 x 150mm)
0 o ___ ___
J:
(psi)
1275
Eq. (2)
3.22
Eq. (4)
3.52
Eq. (5)
3.74
Data from
reference

...,
'0.
0
v o o ,"~ o __ o o--- --- o 3005 2.25 2.45 2.12 5
/ ~ --o 4440 1.71 2.04 1.73
> 4 -/
__.g--~'7 ---0
o -;- \._ fJ
D
[Ll 3.78J
1 95
7180 1.38 1.61 -J.42
~ / /- o 5x6 x 16~in. prisms 1550 2.43 3.26 3.24
9 v.- <;. <~Jf/ (125 x 150 x 400-mm) 2900 2.00 2.49 2.16
0
'I}~ -~~\7 Jf For f~ in: S)'mbols: 5050 1.50 1.92 1.63 7
2.., 0

---;o-0- --~ :~; ~J.: ~ ~ : ~~i~~~;~~~~;;s 6800 1.24 1.65 1.44

4.44 30.6 0 5 x 6 x 16-in. prisms 7600 1.18 1.56 1.39
3.78.26.1
I PSI = 0.006895 MPa.
10 12 14 16
Compressive strength, f~, ksi

The effects of testing conditions and test specimen

Fig. 4-Relationship between f; and /3 for normal characteristics off: have attracted more research atten-
weight concrete tion than any other parameter of the stress-strain rela-
tionship.
Correlation coefficients of 0.93 and 0.94 for the 6 x Hsu et al. 14 showed experimentally that the length of
12 in. and the 3 x 6 in. cylindrical test specimens, re- the strain gages has a definitive effect on the shape of
spectively, imply a significant dependence between fi the descending branch. A similar conclusion is also dis-
and (3. The correlation coefficient for the 5 x 6 x 16 in. cussed in Reference 15. Sangha and Dhir 8 showed ex-
reinforced prisms is 0.64. perimentally that the shape of the stress-strain diagram
Dispersion of the data regarding the regression lines is markedly affected by the following: (1) the confining
indicates that some other important testing variables effect that results from the end restraint of the speci-
and material characteristics were not included in the men by the testing machine plattens; (2) as the material
analysis. approaches f; and beyond, the influence of end re-
Eq. (2) shows that (3, .f.', t;, and E;, are not indepen- straint becomes increasingly important; (3) the maxi-
dent. When (3 is assumed constant, however, the ratio mum size of the aggregate in relation to the specimen
.f.' l(t.f.' E,J also becomes constant. size; (4) the weaker top layers of the specimen; (5) the
The variable nature of theE;, t;lfi ratio was first dis- strain gage length compared with the specimen length
cussed by Saenz, 12 who proposed equations to deter- and slenderness; and (6) the specimen diameter for a
mine both t:
and E;, in terms off:, which yielded the given slenderness ratio.
Ahmad and Shah 16 discussed two experimental meth-
following equation for the Eu t; If; ratio
ods for obtaining a complete stress-strain diagram of
concrete in uniaxial compression, i.e., loading the con-
31.5 - u:Y' (4) crete specimen in parallel with a steel tube and using a
f' u:r r1 + o.oo6 u:w, closed-loop testing system. They concluded that the
steel-tube method gives very reproducible descending
where f: is in psi. For f: in MPa, use 9.08 and 0.87 in portions of the stress-strain curve. Also, it is closer to
lieu of 31.5 and 0.006, respectively. the reinforcing steel-to-concrete strain compatibility
The proposed second-degree parabolic stress-strain and equilibrium conditions than any other testing
relationship 13 results in the Eu t/lfi ratio given by method.
The stress-strain diagrams from each of the testing
methods in Reference 16 show differences in both the
3600
+ 0.92 (5) ascending and descending branches. When using the
f: closed-loop testing system, changes in the size of the
specimen and strain rate may result in significant
where f,~ is in psi. For f: in MPa, use 24.82 MPa in lieu changes in the descending branch and, to a lesser ex-
of 3600 psi. tent, in the ascending branch.
Eq. (2), (4), and (5) are compared in Table 3 with the References 17 and 18 show that the shape of the
data from Smith and Young 5 and from Hognestad et complete strain-stress diagram is affected by the strain
al. 7 for 6 x 12 in. cylindrical specimens. Considering rate during testing. The accepted idea that the initial
that the value of E;, depends on the mathematical portion of the diagram is not affected by the testing
expression used to represent the stress-strain relation- procedure led some authors citing Rasch's work to
ship and that Eq. (4) and (5) are based on different test show the initial portion of his diagrams as unaffected
data and assumptions, the similarity between the values by the strain rate.
in Table 3 is remarkable. The effect of the size and shape of the test specimens
Notice that t; is a variable in Eq. (4), t; = 0.002 in on f: is discussed in References 19 through 22. The size
Eq. (5), and that E,,E :;f:
from Eq. (2) is obtained from correction factors in these references were applied to
the fitted values of (3. the data in Fig. 4 in an attempt to reduce all the test re-
800 ACI JOURNAL I November-December 1985
suits to those of the 6 x 12 in. specimens. Some reduc- Estimating r3 from the initial tangent modulus
tion occurred in the difference between the 6 x 12 in. If some points close to the origin of the stress-strain
( 150 x 300 mm) and the 3 x 6 in. (7 5 x 150 mm) speci- diagram are known, the corresponding value of E;r can
mens, but the improvement was relatively insignificant. be determined graphically. Once E;r is known, E; may be
This confirms that the shape and size of the specimen assumed and (3 can be estimated from Eq. (2). The
are not the only significant factors affecting J: and (3 main disadvantage of this procedure is that the initial
and that they are not independent from the other fac- points of the diagram are very sensitive to the strain
tors discussed. gage seating. Also, no ASTM Standard procedure ex-
ists for this determination.
PARAMETER ESTIMATION FOR DESIGN ASTM C 469 requires the specimens to be loaded at
PURPOSES least twice prior to securing any strain reading. After
The design of reinforced concrete structures is based the seating and behavior of the gages are corrected, at
on compressive strength test results from standard 6 x least two subsequent loadings are recommended to ob-
12 in. cylinders. Four different approaches to find (3 are tain repeatability of test results. The application of
discussed, depending on the information known by the these requirements in ASTM C 469 to strain readings
designer on the concrete characteristics (in terms of/,: from points close to the origin will reduce the strain
only, from fi and En from J: and E;" or in terms of J: gage seating effect on the determination of E;r
and the concrete unit weight.)

Estimating r3 from the specified compressive

Estimating r3 from f: and the concrete unit
weight
strength
The modulus of elasticity for concrete with a unit
During the design stages of most structures, only the
weight between 90 and 155 lb/ft 3 (1440 and 2485 kg/m 3)
specified compressive strength/; is known. Fig. 4 shows
may be estimated with reasonable accuracy by the
a significant interrelationship between J: and (3 for
equation first given in the ACI Building Code (ACI
molded cylindrical test specimens. Therefore, for 6 x 12
318-63). In this equation, E, is calculated in terms of J:
in. cylinder specimens, when only J; is known, (3 may
and the concrete unit weight w. The ACI 318-63 Build-
be estimated using Eq. (6) for J: ~ 5.0 ksi
ing Code adopted the simplest of the equations pro-

f3 =I
L4.7
J:ll + 1.55 (6)
posed by Pauw, 23 who indicated that E, is more sensi-
tive to small changes in the concrete unit weight than to
the compressive strength. If J: and w are known, the
equation for Ec in ACI 318-63 may be used to estimate
the 1:1 (EcE;) ratio
For J: in MPa use 32.4 MPa in lieu of 4. 7 ksi.

Estimating r3 from the modulus of elasticity (8)

The determination of points on the descending
branch requires specialized testing procedures that are
difficult to obtain in most commercial testing labora-
tories. Therefore, it is desirable to have a simple pro- where J: is in psi and w in lb/ft 3 For f' in MPa, use
cedure to determine (3 using standard testing methods 0.043 in lieu of 33.
and equipment. The concrete unit weight, as determined in accord-
The chord modulus of elasticity E, is determined ance with ASTM C 138, "Standard Method for Unit
from the strain corresponding to a stress equal to 0.4 J: Weight, Yield and Air Content (Gravimetric) of Con-
in accordance with ASTM C 469, "Standard Test crete,'' is usually known to the concrete supplier be-
Method for Static Modulus of Elasticity." This method cause of concrete yield determinations. The concrete
requires no specialized equipment and is well-known by unit weight can also be determined from the test speci-
most concrete laboratories. mens prior to their testing in compression. In each case,
Once E, is known, (3 may be obtained from Eq. (7), w shall be corrected for the loss of moisture because of
by assuming E; drying. In most cases this correction is about 5 to I 0
lb/ft 3 (80 to 160 kg/m 3).
Using nonorthogonal regression analysis, Pauw 23 ar-
J: )~
( a E, E; ( J:
-(3 - , - I ) -1=0
E,. E,
(7) rived at another equation that is more accurate than the
equation in the ACI 318-63 Code
where a = 0.40 when Ec is determined in accordance
with ASTM C 469. The value of E; may be measured or (9)
assumed either as 0.002 or as given by Eq. ( 11 ).
For any other point on the ascending branch ( f or
Ec}, the ratio f/ E, may be used in lieu of En and then a where J: is in psi and w is in lb/ft 3 For J: in MPa and
= f!J:. win kg/m 3 use 0.0736 in lieu of 158.1.
ACI JOURNAL I November-December 1985 801
 Table 4 - Estimated r3 from points on the
6 x 12-in. (150 x 300mm)
molded cylinders ascending branch of the stress-strain diagram
Test data from Equation Assumed
1
E; Estimated
Reference 26 number (X 10 ) {3
t: = 4.73 ksi (6) - 2.57
J: = 4.73 ksi (7) 2.0 2.15
E; = 4090 ksi 1.91* 2.28
t: = 4.73 ksi (2) 2.0 2.11
E, = 4490 ksi 1.91* 2.22
J: = 4.73 ksi 1 (7) and (8) 2.0 2.15
w = 148.7 lb/ft (7) and (10) 2.0 2.71
*From Eq. (11).
1 ksi = 6.895 MPa and 1 lb/ft 1 = 16.02 kg/m 1
oTest data
1 ksi 6.895 MPa

10
Strain x 10 4 The CEB-FIP Recommendations 25 adopted E; =
0.002 with their stress-strain relationship, which is a
Fig. 5- Typical stress-strain diagram for concrete in parabola up to f/ followed by a rectangle. In a more
Reference 26 refined diagram, E; = 0.0022 is recommended for 2.9
< f: < 8.7 ksi (20 < f: < 60 MPa). Eq. (11) results in
When this equation is used, Eq. (8) becomes E; < 0.002 when f: < 6.0 ksi (41.4 MPa). The lower
values of E; given by Eq. (11) may be attributed to the
concrete curing conditions prior to testing; that is, Eq.
(10) (11) represents test specimens continuously moist -cured
up to testing, while E; = 0.002 represents concrete al-
The f;!E,E; ratios from Eq. (8) and (10) are substi- lowed to dry prior to testing.
tuted in Eq. (7) to find (3. When test data on E; for a given concrete and from
For a unit weight of 144 lb/ft 3 (2.307 kg/m 3) and f: standard 6 x 12 in. (150 x 300 mm) cylinders are not
up to 3.0 ksi (20.7 MPa), the values of (3 from Eq. (6) available, E; = 0.002 may be used for design purposes
in Eq. (1), (2), (7) through (9), and (10).
are in good agreement with those from Eq. (7), (8), and
(10). For f: higher than 4.0 ksi (27 .6 MPa), Eq. (10)
predicts values significantly higher than those from Eq. COMPARISON WITH TEST DATA FROM
{8). LOADING AND UNLOADING IN THE
A criticism of the ACI 318-63 equation is that it ASCENDING BRANCH
overpredicts E, for high-strength concrete. It has been Fig. 5 shows typical loading and unloading stress-
recommended that an exponential power of Y3 is more strain curves for concrete tested in Reference 26. Each
appropriate than Yz for high-strength concrete.* cylinder was loaded up to approximately 0.45 f:, un-
Gerstle 24 has shown that a power of 0.3 should be used loaded, and then reloaded to a stress level of about 0.6
for high-strength concrete. Therefore, Eq. (10) is rec- f:. Only data from the second loading cycle were con-
ommended instead of Eq. (8) to estimate (3 in terms of sidered in the comparison.
f: and w for high-strength concrete. In Reference 26, f:, Ec, E;, and w are given for Spec-
imens E-28, E-29, and E-30. The strain E; was assumed
Estimation of e; as 0.002 or as given by Eq. (11).
In Eq. (1), compressive strength is the easiest param- The parameters in Eq. (1) that graphically best fit the
eter to determine experimentally; however, the range of data in Fig. 5 for f/ = 4730 psi (32.6 MPa) are E; =
E; is narrower than the range off: and (3. For example, 0.002 for all the specimens, (3 = 2.1 for Specimens E-28
for the data from References 3 and 5 through 11, the and E-30, and (3 = 2.0 for Specimen E-29.
extreme values are 0.0014 and 0.004 for E;, 1.11 and The estimated values of (3 using the four procedures
20.3 ksi (7 .6 and 140 MPa) for f,', and 1.45 and 8.5 for discussed are summarized in Table 4. Fitted parameters
(3. For the 6 x 12 in. specimens, the extremes of E/ are (3 = 2.0 and 2.1 and E: = 0.002 are considered as the
even narrower: 0.0015 and 0.0022. basis for comparison with the values in Table 4 because
Eq. (II) was obtained from a regression analysis on of the larger number of data points used to find them.
the f: versus E; data in References 5 and 7 Results from Eq. (2) and {7) agree closely with the
fitted values of (3 because E;1 and E, were obtained from
E: = (4.88f: + 168) X IQ- 5 (11) points on the same ascending branch of the stress-strain
diagram.
where f: is in ksi. For f: in MPa, use 0. 71 in lieu of The curing conditions in Reference 26 may ex-
4.88. When f: is higher than 3.0 ksi (20.7 MPa), Eq. plain the larger (3 obtained from Eq. (6), (7), and (10)
(11) and the one proposed by Saenz 12 give similar re- when compared with (3 from equations fitted to the
sults. data shown in Fig. 5. These test specimens were kept in
The average value of E/ = 0.002, proposed by
Hognestad 7 13 for f: up to 7.0 ksi (48.3 MPa), has been
*Rusch, K. H.; Jungwirth, D.; and Hilsdorf, H., Private Communication to
widely accepted as a constant for design purposes. ACI Committee 209, Subcommittee II, May 5, 1976.

802 ACI JOURNAL I November-December 1985

their molds, covered with a plastic sheet for 3 days, and (3 may be estimated using Eq. (7). For high-strength
allowed to dry in a 50 percent relative humidity envi- concrete, (3 should be estimated for actual E, test re-
ronment until tested at 121 days. sults to avoid underestimating {3.
As {3 becomes smaller, the nonlinearity of the stress- When the initial modulus of elasticity E;1 is known, (3
strain diagram becomes more pronounced. Long drying may be estimated using Eq. (2). Since E;1 may be ob-
prior to testing causes internal stresses and microcrack- tained at the time E, is determined, the simultaneous
ing from drying shrinkage, which increases the nonlin- estimation of (3 from E, and E; will improve the accu-
1

earity and thus reduces {3 while increasing E ;. Drying racy of the estimated (3.
also reduces the unit weight, which in turn reduces E, The modulus E, depends on the concrete unit weight
and {3. and its compressive strength. When only f; and the
For specimens tested after a long drying period, {3 unit weight ware known, (3 may be estimated from Eq.
may be expected to be lower than for those specimens (7) and (10) for concrete with a unit weight between 90
continuously moist-cured. In accordance with ASTM and 155 lb/ft 3 (1440 and 2485 kg/m 3). The concrete unit
C 192, "Standard Method of Making and Curing Con- weight is usually known by the concrete manufacturer;
crete Test Specimens in the Laboratory," test speci- however, it should be adjusted for moisture loss.
mens are continuously moist-cured until tested. This is
the most frequently used standard when testing
compression specimens. Therefore, it may be assumed REFERENCES
I. Desayi, Prakash, and Krishnan, S., "Equation for the Stress-
that (1) the specimens in References 5, 7, and 23 were Strain Curve of Concrete," ACI JOURNAL, Proceedings V. 61, No.3,
moist-cur~d until tested and that (2) Eq. (6) and (10) Mar. 1964, pp. 345-350.
predict (3 for specimens continuously moist-cured until 2. Todeschini, Claudio E.; Bianchini, Albert C.; and Kesler, Clyde
testing. E., "Behavior of Concrete Columns Reinforced with High Strength
Steels," ACI JouRNAL, Proceedings V. 61, No.6, June 1964, pp. 701-
716.
3. Tulin, Leonard G., and Gerstle, Kurt H., Discussion of "Equa-
CONCLUSIONS tion for the Stress-Strain Curve of Concrete" by Prakash Desayi and
Eq. (1) and (2) are proposed to represent the com- S. Krishnan, ACI JoURNAL, Proceedings V. 61, No. 9, Sept. 1964,
plete stress-strain relationship of concrete in compres- pp. 1236-1238.
4. Popovics, Sandor, "A Numerical Approach to the Complete
sion. These equations fit a wide spectrum of experi- Stress-Strain Curve of Concrete," Cement and Concrete Research, V.
mental data remarkably well and show the ascending 3, No.4, Sept. 1973, pp. 583-599.
and descending branches of the stress-strain curve. 5. Smith, G. M., and Young, L. E., "Ultimate Flexural Analysis
Eq. (I) and (2) are defined in terms of three param- Based on Stress-Strain Curves of Cylinders," ACI JouRNAL, Pro-
eters: E;, J:, and (3 or E;1 All parameters are physically ceedings V. 53, No.6, Dec. 1956, pp. 597-609.
6. Ramaley, D., and McHenry, D., "Stress-Strain Curves for
significant. For design purposes, a fourth parameter,
Concrete Strained Beyond Ultimate Load," Laboratory Report No.
the ultimate strain Eu, may be needed to define the SP-12, U.S. Bureau of Reclamation, Denver, 1947, 23 pp.
strain at which concrete failure is considered. 7. Hognestad, Eivind; Hanson, N. W.; and McHenry, Douglas,
The shape of the stress-strain curve and its parame- "Concrete Stress Distribution in Ultimate Strength Design," ACI
ters are affected by the testing conditions and concrete JouRNAL, Proceedings V. 52, No.4, Dec. 1955, pp. 455-480.
8. Sangha, C. M., and Dhir, R. K., "Strength and Complete
characteristics.
Stress-Strain Relationships for Concrete Tested in Uniaxial
Although Eq. (1) fits a wide range of testing condi- Compression Under Different Test Conditions," Materials and
tions and concretes, no general recommendations are Structures, Research and Testing (RILEM, Paris), V. 5, No. 30,
applicable to every concrete and all testing conditions Nov.-Dec. 1972, pp. 361-370.
because little information regarding testing procedures 9. Kaar, P. H.; Hanson, N. W.; and Capell, H. T., "Stress-Strain
Characteristics of High-Strength Concrete," Douglas McHenry In-
was available.
ternational Symposium on Concrete and Concrete Structures, SP-55,
The following recommendations for estimating E; American Concrete Institute, Detroit, 1978, pp. 161-185.
and (3 are for standard 6 x 12 in. (150 x 300 mm) cylin- 10. Wang, P. T.; Shah, S. P.; and Naaman, A. E., "Stress-Strain
ders that are tested unconfined at controlled strain Curves of Normal and Lightweight Concrete in Compression," ACI
rates. For test specimens continuously moist-cured un- JOURNAL, Proceedings V. 75, No. 11, Nov. 1978, pp. 603-611.
II. Tognon, G.; Ursella, P.; and Coppetti, G., "Design and Prop-
til compression tested, E; may be estimated using Eq.
erties of Concretes with Strength Over 1500 kgflcm'," ACI JouR-
(11). For test specimens allowed to dry prior to testing NAL, Proceedings V. 77, No.3, May-June 1980, pp. 171-178.
and for design purposes, E: = 0.002 is recommended. 12. Saenz, Luis P., Discussion of "Equation for the Stress-Strain
No creep effect is included in these values of E/. Curve of Concrete" by Prakash Desayi and S. Krishnan, ACI JouR-
When only J: is known, (3 may be estimated by us- NAL, Proceedings V. 61, No.9, Sept. 1964, pp. 1229-1235.
13. Hognestad, Eivind, "A Study of Combined Bending and Axial
ing Eq. (6) for concrete specimens continuously moist-
Load in Reinforced Concrete Members," Bulletin No. 399, Engi-
cured until tested. Smaller values of (3 may result if the neering Experiment Station, University of Illinois, Urbana, 1951, !28
specimens are allowed to dry for a long period of time pp.
before testing. 14. Hsu, Thomas T. C.; Slate, Floyd 0.; Sturman, Gerald M.;
The accuracy in the estimation of {3 can be signifi- Winter, George; and Olsefski, Stanley, Authors' reply to the J. Bel-
Her and B. Schneider discussion of "Microcracking in Concrete,"
cantly improved if the elastic modulus E,., as defined by
AC! JOURNAL, ProceedingsV. 60, No. 12, Dec. !963, pp. 1817-1819.
ASTM C 469, is known from the test specimens repre- 15. Carrasquillo, Ramon L.; Nilson, Arthur H.; and Slate, Floyd
senting concrete curing conditions. Once E, is known, 0., "Properties of High Strength Concrete Subject to Short-Term

ACI JOURNAL I November-December 1985 803

Loads," ACI JouRNAL, Proceedings V. 78, No. 3, May-June, pp. 21. Neville, Adam M., "A General Relation for Strengths of Con-
171-178. crete Specimens of Different Shapes and Sizes," ACI JoURNAL, Pro-
16. Ahmad, S. H., and Shah, S. P., "The Complete Stress-Strain ceedings V. 63, No. 10, Oct. 1966, pp. 1095-1110.
Curve of Concrete and Nonlinear Design," Progress Report, Na- 22. Murdock, J. W., and Kesler, C. E., "Effeci of Length to Di-
tional Science Foundation Grant No. PPR 78-22878, Department of ameter Ratio of Specimens on the Apparent Compressive Strength,"
Materials Engineering, University of Jllinois at Chicago Circle, Aug. ASTM Bulletin, Apr. 1957, pp. 68-73.
1979, 29 pp. 23. Pauw, Adrian, "Static Modulus of Elasticity of Concrete as
17. Rasch, Christ fried, "Stress-Strain Relation of Concrete and Affected by Density," ACI JouRNAL, Proceedings V. 57, No.6, Dec.
Stress Distribution in the Flexural Compression Zone at Constant 1960, pp. 679-688.
Strain Rate," Bulletin No. 154, Deutscher Ausschuss fiir Stahlbeton, 24. Gerstle, K. H., "Material Behavior Under Various Types of
Berlin, 1963, 72 pp. (in German) Loadings," Proceedings, Workshop on High Strength Concrete,
18. McHenry, Douglas, and Shideler, Joseph J ., "Review of Data University of Jllinois at Chicago Circle, Dec. 1979, pp. 43-78.
on Effect of Speed in Mechanical Testing of Concrete," Symposium 25. "International Recommendations for the Design and Con-
on Speed in Testing, STP-185, ASTM, Philadelphia, 1956, pp. 72-82. struction of Concrete Structures: Principles and Recommendations,"
19. "Effect of Maximum Size of Aggregate Upon Compressive 2nd Edition, Comite Europeen du Seton/Federation Internationale de
Strength of Concrete," Laboratory Report No. C-1052, U.S. Bureau Ia Pn!contrainte, Paris, 1970, 80 pp.
of Reclamation, Denver, June 1963, II pp. 26. Helgason, T., "Time Dependent Deformation in Plain Con-
20. Blanks, R. F., and McNamara, C. C., "Mass Concrete Tests crete and in Reinforced Concrete Beams," PhD thesis, Department
in Large Cylinders," ACI JouRNAL, Proceedings V. 31, No.3, Jan.- of Civil Engineering, Jllinois Institute of Technology, Chicago, 1973,
Feb. 1935, pp. 280-303. 344 pp.

804 ACI JOURNAL I November-December 1985