CEMENTand CONCRETERESEARCH. Vol. 3, pp. 583-599, 1973.

PergamonPress, Inc
Printed in the United States.

A NUMERICALAPPROACHTO THE COMPLETE

STRESS-STRAIN CURVEOFCONCRETE

Sandor Popovics
College of Engineering
Northern Arizona University
Flagstaff, Arizona 86001

(Communicated by R. E. Philleo)

ABSTRACT
This paper presents the experimental justification of two
previously published formulas, Eqs. 2) and 6), for the
estimation of the complete stress-strain diagram of con-
crete. Eq. 2) combined with Eq. 3) differs from the other
formulas offered in the literature for similar purpose in
that provides more relative curvature in the diagram for
concretes of lower strengths. Also, with Eq. 6), i t can
take the fact into consideration that the value of ¢0 in-
creases with increasing concrete strength. The result of
these refinements is that the stress-strain diagrams cal-
culated by these formulas f i t better the experimentally
obtained diagrams and within wider limits than the similar
formulas available in the literature. (Figs. 5a through 5d,
8a through 8d, and 9.)
Diese Arbeit veranschaulicht die experimentelle Rechtfertigung
zweier frUher verBffentlichter Formeln, Gleichungen 2) und 6)
zur Bestimmung des vollst~ndigen Spannungs-Dehnungs-Diagramms
des Betons. Die mit Gleichung 2 verbundene Gleichung 3 weicht
insofern yon den anderen, bisher in der Literatur bekannten
Formeln der gelichen Richtung ab, als sie mehr relative
KrUmmungen in dem Diagramm fur Beton niederer Festigkeit auf-
zeit. Auch kann man mit Gleichung 6 annehmen, dab der Wert
¢o mit zunehmender Beton festigkeit zunimmt. Das Resultat
d]eser Verbesserungen zeigt, dab die mit diesen Formeln berech-
neten Spannungs-Dehnungs-Diagramme sich besser und im weiteren
Rahmen den experimentell erhaltenen angleichen, als ~hnliche
Formeln, die man der Literatur entnehmen kann.

583
584 Vol. 3, No. 5
STRESS STRAIN CURVES, CONCRETE, THEORY, DATA

Introduction
This paper is the continuation of the previous work of the writer

on the stress-strain diagram. ( I ) - ( 3 ) The purpose of this study is

(a) to present formulas for the estimation of the complete stress-strain

diagram of normal-weight concrete, made with a given aggregate and tested

with a given procedure, under short-term loading either solely from the

f compressive strength or from the combination of the compressive strength

o
and the measured C° unit strain in concrete at the f ultimate stress (Fig.
o
l ) ; and (b) to show the a p p l i c a b i l i t y of these formulas by demonstrating

their goodness of f i t to pertinent experimental results.

The discussion of the stress-strain curve of concrete is timely from

a theoretical point of view because deformations may provide indirect

information concerning the internal structure as well as the failure mech-

anism of concrete. From a practical standpoint, the ultimate-strength

design of reinforced concrete elements brought the stress-strain relation-

ship into focus. Also, a knowledge of the deformability of concrete is

necessary to compute deflections of structures, to compute stresses from

observed strains, to design sections of highway slabs, to compute loss pre-

stress in prestressed elements, etc. (4) In the study of models of concrete

structures, i t is also necessary to know the stress-strain characteristics

of the model material so that dimensional s i m i l a r i t y may be obtained.

Stress-Strain Diagram for Compression

The diagram is influenced considerably by the testing conditions (type

of the testing machine, rate and duration of loading, size and shape of the

specimen, size and location of the strain gages, number of load repetitions,

etc.) as well as by the age and composition of concrete, especially by

the type and quantity of aggregate and by the porosity, in a similar but

not identical way as the concrete strength is influenced by most of these

factors. (2)
Vol. 3, No. 5 585
STRESS STRAIN CURVES, CONCRETE, THEORY, DATA

//~Constant rate of stress

/ ,,,- : -,,
, ,,,
/ / ~Constant . ",
~- / / rate of strain ~.
// i "--...
t I

E Unit Strain
Fig. 1. Two typical stress-strain curves for concrete under uniaxial load.
The top curve is characteristic of a loading process where the
rate of stress increase is kept constant during the testing. The
bottom curve is obtained by keeping the rate of strain increase
constant.
Another interesting fact is that the stress-strain diagrams for stones

and hardened cement pastes under uni-axial loading are practically straight

lines almost up to the ultimate stress. Yet the same diagram is curved

for mortars consisting essentially of the same two comoonents, and even more

curved for concretes (Figs. 2 and 3), as has been pointed out by Gilkey and

Murphy (5). This paradox can be explained in qualitative terms by the

internal cracking (7) and creep of the hardened paste in concrete under

load that are produced by the stress, especially by the stress concentrations,

resulting from the embedded aggregate particles (8). However, the exact

nature of this problem is so complex that only empirical formulas are avail-

able in the literature for numerical approximation of the stress-strain

diagram. These formulas have been discussed elsewhere. (1)

In most of these formulas the E/Eo ratio is a fixed number regardless

of the concrete strength. This restricts the limits of validity of these

586 Vol. 3, No. 5
STRESS STRAIN CURVES, CONCRETE, THEORY, DATA

I.OOi
Pastes
I
M I
m 0.80. I
,,o,- I
or) n~ 12 I
0.60.
Q I
"/7 fo,psi
ti I
........ 5 8 2 5
Q 0.40.
- - - - -- 5 2 2 0
I
IZ:
--435O I
,,..o 8335 I
O.ZO I
I
I
I
O K m
0 0.20 0.40 0.60 0.80 1.00

1.00'
Mortars

w 0.80.
Ik..
,11-,

0.60' fo, psi n

Q
!. 1745 1.9
2. 5450 2.0
_u 0.40 3. 5800 Z I
G 4. 4560 2.4
n,,
5. 4705 2.5
6. 7560 2,8
~ 0.20, 7. 10970 3.0
8. 15470 3.7

0
0 0.20 0.40 0.60 0.80 1.00
~'/~o Relative Strain
Fig. 2. Relative stress-strain diagrams for hardened pastes and mortars of
various strengths along with the best f i t values of n. The curves
were published by Gilkey and Murphy. (5)
Vol. 3, No. 5 587
STRESS STRAIN CURVES, CONCRETE, THEORY, DATA

1.0
Concrete

ee 0.8"
m
Q
t.,
,0-,

0.6" fo ,psi n
._> i. 1200 1.4
¢} 2. 1600 1.7
m
O 0.4- 3. 2250 2.0
4. 3000 2.5
0
5. 4 5 0 0 3.0
6. 6000 4.5
0.2-

0 0.2 0.4 0.6 06 1.0

• / ~ Relative Strain
Fig. 3. Relative stress-strain diagrams for concretes of various compressive
strengths alon~ with the best f i t values of n. The curves were
published by Rusch. (6)

formulas because experimental data show that the E/Eo ratio varies from
near 4 for normal concretes of l,O00 psi to about 1.3 for concretes of

lO,O00 psi. Consequently, when such a formula f i t s a concrete of medium

strength, i t w i l l over estimate the stress for a given strain in the ascend-

ing branch of the stress-strain diagram for the high-strength concretes and

under estimate i t for low-strength concretes. (9) Formulas with variable

E/Eo ratios are more flexible because they can take the composition of the
concrete into consideration directly or i n d i r e c t l y , such as through the

concrete strength. Such a formula is the following (3):

n-l
f = EE l)
n - l + (E/Eo)n

or, considering that at & = EO~E = (fo/Eo)n/(n - I ) ,

f-,o 0
n
n - l + (~/~_)n
2)
588 Vol. 3, No. 5
STRESS STRAIN CURVES, CONCRETE, THEORY, DATA

where the n power can be expressed as an approximate function of the

compressive strength of normal-weight concrete as follows:

nconcrete = 0"4xi0-3 fo + l.O 3)

Similar formulas for cement mortars and pastes are:

nmortar = O'15xlO-3 fo + 1.5 4)

and

npaste = 12. 5)

The long fractions in Eqs. I) and 2) represent the deviation from

the linear e l a s t i c i t y . This might be utilized in the future for the analysis

of crack propagation in concrete. Note also that these equations, combined

with Eq. 3), d i f f e r from the other formulas offered in the literature for

similar purpose in that they provide more relative curvature in the diagram

for concretes of lower strengths.

Figure 4 illustrates Eq. 2) in relative terms for normal-weight con-

cretes and pastes. I t is impossible to produce a single formula similar

to Eq. 3) for lightweight concretes in general because the various types

of such concretes have greatly differing deformabilities. For instance,

gas concretes show nearly linear elasticity while concretes made with ex-

panded blast-furnace slag as aggregate have highly curved stress-strain

diagrams. (lO) I t should also be pointed out that the formulas above

are valid only for standard concrete specimens with a height-width ratio

not less than two, and when the uniaxial compressive load is a short-term

load which is applied at a rate that produces constant rate of strain in

the specimen. (Fig. l)

Vol. 3, No. 5 589
STRESS STRAIN CURVES, CONCRETE,THEORY, DATA

1.0 -- f,,= 2000 psi

~ 0.8 0

/ / / / / A / f _ _ . ~.L~. n _! \ ~6000
. ",0 00
-~ " Hill / . I \
= / -

,.=0.2
~.~
/ --
~ : .....

0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
6/£0 Relative Strain
Fig. 4. Calculated relative stress-strain diagrams for normal-weight
concretes of various compressive strengths and for cement pastes.

Compa.rison to Experimental Data

I t has been shown in qualitative terms (3) that Eqs. 2) through 5)

are more suitable for the description of the stress-strain diagram of a

concrete than the formulas with fixed E/Eo ratio. However, i t has not

been examined yet that to what extent experimental data support these for-

mulas. This examination is presented below for two different situations:

(a) when both the fo and theG o values are available; and (b) the more

practical case, when only the value of fo is available. Incidentally,

Eq. l) is not recommended for practical purposes, although i t is an inter-

esting formula because, for a given n value, i t expresses the stress values

purely in terms of deformations.

In the f i r s t situation, that is, when both fo and Lo are measured

for a concrete, mortar, or paste, the corresponding value of n can be cal-

culated from one of Eqs. 3) through 5), and used with Eq. 2). Since Eqs.

3) through S) were obtained by f i t t i n g the curves of Figure 4 to those in

Figures 2 and 3, the stress-strain diagrams calculated in this way obviously

f i t the pertinent experimental data by Gilkey (5), and by R~sch (6). Other
590 Vol. 3, No. 5
STRESS STRAIN CURVES, CONCRETE,THEORY, DATA

8- 8-

7" 7-

W/C"3340
6- 6
~,.~ w/¢ = 3 3

." S-
'9 //i if-....oo
~4- /S
,23"
.50 ~
~a

I-
.

0 o
I 2 3 4 5
E. Strain, in/In ( X I O 3) L~. ( Strain, in/in (XIO s)

8-

7
~"-'w/~,~. 33
7 ~ w/c, 33

6
//,
5 ~
%
X

?.
4-

3-
lab aoysl
x

~
4
K/
//'I// ~' i 90 daysi

.67 ~
LO0

I too I

o o
o i i i i g o I 2 3 4 5

E Strain, in/in (XIO s) L~: F. Strain, in/in (X I0 3 )

Figs. 5a through 5d. Comparison of experimental stress-strain diagrams
of various concretes (continuous lines) to those plotted from
Eq. 2) with Eq. 3) (dash lines). The experimental curves were
taken from Reference ( l l ) .
Vol. 3, No. 591
STRESS STRAIN CURVES, CONCRETE, THEORY, DATA

Legend

The complete curve was obtained

experimentally.

f:f. . )"
n-I+(F./'&.

n-o.4 x lOS~ol

f. and ( . were obtained experimentally.

0

f:f.~ n-,.r},~.).
n. o., x ,o" f. • I

( ; 2.7 x I0"4

was obtained experimentall~

curves calculated with Eqs. 2) and 3) are compared in Figures 5a through

5d with experimental curves published by Hognestad et. al. (ll) The com-

parison is done with 20 pairs of diagrams representing five water-cement

ratios and four ages. I t can be seen that despite the wide ranges in water-

cement ratio (from 0.33 to l.O by weight) and age at testing (from 7 to

90 days), the calculated diagrams f i t the experimental curves quite well.

In the second situation, when only fo is given, the value of Eo can

be estimated from one of the available formulas in the literature, and

then this value used with Eq. 2) again, as mentioned above. A convenient

form of such a formula was presented earlier-(3) from certain considerations,

but without experimental j u s t i f i c a t i o n , as follows:

--kx1o 4 4 6)
592 Vol. 3, No. 5
STRESS STRAIN CURVES, CONCRETE, THEORY, DATA

+ 7 days
7 x 14 days
6 o= o 28 days
5 ='
o
A 90 days
4 £ , = 2 . 7 X 10"4X

0
X
3 m

._=
r-
° l

.2
o
A
I X
0

I J w i I I (l,og ~ool~),
1.3 2 3 4 5 678
f~ , psi x I0-~
Fig. 6. Example for the relationship between~o and fo for normal-weight
concretes of various ages. The experi6ental data were taken
from a paper by Hognestad et al. ( l l ) .

where k is a function of the type of mineral aggregate used and the applied

test method. As can be seen from Figure 6, Eq. 6) is supported quite well

by the test results on normal-weight concretes that were published by

Hognestad et al. (ll), (although the f i t could be improved by including

the age as an extra variable). Figure 7 shows that test results published

by Watanabe (12) again support Eq. 6). The difference between the defor-

mabilities of lightweight and normal-weight concretes is also illustrated

in this figure.
Returning to the original question, Figures 8a through 8d demonstrate

the goodness of f i t of the curves calculated from the pertinent combination

of Eqs. 2), 3) and 6) (the latter with k = 2.7 in~/Ib~) to the same exper-

imental curves that were discussed in connection with Figures 5a through

5d. Figure 9 shows another comparison, with k = 2.2 in~/Ib~, to a set of
Vol. 3, No. 5 593
STRESS STPJ~IN CURVES, CONCRETE, THEORY, DATA

7' o lightweight concrete

6' '--~oe + normal- weight concrete

C°=2.85 x 10"4,~'~
u
X
¢:

(log scale]
I I I I I : :
2 3 4 5 678
fo, psi x I0-s
Fig. 7. Examples for the relationship between&sand fo for light-weight
as well as normal-weight concretes. The experimental data were
taken from a paper by Watanabe (12).

experimentally obtained stress-strain diagrams by Smith and Young (9).

I t can be seen again that the presented formulas provide a good estimate

of the complete stress-strain diagram of a concrete. As a matter of fact,

the approximation of the proper combination of these formulas is better

within the given wide ranges of water-cement ratio, compressive strength,

and age than the approximations of other formulas recommended in the l i t -

erature for the same purpose. (9) (13) - (16)

The stress-strain diagram for a concrete under short-term uniaxial

tension is similar to the diagram produced by compression, except that the

same curve is applicable for all the relative tensile stress-strain diagrams

regardless of the strength. (17) I t is probably more than coincidental

that this single tension curve is very close to the compression curve
presented in Figure 4 for pastes.
 594 Vol. 3, No. 5
STRESS STRAIN CURVES, CONCRETE,THEORY, DATA

8- 8

7" 7

__ W1C,.33

f'-'-""zt,.-
I/ ,/j
,o .... ~ ¢I~ .-

17 aay,] i C,4 a.sq

"- / .67

I , / / / / "" I Loo
1.0

0 , , , , ~ 0 ~ ~
0 I 2 3 4 -5 0 I 2 :3 4 5
£ Strain, in/in ( X l O 3) ~ E Strain , in/in ( X I O 3 )
8
8- -..----- /~//~/c.=.]3

/""
'.'°''
F • /,./I-
" " ......-.....
//
/ /' ......50

' /.....--"7/~i~° ~ ~ ~ "

I//I
-~" /,7/ ~ " ////
i l/I/ Ize doy,l i ,/,~'. ~., Lgo ~oysl

2 .//
u) 2 l///// J oo
~_
l ~ I
/~/. .... ~ . , ~ , ~ ' - ' - - - ~ - -
oo
,e-
I

0
o, o i 2 3 :4 ~,
o i ~ ~ ~
~. £ S t r a i n , in/in (X I0I ) ~. £ Strain, i n / i n ( X l O I )
Figs. 8a through 8d. Comparison of experimental stress-strain diagrams
of various concretes (continuous lines) to those plotted from
Eq. 2) with Eqs. 3) and 6) (dot lines). The experimental curves
were taken from Reference (l l).
Vol. 3, No. S 595
STRESS STRAIN CURVES, CONCRETE,THEORY, DATA

8
f,, = 7180 psi

t
0
N
4440
x
w
=4
3005
(D

2
1275

" I I i t l I ~ ,
0 I000 2000 3000
Strain, microin/in
Fig. 9. Comparison of stress-strain diagrams of concrete cylinders (con-
tinuous lines) to those plotted from Eq. 2) with Eqs. 3) and 6)
(dash lines). The experimental curves were taken from Reference
(9).

Supplementary Remarks
(a) The area under any of the stress-strain diagrams calculated from

the presented equations can be determined by numerical integration for

which the computer programming is quite simple.

(b) Differentiation of Eq. 2) provides the Et tangent modulus of

elasticity of the concrete as follows:

n - 1 + (E/~o)n - n(E/Eo )n
Et = df/d& = nfo/6 o =
[n - 1 + (~l~o)n] z

1- (,E/~o)n
n(n - 1) fo/£ o 7)
[n - 1 + (~/8o)n] 2
596 Vol. 3, No, 5
STRESS STRAIN CURVES, CONCRETE,THEORY, DATA

This becomes E at ~= O, that is

n 8)
E/Eo = n - 1

which, with Eq. 3), provides the following relationship for normal-weight
concretes:

2,5oo 9)
EIEo : l +
fo

This means that E/Eo is 3.5 when fo = l,O00 psi, and is 1.25 when fo =

lO,O00 psi. These figures are in accordance with experimental data.

(c) Since Eo = fo/&o , the combination of Eq. 9) with Eqs. 3) and 6)

provides the following relationship between the compressive strength and

i n i t i a l modulus of elasticity of concrete:

4
lO fo + 2,500
E- lO)

This equation f i t s reasonably well, within 2,000 and I0,000 psi compres-

sive strength limits, the traditional empirical formula for normal-weight

concretes:

E = K~ o ll)

(d) I t follows from Eq. 8) that

nf o
12)
t o = ' ( n - I)E

Therefore, with the consideration of Eq. 3):

£o = ~ f o + 2,500) 13)
Vol. 3, No. 5 597
STRESS STRAIN CURVES, CONCRETE,THEORY, DATA

I f E is given, or i t can be estimated (18), Eq. 13) provides ~o in

terms of the concrete strength. For instance, by substituting Eq. l l ) for
E:

+ 2,soo > 14>

The peculiarity of Eq. 14) is that the value of Eo decreases with the

decrease of fo up to a point (which is in our case 2,500 psi), then i t

starts increasing so that as fo approaches zero, £o approaches i n f i n i t y .
Since experimental data, such as shown in Figure 6, do not seem to show

such an increase in ~o at low strengths, one can conclude that Eq. 14) may
not hold for low strength normal-weight concretes because Eq. l l ) may not

hold for such concretes.

Conclusions

The proper combination of Eqs. 2), 3), and 6) seems suitable for the
estimation of the complete stress-strain diagram of normal-weight concretes,

made with a given aggregate and tested with a given procedure, either from

the fo and ~o values, or solely from the values of fo" As Figures 5a through
9 demonstrate, the approximation of these formulas is better within the given
wide ranges of water-cement ratio, strength and age than the approximation of

other formulas recommended in the literature for the same purpose. Math-

ematical analysis of Eqs. l) through 6) provides additional relationships

between the strength and deformation of a concrete.

REFERENCES

(I) Popovics, S., ACl Journal, Proc. 67, March 1970, pp. 243 - 248.
(2) Popovics, S., Symposium on Concrete Deformation, Highway Research
Record Number 324, Highway Research Board, 1970, pp. l - 14.
598 Vol. 3, No. 5
STRESS STRAIN CURVES, CONCRETE, THEORY, DATA

(3) Popovics, S., Mechanical Behavior of Materials, Proceedings of the

International Conference on Mechanical Behavior of Materials, Vol.
IV., The Society of Materials Science, Japan, 1972, pp. 172 - 183.

(4) Philleo, R. E., Significance of Tests and Properties of Concrete

and Concrete Making Materials, ASTM STP No. 169-A, Philadelphia,
1966, pp. 160 - 175.

(5) Gilkey, H. J., and Murphy, G., Proceedings ASTM, 38, Part I, 1938,
pp. 318 - 326.

(6) R~sch, H., Versuche zur Festigkeit der Biegedruckzone (Experiments

Concerning the Strength of the Compression Zone in Bending), Deutscher
Ausschuss f~r Stahlbeton, Heft 120, Wilhelm Ernst & Sohn, Berlin, 1955.

(7) Popovics, S., Journal of the Engineering Mechanics Division, Proc.

ASCE, EM 3, June 1969, pp. 531 - 544.

(8) Brown, C. B., and Mostaghel, N., Journal of Materials, 2, No. l ,

March, 1967, pp. 120 - 130.

(9) Smith, G. M., and Young, L. E., ACI Journal, Proc. 53, December 1956,
pp. 597 - 609.

(lO) Rbsch, H., and Sell, R., Deutscher Ausschuss f~r Stahlbeton, Heft 143,
Wilhelm Ernst & Sohn, Berlin, 1961.

(ll) Hognestad, E., Hanson, N. W., and McHenry, D., ACI Journal, Proc. 52,
December 1955, pp. 455 - 479.

(12) Watanabe, F., Mechanical Behavior of Materials, Proceedings of the

International Conference on Mechanical Behavior of Materials, Vol.
IV., The Society of Materials Science, Japan, 1972, pp. 153 - 161.

(13) Liebenberg, A. C., Magazine of Concrete Research, L4, No. 41, London,
July 1962, pp. 85 - 90.

(14) Alexander, S., Indian Concrete Journal, 39, No. 7, July 1965, pp. 274 -
277.

(15) Desayi, P., and Krishnan, S., ACI Journal Proc. 61, March 1964, pp.
345 - 350.

(16) Desayi, P., Publication No. 30, Annual Report of the Department of
Civil and Hydraulic Engineering, Indian Institute of Science, Bangalor,
pp. 79 - 82.
(17) Johnston, C. D., Symposium on Concrete Deformation, Highway Research
Record Number 324, Highway Research Board, 1970, pp. 66 - 76.

(18) Popovics, S., American Ceramic Society Bulletin, 48, No. I I , November
1969, pp. I060 - I064.
Vol. 3, No. 5 599
STRESS STRAIN CURVES, CONCRETE, THEORY, DATA

NOTATION

The l e t t e r symbols used in this paper and their definitions are

as follows:

E = initial modulus of elasticity of the concrete,

E = secant modulus of elasticity at the f ultimate

0 o
stress, that is, Eo = fo/~o '

Et = tangent modulus of elasticity at the ~ strain,

that is, Et = df/d&,

f = a x i a l s t r e s s in the concrete specimen,

f = u l t i m a t e s t r e s s ; i n compression i t means the

0
cylinder strength,

k,K = experimental parameters,

= u n i t s t r a i n i n concrete caused by the f s t r e s s ,

and

= u n i t s t r a i n in concrete at the fo u l t i m a t e s t r e s s .