A mathematical model of the deformational

behavior of concrete under 9eneralised stress

based on fundamental material properties
M. D. Kotsovos (1)

The paper presents a description of the effect of internal fracture processes on the deformational
behaviour of concrete under increasing load and introduces a mathematical model which
describes the stress-strain relationship of the material under short-term 9eneralised states of
stress.

DEFINITIONS expressed in terms of the principal stresses a l, 02, o 3 ,

whereas the rotational variable a defines the direction of
Throughout this paper it is considered that compressi- the deviatoric component on the octahedral plane. The
ve stresses and strains are positive and that the hydrostatic and deviatoric components can also be
stresses 0"1,0"2, 0"3 represent the maximum, intermedia- expressed in terms of the normal (ao) and shear (Zo)
te and minimum principal stresses, respectively. The 'octahedral stresses which are defined as follows:
uniaxial cylinder compressive strength of concrete is
represented by fc and ~1, ~2, e3 are the strains in the ao = z / . , / 3 = ( a l + a2 + a3)/3; )
direction of the principal stresses 0"1, 0"2, 0"3,
Zo = r / x / / 3 ~ (b)
respectively. |

If the orthogonal coordinate system 0"1, 0"2, 0"3 = ( 1 / 3 ) x / ( a 1 - 0"2)2+(0"2 - 0"3)2+(0"3 - 0"1) 2 )
defining the stress space is transformed into a cylindrical Similarly, the normal (Co) and shear (yo) octahedral
coordinate system z, r, 0 such that z coincides with the
strains are defined by the following expressions:
space diagonal (0"1 = 0"2= 0"3) of the original system
and r and a are the radius and rotational variables, gO=(el +~;2+ g3)/3' }
respectively, on the plane perpendicular to the axis z
(octahedral plane), then the two coordinate systems are yO =(1/3) X/(el __ e2)2 +(g2__ g3)Z_b(g3__ el)2" (C)
related by the following expressions:

z=(0"1+ 0.2+ 0.3)/,fi, "l 1. INTRODUCTION

r =(1/x/r3) X/( 0.1 -- 0.2)2 + ( 0.2 -- 0.3)2 +( 0.3 -- 0.1)2,t (a) Due to the lack of sufficient information regarding the
cos 0 = ( 1 / r x/6) ( 0"1 + 0"2-2 0"3). fundamental properties of concrete under generalised
states of stress, most attempts to date to describe
The variables z and r define the hydrostatic and mathematically the strength and deformationaI proper-
deviatoric components, respectively, Of a state of stress ties of the material have been based on the use of
theoretical models such as non linear elastic [1]-[5],
(~) Concrete Materials Research Group, Civil Engineering Depart- elastic-plastic [6], "equivalent time" characterisations
ment, Imperial College, London, UK. [7]-[8], etc.

0025-5432/1980/289/$ 5.00/9 BORDAS-DUNOD 289

 Vol. 13 - N ~ 76 - Mat~riaux et Constructions

of concrete must essentially describe the effect of the

dreciion ot crack /tensile stress concentration
fracture processes on deformation. Such a description
extension = direction ~ ~\ / n e o . r tip of extended crack has led to the formulation of a mathematical model
of maximum principal I ~\ . /
compressive stress i~~'~' "
describing the deformational behaviour of concrete
under stress increasing up to [17]-[18] and beyond [19] the
ultimate strength level.
void "~'k " " " " ~-
The present paper discusses the fundamental concepts
pre-existing microcrack--J/ "~"~ Z:'~ which have formed the basis for modelling the
/ / /--tensile stress concentration deformational behaviour of concrete under short-term
(::: ...... ~---~ microcro,
c n e a tip
r k of pre-existing
generatised states of stress and introduces a mathematical
model which describes the complete stress-strain
,,~ relationship of the material in a unified form suitable for
computer-based techniques of structural analysis.

',[ 2. E F F E C T OF FRACTURE PROCESSES ON DEFORMATION

F i g . 1. - Schematic representation of stress redistribution and void

formation caused by crack extension. The fracture processes of concrete under increasing
stress take the form of crack extension and propagation
in the direction of the maximum principal compressive
stress (or orthogonal to the direction of the maximum
60 -, . . . . . . 1 9 . - " . , , I . . . . . . i . . . . . .
principal tensile stress) [14]. The extension and propaga-
O tion of cracks reduces the high tensile stress concentra-
50 tions which exist near crack tips and causes void
formation within the body of the material [14] (seefig. 1).
~0
The reduction of the high tensile stress concentrations
E may be considered equivalent to the application of an
E
effective internal compressive stress state which tends to
Z 30
I reduce the volume of concrete. On the other hand, void
oo
t~ {!]
formation tends to increase it. The combined effect on
20 deformation, therefore, may be either a decrease or
increase of overall volume depending on whether the
effect of the internal stress state or that of void formation
predominates.

~0 09 16 21. 32
2.1. Effect of internal stress state
Eo x 10 ]

F i g . 2. - Typical % - eo relationship. Each internal stress state may be resolved into a

hydrostatic and a deviatoric component.
Under a pure hydrostatic stress applied to the
However, a comprehensive programme of investiga- boundaries of an element of concrete, the effect of the
tion into the behaviour of concrete under complex states hydrostatic component of the internal stress state on
of stress, in which the Concrete Materials Research deformation is superimposed on that of the applied
Group (CMRG) at Imperial College has been engaged hydrostatic stress and, for the case of a compressive
since 1963, has not only accumulated a large amount of applied hydrostatic stress, the combined effect is
data at the engineering design level [9]-[i2] but also has expressed by the variation of the volumetric strain (e0)
provided evidence regarding the fundamental nature of with increasing applied hydrostatic stress (no) shown in
the material behaviour which forms the basis of the figure 2. In contrast to the hydrostatic component of the
observed engineering characteristics [13]-[16]. This internal stress state, the deviatoric component is
experimental information contradicts most of the above insignificant since the measured deviatoric strain (Vo) has
theoretical assumpti6ns used to model the behaviour of been found by experiment to be negligible [17]. The
concrete and indicates that, for the region of stress space deformational behaviour of concrete under hydrostatic
investigated, the fracture processes of concrete under stress, therefore, is completely defined by the n o - e o
increasing stress are qualitatively similar under any state relationship.
of stress. Under a pure deviatoric stress applied to the
/
Since, therefore, it is generally accepted that the non boundaries of an element of concrete both volumetric
linear behaviour of concrete is dictated by internal and deviatoric strains vary with increasing stress. The
fracture pi'ocesses occurring under increasing stress, any hydrostatic component of the internal stress state causes
mathematical expression of the deformational properties a volume decrease up to a stress level which has been

290
 D. K o t s o v o s

60
I

50

/.0

20~'~_ t K et/0 o 0 72

l .//f~ o + 129
,0/'#/ o ,~ e§

GO[ II L I
2 3
E0 "I0 ~I

F i g . 3. - Typical %- to relationships for various % .

termed onset of unstable fracture propagation transformed into the ai.t - T o relationships of figure 4. A
(OUFP) [14] (seefig. 3). Beyond this stress level, which schematic representation of the approach used to
marks the transition from the "consolidation" to the evaluate flint for a given applied stress state (ao, Zo) is
"volume dilation" stage, the material behaviour is shown in figure 5.
dictated by the void formation processes and this i s By comparing available experimental data obtained
discussed in section 2.2. from tests using various states of stress and stress
paths [1 5], it has been found that the O'int --applied stress
I 61 I I I I l t relationships are effectively independent of the rotational
fr : 31-7 N / r a m 2
variable (63) and stress path effects [17].
o01fc
o
a 0.~9
Contrasting with the effect of the hydrostatic
i
r~ O 72 component of the internal stress state on deformation,
1.2--
0 097 / p. the effect of the deviatoric component is superimposed on
+ 1.29 /
that of the applied deviatoric stress and the combined
effect is reflected in the nonlinear variation of the
deviatoric strain (Yo) with increasing applied deviatoric
\ stress (%) shown in figure 6. For stress levels up to
~O8 _ oY/o OUFP, the Zo-Yo relationship has been found to be
unique in the sense that it is independent of the
hydrostatic component of the applied state of stress ( cro),
the rotational variable (63)and stress path effects
[171, [201.
0~

%
%_bL__~.,___/a~ ~

O~ 08 12
TO / fc
Fig. 4. - T y p i c a l a i . , - ro relationships for various %.

Oint
m Eo

Nominal values of the hydrostatic component of the

internal stress state ( o'int ) for given levels of applied stress /
may be evaluated by using the relationships shown in .o~/
figures 2 and 3. For a value of the volumetric strain (%)
corresponding to a given level of applied stress (ao, %)
below O U F P (see fig. 3) a value of hydrostatic stress .............
which represents a nominal value for the hydrostatic
component of the internal stress state (a~ot) may be go

obtained from figure 2. In this mode, for strcss levels up Fig. 5. - Schematic representation of approach used to evaluate ~h,, for a
to OUFP, the T o - t 0 relationships of figure 3 may be given stress level.

291
Vol. 13 - N ~ 76 - Mat~riaux et Constructions

I I I I

OI > O~ : O ] CI~ = O 2 > 0 3

t.C

E
\
z
9 3C

___.__ :

20: ~o.~ a- o 072

@" A 097
#
#, +
0
I
I 69
29

t, Predicted relationsh,p
with OUFP indicated

i [ I I
5 i0 15 20 5 10
Y0 x 103

Fig. 6. Typical ro -Yo relationships for various ao.

Independence of the rotational variable(0) is behaviour of the model has been considered to consist of
considered to indicate that the fracture processes the following nonlinear components:
occurring under stress increasing up to O U F P are Component A. - defined by the mechanical properties
insuffmient to cause any "damage-induced" anisotropy. of the model assumed as a solid isotropic continuum;
Component B. - dictated by the internal stress state
2.2. Effect of void f o r m a t i o n caused by the fracture processes, and
Component C. - expressing the effect of void
Although void formation may start at low levels of formation occurring during the fracture processes.
applied stress, its effect on deformation becomes
significant when O U F P is exceeded. The void formation
process dictates both the "volume dilation" portion of 3.1. Mechanical properties o f the m o d e l
the T0 - eo relationships (shown in figure 3) as well as the
faster rate of increase of 7o with To exhibited by the The model material has been devised such that its
T o - ~ o relationship (shown in figure 6) for stress levels mechanical properties, when it is considered as a solid
beyond OUFP. Furthermore, void formation increases isotropic continuum, are completely defined by the
the total cross-sectional area of the material (orthogonal a 0 - e o and T o - 7 o relationships obtained for concrete
to the direction of the maximum principal compressive under stress increasing up to O U F P (see figures 2 and. 6,
stress) to such an extent that the overall stress, defined on respectively). As discussed in section 2.1, these relation-
the basis of the total cross-sectional area, decreases with ships are independent of "damage-induced" anisotropy
increasing applied stress whereas the stress defined on the and therefore can be used tO describe the deformational
basis of the solid cross-section may, in fact, continue to behaviour of a solid isotropic continuum. The n o - e o
increase. Such a process eventually leads to disintegra- and To-~'o relationships have been described mathema-
tion of the element of concrete and complete loss of its tically by a regression analysis of available experimental
load carrying capacity. data which have been obtained in previous investigations
In contrast to the effect of the internal stress state on of the behaviour of concrete under multiaxial states of
deformation, the effect of void formation is dependent on stress [17], [18], [20]. Using these relationships the
the rotational variable (O) and this dependence reflects tangent and secant values of the bulk (Kt, K~) and shear
the "damage-induced" anisotropy which is inherent in (Gt, G~) moduli can be easily expressed in the following
concrete [19]. form

3. MODELLING OF CONCRETE BEHAVIOUR

For the mathematical description of the deformational Ks/Ko=.~

I(I[/+
]for l A(~~ 1
ao/f~<2,
1/[1+2 b-lbA (1)
behaviour of concrete a model material with the strength -2b{h--l).,l {cr.//i.)-l]
properties of concrete [21] has been introduced such that
the fracture processes of the model are qualitatively kjor a o / f~> 2.
similar to those of concrete [16]. The deformational G~/Go = 1/[1 + C (r0/fc) a- 1]. (2)

292
 D Kotsovos

K,/Ko=
f ll[l+bA(fl~
for flo/f~<2,
1/(1+2 b-1 bA)
flo/ f~> 2.
for
Gt/Go= l /[l +C d(zo/ f~)e-1].
}
(3)

(4) N=
{1o }
where M and N are given by
M=k/[l + l ( ao/fc)m],

for f~__<31.7N/mm 2,
0.3124+0.0217fc
(8)

Ko and Go are the initial values of the moduli K for f ~ > 3 1 . 7 N / m m 2.

and G, respectively, evaluated in k N / m m z by the
following expressions k, l and m are parameters dependent on fc which are
given by the following expressions
K o = i1.0+0.0032f~,
Go=9.224+O.136fc+3.296xlO_tSfs.z73. } (5) k = 4 . 0 / [ 1 . 0 + 1.087 ( f c - 15.0)~
l = 0.222 + 0.01086 fc - 0.000122 f~,
The parameters A, b, C and d have been expressed in
terms off~ as follows -2.415
0.516 m= x) f~ f c < 3 1 " 7 N / m m 2 ' (9)

I for f~=<3! .7 N / m m 2, } - 3.5308 +0.0352fc

A= 0.516/[1+0.0027(f _31.7)2.39v L f o r fc>31.7 N / m m 2.
for fc>31.7 N/mm z,
As for the expressions of the moduli, numerical values
b = 2 . 0 + 1 . 8 1 x 10 .8 df4.461
c obtained from the above expressions are valid for values
3.573 offc between approximately 15 and 7 0 N / m m 2. A
typical representation of the fli,t-applied stress
J f o r fr N / m m 2, (6)
relationships predicted by equation (7) is also shown in
C= / 3.573/[1 +0.0134 ( f c - 31.7) 1.14] figure 4.
I
(,.for f~>31.7N/mm z,
( 2.12+0.0183f~ L
fc : 31 7N/ram;'
I I I I
.) for f~<31.7 N / m m 2, 12 ~a
d= -% o OT>02 =O3 1
~, Experimental data
tta,~ A s =02>0 3 J
~;7fc> 31.7 N / I I Z .
~ o ~ Predicted relationships

The above expressions are valid for values of f~ o 08

(o
between approximately 15 and 70 N / m m 2. Graphical
representations of the expressions for the moduli K
and G for a typical concrete are given in figure 7. ~ ~ . , ~-aGs/GO
Having established the bulk and shear moduli the
tangent and secant values of the modulus of elasticity (E)
and Poisson's ratio (o) can be easily obtained from the
well known formulae of linear elasticity 0 I I t 7"-
0 0t. 08 12 16
E=gKG/(3K+G); o=(3K-2G)/(6K+2G). "CO/fc

12j t , ~ i
3 . 2 . Effect of internal stress state o o Experimental data
8a o Predicted relotionshlps
The internal stress state caused by the fracture I *

processes of the model is taken to be equivalent to the e o9+ ~ ~176 8

~o~,..0.~.~._ /-'-" Ks/Ko
hydrostatic component of the internal stress state which
develops within concrete when subjected to deviatoric . o

stress (aint). Nominal values for this hydrostatic

component can be evaluated as described in section 2.1 0t. KT / KO " ~ "

and a graphical representation of the variation of flint

with increasing applied stress for a typical concrete is
shown in figure 4.
I I I
By regression of flint on Zo and flo, the variation of flint 0~ ;~ 08 12 16
O0/fc
with Zo for various flo have been expressed as follows
Fig. 7. - Typical variations of secant and tangent values of shear and
fli.t/fc = M (%/f~)u, (7) bulk moduli with increasing stress.

293
Vol. 13 - N ~ 76 - M a t 6 r i a u x et C o n s t r u c t i o n s

STRESS
The effect of 0,~~ on deformation is apparently str(ltn r duo to:-
equivalent to the deformational response of the model o //t int void |ofmQhOn

subjected to three principal stresses f

0 below
ultimate
0"1 ~ 0"2~ 0"3= O'int.
beyond
ultimote
3.3. Effect of void formation

The void formation processes occurring within the /

model are assumed to be similar to those occurring
within concrete. They are caused by crack propagation
t
processes which occur in the direction of tile maximum STRAIN
principal compressive stress and thus affect predomi- Fig. 8. - Schematic representation of approach used to evaluate
deformation at a given stress level.
nantly the deformation in the orthogonal directions.
Such an effect should be reflected on the ratios of the
incremental strains corresponding to the two principal e = - (15.0/f~)- (0.00722/f~) (f~ - 31.7) 2.so7,)
/
stresses orthogonal to the direction of crack propagation f = 3 . 2 8 6 + 0 . 0 0 2 0 6 ( f ~ - 3 1 . 7 ) 2-s~ ~ (13)
with respect to the incremental strain in the direction of /
9 = 0. 6 8 7 - 0.00406 (f~ - 31.7). .}
crack propagation (r2~, r31).
It is assumed, therefore, that the volume dilation which For f c < 3 1 . 7 N / m m 2 and fc>62.1N/mm 2 the
begins at the O U F P level is caused entirely by void above parameters remain constant and equal to their
formation, then the effect of void formation on values for f ~ = 3 1 . 7 N / m m 2 and f c = 6 2 . 1 N / m m 2,
deformation can be defined by the variations of the respectively.
volumetric strain caused by void formation (6 So) and the For stresses beyond the ultimate strength level the
above incremental strain ratios (r21, r3~) with stress mathematical description of the variables 6 So, r 21, r 31
changes beyond the O U F P level. The strains has been based on theoretical consideration of the
(6E1,6E2,6e3) caused by void formation in the fracture mechanism of concrete which have led tO the
direction of the principal stresses can be easily expressed following formulation [19]:
in terms of the variables 6So, r21, r3x as follows
6So=[Fi (Cro)+ Y1 (a~)] ( 0 / 6 0 ) 0.2 6ZO,"~
6sl=36So/(l+r21+r31); 6 e 2 = r 2 1 t~gl; "[
(lO) r21 = t . 0+ [F2 ( ao)-- 1.0)] (c3/60),
6 S a = r 3 1 681.
r31 = 2 . 0 F 2 (0,0) ( 0 / 6 0 ) 0.2 (14)
For stress levels between 0 UFP and ultimate s tren9th a - [ f 2 (0,0)-1.0] (a/60)-- 1.0,
regression analysis of experimental data has resulted in
the following expressions of the variables where
680, r21, r31 [18]:
6So=eo-~
6So= -[71.4(ao/f~)+45.8](6zo/f~) 2, t
is obtained in mm/m;
r21 =(0/60) X1 6% - 0 . 0 2 5 0+ 1.0, (I1)
r31 = ( 1 0 . 9 - 0 . 165 0) X~ 6 % + 0 . 0 2 5 a - 2 . 0 . 6~0=~0-'~

where is expressed in N / m m 2, 0,~, z~, e~ are the values of Cro,

0 and So, respectively, at the ultimate strength level, and
6So = S o - s~ Y1 (0,~), F1 (0,0), F2 (ao) are given by
is obtained in m m / m , (" [ S - T( a"o/f~)] ( ~/f~)2
I

6~o=To-~ ~ / f o r a~/fr T),

YI (0,~) = ~ 4 S a / (27 T 2)
is expressed in N / m m 2, so urP, zouFP are the values of So /
and Zo, respectively, at the O U F P level, and X ~ is given k for 0,~/f~>2S/(3 T).
by F1 ( a o ) = P ~ / [ 1 . 0 + Q x (0,o/fr
P2/[1.O+Q2 (,rolL) R,] (15)
or a o / f ~ > 0 . 3 , for 0,o/f~<0.8,
X1 = (12)
P2/[I .O+Q2 ( 0 . 8 ) R']
F2 (O'o)=
I for 0,o/f~<0.3. + [0.25 ( cro/f~)

The parameters e, f, g have been expressed in terms off~ - 0 . 2 ] / [ ( ~ro/f~) + 0 . 2 ]

as follows for ao/ fc>O.g.
294
 D. Kotsovos

The parameters P 1, Q 1, R 1, P2, Q 2, R 2, S and Thave 4. C O M P A R I S O N OF PREDICTED AND EXPERIMENTAL

STRESS-STRAIN BEHAVIOUR
been expressed in terms offc as follows
( ' 3 . 6 1 6 7 / [ 1 . 0 + 0 . 0 3 8 2 (f~-31.7) 1"9'.48] It may be noted in sections 3.1 to 3.3 that the
|

J f o r f c > 3 1 . 7 N / m m 2, mathematical model of the deformational behaviour of

concrete includes parameters which are expressed in
PI= )3.6167
terms of ft. Such a formation is based on the assumption
~ f o r f ~ < 3 1 . 7 N / m m 2, that for practical purposes a concrete is adequately
Q1 = 0. 546-0.0174f~ + 0.0003 f~, characterised by its uniaxial compressive strength. This
assumption appears to be realistic for a wide range of
R1 = 6. 3822- 0.1953 f~ + 0.0018 f~, concretes likely to be encountered in practice with a
P 2 = - 19.992 + 0.4804 fc - 0.0032 f~, maximum size of aggregates ranging between approxi-
mately 10 and 30 mm.
Q 2 = 20. 897 + 0. 5806 f c - 0 . 0 0 4 3 f~,
R 2 = 6 . 1928-0. 1363 fc +0.001 f~, (16) Using the above mathematical model the deformation
of an element of concrete under a state of stress a
( 1.07)[1.0+0.0186 ( f c - 3 1 . 7 ) 1'89] expressed in terms of principal stresses below ultimate
)for f c > 3 1 . 7 N / m m 2, strength can be evaluated in terms of strains in the
S= direction of principal stresses by adding the deformation
1.07
due to void formation to the deformation of the model
L f o r f , < 3 1 . 7 N / m m z, subjected to a state of stress a + o-int,in order to allow for
( ' 0 . 2 7 6 / [ 1 . 0 + 0 . 0 2 3 (f~-31.7] 1"496] the effect of the internal stress state O'int . On the other
|
hand, the deformation of an element of concrete under a
J f o r f ~ > 3 1 . 7 N / m m 2, state of stress beyond ultimate strength can be evaluated
T= )0.276 by adding the deformation due to void formation to that
I for f~__<3.17N/mm 2. corresponding at ultimate strength evaluated as
described above. A schematic representation of the
The numerical values obtained from the above approach used to evaluate deformation is given in
expressions are valid for a range off~ between 20 and figure 8.
70 N / m m 2. Figures 9 and 10 show typical representations of the
It should be noted, however, that for stress levels stress-strain relationships predicted for concrete under
beyond ultimate strength, a sudden and complete loss of various states of triaxial axisymmetric stress. The figures
the load carrying capacity of concrete occurs when the also include experimental values obtained from tests on
following conditions are fulfilled [9], [I0]: concrete carried out at Imperial College [20] and show a
(a) ~ < f~ / 3 (where ~ is the hydrostatic component very close correlation between predicted and experimen-
of the stress state at ultimate strength); tal relationships.
(b) 0 = 0 (i. e. for a stress state cr1= ~2> ~a). Figure 11 shows the stress-strain relationships
Equations (14), therefore, are valid for a~ > f~/3 and predicted for concrete under biaxial stress conditions
0r together with experimental values obtained at the

180 I IL~..~ I I I 7.o,9.~ I I

150

12(3

60

~ 30

6o -cO -20 0 20 LO 60 80 100

-60
Tensile Compress,re
S~f~In , rnnl/fn
Fig. 9. - Typical stress-strain relationships for concrete under triaxial axisymmetric stress states 0"1 > 0.2 = ~

295
Vol. 13 - N ~ 76 - Mat6riaux et Constructions

50 I I I i t f I

LO

30
-w
-- 01 =~2
?S o
E Stress path ~/'l'/ o
\
z
fc =317 N/mm2 ~.~. ~ o
20
C Fig. 10. - Typical stress-strain relationships
-i a O-./~ik~x/~/
- P r erelatidonshii pc t e~,~1
d Of ~OOh for concrete under triaxial a x i s y m m e t r i c
stress states a I = cr2 > ~r3 ,

8 0
o
9 u 228 N/ram2
-- E3 E1 :E2
~ 9 0 L3-SN/mm2
I I I I I I I I
-i 0 1 2
Tensd E C0mpcesslve
Stram , mm/m

L8 I i I I 1 I f
:317 N/mmz
fc
-- Predtctedrelationships
LO 9 E] ~ 1
E2 El

EI=E 2

32
\" "\?
E 2L
2:
Fig. 11. - Typical Stress-strain relationships
for concrete under various states of biaxial
compression. 16
Oz/OI ~i
a m Q 052
6
A & ~0

i
8 o
-L -3 ~ 2 3
Tersde Compressive
Strclio . mm/m

70 ~ l I

' , Predicted by derived/ /

expressians f~ I I
60 I------ Predicted by /

! ....d
.. / / ,o.,_
/ / // o
Fig. 12. - Typical stress-strain relationships for various concretes under
uniaxiai compression.

IOOy
-8
I
-6
"g~=lE3 }
-~ -2 0
,y/;y_~
2 ~ 6
Tensile Compressive
Strain mm/m
296
 D. Kotsovos

Technical University of Munich [22]. It is apparent from [3] KOPFER H. B., GERSTLE K. H. - Behaviour of concrete
the figure that the predicted relationships provide a good under biaxial stresses. Journal of the Engineering
fit to the data. Mechanics Division. ASCE, Vol. 99, No. EM4, Proc.
Paper 9917, August 1973, pp. 852-866.
The stress-strain relationships predicted for concrete
[4] PALAN1SWAMYR., SHAH S. P. - Fracture and stress-strain
under uniaxial compression has also been found to relationship of concrete under triaxial compression. Journal
provide a very close fit to experimental values obtained of the Structural Division. ASCE, Vol. 100, No. ST5,
by others investigators [23]. Furthermore, for concretes Proc. Paper 10547, May 1974, pp. 901-916.
with f c > 4 0 N / m m 2, they correlate very closely with [5] CEDOL1NL., CRUTZENY. R. J., POLl S. D. - Triaxial stress-
empirical relationships between the applied stress and the strain relationship for concrete. Journal of the Engineering
corresponding strain in the direction of loading proposed Mechanics Division. ASCE, Vol. 103, No. EM 3, Proc.
elsewhere for such a stress state [24]. The above stress- Paper 12969, June 1977, pp. 423-439.
strain relationships are shown in figure 12. [6] CHEN A. C. Z., CHEN W. F. Constitutive relations for
concrete. Journal of the Engineering Mechanics Division.
ASCE, Vol. 101, No. EM4, Proc. Paper 11529, August
1975, pp. 465-481.
5. CONCLUSIONS [7] BAZANTZ. P., BHAT P. D. - Endochronic theory of
inelasticity and failure of concrete. Journal of the
Engineering Mechanics Division. ASCE, Vol. 102,
1. The nonlinear behaviour of concrete is considered to
No. EM4, Proc. Paper 12360, August 1976, pp. 331-344.
be dictated mainly by internal fracture processes
occurring under increasing stress. These processes create [8] BAZAWrZ. P., SmEH C. L. - Endochronic model for non
linear triaxial behaviour of concrete. Nuclear Engineering
voids within the body of concrete and reduce high tensile and Design, Vol. 47, 1978, pp. 305-315.
stress concentrations.
[9] KoTsovos M.D. - Failure criteria under generalised stress
2. For the mathematical description of the deforma- states. Ph. D. Thesis, University of London, 1974, 284 p.
tional behaviour of concrete a model material with the [10] NEWMAN J. B. - Criteria for concrete strength.
strength properties of concrete has been devised such that Ph.D. Thesis, University of London, 1973, 583 p.
its fracture processes are qualitatively similar to those of [11] GEaSTLEK. H., LaNSED. H., BERTACCHIP., KOTSOVOSM.
concrete. D., Ko H.-Y., NEWMAN J. B., Rossl P., SCHICKERTG.,
TAYLOR M . A . , TRAINA L . A . , ZIMMERMAN R . M . ,
3. "The deformational behaviour of the model consists BELEO'I'FIR. -- Strength of concrete under multiaxial stress
of the following components: states. Proceedings, Douglas McHenry International
Component A. - defined by the mechanical properties Symposium on Concrete and Concrete Structures, ACI
of the model assumed as a solid isotropic continuum; Publication SP-55, 1978, pp. 103-131.
[12] GERSTLE K. H., ACHE H., LINSE D. H., BELLOTTIR.,
Component B. - dictated by the reduction of the high BERTACCHIP., KOTSOVOSM. D., Ko H.-Y., NEWMANJ. B.,
tensile stress concentrations caused by the fracture ROSSI P., SCHICKERT G., TAYLOR M. A., TRAINA L. A.,
processes, and ZZMMERMAN R . M . - Behaviour of concrete under
Component C. - expressing the effect of void multiaxial stress states. Submitted for publication in the
Journal of the Engineering Mechanics Division.
formation occurring during the fracture processes.
[13] NEWMANJ. B., KoTsovos M. D. - Failure criteria for
4. Components A and B have been found by concrete under combinations of stress. Second Internatio-
experiment to be independent of " d a m a g e - i n d u c e d " nal Conference on Mechanical Behaviour of Materials,
anisotropy whereas component C essentially expresses Federation of Materials Societies, Boston, Masschusetts,
Paper A 5-4, August 16-20, 1976.
the effect of " d a m a g e - i n d u c e d " anisotropy on
deformation. [14] KOTSOVOSM. D., NEWMANJ. B. - Behaviour of concrete
under multiaxial stress. ACI Journal, Proceedings,
5. The mathematical description of the components A Vol. 74, No. 9, September 1977, pp. 443-446.
and B has been based on an analysis of experimental data [15] KoTsovos M. D. - Effect of stress path on the behaviour of
obtained for concrete under multiaxial stress states concrete under triaxial stress states. ACI Journal,
whereas that of component C has been based on Proceedings, Vol. 76, No. 2, February 1979, pp. 213-223.
theoretical considerations of the fracture mechanism of [16] KoTsovos M. D. - Fracture processes of concrete under
concrete. generalised stress states. Materials and Structures,
Vol. 12, No. 72, November-December 1979.
[17] Koxsovos M. D., NEWMANJ. B. - Generalised stress-strain
relations for concrete. Journal of the Engineering
REFERENCES Mechanics Division. ASCE, Vol. 104, No. EM 4, Proc.
Paper 13922, August 1978, pp. 845-856.
[1] I.au T. C. Y., NILSONA. H., SLATEF. O. - Biaxial stress- [18] KOTSOVOS M. D., NEWMAN J. B. - A mathematical
strain relations Jbr concrete. Journal of the Structural description of the deformational behaviour of concrete under
Division, ASCE, Vol. 98, No. ST 5, Proc. Paper 8095, complex loading. Magazine of Concrete Research, Vol. 31,
May 1972, pp. 1025-1034. No. 107, June 1979, pp. 77-90.
[2] Iau T. C. Y., NILSON A. H., SLATE F. O. Stress-strain [19] Koxsovos M. D., NEWMAN J. B . - Mathematical
response and fracture of concrete in uniaxial and biaxial description of deformational behaviour of concrete under
compression. ACI Journal, Proceedings Vol. 69, No. 5, generalised stress beyond ultimate strength. Submitted for
May 1972, pp. 291-295. publication in the ACI Journal.

297
Vol. 1 3 - N ~ 7 6 - M a t 6 r i a u x et C o n s t r u c t i o n s

[20] KOTSOVOS M. D. - Mathematical description of [22] KUPFER H., HILSDORF H. K., RUSCH H. - Behaviour of
deformational behaviour of concrete under 9eneralised stress concrete under biaxial stresses. ACI Journal, Proceedings,
states. Interim Report, Civil Engineering Department, Vol. 66, No. 8, August 1969, pp. 656-666,
Imperial College of Science and Technology, London, [23] BARNARDP.R. - Researches into the complete stress-strain
1978, 242 p. curve for concrete. Magazine of Concrete Research,
Vol. 16, No. 49, December 1964, pp. 203-210.
[21] KoTsovos M. D. - Mathematical description of the [24] PoPovlcs S. - A numerical approach to the complete stress-
strength properties of concrete under generalised stress, to strain curve of concrete. Cement and Concrete Research,
be published in the Magazine of Concrete Research. Vol. 3, No. 5, September 1973, pp. 583-599.

RESUME tion du moddle comprend les composantes suivantes :

La composante A, dOfinie par les propriOtOs mOcaniques
U n module math~matique de comportement du b~ton en
du modile assimilles ft un corps solide eontinu isotropique,
d~formation sous contrainte g~n6ralis~e qui repose sur les
propri~t6s fondamentales du mat6riau. - Le comporte- La composante B, dOterminOe par la rdduction des fortes
concentrations de con train tes causdes par les processus de
ment non liniaire du b~ton sous contrainte croissante est
rupture.
probablement d6termind par les processus de rupture
interne qui prennent la forme d'une extension et d'une La composante C, qui traduit l'effet de la formation de
propagation des fissures en direction de la contrainte de rides intervenant au cours des processus de rupture.
compression principale maximalel L'extension et la La description mathOmatique des composantes A et B
propagation des fissures r6duisent lesfortes concentrations s'est appuyie sur une dtude des rOsultats exp&imentaux
de contrainte de traction qui se pr~sentent en bout de obtenus sur le bdton mis en Otats de contrainte pluriaxiale
fissure et d~terminent la formation de rides ~ l'int~rieur du alors que celle de la composante C a rdsultO de
mat&iau. considOrations thdoriques sur les m~canismes de rupture du
En rue d'une description math~matique du comporte- bdton. L'itude des donn~es a montrO que les
ment en dOformation du boron, on a rOalisO un modOle du composantes A e t B sont ind~pendantes de l'anisotropie
mat&iau pourvu des propriOt~s de rdsistance du b~ton, et <<induite par ddtdrioration >> alors que la contrainte C
dont la rupture se produit selon des processus traduit essentiellement l'effet sur la ddJbrmation de
qualitativement similaires. Le comportement en d~forma- l'anisotropie <<induite par dOtOrioration >>.

298