Vous êtes sur la page 1sur 10

December, 19b8 ST 12

Journal of the
STRUCTURAL DIVISION
Proceedings of the American Society of Civil Engineers
;;;;;::;:::=
BEHAVIOR OF CONCRETE UNDER COMPRESSIVE LOADINGS
By 1. Demir Karsan,! and James O. Jirsa/ M. ASCE
INTRODUCTION
This paper describes and evaluates an experimental study of the strength
and behavior of plain concrete subjected to repetitions of compressive stress
!Ovarious levels. A total of 46 short rectangular columns were tested under
cyclically varying axial loads to establish stress-strain relations for plain
concrete. The characteristics of the loading and unloading stress-strain re
lationships were studied, and expressions for these relationships were
derived.
BACKGROUND
Early research on plain concrete subjected to variable load histories was
aimed toward obtaining a fatigue limit for the materi?l. Fatigue tests of ce
ment mortar were followed by fatigue tests of plain concrete in which the
reported fatigue limits were generally from 40% to 60% of the staLic cylinder
strength. A decrease in the tangent modulus and the Poisson's ratio with in
creased number of cycles of loading was reported. This early work has been
. reviewed in considerable detail by Nordby (5).3 Murdock and Kesler (4) con-
eluded that there was no Significant fatigue limit for plain concrete under
loads of the order of millions of cycles. However, for a given stress level the
Inumhl>r of ('\7('1,,<: nl'onl1('inO" hillll'P ('olllrl hI' oht"inl'rl.
2545 CONCRETE BEHAVIOR
2544 December, 1969
1. The stress-strain relationships of concrete under
histories possess an envelope curve, which may be considered
identical with the stress-strain curve obtained under constantly
strain.
2. The stress-strain relationships of concrete subjected to cyclic
possess a locus of common pOints which are defined as the point
reloading portion of any cycle crosses the unloading portion. Stresses
the common points produce additional strains, while stresses at Or
these points will result in tIle stress-strain path going into a loop. It Was
observed that the values of the common pOints depended on the
stress in the cycle, i.e., the stress amplitude.
Shah, et al. (7,B,9) reported tests of prismatic specimens subjected to
peated axial compression. Tests showed that the maximum stress of the
of common points appeared to be approximately equal to the critica:lload
which the volume of the concrete uru!l'er compression ceases to decrease
the micracracking in the mortar sharply increases.
Most of the experimental work to date has been aimed toward obtaining
fatigue stress level for concrete. The loadings were generally at high
The effects of acceleration and speed on the behavior were generally
eliminated.
OBJECTIVES
The objectives of this investigation were twofold: (1) To study
tally the behavior of concrete under various compressive loadings in
determine the factors governing the responses of concrete to repeated
ings and examine the mechanism of failure under these loadings; and (2)
develop expressions for the stress-strain relationships of the concrete,
on the experimental results and to use these expressions for predicting
behavior of concrete under other compressive loading histories.
EXPERIMENTAL PROGRAM
Test SPecir;zens.-:-The test specimens were short rectangular COlumllS{
Thedimensions of the column at the critical sectionwere 3-in. X
To confine failure .to the mid?eight of the column, bath ends of the specimellSl
curve. Axial load was applied with a 60-ton hydraulic ram connected
Id_operated high-pressure pump which provided a nearly constant rate
of oil to the ram. The load was transmitted to the specimen through a
yoke resting on a spherical head on the ram. The movable yoke was
to a rigid base plate which distributed the load to the end face
test specimen. On the other end, a similar rigid plate was pin-connected
00 K load cell which was used to monitor the axial load. Both ends of
specimens were grouted with a quick setting high-strength gypsum cement.
lIoriZo
ntal
load was applied through a manually operated screw-type me
'snt. As the mechanism was rotated, a horizontal thrust was developed
the column of the load frame. Horizontal loads were applied only to
:ain
tain
a uniform strain across the specimen.
w
{oJ {bl
FIG. l.-TEST SPECIMEN FIG. 2.-LOAD FRAME
Strain rates were suc? that t?e maximum the
were flared and reInforced WIth No.5 bars. \ jected to steadily increasIng stram was reached In 3 mm to 5 mm. For cyclIc
The concrete mix proportions were constant for all the specimens. The I' loading the peak load on the specimens was applied in 1 min to 2 min and re
concrete was a blend of Type m Portland Cement and 50% Colorado River moved in about 1/2 min to 1 min.
sand and 50% 3/4-in. gravel by weight. The specimens were cast in a vertical instrumentation. -Four O.B-in. wire resistance gages were used to monitor
position. 6-in. x 12-in.. control were cast with each specimen. the strains on the two opposite 3.-in. faces at .the of column.
The speclmens were cured m a mOIst room for 2, 5 or 7 days and testedati Because the specimens were subJected to strams conSIderably In excess of
7, 14 or 21 days. Cylinder strengths varied from 3,500 psi to 5,000 psi. Values
1
( the normal operating range of the strain gages, strains were also measured
of for each specimen and complete details of the experimental program' over a 6-1/2-in. gage length, using two differential transformer displacement
are given in Ref. 3.. , 'I transducers placed at the midsections of the specimens. These transducers
Loading An'angement.-A rigid loading frame (Fig. 2) was constructed in . were placed on the opposite 5-in. faces of the specimen and were 'supported
which .axial Or flexural loads could be applied simultaneously or separately., between light steel frames fastened to the specimen with pOinted set screws.
The silffness of the loading frame was sufficient to avoid problems associated Test Procedure. - Load histories were controlled by monitoring one of two
with the release of energy in the unstable portion of the concrete stress- variables: (0 Incremental strain during a given cycle, or (2) stress level
2547
OJ''
2546 December, 1969
CONCRETE BEHAVIOR
during a given cycle. Loads were applied manually and the magnitude
to produce the specified seress level or strain increment as
by an X-Y recorder plot. A total of 46 specimens were tested in
different series which are identjfied below by the distinguishing feature
load history. The number of tests in each series is given in parentheses.
Series AM1: Steadily Increasing Strain to Failure (13 Specimens).-Speci
.:::
e usually cast in groups of two or four. The strain was steadily
to failure on one specimen in each group. This test was used to
the effects of other loading histories. Typical stress-strain curves
' ...... the test specimens in series AM! are shown.in Figs. 3, 4 and 5. The non-
A; ensiona! coordinates F and S will be presented later.
IllSedes AC2: Cycles to Envelope Curve (9 Specimens).-The concept of an
II Ii'
velope curve for the response of concrete has been proposed by other in
co r Wk1 K< fC""""": '.".............
..
:stlgators (10). The envelope curve can be defined as the limiting curve
S. t/tg
"
"
"
FIG. 3.-CYCLIC LOADING TO ENVELOPE CUR,VE
1.0
--,
I: I I
I ' /;AM,j,,) !
,.
'-06

' ".""! '


'
",.",

, y<
SMITJ-t- YOUNG ............
-.....
' ..... ,.n
'kll:
"
So oc/ to
FIG. 4,-COMPARISOK OF ENVELOPE CURVES \VITH TEST AC4-10
- I
I

, within which all stress-strain curves lie regardless of the load pattern. To
Investigate the validity of the envelope curve, the strains in a given cycle
were increased until the stress-strain path reached the envelope curve. A
-"01
' , stress-strain curve for test AC2-09 is shown in Fig. 3.
Series AC3:Varying Strain Increments (10 Specimens).-The specimens in
series AC2 were subjected to strain cycles in which a specified strain incre
'"",,00,.
) ment was added during each cycle. Stress-strain curves are shown in Fig. 6
"1Ii\l"Ut
r' [or AC3-10 in which strain increments of 0.5 x 10-
3
were added during each
>. cycle. Some of t.h.e specim.ens w.e .. re loaded to an initi.al specified ...s.t. r.a.... in such
as 1 x 10-
3
and then cycled to produce a given incremental st:rElJrr such as
. 0.1 x 10-
3
in each .cycle. In many of these tests the results to
" 5 .. ,/(.
those of series AC2 because the incremental strain was large enotigh.,-i9 pro
duce stress-strain curves which reached the envelope curve.
I
FIG. 5.-COl'l'IPARISO.K OF ENVELOPE CUHVES WITH TEST AC4-13
,: AC4: Cyd" betwn M",lmom and Minlmom 81"" Lov.l, (14
J
0.1-1-- . AI' 'I "'
,.8 II ,I
..I 'ltV r n
I' r 'I
1).6 :.1 I
.4
'1............
02
o. ,V 1/ vq.-r ........--{<
co
1.0 l.a 3.0
5 c/c,
FIG. 6.-CYCLIC LOADING PRODUCING GIVEN STRAIN INCREMENT
2548
2549
December, 1969
Specimens). - In this series load cycles were applied between
e
x 12-in.
and the
ched.
mens
giv-en
levels until the strains stabilized or until the maximum stress lev-el
be sustained, Maxi:num stresses varied between 0.85 fJ and 0,59 JI.
, mum stresses vaned between 0 and 0.70 Stress-strain Curv-es lor
AC4-10 and AC4-13 are shown in Figs. 4 and 5. Additional details Of til
perimental program are given in flef. 3,
BEHAVIOR OF TEST SPECIMENS
Monotonic Loading to Failure.-The specimens in which the strain
steadily increased to failure were used to evaluate the behavior of
subjected to other load histories. In order to facilitate comparisons
test results, stress-strain curves are plotted in normalized coordinates.
stress coordinate F is normali7,ed with respect to I;, the 6-in.
1.0
0.8
'_<.l 0.6
"
0,'
0,'
0,0
0,0 1.0
3.0
S=: /1)
FIG. 7,-l\]OP;OTONIC LOAD/KG TO FAILURE
inder strength. For the specimens under monotonic loading to failure,
median value of the strength was 0.85f; and the mean wasO.S6/; with a
dard deviation of 0.04 r;. The strain coordinate, S, is normalizedwith
to Eo' the strain corresponding to the peak stress. The median value of
the specimens subjected to monotonic loading was 1.68 10-3 ,
was 1.71 10-
3
with a standard deviation of 0.14 )< 10-
3
Strains
specimens subjected to cyclic loads were normalized with respect to the
of Eo for the specimen which was cast from the sa111e batch of concrete
subjected to monotonic loading (0 failure.
The results of several of the tests in series AMI are shown in Figs.
and 5. Points from the stress-strain curves of the 13 tests in. this series
plotted in Fig. 7. Also plotted in Fig. 7 are stress-strain relationshi
gesled by Smith and Young (11), and Hognestad (1,2), The equations for
curves are expressed in terms of the coordinates F and S. The smith-
HOGHESTAO; F= o.ass {Z-sl. fot 5111,0
i
F=O.!lS1-a,I01S,forS!!:to
.L.
t .
---: --; ""-.' I
"', \ -..,.
i -/

F= O.85Se-('i Sl
20
CONCRETE BEHAVIOR
appears to best fit the observed stress-strain relationship and
to approximate the behavior of concrete under monotonic load-
in the age of the concrete at testing and in the strength did not
have an influence on the shape of the stress- strain curves obtained
the specimens under monotonic loading to failure or those under
and were not considered to be significant variables in the range
this investigation.
JlilJlletope Curve. -The test results as represented by the curves in Fig. 3,
aindicate that the stress-strain paths under cyclic loading generally
eKceed the envelope curve. In those cases where the envelope was the
stress-strain curve for the companion specimen under monotonic
to failure, the comparison is excellent.
the specimens tested, the stress-strain relationship became approx
tangent to the envelope curveas.shown by the stress-strain curves
AC2-09 and AC3-10 (.E'xgs. 6). The same behavior was
in specimens AC4-l(i and AC41S (Figs. 4 and 5). It is significant
ulation of strain under constant maximum stress levels produced
when the envelope was reached.
1II Fig. 8 another type of load history is shown. Even though the cycles
quite different than those presented previously, the stress-strain curves
specimen AC2-07 remained within the envelope until very high strains
oints plotted in Fig. 9 are the values of peak stress and strain from
for which the stress-strain path in a given cycle became ap
coincidental with the envelope curve. Although there is some
the Smith-Young expression is a good approximation of the test
the specimens tested, the envelope curve may be defined as the stress-
curve obtained under monotonic loading to failure and approximated by
Smith-Young equation. Failure was observed when a given stress-strain
exceeded the envelope, however, the specimen could be loaded to the
regardless of the strain accumulated prior to a given cycle. Strain
did not appear to reduce the strength to a level below the en
Itshould be remembered that the envelope and the stress strain curves
be altered if the strain rate or the' properties of the concrete were
man Points. -Sinha, GerstIc, and Tulin (W! ind1Catedthat the locus of
oints where the reloading portion 'of any cycle crosses the unloading
maybe defined as a stability limit at which the strains stabilize and a
hysteresis loop is formed in subsequent cycles. Stresses above this
produce additional strains whUe maximum stresses at or below this
cause the stress-strain history to go into a loop, repeating the pre
cycle without further permanent strain. Using this definition, if the
level corresponding to a common point, as indicated in the stress
strain history of specimen AC2-09 (Fig. 3), is not exceeded in subsequent
e)'eles, strains should not exc.eed the value at the common point. .....
'l'hebehavior of the specimens in this investigation suggesfifa"tN;j't'e" rigor
definition of the stability limit. The common cyclic
load tests with various maximum stress levels are plotted in Fig. Hr. In view
Ilfthewide scatter, the common pOints may have a range of values.
2551
.l'fi!1"
2550 December, 1969
The scatter may be explained by examining the stress-strain
specimen AC2-07 [Fig. 8(a)). Four specimens were cycled in this m
all exhibited similar behavior. The location of the common points
obtained.
is
in a detail of the stress-strain history [Fig. 8(b)). For example, after
'0
TIlT AC -07
# oIU,Op6i
0 .
0.6
0.4
0.2
().o
.8 1.0
( ;A Complete Load History (b) POirtt5
FIG. 8.- VARIA TION OF COMMON POINTS
10
'.8
'U
" 0.'
.y. a<
$$ te/Co
02
00 20
'U
.,
s (:/ (0 "
FIG. 9.-POINTS ON ImVELOPE CURVE (J.1EASURED)
20 had been carried out, the specimen was reloaded until the unloading
tion of cycle 20 was reached (point n) and then the specimen was unloaded.
This routine was continued until the common pOint stabilized at points D and
CONCRETE BEHAVIOR
. In general, the magnitude of the reduction of the point of intersectionde
creased with the number of cycles. If the locus of points for the first, second,
third, .. ,common points are drawn, a family of common point curves can be
10
O.
"
o ,
0.1
0.0
o0 10 '0 1.0
s C/ CQ
FIG. lO.-COMMON POINTS (MEASURED)
. ,
'-
Ie
00
I.' Z..Il C /(0
FIG. n.-EFFECT OF 1I1INIMUM STRESS LEVEL ON COMMON POINTS

test results plotted.in Fig. 10 .and the behavior exhibged tn tests such
as AC2-07 (Fig. 8) show that intersecting points of load cycles'"ttf,tije envelope
Curve constituded an upper limit on the common points (hereafter called com
mon point limit). As cycles with lower stress levels were introduced, the
1.0
'"
2553
stress level'

2552 December, 1969
paint of intersection was reduced but stabiltzed at a lower bound
limit),
The effect of the minumum stress level on the common points is iIIull:"
trated in Fig. 11. Specimens AC4-12 and AC4-13 were cast from the salll;
batch of concrete. Both were subjected to the same maximum
but the minimum stress levels were different. The common points for both
specimens were identical. The same behavior was noted in other specimens.
On this basis, it can be assumed that the common points were independent Of
the minimum stress levels in a particular load history.
The dependency of the common pOints on the maximum stress level is
shown in Fig. 12 for 5 tests with cyclic loadings between a zero minimUJll
stress level and different maximum stress levels. In test AC4-12 the maxi.
mum stress level, 0.79 f;, was higher than the peak value of the common
nST F'mQIIi(

AC4- n 0.71

.At'.' 10 U ..
,...
.n ",0
... AC' 03 .... 4040
Ae4 -01
I us!)
0"1
ui /lo::!"'- =t,,_
"
., +-f-----J----
When
'.0 ... 1-----,
liMn

:- G.'
. I I
J.O 1.0 0.0 ...
s C Ie.
FIG. 12.-COMMON POINTS FOR TESTS WITH CONSTANT MAXIMUM STRESS
LEVEL
point limit, and as a result, the points of formed a smooth curve
located approximately on the common paint limit. The maximum stress'level
for test AC4-1O was 0.76 f which was about equal to the peak value of the
common point limit. The points of intersection for this speCimen followed the
common point limit initially, then formed an approximately horizontal
until the strain accumulation reached the common point limit. This trend
also apparent in test AC4-11. Although the strain accumulation was
than expected, this can be explained by the higher envelope curve m
for the companion specimen under monotonic loading to failure.
maximum stress was reduced to 0.63 nand 0.55 f; in tests AC4-03 and
01, the cOmmon points gradually increased but approached the stability
. and strain.accuinulation ceased under continued
The observed behavior may be summarized as
1. The stress and strain at the peak of the load cycle were the prime
CONCRETE BEHAVIOR
abIes in determining the location of the common point. Minimum stress levels
did not appear to have a significant effect on the common points.
2. Peak stress-strain values above the common point limit produced points
of intersection very near this limit. With lower peak values; the points of
intersection fell between the common pOint limit and the stability limit.
0.' .11--__
I I 1111111 H #' H /' :..i"
1.0 +--------- ---,
".<'"
o
"'U
",;
0.4
02
00 J' I'." 'q r'l
0.0 . 2.0
"0 '.0
s .t:/t.
FIG. 13.-LOADINGCURVES
1.0
O.
'-v 0.6
'
..

04
01
0.0
0.0 1.0 20 '.0
S'/'O
FIG. 14.-UNLOADINGCURVES
3, If the stress and strain at the peak of the load cycle wa:t'aoove the sta
bility limit, strains accumulated until failure occurred oruntif' strain accu
mulations reached the stability limit. At this point, strains stabilized and
formed a closed hysteresis loop for subsequent cycles.
2555

2554 December, 1969
Note that if the effects of time were considered, creep strains would
there was
the observed behavior. With reduced strain rates, the stress-strain
would shift toward the strain axis, and it would be difficult to define a
limit (6).
Nonrecoverable Strains. -Nonrecoverable or plastic strains are the strains
corresponding to a zero stress level on loading or unloading stress-strain
curves. The changes observed in the slopes of the stress-strain curves sug.
gest a relationship between the plastic strain ratio Sp and the nature of the
loading curves.
Loading curves from a number of spec1mens subjected to different lOad
histories are plotted in Fig. 13. Each group of curves originated from a
similar plastic strain ratio. It is apparent that the slope of the curves gradu.
ally decreased with increasing values of Sp. The common point limit (the
locus of common points of load cycles to the envelope curve) is also shown in
Fig. 13. It can be seen that the common point limit corresponds approxi
mately to the point at which the slope of loading curves changes significantly.
Previous investigations (7,9) have shown that the change in slope can be at
tributed to a significant increase in microcracking.
Unloading curves from a number of specimens in which the unloading
portion of the cycle started at or near the envelope are plotted in Fig. 14. In
each case the minimum stress level was zero, These plots show that the plas
tic strain ratio was a major variable in determining the shape of the lOading
and unloading curves. The load history preceding a given value of Sp did not
Significantly alter the curves originating at that value of S p.
PREDICTION OF FAILURE
Derivation oj Expressions jor Stress-Strain Curves.- Using the observed
response of the specimens, expressions were developed for loading and un
loading stress-strain curves in order to duplicate the observed response
analytically and predict failure under load histories other than those actually
imposed on the specimens. As shown in Figs. 13 and 14, the shapes of the
loading and unloading curves appear to be functions of the nonrecoverable or
plastic strain ratio. In order to develop expressions for these curves, various
polynomials were compared with the experimental curves, and a second de
gree parabola was selected to represent the shape of the curves. Better ap
proximations with higher order or transcendental expressions might have
been obtained. However, considering the accuracy of the test results, the
advantages of a simple stress-strain relation outweigh the small gain in ac
curacy derived using higher order approx1mations.
To account for the changing shape of the loading and unloading curves with
increasing plastic strains, the stress-strain curves were developed as func
tions of the plastic strain ratio. For a given plastic strain ratio, relationships
between the strain at which a loading curve will intersect the previous un
loading curve (common point) and the envelope curve were obtained.
Envelope Curve.-The equation for the envelope curve, the expression de
veloped by Smith and Young (12), has been presented previously (Fig. 7 and 9)
and is repeated below in terms of the nOrmalized parameter FE and SE'
points on the envelope:
(I-SE)
FE = 0.S5 SE e ........................... (1)
CONCRETE BEHAVIOR
common Points.- The experimental results (Fig. 10)tndicated that although
a variation in the location of the common points, a common point
lilllitand a stability limit could be established. Analysis of the common points
produced exponential expressions of a form similar to the envelope curve:
_ Se [t-Se/(O.315+0.'n(3)]
Fe - {:3 0.315 + 0.77{:3 e ............. (2)
The common point limit, {:3 0.76, and the stability limit, {:3 = 0.63, are plotted
in Fig. 10. The variation in {:3 accounts for the change in the maxima of the
lilllits. Since the common point for a given cycle of load was dependent on the
Illagnitude of stress in the previous cycle, the following distinction must be
Illade as to the value of {:3 for the applicable common point.
1. If the peak lies above the common point limit, {:30.76, the common
point is on the common point limit.
2. If the peak lies in the regiOn between the. common point and stability
limits, {:3 varies betweenO. 76 and 0.63. .
3. If the peak lies below the stability limit, the common point
to the peak, and the stress-strain curves form a closed hystereSiS loop. Note
that this criterion implies that if stresses do not exceed 0.63 n cyclic load
ings will not produce failure.
The values of {:3 for the common point and stability liroits are in the range
of critical stresses reported by other investigators. Shah, et al. (7,9) have
reported the onset of major microcracking at 70% to 90% of the ultimate load.
The value of {:3 at the stability limit is 0.63 n (74% of the specimen strength)
andat the common point limit, is D.76 n (90% of the specimen strength) which
indicates that the behavior of the concrete is controlled primarily by micro
cracking. It is also interesting to note that RUsch (6) has reported that the
sustained load strength of concentrically loaded specimens is 75% to SO% of
the static strength which corresponds with the value 6f {:3 for the stability
limit.
Plastic Strain Ratio (Sp).-Fig. 15 shows the relationship between the plas
tic strain ratioSp and the strain ratio at the common point Se. The expression
for the curves passing through these pOints is the following:
Sp (1.76 - fl)(0.160 S(; + 0.133 Se) ................... (3)
in which 0.63 :S (:3 :S 0.76. Fig. 16(a) shows the relation between the plastic
strain ratio Sp and the strain ratio at the point where a given loading curve
starting at Sp intersects the envelope curve SE' Fig. 16(b) shows the relation
between the plastiC strain ratiO S p and the strain ratio at the point where a
given unloading curve starts on the envelope curve The equations for these
relationships are the following:
Loading: Sp 0.093 Sj: + 0.91 SE ..................... (4)
,', -
Unloading: Sp 0.145 Sf + 0.13 SF
Loading Curves. -The expressions for loading curves are secol'ld degree
parabolas which pass through the following three points: (1) The point at which
the reloading curve or its extension starts (Sp, F =0); (2}the ommonpoint
2557
@lW}Mi'l!fl'if'V')p;
2556 December, 1969


*e
"

,., \=1:,/.

..
.,
ID (b)
"
\",,It.;;
FIG. 15.-RELATIONSHIP BETWEEN PLASTIC STRAIN AND STRAIN AT COMMON
POINTS

.

U
"
1'\
7r.
" '.
$p. o_onsE .. O-lUUf:
UtoII.,J:lAO/fiG ENVELOPE
TO ,Sp_
,.
"
101 \:.. e
E
1<:,
FIG. IS.-RELATIONSHIP BETWEEt\ ENVELOPE STRAINS AND PLASTIC STRAINS
CONCRETE BEHAVIOR
and (3) the point at which the reloading curve or its extension (Sc. Fc);
aches the envelope curve (SE. FE)'
re Unloading Curves. - The three points through which the second degree pa
rabola unloading curves pass are as follows: (1) The pOint at which the un
loading curve or its extension to the envelope originates (SE' FE); (2) the
ollllllon point (Se. Fe); and (3) the plastic strain ratio (Sp. 0) at which the
curve or its extension terminates.
For cycle ABCD[Fig. 17(a)] the three pOints through which the curves pass
are determined in the following manner. The,value of Sp at point B is found
using Eq. 5 (unloading from A); point C is found using Eq. 3 for Se and Eq. 2
for FC; point D is found by solving Eq. 4 for SE and Eq. 1 for FE'
1.0
" ...__-.:1 , .,,,/'NV.""..
0 ..
.."'-..' "
'-
'-,.

0
0.'
COMMON POINT LIMIT
0.' (ll=o.7ol
0,0 r "(.e
0,0 ,,,-.,nl'
,.0
,.\
,.0
,,0
(51411:1 !NVltOP
, ..... /
0.... __ ""'.'.......-'
./ lS".",I f$mQ..
A G ''''"'-.
0.6
/ ,/
1/
I
---.
.
0.'
,
/
0'-1
I ./ /
o.or '(-:
1.0 1.0
fbi
FIG. n.-LOADING AND UNLOADING CURVES (COMPUTED)
For cycle EFGH [Fig. 17(b)] between specified values of FmaxandFmin'
some modification of the preceding procedure is necessary. The unloading
curve EGF is part of curve E' GF', and the value of S p for curve E' GF' is
found by trial and error so that the curve will pass through point E when Eqs.
1, 2, 3, and 5 are satisfied. The unloading curve is terminated at F when F min
Is reached. The loading po.rtion FGH passes through points ..,Fmin)' (Se.
Fe) which was found in the preceding step, and point H' (SE,
using the value of at pOint F' inEq. 4. The curve is terminat-ed'when Fmax
is reached. A similar approach may be used to determine stress'"'strain paths
if a given strain increment is to be added.
A computer program was written to solve the various cases presented:
2561 2560 December, 1969
0<15 <-<
[ _0<T-<
0,60
I :{
j 0<75
< I IEXPfRI.ENTAL (F'::';Nl CURVE, I..
'\....01<
I
1 .... 1
..... -.'
'
, I"L_ I
.l

.--- --.0.__.-"..,-___..
"",0
I .j I
--I I I _
0,1$
P FOtiQuo.l,ml!O<U - __<_
.- I .. I '-' l
0.60
.00 Ht ...
N" Numb., of Cycl
FIG, 22.-NUMBER OF CYCLES TO FAILURE (Fmax " CONSTANT; Fmin " 0)

J
K
08

0.15
0.1
0.&
0<.
0-3
_ 0.2
0,1+ 0.'
00
f=.!;!!.. ___ ___ _____
O' .. I------.-------r-------\..
.0<1 ZOO '00
N =Number 0'1 Cycles
FIG. 23.-'NUl\IBEH OF CYCLES TO FAILURE'(F
max
CONSTANT; Froin I 0)
CONCRETE BEHAVIOR
and the two curves should intersect at some point below D [Fig. 19(b) J. In ad
dition the plastic strain ratios at point G differ considerably. On the basis of
the observed results, the assumption of uniqueness would not appear to be
fl\l.rranted.
Fig. 20 shows the computed and measured response for specimen AC4-10
which was cycled between stress levels of F max = 0.77 and F min O. The
speeimen failed in cycle 21 and failure was predicted in cycle 25. In Fig. 21
the response of specimen AC4-13 is shown. This specimen was subjected to
cyclic loads between stress levels of Fmax 0.79 and Fmin == 0.40. The spec
lInen failed in cycle 28 and failure was predicted in cycle 34. If uniqueness of
the loading and unloading curves was assumed, failure would be predicted
after only three cycles .
The computed number of cycles to fallure for tests in which the load is
varied between a given maximum stress level Fmax and a minimum stress of
zero is shown in Fig. 22. Both measured and computed values are <plotted.
Since the observed maximum of the stability limit was at a stress ratio of
0.63, the experimental curves shown in Fig. 22 will become asymptotic to F
0.63. Fig. 23 shows the computed nuinber of cycles to failure for loadings
between given maximum and minimum stress levels. The maximum stresS
ratio is plotted along the ordinate. For example, the number of cycles to
failure with Fmax == 0.80, and F min == 0040 is approximately 25. Using these
curves (Figs. 22, 23), the number of cycles to failure may be estimated.
SUMMARY AND CONCLUSIONS
A series of 46 short rectangular test specimens (Fig. I) were subjected to
repetitions of compressive stress to various levels to obtain expressions for
the response of plain concrete. The expressions developed are functions of
the ultimate stress and strain values of standard 6 x 12-in. control cylinders
and the loading history. Using these expressions the response of plain con
crete subjected to varying load histories can be estimated.
The following conclusions were obtained:
1. For the specimens tested, the envelope curve coincided with the stress
strain curve for a specimen under monotonic loading to failure (Fig. 7). The
stress-strain path reached the envelope regardless of the strain accumulated
prior to a particular cycle.
2. The location of the common pOints was dependent primarily on the mag
nitude of the maximum stress and strain of the previous load cycle. The com
mon points for loading from nonzero levels were identical to the common
points corresponding to load cycles starting at a stress of zero (Fig. 11).
3. Examination of the location of the common points shows that failure
would be produced under repeated loads with stresses exceeding about 0 .63 f
the maximum of the stability limit. This limit was independent of the mini
mum stress levels in the cycles.
4. Loading and unloading curves starting from .a pOint wtt11tli. tke stress
strain domain were not uriique;-and'the value of stress and strain at the peak
of the previous loading cycle must be known to estimate the re'l!ptmse. '
5. The analytical expressions obtained for the envelope curve, the common
point and the stability limits, and the loading and unloading stress-strain re
2563
'"."""'.
2562 December, 1969
lations produce results that compare well with the experimental results
3, 1B, 20, 21), The formulation of these expressions provides a general lUe"thi
od for estimating the number of cycles to failure under repeated loads (FigS
22, 23) with strairi rates similar to those considered in the investigation, .
APPENDIX I.-REFERENCES
I. Hognestad, E., "A Study of Combined Bending and Axial load in Reinforced Concrete Mein.
bers," University of Illinois Engineering Experimental Station, Bulletin Series No. 399,
195L
2. Hognestad, E., Hanson, N. W., and McHenry, D. "Concrete Stress Distribution in Ultimate
Strength Design," Journal of the American Concrt!tl1lnstitute. Vol. 52, No.4, December, 1955
pp.455-479. i
3. Karsan, I. D., "Behavior of Plain Concrete under Variable load Histories," thesis presented to
Rice University, at Houston, Texas, in 1968. in partial fulfillment of the requirements ror the
degree of Doctor of Philosophy.
4. Murdock, J. W., and Kesler, C. E., "Effect of Range of Stress on Fatigue Strength or Plain
Concrete Beams," Journal of the American Concrete Institute. Vol. 55, No.2, August, 1958, Pp.
221-231.
5. Nordby, G. M., "Fatigue of Concrete-A Review of Research," Journal of the American Con.
crete Institute. Vol. 55, No.2, August, 1958, pp. 191-219.
6. Rusch, H., "Researches toward a General Flexural Theory for Structural Concrete," Journal oJ
the American Concrete Institute. Vol. 57. No. I, July, 1960, pp. 128.
7. Shah, S. P., Sturman, G. M., and Winter, G., "Microcnicking and Inelastic Behavior of Con.
crete," Flexural Mechanics of Reinforced Concrete. ASCE, 1965.. 50, The International Sym.
posium, Miami, Florida, 1964.
8. Shah, S. P., and Winter, G .. "Inelastic Behavior and Fracture of Concrete," Journal of the
American Concrete Institute. Vol. 63, No.9, September, 1966, pp. 925-930.
9. Shah, S. P., and Winter, G., "Response of Concrete to Repeated Loadings," RlLEM. Interna
tional Symposium on the Effects of Repeated Loading on Materials and Structural Elements,
Mexico City, 1966.
10. Sinha. B. P., Gefstle, K. H., and Tulin, l. G., "Stress-Strain Relations for Concrete under
Cyclic Loading," Journal of the American Concrete Institute. VoL 61, No.2, February, 1964,
pp. 195-211.
. Smith, G. M., and Young, l. E., "Ultimate Theory in Flexure by Exponential Function," Jour.
nal of the American Concrete Institute. Vol. 52, No.3, November, 1955, pp. 349.. 359.
APPENDIX II, -NOTATION
The following symbols are used in this paper:
= ultimate compressive strength 'ofstandard 6-in. x 12-in.cylinder;
f = concrete stress;
f max = maximum compressive stress reached in a given cycle;
F = I In = stress ratio;
CONCRETE BEHAVIOR
FaX = maximum stress ratio in a given cycle;
; in minimum stress ratio in a given cycle;
Fe = stress ratio at the common point;
FE = stress ratio on the envelope curve;
S = tlto strain ratio;
maximum strain ratio in a given cycle;
SJ!UllC
minimum strain ratio in a given cycle;
SIIlin
strain ratio at the common point;
Se
pi 0 = plastic or residual strain ratio;
Sp
strain ratio on the envelope curve;
a factor relating the common point with the stress and strain ratios
of the peak of the previous load cycle;
( = concrete strain at I;
(0
concrete strain at I and
(p
plastic strain.