ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 89-815

Shrinkage Cracking in Fully Restrained Concrete Members

by R. lan Gilbert

This paper considers the problem of cracking in fully restrained not exposed and esthetic requirements are of secondary
members subjected to direct tension caused by dry shrinkage. The importance, a larger crack width may be acceptable
mechanism of direct tension cracking is discussed, and some popular
misconceptions concerning the behavior of restrained members are
(say 0.3 to 0.5 mm).
exposed. Paper presents a rational approach for the determination of Most existing techniques for predicting crack widths
the number and spacing of cracks and the average crack width in a involve empirical models that have been calibrated
member, which is fully restrained and subjected only to an axial re- from results obtained in laboratory tests and, as such,
straining force caused by shrinkage. The approach is based on the often fail to accurately predict crack widths in actual
principles of mechanics and is illustrated by worked examples. Pre-
dictions agree well with observed cracking in restrained members. The
structures. It is difficult, therefore, for a structural de-
procedure is used to calculate the quantities of steel required for crack signer to be confident that the cracks in a structure will
control in a number of practical situations. Finally, the results of the in fact satisfy the maximum crack width requirements
investigation are compared with the provisions for shrinkage and of the local building code.
temperature reinforcement in the ACI Building Code (ACI 318-89) In beams and slabs, which have been proportioned to
and AS 3600-1988.
avoid excessive deflection at service loads and which
Keywords: cracking (fracturing); crack width and spacing; creep; deformation; contain sufficient quantities of reinforcement to pro-
reinforced concrete; serviceability; shrinkage; slabs; structural members.
vide adequate ultimate strength and ductility, flexural
In reinforced and partially prestressed concrete crack widths are rarely a problem under normal in-
structures, cracking is to be expected at service loads. service conditions, provided of course that the rein-
Cracks may be caused by the external loads, or by re- forcement bars are not spaced too widely apart. Sensi-
straint to shrinkage and temperature variations, or by a ble reinforcement detailing is the key to flexural crack
variety of other causes. If uncontrolled, cracking may control. It is, therefore, not usually critical if flexural
spoil the appearance of a structure or otherwise ad- crack widths are not specifically checked in design.
versely affect its performance. When flexural members are also restrained at the
Excessively wide cracks in floor systems and walls supports, shrinkage causes a buildup of tension in the
may often be avoided by the inclusion of strategically member, in addition to the bending caused by the ex-
placed contraction (or control) joints, thereby remov- ternal loads. Crack control is still not usually a prob-
ing some of the restraint to shrinkage and temperature lem, since shrinkage is accommodated by small in-
movements and reducing the internal tension. When creases in the widths of the numerous flexural cracks.
cracking does occur, to insure that crack widths remain However, for members not subjected to significant
acceptably small, adequate quantities of well-anchored bending in which restraint is provided to the longitudi-
reinforcement must be included at every location in the nal movement caused by shrinkage and temperature
structure where significant tension is expected. changes, cracks tend to propagate over the full depth of
The maximum crack width, which may be consid- the section. Exessively wide cracks are not uncommon.
ered to be acceptable in a given situation, depends on Such cracks are commonly called direct tension cracks,
the type of structure, the environment, and the conse- since they are caused by direct tension rather than by
quences of excessive cracking. In corrosive and aggres- flexural tension. In fully restrained direct tension mem-
sive environments, some building codes and specifica- bers, relatively large amounts of reinforcement are
tions recommend that crack widths do not exceed about required to control load-independent cracking. The
0.3 mm (0.012 in.). For members with one or more ex- ACI Structural Journal, V. 89, No.2, March-Aprill992.
posed surfaces, a maximum crack width of 0.3 mm Received Apr. I, 1991, and reviewed under Institute publication policies.
Copyright © 1992, American Concrete Institute. All rights reserved, including
should also provide visual acceptability. For the shel- the making of copies unless permission is obtained from the copyright propri-
etors. Pertinent discussion will be published in the January-February 1993 ACI
tered interior of most buildings where the concrete is Structural Journal if received by Sept. I, 1992.

ACI Structural Journal I March-April 1992 141

Jan Gilbert is Professor of Civil Engineering and Head of the Department of
particular cross section first reaches the direct (uniax-
Structural Engineering at the University of New South Wales, Sydney, Austra- ial) tensile strength of concrete j,, full-depth direct ten-
lia. His research interests are in servicability of concrete structures, and he is sion cracking occurs. First cracking may occur less than
the author of over 80 research papers and two books in the area. Dr. Gilbert is
a member of the Concrete Structures and Composite Structures Comittees of
1 week a£ter the commencement of drying. The magni-
Standards Australia and he was actively engaged in the development of the tude of N(t) after cracking and the crack width depend
Australian Standard for Concrete Structures AS 3600-1988. primarily on the amount of bonded reinforcement
crossing the crack. If the member contains no longitu-
dinal steel, cracking causes the restraining force N(t) to
minimum quantities of reinforcement specified in the
drop to zero and a wide, unsightly crack results. If the
Australian Code AS 3600-1988 1 for crack control in re-
member contains only small quantities of reinforce-
strained members are much greater than that required
ment (p = A/bh is less than about 0.003 for ,[y = 400
in the ACI Building Code (metric version) ACI 318M-
MPa), the steel at the crack yields (either immed~ately
892 and are based on Australian research by Base and
or after additional shrinkage), the crack opens widely,
Murray3•4 and others.
and the restraining force drops to a value of AJy,
In this paper, the mechanism of direct tension crack-
which may be only a small fraction of its value prior to
ing is discussed, and a rational approach is presented
cracking. If the member contains relatively large quan-
for the determination of the average crack width in a
tities of reinforcement (p greater than about 0.01 for /y
member that is fully restrained and subjected only to an
axial restraining force caused by shrinkage. The pre-
= 400 MPa), the steel at each crack does not yield, the
crack width remains small, and because the loss of
dicted number and width of shrinkage cracks agree well
member stiffness at cracking is not great, the restrain-
with observed cracking in restrained members. The ap-
ing force remains high. Members containing large
proach is used to calculate the quantities of steel re-
quanities of steel are therefore likely eventually to suf-
quired for crack control in restrained members for a
fer many cracks, but each will be fine and well con-
variety of situations. Finally, the results of the investi-
trolled. For intermediate steel quanities (0.003 < P <
gation are compared with the provisions for crack con-
0.01 when JY = 400 MPa), cracking causes a loss of
trol in AS 3600-1988 1 and ACI 318M-89. 2
stiffness, a reduction of N(t), and a crack width that
may or may not be acceptable.
CRACKING IN DIRECT TENSION MEMBERS
In addition to the quantity of steel, the width of a
Consider a reinforced concrete direct tension mem-
crack in a restrained member depends on the quality of
ber that is prevented from shortening by its supports or
bond between the concrete and the steel, the size and
by adjacent parts of the structure. As the concrete
distribution of the individual reinforcement bars, the
shrinks, an axial tensile restraining force N(t) develops
concrete quality, and whether or not the axial restrain-
with time. When the concrete stress caused by N(t) at a
ing force is accompanied by bending. Direct tension
cracks are more parallel-sided than flexural cracks, and
A 5 = area of reinforcing steel
(
hence the observed crack width is less dependent on the

-
( ~ = area of concrete
amount of concrete cover.
N(t)- ~t===~t=;}_'~======!~ N(t) =~f,
FIRST CRACKING
1--------L------
Description and notation
(a) Just prior to first cracking. Consider the fully restrained member shown in Fig.
1(a). As the concrete shrinks, the restraining force N(t)
gradually increases until the first crack occurs when
N(t) = Acfr· Immediately after first cracking, the re-
1--------L-------1
straining force reduces to Nero and the concrete stress
(b) Just after first cracking. away from the crack is less than the tensile strength of
the concrete j,. The concrete on either side of the crack
shortens elastically and the crack opens to a width w,
as shown in Fig. 1(b). At the crack, the steel carries the
J Region I Region 2 Region I J
entire force Nero and the stress in the concrete is obvi-
(c) Average concrete stress just after flrst cracking.
ously zero. In the region immediately adjacent to the
crack, the concrete and steel stresses vary considerably,
and there exists a region of partial bond breakdown. At
some distance S 0 on each side of the crack, the concrete
and steel stresses are no longer influenced directly by
the presence of the crack, as shown in Fig. l(c) and (d).
In Region 1, where the distance x from the crack is
(d) Steel stress just after first cracking. greater than or equal to S 0 , the concrete and steel
stresses are ac1 and <151 , respectively. Since the steel stress
Fig. 1 - First cracking 'in a restrained direct tension (and hence strain) at the crack is tensile as shown and
member the overall elongation of the steel is zero (full re-
142 ACI Structural Journal I March-April 1992
straint), a,1 must be compressive. Equilibrium requires where
that the sum of the forces carried by the concrete and
the steel on any cross section is equal to the restraining 2 S0
(6)
force. Therefore, with the force in the steel in Region 1 cl = 3 L - 2 So
being compressive, the force carried by the concrete (ac~
Ac) must be tensile and somewhat greater than the re- Prior to cracking, the total concrete strain at any point
straining force (Ncr). In Region 2, where the distance x is zero, since shortening is prevented. Although the to-
from the nearest crack is less than S 0 , the concrete stress tal concrete strain is zero, the individual strain compo-
varies from zero at the crack to ac1 at x = S 0 • The steel nents of creep, shrinkage, and elastic strain are not.
stress varies from as2 (tensile) at the crack to a,1 (com- The creep and elastic strains are tensile (positive) and
pressive) at x = S 0 , as shown. the shrinkage strain is compressive (negative). The sum
To determine the crack width w and the concrete and of the time-dependent creep and shrinkage strain com-
steel stresses in Fig. 1, the distance S 0 , over which the ponents must be equal and opposite to the elastic strain
concrete and steel stresses vary, needs to be known and component. Immediately before the first crack occurs
the restraining force Ncr needs to be calculated. An ap- when the concrete stress just reaches .ft, the sum of the
proximation for S 0 may be obtained using the following creep and shrinkage strain components is therefore
equation, which was proposed by Favre et al. 5 for a
member containing deformed bars or welded wire mesh
(7)

(1)
where Ec is the elastic modulus of the concrete at the
time of first cracking. Immediately after first cracking,
where db is the bar diameter, and p is the reinforcement the magnitude of the elastic component of strain in the
ratio A/Ac. Base and Murray4 used a similar expres- uncracked concrete decreases (as the concrete stress de-
sion. creases), but the creep and shrinkage strain compo-
nents are unaltered. Creep and shrinkage strains only
Calculation of restraining force and internal
change gradually with time.
stresses
In Region 1, at any distance greater than so from the
The following procedure may be used to determine
crack, equilibrium requires that the sum of the forces in
the restraining force Ncr immediately after first crack-
the concrete and the steel immediately after first crack-
ing and the corresponding steel and concrete stresses.
ing is equal to Ncr• That is
The overall elongation of the steel is zero, since the
member is fully restrained and therefore prevented
(8)
from shortening. Integrating the steel strain over the
length of the member [and assuming a parabolic varia-
and substituting Eq. (5) into Eq. (8) and rearranging
tion of stress and, hence, strain in Region 2, as shown
gives
in Fig. 1(d)] gives

CT5 1
-L
E,
+ CTs2 -
E,
CT5 1 (2-s3 + w) = 0
o
(2)
(9)

The compatibility requirement is that the concrete and

Realizing that w is very much less than S0 , Eq. (2) may steel strains in Region 1 are identical, i.e.
be rearranged to give
(10)
-2S0
CT
sl -
-
3 L - 2 S0 CT
s2
(3)
With the concrete strain e1 equal to the sum of the
elastic, creep, and shrinkage strain components, Eq.
At the crack, the restraining force Ncr is carried entirely (10) can be reexpressed as
by the steel. That is

(11)
(4)
Substituting Eq. (5), (7), and (9) into Eq. (11) and
By substituting Eq. (4) into Eq. (3), the steel stress solving for Ncr gives
away from the crack is expressed in terms of the un-
known restraining force
(12)
2 S0 Ncr C Ncr
= 3L - 2 S 0 A,. = (5)
CTsl I A, where p = A/Ac and n = E/Ec.
ACI Structural Journal I March-April1992 143
 pend on the length of the member, the quantity and
distribution of reinforcement, the quality of bond be-
tween the concrete and steel, the amount of shrinkage,
and the concre1e strength. A typical final shrinkage
~-------4-----L------------~ crack pattern is shown in Fig. 2. Let the number of
cracks be m and the final shrinkage-induced restraining
Fig. 2 - Final crack pattern in a restrained direct ten- force be N(oo).
sion member
In Fig. 3(a), a portion of a fully restrained direct
w w tension member is shown after all sh.rinkage has taken
place and the final crack pattern is established. The av-
erage concrete and steel stresses caused by shrinkage are
illustrated in Fig. 3(b) and (c). It is assumed that the
distance S in which the concrete and steel stresses vary
0

(a)Ponion of a restrained member after all cracking. on either side of each crack is the same as the distance
S given by Eq. (1) and previously used in the analysis
0
w w
at first cracking. In Region 1, where the distance x
from the nearest crack is greater than or equal to S the
0,

final concrete and steel stresses are a;1 and a;1 , respec-
tively.

Calculation of final stresses and deformation

(b) Average concrete stress after all shrinkage. The following analysis may be used to determine the
number of cracks m, the final restraining force N(oo),
and the final average crack width w.
For the member containing m cracks, the following
expressions [similar to Eq. (2) and (3)] are obtained by
equating the overall elongation of the steel to zero
w w w
(Jsl
E
. L + m . E . (2Jso +
(Js2 - lis!
(13)
s s
(c) Steel stress after all shrinkage cracking.

Fig. 3 - Final concrete and steel stresses after direct and rearranging gives
tension cracking

When Ncr is calculated from Eq. (12), the concrete and (14)
steel stresses immediately after cracking may be ob-
tained from Eq. (4), (5), and (9). since w is very much less than S Dividing both the nu-
0•

merator and the denominator on the right-hand side of

DETERMINATION OF FINAL CRACK SPACING
Eq. (14) by m and letting the crack spacings = Lim,
AND CRACK WIDTH
gives
Discussion and notation
With the stresses and deformations determined im-
mediately after first cracking, the subsequent long-term (15)
behavior as shrinkage continues must next be deter-
mined. Consider again the fully restrained member
shown in Fig. 1(a) and (b). After first cracking, the where
concrete is no longer fully restrained since the crack
width can increase with time as shrinkage continues. A (16)
state of parital restraint therefore exists after first
cracking. Subsequent shrinkage will cause further
gradual increases in the restraining force N(t) and in the At each crack
concrete stress away from the crack, and a second crack
may develop. Additional cracks may occur as the , N(oo)
(Js2 = ~ (17)
shrinkage strain continues to increase with time. How- s
ever, as each new crack forms, the member becomes
less stiff and the amount of shrinkage required to pro- In Region 1, away from each crack, a typical concrete
duce each new crack increases. The process continues stress history is shown diagrammatically in Fig. 4. The
until the crack pattern is established, usually in the first concrete tensile stress increases gradually with time and
few months after the commencement of drying. The approaches the direct tensile strength of the concrete ft.
number of cracks and the final average crack width de- When cracking occurs elsewhere in the member, the
144 ACI Structural Journal I March-April1992
tensile stress in the uncracked regions drops suddenly as
shown. Although the concrete stress history is continu-
ously changing, for the estimation of creep strain it is
not unreasonable to assume that the average concrete
stress at any time after the commencement of drying uav
is somewhere between uci andj,, as shown in Fig. 4, and
the final creep strain in Region 1 may be approximated
Fig. 4 - Concrete stress history in uncracked Region 1
by
The compatibility requirement is that the concrete and
(18) steel strains are identical

where cJ>' is the final creep coefficient (defined as the

ratio of the final creep strain to elastic strain under the and using Eq. (21)
average sustained stress u.v>·
In this study, it is assumed that Us!
. Uav '
E = E' + e.h (24)
s c

(19) Substituting Eq. (15) and (17) into Eq. (24) and re-
arranging gives
The final concrete strain in Region 1 is the sum of the
elastic, creep, and shrinkage components and may be N(oo) = -~
n'A.(Uav + 'E')
fshe (25)
approximated by
where n* = E/E;.
The crack spacing s (=Lim) must be first deter-
(20) mined to calculate C2 [as defined in Eq. (16)]. Knowing
that u;1 must be less than the tensile strength of con-
crete j, and making use of Eq. (15) and (17), Eq. (23)
The magnitude of the final creep coefficient cJ>' is usu- becomes
ally between 2 and 4, depending on the age at the com-
mencement of drying and the quality of the concrete.
e;h is the final shrinkage strain and depends on the rel- (26)
ative humidity, the size and shape of the member, and
the characteristics of the concrete mix. Numerical esti- Substituting Eq. (16) and (25) into Eq. (26) gives
mates of cJ>' and e;h can be obtained from ACI 2091i and
elsewhere. A number of the more well-known methods 2 S 0 (1 + ~}
s ~ 3~ (27)
for predicting both q,• and e;h have been presented and
compared previously by the author.
Eq. (20) may be expressed as where

(21) (28)

where E; is final effective modulus for concrete and is The number of cracks m ( = Lis) may be taken as the
given by smallest integer that causes Eq. (27) to be satisfied.
With m thus determined, the restraining force N( oo) can
be calculated using Eq. (25) and the steel and concrete
E'=~ (22) stresses in the various regions of the member may be
e 1 + cJ>'
determined from Eq. (15), 17, and (23).
The overall shortening of the concrete is an estimate
In Region 1, at any distance from a crack greater of the sum of the crack widths. The final concrete
than S 0 , equilibrium requires that the sum of the force strain at any point in Region 1 of Fig. 3 is given by Eq.
in the concrete and the force in the steel is equal to (21), and in Region 2, the final concrete strain is
N(oo). That is

(29)

(23) where fn varies beween zero at a crack and unity at S 0

from a crack. If a parabolic variation of stress is as-

ACI Structural Journal I March-April1992 145
sumed in Region 2, the following expression for the av- the slab (As = 750 mm 2/m). The concrete cover to the
erage crack width w is obtained by integrating the con- reinforcement is 30 mm. Estimate the spacing s and fi-
crete strain over the length of the member nal average width w of the restrained shrinkage cracks.
Take
(30)
¢' = 2.5; E;h = - 600 x 10- 6 ; .t; = 2.0 MPa;

The preceding analysis may be used to determine the Ec = 25,000 MPa; Es = 200,000 MPa;
number and width of shrinkage cracks, provided the n = 8; and,[y = 400 MPa.
assumption of linear-elastic behavior in the steel is
valid. However, if the area of steel As is small, yielding The concrete area and reinforcement ratio are
may occur at each crack and the value of N( oo ), calcu-
lated from Eq. (12), will not be correct. In such a case,
oo; is equal to the yield stress of the reinforcement ,[y Ac ::::: Agross = 150,000 mm /m;
2
p = As
A = 0.005
c
and N( oo) is equal to ,[y As. From Eq. (23), the stress in
the concrete away from the crack in Region 1 is now and from Eq. (1)

(31) 12
So = 10 X 0.005 = 240 mm

After the steel at the first crack yields, the tensile The final effective modulus is obtained from Eq. (22)
concrete stress <Tc~ increases only slightly as the com-
pressive steel stress <Ts1 increases with time and the first • 25,000
crack opens. Since the restraining force is constant at Ee = 1 2 = 7143 MPa
all times after yielding of the steel at the first crack (and
+ .5
equal to ,[yAs), the tensile stress in the concrete never and the corresponding effective modular ratio is n*
again approaches the tensile strength of concrete, and E/E; = 28. Eq. (6) gives
subsequent cracking does not occur. The width of the
initial crack is usually unacceptably large as the steel at
the crack deforms plastically. The crack width w may cl = 2 x 240 = o.o331
3 X 5000 - 2 X 240
be found by insuring that the overall elongation of the
steel is zero. That is
and from Eq. (12), the restraining force immediately
u' f' - u' 2 after first cracking is
~ (L - w) + JY 51 - S0 + w = 0 (32)
Es Es 3
8 X 0.005 X 2.0 X 150,000
and since w is very much smaller than L, Eq. (32) can Ncr = 0.0331 + 8 X 0.005 (1 + 0.0331)
be arranged to give = 161,300 N/m

The concrete stress <Tc1 is obtained from Eq. (9)

= 161,300(1 + 0.0331) = 1.1 1 MPa

Since the tensile stress in the uncracked concrete does <Tel 150,000
not change significantly with time, it is reasonable to
assume that the average concrete stress <Tav is given by and from Eq. (19), the average concrete stress may be
Eq. (31) and the final steel stress in Region 1 may be approximated by
obtained by substituting Eq. (31) into Eq. (21) and
simplifying 1.11 + 2.0
1.56 MPa
2

Eq. (28) gives

(34) -28 X 0.005 (1.56 - 0.0006 X 7143)

~ = 28 X 0.005 (1.56 - 0.0006 X 7143) + 2.0
NUMERICAL EXAMPLES = 0.236
Case (a)
Consider a 5-m long and 150-mm thick reinforced and from Eq. (27), the crack spacing must satisfy
concrete slab which is fully restrained at each end. The
slab contains 12-mm diameter deformed longitudinal 2 X 240 (1 + 0.236)
S :!( - ·839 mm
bars at 300-mm centers in both the top and bottom of ""' 3 X 0.236 -
146 ACI Structural Journal I March-April 1992
The minimum number of cracks m is obtained from yield stress and clearly incorrect. The steel at each crack
has yielded and the previous analysis is not valid.
L 5000
m=-:::?:--=596
s 7 839 . .
If the steel at the crack has yielded

Therefore m = 6 and s = Lim = 833 mm.

The constant C2 is obtained from Eq. (16) N(oo) = A.fy = 150,000 N/m

and only a single crack occurs. From Eq. (34)

c2 = 2 x 240 = 0.238
3 X 833 - 2 X 240
28 X 0.0025 X 400 - 600 X 10- 6 X 200,000
and the final restraining force is calculated using Eq.
a;t = 1 + 28 X 0.0025
(25) = -86.0 MPa

and the tensil stress in the concrete away from the crack
N(oo) X 750 (1 56 - 0 0006
= - 280.238 7143) is given by Eq. (31)
. . X
= 240,900 N/m
400 X 375 + 86.0 X 375
a;t = 150,000
1.22 MPa
From Eq. (15), (17), and (23)

u;
2 = 321 MPa; The crack width may be calculated from Eq. (33)
u;1 = -7.643 MPa; and u;1 = 1.99 MPa
W=
- 86.0(3 X 5000 - 2 X 480) + 2 X 480 X 400
The final crack width is determined using Eq. (30)
3 X 200,000
= 1.37 mm
W 1.99 ( 833 - 3
= - [ 7143 2 X 240) - 0.0006 X 833 ]
As expected, when the steel is at yield across the crack,
= 0.31 mm the crack width is excessive.

Case (b) CRACK CONTROL IN RESTRAINED SLABS

Consider the slab of Case (a), with one-half the As mentioned in the introduction, flexural cracks are
quantity of reinforcement, i.e., A, = 375 mm2/m. The rarely a problem in reinforced concrete slabs, provided
number and final average width of the restrained of course that bonded reinforcement at reasonable
shrinkage cracks are to be calcualted. All material spacing crosses the crack and that the member does not
properties are as for Case (a). deflect excessively. In contrast, direct tension cracks
due to restrained shrinkage and temperature changes
For this slab frequently lead to serviceability problems, particularly
in regions of low moment. Such cracks usually extend
completely through the slab and are more parallel-sided
Ac = 150,000 mm 2/m and p = 0.0025. than flexural cracks. If uncontrolled, these cracks can
become very wide and lead to waterproofing and cor-
If 12-mm diameter bars are used, Eq. (1) gives rosion problems. They can also disrupt the integrity
s. = 480 mm. As for Case (a), E; = 7143 MPa and n· and the structural action of the slab.
= 28. Evidence of direct tension-type cracks are common in
From Eq. (6), C 1 = 0.0684 and Eq. (12) gives reinforced concrete slab systems. For example, in a
Nc, = 66,840 N/m. The concrete stress in Region 1 af- typical one-way beam-slab floor system, the load is
ter first cracking is calculated from Eq. (9) to be usually carried by the slab across the span to the sup-
ucl = 0.476 MPa and, using Eq. (19), u•• = 1.24. Eq. porting beams, while in the orthogonal direction the
(28) gives ~ = 0.119 and, from Eq. (27), s ~ 3002 bending moment is small. Shrinkage is the same in both
mm. The number of cracks is therefore m = 2 since directions and restraint to shrinkage usually exists in
Lis ~ 1.67. Therefore s = Lim = 2500 mm. From both directions.
Eq. (16), C 2 = 0.147 and Eq. (25) gives N(oo) = In the span direction, shrinkage will cause small in-
217,900 N. The steel stress at each crack calculated us- creases in the widths of the many fine flexural cracks
ing Eq. (17) is ut = 581 MPa, which is greater than the and may cause additional flexure-type cracks in the
ACI Structural Journal I March-April 1992 147
 r
previously uncracked regions. However, in the orthog- Ncr 194 kN/m, Usl = -5.36 MPa, ucl
onal direction, which is virtually a direct tension situa- 1.33 MPa, and us2 = 206 MPa
tion, shrinkage generally causes a few widely spaced
cracks that penetrate completely through the slab. Fre-
and after all.shrinkage has taken place
quently, more reinforcement is required in the orthog-
onal direction to control these direct tension cracks
than is required for bending in the span direction. As s = 500 mm, N(oo) = 203 kN/m, u;2 = 215 MPA
far as cracking is concerned, it is not unreasonable to u;1 = -73.2 MPa, u;1 = 1.82 MPa
say that shrinkage is a greater problem when it is not
accompanied by flexure. and the final crack width is w = 0.21 mm.
If the amount of reinforcement crossing a direct ten- This would normally be considered to be an accept-
sion crack is too small, yielding of the steel will occur able maximum crack width in a severe exposure condi-
and a wide unserviceable crack will result. To avoid this tion, and the analysis presented herein therefore en-
eventuality, Campbell-Allen and Hughes 7 proposed the dorses this provision of AS 3600. 1
following expression for the minimum steel ratio Pmin For a sheltered environment, in which visible cracks
could cause esthetic problems, AS 3600 suggests that
the minimum area of steel should be taken from Eq.
Pmin = (Ast) = 1.2ft (35) (36). Where cracks are not likely to cause esthetic
bd min Jsy
problems, in a sheltered environment, AS 3600 speci-
fies
where .t; is the tensile strength of immature concrete
(usually about 3 days old) and may be taken as 0.25 1.4
JJ:. For 25-MPa concrete and 400-MPa steel, Eq. (35) Pmin = J;, (37)
gives a minimum reinforcement ratio of Pmin = 0.0038
for the control of shrinkage cracking. This is consider-
ably less steel than contained in the slab of the example This would apply, for example, in the case of an inte-
(p = 0.005), for which the calculated final crack rior slab in which visible cracking could be tolerated or
widths were just over 0.3 mm, and might therefore be in the case of an interior slab that was later to be cov-
considered inadequate for some applications. ered by a floor covering and/or a false ceiling. Eq. (37)
ACI 318M-89 2 suggests minimum ratios of rein- corresponds to a reinforcement ratio of 0.0035 for a
forcement area to gross concrete area in the direction at reinforced slab with 400-MPa steel.
right angles to the principal reinforcement direction in Analyzing the same slab just considered, except that
a one-way slab. For Grade 400 deformed bars, the p = 0.0035, gives
minimum ratio is Pmin = 0.00018. This is not to say
that this is what is required in all situations-it is the Ncr = 109 kN/m, uc1 0. 76 MPa, Usz 207 MPa
absolute mimimun requirement. In many restrained
members, significantly more shrinkage steel is required
to provide crack control. and after all shrinkage
The Australian code AS 3600-1988 1 contains the fol-
lowing provisions for crack control in reinforced con- s = 1250 mm, N(oo) = 191 kN/m, u;2 = 363 MPa
crete slabs. u; 1 = 1.56 MPa and w = 0.53 mm
Where the ends of a slab are restrained and the salb
is not free to expand or contract, the minimum ratio of
reinforcement to gross concrete area in the restrained In a sheltered environment, this size crack may be rea-
direction, when the slab is located in a severe exposure sonable provided visible cracking can be tolerated.
condition, is Where the ends of a slab are unrestrained and the
slab is free to expand or contract, the minimum rein-
forcement ratio is
2.5
Pmin = JY (36)
0.7
Pmin = .{y (38)

which equals 0.0063 for a slab containing 400-MPa

steel. This steel area is recommended, for example, in slabs-
For a 5-m long and 150-mm thick nonprestressed on-ground with control joints at regular centers and is
concrete slab with similar material properties to that similar to the minimum provisions in ACI 318M-89.
analyzed earlier in the numberical examples and with
p = 0.0063 (db = 12 mm), the analysis described in
this paper gives the following results immediately after CONCLUDING REMARKS
first cracking A simple procedure is presented for ·the determina-
148 ACI Structural Journal I March-April 1992
tion of the stresses and deformation after shrinkage forced Concrete and Commentary, (ACI 318M-89/318RM-89.CT92)
(metric version)," American Concrete Institute, Detroit, 1989, 353
cracking in a fully restrained direct tension member. pp.
The predicted number and width of shrinkage cracks 3. Base, G. D., and Murray, M. H., "Controlling Shrinkage
are in accordance with observed cracking in restrained Cracking in Re!ltrained Reinforced Concrete," Proceedings, 9th
members, and the results of the analysis agree with and Conference, Australian Road Research Board, V. 9, Part 4, Bris-
provide endorsement for the direct tension crack con- bane, 1978.
4. Base, G. D., and Murray, M. H., "New Look at Shrinkage
trol provisions of AS 3600-1988. 1 Cracking," Civil Engineering Transactions, Institution of Engineers
Australia, V. CE24, No.2, May 1982, 171 pp.
5. Favre, R., eta!., "Fissuration et Deformations," Manual du
CONVERSION FACTORS Comite Ewo-International du Beton (CEB), Ecole Polytechnique
1mm 0.039 in. Federale de Lausanne, Switzerland, 1983, 249 pp.
1 MPa 145 psi 6. Subcommittee II, ACI Committee 209, "Prediction of Creep,
1 kN/m 5.71lbf/in. Shrinkage, and Temperature Effects in Concrete Structures, 2,"
Draft Report, American Concrete Institute, Detroit, Oct. 1978, 98 pp.
7. Campbell-Allen, D., and Hughes, G. W., "Reinforcement to
REFERENCES Control Thermal and Shrinkage Cracking," Research Report R334,
1. "Australian Standard for Concrete Structures," (AS 3600- School of Civil Engineering, University of Sydney, Nov. 1978, 33 pp.
1988), Standards Association of Australia, Sydney, 1988, 106 pp. 8. Gilbert, R. I., "Time Effects in Concrete Structures," Elsevier
2. ACI Committee 318, "Building Code Requirements for Rein- Science Publishers, 1988, 321 pp.

ACI Structural Journal I March-April 1992 149