Advanced Crack Width Analysis of Reinforced Concrete

Beams under Repeated Loads

Byung Hwan Oh, M.ASCE1; and Se Hoon Kim2

Abstract: Presented is a method which can predict accurately the crack width in reinforced concrete beams under repeated loading. It is
known that the crack widths increase with an increase of the number of load cycles. To evaluate the crack widths realistically, the bond
stress-slip relation and the variation of transfer length under repeated loading have been modeled and incorporated in the analysis model
for crack width. The present study indicates that the transfer length decreases according to the increase of load cycles, which is due to the
increase of residual slip. The calculated crack widths from proposed method agree very well with test data available in the literature, while
the existing method shows large deviation from the test data. The proposed method also allows the calculation of the residual crack width
after repeated loading. The present study provides a realistic base for the prediction of crack width, which is important for the service-
ability of reinforced concrete structures subjected to repeated loads.
DOI: 10.1061/共ASCE兲0733-9445共2007兲133:3共411兲
CE Database subject headings: Cracking; Repeated loads; Slip; Concrete beams; Concrete, reinforced.

Introduction for Structural Concrete 2000兲, a new method of crack width

analysis, which employs the local bond stress-slip relationship
The bond between reinforcing bars and concrete directly affects under repeated loading, must be developed to predict the crack
cracking behavior of reinforced concrete members 共Balázs 1987, widths under repeated loadings accurately. The purpose of the
1991, 1993; Broms 1965; Broms and Lutz 1965; Lovegrove and present paper is, therefore, to propose a rational method of pre-
El Din 1982; Balaguru and Shah 1982; Oh and Kim 2000; Campi dicting crack width for reinforced concrete beams under repeated
et al. 1982; Bhuvasorakul 1974兲. The bond stress versus slip re- loading. The results from the proposed theory are compared with
test results.
lationship is necessary to calculate the crack width in reinforced
concrete members together with the evaluation of transfer length
and stress distributions between the cracks. The transfer length is
Review of Crack Width Analysis under Monotonic
the length that is necessary for transferring the axial force of
Loading
reinforcing bar to concrete. A method of analysis for crack width
using local bond stress-slip relationship under monotonic loading A rigorous formulation of crack width analysis is based on the
has been presented by Balázs 共1993兲. integration of the steel and concrete strains between cracks and is
The crack width is one of the most important factors that affect written as
the serviceability of reinforced concrete members. The crack
widths increase as the number of repeated loads increases. There-
fore, accurate prediction of crack width is very important for re-
inforced concrete members under repetitive cyclic loads such as
w= 冕sw
共␧sx − ␧cx兲dx 共1兲

bridges and offshore structures. Realistic models for crack widths where sw = crack spacing; ␧sx = strain of reinforcing bars at the
which are based on appropriate bond stress-slip relation under position x within the transfer length; and ␧cx = strain of concrete at
repeated loading are not available yet. As bond properties be- the position x within transfer length, lt.
tween reinforcing bar and concrete are changed under repeated In the absence of bond stresses, the crack width is determined
loading 共Oh and Kim 2006; Balázs 1991; International Federation simply by the integral of constant steel strains between cracks as

1 w = sw f s/Es 共2兲
Professor, Dept. of Civil Engineering, Seoul National Univ., San
56-1, Shinrim-dong, Gwanak-gu, Seoul, Korea. E-mail: bhohcon@ where, f s = stress of reinforcing bar and Es = modulus of the elas-
snu.ac.kr ticity of reinforcing bars. The crack spacing sw must be known in
2
Research Associate, Dept. of Civil Engineering, Seoul National order to use Eq. 共2兲. The crack spacing is generally a function of
Univ., San 56-1, Shinrim-dong, Gwanak-gu, Seoul, Korea. several parameters including concrete compressive strength, steel
Note. Associate Editor: Rob Y. H. Chai. Discussion open until August diameter, transfer length of bond stress and so on. However, the
1, 2007. Separate discussions must be submitted for individual papers. To
crack spacing has been simply assumed as a function of ultimate
extend the closing date by one month, a written request must be filed with
the ASCE Managing Editor. The manuscript for this paper was submitted bond strength or concrete cover by many researchers 共Broms
for review and possible publication on February 11, 2004; approved on 1965; Broms and Lutz 1965兲. This is mainly due to the complex-
May 10, 2006. This paper is part of the Journal of Structural Engineer- ity of the bond cracking problem.
ing, Vol. 133, No. 3, March 1, 2007. ©ASCE, ISSN 0733-9445/2007/3- It is reported that stirrups, particularly when closely spaced,
411–420/$25.00. may prevent the opening of splitting cracks that form along em-

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MARCH 2007 / 411

 may change the bond strength in the forthcoming Eq. 共5兲. How-
ever, it is expected that the confinement effect is rather small
because the stress field below the neutral axis of a beam is in
tension longitudinally rather than compression. The effect of
transverse reinforcement on the bond strength in beams is also
beyond the scope of this study and the writers have a plan to
study further subsequently for this subject.
Fig. 1 depicts the distributions of steel and concrete stresses,
strains, slips, and bond stresses for both initial crack and stabi-
lized crack stages. The initial tension cracks form when the ten-
sile stress exceeds the tensile strength of the concrete at weak
sections. At the cracks the concrete is free from stress and the
reinforcement carries all the load 关Fig. 1共a兲兴. Between cracks,
tensile stress exists in the concrete because tension is transferred
from the steel to the concrete by bond. The slip occurs between
the steel and the concrete and it is a maximum at the cracks
共Balázs 1993兲.
At the initial crack formation stage, a portion of the element
between the cracks exists over which the steel and concrete
strains are equal, i.e., the slips and bond stresses are not produced
关see Fig. 1共a兲兴. If the crack spacing is greater than twice the
transfer length of bond, lt, another crack can occur between initial
cracks because the concrete stress may reach the tensile strength
of the concrete. The crack development is stabilized if all crack
spacings vary between lt and 2lt 关see Fig. 1共b兲兴. This phase is
Fig. 1. Distribution of steel stress, concrete stress, strains, slips, and referred to as stabilized crack formation stage.
bond stresses between cracks. 共a兲 initial crack; 共b兲 stabilized crack
stages
Crack Width Analysis Using Local Bond Stress-Slip
Relation under Monotonic Loading
bedded bars and enable greater bond forces to be transmitted if
The equilibrium around the reinforcing bar of a direct tension
shearing stresses are transmitted across the splitting cracks by
member is expressed as 共see Balázs 1993, Figs. 2 and 1兲
means of aggregate interlocks 共Park and Paulay 1975兲. It is, how-
ever, also noted that stirrups cannot prevent splitting cracks,
␶bx␾␲dx = df sxAs or ␶bx␾␲dx = df cxAc 共3兲
which always form when large bars are used in beams 共Park and
Paulay 1975兲. This means that the confinement by transverse re-
inforcement is to prevent a failure along a potential splitting crack Fsx + Fcx = Fs2 共4兲
which is perpendicular to the stirrup and to enforce, if necessary,
Here, ␶bx = bond stress; ␾ = nominal reinforcing bar diameter;
a shear failure.
As = section area of reinforcing bar; Ac = section area of concrete;
The present study, however, deals with, not splitting cracks,
f sx = reinforcing bar stress at the position x within the transfer
but the flexural cracking problem in beams where cracks occur
length; f cx = concrete stress; Fsx = tension force of reinforcing bar
parallel to the stirrups. Therefore, the stirrups do not involve di-
at the position x within the transfer length; Fcx = tension force of
rectly in controlling the vertical flexural cracks in beams. The
concrete; and Fs2 = tension force of reinforcing bar at the crack.
writers tried to find any evidence from the literature if stirrups can
The bond stress distribution can be expressed as a function of
contribute to control the flexural cracks and found that the stirrups
bond slip 共Balázs 1993兲
do not appear to improve crack width control, even though they

冉 冊
may help for some ductile type of bond failure 共Park and Paulay ␣
sx
1975兲. As can be seen in Figs. 1 and 2, the stirrup elements are ␶bx = ␶max 共5兲
not involved in the formulation of cracking analysis because they s p1
are parallel to the cracks. This implies that the stirrups do not where ␶max = bond strength; sx = slip; s p1 = slip corresponding to the
affect the flexural crack width directly, but may affect slightly the bond strength; and 0 艋 ␣ 艋 1 being generally 0.4.
bond strength of concrete by providing some confinement, which The compatibility condition at the interfaces of concrete and
steel can be expressed as follows:

sx = usx − ucx or s⬘x = ␧sx − ␧cx 共6兲

where usx = displacement of reinforcing bar at the position x
within the transfer length and ucx = displacement of concrete at the
same location 共see Fig. 1兲.
The differential equation of the slipping contact is obtained by
combining the compatibility 关Eq. 共6兲兴 and equilibrium equations
关Eq. 共3兲兴 as the following equation, in which nE = modular ratio
Fig. 2. Equilibrium condition in the interface between concrete and 共Es / Ec兲; ␳s,eff = effective reinforcement ratio 共As / Ac,eff兲, and ␶bx
reinforcing bar of direct tension member can be obtained from Eq. 共5兲:

412 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MARCH 2007

 Fig. 3. Details of test members and loading method by authors 共adapted from Oh and Kim 2000兲

4共1 + nE␳s,eff兲 spectively. This ␤ value was also introduced in the American
s⬙x − ␶bx = 0 共7兲 Concrete Institute 共ACI兲 crack width equation to calculate the
␾Es
maximum crack width at the tension face of the beam 共ACI
Eq. 共7兲 can be solved by multiplying both sides by 2s⬘, and 2002兲. If ␤ = 1, it represents the crack width of a beam at the
then separating the variables. After two times of integration, the location of tensile reinforcement 关see Eq. 共11兲兴
slip can be obtained as follows:

sx = 冉
2共1 − ␣兲2共1 + nE␳s,eff兲␶max x2
共1 + ␣兲Es s␣p1␾
冊 1/共1−␣兲
共8兲
w=2 冉 s␣p1␾共1 + ␣兲 f s2 2

8共1 + nE␳s,eff兲 ␶maxEs

冊 1/共1+␣兲
共h − c兲
共d − c兲
共12兲

where h = height of flexural member; d = effective depth of rein-

By substituting Eq. 共8兲 into Eq. 共5兲, the bond stress distribution
forcing bars; c = neutral axis depth; and ␤ = 共h − c兲 / 共d − c兲.
can be obtained as expressed in

f sx = f s1 + K 冉 冊
x1+␣
s␣p1␾
1/共1−␣兲
, Comparison of Balázs’s Crack Width Formula with the
Writers’ Test Results

K = ␶max
1+␣
冉
4共1 − ␣兲 2共1 − ␣兲2共1 + nE␳s,eff兲␶max
共1 + ␣兲Es
冊 ␣/共1−␣兲
共9兲
Fig. 3 shows the details of test members for monotonic loading
tests conducted by authors 共Oh and Kim 2000兲. The cross section
of the beam was 220⫻ 400 mm and the span length 4,000 mm.
where f s1 = steel stress over the portion where ␧s = ␧c 关see Fig. The compressive strength of concrete is 26.5 MPa and the yield
1共a兲兴. The inverse of Eq. 共9兲 gives the length lt over which the strength of reinforcing bar is 392 MPa. The reinforcement ratios
steel stress f s2 can be transferred to the concrete 关Fig. 1共a兲兴 are 0.0149 for the test member C9-D25-2 共Fig. 4兲 and 0.0173 for

lt = 冉 共s␣p1␾兲1/共1−␣兲 f s2
1 + nE␳s,eff K
冊 共1−␣兲/共1+␣兲
共10兲
the test member C9-D22-3 共Fig. 5兲, respectively. The crack
widths were monitored by crack gauges attached over the cracks
after reading the initial values of crack widths with microscopes.
The coefficient K has a dimension of stress. The crack width w Figs. 4 and 5 show the comparisons of Eq. 共12兲 with the test data
at the bar surface can be obtained by substituting Eq. 共10兲 into Eq. of the writers. It can be seen that Eq. 共12兲 by Balázs gives rela-
共8兲 with the condition that x = lt and f sx = f s2 at the crack location tively good correlation with the test data of Oh and Kim 共2000兲.

冉 冊
However, ACI 共2002兲 and Euro-International Concrete Commit-
s␣p1␾共1 + ␣兲 f s2 2 1/共1+␣兲
tee 共1990兲 equations show relatively large deviation from test data
w=2 共11兲
8共1 + nE␳s,eff兲 ␶maxEs and provide unconservative values for crack width.
Eq. 共11兲 is a formula for the crack width at the level of the
steel in a direct tension reinforced concrete 共RC兲 member. The
crack width is twice the slip in case of symmetrical behavior. The
flexural RC members have different crack widths at the level of
the steel bars and at the bottom surface of the member due to the
curvature.
Considering the effect of the curvature, the crack width at the
lower extreme tensile fiber can be increased linearly for flexural
members as Eq. 共12兲. This was derived from the fact that the
crack width at the location of neutral axis of a beam is zero and it
starts to increase linearly up to the extreme tension face which is
located at the bottom of the beam. Considering the law of similar
triangle for crack width below neutral axis, the ratio of relative
increase of crack width between the steel location and bottom
tension face is ␤ = 共h − c兲 / 共d − c兲, where h = total depth; Fig. 4. Comparison of various formulas with test data 共C9-D25-2兲
d = effective depth; and c = depth of neutral axis of the beam, re- 共adapted from Oh and Kim 2000兲

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MARCH 2007 / 413

 Local Bond Stress-Slip Relationship under
Repeated Loading

The local bond stress-slip relation is generally changed under

repeated loading as was clarified by Oh and Kim 共2006兲 and Kim
共2004兲. Therefore, a cycle-dependent bond stress-slip law must be
employed to derive a realistic crack width formula under repeated
loading. In this paper, a cycle-dependent bond stress-slip relation
proposed recently by Oh and Kim 共2006兲 was adopted here
to formulate a realistic equation for crack width under repeated
loading.
The local bond stress versus slip relationship for the first load-
Fig. 5. Comparison of various formulas with test data 共C9-D22-3兲 ing may be expressed as Oh and Kim 共2006兲
共adapted from Oh and Kim 2000兲
␶1共s兲 = ␶max 冉 冊
s − sr0
s p1
␣1
, sr0 = 0 共16兲

where s = slip; sr0 = residual slip before first loading; s p1 = slip

Review of Crack Width Analysis under Repeated value at the peak bond stress= 1.04 mm 共Balázs 1991, 1993;
Loading International Federation for Structural Concrete 2000兲, and sub-
script 1 means first loading.
The crack width formulas under repeated loadings have been The local bond stress versus slip relationship for loading after
based on the results of fatigue tests for RC beams. Lovegrove and 共N − 1兲th repeated loading is then written as
El Din 共1982兲 expressed the crack widths under repeated loadings
as a function of cycle number in which the crack width due to first
loading increases with an increases of cycle number as shown in ␶N共s兲 = ␶max 冉 s − sr共N−1兲
s p1
冊 ␣N
共17兲

in which sr共N−1兲 = residual slip after the 共N − 1兲th loading.

wN = w1共3.82 − 0.0227 log N兲log N 共13兲 The variations of slip and residual slip under repeated loadings
where w1 = crack width due to first loading and N = number of in Eq. 共17兲 may be summarized as follows 共see Oh and Kim
applied load cycles. 共2006兲 and Kim 共2004兲:
Balaguru and Shah 共1982兲 reported a crack width formula for s N = s 1N b, b = 0.098 共18兲
RC beams under repeated loading, which is a function of load
cycle N and the strain difference between steel bar and concrete. srN = sr1Nbr, br = b 共19兲
The following equation represents the maximum crack width for
static loading and the subsequent equation gives the maximum ␣N = ␣1N−0.092共␶rep/␶max兲 共20兲
crack widths under repeated loading, in which f ct = tensile strength
of concrete, N = number of load cycles, and c2 / c1 = 1.2 共Balaguru
sr1 = s1兵− 共␶rep/␶max兲2 + 2共␶rep/␶max兲其 共21兲
and Shah 1982兲:
Here, ␶rep / ␶max = relative fatigue bond stress level;

冉冊
Sr1 = residual slip after unloading of the first loading; ␣N = ␣ after
f ctAc h2 h2 h2 h−c n cycles of repeated loads; and s1 = slip at the first loading which
wmax = 2 共␧s − ␧c兲 = c1共␧s − ␧c兲 , =
s w␾ h1 h1 h1 d−c is obtained from Eq. 共17兲.
共14兲

Variation of Displacement and Slip under Repeated

wmax,N
c2共␧s,N − ␧c,N兲 冉冊
h2
h1
Loading

Eq. 共17兲 indicate that the bond stress is developed only when the
冉冊
N
=
wmax h2 slips, s, exceeds the residual slip, sr. Therefore, the net slip, s px,
c1共␧s − ␧c兲
h1 which is defined here as the total slip sx minus residual slip srx,

c2
冉冊
h2
h1
再 冉 冊冎 冉 冊
␧s,N 1 + 1.22
log10 N
7
2

h2 h − cN
involves in solving the differential equation Eq. 共7兲 for slip:
s px = sx − srx 共22兲

冉冊
N
= , =
c1 h2 ␧s h1 N d − cN Here, the residual slip means the value of slip remained after the
h1 applied load is fully unloaded and the residual slip srx represents
the value of residual slip at the position x of the member 共see
共15兲
Fig. 1兲. Fig. 6 shows the residual slip graphically after complete
The crack width equation 关Eq. 共15兲兴 does not consider the unloading, i.e., the remaining slip at which the bond stress is
effects of repeated loadings on the variation of bond properties reduced to zero after unloading. Substitution of Eq. 共22兲 into
between steel bar and concrete. It is, therefore, necessary to con- Eq. 共7兲 with the condition of Eq. 共5兲 gives
sider effectively the variation of local bond stress-slip relation in
4共1 + nE␳s,eff兲 ␶max ␣N
order to propose a more realistic and accurate form of crack width s⬙px − s px = 0 共23兲
formula under repeated loadings. ␾Es s␣p1N

414 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MARCH 2007

Fig. 6. Local bond stress versus slip relationships for the first and
Nth loadings Fig. 8. Variation of transfer length according to constant amplitude
repeated loading

The solution of Eq. 共23兲 takes the form of

Variation of Transfer Length due to Repeated Loading

s px =冉2共1 − ␣N兲2共1 + nE␳s,eff兲␶max x2

共1 + ␣N兲Es s␣p1N␾
冊 1/共1−␣N兲
共24兲
The transfer length which transfers the steel force to concrete is
affected by the residual slip accumulated during repeated loading,
because the zone of residual slip does not transfer the bond
as expressed in Eq. 共8兲. Then, the total slip value, sx, can be stresses as shown in Fig. 6. Therefore, the transfer length, lt共N兲,
written as after 共N − 1兲 repeated loading may be obtained by subtracting the
total residual slip from the transfer length at the first loading as
sx = s px + srx 共25兲 described in

冕
lt共1兲
lt共N兲 = lt共1兲 − sr共N−1兲dx 共27兲
Variation of Bond Properties within Transfer Length 0
under Repeated Loading
Here, lt共N兲 = transfer length after the 共N − 1兲th loading and
Fig. 1 exhibits the bond stress distribution along the transfer lt共1兲 = transfer length of the first loading. The transfer length, lt共N兲,
length. The substitution of Eq. 共8兲 into Eq. 共5兲 gives the following after repeated loading may be called here as “virtual transfer
form for bond stress distribution, in which C is constant and ␶共lt兲 length,” because it moves inward from the explicit crack as the
is the bond stress at the crack: number of repeated loads increases.

␶共x兲 = Cx2␣/共1−␣兲 = ␶共lt兲 冉冊x

lt
2␣/共1−␣兲
共26兲 Variation of Bond Stress at the End of Virtual Transfer
Length due to Repeated Loading
The local bond stress-slip relation under repeated loading fol- The difference of steel stress between x = lt 共crack location兲 and
lows Eqs. 共17兲–共21兲. All the points within the transfer length x = 0 can be written as follows 共see Figs. 1 and 7兲:

冕
would have different bond stress-slip relations because the mag-
lt
nitudes of bond stresses under repeated loading are different at 1 ␾␲
f s2,rep − f s1 = f s2,rep = ␶共x兲dx 共28兲
each location within the transfer length. Therefore, a representa- 1 + nE␳s,eff 0 As
tive value of ␣N within transfer length, which is defined here as
␣*N, must be derived to be used in the calculation of crack widths The substitution of Eq. 共26兲 into Eq. 共28兲 gives the bond stress at
under repeated loading. the end of transfer length, ␶共lt兲, as shown in

Fig. 7. Transfer length and bond stress distribution for the Nth Fig. 9. Variation of representative ␣*N according to constant
loading amplitude repeated loading

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MARCH 2007 / 415

Table 1. Details of Bhuvasorakul’s Test Member and Repeated Load Levels
Test Dimension 共mm兲 Concrete Loaded stress 共MPa兲
member Bar diameter Number cover
number Width Height Span 共mm兲 of bar 共mm兲 Maximum Minimum Range
RDB1 203 263 3,050 16 2 50.8 276 69 207
RDB2 203 372 3,050 16 2 50.8 276 69 207
RDB3 203 130 3,050 16 1 50.8 276 69 207

1 ␾1+␣1 Variation of Crack Width under Repeated Loading

␶共lt兲 = f s2,rep 共29兲
1 + nE␳s,eff 4 1 − ␣ lt
A simple method of calculating the crack width under repeated
It was found from recent research by Oh and Kim 共2006兲 that loads is to use the crack width w1 under monotonic loading and
the bond strength is not affected greatly by the application of then to consider the number N of load cycles as shown in
repeated loads if bond failure does not occur. Similarly, if the
bond stress at the end of the virtual transfer length is kept con- w n = w 1N b 共32兲
stant, i.e., ␶共lt共1兲兲 = ␶共lt共N兲兲 共see Fig. 7兲, the relationship of ␣1 and However, these methods do not consider the variation of bond
␣*N can be derived as depicted in properties within the transfer length under repeated loading.

冉 冊冒冉 冊
Therefore, the effects of repeated loads on the bond slip behavior
1 + ␣1 lt共N兲 1 + ␣1 lt共N兲 must be considered realistically to obtain more reliable crack
␣*N = −1 +1 共30兲 width values. The crack width is, by definition, just twice the slip
1 − ␣1 lt共1兲 1 − ␣1 lt共1兲
value at the crack location as indicated in
The virtual transfer length after the 共N − 1兲th loading can then
wN = 2s共lt共N兲兲 共33兲
be derived by substituting Eqs. 共19兲, 共21兲, and 共26兲 into Eq. 共27兲
as follows: In order to calculate the slip s共lt共N兲兲, first consider net slip,
s p共lt共N兲兲, from Eq. 共17兲 as follows. The slip s p共lt共N兲兲 can be written
冕
lt共1兲
lt共N兲 = lt共1兲 − sr共N−1兲dx = lt共1兲 from Eq. 共17兲 as

再 冉 冊 冉 冊
0 *

冉 冊
␶共lt共N兲兲 1/␣N
␶共lt共1兲兲 1/␣1
共1 − ␣1兲 ␶共lt共1兲兲 s p共lt共N兲兲 = s p1 共34兲
− s p1共N − 1兲br ␶max
␶max 共3 + ␣1兲 ␶max

冉 冊冎
For each number of repeated load, the slip at the crack location
共1 − ␣1兲 ␶共lt共1兲兲
2
is added step-by-step. For the step between kth loading and Nth
− lt共1兲 共31兲
3共1 + ␣1兲 ␶max loading, the virtual transfer length, lt共N兲, which is needed in Eq.
共34兲 can be written from Eq. 共31兲 as shown in
Figs. 8 and 9 show the variations of virtual transfer length, lt共N兲

冕
lt共k兲
关Eq. 共31兲兴, and the power ␣*N 关Eq. 共30兲兴 according to the number
lt共N兲 = lt共k兲 − sr共N−1兲dx = lt共k兲
of repeated loading cycles. It can be seen from these figures that 0

再 冉 冊
the transfer length and the value of ␣*N decrease linearly in loga-

冉 冊冉 冊
*
rithmic scale with an increase of the number of load cycles. These N−1 br ␶共lt共k兲兲 1/␣k
共1 − ␣*k 兲 ␶共lt共k兲兲
characteristics of transfer length and ␣*N should be considered − s p1
␶max
k 共3 + ␣*k 兲 ␶max

冉 冊冎
realistically in the next section in calculating the crack widths of
reinforced concrete beams under repeated loadings. 共1 − ␣*k 兲 ␶共lt共k兲兲 2
− lt共k兲 共35兲
3共1 + ␣*k 兲 ␶max

Fig. 10. Comparison of analytical crack widths with test results from Fig. 11. Comparison of analytical crack widths with test results from
RDB1 in Bhuvasorakul 共1974兲 RDB2 in Bhuvasorakul 共1974兲

416 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MARCH 2007

Fig. 12. Comparison of analytical crack widths with test results from Fig. 13. Comparison of analytical crack widths with test results of
RDB3 in Bhuvasorakul Kormeling et al. 共1978兲

The value of ␣*N in Eq. 共34兲 can also be described from Eq. 共30兲 The total residual slip after Nth loading is then written as
as shown in sr共lt共N兲兲 = sr共lt共1兲兲共1兲 + ⌬sr共lt共1兲兲共1គ2兲 + ⌬sr共lt共2兲兲共2គ3兲 + ¯

␣*N = 冉 1 + ␣*k lt共N兲

−1 冊冒冉 1 + ␣*k lt共N兲
+1 冊 共36兲
+ ⌬sr共lt共N−1兲兲共N−1គN兲 + ⌬sr共lt共N兲兲共NគN+1兲

冉冉 冊 冊
1 − ␣*k lt共k兲 1 − ␣*k lt共k兲 br
2
The slip, s共lt共N兲兲, in Eq. 共33兲 is then obtained by summing all the = sr共lt共1兲兲 + sr共lt共1兲兲 −1
1

冉冉 冊 冊 冉冉 冊 冊
incremental slips as follows: br br
3 N+1
+ sr共lt共2兲兲 − 1 + ¯ + sr共lt共N兲兲 −1
s共lt共N兲兲 = s p共lt共1兲兲共0−1兲 + ⌬s p共lt共1兲兲共1គ2兲 + ⌬s p共lt共2兲兲共2គ3兲 + ¯ 2 N
+ ⌬s p共lt共N−1兲兲共N−1គN兲 共40兲

冉冉 冊 冊 冉冉 冊 冊
b b Similarly, the residual crack width may be written as in the fol-
2 3
= s p共lt共1兲兲 + s p共lt共1兲兲 − 1 + s p共lt共2兲兲 −1 lowing equation for a flexural member:

冉 冉冉 冊 冊
1 2

冉冉 冊 冊
br
b
h−c 2
N wrN = 2sr共lt共N兲兲 = 2 sr共lt共1兲兲 + sr共lt共1兲兲 −1
+ ¯ + s p共lt共N−1兲兲 −1 共37兲 d−c 1

冉冉 冊 冊
N−1
br
3
The crack width of a reinforced concrete beam is then derived + sr共lt共2兲兲 −1
2
as follows:

wN = 2s共lt共N兲兲
h−c
d−c
冉
= 2 s p共lt共1兲兲 + s p共lt共1兲兲 冉冉 冊 冊
2
1
b
−1
+ ¯ + sr共lt共N兲兲 冉冉 冊 冊冊冉 冊N+1
N
br
−1
h−c
d−c
共41兲

+ s p共lt共2兲兲 冉冉 冊 冊
3
2
b
−1
Procedure of Crack Width Analysis under Repeated

+ ¯ + s p共lt共N−1兲兲 冉冉 冊 冊冊冉 冊 N
N−1
b
−1
h−c
d−c
共38兲
Loading

The procedure of crack width analysis under repeated loading is

summarized as follows.
1 Find the stress of the steel at the crack f s2,rep and neutral axis
Variation of Residual Crack Width under Repeated c by section analysis;
Loading 2. Calculate the transfer length lt共1兲 by Eq. 共10兲 for the stress of
the steel at the crack;
The residual slip, sr共k兲, may be rewritten from Eq. 共21兲 as follows 3. Find the crack width under first loading w1 using Eq. 共11兲 or
for kth loading step: Eq. 共12兲;
4. For each number of repeated load k, calculate the virtual
sr共k兲 = s p共k兲兵2共␶rep/␶max兲 − 共␶rep/␶max兲2其 共39兲 transfer length lt共k兲 using Eq. 共35兲;

Table 2. Details of Kormeling and Reinhardt Test Members and Repeated Load Levels
Test Dimension 共mm兲 Concrete Loaded stress 共MPa兲
member Bar diameter Number cover
number Width Height Span 共mm兲 of bar 共mm兲 Maximum Minimum Range
Beam1 101.6 243.8 2,032 19 2 101.6 84.3 15 69.4

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MARCH 2007 / 417

Table 3. Details of Jagdish and Lal’s Test Members and Repeated Load Levels
Test Dimension 共mm兲 Concrete Loaded stress 共MPa兲
member Bar diameter Number cover
number Width Height Span 共mm兲 of bar 共mm兲 Maximum Minimum Range
Beam1 152.4 254 1,820 29 1 69.5 192.6 69 96

5. For each number of repeated load k, calculate the power 共ROB3兲 of Bhuvasorakul’s tests. The deviation of constant crack
index ␣*K using Eq. 共36兲; and width method from test data becomes larger as the number of
6. Calculate the crack widths for the Nth loading using Eq. 共38兲 repeated loads increases, whereas the proposed method correlates
and the residual crack widths after the Nth loading using reasonably well with test data up to large number of load cycles.
Eq. 共41兲. This is important because the number of repeated load increases
continuously at the service load stage during the life time of struc-
tures. Therefore, correct prediction is important to assure the ser-
Application of Proposed Method for Crack Width viceability of reinforced concrete structures under repeated loads
Prediction such as bridge structures.
Table 2 shows the dimensions and properties of the test mem-
The analysis results by proposed method were compared with test ber for the tests by Kormeling et al. 共1978兲 Fig. 13 shows the
results from other researchers. Table 1 presents the properties of comparison of crack width between theories and test data. It can
the test members by Bhuvasorakul’s study 共1974兲. Fig. 10 shows be seen from this figure that the magnitude of crack width is
the comparison of the proposed method with the test data 共Beam much lower than the previous cases of Bhuvasorakul’s tests
ROB1兲 of Bhuvasorakul’s tests 共1974兲. Fig. 10 indicates that the 共1974兲 because the level of fatigue stress is much lower for this
crack width increases with the increase of the number of load test beam as noted in Table 2. It is again seen that the proposed
cycles and it becomes almost more than twice at the 1 ⫻ 105 load method gives good correlation with test data, whereas the con-
cycle. Fig. 10 indicates that the constant w1 method of Eq. 共33兲 stant w1 method gives large discrepancy as the number of load
共dashed line兲 shows large deviation from test data, while the pro- cycles increases.
posed method 共solid line兲 correlates fairly well with test data. The Table 3 shows the dimensions and properties of the test mem-
reason of large deviation lies in the fact that the constant w1 ber for the tests by Lal 共1970兲. Fig. 14 depicts that the crack width
method does not consider the variation of bond properties within increases with the increase of the number of load cycles. The
the transfer length under repeated loading and considers only the constant crack width method again shows large deviation from
effect of repeated load number N under constant crack width as test data as the number of load cycles increases. The comparisons
shown in Eq. 共33兲. Therefore, the effects of repeated loads on the of proposed theory with test data indicate that the appropriate
bond slip behavior must be considered realistically to obtain more bond stress slip behavior under repeated loads must be incorpo-
reliable crack width values in reinforced concrete beams under rated in order to predict the crack width realistically for rein-
repeated loads. forced concrete beams subject to repeated fatigue loads.
Fig. 11 shows again the comparison of the proposed method
with another test data 共Beam ROB2兲 of Bhuvasorakul’s tests Conclusions
共1974兲. The similar results are also obtained for this case of same
investigators. Here, it can be also seen that realistic results cannot Some important structures such as bridges and highway structures
be obtained without considering the variation of bond properties are subjected to repetitive fatigue loadings. Under these repeated
under repeated loading. Fig. 12 also compares the variation of loadings, the crack width and deformation increase with the in-
crack width under repeated loading for the different beam crease of load cycles. It is, therefore, necessary to accurately es-
timate the crack width increase under repeated loadings in order
to meet the serviceability requirements. The bond stress-slip rela-
tion is directly involved to evaluate the crack width. Therefore, an
appropriate bond stress-slip relation under repeated loading has
been incorporated realistically to derive a method for crack width
analysis in reinforced concrete members.
The concept of transfer length, which is the length of load
transfer from steel to concrete, was introduced to calculate the
bond stress distribution around the steel bar in the present study.
The present study indicates that the transfer length decreases as
the number of load cycles increases. This is due to the accumu-
lation of residual slip according to the repetition of load cycles.
The formulation for crack width analysis involves with the
determination of transfer length, bond stress at crack location, and
bond stress-slip relations under repeated loads. The proposed
method considers all those parameters to obtain accurate and re-
alistic results for crack width. The comparison of proposed theory
Fig. 14. Comparison of analytical crack widths with test results of agrees well with the measured data for crack widths under re-
Lal 共1970兲 peated loadings while the constant crack width method shows

418 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MARCH 2007

large deviation from test data especially as the number of load sr0 , sr1 , srN ⫽ residual slip after unloading of 0th, first, Nth
cycles increases. The proposed theory also enables to calculate repeated loads, respectively, sr0 = 0;
the residual crack widths after unloading of repeated loads. The sw ⫽ crack spacing;
present study provides more realistic and reliable method for sx ⫽ slip at the position x;
evaluation of crack width under repeated loadings. s⬘x ⫽ derivative of slip at the position x;
wmax,N ⫽ maximum crack width after the Nth loading;
wN ⫽ crack width for the Nth loading;
wrN ⫽ residual crack width after the Nth loading;
Acknowledgments w1 ⫽ crack width due to first loading;
␣ , ␣1 , ␣N , ␣*k , ␣*N
The financial support from the National Research Laboratory ⫽ powers in bond stress-slip relations,
共NRL兲 Program and the Development of Manmade Disaster Pre- respectively;
vention 共DMDPT兲 Program of Korea is gratefully acknowledged. ␧cx ⫽ strain of concrete at the position x within the
transfer length;
Notation ␧sx ⫽ strain of reinforcing bar at the position x
within the transfer length;
The following symbols are used in this paper: ␳s,eff ⫽ effective reinforcement ratio 共=As / Ac,eff兲;
␶max ⫽ bond strength;
Ac,eff ⫽ effective cross section area of concrete;
␶N ⫽ bond stress at the Nth loading;
As , Ac ⫽ cross section area of steel and concrete,
␶rep ⫽ repeated bond stress level;
respectively;
␶s2,rep ⫽ repeated bond stress at the crack;
b ⫽ the power of the cyclic number N in the
␶共lt兲 ⫽ bond stress at the crack;
relation of slip and cyclic number;
␶共lt共1兲兲 , ␶共lt共N兲兲⫽ bond stress at the crack for the first and Nth
br ⫽ the power of the cyclic number N in the
relation of residual slip and cyclic number; loading respectively;
C ⫽ constant; ␶共x兲 ⫽ bond stress at the position x; and
c ⫽ kd = neutral axis depth; ␾ ⫽ reinforcing bar diameter.
d ⫽ effective depth of reinforcing bar in a flexural
member; References
Ec , Es ⫽ modulus of the elasticity of concrete and
steel, respectively; American Concrete Institute 共ACI兲. 共2002兲. “Control of cracking in con-
Fcx , Fsx ⫽ tension force of concrete and steel bar at the crete structures.” ACI manual of concrete practice, ACI 224, Detroit.
position x within the transfer length, Balaguru, P., and Shah, S. P. 共1982兲. “A method of predicting crack
respectively; widths and deflections for fatigue loading.” Fatigue of concrete struc-
Fs2 ⫽ tension force of reinforcing bar at the crack; tures, ACI sp-75, ACI, Detroit, 153–175.
f ⬘c ⫽ compressive strength of concrete at the age of Balázs, G. L. 共1987兲. “Bond model with non-linear bond-slip law.” Studie
28 days; ricerche, Post-Graduate School on Design of Reinforced and Pre-
f ct ⫽ tensile strength of concrete; stressed Concrete Structures, Politecnico di Milano, Italy, Vol. 9,
f cx ⫽ concrete stress at the position x within the 157–180.
transfer length; Balázs, G. L. 共1991兲. “Fatigue of bond.” ACI Mater. J., 88共6兲, 620–629.
f r ⫽ bond index or relative rib area of steel bar; Balázs, G. L. 共1993兲. “Cracking analysis based on slip and bond
f s ⫽ stress of reinforcing bar; stresses.” ACI Mater. J., 90共4兲, 340–348.
f s1 ⫽ reinforcing bar stress over the portion where Bhuvasorakul, T. 共1974兲. “Performance of members reinforced with
␧ s = ␧ c; smooth welded wire fabric under static and repeated loads.” Ph.D.
thesis, Oklahoma State Univ., Okla.
f s2 ⫽ reinforcing bar stress at the crack;
Broms, B. B. 共1965兲. “Crack width and crack spacing in reinforced con-
f s2,rep ⫽ repeated tensile stress at the crack;
crete members.” ACI J., 62共10兲, 1237–1256.
f sx ⫽ reinforcing bar stress at the position x within
Broms, B. B., and Lutz, L. A. 共1965兲. “Effects of arrangement of rein-
the transfer length; forcement on crack width and spacing of reinforced concrete mem-
h ⫽ height of flexural member; bers.” ACI J., 62共11兲, 1395–1410.
lt ⫽ transfer length; Campi, V., Eligehausen, R., Bertero, V., and Popov, E. 共1982兲. “Analyti-
lt共1兲 , lt共k兲 , lt共N兲 ⫽ transfer length for the first, kth, Nth loading, cal model for deformed bar bond under generalized excitations.” Rep.
respectively; No. UCB/EERC-82/23, Univ. of Berkeley, Berkeley, Calif.
N ⫽ the number of repeated loading; Euro-International Concrete Committee. 共1990兲. Model code for concrete
ne ⫽ modular ratio 共=Es / Ec兲; structures, EPFL, Lausanne, Switzerland.
s共lt共N兲兲 ⫽ slip at the crack for the Nth loading; International Federation for Structural Concrete. 共2000兲. “Bond of
sN ⫽ slip of the Nth repeated loading; reinforcement in concrete.” Bulletin No. 10, EPFL, Lausanne,
s p1 ⫽ slip corresponding to bond strength; Switzerland.
s p共lt共N兲兲 ⫽ additional slip at the crack for the Nth Kim, S. H. 共2004兲. “Evaluation of crack width and remaining service life
for RC flexural members under repeated loading.” Ph.D. thesis, Seoul
loading;
National Univ., Seoul, Korea, 40–75.
s px ⫽ additional slip at the position x; Kormeling, H. A., Reinhardt, H. W., and Shah, S. P. 共1978兲. “Static and
sr ⫽ residual slip; dynamic testing of concrete beams reinforced with steel fibers and
sr共lt共N兲兲 ⫽ residual slip at the crack for the Nth loading; continuous bars.” Stevin Rep. No. NR 5-78-10, Delft Univ. of Tech-
srx ⫽ residual slip at the position x; nology, Delft, The Netherlands.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MARCH 2007 / 419

Lal, J. 共1970兲. “Dynamic response of reinforced concrete flexural mem- reinforced concrete beams.” KCI J., Proc., 12共2兲,105–108.
bers.” MS thesis, Ohio State Univ., Ohio. Oh, B. H., and Kim, S. H. 共2006兲. “Realistic models for local bond
Lovegrove, J. M., and El Din, S. 共1982兲. “Deflection and cracking of stress-slip of reinforced concrete under repeated loading.” J. Struct.
reinforced concrete under repeated loading and fatigue.” Fatigue of Eng., 133共2兲, 216–224.
concrete structures, ACI sp-75, ACI, Detroit, 133–152. Park, R., and Paulay, T. 共1975兲. Reinforced concrete structures, Wiley,
Oh, B. H., and Kim, S. H. 共2000兲. “Crack width and crack spacing of New York, 392–494.

420 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MARCH 2007