You are on page 1of 9


Title no. 107-S27

Crack Width Estimation for Concrete Plates

by H. Marzouk, M. Hossin, and A. Hussein

This research is focused on evaluating the crack widths and crack splash zone as well as exposure to seawater and sea spray.
properties of thick two-way slabs and plates used for offshore and The design of offshore and nuclear containment structures is
nuclear containment structures. The crack width depends on the controlled by mandatory design codes to ensure structural
quantities, orientation, and distribution of reinforcing steel across safety and integrity. The main objective of this study is to
the crack and characteristics of the bond between the concrete and
investigate the cracking criteria for concrete two-way plates
reinforcement bars in and near the crack. The maximum crack
width that is considered acceptable depends on the type of structure, and develop a rational numerical model to predict the
location within the structure, environment, and consequences of crack width of concrete plates under flexural sustained
excessive cracking. service loads.
A comprehensive experimental and analytical investigation is
presented in this work. The numerical investigation will focus on PREVIOUS RESEARCH
the available code prediction models for estimating the crack AND EXISTING FORMULAS
width of concrete plates. The investigation will focus on the The width of the crack depends on the quantity, orientation,
suitability of available crack width expressions for thick concrete and distribution of reinforcing steel crossing the crack. It
plates used for offshore concrete structure applications and also depends on the deformation characteristic of the concrete
nuclear containment structures. and the bond between the concrete and reinforcing bars.
The experimental work included the investigation of the cracking An extensive statistical research analysis was reported by
behavior, such as examining the effect of increasing concrete cover Geregely and Lutz (1968) of the maximum crack width data
and bar spacing on crack width properties. The crack widths were from a number of sources. The recommended equations are
measured electronically for three series of specimens. The test
results were tabulated to compare test results with the available
considered to best predict the maximum bottom and side
code expressions for calculating crack widths. The tension chord crack width as follows
method was modified to predict the crack width for two-way plates
under flexural loading. –3
w b = 0.091 3 t b A β ( f s – 5 )10 (in.) (1)
Keywords: crack width; high-strength concrete; normal-strength concrete;
offshore concrete walls; reinforced concrete; two-way slabs.
0.091 3 t b A –3
- ( f s – 5 )10 (in.)
w s = ------------------------- (2)
1 + -----
Flexural members, such as beams and slabs, represent the h1
majority of structural elements. In general, concrete can
handle compressive forces very effectively. However, where wb and ws are the most probable crack widths at the
concrete cracks under tension forces. Cracks play an bottom of the beam and level of reinforcement, respectively
important role in concrete’s response to load in both (in.); fs is the reinforcing steel stress (ksi); A is the effective
compression and tension. Moreover, there are extra internal embedment concrete area symmetric with reinforcing steel
forces and stresses that will be developed due to temperatures and divided by the number of bars; tb is the bottom cover to the
shrinkage. If such cracks are too wide, it will destroy the center of the bars (in.); ts is the side cover to the center of the
aesthetics of the structure and cause a significant reduction bars (in.); β is the ratio of the distance between the neutral
in the flexure stiffness of the structural member. axis and the tension face to the distance between the
Cracks may expose bars to the environment, causing neutral axis and the reinforcing steel; and h1 is the distance
corrosion of steel. To protect the structural members from from the neutral axis to the reinforcing steel (in.).
these effects, cracks should be minimized to acceptable
limits under normal service loads. Adequate design and ACI 318 Code
satisfactory reinforcement details are required to limit the A recent study by Frosch (1999) showed that Eq. (1) and
crack width in a concrete offshore structure. Many methods (2) are valid for a relatively narrow range of covers up
have been suggested to control cracking in a marine structure. to 63 mm (2.4 in.). Hence, Frosch’s expression provides a
Most of them, however, are merely empirical rules resulting new equation based on the physical phenomenon for
from observations or testing. Furthermore, there is no determination of the flexural crack widths of reinforcing
agreement on the crack width that should be permitted for concrete members. Section 10.6 in ACI 318-05 (ACI
different types of structures. Hence, the accurate prediction
of crack width is not possible yet.
ACI Structural Journal, V. 107, No. 3, May-June 2010.
RESEARCH SIGNIFICANCE MS No. S-2008-182.R3 received May 22, 2009, and reviewed under Institute publication
policies. Copyright © 2010, American Concrete Institute. All rights reserved, including the
Concrete offshore structures are exposed to harsh making of copies unless permission is obtained from the copyright proprietors. Pertinent
discussion including author’s closure, if any, will be published in the March-April 2011
environmental conditions, including their existence in the ACI Structural Journal if the discussion is received by November 1, 2010.

282 ACI Structural Journal/May-June 2010

width is taken as 60 to 70% more than the average crack
ACI member H. Marzouk is the Chair of the Civil Engineering Department at
Ryerson University, Toronto, ON, Canada. He received his MSc and PhD from width. NS 3473 E provides more detailed regulations for
the University of Saskatchewan, SK, Canada. He is a member of ACI Committees crack width limitations depending on the environmental
209, Creep and Shrinkage in Concrete, and 213, Lightweight Aggregate and Concrete. conditions. Four environment classes are identified;
His research interests include structural and material properties of high-strength
concrete, lightweight high strength, creep, and finite element analysis. namely, aggressive, severely aggressive, moderately
aggressive, and mildly aggressive.
M. Hossin is a Consulting Engineer with MMP Engineering, Calgary, AB, Canada.
He received his MEng in structural engineering from Memorial University of
Newfoundland, St. John’s, NF, Canada, in 2008. He received his BSc from CSA offshore code
Garyounis University, Benghazi, Libya, in 1999. His research interests include The Canadian offshore code (CSA-S474-04 [Canadian
cracking analysis and crack width estimation for offshore structures, and behavior of
two-way slabs.
Standards Organization 2004]) recommends that the average
crack width may be calculated as the average crack spacing
ACI Member A. Hussein is an Associate Professor of civil engineering at Memorial (Eq. (7)) times the total average tensile concrete strain after
University of Newfoundland, St. John’s, NF, Canada. He received his BSc from
Ain-Shams University in 1984, and his MEng and PhD from Memorial University in
considering the contribution of the tension stiffening. Both
1990 and 1998, respectively. His research interests include the use of fiber-reinforced NS 3473 E and CSA-S474-04 provide similar expressions
polymer in concrete, mechanical and structural behavior of high-strength concrete, for calculating crack spacing. CSA-S474-04 estimates the
and nonlinear finite element analysis of concrete structures.
crack width at the surface of the member. However, NS 3473 E
calculates the crack width at the level of steel reinforcement. The
Committee 318 2005) does not make any distinction maximum permissible crack width for offshore structures in
between exterior and interior exposure and it requires that CSA-S474-04 is in the range of 0.25 mm (9.8 × 10–3 in.) in the
for crack control in beams and one way slabs, the spacing of splash zone and up to 0.5 mm (19.6 × 10–3 in.) elsewhere.
reinforcement shall not exceed Usually, Canadian offshore structures exist in a severely
aggressive environment, which limits crack width to a range
S (mm) = [95,000/540fs) – 2.5Cc] (3) of 0.20 to 0.10 mm (7.8 × 10–3 to 3.9 × 10–3 in.).
CSA-S474-04 provides the following expression for
But not to exceed 300(252/fs) mm (4) calculating the crack spacing

Srm = 2.0(C + 0.1S) + k1k2dbehef b/As (7)

where fs is the reinforcing steel stress at service load (MPa);
Cc is the clear cover at tension side (mm); and S is the center-
to-center spacing of flexural tension reinforcement (mm). where Srm is the average crack spacing (mm); C is the
The maximum crack width considered that might not concrete cover (mm); S is the bar spacing of the outer layer
impair the appearance of a structure and will not endanger (mm); k1 is the coefficient that characterizes bond properties
steel reinforcement by corrosion is presented in ACI 224R-01 of bars; k2 is the coefficient to account for strain gradient; dbe
(ACI Committee 224 2001). The guide recommends that is the bar diameter of the outer layer (mm); hef is the effective
crack widths in reinforced concrete under service loads can embedment thickness as the greater of (c + dbe) + 7.5dbe not
vary between 0.41 mm (16.1 × 10–3 in.) for dry air to 0.10 mm greater than the tension zone or half slab thickness (mm), as shown
(3.9 × 10–3 in.) for water-retaining structures. These values in Fig. 1; b is the width of the section (mm); As is the area of
depend on the environment surrounding the structure and reinforcement within the effective embedment thickness (mm2);
various factors, such as the position, length and the surface and ε is the concrete tensile strain in the effective embedment
texture of the crack as well as the illumination in the zone hef.
surrounding area. The commentary on CSA-S474-04 recommends the reduction
of the crack width due to the contribution of tension stiffening
Norwegian code should be included. An expression that was originally
The contribution of concrete in tension between cracks in suggested by Vecchio and Collins (1986) can be used to estimate
European codes was taken as a reduction factor of the total the effect of tension stiffening on the concrete.
concrete strain. The Norwegian Standard NS 3473 E
(Norwegian Council for Building Standardization 1992) f ct = f cr ⁄ ( 1 + 500ε ) MPa (8)
provides the following equation for calculating the crack
width. It uses factor r (Eq. (6)) to account for the tension
stiffening effect.

wk = 1.7wm (5)

wm = rε1Srm (6)

where Srm is the average crack spacing (mm); ε1 is the

average concrete tensile strain in the effective embedment
zone; wk is the maximum characteristic crack width (mm);
wm is the average crack width at the concrete surface (mm); Fig. 1—Effective embedment thickness as defined in Eq. (7)
and r is the tension stiffening modification factor, percentage (CSA 04), where a1 and a2 are distance from centers of bars to
of 1.0, dimensionless coefficient. surface of concrete, mm; dbe is equivalent bar diameter of inner
NS 3473 E and other European codes define the characteristic layer of bars, in mm; dbe′ is equivalent bar diameter of outer layer
crack as the width that only 5% of the cracks will exceed and Srm of bars, in mm; s is center-to-center spacing of outer layer of bars,
is defined as the average crack spacing. This characteristic crack in mm; and c is concrete cover, in mm. (Note: 1 in. = 25.4 mm.)

ACI Structural Journal/May-June 2010 283

f cr = 0.33λ f c′ MPa (9) CEB-FIP (1990) code
The CEB-FIP code (CEB-90) (CEB-FIP 1990) gives the
following equation for calculating the characteristic
where fcr is the modulus of rupture of concrete (MPa); fct is crack width
the tensile strength of concrete (MPa); fc′ is the peak
compressive stress obtained from a cylinder test (MPa); and
w k = l s, max ( ε s2 – βε sr2 – ε cs ) ( mm ) (10)
ε is the concrete tensile strain.
However, that expression ignores the contribution of the
steel reinforcement ratio, depth of the concrete cover, and the where wk is the characteristic maximum crack width (mm);
thickness of the concrete member on the tension stiffening. A wm is the average crack width (mm); εs2 is the steel strain of
tension-stiffening model based on fracture energy and tension the transformed section in which the concrete in tension is
properties of high-strength concrete was developed by ignored; εcs is the free shrinkage of concrete, generally a
Marzouk and Chen (1993). The model can be used to account negative value; εsr2 is the steel strain at a crack, under a force
for the concrete mixture design properties and the steel causing stress equal to fctm within Acef; and β is an empirical
reinforcement contribution through two sets of constants. factor to assess average strain within ls,max.
To account for tension stiffening in the CEB-90, an empirical
shape factor β is used to assess the average strain.

Eurocode 2 (BS EN 1992-1-1:2004) (British

Standards Institution 2004)
The characteristic crack width is estimated as

wk = βSrmξεsm (mm) (11)

where wk is the characteristic crack width; Srm is the average

stabilized crack spacing; ξ is a dimensionless coefficient
between 0 and 1, representing the effect of the participation
of concrete in the tension zone to stiffness of the member;
εsm is the mean strain under relevant combination of loads
and allowing for effects, such as tension stiffening or
Fig. 2—Typical details of test specimen. (Note: 1 in. = 25.4 mm.) shrinkage; and β is the coefficient relating the average crack
width to the design value and equal to 1.7 and 1.3, respectively,
for a section where the minimum dimensions exceed 800 mm
(31.5 in.) or smaller than 300 mm (11.8 in.).
For tension stiffening, BS EN 1992-1-1:2004 uses the
factor ξ, which is a dimensionless coefficient. Moreover,
BS EN 1992-1-1:2004 limits the maximum crack width to
0.30 mm (11.8 × 10–3 in.) for sustained load under normal
environmental conditions, which will not impair the proper
functioning of the structure. During the past 30 years, many
researchers have proposed other analytical models, using the
fracture energy approach to predict the cracking response
and tension stiffening effect in reinforced concrete members.
The tension stiffening values can be recommended based on
the actual measured tension softening values and appropriate
experimentally fitted constants.

Fig. 3—Arrangements of steel reinforcement. EXPERIMENTAL INVESTIGATION

The experimental investigation includes the testing of eight
reinforced concrete slabs in the structural lab at Memorial
University. Eight full-scale, quarter-panel, two-way slab
specimens were cast, instrumented, and tested in the current
research program. The tested slabs were square with a side
dimension of 1900 mm (75 in.) in both directions and were simply
supported along all four edges with the corners free to lift. The test
slab represents the region of negative bending moment around an
interior column in a flat slab system and the simply supported
edges simulate the lines of contra-flexure. A central load was
applied on the slab through a 250 x 250 mm (10 x 10 in.) column
stub. The dimensions and reinforcement details of a typical test
slab are shown in Fig. 2. The steel reinforcement’s arrangements
and the location of steel strain gauges and linear potential
Fig. 4—Location of steel strain gauges and LPDTs. (Note: differential transducers (LPDTs) are shown in Fig. 3 and 4.
1 in. = 25.4 mm.) The maximum nominal size for coarse aggregates was 19 mm

284 ACI Structural Journal/May-June 2010

(0.75 in). The specimens were made with normal- and high- made of three slabs designated as Specimens NSC1, HSC1,
strength concrete of 35 MPa (5000 psi) and 70 MPa and HSC2. All the slabs had the same depth of 200 mm
(10,000 psi), respectively. The specimens were cured under (8 in.), same bar spacing of 150 mm (6 in.), and bar diameter
moisture saturation of plastic sheeting to cover the surface of 25 mm (1 in.) with different concrete covers that were 30 to
for more than 3 weeks until the preparation of instrumentation 60 mm (1.2 to 3.4 in.). Series II—Specimens HSC3 and
and during the fourth week of curing, the specimens were HSC4—was designed to investigate the effect of bar spacing
prepared for testing of panels at an age of 28 days. The
specimens were tested in a specially designed frame in the
vertical position, as shown in Fig. 5. Concrete strain gauges
were glued to the compression concrete face adjacent to the
column stub, as shown in Fig. 6. Transducers and LPDTs
were mounted on the tension concrete face crack, as shown
in Fig. 5. The specimens were loaded initially at 5 to 8% of
the expected ultimate load as soon the crack initiated under
load; the crack gauges were then installed. The specimens
were tested to the failure load, as shown in Fig. 7. A close-
up of the crack displacement transducer to measure the crack
width is shown in Fig. 8.
The punching failure of high-strength concrete slabs can
be classified into two modes, “flexure punching” and
“punching shear” failure. Flexural punching occurred for
slabs with relatively low reinforcement ratio and smaller
thickness. As the steel reinforcement ratio and slab thickness
are increased, slab stiffness increases and deformation
capacity decreases and the slab mode of failure becomes
pure punching shear failure (Marzouk and Hussein 1991).
The test results are divided into three series. The first Fig. 6—Locations of concrete strain gauges. (Note: 1 mm =
series was designed to investigate the effect of concrete 0.0394 in.)
cover on the crack width and crack spacing. The series was

Fig. 7—Typical load deflection for normal- and high-strength

concrete. (Note: 1 mm = 0.0394 in.; 1 kN = 0.2245 kip.)

Fig. 5—Typical specimen (NSC1) during testing.

Table 1—Group specimens’ details

Bar Bar Concrete Slab Steel
Series Slab fc′, size, spacing, cover, thickness, ratio
no. no. MPa mm mm mm mm ρ, %
NSC1 35 25M 150 30 200 2.17
Series I HSC1 69 25M 150 50 200 2.48
HSC2 70 25M 150 60 200 2.68
HSC3 67 25M 200 30 200 1.67
Series II
HSC4 61 25M 250 30 200 1.13
HSC5 70 15M 100 30 150 1.88
Series III NSC2 33 15M 240 30 200 0.52
NSC3 34 10M 210 40 150 0.40
Note: 1 mm = 0.0394 in.; 1 MPa = 145 psi. Fig. 8—Electric crack displacement transducer.

ACI Structural Journal/May-June 2010 285

on the crack width and crack spacing. Specimen NSC1 from used, although some of the original information may also be
Series I was considered for the comparison as a part of this acquired in the form of an analog electrical signal. All
series. The slabs of this series had the same concrete depth of signals were then converted to digital for further processing
200 mm (8 in.), the same concrete cover of 30 mm (1.2 in.), and display. A digital electrical signal has the form of a
the same bar diameter of 25 mm (1 in.), and different bar group of discrete and discontinuous pulses. The digital
spacing. The first and second series were designed to represent signal is either displayed on a digital display device or made
heavily reinforced concrete walls that normally fail under the available for transmission to other digital instruments, such
punching failure mode as is the case for most offshore as a computer for further processing and display. All
structures. The third series, however, was designed to inves- measurements were stored in a computer file and computer
tigate the effect of pure flexure failure and ductile shear failure software (National Instruments Corporation 2006) was used
as recommended by Marzouk and Hussein (1991). The third as a data acquisition system and the data scanning and saving
series included Specimens HSC5, NSC2, and NSC3. A rate was set to record the readings every 3 seconds.
summary of the tested specimens is shown in Table 1. This research is focused on instantaneous loading, the long-
term deformation effects of creep and shrinkage are not
Crack measurements considered in this research work. The high relative humidity
Each slab was carefully inspected at each load step. The in the lab and short duration indicated the shrinkage values are
cracks were marked and the maximum visible crack width insignificant for the case of instantaneous loading. The rela-
was measured using a crack width measuring gauge. The tive humidity of the structural lab at Memorial University in
crack displacement transducer (CDT) shown in Fig. 8 is St. John’s is over 87% year round. The correction factor for
mounted to concrete surface cracks and joints to measure the shrinkage for volume to surface area of the tested specimen
opening displacement. It is a waterproof-enabled gauge. The was approximately 0.88. The estimated shrinkage according
range of the gauge is between ±2 to ±5 mm (78 × 10–3 to to the ACI 209R-92 shrinkage formula was calculated to be 42
196 × 10–3 in.). The accuracy of the measurements improved microstrains. The displacement crack gauge transducer was
as the cracks started to widen. 20 mm (0.79 in.). Therefore, the calculated shrinkage displace-
Electrical strain gauge data, measurements of the LPDT, ment was 0.00084 mm (0.000033 in.), compared to a crack
and the load readings were logged into a computerized data width range of 0.269 to 0.473 mm (0.011 to 0.019 in.).
acquisition system. A digital instrumentation system was A large amount of test data was recorded and the related
graphs were prepared. The behavior of the slabs were
presented in terms of the load-deflection relationship at
different load stages, including service and ultimate load, as
well as crack width-steel strain relationships. Failure modes,
crack patterns, and crack spacing were also depicted by
means of photographs. Herein, the research focuses only on
the data that relates to crack width and serviceability loading
rather than ultimate loading and failure modes.

Crack width
The crack width was measured at each load stage. It was
important to record the crack width at the serviceability
limit. The crack width measurement at the ultimate load,
however, was not structurally significant. The crack width at
the serviceability limit was reported for all specimens, as
Fig. 9—Crack width versus steel strain for Specimen NSC1. given in Table 2. The opening of the crack width was plotted
(Note: 1 mm = 0.0394 in.) versus the steel strain up to the ultimate load, as shown in
Fig. 9 through 13. All measurements reported in Table 2
were taken at the serviceability level at a steel stress level of
250 MPa (equivalent to 0.625fy). All the results of the crack
width prediction equations by different codes presented in
this paper are also reported in Table 2.
In general, the crack width increases as the load and the
deflection is increased. This increase, however, was not very
smooth due the nonlinear behavior of concrete. It was
evident that the crack width versus steel strain was represented
by one straight line up to an average steel strain of 0.001 to
0.0015 for all the specimens. The relation between steel
strain and crack width curve, however, tended to behave
nonlinearly after the steel strain reached a value of 0.0018.
For Specimen NSC2, the crack width continued to increase
after the steel strain reached the yield point, which was an
expected behavior for a slab that failed in pure flexure.
The widths of the primary cracks were examined in the
first two series to determine the effect of reinforcement
Fig. 10—Crack width versus steel strain for Specimen HSC1. spacing and concrete cover. The data showed that as the
(Note: 1 mm = 0.0394 in.) concrete cover increased in Series I, the crack widths became

286 ACI Structural Journal/May-June 2010

larger. The maximum crack width can be influenced by as under pure-punching shear, as defined by Marzouk and
much as 90% when the concrete cover increased from 30 to Hussein (1991). A comparison of the experimental results of
50 mm (1.2 to 2 in.) cover. Moreover, the data showed that the average crack width measurements of the three tested
increasing the concrete cover by 100% to 60 mm (2.4 in.) resulted series indicated that the mode of failure has no effect on the
in increasing the maximum crack width by 106%. crack width size at the serviceability level.
Three specimens of Series II (Specimens NSC1, HSC3,
and HSC4) were designed and specifically tested to NUMERICAL INVESTIGATION
determine the effect of increasing bar spacing on crack Crack width test results versus codes prediction
width. In general, as the bar spacing was increased, the crack The test results and the predicted values of the ACI 318-05,
width was increased, as shown from code predictions in Table 2. CSA-S474-04, and NS 3473 E average crack widths are
The data of Series II showed that for the range of bar spacing shown in Table 2. For Series I and II, Fig. 14 to 17 indicate
tested, the maximum crack width at the serviceability limit that both BS EN 1992-1-1:2004 and CEB-90 are very similar
can be influenced by the bar spacing. When the bar spacing in results for maximum predicted crack width and the value
of Specimen NSC1 (150 mm [6 in.] spacing) was increased
to 200 mm (9 in.) spacing, the crack width was increased by
19%. The measured crack width of Specimens HSC3 and Table 2—Comparison of test results with
HSC4, however, were similar. It is evident from the test predictions of other international codes
results of Series I and II that the effect of concrete cover on Series Slab Experiment, ACI, CSA, NS, CEB-FIP,
the crack width is more profound than the effect of the bar spacing. no. no. mm mm mm mm mm EC2
Series III was designed to investigate the effect of pure NSC1 0.406 0.261 0.227 0.269 0.107 0.135
flexure failure and ductile shear failure modes on crack Series I HSC1 0.772 0.311 0.351 0.354 0.114 0.142
properties. Specimens NSC2 and NSC3 were designed with HSC2 0.950- 0.341 0.438 0.397 0.115 0.143
very low steel reinforcement ratios to fail under flexure- HSC3 0.486 0.329 0.252 0.314 0.143 0.160
Series II
punching, whereas Specimen HSC5 was designed to fail HSC4 0.483 0.399 0.287 0.361 0.174 0.183
HSC5 0.327 0.258 0.248 0.294 0.133 0.165
Series III NSC2 0.248 0.376 0.324 0.430 0.236 0.249
NSC3 — 0.348 0.425 0.473 0.185 0.268
Note: 1 mm = 0.0394 in.

Fig. 11—Crack width versus steel strain Specimen HSC3.

(Note: 1 mm = 0.0394 in.)
Fig. 13—Crack width expansion versus steel strain for
Specimen NSC2. (Note: 1 mm = 0.0394 in.)

Fig. 12—Crack width expansion versus steel strain for Fig. 14—Comparison of maximum crack width for Series I.
Specimen HSC4. (Note: 1 mm = 0.0394 in.) (Note: 1 mm = 0.0394 in.)

ACI Structural Journal/May-June 2010 287

is less than the experimental results by 75%. The CSA-S474-04 strength concrete specimens with small concrete covers of 30 to
and NS 3473 E give closer results than the BS EN 1992-1-1:2004 50 mm (1.2 to 2 in.) cover. Therefore, the BS EN 1992-1-1:2004
and CEB-90, but less than the experiment results. In other and the CEB-90 codes are recommended to be used for building
words, the BS EN 1992-1-1:2004 and the CEB-90 under- with small concrete cover rather than for infrastructure
estimate the maximum crack width by a large percentage for facilities like offshore and nuclear containment structures.
specimens with thick concrete covers of 60 mm (2.4 in.). For The results of ACI 318-05, CSA-S474-04, and NS 3473 E can
Series III, as shown in Table 2, these codes can provide a provide a reasonable estimate for crack widths of such structures.
good prediction for maximum crack width for normal-
Modified tension chord method
The tension chord model (Marti et al. 1998) was used to
calculate the average concrete strain εcm from the simple
analysis of the cross section of a flexural member under
loading, as shown in Fig. 18. Moreover, the tension chord
model gives a better understanding of the cracking mechanism in
reinforced concrete members. The concrete section under
flexural loading was composed of a compression chord
subjected to compressive force (C) and a tension chord
subjected to tensile force (T). The tensile longitudinal zone
between consecutive cracks was subjected to an axial tensile
force (T). The length of the segment represented by Srm, was
equal to crack spacing. Thus, the average concrete strain was

σ cm τ b s rm ρ ef
ε cm = --------
- = -------------------
- (12)
Ec E c d be
Fig. 15—Comparison of maximum crack width for Series II.
(Note: 1 mm = 0.0394 in.) where τb is the bond stress at the steel-concrete interface
(MPa); dbe is the nominal diameter of the tensile reinforcing
bar (mm); Srm is the average crack spacing (mm) from
Eq. (7); ρef = Ast /Act is the effective reinforcement ratio (Ast
is the ratio of tensile reinforcement area to concrete area
[mm2] and Act is the ratio area to the area of the effective
concrete in tension to concrete area [mm2]); εcm is the
average concrete strain; σcm is the average concrete stress
(MPa); εsm is the average steel strain; Act = hef b (mm2); Ec
is the modulus elasticity of concrete (MPa); Es is the modulus
elasticity of steel (MPa); M is the applied service moment
(; kd is the distance from surface to locate the neutral
axis (mm); and h is the effective embedment thickness (mm).

Estimate of bond shear stress (τb)

The bond strength between the concrete and reinforcement
is an important factor in calculating the crack width. The
force in the bar is transmitted to the surrounding concrete by
Fig. 16—Comparison of crack width for Series I. (Note: bond shear stress τb. The bond shear stress depends on
1 mm = 0.0394 in.) several factors, including the concrete tensile strength and
cover, steel stress, bar size and spacing, confining effects,

Table 3—Comparison of test results of crack width

with modified tension chord assumptions and
fracture energy
Bar Concrete
Series Slab fc′, spacing, cover, Experiment MTCA
no. no. MPa mm mm Wk, mm Wk, mm
NSC1 35 150 30 0.406 0.388
Series I HSC1 68.5 150 50 0.772 0.596
HSC2 70 150 60 0.950 0.743
HSC3 66.7 200 30 0.486 0.431
Series II
HSC4 61.2 250 30 0.483 0.490
HSC5 70 100 30 0.327 0.424
Series III NSC2 33 240 30 0.248 0.557
Fig. 17—Comparison of crack width for Series II. (Note: NSC3 34 210 40 — —
1 mm = 0.0394 in.) Note: 1 mm = 0.0394 in.; 1 MPa = 145 psi.

288 ACI Structural Journal/May-June 2010

Fig. 18—Analytical model for flexural cracking in singularly reinforced flexural member.

and load history. Several experimental and theoretical The average crack width at the extreme concrete surface was
investigations were conducted on the behavior of bond for
normal- and high-strength concrete. Marti et al. (1998) f h – kd τ b S rm ρ ef
estimated that the bond stress τb is approximately twice the w m = S rm ⎛ ----s- ⋅ ---------------⎞ – -------------------
- (15)
⎝ E s d – kd⎠ E c d be
concrete direct tensile strength fct at the service load.
Gilbert (2005) indicated that bond stress τb is reduced as
the stress in the reinforcement is increased and proposed where h is the total member thickness (mm); dbe is the effective
the following equation depth (mm); fs is the stress in the steel reinforcement at the
end of the serviceability limit (MPa); kd is the distance from
τb = α1α2 fct (MPa) (13) surface to locate the neutral axis (mm); and h is the effective
embedment thickness (mm).
where α1 depends on the steel stress at crack; α2 is for short- The maximum crack width recorded during each experiment is
or long-term calculations; and fct is the direct tensile strength presented in Table 3 along with the numerical predictions of
of concrete (MPa). the modified tension chord assumptions (MTCA). The
The experimental results of a previous investigation on results of the modified tension cord method are plotted
direct tension test and fracture energy (Marzouk and Chen graphically in Fig. 16 and 17.
1995) and bond test results (Alavi-Fard and Marzouk 2004)
indicated that τb = (1.5 – 2.0)fct at the cracking serviceability SUMMARY AND CONCLUSIONS
limit. Gilbert (2005) recommended using τb = (2.0fct) for The experimental and numerical results can be summarized
short-term calculation and τb = (1.0)fct for long-term calculation as follows:
in determination of the final maximum crack width due to 1. The experimental results indicated that the relationship
creep and shrinkage. The Alavi-Fard and Marzouk (2004) between steel strain and crack width on the tension face is
bond study on high-strength concrete, over 60 MPa (8700 psi), approximately linear up to values that range from 1000 με to
revealed that the bond stress is more appropriately 1800 με of the steel strains. All crack widths were measured
proportional to the cubic root and the bond stress at the experimentally at the serviceability steel stress level of
serviceability limited can be approximately estimated to be 0.625fy (250 MPa [36.25 ksi]).
τb = (1.65)fct for instantiations loading including the effect of 2. The test results of reinforced concrete slabs indicated
temperature and shrinkage. The study also recommended that the concrete cover had a major effect on the crack width,
that the square root of the compressive strength approach and the bar spacing had less of an effect on the crack width
adopted by CSA-S474-04 and ACI 318-05 does not provide size at the serviceability limit. In general, the calculated
a good prediction of the bond strength for the high-strength average crack width was lower than the experimental test results.
concrete. The bond strength of high-strength concrete, over 3. The comparison between different code predictions, as
60 MPa (8700 psi), is more appropriately proportional to the shown in Table 2, indicated that ACI 318-05, CSA-S474-04,
cubic root. All of the European codes recommend that the and NS 3473 E estimates can provide a reasonably close
tension and bond strength of concrete be proportional to the prediction for the crack width of concrete structure members
power of 1/3, while both CEB-90 and NS-92 recommend the with different bar spacing. The effect of concrete cover on
use to the power of 2/3 of the compressive strength. thick concrete members, however, can be improved.
4. A modified numerical model based on the tension chord
Estimate of average steel tensile strain εsm method is recommended for calculating the maximum crack
in tension chord width. The proposed theoretical model allows designers to
Due to the low tensile strength of concrete, cracking in specify concrete covers and reinforcement ratios during the
concrete structures starts at an early stage of loading. For crack design process to control the flexural crack width to an
width analysis, the serviceability limit state governed the crack acceptable limit. The proposed model can be suitable for
width calculations and the stress in steel reinforcement fs was thick high-strength concrete structures with large concrete
selected as the serviceability limit. This value was considered as cover, such as offshore and nuclear containment structures.
the end of the serviceability limit after which the concrete
material behaved nonlinearly. ACKNOWLEDGMENTS
The strain at the concrete surface on the tension concrete The authors are grateful to the Natural Sciences and Engineering
face where the crack width was measured was calculated using Research Council of Canada (NSERC) for providing the funds for the
project. Sincere thanks are due to M. Curtis, S. Organ, D. Pike, and the
technical staff at the Structural Engineering Laboratory at Memorial
f h – kd
ε sm = ⎛ ----s- ⋅ ---------------⎞
University of Newfoundland for their assistance during the preparation and
⎝ E s d – kd⎠ testing of the specimens. Sincere thanks are extended to Capital Ready Mix
Ltd., Newfoundland, for providing the concrete for this project.

ACI Structural Journal/May-June 2010 289

REFERENCES NS 3473 E, 1992, “Concrete Structures, Design Rules,” Norwegian
ACI Committee 209, 1992, “Prediction of Creep, Shrinkage, and Council for Building Standardization, Oslo, Norway, 79 pp.
Temperature Effects in Concrete Structures,” (ACI 209R-92),” American Vecchio, F. J., and Collins, M. P., 1986, “The Modified Compression
Concrete Institute, Farmington Hills, MI, 47 pp. Field Theory Reinforced Concrete Elements Subjected to Shear,” ACI
ACI Committee 224, 2001, “Control of Cracking in Concrete Structures JOURNAL, Proceedings V. 83, No. 2, Mar.-Apr., pp. 219-231.
(ACI 224R-01),” American Concrete Institute, Farmington Hills, MI,
46 pp.
ACI Committee 318, 2005, “Building Code Requirements for Structural APPENDIX A
Concrete (ACI 318-05) and Commentary (318R-05),” American Concrete Numerical example
Institute, Farmington Hills, MI, 430 pp. Sample calculations using modified tension chord
Alavi-Fard, M., and Marzouk, H., 2004, “Bond Behaviour of High
Strength Concrete,” Magazine of Concrete Research, V. 56, No. 9, pp. 545-557.
assumption (MTCA) principles for crack width estimation
CEB-FIP, 1990, “Model Code for Concrete Structures (CEB-90),” Euro are presented for Specimen HSC1.
Internationale deBeton/Federation Internationale de la Procontrainte, The following mechanical properties were recorded from
Lausanne, Swittzerland, 460 pp. Marzouk and Chen (1995) direct tension and fracture energy
BS EN 1992-1-1:2004, 2004, “Eurocode 2: Design of Concrete Structures.
General Rules and Rules for Buildings,” British Standards Institutions, London,
laboratory measurements: Ec = 51,400 (7450 ksi); Es =
UK, 230 pp. 200,000 (29,000 ksi); ft′ = 3.30 MPa (480 psi), using
CSA-S474-04, 2004, “Concrete Offshore Structures,” Canadian Marzouk and Chen (1995) tension stiffening model; and
Standards Association, Mississauga, ON, Canada, 78 pp. εto = 118.6 με.
Frosch, R. J., 1999, “Another Look at Cracking and Crack Control in
Reinforced Concrete,” ACI Structural Journal, V. 96, No. 3, May-June, The averaged bond stress τb of high-strength concrete at
pp. 437-442. the serviceability limit was obtained from the experimental
Gilbert, R. I., 2005, “Time-Dependent Cracking and Crack Control in results of Alavi-Fard and Marzouk (2004), τb = 5.4 MPa
Reinforced Concrete Structure,” Serviceability of Concrete: A Symposium (783 psi) (equivalent to 1.65fct).
Honoring Dr. Edward G. Nawy, SP-225, American Concrete Institute,
Farmington Hills, MI, pp. 223-240 The average crack spacing Srm can be calculated from the
Gergely, P., and Lutz, L. A., 1968, “Maximum Crack Width in following, (Eq. (4)).
Reinforced Concrete Flexural Members,” Causes, Mechanism, and
Control of Cracking in Concrete, SP-20, American Concrete Institute,
Farmington Hills, MI, pp. 87-117. Srm = 2.0(c + 0.1s) + k1k2dbehefb/As
LabVIEW, 2006, “Acquiring Data and Processing Signals,” National
Instruments Corporation,
Marti, P.; Alvarez, M.; Kaufmann, W.; and Sigrist, V., 1998, “Tension
where Srm = 167 mm (6.57 in.)
Chord Model for Structural Concrete,” Structural Engineering International, Crack width calculation
pp. 287-298.
Marzouk, H., and Hussein, A., 1991, “Experimental Investigation on the
f h – kd τ b S rm ρ ef
w m = S rm ⎛ ----s- ⋅ ---------------⎞ – -------------------
Behavior of High-Strength Concrete Slabs,” ACI Structural Journal, V. 88,
No. 6, Nov.-Dec., pp. 701-713. ⎝ E s d – kd⎠ Ec db
Marzouk, H., and Chen, Z. W., 1993, “Nonlinear Analysis of Normal
and High-strength Concrete Slabs,” Canadian Journal of Civil Engineering,
V. 20, No. 4, pp. 696-707.
Marzouk, H., and Chen, Z. W., 1995, “Fracture Energy and Tension
Therefore, the average width crack wm = 0.351 mm (0.138 in.);
Properties of High-Strength Concrete,” Journal of Material in Civil Engineering, and the characteristic crack width wk = 1.7wm = 0.596 mm
ASCE, V. 7, No. 2, pp. 108-116. (0.023 in.) at a steel stress level of 250 MPa (36 ksi).

290 ACI Structural Journal/May-June 2010