Vous êtes sur la page 1sur 11

Technical Paper

Wael Kassem

DOI: 10.1002/suco.201400045

Shear strength of deep beams:


a mathematical model and design formula
This paper presents a proposal for estimating the shear capacity
of reinforced concrete deep beams. The proposed model is
based on the fixed-angle softened truss model and utilizes a newly proposed formula for the effective transverse compressive
stress acting on the beam web. The proposed formula is developed using a stepwise multiple linear regression analysis of the
reported shear capacity values of 445 experimental deep beams.
The validity of the mathematical model is examined by comparing
its response with the experimental results as well as the predictions of other formulas available in the literature, and it results in
the one best fitting the measured shear strengths. The mathematical model leads to an explicit single closed-form expression for
computing the shear strength of deep beams. The proposed expression is dimensionless and contains four variables that express the horizontal and vertical reinforcement ratios, the concrete strength and the shear span-to-depth ratio. On the basis of
the results of this paper, a design formula is proposed with predictions that are more consistent and also more reliable than
those of the ACI Code and the Eurocode.
Keywords: reinforced concrete, deep beam, shear, fixed-angle, truss model

Introduction

Reinforced concrete (RC) deep beams are common structural elements that are often used in pile caps, tanks, folded plates, foundation walls, bins, folded plate roof structures and offshore structures as well as in tall buildings as
transfer girders [1, 2]. They are identified as being relatively short and deep, with a thickness that is small relative to
their span or depth. Owing to the relatively small span-todepth ratio, a complex strain state develops in deep beams
and so traditional sectional design approaches based on
plane sections theory are violated and unsuitable for their
design. Deep beams according to ACI 318-11 [3] are structural members in which either the clear span is 4 times
the overall depth, or the concentrated loads are applied
within a distance 2 times the depth from the face of the
support. They are frequently designed using either semiempirical methods [47], non-linear finite element analysis
[8] or by using compatibility-aided truss models [911], in-

* Corresponding author: wakassem@uqu.edu.sa


Submitted for review: 5 June 2014
Revised: 22 October 2014
Accepted for publication: 11 November 2014

184

cluding both strut-and-tie models [6] and truss-based models [1215].


The strut-and-tie model (STM) is gaining in popularity and respect as a rational method for designing deep
beams and has been included in the most recent design
guidance as provided by ACI-ASCE Committee 445 [16]
and fib Bulletin [17]. The STM is based on the lower
bound theorem of plasticity. Consequently, there is an unlimited number of possible solutions and designers must
exercise greater care when selecting an appropriate model
that will arrive at a safe, optimum solution. Several STMs
are usually possible and so there is no unique design solution as typically occurs when using sectional design procedures. This flexibility may lead to inadequate strength to
support the applied loads if an inappropriate model is selected, which reduces the confidence in the design
method, and might be considered as an approximate approach with uncertain accuracy [11]. Despite the guidance
on the development of STMs and the calculation of member dimensions and axial resistance provided by design
standards, difficulties can arise in the dimensioning of
nodes when their boundaries are not defined by bearing
plates [18].
Prior studies [2, 6] have indicated that empirical
equations available in the literature do not provide unbiased estimates of the shear capacity of deep beams and
their predictions yield significant scatter. Moreover, the
ACI 318-11 [3] method for computing the shear strength of
deep beams is too laborious to apply and can lead to very
over-conservative shear strength predictions [6]. Addressing these limitations points to the importance of providing
a more accurate method that can be used to predict the
shear capacity of deep beams accurately.
On the other hand, truss-based models include both
the Rotating Angle Softened Truss Model (RA-STM) and
the Fixed-Angle Softened Truss Model (FA-STM) [19]. The
fundamental discrepancies between the two models lie in
the way they handle the direction of cracks. The RA-STM
is based on the assumption that the direction of cracks is
perpendicular to the principal tensile stress in the concrete element, whereas the FA-STM is based on the assumption that the direction of subsequent cracks is perpendicular to the applied principal tensile stress [20]. An
analysis method based on the RA-STM has previously
been developed for predicting the shear strength of deep
beams [9]. The assumption of a rotating angle leads to the

2015 Ernst & Sohn Verlag fr Architektur und technische Wissenschaften GmbH & Co. KG, Berlin Structural Concrete (2015), No. 2

W. Kassem Shear strength of deep beams: a mathematical model and design formula

the designer to avoid the current tedious and time-consuming computing procedures. The proposed theoretical
model consists of the application of the FA-STM and utilizes a newly proposed formula for the effective transverse
compressive stress acting on the beam web. A design formula for estimating the shear strength of deep beams, and
suitable for implementing in codes and standards, is proposed.

conclusion that the shear strength is due entirely to the


steel, i.e. concrete makes no contribution, which seems to
be unrealistic and leads to the shear strengths of deep
beams being underestimated [2]. This drawback was overcome in the FA-STM by assuming that the cracks in the
concrete are oriented at a fixed angle; hence, it can predict
the contribution of the concrete. Unlike the RA-STM,
however, the concept of the FA-STM has only been employed to predict the shear strength of RC panels [19], not
deep beams.
In the present paper, a rational model based on the
FA-STM is proposed for predicting the shear capacity of
RC deep beams while employing the three fundamental
principles of mechanics of materials: the stress equilibrium condition, the strain compatibility condition and the
constitutive laws of concrete and steel. The applicability
of the proposed method has been validated by comparing
the calculated shear capacity with the experimental results
of 445 RC deep beams available in the literature. A single
closed-form expression that is more consistent and more
accurate than existing formulas or tedious and time-consuming computing procedures is presented. Moreover, an
explicit general design formula, whose predictions are adequately on the safe side, is also proposed.

Fig. 1 shows a typical RC deep beam of rectangular crosssection that is loaded on top by two vertical forces Vn
applied at a distance a from the midpoint of the closest
support. The longitudinal main reinforcement of crosssectional area Ast is placed at a distance (hd) from the
beam bottom surface, where d is the effective depth of the
beam and h is the overall beam depth. The beam is reinforced vertically by uniformly distributed stirrups at horizontal spacing Sv in the shear span. The horizontal web reinforcement is provided by uniformly distributed rebars at
the vertical spacing Sh.
According to the RA-STM model proposed by Mau
and Hsu [9], the deep beam within the shear span may be
visualized as an assembly of three different elements: the
top element with a thickness d (equal to distance from
top surface of beam to centre of flexural compression
steel), including the concrete and top compression steel;
the bottom element, consisting of bottom tension steel
only; and the middle element, including web concrete,

Research significance

The aim of the present study is to resolve problems in predicting the shear strength of deep beams by means of an
explicit, sufficiently accurate single expression that allows

Vn

Vn

Asc

d'

Description of model

Diagonal
crack

Av
dv

Sh

h
d

Ah
1

Sv

Web reinforcement

Ast

Ln

c2

lt

c2
l
lt

Shear
element

l f l

Vn

1c

c2
c
12

c
12

1c
c
12

1c
Rebars

1c
c2

Vn

c2
t f t

Fig. 1. Geometry of reinforced concrete deep beam with the stresses in the shear element

Structural Concrete (2015), No. 2

185

W. Kassem Shear strength of deep beams: a mathematical model and design formula

web reinforcement and both top and bottom longitudinal


steel. The top and bottom elements resist the longitudinal
compressive and tensile stresses resulting from the sectional moment respectively, while the middle element
resists the sectional shear. The middle element, indicated
by the dashed lines in Fig. 1, behaves like a RC membrane element subjected to in-plane normal and shear
stresses.

age normal strains in the l and t directions of the two-dimensional deep beam element can be expressed in matrix
form as follows [21]:

3.1. Stress equilibrium equations

where:
l, t smeared uniaxial strain in l and t directions respectively
lt
smear shear strain at l-t coordinate
2, 1 smeared strain in 1-direction and 2-direction respectively
12
smear shear strain at 12 coordinate

The application of the vertical loading causes concrete


cracking that splits the concrete into a series of diagonal
struts at an angle to the longitudinal direction which resist not only compressive stresses along their axis 2c and
tensile stresses perpendicular to the axis 1c , but also shear
c
stress 12
in the direction of the cracks, see Fig. 1. By assuming that the flexural strength of the deep beam is sufficiently greater than its shear strength, the shear strength lt
V
can be defined as n , where b is the beam width and
bdv
dv is the beam effective depth, see Fig. 1. Any point
of stress on the beam web in the shear span should conform to the requirement of force equilibrium of the twodimensional stress element, whereas any point of strain
should meet the requirement of Mohrs compatibility. The
point stress equilibrium satisfies the three algebraic
equations, which can be expressed in a matrix form as
follows:

l
t

lt

2c

= [ S( )] 1 +
c

12

f
ll
t ft

(1)

where:
[S()]is the transformation matrix defined as [21]:

cos 2( )
sin2( )
2 cos( ) sin( )

2
2
[ S( )] =
2 cos( ) sin( ) (2)
sin ( )
cos ( )
cos( ) sin( ) cos( ) sin( ) cos 2( ) sin2( )

and
l, t normal stresses in the RC element in the longitudinal (horizontal) direction l and transverse (vertical)
direction t respectively (positive for tension)
lt
shear stress at l-t coordinate
l, t reinforcement ratios in l and t directions respectively
fl , ft smeared steel stress in longitudinal and transverse
directions respectively

fixed angle between applied compression stress (2direction) and longitudinal steel bars (assumed to
be equal to the inclination of the line connecting
centre-line of loading plate and centre-line of support)

t
0.5
lt

= [ S( )] 1

0.5
12

(3)

3.3. Constitutive laws of materials


The stress and strain in the concrete in the 2-direction are
assumed to obey the following material law for the softened concrete [20]:

2c = fc [2( 2 / 0 ) ( 2 / 0 )2 ]

( 2 / 0 ) 1

(4a)

2c = fc [1 (( 2 / 0 ) 1/(4/ 1))2 ]

( 2 / 0 ) 1

(4b)

with = (0.9/ 400 + 1 )

(5a)

f 20

0 = 0.002 0.001 c
for 20 fc 120 (MPa)
80

(5b)

where:
0 concrete cylinder strain corresponding to peak cylinder strength fc
softening coefficient
The constitutive relationship for concrete in tension can
be expressed mathematically using the following [22]:

1c = Ec 1

1 cr

(6a)

1c = fcr cr 0.4

1 > cr

(6b)

where Ec is the elastic modulus of concrete


Ec = 3875
fc
(M
Pa) and fcr and cr are the cracking stress
and strain in the concrete respectively, taken as
fcr = 0.31
fc
(M
Pa) and cr = 0.00008.
Several expressions were proposed for describing the
concretes stress-strain behaviour in shear, among them
the relationship proposed by Zhu et al. [23] to correlate
c and the shear
between the shear stress in the concrete 12
strain 12:
c = ( c c )/2( )
12
12 1
2
1
2

(7)

3.2. Strain compatibility equations


Similar to stresses, the in-plane strain conditions should
fulfil the requirements of Mohrs compatibility. The aver-

186

Structural Concrete (2015), No. 2

The rebars in the proposed model were assumed to resist


axial stresses only, neglecting any possible dowel action
[20]. The steel stresses in Eq. (1) are related to the longitu-

W. Kassem Shear strength of deep beams: a mathematical model and design formula

dinal and transverse strains based on a linear elastic perfectly plastic relationship:
fs = Es s

s y

(8a), (9a)

fs = fy

s > y

(8b), (9b)

where fs and s are the smeared stress and strain of the


steel bars respectively and fy is the yield stress of the steel
rebars. They change to fl, l and ft, t when applied to longitudinal and transverse steel respectively, and Es is the
modulus of elasticity of steel rebars.
In calculating the longitudinal steel ratio of the shear
element l the longitudinal steel reinforcement provided
at the bottom and the top of the beam is also included.
This is because the expansion of the element in the longitudinal direction due to shear is restrained by the longitudinal top and bottom bars, as previously proposed in the
RA-STM [9]. Moreover, the beam effective depth dv is assumed to be equal to the distance between the centre of
the compression steel and the centre of the tension steel.
In specimens without compression steel reinforcement,
depth d is calculated as one-half of the depth needed to
provide a compression force in the concrete to balance
the tension in the bottom tension steel at yield [9].

The analysis of the FA-STM for predicting the shear capacity of deep beams incorporates 15 unknowns: l, t, lt,
c
2c , 1c , 12
, fl, ft, l, t, lt, 2, 1, 12 and . Twelve equations
have already been defined, including six equilibrium and
compatibility equations and six related to the constitutive
laws of materials. The degree of indeterminacy can be reduced by specifying strain 2 for each load stage and calculating the effective normal stress in the horizontal direction l from the applied axial load N and deep beam
cross-section. Since axial loading is rarely applied to deep
beams, the value of the normal stress l vanishes.
The application of the vertical loads along with the
support reactions creates a large compressive stress field t
normal to the beams horizontal axis which is responsible
for the arch action unique to deep beams. Mau and Hsu [9]
estimated the transverse stress t using a transverse stress
intensity factor K, defined as a function of the shear span to
overall depth ratio a/h and beam effective depth dv, defined in Fig. 1 and expressed mathematically as

t = K lt

K1 =

2dv
h

(10)

a
K1 0 < 0.5
h

K = K2 0.5 < a 2
h

0 >2

h
K2 =

dv
h

l fl + t ft = ( l + t ) ( 2c + 1c )

(12)

c sin(2 ) (13)
l fl t ft = ( l t ) ( 2c 1c ) cos(2 ) 2 12

With the help of a computer this procedure is used to


trace the response history of the shear element and to locate the maximum shear sustained by the shear element.

3.4. Solution technique

where

plication of RA-STM for predicting deep beam shear


strengths, while considering the transverse stress intensity
factor K, showed that the method generally leads to unsafe
predictions [2, 6]. This is possibly due to the exclusion of
the shear resistance of the concrete struts as well as the
implementation of the transverse stress intensity factor K,
which is not based on experimental or analytical evidence
[24]. This limitation thus represents a weakness in an otherwise sound analysis method for deep beams, and it is anticipated that if the FA-STM were to be used to include the
shear resistance of concrete struts, and a more realistic
value of the transverse stress intensity factor K could be
developed, the overall accuracy of the model would be improved.
The solution algorithm of the proposed method of
analysis is similar to that of the fixed-angle solution proposed by Hsu and Mo [20]. The first two basic equilibrium
equations in Eq. (1) are summed and subtracted to obtain
the following two equations, which are used as the convergence criteria for the solution procedure [1, 25]:

(11)

h 4 2a

a 3 3h

Eq. (10) supplies the remaining condition required to


make the analysis possible. Earlier assessment of the ap-

Regression analysis of experimental data

With the aim of determining the value of the transverse


stress intensity factor K on the basis of experimental results, a database of 445 deep beams was compiled. The data for the 445 deep beams used in this study were gathered, revised and organized by (in chronological order):
Birrcher [26], Brown et al. [27], Clark [28], de Paiva and
Siess [29], Foster and Gilbert [30], Hara [31], Islam [2],
Kong et al. [32], Matsuo et al. [33], Moody et al. [25], Oh
and Shin [34], Ramakrishnan and Ananthanarayana [35],
Rogowsky et al. [1], Smith and Vantsiostis [36], Subedi et
al. [37], Tan and Lu [38], Tan et al. [39, 40], Tanimura and
Sato [41] and Walraven and Lehwalter [42]. All the beams
selected were simply supported and tested under concentrated loading, did not fail in flexure or bearing modes,
had shear span to depth ratios a/d < 2.0 and effective
length-depth ratios Ln/h < 4.0 (according to ACI 318-11 [3])
and had uniformly distributed web reinforcement. Fig. 2
shows a summary of the information associated with experimental parameters in the form of histograms for the
445 deep beams considered in this study.
The data extracted was used in a stepwise multiple
linear regression analysis to determine a suitable value of
the transverse stress intensity factor K. The study was
based on the expected statistical association between the
K value and a number of parameters thought likely to influence its value, including geometric dimensions, reinforcement ratios and concrete strength. Mau and Hsu [9]
suggested that the transverse stress intensity factor is dependent on the span to overall depth ratio a/h. They postulated that for a/h = 0, the transverse stress at mid-depth
of the beam is a maximum on the centre-line of support

Structural Concrete (2015), No. 2

187

260

240

180

195

180

120
60

1000

50
25

0.5

1.0
a/d

1.5

135

Frequency denisty

270
180
90

2
c (%)

800

1000

30

60
90
f`c (MPa)

60
0

500

45

0.0 0.5 1.0 1.5 2.0 2.5 3.0


h (%)

1000 1500
h (mm)

2000

90
60
30
0

120

90

120

120

60
0

2.0

400 600
b (mm)

120

180

200

180

360

240

75

0
0.0

Frequency denisty

4000

Frequency denisty

Frequency denisty

100

2000 3000
a (mm)

65

Frequency denisty

130

0.0

0.5

1.0

1.5

2.0

2.5

3.0

0.0

0.5

1.0

1.5

2.0

2.5

3.0

260

Frequency denisty

Frequency denisty

240

Frequency denisty

Frequency denisty

W. Kassem Shear strength of deep beams: a mathematical model and design formula

b (%)

195
130
65
0

t (%)

Fig. 2. Histograms of geometric and other properties of the 445 deep beams*
*Note: a, b, c and d are identified in Fig. 1, b is the reinforcement ratio of flexural bottom tensile steel, taken as b = Ast /(bd), c is the reinforcement ratio of
flexural top compression steel, taken as c = Asc /(bd), c is the ratio of vertical web steel (stirrups), taken as t = Av /(bSv), and h is the ratio of horizontal
web steel, taken as h = Ah /(bSh). (Note: 1 MPa = 145 psi, 1 mm = 0.03937 inch)

action, gradually decreases as we move away from it and


vanishes at a/h = 2, where shear behaviour approaches
that of a slender beam. Based on this assumption, the deep
beams in the database were divided into two groups: a first
group (containing 90 specimens) in the range 0 < a/h 5
and a second group (containing 355 specimens) in the
range 0.5 < a/h 2. The value of the transverse stress intensity factor that provided a match between the mathematical and experimental shear strengths was recorded for
each specimen and then a stepwise multiple linear regression analysis was carried out to determine the parameters
that influence this factor. In order to arrive at generic values of the transverse stress intensity factor suitable for future deep beam design, the statistical association between
the transverse stress intensity factor values and several parameters related to concrete strength fc, geometric dimensions a, b, d, dv, a/d, a/dv, dv/h and a/h, reinforcement ratios l, t and reinforcement stresses l fyl and t fyt was
assessed.
The goal of using the stepwise multiple linear regression analysis was to determine which subset of independent variables contributes significantly to explaining the
variability of the dependent variable K by minimizing the

188

Structural Concrete (2015), No. 2

residual mean square, which would in turn maximize the


multiple correlation value R2. The proposed stepwise multiple linear regression model takes the following form:
Y = A0 + A1 X1 + A2 X2 + A3 X3 ++ Ap X p +

(14)

where p is the number of independent variables. The


residual mean square error is defined as follows [43]:
n

2i = [Yi ( A0 + A1 X1 + A2 X2 + A3 X3 + + Ap Xp)]
i =1

(15)

where n is the number of data points. The results show


that the association between K and parameters K1 and K2
assumed in the RA-STM (Eq. (11)) is very weak (R2 = 0.28)
for deep beams in the first range (0 < a/h 0.5) and slightly higher in the second range (0.5 < a/h 2), where R2 increases to 0.40, see Fig. 3. For the specimens considered in
first range, a stronger association (R2 = 0.71) was found
d
when the parameters l fyl, t, v , t and were conh
sidered, see Fig. 3. On the other hand, considering the

W. Kassem Shear strength of deep beams: a mathematical model and design formula

a
, f , f , f and h resulted
d t yt l yl c
in a robust association (R2 = 0.60) with the transverse
stress intensity factor K for specimens considered in the
second range, see Fig. 3. The stepwise multiple linear regression analysis showed that the transverse stress intensity factor K should best be expressed as shown in Eq.(16)
instead of Eq. (11):
effect of the parameters

a
K3 0 < 0.5
h

K = K4 0.5 < a 2
h

0 >2

(16)

K3 = 98.02 0.18 l fyl 580.2 t +


+ 130.30

dv
224.445 h 0.006 a 0
h

a
0.045 t fyt 0.012 l fyl +
d
+ 0.003 fc 0.001 h 0

K4 = 1.29 0.62

18

In Eq. (16), fc, fyl and fyt are in MPa, a and h in mm. Solutions with higher-degree polynomials produced a better fit
than Eq. (16), but they were rejected because of their excessive length. Simplicity is a requirement, and as the
complexity of the model increases, so its ability to generalize can be affected by the risk of over-fitting the data [44].
Eq. (16) implies that the transverse stress intensity factor is
dependent on the shear span to overall depth ratio. As the
a/h ratio increases, it is more likely that the effective transverse stress will decrease due to the greater distance between the applied load and the nearest support. However,
the current study revealed that other factors, in addition to
the a/h ratio, affect the factor K and should be considered
when calculating the transverse compressive stress, including longitudinal reinforcement ratio, horizontal shear
reinforcement, concrete strength and size effect. The
shear strength of a deep beam increases with the increase
in the longitudinal reinforcement ratio up to a certain limit depending on the shear span to depth ratio and concrete
compressive strength fc. This may be attributed to the better bond conditions, which results in increases in the
transverse compression in the direct strut and the dowel
action becoming more preponderant in deep beams with

18

10

15
18

R2=0.71

R2=0.28
12

12

12

0
1.6

1.7

1.8

1.9

2.0

1.5

2.0
1.5

5
10
Parameter (K3)

Parameter (K1)
0.0
1.5

0.5

1.0

0.0
1.5

0.2

0.4

0
15

0.6
1.5

R2=0.60

R2=0.40
1.0

0.5

0.5

1.0

1.0

0.0
0.0

0.5

1.0
Parameter (K2)

1.5

0.0
2.0

1.0

0.5

0.5

0.0
0.0

0.2

0.4

0.0
0.6

Parameter (K4)

Fig. 3. Correlation between transverse stress intensity factor K and four deep beam parameters in Eqs. (11) and (16)

Structural Concrete (2015), No. 2

189

W. Kassem Shear strength of deep beams: a mathematical model and design formula

Table 1. Stepwise multiple linear regression statistics

0<

Model

Sum of
squares

Degree of
freedom

Mean
square

Sig.

a
0.5
h

Regression
Residual
Total

1323.1
547.7
1870.8

6
84
90

264.6
7.4

35.7

.000

a
2
h

Regression
Residual
Total

19.1
12.8
32.0

8
347
355

2.3
.03

64.5

0.5 <

large amounts of longitudinal reinforcement, especially


when the longitudinal reinforcement is distributed over
several layers [43]. Moreover, as the horizontal shear reinforcement increases, confinement of the concrete in the
direct strut mechanism increases, which, consequently, increases the effective transverse compression in the direct
strut. Finally, as the beams get deeper, and have a greater
size, the cracking pattern will develop more quickly and it
is more likely that the transverse compression will decrease in this environment. A summary of the stepwise
multiple linear regression analysis results is given in
Table 1.

Verification of proposed method of analysis

A method of analysis based on the FA-STM and incorporating the newly proposed transverse stress intensity factor was used to predict the shear capacity of the 445 deep
beams for which the experimental results are available in

R2

0.84

0.71

0.77

0.60

.000

the literature. To assess the effectiveness of the proposed


model with respect to those in existence, the shear
strength formulae provided by the RA-STM [9], Mau and
Hsu [4], Siao [7], Hwang and Lee [45] and Russo et al. [6]
were considered while using a strength-reduction factor of
unity and the actual material properties obtained from the
testing. The shear strength formulae considered for estimating the shear capacity of RC deep beams are summarized in Table 2. A statistical presentation of the ratios of
the experimental test to calculated calc shear strengths for
the 445 deep beams considered is summarized in Table 3.
The diagrams representing the experimental versus calculated shear strengths relative to the proposed analysis
method and other formulae are also illustrated in Fig. 4,
together with the average (Avg) and coefficient of variation (CoV) values. A careful examination of the results
confirmed the inability of the RA-STM, Mau and Hsu [4],
Siao [7] and Hwang and Lee [45] formulae to produce safe
predictions of the shear capacity of deep beams. On the

Table 2. Shear strength formulae and procedures for calculating shear capacity

1. Mau and Hsu [4]

n1 = 0.5 fc [K (0.03 + l ) + K 2(0.03 + l )2 + 4(0.03 + l )(0.03 + t )] 0.3 fc [MPa]


where K, l and t are defined in Eqs. (10) and (25). Limitations of l 0.26 and t 0.12 were implemented in the analysis.
2. Siao [7]

1.04 fc [1 + n( h sin 2( ) + ( h cos2( )] [MPa]


n2 = max.
1.8 fy [( h sin 2( ) + ( v cos2( )]

where ft is the allowable concrete tensile stress, h and v are the steel ratios of the horizontal and vertical web reinforcement
respectively, is the angle between inclined strut and horizontal direction, n is the modular ratio and fy is the steel yield strength
of the web reinforcement.
3. Hwang and Lee [45]

n3 = (Kh + Kv 1) fc Astr cos( ) [MPa];


3.35
=
0.52; Astr = tw as; Kh & Kv are a function of the web steel reinforcement
fc
4. Russo et al. [6]

a
n4 = 0.76 k fc cos + 0.25 h fyh cot + 0.35 v fyv [MPa]
d

where k = (n f )2 + 2n f n f , = 0.74 3 1.28 2 + 0.22 + 0.87, =

a
fc
, = tan 1
,
105
0.9d

n is the modular ratio, a is the shear span, h and v are the steel ratios of the horizontal and vertical web reinforcement respectively,
fyh and fyv are the steel ratios of the horizontal and vertical web reinforcement respectively and d is the beam depth.
Note: 1 MPa = 145 psi, 1 mm = 0.03937 inch

190

Structural Concrete (2015), No. 2

W. Kassem Shear strength of deep beams: a mathematical model and design formula

Table 3. Statistics of ratio of experimental shear strength to calculated


shear capacity

Strength ratio
( test /calc)

Avg

SD

CoV

Max.

Min.

Current study

1.13

0.33

0.22

2.14

0.5

RA-STM [9]

0.77

0.25

0.33

2.00

0.28

Mau and Hsu [4]

0.76

0.22

0.29

1.62

0.33

Siao [7]

0.85

0.31

0.36

2.33

0.24

Hwang and Lee [45]

0.76

0.26

0.34

2.00

0.40

Russo et al. [6]

1.13

0.46

0.41

3.00

0.41

Closed-form expression for shear strength

In an attempt to develop a closed-form expression for estimating the shear strength of deep beams which is suitable
for implementation in codes and standards, the laborious
iterative procedures given in the preceding section were
modified. The three equilibrium conditions given in Eq.
(1) were manipulated as follows:
The first two equations may be rewritten using the
Pythagorean identity:
cos 2 =

sin 2 =

where:

c
C1 = c2 12

Mau and Hsu (1989)

18

18

12

18

24

12

test (MPa)

30

12

test / calc
Avg= 0.77
CoV= 33 %

30

calc (MPa)

18

30

Siao (1994)

12

18

test (MPa)

24

30

test / calc

Avg= 0.85
CoV= 36 %
18

test (MPa)

24

12

30

12

test / calc
Avg= 0.76
CoV= 34 %

12

18

24

12

18

24

30

test (MPa)

calc (MPa)

18

calc (MPa)

18

12

Russo et al. (2006)

18

12

30
24

Avg= 0.76
CoV= 29 %

Hwang and Lee (2002)


24

test / calc

24

c
C2 = c2 + 12

30
24

Avg= 1.13
CoV= 22 %

(20)

Dividing Eq. (20) by fc and assuming yielding of the steel,


the variables l fl and t ft in Eq. (20) can be replaced by a
dimensionless reinforcement index as follows:

calc (MPa)

calc (MPa)

K (C1 + l fl ) +

lt = 0.5 2
K (C1 + l fl )2 + 4(C1 + l fl )(C 2 + t ft )

24

(19)

The solution of Eq. (19) can be obtained from:

RA-STM

test / calc

(18)

c )(K + f + c + c )
( lt )2 = ( l fl + 2c 12
lt
tt
12
2

24

calc (MPa)

c cos sin )
(K lt + t ft + 2c + 2 12
( 1c 2c )

30

Current study

12

(17)

( 1c 2c )

and substituting in the third equilibrium condition given


in Eq. (1) gives a quadratic equation in terms of lt:

contrary, the proposed analysis method and Russo et al.


[6] formula yielded predictions on the safe side. Further,
the average experimental to calculated strength ratios using the proposed method and Russo et al. [6] formula are
equal, but the coefficient of variation of the proposed
model is much smaller (22 %). Given the breadth of the
database, these results are considered to be sufficiently
good to confirm that the proposed analysis method provides a reliable and safe means of predicting the shear capacity of deep beams.
Since the effect of support and loading plates widths
on shear strength of deep beams cannot be ascertained by
this method, although the significant effect of support
width on shear strength is evident from the tests of [2], this
point can be considered as the main limitation of the proposed analysis method.

30

c cos sin )
( l fl + 2c 2 12

test / calc

30

Avg= 1.13
CoV= 41 %
0

test (MPa)

12

18

24

30

test (MPa)

Fig. 4. Shear capacity calculated using the proposed analysis method and different methods versus measured shear strength

Structural Concrete (2015), No. 2

191

W. Kassem Shear strength of deep beams: a mathematical model and design formula

l fyl

l =

fc

and t =

t fyt

(21)

fc

Eq. (20) will then be:


K (1 + l ) +

lt = 0.5fc 2
K (1 + l )2 + 4(1 + l )( 2 + t )
where: 1 =

(22)

C1
C
and 2 = 2
fc
fc

The ultimate shear strength is controlled by the yielding of


the steel if l and t are limited to a maximum value as follows:
l Cl

t Ct

(23)

The values of constants 1, 2, Cl and Ct in Eqs. (22) and


(23) are determined based on optimization using test data
available in the database.
A constrained non-linear optimization function
fmincon included in MATLABs Optimization Toolbox
[46] was employed, which attempts to find a constrained
minimum of a scalar function of several variables starting
with an initial estimate within a region specified by linear
constraints and bounds. The constraint utilized in this
study was that the average (Avg) value of the ratio of shear
strength measured in the test and the calculated shear
strength (test/n) was set to 1.0 while minimizing the
coefficient of variation (CoV ) associated with the ratios of
experimental to calculated shear strength given by
Eq. (24):
S(i) =

test(i)
1
; Avg =
calc(i)
n

S(i); CoV =

i =1

1
n 1

[S(i) Avg]2 (24)


i =1

calc fc

(25)

This limit is imposed to control cracking under service


loads and guard against diagonal compression failures in
deep beams. A similar approach was proposed by the ACI
code [3] with a limit of 0.85
fc
(M
Pa) on the shear
strength of deep beams.
A search for the coefficients that satisfy the constraints and minimize the variation leads to the set
1 = 0.024, 2 = 0.049, Cl = 0.296, Ct = 0.307 and = 0.43.
Substituting the values of constants into Eq. (22), the proposed closed-form expression is as follows:
K (0.024 + l )

K 2 (0.024 + l )2 +
0.43fc
+
4(0.024 + )(0.049 + )
l
t

with the limitations l 0.296 and t 0.307.

192

Structural Concrete (2015), No. 2

Design formula

Despite the success of the proposed closed-form expression for predicting the shear strength of deep beams, as
expressed by the minimum coefficient of variation and average strength ratio of unity, when compared with experimental results and other formulae available in the literature, it is not possible to use the expression directly as a
code-oriented design equation. The reason for that is the
exclusion of any factor of safety in the predictions plus the
fact that it leads to a calculated shear strength increase
greater than the measured one for 52 % of the beams. In
order to achieve conservative predictions of shear strength
using the proposed expression, Eq. (26) can be multiplied
by a factor that can increase the average strength ratio
without affecting the coefficient of variation. The factor
value is determined using the Eurocode [47] rules, by leaving no more than 5 % of the specimens on the unsafe side.
The factor value, based on the experimental results of the
445 specimens, was found to be equal to 1.85. Consequently, the characteristic expression obtained in this way
is chosen as a design formula:
K (0.024 + l )

Rk = 0.93fc K 2 (0.024 + l )2 +
0.43fc [MPa] (27a)
+
4(0.024 + )(0.049 + )
l
t

where n is the number of specimens. Earlier studies postulated that the shear strength tends to increase only marginally beyond a certain value [4]. Thus, for practical purposes, the shear strength may also be limited by

calc = 0.5fc

Based on the optimization process, the average ratio


between test and calc is equal to 1 and the coefficient of
variation (CoV ) was found to be equal to 21 %. The proposed closed-form expression is a revised form of the formula proposed by Mau and Hsu [4], with more accurate
and consistent predictions.

(26)

K (0.024 + l )

Rk = 134.1fc K 2 (0.024 + l )2 +
0.43fc [psi] (27b)
+
4(0.024 + )(0.049 + )
l
t

To check the reliability and safety of the proposed design


formula when predicting shear strengths, Eq. (27) has
been compared with the predictions obtained from the
American [3] and European [47] codes. The ACI 318-11 [3]
method for designing deep beams was computed with the
strut-and-tie model described in Appendix A of this code
by utilizing the iterative procedure proposed by Russo et
al. [6]. The diagrams representing test/Rk versus the concrete strength relative to the proposed design formula
(Eq. (27)), ACI Code [3] and Eurocode [47] are shown in
Fig. 5, together with the corresponding Avg and CoV values. From the figure it is apparent that the predictions of
the proposed design formula are better than the others
since it leads to the most uniform prediction (i.e. the most
accurate prediction in the overall range of the parameters
considered), as supported by the lowest CoV value, and
the lowest number of non-conservative predictions.
The proposed design formula has a wide range of applications. In particular, the ranges of the individual parameters are:
0.1 a/d 2.0,
50 mm a 4375 mm,

W. Kassem Shear strength of deep beams: a mathematical model and design formula

8
test / Rk

4
2
0

25

50

75

100

f`c (MPa)

125

4
2
0

test / Rk

EC2

Avg= 1.90
CoV= 53 %

Avg= 1.85
CoV= 21 %

test /Rk

test /Rk

ACI (318-11)

25

50

75

100

f`c (MPa)

test / Rk
Avg= 2.48
CoV= 81 %

test /Rk

Design Formula

125

4
2
0

25

50

75

100

125

f`c (MPa)

Fig. 5. Effect of concrete strength fc on shear strength predictions using the proposed design formula, ACI Code [3] and Eurocode [47]

50 mm b 915 mm,
178 mm h 2050 mm,
152 mm d 2000 mm,
0.00 % h 3.25 %,
0.00 % c 4.25 %,
0.00 % h 3.20 %,
0.00 % t 2.90 %,
12.5 MPa fc 120 MPa,
280 MPa f yl 1330 MPa and
280 MPa f yt 1050 MPa.

Conclusions

A method of analysis based on the fixed-angle softened


truss model (FA-STM) and utilizing a newly proposed formula to estimate the transverse stress intensity factor has
been developed to predict the theoretical shear capacity of
deep beams. The proposed transverse stress intensity factor was based on a stepwise multiple linear regression
analysis involving the experimental results of 445 deep
beams in order to determine a suitable value for future
deep beam design. According to the results obtained from
comparing the experimental shear capacity of 445 deep
beams available in the literature with the proposed analysis method, the following conclusions can be drawn:
1. The proposed analysis model for computing the shear
capacity of deep beams leads to a single closed-form expression that is more reliable than others in existence.
It is not only the most accurate for the largest beam
specimens considered up to now, but it also provides
the lowest variation.
2. The proposed closed-form expression is a revised form
of the equation proposed by Mau and Hsu [4] and is
based on fitting with a large amount of data. The expression expresses explicitly the contributions of the
shear-span ratio a/h, horizontal reinforcement, web
vertical reinforcement and concrete strength.
3. By utilizing this closed-form expression, a general, conservative design formula for predicting the shear capacity of deep beams has been developed. The latter, besides being an explicit expression, is also more
consistent and more reliable than those in the ACI
Code [3] and Eurocode [47].

References
1. Rogowsky, D. M., MacGregor, J. G., Ong, S. Y.: Tests of reinforced concrete deep beams. ACI Journal Proceedings, 1986,
83 (4).
2. Islam, R.: Effect of web reinforcement on shear strength of
R/C deep beams. PhD thesis, Division of Civil Engineering,
National University of Singapore, 2003, p. 262.
3. American Concrete Institute: Building Code Requirements
for Structural Concrete (ACI 318-11) and Commentary (ACI
318R-11), Farmington Hills, MI, 2011.
4. Mau, S., Thomas, S. T. C.: Formula for the shear strength of
deep beams. ACI Structural Journal, 1989, 86 (5).
5. Mihaylov, B. I., Bentz, E. C., Collins, M. P.: Two-Parameter
Kinematic Theory for Shear Behavior of Deep Beams. ACI
Structural Journal, 2013, 110 (3).
6. Russo, G., Venir, R., Pauletta, M.: Reinforced Concrete
Deep Beams-Shear Strength Model and Design Formula.
ACI Structural Journal, 2005, 102 (3).
7. Siao, W. B.: Shear strength of short reinforced concrete
walls, corbels, and deep beams. ACI Structural Journal, 1994,
91 (2), pp. 123132.
8. Mohammed, A. K., Gomaa, S. T.: Finite element modelling
of deep beams. Computers & Structures, 1993, 48 (1), pp.
6371.
9. Mau, S. T., Hsu, T. T. C.: Shear strength prediction for deep
beams with web reinforcement. ACI Structural Journal, 1987,
84 (6), pp. 513523.
10. Hwang, S.-J., Lu, W.-Y., Lee, H.-J.: Shear strength prediction
for deep beams. ACI Structural Journal, 2000, 97 (3).
11. Park, J., Kuchma, D.: Strut-and-tie model analysis for
strength prediction of deep beams. ACI Structural Journal,
2007, 104 (6).
12. Gupta, A., Rangan, B. V.: High-Strength Concrete (HSC)
Structural Walls. ACI Structural Journal, 1998, 95 (2), pp.
194205.
13. Kassem, W., Elsheikh, A.: Estimation of Shear Strength of
Structural Shear Walls. Journal of Structural Engineering,
2010, 136 (10), pp. 12151224.
14. Mau, S. T., Hsu, T. C.: Shear Behaviour of reinforced Concrete Framed Wall Panels with Vertical Loads. ACI Structural Journal, 1987, 84 (3), pp. 228234.
15. Hsu, T. T. C., Mo, Y. L.: Softening of concrete in low-rise
shearwalls. ACI Journal, 1985, 82 (6), pp. 883889.
16. Joint ACI-ASCE committee 445. SP-273: Further examples for
the design of structural concrete with strut-and-tie models.
American Concrete Institute, Farmington Hills, MI, 2010.

Structural Concrete (2015), No. 2

193

W. Kassem Shear strength of deep beams: a mathematical model and design formula

17. fib: Design examples for strut-and-tie models. fib Bulletin


No. 61, fib, Lausanne, Switzerland, 2011.
18. Vollum, R. L., Parker, D.: External beamcolumn joints: design to Eurocode 2. Magazine of Concrete Research, 2000, 7
(60), pp. 511521.
19. Pang, X. B., Hsu, T. T. C., Fixed angle softened truss model
for reinforced concrete. ACI Structural Journal, 1996, 93 (2),
pp. 197207.
20. Hsu, T. T. C., Mo, Y. L.: Unified theory of concrete structures, Wiley, 2010.
21. Jeng, C. H., Hsu, T. T. C.: A softened membrane model for
torsion in reinforced concrete members. Engineering Structures, 2009, 31 (9), pp. 19441954.
22. Hsu, T. T. C., Zhu, R. R. H.: Softened membrane model for
reinforced concrete elements in shear. ACI Structural Journal, 2002, 99 (4), pp. 460469.
23. Zhu, R. R. H., Hsu, T. T. C., Lee, J. Y.: Rational shear modulus for smeared-crack analysis of reinforced concrete. ACI
Structural Journal, 2001, 98 (4), pp. 443450.
24. Bakir, P.: Mechanical behaviour and non-linear analysis of
deep beams using softened truss model. Journal of advanced
materials, 2007, 39 (2), pp. 4864.
25. Moody, K., et al.: Shear Strength of Reinforced Concrete
Beams Part 1 Tests of Simple Beams. ACI Journal Proceedings, 1954, 51 (12).
26. Birrcher, D. B.: Design of reinforced concrete deep beams
for strength and serviceability. PhD thesis, The University of
Texas at Austin, Texas, USA, 2009, p. 391.
27. Brown, M. D., Bayrak, O.: Investigation of Deep Beams with
Various Load Configurations. ACI Structural Journal, 2007,
104 (5), pp. 611620.
28. Clark, A. P.: Diagonal tension in reinforced concrete beams.
ACI Journal Proceedings, 1951, 48 (10).
29. de Paiva, H. A. R., Siess, C. P.: Strength and behavior of
deep beams in shear. Proc. ASCE, 1965, 91, ST5, pp. 1941.
30. Foster, S. J., Gilbert, R. I.: Experimental studies on highstrength concrete deep beams. ACI Structural Journal, 1998,
95 (4).
31. Hara, T.: Shear Strength of Reinforced Concrete Deep
Beams. Transactions of the Japan Concrete Institute, 1984, 6,
pp. 395402.
32. Kong, F. K., Robins, P. J., Cole, D. F.: Web reinforcement effects on deep beams. ACI Journal Proceedings, 1970, 67 (12).
33. Matsuo, M., et al.: Shear Behavior in RC Deep Beams with
Stirrups. Transactions of the Japan Concrete Institute, 2001,
23, pp. 385390.
34. Oh, J. K., Shin, S. W.: Shear strength of reinforced highstrength concrete deep beams. ACI Structural Journal, 2001,
98 (2), pp. 164173.

194

Structural Concrete (2015), No. 2

35. Ramakrishnan, V., Ananthanarayana, Y.: Ultimate strength


of deep beams in shear. ACI Journal Proceedings, 1968, 65
(2).
36. Smith, K., Vantsiotis, A.: Deep beam test results compared
with present building code models. ACI Journal Proceedings,
1982, 79 (4).
37. Subedi, N., Vardy, A. E., Kubotat, N.: Reinforced concrete
deep beams some test results. Magazine of Concrete Research, 1986, 38 (137), pp. 206219.
38. Tan, K., Lu, H.: Shear behavior of large reinforced concrete
deep beams and code comparisons. ACI Structural Journal,
1999, 96 (5).
39. Tan, K.-H., et al.: Main tension steel in high strength concrete deep and short beams. ACI structural journal, 1997, 94
(6).
40. Tan, K.-H., et al.: Effect of web reinforcement on highstrength concrete deep beams. ACI Structural Journal, 1997,
94 (5).
41. Tanimura, Y., Sato, T.: Evaluation of shear strength of deep
beams with stirrups. Quarterly Report of RTRI, 2005, 46 (1),
pp. 5358.
42. Walraven, J., Lehwalter, N.: Size effects in short beams
loaded in shear. ACI Structural Journal, 1994, 91 (5).
43. Bakir, P. G., Boduroglu, H. M.: Mechanical behaviour and
non-linear analysis of short beams using softened truss and
direct strut & tie models. Engineering Structures, 2005, 27
(4), pp. 639651.
44. Ashour, A. F., Alvarez, L. F., Toropov, V. V.: Empirical modelling of shear strength of RC deep beams by genetic programming. Computers & Structures, 2003, 81 (5), pp.
331338.
45. Hwang, S.-J., Lee, H.-J.: Strength prediction for discontinuity
regions by softened strut-and-tie model. Journal of Structural
Engineering, 2002, 128 (12), pp. 15191526.
46. MATLAB Version 7.12.0, T.M., Inc., Natick, MA, 2011.
47. British Standards Institution: BS EN 1992-1-1:2004. Eurocode 2: Design of concrete structures- Part 1-1: General
rules and rules for buildings, 2008.

Wael Kassem
Division of Construction Engineering
Umm Al-Qura University
College of Engineering at Al-Qunfudah
P.O. BOX 288, Al-Qunfudah 21912, Saudi Arabia
Tel. 00966 564 780 673
wakassem@uqu.edu.sa

Vous aimerez peut-être aussi