Engineering Structures 108 (2016) 104–112

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Modeling of the strut-and-tie parameters of deep beams for shear

strength prediction
Erwin Lim, Shyh-Jiann Hwang ⇑
Department of Civil Engineering, National Taiwan University, Taipei, Taiwan, ROC

a r t i c l e i n f o a b s t r a c t

Article history: A generally accepted rational model to predict the shear strength of a structural member needs to satisfy
Received 14 March 2015 Navier’s three principles (force equilibrium, strain compatibility, and constitutive equations). However, a
Revised 26 November 2015 shear model satisfying these three principles usually requires rigorous computational effort. This study
Accepted 27 November 2015
reveals that, if involved structural parameters are properly considered, a simple strut-and-tie model that
Available online 17 December 2015
merely satisfies force equilibrium can give similar accuracy compared to the sophisticated strain-
compatible model. This finding was verified against 118 deep beam specimens tested in the laboratory.
Keywords:
The important structural parameters identified are the definition of a shear element that is consistent
Deep beam
Structural parameter
with force discontinuity, the consideration of elastic behavior in estimating the width of a strut, the
Shear strength prediction dimensions of a nodal zone influenced by a loading plate, and the proper selection of the probable failure
Strut-and-tie model modes.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction stub, the softened strut-and-tie model was further adjusted to

Since its introduction by Schlaich et al. [1], the strut-and-tie
method (STM), which is based on the lower bound theory of
plasticity, has served as a design and analytical tool for structural
members, especially those with complicated flows of forces or
deep members. The STM has been widely used by both engineers
and researchers because the method provides a clear load path
and is simple in terms of both the solution algorithm and the equa-
tions involved. Moreover, a computer program to assist engineers
to analyze and design using STM was developed by Tjhin and
Kuchma [2]. The main implication of the wide acceptance of the
STM is its adoption into the building codes of many countries, such
as Canada [3], the European Union [4], New Zealand [5], and the
United States [6].
By contrast, several analytical models have also been derived.
Zhang and Tan [7] proposed a strut-and-tie model that was based
on the Mohr Coulomb’s failure criterion. Several other researchers
developed strut-and-tie models which satisfied Navier’s three
principles, such as the compatibility-based strut-and-tie that uses
the secant stiffness formulation [8] and the softened strut-and-tie
model [9,10]. For deep beam specimens loaded through a column
⇑ Corresponding author at: Department of Civil Engineering, National Taiwan
University, No. 1, Sec. 4, Roosevelt Road, Taipei 106, Taiwan, ROC.
Tel.: +886 2 33664351; fax: +886 2 2362 2975.
E-mail address: sjhwang@ntu.edu.tw (S.-J. Hwang).
include the effect of boundary condition due to presence of a col-
umn stub [11]. These analytical models allow engineers to produce
one unique solution that satisfies not only force equilibrium but
also strain compatibility and a stress–strain relationship for
cracked reinforced concrete. Some other models are based on a
mechanical approach, such as the two-parameter kinematic theory
[12], which was derived satisfying the kinematic of deep beam’s
deformation. These available models provided reasonable accuracy
to the existing database, but they required rigorous computational
effort.
It is commonly assumed that sophisticated solution algorithms
produce better accuracy when predicting the shear capacity of
deep beam specimens compared to the simple ACI 318 strut-
and-tie equations. This assumption may be a misconception
because not all of the available algorithms use the same macro
model. In this case, the macro model corresponds to the idealized
visualization of the load path and the geometry of the struts and
ties that reflect the major parameters influencing the structural
behavior of a deep beam.
This study begins with the strut-and-tie model (STM) provision
described in ACI 318 [6] in which a direct force transfer mechanism
is assumed. The analysis result indicates that the simple ACI 318
STM provides a too conservative shear strength prediction when
gauged against a group of deep beam databases. A further investi-
gation of this over conservative prediction suggests that it does not
come from the simplicity of the solution algorithm, but rather from

http://dx.doi.org/10.1016/j.engstruct.2015.11.024
0141-0296/Ó 2015 Elsevier Ltd. All rights reserved.
 E. Lim, S.-J. Hwang / Engineering Structures 108 (2016) 104–112
the improper modeling of the structural parameters inherent
within the macro model. Therefore, the original macro model of
ACI 318 is modified to include a structural behavior that represents
the testing conditions and behavior of the deep beam specimens
tested in the laboratory as well as the proper justification of the
most probable failure modes. The results of the final modification
of the model indicate that when all of these parameters are prop-
erly considered, the simple ACI 318 strut-and-tie equations pro-
vide a similar accuracy compared to the softened strut-and-tie
model (SST). Meanwhile, further discussions related to the differ-
ences between ACI 318 strut-and-tie and SST model can be seen
elsewhere [13].
2. Strut-and-tie model based on ACI 318
ACI 318 STM is presented in a simple form, as shown in Fig. 1.
Once all of the beam dimensions, reinforcement detailing, material
properties, and testing parameters are known, one may develop
any macro model of a strut-and-tie that satisfies force equilibrium.
As given by Eq. (1), the calculated shear strength of a deep beam
(Vn) is determined as being the smallest of the following: the
strengths of the diagonal strut at the top (V1) and at the bottom
(V2), the strengths of the nodal zone at the top (V3) and at the bot-
tom (V4), and the yielding of tension ties (V5):
V n ¼ minfV 1 ; V 2 ; V 3 ; V 4 ; V 5 g ð1Þ
The strengths of the struts are empirically determined using
strut efficiency factor 0.85 bs, depending on the amount of the ver-
tical and horizontal shear reinforcements crossing the strut [6]. The
strengths of the nodal zone at the upper part (V3) and the lower
part (V4) are determined using the bn factor for the CCC (resists
three compressive forces) and CCT (resists two compressive forces
and one tensile force) nodes, respectively. Meanwhile, the strength
of the tension tie (V5) is taken as the yielding strength of the flex-
ural reinforcement.
2.1. Macro model
Although a STM provides a clear force transfer mechanism of a
modeled structure or a region, it does not rigidly specify how the
load is transferred from the loading point (actuator) to the support.
One may use a direct STM in which the load is transferred directly
from the loading plate to the reaction plate or other truss models to
consider the additional load paths due to the presence of vertical
stirrups. The simplest STM, also adopted in this paper, uses a direct
transfer mechanism in which the force from the loading actuator is
directly transferred to the support reaction (Fig. 1). Brown and
Bayrak [14] also concluded that the direct transfer mechanism
might be considered as an appropriate mechanism, especially for
deep beams with a shear span to depth ratio that is less than 2.
After determining the load path, one must define the macro
model of the strut-and-tie, which includes the determination of
the width of the horizontal strut ws, the width of the horizontal
tie wt, and the inclination angle of the diagonal strut h. The afore-
mentioned width of the horizontal strut represents the depth of
the compression zone at the constant bending moment region.
According to Tjhin and Kuchma [15], the depth of the compression
zone is taken as the plastic compression zone presented in Eq. (2):
As f y
ws ¼ 0 ð2Þ
0:85bs f c b
where Asfy is the yield strength of the main flexural reinforcement,
bs is a factor used to account for the effect of cracking on the effec-
0
tive compressive strength of a concrete strut given by ACI 318-14, f c
is the compressive strength of concrete, and b is the width of the
105
beam. In addition, the width of the horizontal tie is calculated by
following the recommendation in ACI 318-14, as presented in
Eq. (3):
As f y
wt ¼ 0 ð3Þ
0:85bn f c b
where bn in this case is the factor used to account for the effect of
the anchorage ties on the effective compressive strength of a nodal
zone taken at the lower part (CCT node).
The inclination angle relative to the horizontal axis h of this
concrete strut is given by Eq. (4):



jd d  ws =2
h ¼ tan1 ¼ tan1 ð4Þ
a ‘b =2 þ a0 þ ap =2
where jd is the force lever arm, a is the shear span, d is the effective
depth of the beam, ‘b is the width of the bearing plate, ap is the
width of the loading plate, and a0 is the clear shear span.Finally,
after these three parameters are determined, the geometry of the
nodal zone and the strut area at the top and bottom parts of the
diagonal strut can be determined, as illustrated in Fig. 2(a). The
nodal zone areas (Acn;top ; Acn;bot ), which are the same as the strut
areas (Acs;top ; Acs;bot ), are defined such that they are perpendicular
to the strut:
Acs;top ¼ Acn;top ¼ ðws cos h þ ap sin hÞb ð5Þ
Acs;bot ¼ Acn;bot ¼ ðwt cos h þ ‘b sin hÞb ð6Þ
2.2. Verification of ACI 318 STM
To verify the accuracy of the STM of ACI 318 (Analysis 1), a data-
base of the deep beam specimens that failed in shear was collected
from the literature [16–21] and is presented in Table 1. The
specimens used in this study were collected from the available
literature, ensuring that the complete information of the test setup
was provided. The database covers a wide range of concrete
0
compressive strengths f c along with different layouts of
reinforcements.
The shear strength ratios (V test =V calc ) are plotted in Fig. 2(b),
with each specimen represented by a number to indicate the pre-
dicted failure mode. The STM defined using the ACI 318 parameters
predicts that the majority of specimens would fail in the upper part
of the strut (failure mode 1). The STM also provides a very conser-
vative and scattered strength prediction, as indicated by its aver-
age value of 1.54, and a coefficient of variation (COV) of 0.41.
Fig. 2(b) also shows that the strength ratios for Kong et al.’s speci-
mens are rather poor compared to others. This finding is because
majority of Kong et al.’s specimens were detailed with small
amount of longitudinal reinforcement ratio (q 6 0:8%). A further
look into this finding suggests that the accuracy of the current
ACI 318 STM is not so reliable for specimens with relatively low
ratio of longitudinal reinforcement (Fig. 3). Therefore, the authors
argue that in order to improve the accuracy of the current ACI
318 STM (Analysis 1), the macro model should reflect the struc-
tural behavior of deep beam specimens tested in the laboratory.
In the following section, a series of parametric analyses, with
each analysis representing a macro model closer to the structural
behavior of deep beam specimens, is performed. These analyses
include the effect of force discontinuity on the appropriate determi-
nation of the shear element (Analysis 2), the selection of the com-
pression zone to represent the width of the strut (Analysis 3), the
influence of the steel loading plate on the dimension of the nodal
zone (Analysis 4), and the selection of the probable failure mode
(Analysis 5). In addition, using the same macro model as that used
in the latest analysis (Analysis 5), this research also verifies the
106 E. Lim, S.-J. Hwang / Engineering Structures 108 (2016) 104–112

Fig. 1. Flow chart for the ACI 318-14 strut-and-tie model [6].

applicability of a sophisticated strain-compatible model, i.e., the
softened strut-and-tie model, as described in Analysis 6.
3. Parametric modeling
3.1. Consideration of the force discontinuity (Analysis 2)
Because perfect bonding does not exist between the steel plate
and the concrete surface, ACI 318-14 indicates that the D-region of
the deep beam specimens tested using two-point loading is
bounded by the discontinuity of force. In the original ACI 318
model (Analysis 1), the geometry of the nodal zone at the upper
part of the diagonal strut is determined using the entire width of
the loading plate ap (Eq. (5)). However, taking the entire width of
the loading plate may not be appropriate.
Due to the load is transferred through the inclined force,
the shear strength should be determined by the nodal/strut
properties associated with the inclined force diagonal. Considering
 E. Lim, S.-J. Hwang / Engineering Structures 108 (2016) 104–112 107

Fig. 2. Macro model for ACI 318-14 and its experimental verification.

Table 1
Deep beam database.

Authors
Specimen number: 1–4
Authors: de Paiva and Siess [16]
Specimen number: 5–39
Authors: Kong et al. [17]
Specimen number: 40–91
Authors: Smith and Vantsiotis [18]
Specimen number: 92–93
Authors: Aguilar et al. [19]
Specimen number: 94–114
Authors: Yang et al. [20]
Specimen number: 115–118
Authors: Alcocer and Uribe [21]
a
Specimens reinforced with neither vertical nor horizontal shear reinforcement.
Specimen ID
G33S-11a, G33S-31a, G332-12, G332-32
1-30, 1-25, 1-20, 1-15, 1-10, 2-30, 2-25, 2-20, 2-15, 2-10, 3-30, 3-25, 3-20, 3-15, 3-10, 4-30, 4-25, 4-20, 4-15, 4-10, 5-30, 5-25,
5-20, 5-15, 5-10, 6-30, 6-25, 6-20, 6-15, 6-10, 7-30Aa, 7-30B, 7-30C, 7-30D, 7-30E
0A0-44a, 0A0-48a, 0B0-49a, 0C0-50a, 0D0-47a, 1A1-10, 1A3-11, 1A4-12, 1A4-51, 1A6-37, 2A1-38, 2A3-39, 2A4-40, 2A6-41, 3A1-
42, 3A3-43, 3A4-45, 3A6-46, 1B1-01, 1B3-29, 1B4-30, 1B6-31, 2B1-05, 2B3-06, 2B4-07, 2B4-52, 2B6-32, 3B1-08, 3B1-36, 3B3-
33, 3B4-34, 3B6-35, 4B1-09, 1C1-14, 1C3-02, 1C4-15, 1C6-16, 2C1-17, 2C3-03, 2C3-27, 2C4-18, 2C6-19, 3C1-20, 3C3-21, 3C4-
22, 3C6-23, 4C1-24, 4C3-04, 4C3-28, 4C4-25, 4C6-26, 4D1-13
STM-H, STM-M
L5-40a, L5-60a, L5-60Ra, L5-75a, L5-100a, L10-40a, L10-40Ra, L10-60a, L10-75a, L10-75Ra, L10-100a, UH5-40a, UH5-60a, UH5-75a,
UH5-100a, UH10-40a, UH10-40Ra, UH10-60a, UH10-75a, UH10-75Ra, UH10-100a
MT, MR, CT, CR

that the actuator is applied at the middle of the steel plate at the
upper part (Fig. 4(a)), half of the plate’s width (ap =2) to the left of
the actuator belongs to the constant moment (zero shear) region.
In a constant moment region, only horizontal concrete stresses
can be found and no shear stress exists. Therefore, the left half of
the bearing plate should have no influence on the nodal/strut
dimension. Only the other half of plate to the right of the actuator,
which is located within the shear transfer region, is considered to
be effective as the strut area. Consequently, the area of the upper
strut is adjusted to:
 ap 
Acs;top ¼ ws cos h þ sin h b ð7Þ
2
Meanwhile, at the lower node, where the moment is zero, the
stress is consequently more uniformly distributed. Therefore, the
width of strut area at the support region remains unchanged (Eq.
(6)).
Analysis 2 predicts that all specimens failed due to the crushing
of the upper strut and indicates that its accuracy is reduced com-
pared to Analysis 1, as indicated in Fig. 4(b). For the majority of
specimens tested by Kong et al. [17], the strength predictions are
scattered and too conservative. The average shear strength ratio
of the collected 118 specimens increases up to 2.12, with a COV
of 0.56. This result suggests that the geometrical modeling of the
strut area at the upper part still neglects some important consider-
ations. In the two subsequent parametric studies (Analysis 3 and
4), the authors suggest two important factors should be considered
when modeling the width of the strut and the dimensions of the
nodal zone in the upper part. These factors are the consideration
of the elastic behavior of a deep beam and the effect of a steel load-
ing plate.

3.2. Consideration of the compression zone (Analysis 3)

Previous analysis (Analysis 1) suggested that the shear strength

calculated using ACI 318 STM gave large discrepancies, especially
for specimens with q 6 0:8%. Part of the reason is because the cal-
culated depth of concrete strut was very small (Eq. (2)), resulting in
Fig. 3. Accuracy of ACI 318 STM plotted against longitudinal reinforcement ratio. the strength of diagonal concrete strut at top V1 governed for most
108 E. Lim, S.-J. Hwang / Engineering Structures 108 (2016) 104–112

Fig. 4. Parametric models and experimental verifications.

of those specimens. Moreover, the change indicated in Analysis 2 In Analysis 3, the width of the horizontal strut is taken as the
showed poor accuracy and scattered predictions of the shear elastic compression zone kd, as given by Eq. (8):
strength. One of the possible explanations of these results is the
ws ¼ kd ð8Þ
inappropriate use of plastic compression zone depth as the width
of the horizontal strut ws. In all of the collected specimens, the where k is derived from a singly reinforced beam section:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
flexural reinforcement was detailed such that the shear capacity
k ¼ ðnqÞ2 þ 2nq  nq, and n is the elastic modulus ratio of steel
was reached prior to flexural failure, as reported by the test results.
to concrete.
In addition, a simple conventional flexural analysis indicates that
A similar method for choosing the elastic compression zone kd
the majority of collected specimens failed before reaching their
has been used by other researchers [8]. Following this modifica-
nominal flexural strength Mn which is defined as the bending
tion, the macro model of the STM is re-defined as shown in Fig. 4
moment calculated when the strain in the extreme concrete com-
(c) and the strut inclination angle changes to:
pression fiber reaches 0.003. As a consequence, the use of plastic



compression zone may underestimate the width of horizontal strut jd d  kd=3
since the concrete in flexure is still primarily in the elastic range.
h ¼ tan1 ¼ tan1 ð9Þ
a a
 E. Lim, S.-J. Hwang / Engineering Structures 108 (2016) 104–112 109

The strut area at the upper part is modified, as given by Eq. (10): strut at the lower part (V2 and V4) is less likely to govern the overall
  behavior of a deep beam specimen.
ap In addition, the flexural capacity of a beam is more appropri-
Acs;top ¼ kd cos h þ sin h b ð10Þ
2 ately represented using its nominal flexural strength (Mn) rather
The average shear strength ratio becomes 1.63, with a COV of than its yielding strength (V5). As a result, in the following analysis
0.33, as shown in Fig. 4(d). Although the average shear strength (Analysis 5), failure modes 2, 4, and 5 are excluded from the gov-
ratio is still too conservative, the consideration of the elastic com- erning strength in determining the shear capacity of a deep beam.
pression zone to model the geometry of the strut area at the upper
part slightly improves the accuracy of the strength prediction. 3.4. Consideration of the most probable failure mode (Analysis 5)
The main reason for this over-conservatism can be argued as
being the result of neglecting the beneficial effect of the loading Analysis 5 used the same macro model as the one described in
plate. The presence of a steel loading plate should smooth out Analysis 4. However, these models differ in the selection of the
the spread of force and enlarge the dimensions of the nodal zone. probable controlling failure mode for the majority of deep beam
This effect is considered in the following analysis (Analysis 4). specimens. As mentioned before, the presence of a plate to anchor
the longitudinal flexural reinforcement bars can help widen the
strut area and spread out the stress at the lower node. In addition,
3.3. Consideration of the effect of a steel loading plate (Analysis 4)
the moment diagram also suggests a zero moment at the lower
part (support reaction). This observation implies that the stress
The main underlying concept for the redefinition of the macro
concentration at the lower part (CCT node) is relatively small com-
model of the strut-and-tie in Analysis 4 is the consideration of
pared to that at the upper part (CCC node) and is therefore less crit-
how the load spreads throughout the steel plate into the beam
ical. Hence, in Analysis 5, failure modes 2 and 4 are excluded.
specimen. As the load is applied from the actuator, a very high
In addition, to represent a beam’s flexural capacity, it is com-
stress concentration occurs at the nodal zone beneath it. This stress
monly accepted that shear corresponding to the nominal flexural
concentration would spread throughout the steel plate to the beam
strength (Vf) is more representative than shear corresponding to
specimen and reach a more uniform stress at the middle of the
the first yielding moment (V5); the value of Vf is calculated using:
diagonal strut. Hence, a proper redefinition of the nodal zone at
the upper and lower parts of the beam with consideration of the V f ¼ Mn =a ð13Þ
spreading of the force may play an important role because failure
Finally, the strength of a simply supported deep beam in Anal-
may occur at these points. The significance of the loading plate’s
ysis 5 is determined by either the crushing of the strut and the
size in affecting the geometry of strut and nodal zone was also rec-
nodal zone at the upper part or the shear corresponding to the
ognized by Zhang and Tan [22].
nominal flexural strength Vf, as given by Eq. (14):
In Analysis 4, the authors assume that the force spreads out
with a ratio of 1:2 (vertical:horizontal). A similar ratio can also Vn ¼ minfV 1 ; V 3 ; V f g ð14Þ
be obtained from the principle stress trajectory using finite ele-
ment analysis for a deep beam [23]. The strut area is defined at The shear strength prediction of 118 specimens using Analysis 5
the location where the spreading of the vertical load reaches the provides an average strength ratio of 1.28 with a COV of 0.19
resultant horizontal compressive force C, as illustrated in Fig. 4 (Fig. 5).
(e). Correspondingly, a larger area of concrete below the loading As observed in Fig. 5, a quite scattered strength prediction result
plate participates in resisting the concrete crushing. The strut area was produced for the 29 specimens tested by Yang et al. [20]. The
is then defined accordingly: average strength ratio for these 29 specimens is 1.06, with a COV of
0.30. The main reason for this result is argued to be the absence of

   
2kd ap  kd vertical and horizontal shear reinforcements within the deep
Acs;top ¼ b cos h þ b min 2t p ; þ2 sin h ð11Þ beam. The absence of these reinforcements could possibly cause
3 2 3
premature failure of the specimen prior to the mature develop-
Similarly, the strut area at the lower part is: ment of the strut-and-tie mechanism. Thus, with the exclusion of
these 29 specimens, the average strength ratio given by Analysis
Acs;bot ¼ b½wt  ðh  dÞ cos h 5 for the remaining 89 specimens is 1.34, with a COV of 0.14.

 
‘b ‘b
þ b min 2t p ; þ 2ðh  dÞ þ sin h ð12Þ
2 2 4. Softened strut-and-tie (SST) model
The accuracy of the shear strength prediction using Analysis 4 is
found to be greatly improved over the previous versions of the The softened strut-and-tie (SST) model is considered one of sev-
models, with an average value of 1.35 and a COV of 0.21, as indi- eral available strain-compatible analytical models. This model
cated in Fig. 4(f).
The failure mode prediction of Analysis 4 indicates that approx-
imately one-third of the collected 118 specimens failed due to the
crushing of concrete at the lower strut (failure mode 2). Examina-
tion of the details of these specimens indicates that the longitudi-
nal flexural reinforcement bars were properly anchored to the steel
plate [17,20]. With the presence of the anchorage plate, the stress
can be more uniformly distributed toward the location of the
anchorage plate. In reality, this beneficial effect should be consid-
ered in the modeling of the strut area at the lower node; for exam-
ple, for the last term of Eq. (12), ‘b =2 could be further increased by
considering a spreading of the reaction toward the anchorage plate
and the corresponding widening of the strut area. However, for
simplicity, the authors assume that the crushing of the diagonal Fig. 5. Experimental verification of Analysis 5.
110 E. Lim, S.-J. Hwang / Engineering Structures 108 (2016) 104–112

determines the shear strength of a deep beam by calculating the
crushing strength of concrete at the upper strut, adjacent to the
nodal zone. The concrete crushing strength VSST, is evaluated using
the softened behavior of the reinforced concrete by considering the
force equilibrium and strain compatibility within the shear
element, as shown in Fig. 6.
4.1. Macro model
The macro model of SST starts with the determination of the
size of the shear element represented by the strut inclination
angle h, which is bounded by the moment couple arm and the
shear span in the vertical and horizontal dimensions. The same

Fig. 6. Solution procedure of the softened strut-and-tie model [9,10].

 E. Lim, S.-J. Hwang / Engineering Structures 108 (2016) 104–112 111

macro model as that used in Analysis 5 (Fig. 4(e)) is also used in

this SST analysis.

4.2. Force distribution

This model assumes that there are three possible mechanisms

to transfer the load from the loading plate to the bearing plate
within the D-region, which includes diagonal, horizontal, and ver-
tical mechanisms, as illustrated in Fig. 6. In the absence of vertical
and horizontal shear reinforcements, the load transfer passes
mainly through the main diagonal concrete strut (D). Vertical
and/or horizontal shear reinforcement can act as vertical and Fig. 7. Experimental verification of Analysis 6.
horizontal tension ties, represented by Fv and Fh, respectively. With
the presence of these tension ties, additional load paths may be
created to include more concrete through the vertical and/or
horizontal sub-struts indicated by the dashed line in Fig. 6. The Mean= 1.12
force distributed among these struts and ties follow the elastic COV = 0.10
spring concept and are always in equilibrium at the nodes [9,10].

4.3. Constitutive laws and strain compatibility

Having determined the cross-sectional area and the distributed

force of each strut and tie element, one can calculate the strains in
the horizontal tie eh and vertical tie ev using an elastic perfectly
plastic stress–strain relationship for steel. The crushing strength
of the diagonal strut can be obtained through the iterative process
[9,10] using the softened reinforced concrete stress–strain rela-
tionship [24], as described in Eqs. (15) and (16):
"

2 #
ed ed ed Fig. 8. Comparison of the strut efficiency factor and the softening coefficient.
rd ¼ ff 0c 2  for 61 ð15Þ
feo feo feo

5:8 1 0:9
f ¼ qffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 6 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
0
fc 1 þ 400 e r 1 þ 400er
where rd is the average principal compressive stress of concrete, f is
the softening coefficient, eo is the concrete cylinder strain corre-
sponding to the cylinder strength f c , ed is the principal compressive
0 and the softening coefficient f of SST is 1.12 with a COV as small
strain of concrete, and er is the principal tensile strain of concrete.
In addition, strain compatibility must be maintained through-
out the solution using Eq. (17):
er ¼ eh þ ev  ed
The solution procedure of this softened strut-and-tie model
algorithm is shown in the flow chart in Fig. 6
4.4. Consideration of a strain-compatible rational model (Analysis 6)
Using the macro model described in Fig. 4(e), the softened strut-
and-tie model is used in Analysis 6 to calculate the shear strength
of the collected deep beam specimens. Because the SST solution
algorithm only calculates the crushing strength of a strut at the
most critical location, namely, at the upper strut, the governing
strength is determined by:
V n ¼ minfV SST ; V f g
The result, as shown in Fig. 7, indicates that the average shear
strength ratio is 1.27, with a COV of 0.16. The macro model pro-
posed in this research is also able to appropriately model the pres-
ence of a wide loading plate used in specimens MT, MR, CT, and CT
tested by Alcocer and Uribe [21]. The calculation using the original
macro model of the SST provided an average shear strength ratio as
low as 0.75 [21]. However, by using the macro model proposed in
this research (Fig. 4(e)), the contribution of the loading plate can be
properly attributed and resulted in the average shear strength ratio
ð16Þ
for these four specimens becomes 0.92.
A further look into the behavior of cracked reinforced concrete
is shown in Fig. 8. Fig. 8 suggests that the average ratio between
empirically determined strut efficiency factors 0.85 bs of ACI 318
as 0.10. It implies that the strut efficiency factor 0.85 bs could rea-
sonably capture the strain field represented by the softening coef-
ficient f within the shear element. This finding is also consistent
with the report by Quintero-Febres et al. [25], which concluded
ð17Þ that the strut efficiency factor recommended by ACI 318 is reason-
able for normal strength concrete.
Although many assumptions were made in the derivation of the
macro model, the results of Analysis 5 and 6 suggest that the ACI
318 STM provisions to calculate the strength of a diagonal concrete
strut provides similar accuracy compared to a rigorous SST proce-
dure, as long as the appropriate structural parameters are consid-
ered in the macro model. This study emphasizes the importance
of recognizing crucial structural parameters in the modeling.
5. Conclusions
A series of parametric studies were evaluated to determine the
most suitable macro model to reflect the shear strength behavior of
ð18Þ
a deep beam. Using data collected from 118 deep beam specimens,
this study indicated that:
1. The macro model should reflect the structural behavior of a
deep beam.
2. The most important structural parameters used to model a deep
beam specimen tested in the laboratory are the definition of a
shear element that is consistent with the force discontinuity,
112 E. Lim, S.-J. Hwang / Engineering Structures 108 (2016) 104–112
the consideration of the elastic behavior in defining the width of
a strut, the effect of a steel loading plate on the spreading of the
force, and the proper selection of the probable failure modes.
3. Using the same macro model, the simple and straightforward
solution procedure of ACI 318-14 STM yields agreeable accuracy
compared with the sophisticated and rigorous procedure of a
strain-compatible model, such as the softened strut-and-tie
model.
Acknowledgments
We express our gratitude for the support funds received from
the National Taiwan University, the Ministry of Science and
Technology of Taiwan, and the National Center for Research on
Earthquake Engineering (NCREE) throughout this study.
References
[1] Schlaich J, Schäfer K, Jennewein M. Toward a consistent design of structural
concrete. PCI J 1987;32(3):74–150.
[2] Tjhin TN, Kuchma DA. Integrated analysis and design tool for the strut-and-tie
method. Eng Struct 2007;29:3042–52.
[3] CSA Committee A23.3. Design of concrete structures. Mississauga: Canadian
Standard Association; 2004.
[4] Comite Euro-International du Beton (CEB) – Federation International de la
Precontrainte (FIP). CEB-FIP Model Code 1990. London: Thomas Telford; 1993.
[5] NZS 3101:2006. Concrete structures standard, Part 1 – the design of concrete
structures. Wellington: Standards Association of New Zealand; 2006.
[6] ACI Committee 318. Building code requirements for structural concrete (ACI
318-14) and Commentary (318R-14). Farmington Hills: American Concrete
Institute; 2014.
[7] Zhang N, Tan KH. Direct strut-and-tie model for single span and continuous
deep beams. Eng Struct 2007;29:2987–3001.
[8] Park JW, Kuchma D. Strut-and-tie model analysis for strength prediction of
deep beams. ACI Struct J 2007;104(6):657–66.
[9] Hwang SJ, Lee HJ. Analytical model for predicting shear strength of exterior
reinforced concrete beam-column joints for seismic resistance. ACI Struct J
1999;96(5):846–58.
[10] Hwang SJ, Lu WY, Lee HJ. Shear strength prediction for deep beams. ACI Struct J
2000;97(3):367–76.
[11] Lu WY, Lin IJ, Yu HW. Shear strength of reinforced concrete deep beams. ACI
Struct J 2013;110(4):671–80.
[12] Mihaylov BI, Bentz EC, Collins MP. Two-parameter kinematic theory for shear
behavior of deep beams. ACI Struct J 2013;110(3):447–56.
[13] Kuo WW, Cheng TJ, Hwang SJ. Force transfer mechanism and shear strength of
reinforced concrete beams. Eng Struct 2010;32:1537–46.
[14] Brown MD, Bayrak O. Design of deep beams using strut-and-tie models – Part
I: evaluating U.S. provisions. ACI Struct J 2008;105(4):395–404.
[15] Tjhin TN, Kuchma DA. Example 1b: alternative design for non-slender beam
(deep beam). ACI SP-208. Farmington Hills: American Concrete Institute; 2002.
[16] de Paiva HAR, Siess CP. Strength and behavior of deep beams in shear. J Struct
Eng 1965;91(ST5):19–41.
[17] Kong FK, Robins PJ, Cole DF. Web reinforcement effects on deep beams. ACI J
1970;67(12):1010–7.
[18] Smith KN, Vantsiotis AS. Shear strength of deep beams. ACI J 1982;79
(3):201–13.
[19] Aguilar G, Matamoros AB, Parra-Montesinos GJ, Ramires JA, Wight JK.
Experimental evaluation of design procedures for shear strength of deep
reinforced concrete beams. ACI Struct J 2002;99(4):539–48.
[20] Yang KH, Chung HS, Lee ET, Eun HC. Shear characteristics of high-strength
concrete deep beams without shear reinforcements. Eng Struct
2003;25:1343–52.
[21] Alcocer SM, Uribe CM. Monolithic and cyclic behavior of deep beams designed
using strut-and-tie models. ACI Struct J 2008;105(3):327–37.
[22] Zhang N, Tan KH. Size effect in RC deep beams: experimental investigation and
STM verification. Eng Struct 2007;29:3241–54.
[23] Cook WD, Mitchell D. Studies of disturbed regions near discontinuities in
reinforced concrete members. ACI Struct J 1988;85(2):206–16.
[24] Zhang LXB, Hsu TTC. Behavior and analysis of 100 MPa concrete membrane
elements. J Struct Eng 1998;124(1):24–34.
[25] Quintero-Febres CG, Parra-Montesinos G, Wight JK. Strength of struts in deep
concrete members designed using strut-and-tie method. ACI Struct J 2006;103
(4):577–86.