ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 97-S40

Shear Strength Prediction for Deep Beams

by Shyh-Jiann Hwang, Wen-Yao Lu, and Hung-Jen Lee
A softened strut-and-tie model for determining the shear strengths
of deep beams is proposed in this paper. The proposed model origi-
nates from the strut-and-tie concept and satisfies equilibrium, com-
patibility, and constitutive laws of cracked reinforced concrete. The
shear strength predictions of the proposed model and the empirical
formulas of the ACI 318-95 Code are compared with the collected
experimental data of 123 deep beams. The comparison shows that
the performance of the softened strut-and-tie model is better than the
ACI Code approach for all the parameters under comparison. The
parameters reviewed include the ratios of horizontal and vertical Fig. 1 Typical shear failures in deep beams.
reinforcement, concrete strength, and the shear span-depth ratio. The
softened strut-and-tie model can be further developed to improve the
current deep beam design procedures by incorporating the actual tion. In this disturbed region, it is inappropriate to assume
shear resisting mechanisms in predicting shear strength supply of that the shear stress is uniform. Therefore, the strut-and-tie
deep beams. model is believed to be a better choice in modeling the flow of
the forces of the deep beam, with compressive struts repre-
Keywords: deep beams; reinforced concrete; shear strength. senting the flow of concentrated compressive stresses in the
concrete and tension ties representing the reinforcing steel.
INTRODUCTION The purpose of this paper is to present a softened strut-
Reinforced concrete deep beams with a shear span-depth and-tie model for determining the shear strengths of rein-
ratio less than 2.51 find wide applications in tall building con- forced concrete deep beams. A similar model for the shear
struction, offshore structures, and complex foundation sys- strength predictions of beam-column joints and squat walls
tems. The strength of deep beams is usually controlled by has been proposed.5-7 This model is based on the strut-and-tie
shear rather than flexure, provided a normal amount of longi- concept and is derived to satisfy equilibrium, compatibility,
tudinal reinforcement is used. An understanding of the shear and constitutive laws of cracked reinforced concrete. The
strength behavior of deep beams is an essential prerequisite word softened emphasizes the importance of the compression
for achieving optimum design and proportioning of such softening phenomenon, which means that cracked reinforced
members. concrete in compression exhibits lower strength and stiffness
The shear action in the beam web leads to compression in than uniaxially compressed concrete.8 It is believed that the
a diagonal direction and tension in a direction perpendicular shear failure relating to concrete crushing should be governed
thereto. A tied arch action is usually recognized as the force by the softening effect of concrete.
transferring mechanism of deep beams, and the possible fail- The theory presented in this paper is an extension of the
ure mode may be the web compression failure (crushing of softened strut-and-tie model that specifically predicts shear
concrete) or the shear tension failure (loss of beam bar strength of deep beams associated with web compression fail-
anchorage), as shown in Fig. 1. By carefully detailing the end ure. In the following, the model and its theory will first be
anchorage of beam bars, shear tension failure can be efficient- presented and developed. The validity and accuracy of the
ly avoided. In safeguarding against web compression failure, proposed model is then tested against available experimental
the ACI 318-95 Code2 adopts empirical formulas derived results. Data from a total of 123 single-span top-loaded deep
using test results. beams were compiled and compared with predictions from the
The current ACI 318-95 Code2 design criterion for the proposed model and the ACI 318-95 Code.
shear strength of deep beams is known to be conservative
with a large safety margin.3,4 Such conservatism is often RESEARCH SIGNIFICANCE
unavoidable when the contribution of the various factors to This paper presents a new treatment for the prediction of
the shear strength is not explicitly derived from a clear failure shear strength in deep beams. The theoretical model consists
mechanism. Although the current code formulas have identi- of the application of a softened strut-and-tie model. The pro-
fied all of the influencing factors such as the strength of con- posed method adequately predicts the shear strength of deep
crete and steel, the longitudinal and transverse steel areas, beams of different concrete strengths, shear span-depth ratios,
and the shear span-depth ratio, the accuracy of these code for- and horizontal and vertical web reinforcement.
mulas is not satisfactory. 3,4 Clearly, design methods based on
a fundamental understanding of the shear strength behavior SOFTENED STRUT-AND-TIE MODEL
of deep beams would be preferable to the present procedures. Before formulating a mathematical model, the forces
To predict the shear strength of deep beams, the softened around and within a deep beam should be identified. Consider
truss model was developed by Mau and Hsu.3 In the softened a typical deep beam of a rectangular cross section loaded on
truss model, the state of stresses in the web shear element is
assumed to be uniform and the flow of compressive stresses is ACI Structural Journal, V. 97, No. 3. May-June 2000.
MS No. 98-277 received December 30, 1998, and reviewed under Institute pub-
idealized by a series of parallel compressive struts. The inter- lication policies. Copyright ' 2000, American Concrete Institute. All rights
nal stress flow of the beam web, however, is highly disturbed reserved, including the making of copies unless permission is obtained from the
copyright proprietors. Pertinent discussion will be published in the March-April
by the presence of the top load and the bottom support reac- 2001 ACI Structural Journal if received by November 1, 2000.

ACI Structural Journal/May-June 2000 367

ACI member Shyh-Jiann Hwang is a professor of construction engineering at the National
Taiwan University of Science and Technology, Taipei, Taiwan, Republic of China. He
received his MS and PhD from the University of California at Berkeley. His research inter-
ests include seismic behavior of beam-column joints, shear strengths of reinforced concrete
members, and bond and anchorage.

Wen-Yao Lu is an instructor of civil engineering at Chung Kuo Insititute of Technology

and Commerce, Taipei, Taiwan, Republic of China. He received his MS from the National
Taiwan University of Science and Technology. His research interests include the probability
of reinforced concrete members and shear behavior of nonflexural members.

Hung-Jen Lee is a PhD candidate in construction engineering at the National Taiwan

University of Science and Technology. His research interests include behavior and design of
RC beam-column joints.

Fig. 2 Forces in web of deep beams.

top and simply supported at the bottom, as shown in Fig. 2.

By taking into account the distances between couples (Fig. 2),
it will be sufficiently accurate to express the following rela-
tionship between vertical and horizontal shears

Vbv jd
≈ (1)
Vbh a

where Vbv and Vbh are the vertical and horizontal beam shear
forces, respectively; jd is the distance of the lever arm from the
resultant compressive force to the centroid of the longitudinal
tension reinforcement; and a is the shear span measured cen-
ter-to-center from load to support. The horizontal shear force Fig. 3 Shear resisting mechanisms of deep beams.
Vbh is equal to the resultant tensile force T or the resultant
compressive force C at the section due to flexure (Fig. 2). using the elastic bending theory in Eq. (2) and (3) is to simu-
According to linear bending theory, the lever arm jd of a late the situation where the deep beam fails in shear but keeps
doubly reinforced rectangular section can be estimated as a reserve of flexural capacity. This situation is quite common
for the laboratory deep beam specimen.
Statically indeterminate strut-and-tie load paths are pro-
kd
jd = d − (2) posed to model the force transference within the web of the
3 deep beam. The web compression failure is defined as the
crushing of concrete within the inclined strut that forms
between the load point and the support.
where d is the effective depth of the deep beam; kd is the depth
of the compression zone at the section; and coefficient k can be Macromodel
defined as9 After the development of the first cracking pattern in the
deep beam web, the steel bars are subjected to tension and the
concrete acts as the compressive struts, thus forming a strut-
and-tie action. Figure 3 shows the proposed softened strut-
k = [nρ + (n − 1) ρ ′]2 + 2[nρ + (n − 1) ρ ′d ′ / d ]
and-tie model, which is composed of diagonal, horizontal, and
− [nρ + (n − 1) ρ ′] (3) vertical mechanisms. The diagonal mechanism (Fig. 3(a)) is a
diagonal compression strut whose angle of inclination θ is
defined as
where n is the modular ratio of elasticity; ρ is the ratio of the
tension reinforcement; ρ’ is the ratio of the compression rein-
forcement; and d’ is the distance from the extreme compres-  jd 
θ = tan −1   (4)
sion fiber to the centroid of the compression reinforcement.  a 
Since the effect of the horizontal web reinforcement is
neglected, Eq. (2) and (3) are approximate estimations for the
internal lever arm and the neutral axis depth. The reason for The direction of the principal compressive stress of the

368 ACI Structural Journal/May-June 2000

concrete is also assumed to coincide with the direction of the
diagonal concrete strut.
The effective area of the diagonal strut Astr is defined as

Astr = as × bs (5)

where as is the depth of the diagonal strut; and bs is the width

of the diagonal strut that can be taken as the width of the
beam web.
The depth of the diagonal strut (as) depends on its end con-
dition, which is provided by the compression zone at the sec-
tion and the bearing block. It is intuitively assumed that

as = (kd )2 + ab2 (6)

where ab is the depth of the bearing block.

The horizontal mechanism (Fig. 3(b)) consists of one hor-
izontal tie and two flat struts. The horizontal tie is made up of
the horizontal web reinforcement. When computing the cross
area of the horizontal tie, it is roughly assumed that the hori- Fig. 4 S t rut-and-tie model for a deep beam.
zontal web reinforcement within the center half of the web is
fully effective, and the other horizontal web steel is included
as 50% effective.5-7 where Rd, Rh, and Rv are the beam shear ratios resisted by the
The vertical mechanism (Fig. 3(c)) includes one vertical diagonal, horizontal, and vertical mechanisms, respectively
tie and two steep struts. The vertical web reinforcement con- (Fig. 3). Equation (9) can be rearranged as
stitutes the vertical tie. The way to estimate the cross-sec-
tional area of the vertical tie is similar to that of the horizon- −1 Rd
tal tie. D= × × Vbv
The proposed model is a statically indeterminate system. sin θ (Rd + Rh + Rv ) (10)
Deep beams without horizontal or vertical web reinforcement
can resist shears using reduced mechanisms, namely, diagonal
plus vertical mechanisms, diagonal plus horizontal mecha- 1 Rh
Fh = × × Vbv
nisms, or the diagonal mechanism alone. For web compression tan θ (Rd + Rh + Rv ) (11)
failure (Fig. 1), the shear strength of the deep beams is
reached when the concrete compression capacity at the strut
near the nodal zone is exceeded at ultimate load. The bound- Rv
ary of the nodal zone coincides with the diagonal strut bound- Fv = × Vbv
ary, but the concrete bearing force to be examined is the sum- (Rd + Rh + Rv ) (12)
mation of compressions from the diagonal, flat, and steep
struts (Fig. 3).

Equilibrium conditions The values of Rd,, Rh , and Rv are defined as5

Figure 4 shows the forces in equilibrium within the shear
span of the deep beam using a strut-and-tie perspective. The
resistance against the horizontal and vertical beam shears is
estimated as (Fig. 4) Rd =
(1 − γ h )(1 − γ v ) (13)
1 − γ hγ v
Vbv = − D sin θ + Fh tan θ + Fv (7)
γ h (1 − γ v )
Rh = (14)
Vbh = − D cosθ + Fh + Fv cot θ (8) 1 − γ hγ v

γ v (1 − γ h )
where D is the compression force in the diagonal strut; Fh is Rv = (15)
the tension force in the horizontal tie; and Fv is the tension 1 − γ hγ v
force in the vertical tie. The ratio Vbv/Vbh = tan θ is always
preserved in the proposed model.
Three load paths exist in the beam web, and the shear where γh is the fraction of horizontal shear transferred by the
forces must be apportioned to the resisting mechanisms. The horizontal tie in the absence of the vertical tie; and γv is the
ratios of the vertical shear assigned among the three mecha- fraction of vertical shear carried by the vertical tie in the
nisms are assumed as5 absence of the horizontal tie. According to Sch fer,10 the val-
ues of γh and γv are defined as
− D sin θ : Fh tan θ : Fv = Rd : Rh : Rv (9)

ACI Structural Journal/May-June 2000 369

 2 tan θ − 1
γh = for 0 ≤ γ h ≤ 1 (16)
3

2 cot θ − 1
γv = for 0 ≤ γ v ≤ 1 (17)
3

If the horizontal shear reinforcement does not exist or the

horizontal tie yields, the shear forces (or increments) are then
resisted by the reduced mechanisms, such as the diagonal and
vertical mechanisms. If the horizontal reinforcement is absent
or yields, by assigning γh = 0 the value of Rv converts to γv for
the reduced mechanisms. The same situation occurs between
Rh and γh.
The maximum compressive stress σd,max resulting from Fig. 5 Softening of compressive stress-strain curve due to trans-
the summation of the compressive forces from the diagonal, verse tensile strains.
flat, and steep struts (Fig. 4) on the nodal zone can be esti-
mated as
where σd is the average principal stress of concrete in the d-
direction; ζ is the softening coefficient; f c′ is the compres-
   jd   
 cosθ − tan −1     sive strength of a standard concrete cylinder in units of MPa;
εd and εr are the average principal strains in the d- and r-
D −   2a    directions, respectively; and ε0 is the concrete cylinder strain
Fh
  −1  jd    corresponding to the cylinder strength. The value of εo can
 cos tan     be defined approximately as12
1    2a   
σ d ,max =   (18)
Astr   −1 2 jd   

cos tan   −θ   f ′ − 20 
F 
ε 0 = 0.002 + 0.001 c 
  a   80 
− v  (22)
  −1  2 jd    for 20 ≤ f c′ ≤ 100MPa
sin  tan  
   a  
If the horizontal and vertical shear reinforcement are
assumed to be elastic-perfectly-plastic, the stress-strain rela-
If the bearing pressure on the nodal zone σd,max reaches the tionships are
capacity of the concrete, the shear strength of the deep beam
is attained. f s = Esε s for ε s < ε y (23)
Constitutive laws
Cracked reinforced concrete in compression has been f s = f y for ε s ≥ ε y (24)
observed to exhibit lower strength and stiffness than uniaxi-
ally compressed concrete, as shown in Fig. 5.8,11 The extent
of reduction can be related to the value of the principal tensile where Es is the elastic modulus of the steel bars; and fs and εs
strain in the concrete.8,11 This phenomenon of strength and are stress and strain in the mild steel, respectively. fs becomes
stiffness reduction is called the softening of concrete. fh or fv, εs becomes εh or εv, and fy becomes fyh or fyv when
Successful application of the strut-and-tie model with a soft- applied to horizontal or vertical web reinforcement, respec-
ened concrete stress-strain relationship has been obtained in tively.
several different problem areas.5-7 It is believed that the shear Based on the simplified constitutive equations for steel, the
strength of the deep beam should also be governed by the relationship between forces and strains of the tension ties
softened effect of concrete. become
According to Zhang and Hsu,11 the softened stress-strain
curve of the cracked concrete (Fig. 5) is represented as follows
Fh = Ath E sε h ≤ Fyh (25)
  −ε   − εd 
2
 − εd
σ d = −ζf c′2 d  −    for ≤1 (19) Fv = Atv Esε v ≤ Fyv (26)
  ζε 0   ζε 0   ζε 0

where Ath and Atv are the areas of the horizontal and vertical
  − ε / ζε − 1  2  − εd ties, respectively; and Fyh and Fyv are the yielding forces of the
σ d = −ζf c′1 −  d 0
  for >1 (20) horizontal and vertical ties, respectively.
 
2 / ζ − 1   ζε 0
Compatibility condition
5 .8 1 0 .9 A rigorous analysis dealing with two dimensional mem-
ζ = ≤ (21) brane elements should satisfy Mohr s circular compatibility
f c′ 1 + 400ε r 1 + 400ε r condition (Fig. 6). The basic compatibility equation used in

370 ACI Structural Journal/May-June 2000

Fig. 6 Compatibility conditions for diagonally cracked concrete.

this paper is the first strain invariant9

εr + εd = εh + εv (27)

Equation (27) is used to estimate the value of the principal

tensile strain, which is directly related to the extent of soften-
ing of the concrete, as per Eq. (21).

Proposed solution procedure

The algorithm in Fig. 7 starts with a selection of the ver-
tical beam shear Vbv and can be divided into five major steps
as follows:
1. According to the forces D, Fh, and Fv due to the select-
ed Vbv, the values of σd,max, εh, and εv can be estimated by Eq.
(18), (25) and (26);
2. By selecting a value for εd, the associated εr should be
estimated to determine the softening effect of the concrete.
For the cases of 1/2 < tan θ < 2, the values of the εr can be
calculated by εh, εv, and εd. For the cases of tan θ ≤ 1/2, the Fig. 7 Flow chart showing solution algorithm.
value of the εr is determined solely by εv and εd because the
horizontal mechanism is inactive for the load transmission
(that is, εh = 0). Similarly, the value of the εr is calculated only
by εh and εd for the cases of tan θ ≥ 2 (that is, εv = 0). When reported depth; and
the web reinforcement is yielded or not detailed, the yielding 4. Horizontal and vertical web reinforcement, if detailed,
strain (εh, = 0.002, or εv = 0.002) is used to estimate the lim- were distributed uniformly.
iting value of εr 8 Table 1 compares the test failure loads with the predic-
3. A concrete softening coefficient ζ can be estimated using tions of the softened strut-and-tie model previously presented
Eq. (21). The allowable compressive stress σd corresponding and the ACI 318-95 Code.2 Accuracy for the analytical mod-
to the strain εd in Step 2 is calculated by Eq. (19); els is gaged in terms of a strength ratio, which is defined as
4. If the acting stress σ d ,max determined in Step 1 is less the ratio of the measured to the computed strengths. Failure
than the allowable stress σ d in Step 3, iteration continues is predicted when the strength ratio reaches a value of unity.
from Step 1 by increasing the value of Vbv; and The softened strut-and-tie model reproduced the 123 test
5. If the stress σ d ,max is larger than the strength σ d , itera- results with reasonable accuracy. The mean of the test-to- cal-
tion continues from Step 2 by sweeping through possible val- culated strength ratio was 1.15 with a coefficient of variation
ues of εd to find a larger value of σ d . The largest value of (COV) of 0.16 (Table 1). A more conservative and scattered
σd obtained during the procedure is the strength. prediction was obtained from the ACI 318-95 Code.2 The
mean of the test-to-calculated strength ratio was 1.87 with a
EXPERIMENTAL VERIFICATION COV of 0.33 (Table 1).
A total of 123 test specimen results available in the litera- The test specimens with maximum moments at failure
ture are used to verify the proposed model. These are the test exceeding their yielding moments, a number of 28 out of 123
results by de Paiva and Siess,13 Kong et al.,14 Smith and specimens (23%), are specially marked in Table 1. Equation (2)
Vantsiotis,15 Fang et al.,16 and Chen17 that are listed in Table and (3) seem to have no application to these specimens.
1 in chronological order for easy reference. In selecting these However, the strength ratios of these specimens (Table 1)
test data, a number of screens were applied. They were as fol- indicate that the proposed model can yield reasonable estima-
lows: tions for these specimens despite of their nonlinear flexural
1. The test specimens were reported to have failed because behavior. The interaction between the shear failure and the
of web compression failure, not shear tension, bearing, or nonlinear flexural behavior is interesting, but was not pur-
flexural failures; sued further in this paper. Future research on this matter,
2. The shear span-depth ratio a/d was less than 2.5; however, is needed.
3. The test specimens were simply supported. Loads and The major factors influencing shear strength of deep
reactions were applied through the bearing blocks with beams have been found to be horizontal and vertical steel
areas, the strength of concrete and steel, and the shear span-

ACI Structural Journal/May-June 2000 371

Table 1—Experimental verification

depth ratio.18 The test-to-theory comparisons in this paper Fig. 8(b)).

use parametric study to further assess the suitability of the Figure 9(a) and (b) plot the effect of the vertical web rein-
softened strut-and-tie model and the ACI 318-95 Code for forcement ρv fy on the shear strength predictions of the two
deep beams. models, and a trend similar to the one seen in Fig. 8 is
Figure 8(a) and (b) show the effect of the horizontal web obtained.
reinforcement ρh fy on the shear capacity predictions of the Comparing the computed shear strengths according to the
proposed model and the ACI 318-95 Code. On the whole, the ACI Code with the measured shear strengths indicates that
predictions of the softened strut-and-tie model are very con- the ACI Code method considerably underestimates the shear
sistent for a broad range of horizontal web reinforcement strength of a deep beam without web reinforcement (Fig. 8(b)
(Fig. 8(a)). A greater scattering, however, is found for the ACI and 9(b)). It is thus concluded that the ACI 318-95 Code
Code predictions, with some overly conservative estimations underestimates the contribution of concrete and, in conse-
for beams without any horizontal web reinforcement (ρh = 0, quence, overestimates the contribution of web reinforcement

372 ACI Structural Journal/May-June 2000

Table 1—(continued)

ACI Structural Journal/May-June 2000 373

Table 1—(continued)

✝ : specimens with failure moment exceeding yielding moment

on the shear strength of deep beams. This point had been and expresses the function of web reinforcement reasonably
raised by de Paiva and Siess13 and by Smith and Vantsiotis.19 well.
The major discrepancy between the predictions of the The ACI Code method assumes that the shear strength of
proposed model and the ACI 318-95 Code comes from the dif- deep beams is derived from the combination of the tensile
ferent hypothesis of shear strength for deep beams. Based on strength of the concrete and the yielding strength of the web
the observation of concrete crushing in the diagonal direction, reinforcement, but this does not correlate well with the
the softened strut-and-tie model assumes that the shear observed failure phenomenon of concrete crushing in the web.
strength of deep beams depends on the softened compressive Figure 8 and 9 reveal that the empirical formulas of the ACI
behavior of concrete within the diagonal strut. At the same Code miscalculate the contribution of the concrete and web
time, web reinforcement plays two roles. One is to form ten- reinforcement to the shear strength; therefore, information on
sion ties and provide shear transferring paths. The other is to the proportioning and detailing of deep beams will be mis-
control the crack widths and retard the softening process of leading.
the cracked concrete. Figure 8 and 9 indicate that the softened As for the relative effectiveness of the horizontal and ver-
strut-and-tie model possesses a sound strength hypothesis tical web reinforcement, both the ACI Code method and the

374 ACI Structural Journal/May-June 2000

Fig. 8 E ffect of horizontal web reinforcement on shear strength Fig. 1 0 E ffect of concrete strength on shear strength predictions.
predictions.

Fig. 9 E ffect of vertical web reinforcement on shear strength pre- Fig. 1 1 E ffect of shear span-depth ratio on shear strength predic-
dictions. tions.

proposed model agree that the horizontal bars become more the diagonal strut is measured mainly by the a/d ratio as stat-
effective as a/d or ln/d becomes smaller but they disagree ed in Eq. (4). Therefore, the prediction of shear strength using
about the weighting factors. In the ACI Code provisions, ver- the proposed model is very sensitive to the value of a/d and
tical and horizontal web reinforcement are taken to be equal- higher shear strength will be predicted for a smaller a/d
ly effective at ln/d = 5.2 For a single span deep beam subject- value. Apparently, the ACI Code approach doesn t adequately
ed to central loading, this corresponds to an a/d ratio of 2.5. simulate the highly influential factor of the a/d ratio.
The critical threshold, however, is set at a/jd = 1 in the pro- The softened strut-and-tie model performance is better
posed model according to Eq. (16) and (17). Recently, Tan et than the ACI Code approach for all parameters under com-
al.4 pointed out that the ACI Code tends to overestimate the parison (Fig. 8 to 11). It is therefore recommended that the
shear strength contribution from horizontal web reinforce- current shear design procedures for deep beams be reformed
ment because the critical threshold is set at ln/d = 5. Based on to incorporate the actual shear resisting mechanisms as pos-
the experimental studies, Tan et al.20 recommended that the tulated by the softened strut-and-tie model.
critical threshold should be at an a/d ratio of 1.13, which is
quite close to the recommendation of the proposed model. CONCLUSIONS
Figure 10 shows the effect of concrete strength f c′ on the A proposal for determining the shear strength of deep
shear capacity predictions of the proposed model and the ACI beams has been made. The proposed softened strut-and-tie
318-95 Code for deep beams. Again, the softened strut-and-tie model is derived from the concept of struts and ties, and it also
model is very consistent for a broad range of concrete satisfies the equilibrium, compatibility, and stress-strain rela-
strengths (Fig. 10(a)). The ACI Code predictions become tionship of cracked reinforced concrete. Based on test results
more conservative with increasing concrete strength (Fig. in the literature and comparison of these results with the pro-
10(b)). Since the ACI Code method underestimates the contri- posed model and the ACI Code formulas, the following con-
bution of concrete, the more pronounced conservatism is clusions can be made:
expected for the high-strength concrete beams. 1.The softened strut-and-tie model consistently repro-
Figure 11 presents the effect of the a/d ratio on the shear duced 123 deep beam measured shear strengths with reason-
strength predictions using various methods. A fairly consis- able accuracy for a wide range of horizontal and vertical web
tent prediction is obtained by the softened strut-and-tie model reinforcement ratios, concrete strengths, and shear span-
for an a/d ratio between 0.35 and 2.1 (Fig. 11(a) and Table 1). depth ratios;
The ultimate strength predictions by the ACI Code, however, 2.The ACI 318-95 Code empirical equations were found to
tend to be more conservative with a decreasing a/d ratio (Fig. underestimate the contribution of concrete and overestimate
11(b)). Tan et al.4 also reported that the predictions of ACI the contribution of web reinforcement on the shear strength
Code can be very conservative at the low end of the a/d ratio. of deep beams;
In the softened strut-and-tie model, the angle of inclination of 3. In general, the ACI Code s predictions are conservative

ACI Structural Journal/May-June 2000 375

for the selected test results in this paper and more pronounced ζ = softening coefficient of concrete in compression
σd, σr = average normal stresses in d- and r- directions, respectively, (posi-
conservatism can be found for deep beams without web rein- tive for tension)
forcement and with high-strength concrete and a low a/d = assumed principal stresses
ratio; and σdmax = maximum compressive stress exerted on nodal zone in d-direction
4. The proposed model can provide valuable insights into
the shear strength and behavior of reinforced concrete deep REFERENCES
beams and may be incorporated into current deep beam 1. ACI-ASCE Committee 426, Shear Strength of Reinforced
Concrete Members, Proceedings, ASCE, V. 99, No. ST6, June 1973, pp.
design methods. 1091-1187. (Reaffirmed in 1980 and published by ACI as Publication No.
426R-74).
ACKNOWLEDGMENTS 2. ACI Committee 318, Building Code Requirements for Structural
This research study was partially sponsored by the National Science Concrete (ACI 318-95) and Commentary (318R-95), American Concrete
Council of the Republic of China under Project NSC 88-2211-E-011-011. Institute, Farmington Hills, 1995, 369 pp.
The authors would like to express their gratitude for this support. 3. Mau, S. T., and Hsu, T. T. C., Shear Strength Prediction for Deep
Beams with Web Reinforcement, ACI Structural Journal, V. 84, No. 6,
Nov.-Dec. 1987, pp. 513-523.
NOTATIONS 4. Tan, K. H.; Kong, F. K.; and Weng, L. W., High-Strength
a = shear span measured center-to-center from load to support Reinforced Concrete Deep and Short Beams: Shear Design Equations in
ab = depth of bearing block North American and UK Practice, ACI Structural Journal, V. 95, No. 3,
as = depth of diagonal strut May-June 1998, pp. 318-329.
Astr = effective area of diagonal strut
Ath, Atv = areas of horizontal and vertical ties, respectively
5. Hwang, S. J., and Lee, H. J., Analytical Model for Predicting Shear
bs = width of diagonal strut Strengths of Exterior Reinforced Concrete Beam-Column Joints for
C = resultant compressive force at section due to flexure Seismic Resistance, ACI Structural Journal, V. 96, No. 5, Sept.-Oct. 1999,
d = direction of diagonal concrete strut pp. 846-857.
= assumed direction of principal compressive stress of concrete 6. Hwang, S. J., and Lee, H. J., Analytical Model for Predicting Shear
= effective depth of deep beam Strengths of Interior Reinforced Concrete Beam- Column Joints for
d = distance from extreme compression fiber to centroid of compression Seismic Resistance, ACI Structural Journal, V. 97. No. 1, Jan.-Feb. 2000,
reinforcement pp. 34-44.
D = compression force in diagonal strut (negative for compression)
Ec = elastic modulus of concrete
7. Hwang, S. J.; Fang, W. H.; Lee, H. J.; and Yu, H. W., Analytical
Es = elastic modulus of steel bar Model for Predicting Shear Strength of Squat Walls, submitted for pub-
= compressive strength of standard concrete cylinder lication, Journal of Structural Engineering, ASCE, 1999.
ffhc,′ fv = steel stresses in the h- and v- directions, respectively 8. Vecchio, F. J., and Collins, M. P., Compression Response of Cracked
fs = average tensile stress of mild steel bars, taken as fh and fv in the h- Reinforced Concrete, Journal of Structural Engineering, ASCE, V. 119, No.
and v- directions, respectively 12, Dec. 1993, pp. 3590-3610.
fy = yield strength of bare mild steel 9. Hsu, T. T. C., Unified Theory of Reinforced Concrete, CRC Press, Inc.,
fyh, fyv = yield strength of bare mild steel of horizontal and vertical web rein- Boca Raton, Fla., 1993, 336 pp.
forcement, respectively
Fh, Fv = tension forces in horizontal and vertical ties, respectively (positive
10. Sch fer, K., Strut-and-Tie Models for the Design of Structural
for tension) Concrete, Notes of Workshop, Department of Civil Engineering,
Fyh, Fyv = yielding forces of horizontal and vertical ties, respectively National Cheng Kung University, Tainan, Taiwan, Mar. 1996, 140 pp.
h = direction of horizontal web reinforcement 11. Zhang, L. X. B., and Hsu, T. T. C., Behavior and Analysis of 100
j, k = coefficients MPa Concrete Membrane Elements, Journal of Structural Engineering,
jd = distance of lever arm from resultant compressive force to centroid ASCE, V. 124, No. 1, Jan. 1998, pp. 24-34.
of longitudinal tension reinforcement 12. Foster, S. J., and Gilbert, R. I., The Design of Nonflexural
kd = depth of compression zone at section Members with Normal and High-Strength Concretes, ACI Structural
ln = clear span of supports measured face-to-face
n = modular ratio of elasticity
Journal, V. 93, No. 1, Jan.-Feb. 1996, pp. 3-10.
= Es/Ec 13. de Paiva, H. A. R., and Siess, C. P., Strength and Behavior of Deep
r = direction perpendicular to d Beams in Shear, Journal of the Structural Division, Proceedings, ASCE, V.
= assumed direction of principal tensile stress 91, No. ST5, Oct. 1965, pp. 19-41.
Rd, Rh, Rv = beam shear ratios carried by diagonal, horizontal, and vertical 14. Kong, F. K.; Robins, P. J.; and Cole, D. F., Web Reinforcement
mechanisms, respectively Effects on Deep Beams, ACI JOURNAL, Proceedings V. 67, No. 12, Dec.
T = resultant tensile force at section due to flexure 1970, pp. 1010-1017.
v = direction of vertical web reinforcement 15. Smith, K. N., and Vantsiotis, A. S., Shear Strength of Deep
Vbh, Vbv = horizontal and vertical beam shear forces, respectively
γh = fraction of horizontal shear transferred by horizontal tie in absence
Beams, ACI JOURNAL, Proceedings V. 79, No. 3, May-June 1982, pp. 201-
of vertical tie 213.
γv = fraction of vertical shear carried by vertical tie in absence of hori- 16. Fang, I. K.;Horng, K. L.; and Chen, J. C., Shear Behavior of High-
zontal tie Strength Concrete Thin-Web Beams, National Science Council Project
εo = strain at peak stress of standard concrete cylinder Report No. 82-0410-E006-110, Taiwan, 1993, 145 pp. (in Chinese)
= 0.002+0.001( -20)/80 ( in units of MPa) 17. Chen, J. C., Shear Behavior of Simply-Supported Thin-Web Deep
εd, εr = average normal f c′ strains in d-
f c′ and r- directions, respectively (positive Beams Using High-Strength Concrete, Master s thesis, Department of
for tensile strain) Civil Engineering, National Cheng Kung University, Tainan, Taiwan,
= assumed principal strains
εh, εv = average normal strains in h- and v- directions, respectively (positive
1994, 156 pp. (in Chinese)
for tensile strain) 18. Mau, S. T., and Hsu, T. T. C., Formula for the Shear Strength of
εs = average tensile strain of mild steel bars, taken as εh and εv in h- and Deep Beams, ACI Structural Journal, V. 86, No. 5, Sept.-Oct. 1989, pp.
v- directions, respectively 516-523.
εy = yield strain of bare mild steel 19. Smith, K. N., and Vantsiotis, A. S., Deep Beam Test Results
θ = angle of inclination of h-axis with respect to d-axis Compared with Present Building Code Models, ACI JOURNAL,
ρ = ratio of tension reinforcement Proceedings V. 79, No. 4, July-Aug. 1982, pp. 280-287.
ρ′ = ratio of compression reinforcement 20. Tan, K. H.; Kong, F. K.; Teng, S.; and Weng, L. W.; Effect of Web
ρh = ratio of horizontal web reinforcement
ρv = ratio of vertical web reinforcement
Reinforcement on High-Strength Concrete Deep Beams, ACI Structural
Journal, V. 94, No. 5, Sept.-Oct. 1997, pp. 572-582.

376 ACI Structural Journal/May-June 2000