A CODE-BASED COMPARATIVE STUDY ON RC DEEP

BEAMS BEHAVIOR WITH SHEAR SPAN TO DEPTH

RATIO BETWEEN 0.5 AND 1.5

MOHAMMAD REZA SALAMY1, HIROSHI KOBAYASHI2, SHIGEKI UNJOH3, KENJI

KOSA4, TSUTOMU NISHIOKA5
1
Member of JSCE, Dr. of Eng., Public Works Research Institute, Earthquake Engineering Research Team
Earthquake Disaster Prevention Research Group, Minamihara 1-6, Tsukuba-shi, Japan 305-8516
2
Member of JSCE, Public Works Research Institute, Earthquake Engineering Research Team
3
Member of JSCE, Dr. of Eng., Public Works Research Institute, Leader of Earthquake Engineering Research Team
4
Member of JSCE, PhD, Kyushu Institute of Technology
5
Member of JSCE, Dr. of Eng., Hanshin Expressway Public Corporation

1. Introduction prediction in comparison with JSCE with an adequate

safety margin.
In order to investigate RC deep beams behavior and
lateral reinforcement effects in improving shear behavior 2. Experiment setup and specimens details
of those beams, a study is undergoing in Public Works
research Institute (PWRI) based on the experiments To evaluate analytical results of FEM as well as code-
conducted during the year 2003 and 2004. Three sets of based design, the following sets of experiments are
specimens comprise of nineteen RC beams including the carried out at PWRI and Kyushu Institute of Technology.
experiments carried out on a joint research basis with However analytical evaluation by FEM will be presented
Kyushu Institute of Technology (KIT) and Hanshin in future in some other publications and here only the
Expressway Public Corporation (HEPC) are investigated
Table 1: Steel Properties of specimens
in this study. The beams have shear span to depth ratio
between 0.5 and 1.5 and effective depth size from 400
Beam
ρw ρ st fy Ast
Stirrups
mm to 1400 mm. The longitudinal tensile reinforcement Asc
% % MPa
ratio is kept almost constant in about 2% for all
specimens while lateral reinforcement (stirrups) ratio B-2 0.0
varies by 0.0%, 0.4% and 0.8% in shear span. The results B-3 0.4 D6@65
of experiment compared with Japanese design code such B-4 0.8 D10@75
as Japan Road Association1, 2 (JRA) and Japan Society of B-6 0.0
5D22
Civil Engineers3 (JSCE). In order to trace compressive B-7 0.4
2.02 376
D6@65
force path in RC beams, numbers of acrylic bars are B-8 0.8 2D10 D10@75
located in between loading plates and supports to B-10-1
0.0
measure strain in concrete in the designated path. The B-10-2
objective of using this method in experiment is to verify B-11 0.4 D6@65
the validation of strut-tie model and also characterize and B-12 0.8 D10@75
measure actual strain in the location with highest B-10.3-1 388 9D25
0.0 2.11
possibility of crushing or cracking in any kind of shear B-10.3-2 371.7 2D16
failure occurs in RC beams with low a/d ratio. The results B-13-1 10D32
presented in this paper are part of a larger study on shear 0.0 2.07 398
2D13
behavior of RC deep beams including size effect B-13-2
experimentally and numerically. In this regard, only B-14 0.0 14D32
2.04 398
experimental results in comparison with codified design B-17 0.4 4D13 D13@100
results. It is found however that by increasing a/d in both 18D35
B-15 0.0 1.99 402
design-codes, shear strength of the member will be 2D13
decreased which agree well with experimental B-16 0.0 394 18D41
2.05
observation. On the other hand, JRA code yields better B-18 0.4 397.5 2D13 D16@120

1
 bs c a
B16,18 B15 B14,17
bs

h
d
B13 B10.3 B2-B12
L

Fig.1. Detail of specimens with and without stirrups

results of design codes are evaluated by experimental than those in shear span due to the reason stated above4.
observation. Experiments consist of nineteen RC beams Further study is however necessary to confirm the effect
with geometric characteristic and material properties of mid-span stirrups experimentally.
given in Fig.1, Table 1 and Table 2. In Table 2, b is specimen width, a/d and fc′ are shear span
In Table 1, ρ w , ρ s , fy , Ast and Asc are shear span, stirrups to depth ratio and compressive stress respectively.
ratio, longitudinal tensile reinforcement ratio and their Maximum load capacity and related deflection as well as
yield stress, cross section area of tensile and compressive shear crack initiation load and maximum deflection are
reinforcement respectively. All specimens, with or noted as Pmax , Pcrsh , δ peak and δ max respectively. Other
without stirrups in shear span, have a minimum lateral
geometrical parameters of Table.2 are schematically
reinforcement in mid-span and out of span. Despite
determined in Fig.1. All specimens are subjected to four
absence of shear stress in this part, which in the first look
points monotonic static load condition. Experimental data
implies un-necessities of shear reinforcement, they may
acquisition is mainly focused on mode of failure; crack
delay or in some case prevent the propagation of diagonal
patterns, load-displacement relationship as well as steel
crack to compression zone. It is believed even that
and concrete strain in some designated locations to
reinforcements in mid-span sometimes are more effective
evaluate the analytical results.

Table 2: Geometric and material Properties of specimens

Beam Geometry size (mm) f c′ Pmax δ peak Failure

a/d Pcrsh Mode
L c a d h b bs
MPa KN (mm )
B-2 1550 525 3.16 II
36.2
B-3 0.5 700 200 1536 625 4.78 II
B-4 31.3 1951 700 1.85 II
B-6 1050 400 2.77 II
31.3
B-7 1.0 1100 400 1181 400 2.83 II
300 400 475 240 100
B-8 37.8 1501 600 3.26 II
B-10-1 29.2 616 325 3.82 II
B-10-2 23 703 278 5.28 II
1500 600
B-11 29.2 1025 350 15.96 II
B-12 31.3 1161 300 7.05 II
B-10.3-1 37.8 1960 700 6.62 II
2250 450 900 600 675 360 150
B-10.3-2 31.15 1787 527 8.62 II
B-13-1 1.5 31.63 2985 500 11.87 II
3000 600 1200 800 905 480 200
B-13-2 24 2257 807 9.33 II
B-14 31 3969 1100 9.27 II
3750 750 1500 1000 1105 600 250
B-17 28.7 5214 1600 11.92 II
B-15 4500 900 1800 1200 1305 720 300 27 5390 1500 11.91 II
B-16 27.3 5975 1900 10.57 II
5250 1050 2100 1400 1505 840 350
B-18 23.5 8396 2400 15.79 II

2
3. Shear failure mechanism of RC beams

Failure modes are so determined in two main categories B18

of flexural failure mode (Mode I) and shear failure mode
(Mode II) with three subcategories for Mode II failure as
followings: B15
Mode II-1: Diagonal tension failure, which in the line of
thrust become so eccentric and give rise to flexural B17
failure in compressive zone. It is important however to
mention that this kind of failure is a result of tensile
crack extension in compressive zone due to the flexural
B13-2
load.
Mode II-2: Shear compression failure where RC beam B10-2
fails due to the development of diagonal crack into the
compressive zone and reduces the area of resisting
region excessively and beam crushes once generated Fig.3. Experimental results of beams with large load
compressive stress exceed compressive strength of reserve after peak load.
concrete.
Mode II-3: Shear proper or compressive failure of struts,
which is often observed in beams with very small shear larger a/d ratios.
span to depth ration a/d (about a/d<1.5). In this case due
to the small a/d ratio, the line of thrust will be so steep 4. Design codes proposed equations for shear
and arch action not only reserve flexural capacity in most
strength of RC members
cases but also efficiently sustains required shear force.
Arch formation is clearly observed in those beams and
Design codes JSCE (Japan Society of Civil Engineers)
finally beams fail due to either sudden tensile crack
and JRA (Japan Road Association) are employed to
formation parallel to the strut axes or compressive crush
design tested beams for comparison with experiment by
in normal direction to the strut axes. The latter case
means of shear span to depth ratio as well as effective
shows more reserved load after crushing (for instance
depth size effect. Since JRA code is not formulized like
Beams 13, 10.3 and B18). Figure 2 depicts crack pattern
other codes and design values are presented in a tabular
of B-18 at the last stage of loading where the beam failed
form, here only shear design procedure of JSCE code is
as a result of strut compressive failure in the location
presented briefly. It is noted however that JRA design
stated in the figure. Thrust zone is schematically shown
code henceforth means design through part five2 taking
in this figure. Despite compressive failure in strut the
into account the effect of deep beam (Cdc and Cds)
beam sustained almost 80% of peak load and a plateau
through part four1 (page 230).
formed after small drop of the peak load. This
JSCE code: Based on this code, shear capacity of
phenomenon happened in some other beams such as B17
concrete in RC members Vcd (concrete contribution to
and B15, which is in contrary to what shear failure
naturally implies as a sudden failure. Some results for shear capacity) can be obtained by the following
specimens with large reserved load capacity after peak equations. L and h are beam’s span and height.
load are shown in figure 3. Note however that concrete is In case of L ≥ 2 for simple beams:
a rate dependent material so loading rate is supposed to h
have significant effect on RC member behavior and
should be taken into account when numerical simulation β d β p β n f vcd bw d
Vcd = (1)
is performed. γb
It is also worth to note that arch action requires a
1000
substantial horizontal reaction at the support to be βd = 4 ≤ 1 .5 (2)
formed. To satisfy this condition tensile bars in all d
specimens are well anchored with a rather long hook β p = 3 100 ρ s ≤ 1.5 (3)
beyond support region. In this study only the effect of
shear span stirrups have been considered though it would
ρ s = As (bw .d ) (4)
be of interest to study on influence of stirrups in mid-span f vcd = 0.2 3 f cd′ ≤ 0.72 ( N / mm ) 2
(5)
in preventing diagonal crack to compressive zone for

Compressive Failure
Parameters As , f cd′ , γ b , bw and d are longitudinal tensile
reinforcements area, concrete ultimate compressive
stress, uncertainty parameter which in general case will
be 1.3, member web width and effective depth in critical
section respectively. Since the nominal shear strength is
used for comparison with experiments, γ b = 1 is
supposed to set in all calculation. However the parameter
Fig.2. Crack Pattern of Beam 18
for material uncertainty is not explicitly stated in JRA

3
code therefore to have a meaningful comparison between
two codes γ b = 1.3 is conducted. On the other hand JSCE JSCE (all specimens are designed as deep beam)

experimental results are also calibrated by the same 9000

reduction factor equal to 1.3. The value β n = 1 is also 8000

adopted due to the code definition for simply supported 7000

beams. 6000
In case of L < 2 for simple deep beams:

Test Pu (KN)
5000
h 2000
4000

β d β p β a f dd bw d 3000
Vcd = (6) 1000
γb 2000
a/d<1.5

f dd = 0.19 f cd′ ( N / mm 2 ) (7) 1000 0

0 1000 2000
5 0
βa = (8)
1 + (a / d ) 2
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
JSCE Design Code Pu (KN)
φ = −0.17 + 0.3(a / d ) + 0.33 / ρ wb ≤ 1.0 (9)
Fig.4. JSCE design code ultimate load versus test result
Vsdd = φ.Vsd (10)
Aw f wyd (sin α s + cos α s ) / s s (11)
Vsd = .z 9000
γb 8000

where z ≈ d / 1.15 and f wyd ≤ 400 MPa for normal 7000

6000
strength concrete. Lateral reinforcement contribution to
Test Pu (KN)

shear capacity is denoted by Vsd and is calculated by 5000

Eq.11 for any values of L ratio. 4000

2000
h 3000
Finally the shear capacity of entire section V yd is 1000
2000 a/d<1.5
calculated through Eq.12 as below.
1000 0
0 1000 2000

V yd = Vcd + V sdd (12) 0

0 1000 2000 3000 4000 5000 6000 7000 8000 9000
JRA Design Code Pu (KN)
5. Comparison of the Results
Fig.5. JRA design code ultimate load versus test result
Both codes are applied for shear load capacity
calculation. The results of codified calculation and
experimental observation are shown in figures of this definition and the results show a kind of scatter
section. distribution around both solid and dotted lines. It is
however observed that no data points fall significantly
(1) Load capacity below either lines for any of the codes. Experiment
As mentioned in section 4, only simply supported beams showed that shear crack initiated at about 40% of the
with L/h<2 are considered as deep beam by JSCE design ultimate load and full shear crack will be formed
code. In other words despite most other design codes approximately in 0.5Pu but still beam sustained load
including JRA which recognize beams with a/d<2.5 as capacity to about 80-90% of the ultimate load. Afterward
deep beam, only the specimens tested in this study with shear cracks were severely widened and extended to
a/d=0.5 are recognized by JSCE code as deep beam. compressive zone. Shear sliding of concrete pieces
Nevertheless in this study, beams are designed in both around shear crack could be clearly observed with bare
cases of only following JSCE regulation as well as eyes. This point is considered the ultimate capacity of
considering all beams as deep beams and applying JSCE beam in shear by a number of design codes, which the
deep beam criterions. beam is in serious irreversible circumstances. Aggregate
Figures 4 and 5 show ultimate loads of specimens along interlock, which is the backbone of current codes are
with those obtained by the mentioned design codes. almost exhausted in this stage. Figure 6 shows code
Dotted lines in figures illustrate reduced ultimate loads by prediction capacities in terms of above-mentioned
means of JSCE reduction factor γ b = 1.3 to calibrate test proportions to the experiment peak load.
The figure shows that JSCE code have usually estimated
results for the sake of comparison with reduced design
the load capacity of members around shear crack
codes predictions. As can be seen in both figures, JRA
initiation load while JRA code yields the results near
code gives much better agreement to the experiment than
practical ultimate capacity of beams (about 0.8Pu). In
JSCE code with an acceptable safety margin. JSCE code
other words, for beams with a/d>0.5 JRA allows shear
seems however has no consistency in terms of deep beam

4
crack occur and extend but JSCE allows only shear crack 9000 0.4Pu 0.8Pu Pu Pexp
form but not extend. It is acceptable in essence if the JSCE
philosophy of JSCE code like some other design codes is 8000 JRA
to ensure the safety of structures before initiation of shear 7000
cracks not to reach to the ultimate load. Nevertheless the Pexp
discrepancy in the results is for beams with a/d<1.0 6000 Pu =
1.3
which gave rise to a jump in predicted shear capacity of

Test Pu (KN)
5000
the member by JSCE and despite a big safety margin for
2000
specimens with larger a/d ratio, these beams seem to be 4000

overestimated. Figure 7 gives a comparison between load

3000
capacity observed in experiment and design codes. For 1000
a/d=0.5, JSCE and JRA yield almost same prediction but 2000

two codes differ more as long as a/d increases. Shear

1000
crack load is also presented in this figure to clarify and 0
0 1000 2000
follow the discussion made on figure 6. On the other 0

hand, complementary JSCE standard specification4 0 1000 2000 3000 4000 5000 6000 7000 8000 9000
Design Code Pu (KN)
present a rigorous procedure in design of deep beams
with a/d<3. Equations 1 to 5 are applied for design while Fig.6. Design code ultimate load versus test result
a new coefficient taking into account the effect of a/d
ratio as follows. 9000
Experiment
8000
1.4 (13)
β a = 0.75 + 7000
a a/d=1.5
d 6000
JRA
5000
Effect of β a is illustrated in figures 8 and 9. Although the

B-11

B-12
B-10-1

B-10-2
JSCE 4000
results show better agreement with experiment and also
3000
clod to JRA prediction but for beams with a/d=0.5 load
capacity goes beyond experiment load even with no 2000

reduction factor. 1000

Shear crack load
0
B-10-1

B-10-2

B-11

B-12

B-10.3-1

B-10.3-2

B13-1

B13-2

B-14

B-17

B-15

B-16

B-18
(2) Size effect
In order to study size effect in shear capacity of beams
with low a/d ratio, both code examined and verified with
2000
experimental results. Test specimens cover a wide range a/d=0.5 a/d=1
of effective depth from 400mm to 1400mm. Accordingly 2000

variation of average shear stress taking into account JRA 1000

concrete compressive strength ( V u / b.d .3 f c′ ) in terms of 1000 JSCE JRA

effective depth is shown in figure 10. To eliminate a/d JSCE

effect on ultimate shear stress of the beams, only a/d=1.5 0
0
B-6R

B-7R

B-8
B-2

B-3

B-4

is considered here. It is clear that as long as the effective

depth increases, the shear strength of the section
decreases. The regression curve is assumed to be a power Fig.7. Experiment and design code load for all specimens
function of effective depth d in order to adjust to the size
effect function proposed by JSCE and JRA. The equation 9000 0.8Pu Pu Pexp
is round off and rewritten in the following form JSCE Complementary

f (vu ) = λ (d −0.22 ) (14)

JRA

6000 Pexp
Pu =
Test Pu (KN)

where coefficient λ is a function of a/d ration, 1. 3

reinforcement ration and member’s boundary condition. 2000
Since the three aforementioned parameters are constant
3000
for the beams used for producing figure 10 consequently 1000
λ = 4.77 is determined to best fit to the experiment data
points. According to JSCE, shear stress varies in a form
1 0
given by Eq.2 in terms of d 4 . On the other hand JRA1 0 0 1000 2000
1 0 3000 6000 9000
proposed procedure can be estimated by a function of d 3
Design Code Pu (KN)
to take into account size of specimen. Although the
foregoing expression of figure 10 is a crude Fig.8. Design code ultimate load versus test result

5
approximation, the form of equation 14 agrees well with 2
that of Eq.12 of JSCE code. It is however not a Pexp
significant differences between JSCE and JRA size effect
1.8
Pu = JSCE
1.6 1 .3 JSCE Complementary
expression as can be seen in figure 11 and both 1.4
JRA

Test Pu (KN)
expressions are attributed to a reasonable estimation of 1.2
member depth effect. The maximum values for this 1 Pexp
coefficient set 1.0 and 1.5 by JRA and JSCE code 0.8 Pu
respectively. In other words, in spite of JSCE code, which 0.6 0.8Pu
attributes 50% increase in shear strength capacity to size 0.4
effect, JRA however does not allow any increase in shear 0.4Pu
0.2
strength. One reason for this might be the fact that JRA is 0
usually dealing with structures with large components 0 0.5 1 1.5 2
most of them larger than one-meter depth but JSCE Design Code Pu (KN)
should cover wider range of element size since it is to Fig.9. Design code ultimate load to test ratio versus a/d
design various structures too.
2
6. Conclusion

A comparative study between experiment, JSCE and JRA

Average shear stress

design codes has carried out by means of ultimate loads,
1
shear crack loads as well as size effect issues proposed by
either codes. It is found that JRA code has a consistence -0.22
design procedure for RC beams with low shear span to y = 4.77 x
depth ratio. This code assumes beams with a/d<2.5 as
deep beams while JSCE has larger limit for deep beam as 0
L/h<2 where L and h are beam’s span and height 200 600 1000 1400

respectively (more than two time bigger than JRA limit). Effective depth d (mm)
Therefore only beams with a/d=0.5 of this experiment are
designated by JSCE to follow the deep beam design Fig.10. Effective depth versus Vu for Av=0 and a/d=1.5
procedure. Since deep beams usually have higher shear b.d .3 f c′
strength due to the resisting mechanism of such beams
against external loads by means of compressive arch 2
JRA JSCE Experiment (Eq.13)
formation, there will be a discrepancy between
experimental observation and estimated strength by JSCE
Size effect coefficient

code. Estimated shear load capacity by JSCE is around

shear crack load of experiments while JRA code allows
1
shear cracks form and extend to a certain level with
higher load capacity prediction. In this sense it can be
concluded that JSCE design code yields much
conservative results than that of JRA except for very
small a/d ratio say 0.5 where JSCE amplifies the 0
predicted shear strength by means of a function of a/d 0 2 4 6 8 10

ratio (Eq.8). It is noted however that no code data points d (m)

go beyond experiment ultimate loads. Application of Fig.11. Effective depth versus size effect coefficient
Complementary JSCE standard specification is also
examined for tested beams. Despite better agreement with
experiment and closer results to JRA prediction, data
points for a/d=0.5 drop below the line (Fig.8) or in other REFRENCES
words, predicted load capacity goes beyond obtained load
1. Japan Road Association. Design specifications of highway
by experiment. bridges. Part IV: substructures; (in Japanese), 2002.
Concerning size effect in shear strength of RC beams 2. Japan Road Association. Design specifications of highway
with low a/d ratio, experimentally observed size effect by bridges. Part V: seismic design; (in Japanese), 2002.
means of effective depth variation confirmed that both 3. Japan Society of Civil Engineers. Standard specifications
code have included this phenomenon in design procedure for concrete structures-2002, Structural performance
adequately thought JSCE equation has better agreement verification. March 2002, Tokyo, Japan (in Japanese).
with experiment. The main difference between the codes 4. Behavior of Reinforced Concrete Beams with a Shear Span
lies on the beams with depth smaller than 1000mm which to Depth Ratio Between 1.0 and 2.5, Michael D. Kotsovos,
JRA limits the coefficient to one but JSCE goes as far as ACI Journal, May-June 1984
5. Japan Society of Civil Engineers. Complementary for
1.5 and attributes shear strength to the size effect up to standard specifications for concrete structures-2002,
50% higher in members with depth less than one meter. March 2-002, Tokyo, Japan (in Japanese).