Engineering Structures 182 (2019) 501–509

Contents lists available at ScienceDirect

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

A kinematics-based approach for the shear strength of short fibre-reinforced T

concrete coupling beams
Boyan Mihaylov
University of Liege, Department of ArGEnCo, Building B52, Quartier Polytech 1, Allée de la Découverte 9, B-4000 Liège, Belgium

ARTICLE INFO ABSTRACT

Keywords: Short coupling beams are susceptible to brittle shear failures that are typically suppressed with dense transverse
Coupling beams and/or diagonal reinforcement. To reduce the amount of shear reinforcement and improve the service behavior
Fibre-reinforced concrete of the beam, researchers have proposed a solution with steel fiber-reinforced concrete (FRC). However, while
Shear this solution is promising, there are no sufficiently simple mechanical models capable of capturing the shear
Kinematics
strength and displacement capacity of short FRC coupling beams without diagonal reinforcement. This paper
proposes such a model based on first principles: kinematics, equilibrium, and constitutive relationships for the
mechanisms of shear resistance. The model accounts in an explicit manner for five shear mechanisms across the
critical shear cracks: diagonal compression in the critical loading zones, aggregate interlock, tension in the
stirrups and in the steel fibres, and dowel action of the longitudinal reinforcement. These mechanisms are
predicted and the results are compared to 20 tests from the literature as well as to FEM predictions. It is shown
that the proposed approach models well the effect of beam aspect ratio, concrete strength, stirrup ratio, and
amount of steel fibres. Furthermore, the model is used to develop relationships outlining the effectiveness of steel
fibres to reduce conventional stirrup reinforcement in coupling beams with various properties.

1. Introduction is a need for a displacement-based approach which addresses these is-

This paper focuses on the modelling of short coupling beams with
clear-span-to-depth ratios smaller than about 2.5 (Fig. 1a). Such beams
fall into the category of disturbed regions/members (D-members) as
they do not obey the classical plane-sections-remain-plane hypothesis
and are typically designed with strut-and-tie models. In addition, due to
their large stiffness, they typically work with high shear stresses asso-
ciated with double-curvature bending, and therefore require large
amounts of stirrups and/or diagonal reinforcement. Therefore, experi-
mental research has been performed to find more efficient reinforce-
ment layouts and to reduce the amount of shear reinforcement [2–5].
More recently, researchers have proposed the use of steel fibres in the
concrete that bridge the critical shear cracks and enhance the strength
of the member [5–8]. However, while tests have indicated that this is a
viable solution, extending the strut-and-tie models for reinforced con-
crete (RC) coupling beams to fibre-reinforced concrete (FRC) members
is a challenging problem. This is mainly because the strut-and-tie ap-
proach neglects the tension in the concrete, while enhanced crack
control and ductility in tension is the main advantage of FRC. In ad-
dition, strut-and-tie models are not well suited for evaluating de-
formations and displacement capacity which are key for the perfor-
mance-based seismic design of coupled-wall structures. Therefore, there
sues in a rational and computationally efficient manner.
A kinematics-based framework has already been developed for the
modelling of various reinforced concrete D-members such as deep
beams, short walls, and short coupling beams [9–12]. The same fra-
mework was used recently to model the complete shear response of
short FRC coupling beams, including their post-peak behavior [13]
(Fig. 1b). This approach is based on a kinematic model which describes
the complete deformation patterns of short coupling beams with the
help of only two degrees of freedom (DOFs). The kinematic model
provides conditions for compatibility of deformations which are com-
bined with equilibrium equations and constitutive relationships for the
mechanisms of shear resistance across the critical shear cracks. As
evident from Fig. 1b, the kinematics-based approach accounts for five
components of shear resistance: critical loading zone, VCLZ, aggregate
interlock, Vci, stirrups Vs, dowel action of the longitudinal reinforce-
ment, Vd, and steel fibres Vf. The main goal of this paper is to simplify
this approach to predict not the complete response, but only the peak
resistance of short FRC coupling beams. The main assumption which is
used in the derivation of the model stems directly from Fig. 1b and
other similar analyses: the failure of the beam is triggered by the failure
of the critical loading zone (CLZ). In other words, the peak resistance of
the beam occurs almost simultaneously with that of the CLZ. Apart from

E-mail address: boyan.mihaylov@uliege.be.

https://doi.org/10.1016/j.engstruct.2018.11.066
Received 1 May 2018; Received in revised form 31 August 2018; Accepted 26 November 2018
0141-0296/ © 2018 Elsevier Ltd. All rights reserved.
B. Mihaylov Engineering Structures 182 (2019) 501–509

θ strain in the longitudinal reinforcement εt,avg measured along the

cracked length of the member lt. When the longitudinal reinforcement
V elongates by εt,avglt, the critical crack opens as shown in the top diagram
V in Fig. 2a (crack width w). In addition to the critical crack, a series of
radial cracks also open above and below the diagonal of the beam,
outlining two fans of rigid radial struts. The struts are pinned at the
bottom-left and top-right corners of the beam and are connected to the
longitudinal reinforcement. As the opening of the fans is mainly caused
by the bending moments M, this deformation pattern can be associated
with flexure.
While the deformation pattern controlled by DOF εt,avg is associated
with flexure, the bottom deformation pattern in Fig. 2a is associated
with shear. In this pattern the critical crack undergoes both opening
and slip displacements w and s associated with a vertical displacement
Δc between the crack faces. Namely displacement Δc is the second DOF
of the kinematic model. In the ends of the critical crack Δc results in
shear deformations along length lk where the longitudinal reinforce-
ment works in double curvature, and diagonal compressive strains in
the critical loading zone (CLZ) where the concrete crushes at failure.
(a) The geometry of the CLZ is determined by its characteristic length lb1e
and the angle of the diagonal of the beam α. The expressions for lt, lk,
lb1e and α as well as the remaining geometrical properties of the ki-
300 nematic model have been derived elsewhere [12] and are summarized
in Fig. 2b.
The assumptions and geometry of the kinematic model were used to
250 pred. derive the complete displacement field of the coupling beam based on
small-displacement kinematics (Eqs. 5–8 in Fig. 2c). In this way, the
200 exp. horizontal and vertical displacements δx and δz of each point of the
beam are expressed as functions of DOFs εt,avg and Δc. More im-
portantly, the two DOFs are used to express key deformations in the
V, kN

150
Vs
100
VCLZ
50 Vf
Vci
0 Vd
0 5 10 15
θ, rad x10-3
(b)
Fig. 1. Short FRC coupling beams. (a) Short coupling beams in a coupled wall
system (adapted from [1]). (b) Predicted response of a short FRC coupling beam
based on a two-parameter kinematic approach (adapted from [13]).
the derivation of the model, the paper presents comparisons with 20
tests as well as with nonlinear finite element simulations. Validated in
this manner, the kinematics-based approach is used to evaluate the
effectiveness of steel fibres to reduce conventional stirrups.
2. Kinematics of short coupling beams
The deformation patterns of short coupling beams failing in diag-
onal tension are modelled with the kinematic model depicted in Fig. 2a.
This model is adapted from earlier models developed for deep re-
inforced concrete beams and short reinforced concrete coupling beams
under double curvature [11,12]. It consists of two simple deformation
patterns, each of which a function of a single degree of freedom (DOF).
The shear failure is assumed to occur along a straight shear crack which
extends along the diagonal of the coupling beam. It is also assumed that
the load and reinforcement are symmetrical, and therefore the strains in
the top and bottom longitudinal reinforcement are equal. The complete
deformation pattern of the beam is obtained as a linear combination of
the two simple patterns.
The first degree of freedom of the kinematic model is the average
critical diagonal crack, i.e. stirrup strain εv, crack width w and crack slip
s half way along the crack (Eqs. 9–11) [11,12]. These deformations will
be used to express the mechanisms of shear resistance across the crack,
also as functions of DOFs εt,avg and Δc.
3. Mechanisms of shear resistance
There are at least five mechanisms of shear resistance which de-
velop in short FRC coupling beams (Fig. 3). As characteristic of short
members, a significant portion of the shear is carried in the critical
loading zone where the concrete is subjected to diagonal compression
associated with DOF Δc. As mentioned earlier, in addition to shear
component VCLZ, shear is also carried by aggregate interlock, Vci, ten-
sion in the stirrups, Vs, dowel action of the longitudinal reinforcement,
Vd, and the contribution of the steel fibres, Vf. Component Vci depends
mainly on the width of the critical crack, Vs on the strain in the stirrups,
and Vd on strain εt,avg in the bar-dowels of length lk. These four me-
chanisms have been derived elsewhere [11,12] and are summarized in
Fig. 3 (Eqs. 12–15), while component Vf is discussed in more detail
below.
As demonstrated in [13], the steel fibres influence the shear beha-
vior of short coupling beams mainly in two ways: they transfer tension
across the critical diagonal cracks and also enhance the ductility of the
critical loading zones where the concrete works in compression. How-
ever, as the latter effect has been shown to impact mainly the post-peak
behavior of the beam and not its strength, it is neglected here for the
sake of simplicity. Therefore, Eq. 12 in Fig. 3 for the shear carried in the
CLZ is adopted directly from the shear strength models for deep RC
beams in double curvature and short RC coupling beams [11,12]. This
expression was derived based on the assumption that the CLZ is at
failure under principal compressive stresses inclined at angle α. Because
the CLZ is crossed by the top longitudinal reinforcement which works in
tension, the strains in the reinforcement εt,avg damage the concrete and
reduce its compression capacity. This compression softening effect is
modelled with factor kc(εt,avg) proposed by Vecchio and Collins [14].
The tension behaviour of steel fibres across cracks has been studied

502
B. Mihaylov Engineering Structures 182 (2019) 501–509

(a) Deformation patterns and DOFs (b) Geometry (c) Deformations

DOF t,avg Characteristic length of critical Displacement field
loading zone
• below critical crack
(1) lb1e = 0.11 a + h 2 2 (5) δ x = ε t , avg x
V M
M Angle of critical crack ε t , avg x 2

d
h
(6) δ z =

w
V h−z
(2) α1 ≈ α ≥ 35o
1
db scr • above critical crack
h (7) δ x = ε t , avg ( h cot α + x − a )
a+ t,avglt (2) α = atan
a
2
+ az cot α − ( x − a )
Dowel length (8) δ z = ∆ c + ε t , avg
DOF c z
lb1e
(3) lk = l0 + d (cot α − cot α1) Stirrup strain
c
CLZ l0 = 1.5( h − d ) cot α1 ≥ scr ∆ c + 0.25ε t , avg d cot 2 α1
(9) ε v = 2
(3) 0.28db 2.5 ( h − d ) 0.9 d
w scr =
v ρl d
s Crack width and slip
Cracked length along ε t , avg lk
z
longitudinal reinforcement (10) w = + ∆c cosα1
lk
x 2sin α1
(4) lt = d cot α1 + lk − l0
(11) s = ∆ c sin α1
Fig. 2. Two-parameter kinematic model for short coupling beams [12,13].

extensively by different authors [15–17], particularly for the case of

pure crack opening (Mode I fracture). It is now generally accepted that
the relationship between crack opening and normal stress across the
crack can be expressed as a superposition of two stresses (Fig. 4): ten-
sion transferred directly between the two crack faces (tension softening
of concrete), and tension carried by the steel fibres which are bonded
(anchored) on each side of the crack [15]. As in the kinematic model
the coupling beam is assumed fully cracked, the tension softening is
neglected and the tensile stress transferred by the fibres is expressed
according to a variable engagement model proposed by Voo and Foster
[15]:

f =
0.396 fc f lf tan
1
( )1 w
df / 3.5 2w
2

df lf (16) Fig. 4. Tension behaviour of FRC (adapted from Voo and Foster 2003).

where 0.396√fc is the bond stress between the fibres and concrete, ρf

(12)

VCLZ
V (13)
vci
M
(14)
Vs , Vf

(15)
Vd

Fig. 3. Mechanisms of shear resistance in short coupling beams [11,12]. kc = compression softening factor; fc = concrete compressive strength; b = section width;
ε1 = principal tensile strain in CLZ; ag = maximum diameter of coarse aggregates; Es = modulus of elasticity of longitudinal reinforcement; Av = area of stirrups;
ρv = stirrup ratio; fyv = yield strength of stirrups; α1 = angle of critical crack; nb = number of top/bottom longitudinal bars; db = diameter of longitudinal bars;
lk = length of bar-dowels; fy = yield strength of longitudinal bars.

503
B. Mihaylov
is the volumetric ratio of steel fibres, lf is the length of the fibres, df is
the fibre diameter, and w is the crack width.
However, because in coupling beams the critical crack undergoes
both opening and slip displacements (mixed Mode I and II fracture),
shear component Vf is not governed only by the width of the crack w.
Therefore, it is proposed to replace w in Eq. (16) with the vertical
displacement in the crack wv which dominates the crack movement
near shear failure. Displacement wv is expressed with the two DOFs of
the kinematic model as follows:
wv = 0.5 t , avg lk cot 1 + c
and the shear contribution of the steel fibres across the critical crack is
Vf = f (wv ) bd/sin 1 (18)
where b is the width of the beam section, d is the effective depth, α1 is
the angle of the critical crack, and d/sinα1 is the length of the crack. In
the derivation of Eq. (18), it is assumed that the fibres generate the
same force for the same magnitude of crack displacement, regardless of
the direction of the displacement. In addition, it is assumed that the
force generated by the fibres is parallel to the crack displacement. In
other words, if the crack undergoes a vertical displacement wv or a pure
opening w = wv, the fibres will generate the same force, but the force
will be oriented either in the vertical direction or perpendicular to the
crack, respectively. Therefore, the expression for Vf represents the
tensile force across the critical crack obtained as if the crack undergoes
a pure opening w = wv, but Vf is oriented in the vertical direction
parallel to the crack displacement.
According to Eqs. 12–18, all mechanisms of shear resistance are
expressed directly or indirectly with DOFs εt,avg and Δc. Therefore, in
order to predict the shear strength of short FRC coupling beams, it is
necessary to predict the values of the DOFs at failure.
4. Shear strength prediction
Similarly to the derivation of VCLZ, DOF Δc is derived by assuming
that the failure of short coupling beams is triggered by crushing of the
critical loading zone. As the CLZ is at crushing, the strain along its
bottom inclined face is assumed equal to -0.0035kc, where kc is the
same compression softening factor used in Eq. 12. By making also ap-
propriate assumptions for the geometry of the CLZ, Δc has been ex-
pressed as [11]:
c = 3 × 0.0035kc lb1e cot
where lb1e is the characteristic length of the CLZ (Eq. 1) and α is the
angle of the diagonal of the beam (Eq. 2).
400
350
Vi
300 Vf
250 Vci
Shear forces, kN
200
150
100
50
Vd
0
0 0.5
t,avg,
Fig. 5. Solution of model equations applied to specimen 40-1-1 [8].
Engineering Structures 182 (2019) 501–509
The derivations so far considered two sets of fundamental equations:
compatibility of deformations (kinematics) and constitutive relationships
for the mechanisms of shear resistance. To predict the second DOF of the
kinematic model εt,avg, it is also necessary to consider the equilibrium of
the internal and external forces acting on the coupling beam. This is de-
monstrated graphically in Fig. 5 where on the horizontal axis is strain εt,avg
and on the vertical axis are the shear forces. The calculations begin by
determining the geometry of the kinematic model from Eqs. 1–4. Next, for
a given value of DOF εt,avg, DOF Δc is determined from Eq. (19), and the
deformations along the critical diagonal crack are obtained from Eqs. 9–10
(17)
and 17. In turn, these deformations are substituted in Eqs. 12–15 and 18 to
determine shear components VCLZ, Vci, Vs, Vd, and Vf. The sum of the five
components results in the shear capacity curve ΣVi in Fig. 5 (thick con-
tinuous line). As evident from the plot, the shear capacity decreases with
increasing strain in the longitudinal reinforcement, and therefore short
coupling beams exhibit strain effect in shear similarly to slender beams.
This effect is explained mainly with the compression softening of the CLZ,
and in some cases with the diminishing of the aggregate interlock re-
sistance due to the opening of the critical diagonal crack with increasing
εt,avg. The dowel action Vd, which typically has a relatively minor con-
tribution to the shear resistance, also diminishes with increasing strain.
When εt,avg reaches the yield strain of the steel, the longitudinal bars yield
in tension, and therefore do not provide any resistance to transverse dis-
placements.
To determine εt,avg at failure, the shear capacity must equal the
shear expressed from the tension in the longitudinal reinforcement. For
simplicity, the reinforcement is assumed linear elastic and the tension
stiffening effect of the concrete around the reinforcement is neglected,
therefore resulting in:
V = 2Es As t , avg (0.9d )/ a (20)
where EsAs is the axial stiffness of the reinforcement, EsAsεt,avg is the
tensile force in the reinforcement, 0.9d is the approximate lever arm of
the longitudinal forces in the end sections, and a is the length of the
beam. This shear force is derived by considering the moment equili-
brium of the coupling beam and is plotted with a thick dashed line in
Fig. 5. Graphically, the solution of the equations of the proposed model
lies at the intersection of the two thick curves where the shear forces are
in equilibrium. The solid curve is the shear capacity while the dashed
line can be viewed as the shear “demand” on the beam. The abscissa of
the intersection point is the predicted strain at failure and the ordinate
is the predicted shear strength. The calculations illustrated in Fig. 5
(19)
were performed for an FRC coupling beam tested by Cai et al. [8], see
beam 40-1-1 in Table 1. For this beam, the obtained shear strength
experimental-to-predicted ratio is Vexp/Vpred = 295 kN/277 kN = 1.06.
Equilibrium
VCLZ
V
vci
M
Vs Vs , Vf
VCLZ Vd
1 1.5 2
x10-3
504
B. Mihaylov Engineering Structures 182 (2019) 501–509

Table 1
Tests of short FRC coupling beams [7,8] and shear strength predictions.

Beam name a/d d, mm b, mm ρl, % fy, MPa ρv, % fyv, MPa fc′, MPa ρf, % lf, mm df, mm Load type Vexp, kN VFEM, kN Vpred, kN

*
30-2-1 2.32 345 150 1.21 363 0.56 296 40.5 1.0 32 0.76 C 227 220 205
40-2-1 43.1 238 227 208
50-2-1 52.9 243 238 220
60-2-1 66.7 250 253 235
70-2-1 70.1 253 256 239
80-2-1 80.7 255 266 249

40-1-1 1.16 345 150 1.21 363 0.56 296 43.1 1.0 32 0.76 C 295 355 277
40-1.5-1 1.74 292 278 246
40-2.5-1 2.90 190 185 185

50-2-0 2.32 345 150 1.21 363 0.56 296 55.6 0 32 0.76 C 203 220 188
50-2-0.5 54.5 0.5 238 236 204
55-2-1 54.8 1.0 244 242 222
50-2-1.5 55.9 1.5 250 252 242
50-2-2 55.3 2.0 256 253 257
50-2-2.5 54.1 2.5 257 255 255

S-10/M 1.11 360 100 2 469 1 384 37.5 1.0 50 1.05 M 299 324 257
S-15/M 1.67 1 38.2 282 302 243
S-20/M 2.22 1 38.7 262 288 236
S-15/W 1.67 0 45.0 130 114 149
S-15/S 1.67 2 45.2 358 364 301

Avg. (Vexp/VFEM) = 0.99 with COV = 6.4%.

Avg. (Vexp/Vpred.) = 1.08 with COV = 7.0%.
a = clear span; d = effective depth; b = width of cross section; ρl = ratio of top/bottom longitudinal reinforcement; fy = yield strength of longitudinal reinforce-
ment; ρv = stirrup ratio; fyv = yield strength of stirrups; fc = compressive cylinder strength of concrete; ρf = volumetric ratio of steel fibres; lf = fibre length;
df = fibre diameter; C/M = cyclic/monotonic load; Vexp = measured shear force at failure; Vpred = predicted strength.
* For brevity the name of beam CCB3-30-2-1FS is abbreviated as 30-2-1, similarly for the other 14 beams by Cai et al. [8].

5. Effect of test variables
Similar calculations were performed with a total of 20 beams listed
in Table 1 (15 beams by Cai et al. [8] and 5 tests reported by Kuang and
Baczkowski [7]). These tests allow for a systematic examination of four
important variables on the shear strength of short FRP coupling beams:
shear-span-to-effective-depth ratio a/d, concrete compressive strength
fc, stirrup ratio ρv and, most importantly, the volumetric ratio of steel
fibres ρf. As evident from Table 1, the beams had a/d ratios between
1.11 and 2.90, effective depths of about 350 mm, longitudinal re-
inforcement ratios of 1.21% or 2%, stirrup ratios between 0 and 2%,
and steel fibre ratios between 0 and 2.5%. The first 15 beams were
tested under reversed cyclic loading, while the remaining 5 were loaded
monotonically.
5.1. Effect of a/d ratio
Fig. 6a shows the measured and predicted effect of the a/d ratio on
the shear strength of four coupling beams tested by Cai et al. [8]. As can
be expected, the experimental points show that the strength decreases
significantly as the beams transition from a very short coupling beam
with a/d of 1.16 to a slender beam with a/d of 2.90. The kinematics-
based approach models well this trend, even though it produces slightly
conservative strength predictions. It can be seen from the predicted
components of shear resistance that the strength reduction is mainly
due to the diminishing of shear components VCLZ and Vci. As the span a
increases for a constant d, the flexural strains in the longitudinal re-
inforcement also increase, resulting in increased compression softening
of the CLZ as well as wider cracks with less aggregate interlock re-
sistance. In addition, as the span a increases, the critical shear crack
becomes flatter, resulting in a more slender and weaker critical loading
zone. Flatter cracks are also associated with larger amounts of stirrups
and steel fibres engaged along the cracks, and therefore larger shear
components Vs and Vf. According to the predictions in Fig. 6a, the
contribution of the fibres with a volumetric ratio of 1% varies from
about 10% of the shear strength at a/d of 1 to about 20% at a/d of 3. It
can also be seen that the dowel action contribution Vd is negligible
across the entire range of a/d ratios.
In addition to experimental results and kinematics-based predic-
tions, Fig. 6a also contains shear strength predictions obtained with a
nonlinear finite element model (FEM). The FE analyses were performed
in an earlier study [13] and are reported here for the sake of compar-
ison. Program VecTor2 based on the Disturbed Stress Field Model
(DSFM [18]) was used to model the beams with plane stress finite
elements. The DSFM is a smeared rotating crack model that originates
from the Modified Compression Field Theory for reinforced concrete
elements subjected to shear [14]. In the DSFM, the cracks are assumed
parallel to the principal compressive stress directions in the concrete,
while the principal strain directions deviate from the stress directions
due to slip displacements in the cracks. The slip displacements and
crack widths are used to calculate aggregate interlock stresses trans-
ferred across the cracks. To model FRC, this formulation has been ex-
tended to also account for the resistance of the steel fibers crossing the
cracks [15,19]. It can be seen from Fig. 6a that this approach over-
estimated the strength of the shortest beam by about 20%, but pro-
duced excellent predictions for the other three test specimens. It can
also be seen that the kinematics-based approach, which uses only two
DOFs to describe the deformation patterns of short coupling beams,
produces similarly accurate results to those of the FEM with thousands
of DOFs.
Similar observations for the effect of the a/d ratio can be made from
Fig. 6b which includes three tests by Kuang and Baczkowski [7]. Ac-
cording to the experimental points, the shear strength decreases with
increasing a/d ratios, even though this trend appears less pronounced
than in Fig. 6a. The kinematics-based approach predicts a nearly con-
stant trend up to an a/d ratio of about 1.7, which can be explained with
the predicted components of shear resistance. As before, components
VCLZ, Vci and Vd diminish, and Vs and Vf increase with increasing a/d
ratios. However, the test specimens by Kuang and Baczkowski had a
significantly higher stirrup ratio (ρv = 1% vs. 0.56%), and therefore the

505
B. Mihaylov Engineering Structures 182 (2019) 501–509

400 350

350 FEM 300

FEM
300 exp. exp.
250
pred.

Shear strength, kN
250 pred.
Shear strength, kN

200 Vf
200
Vf
Vci
Vci 150
150

100
100 Vs Vs
50 50
VCLZ VCLZ
0 Vd 0 Vd
1 1.5 2 2.5 3 35 45 55 65 75 85
a/d fc, MPa
(a)
(a)
350
FEM 400
300
exp.
350
pred.
250 Vf 300 Vf
Vci
Shear strength, kN

Shear strength, kN

200 250
Vci
150 sliding 200

Vs shear
150
Vs
100

50
VCLZ
Vd
0
1 1.5 2 2.5
a/d
(b)
Fig. 6. Measured and predicted effect of beam aspect ratio. (a) Tests 40-1-1, 40-
1.5-1, 40-2-1 and 40-2.5-1 [8]. (b) Tests S10-M, S15-M and S20-M [7].
increase of shear component Vs is predicted to compensate for the de-
crease of VCLZ, Vci and Vd in the range of a/d = 1–1.7. Furthermore, it
appears that the 1% stirrups were nearly sufficient to suppress shear
failures along diagonal cracks, and to force sliding shear failures in the
end sections of the beams. This is evident from the dashed line in
Fig. 6b which represents the resistance to sliding shear evaluated based
on a recently-proposed model for RC coupling beams [20]. The small
difference between the predictions of this model and those of the ki-
nematics-based approach are consistent with the failure modes ob-
served in the tests. The failures occurred neither at the vertical end
sections with yielding of the reinforcement and crushing of the con-
crete, nor along diagonal cracks as assumed in the kinematics-based
approach, but with the opening of steep flexure-shear cracks located in
between. Nevertheless, both prediction lines in Fig. 6b approximate
reasonably well the experimental results. It is also noted that the FEM
predictions lie almost perfectly on the shear sliding line.
100
VCLZ
50
0 Vd
35 45 55 65 75 85
fc, MPa
(b)
Fig. 7. Effect of concrete strength on shear capacity. (a) Measured and pre-
dicted strength of beams with a/d=2.32 (tests 30-2-1, 40-2-1, 50-2-1, 60-2-1,
70-2-1 and 80-2-1 [8]). (b) Predicted strength of beams with a/d=1.16
5.2. Effect of concrete strength
In addition to the a/d ratio, another important factor affecting the
shear strength of short FRC coupling beams is the compressive strength
of the concrete. Fig. 7 shows this effect observed in six tests of beams
with a/d of 2.32, longitudinal reinforcement ratio of 1.21%, stirrup
ratio of 0.56%, and steel fibre ratio of 1% [8] (Table 1). This plot is very
instructive as concrete strengths of up to 80–90 MPa become more and
more common in the construction of high-rise buildings. According to
the experimental points, as the concrete strength was doubled from
40.5 MPa to 80.7 MPa, the measured shear strength increased by about
12%. This trend is modelled very well by the nonlinear FEM, even
though the model overestimated slightly the shear strength of the
specimen with the strongest concrete. As also evident from the plot, the
kinematics-based approach is slightly conservative for the lower

506
B. Mihaylov Engineering Structures 182 (2019) 501–509

400 by 33%. This is explained with the larger contribution of the CLZ whose
strength is proportional to fc0.8 and sin2α (Eq. 12). Therefore, the
350 exp. shorter is the beam and the larger is the inclination of its diagonal α, the
larger is the shear component VCLZ as compared to the other compo-
FEM
300 nents and the larger is the effect of fc.

250 Vf
Shear strength, kN

5.3. Effect of stirrup ratio

200 Vci
150
Vs
100
50
VCLZ
0 Vd
0 0.5 1 1.5
v, % followed by a horizontal plateau for larger ratios. The linear branch
Fig. 8. Measured and predicted effect of stirrup ratio (tests S-15/W, S-15/M and
S-15/S by [7]).
concrete strengths and approaches the experimental results as fc ap-
proaches 90 MPa. The model predicts that shear components VCLZ, Vci
and Vf increase in a very similar manner, while Vs remains constant and
equal to the yield strength of the stirrups crossing the critical diagonal
crack. It should be noted that the continuous prediction line is obtained
without taking into account that the cracks in high-strength concretes
propagate through the coarse aggregates, and therefore have less ca-
pacity to interlock. This effect can be modelled by reducing the max-
imum aggregate size ag in Eq. 13 from its actual value to 0 as the
concrete strength increases from 60 MPa to 70 MPa [21]. The results
obtained based on this approach are shown with a dashed line in Fig. 7.
As the effect of the concrete strength in Fig. 7a is relatively modest,
the kinematics-based approach is used to demonstrate cases where this
effect is significantly larger. Fig. 7b is generated with the same beam
Fig. 8 shows how the shear strength of coupling beams increases
with increasing amounts of stirrups. The a/d ratio of the beams was
1.67, the concrete strength about 42 MPa, the longitudinal reinforce-
ment ratio 2%, and the fibre volumetric ratio 1% [7] (Table 1). Ac-
cording to the experimental points, as the stirrup ratio was increased
from 0 to 2%, the shear strength increased by a factor of 2.8. The
prediction curve approximates well the first two experimental points
and underestimates by 19% the shear capacity of the test specimen with
maximum stirrup ratio. It can be seen that the shear strength is pre-
2 dicted to increase almost linearly up to a stirrup ratio of about 1.3%,
corresponds to diagonal tension failures modelled with the kinematics-
based approach, while the upper bound on this branch corresponds to
failures in the end sections calculated according to [20]. Within the
linear branch, the stirrup shear contribution increases linearly, while
the rest of the shear mechanisms remain approximately constant.
Within the plateau, the longitudinal reinforcement in the end sections is
predicted to yield, and the concrete in the compression zones to crush
under the combined action of flexural compression and shear.
250
200
Vf
150 Vci
Shear strength, kN

properties as Fig. 7a, but the length of the beams is reduced by one half,
and therefore the a/d ratio is 1.16. It can be seen that in this case as fc is
doubled from 40 to 80 MPa, the shear strength is predicted to increase
100
Vs
350
50
300 VCLZ
FEM
exp. Vd
250 0
0 0.5 1 1.5 2 2.5
Vf
Shear strength, kN

200 1

150 Vci 0.8

0.6
%

100
Vs
v,

0.4
50
0.2
VCLZ
0 Vd 0
0 0.5 1 1.5 2 0 0.5 1 1.5 2 2.5
f, % f, %

Fig. 9. Measured and predicted effect of steel fibre volumetric ratio (tests 50–2- Fig. 10. Combinations of ρf and ρv to achieve the same shear strength (beam
0, 50–2-0.5, 55–2-1, 50–2-1.5, 50–2-2, and 50–2-2.5 by [8]). properties identical to those of specimen S-10/M).

507
B. Mihaylov Engineering Structures 182 (2019) 501–509

0.35 material properties.

To perform more rigorous comparisons, Fig. 10 shows results from a
0.3
series of beams with a constant shear strength of 200 kN, but different
0.25 fc = amounts of stirrups and steel fibres. Apart from the ρv and ρf ratios, the
90 MPa properties of the beams are identical to those of test specimen S/15-M
0.2 (Table 1). The steel fibre ratio is increased from 0 to 2.5% (though 2.5%
v/ f

60
0.15 is rare in practice due to corrosion problems), and the stirrup ratio is
30 decreased appropriately to maintain a constant shear strength. It can be
0.1 h=1000 mm seen from the top plot that as the shear contribution of the fibres Vf
d=950 mm lf=50 mm increases linearly with increasing ρf, the contribution of the stirrups Vs
0.05
fyv=400 MPa df=1.05 mm decreases linearly with the same rate, and components VCLZ, Vci and Vd
0 remain constant. The bottom plot shows that the stirrup ratio decreases
0.5 1 1.5 2 2.5 3 linearly from 0.65% to 0.2%, and that 2.5% of fibres are sufficient to
replace 0.45% of stirrups. Therefore, the efficiency of the fibres can be
0.35 quantified by the ratio of 0.45%/2.5% = 0.18, where a unit value
would correspond to equal efficiency of fibres and stirrups.
0.3
It is also of interest to use the kinematics-based approach to identify
0.25 the main factors affecting the efficiency of the fibres, and to study
systematically the effect of these factors. To express the fibre efficiency
fc=
0.2 ratio, it is necessary to equate the shear strength contribution of the
v/ f

90 MPa
0.15 60
0.1 h=1000 mm 30
d=950 mm lf=32 mm
0.05
fyv=400 MPa df=0 .76 mm
0
0.5 1 1.5 2 2.5 3 where σf is the stress transferred by fibres, fyv is the yield strength of the
a/d
Fig. 11. Effectiveness of steel fibre reinforcement as compared to stirrups.
5.4. Effect of fibre volume ratio
Fig. 9 shows how the shear strength of coupling beams increases
with increasing amounts of steel fibres. The six tests specimens used for
the plot had an a/d ratio of 2.32, concrete strength of about 55 MPa,
longitudinal reinforcement ratio of 1.21%, and stirrup ratio of 0.56%
[8] (Table 1). According to the experimental points, as the volumetric
fibre ratio was increased from 0 to 2.5%, the shear strength increased
by a factor of 1.27. The predicted shear strengths agree well with the
test results at small and large volumes of fibres, and slightly under-
estimate the shear strength for intermediate volumes. It has been
identified that this slight discrepancy is mainly due to the modelling of
the compression softening effect which is based on an expression de-
rived from shear panels without fibres (factor kc in Eq. 12). As evident
from the predicted shear components, all shear contributions remain
nearly constant, except for the contribution of the fibres which in-
creases linearly with ρf. As in Fig. 8, the plateau of the prediction curve
for ρf > 2% is due to the yielding of the longitudinal reinforcement and
crushing of the concrete in the end sections (sliding shear failure) [20].
6. Effectiveness of steel fibre reinforcement as compared to
stirrups
By comparing Figs. 8 and 9, it can be seen that the stirrup ratio and
steel fibre ratio affect the shear capacity in a very similar manner. In
both cases the contribution of the shear reinforcement to the shear
capacity of the members increases linearly with the reinforcement ratio,
while the remaining shear mechanisms remain approximately constant.
Furthermore, both stirrups and steel fibres can be used to suppress di-
agonal tension failures and to shift the failure to the end sections.
However, the steeper linear branch in Fig. 8 shows that stirrups are
more effective than steel fibres in increasing the shear strength of short
coupling beams. This conclusion is of general validity, even though the
beams in the two figures did not have the same geometrical and
fibres Vf (Eq. (18)) to the contribution of the stirrups Vs (Eq. 14). After
performing some rearrangements, and also assuming that the stirrups
yield at failure, the efficiency ratio becomes
(w v )
v
=
f
0.9fyv cos
f 1 (21)
stirrups and α1 is the angle of the critical crack. To evaluate the right-
hand-side of Eq. (21), it is necessary to make further simplifications and
to select beam properties of practical relevance. To evaluate cos α1
based on Eq. 2, the analysis will focus on a series of beams with a
constant depth h = 1000 mm (d ≈ 950 mm) and variable span a. In
addition, it is necessary to simplify the expression for the crack dis-
placement wv which depends on both DOFs of the kinematic model and
controls the stress in the fibres. To that end, it is assumed that the strain
in the longitudinal reinforcement εt,avg ≈ 0.002 and the length of the
reinforcement contributing to wv is lk 2l 0 = 2 × 1.5(h d )cot 1, re-
sulting in:
0.00116 a2 + h2 cot
wv = 0.003(h d )cot2 +
1
0.8 + 0.34(1 + cot2 ) (22)
With this expression substituted in Eq. (16), the ρv/ρf ratio is eval-
uated for a series of beams with a yield strength of the stirrups
fyv = 400 MPa and concrete strengths fc = 30–90 MPa. Fig. 11 presents
the results from these calculations for two types of fibres: fibre diameter
df = 1.05 mm and length lf = 50 mm ≈ 48df (top plot), and
df = 32 mm and lf = 32 mm ≈ 42df (bottom plot). The longer fibres
were used in the tests by Kuang and Baczkowski [7] while the shorter
ones in the tests by Cai et al. [8]. It can be seen from both plots that the
effectiveness of the fibres is affected by the a/d ratio of the member as
well as the concrete strength. The ρv/ρf ratio decreases with increasing
beam aspect ratio up to a/d of about 1.5, and remains approximately
constant for more slender coupling beams. This trend is explained
mainly by the presence of cosα1 in the denominator of Eq. (22), as for
members with a/d > 1.5 the angle of the critical shear crack remains
constant at α1 = 35° (Eq. 2). A more physical explanation of this trend
has to do with the properties of FRC: steel fibres will transfer shear even
if the critical crack is very steep (i.e. small a/d and large α1) as long as
there are slip displacements in the crack associated with DOF Δc. Stir-
rups, in contrast, are not as effective in members with small a/d ratios
because the steep critical cracks cross very few stirrup legs.
In addition to small a/d ratios, the effectiveness of steel fibres is also
increased by high concrete strengths. It can be seen from Fig. 11 that as
fc is increased from 30 MPa to 90 MPa, the ρv/ρf ratio increases by a
factor √(90/30) = 1.73. It is evident that this effect is due to the better

508
B. Mihaylov
2 2
1.8
1.6
1.4 monotonic cyclic
Vexp/Vpred
1.2
1 1
0.8
0.6
0.4
0.2
0 0
1 1.5 2 2.5
a/d
Fig. 12. Experimental-to-predicted strength ratios for 20 tests; Avg. = 1.08 and COV = 7.0%
bond between the fibres and the concrete, which requires a larger stress
to pull the fibres out of the concrete matrix. By comparing the two plots
in Fig. 11, it can be concluded that the longer fibres with larger lf/df
ratio are more effective in replacing conventional transverse re-
inforcement.
Finally, to complete the validation of the proposed kinematics-based
approach, Fig. 12 shows the shear strength experimental-to-predicted
ratios Vexp/Vrped for all 20 tests listed in Table 1. The ratios are plotted
as functions of a/d and the steel fibre ratio showing no visible bias
across the entire range of the test variables. For this test database, the
average Vexp/Vrped ratio is 1.08 and the coefficient of variation (COV) is
7.0%. For comparison, the nonlinear FEM produced an average ratio of
0.99 with a COV of 6.4%.
7. Conclusions
This paper presented a kinematics-based approach for evaluating
the shear strength of short fibre reinforced concrete (FRC) coupling
beams. This approach was validated with 20 tests from the literature
and its accuracy was compared to that of a nonlinear FEM im-
plementing the disturbed stress field model. As the kinematics-based
approach is relatively simple and predicts explicitly the contribution of
five mechanisms of shear resistance, it was used to study the effec-
tiveness of steel fibres in comparison to that of conventional stirrups.
The main conclusions from these studies are the following:
(1) According to the kinematics-based approach, short coupling beams
exhibit strain effect in shear, and their shear strength can be pre-
dicted by intersecting a shear capacity curve with a shear “demand”
line. The strain effect consists of decreasing shear capacity as the
strains in the longitudinal reinforcement increase and the critical
shear cracks widen.
(2) As demonstrated in the parametric studies, the kinematics-based
approach can model adequately the effect of a/d ratio, concrete
strength, stirrup ratio and steel fibre ratio on the shear strength of
short FRC coupling beams. For the 20 tests, the kinematics-based
approach produced shear strength experimental-to-predicted ratios
with an average value of 1.08 and a coefficient of variation of only
7.0%. More tests are however needed to further evaluate the re-
liability of the model.
(3) The kinematics-based approach, which uses only two degrees of
freedom to describe the deformation patterns of short coupling
beams, predicts the shear strength with approximately the same
accuracy as the complex FEM with thousands of DOFs.
(4) With a parametric study involving beams with an a/d ratio of 1.67,
it was shown that the required amount of stirrups to achieve the
same shear strength decreases linearly with the increase of the steel
fibre ratio. It was also shown that a member with 0.65% of stirrups
and no fibres possesses the same shear strength as a member with
Engineering Structures 182 (2019) 501–509
1.8
1.6
1.4
Vexp/Vpred
1.2
0.8
0.6
0.4
0.2
3 0 0.5 1 1.5 2 2.5
f, %
0.2% of stirrups and 2.5% of fibres.
(5) According to the kinematics-based approach, the efficiency of steel
fibres as compared to that of conventional stirrups depends mainly
on the a/d ratio of the beam, the concrete compressive strength and
the properties of the fibres. The fibre efficiency decreases with in-
creasing a/d and increases with increasing concrete strength.
References
[1] Moehle JP, Ghodsi T, Hooper JD, Fields DC, Gedhada R. Seismic design of cast-in-
place concrete special structural walls and coupling beams. NEHRP Seismic Design
Technical Brief No. 6, produced by the NEHRP Consultants Joint Venture, a part-
nership of the Applied Technology Council and the Consortium of Universities for
Research in Earthquake Engineering, for the National Institute of Standards and
Technology, Gaithersburg, MD, NIST GCR 11-917-11REV-1; 2011.
[2] Barney GB, Shiu KN, Rabbat BG, Fiorato AE, Russell HG, Corley WG. Behavior of
coupling beams under load reversals. Research and Development Bulletin RD068.
01B. Portland Cement Association; 1980. p. 22.
[3] Tassios TP, Moretti M, Bezas A. On the behavior and ductility of reinforced concrete
coupling beams of shear walls. ACI Struct J 1996;93(6):711–20.
[4] Galano L, Vignoli A. Seismic behavior of short coupling beams with different re-
inforcement layouts. ACI Struct J 2000;97(6):876–85.
[5] Canbolat B, Parra-Montesinos G, Wight J. Experimental study on seismic behavior
of high-performance fiber-reinforced cement composite coupling beams. ACI Struct
J 2005;102(1):159–66.
[6] Yun H-D, Kim S-W, Jeon E, Park W-S, Lee Y-T. Effects of fiber-reinforced cement
composites’ ductility on the seismic performance of short coupling beams. Mag
Concr Res 2008;60(3):223–33.
[7] Kuang JS, Baczkowski BJ. Steel-fiber-reinforced concrete coupling beams subjected
to monotonic loading. Mag Concr Res 2009;61(1):35–41.
[8] Cai G, Zhao J, Degée H, Vandoren B. Shear capacity of steel fiber reinforced con-
crete coupling beams using conventional reinforcements. Eng Struct
2016;128:428–40.
[9] Mihaylov BI, Bentz EC, Collins MP. Two-parameter kinematic theory for shear be-
havior of deep beams. ACI Struct J 2013;110(3):447–56.
[10] Mihaylov BI, Hannewald P, Beyer K. Three-parameter kinematic theory for shear-
dominated reinforced concrete walls. J Struct Eng 2016;142(7).
[11] Mihaylov BI, Hunt B, Bentz EC, Collins MP. Three-parameter kinematic theory for
shear behavior of continuous deep beams. ACI Struct J 2015;112(1):47–57.
[12] Mihaylov BI, Franssen R. Three-parameter kinematic approach for shear behaviour
of short coupling beams with conventional reinforcement. fib Bulletin
2018;85:283–98.
[13] Mihaylov BI, Liu J, Lobet R. A kinematic approach for the complete shear behavior
of short FRC coupling beams. ACI Struct J Spec Publ 2018;SP-328:8.1–8.20.
[14] Vecchio FJ, Collins MP. The modified compression field theory for reinforced
concrete elements subjected to shear. ACI J 1986;83(2):219–31.
[15] Voo JYL, Foster SJ. Variable engagement model for fibre-reinforced concrete in
tension. Uniciv Report No. R-420, University of New South Wales, School of Civil
and Environmental Engineering; 2003. p. 86.
[16] Lee S-C, Cho J-Y, Vecchio FJ. Diverse embedment model for steel fiber-reinforced
concrete in tension: model development. ACI Mater J 2011;108(5):516–25.
[17] Lee S-C, Cho J-Y, Vecchio FJ. Simplified diverse embedment model for steel fiber
reinforced concrete elements in tension. ACI Mater J 2013;110(4):403–12.
[18] Vecchio FJ. Disturbed stress field model for reinforced concrete: formulation. J
Struct Eng 2000;126(9):1070–7.
[19] Lee S-C, Cho J-Y, Vecchio FJ. Analysis of steel fiber-reinforced concrete elements
subjected to shear. ACI Struct J 2016;113(2):275–85.
[20] Mihaylov BI, Franssen I. Shear-flexure interaction in the critical sections of short
coupling beams. Eng Struct 2017;152(12):370–80.
[21] AASHTO, AASHTO LRFD bridge design specifications, 4th ed. Washington, DC:
American Association of State Highway Officials; 2007. 1526p.
509