You are on page 1of 21

Finite Elements in Analysis and Design 50 (2012) 266–286

Contents lists available at SciVerse ScienceDirect

Finite Elements in Analysis and Design


journal homepage: www.elsevier.com/locate/finel

Nonlinear finite element analysis of steel fiber-reinforced concrete members


using direct tension force transfer model
Deuck Hang Lee a,1, Jin-Ha Hwang a,1, Hyunjin Ju a,1, Kang Su Kim a,n, Daniel A. Kuchma b,2
a
Department of Architectural Engineering, University of Seoul, 90 Jeonnong-dong, Dongdaemun-gu, Seoul 130-743, Korea
b
Department of Civil Engineering, University of Illinois at Urbana Champaign, 2106 Newmark Laboratory, 205 N. Mathews Ave., Urbana, IL 61801, USA

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

Article history: Our understanding on the shear behavior of SFRC members is still quite limited due to the complex
Received 25 August 2011 inter-combinatorial mechanisms of shear transfer. Some efforts have been made to extend the
Received in revised form application of modified compression field theory and the soften truss model to SFRC members. It is
20 October 2011
difficult, however, for such models to reflect some of the key behavioral characteristics of SFRC, such as
Accepted 22 October 2011
the fiber directionality at the crack interfaces, the differences of bond strengths by fiber types, and the
Available online 8 November 2011
pull-out failure of fibers at crack. Thus, in this study, a nonlinear finite element analysis procedure has
Keywords: been developed that sufficiently reflects the material properties and behavioral characteristics of SFRC,
NLFEM and can be applied to obtain the shear behavior. The accuracy and rationality of the proposed model
Steel fiber
were also verified in this study by comparing to the recent experimental results on the SFRC shear
SFRC
panels.
Shear
Secant modulus & 2011 Elsevier B.V. All rights reserved.
DTFTM

1. Introduction etc., and most studies have merely provided empirical methods on
shear strength of SFRC members based on experimental results.
Steel fiber-reinforced concrete (SFRC) is a composite material, Recently, Susetyo [8], the research group at the University of
which improves the brittle material characteristics of conventional Toronto, reported the effect of inclusion of steel fibers on shear
concrete by adding steel fibers. According to the many experi- strength enhancement of SFRC shear panels, whose results was
mental and theoretical researches on shear strength of SFRC also reflected on a non-linear finite element analysis (NLFEA) by
[1–10], it is reported that inclusion of a sufficient amount of steel modifying the existing concrete tension stiffening and tension
fiber in concrete improves the tensile strength of concrete before softening curves [27] to be adjusted for SFRC as the composite
and after cracking [1–5], ductility of members [5–8], and concrete material. Such a method, however, which is based on the simple
shrinkage control capacity [8–10]. Since the steel fibers at crack adjustment of the constitutive equation of concrete in tension,
interfaces control the spacing, distribution, and widths of cracks cannot sufficiently reflect the fiber directionality and the pull-out
[11,12], it has been emphasized that the shear resistance at crack failure mode at the crack interface, and requires too many panel
interfaces of SFRC members is greatly improved [13–18]. As the tests to consider the effects of all the types of steel fibers, the
advantages of SFRC are being reported, there have been many volume fraction of fibers (Vf), the concrete compressive strengths
0
studies on investigating the shear strength enhancement of SFRC (f c ), etc. Therefore, to effectively reflect the bond characteristics
members [13–24]. The shear behavior of SFRC members, however, between fibers and surrounding concrete, the directionality and
has yet to be clearly understood due to its complex mechanisms, reinforcing effect of fibers on estimation of shear behavior of SFRC
such as shear transfer at crack interfaces (aggregate interlock), members, in this study, the direct tension force transfer model
directionality of fibers at crack interfaces [3,26], bond behavior (DTFTM), in which steel fibers were modeled as a simple tension
between fibers and surrounding concrete [21], dowel action [25], force transfer element at crack interfaces, was derived to appro-
priately evaluate the pull-out failure of steel fibers from sur-
rounding concrete, and it was also applied to a non-linear finite
n
Corresponding author. Tel.: þ82 2 2210 5707; fax: þ82 2 2248 0382. element analysis (NLFEA). Modeling the steel fibers in this way
E-mail addresses: dklee@uos.ac.kr (D.H. Lee), makes it possible to estimate the shear behavior of SFRC members
asorange@hanmail.net (J.-H. Hwang), fis00z@uos.ac.kr (H. Ju),
kangkim@uos.ac.kr (K.S. Kim), kuchma@illinois.edu (D.A. Kuchma).
with minimum experimental efforts, such as a simple bond
1
Tel.: þ82 2 2210 5374; fax: þ82 2 2248 0382. strength test of steel fibers. The nonlinear finite element analysis
2
Tel.: þ1 217 333 1571; fax: þ1 217 333 9464. model, proposed in this study, can reflect the differences of bond

0168-874X/$ - see front matter & 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.finel.2011.10.004
D.H. Lee et al. / Finite Elements in Analysis and Design 50 (2012) 266–286 267

Notations Esi secant modulus of reinforcement in i direction


0
fc specified compressive strength of concrete
a2 angle of inclination of crack in element coordinate fcr stress in concrete at cracking
ai orientation of reinforcement relative to element coor- fct tensile strength of SFRC
dinate system fsi stress in i th reinforcement direction
bi orientation of element x, y axes relative to global fy yield stress of reinforcing steel
coordinate system ffu ultimate strength of steel fiber
ec1 principal tensile strain in concrete Gc secant shear modulus of concrete
0
ec2 principal compressive strain in concrete Gc secant shear modulus of SFRC
eco strain at specified compressive strength of concrete Lf length of fiber
ecr cracking strain nf effective number of steel fiber across crack
es strain in steel reinforcement nw number of steel fiber across crack per unit area
z rotation angle in transformation matrix q crack spacing control factor
l direction coefficient of fiber in arbitrary space Vf volume fraction of steel fiber
u poisson’s ratio smy average shear crack spacing
ri reinforcement ratio in ith direction smx average crack spacing of xelem direction in element
sc1 principal tensile stress in conrete coordinate system
sc2 principal compressive stress in concrete smy average crack spacing of yelem direction in element
sc2,max maximum principal compressive stress coordinate system
ssf,1 stress in steel fiber in principal direction not con- T sf tensile force in steel fibers between cracks
sidering effective area of fiber w crack width of convertional reinforced
ssfrp,1 stress in steel fiber in principal direction considering concrete member
effective area of fiber wmax maximum crack width of SFRC member when fiber
s1sf rp,max maximum stress in steel fiber when bond stress pull-out failure
reaches to maximum IBS wsfrc crack width of SFRC member
tmax maximum bond strength between concrete and fibers feg element strain matrix
tu bond strength between concrete and fibers fsg element stress matrix
Acs area of inclined crack plane {D} nodal displacement matrix of structure
Afp surface area of fiber that bond stress is developed ½B element shape matrix
Asf area of one piece of steel fiber [D] composite material stiffness matrix in global coordi-
Asx area of reinforcing steel in xelem direction nate system
df shape coefficient of fiber [Dc] concrete material stiffness matrix in global coordi-
D diameter of fiber nate system
Ec modulus of elasticity of concrete [Dc]1  2 concrete material stiffness matrix in principal coordi-
Es modulus of elasticity of reinforcement and structural nate system
steel ½Dc 012 SFRC material stiffness matrix in global coordinate
Esf modulus of elasticity of steel fiber system
Ec1 secant modulus of concrete in principal tensile strain [De] elastic material stiffness matrix
direction [Ds]i reinforcement material stiffness matrix in global
0
Ec1 secant modulus of SFRC in principal tensile strain doordinate system
direction ½Ds 0i reinforcement material stiffness matrix in ith
Ec2 secant modulus of concrete in principal compressive direction
strain direction [F] applied nodal force matrix
0
Ec2 secant modulus of SFRC in principal compressive ½k element stiffness matrix
strain direction ½K structure stiffness matrix
Esf rp secant modulus of steel fibers in principal compres-
sive strain direction

strengths depending on the effective areas of steel fibers at crack et al. [25] also experimentally verified that the shear strength of
interfaces and the types (or shapes) of steel fibers. The accuracy SFRC members considerably improved compared to those of RC
and rationality of the proposed model were verified by comparing members without shear reinforcement. Sharma [14], based on
to the experimental results of SFRC shear panel tests subjected to experiments with various amounts of shear reinforcements,
pure shear. proposed a relatively simple shear strength equation, in which
the shear span-depth ratio (a/d) and the tensile strength of SFRC
were reflected, and ACI Committee 544 [1] have recommended
2. Review of previous researches and specified this equation for shear strength evaluation of SFRC
members in the ACI manual of concrete practice 2008 [27].
The SFRC has been used as structural material since the 1960s Narayanan and Darwish [15] also conducted tests on 49 SFRC
[1–3]. Batson et al. [13] conducted shear tests on 102 SFRC beams beams, and proposed a shear strength equation considering arch
with the volume fraction of steel fibers (Vf) and the shear span- action that were influenced by the shear span-to-depth ratio (a/d)
depth ratios (a/d) as the key parameters, and reported that of SFRC beams. Ashour et al. [18] proposed a shear strength
inclusion of steel fibers in concrete greatly improved the shear equation for high-strength concrete members with steel fibers,
strength compared to conventional reinforced concrete. Swamy which considered the shear span-depth ratio (a/d), fiber volume
268 D.H. Lee et al. / Finite Elements in Analysis and Design 50 (2012) 266–286

fractions (Vf), longitudinal reinforcement ratios (rs), etc., as key element analysis algorithm, and the accuracy and rationality of
variables. Khuntia et al. [19] also proposed a simple equation for the proposed model are also verified by comparing to the recent
shear strength estimation of SFRC members, assuming 45 degrees experimental results on the SFRC shear panels.
of cracking angle, that reflects arch action, fiber volume fractions
(Vf), fiber orientation, and effective embedded length of fibers at
cracks. Recently, Oh et al. [20] proposed a simple shear strength 3. Direct tension force transfer model (DTFTM)
equation, in which the longitudinal reinforcement ratio (rs), fiber
volume fraction (Vf), and shear span-depth ratio (a/d) were As shown in Fig. 1(a), a certain number of steel fibers according
considered as influencing parameters. to the volume fraction of the steel fiber (Vf) resist tensile stresses
Tan et al. [16] made a pioneering attempt to expand the applica- in perpendicular direction to the cracks. Also, since steel fibers can
tion of rotated angle-softened truss model (RA-STM), which was be assumed to behave in perfectly bond with concrete before
originally developed by Hsu [29], into the shear behavior of SFRC pulled-out, the stresses of the steel fibers, ssf,1 until they reached
members. Although their analysis showed a good agreement with the ultimate interfacial bond strength, can be expressed as
test results, however, the compression softening curve and the ssf ,1 ¼ Esf ec1 ð1Þ
tension-stiffening curve of SFRC subjected to biaxial stresses, which
were main concept in their proposed model, were lack of experi- where Esf is the modulus of elasticity of steel fibers and 200 GPa is
mental supports at that time. The test specimens used for the used in this study [21], and ec1 is the average principal tensile strain
verification of the proposed model had relatively low compressive in concrete in the rotated-crack angle model [28–31], assuming that
strength. Later, Tan et al. [17] also proposed an analytical model for the average principal tensile strain in concrete is identical to the
SFRC members that was based on the modified compression field average tensile strain in steel fibers, which is reasoned by the
theory (MCFT) proposed by Vecchio and Collins [30]. They [33] also consideration of the directionality of steel fibers in crack direction
expanded their research to evaluate the shear strength of partially that are explained later in detail. Since not all steel fibers resist the
prestressed steel fiber-reinforced concrete beams using numerical tensile stresses in the principal tensile direction, the effective average
analysis method presented by Vecchio [31,32]. tensile stresses of the steel fibers in the crack direction should be
Continuous efforts on the development of the numerical estimated. Therefore, if the number of steel fibers across cracks (in 1–
analysis model for SFRC have been made. Lihua et al. [34] showed 2 direction shown in Fig. 1) is defined as nf, the tension force in the
relatively accurate analysis results on the shear behavior of SFRC steel fibers at the crack interface (Tsf) is
deep beams using the commercial finite element analysis pro- T sf ¼ ssf ,1 nf Asf ð2Þ
gram. Their analysis model, however, was quite complicated due
to the three-dimensional elements used in their model, was lack where Asf is the sectional area of a steel fiber. Then, the effective
of experimental evidences on the material constitutive models, average tensile stresses in steel fibers at crack interface (ssfrp,1) can be
and did not reflect the primary influential factors of SFRC, such as calculated as following:
directionality, the fiber volume fraction, and the type of fibers. T sf ssf ,1 nf Asf
Susetyo [8] proposed material models suitable for SFRC, based on
ssf rp,1 ¼ ¼ ð3Þ
Acs Acs
the results of experiments on eight SFRC panels. They performed
where Acs is the area of the inclined crack plane. The effective number
the nonlinear finite element analysis based on the proposed
of steel fibers across the cracks (nf) at the crack interface in Eq. (3) can
material models, and had good results of the estimation of shear
be determined by multiplying the area of the inclined crack plane
strength. In addition, they also mentioned that it would be
necessary to develop an analysis model that could more accu-
rately simulate the pull-out failure of SFRC between steel fibers
and concrete at the crack interfaces.
Crack
Although the simple shear strength equations [14,15,18–20] Y Steel fiber
quite simply reflect the key variables such as the shear span- 1 2
Reinforcement
depth ratio (a/d), the volume fraction of the steel fiber (Vf), and
the longitudinal reinforcement ratio (rs), etc., they are lack of the X
theoretical bases as the effect of main influential factors are x : Arbitrary length
empirically determined, based on the results of the experiments
in the aforementioned studies. On the other hand, the shear
behavior estimation models [16,17] and the numerical analysis
models [33,34], which were based on the smeared truss model
[29,30], showed relatively good agreements with experimental
results of SFRC members. These shear behavior models, however,
cannot capture the clear shear transfer mechanism of SFRC d f = 0.5 d f = 0.75 d f = 1.0
members, including the pull-out failure of steel fiber from Straight type crimped-shaped hooked-shape
surrounding concrete at the crack interfaces, [8] and require Approximately L =L /4
heavy experimental efforts, such as panel tests, for the modifica- Aggregate
tion of the constitutive relations of SFRC due to the mixed
behavioral characteristics of steel fibers and concrete. [21] There-
fore, in this study, the direct tension force transfer model
(DTFTM) is introduced, in which randomly distributed fibers are
modeled as the direct tensile stress transfer elements by convert- : Bond stress
ing the number of fibers in the direction perpendicular to cracks Crack width = 0.25 f
into the equivalent effective area, and the differences of bond τ =τd
strengths by the types of fibers are also considered. Furthermore, Fig. 1. Concept of direct tensile contribution model. (a) Steel fiber behavior
a shear behavior model for SFRC members subjected to shear is between cracks; (b) Type of steel fibers and shape factors; (c) Inclined cracks
proposed by inserting the DTFTM into the nonlinear finite and steel fibers between cracks.
D.H. Lee et al. / Finite Elements in Analysis and Design 50 (2012) 266–286 269

(Acs) by the number of steel fibers across cracks per unit area, nw, as in IBS (tmax ) that considers the shape of steel fibers can be expressed
as follows:
nf ¼ nw Acs ð4Þ
tmax ¼ tu df ð9Þ
This study used the effective number of steel fibers per the
unit area (nw) proposed by Romualdi et al. [3], and Soroushian where the shape coefficient of the fiber (df) is, as shown in
and Lee [26], which is expressed as Figs. 1(b), 1.0 for the hooked type, 0.75 for the crimped type,
Vf and 0.5 for the straight type, as proposed by Narayanan and
nw ¼ l ð5Þ Darwish [15]. Therefore, the maximum tensile stress (s1sf rp,max ) in
Asf
the steel fiber at the crack interface, when the bond stress reaches
where Vf is the volume fraction of steel fibers and l is the fiber the maximum IBS expressed in Eq. (7), is modified as
orientation factor in arbitrary space, for which 0.41 was used in
this study as proposed by Romualdi et al. [3] By substituting the tu df Af p nf
s1sf rp,max ¼ ð10Þ
effective number of steel fibers at the crack interface (nw), which Acs
was calculated using Eqs. (4) and (5), into Eq. (3), the effective
As the steel fibers are randomly distributed, however, the
average tensile stress in the steel fibers at crack interface (ssfrp,1)
embedded length of the steel fiber (lb) in the concrete when
can be estimated as
cracks occur differs by fibers. In this study, the effective
ssf rp,1 ¼ 0:41ssf ,1 V f ð6Þ embedded length of steel fiber was assumed as 1/4 of the fiber
length (Lf) on average [16], as shown in Fig. 1(c). Thus, Eq. (10) can
Since the existing analysis models [8,16,17,33,34] reflected the
be expressed as
inclusion and strengthening effect of the steel fibers by adjusting
the constitutive equation in tension to be appropriate with SFRC as tmax ðpDLf =4Þð0:41ðV f Acs =Asf ÞÞ
s1sf rp,max ¼ ð11Þ
a composite material, they had difficulties in considering the pull- Acs
out failure of steel fibers due to the loss of the bond resistance
where D is the diameter of the fiber. Also, Eq. (11) can be further
between fibers and concrete, and tended to overestimate the shear
expressed simply as follows:
strength and the deformation capacity of SFRC members. In this
study, steel fibers were modeled as direct tensile stress transfer Lf
elements so that the bond strength of steel fibers at crack s1sf rp,max ¼ 0:41V f tmax ð12Þ
D
interfaces can be directly considered; based on the bond test
results of Lim et al. [21] as shown in Fig. 2, it was assumed that In this study, steel fibers at the crack interfaces are modeled as
once the bond stress of steel fibers reaches the ultimate bond direct tensile stress transfer elements and utilized in a smeared
strength, there would be no tensile stress increase in steel fibers. crack truss model for the estimation of shear behavior of SFRC
members. In this way, the bond characteristics of steel fibers can
Therefore, the maximum tensile stress in steel fibers (s1sf rp,max ),
be easily counted for in more detail, considering shape, amount,
limited by the ultimate bond stress, can be expressed as following:
and directionality of steel fibers, and the pull-out failure mechan-
tmax Af p nf ism of steel fibers embedded in concrete, which is an important
s1sf rp,max ¼ ð7Þ
failure mode of SFRC members, can be properly described.
Acs
where tmax is the maximum interfacial bond stress (IBS) between
concrete and steel fibers, and Afp is the surface area of a steel fiber.
In this study, the bond strength of a steel fiber (tu) suggested by 4. Constitutive models and material stiffness matrix
Voo and Foster [35] is used, and expressed as follows:
The stress–strain relationship in the plane stress state of the
tu ¼ 0:25f ct ð8Þ
linear elastic isotropic material can be defined as follows [36,37]:
qffiffiffiffiffi
0
where fct is the tensile strength of concrete and 0:33 f c is fsg ¼ ½De feg ð13aÞ
typically used in this study. Also, the bond strength differs
according to the types of steel fibers, for which the shape 8 9 2 38 9
> s = 1 n 0 e
< x > Ec 6
>
< x >
=
coefficient of fiber (df) can be introduced. Then, the maximum sy ¼ n 1 0 7 ey
4 5 ð13bÞ
; ð1n Þ 0
> 2
:v > 0
>
ð1nÞ=2 : g
>
;
xy xy
8
30mm, hooked end
7 50mm, hooked end where [D] is the elastic material stiffness matrix. To reflect the
30mm, straight nonlinear behavior of SFRC members, the elastic material stiffness
50mm, straight
6 matrix, [D] should be modified to satisfy the nonlinear constitu-
Bond stress (MPa)

tive relation of the material.


5 As shown in Fig. 3(a), Collins’s model [38] and Popovics’s
model [39] are used to describe the stress–strain relationship of
4
concrete in the direction of the principal compressive stress for
3 normal and high-strength concrete, respectively. Also, the Vec-
chio and Collins’s model [29] from MCFT is used to evaluate the
2 compression softening phenomenon of concrete due to cracking
in perpendicular direction. That is, the principal compressive
1
Idealized bond strength stress in concrete (sc2 ) can be expressed as follows:
0 For normal-strength concrete
0 1 2 3 4 5 6 7
"     #
Slip (mm) ec2 ec2 2
: sc2 ¼ sc2,max 2  ð14aÞ
Fig. 2. Bond behavior of SFRC.
eco eco
270 D.H. Lee et al. / Finite Elements in Analysis and Design 50 (2012) 266–286

σc 2 and strain (¼fcr/Ec) of concrete, respectively. The stress–strain


Uniaxial behavior
relationship of the reinforcement is shown in Fig. 3(c), assuming
Eq. (14b)
its elasto-plastic behavior, and in the following equation:
f ' (HSC)
f s ¼ Es es rf y ð17Þ
σc 2,max Eq. (14b) & (14c)
where fs is the stress in reinforcement, and Es and fy are the
Eq. (14a) & (14c)
f ' (NSC) modulus of elasticity and the yield strength of the reinforcement,
σc 2,max respectively. As shown in Fig. 3(c), the behavior of steel fibers in
Biaxial behavior
the principal stress direction can be described using DTFTM
aforementioned, as following:
Eq. (14a)
ssf rp,1 ¼ 0:41ssf ,1 V f r s1sf rp,max ð18Þ
ε
ε ε As shown in Fig. 4(a), the angle between the local coordinate
system(xelem yelem coordinate) and the global coordinate sys-
σ tem(X Y coordinate) is defined as bi, and the nonlinear material
stiffness matrix of conventional concrete in the principal stress
f cr direction of each element ([Dc]1  2), that is, the 1–2 axis direction
in Fig. 4(b), can be expressed as
2 3
Ec2 0 0
6 7
½Dc 12 ¼ 4 0 Ec1 0 5 ð19Þ
Eq. 15 0 0 Gc
Eq. 16
E = 2 f ' /ε where Ec2 , Ec1 , and Gc are the secant modulus of the principal
compressive stress, the principal tensile stress, and the shear
stress, respectively, and can be defined as
ε cr ε1
f c2
f (f ) Ec2 ¼ ð20Þ
ec2
f c1
, Ec1 ¼ ð21Þ
ec1
Bond stress between concrete and fiber reaches
E at max. IBS calculated from eq. 9. Ec1 Ec2
Gc ¼ ð22Þ
f (f or f ) Ec1 þ Ec2
As explained in the previous section, steel fiber was considered
Eq. 17
as the direct tensile stress transfer element in this study, and
E therefore, by reflecting steel fiber contribution on the nonlinear
ε
ε (ε ) material stiffness matrix, final material stiffness matrix in secant
ε
form can be derived as follows:
Fig. 3. Stress–strain relationship of materials. (a) Concrete compressive stress- 2 3
strain relationship; (b) Concrete tensile stress-strain relationship; (c) stress-strain 2 0 3 Ec2 0 0
Ec2 0 0 6 7
relationship of reinforcement and steel fiber. 6 7
6 0 7 6 7
0 6 7
½Dc 12 ¼ 4 0 Ec1 0 5 ¼ 6 0 6 Ec1 þEsf rp 0 7 ð23Þ
7
" # 0 6 7
0 0 Gc 4 Ec2 ðEc1 þ Esf rp Þ 5
nðec2 =eco Þ 0 0
For high-strength concrete : sc2 ¼ sc2,max Ec1 þ Esf rp þ Ec2
n1þ ðec2 =eco Þnk
where Esf rp is the secant modulus of the steel fibers, which is
ð14bÞ
expressed as
where ssf rp,1
0 Esf rp ¼ ð24Þ
f c 0 ec1
sc2,max ¼ Zf c ð14cÞ
0:8ð0:34ec1 =eco Þ The material stiffness matrix of reinforcement arranged in
and, ec2 is the principal compressive strain in concrete, eco is the each direction (½Ds 0i ), can be expressed as follows:
strain at the specified compressive strength of concrete (fc0 ), n is 2 3
ri Esi 0 0
0.8 þfc0 /17, and k is 0.67þfc0 /62Z1. The stress–strain relationship 6 7
½Ds 0i ¼ 4 0 0 05 ð25Þ
of concrete in tension proposed by Vecchio and Collins’s model
[30] shown in Fig. 3(b) is used in this study, which considers the 0 0 0
tension stiffening effect of concrete, as in
sc1 ¼ Ec ec1 , 0 r ec1 r ecr ð15Þ y
Y y x
y 2 1
f cr x
sc1 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi, ecr r ec1 ð16Þ
1þ 200ec1 y
α α
β
where sc1 and ec1 are the average principal tensile stress and X x x
strain in concrete, respectively, Ec is the modulus of elasticity of
qffiffiffiffiffi Fig. 4. Coordinate system in finite element analysis. (a) Element coordinate
0 0
concrete ( ¼ 2f c =eco ), fcr and ecr are the cracking stress ( ¼ 0:33 f c ) system; (b) Principal coordinate system; (c) Reinforcement.
D.H. Lee et al. / Finite Elements in Analysis and Design 50 (2012) 266–286 271

where ri is the reinforcement ratio in the ith direction, and Esi is direction of the reinforcement, can be derived as following:
the secant modulus of reinforcement in the ith direction, which is X
n
½D ¼ ½Dc  þ ½Ds i ð27Þ
f si i¼1
Esi ¼ ð26Þ
esi where [Dc] and [Ds]i are the stiffness matrix for concrete and
reinforcement in the global coordinate system, which are expressed as
Therefore, the material stiffness matrix in the global coordi-
nate system, considering the local coordinate system and the ½Dc  ¼ ½T c t ½Dc 012 ½T c  ð28Þ

½Ds i ¼ ½T s ti ½Ds 0i ½T s i ð29Þ


8 10 8 10
and the transformation matrix, [T] (i.e., [Tc] or [Ts]) is defined as
(x , y ) 7 (x , y ) 9 (s ,t ) 7 (s ,t ) 9
2 3
Node 4 Node 3 Node 4 Node 3 cos2 z sin2 z cos z sin z
6 2 2 7
½T ¼ 4 sin z cos z cos z sin z 5 ð30Þ
y −y Element 2
2 Element 4 2 4 1 2 cos z sin z 2 cos z sin z cos2 zsin z
1 3 1 3
where the rotation angle z in the transformation matrix is, as shown
Node 1 Node 2 Node 1 Node 2 in Fig. 4, defined in each case of the concrete element or the
(x , y ) (x , y ) (s ,t ) (s ,t )
reinforcement element, respectively, as follows:
x −x 1
z ¼ 180 þ bi a2 ð31Þ
Fig. 5. 4 node isotropic element. (a) Element in original coordination; (b) Element
in natural coordination. z ¼ ai þ bi ð32Þ

R σ
Iteration
1 3 5 Iteration
P
Compressive stress
External load

5
3
1
6 6
P : Imposed load 4
4 : Load-deformation response 2
Secant stiffness, E
2
: Stress-strain relation of concrete in compression

r ε
Deformation Compressive strain

σ σ Iteration
: Stress-strain relation of concrete in tension 5
Iteration
Reinforcement stress

3 5
Tensile stress

3, 4 6
1, 2

4
6 1, 2 Secant stiffness, E

: Stress-strain behavior of Reinforcement


Secant stiffness, E
ε ε
Tensile strain Reinforcement strain

σ Iteration

5
3
Steel fiber stress

4 6 Maximum IBS criteria

1, 2 Secant stiffness, E

: Stress-strain behavior of steel fiber


: Realistic behavior of steel fiber considering max. IBS

ε
Steel fiber strain

Fig. 6. Non-linear secant stiffness iteration method. (a) Member response; (b) Concrete in compression; (c) Concrete in tension; (d) Reinforcement; (e) Steel fiber.
272 D.H. Lee et al. / Finite Elements in Analysis and Design 50 (2012) 266–286

where bi is the orientation of elements from the global coordinate 6. Implementation of iteration method for nonlinear analysis
system to the local coordinate system, and a2 is the angle of
inclination of the crack in the element with respect to local coordinate, The nonlinear analysis procedures for the steel fiber-reinforced
and ai is the orientation of the reinforcement relative to the local concrete members are shown in Figs. 6 and 7 in detail. In this
coordinate system. study, the secant stiffness iteration method was used [31,32,42],
as it offers excellent convergence and stability in the iterative
calculation process. By determining the element stiffness based
5. Element stiffness matrix on the elastic material stiffness ([k]) in Eq. (13) at a specific load
state, and combining the element stiffness matrix [k] of the i
As shown in Fig. 5, a simple four-node isotropic element is number of each element, the stiffness matrix of system (or
used in this study. The coordinates of each node, xi,yi (i¼1 4), structure), [K], is formed as follows:
are shown in Fig. 5(a), and the stiffness matrix of the four-node
X
k
isotropic element ([k]) can be expressed in integral form as ½K ¼ ½ki ð34Þ
[31,32, 40,41] i¼1
Z
½k ¼ ½Bt ½D½BdV ð33Þ Then, the nodal displacement matrix can be calculated as
follows:
where [B] is the shape matrix of element. It is not possible to
fDg ¼ ½K1 fFg ð35Þ
explicitly integrate the Eq. (33), the four-point Gaussian quadrate
in Appendix 1 is used to derive the element stiffness matrix where {F} is the force matrix.
numerically. When a higher order element than the one used in The stresses in concrete, steel fiber, and reinforcement, and
this study is applied, the same method[37,40,41] explained in their secant moduli can be calculated using the nonlinear stress–
Appendix 1 can be used to derive the element stiffness matrix. strain relationship shown in Fig. 6, and then the convergence is

START

Determine the total global


Input structural stiffness matrix
and material properties

Input external nodal load, Determine the nodal


displacements

for i = 1 to m
for j = 1 to n
for i = 1 to m
Estimate the secant modulus of for j = 1 to n
each element
Determine the element average
, and strains and stresses

and
Determine the material stiffness
of each element
Calculate the updated secant
and moduli

, and
Determine the each element
stiffness matrix
No
Secant moduli
converged?

Determine the composite stiffness yes


of each elements

No
Check crack width criteria

sfrc max

m : No. of elements yes


n : No. of reinforcement direction
END

Fig. 7. Analysis procedure for non-linear iterative method.


D.H. Lee et al. / Finite Elements in Analysis and Design 50 (2012) 266–286 273

checked by comparing the secant moduli in current stage with and it is reflected on the analysis that pull-out failure of SFRC
those in the previous stage in the iterative calculation process. If a occurs if the crack width (wsfrc) estimated from Eq. (39) exceeds
certain level of convergence is not secured, each secant modulus the maximum crack width criterion of SFRC (wmax ) presented in
is updated and the iterative calculation is performed continu- Eq. (36).
ously. Once the convergence criteria is satisfied, the strains at
each node can be calculated based on the determined nodal
displacements, and the average nodal strains and stresses can 8. Verifications of the proposed nonlinear finite element
be determined at each node. Furthermore, the average strain and analysis
stress at each node can be calculated within the elements.
To verify the proposed non-linear finite element analysis
(NLFEA), the analysis results were compared with the results of
7. Pull-out failure criteria of SFRC the SFRC panel tests subjected to pure shear[8], which were
recently performed at the University of Toronto. Table 1 shows
Steel fibers typically have the high tensile strengths over the material properties and dimensional details of the specimens,
1000 MPa. As shown in Fig. 2, however, once the bond stresses in which the main experimental parameters are concrete com-
developed between steel fibers and surrounding concrete at the pressive strength (fc0 ), types of fibers, and fiber volume fraction
crack interface reach the maximum IBS (tmax ), the tensile stresses (Vf). The mesh details and boundary conditions in the analysis are
in fibers are maintained until a certain deformation, i.e., their shown in Fig. 8, and pure shear stresses are applied by loading the
pull-out failures. Normally, most steel fibers reach their bond forces along the edges. In all the loading stages, the convergence
strengths before they reach their tensile strengths as the crack condition of the iterative procedures is that the ratio of the secant
widths increase; and therefore, to accurately describe the failure moduli in the current stage (i) to those in the previous stage (i  1)
modes of SFRC, such pull-out failures of steel fibers should be satisfies less than 1%. As aforementioned, the failure criterion in
reflected in the analytical model. analysis of SFRC members with respect to the pull-out failure was
Lim et al. [21] proposed the maximum crack width of SFRC applied once the estimated crack width (wsfrc) by Eq. (39) reaches
based on experimental research as the maximum crack width (wmax ) in Eq. (36). Furthermore, Fig. 9
Lf additionally shows the points, calculated from the proposed
wmax ¼ ð36Þ NLFEA, at which the stresses in steel fibers reach their maximum
16
interfacial bond stresses, because they allow observation of the
where Lf is the length of the fiber. The crack width of a conven- additional deformation capacity of the steel fiber-reinforced
tional reinforced concrete member can be estimated, as follows: concrete members after they reach the maximum interfacial bond
w ¼ smy ec1 ð37Þ stress until their pull-out failures.
Fig. 9(a) shows the comparison of the analysis results and the
where the average shear crack spacing (smy) [30] can be calcu-
test results of the specimen C1F1V1. The concrete compressive
lated, as follows:
strength of the specimen C1F1V1 was 51.4 MPa, and the length
1 and diameter of fibers were 50 mm and 0.62 mm, respectively.
smy ¼ ð38Þ
ððsin a2 =smx Þ þðcos a2 =smy ÞÞ The fiber volume fraction (Vf) was 0.5%, which was relatively
lower than other specimens. While NLFEA with DTFTM evaluated
where smx and smy are the average crack spacing in the direction of
the shear strength of the panel specimen accurately, it tended to
xelem and yelem, respectively, as shown in Fig. 4, which can be
underestimate the shear stiffness. The analysis result implies that
calculated using the method proposed by Collins and Mitchell
the pull-out failure occurred when the bond stress developed
[37]. The crack spacing of SFRC is smaller than that of conven-
between steel fiber and adjacent concrete reached the maximum
tional reinforced concrete due to the crack control capacity of
IBS, at the same time, the crack width also reached the maximum
steel fibers. In this study, therefore, the crack spacing control
crack width. It is also shown that both the strength and the
factor (q), which was proposed by Dupond and Vandewalle [12],
deformation capacity of the panel C1F1V1 were very accurately
is introduced, and the crack width of SFRC can be expressed by
estimated by the proposed method. Fig. 10(a) shows the esti-
modifying Eq. (37), as follows:
mated stresses of reinforcement of the C1F1V1 panel compared to
wsf rc ¼ smy ec1 q ð39Þ experimental results measured from the strain gage in reinforce-
ment and LVDTs, respectively. The reinforcement was not yield at
50 ultimate, and NLFEA with DTFTM very accurately estimated
q¼ ð40Þ
Lf =df the stress of reinforcement in the longitudinal direction. Also,

Table 1
Summary of panel specimens used for verification of NLFEM with DTFTM.

Specimen names Failure mode Concrete Steel fiber

0
f c ðMPaÞ e0cu ð103 Þ ffu (MPa) L (mm) D (mm) Vf (%)

C1F1V1 Interlock failure 51.4 2.150 1050 50 0.62 0.5


C1F1V2 Interlock failure 53.4 2.670 1050 50 0.62 1.0
C1F1V3 Interlock failure 49.7 2.500 1050 50 0.62 1.5
C1F2V3 Interlock failure 59.7 3.280 2300 30 0.38 1.5
C1F3V3 Interlock failure 45.5 2.340 1100 35 0.55 1.5
C2F1V3 Interlock failure 79.4 2.770 1050 50 0.62 1.5
C2F2V3 Interlock failure 76.5 2.220 2300 30 0.38 1.5
C2F3V3 Interlock failure 62.0 2.030 1100 35 0.55 1.5

Note: Size of panel, yield strength and amount of longitudinal bar are same in all specimens as 890 mm  890 mm  70 mm, 2063 mm2(rsx ¼3.31%) and 552 MPa,
respectively, and all panels were not reinforced transversely.
274 D.H. Lee et al. / Finite Elements in Analysis and Design 50 (2012) 266–286

14 16 18 P

7 13 8 15 9 17 P P
P

8 10 12 P

890 mm 7 9 11
4 5 6

2 4 6 P

1 3 5

1 2 3

890 mm
Fig. 8. Analysis details of panel specimen. (a) Mesh and dimensions of panel; (b) Loading and boundary condition.

Fig. A2(a) visually shows the nonlinear finite element analysis The analysis results show good agreements with the shear
results of C1F1V1 by contour at the ultimate stage. Since no strength of the specimen, and accurately estimated the deforma-
transverse reinforcements were provided in the C1F1V1 panel, tion capacity of the tested panel within the difference of about
the transverse strain in the Y–Y direction was larger than the 20%. It is believed that this is because the DTFTM proposed in this
longitudinal strain in the X–X direction. study appropriately reflected the difference in the bond charac-
Fig. 9(b) shows the shear behavior of the C1F1V2 specimen teristics according to the type of the fibers. Fig. A2(d) shows the
whose concrete compressive strength was 53.4 MPa, fiber length stresses, strains, and deformations contour evaluated by NLFEA
and diameter were 50 mm and 0.62 mm, respectively, and fiber with DTFTM, in which the distribution trends of the stresses and
volume fraction (Vf) was 1.0%. NLFEA with DTFTM accurately deformations are similar to those of the previous specimens.
estimated the shear strength of the specimen, and the analysis Although the average strain and stress in the longitudinal rein-
result indicated that the bond stress developed between fiber and forcement was 0.0015, and 290 MPa, respectively, which differ
concrete reached the maximum IBS at 4.8 MPa of the shear stress about 20% compared to the 261 MPa measured from the test, as
prior to the pull-out failure of fibers. This corresponds to 66% of shown in Fig. 10(d), and the overall behavior of the reinforcement
the ultimate deformation capacity, which implies that the failure was accurately estimated.
criterion defined by the maximum crack width is more appro- Fig. 9(e) shows the comparison of the analysis results and the
priate than that by the maximum interfacial bond stress. The test results of the C1F3V3 specimen, whose concrete compressive
shear stress evaluated at maximum IBS, however, is very similar strength was 45.5 MPa and volume fraction of fiber (Vf) was 1.5%;
with shear strength at ultimate. Compared to the specimen and fiber aspect ratio (Lf/D) was the smallest in all the panel
C1F1V1, the specimen C1F1V2 showed almost twice the strain specimens with 35 mm and 0.55 mm in length and diameter,
capacity, and had the same pullout failure of steel fibers on the respectively. The analysis results underestimated the stiffness of
cracked surface. This means the fiber volume fraction (Vf) is a key the specimen and the strength about 10%. As the test results
influential factor in the determination of the deformation capacity show, however, the proposed model very accurately estimated
of SFRC members. The estimated responses by NLFEA with DTFTM the deformation capacity of the specimen at the failure, in which
show that the pull-out failure occurred at about 0.006 of the the specimen showed more brittle behavior than the other
average shear strain, which agree well with test results of speci- specimens. As shown in Fig. A2(e), the longitudinal reinforcement
men. Also, as shown in Fig. A2(b), the estimated strain distribu- did not yield in any part of panel; and as shown in Fig. 10(e), the
tion of C1F1V2 panel were mostly below the yield strain of proposed model accurately estimated the stress change in the
reinforcement, except the small region at the right end of the reinforcement at various loading stages.
specimen. As shown in Fig. 10(b), the estimated stress of the Fig. 9(f) shows the comparison of the analysis and test results of
reinforcement is also very similar to the test result. the specimen C2F1V3 cast with high-strength concrete, whose
The specimen C1F1V3, shown in Fig. 9(c), has a concrete concrete compressive strength was 79.4 MPa, fiber length and
compressive strength of 49.7 MPa, a fiber length and diameter diameter were 50 mm and 0.62 mm, respectively, and fiber volume
of 50 mm and 0.62 mm, respectively, and a 1.5% volume fraction fraction (Vf) was 1.5%. NLFEA with DTFTM well estimated the shear
of fiber (Vf). NLFEA with DTFTM estimated the shear strength of behavior and the ultimate strength of the specimen, and also
the specimen with a 13% difference and the deformation capacity accurately evaluated the failure mode and deformation capacity,
with about 11% difference. The detailed results from the finite which means that the proposed model can appropriately predict
element analysis, shown in Fig. A2(c), show that the transverse the shear behavior of SFRC members with high-strength concrete.
strains are larger than the horizontal strains. Also, while the Fig. A2(f) shows that the longitudinal strain was larger than that of
estimated strains exceeded the yield strain at around the right the previous specimens, and Fig. 10(f) shows that the average
end of the specimen, the average strain of the shear panel in the strain in the reinforcement was somewhat underestimated.
direction of longitudinal reinforcement, as shown in Fig. 10 (c), The specimen C2F2V3 is a high-strength FRC specimen with a
was around 210 MPa, which is very close to the stress in the concrete compressive strength of 76.5 MPa, a fiber volume fraction
reinforcement that was measured from the test. (Vf) of 1.5%, and a fiber length and diameter of 30 mm and
The C1F2V3 specimen in Fig. 9(d) has a concrete compressive 0.38 mm, respectively. As shown in Fig. 9(g), the proposed NLFEA
strength of 59.7 MPa, a fiber volume fraction (Vf) of 1.5%, and a with DTFTM very accurately estimated the shear strength and
fiber length and diameter of 30 mm and 0.38 mm, respectively, behavior of the specimen, and show that the proposed model can
which are smaller than the fibers used in the previous specimens. accurately estimate the behavioral differences due to the change in
D.H. Lee et al. / Finite Elements in Analysis and Design 50 (2012) 266–286 275

5.0 7.0
4.5 maximum IBS criteria
6.0
4.0 max. crack width criteria

Shear stress (MPa)


Shear stress (MPa)
3.5 5.0
3.0 4.0 maximum IBS criteria
2.5 max. crack width criteria
2.0 3.0
C1F1V2 Panel
1.5 C1F1V1 Panel 2.0 NLFEA with DTFTM
1.0 NLFEA with DTFTM
1.0
0.5
0.0 0.0
0.000

0.001

0.002

0.003

0.004

0.005

0.006

0.000

0.001

0.002

0.003

0.004

0.005

0.006

0.007

0.008

0.009

0.010
Shear strain (mm/mm) Shear strain (mm/mm)

8.0 8.0
7.0 7.0
6.0 6.0
Shear stress (MPa)

Shear stress (MPa)


5.0 5.0
maximum IBS criteria maximum IBS criteria
4.0 4.0 max. crack width criteria
max. crack width criteria
3.0 C1F1V3 Panel
3.0 C1F2V3 Panel
NLFEA with DTFTM NLFEA with DTFTM
2.0 2.0
1.0 1.0
0.0 0.0
0.000

0.001

0.002

0.003

0.004

0.005

0.006

0.007

0.008

0.009

0.010

0.000

0.001

0.002

0.003

0.004

0.005

0.006

0.007

0.008

0.009

0.010
Shear strain (mm/mm) Shear strain (mm/mm)

7.0 8.0
maximum IBS criteria
6.0 max. crack width criteria 7.0
6.0
Shear stress (MPa)
Shear stress (MPa)

5.0
5.0
4.0 maximum IBS criteria
4.0 max. crack width criteria
3.0
3.0
C2F1V3 Panel
2.0 C1F3V3 Panel 2.0 NLFEA with DTFTM
NLFEA with DTFTM
1.0 1.0
0.0 0.0
0.000

0.001

0.002

0.003

0.004

0.005

0.006

0.007

0.008

0.009

0.010

0.000

0.001

0.002

0.003

0.004

0.005

0.006

0.007

0.008

0.009

0.010

Shear strain (mm/mm) Shear strain (mm/mm)

8.0 8.0
maximum IBS criteria
7.0 7.0 max. crack width criteria

6.0 6.0
Shear stress (MPa)
Shear stress (MPa)

5.0 5.0
maximum IBS criteria
4.0 max. crack width criteria 4.0
3.0 C2F2V3 Panel
3.0
C2F3V3 Panel
NLFEA with DTFTM
2.0 2.0 NLFEA with DTFTM

1.0 1.0
0.0 0.0
0.000

0.001

0.002

0.003

0.004

0.005

0.006

0.007

0.008

0.009

0.010

0.000

0.001

0.002

0.003

0.004

0.005

0.006

0.007

0.008

0.009

0.010

Shear strain (mm/mm) Shear strain (mm/mm)

Fig. 9. Verification of shear behavior of proposed NLFEM with DTFTM. (a) C1F1V1 panel (b) C1F1V2 panel; (c) C1F1V3 panel (d) C1F2V3 panel; (e) C1F3V3 panel (f) C2F1V3
panel; (g) C2F2V3 panel (h) C2F3V3 panel.
276 D.H. Lee et al. / Finite Elements in Analysis and Design 50 (2012) 266–286

200 250
Ave. stress in reinforcement (MPa)

Ave. stress in reinforcement (MPa)


Zuruch gage

150
LVDT 200
NLFEM with DTCM
150
100
100
Zuruch gage
50 LVDT
50 NLFEM with DTCM

0 0

-50 -50
-0.50 0.00 0.50 1.00 1.50 2.00 2.50 3.00 -0.50 0.50 1.50 2.50 3.50 4.50 5.50 6.50
Ave. shear strain (mm/mm) Ave. shear strain (mm/mm)

250 400
Ave. stress in reinforcement (MPa)

Ave. stress in reinforcement (MPa)


350
200
300
150 250
200
100
Zuruch gage 150 Zuruch gage
LVDT LVDT
50 NLFEM with DTCM 100 NLFEM with DTCM
50
0
0
-50 -50
-0.50 0.50 1.50 2.50 3.50 4.50 5.50 6.50 -0.50 1.50 3.50 5.50 7.50 9.50
Ave. shear strain (mm/mm) Ave. shear strain (mm/mm)

250 400
Ave. stress in reinforcement (MPa)

Ave. stress in reinforcement (MPa)

350
200
300
150 250
200
100
Zuruch gage 150 Zuruch gage
LVDT LVDT
50 NLFEM with DTCM
100 NLFEM with DTCM
50
0
0
-50 -50
-0.50 0.50 1.50 2.50 3.50 4.50 5.50 -0.50 0.50 1.50 2.50 3.50 4.50 5.50 6.50 7.50
Ave. shear strain (mm/mm) Ave. shear strain (mm/mm)

400 400
Ave. stress in reinforcement (MPa)

Ave. stress in reinforcement (MPa)

350 Zuruch gage 350 Zuruch gage


LVDT LVDT
300 300
NLFEM with DTCM NLFEM with DTCM
250 250
200 200
150 150
100 100
50 50
0 0
-50 -50
-0.50 0.50 1.50 2.50 3.50 4.50 5.50 -0.50 0.50 1.50 2.50 3.50 4.50 5.50
Ave. shear strain (mm/mm) Ave. shear strain (mm/mm)

Fig. 10. Verification of proposed NLFEM with DTFTM. (a) C1F1V1 panel (b) C1F1V2 panel; (c) C1F1V3 panel (d) C1F2V3 panel; (e) C1F3V3 panel (f) C2F1V3 panel;
(g) C2F2V3 panel (h) C2F3V3 panel.

the fiber shape even in the case of shear panel cast with high- stress behavior in the longitudinal steel reinforcement very closely
strength concrete. The analysis result also showed that the bond to the test results.
stress developed between concrete and fibers at crack interface Fig. 9(h) shows the shear behavior of the specimen C2F3V3, in
reached the maximum IBS simultaneously with the pulled-out which concrete compressive strength was 62.0 MPa, fiber length
failure of fibers. As shown in Fig. A2(g), the deformation capacity of and diameter were 35 mm and 0.55 mm, respectively, and
the C2F2V3 specimen was smaller than that of the C2F1V3 speci- volume fraction of fiber (Vf) was 1.5%. The results of NLFEA with
men, and as shown in Fig. 10(g), the proposed model estimated the DTFTM were almost identical to the test results. The proposed
D.H. Lee et al. / Finite Elements in Analysis and Design 50 (2012) 266–286 277

model predicted that the steel fiber first reached the maximum Appendix 1. Element stiffness matrix
IBS at a shear stress of around 4.8 MPa; and the pull-out failure of
fibers occurred at a shear strength of around 5.5 MPa, which were The stiffness matrix of the four node isotropic element shown
identical to the test results. In addition, the proposed NLFEA with in Fig. 5 can be expressed, as follows:
the maximum IBS as the failure criterion conservatively estimated Z
the shear strength and deformation capacity of the specimen. ½k ¼ ½Bt ½D½BdV ð33Þ
Fig. A2(h) shows the detailed analysis results of the specimen, and
its behavior was very similar to those of the previously mentioned where [B] is shape matrix, which varies by element types. As
high-strength specimens. As for the longitudinal average strain in Eq. (33) cannot be directly integrated, the numerical integration is
the analysis results, the estimated strains did not exceed the yield required, based on the procedure shown in Fig. A1 using 4 points
strain of steel in any part, and as shown in Fig. 10(h), the behavior Gaussian quadrate. Then, the stiffness matrix can be obtained as
of the longitudinal reinforcement was reasonably well evaluated Z 1Z 1
by the proposed model. ½k ¼ ½Bðs,tÞt ½D½Bðs,tÞ9J9h dsdt ðA1Þ
1 1
As discussed above, the comparison of the test results and the
analysis of the SFRC panels, whose experimental parameters were where 9J9 is the determinant of Jacobian matrix, (s,t) are normal-
the compressive strength of the concrete (fc’), the volume fraction ized coordinate values in natural coordinate system, and h is the
of steel fibers (Vf), and the fiber aspect ratio (Lf/D), shows that the thickness of plane stress element. The element stiffness matrix,
proposed NLFEA with DTFTM estimated the shear strength and therefore, can be derived explicitly, as following:
behavior of the fiber-reinforced concrete members accurately.
½k ¼ ½Bðs1 ,t 1 Þt ½D½Bðs1 ,t 1 Þ9Jðs1 ,t 1 Þ9hW 1 W 1
Particularly, the pull-out failure criterion, which was proposed in
this study, estimated the deformation capacity of the SFRC þ½Bðs2 ,t 2 Þt ½D½Bðs2 ,t 2 Þ9Jðs2 ,t 2 Þ9hW 2 W 2
members very accurately, and the proposed NLFEA with DTFTM
þ½Bðs3 ,t 3 Þt ½D½Bðs3 ,t 3 Þ9Jðs3 ,t 3 Þ9hW 3 W 3
also provided good analysis results for the high-strength SFRC
members. þ½Bðs4 ,t 4 Þt ½D½Bðs4 ,t 4 Þ9Jðs4 ,t 4 Þ9hW 4 W 4 ðA2Þ

where si and ti (i ¼1 4) are Gaussian location factor, which are


s1 ¼s2 ¼ t1 ¼t3 ¼  0.5773 and s3 ¼s4 ¼t2 ¼t4 ¼0.5773, respectively,
9. Conclusion and Wi (i¼1–4) are Gaussian weight factor, which are 1.0 in all
cases (i.e. W1 ¼ W2 ¼W3 ¼W4 ¼1.0). Also, the shape matrix can be
In this study, the non-linear finite element analysis with direct
tension force transfer model (NLFEA with DTFTM), which can
predict the shear behavior of SFRC members, was developed, in
which, unlike previous studies, steel fibers were considered as a Read in four Gauss points and weight functions
direct tension force transfer elements. Also, the proposed numer-
ical analysis model for the shear behavior of the SFRC members
was verified by comparing to the results of shear panel tests, from
which the following conclusions can be made:
Zero
1. It was verified that the proposed numerical analysis model
with DTFTM can appropriately consider the directionality of
fibers, difference of bond characteristics according to the types
of fibers, and pull-out failure mechanism of fibers with a
relatively simple approach. for
2. It is considered that the pull-out failure of the steel fiber is a
dominant mechanism at the crack interface of SFRC members,
and the proposed model estimated the deformation capacity of
the SFRC members very accurately using the simple pull-out
failure criterion. Let
3. The numerical analysis model proposed in this study only
requires a simple bond test of steel fibers for the estimation of
pull-out failure, even if the type of the steel fibers or the
strength of the concrete changes; and it is believed that this Compute
will greatly reduce the experimental efforts required to
develop the existing constitutive equations in tension for SFRC
composite material.
4. The proposed analysis model showed relatively good estima- Compute
tion on the shear behavior of SFRC members with various
concrete compressive strengths or different types of steel
fibers, and also provided good analysis results for the shear
behavior of the high-strength SFRC members. Update

Acknowledgments
END
This work was supported by the Korea Research Foundation
Grant funded by the Korean Government (MOEHRD, General Fig. A1. Flow-chart for evaluation of the element stiffness matrix by Gaussian
Research Promotion Fund) (KRF-2009–0070821). quadrate.
278 D.H. Lee et al. / Finite Elements in Analysis and Design 50 (2012) 266–286

Fig. A2. Stresses, strains and displacement contour from analysis results. (a) C1F1V1 panel; (b) C1F1V2 panel; (c) C1F1V3 panel; (d) C1F2V3 panel; e) C1F3V3 panel;
(f) C2F1V3 panel; (g) C2F2V3 panel; (h) C2F3V3 panel.
D.H. Lee et al. / Finite Elements in Analysis and Design 50 (2012) 266–286 279

Fig. A2. (continued)


280 D.H. Lee et al. / Finite Elements in Analysis and Design 50 (2012) 266–286

Fig. A2. (continued)


D.H. Lee et al. / Finite Elements in Analysis and Design 50 (2012) 266–286 281

Fig. A2. (continued)


282 D.H. Lee et al. / Finite Elements in Analysis and Design 50 (2012) 266–286

Fig. A2. (continued)


D.H. Lee et al. / Finite Elements in Analysis and Design 50 (2012) 266–286 283

Fig. A2. (continued)


284 D.H. Lee et al. / Finite Elements in Analysis and Design 50 (2012) 266–286

Fig. A2. (continued)


D.H. Lee et al. / Finite Elements in Analysis and Design 50 (2012) 266–286 285

Fig. A2. (continued)


286 D.H. Lee et al. / Finite Elements in Analysis and Design 50 (2012) 266–286

expressed in natural coordinate system, as follows: [11] J.P. Romualdi, G.B. Batson, The mechanics of crack arrest in concrete, Journal
of Engineering Mechanics Division, Proceedings of the ASCE 89 (EM 3) (1963)
1   147–168.
½Bðs,tÞ ¼ B1 B2 B3 B4 ðA3Þ
9J9 [12] D. Dupont, L. Vandewalle, Calculation of Crack Widths with the s-e Method,
in: Test and Design Methods for Steel Fibre Reinforced Concrete: Background
where Bi is and Experiences—Proceedings of the RILEM TC162-TDF Workshop, RILEM
2 3 Technical Committee 162-TDF, Bochum, Germany, 2003, pp. 119–144.
aðN i,s ÞbðNi,t Þ 0
[13] G. Batson, E. Jenkins, R. Spatney, Steel fibers as shear reinforcement in beams,
6 7
½Bi  ¼ 4 cðN i,t ÞdðN i,s Þ 0 5 ðA4Þ ACI Journal Proceedings 69 (10) (1972) 640–644.
cðN i,t ÞdðN i,s Þ aðN i,s ÞbðN i,t Þ [14] A.K. Sharma, Shear strength of steel fiber reinforced concrete beams, ACI
Journal, Proceedings 83 (4) (1986) 624–628.
where [15] R. Narayanan, I.Y.S. Darwish, Use of steel fibers as shear reinforcement, ACI
Structural Journal 84 (3) (1987) 216–227.
  [16] K.H. Tan, M.A. Mansur, Shear transfer in reinforced fiber concrete, Journal of
a ¼ 14 y1 ð1 þ sÞ þ y2 ð1sÞ þ y3 ð1þ sÞ þy4 ð1sÞ ðA5  aÞ
Materials in Civil Engineering, ASCE 2 (4) (1990) 202–214.
  [17] K.H. Tan, K. Murugappan, P. Paramasivam, Shear behavior of steel fiber
b ¼ 14 y1 ð1þ tÞ þy2 ð1tÞ þ y3 ð1 þtÞ þ y4 ð1tÞ ðA5  bÞ
reinforced concrete beams, ACI Structural Journal 89 (6) (1992) 3–11.
[18] S.A. Ashour, G.S. Hassanain, F.F. Wafa, Shear behavior of high-strength fiber
c ¼ 14½x1 ð1þ tÞ þx2 ð1tÞ þx3 ð1 þtÞ þ x4 ð1tÞ ðA5  cÞ
reinforced concrete beams, ACI Structural Journal 89 (2) (1992) 176–184.
[19] M. Khuntia, B. Stojadinovic, S.C. Goel, Shear strength of normal and high-
d ¼ 14½x1 ð1 þsÞ þ x2 ð1sÞ þ x3 ð1 þ sÞ þ x4 ð1sÞ ðA5  dÞ
strength fiber reinforced concrete beams without stirrups, ACI Structural
Ni,s and Ni,t are derivatives of the shape function (Ni) respect to s Journal 96 (2) (1999) 282–289.
[20] Y.H. Oh, J.H. Kim, Estimation of flexural and shear strength for steel fiber
and t, respectively, and the shape function, Ni can be established reinforced flexural members without shear reinforcements, Journal of the
linearly, as following: Korea Concrete Institute 20 (2) (2008) 257–267.
[21] T.Y. Lim, P. Paramsivam, S.L. Lee, Shear and moment capacity of reinforced
ð1sÞð1tÞ steel-fiber-concrete beams, Magazine of Concrete Research 39 (140) (1987)
N1 ¼ ðA6  aÞ
4 148–160.
[22] R.N. Swamy, H.M. Bahia, Effectiveness of steel fibers as shear reinforcement,
ð1 þ sÞð1tÞ Concrete International: Design and Construction 7 (3) (1985) 35–40.
N2 ¼ ðA6  bÞ
4 [23] R.N. Swamy, R. Jones, A.T.P. Chinam, Influence of steel fibers on the shear
resistance of lightweight concrete T-beams, ACI Structural Journal 90 (1)
ð1 þ sÞð1 þ tÞ (1993) 103–114.
N3 ¼ ðA6  cÞ [24] Y.K. Kwak, M.O. Eberhard, W.S. Kim., J.B. Kim, Shear strength of steel fiber-
4
reinforced concrete beams without stirrups, ACI Structural Journal 99 (4)
(2002) 530–538.
ð1sÞð1 þ tÞ
N4 ¼ ðA6  dÞ [25] R.N. Swamy, H.M. Bahia, Influence of fiber reinforcement on dowel resistance
4 to shear, ACI Journal Proceedings 76 (2) (1979) 327–355.
[26] P. Soroushian, C.D. Lee, Distribution and orientation of fibers in steel fiber
reinforced concrete, ACI Material Journal 87 (5) (1988) 433–439.
[27] American Concrete Institute, ACI Manual of Concrete Practice 2008, part 6,
ACI 504R-90(97) to ACI ITG-4.3R-0.7, 2008.
Appendix 2. Analysis results [28] F.J. Veccho, Disturbed stress field model for reinforced concrete: formulation,
Journal of Structural Engineering, ASCE 126 (9) (2000) 1070–1077.
Fig. A2. [29] T.T.C. Hsu, Softened truss model theory for shear and torsion, ACI Structural
Journal 85 (5) (1988) 624–635.
[30] F.J. Vecchio, M.P. Collins, Modified compression field theory for reinforced
References concrete elements subjected to shear, ACI Journal Proceedings 83 (2) (1986)
219–231.
[31] F.J. Vecchio, Nonlinear finite element analysis of reinforced concrete mem-
[1] ACI Committee 544, Design consideration for steel fiber reinforced concrete branes, ACI Structural Journal 86 (1) (1989) 26–35.
(ACI 544.4 R-88), ACI Structural Journal 85 (5) (1988) 563–580. [32] F.J. Vecchio, Reinforced concrete membrane element formulation, Journal of
[2] O. Janis, New frontiers for steel fiber-reinforced concrete, Concrete Interna-
Structural Engineering, ASCE 116 (3) (1990) 730–750.
tional 30 (5) (2008) 45–50.
[33] P. Paramasivam, K.H. Tan, K. Murugappan, Finite element analysis of partially
[3] J.P. Romualdi, J.A. Mandel, Tensile strength of concrete affected by uniformly
prestressed steel fiber concrete beams in shear, Advanced Cement Based
distributed and closely spaced short lengths of wire reinforcement, ACI
Materials 2 (6) (1995) 231–239.
Journal Proceedings 61 (6) (1964) 657–671.
[34] X. Lihua, C. Yin, X. Dongtao, Nonlinear finite element analysis of steel fiber
[4] P. Adebar, S. Mindess, D. St-Pierre, B. Olund, Shear tests of fiber concrete
reinforced concrete deep beams, Journal of Natural Science 13 (2) (2008)
beams without stirrups, ACI Structural Journal 94 (1) (1997) 68–76.
[5] L. Vandewalle, Influence of Tensile Strength of Steel Fibre on Toughness of 201–206 Wuhan University.
High Strength Concrete, in: H.W. Reinhardt, A.E. Naaman (Eds.), Proceedings [35] J.Y.L. Voo, S.J. Foster, Variable Engagement Model for Fibre Reinforced
of Third International Workshop on High-Performance Cement Composites, Concrete in Tension, UNICIV Report No. R-420 June 2003, The University of
RILEM Publications, 1999, pp. 331–337. New South Wales, Sydney, Australia, 2003, pp. 1–86.
[6] S.P. Shah, B.V. Ragan, Fiber reinforced concrete properties, ACI Journal 68 (2) [36] I.M. Smith, D.V. Griffiths, Programming the Finite Element Analysis, 4th Ed.,
(1971) 126–137. John Wiley and Sons, Ltd, 2004, pp. 628.
[7] G. Batson, Steel fiber reinforced concrete, Materials Science and Engineering [37] D.L. Logan, A First Course in the Finite Element Method, 4th Ed., Thomson,
25 (1976) 53–58. 2007 808 pp.
[8] J. Susetyo, Fibre Reinforcement for Shrinkage Crack Control in Prestressed, [38] M.P. Collins, D. Mitchell, Prestressed Concrete Structures, Prentice Hall, 1991
Precast Segmental Bridges, Ph.D. Dissertation, University of Toronto, 2009, 766 pp.
307 pp. [39] S. Popovics, A numerical approach to the complete stress–strain curve and
[9] S.P. Shah, J. Weiss, W. Yang, Shrinkage cracking—can it be prevented? concrete, Cement and Concrete Research 3 (5) (1973) 583–599.
Concrete International 20 (4) (1998) 51–55. [40] K.J. Bathe, Finite Element Procedure, Prentice-Hall, 1996 1037 pp..
[10] K.C.G. Ong, P. Paramasivam, Cracking of Steel Fibre Reinforced Mortar due to [41] T.Y. Yang, Finite Element Structural Analysis, Prentice-Hall, 1986 543 pp..
Restrained Shrinkage, in: R.N. Swamy, B. Barr (Eds.), Fiber Reinforced [42] A.C. Scordelis, Finite Element Modeling of Reinforced Concrete Structures, in:
Cements and Concretes: Recent Developments, Elsevier Science Publishers Proceedings of the Symposium on Analysis of Reinforced Concrete Structures
Ltd, 1989, pp. 179–187. by Finite Element Method, Politecnico di Milano, 1978.

You might also like