See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/315171515

A flexure-shear Timoshenko fiber beam element based on softened damage-

plasticity model

Article  in  Engineering Structures · June 2017

DOI: 10.1016/j.engstruct.2017.02.066

CITATIONS READS

20 431

4 authors, including:

De-Cheng Feng Zeyang Sun

Southeast University (China) Southeast University (China)
44 PUBLICATIONS   235 CITATIONS    21 PUBLICATIONS   204 CITATIONS   

SEE PROFILE SEE PROFILE

Ji-Gang Xu
Southeast University (China)
2 PUBLICATIONS   20 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Uncertainty quantification by advanced PDF derivation techniques View project

Advanced finite beam elements for efficient analysis of large-scale structures View project

All content following this page was uploaded by De-Cheng Feng on 28 November 2018.

The user has requested enhancement of the downloaded file.

 A flexure-shear Timoshenko fiber beam element based on softened
damage-plasticity model

De-Cheng Fenga,∗, Gang Wua , Ze-Yang Suna , Ji-Gang Xua

a Key Laboratory of Concrete and Prestressed Concrete Structures of the Ministry of Education, Southeast University,
Nanjing 210096, China

Abstract
In this paper, a new displacement-based fiber beam element including flexure-shear interaction is developed.
The element is based on the conventional Timoshenko beam theory, and the section behavior is modeled
with the well-known fiber section approach, where the section is divided into steel fibers and concrete fibers.
The Menegotto-Pinto model is used for steel fibers and and a multi-dimensional softened damage-plasticity
model, which accounts for the compression-softening effect of reinforced concrete, is adopted for concrete
fibers. The axial-flexure-shear interaction can be well reflected in both section level and material level
since the normal-shear coupling is reflected naturally in the multi-dimensional material constitutive law.
Besides, the concrete material parameters are related to fracture energy to avoid mesh-sensitivity issues of
softening problems. The numerical implementation of the proposed element, including the finite element
approximation and fiber state determination, is also discussed in detail. The element is validated through
the test results for a series of simply supported reinforced concrete beams under monotonic loading and
reinforced concrete columns and wall under cyclic loading, and the results indicate that the element is
capable to reproduce the shear behavior of reinforced concrete members under different loading cases.
Keywords: fiber element, flexure-shear interaction, Timoshenko beam, damage-plasticity model,
compression-softening, mesh-objectivity

1. Introduction

Beam-column element, or referred to as frame element, is widely used in nonlinear analysis of reinforced
concrete (RC) frame structures due to its combination of numerical accuracy and computational efficiency.
In the past two decades, several remarkable elements have been developed, among which the fiber beam
element is most preferred in application [1, 2, 3]. The section is discretized with steel fibers and concrete
fibers, and each fiber is assigned with a uniaxial material constitutive law. The axial-flexure interaction can
be well reflected by the fiber section model, and the confinement effect of stirrups can also be considered.
However, most fiber beam elements are settled in the framework of Euler-Bernoulli beam theory, where the
shear deformations are neglected. This feature makes the elements not suitable for modeling RC members
with obvious shear deformation, i.e., columns with shear span ratio lower than 2 [4].
In order to include the flexure-shear interaction in the beam-column elements, several approaches have
been proposed. A detailed review can be found in [5]. The first way is to introduce flexure springs and/or
shear springs at the end of the element such that the flexure-shear interaction can be obtained in the
element level [6, 7, 8, 9, 10, 11]. This method is simple and easy to be implemented, however, the element
is actually a macroscopic phenomenological model and the parameters should be calibrated or identified
according to the section properties (e.g., section area and reinforcing details) since the spring behavior is

∗ Correspondingauthor.
Email address: dcfeng@seu.edu.cn (De-Cheng Feng)

Preprint submitted to Engineering Structures November 28, 2018

based on empirical force-deformation relations. The second way is to construct the element formulation
based on the Timoshenko beam theory, and uses a multi-dimensional constitutive law for concrete fibers,
thus the flexure-shear interaction can be obtained at both the material level and section level. Petrangeli et
al. [12] first proposed a force-based model that accounts for shear effects using a constitutive model based
on the micro-plane concrete model. Then the smeared-crack models are introduced into the Timoshenko
beams, both displacement-based [13, 14, 4] and forced-based [15]. Some researchers also developed new
Timoshenko beams based on damage mechanics [16]. All of such elements seem to be more complicated
compared with those in the first way, however, they are physical meaningful and more accurate, thus can
be generally used in engineering practice. Moreover, the detailed stress and strain state of section fibers
can be achieved. Obviously, it is the tendency in structural engineering to combine Timoshenko fiber beam
with multi-dimensional material constitutive laws to reflect flexure-shear interaction with the development
of numerical methods and computer technology.
Although several models can be used for concrete constitutive law, the smeared-crack models, i.e., the
modified compression field theory (MCFT) family [17, 18] and the softened truss model (STM) family
[19, 20, 21, 22], and the damage-plasticity models [23, 24, 25, 26] seem to be the favorite ones. The
MCFT and STM rely on the experimental studies of RC panels subjected to shear, and establish the
constitutive model for cracked reinforced concrete in terms of average stress and average strain. The two
typical characteristics of reinforced concrete subjected to shear, i.e., tension-stiffening and compression-
softening, can be represented by the MCFT and STM. The models are the pioneer work in determining
the realistic shear behavior of RC structures, and their main contribution is the introduction of softening
coefficient that considers compression-softening. However, these models are based on the empirical uniaxial
stress-strain relationships of concrete, whose hysteretic rules are rather complex, leading to some numerical
issues from the computational point of view, especially for cyclic loading and dynamic loading [27, 28].
On the other hand, damage mechanics offers an alternative way to construct the concrete constitutive
model. The degradation of material is represented by a generalized internal variable, i.e., damage, and
the residual deformations are explained by the plastic strains. Based on the thermodynamics, damage
and plasticity of concrete are organised in a unified framework. Damage-plasticity models have a solid
theoretical foundation, ensuring that they can be used as a standard tool for nonlinear analysis of RC
structures [27, 28, 29]. Unfortunately, the compression-softening effect of reinforced concrete in shear has
not been considered in damage models up till now, causing an overestimate of the shear behavior of RC
structures.
When using local constitutive models in beam elements for nonlinear analysis of RC members, the
localization issues may arise for strain-softening problems. The plastic deformations will be concentrated on
a single element (for displacement-based element) or a single integration point (force-based element) [30].
Several regularization methods have been developed for both kinds of elements, i.e., constant fracture energy
methods [30], nonlocal integral models [31, 32], gradient models [33], and plastic hinge integration methods
[3, 34, 35]. However, most approaches have been applied to Euler-Bernoulli beams only. The localization
issues for Timoshenko beams, especially the flexure-shear fiber beam, have rarely been discussed [5].
Based on the above-mentioned background, this paper aims at developing a new displacement-based Ti-
moshenko fiber beam element that including flexure-shear interaction. A softened damage-plasticity model,
in which the softening coefficient in MCFT and STM is adopted to modify the compressive damage variable
to reflect compression-softening, is used for concrete. In addition, the model parameters is related to the
fracture energy to avoid mesh-sensitivity in strain-softening problems. The paper is stuctured as follows:
in Section 2, the softened damage-plasticity model is first introduced. Section 3 gives the formulation and
finite approximation of the classical Timoshenko beam element. The fiber section state determination is
discussed in Section 4. In Section 5, a series of RC beams under monotonic loading and RC columns and
wall under cyclic loading are simulated by the proposed element and Euler-Bernoulli element, demonstrating
the performance and advantage of the proposed element. Finally, the conclusions are drawn in Section 6.

2
2. Softened damage-plasticity model for concrete

2.1. Theoretical framework

The bi-scalar damage-plasticity model developed by Wu et al. [26] is adopted as the framework in the
present paper. According to the damage-plasticity theory, the strain tensor  is split to a elastic part and
plastic part

 = e + p (1)
where the superscripts ”e” and ”p” denote the elastic part and plastic part, respectively.
Calling for the hypothesis of strain equivalence [23], i.e., the strain undergoing a damaged state in the
Cauchy stress space is equivalent to the undamaged state in the effective stress space, the effective stress σ
of concrete can be expressed as

σ = E0 : e = E0 : ( − p ) (2)
where E0 is the fourth-order initial elastic stiffness tensor.
Considering the different behaviors of concrete material under tension and compression, the effective
stress can be further decomposed into a positive (tensile) part and a negative (compressive) part, i.e.,

σ = σ+ + σ− (3)
where the superscripts ” + ” and ” − ” denote the positive part and negative part, respectively. The spectral
decomposition method [26] can be followed for Eq. 3 to obtain the positive and negative effective stresses
(
σ + = P+ : σ
(4)
σ − = P− : σ
where P± are the projective tensors, which define the eigen-tensors of the fourth order damage tensor, and
are expressed by the principal directions of effective stress [25, 26]
 X
P+ = H(σ̂ i )pi ⊗ pi ⊗ pi ⊗ pi
t (5)
 −
P = I − P+

in which σ̂ i is the i − th principal stress of the effective stress and pi is the associated principal direction;
H(·) is the Heaviside function; I is the identity tensor.
In order to establish the constitutive law through a thermodynamic manner, the Helmholtz free energy
(HFE) potential should be defined. Meanwhile, introducing two damage variables d+ and d− to represent
the tensile and compressive damages respectively, the HFE ψ can be expressed as

ψ = (1 − d+ )ψ0+ + (1 − d− )ψ0− (6)

±
where damage d are the continuum measures of material degradation induced by cracks and defects in the
sub-scale; ψ0± are the initial HFE, which can be also split to a elastic part ψ0e± and plastic part ψ0p± , and
the expressions for each part are

ψ e± = 1 σ ± : e
0
2 (7)
ψ p± = R σ ± : dp
0

According to the second principle of thermodynamics, the Clausius-Duhem inequality can be expressed
as

γ̇ = −ψ̇ + σ : ˙ ≥ 0 (8)

3
 Substitution of Eqs .6-7 in Eq. 8 leads to the final expression of the constitutive relation, i.e.,

σ = (I − D) : σ = (I − D) : E0 : ( − p ) (9)
where σ is the second-order Cauchy stress tensor; D is the fourth-order damage tensor, and

D = d+ P+ + d− P− (10)
The detailed derivation of above Eqs. 8-10 can be found in Appendix A.
Obviously, the evolution of plastic strain and damage variables in Eq. 9 should also be defined. Actually,
the strain equivalence hypothesis in Eq. 2 enables us to consider the plasticity and damage separately. In
the plasticity sub-space, the plastic strain evolution can be determined in the effective stress space through
either a theoretical way or an empirical way. The theoretical model [23, 26] relies on the classical plasticity
theory and needs an iterative process to calculate the plastic strain at each step, which will increase the
computational cost, especially for the large scale structural analysis [25, 28], while the empirical model
[25, 28, 36] regards the plastic strain as an ”overall effect” and uses an explicit expression to describe the
plastic strain, thus the numerical efficiency is greatly enhanced. Considering these aspects, the empirical
plastic model proposed by Faria et al. [25] and then modified by Wu [36] is adopted here, i.e.,

˙ p = bp σ (11)
p
where b denotes the empirical plastic flow parameter
e
h : i
˙
bp = ξ p E0 H(d˙− ) ≥0 (12)
σ:σ
where ξ p ≥ 0 denotes a plastic coefficient that controls the plastic strain rate, and usually between 0.1 to 0.3;
E0 denotes the initial Young’s modulus. Note that the tensile plastic strain is neglected since it is relatively
small.
On the other hand, in the damage sub-space, the damage energy release rates Y + and Y − are treated
as the damage driven force, which can be derived from Eq. 6
∂ψ
Y± =− = ψ0± (13)
∂d±
and its simplified expressions are [26]

+ 1
σ + : E−1


Y + = ψ 0 ≈

2 0 :σ
2
(14)
 q
− −
Y −
 = ψ0− ≈ b0 αI 1 + 3J 2

− −
where I 1 denote the first invariant of the compressive effective stress σ − ; J 2 denote the second invariant
of the deviator s− of the compressive effective stress σ − ; b0 and α are the material parameters [26].
Furthermore, Li and Ren [37] proposed the energy equivalent strains according to the damage consistent
condition, making it easier to calculate the the multi-axial damage evolution through a uniaxial damage
evolution function. The energy equivalent strains are expressed as
 q
eq+ = 2Y +
E0 q
(15)
eq− = 1 Y−
E0 (1−α) b0

The uniaxial damage evolution functions are obtained from some experimental results [38, 39]

ρ± n±
1 − ± x± ≤ 1
±
d = n −1+(x± )n
±
(16)
1 − ± ±ρ± 2 ± x± > 1
α (x −1) +x

4
 eq± fc± Ec ±
x± = , ρ± = , n± = c
(17)
±
c E c ±
c E c ± ±
c − fc

where Ec is the elastic modulus of concrete; fc± and ±c are the tensile/compressive strength and the cor-
responding strain, respectively; α± are the descending parameters that control the shape of the descending
part of stress-strain curves.

2.2. Compression-softening effect

Although the previous damage-plasticity model unifies the damage and plasticity in one framework, and
the model can accurately capture the typical behaviors of plain concrete with great numerical efficiency,
some limitations still remain. As shown in Fig. 1, when RC structures subjected to shear, the principal
compressive strength will be softened by the transverse tensile cracks and stresses, i.e., the compression-
softening effect, can not be reflected by the damage-plasticity model. Therefore, there is a need to add the
compression-softening in the framework of damage-plasticity model to make it suitable for analysis shear
behaviors of RC structures.
2

Plain concrete
Cracked
2
reinforced concrete
1 2
2

2 1
Un-softened
Softened

2

Figure 1: Compression-softening effect of reinforced concrete

In fact, as shown in Fig. 2, the effective stress decomposition defines tensile and compressive stress state
respectively, and the projective tensors P± in Eq. 5 are expressed by the principal effective stress directions.
Consequently, the damage-plasticity model is an equivalent form of the rotating angle smeared-crack models,
i.e., MCFT and rotating-angle STM, since rotation of the principal effective stress yields the rotation of the
principal damage [29]. Thus, the compression-softening effect can be easily implemented in the model.
Introducing the softening coefficient β in the principal compressive direction, the stress can be written
as

σ2s = 1 − ds− σ̂ 2 = β 1 − d− σ̂ 2



(18)
where the superscript ”s” denotes the compression-softening effect, and the new damage variable ds− is
expressed as

ds− = 1 − β 1 − d−


(19)

5
 ˆ 2 ̂ 1 ̂ 1 ˆ 2

x2 p1 p1 p1

x1
p2 p2 p2
̂ 1 ˆ 2 ̂ 1 ˆ 2

σ σ in principal directions σ σ

Figure 2: Effective stress decomposition

The softening coefficient can be obtained from curve fitting of experimental data. Currently, two options
are available, i.e., the one in MCFT [17] and the one in STM [19]
(
1
βM CF T = 0.8+170 ≤1
1
1
(20)
βST M = √1+400 ≤ 1
1

where 1 denotes the principal tensile strain.

Actually, the two coefficients in the two theories have the same trend and shape, the difference may
be just caused by the fact that they are fitting from different experimental data, hence either one can be
adopted herein. In this paper, The βST M is adopted to represent the softening effect since the curve is more
smooth and continuous compared with βM CF T . The principal tensile strain 1 is replaced by the max tensile
energy equivalent strain e+
max of the loading history here, which makes it more convenient to calculate under
complex multi-axial stress state and accounts for the accumulated influence of compression-softening under
reverse loading, and the monotonic character of the damage variable can be guaranteed, that is
1
β=p (21)
1 + 400e+
max

Combining the constitutive relation in Eq. 9, the Cauchy stress now becomes

σ = 1 − d+ σ + + 1 − ds− σ −



(22)
Thus, the softened damage-plasticity model is derived as
(
σ = (I − Ds ) : E0 : ( − p )
(23)
Ds = d+ P+ + ds− P−
where Ds denotes the modified fourth-order damage tensor considering compression-softening effect.
The softened compressive strength with respect to principal tensile strain by the model is shown in
Fig.3(a). Meanwhile, the comparison of the stress-strain curves by STM, MCFT and proposed model under
different tensile principal strain 1 are also given in Fig.3(b). Although there are some small differences
between the curves, the shape and trend of the curves are the same. Principal tensile stress not only cause
softening of the compressive strength, but also has an effect on slope of the curve before and after the peak
strength, which is also very important for the calculation of the realistic behavior of structures.
Furthermore, the typical performance of the proposed concrete model during a uniaxial tension-compression
cyclic test is shown in Fig. 4. A complex cyclic loading scheme was imposed on the test specimen. It was
first under tension to the peak strength (path O-A), and then entering the descending branch (path A-B),
tensile damage developed during the whole process. Loading is then reversed returning to point O and a
subsequent incursion into compression up to point C, where compressive damage and plastic deformation
are observed. Then a new reverse was applied, and the response was along path C-D-E-F. Between points

6
 Compressive strength (MPa)

30

20

10

0
0
0.002 0.01
0.004 0.008
Ten 0.006 0.006
sile 0.004
0.008 in
str 0.002 e stra
ain 0.01 ressiv
0 Comp

(a) Softened stress-strain relationship by proposed model

3 5

3 0 S T M
M C F T
2 5
P ro p o se d m o d e l

2 0
S tre s s (M P a )

1 5

1 0
ε1 = 0 , 0 . 0 0 3 , 0 . 0 0 5
5

0
0 .0 0 0 0 .0 0 2 0 .0 0 4 0 .0 0 6 0 .0 0 8 0 .0 1 0
S tra in
(b) Comparison of stress-strain curves by STM, MCFT and proposed model

Figure 3: Stress-strain relationship by proposed model

7
E and F further tensile damage occurs. The proposed model can successfully reproduce the softening be-
haviour under tension, as well as the hardening and softening which occurs in concrete under compression.
The stiffness of unloading paths B-O, F-D, and C-D are obviously smaller than the initial stiffness (path
O-A), indicating that stiffness degradation due to damage can be well reflected by the model. Note also the
stiffness recovery due to cracks closure that takes place during paths B-O-C or F-D-G can also be captured
by the proposed model. Moreover, the constitutive model has also the capability of maintaining the plastic
deformations induced during previous compressive damaging, as demonstrated by the horizontal shift D-O
experienced by curve D-E-F, corresponding to an irreversible strain.

A T e n s io n
E B
D F
O
S tre s s

G
C o m p r e s s io n C

S tra in
Figure 4: Typical cyclic behavior of the model

2.3. Regularization for mesh-objectivity

Concrete is a strain-softening material and localization issues may occur when using local constitutive
model in nonlinear finite element analysis. In the past two decades, a variety of regularization methods
have been proposed to overcome this problem, i.e., crack band models, the nonlocal integral models, and
the gradient models, etc. Among these regularization techniques, the crack band model seems to be the
most simple and efficient one, and has been successfully applied to damage models [40]. By introducing the
constant tensile and/or compressive fracture energy, the material parameters can be calculated according
to the element mesh size, thus the mesh-sensitivity can be partially removed. However, in most models
only the tensile parameters are regularized, since the constant compressive fracture energy is a controversial
problem in structural engineering, and the theoretical derivation of the fracture energy of the constitutive
model is also very complicated.
In this paper, a simple method is developed based on the fracture energy regularization technique. As
shown in Fig. 5, the specific fracture energies are represented by the area under the uniaxial stress-strain
relation curves, i.e.,
 +
 Gf = R σ + d+
lch
− (24)
 Gf = R σ − d−
lch

where the G+
f and G−
f are the tensile and compressive fracture energy, respectively; lch is the characteristic
length related to the dimension of the element mesh, and in one-dimensional problems it is assumed as the
element length.
8
 

Tensile fracture energy

ft
 20 0
t  tu 
0.2 f c

fc
Compressive fracture energy

Figure 5: Fracture energies

Due to the integral in Eq. 24 of the proposed damage model is difficult to derive, here the bilinear model
in tension and Kent-Scott-Park model in compression [30] as shown in Fig.5 are used to derive Eq. 24, i.e.,
 +
 Gf = 1 ft tu
lch 2
−   (25)
 Gf = 0.6fc 20 − 0 + 0.8 fc
lch Ec

where Ec is the elastic modulus; ft is the tensile strength; tu is the ultimate tensile strain; fc is the
compressive strength; 0 is the compressive strain corresponding to the peak strength; 20 is the compressive
strain corresponding to the residual strength, i.e., 20% of the peak strength.
Hence, the uniaxial curves under different mesh types may be obtained and the material parameters of
the softened-damage model under different mesh sizes may be determined by imposing the one-dimensional
numerical curves to fit the curves by the above-mentioned method.
Although the method eliminates the mesh-sensitivity in a general sense, it is rather simple and can be
directly used in the finite element analysis without additional computational cost, and avoids a complicated
implementation of a nonlocal form or a gradient form of the proposed softened damage-plasticity model.

3. Formulation for Timoshenko beam element

3.1. Beam kinematics

The proposed element is based on the classical two-dimensional displacement-based Timoshenko beam
theory without geometrical nonlinearity, while the extension for a three-dimensional case is straightforward.
The element has a length L and the generalized displacement field u is given by
T
u (x) = {u (x) , w (x) , ϑ (x)} (26)
where u (x) is the longitudinal displacement along axis x; w (x) is the transverse deflection along axis y;
ϑ (x) is the rotation about axis z. The corresponding strain field of the beam section is
T
ε (x) = {ε (x) , γ (x) , κ (x)} (27)

9
where ε (x), γ (x), and κ (x) are the longitudinal extension, shear deformation and curvature of the section,
respectively.

w
x

 

w( x )

y Middle axis
u( x )

 ( x)
x
z
Figure 6: Timoshenko beam kinematics

According to the Timoshenko beam theory, the section of the beam remains plane but not necessarily
normal to the deformed longitudinal axis, as shown in Fig. 6, hence the deformations of the section at
position x are expressed as
 ∂u(x)
ε (x) = ∂x

γ (x) = ∂w(x)
∂x − ϑ (x)
(28)
 ∂ϑ(x)
κ (x) = ∂x


Therefore, the generalized expression of the beam strain field is

ε (x) = Lu (x) (29)

where L (·) is a kinematic operator as
∂
 
∂x 0 0
∂
L (·) =  0 ∂x −1 (30)
∂
0 0 ∂x

3.2. Finite element approximation

There are several choices for the finite element approximation of the above-mentioned Timoshenko beam.
The simplest way is to use a two-node element with linear interpolation functions approximating the strain
field. However, this kind of element suffers from the shear-locking problem when modeling slender beams.
The problem is actually caused by the inability of the element to produce a constant transverse shear strain.
Normal solution to this problem is to use the reduced integration method, i.e., using only one integration
point at the beam center [41]. Although the shear-locking can be avoided by the above method, its numerical
performance is really poor. An alternative approach is to add a bubble term at the middle of the element
and use a mixed interpolation to approximate the strain field. With this idea, the displacement field of the
element is approximated as [42]

x x


u (x) = 1 − L
u1 + L u2

x x
w2 − x2 1 − L
x
(ϑ2 − ϑ1 ) + 2x 2x x




w (x) = 1 − L w1 + L 3 L −1 1− L ϑ3 (31)
x x x x



ϑ (x) = 1 − L ϑ1 + L ϑ2 + 4 L 1− L ϑ3


10
where ui , wi , ϑi , i = 1, 2 are the nodal displacements at the element ends; ϑ3 is the bubble term in rotation
at the middle of the element. Eq. 31 can be written in a generalized form
2
X
u (x) = Ni (x) di + N3 (x) d3 (32)
i=1
T
where di = {ui , wi , ϑi } are the nodal displacements at the two end-node and d3 = ϑ3 is bubble rotation;
the corresponding interpolation functions are

x
 
1− L 0 0

x x x 
N1 (x) =  0 1− L 2 1− L (33a)
x
0 0 1− L
x 
L 0 0
x
− x2 1 − Lx 


N2 (x) =  0 L (33b)
x
0 0 L
 
0

N3 (x) =  2x 2x x


3 L − 1 1
− L
 (33c)
x x
4L 1− L

Accordingly, the strain field of the element can be obtained by substituting Eq. 33 to Eq. 29
2
X
ε (x) = Lu (x) = Bi (x) di + B3 (x) d3 (34)
i=1

where Bi denotes the compatible matrices, which are given by

 
1 0 0
1
B1 (x) = LN1 (x) = − 0 1 L2  (35a)
L
0 0 1
 
1 0 0
1
B2 (x) = LN2 (x) = 0 1 − L2  (35b)
L
0 0 1
 
0
1
B3 (x) = LN3 (x) =  − 2L 3
 (35c)
L
4 1 − 2x


L

Then the element stiffness K ele and internal force vector F int can be readily determined through a
displacement-based finite element method
Z
ele
K = B T (x) K sec (x) B (x) dx (36)
L
Z
F int = B T (x) F sec (x) dx (37)
L

where K sec (x) is the section stiffness; F sec (x) is the section resisting force vector; the matrix B (x) has
the following form
 
B (x) = B1 (x) B2 (x) B3 (x) (38)

11
4. Fiber section state determination

4.1. Fiber section modeling

To compute the section stiffness and resisting force, the fiber section is adopted. The model discretizes
the section into steel fibers and concrete fibers, and the uniaxial Menegotto-Pinto model [43] is used to
describe the material behavior of steel fibers while the softened damage-plasticity model is used for concrete
fibers. As shown in Fig. 7, the fiber strains  (x) at height y are obtained according to the plane section
hypothesis

 (x, y) = I (y) ε (x) (39)

T
where  (x, y) = {xx , γxy } , and I (y) is the section compatibility matrix
 
1 0 −y
I (y) = (40)
0 1 0
Note that the shear strain is assumed to be uniformly distributed along the section height, and the shear
strain of steel fibers is neglected.

Concrete fiber

 yy ,  yy
z
 xy ,  xy

 xx ,  xx
Steel fiber

Figure 7: Fiber section

Call for the material constitutive law, the stress σ (x, y) and stiffness E (x, y) of the fibers can be obtained
by
(
σ (x, y) = σ ()
(41)
E (x, y) = ∂σ/∂
T
where σ (x, y) = {σxx , τxy } .
Integrating over the entire section, the section resisting force can be given by
   R 
 N (x)   RA σxx dA  Z
F sec (x) = V (x) = R A τxy dA = I T (y) σ (x, y) dA (42)

M (x)
 
−yσxx dA
 A
A
where N (x) , V (x) , M (x) are the section axial force, shear and bending moment, respectively; A is the area
of the section.
Therefore, the section stiffness can be also derived
∂F sec
Z Z
sec T ∂σ
K (x) = = I (y) dA = I T (y) E (x, y) I (y) dA (43)
∂ε A ∂ε A
The integral in Eq. 42 and Eq. 43 can be computed using the normal rectangular method.
12
4.2. Condensation for concrete fibers
As shown in Fig. 7, the fiber section offers two components of the strain tensor in the two-dimensional
beam problem, i.e., the axial strain xx and the shear strian xy = γxy /2 for each concrete fiber, however,
the transverse strain yy is unknown. In order to determine the value of yy , the internal equilibrium
between steel fibers and concrete fibers is imposed by setting the stress component σyy = 0. In general, the
incremental form of the constitutive relation can be written as
    
∆σxx  D11 D12 D13 ∆xx 
∆σyy = D21 D22 D23  ∆yy (44)
∆σxy D31 D32 D33 ∆xy
   

where Dij , i, j = 1, 2 is the components of the material stiffness matrix determined according to the softened
damage-plasticity model. Thus the the transverse strain is obtained by
∆xx D21 + ∆xy D23
∆yy = − (45)
D22
Substitution of Eq. 45 in Eq. 44 leads to the condensed stiffness matrix for concrete fiber
    
∆σxx k11 k12 ∆xx
= (46)
∆σxy k21 k22 ∆xy
where

k11 = D11 − D12 D21 /D22 k12 = D13 − D12 D23 /D22
(47)
k21 = D31 − D32 D21 /D22 k22 = D33 − D32 D23 /D22

However, the above process needs an iterative procedure to get the correct yy . A Newton-Raphson
algorithm is adopted here and the detailed iterative procedure are listed below:
Step 1. For a iterative step i, give the known components of the strain increments ∆ixx and ∆ixy , and
assume the transverse strain increment ∆iyy = ∆i−1yy .
Step 2. Update the strain tensor i = i−1 + ∆i , then calculate the stress tensor σ i and stiffness tensor
D i based on the damage model.
i
Step 3. Judge σyy < tolerance, if yes then exit and compute the condensed stiffness through Eq. 46;
if not then update the transverse strain increment as ∆i+1 i i i
yy = ∆yy − ∆σyy /D22 and return to Step 2.

5. Numerical examples

The proposed Timoshenko beam element is implemented in ABAQUS through the subroutine user-
defined element (UEL) [44]. To validate the performance of the proposed element, several numerical examples
are demonstrated in this section. First a series of RC simply supported beams with different failure types is
simulated by the element. As a comparison, the beams are also simulated by the Euler-Bernoulli element.
Then the RC columns and walls under cyclic loading is analyzed to represent the capacity of the element in
modeling hysteretic behaviors of the RC structural members.

5.1. RC simply supported beams

Bresler and Scordelis [45] have conducted a series of RC simply supported beams that with different shear-
span, section dimension, reinforcing detail, material property, etc., to investigate their typical behaviors. The
test is treated as a benchmark in literatures since the shear critical behavior has been observed in some of
the specimens. Totally 12 beams of 3 span lengths are tested under mid-span concentrated load. The
geometrical dimensions and the section reinforcement details of the beams are shown in Fig. 8. Note that
in the name the beam specimens, letters ”OA, A, B, C” represent four kinds of section dimensions, and
numbers ”1, 2, 3” represent the 3 span lengths. Three kinds of reinforcement bars are used in the specimens,
13
 310 305 307

461

466

462
OA1 OA2 OA3

556

561

556
Series 1

307 305 307

220 3660 220

466

464

466
A1 A2 A3

561

559

561
Series 2
231 229 229

466
461

461
B1 B2 B3

561
556

556
220 4570 220

155 152 155

Series 3 No.4
bars

464
464

459
C1 C2 No.2 C3

559
559

554
bars

220 6400 220 No.9

bars

Figure 8: Simply supported RC beam tests

No.9 bars are used for the bottom longitudinal bars, No.4 bars are used for the top longitudinal bars, and
No.2 bars are used for the stirrups. The material properties for concrete and steel are listed in Table 1.
−
Note the fracture energies are set as G+ f =0.6 N/mm, Gf =60 N/mm for series 1-2 (the compressive strength
+ −
is around 22 MPa), and Gf =0.8 N/mm, Gf =80 N/mm for series 3 (the compressive strength is around 37
MPa).
The beams are modeled with both the proposed Timoshenko element and Euler-Bernoulli element. Five
elements are used in the simulation, and the section are discretized with 20 concrete fibers. The steel fibers
are assigned according to their positions and areas for different specimens. The analysis is terminated when
convergence has not been achieved, which also can be treated as the failure of the specimen. The numerical
results, i.e., load-deflection relationships at mid-span, by finite element models and the experimental results
are compared in Fig. 9. It can be found that excellent agreement is achieved between the experimental data
and numerical predictions by the Timoshenko beam for all cases.
Generally, the results by the Timoshenko element are better than that by the Euler-Bernoulli element.
Especially, the differences between the results by Timoshenko and Euler-Bernoulli element become larger
with the decrease of the beam span, as expected, since shear deformations are larger in beams of shorter
spans. For series 1 (OA1-C1), whose span is the shortest, the Timoshenko beam can capture the experimental
behavior, and even the failure load which is caused by diagonal tension and inclined cracks. On the other
hand, the Bernoulli element will overestimate the failure loads for series 1, since it could not capture the
shear failure mode. For series 2 (OA2-C2), whose has a moderate span, the differences become smaller
but the failure loads by Euler-Bernoulli element are still larger than the experimental results. For series
3 (OA3-C3), the results by the two elements are nearly the same since it has a relatively long span and
the shear deformation is no longer evident. Meanwhile, for the OA- specimens, which have no stirrups,
the simulated failure loads by the Timoshenko element match the experimental results very well, while the
results by Euler-Bernoulli element is much higher than the test data. For example, tested failure load of
specimen OA1 is 333.88 kN, while the numerical results by Timoshenko and Euler-Bernoulli element are
380 kN and 597 kN, respectively. The reason is that OA1 is failed due to diagonal tension with the opening
of inclined cracks, which is a typical shear failure and cannot be captured by Euler-Bernoulli element.
In addition, the responses of beams B1, B2 and B3 predicted by the damage-plasticity model with and
without compression-softening effect are also given in Fig.10. From the figures we can find that for the
beams with a small span-to-depth (B1 and B2), which means the shear behavior is obvious, the responses,
14
 6 0 0 7 0 0

6 0 0
5 0 0

5 0 0
4 0 0

4 0 0
L o a d (k N )

L o a d (k N )
3 0 0
3 0 0

2 0 0
2 0 0
E x p . d a ta E x p . d a ta
1 0 0 E u le r-B e rn o u lli 1 0 0
E u le r-B e rn o u lli
T im o s h e n k o T im o s h e n k o

0 0
0 2 4 6 8 1 0 1 2 1 4 1 6 0 2 4 6 8 1 0 1 2 1 4 1 6 1 8 2 0
D e fle c tio n (m m ) D e fle c tio n (m m )

(a) OA1 (b) A1

5 0 0 6 0 0

5 0 0
4 0 0

4 0 0
3 0 0
L o a d (k N )

L o a d (k N )

3 0 0

2 0 0
2 0 0

E x p . d a ta E x p . d a ta
1 0 0
E u le r-B e rn o u lli 1 0 0 E u le r-B e rn o u lli
T im o s h e n k o T im o s h e n k o

0 0
0 2 4 6 8 1 0 1 2 1 4 1 6 0 5 1 0 1 5 2 0 2 5 3 0
D e fle c tio n (m m ) D e fle c tio n (m m )

(c) OA2 (d) A2

5 0 0 6 0 0

5 0 0
4 0 0

4 0 0
3 0 0
L o a d (k N )

L o a d (k N )

3 0 0

2 0 0
2 0 0

E x p . d a ta E x p . d a ta
1 0 0
E u le r-B e rn o u lli 1 0 0 E u le r-B e rn o u lli
T im o s h e n k o T im o s h e n k o

0 0
0 3 6 9 1 2 1 5 1 8 2 1 2 4 2 7 3 0 3 3 3 6 0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 0
D e fle c tio n (m m ) D e fle c tio n (m m )

(e) OA3 (f) A3

15
 6 0 0 4 0 0

5 0 0
3 0 0
4 0 0
L o a d (k N )

L o a d (k N )
3 0 0 2 0 0

2 0 0

E x p . d a ta 1 0 0
E x p . d a ta
1 0 0 E u le r-B e rn o u lli E u le r-B e rn o u lli
T im o s h e n k o T im o s h e n k o
0 0
0 2 4 6 8 1 0 1 2 1 4 1 6 1 8 2 0 0 2 4 6 8 1 0 1 2 1 4 1 6 1 8 2 0 2 2 2 4
D e fle c tio n (m m ) D e fle c tio n (m m )

(g) B1 (h) C1

5 0 0 4 0 0

4 0 0
3 0 0

3 0 0
L o a d (k N )

L o a d (k N )

2 0 0

2 0 0

E x p . d a ta 1 0 0 E x p . d a ta
1 0 0 E u le r-B e rn o u lli E u le r-B e rn o u lli
T im o s h e n k o T im o s h e n k o

0 0
0 4 8 1 2 1 6 2 0 2 4 2 8 0 4 8 1 2 1 6 2 0 2 4
D e fle c tio n (m m ) D e fle c tio n (m m )

(i) B2 (j) C2

5 0 0 3 5 0

3 0 0
4 0 0

2 5 0

3 0 0
2 0 0
L o a d (k N )

L o a d (k N )

1 5 0
2 0 0

1 0 0
E x p . d a ta E x p . d a ta
1 0 0
E u le r-B e rn o u lli 5 0
E u le r-B e rn o u lli
T im o s h e n k o T im o s h e n k o

0 0
0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 0 0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 0
D e fle c tio n (m m ) D e fle c tio n (m m )

(k) B3 (l) C3

Figure 9: Numerical and experimental mid-span load-deflection curves of the beams

16
 Table 1: Material properties of Bresler-Scordelis beams

Reinforcement
Bar type
d (mm) fy (MPa) fu (MPa) Es (MPa) hardeing ratio
No.2 6.4 325 430 190000 0.01
No.4 12.7 345 542 201000 0.01
No.9 28.7 555 933 218000 0.01
Concrete
Beam number
+
c fc+ (MPa) −
c fc− (MPa) α+ /α−
OA1 1×10−4 1.80 0.002 22.6 0.5/2.0
OA2 1×10−4 1.89 0.002 23.7 0.8/3.2
OA3 1×10−4 2.89 0.002 37.6 1.5/4.8
A1 1×10−4 1.92 0.002 24.1 0.5/2.4
A2 1×10−4 1.94 0.002 24.3 0.8/3.8
A3 1×10−4 2.72 0.002 35.1 1.4/4.2
B1 1×10−4 1.98 0.002 24.8 0.5/2.8
B2 1×10−4 1.85 0.002 23.2 0.8/3.0
B3 1×10−4 2.96 0.002 38.8 1.7/5.0
C1 1×10−4 2.34 0.002 29.6 0.6/3.8
C2 1×10−4 1.90 0.002 23.8 0.8/3.0
C3 1×10−4 2.72 0.002 35.1 1.4/4.2

as well as the failure loads, may be overestimated if compression-softening is not considered. For the beam
B3, whose shear behavior is not evident, the results by the two models are close to each other.

5.2. RC columns
The proposed element is also applied to the RC structural members under cyclic loading. Six RC
columns of different cases are firstly simulated, i.e., two shear (S) dominant columns, two flexure-shear (FS)
columns and two flexure dominant (F) columns. The columns are named by their failure types (S, FS or
F) and number (1 or 2). For instance, Col FS 1 represents no.1 specimen of the flexure-shear columns.
The geometrical information, reinforcement profiles, and the material properties of the columns are given in
Table 2.
Table 2: Geometrical and material properties of the columns

Specimen Col S 1 Col S 2 Col FS 1 Col FS 2 Col F 1 Col F 2

Reference [46] [47] [48] [49] [50] [51]
Aspect ratio L/h 1.25 1.65 2 3.22 4 4.7
Column length (mm) 225 825 500 1473.2 2438.4 1645
Section b×h (mm×mm) 180×180 400×500 250×250 457×457 609.6 (spiral) 350×350
Axial force (kN) 476 392 161 503 654 961
Concrete strength fc (MPa) 33.0 27.1 31.6 33.1 31 34
Peak strain c 2.4×10−3 2.1×10−3 2.0×10−3 2.0×10−3 2.0×10−3 2.0×10−3
Longitudinal bars (mm) 8 φ12.7 14 φ22 8φ12.7 8φ25.4 44φ15.9 12φ19.5
Yielding strength fyx (MPa) 340 318 382 331 462 455.6
Stirrups (mm) φ4@64.3 φ9@100 φ3.9@37 φ9.5@457.2 φ6.4@31.8 φ6.6@76
Yielding strength fyy (MPa) 249 336 386.5 399.9 606.8 580

The columns are all simulated with three elements, and the cross-section is divided to 20 concrete fibers
and several steel fibers, whose number and area are determined with each column. The analysis procedure

17
 6 0 0 5 0 0

5 0 0
4 0 0

4 0 0
3 0 0
L o a d (k N )

3 0 0 L o a d (k N )
2 0 0
2 0 0

E x p . d a ta E x p . d a ta
1 0 0
1 0 0 W ith o u t s o fte n in g W ith o u t s o fte n in g
W ith s o fte n in g W ith s o fte n in g
0 0
0 2 4 6 8 1 0 1 2 1 4 1 6 1 8 0 5 1 0 1 5 2 0 2 5 3 0
D e fle c tio n (m m ) D e fle c tio n (m m )
(a) B1 (b) B2

5 0 0

4 0 0

3 0 0
L o a d (k N )

2 0 0

1 0 0
E x p . d a ta
W ith o u t s o fte n in g
W ith s o fte n in g
0
0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 8 0
D e fle c tio n (m m )
(c) B3

Figure 10: Comparison of results by damage-plasticity model with and without compression-softening

18
includes two stages: first the axial load is imposed at the top of the column through force control and then
the cyclic lateral load is applied through displacement control. The results of top displacement vs. lateral
load curves of the columns by Euler-Bernoulli element and Timoshenko element are compared with the
experimental data in Fig. 11, respectively.
For shear dominant columns (Col S 1 and Col S 2), whose aspect ratios are 1.25 and 1.65, respectively,
the shear deformations are obvious and have a great influence on the behavior of columns so that the
Euler-Bernoulli element overestimates the initial stiffness, loading capacity, unloading stiffness and energy
dissipation of the columns. On the other hand, the Timoshenko element can capture the loading capacity
and pinching effect of the hysteretic behavior of the columns since the shear contribution can be considered
in both section and material levels.
For flexure-shear columns (Col FS 1 and Col FS 2), whose aspect ratios are 2 and 3.22, respectively, the
results by Timoshenko are still superior to those by Euler-Bernoulli element since flexure and shear actions
are similarly influential on the behavior of columns. Although the performance of Euler-Bernoulli element
are much better compared with the shear dominant cases, the peak strength, unloading stiffness and energy
dissipation by Euler-Bernoulli element are higher than the experimental results because the stiffness loss
due to inclined cracks cannot be accounted for.
For flexure dominant columns (Col F 1 and Col F 2), whose aspect ratios are 4 and 4.7, respectively,
the results by the two elements are both general agree well with the experimental data because the shear
deformations are not obvious in these cases, thus shear effect can be neglected in these cases.
Furthermore, since Col S 1 demonstrates an obvious softening response, it is selected to show the mesh-
objectivity of the proposed model. Three kinds of meshes are used, i.e., 3, 5, and 10 elements. The
compressive fracture is set as G− f =70 N/mm and the corresponding descending parameters for the three
meshes (3, 5, 10 elements) are α− =0.05, 0.03 and 0.01, respectively. The numerical responses are shown in
Fig. 12. It is observed that the results of different mesh types are relatively objective, indicating that the
proposed method to eliminate mesh-sensitivity is useful.

5.3. RC shear wall

The shear wall specimen RW2 tested by Thomsen and Wallace [52] is also analyzed by the proposed
Timoshenko element here. The wall is 3658 mm in height and 1219 mm in width, and a constant axial load
of 0.07Ag fc0 was applied at the top of the wall. The experimental details, e.g., the geometrical dimensions,
reinforcing information and material properties, can be found in [52].
The wall is modeled with 5 elements and the section is divided into 20 concrete fibers, the reinforcement
fibers are arranged according to their own positions and areas. The simulated top displacement versus lateral
load curve of the specimen is compared with the experimental result in Fig. 13. It can be found from the
figure that the numerical result matches the experimental result very well. Furthermore, the vertical strain
distributions of the base section at different drift levels obtained by proposed element are also given in Fig.
14 against the experimental ones. Evidently the numerical results agree with the experimental results well
again.

6. Conclusions

In this paper, a new displacement-based fiber element considering flexure-shear interaction is developed
for nonlinear analysis of RC structural members. The element is settled on the framework of classical
Timoshenko beam theory, and is free from shear-locking. The multi-dimensional damage-plasticity model,
in which a softening coefficient is introduced to account for the compression-softening effect of reinforced
concrete caused by shear, is used for concrete material. Meanwhile, the model parameters are related to
fracture energy to obtain mesh-objectivity in numerical simulations. The fiber section state determination,
including a condensation process for the multi-dimensional concrete constitutive model, is also discussed in
detail.
The element is validated through simulation of a series of RC simply supported beams under monotonic
loading and RC columns and wall under cyclic loading. The typical shear failure of the beam, and pinching

19
effect in the hysteretic behavior of the columns can be well reproduced by the element. The results also
indicate that the proposed element is significantly superior to Euler-Bernoulli element, which cannot reflect
flexure-shear interaction, in the cases where the shear deformation is evident, and the compression-softening
should be considered in these problems.

Acknowledgements

The authors would like to acknowledge financial supports from the National Key Research and Deve-
lopment Program of China (No. 2016YFC0701400), and the National Natural Science Foundation of China
(Nos. 51408126, 51528802, 51525801).

Appendix A. Derivation of the constitutive relation

Call for the second principle of thermodynamics, the Clausius-Duhem inequality can be expressed as

γ̇ = −ψ̇ + σ : ˙ ≥ 0 (A.1)
where the Helmholtz free energy potential ψ has the form

ψ = ψ e e , d+ , d− + ψ p p , d+ , d−



(A.2)
Differentiating both sides of Eq. A.2 with respect to time yields
∂ψ e e ∂ψ p
˙+ + ∂ψ : d˙− + ∂ψ : ˙p
ψ̇ = : ˙ + : d (A.3)
∂e ∂d+ ∂d− ∂p
Substituting Eq. A.3 into Eq. A.1 leads to

∂ψ e ∂ψ p
       
∂ψ ˙ ∂ψ ˙
σ− ˙e +
: + − + :d + − − :d + σ− − : ˙p ≥ 0 (A.4)
∂e ∂d ∂d ∂p
Referring to the standard thermodynamics arguments, along with the assumption that damage and
plastic unloading are elastic processes, for any admissible process the following conditions have to be fulfilled:
∂ψ e
σ= (A.5)
∂e
Considering the definition for the elastic HFE potential expressed in Eq. 7, Eq. A.5 leads to

∂ψ0e+
∂ψ0e−
σ = 1 − d+ e
+ 1 − d− (A.6)
∂ ∂e
then we obtain

σ = 1 − d+ σ + + 1 − d− σ − = 1 − d+ P+ : σ + 1 − d− P− : σ = (I − D) : σ





(A.7)
with the fourth-order damage tensor D

D = d+ P+ + d− P− (A.8)
Referring to the definition of the effective stress in Eq. 2, one obtains the final form of the constitutive
relation

σ = (I − D) : σ = (I − D) : E0 : ( − p ) (A.9)

20
References
[1] F. Taucer, E. Spacone, F. C. Filippou, A fiber beam-column element for seismic response analysis of reinforced concrete
structures, Vol. 91, Earthquake Engineering Research Center, College of Engineering, University of California Berkekey,
California, 1991.
[2] E. Spacone, F. C. Filippou, F. F. Taucer, Fibre beam-column model for non-linear analysis of r/c frames: Part i. formu-
lation, Earthquake Engineering & Structural Dynamics 25 (7) (1996) 711–726.
[3] M. H. Scott, G. L. Fenves, Plastic hinge integration methods for force-based beam–column elements, Journal of Structural
Engineering 132 (2) (2006) 244–252.
[4] Z. X. Li, Y. Gao, Q. Zhao, A 3d flexurecshear fiber element for modeling the seismic behavior of reinforced concrete
columns, Engineering Structures 117 (2016) 372C383.
[5] P. Ceresa, L. Petrini, R. Pinho, Flexure-shear fiber beam-column elements for modeling frame structures under seismic
loading state of the art, Journal of Earthquake Engineering 11 (sup1) (2007) 46–88.
[6] A. D’Ambrisi, F. C. Filippou, Modeling of cyclic shear behavior in rc members, Journal of Structural Engineering 125 (10)
(1999) 1143–1150.
[7] E. Spacone, A. Marini, Analysis of reinforced concrete elements including shear effects, Aci Structural Journal 103 (5)
(2006) 645–655.
[8] P. E. Mergos, A. J. Kappos, A distributed shear and flexural flexibility model with shearcflexure interaction for r/c
members subjected to seismic loading, Earthquake Engineering & Structural Dynamics 37 (12) (2008) 1349C1370.
[9] S. Y. Xu, J. Zhang, Hysteretic shearcflexure interaction model of reinforced concrete columns for seismic response asses-
sment of bridges, Earthquake Engineering & Structural Dynamics 40 (3) (2011) 315–337.
[10] S. Y. Xu, J. Zhang, Axialcshearcflexure interaction hysteretic model for rc columns under combined actions, Engineering
Structures 34 (1) (2012) 548–563.
[11] M. S. Lodhi, H. Sezen, Estimation of monotonic behavior of reinforced concrete columns considering shear-flexure-axial
load interaction, Earthquake Engineering & Structural Dynamics 41 (41) (2012) 2159C2175.
[12] M. Petrangeli, P. E. Pinto, V. Ciampi, Fiber element for cyclic bending and shear of rc structures. i: Theory, Journal of
Engineering Mechanics 125 (9) (1999) 994–1001.
[13] P. Ceresa, L. Petrini, R. Pinho, R. Sousa, A fibre flexurecshear model for seismic analysis of rc-framed structures, Eart-
hquake Engineering & Structural Dynamics 38 (5) (2009) 565C586.
[14] R. S. B. Stramandinoli, H. L. L. Rovere, Fe model for nonlinear analysis of reinforced concrete beams considering shear
deformation, Engineering Structures 35 (2012) 244–253.
[15] T. R. S. Mullapudi, A. S. Ayoub, Analysis of reinforced concrete columns subjected to combined axial, flexure, shear and
torsional loads, Journal of Structural Engineering 139 (4) (2013) 561–573.
[16] X. Long, J. Q. Bao, K. H. Tan, C. K. Lee, Numerical simulation of reinforced concrete beam/column failure considering
normal-shear stress interaction, Engineering Structures 74 (2014) 32–43.
[17] F. J. Vecchio, M. P. Collins, The modified compression-field theory for reinforced concrete elements subjected to shear,
in: ACI Journal Proceedings, Vol. 83, pp. 219–231.
[18] F. J. Vecchio, Disturbed stress field model for reinforced concrete: Formulation, Journal of Structural Engineering 126 (9)
(2000) 1070–1077.
[19] T. T. Hsu, Softened truss model theory for shear and torsion, ACI Structural Journal 85 (6).
[20] X. B. Pang, T. T. C. Hsu, Fixed angle softened truss model for reinforced concrete, Aci Structural Journal 93 (2) (1996)
197–207.
[21] T. T. Hsu, R. R. Zhu, Softened membrane model for reinforced concrete elements in shear, ACI Structural Journal 99 (4).
[22] Y. L. Mo, J. Zhong, T. T. C. Hsu, Seismic simulation of rc wall-type structures, Engineering Structures 30 (11) (2008)
3167–3175.
[23] J. Ju, On energy-based coupled elastoplastic damage theories: constitutive modeling and computational aspects, Interna-
tional Journal of Solids and structures 25 (7) (1989) 803–833.
[24] J. Lee, G. L. Fenves, Plastic-damage model for cyclic loading of concrete structures, Journal of engineering mechanics
124 (8) (1998) 892–900.
[25] R. Faria, J. Oliver, M. Cervera, A strain-based plastic viscous-damage model for massive concrete structures, International
Journal of Solids and Structures 35 (14) (1998) 1533–1558.
[26] J. Y. Wu, J. Li, R. Faria, An energy release rate-based plastic-damage model for concrete, International Journal of Solids
and Structures 43 (3) (2006) 583–612.
[27] A. Saritas, Mixed formulation frame element for shear critical steel and reinforced concrete structural members, Phd
thesis, University of California, Berkeley, CA, USA (2006).
[28] L. Tesser, F. Filippou, D. Talledo, R. Scotta, R. Vitaliani, Nonlinear analysis of r/c panels by a two parameter con-
crete damage model, in: III ECCOMAS Thematic Conference on Computational Methods in Structural Dynamics and
Earthquake Engineering, 2011, pp. 25–28.
[29] X. Ren, S. Zeng, J. Li, A rate-dependent stochastic damage–plasticity model for quasi-brittle materials, Computational
Mechanics 55 (2) (2015) 267–285.
[30] J. Coleman, E. Spacone, Localization issues in force-based frame elements, Journal of Structural Engineering 127 (11)
(2001) 1257–1265.
[31] A. R. Khaloo, S. Tariverdilo, Localization analysis of reinforced concrete members with softening behavior, Journal of
Structural Engineering 128 (9) (2002) 1148–1157.

21
[32] H. R. Valipour, S. J. Foster, Nonlocal damage formulation for a flexibility-based frame element, Journal of Structural
Engineering 135 (10) (2009) 1213–1221.
[33] D. Feng, X. Ren, J. Li, Implicit gradient delocalization method for force-based frame element, Journal of Structural
Engineering (2015) 04015122.
[34] M. Scott, O. Hamutçuoğlu, Numerically consistent regularization of force-based frame elements, International Journal for
Numerical Methods in Engineering 76 (10) (2008) 1612–1631.
[35] J. Almeida, S. Das, R. Pinho, Adaptive force-based frame element for regularized softening response, Computers &
Structures 102 (2012) 1–13.
[36] J. Y. Wu, Damage energy release rate-based elastoplastic damage constitutive model for concrete and its application to
nonlinear analysis of structures, Phd thesis, Tongji University, Shanghai (2004).
[37] J. Li, X. Ren, Stochastic damage model for concrete based on energy equivalent strain, International Journal of Solids
and Structures 46 (11) (2009) 2407–2419.
[38] D. Feng, J. Li, Stochastic nonlinear behavior of reinforced concrete frames. ii: Numerical simulation, Journal of Structural
Engineering 142 (3).
[39] D. Feng, C. Kolay, J. M. Ricles, J. Li, Collapse simulation of reinforced concrete frame structures, Structural Design of
Tall & Special Buildings 25 (12) (2016) 578–601.
[40] L. Berto, A. Saetta, R. Scotta, D. Talledo, A coupled damage model for rc structures: Proposal for a frost deterioration
model and enhancement of mixed tension domain, Construction & Building Materials 65 (9) (2014) 310–320.
[41] J.-Y. Wu, New enriched finite elements with softening plastic hinges for the modeling of localized failure in beams,
Computers & Structures 128 (2013) 203–218.
[42] D. Ehrlich, F. Armero, Finite element methods for the analysis of softening plastic hinges in beams and frames, Compu-
tational Mechanics 35 (4) (2005) 237–264.
[43] M. Menegotto, Method of analysis for cyclically loaded r. c. plane frames including changes in geometry and non-elastic
behavior of elements under combined normal force and bending, in: Proc.of IABSE Symposium on Resistance and Ultimate
Deformability of Structures Acted on by Well Defined Repeated Loads, 1973, pp. 15–22.
[44] Hibbett, Karlsson, Sorensen, ABAQUS/standard: User’s Manual, Vol. 1, Hibbitt, Karlsson & Sorensen, 1998.
[45] B. Bresler, A. C. Scordelis, Shear strength of reinforced concrete beams, in: ACI Journal Proceedings, Vol. 60, ACI, 1963.
[46] T. Arakawa, Y. Arai, M. Mizoguchi, M. Yoshida, Shear resisting behavior of short reinforced concrete columns under
biaxial bending-shear, Transactions of the Japan Concrete Institute 11 (1989) 317–324.
[47] H. Imai, Yamamoto, A study on causes of earthquake damage of izumi high school due to miyagi-ken-oki earthquake in
1978, Transactions of the Japan Concrete Institute 6 (1986) 405–418.
[48] B. R. Institute, A list of experimental results on deformation ability of reinforced concrete columns under large deflection
(No. 3), no. 21, Ministry of Construction, Japan, 1978.
[49] A. Lynn, Seismic evaluation of existing reinforced concrete building colums, Phd thesis, University of California, Berkeley,
CA, USA (1999).
[50] D. E. Lehman, J. P. Moehle, Seismic performance of well-confined concrete bridge columns, Pacific Earthquake Engineering
Research Center, 2000.
[51] M. Saatcioglu, M. Grira, Confinement of reinforced concrete columns with welded reinforcement grids, ACI Structural
Journal 96 (1) (1999) 29–39.
[52] J. H. Thomsen, J. W. Wallace, Displacement-based design of slender reinforced concrete structural walls-experimental
verification, Journal of Structural Engineering 130 (4) (2004) 618–630.

22
 1 8 0 1 8 0

1 2 0 1 2 0

6 0 6 0
L o a d (k N )

L o a d (k N )
0 0

-6 0 -6 0

E x p . d a ta E x p . d a ta
-1 2 0 -1 2 0
S im u la tio n S im u la tio n

-1 8 0 -1 8 0
-4 -3 -2 -1 0 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4
D ia p la c e m e n t (m m ) D ia p la c e m e n t (m m )
(a) Col S 1 Euler-Bernoulli (b) Col S 1 Timoshenko

6 0 0 6 0 0

4 0 0 4 0 0

2 0 0 2 0 0
L o a d (k N )

L o a d (k N )

0 0

-2 0 0 -2 0 0

E x p . d a ta E x p . d a ta
-4 0 0 -4 0 0
S im u la tio n S im u la tio n

-6 0 0 -6 0 0
-2 0 -1 5 -1 0 -5 0 5 1 0 1 5 2 0 -2 0 -1 5 -1 0 -5 0 5 1 0 1 5 2 0
D ia p la c e m e n t (m m ) D ia p la c e m e n t (m m )
(c) Col S 2 Euler-Bernoulli (d) Col S 2 Timoshenko

1 5 0 1 5 0

1 0 0 1 0 0

5 0 5 0
L o a d (k N )

L o a d (k N )

0 0

-5 0 -5 0

E x p . d a ta E x p . d a ta
-1 0 0 -1 0 0
S im u la tio n S im u la tio n

-1 5 0 -1 5 0
-3 0 -2 0 -1 0 0 1 0 2 0 3 0 -3 0 -2 0 -1 0 0 1 0 2 0 3 0
D ia p la c e m e n t (m m ) D ia p la c e m e n t (m m )
(e) Col FS 1 Euler-Bernoulli (f) Col FS 1 Timoshenko

23
 3 0 0 3 0 0

2 0 0 2 0 0

1 0 0 1 0 0
L o a d (k N )

L o a d (k N )
0 0

-1 0 0 -1 0 0

E x p . d a ta E x p . d a ta
-2 0 0 -2 0 0
S im u la tio n S im u la tio n

-3 0 0 -3 0 0
-5 0 -4 0 -3 0 -2 0 -1 0 0 1 0 2 0 3 0 4 0 5 0 -5 0 -4 0 -3 0 -2 0 -1 0 0 1 0 2 0 3 0 4 0 5 0
D ia p la c e m e n t (m m ) D ia p la c e m e n t (m m )
(g) Col FS 2 Euler-Bernoulli (h) Col FS 2 Timoshenko

6 0 0 6 0 0

4 0 0 4 0 0

2 0 0 2 0 0
L o a d (k N )

L o a d (k N )

0 0

-2 0 0 -2 0 0

E x p . d a ta E x p . d a ta
-4 0 0 -4 0 0
S im u la tio n S im u la tio n

-6 0 0 -6 0 0
-2 0 0 -1 5 0 -1 0 0 -5 0 0 5 0 1 0 0 1 5 0 2 0 0 -2 0 0 -1 5 0 -1 0 0 -5 0 0 5 0 1 0 0 1 5 0 2 0 0
D ia p la c e m e n t (m m ) D ia p la c e m e n t (m m )
(i) Col F 1 Euler-Bernoulli (j) Col F 1 Timoshenko

2 5 0 2 5 0

2 0 0 2 0 0

1 5 0 1 5 0

1 0 0 1 0 0

5 0 5 0
L o a d (k N )

L o a d (k N )

0 0

-5 0 -5 0

-1 0 0 -1 0 0

-1 5 0 E x p . d a ta -1 5 0 E x p . d a ta
S im u la tio n S im u la tio n
-2 0 0 -2 0 0

-2 5 0 -2 5 0
-1 5 0 -1 2 0 -9 0 -6 0 -3 0 0 3 0 6 0 9 0 1 2 0 1 5 0 -1 5 0 -1 2 0 -9 0 -6 0 -3 0 0 3 0 6 0 9 0 1 2 0 1 5 0
D ia p la c e m e n t (m m ) D ia p la c e m e n t (m m )
(k) Col F 2 Euler-Bernoulli (l) Col F 2 Timoshenko

Figure 11: Comparison of the numerical and experimental top displacement-lateral load curves

24
 1 5 0

1 0 0

5 0
L o a d (k N )

-5 0

3 e le m e n t
-1 0 0 5 e le m e n t
1 0 e le m e n t
-1 5 0
-3 -2 -1 0 1 2 3
D ia p la c e m e n t (m m )

Figure 12: Numerical results by different finite element meshes

2 0 0

1 5 0

1 0 0

5 0
L o a d (k N )

-5 0

-1 0 0
E x p . d a ta
-1 5 0 S im u la tio n

-2 0 0
-1 0 0 -8 0 -6 0 -4 0 -2 0 0 2 0 4 0 6 0 8 0 1 0 0
D ia p la c e m e n t (m m )

Figure 13: Comparison of the experimental and numerical top displacement-lateral load curves of RW2

25
 0 .0 4

0 .0 3
D r ift 2 .0 %

0 .0 2 D r ift 1 .5 %
V e rtic a l s tra in

0 .0 1
D r ift 1 .0 %

0 .0 0

E x p . d a ta
- 0 .0 1 S im u la tio n

0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 1 2 0 0
S e c tio n w id th (m m )

Figure 14: Vertical strain distributions at different drift levels along the section width

26

View publication stats