Construction and Building Materials 156 (2017) 761–772

Contents lists available at ScienceDirect

Construction and Building Materials

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

Stiffness matrix for the analysis and design of partial-interaction

composite beams
Jian-Ping Lin a, Guannan Wang b,⇑, Guangjian Bao c, Rongqiao Xu d
a
College of Civil Engineering, Huaqiao University, Xiamen, Fujian Province 361021, China
b
Department of Mechanical Engineering, Texas Tech University, Lubbock, TX 79409, USA
c
Transportation Bureau, Tonglu County, Hangzhou 311500, China
d
Department of Civil Engineering, Zhejiang University, 866 Yuhangtang Road, Hangzhou 310058, China

highlights

● Stiffness matrix is derived for partial-interaction Timoshenko composite beam.

● Explicit expressions will benefit structural calculations and designs.
● Explicitly expressed stiffness matrix has easy connection with commercial packages.
● The stiffness matrix is validated through numerical results.
● Particle swarm optimization is adopted for shear connector distribution.

articleinfo
abstract
Article history:
Received 15 March 2017 Compared to the classical Rayleigh-Ritz method and other analytical solutions, finite element (FE)
Received in revised form 18 July 2017 method is more efficient and capable in calculating the deformations and stress states of partial-
Accepted 29 August 2017 interaction composite beams, as well as manipulating the material and geometrical parameters for better
engineering designs. Stiffness matrix of composite beams considering the interlayer slips is derived
based on the kinematic assumptions of the Timoshenko’s beam theory by taking into account of the
Keywords: transverse shear deformations. A detailed derivation is elaborated to obtain the local stiffness matrix for
Composite beam a composite beam element, while the higher-order interpolation functions are adopted for the
Partial interaction displacement fields (deflection, rotation, and interlayer slip). Then a finite element program is
Finite element analysis developed by assembling the local stiffness matrices and applying corresponding equivalent nodal
Explicitly-expressed stiffness matrix
stresses. Several numerical results are presented and compared against the analytical solutions available
Interfacial slip
in the literature to demon- strate the accuracy of the proposed FE stiffness matrix. Finally, a design
Shear deformation
Particle swarm optimization procedure by connecting particle swarm optimization technique with the present FE analysis is created
to reduce the deformations of sim- ply supported composite beams while the quantity of shear
connectors remains the same, to prove the superior simulation capacity and efficiency of the derived FE
stiffness matrix with other techniques. Compared to the analytical methods, the proposed finite element
is more convenient and applicable in the analysis of partial-interaction composite beams under more
complicated loading and boundary con- ditions. In the meantime, the explicitly expressed local stiffness
matrix can be easily implemented into other commercial software packages as a subroutine for both
professional and personal engineering designs and calculations.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction
Steel-concrete composites combine the best attributes of their
component materials. In this case, the high tensile strength and
shear efficiency of steel and the compressive strength of concrete
⇑ Corresponding author.
E-mail address: guannan.wang@ttu.edu (G. Wang).
can be fully utilized. Experimental studies [1] show that interfacial
slip between the concrete slab and the steel girder occurs even at
low-level load, due to the finite rigidity of shear connectors. This
phenomenon is called ‘‘partial interaction”. Analytical studies have
been carried out to investigate the static and dynamic behavior of
composite beams with partial interactions based on Euler-
Bernoulli beam theory coupled with interlayer slips [2–7]. In order
to take into account of the effects of shear deformations, Xu and

http://dx.doi.org/10.1016/j.conbuildmat.2017.08.154
0950-0618/© 2017 Elsevier Ltd. All rights reserved.
762 J.-P. Lin et al. / Construction and Building Materials 156 (2017) 761–772

Wu [8] used Timoshenko’s beam theory together with considera- slips [30–32]. According to the authors’ experience, numerical
tion of interlayer slip to investigate the static, dynamic, and buck-
ling behavior of partial interaction composite beams. Then Xu and
Wang [9] formulated the minimum potential and complementary
energy principles for composite beams, as well as the variational
principles for the free vibration and buckling phenomenon, and
proposed approximate solutions under different boundary condi-
tions. Schnabl et al. [10] also gave the analytical solutions of a two-
layer beam by taking into account of shear deformation. In
addition, Xu and Wang [11] derived the relations of solutions of
partial-interaction composite beams between Timoshenko and
Euler-Bernoulli theories based on the constitutive equations.
Numerical methods, mainly based on the finite element
method, have also been developed to investigate the partial-
interaction composite beams [12–15]. What is more important,
Thompson et al. [16] developed a finite element method for the
analysis of layered wood systems. Ayoub and Filippou [17] derived
an inelastic beam element from a two-field mixed formulation
with independent approximation of internal forces and transverse
displacements, and the partial interaction was accounted for by an
interface model with distributed force transfer characteristics.
Faella et al. [18] presented a displacement-based finite element
model in which the stiffness matrix and the fixed-end nodal force
vector are directly derived from the ‘‘exact” solution of Newmark’s
differential equation. Dall’Asta and Zona [19] proposed a three field
mixed finite element for the non-linear analysis of composite beam
with deformable shear connection. Cˇ as et al. [20] presented a
new
finite element formulation by employing a modified principle of
virtual work for the non-linear analysis of two-layer composite
planar frames with an interlayer slip. In the finite element formu-
lation the unknown functions are axial extensional strains of each
layer and the pseudocurvature of the reference axis of the compos-
ite beam. Ranzi et al. [21] proposed an alternate formulation based
on a direct stiffness approach that utilizes an internal solution for
the slip in the same way as in Newmark et al.’s work [6]. Ranzi and
Bradford [22] presented a stiffness formulation which is based on
the direct stiffness method for the analysis of composite steel–con-
crete beam-columns with partial shear interaction. Ranzi and Zona
[23] proposed an analytical model which was obtained by
coupling the Euler–Bernoulli assumption for the reinforced
concrete slab to the Timoshenko theory for the steel girder, for
the analysis of steel–concrete composite beams with partial
interaction including the shear deformability of the steel
component. Valipour and Brad- ford [24] derived a flexibility-
based element in the framework of the total secant approach for
one dimensional composite elements with partial shear
interaction. Jiang et al. [25] developed a two- node linear
composite beam element for the steel–concrete com- posite
beam with discrete shear connection. The element is derived by
using the total potential energy method based on the
Timoshenko beam theory and the linear Lagrangian
interpolation function. Nguyen et al. [26] derived the exact
stiffness matrix for a two-layer Timoshenko beam element with
partial interaction basically inspired by the direct stiffness
method. Martinelli et al.
[27] presented closed-form analytical formulation of the stiffness
matrix and the vector of equivalent nodal forces for analyzing
shear-flexible steel-concrete composite beams in partial interac-
tion. Brighenti and Bottoli [28] presented a finite element for the
analysis of composite cross-section beams. The stiffness matrix of
the finite element was obtained by using the direct stiffness
method based on the theoretical solution of the problem provided
in the literature. Taig and Ranzi [29] proposed a partial interaction
formulation based on the generalized beam theory to study the
partial shear interaction behavior of composite steel–concrete
members. Other methods for the simulations of the partial interac-
tion were using the interface elements to model the interfacial
techniques (mainly FE method) are more applicable in broader
engineering problems by involving efficient matrix implementa-
tions and avoiding tedious derivations.
Considered as an extension of the work by Xu and Wang [9],
the current research employs Timoshenko’s beam theory
together with the consideration of the interlayer slip to derive
the stiffness matrix of partial interaction composite beams.
The present deriva- tion distinguishes itself from other work in
that it provides explicit expressions of the stiffness matrix of
composite beams, which has not likely been provided
previously. The stiffness matrix presented benefits not only
professionals but also inexperienced designers or engineers for
both qualitative learning and quantitative designs, without
going through the process of physical inspirations and
mathematical derivations. In addition, the explicit expressions
can also be easily implemented into ABAQUS or ANSYS user
subrou- tines as composite beam elements for more
sophisticated struc- tural analysis instead of building the
elements from scratch. Finally, the explicit expressions can
significantly improve the com- putational efficiency of the
simulations, especially when it involves large-scale calculations
or optimizations.
The organization of this work follows as: Section 2 provides
the basic assumptions and formulations for the partial-
interaction composite structures. The stiffness matrix is
derived following the standard procedure based on the
principle of minimum poten- tial energy. The higher-order
interpolation functions for the dis- placements are assumed.
Detailed derivation of the stiffness matrix is provided in
Section 3, and explicit expressions of stiffness matrix elements
are presented in the Appendix A. A finite element program is
then implemented based on the derivations. In Sec- tion 4,
numerical results are also presented to illustrate the accu- racy
and efficiency of the proposed finite element method in the
analysis of partial-interaction composite beams under
complicated loading and boundary conditions. A new design
procedure is intro- duced in Section 5 by employing particle
swarm optimization tech- nique connected with the FE
program, aiming at reducing the deformations of composite
beams and showing the advantage of the present FE stiffness
matrix. Section 6 concludes this work.

2. Structural model formulation

2.1. Basic description and assumptions

The partial-interaction composite beam is composed of two

sub-structures with different materials (steel-concrete) in the
x-z

Fig. 1. Composite members and coordinate system: (a) elevation; (b) cross
section.
 J.-P. Lin et al. / Construction and Building Materials 156 (2017) 761–772 763

plane, as shown in Fig. 1. The x axis is constructed coinciding with dM

the centroid of the whole cross-section. It can be noticed that the ¼Qþm ð2Þ
dx
Ei, Gi, Ii, Ai, and qi ði ¼ 1; 2Þ stand for Young’s modulus, shear
dQ
mod- ulus, moment of inertia, cross-sectional area and the mass
¼ —q ð3Þ
density of the sub-structures, respectively, while L and H denote dx
the length In addition, in accordance with the equilibrium of forces in the
and height of the composite beam. h1 and h2 are the distances of x-direction, it can be obtained that
centroids between each sub-structure and the whole section, and
N 1 þ N2 ¼ 0 ð4Þ
here is the relation h = h1 + h2.
Several assumptions need to be established to keep the consis-
tency of this work for partial-interaction composite beams with
dN 1 d ð5Þ
small deformations, as already been repeated several times in the Q s ¼— N2
¼
literature [8,9]: dx dx
Also, the bending moment of the whole element can be
(1) All of the constitutive materials considered have linear expressed as
behavior and small deformations;
M ¼ M1 þ M2 — N 1 h ð6Þ
(2) The shear connection is continuous along the x directions;
(3) The longitudinal shear force between upper and lower sub- in which Ni and Mi ði ¼ 1; 2 are the axial forces and bending
structures is proportional to the interlayer slip [33,34]; moments of sub-structures.
(4) No transverse separation (opening) happens on the interface Constitutive relationship: As assumed in Section 2.1, the inter-
between sub-structures, which means the curvature is same layer distributing force has a linear relationship with the rigidity of
at any cross-sections [35,36]; the shear connectors:
(5) Timoshenko’s beam theory is employed for the composite
Q s ¼ ks us ð7Þ
beams, which means that the transverse shear deformations
of the cross section are allowed and are identical in the two Based on Timoshenko’s beam theory, the relationship between
sub-structures. the stress resultants and deformations are
dw dw
2.2. Theoretical M1 ¼ —E1I1 ; M2 ¼ —E2I2 ð8Þ
framework
d d
dw x x
The kinematical relationship, equilibrium equations and consti- Q ¼ C. — wΣ ð9Þ
tutive relationship have already been presented in details by Xu d
and Wang [9] based on Timoshenko’s beam theory assumption, x
where C is the shear rigidity of the composite beam, and
which are not totally repeated in this section. Only the necessary
equations are borrowed here to clarify the model establishment. C ¼ j1 G1 A1 þ j2 G2 A2 ð10Þ
Readers can refer to Xu and Wang [9] for a detailed description. in which j1 and j2 are the shear correctors, whose values depend
Kinematical relationship: the relationship between interlayer on the shape of the cross section of the sub-structures according
slip us and rotation angle ðwÞ of the cross section is to the Timoshenko’s beam theory.
us ¼ u2 — u1 þ wh ð1Þ The axial forces of each sub-structure can be expressed as the
function of the corresponding axial displacements
in which u1 and u2 are the longitudinal displacement of the upper
and lower sub-structures, respectively, as shown in Fig. 2. d u1 d u2
Fig. 3 shows a free-body diagram of an infinitesimal element N1 ¼ ; N2 ¼ 2
ð11Þ
E1A
1
dx E2A dx
with the length of dx, M, Q denote the bending moment and shear
force, while q and m indicate distributed load and bending
2.3. Principle of minimum potential energy
moment. Thus, according to the equilibrium in the z-direction,
we have
The potential energy of a partial-interaction composite beam
can be expressed as
p ¼ p1 þ p2 þ p3 — W q ð12Þ
which includes four components:

Fig. 2. Geometrical relationship of us, w and u. Fig. 3. An infinitesimal element of the composite beam.
764 J.-P. Lin et al. / Construction and Building Materials 156 (2017) 761–772

ZL
1 . Σ2 ZL . ZL .
dw 1 1 1 1
p1 ¼ 0 1Σ2 2Σ2 a x x b x x 20
2 d du du
¼ ðj— Þ ¼ ð — jÞ Þ
0 0
dx dx d
2 d 2 d l ; l
þ þ x
ð13Þ The potential energy of the eth element can be obtained from
Eqs (12)–(17) from Section 2.3:
p2 ¼ Z L . Σ2
d 1 1
1 C
w — w dx ð14Þ p Z xj Σ EI u 2
C wu 2
qw muΣdx
0 0
¼ i
0 2 dx e 0ð Þþ 0ð — Þ — —
ZL þx1 Z2 xj EA
ð 0
—
u u
2 Þ2 þ 1
h dx Z xj
2
s s
k u dx ð 21Þ
1 2 x s 2 x
p 3 ks u2 dx ð15Þ
¼ 2 i i
0

W ZL As is well known, the following expression can be derived after

qwq ¼ dx þð Q¯ w— N¯1 u the integration of Eq. (21), no matter which type of interpolation
0 þ mwÞ w sÞj0 16 Þ
function is adopted in the calculation.
ð — L

M¯

in which EI0 ¼ E1 I1 þ E2 I2 , and M¯ , Q¯ and N¯ 1 are the external

forces 1
applied on the boundaries. pe¼ u KTeu —
e
eF u e ð22Þ
2
The four components in Eqs. (13)–(16) indicate strain energy of where ue ¼ ½wi ; ui ; usi ; wj ; uj ; usj ] , and K e , F e are usually called as
T

bending ðp1 Þ, the strain energy of shearing ðp2 Þ, the strain energy
of shear connection ðp3 Þ, and the work done by external forces stiffness matrix, generalized load vector. What should be noticed
ðWqÞ, respectively. is that the accuracy and stability of stiffness matrix directly depends
Substituting the kinematical relationship Eq. (1), equilibrium on the forms of the interpolation functions for the displacement
equation Eq. (4), as well as constitutive relationship Eq. (11) into fields. Higher-order functions for the displacements are then
Eq. (13), the strain energy of bending can be expressed in a simpler assumed:
pattern:
wðxÞ¼ wia þ wjb þ n1ab þ n2ða2 b — ab2Þ
ZL . Σ2 Z L Σ uðxÞ¼ ui a þ uj b þ g1 ð23Þ
p1 ¼ 1 dw d 1
2 d ab
dx EA ðus — dx ð17Þ
Σ2 2 dx
0 þ 0 u s ðxÞ¼ usia þ usjb þ g2ab
whÞ
where EA ¼ EE1 AA1 ×E2 A2
. The simplification makes the finite element As seen above, a three-order interpolation is employed for
þE A1 1 2 2

analysis more easily, and benefits the following derivations. deflection, while two-order functions are used for the rotary dis-
placement and interlayer slip. However, four new unknown coeffi-
3. Finite element analysis of composite beams cients n1; n2; g1; g2 are also introduced in the function. These
coefficients, however, just affect the deformation of the local ele-
Compared to the classical Rayleigh-Ritz method and other ment rather than the external nodes or other elements, leading
approximate analytical solutions [7,9], finite element method is a to their notations as ‘‘internal degrees of freedom (dofs)”. The val-
more accurate and efficient technique in calculating deflections, ues of the internal dofs n1; n2; g1; g2 should also be determined
rotary displacements, and interlayer slips of partial-interaction using the principle of minimum potential energy in the local ele-
composite beams. ment. Thus, Eq. (21) for local element can be further simplified
First of all, the as follows to calculate the internal dofs:
small elements, eachcomposite
of them is beam can as
indicated be 1,divided
2, 3,.. .,into several
N, respec-

Z1 xj EA0 u2 u1h
Z 1xj Σ 1 2Σ
p0e dx0 Z xj k u dx 2
24
tively. Fig. 4 shows the eth element, in which the coordinates of EI0ðu Þ þ Cðw — þ d
the starting point and the ending point are xi and xj in x- x 2 uÞ 2 x
¼ 2
i 2
ðÞ
ð 0s — 0 Þ ss
direction. Thus the length of the element can be expressed as
le ¼ xj — xi ð18Þ þ
Substituting
xi
Eq. (23) into the new energy expression Eq. (24)
xi
Here we assume x can be the coordinate of any point within the 2
element, leading to the non-dimensionalized coordinate as makes
p T ! 25
ae 1 e
K L
T

0
1 1 e ¼ ðu0e Þ u0e ð Þ
¼ ðx — xÞ; x — xj Þ
b ð19Þ 2 L H
e ¼
whichl can also be called
e e ð
the shape function of the local element.
el can be ignored for the simplification of the
T
Then the subscript in which u0e ¼ ½wi ; ui ; us ; wj ; uj ; su ;i n ;i g j; n j; g ] ð26Þ
i j
notations: In addition, the stiffness matrix Ke and the other two compo-
nents L, H of the matrix are
0
0 l
C
2
C
l 2
C
1
. Σ 0 —C . Σ
C
D¯ 1 D¯
B 1þ — EAh —C —1 þ 1 EAh
C
2 l 3c l 2 l 6c l
 l .l Σ
3c2 l .l Σ
6c

B 0 — EAh EA
1þ 1
0 C
EAh EA
—1 þ 1 2
C
— —
0 0
Ke ¼ B
C. 2
Σ
l
. 2
Σ
C
D¯ C D¯ C
B—
B —1 þ 1 EAh
—C 1þ 1
— EAh C
l @C . Σ . Σ A
B 2
0l EAh 6c EA
—1l þ 1
02 l 3c
— EAh EA
1 þl 1 C
1 1
Fig. 4. The coordinates of the eth element. l l 6c2 l l 3c2 ð27Þ

0 J.-P. Lin et al. / Construction and Building Materials 156 (2017) 761–772 765
0—C
C 1
6 0 0 C
0 The explicit expressions of the symmetric local stiffness matrix
6

06
L 0—C ð28 elements cantobethe
refereed
6 ¼ B
0 0
Cl 0 0
Cl
12
00 C
Þ corresponds linear in the in
terms Appendix A. Theenergy
the potential generalized load
expression
12
12
12
@0 0 ks l
0 0 ksl A in Eq. (21),Z namely
T
0 C ðF e Þ ue ¼ xj ðqw þ muÞdx ð35Þ
3
0 0 0 x
1 l
i

B 0 1þ10r30Cl C 1 — EAh C Submitting the new interpolation functions Eq. (33) into Eq.
30 3
¼
H 0 C C l C ð29Þ (35), the generalized load Fe can also be obtained.
B@ 30 5l
0 A The global stiffness matrix and global load vector of the com-
0 EA
h
3l 0 1þ10c2 kl30 posite beams can be organized as following

where D ¼ EI0 þ h2EA, ¼ Cl2 , and c2 EA

2 K
X¼ N F
X¼ N Fe ð36Þ
¼ ksl
c1 D¯ Ke
. ;
As already discussed earlier, the internal dofs n1; n2; g1; g2
e¼1 e¼1
should minimize the potential energy p0e in local element, thus
@ p0e @ p0e @ p0 @ p0e which leads to the total potential energy as
e

1 ¼ 0; 2 ¼ 0; ¼ 0; ¼0 ð30Þ T
@n @n @1 g @2g p ¼ 1 uT Ku — F u ð37Þ
which can be re-expressed as 2
T T
L½w i; ui ; us ; wj ; uj ; su ] þ H½n
1 ;1g ;2 n 2; g ] ¼ 0 ð31Þ
i j The new equations between node displacements and stresses
from which the relationship between internal dofs and the displace- can be obtained by still employing the principle of minimum
ments on the nodes of the local element is obtained as potential energy:
K·u¼F ð38Þ
ui ¼ —H—1L ·
ue in which
u wu u w u
u
T
, and
0 1 ¼ ½ 1; 1; s1; ... ; Nþ1 ; Nþ1 ; sðNþ1Þ ]
ð1—
0 2
0 0 — ð1—2 lÞl 0 F ¼ ½Q 1 ; M 1 ; N1ð1Þ ; . . . ; Q N 1 ; M Nþ1 ;
l
BÞl C T þ
ð1þ10c2 1
2
N1ðNþ1Þ ] .
ð1þ10c2
Þl
Þl 2
ð1þ10c 2 ð1þ10c2 C
Þl Þl 2
1— C— C
1 B h2 c2 c3
20 h2 c2 c3
40
— h2 c2 c3
40 4c2 h
4c2 h A
B— 20c2c3h2
3

¼ 1—l
@
B
ð1þ10c2 Þl 2
ð1þ10c2 Þl3 l ð1þ10c2 Þl — ð1þ10c2
Þl 2 ·
— uel 4. Validation and applications
120c2 c3 240c2 c3 24c2 120c2 c3 240c2 c3 24c2 h Based on the stiffness matrix derived above, a finite element
h2
l
h2 h h2 h2 ð1þ12c1 Þl program was implemented using MATLAB. Partial interaction com-
— 2cl h
2

2c3 h
l2
4c3 h
ð1þ12c1
3
l2
4c3 h 48c2 c3 h2 posite beams with different shape of cross sections are analyzed
Þl
T
ð32Þ
48c2 c3 h2
where
c u ¼ ½n 1 ; g1 ; n2 ; g2 ] , and
i
under different types of boundary conditions. The FE simulation
, Cl ¼ ð1þ12c1 Þð1þ10c2 Þl2 .
3 EA
¼
120c2 c3
h
Then the assumed displacement expressions in Eq. (23) can be
2
results are validated against the corresponding analytical formu-
modified using the relationship Eq. (32) as lae, and further examples are employed to prove the capabilities
of the present technique.
l
wðxÞ ¼ ½a — mabða — bÞ]wi þ ½ab — mabða — bÞ] i w
2 4.1. Specimen of Girhammar and Gopu [4]
— pabða — bÞusi þ ½b þ mabða — bÞ]wj
l First of all, a single-span composite beam with the length of
þ ½—ab — mabða — bÞ] wj — pabða — bÞusj wðxÞ L = 4 m is investigated under different boundary conditions. The
wi 2 usi wj
cross-section of the beam shown in Fig. 5 is the same as that of
¼ 6mab þ ða þ 3mabÞwi þ 6pab — 6mab Girhammar and Gopu [4], and the main properties are
l u l 12c l
þ ða þ 3mabÞw þ 6pab sj u s ðxÞ¼ 2 pabwi E1 = 12 GPa, E2 = 8 GPa, ks = 50 MPa, h1 = 0.05 m, h2 = 0.15 m and
j
l c3 Poisson’s ratios are supposed to be m1¼ 0:2 and m2 0:3, respec-
12c2 l 12c2 tively. In addition, the width of the slab and girder are

þ 3
c
 12c2 pab wi þ ða þ nabÞusi — 3 pabwj b1 = 0.30 m, b2 = 0.05 m. An uniformly distributed load is applied
þ 3 2 c at the top side of the concrete slab with the magnitude of
c l
pab w þ ða þ nabÞusj q0 = 1 kN/m. Not only the simply-supported (SS) boundary condi-
2j tion, but also clamped-free (CF), clamped-simply supported (CS),
ð33Þ
1
and m ¼ ð1þ10c2 Þl2 ð1þ12c1 Þl2 and clamped-clamped (CC) boundary conditions are analyzed.
, n ¼1—1l , p ¼ 1—1l l
,q¼ 1l
.
1—l 120c2 c3 48c2 c3 24c2 1—l 2r3 h The deflection is one of the most significant factors needed to be
2 2 h
h h
The new interpolation functions for displacement fields are the considered in the engineering calculations and designs. It is crucial
exact solutions without any internal loading, from which the inter- in evaluating the capacity of the beam structure. The analytical
nal stresses satisfy the equilibrium equations. Finally, the new expressions of deflections for a Timoshenko’s beam under uni-
functions Eq. (33) can be taken back to the potential energy in formly distributed loading with different boundary conditions
Eq. (21) to give the corresponding local stiffness matrix: can be referred to [8].
0 1 Fig. 6(a)–(d) shows deflections of composite beams using both
k11 k12 k13 k14 k15 k16
present FE analysis and analytical results [8] under four different
k12 k22 k23 k24 k25 k26 boundary conditions: SS, CF, CS, CC. The deflections show different
B C
K ¼ B k1 k23 k33 k34 k35 k36
C ð34Þ patterns under different boundary conditions, and good agree-
3
k14 k24 k34 k44 k45 k46 ments are always satisfied. Different number of elements have
Bk k
@15 25 k35 k45 k55 k56 been adopted to test the convergence of the present method by
validating the mid-span deflections against analytical results. It
C k16 k26 k36 k46 k56 k66
A can be seen from Table 1 that about 50 elements are good
enough
766 J.-P. Lin et al. / Construction and Building Materials 156 (2017) 761–772

Table 1
The mid-span deflections of the single-span composite beam under uniformly
distributed load.

SS (mm) CF (mm) CS (mm) CC (mm)

Analytical solution 7.59 23.62 3.88 2.35
FE analysis (N = 5) 8.88 27.50 4.93 2.99
FE analysis (N = 10) 8.05 24.67 4.31 2.62
FE analysis (N = 20) 7.70 23.89 4.03 2.42
FE analysis (N = 50) 7.60 23.66 3.95 2.36
FE analysis (N = 100) 7.59 23.62 3.94 2.35
FE analysis (N = 200) 7.59 23.62 3.94 2.35

direction are shown in Fig. 7(a)–(d). Similar to the comparisons of

deflections, the FE simulation results again agree well with the
analytical results, which demonstrate the accuracy of the proposed
FE method. For the second time, the convergence of the new solu-
tion is tested by calculating the 1/4 span (x = 1 m in Fig. 7(a)–(d))
interlayer slips, and 50 elements are enough to generate converged
results with three decimal places (Table 2).

4.2. Specimen of Chapman and Balakrishnan [37]

The second scenario is to analyze a simply supported composite

Fig. 5. A simply supported composite beam of concrete and wood.
to generate converged results with three decimal places, and it
takes about 0.08 s (Intel Core i5-4200 M CPU @ 2.50 GHz) for the
execution of the program, due to the explicit stiffness matrix
expressions. The quick execution of the present programs also
facilitates the other use of FE analysis, such as optimization and
design.
The interlayer slip is then evaluated under the same four
boundary conditions, under which the analytical expressions can
be further referred to [3,4]. The interfacial slips along longitudinal
beam with the length of 6.05 m subjected to a concentrated force
loaded at the mid-span, Fig. 8. The cross section is the same as
the specimen Beam E1 tested by Chapman and Balakrishnan at
Imperial College London [37], the thickness and width of the con-
crete slab is 0.152 m and 1.22 m, respectively, and the type of steel
girder is 1200× ×600 44 lb/ft B.S.B. The Young’s moduli of
concrete and steel are 30.5 GPa and 205 GPa, respectively, while the
Pois- son’s ratios are 0.2 and 0.3. The concentrated loading is P
= 60 kN. The analytical expressions of deflections and interlayer
slip of a simply supported Euler-Bernoulli beam with a concentrated
force at the mid-span were obtained by [3,5], which are not
repeated here.

8
70
7 FE analysis Analytical result
6 60
Deflection (mm)

Deflection (mm)

5 50
4 40
FE analysis
3 Analytical result 30
2
20
1
10
0
0 1 2 3 0
4 1 2 3 4
x (m) 0 x (m)
(a) Simply-Supported (SS) (b) Clamped-Free (CF)

5
2.5
4
Deflection (mm).

2.0
Deflection (mm).

3
1.5
FE analysis
2 1.0 Analytical result
FE analysis
1 Analytical 0.5
result

0
0 1 2 3 0.0
4 0 1 2 3 4
x (m) x (m)
(c) Clamped-Simply supported (CS) (d) Clamped - Clamped (CC)
Fig. 6. Deflections along longitudinal direction under four boundary conditions.
 J.-P. Lin et al. / Construction and Building Materials 156 (2017) 761–772 767

0.3 0.4
FE analysis
0.2 0.3
Analytical result
0.1

Slip (mm)
Slip (mm)
0.2
0.0
0.1
-0.1
0.0 FE analysis Analytical result
-0.2
-0.3 -0.1
0 1 2 3 0 1 2 3 4
4 x (m)
x (m)
(a) Simply-Supported (SS) (b) Clamped-Free (CF)

0.3 0.15

0.2 0.10
0.05

Slip (mm)
Slip (mm)

0.1
0.00
0.0
FE analysis - 0.05
-0.1 Analytical result
-0.10
FE analysis Analytical result
-0.2
0 1 2 3 -0.15
4 0 1 2 3 4
x (m) x (m)
(c) Clamped-Simply supported (CS) (d) Clamped-Clamped (CC)

Fig. 7. Interlayer slips along longitudinal direction under four boundary conditions.

Table 2 Fig. 9(a) shows the comparison of deflections between the pre-
The interfacial slips at 1/4 span of the single-span composite beam under point load. sent numerical results and analytical solutions when the shear
connector stiffness equals to 100 MPa. It should noted that the
SS (mm) CF (mm) CS (mm) CC (mm)
shear connector stiffness depends on several aspects, such as
Analytical solution 0.142 0.377 0.181 0.114
dimension and shape of the connectors, and the spacing of their
FE analysis (N = 5) 0.202 0.524 0.237 0.154
FE analysis (N = 10) 0.157 0.414 0.191 0.124 distributions, etc.. ., according to Eurocode 4 [38]. As the analytical
FE analysis (N = 20) 0.146 0.388 0.181 0.117 results based on Euler-Bernoulli beam theory neglect the shear
FE analysis (N = 50) 0.142 0.379 0.176 0.115 deformation, its values are slightly lower than the present numer-
FE analysis (N = 100) 0.142 0.378 0.176 0.114
ical results which include the shear deformation. Comparison
FE analysis (N = 200) 0.142 0.377 0.176 0.114
between the present numerical results and analytical solutions of
interfacial slip distributions is illustrated in Fig. 9(b), which shows
no difference between the two results even if they are under differ-
ent beam assumptions.
In order to learn the influence of the transverse shear assump-
tion on the deformations of composite beams, which has been
ignored by many literature, Table 3 lists the maximum deflections
with different height-to-length ratios (H/L) from 0.05 to 0.25, and
fixed-valued (100 MPa) shear connector stiffness. It shows that
the influence of the shear deformation becomes more evident with
the increase of the height to-length ratios. For this simulation case,
the maximum deflection without shear deformation is 6.40 per-
cent smaller than its value including shear deformation when H/L
equals to 0.25, which means significant underestimates could hap-
pen to high H/L ratio beams if shear deformation is neglected, and
might lead to dangerous real practice design.
Fig. 10 shows the calculated results of the maximum deflections
at the middle point of a simply supported steel-concrete composite
beam as the shear connector stiffness varies, while other material
properties and geometrical characteristics are the same as shown
in Fig. 8. When the shear connector stiffness is small the maximum
deflection can reach 5.5 mm, while its value approaches to almost
2.0 mm when the shear connector stiffness approaches infinite.
Results demonstrate that, when the shear stiffness is smaller
Fig. 8. Geometrical characteristics of a simply supported steel-concrete composite than 1 MPa, the value of deflection will not increase as the shear
beam. stiff- ness decreases, while its value will not decrease when the
shear
768 J.-P. Lin et al. / Construction and Building Materials 156 (2017) 761–772

5
stiffness is larger than 10 GPa. The shear connector stiffness has
a sensitivity range between 1 MPa and 10 GPa, within which the
4 maximum deflection at the middle point of span dramatically
var- ies from 5.5 mm to 2.0 mm.
Deflection (mm)

3 Fig. 11 shows the variations of maximum interfacial slip at the

ends of beam versus the shear connector stiffness. Similar
pattern of curve is illustrated as before. Within the sensitivity
2 range of 1 MPa–10 GPa, the maximum interfacial slip rapidly
FE analysis Analytical result varies from
1 0.68 mm to zero. From Figs. 10 and 11, it should be learnt that
when the shear connector stiffness is below 1 MPa, largest
deflec- tion and interlayer slip are obtained, which means
0
0 1 2 3 4 5 minimum shear interaction between sub-elements is achieved,
x (m) even though they are still attached according to the basic
(a) Deflections assumptions. In the mean- time, when shear connector stiffness
is above 10GPa, almost per- fect connection is gained between
0.6 sub-elements, which can be proved by the fact that zero
FE analysis Analytical result
interlayer slips appear within this range. This phenomenon
0.4 described not only provides guidance of engineering calculations
but also inspires us to reconsider the designs of the shear
0.2 connectors by changing their distributions along the spans of
Slip (mm)

beams, as is shown in Section 5.

0.0
4.3. Continuous composite beam with box section
-0.2
A two span continuous steel-concrete composite beam shown
-0.4 in Fig. 12 is analyzed. The length and width of concrete slab is

-0.6 0.8
0 1 2 3 4 5
x (m)
0.6
(b) Interlayer slips
Slip (mm)

Fig. 9. Deflections (a) and interlayer slips (b) along longitudinal direction of I shape 0.4
girder-concrete slab composite beam. FE analysis Analytical result
0.2

Table 3 0.0
The influence of shear deformation on the maximum deflection.

-0.2
1.E+01 1.E+03 1.E+05 1.E+07 1.E+09 1.E+11 1.E+13
H/L Including shear Without shear Relative Shear connector stiffness ks (Pa)
deformation (mm) deformation (mm) error (%)
Fig. 11. Maximum interfacial slip versus rigidity of shear connectors.
0.05 14.956 14.852 —0.70
0.10 2.564 2.525 —1.51
0.15 0.857 0.834 —2.74
0.20 0.385 0.368 —4.38
0.25 0.206 0.193 —6.40

t=6mm
6 t=5mm
t=4mm
t=8mm
5
FE analysis
Deflection (mm)

Analytical result
4
q0=10 kN/m
3

1
1.E+01 1.E+03 1.E+05 1.E+07 1.E+09 1.E+11 1.E+13 L=9000 L=9000
Shear connector stiffness ks (Pa)

Fig. 10. Maximum deflection versus rigidity of shear connectors. Fig. 12. Geometrical characteristics of the continuous steel-concrete composite
beam (Unit: mm).
 J.-P. Lin et al. / Construction and Building Materials 156 (2017) 761–772 769

18 m and 1.400 m, respectively. The thickness of concrete slab that
is close to the steel web is 100 mm, while its value is 80 mm at the
flange and middle part of concrete slab. The thicknesses of the bot-
tom plate, steel web, stiffener, and top flange of steel girder are
8 mm, 4 mm, 4 mm, and 6 mm, respectively. The Young’s moduli
of concrete and steel are 30.5 GPa and 205 GPa, respectively. Pois-
son’s ratios are supposed to be 0.2 and 0.3 for concrete and steel.
The uniformly distributed loading is q0 = 10 kN/m.
When the shear stiffness is chosen as 100 MPa, comparisons of
deflections and interfacial slips are illustrated in Fig. 13(a)–(b)
between the present numerical results and analytical solutions
based on Euler-Bernoulli beam theory [3]. It is not surprised to
observe that analytical results of deflection are lower than the pre-
sent numerical results (with a small magnitude), which takes the
shear deformation into account. However, the interlayer slips
between sub-structures are well matched between the two results.
3.0
FE analysis
Chen (2012)
2.5
Deflection (mm)
Both Examples in Section 4.2 and 4.3 indicate that the shear defor-
mation assumption has little effect on the interlayer slips. In this
case, the proposed finite element analysis is more efficient than
analytical methods in calculating the structural response of partial-
interaction composite beams under more complicated boundary
and loading conditions, by avoiding complicated struc- tural
analysis and tedious mathematical derivations.
5. A new design of shear connectors for simply
supported composite beams with particle swarm
optimization
During the engineering design and construction, the shear con-
nectors are always uniformly distributed along the longitudinal
direction of composite beam for convenience. This kind of simpli-
fications will inevitably cause that the shear connectors to be over
distributed at certain positions while not adequate at some others.
The uniform distribution of shear connectors without considering
boundary conditions will not only be a waste of construction mate-
rials, but also prevent the normal service life of structures. Jasim
[39] has developed the differential equations governing the behav-

2.0 ior of composite beams, in which the distribution of connectors

along the span is double triangular shaped under simply supported
1.5 (SS) boundary condition while keeping the same quantity as uni-
form distribution, Fig. 14(a)–(b), based on the fact that the inter-
layer slip becomes largest at both end points and turn to zero in
1.0
the middle locations. The exact solution of the deflections derived
demonstrated superiority of the new design which effectively
0.5
decreased the deflections and interlayer slips and utilized the dis-
0.0 tributions of shear connectors.
0 3 6 9 12 15 18 The shear connector stiffness k~dt with double triangular distri-
x (m) s
butions along the length of beam can be expressed as:
.
~dt ¼ 2ks ð1 — 2x=LÞ 06x6
(a) Deflections L=2
k ð39Þ
s
2ks ð2x=L — 1Þ L=2 6 x 6 L
0.3
FE analysis where L denotes the span of beam.
Chen (2012) The influence of the new design with double triangular distribu-
0.2
tion of shear connectors on the interfacial slip can be easily ana-
lyzed with the FE technique. However, in addition to the uniform
0.1 and double triangular patterns, the shear connectors can be effi-
Slip (mm)

ciently designed by adopting a more sophisticated design process

0.0 that is achieved by connecting the present stiffness matrix with
optimization technique, helping to obtain converged optimized
-0.1 parameters that minimize the target function. The target function
for a simply supported composite beam in this section is to mini-
-0.2 mize the interlayer slip at the end points where largest deforma-
tion usually appears:
-0.3
0 3 6 9 12 15 18 Obj ¼ minimize ½us ðx ¼ 0 or LÞ] ð40Þ
x (m) with the constraint condition of keeping the total quantity of shear
(b) Interlayer connectors unchanged with a new quadratic functions defined as
slips (
2
að1 — 2x=LÞ þ bð1 — 2x=LÞþ c 0 6 x 6
Fig. 13. Deflections (a) and interlayer slips (b) along longitudinal direction of k~s opt 2 ð41Þ
continuous beam. L=2 að2x=L — 1Þ þ bð2x=L — 1Þ þ c L=2 6
¼
x6L
Fig. 14. Shear connectors with (a) uniform, (b) double-triangular, and (c) quadratic distributions along the length of composite beam.
770 J.-P. Lin et al. / Construction and Building Materials 156 (2017) 761–772

with the shape of Fig. 14(c), and the area of shadow regions always
keeps same.
Here we adopt particle swarm optimization (PSO) which was
developed by Kennedy and Eberhart in 1995 [40] to imitate the
intelligence and movement of swarms of birds or fish. The tech-
nique has been extensively used in different types of problems,
such as [41–42], proving its stability and efficiency in structural
and material design process.
A brief explanation of the PSO algorithm based on the present
FE calculation is illustrated in Fig. 15, while the details can be
referred to Wang and Pindera [42]. First, the positions of a swarm
of particles (unknown coefficients in Eq. (41)) are considered
within the considered range, aiming at keeping the quantity of
shear connectors same. The position of ith particle is denoted as X i
¼ ðx1 ; x2 ; . . . ; xD Þi , where D stands for the dimension of variables,
(a) Deflections
which is also the designated parameters that need to be optimized.
The velocity of the same particle is V i ¼ ðv 1 ; v 2 ; . . . ; v D Þi . pBest and
gBest indicate the best experience of ith particle and other parti-
cles, respectively, used to update the positions continuously. The
updating algorithms are expressed as

v kþ1
i ¼ v½xv ki þ a1 randk1 ðÞðpBestik — xik Þ þ a2 rand2k ðÞðgBestik — xi k Þ]
d
kþ1
xid ¼ xidk
þ vdkþ1
id
d d d d

ð42Þ

in which superscript k and k þ 1 are the iteration numbers. randk

ðÞ 1

and rand2k ðÞ are random numbers with uniform distributions in

the interval [0,1], and a1; a2 are acceleration constants. The
parameter x is the inertia weight parameter, and the parameter v
introduced to ensure convergence is called constriction factor.
To demonstrate the efficiency of the new design, comparisons
are carried out when the shear connector stiffness is uniformly,
double-triangularly and optimized quadratic distributed. The
dimensions and material properties (except shear connector
stiff- ness) here are the same as that of Girhammar and Gopu [4]
which have already been provided in Section 4.1. The uniformly
dis- tributed load on the simply supported beam is q0 = 1 kN/m.
It should be mentioned that 50 elements are employed to
generate the accurate results, where the values of shear
connector stiffness are assigned as instructed in Eqs. (39) or
(41). 15 particles and 15 maximum iterations are designated for
the optimization. After
15.3 s execution, the converged parameters are generated after
5th iteration (Intel Core i5-4200 M CPU @ 2.50 GHz). The final
opti- mized parameters are ¼ a 3ks,¼b 0, c 0 in Eq. (41).
¼
Fig. 16(a)–(b) show the calculated results of deflections and
interlayer slips before and after the design. Results show that the
double triangular distribution of shear connecters can significantly
decrease the interfacial slips of partial-interaction composite
beam, while the optimized parameters provide even smaller inter-
layer slips for quadratic distributions.
(b) Interlayer slips
Fig. 16. Deflections (a) and interlayer slips (b) along longitudinal direction of the
simply supported beam for uniform, double-triangular, optimized quadratic shear
connector distributions.
In real engineering practice, based on the analysis of the safety
state of structure, the shear connectors of the partial-interaction
composite beam could be properly increased or reduced to realize
the balance between economy and safe capacity, depending on the
boundary conditions. More complicated situations can be designed
through the similar procedure, which will be explained in future
research work.
6. Conclusions
In this work, the local stiffness matrix of composite beams con-
sidering the interfacial slips is derived based on the kinematic
assumptions of Timoshenko’s beam theory. Higher-order interpo-
lation functions are adopted for the displacements in order to
obtain accurate results, and explicit expressions of the stiffness

Fig. 15. PSO algorithm based on FE analysis.

 J.-P. Lin et al. / Construction and Building Materials 156 (2017) 761–772 771

matrix elements are provided in this paper. A finite element pro-

gram is developed based on the derivations. The accuracy of pre- 6EI0 C EA h h EAh
sent numerical method are demonstrated with several numerical
þ ð1 þ mÞp þ .q — 6m Σ.n — 6p Σ þ
k23 ¼ 2 mp 2 6 l l
l k l2 l
ks l2
examples, for both single-span and double-span beam structures þ
s
qn þ
with different boundary conditions. The explicitly expressed stiff- 60 24
ness matrix elements derived in this paper can also be easily
6EI0 2 C 2 EA. h 2 ksl2 2
implemented into commercial software such as ABAQUS, ANSYS k24 ¼ — m — ð1 þ mÞ — qΣ— 6m — q
or for other scientific and engineering designs. The finite element 2
2 6 l 60
program is then connected to a particle swarm optimization tech- l .
nique to create a sophisticated design procedure for the shear con- 3EI0 2 EI0 Cl EAl Σ2 EAh2
2
h
k25 ¼ m — þ ð1 þ mÞ þ q — 6m —
nector distributions, in order to reduce both the interlayer slips
l 3 l 4 12 l l
and deflections of composite beams under simply supported ksl 2
boundary condition, trying to balance the economical practice þ q
and service capacities. The proposed explicitly expressed finite ele- 120
ment model is convenient and applicable in the analysis and . Σ. Σ
6EI0 C EA h h EAh
design of partial-interaction composite beams for both profession- k26 ¼ 2 mp þ ð1 þ mÞp þ q — 6m —
l 2 6 l n — 6p l
als and non-professionals alike, as well as provides another stan-
dard against other analytical and numerical techniques. ksl2 k sl2 l
Acknowledgements þ qn þ q
60 24 . Σ2
12EI0 C EA h EA k sl k sl k sl
2

k33 ¼ l
The first author gratefully acknowledge the support by the
National Natural Science Foundation of China (No. 51608211), 3 p2 þ p2 þ n — 6p þ þ nþ þ n
the Provincial Natural Science Foundation of Fujian Province (No. l C3l l EA l 30
h 3 h6
12EI0 1 þ mÞp — .q — 6m Σ.n — 6p Σ
2017J05083), and the Scientific Research Funds of Huaqiao k34 ¼ —
Univer- mp3 — 3 l l
l l
sity (No. 16BS403). The fourth author is supported by the National k sl ksðl l
Natural Science Foundation of China (No. 51478422 and No. — qn — q
11172266). 30 12
6EI0 C EA h h EAh
Appendix A k35 ¼ þ ð1 þ mÞp þ .q — 6m Σ.n — 6p Σ —
mp
The stiffness matrix elements of partial-interaction composite 2 2 2 6 l l
l k sl
ksl2 l
beams: þ qn þ q
60 24
12EI0 EA Σ2
2
þ ð1 þ 2
h þk sl 2 p2 EA Σ2 s
k11 ¼ l . 30 12EI0 n— h — EA þ k s n2 þ k sln þ k6l
3m k36 ¼ p2 þ . 30 6
q þ 6p l
Cl mÞ l
C 3 l
3
l
þ q — 6m 2 l
3l l k4 ð1 þ mÞ2 þ s
q2
30
6EI0 EA Σ2
2
h ksl2 . Σ2
k12 ¼ . 12EI C EA h kl
¼ þ
C 3 3 l
þ l0
m2
2
l þ 2 ð1 þ mÞ þ 6 q — 6m l l m2 þ q—
60
q 6m
k ¼ 12EI0 C EA h h kl .
1 þ mÞp þ .q — 6m Σ.n — 6p Σ þ s qn 6EI0 2 C Σ2 ksl2 2 2 EA
mp h — q
13 3 þ ð 3l l l 30 k45 ¼ — 2 m — ð1 þ mÞ — q — 6m
l l 60
ksl l 2 6 l
þ . Σ. Σ
12 12EI0 C EA h h
q k46 ¼ — mp ð1 þ mÞp q— n—
3 3 6m l 6p l
l— —
l
— qn — s q
s
— ð1 þ — q — 6m —
l 3l
EA l Σ 30 30 12l
12EI0 2 2 kl kl k
k14 ¼— 3m
2 mÞ . h q s 2
l
C
.
6EI0 C EA Σ2 ksl2 3EI0 2 EI0 2 EAl Σ2 EAh2
2 h . h
2
lþ Cl

2 þ
q 60 k55 ¼
k15 ¼ m2 ð1 þ mÞ þ q — 6m
2 mþ þ ð1 þ mÞ þ q — 6m þ
6 l 3 l 4 12 l l
ksl
l þ
 Σ
k16 ¼ 12EI0 C EA. h h 2

þ ð1 þ mÞp
—q Σ.n — 6p ksl
þ 30 qn 120
mp
3
l
þ 6m 3l
l l l 6EI0 C EA h h EAh
þ sl q
k k56 ¼ mp 1 þ mÞp þ .q — 6m Σ.n — 6p Σ þ
12 l2 þ ð
ksl2 2 2 6 l l l
ksl
Σ2
k22 ¼ 3EI0 2 EI0 Cl EA EAh2 þ qn þ q
l .q — 6m h
2
m þ ð1 þ mÞ þ 60 24
l l 4 þ l l
þ 12 . Σ
C EA ks 2 ksl ksl
þ p2 þ EA—n
ksl3 12EI h
þ k66 ¼ p2 0
3l
6p
l þ þ l n nþ 3
120 3 l l 30 þ 6
q l
2 2
772 J.-P. Lin et al. / Construction and Building Materials 156 (2017) 761–772
References
[1] E. Ellobody, B. Young, Performance of shear connection in composite beams
with profiled steel sheeting, J. Constr. Steel Res. 62 (7) (2006) 682–694.
[2] Y.F. Wu, D.J. Oehlers, M.C. Griffith, Partial-interaction analysis of composite
beam/column members, Mech. Struct. Mach. 30 (3) (2002) 309–332.
[3] D. Chen, Variational Principles of Partial-Interaction Composite Beams and Modified
Reduced Stiffness Method for Calculating its Deflection Master Thesis, Zhejiang
University, Hangzhou, 2012.
[4] U.A. Girhammar, V.K. Gopu, Composite beam-columns with interlayer slip- exact
analysis, J. Struct. Eng. 119 (4) (1993) 1265–1282.
[5] U.A. Girhammar, D.H. Pan, Exact static analysis of partially composite beams and
beam-columns, Int. J. Mech. Sci. 49 (2) (2007) 239–255.
[6] N. Newmark, C. Siess, I. Viest, Test and analysis of composite beams with incomplete
interaction, P. Soc. Exp. Stress Anal. 9 (1) (1951) 75–92.
[7] R. Xu, D. Chen, Variational principles of partial-interaction composite beams, J. Eng.
Mech. 138 (5) (2011) 542–551.
[8] R. Xu, Y. Wu, Static, dynamic, and buckling analysis of partial interaction
composite members using Timoshenko’s beam theory, Int. J. Mech. Sci. 49 (10)
(2007) 1139–1155.
[9] R. Xu, G. Wang, Variational principle of partial-interaction composite beams using
Timoshenko’s beam theory, Int. J. Mech. Sci. 60 (1) (2012) 72–83.
[10] S. Schnabl, M. Saje, G. Turk, I. Planinc, Analytical solution of two-layer beam taking
into account interlayer slip and shear deformation, J. Struct. Eng. 133 (6) (2007)
886–894.
[11] R. Xu, G. Wang, Bending solutions of the Timoshenko partial-interaction
composite beams using Euler-Bernoulli solutions, J. Eng. Mech. 139 (12) (2013)
1881–1885.
[12] Y. Wu, M.C. Griffith, D.J. Oehlers, Numerical simulation of steel plated RC columns,
Comput. Struct. 82 (2004) 359–371.
[13] Q. Nguyen, M. Hjiaj, S. Guezouli, Exact finite element model for shear-
deformable two-layer beams with discrete shear connection, Finite Elem. Anal.
Des. 47 (2011) 718–727.
[14] M.R. Salari, E. Spacone, Finite element formulations of one-dimensional
elements with bond-slip, Eng. Struct. 23 (7) (2001) 815–826.
[15] Z. Shen, H. Zhong, Static and vibrational analysis of partially composite beams
using the weak-form quadrature element method, Math. Probl. Eng. 2012 (2012)
1–23.
[16] E. Thompson, J. Goodman, M. Vanderbilt, Finite element analysis of layered
wood systems, J. Struct. Div. – ASCE 101 (12) (1975) 2659–2672.
[17] A. Ayoub, F.C. Filippou, Mixed formulation of nonlinear steel-concrete
composite beam element, J. Struct. Eng. 126 (3) (2000) 371–381.
[18] C. Faella, E. Martinelli, E. Nigro, Steel and concrete composite beams with flexible
shear connection: ‘‘exact” analytical expression of the stiffness matrix and
applications, Comput. Struct. 80 (2002) 1001–1009.
[19] A. Dall’Asta, A. Zona, Three-field mixed formulation for the non-linear
analysis of composite beams with deformable shear connection, Finite Elem.
Anal. Des.
40 (2004) 425–448.
[20] B. Cˇ as, M. Saje, I. Planinc, Non-linear finite element analysis of
composite planar frames with an interlayer slip, Comput. Struct. 82 (2004)
1901–1912.
[21] G. Ranzi, M.A. Bradford, B. Uy, A direct stiffness analysis of a composite beam with
partial interaction, Int. J Numer. Meth. Eng. 61 (5) (2004) 657–672.
[22] G. Ranzi, M.A. Bradford, Direct stiffness analysis of a composite beam-
column element with partial interaction, Comput. Struct. 85 (2007) 1206–
1214.
[23]
G. Ranzi, A. Zona, A steel–concrete composite beam model with partial interaction
including the shear deformability of the steel component, Eng. Struct. 29 (11) (2007) 3026–
3041.
[24] H.R. Valipour, M.A. Bradford, A steel-concrete composite beam element with
material nonlinearities and partial shear interaction, Finite Elem. Anal. Des. 45
(2009) 966–972.
[25] S.F. Jiang, X. Zeng, D. Zhou, Novel two-node linear composite beam element
with both interface slip and shear deformation into consideration:
Formulation and validation, Int. J. Mech. Sci. 85 (8) (2014) 110–119.
[26] Q. Nguyen, E. Martinelli, M. Hjiaj, Derivation of the exact stiffness matrix for a
two-layer Timoshenko beam element with partial interaction, Eng. Struct. 2 (33) (2011) 298–
307.
[27] E. Martinelli, C. Faella, G.D. Palma, Shear-flexible steel-concrete composite
beams in partial interaction: closed-form ‘‘exact” expression of the stiffness
matrix, J. Eng. Mech. 138 (2) (2012) 151–163.
[28] R. Brighenti, S. Bottoli, A novel finite element formulation for beams with composite
cross-section, Int. J. Mech. Sci. 89 (2014) 112–122.
[29] G. Taig, G. Ranzi, Generalised beam theory (GBT) for composite beams with
partial shear interaction, Eng. Struct. 99 (2015) 582–602.
[30] J.P. Lin, J.F. Wang, R.Q. Xu, Cohesive zone model based numerical analysis of
steel-concrete composite structure push-out tests, Math. Probl. Eng. 2014 (2014)
1–12.
[31] J.B.M. Sousa, A.R. Da Silva, Nonlinear analysis of partially connected composite
beams using interface elements, Finite Elem. Anal. Des. 43 (2007) 954–964.
[32] A.R. Da Silva, J.B.M. Sousa Jr., A family of interface elements for the analysis of
composite beams with interlayer slip, Finite Elem. Anal. Des. 45 (2009) 305– 314.
[33] Composite construction in structural steel and concrete: British standard
code of practice CP 117: Part I. The Council for Codes of Practice British
Standards Institution, 1965.
[34] J.B. Menzies, CP 117 and shear connectors in steel-concrete composite made with
normal-density or lightweight concrete, The Struct. Engineer 49 (3) (1971) 137–
154.
[35] K. Abel-Aziz, J.M. Aribert, Calcul des poutres mixtes jusqu’à l’état ultime avec un
effet de soulèvement à l’interface acier-béto, Constr. Métallique 4 (1985) (in
French).
[36] H. Robinson, K.S. Naraine, Slip and uplift effects in composite beams, in:
Proceedings of the Engineering Foundation Conference on Composite
construction (ASCE) (1988) 487-497.
[37] J.C. Chapman, S. Balakrishnan, Experiments on composite beams, The Structural
Engineer 42 (11) (1964) 369–383.
[38] R.P. Johnson, Designers’ guide to eurocode 4: design of composite steel and
concrete structures (2nd Edition), London: ICE (Designers’ guides to the
Eurocodes), 2012.
[39] N.A. Jasim, Deflections of partially composite beams with linear connector
density, J. Constr. Steel Res. 49 (1999) 241–254.
[40] J. Kennedy, R. Eberhart, Particle swarm optimization, Proceedings IEEE
International Conference on Neural Networks (1995) 1942–1948.
[41] W. Tu, M.-J. Pindera, Targeting the finite-deformation response of wavy
biological tissues with bio-inspired material architectures, J. Mech. Behav.
Biomed. Mater. 28 (2013) 291–308.
[42] G. Wang, M.-J. Pindera, Elasticity-based microstructural optimization: an integrated
multiscale framework, Mater. Des. 132 (2017) 337–348.