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

net/publication/311947006

Practical nonlinear inelastic analysis method of composite steel-concrete

beams with partial composite action

Article  in  Engineering Structures · March 2017

DOI: 10.1016/j.engstruct.2016.12.017

CITATIONS READS

27 8,495

2 authors:

Cosmin G Chiorean Stefan M Buru

Universitatea Tehnica Cluj-Napoca Universitatea Tehnica Cluj-Napoca
46 PUBLICATIONS   433 CITATIONS    17 PUBLICATIONS   39 CITATIONS   

SEE PROFILE SEE PROFILE

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

Computational Efficient Models for Non-Linear Inelastic Analysis of Building Frameworks View project

All content following this page was uploaded by Cosmin G Chiorean on 15 January 2018.

The user has requested enhancement of the downloaded file.

 Engineering Structures 134 (2017) 74–106

Contents lists available at ScienceDirect

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

Practical nonlinear inelastic analysis method of composite steel-concrete

beams with partial composite action
Cosmin G. Chiorean ⇑, Stefan M. Buru
Faculty of Civil Engineering, Technical University of Cluj-Napoca, 15 C Daicoviciu Str., 400020 Cluj-Napoca, Romania

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

Article history: This paper presents an efficient computer method for nonlinear inelastic analysis of composite steel-
Received 14 June 2016 concrete beams with partial composite action. The proposed formulation is intended to model the com-
Revised 30 November 2016 bined effects of partial composite action and distributed plasticity using only one 2-noded beam-column
Accepted 12 December 2016
element per structural member. Based on elasto-plastic cross-sectional analyses the behaviour model is
able to take into account the effects of partial composite action between the concrete slab and the steel
beam. Gradual yielding throughout the cross-section is described through basic equilibrium, compatibil-
Keywords:
ity, material and shear connection nonlinear constitutive equations. Tangent flexural and axial rigidity of
Composite beams
Partial composite action
the cross-section are derived and then using the flexibility approach the elasto-plastic tangent stiffness
Degree of composite action matrix and equivalent nodal loads vector of the beam-column element including the shear deformability
Nonlinear inelastic analysis of the partially connected composite beam has been developed. The proposed nonlinear analysis formu-
Cross-sectional analysis lation has been implemented in a general nonlinear static purpose computer program, NEFCAD.
Distributed plasticity Advanced finite element simulations have been conducted by using the specialized software for nonlin-
Shear deformability ear analysis of structures, ABAQUS. Several computational examples are given to validate the accuracy
and efficiency of the proposed method by comparing the results predicted by NEFCAD with those given
by the ABAQUS software and other results retrieved from the open literature.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction
Composite systems have seen widespread use in recent decades
due to multiple benefits that occur by combining the individual
mechanical properties of the main component materials, steel
and concrete. The structural steel experiences high strength and
ductility in tension and compression, while the concrete experi-
ences high stiffness and robustness in compression. The composite
action between the concrete slab and the steel beam is achieved
through mechanical connectors such as shear studs. The resulted
composite beam-column provides an increase of the rigidity,
strength and the ultimate moment capacity of the composite ele-
ment, compared with the independent use of each material. An
important aspect in the structural behaviour of steel-concrete
composite beams is represented by the level of shear connection
between the concrete slab and steel beam, which is defined as
the ratio between the shear connection capacity provided by the
studs and the weakest component capacity (concrete slab or steel
beam). If the disposed number of shear connectors at the steel-
concrete interface of the composite beam is lower than the number
⇑ Corresponding author.
E-mail address: cosmin.chiorean@mecon.utcluj.ro (C.G. Chiorean).
that provides full shear connection, then the stiffness and ultimate
capacity decreases, while the ductility of the composite beam may
be enhanced [1–4].
In recent years, have witnessed significant advances in nonlin-
ear inelastic analysis methods for steel and composite steel-
concrete beams and framed structures and integrate them into
the new and more rational advanced analysis and design proce-
dures (e.g., [1,5–15]). Reliable nonlinear analysis tools are, for
instance, essential in performance-based earthquake engineering
and advanced analysis methodologies, that involves accurate pre-
dictions of inelastic limit states, up or beyond, to structural
collapse.
There currently exist several methods and computer programs
devoted to the analysis of composite steel-concrete beams and
frame structures, with full and partial composite action, able to
reveal both linear-elastic and nonlinear inelastic behaviour of
these structures. At one extreme, two- and three dimensional finite
elements enhanced with advanced material and shear connection
nonlinear constitutive laws were used to investigate the nonlinear
response of composite steel-concrete elements with partial com-
posite action (e.g., [3,16–19]). All these available tools for such
advanced analyses are general purpose FE programs that require
very fine-grained modelling, extensive calibration and mesh

http://dx.doi.org/10.1016/j.engstruct.2016.12.017
0141-0296/Ó 2016 Elsevier Ltd. All rights reserved.
 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106
Nomenclature
aL non-dimensional partial composite action parameter
c elastic degree of composite action
ceff inelastic member effective degree of composite action
f(ceff) a function of the inelastic member effective degree of
composite action
e resultant stain under the assumption of full composite
action
ec, es resultant stains in concrete slab and steel beam under
partial composite action
eslip slip strain
g degree of shear connection
l buckling length
sxy, c(s) shear stress in concrete and steel components respec-
tively
weq equivalent transverse shear angle
D denotes an incremental quantity
/ common curvature in concrete and steel components
ft, kt tangent flexibility and stiffness matrices of the cross-
section
fr, kr instantaneous flexibility and stiffness matrices of beam-
column element without rigid body modes
ic longitudinal spacing between shear connectors
k shear connection stiffness
k50% shear connection stiffness corresponding to a value of
connector shear force (P) equal to 50% of shear connec-
tor strength capacity (Psc)
ksec secant shear connection stiffness
kc, ks transverse shear coefficients of concrete slab and steel
profile
n effective number of shear connectors
nf number of shear connectors that provides full shear
connection
r distance between the centroids of concrete slab and
steel profile
s shear slip
u axial strain under the assumption of full composite ac-
tion
uc, us axial strains in concrete slab and steel beam under par-
tial composite action
Ac, As areas of concrete slab and steel profile respectively
generation studies that are often impractical to the structural engi-
neer. At the other extreme, the line elements (1-D) approach in
conjunction with either distributed or concentrated plasticity
models, have been devoted to the development of nonlinear anal-
ysis tools for two-layer composite beams with interlayer slip that
provide a desirable balance between accuracy and computational
efficiency (e.g., [1,7,10,20–27]) among others.
Analytical procedures based on the key hypothesis of New-
mark’s model [28] have been proposed, in recent years, for the sta-
tic response of two-layer beams with interlayer slip in the linear-
elastic regime and based on Euler-Bernoulli beam theory (e.g.,
[21,22,29–31]) among others. Shear flexibility of both concrete slab
and steel beam according to Timoshenko’s theory has been taken
into account, in several analytical and finite element-based proce-
dures to predict linear and nonlinear inelastic behaviour of com-
posite beams with partial composite action (e.g., [1,32–36]) A
detailed discussion about various Euler-Bernoulli and
Timoshenko-based models that were proposed for the analysis of
composite beams with partial composite action can be found in
[1,37]. Moreover, the analytical and finite element formulation
can be cast within the framework of discrete-bond model (i.e.
shear connection is modelled using the concentrated springs at
75
Arsi area of the i-th reinforcement bar
Ashear, c(s) shear area of concrete slab and steel profile respectively
EcAc, EsAs axial rigidity of concrete slab and steel profile respec-
tively
EA⁄ ratio between the product and the sum of axial stiffness
EcAc and EsAs
(EA)t tangent axial rigidity of composite cross-section under
partial composite action
EcIc, EsIs elastic flexural rigidity of concrete slab and steel profile
respectively
(EI)0 elastic flexural rigidity of composite cross-section under
no composite action
(EI)1 elastic flexural rigidity of composite cross-section under
full composite action
(EI)el elastic flexural rigidity of composite cross-section under
partial composite action
(EI)eff effective flexural rigidity of composite cross-section un-
der partial composite action
(EI)t tangent flexural rigidity of composite cross-section un-
der partial composite action
Et tangent modulus of elasticity
Gc, Gs shear modulus of concrete and steel materials
(GA)eq elastic equivalent shear stiffness of composite cross-
section
K50% shear connector stiffness corresponding to a value of
connector shear force (P) equal to 50% of shear connec-
tor strength capacity (Psc)
Ksec secant shear connector stiffness
Ncint , Nsint internal axial force in concrete slab and steel profile
Ncf internal axial force in concrete slab under full composite
action
Necf elastic internal axial force in concrete slab under full
composite action
Nrs number of conventional steel reinforcement bars
Psc shear connector capacity
P interface shear force attributed to a single shear connec-
tor
Sshear, c(s) first moment of the sheared area of concrete slab and
steel profile
connector locations) [26,38] and continuous-bond model (i.e. shear
connection is modelled using the equivalent distributed spring
stiffness) [2,39]. A useful brief review about this issue has been
given by [38]. There are three main approaches that have been
used to model the gradual plastification of partially connected
members in a nonlinear inelastic analysis, one based on the dis-
placement method or finite element approach (e.g., [2,23]), the sec-
ond one based on the force or flexibility method (e.g., [25]), and the
third one refers to mixed or hybrid approach (e.g., [24,26,39,40]).
Despite of the simplicity and ease of implementation, because
displacement-based elements implicitly assumed linear curvatures
along the element length, accuracy in this approach when material
nonlinearity is taken into account can be obtained only using sev-
eral elements in a single structural member, thus the computa-
tional effort is greatly enhanced and the method becomes
prohibited computational in the case of large scale frame struc-
tures. Moreover, these formulations may suffer from the shear-
locking and slip-locking phenomena [2,25]. On the other hand in
the flexibility based approach only one element per physical mem-
ber can be used to simulate the gradual spread of yielding through-
out the volume of the members but the complexity of these
methods derives from their implementation in a finite element
76 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106
analysis program (i.e. state determination procedure) and the
inclusion of the element geometrical effects [41]. However, in
order to allow the concrete and steel to have independent displace-
ments (i.e. partial composite action) all the above-mentioned
methods include additional degrees of freedom at the element
ends, thus the computational effort is greatly enhanced in the case
of large scale frame structures. When modelling the semi-rigid
composite frameworks some difficulties may arise enforcing the
compatibility conditions at the semi-rigid composite connections
[7]. In spite of the availability of such nonlinear inelastic algo-
rithms and powerful computer programs, the nonlinear inelastic
analysis of real large-scale frame structures with partially con-
nected members still possess high demands on the most powerful
computers available and still represents unpractical tasks to most
designers.
The present work attempts to develop accurate yet computa-
tional efficient tools for the nonlinear inelastic analysis of partially
connected composite steel-concrete beams fulfilling the practical
and advanced analysis requirements. Within the framework of
flexibility-based formulation a 2-noded 2D frame element with 6
DOF able to take into account the distributed plasticity and partial
composite action is developed. The nonlinear effects of partial
composite action between the reinforced concrete slab and the
steel beam are taken into account innovatively by introducing
the internal axial force of the concrete slab as function of the mem-
ber effective degree of composite action. With the intention of
quantifying the effects of partial composite action on the strength
and rigidity of composite beams, the internal axial force in the con-
crete slab is assumed to be a fraction of the axial force in the con-
crete slab under full composite action and is estimated as a
nonlinear function of the degree of composite action of the mem-
ber. Hence, the inelastic response, at the cross-sectional level,
can be formulated by means of three equilibrium equations. In this
way, gradual yielding throughout the cross-section subjected to
the combined action of axial force and bending moment is
described through basic equilibrium, compatibility, material and
shear connection nonlinear constitutive equations, the states of
strain, stress and yield stress are monitored explicitly during each
step of the analysis. Tangent flexural and axial rigidity of the cross-
section are derived and then using the flexibility approach the
elasto-plastic tangent stiffness matrix and equivalent nodal load
vector of the beam-column element including the shear deforma-
bility of the composite beam is developed. In this respect, equiva-
lent transverse shear stiffness for partially connected composite
beam has been introduced and used in the general force-
displacement relationships. The derived flexural, axial and trans-
verse shear stiffnesses reflects the influence of the inter layer slip
and depends of the shear connector nonlinear behaviour and
cross-sectional geometry and adopted nonlinear constitutive rela-
tionships for concrete, steel and reinforced bars. The inelastic
response of a composite beam is computed as the weighted sum
of the response of a discrete number of cross-sections (i.e. stations)
that are located at the numerical integration scheme points by
integrating the variable section flexural EIt and axial EAt rigidity
along the member length, depending on the bending moments
and axial force level, cross-sectional shape and nonlinear constitu-
tive relationships for concrete, steel, reinforcements and shear con-
nectors. Besides the equivalent transverse shear stiffness for
partially composite beam is used for the computation of the gener-
alized displacements in the general force-displacement relation-
ship derived at the element level avoiding in this way the
consideration of additional nodes and degrees of freedom to
describe the behaviour of shear-deformable of two-layer beam
with partial interaction as in [32]. In this way the elements of
the stiffness matrix and equivalent nodal loads can be obtained
analytically and readily evaluated by computing the correction
coefficients that affect the elastic flexibility coefficients and equiv-
alent loads and numerical integrations are required only to evalu-
ate these correction coefficients.
Comparing the proposed method with the related distributed
plasticity methods (e.g., [2,25]) the benefits of the present
approach are two folds. First, the proposed formulation takes
advantage of using only one 2-noded beam-column element with
only 6 DOF, as in the classical flexibility-based elements used to
model members with full composite action, since in the derivation
of the force-displacement relationship at the element level, the
cross-section tangent stiffnesses with full composite action can
be replaced with the cross-section tangent stiffnesses accounting
for the effect of partial composite action. Hence, the present
approach does not require additional nodes and degrees of free-
dom to simulate partial shear connection between the reinforced
concrete slab and the steel profile with a beneficial effect over
computational effort and numerical modelling. This is an essential
requirement to approach real large-scale structures, combining
modelling benefits, computational efficiency and reasonable accu-
racy and may circumvent some of the difficulties that may arise
enforcing the compatibility conditions at the semi-rigid composite
connections. Second, since the basic idea of the proposed approach
consists in quantifying the nonlinear effects of partial composite
action and spread of plasticity at the cross-sectional level in which
realistic material properties and cross-sectional shapes are incor-
porated, the proposed inelastic cross-section analysis represents
a valuable tool to assess the behaviour of composite steel-
concrete cross-sections with partial composite action. Based on
this approach sensitivity and parametric studies can be carried
out for a rapid and exact evaluation of different factors that affect
qualitatively the elastic and inelastic response of cross-sections.
A predictor/corrector strategy has been implemented, corre-
sponding to the proposed flexibility-based element, in order to find
the nodal displacements and the element resisting forces. The pro-
posed nonlinear analysis formulation has been implemented in a
general nonlinear static purpose computer program, NEFCAD.
Advanced finite element simulations have been conducted by using
the specialized software for nonlinear analysis of structures, ABA-
QUS [42]. Advanced numerical models have been developed consid-
ering the combination of three-dimensional solid elements (for
concrete volumes), shell elements for steel elements, and connector
elements, with nonlinear behaviour, to represent the shear
connectors.
The paper is organized as follows. In Section 2.1, basic assump-
tions and constitutive material and shear connection models
adopted in the present formulation are presented. The procedure
developed for elasto-plastic analysis of composite beam cross-
sections, with full and partial composite action is described in Sec-
tions 2.2 and 2.3 respectively. A comprehensive study concerning
the numerical procedure developed for evaluation of the tangent
flexural and axial rigidity of the beam cross-section as function
of the inelastic member effective degree of composite action is pre-
sented in Section 2.2. In order to simplify the analysis of the shear
deflections for composite beams equivalent transverse shear stiff-
ness for partially connected composite beam is derived in Sec-
tion 2.4. Having axial, flexural and shear rigidities defined,
flexibility-based method is used to formulate the distributed plas-
ticity model of a 2D beam-column element (6 DOF). This procedure
is detailed in Sections 2.5 and 2.6 respectively. In Section 3 the key
elements for advanced finite element modelling using Abaqus soft-
ware are presented. Several computational examples dealing with
simply supported, two-span continuous composite steel-concrete
beams and a steel-concrete plane frame with composite beams
are presented in Section 4 in order to assess the performance of
the proposed model. The main conclusions of this study are drawn
in Section 5.
 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106 77

2. Mathematical formulation of the proposed method where



 "

 #
Eci ec lim 2 Eci ec lim 2 ec lim Eci
2.1. Basic assumptions and constitutive material and shear connection n¼4 1þ 2 2 þ2  ;
Ec1 ec1 Ec1 ec1 ec1 Ec1
models
2=3
f
ec1 ¼ 0:0022; f ctm ¼ 1:4 ck ð2Þ
In this paper, the following general assumptions are adopted in 10
the formulation of analytical model: (1) Plane sections remain where fcm, fctm and fck are the mean compressive, the mean tensile
plane for entire cross-section after flexural deformation; through- and the characteristic compressive concrete strength respectively;
out the depth of the cross-section, the strain distribution is linear, ec1 is the strain at the peak compression stress; eclim is the compres-
but a discontinuity exists at the concrete slab-steel beam interface sion strain corresponding to half the peak stress; Eci is the initial
due to slip; (2) frictional effects and uplift are neglected, the inter- tangent modulus while Ec1 is the secant modulus at the peak stress.
face slip and strains are assumed to be small; (3) discretely located Multi-linear elasto-plastic stress-strain relationships, both in ten-
interlayer connectors with uniform spacing are regarded as sion and in compression, are assumed for the structural steel and
continuous and ductile with an nonlinear elasto-plastic behaviour; reinforcement bars. In this way the strain-hardening effect may
and (4) torsional buckling do not occur; lateral torsional buckling be included in analysis and different stress-strain relationships
is prevented; warping and cross-section distortion are not can be used to simulate elasto-plastic behaviour for the web and
considered. the flange of the steel beam. The stress-strain relationship of the
As a consequence of the assumption (2) the vertical displace- conventional steel reinforcements in tension concrete region could
ment and the curvature of the different subcomponents (concrete be calculated as (Fig. 1b):
slab and structural steel) are assumed to be the same. The sagging (
Es e; e 6 enr
and hogging flexural behaviour of the composite cross-section is fr ¼   ð3Þ
similar in the elastic domain but becomes different in the elasto- f yr ð0:91  2BÞ þ ð0:02 þ 0:25Be=eyr Þ ; e > enr
plastic field due to different nonlinear constitutive relationships 1:5
where the parameter B ¼ ðf cr =f yr Þ q1 , q = longitudinal reinforce-
assumed for concrete in tension and compression. Transverse
shear deformations, associated to the transverse shear forces are ment steel ratio (limited to a minimum of 0.25%), the modified yield
taken into account on the element stiffness in the proposed strain of the steel bar is enr ¼ eyr ð0:93  2BÞ=ð1  0:25BÞ, the modi-
approach but are neglected in the plastic constitutive relationships. fied yield stress is f nr ¼ Es enr , and the ultimate average strain is
The model suggested by the CEB-FIB Model Code 90 [43], is eur ¼ eyr ð0:07 þ 2BÞ=ð0:25BÞ. In this way can be simulated the inho-
adopted in the present paper to model the concrete under mogeneous distribution of stress of the steel bar along the longitu-
compression and tension (Fig. 1a). According to this model the dinal direction in the tension concrete region [44].
stress-strain relationships are approximated by the following The shear connectors are evenly spaced and assumed to produce
functions: uniformly distributed shear forces (i.e. continuous-bond model).
8 " The constitutive law of the shear connection, that describes the
>
 !

 #1
>
> ec1 ec1 2 ec 2 4ec1 ec relationship between the slip and the interface shear force, usually
>
>  n2 þ n  f cm ;
>
> e e e e e is obtained from experimental push-out tests on shear studs, or,
>
> c lim c lim c1 c lim c1
>
> alternatively, when experimental data are not available, it can be
>
> ec 6 ec lim ;
>
>
>
> 
 1 "
2 # described by widely accepted Ollgaard’s constitutive model [45].
>
> Eci ec Eci ec ec In the proposed model nonlinear elasto-plastic behaviour is
>
>  1 þ  2   f cm ;
>
> Ec1 e Ec1 ec1 e adopted assuming a secant connector stiffness evaluated as func-
>
> c1 c1
< tion of the level of the internal axial force in the concrete slab as will
rc ¼ ec lim 6 ec 6 0;
>
> be briefly described in the next section of the paper.
> Eci ec ;
> 0 < ec 6 0:9f ctm =Eci ;
>
> A beam member (Fig. 2a) is idealized as 2-noded 2D beam-
>
> 0:1f ctm
>
> f ctm  ð0:00015  ec Þ; column element with 6 degrees of freedom (DOF). Each of the ele-
>
> 0:00015  0:9f ctm =Eci
>
> ment nodes has 3 DOF (2 displacements and 1 rotation). The pro-
>
>
>
> 0:9f ctm =Eci < ec 6 0:00015; posed approach is based on the most refined type of second
>
>
>
> order inelastic analysis, distributed plasticity model, where
> rc ¼ 0;
>
>
: elasto-plastic behaviour is modelled accounting for spread-of plas-
ec > 0:00015;
ticity effect in sections and along the element length and employs
ð1Þ modelling of structures with only one line element per structural

Fig. 1. (a) CEB-FIB Model Code 90 relationships for concrete and (b) stress-strain relationship for reinforcing bars.
78 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106
(a)
(Stations)
(reference axis)
Fig. 2. (a) 2-noded 2D beam-column element (6 DOF) and (b) beam-column element with rigid-body modes removed (3 DOF).
member, which reduces the number of degree of freedom involved
and the computational time. The above assumptions allow the for-
mulation details to be considered on two distinct levels, namely,
the cross-sectional level and the member longitudinal axis level.
The key elements of the elasto-plastic formulation are first estab-
lished at the cross-sectional level to reflect the elasto-plastic flex-
ural and axial rigidity of cross-section, upon which, the nonlinear
inelastic response of a beam-column element can be evaluated as
a weighted sum of the response of a discrete number of monitored
cross-sections (i.e. stations) that are located at the numerical inte-
gration scheme points (Fig. 2a).
2.2. Elasto-plastic analysis of beam cross-sections
Composite beams are usually composed of a steel joist which is
connected to a concrete slab by means of shear connectors and the
composite action between the steel beam and the concrete slab
depends on the performance of shear connectors at their interface.
The elastic and inelastic behaviour of these members is quite com-
plex because the shear connectors generally permit the develop-
ment of only partial composite action between the individual
components of the member, and their analysis requires the consid-
eration of the interlayer slip between the subcomponents. Full
composite action (infinite connection stiffness) and non-
composite action (zero connection stiffness) represent upper and
lower bounds for the partial composite action [29]. Partial compos-
ite action involves the influence of both the partial-stiffness and
the partial-strength of shear connection according to the degree
of composite action. Usually the degree of composite action of
steel-concrete composite members can be defined into two terms,
namely the degree of interaction and the degree of shear connec-
tion. The degree of interaction is related to the stiffness-based
property while the degree of shear connection is the strength
based property, and characterizes partial-shear interaction beha-
viour, and partial-shear connection behaviour, respectively, but
both are directly related to each other. At the interface between
the steel beam and the concrete slab, if the shear strength is
increased, the shear stiffness is also increased [3,4]. Usually, for a
given composite beam, the full shear connection is defined as the
least number of shear connectors (nf) such that the bending resis-
tance of the beam would not be affected if more shear connectors
are provided, whereas partial shear connection occurs when the
number of connectors (n) used in a beam is lower than (nf).
In order to analyze and design the composite beams, simplified
methods are very useful and such methods are proposed in inter-
national literature and in some codes [46]. For instance in the Euro-
code 4 [46] the concept of the degree of shear connection g ¼ n=nf
is used and the ultimate bending strength capacity of cross-section
(b)
is evaluated by simple equilibrium of stresses with a prescribed
compressive axial force in the concrete slab N c ¼ nP sc , where
n < nf represents the number of shear connectors and Psc is the con-
nector strength. It is important to highlight that the full shear con-
nection and implicitly the associated number of shear connectors
(nf) can be defined only in relation to particular methods of design
and of actual material properties measured during the experimen-
tal procedure [3,46].
The composite beam cross-section considered here consists of a
concrete solid slab connected to a steel beam as presented in Fig. 3.
We assume that the geometrical shape of the cross-section and the
distribution of material (i.e. concrete and steel) are symmetric with
respect to the vertical axis (y) and the origin of the reference axis is
considered in the geometric centroid of the structural steel. Loads
are symmetrical with respect to the (yx) plane which represents
the plane of bending. No torsion and out of plane bending are con-
sidered. Frictional effects at the steel-concrete interface are
neglected and no separation between steel beam and concrete slab
is allowed, they are consistent by having the same curvature (/).
The conventional steel reinforcements are assumed to be discrete
points and full strain compatibility exists between the steel rein-
forcement and the surrounding concrete (i.e. bond-slip between
concrete and reinforcement is not considered).
2.2.1. Full composite action
In this case full composite action between structural steel and
concrete is assumed, plane sections remain plane after deforma-
tions (Fig. 3.a). Let us consider the cross-section subjected to the
action of the external bending moment (M), and axial force (N)
as shown in Fig. 3.a, where typical steel-concrete composite beam
cross-section is depicted. Under the above assumptions the resul-
tant strain distribution corresponding to the curvature / and the
axial strain u (evaluated at the centroid of structural steel) can
be expressed, throughout the depth of cross-section, in a linear
form as:
e¼uþuy ð4Þ
The equilibrium is satisfied when the external forces (N, M) are
equal to the internal ones. These conditions can be represented
mathematically in terms of the following nonlinear system of equa-
tions as:
8
> R R X
N rs
>
>
> As rðeðu; uÞÞdAs þ Ac rðeðu; uÞÞdAc þ
<
rðei ðu; uÞÞArsi N ¼ 0
i¼1
>
> R R X
Nrs
>
>
: As rðeðu; uÞÞydAs þ Ac rðeðu; uÞÞydAc þ rðei ðu; uÞÞyi Arsi M ¼ 0
i¼1
ð5Þ
 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106 79

(a)

(Origin of reference
axis)

(b)

(Origin of reference
axis)

Fig. 3. Composite beam: (a) full composite action and (b) partial composite action.

in which u; u represent the unknowns, defined implicitly via resul-
tant strain function eðu; uÞ, and the surface integral is extended over
steel (As) and concrete areas (Ac), the contribution of slab reinforce-
ment is explicitly considered, Arsi (i = 1, Nrs) denotes the resisting
cross-section area of the i-th bar located at yi distance about the ref-
erence axis (z), Nrs denotes the total number of conventional steel
reinforcements. The above system can be solved numerically using,
for instance, the Newton iterative method and taking into account
the fact that the stresses are implicit functions of the axial strain
and curvature through the resultant strain distribution given by
the Eq. (4) [15]. In this way, for given bending moment and axial
force we can obtain the strain and stress distribution throughout
the cross-section and then the axial and flexural rigidity of the
cross-section can be computed as described in [15]. Moreover, the
resultant stresses for each component (steel beam and concrete)
can be evaluated. For instance, the internal axial force in concrete,
in which the contribution of the conventional steel reinforcements
is included, can be computed as:
Z X
N rs
Ncf ¼ rðeðu ; u ÞÞdAc þ rðei ðu ; u ÞÞArsi
Ac i¼1
where ðu ; u Þ represents the solution of the nonlinear system (5).
2.2.2. Partial composite action
2.2.2.1. Elasto-plastic cross-sectional stiffness. In this case the distri-
bution of strains throughout the depth of both the steel section and
the concrete slab is assumed to be linear with a discontinuity at the
steel-concrete interface (Fig. 3b); the slip between the steel section
composite action ceff , whose expression will be detailed and dis-
cussed in the next section, and which can be defined also, implicitly,
as a function of the degree of shear connection g already defined.
The composite beam is assumed to be subjected to the action of a
positive or sagging bending moment (M) and axial force (N), as
shown in Fig. 3b, even though it will be seen that the analysis pro-
cedures developed here can be applied in a similar way for the cases
when the composite beam is subjected to a negative (or hogging)
bending moment. Under the above assumptions, the resultant
strain distribution, corresponding to the curvature / and the axial
strains uc and us, evaluated at the centroid of concrete slab and
structural steel respectively, can be expressed in a linear form as:
ec ¼ uc þ u  ðy  rÞ; es ¼ us þ u  y ð7Þ
where ec and es represents the strains in concrete slab (including the
conventional reinforcements) and steel beam respectively and r
represents the distance from the central axis of the concrete slab
to that of the steel beam. The slip strain (eslip ) is defined as the rel-
ative strain between the concrete slab strain (ec ) and the steel beam
ð6Þ strain (es ) at the interface (Fig. 3b):
eslip ¼ ec  es ¼ uc  us  u  r ð8Þ
The equilibrium is satisfied when the external forces are equal to
the internal ones. These conditions can be represented mathemati-
cally in terms of the following nonlinear system of equations as:
8
> R R X
N rs
>
> As rðes ðus ; uÞÞdAs þ Ac rðec ðuc ; uÞÞdAc þ rðei ðuc ; uÞÞArsi  N ¼ 0
>
<
i¼1

and concrete slab is quantified in term of slip strain. Let us consider >
> R R X
N rs
>
>
the cross-section subjected to the action of the external bending : As rðes ðus ; uÞÞydAs þ Ac rðec ðuc ; uÞÞydAc þ rðei ðuc ; uÞÞyi Arsi  M ¼ 0
i¼1
moment (M), and axial force (N) as shown in Fig. 3b. ð9Þ
Nonlinear effects of partial shear connection of composite
beams are taken into account innovatively by introducing the inter- in which uc, us and / represent the unknowns, Arsi (i = 1, Nrs), Nrs
nal axial force of the concrete slab as function of the degree of denotes the area and the number of conventional steel
80 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106

2 3
reinforcement bars, respectively, and the surface integrals are Nc
extended over concrete slab (Ac) and structural steel areas (As). ext 6 7
f ¼ 4 N  Nc 5 ð16Þ
The internal axial force and the bending moment respectively, in
M
the concrete slab, in which the contribution of the conventional
steel reinforcements is included, are expressed as: The above system can be solved numerically using, for instance, the
Z X
N rs
Newton iterative method and taking into account the fact that the
Ncint ¼ rðec ðuc ; uÞÞdAc þ rðei ðuc ; uÞÞArsi ð10:aÞ stresses are implicit functions of the axial strains and curvature,
Ac i¼1 for each component, through the resultant strain distribution given
by the Eq. (7). According to the Newton iterative method, the iter-
Z X
N rs ative changes of unknowns vector X can be written as:
M cint ¼ rðec ðuc ; uÞÞydAc þ rðei ðuc ; uÞÞyi Arsi ð10:bÞ
1
Ac i¼1 Xkþ1 ¼ Xk  F0 ðXk Þ FðXk Þ; kP0 ð17Þ
whereas the internal axial force and the bending moment, respec- 0
where F represents the Jacobian (or tangent cross-section stiffness
tively, in the structural steel beam are determined by the following matrix) of the nonlinear system (14) and can be expressed as:
relationships: 2 3
Z @Ncint @N cint @Ncint
6 @uc @us @u 7
Nsint ¼ rðes ðus ; uÞÞdAs ð11:aÞ 6 7

 6 7
As @F 6 s
@Nint s
@N int s
@Nint 7
F0 ¼ ¼66
7
7
Z @X 6 @uc @us @u 7
6 7
M sint ¼ rðes ðus ; uÞÞydAs ð11:bÞ 4 @Ms @M c
@M s
@M c
@M s
@M c 5
ð18Þ
As
int
þ int int
þ int int
þ int
@uc @uc @us @us @u @u
In order to solve the above nonlinear system of equations the inter- 2 3 2 3
ðEt AÞc 0 ðEt SÞc  rðEt AÞc k11 0 k13
nal axial force in the concrete slab (N cint ), under partial composite 6 7 6 7
¼64 0 ðEt AÞs ðEt SÞs 7 ¼ 6 0 k22 k23 7
5 4 5
action, is assumed to be a fraction of the axial force in the concrete
slab under full composite action (Ncf) and the amount is defined by ðEt SÞc ðEt SÞs ðEt IÞs þ ðEt IÞc  rðEt SÞc k31 k32 k33
a function of the degree of composite action ceff as (Fig. 3b):
Explicitly the expressions of the Jacobian’s coefficients are given in
Ncint ¼ f ðceff Þ  Ncf ð12Þ Appendix A, Eq. (A1), ðEt AÞc and ðEt AÞs denote the axial rigidity of
concrete slab and steel component, ðEt SÞc , ðEt SÞs represents the first
where Ncf represents the internal axial force in the concrete slab of moment of the axial rigidity of the concrete slab and structural steel
the cross-section subjected to the same external bending moment respectively, in respect with the geometric centroid of the structural
M and axial force N but under the assumption of full composite steel, ðEt IÞc , ðEt IÞs represents the second moment of the axial rigidity
action between the steel beam and concrete slab, f ðceff Þ represents of concrete slab and structural steel, respectively, in respect with
a function of the effective degree of composite action whose expres- the geometric centroid of the structural steel. These coefficients
sion will be detailed and discussed in the next section. The internal are expressed in terms of the tangent modulus of elasticity Et = -
axial force (Ncf) is obtained by applying the procedure described in dr/de. The iterative procedures starts with the all unknowns uc,
the preceding section (see Eq. (6)). It is important to emphasize here us and / set to zero and the solutions will be computed in just a
that for composite beams with full shear connection (g P 1, few iterations by applying the rapid locally convergent Newton iter-
ceff ! 1), f ðceff Þ ! 1 and consequently Ncint ! Ncf , whereas in the ative procedure given by Eq. (17). The convergence criterion is
case of no shear connection (ceff ¼ 0, g ¼ 0), Ncint ¼ 0. In this way expressed as a ratio of the norm of the out-of-balance force vector
with Nc ¼ f ðceff Þ  N cf and N s ¼ N  N c ¼ N  f ðceff Þ  N cf the nonlin- to the norm of the total applied load [15]. In this way, for given
bending moments we can obtain the strain and stress distribution
ear system (9) becomes:
8 throughout the cross-section and then the tangent flexural rigidity
> R XN rs
of the cross-section can be computed, as will be briefly described
>
> A rðec ðuc ; uÞÞdAc þ rðei ðuc ; uÞÞArsi  Nc ¼ 0
>
>
>
>
c below. The tangent stiffness matrix for the cross section relates
<R i¼1

rðes ðus ; uÞÞdAs  ðN  Nc Þ ¼ 0 small changes in the deformations to small changes in the actions
>
>
As
sustained. The incremental relationships between incremental
>
>R R XNrs
>
> efforts and incremental deformations can be expressed:
>
: As rðes ðus ; uÞÞydAs þ Ac rðec ðuc ; uÞÞydAc þ rðei ðuc ; uÞÞyi Arsi  M ¼ 0
i¼1
2 3 2 3 2 3
ð13Þ k11 0 k13 Duc DN c
6 7 6 7 6 7
The above nonlinear system of equations can be represented in 4 0 k22 k23 5  4 Dus 5 ¼ 4 DNs 5 ð19Þ
vectorial form as: k31 k32 k33 Du DM

int ext
FðXÞ ¼ f f ¼0 ð14Þ where the coefficients of the tangent stiffness matrix kij given in Eq.
(19) are detailed in Appendix A (Eq. (A1)), and are evaluated consid-
where X ¼ ½ uc us u T represents the unknown vector, the exter- ering the strains and stresses at equilibrium for given external
nal and internal loading vectors can then be represented by: actions applied on the composite section, DN c , DNs ¼ DN  DNc rep-
2 3 resents the incremental axial forces in concrete slab and steel beam
c R X
N rs
6 N int ¼ Ac rðec ðuc ; uÞÞdAc þ rðei ðuc ; uÞÞArsi 7 respectively and DM represents the incremental bending moment.
6 7
6 i¼1 7 We define the tangent flexural rigidity of cross-section as a ratio
6 R 7
int 6 s
f ¼ 6N int ¼ As rðes ðus ; uÞÞdAs 7 between incremental bending moment and incremental curvature
7
6 7
while keeping constant the axial force (DN ¼ 0) as:
6 R R X
N rs 7
4
M int ¼ As rðes ðus ; uÞÞydAs þ Ac rðec ðuc ; uÞÞydAc þ rðei ðuc ; uÞÞyi Arsi 5 
DM 
i¼1
ðEIÞt ¼ ð20Þ
ð15Þ Du DN¼0
 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106 81

The system given by the Eq. (19) can be detailed as: in which ðEs As Þ and ðEc Ac Þ represents the axial stiffness of the steel
8 and concrete (including the contribution of conventional steel rein-
< k11 Duc þ k13 Du ¼ DNc ¼ f ðceff Þ  DNcf
>
forcements) components respectively, and employing the definition
k22 Dus þ k23 Du ¼ DNs ¼ f ðceff Þ  DNcf ð21Þ for flexural rigidities of cross-section with and without composite
>
:
k31 Duc þ k32 Dus þ k33 Du ¼ DM action ðEIÞ1 and ðEIÞ0 respectively, we can simplify the expression
and solving for Duc from the first equation and for Dus from the sec- of the bending stiffness as:
ond equation in function of the incremental curvature Du, the third ðEIÞ0
equation of the system (19) becomes: ðEIÞel ¼ 1 0
ð30Þ
1  f ðceff Þ ðEIÞðEIÞðEIÞ
1
f ðceff Þ  DNcf  k13 Du f ðceff Þ  DNcf  k23 Du
k31 þ k32 þ k33 Du ¼ DM
k11 k22 When the connection is completely rigid (infinite degree of com-
ð22Þ posite action, ceff ! 1), f ðceff Þ ¼ 1 and ðEIÞel ¼ ðEIÞ1 . When there
is no connection (zero degree of composite action, ceff ¼ 0),
Then, by simply rearranging the terms in the above equation and
taking into account the Eq. (20) the tangent flexural rigidity of the f ðceff Þ ¼ 0 and ðEIÞel ¼ ðEIÞ0 .
cross-section, with partial composite action, can be finally The tangent axial rigidity of the cross-section is computed as
expressed in the following form: algebraic sum of the tangent axial rigidities of the steel and con-
crete components, namely:
a
ðEIÞt ¼ DNcf
ð23Þ ðEAÞt ¼ ðEAÞct þ ðEAÞst ð31Þ
1  bf ðceff Þ DM
where tangent axial rigidity of each component is defined as a ratio
where
between incremental axial force and corresponding incremental
k31 k13 k32 k23 k31 k32 axial deformation while keeping constant the bending moment
a ¼ k33   ; b¼  ð24Þ
k11 k22 k11 k22 (DM = 0):
 
and DNcf represents the incremental axial force in the concrete slab DNc  DNs 
ðEAÞct ¼ ; ðEAÞ ¼ ð32Þ
under the assumption of full composite action computed as a differ- Duc DM¼0 st
Dus DM¼0
ence between the axial force in the concrete slab associated at an
incremented bending moment M + DM and the axial force at given Solving the system (21) considering the above constraint the fol-
value of the bending moment M: lowing relationships are derived for the tangent axial rigidities for
concrete and steel component respectively:
DNcf ¼ N cfMþDM  NM
cf ð25Þ 2
k11 ðk23  k22 k33 Þ þ k13 k22 k31
ðEAÞct ¼ 2
;
The relationship (23) can be applied even for evaluation of the elas- k23  k22 k33 þ k13 k23
tic bending stiffness of the composite cross-section with partial 2
k22 ðk11 k33  k13 k31 Þ  k11 k23
composite action. Indeed, considering the origin of the reference ðEAÞst ¼ ð33Þ
k11 k33  k13 k31  k31 k23
axis in the geometric centroid of the steel beam, when apply Eq.
(23) the expression of the bending stiffness can be reduced to the and where has been taken into account that k23 ¼ k32 . It can easily be
following relationship: observed that when elastic behaviour is assumed the total axial stiff-
ness ðEAÞel of the cross-section is reduced to the algebraic sum of axial
Es Is þ Ec Ic
ðEIÞel ¼ DNecf
ð26Þ rigidities of the concrete slab (Ec Ac ) and steel components (Es As ):
1  r  f ðceff Þ DM
ðEAÞel ¼ ðk11 þ k22 Þel ¼ Ec Ac þ Es As ð34Þ
which represents the elastic flexural rigidity of the composite beam
Fig. 4 shows a simplified analysis flowchart of the procedure devel-
cross-section with partial composite action and where EsIs, EcIc are
oped for inelastic analysis of cross-section with partial composite
flexural rigidity of steel and concrete slab components (including
action. When the tangent flexural ðEIÞt and axial rigidities ðEAÞt
the contribution of conventional steel reinforcements) respectively,
are computed for each cross-section located along the element
computed about their centroidal axes and DN ecf represents the elas-
length accordingly with the adopted numerical scheme, the inelas-
tic internal axial force in concrete when full composite action is tic behaviour of the beam element can be determined as a weighted
assumed. It can easily be observed that when f(ceff) = 0 the equation sum of inelastic responses of these cross-sections, as will be briefly
is reduced to the well-known relationship used for computing the described in Section 2.5.
elastic flexural rigidity of the composite beam section without com- It is important to point out here that the accuracy of the pro-
posite action (zero interaction) as algebraic sum of the flexural posed approach, for inelastic behaviour evaluation of the compos-
rigidities of the steel and concrete components: ite cross-section, is influenced only by the approximate manner in
which axial force in the concrete slab is predicted along the ele-
ðEIÞ0 ¼ Es Is þ Ec Ic ð27Þ
ment length as a function of the effective degree of composite
Additionally, observing that the elastic internal axial force in con- action. The next section deals with the development of such an
crete can be expressed as: approximate procedure.

DM ðEc Ac ÞðEs As Þ 2.2.2.2. Evaluation of the axial force in the concrete slab. Member
DNecf ¼ r ð28Þ
ðEIÞ1 ðEc Ac Þ þ ðEs As Þ effective degree of composite action. In the preceding subsection,
effective elastic and elasto-plastic cross-sectional rigidities have
where elastic flexural rigidity of the composite beam with full been derived to model cross-sectional behaviour of the beam with
composite action is expressed as: partial composite action. In attempting to find approximate but
reliable and general solutions for computing the elastic deflections
ðEc Ac ÞðEs As Þ
ðEIÞ1 ¼ Es Is þ Ec Ic þ r 2 ð29Þ of composite beams with partial composite action some equations
ðEc Ac Þ þ ðEs As Þ
have been proposed by several researchers [20]. Individual
82 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106

INPUT
Start with known bending moment (M ) and axial force
(N), detailed geometry of cross-section, stress-strain
relationships and elastic (or effective) degree of composite
action γ (or γ eff )

Compute the axial force in concrete slab under the assumption of full composite action
N cfM (Eq.6) for combination ( N and M ) by solving the nonlinear system (5). Repeat the
M + ΔM
procedure for combination ( N and M + ΔM ) resulting N cf . Determine the incremental axial
force in concrete slab ΔN cf = N cfM + ΔM − N cfM .

Compute the axial force in the concrete slab as a fraction of N cf : N c = f (γ eff ) N cf

M M

Define the external force vector: f = [N c N − N c M ]

ext T

Initialize deformation vector (axial strain in concrete uc and steel us and curvature φ )
X = [uc u s φ ] = 0 and put k=0

Compute/Update the cross-section stiffness matrix F (X ) (kij, i,j=1,2,3) (Eq. A1) and internal
' k

T
force vector f = [N int M int ] (Eq. 15).
int c s
N int
For each iteration (k=k+1)

( k)
Compute out of balance force vector: F( X ) = f int − f ext

Compute updatet values of deformation vector (axial strains and curvature), Eq. 17:
−1
X (k +1) = X (k ) − F ' (X (k ) ) F(X (k) )

Check
NO convergence
FT F f extT f ext < tol

YES

Update the cross-section stiffness matrix coefficients ( kij, i,j=1,2,3) (Eq. A1)

Compute tangent flexural rigidity ( EI t ) by using the Eq. (23) and tangent axial rigidity (EAt ) by
using the Eq. (31).

Fig. 4. Simplified analysis flowchart of composite cross-sections with partial composite action.

equations have been derived for different loading scenarios and hyperbolic trigonometric functions of the non-dimensional partial
materials which can all be written in the following general form: composite action parameter aL (see Eq. (37)) which can be closely

1  approximated by [20]:
EI
d ¼ d1 1 þ f ðaLÞ  1 ð35Þ 10
EI0 f ðaLÞ ¼ ð36Þ
10 þ ðaLÞ2
where d is the deflection of the beam, d1 is the deflection when full Noting that since d=d1 ¼ EI1 =EIel the Eq. (35) can be rewritten sim-
composite action is assumed, and f ðaLÞ is a constant involving ilarly as in Eq. (30). Recently, an effective elastic bending stiffness
 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106
valuable for analysing approximate deflections and internal actions
or stresses in composite beams with partial shear connection has
been derived in [29–31]. This approach it is general in nature and
can be applied to arbitrary support, loading conditions, different
types of slabs, shear connectors and materials (steel-concrete or
layered wood beam systems), and geometry parameters. This effec-
tive bending stiffness depends on two non-dimensional parame-
ters: the non-dimensional partial composite action parameter aL,
and the non-dimensional relative bending stiffness parameter
ðEIÞ0 =ðEIÞ1 as:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
k ðEIÞ1
aL ¼  L
EA ðEIÞ0
where k represents the shear connection stiffness constant, L repre-
sents the member length, ðEIÞ0 represents the elastic bending stiff-
ness of the composite section without composite action given by
the Eq. (27) and ðEIÞ1 is the elastic bending stiffness of the fully
composite section given by the Eq. (29) which can be further
expressed as in Eq. (38) in which EA represents the ratio between
the product and the sum of the axial stiffnesses of the concrete and
steel components of the beam cross-section:
ðEIÞ1 ¼ Es Is þ Ec Ic þ r 2 EA
The approximate effective bending stiffness for partially composite
beams, with linear elastic behaviour, can be introduced as [29]:
" #1
ðEIÞ1 =ðEIÞ0  1
ðEIÞeff ¼ 1 þ  ðEIÞ1
1 þ ðl=pÞ2 ðaLÞ2
This approximate effective bending stiffness has been proved to
be very efficient and general to predict the deflections, internal
actions and stresses by simply replacing the fully composite bend-
ing stiffness with the effective one given in Eq. (39) in the expres-
sions for these quantities in the corresponding fully composite
beam. The buckling length coefficient l introduced in Eq. (39)
can be taken as the same as those valid for fully composite ele-
ments [29,30]. For instance l ¼ 1 for simply supported element,
l ¼ 0:5 for clamped-clamped element, and l ¼ 0:7 for pinned-
clamped element. The effective bending stiffness given in Eq.
(39) can be further expressed in the following form:
ðEIÞ0
ðEIÞeff ¼ " #
ðEIÞ1 ðEIÞ0
1 1
2 ðEIÞ1
1þ p 2
ðlaLÞ
Using the above equation and comparing with that obtained in the
current paper for elastic flexural rigidity of composite beam
Fig. 5. Nonlinear constitutive law for the shear connection. The connection secant stiffness.
83
cross-sections (see Eq. (30)) it is easy to deduce the corresponding
expression for the function f ðcÞ:
1 1
f ðcÞ ¼ ¼ ð41Þ
1 þ ðcÞ2 1 þ ðlpaLÞ2
p2 2
where
c ¼ laL ð42Þ
and we will consider here as an elastic degree of composite action.
The degree of composite action used in expression given by the
function f ðcÞ can be used to obtain approximate values for the inter-
ð37Þ nal actions in partially composite beams in a way already illustrated
in preceding section. For instance the linear elastic axial force in the
concrete slab can approximately be expressed as:
Nec ¼ f ðcÞNecf ð43Þ
where Necf represents the elastic axial force in concrete slab for fully
composite action.
This approach, based on elastic effective composite bending
stiffness, may be now extended in order to account for the nonlin-
ear phenomena (distributed plasticity) exhibited at the member
level and shear connection nonlinear behaviour. In general, when
ð38Þ yielding occurs in a member, the cross-sectional flexural rigidities
varies along the longitudinal axis of the member and consequently
the degree of composite action as defined above will vary as well. A
current cross-section located at distance x from one end of the
member is characterized by the values of shear connection stiff-
ness kðxÞ and the parameter aðxÞ defining the non-dimensional
ð39Þ
partial composite action parameter by means of Eq. (44). These
characteristics are suitably modified step by step, accordingly with
the incremental analysis performed at the element level. Conse-
quently, for each cross-section, along the member length, we can
define the cross-sectional degree of composite action
cðxÞ ¼ laðxÞL where:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
kðxÞ ðEIÞ1
aðxÞ ¼  x
ð44Þ
ðEAÞx ðEIÞ0x
In the proposed approach the discrete shear connectors are
regarded as continuous and ductile with a nonlinear elasto-
plastic behaviour (Fig. 5). However, the connection is generally
realized by studs that are assumed to be evenly spaced at distance
ic. The internal axial force in the concrete slab (N cint ), see Eq. (12), in
ð40Þ
a given cross-section is equal to the sum of the shear forces that
develop in all the connectors located on the one or the other side
of the respective cross-section. In other words, the shearing forces
loading the shear connectors represent the difference between the
axial forces that develop in the concrete slab in the cross-sections
84 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106

Z Z X
N rs
adjacent to the connectors. Under the above assumption the inter-
face shear force acting to a single connector, at the current cross- ðEIÞ0x ¼ ETs y2 dAs þ ETc ðy  rÞydAc þ ETrs;i ðyi  rÞyi Arsi
As Ac i¼1
section, can be approximated as: Z Z
X
N rs
DNcint ðEAÞx;c ¼ ETc dAc þ ETrs;i Arsi ; ðEAÞx;s ¼ ETs dAs ;
P ð45Þ Ac i¼1 As
nc
ðEAÞx;s  ðEAÞx;c
where (P) represents the interface shear force attributed to a single ðEAÞx ¼ ; ðEIÞ1 0 
x ¼ ðEIÞx þ r  ðEAÞx
2
ðEAÞx;s þ ðEAÞx;c
connector, DN cint ¼ N cint;ðic =2Þ  Ncint;ðþic =2Þ denotes difference between
ð48Þ
the axial forces that develop in the concrete slab in the cross-
sections adjacent to the current cross-section, nc represents the These coefficients are expressed in terms of the tangent modulus of
number of connectors in a single row at the current cross-section, elasticity Et = dr/de computed for concrete (c), steel (s) and rein-
and ic represents the connector spacing. In the above relationship forcements (rs), respectively. It is worth noting that in elasto-
N cint;ðic =2Þ represents the internal axial forces in the concrete slab, plastic domain the relative bending ðEIÞ1 x , expressed in the above
computed at the cross-sections located at the distances ic =2 from equations, does not represents the tangent flexural rigidity of beam
current cross-section assumed as a control section (integration cross-section with full composite action. The tangent flexural rigid-
point). ity of beam cross-section with full composite action can be deter-
For a shear force (P) less than connector strength (Psc) but mined by using the procedure described in [15].
higher than 50% of Psc, a secant value for the shear connection stiff- In an attempt to avoid the computational complexity associated
ness (ksec) is considered (Fig. 5). For instance, when the Ollgaard to the complicated functional form of cðxÞ we shall introduce the
et al. [45] nonlinear relationship between the shear force (P) in concept of inelastic member effective degree of composite action
the connector and relative slip (s) is assumed, the shear connection for the present analysis. This member effective degree of compos-
stiffness (ksec) can be computed as: ite action is assumed to be a constant over the length of the mem-
a ber and it is defined by the following relationship:
K sec P Psc ð1  ebs Þ
Z L 
Z 1 
ksec ¼ ¼ ¼ ð46Þ 1 x
ic ic  s ic  s ceff ¼ lL aðxÞdx ¼ lL aðnLÞdn ; n¼ ð49Þ
L 0 0 L
where Psc is the connector strength and a and b represents two con-
stants usually determined by a standard push-out tests [45]. The and represents an average of the cross-sectional degree of compos-
secant slip (s) corresponding to the shear force P used in Eq. (46) ite action cðxÞ along the member length (Fig. 6). The above integra-
can be derived by the Ollgaard nonlinear relationship and consid- tion is carried out by applying the Gauss-Lobatto rule as will be
ered in Eq. (46): described next in Section 2.5 of this paper.

a1 In summary, to carry out the proposed method of elasto-plastic
1 P analysis of partially composite cross-sections, the member effec-
s¼ lnð1  Þ ð47Þ
b Psc tive degree of composite action has to be computed first, as already
described, taking into account the partially composite characteris-
In this way the connection secant stiffness ksec may be updated at
tics along the member length evaluated based on the information
each load increment as function of the level of the internal axial
retrieved from the previous loading increment. In this respect the
force in the concrete slab at the current cross-section. For shear
cross-sectional effective degree of composite action is evaluated
forces (P) less than 50% of Psc a constant value for the shear connec-
at each control section (i.e. integration point) along the member
tion stiffness (k50% ) is considered (Fig. 5), assuming a secant connec-
length. The member effective degree of composite action ceff is
tor stiffness corresponding to 50% of Psc which is currently assumed
as a reference [20]. The stiffness reduction for shear connectors in then calculated from Eq. (49). This member effective degree of
hogging bending regions may be modelled, generalizing the above composite action is to be used in place of elastic degree of compos-
nonlinear relationship as suggested in [20]. In Eq. (44) the relative ite action c in Eq. (42) in subsequent analysis and will be updated
at each load increment for each member. To compute the axial
bending ðEIÞ1 0 
x , ðEIÞx and axial ðEAÞx stiffnesses may be updated at
force in the concrete slab, as a fraction of the concrete force under
every load cycle, and are evaluated considering the strains and
full composite action (Ncf), the following expressions will be used:
stresses throughout the cross-section at equilibrium for given
external actions applied on the composite section from the previous Ncint ¼ f ðceff Þ  Ncf ð50Þ
load increment. The strains and stresses are determined by applying
the numerical procedure for solving the nonlinear system (13), 1
described in the preceding subsection of this paper. For the sake
f ðceff Þ ¼ ð51Þ
1 þ ðcp Þ2
2

of completeness these expressions are listed below for reference: eff

γ(x) L γ(x)
∫ γ (x )dx = γ
0
eff ⋅L

γeff
x

0 L

Fig. 6. Inelastic member effective degree of composite action.

 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106 85

Fig. 7 summarizes the proposed procedure for determining the adaptive quadrature rule in each subinterval [48]. The conven-
inelastic member effective degree of composite action and tangent tional steel reinforcements are assumed to be discrete points with
stiffnesses within a single load increment. effective area, co-ordinates and stresses. To avoid double counting
As already mentioned the degree of shear connection represents of the concrete area that is displaced by the steel bars, the concrete
a strength-based property and characterize partial-shear connec- stress at the centroid of the steel bars is subtracted from the rein-
tion behaviour, and is directly related with the degree of interac- forcement bar force.
tion. Therefore the degree of shear connection g can be easily
related to the non-dimensional partial composite action parameter 2.4. Derivation of equivalent transverse shear stiffness for composite
aL as a load-slip curve is assumed for shear connectors. The mem- beam
ber effective degree of composite action ceff can be defined implic-
itly by the degree of shear connection g. For instance, if the shear To simplify the analysis of the shear deflections for composite
strength is increased by increasing the number of connectors (n), beams, it is convenable to associate the transverse shear force
the shear stiffness is also increased due to the fact that increasing TðxÞ at a given cross-section (x) along the element length (L) with
the number of shear connectors the distance ic between them an equivalent (average) shear angle weq ðxÞ, according to
decreases, accordingly with the Eq. (46). When g ¼ n=nf P 1 Timoshenko model [34]. An energy approach is used to develop
(nf =number of connectors associated to full shear connection) an explicit formulation for the equivalent transverse shear stiffness
the effective degree of composite action ceff ! 1 and consequently for composite beam [34]. The equivalent shear angle, associated
f ðceff Þ ! 1. The design parameters assumed by the proposed with the shear force T can be defined in the following form:
approach to describe the partial shear connection of the composite 6 TðxÞ
beams are similar with those required in the current design codes weq ðxÞ ¼ ð52Þ
5 ðGAÞeq
[46]. For partial shear connection, in the most of the design codes
the concept of the degree of shear connection g ¼ n=nf is associ- where ðGAÞeq denotes the elastic equivalent shear stiffness of com-
ated to a beam and used to compute the elastic stiffness and ulti- posite cross-section and 6/5 is the average shear strain coefficient
mate strength capacity. In the proposed approach this parameter is of a rectangular cross-section for a homogenous, single-layer
replaced with the elastic degree of composite action c and is com- beam. Hence the energy of the equivalent shear angle can be
puted based on the number of shear connectors with uniform spac- expressed as:
ing along the beam length and the constitutive law assumed for
Z Z
nonlinear behaviour of shear connectors and the material and geo- 1 L
1 L
6 T 2 ðxÞ
U¼ TðxÞweq ðxÞdx ¼ dx ð53Þ
metrical cross-sectional characteristics. It is worth noticing that 2 0 2 0 5 ðGAÞeq
while the current design codes allows only to evaluate the ultimate
strength capacities of the beams with partial shear-connection in On the other hand, the shear stresses, for steel sxy;s and concrete sxy;c
the proposed approach gradual yielding throughout the cross- components, assuming a uniform distribution both over the whole
section and along the member length can be described through width of the steel layer bs ðyÞ at the y level and the concrete layer
basic equilibrium, compatibility, material and shear connection bc ðy0 Þ at the y0 level, respectively, can be expressed as:
nonlinear constitutive equations, the states of strain, stress and " ! #
yield stress can be monitored explicitly during each step of the ðEIÞ0 Ashear;s ðyÞ Es Sshear;s ðyÞ
sxy;s ¼ TðxÞ 1 þ ð54:aÞ
analysis. ðEIÞel As rbs ðyÞ ðEIÞel bs ðyÞ

" ! #
ðEIÞ0 Ashear;c ðy0 Þ Ec Sshear;c ðy0 Þ
2.3. Evaluation of tangent stiffness and stress resultant sxy;c ¼ TðxÞ 1 þ ð54:bÞ
ðEIÞel Ac rbc ðy0 Þ ðEIÞel bc ðy0 Þ
In the above analysis procedures the integration process over The above relationships have been derived by solving the system
the concrete and steel sections is necessary. The quadrature (13) with N = 0 and assuming elastic behaviour of each layer (con-
method adopted in the current paper is the boundary integration crete and steel), which yields the following relationships for normal
method based on Green’s theorem in plane. This approach has been stresses in steel and concrete components:
proved to be extremely rapid and exact because stress integrals " #
need only be evaluated at a small number of points on the section ðEIÞ0 1 Es y
rs ¼ Es ðus þ uyÞ ¼ MðxÞ ð1  Þ þ ð55:aÞ
boundary [47,48]. According to Green’s integration formula the ðEIÞel As r ðEIÞel
domain integrals appearing in the evaluation of internal resultant
efforts and tangent stiffness matrix coefficients of the section can " #
be evaluated in terms of boundary integral. Noticing that the stress ðEIÞ0 1 Ec y0
rc ¼ Ec ðuc þ uy0 Þ ¼ MðxÞ ð1  Þ þ ð55:bÞ
field is always uniform in a direction parallel with neutral axis, ðEIÞel Ac r ðEIÞel
strains, stresses and tangent modulus of elasticity can be expressed
as function of the single coordinate (y), eðyÞ, rðyÞ, ET ðyÞ. The stres- then the shear stresses sxy;sðcÞ in Eq. (54) are readily derived using
ses are evaluated using the fibre strains and the assumed nonlinear Jouravski’s shear stress formula and considering that
constitutive relations. It is important to highlight that because the TðxÞ ¼ dMðxÞ=dx. In the above relationships, ðEIÞ0 represents the
stress field is defined usually by a piecewise function and there is elastic flexural rigidity of composite beam without composite
no continuity in the derivative, the interpolatory integration meth- action (Eq. (27)), ðEIÞel denotes the elastic flexural rigidity of com-
ods can produce important integration errors [49]. In the proposed posite beam with partial composite action (Eq. (30)), As , Es , Ac , Ec
approach, an adaptive strain-mapping quadrature strategy is used denote the cross-sectional area and Young’s modulus of elasticity
[50]. This procedure, combined with the Gauss–Lobatto integration for the steel and concrete layers, respectively, Sshear;sðcÞ represents
rule, can be successfully applied for each function part within each the first moment of the sheared area Ashear;sðcÞ of the steel and con-
side (Li) by dividing the interval of integration according with crete component, respectively, at the y(y0 ) coordinate where the
changes in the definition of the stress function, and then use the shear stresses are to be calculated.
86 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106

For each member (e) the effective degree of composite action γ eff (Eq. 49) or elastic degree
of composite action γ (for the first load increment) (Eq.42) is assumed to be known. Axial
force N for member is known. (e=1,NB).

Compute for each cross-section (control node) along the member length (n control nodes) the
bending moment M k (x k ) = M i ⎛⎜ k − 1⎞⎟ + M j k + qL k ⎛⎜ k − 1⎞⎟ (k=1,n) (Eq. 60)
2
x x x x
⎝ L ⎠ L 2 L ⎝ L ⎠

Cross-section analysis-full composite action

Compute the axial force in concrete slab under the assumption of full composite action
For each integration point (k=k+1) (monitored cross-section)

N cfM k (Eq.6) for combination (N and M k ) by solving the nonlinear system (5).

Compute the axial force in concrete slab under the assumption of full composite action
N cfM k + ΔM (Eq.6) for combination (N and M k +∆M) by solving the nonlinear system (5).

For each element (e=e+1)

M + ΔM
Determine the incremental axial force in concrete slab ΔN cf = N cf k − N cfM k .

M
( )
Compute the axial force in the concrete slab as a fraction of N cf k : N c = f γ eff N cf k , Eq.
M

(50).

Cross-section analysis-partial composite action

Solve the nonlinear system Eq. (13) with the unknowns uc, us and φ using the iterative
procedure described at section 2.2.2.1 (see Fig. 4).

Compute the cross-section tangent stiffness coefficients kij , Eq. A.1, from Appendix 1

Compute tangent flexural rigidity of the cross-section ( EI ) t ,k , Eq. (23) and tangent axial
rigidity ( EA) t ,k using the Eq. (31).

Compute the relative bending stiffnesses (EI )k , (EI )k and axial stiffness (EA)k , Eq. (48)
∞ 0 ∗

and the secant shear connection stiffness (ksec) using Eq. (46) or k sec = k 50% . Then compute
the parameter α k ( x k ) with Eq. (44).

Use the tangent flexural ( EI ) t and axial ( EA) t rigidities of each monitored cross-section to
form the element stiffness matrix and equivalent nodal loads, as a weighted sum of inelastic
responses of these cross-sections (see Section 2.5).

Compute (update) the new inelastic member effective degree of composite action by
applying the Gauss-Lobatto rule for integration of each parameter α k ( x k ) along the member
length, γ effnew with Eq. (49).

Fig. 7. Evaluation of the inelastic member effective degree of composite action and tangent stiffnesses.

The total shear strain energy of composite beam is determined Z " ! #2

ðEIÞ0 Ashear;s ðyÞ Es Sshear;s ðyÞ
explicitly as: ks ¼ As 1 þ dAs ð57:aÞ
As ðEIÞel As rbs ðyÞ ðEIÞel bs ðyÞ
Z Z
1 s2xy;s 1 s2xy;c
U¼ dV s þ dV c
2 Vs Gs 2 Vc Gc Z " ! #2
Z   ðEIÞ0 Ashear;c ðy0 Þ Ec Sshear;c ðy0 Þ
1 L
ks TðxÞ kc TðxÞ kc ¼ Ac 1 þ dAc ð57:bÞ
¼ TðxÞ þ dx ð56Þ Ac ðEIÞel Ac rbc ðy0 Þ ðEIÞel bc ðy0 Þ
2 0 Gs As Gc Ac

where Gs , Gc denote the concrete and steel shear modulus respec- When full composite action is assumed the above relationships can
tively, whereas the coefficients ks and kc are defined by the follow- be modified by simply replacing the elastic flexural rigidity of com-
ing expressions: posite beam with partial composite action ðEIÞel with the flexural
 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106 87

Z Z
rigidity of cross-section with full composite action ðEIÞ1 given by L
@ DMðxÞ DMðxÞ L
@ DTðxÞ DTðxÞ
Dhi ¼  dx þ  dx ð61Þ
the Eq. (38). 0 @ DMi EIt ðxÞ 0 @ DM i 56 ðGAeq Þ
The equivalent shear stiffness can be determined by expressing
the equality between the total shear strain energy of the equivalent Z Z
L
@ DMðxÞ DMðxÞ L
@ DTðxÞ DTðxÞ
(average) shear angle (Eq. (53)) and of the corresponding real situ- Dhj ¼  dx þ  dx ð62Þ
0 @ DM j EIt ðxÞ 0 @ DM j 56 ðGAeq Þ
ation (Eq. (56)). In this way the equivalent shear stiffness can be
defined as: or in a condensed matriceal form as:
 1   Z L
6 ks kc Dh i
ðGAÞeq ¼ þ ð58Þ Du r ¼ ¼
T
b ðxÞDeðxÞdx ð63Þ
5 Gs As Gc Ac Dhj 0

in which:
2.5. Inelastic member analysis 2 3
@ DMðxÞ @ DTðxÞ
@ DM i @ DM i
h iT
b ¼ 4 @ DMðxÞ 5;
T DMðxÞ DTðxÞ
@ DTðxÞ
DeT ðxÞ ¼ ½ Du Dw T ¼ EIt ðxÞ 5ðGA Þ
Flexibility-based method is used to formulate the distributed 6 eq
@ DM j @ DM j
plasticity model of a 2D beam-column element (6 DOF) under
the above assumptions. An element is represented by several cross ð64Þ
sections (i.e. stations) that are located at the numerical integration and where DeðxÞ represents the incremental deformation vector at
scheme points, as shown in Fig. 2a. The spread of inelastic zones current cross-section (x) along the element length. Assuming the
within an element is captured considering the variable section above expressions for the bending moments and shear forces along
flexural ðEIÞt and axial ðEAÞt rigidity along the member length (L), the element length, the relationship between nodal displacements
depending on the bending moments and axial force level, cross- (Dur ) and the nodal efforts (Dsr ) could be further expressed by
sectional shape and nonlinear constitutive relationships as already defining the cross-section tangent flexibility matrix (f st ), element
described. The transverse shear response is assumed linearly- flexibility matrix (f r ) and nodal displacements (Ddr ), given by the
elastic in the proposed approach and equivalent transverse shear uniform distributed loads, as:
stiffness ðGAÞeq for composite beam is used in the general force- 8 # 2 1 3 " 9
  <Z L " x 0
#
=
displacement relationship. Fig. 2b shows the deformed shape of a Dhi 1 1
EIt ðxÞ
x
1 x
¼ L L
4 5 L L
dx 
2D beam-column element in local system attached to the initially Dh j : 0 x 1 0 1
5ðGA Þ
1 1 ;
straight centre line, with the rigid body modes removed. Non-
L L 6 eq L L
ð65Þ
 
linear analysis by the proposed method requires incremental load- DM i
þ Dd r
ing, i.e. the inelastic behaviour is approximated by a series of elas- DM j
tic analyses. The element incremental flexibility matrix fr which
where the incremental nodal displacements (Ddr ) resulting from
relates the end incremental displacements Dur to the actions Dsr
loads acting along the member length can be expressed as:
can be derived by applying the Castigliano’s theorem for computa-
tion of generalized displacements. Assuming elastic behaviour Z " # 2 1 3 " #
1 0 DqL2 x x
5  2 L L  1 dx
L x 1
EIt ðxÞ
within a load increment, and no coupling of axial and flexural Dd r ¼ L
x
L
1
4 1
ð66Þ
responses at the element level, the generalized displacement at 0 L L
0 5ðGA Þ
DqL
2 xL  1
6 eq 2
the point i of the member produced by the force DP (DMi, DMj)
could be expressed as in Eq. (59) where DMðxÞ and DTðxÞ repre- or in a matriceal condensed form as:
sents the incremental bending moment and shear force respec- Dur ¼ f r Dsr þ Ddr ð67Þ
tively, at the current location x along the element length:
Z Z L where
@ DMðxÞ DMðxÞ
L
@ DTðxÞ DTðxÞ Z
Ddi;DP ¼  dx þ  dx ð59Þ L
0 @ DP EIt ðxÞ 0 @ DP 56 ðGAeq Þ fr ¼
T
Dsr ¼ ½ DM i DMj  ;
b f st Bdx;
T
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
bending moment component shear force component
0
Z L " #
T DqL2 b5 ðxÞ
The second term in Eq. (59) introduces the additional effect of Dd r ¼ b f st Dqdx; Dq ¼ ;
0 DqLb6 ðxÞ
transverse shear deformations where ðGAÞeq denotes the equivalent " #  
h i x
 1 xL b1 ðxÞ b2 ðxÞ
shear stiffness of the composite beam, evaluated as already 1 1
f st ¼ diag EIt ðxÞ 5ðGAeq Þ ; b ¼ L
¼
described in the previous section of this paper. Supposing that the 6 1
L
1
L b3 ðxÞ b4 ðxÞ
element is subjected to uniform distributed loads Dq the incremen- ð68Þ
tal bending moments and shear forces, in the first order theory (lin-
ear geometrically) can be expressed as functions of the incremental For instance the flexibility matrix f r could be detailed as, with
nodal element forces, and are given by the following expressions: n ¼ x=L:
x  2 3
x DqL2 x x  Z L 
b1 ðxÞ b3 ðxÞ
 1
0 
b1 ðxÞ b2 ðxÞ

DMðxÞ ¼ DMi  1 þ DM j þ 1 ð60:aÞ EIt ðxÞ
L L 2 L L fr ¼ 4 5
0b2 ðxÞ b4 ðxÞ 0 1
5ðGA Þ
eq
b3 ðxÞ b4 ðxÞ
1 DqL  x  6
@ DMðxÞ 1 2 2 2
3 ð69Þ
DTðxÞ ¼ ¼ DM i þ DM j þ 2 1 ð60:bÞ Z 1 b1 ðnLÞ b3 ðnLÞ
þ 5ðGA b1 ðnLÞb2 ðnLÞ
þ b3 ðnLÞb4 ðnLÞ
@x L L 2 L 6 EIt ðnLÞ eq Þ EIt ðnLÞ 5 ðGAeq Þ 7
dx ¼ L 4
6 6
5
b1 ðnLÞb2 ðnLÞ b3 ðnLÞb4 ðnLÞ b22 ðnLÞ b24 ðnLÞ
The total bending moments M(x) and shear forces T(x) associ- 0
EIt ðnLÞ
þ 5ðGA Þ EIt ðnLÞ
þ 5ðGA Þ
6 eq 6 eq
ated to the total nodal forces (Mi, Mj) and uniform distributed loads
(q) can be obtained by substituting the incremental values of the It is important to note that the effects of the shear deformations
nodal bending moments (DMi, DMj) and distributed load (Dq) with are taken into account both in the tangent stiffness matrix coeffi-
the corresponding total ones in the above relationships. Explicitly cients and equivalent nodal loads, but the response in shear is
for nodal displacements (rotations at ends i and j) (Dur ) the above assumed linearly-elastic in the present paper and, as a conse-
relationships could be expressed in a detailed manner as: quence, shear rigidity is constant along the member length. Under
88 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106

the above considerations and in order to simplify the expressions ~ rð33Þ Du

D~srð31Þ ¼ k ~ rð31Þ  Dq
~ rð31Þ ; ð78Þ
of the flexibility coefficients the following ‘‘correction coefficients”
are introduced: ~ rð31Þ is obtained by adding the axial deformation to the
where Du
Z 2 Z 2 ~ rð31Þ ¼ ½ Du
displacement vector Dur , Du Dur T , axial force is added
1
b1 ðnÞ 1
b2 ðnÞ
c1 ¼ 3 dn; c2 ¼ 3 dn;
0 EIt ðnLÞ=EIel 0 EIt ðnLÞ=EIel to vector Dsr resulting the vector D~srð3x1Þ ¼ ½ DN Dsr T and
Z ~ rð31Þ ¼ ½ 0 Dqr T . To include rigid body modes, the stiffness
1
b1 ðnÞb2 ðnÞ Dq
c3 ¼ 6 dn ð70Þ
0 EIt ðnLÞ=EIel matrix is pre- and post-multiplied by a transformation matrix to
result in the required 6  6 matrix [51]. A similar treatment is
which are numerically calculated by the Gauss-Lobatto integration
applied for incremental elasto-plastic equivalent nodal load vector
rule and in which EIel denotes the elastic flexural rigidity of compos-
following a detailed procedure given in [51]. As already mentioned,
ite beam. In this way the inelastic nonlinear force-based flexibility
to determine the bending moments along the member length, each
sub-matrices f ryðzÞ could be detailed as:
2 3 element, at each load increment, is divided in a few stations and the
L
c1 þ L5ðGA
1 L
c3 þ 3 L5ðGA
1 bending moments and shear forces in each station are computed by
3EIel eq Þ 6EIel eq Þ
f ryðzÞ ¼ 4 6 6 5 ð71Þ using Eq. (60) as functions of the total nodal element forces. Thus,
L
c3 þ L5ðGA
1 L
c2 þ L5ðGA
1
6EIel
6 eq Þ 3EIel
6 eq Þ the inelastic response of a beam-column element is typically com-
puted as the weighted sum of the response of a discrete number
To produce the deformational-stiffness relation, Eq. (67) is
of cross-sections by computing the element stiffness matrix by
inverted, resulting the following deformational-stiffness equation:
numerical integration (Eq. (74)). The cross-sections (i.e. stations)
Dsr ¼ kr Dur  Dqr ; are located at control points whose number and location depends
ð72Þ on the numerical integration scheme. In this work, the Gauss-
Dqr ¼ kr Ddr
Lobatto rule for element quadrature is adopted. Though this rule
where the vector Dqr is the incremental equivalent load vector, has a lower order of accuracy than customary Gauss-Legendre rule,
whereas kr represents the instantaneous element stiffness matrix it has integration points at each end of the element, where the plas-
of the beam-column element without rigid body modes, deter- tic deformation is important, and hence performs better in detect-
mined by matrix inversion of the flexural matrix fr: ing yielding [41]. If the state of forces at any cross-section along
1 the beam column element equals or exceeds the plastic section
kr ¼ fr ð73Þ
ð22Þ capacity (i.e. when the strain in the extreme fibre, attains the ulti-
where mate value), the flexural stiffness at the respective location
" 4EI 2EIel
# approaches zero or becomes negative when the strain-softening
el
a b
kr ¼ 2EIL L
4EIel
effect for the concrete in compression is taken into account. In order
ð22Þ el
b c to avoid numerical instabilities, for this cases, the sectional harden-
L L
 
3ðc2 þ 4t Þ 3 c3 þ 2t 3 c1 þ 4t ð74Þ ing is activated, thus a residual value of the tangent flexural rigidity
a¼ ; b¼ ; c¼ ; of the cross section is considered to be EIt =EIel ¼ 0:001.
ZþtS ZþtS ZþtS
12EIel
t¼ ; Z ¼ 4c1 c2  c23 ; S ¼ c1 þ c2  c3
GAeq L2 2.6. Force recovery and analysis algorithm
The resulting correction factors (Eq. (74)) are sub unities and
reflect the change in member rigidity as plastic zone develops In the incremental-iterative approach adopted in the current
along the member length. The elasto-plastic equivalent nodal paper, at each load increment, a Newton iterative method is
forces, transferred to the nodes, from the member loads, will not applied to compute the nonlinear load-deformation path. A predic-
be constant during the analysis, and will be dependent on the vari- tor/corrector solution scheme has been implemented to find the
able flexural rigidity along the member according with the process nodal displacements and element resisting forces under a given
of gradual formation of plastic zones. The incremental equivalent set of applied loads. Using an updated Lagrangian formulation
load vector Dqr in this case can be expressed as: (UL), the global geometrical effects are considered, updating the
2 3 element forces and the transformation matrix corresponding to
DqL2
ð48ac4 þ 24bc5 Þ equations from local to global coordinates, at each load increment
Dqr ¼ 4 12 5 ð75Þ
DqL2
ð48cc5 þ 24bc4 Þ [51]. The proposed approach, for the element force recovery con-
12
sists in several steps that are briefly described next for a current
where the correction coefficients in this case are: iteration ‘‘j” The proposed approach, which is based on the flexibil-
Z Z ity method, differs from the previous numerical strategies [41] in
1
b1 ðnÞb5 ðnÞ 1
b2 ðnÞb5 ðnÞ
c4 ¼ dn; c5 ¼ dn ð76Þ that the section forces equilibrium is enforced during local itera-
0 EIt ðnLÞ=EIel 0 EIt ðnLÞ=EIel
tion at the element level but the residual element displacements
The resulting stiffness matrix is a 2  2 matrix, and does not include are dissipated at the global iteration phase. The first step consists
axial degree of freedom. Axial stiffness component are then added in determination of the incremental nodal element forces
to result in the required 3  3 stiffness matrix: (D~srð31Þ
j
) from the current element displacement increments using
2 EA 3
el
cx 0 the element stiffness matrix and then the total element nodal
L
~r 6 ð12Þ 7 ~j
forces are determined (S ):
k ¼4 5 ð77Þ rð31Þ
ð33Þ 0 kr
ð21Þ ð22Þ
D~srð31Þ
j
¼k~ j1 Du ~ rð31Þ
j
~ j1
 Dq
R1 EAt ðnLÞ
rð33Þ rð31Þ
ð79Þ
where cx ¼ 0 EAel
dn introduces the effects of variable axial rigid- ~j
S ¼S~j1 þ D~s j
rð31Þ rð31Þ rð31Þ
ity along the member lengths. In the same way the relationship
between nodal displacements (Du ~ r ) and the nodal efforts (D~sr ) con- Starting from the element nodal forces the force field is gener-
sidering all three degrees of freedom of the element represented in ated along the element length then the strain field
local system with the rigid body modes removed (Fig. 2b), could be e~j ðxÞ ¼ ½ e0 u w T (cross-section deformations) is computed
expressed as: from the given section forces by using the iterative procedure
 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106 89

described in Section 2.2.2.1 and detailed in the flowchart depicted stiffness matrix. The structure resisting forces are compared with
in Fig. 4. The axial strain is evaluated as: e0 ¼ N j =EAt ðxÞ where the the total applied load and if the differences are not within the spec-
axial rigidity EAt ðxÞ and curvature u of current cross-section are ified tolerance, a new global iteration begins.
evaluated using the iterative procedure described at Section 2.2.2.1.
The shear angle w ¼ 6=5TðxÞ=GAeq is computed assuming the 3. Advanced finite element modelling using Abaqus software
response in shear linearly-elastic and the equivalent shear stiffness
GAeq of the beam. In this way the tangent flexural and axial rigidity In this study, the advanced numerical simulation is conducted
of monitored cross-sections are evaluated and then element tan- with ABAQUS v.6.11 [42] software, which is able to take into
gent stiffness matrix and equivalent nodal loads are updated. account most of the factors that govern the structural behaviour
Residual element displacements (du ~ j ) are computed by integrating of steel-concrete composite beams with full and partial composite
the incremental strain field along the element length. The incre- action. The numerical model described here consists in a combina-
mental strain field De ~j ðxÞ ¼ e
~j ðxÞ  e
~j1 ðxÞ is integrated along the tion of 3-D solid elements (for concrete volumes), shell elements
element to obtain the incremental displacement vector such that for steel elements and linear truss elements for reinforcing bars
the compatibility condition is satisfied along the element: as shown in Fig. 8a. Eight-node brick elements, C3D8, are picked
2 3 from element library to model the concrete slabs, this element
Z L 1 0
~ T De ~¼4 ð12Þ
5 being able to successfully manage the hourglassing problems
~ rjðeÞ ¼
Du b ~j ðxÞ; b ð80Þ
0 0 b which may arise in high stress concentration regions. The struc-
ð21Þ ð22Þ
tural steel is represented by means of four-node, general purpose
The presence of residual element displacements violates the com- shell element S4R, which use reduced integration (one integration
patibility of the element. The residual displacements are the errors point) to form the stiffness matrix and takes into account the effect
made in the linearization process of the element force–deflection of shear deformations, thus providing accurate solutions to both
relationship: thin and thick shell problems. The reinforcing bars are defined as
two-node three-dimensional linear truss elements, T3D2, which
~ j ¼ Du
du ~ rj  Du
~ rjðeÞ ð81Þ assume linear interpolation and constant stress over the element
Then the residual nodal forces are computed based on the updated length. The steel joist - concrete slab connection is provided by
element tangent stiffness matrix. In order to restore the element special purpose connector elements CONN3D2 which are discretely
placed at specified spacing. In this study, Cartesian + Align connec-
compatibility residual nodal forces (corrective forces) d~f must be j
tor type is selected, hence the connector second’s node can inde-
applied at the ends of the element:
pendently translate in three local Cartesian directions while the
d~f j ¼ k
~ j du
r
~j ð82Þ rotations are prohibited. For the two in-plane translational direc-
tions the relative behaviour is defined by a force-displacement
These residual nodal forces are used to determine the element constitutive law, while a rigid connection is assumed in a direction
resisting forces in natural coordinates systems and then are trans- normal to the steel-concrete interface (the uplift is neglected).
formed and assembled in the local element coordinates system: The constitutive law, that describes the in-plane translational
~j ~j1 ~j ~ behaviour, is usually obtained from experimental push-out tests
Srð31Þ ¼ Srð31Þ þ Dsrð31Þ þ df jð31Þ ð83Þ
on shear studs, or, alternatively, when experimental data are not
~ available, it can be described by Ollgaard’s [45] model given by
~j ~j1 T
~j
Sð61Þ ¼ Sð61Þ þ Tð63Þ ðDsrð31Þ þ df jð31Þ Þ þ Dqð61Þ ð84Þ the following exponential function:
where Dqð61Þ represents a vector resulted from the equilibrium a
P ¼ Psc ð1  ebs Þ ð85Þ
conditions and depends on the type of element loads to be
accounted for and T represents the kinematic transformation matrix where Psc represents the ultimate shear capacity of a single shear
given in [51]. The element resisting forces are finally computed for stud, which is experimentally obtained or, is evaluated according
each element and they are transformed and assembled in the global to Eurocode 4 provisions [46], s refers to the relative slip between
equilibrium equations and accounted for during the global iteration the two components of the composite beam while a and b are
process. The element resisting forces are only computed once for parameters that describe the curve’s shape, namely they control
each global iteration, as in the case of the conventional stiffness- the slope (stiffness) of the curve. Fig. 8b depict the constitutive
based element. The element force vectors are assembled to form curves drawn with the parameters mentioned by Ollgaard et al.
the updated structure resisting forces. The new structure stiffness [45], Aribert et al. [52] and Johnson et al. [53]. As can be seen, the
matrix is formed by assembling the current (updated) element curves are highly nonlinear and greatly influenced by the shape

Fig. 8. (a) Advanced FEM modelling in Abaqus and (b) nonlinear constitutive law for the shear connection.
90 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106
parameters but all of them converge asymptotically to the ultimate
shear capacity as the slip increases. The mesh size significantly
influences the computational effort and the accuracy of the results.
Concerning this, sensitivity studies have indicated that mesh
dimensions aspect ratio close to 1.0 and an average size of 50 mm
provides accurate results. The structural steel and reinforcing steel
is modelled as an elasto-plastic-linear hardening material, using the
classical metal plasticity model which use Mises yield surface and
allow for isotropic hardening behaviour. The elastic branch of
concrete under compression is defined by Young’s modulus and
Poisson’s coefficient and its limit stress is chosen as 0:4f cm where
f cm is taken as the actual cylinder strength test value. The inelastic
behaviour of concrete is defined with CDP-Concrete Damaged Plastic-
ity model which assumes that the two main failure mechanisms are
tensile cracking and compression crushing of concrete material. The
analyses are conducted with the nonlinear static arc-length based
Riks procedure. This method can provide solutions even in cases
of complex, unstable response or, where, the load-displacement
response exhibit negative stiffness [42]. In order to establish the
ultimate load capacity two failure criteria have been considered.
The developed model is able to predict the failure modes associated
with either slab crushing of the concrete in compression, which is
assumed when the ultimate compressive strain is reached, or stud
failure which is assumed when the composite beam’s most heavily
loaded stud reached its ultimate load.
4. Computational examples
The accuracy of the analytic procedure and the computer pro-
gram developed here (NEFCAD) has been extensively tested
through comparisons with experimental and numerical studies
retrieved from international literature. In this paper, several repre-
sentative benchmark problems are selected and discussed to vali-
date and assess the accuracy and effectiveness of the proposed
approach. The obtained results are compared with experimental
data, advanced three-dimensional FEM model developed by the
authors in Abaqus software package and other results reported
by researchers using independent finite element solutions. The
influence of both partial composite action and shear-flexibility of
the composite beams is studied.
4.1. Elastic analysis of simply supported beam. The effect of shear-
flexibility
The proposed model in conjunction with the equivalent trans-
verse shear stiffness derived here is used to predict the elastic
Fig. 9. (a) Simply supported beam layout and (b) load-deflection curves.
deflection of simply-supported composite beam tested by Aribert
et al. [54,32]. The geometry, section properties, loading and the
main material and shear connection properties are given in [32]
(Fig. 9a). The shear connection stiffness is k = 450 MPa resulting
the equivalent transverse shear stiffness (GA)eq = 4.38  105 kN. In
order to compare the performance of the proposed model against
the existing models the results retrieved from [32] has been used.
In [32] a closed-form solution of the governing equations describ-
ing the behaviour of shear-deformable two-layer beam with partial
interaction has been presented and the corresponding ‘‘exact” stiff-
ness matrix has been derived and implemented in a general dis-
placement based F.E. code. Fig. 9b shows the comparative load
deflection curves underneath the load point with and without con-
sidering the effect of shear-flexibility. The experimental results
under the considered load level are also depicted in Fig. 9b. As it
can be seen the proposed model agrees very well with those pub-
lished in [32]. It can be observed also that the both theoretical
approaches are slightly stiffer than the experimental response
but the models based on Timoshenko’s beam theory are closer to
the experimental data. A parametric analysis has been carried
out to study the influence of both shear flexibility and partial com-
posite action on the global behaviour of composite beam.
Fig. 10 shows the comparative mid-span deflections ratio
obtained with the proposed model and compared against those
obtained in [32] for different span-to-depth ratios (L/H) and shear
connection stiffness (k = 1, 100, 1000, 10,000 MPa). As expected,
the deflection predicted by the proposed model when shear flexi-
bility effect is considered (Timoshenko’s model-dT) is larger than
the corresponding one evaluated according to Bernoulli’s model
(dEB), for any value of the span-length-to-beam-depth ratio L/H.
As it can be seen the proposed results are in close agreement with
those published in [32]. These results confirms also that shear-
flexibility plays a more important role in the case of low L/H ratios
and is generally more important for composite beams character-
ized by substantial shear interaction.
4.2. Simply supported beams tested by Chapman and Balakrishnan
The proposed numerical model is validated by comparisons
against Chapman and Balakrishnan [55] experimental tests on sev-
eral simply supported composite beams, as well as against the
results predicted by ABAQUS software and those published by
other authors [3,17,56].
The geometry and section properties of the analysed beams are
depicted in Fig. 11 and the main material and shear connection
properties are given in Table 1. The nonlinear constitutive material
 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106 91

Fig. 10. Midspan deflection ratio versus the span-to-depth ratio for different shear connection stiffness.

Fig. 11. Simply supported beams layout.

relationships described at Section 2.1 of this paper have been used
in the proposed approach. The concrete elastic modulus is evalu-
ated according to the CEB-FIB Model Code 90 [43]. Due to the fact
that in the proposed approach the key elements of the elasto-
plastic formulation are established at the cross-sectional level, the
first set of tests is conducted at the cross-section level. In order to
prove the reliability, efficiency and numerical stability of the pro-
posed approach several moment-curvature analyses have been per-
formed considering the material and shear connection properties of
beam cross-sections (A3, A5 and A6) used in the experimental tests
by Chapman and Balakrishnan [55]. The connector’s behaviour was
defined using the experimental load-slip curve retrieved from [55]
and assuming a constant value for secant stiffness Ksc,50% for each
specimen as depicted in Table 1. Based on these values the corre-
sponding shear connections stiffnesses k50% and elastic degrees of
composite action c have been computed (Table 1). The results of
the proposed approach for the moment-curvature response are
compared with test results in Fig. 12. For the specimens A5 and
A6 the results are also compared with an analytical model
developed in [56] for elasto-plastic cross-sectional behaviour of
composite beams with partial shear-connection.
In [56] the partial shear connection is defined at cross-sectional
level assuming the slip strain at the steel-concrete interface to be a
function of the degree of shear connection. As can be seen the
agreement between the proposed approach, test results [55] and
analytical results from [56] is very good. The convergent rate of
the proposed approach is very good as usually convergence is
achieved within only few iterations. For instance considering the
A3 cross-section subjected to a bending moment M = 550 kN m
the equilibrium is achieved after only six iterations despite the fact
that the iterative process has been started with the initial values of
axial strain in concrete and steel and curvature set to zero and a
restrictive equilibrium tolerance (i.e. tol = 1E4), has been consid-
ered. Table 2 depicts the iterative procedure applied in this case
showing the evolution of the main parameters involved in the
inelastic analysis of cross-section.
In the following a comprehensive parametric and sensitivity
analysis of cross-sectional behaviour is presented. Fig. 13a shows
92 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106

Table 1
Material properties for A3, A5, A6 and E1.

Beams
A3 A5 A6 E1
Concrete Compressive strength (MPa) 18.27 24.75 23.72 32.68
Young Modulus (MPa) 26,284 29,082 28,673 31,905
Steel Yield stress (MPa) Flange 256.2 235.5 240.2 249.4
Web 307.3 295 267.7 296.5
Ultimate stress (MPa) Flange 454.1 451 423.2 464.9
Web 451 451 466.4 460.2
Young modulus (MPa) Flange 203,864 206,953 206,953 208,497
Web 197,686 199,231 214,675 202,320
Strain-hardening strain Flange 0.0013 0.0011 0.0020 0.0026
Web 0.0016 0.0015 0.0019 0.0015
Strain-hardening modulus (MPa) Flange 3500 3500 3500 3500
Web 3500 3500 3500 3500
Longitudinal reinforcement 8 mm diameter at 304.8 mm at top and bottom of slab
Yield stress (MPa) 320
Ultimate stress (MPa) 320
Young modulus (MPa) 205,000
Transverse reinforcement 12.7 mm diameter at 152.4 mm at top of slab
12.7 mm diameter at 304.8 mm at bottom of slab
Shear connection Number of studs 68 44 32 100
Number of rows@Spacing (mm) 2@180 2@285 2@395 2@121
Ultimate load of connectors (kN) 126.5 118.5 124.5 64.67
Secant connector stiffness Ksc.50% (N/mm) 221,577 237,511 136,082 122,701
Shear connection stiffness k50% (N/mm2) 2462 1667 689 2028
Degree of composite action c 12.94 10.45 6.70 11.35
Function of the degree of composite action f(c) 0.944 0.917 0.820 0.929

Constitutive models

moment-curvature response for beam cross-section A5 by means
of adopting different values for secant stiffness Ksc,50% associated
to the connector’s experimental load-slip curves (obtained from
push-out tests for specimens PA2, PA4 and PA5 [55]) for the shear
connector representation. As it can be seen, the stiffness of connec-
tors appears to have a small influence on overall behaviour of the
cross-sections. Fig. 13b shows moment-curvature response of
cross-section A5 by means of adopting different values for concrete
strength. As it can be seen, increasing the concrete strength the
corresponding moment-curvature curve becomes stiffer, but this
effect is not very significant. Fig. 13c shows that the steel ratio
between the strain at strain hardening and yield strain did not sig-
nificantly affect the overall response of the composite cross-
section. Figs. 14 and 15 show the cross-sectional response for spec-
imens A5 and A6 respectively for a fixed slab concrete strength and
adopting different values for either the flanges or the web yield
stresses. As it can be seen, for the composite cross-sections anal-
ysed, the strength properties for the steel profile, significantly
affect the overall response of the composite cross-section. In the
following the proposed moment-curvature analysis is used to
assess the influence of the effects of partial shear connection.
Fig. 16 shows the comparative bending moment-curvature and
bending moment-tangent flexural rigidity diagrams of cross-
section A6 considering different values for the degree of composite
action by means of adopting different values for the function f ðcÞ
ranging from f ðcÞ ¼ 1 (i.e. full composite action) and f ðcÞ ¼ 0 (i.e.
no composite action). It can be observed that, as expected, on
decreasing the level of shear connection (i.e. decreasing the values
of f ðcÞ) the strength and stiffness of the cross-section is reduced.
However, the decrease in the ultimate strength capacity did not
seem to be as significant as the decrease in the tangent flexural
rigidity. Based on the outcomes of the sensitivity analysis carried
above, concerning the assessment of the influence of small varia-
tions in concrete and steel material properties, it can be concluded
that the strength properties of steel profile are the key structural
parameters in the definition of the overall shape of the moment–
curvature curve of the composite cross-section analysed.
Furthermore these parametric tests proves the numerical stabil-
ity and sensitivity of the proposed approach to several parametric
changes and the proposed approach could represents an useful tool
for such qualitative analyses of composite steel-concrete cross-
sections with partial composite action.
 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106 93

Fig. 12. Comparison with test results for moment-curvature response: (a) A3; (b) A5; and (c) A6.

Table 2
Main parameters involved in the iterative process of A3 cross-section.
pffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Iteration Axial strain concrete (uc  104) Axial strain steel (us  103) Curvature (/  104 cm1) Convergence error FT F= f
extT ext
f
Initial 0 0 0 1
1 6.680 1.0877 0.4277 0.1906
2 8.351 2.093 1.220 4.946E2
3 9.409 3.056 1.877 1.449E3
4 9.685 3.988 2.497 3.022E3
5 9.872 4.409 2.793 7.798E4
6 9.911 4.459 2.829 1.0993E5

Fig. 13. (a) Influence of shear connector’s stiffness. (b) Influence of concrete strengths. (c) Influence of the ratio of strains at strain hardening and at yield.
94 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106

Fig. 14. Influence of the yield stress of the web (a) and flanges (b) for A5 beam cross-section.

Fig. 15. Influence of the yield stress of the web (a) and flanges (b) for A6 beam cross-section.

Fig. 16. Effect of degree of composite action: (a) moment-curvature response; and (b) moment-tangent flexural rigidities.

The following set of tests refers to the global member analysis
and in this respect the nonlinear behaviour of beam E1 is dis-
cussed. In the finite element model developed in Abaqus as well
as in the proposed approach, the connector’s behaviour was
defined using the experimental load-slip curve retrieved from
[55]. In the proposed procedure, a value of 122,701 N/mm was
assumed for the secant stiffness Ksc,50% while the corresponding
shear connection stiffness k50% is 2028 N/mm2 resulting the elastic
degree of composite action c = 11.35 and the equivalent shear stiff-
ness (GA)eq = 8.13  105 kN. The effect of overhang region (region
beyond supports) is ignored in all the analyses discussed in the fol-
lowing. In the proposed approach, one half of the beam is modelled
by a single inelastic flexibility- based element with 7 Gauss-
Lobatto integration points along the element length. It is worth
noting that since the proposed model is based on the flexibility
approach considering two or more elements per physical member
does not improve significantly the accuracy of results. Moreover
the necessity to divide the beam into two elements is due only
to have a node at mid-span of the beam where load-deflection
curve is explicitly monitored. Besides, for this computational
example, increasing the number of the integration points, per
one half of the beam, beyond the number mentioned above exerts
negligible effects on the accuracy of the results. The mid-span
deflection vs. applied load comparative curves is plotted in
Fig. 17. As it can be seen the behaviour of the composite beam pre-
dicted by the proposed analysis procedure and the advanced three-
dimensional finite element model implemented in Abaqus,
described in Section 3, is in close agreement with that of experi-
mental test. Fig. 17 indicates that the proposed model successfully
predicts the elastic part stiffness and the ultimate load capacity of
the composite beam, while a slightly stiffer behaviour is observed
on the first part of the post-elastic portion of the curve. The results
are also consistent with those published by Queiroz et al. [3] and
El-Lobody and Lam [17], in which the nonlinear behaviour of all
components of the cross-section is explicitly modelled by using
advanced three-dimensional finite element approaches.
The results obtained using the proposed approach when the
influence of the shear deformability of the beam is ignored is also
 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106
Fig. 17. Load vs. mid-span deflection for simply supported beam E1.
depicted in Fig. 17. As it can be seen, when the influence of shear
deformability of the beam is ignored, small but not negligible dif-
ferences in overall beam stiffness has been obtained comparing
with the case when the shear deformability is taken into account.
As expected, in this case, the model produces a slightly stiffer
response. Fig. 18 presents the comparative load-deflection curves
obtained by the proposed approach when the full composite action
is assumed and those obtained with the advanced FEM model con-
sidering different higher levels of shear connections. In this study,
the levels ranging from 100% to 136% are analysed using advanced
three-dimensional FEM model, by varying the number of shear
connectors. It can be observed that, as expected, increasing the
level of shear connection the system became more rigid with
increased strength and stiffness and the inelastic behaviour
becomes similar with the response predicted by the proposed
approach (full composite action). However, this study shows that
the effects of partial shear connection (PSC) with higher levels of
the degree of shear connection can be neglected, in this case, for
levels above 100% as no significant improvement in terms of either
stiffness or strength of the beam was observed.
The effectiveness of the proposed procedure is further assessed
by varying the degree of shear connection (by means of varying the
number of shear connectors-partial shear connection) and compar-
ing the predicted curves with those obtained with more complex
finite element analysis, as shown in Figs. 19–22. Two load cases
are considered: mid-span point load and uniformly distributed
Fig. 18. Load-deflection curves for different higher levels of shear connection.
95
load [3]. The levels of shear connection, ranging from 47% to
136%, are determined using the material and shear connector prop-
erties retrieved from experimental procedure as in [3]. For mid-
span point load case, the full load-mid-span deflection curves for
all cases having different level of shear connection are shown in
Fig. 19. In order to illustrate the effect of the degree of shear con-
nection over initial stiffness of the beam, and to compare the
results obtained by the proposed approach with those given in
[3], Fig. 20 presents these curves just for the initial branch up to
300 kN. As expected on decreasing the level of shear connection
the beam became more flexible, with reduced strength and stiff-
ness. For uniformly distributed load case, the full load vs. mid-
span deflection curves are shown in Fig. 21 whereas the initial
branches of the curves are depicted in Fig. 22 for a load level up
to 100 kN/m. By comparing the load-deflection curves depicted
in Figs. 19–22, it can be observed, that the proposed method (Nef-
cad) predicts fairly well the nonlinear behaviour and ultimate load
capacity of the system when lower levels of shear connection are
considered.
4.3. Continuous composite beams tested by Ansourian
Two-span continuous beam experimentally tested by Ansourian
[57] were loaded by a point load applied at the centre of the left
span while the right span remained unloaded (CTB1) and also
another loading scheme assume point loads applied at the centres
of both spans (CTB4 and CTB5). Their layouts, loading conditions
and main material properties are shown in Fig. 23 and Table 3
respectively.
The nonlinear constitutive material relationships described at
Section 2.1 of this paper have been used in the proposed approach.
The concrete elastic modulus is evaluated according to the CEB-FIB
Model Code 90 [43]. In the finite element model developed in Aba-
qus as well as in the proposed approach, the connector’s behaviour
was described by Olgaard’s constitutive model using the shape
parameters reported in [58], a = 0.85 and b = 1.2 mm1. The values
assumed in the proposed procedure for the secant stiffness Ksc,50%,
shear connection stiffness k50%, elastic degree of composite action c
and equivalent shear stiffness (GA)eq respectively, are presented in
Table 3 for each specimen. Zona and Ranzi [1] developed advanced
1-D finite element models for the nonlinear analysis of composite
members with partial shear connection in combined bending and
shear. Three different beam models are derived in [1] by coupling
with a deformable shear connection two Euler-Bernoulli (EB-EB)
96 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106
Fig. 19. Load-deflection curve - Mid-span concentrated load.
Fig. 20. Load-deflection curve. Initial stiffness - Mid-span concentrated load.
beams for concrete slab and steel beam respectively, an
Euler-Bernoulli beam for concrete slab to a Timoshenko beam for
steel beam (EB-T) and two Timoshenko beams for both concrete
slab and steel beams (T-T). In these approaches shear failure mode
and shear deformability have been included explicitly in the
Fig. 21. Load-deflection curve - Uniformly distributed load.
136%
118%
100%
89%
71%
47%
formulation of the Timoshenko beam. The mid-span deflection
vs. applied load comparative curves are plotted in Figs. 24 and 25
for both beams CTB1 (Fig. 24) and CTB4 (Fig. 25).
In the proposed approach, one half of the beam is modelled by a
single inelastic flexibility-based element with 7 Gauss-Lobatto
integration points along the element length. As it can be seen the
behaviour of the composite beams predicted by the proposed anal-
ysis procedure is in close agreement with that of experimental
tests, advanced three-dimensional finite element model imple-
mented in Abaqus, and the advanced finite element models pro-
posed in [1].
However, the proposed approach gives a slightly stiffer
response as compared with those obtained in [1] when the model
couples two Timoshenko beams both for steel profile and rein-
forced concrete slab beam. The reference [1] reported very small
differences between EB-T model and T-T model respectively;
therefore this result is not depicted in Figs. 24 and 25. This slightly
stiffer response could be explained by the fact that in our approach
the response in shear is assumed to be linear-elastic (no shear
degradation is allowed) whereas in the advanced finite element
model proposed in [1] different shear strengths are used in the sag-
ging and hogging regions, i.e., for the slab in compression and for
the slab in traction, respectively. However, when the equivalent
136%
118%
100%
89%
71%
47%
 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106
Fig. 22. Load-deflection curve. Initial stiffness - Uniformly distributed load.
shear stiffness in the hogging regions is computed taken into
account only the contribution of the steel beam ðGAÞeq ¼ GAsteel eq
the differences between our model and T-T model reduced signif-
icantly (Figs. 26 and 27). The hogging regions are assumed to be
15% of the span length on each side of the internal support for
beam CTB4 and 15% of the first span length in addition to the sec-
ond span length for CTB1. In the proposed model the sagging and
hogging regions are modelled by using two elements and one ele-
ment respectively for each beam with 5 integration points along
the elements length.
Figs. 26 and 27 report the comparative force and mid-span
deflection relationships when the effect of shear deformability is
taken into account in different ways: (a) when the equivalent shear
stiffness is computed for actual degree of composite action c of the
beam (ðGAÞeq ¼ ðGAÞceq ) both in hogging and sagging regions and (b)
when the equivalent shear stiffness is computed taken into account
only the contribution of the steel beam (ðGAÞeq ¼ GAsteel
eq ) in the hog-
ging regions whereas ðGAÞeq ¼ ðGAÞceq for sagging regions. All these
curves are compared with the case when the effect of shear
deformability of the composite beam is ignored and with those
reported in [1] and [57]. No significant differences have been
obtained between the case (a) and when the equivalent shear stiff-
Fig. 23. Ansourian’s continuous beams layout.
97
ness is computed for the case of full composite action of the beam
ðGAÞeq ¼ ðGAÞ1 eq assumed both in sagging and hogging regions;
therefore this result is not depicted in Figs. 26 and 27. As it can be
seen the effect of shear deformability is small but not negligible
in evaluation of the overall beam stiffness. Moreover, this compar-
ative study confirms that, in a conservative approach, the shear
deformability of the concrete slab maybe neglected in the hogging
regions in the evaluation of the overall shear stiffness of the com-
posite steel-concrete beam. This study is extended in Fig. 28 where
the combined effect of partial shear connection (i.e. assuming dif-
ferent number of shear studs) and transverse shear deformability
of composite beams is presented. In these analyses different equiv-
alent shear stiffnesses have been considered in hogging and sagging
regions respectively according with the case (b) defined above.
As expected, the deflection predicted by the proposed model
when shear flexibility effect is considered is larger than the corre-
sponding one evaluated according to Bernoulli’s model for any level
of the shear connection, ranging from full composite action to sim-
ple bare steel beams, and this effect is generally more important for
composite beams characterized by substantial shear interaction.
Ranzi and Bradford [2] developed an advanced 1-D finite element
model for the nonlinear behaviour of composite steel-concrete
beams with partial shear interaction. The model explicitly accounts
of the slip between concrete slab and steel profile and consider non-
linear constitutive law for the shear connection.
However in order to achieve convergence the model require
many finite elements to model one beam. Fig. 29 presents the com-
parative results in terms of the mid-span deflection vs. applied
load, for the beam CTB5, obtained by the FEM model developed
in [2], experimental results [57] and those obtained with the pro-
posed approach and advanced three dimensional finite element
model (Abaqus) developed in the present paper. As it can be seen
the present method accurately predicts the nonlinear behaviour
and ultimate load capacity.
Fig. 30 presents the combined effect of partial shear connection
and transverse shear deformability for beam CTB5. Once again it is
proved that the deflection predicted by the proposed model when
shear flexibility effect is considered is larger than the correspond-
ing one when this effect is ignored for any level of the shear con-
nection, and this effect is generally more important for
composite beams characterized by substantial shear interaction.
98 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106

Table 3
Material properties for CTB1, CTB4, CTB5.

Beams
CTB1 CTB4 CTB5
Concrete Compressive strength (MPa) 24.6 27.9 23
Young modulus (MPa) 31,879 32,920 31,349
Steel Yield stress (MPa) Flange 277 236 265
Web 340 238 278
Ultimate stress (MPa) Flange 421 393 442
Web 440 401 428
Young modulus (MPa) Flange 206,000
Web 206,000
Strain-hardening strain 0.012 0.018 0.019
Strain-hardening modulus (MPa) 6000 3000 4800
Longitudinal reinforcement Hogging top/bottom (mm2) 800/316 804/767 1260/470
Sagging top/bottom (mm2) –/160 160/160 320/160
Yield stress (MPa) 430
Ultimate stress (MPa) 533
Young Modulus (MPa) 206,000
Strain-hardening strain 0.012 0.018 0.019
Strain-hardening modulus (MPa) 6000 3000 4800
Transverse reinforcement 10 mm diameter at 100 mm at bottom of slab
Shear connection Number of 19  75 studs 66 84 60
Number of rows@Spacing (mm) 2@280 3@330 2@300
Ultimate load of connectors (kN) 110 110 110
Secant connector stiffness Ksc.50% (N/mm) 112,980 112,980 112,980
Shear connection stiffness k50% (N/mm2) 807 1027 753.2
Degree of composite action c 8.78/10.97 8.61 7.99
Function of the degree of composite action f(c) 0.886/0.924 0.882 0.866
Equivalent shear stiffness (GA)eq (105 kN) 3.144/3.178 2.748 3.644

Constitutive models

Fig. 24. Load versus mid-span deflection curves for beam CTB1 tested by Ansourian. Fig. 25. Load versus mid-span deflection curves for beam CTB4 tested by Ansourian.
 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106
Fig. 26. Load versus mid-span deflection curves for beam CTB1. The effect of shear
deformability.
Fig. 27. Load versus mid-span deflection curves for beam CTB4. Shear deformability
effects.
Fig. 28. The combined effect of partial composite action and transverse shear deformability for beam CTB4.
99
4.4. Simply supported beam tested by Fabbrocino and Pecce
Beam Type C is part of the experimental study reported by
Fabbrocino and Pecce [59] which was aimed to investigate the
structural behaviour of steel-concrete composite beams under
negative (hogging) bending moment. The beam is simply sup-
ported with a span length of 3610 mm made of 800 mm wide
and 120 mm thick concrete slab longitudinally reinforced by 4
steel bars of 14 mm diameter at the centre. The steel section of
beam is HEB 180 and is loaded by a concentrated force at mid-
span. The beam layout and the main material properties are
depicted in Fig. 31 and Table 4 respectively. The nonlinear consti-
tutive material relationships described at Section 2.1 of this paper
have been used in the proposed approach. The concrete elastic
modulus is evaluated according to the CEB-FIB Model Code 90
[43].
In the finite element model developed in Abaqus as well as in
the proposed approach, the connector’s behaviour was described
by Olgaard’s constitutive model using the shape parameters pro-
vided by Aribert [52], a = 0.8 and b = 0.7 mm1. Furthermore, in
the proposed procedure, a value of 47,302 N/mm was assumed
for the secant stiffness Ksc,50% while the corresponding shear con-
nection stiffness k50% is 92 N/mm2 resulting the elastic degree of
composite action c = 1.94, and the equivalent shear stiffness
(GA)eq = 3.35  105 kN. In the proposed approach, one half of the
beam is modelled by a single inelastic flexibility- based element
with 7 Gauss-Lobatto integration points along the element length.
The aim of this study is to assess the capability of the proposed
nonlinear analysis to satisfactorily predict the structural behaviour
of composite beams with partial composite action subjected to
negative (hogging) bending moment. The proposed numerical
model is validated by comparisons against experimental results
[59] as well against the results predicted by ABAQUS software
and those published by Nguyen et al. [26]. The force and mid-
span deflection relationships and the load versus rotation curves
at the support are shown in Figs. 32 and 33 respectively. The effect
of shear deformability of composite beam is also reported when
equivalent shear stiffness is considered for partial composite action
100 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106

Fig. 29. Load versus mid-span deflection curves for beam CTB5 tested by Ansourian.

Fig. 30. The combined effect of partial composite action and transverse shear deformability for beam CTB5.

Fig. 31. Simply supported beam Type C layout.

and for the case when equivalent shear stiffness is computed by
neglecting the concrete contribution. The case when shear
deformability is ignored is also depicted in Fig. 32. It can be seen
that the global behaviour of the beam both in elastic and post-
elastic field as well as the ultimate loading capacity predicted by
the proposed approach, correlate reasonable well with the experi-
mental results and advanced mixed finite element model devel-
oped in [26]. When the shear deformability is ignored or when
composite action is considered in evaluation of the equivalent
shear stiffness of the beam, the proposed approach gives a slightly
stiffer response as compared with experimental and advanced
finite element models.
When the shear stiffness is computed considering only the con-
tribution of the steel beam, simulating in this way the effect of con-
crete cracking in hogging moment regions over shear stiffness of
the beam, the predicted behaviour of the beam approaches very
close to experimental one in terms of load-displacement relation-
ship (Fig. 32).
 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106 101

Table 4
Material properties for beam type C.

Beam type C
Concrete Compressive strength (MPa) 24
Tensile strength (MPa) 2.5
Young modulus (MPa) 31,683
Steel Yield stress (MPa) Flange 375
Web 375
Ultimate stress (MPa) Flange 475
Web 475
Young modulus (MPa) Flange 200,000
Web 200,000
Strain–hardening strain Flange 0.0125
Web 0.0125
Hardening modulus (MPa) Flange 2666
Web 2666
Longitudinal reinforcement 14 mm diameter at 250 mm at middle of slab (4 bars)
Yield stress (MPa) 540
Ultimate stress (MPa) 635
Young modulus (MPa) 200,000
Strain –hardening strain 0.0125
Hardening modulus (MPa) 2666
Transverse reinforcement 14 mm diameter near to sections were shear studs are placed
Number of 16  100 studs 8
Number of rows@Spacing (mm) 1@515
Ultimate load of connectors (kN) 73.75
Shear Connection Secant connector stiffness Ksc.50% (N/mm) 47,302
Shear connection stiffness k50% (N/mm2) 92
Degree of composite action c 1.94
Function of the degree of composite action f(c) 0.276

Constitutive models

Fig. 33. Load versus rotation curves at the support for simply supported beam Type
Fig. 32. Load versus deflection curves for simply supported beam Type C. C.
102 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106
4.5. Vogel’s six-story frame with composite beams and steel columns
The geometry of Vogel’s six-story two-bay frame with propor-
tionally applied distributed gravity loads and concentrated lateral
loads, is depicted in Fig. 34. This steel frame studied previously
by other researchers [60] has been modified by converting the bare
steel beams to composite beams and included in this verification
study [60]. The geometry of the composite beams is depicted in
Fig. 34 and single and double rows of shear stud of spacing
250 mm are considered in the analysis [60]. The main material
properties are presented in Table 5 where the adopted constitutive
laws for steel and concrete are also depicted. The shear studs are
assumed to behave elastic throughout the complete loading pro-
cess with an elastic stiffness of 100 kN/mm as in [60].
In the analyses performed in this example both cracked and
uncracked cases have been considered. For cracked case the Young’s
modulus of elasticity for concrete in tension vanishes when the
strain in tensioned fibres is larger than the concrete tensile crack
strain whereas for uncracked case the Young’s modulus of elasticity
for concrete in tension is kept constant throughout the complete
loading history [60]. This frame has been analysed in [60] using a
fibre-finite element approach which allows for the spread of plas-
ticity and partial shear connection along the element length.
Assuming elastic behaviour of shear studs the entire composite
beam is modelled by a single element with a hybrid mixed-shape
function for the transverse deformation. The serviceability response
of the frame under 60% of the ultimate load is studied in [60]. In the
present approach, one inelastic flexibility-based element with 7
Gauss-Lobatto integration points has been used to model each col-
umn and beam. Fig. 35 shows load-lateral deflection curves for
monitored point (Fig. 34) considering full and partial composite
action of the beams.
As it can be seen the results agrees closely with those reported
in [60] and the lateral deflection is sensitive to the stud spacing
Fig. 34. Vogel’s six-story frame with composite beams and steel columns.
(composite action effect) and the effect of cracked and uncracked
concrete cross-sections plays a minor role over global nonlinear
behaviour. The analysis of the frame is extended here by studying
the complete nonlinear behaviour of the steel frame with different
levels of composite action of the composite beams ranging from
full composite action to simple bare steel frame.
A comparison of the load vs. lateral deflection curves obtained
with different composite beams is given in Fig. 36. For these cases,
[60] does not present comparative results. For all these analyses has
been reported very small effects of the transverse shear deformabil-
ity over global nonlinear response therefore these comparative
results are not depicted in Fig. 36. From the results given in
Fig. 36, it is clear that the composite action coming from concrete
slabs enhances the lateral stiffness and ultimate load factor of the
steel frame but the ductility of the frame is drastically reduced in
this case. Such behaviour can be explained in this case due to the
fact that the ultimate strength capacity of the steel beams is
enhanced in comparisons with the pure steel columns and a plastic
mechanism is developed at the lower levels of the frame along the
steel columns. Running the present computer program on a laptop
computer with 2 GHz processor, the present analysis, when partial
composite action of the beams has been considered, was performed
in almost 30 s, despite the fact that the analysis has been launched
within the graphical user interface. This result proves the high com-
putational efficiency of the proposed approach.
5. Summary and conclusions
This paper has presented an efficient nonlinear inelastic
analysis method for composite steel-concrete beams that is able
to take into account the combined effects of distributed plasticity
and partial composite action between the concrete slab and the
steel beam. The model also presented an additional ability to
include the effects of shear deformation in the partially connected
 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106 103

Table 5
Material properties for Vogel’s frame with composite beams[60].

Vogel’s frame
Concrete Compressive strength (MPa) 30
Tensile strength (MPa) 2.1
Young modulus (MPa) 30,890
Steel Yield stress (MPa) 220
Young modulus (MPa) 200,000
Strain-hardening strain 0.0175
Strain-hardening modulus (MPa) 4000
Shear connection Number of rows@Spacing (mm) 1@250 2@250
Connector elastic stiffness Ksc.50% (N/mm) 100,000
Connection elastic stiffness k50% (N/mm2) 400 800
Degree of composite action c 1st floor 2.743 3.879
2nd floor 2.924 4.135
3rd floor 2.924 4.135
4th floor 3.328 4.707
5th floor 3.328 4.707
6th floor 3.760 5.318
Function of the degree of composite action f(c) 1st floor 0.433 0.604
2nd floor 0.464 0.634
3rd floor 0.464 0.634
4th floor 0.529 0.692
5th floor 0.529 0.692
6th floor 0.589 0.741

Constitutive models

Fig. 35. Load-deflection curves under 60% of the ultimate load: (a) Uncracked and (b) cracked.

composite beam. In this respect, equivalent transverse shear stiff-
ness has been derived by using the energy relations.
The concept of the inelastic member effective degree of com-
posite action has been introduced. Partial composite action was
considered computing the axial force in the concrete slab along
the beam length as a nonlinear function of the inelastic member
effective degree of composite action. Hence, the inelastic response,
at the cross-sectional level, can be formulated by means of three
equilibrium equations. Gradual yielding throughout the cross-
section subjected to the combined action of axial force and bending
104 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106

Fig. 36. Load-deflection curves considering different levels of composite action of the composite beams.

moment is described through basic equilibrium, compatibility, beam length. However, as already mentioned, further investiga-
material and shear connection nonlinear constitutive equations. tions are still needed due to the limitation of current work.
Tangent flexural and axial rigidity of the cross-section are derived The proposed formulation takes advantage of using only one 2-
and then, using the flexibility approach, the elasto-plastic tangent noded beam-column element with 6 DOF being able to take into
stiffness matrix and equivalent nodal loads vector of the beam- account the distributed plasticity and the effect of partial compos-
column element including the shear deformability of the compos- ite action and features, in this way, the ability to be used for prac-
ite beam has been developed. A predictor/corrector strategy has tical applications by combining modelling benefits, computational
been implemented, corresponding to the proposed flexibility- efficiency and reasonable accuracy. This is an essential require-
based element. ment for nonlinear analysis methods to approach real large-scale
The accuracy of the analytic procedure and the computer pro- composite frame structures with partial composite action and
gram developed here, has been evaluated using several benchmark semi-rigid joints and to be implemented into design practice for
problems dealing with simply supported two-span continuous advanced analysis and seismic performance evaluation of frame
composite steel-concrete beams, and a steel-concrete plane frame structures. Ongoing research will focus on this direction concern-
with composite beams and steel columns, by comparing the results ing the application of the proposed approach for nonlinear inelastic
with experimental data, advanced three-dimensional FEM model analysis of 3D composite steel-concrete frameworks with partial
developed by the authors in Abaqus software package and other composite action and semi-rigid connections.
results reported by researchers using independent finite element
solutions. Appendix A. Tangent stiffness matrix coefficients of cross-
It can be concluded that the global behaviour of the composite section with partial composite action
beams with partial composite action, both in elastic and post-
elastic field as well as the ultimate loading capacity predicted by The coefficients of the tangent stiffness matrix kij ði; j ¼ 1; 2; 3Þ of
the proposed approach, correlate reasonable well with the experi- the cross-section when partial composite action is assumed are
mental results and advanced finite element models but with compu- given by the relationships given in Eq. (A1). In all of the below rela-
tational efficiency, usually only one element per member is tionships ðEt AÞc and ðEt AÞs denote the axial rigidity of concrete slab
necessary to analyze. It has been demonstrated that this model is and steel component, ðEt SÞc , ðEt SÞs represents the first moment of
applicable to composite beams using solid slabs with different mate- the axial rigidity of the concrete slab and structural steel respec-
rial properties, various geometry ratios for slabs, varied cross section tively, in respect with the geometric centroid of the structural
parameters, different beam lengths and various loading scenarios. steel, ðEt IÞc , ðEt IÞs represents the second moment of the axial rigid-
Although this study is limited to composite beams with solid ity of concrete slab and structural steel respectively, in respect
slabs, the present approach as is formulated here seem to suggest with the geometric centroid of the structural steel.
that such a method may be also be applied for beams with profiled
steel decking and irregular distribution of shear connectors. Since "Z #
@Nint @ X
N rs
the basic idea of the present formulation consists in quantifying k11 ¼ c
¼ rðec ðuc ; uÞÞdAc þ rðei ðuc ; uÞÞArsi
the nonlinear effects of partial composite action on the cross-
@uc @uc Ac i¼1
section flexural strength and stiffness by means of an advanced Z X
Nrs Z XNrs
@ r @ ec @ r @ ei
inelastic analysis of arbitrary cross-section, in which realistic ¼ dAc þ Arsi ¼ ETc dAc þ E A
i¼1 Trs;i rsi
Ac @ ec @u c @ ei @uc Ac
material and shear connection properties and cross-sectional i¼1 |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
ðET AÞc concrete slab and re inf :
shapes are incorporated, it appears to be feasible that the present
approach could be applied also to different types of slabs, shear ¼ ðET AÞc
connectors and materials and geometry parameters, but this issues
requires further investigations and calibrations and will be treated
"Z #
in a future work. Moreover, by simply dividing the beam according @Nint @ X
Nrs

with the variable distribution of shear connectors and treating k12 ¼ c

¼ rðec ðuc ; uÞÞdAc þ rðei ðuc ; uÞÞArsi
@us @us Ac i¼1
those segments as beams with uniform distribution of shear con-
Z X
Nrs
nectors, as in the approach developed here, could represents a @ r @ ec @ r @ ei
¼ dAc þ A ¼0
direct and simple way to extend the proposed approach to the Ac @ ec @us i¼1
@ ei @us rsi
cases of non-uniform distribution of shear connection along the
 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106 105

"Z #
@Nint @ XNrs
References
k13 ¼ c
¼ rðec ðuc ; uÞÞdAc þ rðei ðuc ; uÞÞArsi
@u @ u Ac i¼1 [1] Zona A, Ranzi G. Finite element models for nonlinear analysis of steel–concrete
Z X
N rs composite beams with partial interaction in combined bending and shear.
@ r @ ec @ r @ ei
¼ dAc þ Arsi Finite Elem Anal Des 2011;47:98–118.
Ac @ ec @ u i¼1
@ ei @ u [2] Ranzi G, Bradford MA. Nonlinear analysis of composite beams with partial
shear interaction by means of the direct stiffness method. Steel Compos Struct
Z XN rs
2009;9:131–58.
¼ ETc ðy  rÞdAc þ ETrs;i ðyi  rÞArsi [3] Queiroz FD, Vellasco PCGS, Nethercot DA. Finite element modelling of
Ac i¼1 composite beams with full and partial shear connection. J Constr Steel Res
0 1 2007;63:505–21.
Z XNrs BZ XNrs C
[4] Jeong YJ. Simplified model to predict partial-interactive structural
B C performance of steel–concrete composite slabs. J Constr Steel Res
¼ ETc ydAc þ E Trsi y A rsi  r B ETc dAc þ E A C
i¼1 Trsi rsi A 2008;64:238–46.
Ac
i¼1 i
@ Ac
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} [5] El-Tawil S, Deierlein GG. Nonlinear analysis of mixed steel-concrete frames. I:
ðET SÞc concrete slab and re inf : ðET AÞc concrete slab and re inf : element formulation. J Struct Eng ASCE 2001;127:647–55.
[6] Izzudin BA, Siyam AAFM, Lloyd-Smith D. An efficient beam-column
¼ ðET SÞc  rðET AÞc formulation for 3D reinforced concrete frames. Comput Struct
2002;80:659–76.
Z  Z [7] Liew JYR, Chen H, Shanmugam NE. Inelastic analysis of steel frames with
@Nint @ @ r @ es composite beams. J Struct Eng 2001;127(2):194–202.
k21 ¼ s
¼ rðes ðus ; uÞÞdAs ¼ dAs ¼ 0
@uc @uc As As @ es @uc [8] Li GQ, Li JJ. Advanced analysis and design of steel frames. John Wiley & Sons;
2007.
Z  Z
@Nint @ @ r @ es [9] Ngo-Huu C, Kim SE. Practical nonlinear analysis of steel-concrete composite
k22 ¼ s
¼ rðes ðus ; uÞÞdAs ¼ dAs frames using fiber-hinge method. J Constr Steel Res 2012;74:90–7.
@us @us As As @ es @us [10] Iu CK. Inelastic finite element analysis of composite beams on the basis of the
Z
plastic hinge approach. Eng Struct 2008;30:2912–22.
¼ ETs dAs ¼ ðET AÞs [11] Iu CK, Bradford MA, Chen WF. Second-order inelastic analysis of composite
As framed structures based on the refined plastic hinge method. Eng Struct
Z  Z 2009;31:799–813.
@Nint @ @ r @ es [12] Iu CK. Nonlinear analysis for the pre- and post-yield behaviour of a composite
k23 ¼ s
¼ rðes ðus ; uÞÞdAs ¼ dA
@u @ u As As @ es @ u s structure with the refined plastic hinge approach. J Constr Steel Res
Z 2016;119:1–16.
[13] Lemes ÍJM, Silva ARD, Silveira RAM, Rocha PAS. Determination of structural
¼ ETs ydAs ¼ ðET SÞs ðA1Þ composite elements strength through the use of the refined plastic hinge
As
method. Revista Internacional de Métodos Numéricos para Cálculo y Diseño en
Z Ingeniería 2015. http://dx.doi.org/10.1016/j.rimni.2015.10.003.
@Mint @M int @ [14] Liu SW, Lui YP, Chan SL. Advanced analysis of hybrid steel and concrete frames.
k31 ¼ s
þ c
¼ rðes ðus ; uÞÞydAs
@uc @uc @uc As #
Part I: cross section analysis technique and second-order analysis. J Construct
Z X
N rs Steel Res 2012;70:326–36.
[15] Chiorean CG. A computer method for nonlinear inelastic analysis of 3D
þ rðec ðuc ; uÞÞydAc þ rðei ðuc ; uÞÞyi Arsi
Ac
composite steel-concrete frame structures. Eng Struct 2013;57:125–52.
i¼1
Z XN rs
[16] Nie J, Tao M, Cai CS, Chen G. Modeling and investigation of elasto-plastic
@ r @ ec @ r @ ei behavior of steel–concrete composite frame system. J Constr Steel Res
¼ ydAc þ yA 2011;67:1973–84.
@ e c @u c @ ei @uc i rsi
Z Ac XNrs
i¼1 [17] El-Lobody E, Lam D. Finite element analysis of steel-concrete composite
girders. Adv Struct Eng 2003;6:267–81.
¼ ETc ydAc þ E y A ¼ ðET SÞc
i¼1 Trsi i rsi [18] El-Lobody E, Young B. Numerical simulation of concrete encased steel
Ac
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} composite columns. J Constr Steel Res 2011;67:211–22.
ðET SÞc concrete slab and re inf : [19] Wang Y, Nie J, Cai CS. Numerical modeling on concrete structures and steel–
concrete composite frame structures. Composites B 2013;51:58–67.
Z [20] Faella C, Martinelli E, Nigro E. Shear connection nonlinearity and deflections of
@Mint @M int @ steel-concrete composite beams: a simplified method. J Struct Eng 2003;129
k32 ¼ s
þ c
¼ rðes ðus ; uÞÞydAs (1):112–20.
@us @us @us As #
Z XN rs
[21] Faella C, Martinelli E, Nigro E. Analysis of steel–concrete composite PR-frames
in partial shear interaction: a numerical model and some applications. Eng
þ rðec ðuc ; uÞÞydAc þ rðei ðuc ; uÞÞyi Arsi Struct 2008;30:1178–86.
Z Ac Z i¼1 [22] Faella C, Martinelli E, Nigro E. Steel concrete composite beams in partial
@ r @ es interaction: closed-form ‘‘exact” expression of the stiffness matrix and the
¼ ydAs ¼ ETs ydAs ¼ ðET SÞs
As @ es @us As
vector of equivalent nodal forces. Eng Struct 2010;32:2744–54.
[23] Čas B, Saje M, Planinc I. Non-linear finite element analysis of composite planar
frames with an interlayer slip. Comput Struct 2004;82:1901–10.
@M int @Mint [24] Ayoub A, Filippou FC. Mixed formulation of nonlinear steel-concrete
k33 ¼ s
þ c
@u @u composite beam element. J Struct Eng 2000;126(3):371–81.
"Z Z # [25] Valipour HR, Bradford MA. A steel-concrete composite beam element with
@ XNrs
material nonlinearities and partial shear interaction. Finite Elem Anal Des
¼ rðes ðus ; uÞÞydAs þ rðec ðuc ; uÞÞydAc þ rðei ðuc ; uÞÞyi Arsi 2009;45:966–72.
@ u As Ac i¼1 [26] Nguyen QH, Hjiaj M, Uy B, Guezouli S. Analysis of composite beams in the
Z Z XNrs hogging moment regions using a mixed finite element formulation. J Constr
@ r @ es @ r @ ec @ r @ ei
¼ ydAs þ ydAc þ y Arsi Steel Res 2009;65:737–48.
As @ es @ u Ac @ ec @ u i¼1
@ ei @ u i [27] Turmo J, Lozano-Galant JA, Mirambell E, Xu D. Modeling composite beams
Z Z XNrs
with partial interaction. J Constr Steel Res 2015;114:380–93.
[28] Newmark NM, Siess CP, Viest IM. Tests and analysis of composite beams with
¼ ETs y2 dAs þ ETc ðy  rÞydAc þ ETrsi ðyi  rÞyi Arsi incomplete interaction. Proc Soc Exp Stress Anal 1951;9(1):75–92.
As Ac i¼1
Z Z XNrs
[29] Girhammar UA, Pan DH. Exact static analysis of partially composite beams and
beam-columns. Int J Mech Sci 2007;49:239–55.
¼ ETs y2 dAs þ ETc y2 dAc þ i¼1
ETrsi y2i Arsi [30] Girhammar UA. Composite beam–columns with interlayer slip—approximate
As Ac
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} analysis. Int J Mech Sci 2008;50:1636–49.
ðET IÞc concrete slab and reinf : [31] Girhammar UA. A simplified analysis method for composite beams with
0 1 interlayer slip. Int J Mech Sci 2009;51:515–30.
BZ XNrs C [32] Nguyen QH, Martinelli E, Hjiaj M. Derivation of the ‘‘exact” stiffness matrix for
B C a two-layer Timoshenko beam element with partial interaction. Eng Struct
 r B ETc ydAc þ ETrsi yi Arsi C ¼ ðET IÞs þ ðET IÞc  rðET SÞc
@ Ac i¼1 A 2010;33(2):298–307.
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
ðET SÞc concrete slab and reinf :
106 C.G. Chiorean, S.M. Buru / Engineering Structures 134 (2017) 74–106
[33] Ranzi G, Zona A. A steel–concrete composite beam model with partial
interaction including the shear deformability of the steel component. Eng
Struct 2007;29(11):3026–41.
[34] Girhammar UA, Atashipour SR. Analysis of shear deflections of deep composite
box-type of beams using different shear deformation models. Comput Struct
2015;155:42–53.
[35] Santos HAFA, Silberschmidt VV. Hybrid equilibrium finite element formulation
for composite beams with partial interaction. Compos Struct
2014;108:646–56.
[36] Neagoe CA, Gil L. Analytical procedure for the design of PFRP-RC hybrid
beams including shear interaction effects. Compos Struct
2015;132:122–35.
[37] Martinelli E, Nguyen QH, Hjiaj M. Dimensionless formulation and comparative
study of analytical models for composite beams in partial interaction. J Constr
Steel Res 2012;75:21–31.
[38] Nguyen QH, Hjiaj M, Guezouli S. Exact finite element model for shear-
deformable two-layer beams with discrete shear connection. Finite Elem Anal
Des 2011;47:718–27.
[39] Dall’Asta A, Zona A. Comparison and validation of displacement and mixed
elements for the non-linear analysis of continuous composite beams. Comput
Struct 2004;82:2117–30.
[40] Ayoub A. Analysis of composite frame structures with mixed elements-state of
the art. Struct Eng Mech 2012;41:157–81.
[41] Neuenhofer A, Filippou FC. Evaluation of nonlinear frame finite-element
models. J Struct Eng ASCE 1997;123:958–66.
[42] ABAQUS/standard user’s manual, version 6.11. Hibbitt, Karlsson & Sorensen
Inc., 2011.
[43] CEB-FIP (Comité Euro-International du Béton). Model Code 1990: Design Code.
London (UK): Thomas Thelford; 1993.
[44] Mansour M, Lee JY, Hsu TTC. Cyclic stress-strain curves of concrete and steel
bars in membrane elements. J Struct Eng ASCE 2001;127:1402–11.
[45] Ollgaard JG, Slutter RG, Fisher JW. Shear strength of stud connectors in
lightweight and normal weight concrete. AISC Eng J 1971;8(2):55–64.
[46] European Committee for Standardization, Eurocode 4: Design of composite
steel and concrete structures - part 1-1: general rules and rules for buildings,
EN 1994-1-1. Brussels (Belgium), December 2004.
View publication stats
[47] Rotter JM. Rapid exact inelastic biaxial bending analysis. J Struct Eng, ASCE
1985;111(12):2659–67.
[48] Chiorean CG. Computerised interaction diagrams and moment capacity
contours for composite steel-concrete cross-sections. Eng Struct
2010;32:3734–57.
[49] Bonet JL, Romero ML, Miguel PF, Fernandez MA. A fast stress integration
algorithm for reinforced concrete sections with axial loads and biaxial
bending. Comput Struct 2004;82:213–25.
[50] Papanikolau VK. Analysis of arbitrary composite sections in biaxial bending
and axial load. Comput Struct 2012;98–99:33–54.
[51] Chiorean CG. A computer program for nonlinear inelastic analysis of 3D semi-
rigid steel frameworks. Eng Struct 2009;31:3016–33.
[52] Aribert JM, Labib AG. Modele de calcul elasto-plastique de poutre mixtes a
connexion partielle. Construction Metallique 1982;4:3–51.
[53] Johnson RP, Molestra IN. Partial shear connection in composite beams for
buildings. Proc Inst Civil Engineers 1991;79(2):679–704.
[54] Aribert JM, Labib AG, Rival JC. Etude numerique et experimental de l’influence
d’une connexion partielle sur le comportement de poutres mixtes.
Communication presentee aux journees A.F.P.C., mars 1983, theme 1, sous-
theme [in French].
[55] Chapman JC, Balakrishnan S. Experiments on composite beams. Struct Eng
1964;42(11):369–83.
[56] Ban H, Bradford MA. Elastoplastic cross-sectional behavior of composite beams
with high-strength steel: analytical modeling. J Struct Eng (ASCE)
2014;12:1–12.
[57] Ansourian P. Experiments on continuous composite beams. Proc Inst Civil Eng,
Part 1981;2(71):25–51.
[58] Loh HY, Uy B, Bradford MA. The effects of partial shear connection in the
hogging moment regions of composite beams Part II – analytical study. Finite
Elem Anal Des 2011;47:98–118.
[59] Fabbrocino G, Pecce M. Experimental tests on steel-concrete composite beams
under negative bending. In: Third structural specialty conference of the
Canadian society for civil engineering; 2000.
[60] Fang LX, Chan SL, Wong YL. Numerical analysis of composite frames with
partial shear-stud interaction by one element per member. Eng Struct
2000;22:1285–300.