00b49525c79079cc8a000000 (1) (1).pdf

download 00b49525c79079cc8a000000 (1) (1).pdf

of 17

  • date post

    02-Jun-2018
  • Category

    Documents

  • view

    244
  • download

    0

Transcript of 00b49525c79079cc8a000000 (1) (1).pdf

  • 8/10/2019 00b49525c79079cc8a000000 (1) (1).pdf

    1/17

    Journal of Constructional Steel Research 63 (2007) 505521www.elsevier.com/locate/jcsr

    Finite element modelling of composite beams with full and partial shearconnection

    F.D. Queiroza,, P.C.G.S. Vellascob, D.A. Nethercot a

    aDepartment of Civil and Environmental Engineering, Imperial College London, SW7 2AZ, United Kingdomb Structural Engineering Department, State University of Rio de Janeiro, RJ, Brazil

    Received 25 January 2006; accepted 8 June 2006

    Abstract

    The present investigation focuses on the evaluation of full and partial shear connection in composite beams using the commercial finite element

    (FE) software ANSYS. The proposed three-dimensional FE model is able to simulate the overall flexural behaviour of simply supported composite

    beams subjected to either concentrated or uniformly distributed loads. This covers: load deflection behaviour, longitudinal slip at the steelconcrete

    interface, distribution of stud shear force and failure modes. The reliability of the model is demonstrated by comparisons with experiments and

    with alternative numerical analyses. This is followed by an extensive parametric study using the calibrated FE model. The paper also discusses in

    detail several numerical modelling issues related to potential convergence problems, loading strategies and computer efficiency. The accuracy and

    simplicity of the proposed model make it suitable to predict and/or complement experimental investigations.

    c 2006 Elsevier Ltd. All rights reserved.

    Keywords:Composite beams; Finite element method; Material nonlinearity; Shear connection; Parametric analysis; ANSYS

    1. Introduction

    Composite steelconcrete construction, particularly for

    multi-storey steel frames, has achieved a high market share in

    several European countries, the USA, Canada and Australia.

    This is mainly due to a reduction in construction depth, to

    savings in steel weight and to rapid construction programmes.

    Composite action enhances structural efficiency by combin-

    ing the structural elements to create a single composite sec-

    tion. Composite beam designs provide a significant economy

    through reduced material, more slender floor depths and faster

    construction. Moreover, this system is well recognised in termsof the stiffness and strength improvements that can be achieved

    when compared with non-composite solutions.

    A fundamental point for the structural behaviour and design

    of composite beams is the level of connection and interaction

    between the steel section and the concrete slab. The term full

    shear connection relates to the case in which the connection

    between the components is able to fully resist the forces applied

    Corresponding author. Tel.: +44 (0) 20 7594 6097.E-mail addresses:[email protected](F.D. Queiroz),

    [email protected](D.A. Nethercot).

    to it. This is possibly the most common situation; however, over

    the last two decades the use of beams in building construction

    has led to many instances when the interconnection cannot

    resist all the forces applied (partial shear connection). In this

    case, the connection may fail in shear before either of the other

    components reaches its own failure state.

    In the case of the serviceability limit state of composite

    beams, the condition when the connection between the

    components is considered as infinitely stiff is said to comprise

    full interaction. Whilst this is often assumed in design, it

    is theoretically impossible and cases where the connection

    has more limited stiffness (partial interaction) often need

    to be considered. In this case, the connection itself may

    deform, resulting in relative movement along the steelconcrete

    interface and the effect of increased shear deformation in the

    beam as a whole. Therefore, partial interaction occurs to some

    extent in all beams whether fully connected or not [1]. However,

    studies [1,2] have shown that any flexibility in the connection

    may be ignored for beams designed for full connection.

    The use of partial connection provides the opportunity to

    achieve a better match of applied and resisting moment and

    some economy in the provision of connectors. Generally, the

    effects of partial interaction, which are increased by the use

    0143-974X/$ - see front matter c 2006 Elsevier Ltd. All rights reserved.doi:10.1016/j.jcsr.2006.06.003

  • 8/10/2019 00b49525c79079cc8a000000 (1) (1).pdf

    2/17

    506 F.D. Queiroz et al. / Journal of Constructional Steel Research 63 (2007) 505521

    of partial shear connection, will result in reduced strength

    and stiffness, and potentially enhanced ductility of the overall

    structural system [2].

    It is widely known that laboratory tests require a great

    amount of time, are very expensive and, in some cases, can

    even be impractical. On the other hand, the finite element

    method has become, in recent years, a powerful and useful

    tool for the analysis of a wide range of engineering problems.

    According to Abdollahi [3], a comprehensive finite element

    model permits a considerable reduction in the number of

    experiments. Nevertheless, in a complete investigation of any

    structural system, the experimental phase is essential. Taking

    into account that numerical models should be based on reliable

    test results, experimental and numerical/theoretical analyses

    complement each other in the investigation of a particular

    structural phenomenon [4].

    Previous numerical studies have been conducted to

    investigate the behaviour of composite beams. Nevertheless,

    most of them are based on two-dimensional analytical models

    (e.g., Gattesco [5] and Pi et al. [6]), and are thus not

    able to simulate more complex aspects of behaviour, which

    are intrinsic for three-dimensional studies; for instance: full

    distribution of stresses and strains over the entire section

    of the structural components (steel beam and concrete slab),

    evolution of cracks and local deformations in the concrete slab.

    In addition, in the particular case of the model developed by

    Pi et al. [6], it was assumed that the shear connectors were

    uniformly distributed along the length of a composite member.

    A three-dimensional finite element model has been developed

    by El-Lobody and Lam [7] using the package ABAQUS [8],

    in which the mode of failure of the beams is detected by

    a manual check of the compressive concrete stress and stud

    forces for each load step. Nevertheless, just two beams were

    used to validate the proposed model for composite beams with

    solid slabs. All these studies [57] were focused just on the

    presentation and validation of their corresponding models, but

    these models were not used to investigate in more detail either

    the effect of particular structural parameters or other aspects

    of the system behaviour. It is only very recently that papers

    on finite element analyses of composite systems have started

    to contain parametric studies (e.g., investigations related to the

    behaviour of individual shear connectors [9,10]).In order to obtain reliable results up to failure, finite element

    models must properly represent the constituent parts, adopt

    adequate elements and use appropriate solution techniques.

    As the behaviour of composite beams presents significant

    nonlinear effects, it is fundamental that the interaction of all

    different components should be properly modelled, as well as

    the interface behaviour. Once suitably validated, the model can

    be utilised to investigate aspects of behaviour in far more detail

    than is possible in laboratory work. For instance, it permits

    the study of the sensitivity of response to variability of key

    component characteristics, including material properties and

    shear stud layout. Consequently, different spacing in distinct

    parts of the beam can be adopted, allowing the investigation ofpartial interaction effects.

    The present investigation focuses on the modelling of

    composite beams with full and partial shear connection using

    the software ANSYS [11]. A three-dimensional model is

    proposed, in which all the main structural parameters andassociated nonlinearities are included (concrete slab, steel beam

    and shear connectors). Test and numerical data available in

    the literature are used to validate the model, which is able to

    deal with simply supported systems with I-beams and solid flat

    slabs. Other features such as steel profiled sheeting, different

    types of slab (e.g., precast slabs) and distinct end-connectivities

    are not included in the present study.

    Based on the validated model, an extensive parametric

    analysis of composite beams is performed, specifically aimed

    at:

    studying the effect of the continuation of shear connection

    beyond the supports of simply supported composite beams; investigating the overall structural system behaviour when

    different concrete compressive strengths are used in the slab

    and in the associated push-out tests. This situation has often

    been observed in reported laboratory studies but has not

    previously received any systematic study, being important,

    for instance, for the definition of the loadslip curves used

    for the shear connectors;

    analysing the influence of small variations in key input

    parameters (i.e., concrete and steel material properties) on

    the structural behaviour of uniformly loaded composite

    beams. This sensitivity study can be used to identify the

    variables (e.g., web and flange yield stresses, concrete

    strength, etc.) which are more important in terms ofdefinition of the overall response of the system and therefore

    can be helpful in terms of establishing possible differences

    between numerical and test results;

    assessing the influence of the effects of partial shear

    connection and partial interaction not only on the overall

    flexural behaviour of composite beams (represented by the

    loaddeflection curve), but also on the associated failure

    modes for either slab crushing or stud failure, and on the

    distribution of stud shear forces along the beam length.

    This paper also discusses several numerical modelling issues

    related to convergence problems that arise when the concrete

    material is considered, loading strategies for the simulation of

    distributed loads and a comparison between the load controland the displacement control methods in terms of computer

    efficiency.

    2. Finite element model

    2.1. Software, element types and mesh construction

    Advances in computational features and software have

    brought the finite element method within reach of both

    academic research and engineers in practice by means of

    general-purpose nonlinear finite element analysis packages,

    with one of the most used nowadays being ANSYS. The

    program offers a wide range of options regarding element types,material behaviour and numerical solution controls, as well as

  • 8/10/2019 00b49525c79079cc8a000000 (1) (1).pdf

    3/17

    F.D. Queiroz et al. / Journal of Constructional Steel Research 63 (2007) 505521 507

    Fig. 1. Typical composite beam FE mesh.

    graphic user interfaces (known as GUIs), auto-meshers, andsophisticated postprocessors and graphics to speed the analyses.

    In this paper, the structural system modelling is based on the use

    of this commercial software.

    The finite element types considered in the model are as

    follows: elastic-plastic shell (SHELL43) and solid (SOLID65)

    elements for the steel section and the concrete slab,

    respectively, and nonlinear springs (COMBIN39) to represent

    the shear connectors. Both longitudinal and transverse

    reinforcing bars are modelled as smeared throughout the solid

    finite elements.

    The element SHELL43 is defined by four nodes having

    six degrees of freedom at each node. The deformation shapes

    are linear in both in-plane directions. The element allows forplasticity, creep, stress stiffening, large deflections, and large

    strain capabilities [11]. The element SOLID65 is used for three-

    dimensional modelling of solids with or without reinforcing

    bars (rebar capability). The element has eight nodes and three

    degrees of freedom (translations) at each node. The concrete is

    capable of cracking (in three orthogonal directions), crushing,

    plastic deformation, and creep [11]. The rebars are capable of

    sustaining tension and compression forces, but not shear, being

    also capable of plastic deformation and creep.

    The element COMBIN39 is defined by two node points

    and a generalized forcedeflection curve and has longitudinal

    or torsional capability. The longitudinal option is a uniaxial

    tensioncompression element with up to three degrees offreedom (translations) at each node.

    Symmetry of the composite beams is taken into account by

    modelling only one half of the beam span. A typical FE mesh

    for a composite beam is shown inFig. 1.

    2.2. Material modelling

    The von Mises yield criterion with isotropic hardening rule

    (multilinear work-hardening material) is used to represent the

    steel beam (flanges and web) behaviour. The stressstrain

    relationship is linear elastic up to yielding, perfectly plastic

    between the elastic limit (y ) and the beginning of strainhardening and follows the constitutive law used by Gattesco [5]

    for the strain-hardening branch:

    = fy+ E

    h(

    h)1 E

    h

    h

    4(fu fy) (1)

    where fy and fu are the yield and ultimate tensile stresses of

    the steel component, respectively; Eh and h are the strain-

    hardening modulus (i.e., 3500 N/mm2) and the strain at strain

    hardening of the steel component, respectively.

    The von Mises yield criterion with isotropic hardening rule

    is also used for the reinforcing steel. An elastic-linear-work-

    hardening material is considered, with tangent modulus being

    equal to 1/10 000 of the elastic modulus, in order to avoid

    numerical problems. The values measured in the experimental

    tests for the material properties of the steel components (steel

    beam and reinforcing bars) are used in the finite element

    analyses.The concrete slab behaviour is modelled by a multilinear

    isotropic hardening relationship, using the von Mises yield

    criterion coupled with an isotropic work hardening assumption.

    The uniaxial behaviour is described by a piece-wise linear

    total stresstotal strain curve, starting at the origin, with

    positive stress and strain values, considering the concrete

    compressive strength (fc) corresponding to a compressive

    strain of 0.2%. The stressstrain curve also assumes a total

    increase of 0.05 N/mm2 in the compressive strength up to

    the concrete strain of 0.35% to avoid numerical problems due

    to an unrestricted yielding flow. The concrete element shear

    transfer coefficients considered are: 0.2 (open crack) and 0.6

    (closed crack). Typical values range from 0 to 1, where 0represents a smooth crack (complete loss of shear transfer) and

    1 a rough crack (no loss of shear transfer). The default value

    of 0.6 is used as the stress relaxation coefficient (a device that

    helps accelerate convergence when cracking is imminent). The

    crushing capability of the concrete element is also disabled to

    improve convergence.

    The concrete slab compressive strength is taken as the actual

    cylinder strength test value. The concrete tensile strength and

    the Poissons ratio are assumed as 1/10 of its compressive

    strength and 0.2, respectively. The concrete elastic modulus is

    evaluated according to Eurocode 4 [12], i.e.:

    Ec = 9500(fc + 8)1/3c24

    1/2 (2)

    where:c is equal to 24 kN/m3.

    The model allows for any pattern of stud distribution to be

    considered, for instance: the conventional uniform arrangement

    and a triangular spacing scheme where the stud distribution

    follows the nominal elastic shear diagram [13]. In all analyses,

    the number/spacing of studs adopted in the experimental

    programmes is utilised. As far as the shear connector behaviour

    is concerned, the loadslip curves for the studs are used

    (obtained from available push-out tests) by defining a table

    of force values and relative displacements (slip) as input data

    for the nonlinear springs. These springs are modelled at thesteelconcrete interface, as shown inFig. 2.

  • 8/10/2019 00b49525c79079cc8a000000 (1) (1).pdf

    4/17

    508 F.D. Queiroz et al. / Journal of Constructional Steel Research 63 (2007) 505521

    Fig. 2. Modelling of shear connectors (longitudinal view). (a) Shear studs in a typical composite beam. (b) Shear studs in a typical composite beam finite element

    mesh. (c) Representation of the shear stud model.

    2.3. Application of load and numerical control

    Regarding application of load, concentrated loads are

    incrementally applied to the model by means of an

    equivalent displacement to overcome convergence problems

    (displacement control). For the convergence criterion, the L2-

    norm (square root sum of the squares) of displacements is

    considered. Uniformly distributed loads are represented bymeans of point loads applied at all mid-section concrete

    nodes. These concentrated loads are also applied to the model

    incrementally using the load control strategy and the L2-

    norm. The tolerance associated with this convergence criterion

    (CNVTOL command of ANSYS) and the load step increment

    are varied in order to solve potential numerical problems.

    Whenever the solution does not converge for the set of

    parameters considered, as far as load step size and converge

    criterion are concerned, the RESTART command is used in

    conjunction with the CNVTOL option. ANSYS allows two

    different types of restart: the single-frame restart and the multi-

    frame restart, which can be used for static or full transient

    structural analyses. The single-frame restart only allows theuser to resume a job at the point it stopped. The multi-frame

    restart can resume a job at any point in the analysis for

    which information is saved. This capability enables multiple

    model analyses, presenting more options for data retrieval

    after an undesired aborted solution. The second approach

    is used throughout the present analyses and the associated

    error is controlled by comparisons between applied forces and

    reactions (balance of forces) for each load step.

    The load control strategy is adopted due to the fact that,in structural problems in which significant nonlinear effects

    occur, it is difficult to derive a relationship between loads

    and associated displacements for the case of distributed loads,

    mainly for the plastic range of behaviour. For the case in

    which only one point load is applied to the system, there is

    a direct relationship between force and displacement, making

    the displacement control method easier to be utilised. The load

    control method is, however, less efficient than the displacement

    control method in nonlinear analyses. This fact is observed

    especially when the applied load approaches the ultimate load

    of the system, as an incremental increase in the load leads

    to a significant increase in the corresponding displacements,

    causing difficulties in terms of numerical convergence. Forthe type and size of the finite element problem investigated,

  • 8/10/2019 00b49525c79079cc8a000000 (1) (1).pdf

    5/17

    F.D. Queiroz et al. / Journal of Constructional Steel Research 63 (2007) 505521 509

    the load control method demanded, on average, 40% more

    disk space and took 140% longer to be processed than similar

    displacement control solutions.

    Preliminary attempts to overcome the convergence problemsarising from the use of the load control method included the

    specification of different types of equation solvers. The best

    approach in terms of numerical performance was the option

    in which the software ANSYS selects a solver based on the

    physics of the problem. The arc-length method was also tested

    but proved not to be the best option for this particular type of

    analysis.

    2.4. Failure criterion

    Two limits are established to define the ultimate load for

    each finite element investigation: a lower and an upper bound,

    corresponding to concrete compressive strains of 0.2%, and0.35%, respectively. These two limits define an interval in

    which the composite beam collapse load is located. A third

    limit condition, hereinafter referred to as the stud failure point,

    can also be reached when the composite beams most heavily

    loaded stud reaches its ultimate load, as defined from the

    appropriate push-out tests.

    If the stud failure point is located before the lower bound

    of concrete (i.e., the corresponding load of the stud failure

    point is smaller than the lower bound load) then the mode

    of failure of the composite beam is considered as being stud

    failure. Conversely, if the stud failure point is located after the

    upper bound of concrete, the mode of failure is assumed as

    being concrete crushing. For the intermediate case, where thestud failure point lies between the lower and upper bounds of

    concrete, than the mode of failure could be either of them.

    Therefore, the proposed finite element model is able to

    predict the failure modes associated with either slab crushing

    or stud failure.

    3. Validation of the model

    The present model is validated by comparisons against

    Chapman and Balakrishnan tests [13], as well as against

    alternative numerical studies (Gattesco [5], Pi et al. [6] and El-

    Lobody and Lam [7]).

    3.1. Chapman and Balakrishnan tests

    The tests performed by Chapman and Balakrishnan

    successfully illustrate the behaviour of the composite system

    which is being investigated. The beams spanned 5490 mm

    with an I-shaped steel member 305 mm deep (12 6

    44 lb/ft BSB) and a concrete slab 152 mm thick1220 mm

    wide. The number and type of studs, as well as the steel and

    concrete strengths, varied according to the tested composite

    beam. The slab was longitudinally reinforced with four top

    and four bottom 8 mm bars. The transverse reinforcement

    incorporated top and bottom bars of 12.7 mm @ 152 mm

    centres and 12.7 mm @ 305 mm centres, respectively. Thetensile strength, the Youngs modulus and the Poissons ratio

    Fig. 3. Simply supported beam layout (dimensions in mm).

    Fig. 4. Load (kN) vs. midspan deflection (mm) beams A2 and A3.

    of the reinforcing steel bars were 320 N/mm2, 205000 N/mm2

    and 0.3, respectively. A list of material properties for all beams

    is given in [1316] and a full description of these beams is

    presented inFig. 3andTable 1.Based on the composite section strength of the concrete

    slab, steel components and shear connectors, the level of

    shear connection could be determined. This value is defined

    as the ratio between the shear connection capacity and the

    weakest element capacity (concrete slab or steel beam).Table 2summarises the level of shear connection for all the composite

    beams, considering two different approaches. The first one uses

    the nominal values presented by [13] for the stud strength and

    steel yield stress. In the second one, the material properties

    are taken as the actual measured values [13]. Considerable

    differences among the levels of shear connection according to

    these two approaches are noticed, leading to the conclusion that,

    in order to calculate the level of shear connection of composite

    systems, the actual material properties of the components

    (measured values), related to each experimental programme,

    should be used.For all composite beams shown in Table 1, loadmidspan

    deflection curves are compared to the test results. Figs. 48(beams A2 to E1) and Figs. 911 (beams U1 to U4)

    depict comparisons between the FE model results and the

    experimental data for the midspan concentrated and uniformly

    distributed loaded composite beams, respectively. The limit

    points for the concrete are represented by a full triangle (lower

    bound) and by a full square (upper bound), and the stud failure

    point is represented by a full circle (item 2.4). Good agreement

    was obtained between test and numerical results.In order to illustrate finite element data for local results,

    the numerical and test values regarding the slip at the

    steelconcrete interface along the beam axis for the cases E1

    and U4 are plotted inFigs. 12and13, respectively. The graphs

    show that the proposed model can predict the slip distributionwith good precision. From these figures, it can be noticed that

  • 8/10/2019 00b49525c79079cc8a000000 (1) (1).pdf

    6/17

    510 F.D. Queiroz et al. / Journal of Constructional Steel Research 63 (2007) 505521

    Table 1

    Details of composite beams tested by [13]

    Beam A2 A3 A4 A5 A6 B1 C1 D1 E1 U1 U3 U4

    Stud diam. (mm) 19

    12.7

    Stud overall length (mm) 102

    76

    50

    Number of studs 100

    76

    68

    56

    44

    32

    Spacing in pairs (mm) 121 a

    159 a

    178 a216 a

    274 a

    378 a

    Mode of failure Slab crushing

    Stud failure

    Load type Midspan concentrated Uniformly

    distributed

    a Triangular spacing.

    Table 2

    Level of shear connection of the composite beams (%)

    Beam A2 A3 A4 A5 A6 B1 C1 D1 E1 U1 U3 U4

    Nominal values 238 213 175 138 101 138 138 313 313 175 175 101

    Measured values 231 186 137 123 95 116 114 136 148 155 177 90

    Fig. 5. Load (kN) vs. midspan deflection (mm) beams A4 and A5.

    Fig. 6. Load (kN) vs. midspan deflection (mm) beams A6 and B1.

    Fig. 7. Load (kN) vs. midspan deflection (mm) beams C1 and D1.

    Fig. 8. Load (kN) vs. midspan deflection (mm) beam E1.

  • 8/10/2019 00b49525c79079cc8a000000 (1) (1).pdf

    7/17

    F.D. Queiroz et al. / Journal of Constructional Steel Research 63 (2007) 505521 511

    Fig. 9. Load (kN/m) vs. midspan deflection (mm) beam U1.

    Fig. 10. Load (kN/m) vs. midspan deflection (mm) beam U3.

    Fig. 11. Load (kN/m) vs. midspan deflection (mm) beam U4.

    the slip is not uniform along the beam length, even where the

    externally applied shear force is uniform (midspan concentrated

    load case). The maximum slip value tends to occur when X/L

    equals 0.4 (point load case) and X/L equals 0.3 (uniformly

    distributed load case UDL). According to [13], the large slip

    which occurs near midspan is due to the high interface shear inthe plastic region.

    Table 3 presents the ultimate load for each of the studied

    composite beams. This load is expressed in terms of the test

    results and both lower and upper bound limits (FE analysis).

    This table also presents the ratio between the numerical and

    test results for each limit point and their associated dispersion

    values. It can be noticed that the proposed model was able to

    predict the experimental ultimate load very accurately, as well

    as the associated mode of failure (slab crushing or stud failure).

    3.2. Comparisons with previous numerical studies

    Gattesco [5] presented an analytical procedure for theinvestigation of composite beams, in which the nonlinear

    Fig. 12. Slip distribution along span beam E1.

    Fig. 13. Slip distribution along span beam U4.

    behaviour of all materials was considered. The parametric

    analysis demonstrated that the numerical program was able to

    model full and partial shear connection. El-Lobody and Lam [7]

    used the ABAQUS FE software to undertake a numericalanalysis of composite girders with solid and precast hollow

    core slabs. Both models included the material nonlinearities,

    as well as the stud nonlinear loadslip characteristics. Partial

    interaction between the steel and concrete components was

    also incorporated in the total Lagrangian finite element

    model formulated by Pi et al. [6]. The model was validated

    by comparisons against simply supported and continuous

    composite beams tests.

    In Figs. 14 and 15 the loadmidspan deflection curves

    obtained from these previous numerical models are compared

    with the present study results for the composite beams E1 and

    U4. Good agreement between the curves was obtained.

    3.3. Effect of the overhang region (region beyond supports)

    As it can be seen in Fig. 3, the experiments conducted

    by Chapman and Balakrishnan concerned composite beams in

    which a number of shear connectors were used in the overhang

    regions. According to [13], a more rational test procedure

    would have been to limit the slab length to the distance between

    supports. Nevertheless, it was also stated by [13] that, in

    practice, the concrete slabs do not always necessarily terminate

    at the reactions. In this section, the effect of the overhang

    regions (i.e., the regions beyond supports) will be assessed in

    terms of both overall and local behaviour. None of the already-

    mentioned previous studies on the Chapman and Balakrishnantests [57] investigated this aspect.

  • 8/10/2019 00b49525c79079cc8a000000 (1) (1).pdf

    8/17

    512 F.D. Queiroz et al. / Journal of Constructional Steel Research 63 (2007) 505521

    Table 3

    Ultimate load results for the experimental and numerical analyses

    Beam Pexp PLB(0.2%) PUB(0.35%) 1 = PLB/Pexp 2 = PUB/Pexp % = (2 1) 100

    A2 448 429 469 0.96 1.05 9

    A3 449 425 447 0.95 0.99 4

    A4a 523 444 470 0.85 0.90 5

    A5 468 462 479 0.99 1.02 3

    A6 430b 449b 1.04

    B1 486 459 468 (466b ) 0.94 0.96 (0.95b ) 2

    C1 448 445 474 0.99 1.06 7

    D1 481 457 475 0.95 0.99 4

    E1 513 520 548 1.01 1.07 6

    U1 191 171 178 0.90 0.93 3

    U3 185 166 182 0.90 0.98 8

    U4 176b 179b 1.02

    Pexp, PLBand PUBare the test ultimate load, lower and upper bound loads, respectively (midspan load [kN]; uniformly distributed load [kN/m]).a The strength of the concrete used in the push-out test was much less than the concrete strength of the composite beam;b

    Stud failure, so bounds not applicable.

    Fig. 14. Load vs. midspan deflection Other similar studies (beam E1).

    Fig. 15. Load vs. midspan deflection Other similar studies (beam U4).

    In the following (Figs. 1625), the analyses in which theseregions are included in the model will be hereinafter referred to

    as the extended case (results presented in item 3.1), and the

    ones in which only the regions between supports are modelled

    will be referred to as the standard case.

    The loaddeflection curves for the composite beams A4, D1,

    E1 (midspan concentrated load) and U4 (UDL) are shown in

    Figs. 1619, respectively. It can be observed that an increase

    in the system stiffness was present for the extended case when

    compared with the standard one. The connectors placed beyond

    the supports have an anchorage effect on the beam, leading to

    an improvement in the composite effect of the system.

    This behaviour is in accordance with the observations by

    Goodman and Popov [17], who investigated the behaviour oflayered wood beam systems with interlayer slip. Results of

    Fig. 16. Load (kN) vs. midspan deflection (mm) beam A4.

    Fig. 17. Load (kN) vs. midspan deflection (mm) beam D1.

    Fig. 18. Load (kN) vs. midspan deflection (mm) beam E1.

    experiments with layered beams connected with nails, withand without glued ends, showed that there was a decrease

  • 8/10/2019 00b49525c79079cc8a000000 (1) (1).pdf

    9/17

    F.D. Queiroz et al. / Journal of Constructional Steel Research 63 (2007) 505521 513

    Fig. 19. Load (kN/m) vs. midspan deflection (mm) beam U4.

    Fig. 20. Ratio of stud forces vs. relative position of the stud beam A4.

    Fig. 21. Ratio of stud forces vs. relative position of the stud beam D1.

    Fig. 22. Ratio of stud forces vs. relative position of the stud beam E1.

    in the beam deflection values for the former case (i.e.,

    the structural system was stiffer). Although the conclusions

    presented were illustrated for the case of layered systems of

    wood, any mechanically-connected layered beam system (e.g.,

    steelconcrete beams connected by means of shear studs) canbe analysed in the same manner. The effect of the use of

    Fig. 23. Ratio of stud forces vs. relative position of the stud beam U4.

    Fig. 24. Slip distribution along span beam E1.

    Fig. 25. Slip distribution along span beam U4.

    connectors beyond the beam supports could be linked to the

    situation of glued beam ends described above.

    For the case of beams subjected to uniformly distributedload (as exemplified by Fig. 19 for beam U4), the increase

    in the stiffness of the system is more significant than for the

    case of beams subjected to midspan point loads (Figs. 1618).

    Considering the distribution of flexural moments in the regions

    next to the supports, the moments for a beam subjected to a

    UDL present proportionally higher values than the moments for

    a beam subjected to a midspan concentrated load, making the

    shear connectors in this region more important than the ones

    towards the midspan. In the point load case, the studs work in

    a more uniform manner along the beam. The overhang regions

    lead to a more rapid development of the effective breadth of

    the concrete slab (it is known that, for the standard case, the

    effective breadth is theoretically zero at the support), resultingin a more accentuated increase in the stiffness.

  • 8/10/2019 00b49525c79079cc8a000000 (1) (1).pdf

    10/17

    514 F.D. Queiroz et al. / Journal of Constructional Steel Research 63 (2007) 505521

    Table 4

    Level of shear connection (%) extended vs. standard beams

    Beam A2 A3 A4 A5 A6 B1 C1 D1 E1 U1 U3 U4

    Extended 231 186 137 123 95 116 114 136 148 155 177 90

    Standard 207 164 127 112 83 106 104 125 136 144 164 79

    Table 5

    Ultimate load results for the experimental and numerical analyses standard case

    Beam Pexp PLB(0.2%) PUB(0.35%) 1 = PLB/Pexp 2 = PUB/Pexp % = (2 1) 100

    A2 448 431 468 0.96 1.04 8

    A3 449 426 445 0.95 0.99 4

    A4a 523 444 469 0.85 0.90 5

    A5 468 448 467c 0.96 1.00 4

    A6 430b 444b 1.03

    B1 486 444 (440b ) 452 0.91 (0.91b ) 0.93 2

    C1 448 446 469 1.00 1.05 5

    D1 481 462 472 0.96 0.98 2

    E1 513 518 540 1.01 1.05 4

    U1 191 162 171 0.85 0.90 5

    U3 185 154 169 0.83 0.91 8

    U4 176b 166b 0.94

    Pexp, PLBand PUBare the test ultimate load, lower and upper bound loads, respectively (midspan load [kN]; uniformly distributed load [kN/m]).a The strength of the concrete used in the push-out test was much less than the concrete strength of the composite beam;b Stud failure, so bounds not applicable;c ANSYS analysis terminated.

    In addition, it is worth pointing out that the shear connectors

    in the overhang regions should be taken into account when

    calculating the level of shear connection, as they have an

    influence on the system behaviour. Therefore, if only the region

    between supports is considered, the connection level decreases,as shown inTable 4.

    In order to illustrate the results in terms of stud force

    distribution, tests A4, D1, E1 and U4 are considered. In

    Figs. 2023, stud force distribution graphs are plotted relating

    force ratio to stud position. It can be seen that, for the standard

    cases, there is a disturbance in the stud force distribution in

    the region near the supports, caused by the reduction of the

    effective breadth of the slab in this region. For the extended

    cases, this effect is not so significant, as the effective breadth

    of the slab is almost completely developed in the support

    region (it can be noticed that the tendency of the curve is

    maintained along the beam). In the region between the supports

    and midspan, only the steel and concrete deformation patternsdefine the behaviour.

    The slip at the steelconcrete interface along the beam

    axis for the cases E1 and U4 is plotted in Figs. 24 and

    25, respectively, for both extended and standard cases. It is

    evident from these graphs that the overhang regions also have

    a considerable effect on the slip values, mainly in the regions

    next to the supports. The disturbance of the slip distribution

    near the supports caused by the effective breadth of the slab

    can be noticed in Fig. 24, as well. In addition, it can be

    observed in Fig. 24 that the curve corresponding to the extended

    case is above the experimental curve. This fact may be due

    to a difference between the loadslip behaviour of the shear

    connectors in the tested composite beam and the loadslipbehaviour of the push-out test (used in the FE analysis).

    Moreover, the curves related to the standard case are also above

    the ones for the extended situation. Nevertheless, the sum of the

    stud forces is the same, provided that the connectors beyond the

    supports are considered.

    Table 5 presents the ultimate loads for all analysesconsidering the standard case (beams between supports). It can

    be observed that the values predicted by this model are very

    close to the ones obtained for the extended composite beam

    case (Table 3).

    The complete results for all standard beams are presented

    and discussed in Queiroz et al. [14,15], including a detailed

    study in terms of the absolute force carried by the shear studs

    for three different stud overall lengths (for a fixed diameter and

    spacing), and the absolute force carried by the shear studs for

    five different connector spacing values (for a fixed diameter and

    length).

    Based on this investigation, it can be seen that the

    continuation of the shear connection beyond the beam supportscan affect not only the overall response (e.g., shape of the

    loaddeflection curve), but also local results (e.g., slip and

    stud force distributions along the beam). Nevertheless, for

    the composite beams considered, the ultimate load was not

    significantly influenced.

    3.4. Effects of concrete slab strength and concrete strength for

    push-out specimens

    This section describes a comprehensive parametric analysis

    of the overall structural behaviour of steelconcrete composite

    beams when different concrete compressive strengths are

    present in the slab and in the associated push-out tests. Itwill always be the case, for example, for composite beams

  • 8/10/2019 00b49525c79079cc8a000000 (1) (1).pdf

    11/17

    F.D. Queiroz et al. / Journal of Constructional Steel Research 63 (2007) 505521 515

    Table 6

    Influence of the slab concrete strength

    Beam Diameter

    (mm)

    Overall length

    (mm)

    Spacing in pairs

    (mm)

    Cylinder strength of the concrete (N/mm2)

    Push-out (used in the shear connector

    springs)

    Slab (used in the slab finite elements)

    25.0

    26.9

    A2 19 102 159 30.7 28.8

    30.7

    32.6

    23.1

    24.8

    A5 19 102 274 28.0 26.4

    28.0

    29.6

    Table 7

    Influence of the concrete strength used in the push-out tests

    Beam Diameter

    (mm)

    Overall length

    (mm)

    Spacing in pairs

    (mm)

    Cylinder strength of the concrete (N/mm2)

    Slab (used in the slab finite

    elements)

    Push-out (used in the shear connector springs)

    19.7 (A4)

    A2 19 102 159 26.9 28.0 (A5)

    30.7

    19.7

    A4 19 102 216 20.1 28.0 (A5)

    30.7 (A2)

    19.7 (A4)

    A5 19 102 274 24.8 28.030.7 (A2)

    with precast slabs, for which results from push-out tests with

    solid concrete slabs are usually used. Two different effects

    are considered, utilising the standard model discussed in item

    3.3: one for which the slab concrete strength is fixed and

    the concrete strength for the associated push-out specimens

    is varied and another assuming the opposite situation. Results

    relating to the distribution of stud forces along the beam

    length and to the moment capacity of the composite system

    are discussed. The cases used for the parametric study are

    summarised inTables 6and7.

    Table 6 presents the scope of the analysis adopted toinvestigate the effect of the slab strength on the beam structural

    response. For each beam (A2 and A5) five different slab

    strengths are considered. The relative differences between two

    consecutive slab concrete strengths are approximately the same.

    For both beams, the diameter and overall length of the studs

    are also equal. The concrete strength of the associated push-out

    test carried out by [13] is fixed for each beam. Nevertheless, the

    shear connector spacing of each beam is distinct.Table 7 presents the range of values adopted in order

    to assess the influence of the concrete strength used in the

    associated push-out tests. For each beam (A2, A4 and A5)

    three different push-out strengths are assumed, i.e.: the concrete

    strength of the beams A2, A4 and A5. For all beams, thediameter and overall stud length are also equal. The slab

    concrete strength and the shear connector spacing are fixed and

    distinct for each beam.

    3.4.1. Influence of the slab concrete strength on the composite

    beam response

    The main focus of discussion is centred on: load versus

    midspan deflection curves, distribution of stud forces along the

    beam lengths and absolute forces carried by the shear studs for

    different slab strengths, with the same concrete strength being

    used in the associated push-out specimen.

    Figs. 2628andFigs. 2931present the FE model resultsfor the beams A2 and A5, respectively. By analysing the

    loaddeflection curves (Figs. 26 and 29), it is possible to

    observe that an increase in the slab concrete strength resulted in

    a stiffer system and in an increase in the moment capacity of the

    beam, represented by the ultimate bound points. Nevertheless,

    the former effect seems to be less significant than the latter

    one.

    According to Fig. 27 (absolute force carried by the shear

    studs for different slab strengths), an increase in the stud

    forces is observed as the slab concrete strength increases, most

    noticeably for the two beams with the highest slab concrete

    strengths (30.7 N/mm2 and 32.6 N/mm2). Based onFig. 30, a

    similar conclusion may be drawn for beam A5, i.e.: an increasein the stud forces occurs as the slab concrete strength increases.

  • 8/10/2019 00b49525c79079cc8a000000 (1) (1).pdf

    12/17

    516 F.D. Queiroz et al. / Journal of Constructional Steel Research 63 (2007) 505521

    Fig. 26. Load (kN) vs. midspan deflection (mm) for different slab concrete

    strengths beam A2.

    Fig. 27. Stud force vs. slab concrete strength beam A2 (lower and upper

    bounds).

    Fig. 28. Ratio of stud forces vs. relative position of the stud beam A2.

    The mode of collapse for the beams with the highest slab

    concrete strengths (28.0 N/mm2 and 29.6 N/mm2) was related

    to stud failure. The composite beam with slab concrete strength

    of 29.6 N/mm2 presented a lower bound closer to the stud

    failure point than did the composite beam with a 28.0 N/mm2-

    slab concrete strength (Fig. 29).

    Two concrete strengths are chosen for each beam (A2 and

    A5) to illustrate the distribution of stud forces along the beam

    lengths (Figs. 28 and 31, respectively). It can be noticed that, for

    the beam A2, the stud forces decrease towards the beam ends.

    This was not the case for the beam A5, where the stud forces

    did not vary significantly along the beam length.Fig. 30showsthat the variation in the stud forces for the beam A5 is smaller

    Fig. 29. Load (kN) vs. midspan deflection (mm) for different slab concrete

    strengths beam A5.

    Fig. 30. Stud force vs. slab concrete strength beam A5 (lower and upper

    bounds).

    Fig. 31. Ratio of stud forces vs. relative position of the stud beam A5.

    than the variation for the beam A2 (Fig. 27), therefore resultingin a more uniform distribution of forces along the beam.

    The analysis shows that, therefore, an increase in the slab

    concrete strength can have an influence not only on general

    aspects of behaviour (e.g., overall stiffness of the system and

    ultimate moment capacity) but also on local results (e.g., stud

    forces).

    3.4.2. Effect of the push-out concrete strength on the composite

    beam response

    The main focus of discussion is centred on varying the push-

    out test concrete strengths (by means of adopting different

    loadslip curves for the shear connector representation)

    maintaining the same slab concrete strength. Figs. 3234,3537,3840present the FE model results for the beams A2,

  • 8/10/2019 00b49525c79079cc8a000000 (1) (1).pdf

    13/17

    F.D. Queiroz et al. / Journal of Constructional Steel Research 63 (2007) 505521 517

    Fig. 32. Load (kN) vs.midspandeflection (mm) fordifferent concrete strengths

    beam A2.

    Fig. 33. Stud force vs.concrete strength beam A2 (lowerand upper bounds).

    Fig. 34. Ratio of stud forces vs. relative position of the stud beam A2.

    A4 and A5, respectively. Two concrete strengths are chosen foreach beam to illustrate the distribution of stud forces along the

    beam lengths (Figs. 34,37and40).

    The concrete strength used in the push-out test appears to

    have a small influence on both the overall behaviour of the

    composite beam and the stud force distribution, as can be

    seen in the graphs of applied load versus midspan deflection

    and absolute force carried by the shear studs versus concrete

    strengths used in the push-out tests (beam A2:Figs. 32and33;

    beam A4:Figs. 35and36; beam A5:Figs. 38and39).

    4. Sensitivity study

    In this section, an investigation is performed aimed atassessing the sensitivity of the overall response of composite

    Fig. 35. Load (kN) vs.midspandeflection (mm) fordifferent concrete strengths

    beam A4.

    Fig. 36. Stud force vs.concrete strength beam A4 (lowerand upper bounds).

    Fig. 37. Ratio of stud forces vs. relative position of the stud beam A4.

    Fig. 38. Load (kN) vs.midspandeflection (mm) fordifferent concrete strengths

    beam A5.

    beams (represented by loaddeflection curves and ultimate

    moment capacity) to likely variations in material strengths.

    Beams U1, U3 and U4 described in item 3 are used forthis analysis, as well as the standard model discussed in item

  • 8/10/2019 00b49525c79079cc8a000000 (1) (1).pdf

    14/17

    518 F.D. Queiroz et al. / Journal of Constructional Steel Research 63 (2007) 505521

    Fig. 39. Stud force vs.concrete strength beam A5 (lowerand upper bounds).

    Fig. 40. Ratio of stud forces vs. relative position of the stud beam A5.

    3.3. The numerical modelling of composite beams subjected

    to uniformly distributed loads is not a straightforward process,

    particularly due to their highly nonlinear behaviour. Hence, by

    identifying the key structural variables which can affect thesystem response, possible differences between numerical and

    tests results could be further comprehended.The range of steel and concrete strengths is assumed based

    on observed variations for nominally identical samples. The

    material properties studied are: (a) steel: web and flanges yield

    stresses and ratios between the strain at strain hardening and the

    yield strain; (b) concrete: compressive strength.The main results are as follows (a comprehensive description

    of the material properties adopted and the case studies

    considered can be found in Queiroz et al. [16]).

    4.1. Influence of the slab concrete strength

    It was noticed that, increasing the concrete strength, for afixed combination of yield stresses for the flanges and web:

    the lower and upper bounds increase in terms of both the

    ultimate load and the associated deflection. In some cases,

    the failure mode of the composite beam can change from

    slab crushing to stud failure; the corresponding loaddeflection curve becomes stiffer, but

    this effect is not very significant when compared with the

    effect previously mentioned.

    These results confirm the outcomes discussed in item3.4.1.

    4.2. Influence of the yield stress of the flanges and web

    It could be observed that, for a fixed slab concrete strengthand increasing either the flanges or the web yield stresses:

    the lower and upper bounds increase in terms of the ultimate

    load;

    the corresponding loaddeflection curve becomes stiffer. For

    the range of material properties considered in the parametricanalysis, the change in the loaddeflection curve was more

    significant for the cases in which the web yield stress was

    modified.

    4.3. Influence of the ratio of strains at strain hardening and at

    yield

    For the range of values considered, the steel ratio between

    the strain at strain hardening and the yield strain did not

    significantly affect the overall response of the composite beam.

    Based on the outcomes of the sensitivity analysis, it can

    be concluded that the web yield stress is the key structural

    parameter in the definition of the overall shape of the

    loaddeflection curve of the composite beams analysed. An

    increase in this property makes the curve become stiffer. In

    addition, the failure mode of the composite beams can be

    influenced by the concrete strength.

    5. Effect of partial shear connection

    In this section, the proposed finite element model is used

    to assess the influence of the effects of partial shear connection

    and partial interaction not only on the overall flexural behaviour

    of the structural system (represented by the loaddeflection

    curve), but also on the associated failure modes for either slabcrushing or stud failure, and on the distribution of stud shear

    forces along the beam length. In order to isolate the effect of

    the level of shear connection, the material properties (steel,

    concrete and shear connectors) and dimensions of the beam

    E1 tested by [13] are used for all analyses. Two load cases

    are considered: midspan point load and uniformly distributed

    load. The levels of shear connection are calculated using the

    actual material properties measured during the experimental

    procedure. In the present study, levels ranging from 47% to

    136% (by means of varying the number of shear connectors)

    are analysed using the standard finite element model

    (item 3.3).

    As the present model of composite beams adopts a solidelement to represent the concrete material and a nonlinear

    spring to model the shear connectors, the interaction between

    the slab and the studs cannot be investigated explicitly. A

    more refined model which adopts three-dimensional solid

    elements for the shear connectors would be the ideal option

    in order to capture this effect. Nevertheless, such an approach

    would certainly lead to complications as far as the total size

    of the finite element mesh (which can lead to a significant

    increase in running times) and numerical convergence problems

    are concerned, especially if the crushing capability of the

    concrete material is not disabled. Based on comparisons with

    experimental tests, the use of springs to model the studs has

    proved to be, in spite of its simplicity, very efficient in terms offinite element analyses.

  • 8/10/2019 00b49525c79079cc8a000000 (1) (1).pdf

    15/17

    F.D. Queiroz et al. / Journal of Constructional Steel Research 63 (2007) 505521 519

    Fig. 41. Load-deflection curve Midspan concentrated load.

    Fig. 42. Load-deflection curve Midspan concentrated load Initial

    stiffness.

    5.1. Midspan point load case

    The loadmidspan deflection curves for all cases (each one

    having a different level of shear connection) are shown inFig. 41.Fig. 42presents these curves just for the initial branch,

    in order to better illustrate the differences between the curves

    in terms of initial stiffness. It can be observed that, as expected,

    on decreasing the level of shear connection the system became

    more flexible, with reduced strength and stiffness. However, the

    decrease in stiffness did not seem to be as significant as the

    decrease in the ultimate load.

    The three higher levels (118%, 130% and 136%) resulted

    in very similar curves, in terms of both stiffness and ultimate

    load, represented by the lower and upper bound values, with

    the mode of failure being slab crushing at the midspan of

    the beam (location of maximum bending moment). Regarding

    level 100%, its mode of failure could be either slab crushingor stud failure, as a stud failure point was obtained between

    the lower and upper bounds. As the three lower levels (47%,

    71% and 89%) resulted in stud failure, the level 100% was an

    intermediate level between the two possible modes of failure.

    The ratio between the ultimate load and the load at the end

    of the elastic behaviour was not much affected by the distinct

    levels of shear connection, being in the range from 1.5 to 2

    for all cases, which is a common value as far as steelconcrete

    composite beams are concerned.

    The stud force distribution related to the composite beam

    ultimate load is plotted relating force ratio to stud position in

    Fig. 43. It can be seen that the variation of the stud force ratio

    of all curves is approximately 10%. Moreover, if the 136% levelis not considered, this variation is only 5%. This means that, at

    Fig. 43. Ratio of stud forces vs. relative position of the stud.

    Fig. 44. Load-deflection curve UDL.

    Fig. 45. Load-deflection curve UDL Initial stiffness.

    ultimate load and regardless of the level of shear connection, the

    connectors were able to redistribute the load almost uniformly

    among them.

    The results demonstrate that, therefore, the effects of partial

    interaction, which are increased by the use of partial shearconnection, can be neglected for levels of shear connection

    above 100%, as no significant improvement in terms of either

    strength or stiffness of the beam was observed.

    5.2. Uniformly distributed load case

    The loadmidspan deflection curves for all cases are shown

    inFig. 44. It can be observed that, as occurred for the midspan

    load case, decreasing the level of shear connection makes

    the system become more flexible, with reduced strength and

    stiffness. Once more, the decrease in stiffness did not seem to

    be as significant as the decrease in the ultimate load (Fig. 45).

    The two higher levels (130% and 136%) resulted in verysimilar curves, in terms of both stiffness and ultimate load,

  • 8/10/2019 00b49525c79079cc8a000000 (1) (1).pdf

    16/17

    520 F.D. Queiroz et al. / Journal of Constructional Steel Research 63 (2007) 505521

    Fig. 46. Ratio of stud forces vs. relative position of the stud.

    represented by the lower and upper bound values, with the

    mode of failure being slab crushing at the midspan of the

    beam (location of maximum bending moment). Regarding level

    118%, its mode of failure could be either slab crushing or studfailure, as a stud failure point was obtained between the lower

    and upper bounds. As the other four levels (47%, 71%, 89%

    and 100%) resulted in stud failure, the level 118% was an

    intermediate level between the two possible modes of failure.

    The ratio between the ultimate load and the load at the end

    of the elastic behaviour was similar to the one observed for the

    midspan point load case (from 1.5 to 2 for all cases).

    The stud force distribution graph (Fig. 46) is plotted relating

    force ratio to stud position. It can be noticed that the connectors

    were able to plastify and redistribute the load almost uniformly

    among them for a length of approximately 30% of the span

    from each of the supports (0 X/L 0.3). In the central

    region of the beam (0.3 X/L 0.5), the studs were lessheavily loaded. This behaviour occurred for all levels of shear

    connection, except the two highest ones (130% and 136%), for

    which the length of the central region increased by 10%.

    The results show that, for the UDL case, the effects of

    partial shear connection can be important for levels of shear

    connection above 100%. It was observed that the connectors

    were not loaded uniformly along the beam length (Fig. 46), with

    the stud forces decreasing towards the midspan of the beam.

    In addition, this behaviour is augmented by local deformations

    of the concrete slab. Consequently, more shear connectors

    are needed in order to obtain a total interaction level, for

    which the effects of partial interaction may be neglected.

    In spite of the fact that current design codes assume thesame strength for all shear connectors, as well as a uniform

    distribution of load among them for the ultimate limit state,

    an 18% difference between the total-interaction levels for the

    concentrated load (100%) and UDL (118%) cases was obtained.

    Nevertheless, this difference is not sufficiently significant to

    propose modifications to well accepted design methods.

    6. Conclusions

    A three-dimensional finite element model of composite

    beams is proposed based on the use of the commercial

    software ANSYS. It has proved to be effective in terms of

    predicting the loaddeflection response for beams subjected to

    concentrated or uniformly distributed loads, longitudinal slipat the steelconcrete interface, shear force carried by the studs

    and the mode of failure (stud failure or concrete crushing). It is

    also able to investigate beams with either full or partial shear

    connection. Comparisons against experimental results and

    against alternative numerical analyses indicate that the presentmodel can be used to perform extensive parametric studies.

    Based on the use of this model, it was shown that

    the continuation of the shear connection beyond the beam

    supports of simply supported beams can affect not only the

    overall system response, but also the slip and the stud force

    distributions along the beam.

    Results from a parametric analysis, designed to investigate

    the overall structural behaviour of composite beams when

    different concrete compressive strengths are used in the slab

    and in the associated push-out tests, revealed that the slab

    concrete strength (for a fixed concrete strength for the push-out

    specimen) can have an influence not only on the overall stiffness

    of the system and on its ultimate moment capacity, but also onthe stud forces. In addition, for a fixed slab concrete strength,

    it was observed that the concrete strength of push-out tests

    appears to have a small influence on both the overall behaviour

    of the composite beam and the stud force distribution.A sensitivity study was also undertaken, focused on the

    assessment of the influence of small variations in key input

    parameters (i.e., concrete and steel material properties) on the

    overall structural behaviour of composite beams. For the range

    of material properties considered, it was noticed that the web

    yield stress was the main structural parameter influencing the

    definition of the overall shape of the loaddeflection curve of

    the composite beams analysed. In addition, it could be observed

    that the slab concrete strength can affect the mode of failure ofthe composite beam, confirming the outcomes obtained in the

    previous parametric study.Finally, a study was carried out on the effects of partial

    shear connection/partial interaction. It was demonstrated that,

    by decreasing the level of shear connection, the composite sys-

    tem becomes more flexible, with reduced strength and stiffness,

    mainly for beams with less than 100%, for which the partial in-

    teraction effects are significant and must be taken into account.The proposed three-dimensional model provides the

    opportunity to develop insights that would be virtually

    impossible using experimental tests, due to costs and,

    especially, the dispersion of material properties that inevitably

    occurs in laboratory work. For instance, in the investigation

    of the influence of the level of shear connection, the concrete

    strength effect could be easily disregarded by the consideration

    of a fixed value for this property. Despite the fact that the

    proposed three-dimensional model is able to accurately provide

    a wide range of results, including the detection of local

    aspects of behaviour (e.g., local deformations of the concrete

    slab), a two-dimensional model could be the solution for

    more complex structural systems (e.g., beams with different

    degrees of continuity), due to numerical convergence aspects

    and processing times.

    Acknowledgment

    The authors would like to acknowledge the support providedby the Brazilian Foundation CAPES.

  • 8/10/2019 00b49525c79079cc8a000000 (1) (1).pdf

    17/17

    F.D. Queiroz et al. / Journal of Constructional Steel Research 63 (2007) 505521 521

    References

    [1] Nethercot DA. Composite construction. London: Spon Press; 2003.

    [2] Uy B, Nethercot DA. Effects of partial shear connection in semi-continuous composite beams. In: Hancock, et al., editors. Advances in

    structures. 2003. p. 7116.

    [3] Abdollahi A. Numerical strategies in the application of the FEM to RC

    structures-I. Computers and Structures 1996;58(6):117182.

    [4] Nethercot DA. The importance of combining experimental and numerical

    study in advancing structural engineering understanding. Journal of

    Constructional Steel Research 2002;58:128396.

    [5] Gattesco N. Analytical modelling of nonlinear behaviour of composite

    beams with deformable connection. Journal of Constructional Steel

    Research 1999;52:195218.

    [6] Pi YL, Bradford MA, Uy B. Second order nonlinear inelastic analysis of

    composite steelconcrete members. II: Applications. Journal of Structural

    Engineering, ASCE 2006;132(5):76271.

    [7] El-Lobody E, Lam D. Finite element analysis of steelconcrete composite

    girders. Advances in Structural Engineering 2003;6(4):26781.

    [8] ABAQUS user manual, version 5.3.1. 1080 Main Street, Pawtucket, RI02860-4847, USA: Hibbitt, Karlsson & Sorensen, Inc., 1994.

    [9] Lam D, El-Lobody E. Behaviour of headed stud shear connectors in

    composite beam. Journal of Structural Engineering, ASCE 2005;131(1):

    96107.

    [10] El-Lobody E, Young B. Performance of shear connection in composite

    beams with profiled steel sheeting. Journal of Constructional Steel

    Research 2006;62(7):68294.

    [11] Swanson Analysis Systems, ANSYS. Online manual, version 7.0 andTheory Reference. 7th ed. Swanson Analysis Systems, s.1., s.d.

    [12] Eurocode 4, ENV 1994-1-1:1992 Design of composite steel and

    concrete structures Part-1-1: General rules and rules for buildings.

    Brussels; 1992.

    [13] Chapman JC, Balakrishnan S. Experiments on composite beams. The

    Structural Engineer 1964;42(11):36983.

    [14] Queiroz FD, Vellasco PCGS, Nethercot DA. Structural assessment

    of composite beams using the Finite Element Method. In: Eurosteel

    conference of steel and composite structures, Eurosteel05, vol. B. 2005.

    p. 4.3-494.3-58.

    [15] Queiroz FD, Vellasco PCGS, Nethercot DA. A non-linear evaluation of

    steelconcrete composite beams by the Finite Element Method. In: Fourth

    international conference on advances in steel structures. 2005. p. 88102.

    [16] Queiroz FD, Vellasco PCGS, Nethercot DA. Numerical modelling

    of steelconcrete composite beams subjected to uniformly distributedloads. In: The tenth international conference on civil, structural and

    environmental engineering computing. 2005. paper 36. p. 118.

    [17] Goodman JR, Popov P. Layered beam systems with interlayer slip.

    Journal of the Structural Division 1968;253547.