ARTICLE IN PRESS

International Journal of Mechanical Sciences 49 (2007) 239–255

www.elsevier.com/locate/ijmecsci

Exact static analysis of partially composite beams and beam-columns

Ulf Arne Girhammar, Dan H. Pan1
Department of Applied Physics and Electronics, Civil Engineering, Umeå University, SE-901 87 Umeå, Sweden
Received 28 May 2002; received in revised form 3 July 2006; accepted 12 July 2006
Available online 9 November 2006

Abstract

The ordinary differential equations and general solutions for the deflection and internal actions and, especially, the pertaining
consistent boundary conditions for partially composite Euler–Bernoulli beams and beam-columns are presented. Static loading
conditions, including transverse and axial loading and first- and second-order analyses are considered. The theoretical procedure is
applicable to general loading and boundary conditions for uniform composite beams and beam-columns with interlayer slip. Further, the
exact closed form characteristic equations and their associated exact buckling length coefficients for composite columns with interlayer
slip are derived for the four Euler boundary conditions. It is shown that these coefficients are the same as those for ordinary fully
composite (solid) columns, except for the Euler clamped-pinned case. For the clamped-pinned case, the difference between the exact
buckling length coefficient and the corresponding value for solid columns is less than 1.8% depending on the so-called composite action
parameter and relative bending stiffness parameter. Correspondingly, the maximum deviation between the exact and approximate
buckling load is at most 2.5%. These small differences can in most practical cases be neglected. Also, the maximum theoretical range for
the relative bending stiffness for partially composite beams and beam-columns is derived. An effective bending stiffness, valuable in the
determination of the critical buckling load for partially composite members, is derived. This effective bending stiffness is also suitable for
analysing approximate deflections and internal actions or stresses in composite beams with flexible shear connection. The beam-column
analysis is applied to a specific case. The difference in the approaches to the first- and second-order analysis is illustrated and the results
clearly show the magnification in the actions and displacements due to the second-order effect. The magnification of the internal axial
forces is different from magnifications obtained for the other internal actions, since only that portion of an internal axial force that is
induced by bending is magnified by the second-order effect.
r 2006 Published by Elsevier Ltd.

Keywords: Composite beams/columns; Partial interaction; Interlayer slip; Elastic buckling; Buckling length coefficients; Variational methods

1. Introduction static applications and a companion paper considers the

corresponding dynamic applications, [1].
Composite structures of different materials are fre- The behaviour of these members is quite complex
quently used in building, bridge and shelter construction. because the shear connectors generally permit the devel-
In building construction, composite structures are often opment of only partial composite action between the
used as floor and wall elements, e.g. timber–concrete individual components of the member, and their analysis
elements composed of thin concrete plates attached to requires the consideration of the interlayer slip between the
wood studs by means of shear connectors. These structures subcomponents.
can be subjected to different kinds of static and dynamic The fundamental equations of the theory for one-
transverse loadings, and when used as wall panels they are dimensional, linear elastic, partial composite action for
also subjected to axial loads. This paper deals with the beams and columns subjected to static loads was first
developed by Stüssi [2], Switzerland, Granholm [3],
Sweden, Newmark et al. [4], USA, and Pleshkov [5],
Corresponding author. Tel.: +46 90 7867495; fax: +46 90 7866469. Russia, independently. The theory has been applied by
E-mail address: [email protected] (U.A. Girhammar). many analysts in the analysis of composite beams and
1
Present address: Ericsson AB, Stockholm, Sweden. columns with interlayer slip, e.g. Goodman [6], Amana and

0020-7403/$ - see front matter r 2006 Published by Elsevier Ltd.

doi:10.1016/j.ijmecsci.2006.07.005
 ARTICLE IN PRESS
240 U.A. Girhammar, D.H. Pan / International Journal of Mechanical Sciences 49 (2007) 239–255

Booth [7], concerning static loads, and Henghold [8], constant slip value (for example, zero) is a necessary and
concerning dynamic loads, to mention a few of the early consistent boundary condition in order to render a well-
works. One of the authors applied the partial composite posed formulation.
action theory to beams subjected to static loads, Girham- The strength of variational methods is that the particular
mar [9], and to dynamic loads, Girhammar [10], and to function that gives the minimum value of the potential of
beam-columns subjected to static loads, Girhammar [11]. the problem is the solution to the governing differential
Earlier analysis of composite beam-columns conducted by equation. The procedure gives at the same time all the
Girhammar and Gopu [12] dealt with only one particular admissible boundary conditions. From these boundary
axial loading case; they extended and generalized their conditions a well posed formulation, mechanically and
theory later, Girhammar and Gopu [13]; and Girhammar mathematically, may be chosen. These methods are well
and Pan [14] analysed composite members subjected to established and widely used, see e.g. Washizu [30] and
dynamic loads; see also Pan [15]. Recently, Ranzi et al. [16] Oden and Reddy [31].
presented general analytical solutions for statically loaded Girhammar and Gopu [13] developed in the classical
composite beams with partial interaction and Faella et al. way the governing sixth-order ordinary differential equa-
[17] an ‘‘exact’’ finite element model based on ‘‘exact’’ tion and its general solution for composite beam-columns
analytical expression of the stiffness matrix. Ranzi and with interlayer slip subjected to static loading, but did not
Bradford [18] and Jurkiewiez et al. [19] presented analytical discuss the proper boundary conditions. Also, they did not
solutions and a numerical model, respectively, for time- evaluate the exact buckling length coefficients or exact
dependent behaviour of partially composite beams. Dif- buckling loads as a function of the degree of composite
ferent kinds of numerical and finite element formulations action and the relative bending stiffness for the various
for the analysis of composite beams with interlayer slip Euler cases. They only proposed to use the buckling length
have been suggested, e.g. by Ayoub [20,21], Ranzi et al. [22] coefficients valid for ordinary fully composite (solid)
and Cas et al. [23]. Gara et al. [24] developed a finite columns in the evaluation of approximate buckling loads
element model for composite beams with both longitudinal for partially composite columns. A similar approximate
slip and vertical uplift of the interface connection. procedure is adopted for buckling of sandwich columns,
Dall’Asta [25] developed a three-dimensional theory for Plantema [32].
composite beams with shear weak connection dealing with This paper is an extension and generalization of the
combinations of bending in the symmetry plane, and theories and procedures given in Girhammar and Gopu
torsion and transverse bending in the plane parallel to the [13]. The purpose of this paper is to derive, by using
shear connector interface. variational methods, the ordinary differential equations for
The classical approach to the solution of structural and the deflection and internal actions and all the pertaining
solid mechanics problems, like the one mentioned above, is admissible boundary conditions for partially composite
to formulate the governing differential equation and obtain Euler–Bernoulli beams and beam-columns, both of first
the analytical solution. The corresponding boundary and second order. In particular, the aim is to derive
conditions are stated so as to satisfy the apparent the exact closed form characteristic equations within the
geometrical and mechanical conditions at the boundaries. validity of the Euler–Bernoulli beam theory and the
The stated set of boundary conditions, which reflects the associate buckling length coefficients for partially compo-
investigators view of the conditions at the boundary, need site beam-columns. These equations and coefficients are
not only be mechanically correct, the resulting formulation derived for the four different Euler cases as a function of
needs also be mathematically well posed, i.e. mathemati- the degree of composite action and the relative bending
cally consistent. A number of such boundary value stiffness. The exact buckling length coefficients are also
problems in structural mechanics, whose intuitive formula- compared to the corresponding coefficients for fully
tion led to incorrect boundary and matching conditions, composite (solid) beam-columns. In this paper, the
have been published, see e.g. Kerr [26]. Examples of coefficients for solid beam-columns are called approximate
attention to this critical issue of boundary conditions coefficients when they are applied also to partially
derived consistently with the energy functional in other composite beam-columns. An evaluation of the accuracy
related fields are the works of Askes and Metrikine [27] and of using these approximate buckling length coefficients in
Landis [28]. Thus, for some problems it is not a straight the determination of the fundamental buckling loads for
forward procedure to mathematically describe the geome- partial composite columns is also made. The theory is
trical and boundary conditions in a consistent way. applied to simply supported beam-columns in order to
For composite members with interlayer slip, boundary illustrate the magnification in the internal actions and
conditions chosen by inspection of what have appeared to displacements due to the second-order effects. An example
be the physical conditions of the body have later been of derivation of exact characteristic equations in the
proved to be wrong. For example, for a built-in end of a dynamics of vibrations in another related field is the work
partially composite member, the conditions of zero of Hashemi and Arsanjami [33]. For a corresponding
deflection and zero angular rotation, but non-constant slip analysis with respect to deriving the governing equations,
were assumed, [29]. It will be proved in this paper that a consistent boundary conditions and exact characteristic
 ARTICLE IN PRESS
U.A. Girhammar, D.H. Pan / International Journal of Mechanical Sciences 49 (2007) 239–255 241

equations for the critical load, see e.g. Papargyri-Beskou q (x)

et al. [34].
F F
cg,∞
2. Basic assumptions and conditions x, u

The present analysis of partially composite beams is L

based on the Euler–Bernoulli beam theory (for each part of
z, w
the composite beam), which neglects the effect of shear
deformation. The beams are subjected to only transverse Fig. 2. Composite beam-column with arbitrary boundary conditions
loading and the beam-columns to transverse loading and/ subjected to transverse, q(x), and axial loads, F (x, z ¼ coordinates, u,
w ¼ displacements and L ¼ length of composite component). The symbol
or external axial forces. The analyses are carried out on the () at the ends represents general geometrical and mechanical boundary
basis of the following simplifying assumptions: conditions (the mechanical conditions expressed in terms of general

functions, V x0 ¼ VðwÞx¼x and M x0 ¼ Mðdw=dxÞx¼x at the boundary,
0 0
(1) A relative slip can occur at the interface. Frictional x ¼ x0 ).
effects and uplift are neglected (cf. p. 4 below). The
interface slip is small.
(2) Discretely located interlayer connectors are regarded as loads. The x-axis of the co-ordinate system is located in the
continuous. The connector load (per unit length)-slip centroid of the fully composite section. The displacements
behaviour is linear elastic with a constant slip modulus in the x- and z-direction are denoted u and w, respectively.
K [N/m2]. The generic transverse loading function includes possible
(3) The constituent materials behave elastically in tension partially distributed loads, point forces and point mo-
and compression. ments, which can be represented by using Macauley’s
(4) The curvature of the different subcomponents is singularity functions (see e.g. Refs. [35,36]).
assumed to be the same. Axial tension loads are denoted by F and axial
(5) Axial loads are applied at the centre of gravity of the compression loads by P (F ¼ P). The axial load F, acting
fully composite section. at the centroid of the transformed cross-sectional area of
the corresponding fully composite member—referred to as
Full composite action (infinite slip stiffness, K-N) and the centroid of the member, for convenience—is the
non-composite action (zero slip stiffness, K-0) represent resultant of the axial loads, F1 and F2, acting at
upper and lower bounds for the partial composite action. the centroid of the top and bottom elements, respectively.
Moment, shear force, normal force, slip force per unit The location of F at the centroid implies that F1 and F2 are
length, axial displacement, interlayer slip and deflection are proportioned in accordance with the relative axial stiff-
denoted by M, V, N, Vs, u, Du, and w, respectively. nesses of the elements they are acting on. This type of axial
Fig. 1 shows the geometric parameters defining a typical load application ensures that the axial loads induce
composite beam with two sub-elements of different uniform axial strain in the member and do not contribute
materials. The subscripts ‘1’ and ‘2’ refer to the top and to the bending of the member.
bottom elements of the cross-section, respectively. Fig. 2 The geometrical and mechanical end conditions are not
shows a typical beam subjected to transverse loading, a specified, because they can be of any kind. Only the
general distributed load with an intensity q(x,t) and axial locations of the supports are represented in Fig. 2 by solid
circles. For the sake of clarity and illustration, the
mechanical boundary conditions (or boundary loads) can
b1 generally be expressed as functions of the deflection
w ¼ w(x) and  rotation dw/dx ¼ dw(x)/dx at the ends, i.e.
cg,1
V x0 ¼ V ðwÞx and M x0 ¼ Mðdw=dxÞx , describing shears
h1 0 0
1 (V) and moments (M) at the boundary (x0). If
r1 
cg,∞
zcg,∞ W B ðw; dw=dxÞx is a general form of work done by these
r 0
r2 shear forces and moments at the boundary, the shear force
r-zcg,∞
h2 cg,2 at the boundary is obtained by

2 qW B ðw; dw=dxÞ
V x0 ¼  (1)
qw x¼x0
b2
and the moment at the boundary by

Fig. 1. Geometric parameters of composite beam (cg,1, cg,2 and


cg,N ¼ centroid of sub-element 1, 2 and of fully composite section, qW B ðw; dw=dxÞ
respectively, zcg,N ¼ distance of the centroid of the fully composite section M x0 ¼ . (2)
to the centroid of sub-element 1).
qðdw=dxÞ x¼x0
 ARTICLE IN PRESS
242 U.A. Girhammar, D.H. Pan / International Journal of Mechanical Sciences 49 (2007) 239–255

To illustrate, the following two examples are given: relationships must be valid [13]:
N 1 þ N 2 ¼ 0, (5)
(1) A general end point loaded beam: For a constant shear
force, V x0 ¼ V 0 and a constant moment, M x0 ¼ M 0 d2 w d2 w
at x ¼ x0 , we get M ¼ M 1 þ M 2 þ N 1r ¼ E1I 1 þ E 2 I 2 þ N 1r
dx2 dx2
   2
dw  dw dw
W B w; ¼ V wj þ M . (3) ¼ EI 0 2 þ N 1 r, ð6Þ
dx  dx 
0 x¼x0 0
x¼x0 x¼x0 dx
 
dw dN 1 dN 2
V s ¼ K Du ¼ K u2  u1 þ r ¼ ¼ , (7)
(2) A general elastically end restrained beam: For a dx dx dx
vertical shear force, V x0 ¼ k1 w, with a vertical
translational spring coefficient k1, and a moment, dM
V ¼ ¼ V1 þ V2
M x0 ¼ k2 ðdw=dxÞ, with a rotational spring coefficient dx    
k2, at x ¼ x0 , we get dM 1 dM 2
¼  þ V s r1 þ  þ V s r2
    2  dx dx
dw  1 
2 1 dw 
W B w; ¼ k1 w  þ k2  . d3 w
dx  x¼x0 2 2 x¼x0dx  þ V s r.
¼  EI 0 ð8Þ
x¼x0
dx3
(4) The differential equation for the internal actions then
becomes
As later will be shown, it is evident that these general
d2 N 1
mechanical conditions alternatively can be regarded as  a2 N 1 ¼ bM (9)
inherent in the expressions for the total moment and shear dx2
force at the boundaries. or
In this study, a uniform beam is assumed, i.e. the bending  
d2 N 1 2 br d2 w
stiffness of the beam is assumed constant, EðxÞIðxÞ ¼ EI.  a 1  N 1 þ bEI 0 ¼ 0, (10)
dx2 a2 dx2
Likewise, the slope along the beam is assumed continuous.
In case of discontinuities in the flexural stiffness and the where
slope of the deflected curve, the present model need to be  
2 1 1 r2
extended, e.g. by expressing the discontinuities by means of a ¼K þ þ
E 1 A1 E 2 A2 EI 0
superimposition of Macauley’s singularity functions to the  
uniform one dimensional field of stiffness and slope, EA0 EAp r2
¼K 1þ
respectively (see e.g. [37]). EAp EA0 EI 0
Kr2
¼  , ð11Þ
EI 0 1  EI 0 =EI 1
3. Analysis of partially composite beams (F ¼ 0)
Kr
3.1. Basic equations b¼ , (12)
EI 0
)
Consider the differential element in Fig. 3, but without EA0 ¼ E 1 A1 þ E 2 A2
axial loading (F ¼ 0). For the element to be at equilibrium , (13a,b)
EAp ¼ E 1 A1  E 2 A2
and the displacements to be compatible, the following

q(x)
M 1 + dM 1
M V M1 V1 V1 + dV1 M + dM
cg,1
N1 N1 + dN1
Vs
N=F N=F
V2 cg,∞ V2 + dV2 M2 + dM2
M2

N2 cg,2 N 2 + dN2
V + dV
x dx

Fig. 3. Differential element in composite beam-column subjected to axial load, F ¼ F1+F2, and distributed transverse load, q(x), acting as shown. Shear
 z-axis direction, which gives qðxÞ ¼ dV =dx. The direction of moments is defined in such a way as to produce
forces at the right hand side is positive in the
a positive curvature of the beam, i.e. d2 w dx2 ¼ M=EI. (The notation used for the quantities corresponds to the case without axial forces (F ¼ 0). The
dash notation () used for first-order quantities is not shown in the figure.)
 ARTICLE IN PRESS
U.A. Girhammar, D.H. Pan / International Journal of Mechanical Sciences 49 (2007) 239–255 243

9
EI 0 ¼ E 1 I 1 þ E 2 I 2 >
= i.e.
EI 0 EAp r2 . (14a,b)
EI 1 ¼ ¼ EI 0 þ >
1  br=a2 EA0 ; dP ¼ d U b ½M i ðwÞ þ U a ½N i ðui Þ þ U s ½V s ðui Þ
 
Subscripts 0 and N denote the properties for non- dw
þl½ f ðui ; wÞ  W w; ¼ 0, ð21Þ
composite and fully composite beams, respectively. The dx
parameter a2 is proportional to the slip modulus K and
where Ub is the strain energy due to internal bending moments
therefore expresses the degree of composite action. For
Mi, Ua the strain energy due to internal axial forces Ni, Us the
convenience, a non-dimensional composite action para-
strain energy due to connector deformations or interlayer slip
meter, aL, and a non-dimensional relative bending stiffness
force Vs, l a Lagrange multiplier, f the constraint condition
parameter, EI 0 =EI 1 , are introduced, cf. [13,14]. According
expressing the relation between the internal axial strain dui/dx
to Eq. (11), the composite action parameter is defined as
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi and the deflection w(x) corresponding to Eq. (20), and W the
  potential energy due to external loadings. These quantities can
1 1 r2
aL ¼ K þ þ L be expressed as follows:
E 1 A1 E 2 A2 EI 0
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Z L Z  2 2
Kr2 1 d2 w 1 L d w
¼   L. ð15Þ Ub ¼ ðM 1 þ M 2 Þ 2 dx ¼ EI 0 dx;
EI 0 1  EI 0 =EI 1 0 2 dx 2 0 dx2
(22)
When the beam has partial interaction, the practical range
for this parameter is for many applications of the order of Z  L 
1 du1 du2
0:1oaLo100 or often even more narrow, 1oaLo10. For Ua ¼ N1 þ N2 dx
0 2 dx dx
lesser values, the member approaches a non-composite Z  2
beam and for higher values a fully composite beam. Using 1 L E 1 A1 du1
¼ EA0 dx, ð23Þ
Eq. (14b), the relative bending stiffness parameter can be 2 0 E 2 A2 dx
defined as
Z L Z
EI 0 br 1 1 1 L V 2s
¼1 2 ¼ . (16) Us ¼ V s Du dx ¼ dx
EI 1 a 1 þ EAp r2 =ðEA0 EI 0 Þ 0 2 2 0 K
Z  2
For sub-elements of rectangular cross-sections, the 1 L ðE 1 A1 Þ2 d2 u1
¼ dx, ð24Þ
interval for the relative bending stiffness parameter can 2 0 K dx2
be shown to be (Appendix A) Z  
L
d3 u1 2 br du1
1 EI 0 lf ¼ l E 1 A1  E A a 1 
p o1. (17) 0 dx3
1 1
a2 dx
4 EI 1
d2 w
For the limit case EI 0 =EI 1 ¼ 1, the two-member compo- þbEI 0 dx, ð25Þ
nent reduces to a one-member component. dx2
To obtain the governing differential equations the Z    
problem will subsequently be approached by assuming
L
dw  dw 
W¼ qw dx þ W B w; þ W w; ;
dx x¼0 dx x¼L
B
the displacements as unknown. Expressed in displace- 0
ments, the internal axial and slip forces can be written as (26)
dui where WB is the general form of work done by the shear
N i ¼ E i Ai ; i ¼ 1; 2, (18)
dx forces and moments at the boundary. The strain energy
produced by the shear force V (or V1 and V2) is neglected
d2 u1 d2 u2 according to the Euler–Bernoulli beam theory. Inserting
V s ¼ K Du ¼ E 1 A1 ¼ E 2 A 2 (19)
dx2 dx2 expressions (22)–(26) in Eq. (21), differentiating, integrating
and Eq. (11) becomes by parts, and collecting terms we obtain
  Z L 
d3 u1 br du1 d2 w d4 w d2 l
E 1 A1 3  E 1 A1 a2 1  2 þ bEI 0 2 ¼ 0. (20) EI 0 4 þ bEI 0 2  q dw dx
dx a dx dx 0 dx dx
Z L
EA0 d2 u1 E 1 A1 d4 u1 d3 l
þ E 1 A1  þ 
3.2. Problem formulation 0 E 2 A2 dx2 K dx4 dx3
  Z L
br dl d3 u1
The variational approach used here is a procedure based þa2 1  2 du1 dx þ E 1 A1 3
a dx 0 dx
on the principle, which states that the displacements at the   2
equilibrium position occur such that the potential energy of br du1 d w
E 1 A1 a2 1  2 þ b EI 0 2 dl dx
a stable system is a minimum value (see e.g. [38, p. 318]), a dx dx
 ARTICLE IN PRESS
244 U.A. Girhammar, D.H. Pan / International Journal of Mechanical Sciences 49 (2007) 239–255

d3 w dl qW B equations we arrive at
þ EI 0 3
 bEI 0  dwjL0
dx dx qw d6 w d4 w a2 1 d2 q
L  a2 4 ¼  qþ . (33)
d2 w qW B dw dx 6 dx EI 1 EI 0 dx2
þ EI 0 2 þ bEI 0 l  d
dx qðdw=dxÞ dx 0  
3 2 du1 EI 1 d4 w 2 EI 0 d2 w q
EA0 du1 E 1 A1 d u1 d l ¼ 2  4 þa 1 þ . (34)
þ E 1 A1  þ dx a rE 1 A1 dx EI 1 dx2 EI 0
E 2 A2 dx K dx3 dx2
  L   Explicit expressions for N1, N2, M, M1, M2, Vs, V, V1
br  E 1 A1 d2 u1 dl
2 
a 1  2 l du1  þ E 1 A1  and V2 are readily available by Eqs. (5)–(8) and (18)
a K dx2 dx
0 and (19). It is evident that the above differential equations
 L
du1 L d2 u1  for the deflection and internal actions are equivalent
d þ E 1 A1 l d 2  ¼ 0. ð27Þ
dx 0 dx 0 to those proposed in an earlier paper by Girhammar
and Gopu [13] (for the case without axial forces) and
Then using standard arguments from the calculus of Girhammar and Pan [14] (for the case without dynamic
variations, each bracketed factor within the integrals must loads).
be identically zero and the remaining evaluated terms zero. For the boundary conditions (31) or (32) to hold, we
This applied to the second, seventh and sixth term, gives get the following alternatives at each end (cf. also Eqs. (1)
and (2)):
E 1 A1 du1
l¼ (28) dw
K dx M  M x0 ¼ 0 or ¼ prescribed constant
dx
and the coupled governing equations for the composite beam
at x ¼ 0 and x ¼ L, ð35a; bÞ
are given by

d4 w d3 u1 V  V x0 ¼ 0 or w ¼ prescribed constant
EI 0 þ rE 1 A1  q ¼ 0, (29)
dx4 dx3 at x ¼ 0 and x ¼ L, ð36a; bÞ

  N 1 ¼ 0 or Du ¼ prescribed constant
d3 u1 2 br du1 d2 w
E 1 A1 3  E 1 A1 a 1  2 þ bEI 0 2 ¼ 0 (30) at x ¼ 0 and x ¼ L. ð37a; bÞ
dx a dx dx
and the pertaining boundary conditions These conditions will be called general admissible
boundary conditions. The special cases, when the constants

d2 w du1 qW B dwL in Eqs. (35b)–(37b) equal zero, will be called restricted
EI 0 2 þ rE 1 A1  d ¼ 0,
dx dx qðdw=dxÞ dx 0 admissible boundary conditions. Special cases of those
L restricted admissible boundary conditions of particular
d3 w d2 u1 qW B  engineering interest are pinned, clamped and free ends,
EI 0 3  rE 1 A1 2  dw ¼ 0,
dx dx qw 0 here called engineering admissible boundary conditions.
 2 L For a clamped end, for example, it is evident from
du1 E 1 A1 d u1 
E 1 A1 d ¼0 ð31a; b; cÞ Eqs. (37a) and (37b) that either the internal axial force
dx K dx2 0 must equal zero, i.e. an axially rolling or sliding clamped
or, in view of Eqs. (6)(8) and (18)(20), support, or the slip must equal a prescribed constant (e.g.
zero), i.e. an axially non-displaceable or fixed clamped end.
L
qW B dw The latter condition is not fully self-evident if only based
M d ¼ 0,
qðdw=dxÞ dx 0 on inspection, cf. e.g. [29].
L For convenience, the moment and shear force for the
qW B 
V dw ¼ 0, whole cross-section, and the axial force in the first
qw 0 component in Eqs. (6)–(8) are explicitly expressed in terms
L
N 1 dðDuÞ0 ¼ 0. ð32a; b; cÞ of w as

From Eqs. (32a) and (32b) it is evident that the boundary EI 1 d4 w 2

2d w q
M¼  þ a þ , (38)
conditions expressed by the function W B ðor M x0 and V x0 Þ, a2 dx4 dx2 EI 0
also could have been regarded as part of the total moment, M,
and the total shear force, V, at the boundary. dM EI 1 d5 w 3
2d w 1 dq
For the limit case of non-composite action (Vs ¼ 0; aL- V ¼ ¼ 2 5
 a 3
 , (39)
dx a dx dx EI 0 dx
0), du1/dx-0 and for the limit case of fully composite
action (Du ¼ 0; aL-N), d2 u1 dx2 ! 0. Eq. (29) then  
reduces to the conventional fourth-order differential EI 1 d4 w 2 EI 0 d2 w q
N1 ¼ 2  4 þ a 1  2
þ . (40)
equation valid for solid beams. Combining the coupled a r dx EI 1 dx EI 0
 ARTICLE IN PRESS
U.A. Girhammar, D.H. Pan / International Journal of Mechanical Sciences 49 (2007) 239–255 245

The engineering admissible boundary conditions are Q

then expressed as follows: For a pinned end ðw ¼ M ¼
N 1 ¼ M x0 ¼ 0 at x ¼ x0 Þ, they are given by

w ¼ 0,
d2 w
¼ 0,
dx2
d4 w q
¼ . ð41a; b; cÞ L /2 L /2
dx4 EI 0
For a clamped end ðw ¼ dw=dx ¼ Du ¼ 0 at x ¼ x0 Þ, by
Fig. 4. Simply supported composite beam subjected to a point load at
w ¼ 0, midspan.

dw
¼ 0,
dx
 
d5 w 2 EI 0 d3 w 1 dq respect to resilience or sometimes called springiness:
 a 1  ¼ . ð42a; b; cÞ
dx5 EI 1 dx3 EI 0 dx wmax p1:5 mm for Q ¼ 1 kN: (46)
And for a free end ðM ¼ V ¼ N 1 ¼ M x0 ¼ V x0 ¼ 0 at
For a floor, considered in a simplified way, as a
x ¼ x0 Þ, by
composite joist-flooring layer system, the deflection w for
d2 w this particular case is obtained as
¼ 0,  
dx2 wslip;max
wmax ¼ w1;max 1 þ
d4 w q w1;max
4
¼ , "  
dx EI 0 QL3 12 EI 1
d5 w 3 ¼ 1þ  1
2d w 1 dq 48EI 1 ðaLÞ2 EI 0
 a ¼ . ð43a; b; cÞ
dx 5 dx 3 EI 0 dx  !#
4 sinh2 aL=2
It is evident that the mathematically consistent boundary  1 , ð47Þ
conditions above that are of basic engineering interest aL sinhðaLÞ
verify those proposed in an earlier paper by Girhammar where wN,max is the maximum deflection for a fully
and Pan [14] (in case of only static loading). composite section.
The general solution of Eq. (33) is given by Girhammar Fig. 5 shows the deflections w as a function of the non-
and Gopu [13] as dimensional composite action parameter aL and the
w ¼ a1 sinhðaxÞ þ a2 coshðaxÞ þ a3 x3 þ a4 x2 relative bending stiffness parameter EI0/EIN.
It is evident from Fig. 5 that the effect of composite
þ a5 x þ a6 þ wps , ð44Þ
action is substantial for composite action parameters in the
where a1 y a6 ¼ constants that depend on the boundary range 1oaLo10. For a composite section according to
conditions. The particular solution wps in Eq. (44) is then Fig. 6, the maximum deflection according to Eq. (47)
given by becomes wmax ¼ 2.22  1.386 ¼ 3.08 mm, which makes it
Z x not suitable for a residential floor without structural
1 EI 1 d2 qðsÞ
wps ¼ 5 a2 qðsÞ  changes.
a EI 1 0 EI 0 ds2
a3
 aðx  sÞ þ ðx  sÞ3  sinh½aðx  sÞ ds, ð45Þ 4. First-order analysis of partially composite beam-columns
6
where s is a dummy variable. Knowing the solution for w 4.1. Governing equations and boundary conditions
for a given set of boundary conditions, we can easily
evaluate the various internal actions as mentioned above. In this section, the axial load effects are evaluated by a
first-order analysis. First-order quantities are distinguished
3.3. Application of analysis procedure from the previous case of a beam with no axial loads by a
dash (–). In the first-order case, the applied axial forces
The partial composite action theory developed in the will only affect the internal normal forces, but not the
preceding section is applied to a simply supported deflection. With axial tensile forces (F) acting on the
composite beam subjected to a point load Q in the midspan element (see Fig. 3), Eq. (5) changes to
as shown in Fig. 4. This simple example is of practical    
N̄ 1 þ N̄ 2 ¼ F ¼ F 1 þ F 2 or N̄ 1  F 1 þ N̄ 2  F 2 ¼ 0,
interest. Eurocode 5 stipulates that for residential wooden
floors the following requirement should be satisfied with (48a,b)
 ARTICLE IN PRESS
246 U.A. Girhammar, D.H. Pan / International Journal of Mechanical Sciences 49 (2007) 239–255

EI0 1
=

Relative Maximum Deflection

(1) (1)
EI∞ 4
3 EI0 1
wmax/w∞,max (2) =
EI∞ 2

(2) EI0 1
2 (3) =
EI∞ 4/3

(3) EI0
(4) =1
EI∞
(4)
1
0.1 1 10 100
Composite Action Parameter αL
α

Fig. 5. Dimensionless deflection for a simply supported partially composite beam subjected to a point load at midspan obtained by exact analysis vs. the
partial composite action parameter aL and the relative bending stiffness parameter EI 0 =EI 1 . It is evident that the influence of partial interaction is most
prevalent for partial composite action parameters in the range, 1oaLo10. (Note that the value of the stiffness parameter, EI 0 =EI 1 ¼ 1=n, gives the value
of the relative deflection as, wmax/wN,max ¼ n).

where differential equations for the deflection:

E 1 A1 d6 w̄ 4
a2 1 d2 q
F1 ¼ F, 2 d w̄
EA0  a ¼  q þ , (55)
dx6 dx4 EI 1 EI 0 dx2
E 2 A2
F2 ¼ F. ð49a; bÞ  
EA0 dū1 F1 EI 1 d4 w̄ EI 0 d2 w̄ q
 ¼ 2  4 þ a2 1  þ .
The other equations given in Section 3.1 are thus still dx E 1 A1 a rE 1 A1 dx EI 1 dx 2 EI 0
valid if only the following transformations are made: (56)
N 1 ! N̄ 1  F 1 and N 2 ! N̄ 2  F 2 , (50a,b)
Likewise, the following alternative boundary conditions
must hold at each end
dN 1 dN̄ 1 dN 2 dN̄ 2
! and ! , (51a,b) dw̄
dx dx dx dx M̄  M x0 ¼ 0 or ¼ prescribed constant
dx
M; V ; V s ; w; ui ; Du ! M̄; V̄ ; V̄ s ; w̄; ūi ; Dū (52) at x ¼ 0 and x ¼ L, ð57a; bÞ

In particular, the constraint condition (10) or (20)

becomes V̄  V x0 ¼ 0 or w̄ ¼ prescribed constant
   at x ¼ 0 and x ¼ L, ð58a; bÞ
d3 ū1 2 br dū1 F1
E 1 A1 3  E 1 A1 a 1  2 
dx a dx E 1 A1
N̄ 1  F 1 ¼ 0 or Dū ¼ prescribed constant
d2 w̄
þ bEI 0 2 ¼ 0. ð53Þ at x ¼ 0 and x ¼ L, ð59a; bÞ
dx
RL  
Hence, the additional term, dPf̄ ¼ 0 a2 1  br=a2  where the internal actions are given by
F 1 dl̄ dx, needs to be added to the equation corresponding EI 1 d4 w̄ 2
2 d w̄ q
to Eq. (27). The work done by the applied axial loads, M̄ ¼ 2
 4
þ a 2
þ , (60)
a dx dx EI 0
N x0 ¼ F at x ¼ 0 and x ¼ L, is given by
F 1L F 2L F 2L dM̄ EI 1 d5 w̄ d3 w̄ 1 dq
W B ðuÞ ¼ F 1 u1 þ F 2 u2 ¼ F 1 þ F2 ¼ . V̄ ¼  ¼ 2  a2 3  , (61)
E 1 A1 E 2 A2 EA0 dx a dx 5 dx EI 0 dx
(54)
 
Since, dPW B ¼ 0, the potential energy of the system is EI 1 d4 w̄ 2 EI 0 d2 w̄ q
N̄ 1  F 1 ¼ 2  4 þ a 1  2
þ .
not changed. The various equations, governing differential ar dx EI 1 dx EI 0
equations, boundary conditions and the general solution (62)
according to Eqs. (21)–(45), presented in Section 3.2, are
still applicable if only the transformations are made Eqs. (41)–(43) still hold if dash notation for first-order
according to Eqs. (50)–(52). We thus get the following quantities is used.
 ARTICLE IN PRESS
U.A. Girhammar, D.H. Pan / International Journal of Mechanical Sciences 49 (2007) 239–255 247

4.2. Application of first-order analysis procedure according to Fig. 6. The deflection, slip and internal actions
are given in the table. These exact values will later be
The theory in the previous section is applied to a partial compared to the exact second-order results (Section 5.5)
composite beam-column subjected to an axial compression and to the approximate values according to the procedures
load, F ¼ P, and a uniformly distributed load, q0, as discussed in Section 5.6.
shown in Fig. 6. The first-order analysis method is applied
to determine the deflection and the internal actions. The 5. Second-order analysis of partially composite
homogeneous and particular solution for the first-order beam-columns
case for the beam-column are obtained by solving Eqs. (44)
and (45) to give the following general solution: 5.1. Governing equations and boundary conditions
 4 
q L4 x 2x3 x
w̄ ¼ 0  þ In this section, the axial load effects are evaluated by a
24EI 1 L4 L3 L second-order analysis. Second-order quantities are distin-
  
24 EI 1 cosh aðx  L=2Þ guished from the previous ones by a tilde (). The second-
þ 1
ðaLÞ4 EI 0 coshðaL=2Þ order analysis is also based on the key assumptions utilized
ðaLÞ2 x2 ðaLÞ2 x in the first-order case. The deformed geometry of the
 þ 1 . ð63Þ member is considered in this case to establish the second-
2 L2 2 L
order effects, Fig. 7. Since the slope of the beam-column is
It is obvious from Eq. (63) that the axial forces do not small, the component of the applied axial forces along the
influence the deflection in the first-order case. Knowing the deformed axis of the element can be taken as approxi-
solutions in terms of the displacements in Eq. (63), the mately equal to their horizontal values. The influence of q
internal actions are readily given by Eqs. (6)–(8) and (48). along the deformed axis of the element can be ignored.
Table 1 presents the results obtained from the first-order Eqs. (48)–(54) are still valid; the dash (–) denoting the
analysis applied to the composite beam-column example first-order quantities are changed to the tilde () to denote

q0 = 1kN/m

P = 50 kN P = 50 kN
cg,∞ x, u

L=4m

(a) z, w

b1 = 0.300 m

cg,1
h1 = 0.050 1
r1 = 0.025 m zcg,∞ = 0.025 m
cg,∞
r2 = 0.075 m r = 0.100 m
h2 = 0.150 cg,2
2

(b) b2 = 0.050 m

Modulus Axial Stiffness Bending Stiffness Buckling Magnification

Parameters
[GPa] [MNm] [MNm2] Loads [kN] Factors

K = 0.05 E1 A1 = 180 E1I1 = 0.0375 α = 2.10819 m

-1
Pcr = 271.0 ν = 1.226
E1 = 12* E2 A2 = 60 E2 I 2 = 0.1125 Pcr , ∞ = 370.1 ν ∞ = 1.156
α L = 8.433
E2 = 8 EA0 = 240 EI 0 = 0.150 θ 1 = 2.04936 m
-1 P/Pcr , ∞ = 0.185
EI ∞ = 0.600 -1
θ 2 = 0.29698 m
EI eff = 0.732 EI ∞
∗ For structural concrete, a higher E-value should be used.

Fig. 6. Uniformly loaded, simply supported composite beam-column of timber and concrete with the relative stiffness parameter EI 0 =EI 1 ¼ 0:250
(the concrete is assumed to be uncracked) (P1 ¼ 37:5 kN and P2 ¼ 12:5 kN).
 ARTICLE IN PRESS
248 U.A. Girhammar, D.H. Pan / International Journal of Mechanical Sciences 49 (2007) 239–255

Table 1
Exact first- and second-order results for the partially composite beam-column according to Fig. 6

Displace-ment/internal action First-order analysis Second-order analysis Exact Approx.

magnification magnification
factor factor
9
wmax 7.560 mm 9.280 mm 1.228 >>
> 1.226d
M1,max 0.1659 kNm 0.2053 kNm 1.237 >>
>
>
=
M2,max 0.4978 kNm 0.6160 kNm 1.237
Mmax 2.000 kNm 2.463 kNm 1.232 >>
>
N1,max 50.862 kN ¼ 13.362a37.5a 53.927 kN ¼ 16.427a37.5b 1.229c >
>
>
>
c ;
N2,max 0.862 kN ¼ 13.362a–12.5b 3.927 kN ¼ 16.427a12.5b 1.229
9
Vs,max 11.444 kN/m 13.887 kN/m 1.213 >> 1.226e
=
V1,max 0.5000 kN 0.5940 kN 1.188
V2,max 1.500 kN 1.782 kN 1.188 >>
;
Vmax 2.000 kN 2.376 kN 1.188
a
The term refers to that part of the normal force that is induced by the bending moment.
b
The term refers to that part of the normal force that is induced by the applied axial load.
c
The magnification of the internal axial forces is different from the magnifications obtained for the other internal actions since only that portion of the
internal axial force that is induced by bending is magnified by the second-order effect.
d
The approximate magnification factor for ordinary beams is given v ¼ 1/(1P/Pcr) ¼ 1.226.
e
The approximate magnification factor is not fully applicable for slip and shear forces since these quantities are obtained by differentiation of the normal
forces and moments, respectively. For further details, see [40].

M
zcg ,∞ V M1 q(x)
V1

N1
N F cg ,1 V1 + dV1
M2
V2 M 1 + dM 1
w' N1 + dN1 M + dM
N2 cg ,∞ Vs

w '+ w '' dx
cg , 2 M 2 + dM 2
N2 + dN 2 N F
V2 + dV2
dx V + dV
x

Fig. 7. Deformed differential element in composite beam-column subjected to an axial load, F1+F2 ¼ F, and distributed transverse load of intensity, q(x).
(The tilde notation () used for second-order quantities is not shown in the figure.)

the second-order quantities. The applied axial forces will, of the deformed beam-column, V~ F ;x0 , equals
 the compo-
~ 
in the second-order case, affect both the axial displace- nent of the applied axial load,  F ðdw=dxÞ x¼x0
, at the free
ments and the deflection. end, i.e. V~ F ;x0 ¼ F ðdw=dxÞ
~  .
x¼x0
Eqs. (6)–(10) are still valid for the second-order case (if As mentioned, the constraint condition and its variation
Eq. (50) with tilde notation is taking into account). The in the second-order case (with tilde () notation) are not
difference is that all actions in the second-order case changed in relation to the first-order case (with dash
correspond to the deformed shape of the beam-column. notation according to Eq. (53)). However, the change (D)
Due to the effect of the applied axial forces on the internal of the loading potential and its variation in the second-
shear forces, the second-order shearing force (perpendicu- order case, due to the axial displacement of the beam-
lar to the deflected axis of the beam) is related to the column, in addition to the changes in the first-order case
corresponding first-order one (perpendicular to the unde- according to Eq. (54), becomes
flected axis of the beam) by the expression (cf. [39]) Z L  2
dw~ dw~ ~ ¼ 1 dw~
V 2nd ¼ V~ ¼ V 1st  F ¼ V̄  F (64a) DW F dx: (65)
dx dx 2 0 dx
or Z L
  dw~ dw~
dV~ dV̄ d2 w~ d2 w~ ~ ¼ 
D dW d F dx
¼  F 2 ¼ q  F 2 (64b) 0 dx dx
dx dx dx dx L Z L
For example, the boundary condition for a cantilever dw~  d2 w~
¼ F dw~  þ F 2 dw~ dx: ð66Þ
subjected to an axial force at the end, the true shear force dx dx
0 0
 ARTICLE IN PRESS
U.A. Girhammar, D.H. Pan / International Journal of Mechanical Sciences 49 (2007) 239–255 249

 
The governing differential equation for the composite For a clamped end w~ ¼ dw=dx
~ ¼ Du~ ¼ 0 at x ¼ x0 , as
system (55) and the corresponding equation for the axial
w~ ¼ 0,
strain (56) in the first-order case change to
dw~
  ¼ 0,
d6 w~ 2 F d4 w~ a2 F d2 w~ dx
 a þ þ   3
dx6 EI 0 dx4 EI 1 dx2 d5 w~ 2 EI 0 F d w~ 1 dq
a 1 þ ¼ . ð76a; b; cÞ
a2 1 d2 q dx5 EI 1 a2 EI 0 dx3 EI 0 dx
¼ qþ , ð67Þ 
EI 1 EI 0 dx2 ~ ¼ V~ ¼ N~ 1  F 1 ¼ M x ¼ V x ¼ 0
And for a free end M 0 0

  at x ¼ x0 Þ, as
du~ 1 F1 EI 1 d4 w~ EI 0
 ¼ 2  4 þ a2 1 d2 w~
dx E 1 A1 a rE 1 A1 dx EI 1 ¼ 0,
dx2
F d2 w~ q d4 w~ q
þ þ . ð68Þ ¼ ,
a2 EI 0 dx2 EI 0 dx4 EI 0
  3
The pertaining consistent boundary conditions are given d5 w~ 2 F d w~ a2 F dw~ 1 dq
 a 1 þ þ ¼ . ð77a; b; cÞ
by dx5 a2 EI 0 dx3 EI 1 dx EI 0 dx

~  M x ¼ 0 or dw~ ¼ prescribed constant

M To find the solution of Eq. (67) and evaluate the response
0
dx of the beam-column, it is necessary to distinguish between
at x ¼ 0 and x ¼ L, ð69a; bÞ axial tension loading (F) and axial compression loading
ðF ¼ PÞ. The general solutions of the two cases have been
 given by Girhammar and Gopu [13]. Here we will limit
~ dw~ 
V þF   V x0 ¼ 0 or w~ ¼ prescribed constant ourselves to the case of bending and axial compression.
dx x¼x0
at x ¼ 0 and x ¼ L, ð70a; bÞ 5.2. Bending and axial compression

For composite beam-columns in axial compression

N~ 1  F 1 ¼ 0 or Du~ ¼ prescribed constant
ðF ¼ PÞ, the general solution is given by Girhammar
at x ¼ 0 and x ¼ L, ð71a; bÞ and Gopu [13] as
where the internal actions are given by w~ ¼ e1 sinhðy1 xÞ þ e2 coshðy1 xÞ þ e3 sinðy2 xÞ
  2 þ e4 cosðy2 xÞ þ e5 x þ e6 þ w~ ps . ð78Þ
~ EI 1 d4 w~ 2 F d w~ q
M ¼ 2  4 þa 1þ 2 2
þ , (72)
a dx a EI 0 dx EI 0 The particular solution can be obtained as
Z x
  3 1 EI 1 d2 qðsÞ
~
dM EI 1 d5 w~ F d w~ w~ ps ¼ a2 qðsÞ 
V~ ¼  ¼ 2 2
a 1þ 2 EI 1 0 EI 0 ds2
dx a dx 5 a EI 0 dx3 ( )
x  s sinh½y1 ðx  sÞ sin½y2 ðx  sÞ
1 dq  2 2  3 2   3 2  ds. ð79Þ
 ð73Þ y1 y2 y1 y1 þ y22 y2 y1 þ y22
EI 0 dx
The roots are given by
  vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
EI 1 d w~ 4
EI 0 u 8  s ffi9ffi
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
N~ 1  F 1 ¼ 2  4 þ a2 1 u < 2
ar dx EI 1 u1 P P P =
y1 ¼ t a2  þ a2  þ 4a2 ,
F d2 w~ q 2: EI 0 EI 0 EI 1 ;
þ 2 2
þ . ð74Þ
a EI 0 dx EI 0
(80)
The engineering admissible boundary conditions  are vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
8 ffi9
then expressed as follows: For a pinned end w~ ¼ M ~ ¼ u
u   s ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
N~ 1  F 1 ¼ M x0 ¼ 0 at x ¼ x0 Þ, they are given by u 1< 2 P P P =
y2 ¼ t a   a2  þ 4a2 .
2: EI 0 EI 0 EI 1 ;
w~ ¼ 0,
d2 w~ (81)
¼ 0,
dx2 The roots y21 and y22 are real, unequal, and positive and
4
d w~ q negative, respectively. The buckling load, Pcr, is related to
¼ . ð75a; b; cÞ
dx4 EI 0 the eigenvalue y22;cr .
 ARTICLE IN PRESS
250 U.A. Girhammar, D.H. Pan / International Journal of Mechanical Sciences 49 (2007) 239–255

For the limit case of single-component column where

ðEI 0 =EI 1 ! 1Þ, the buckling load becomes Pcr;single ¼ 8
2 Euler case 1; clampedfree;
y2 EI 0 ¼ y22;cr EI 1 ; for a non-composite column >
>
>
 2;cr  < 1 Euler case 2; pinnedpinned;
V~ s ¼ 0; K ! 0; aL ! 0 , it becomes Pcr;0 ¼ y2 EI 0 ; 2;cr;0 msolid ¼
and for a fully composite column ðDu~ ¼ 0; K ! 1; >
> 0:699 Euler case 3; pinnedclamped;
>
:
aL ! 1Þ, it becomes Pcr;1 ¼ y22;cr;1 EI 1 . 0:5 Euler case 4; clampedclamped
(86a2d)
5.3. Buckling load
for the four Euler cases. The exact solutions for the buckling
If q ¼ 0, Eq. (67) reduces to that valid for a pure column. length coefficients for the four Euler cases given below, will
The critical buckling load, valuable in the design of be compared to these approximate or tentative values.
partially composite members, can be obtained from
Eq. (81) as 5.4. Exact buckling length coefficients for the four Euler
cases
y22;cr EI 1
Pcr ¼
1 þ ðEI 1 =EI 0  1Þ=ð1 þ a2 =y22;cr Þ In Ref. [13] only approximate buckling length coeffi-
ðy22;cr =y22;cr;1 ÞPcr;1 EI eff cients were employed. In this paper, the exact buckling
¼ 2 2
¼ Pcr;1 length coefficients for the four Euler cases are evaluated.
1 þ ðEI 1 =EI 0  1Þ=½1 þ ðm=pÞ ðaLÞ  EI 1
In the general case of arbitrary boundary conditions the
ð82Þ eigenvalues y2,cr are obtained from large, often transcen-
dental, equations, which require numerical solutions. The
where an effective bending stiffness for the composite iteration is conveniently carried out in the following way:
beam-column with incomplete interaction has been intro-
duced and which can be expressed as (a generalization of
(a) assume P;
the effective bending stiffness presented by Girhammar and
(b) calculate y1 and y2 from Eqs. (80) and (81) and
Gopu [13])
substitute into Eq. (78);
  (c) satisfy the boundary conditions at each end, e.g. for the
y2;cr 2 EI 1
EI eff ¼ four Euler cases, one group of the boundary conditions
y2;cr;1 1 þ ðEI 1 =EI 0  1Þ=ð1 þ a2 =y22;cr Þ
(75)–(77), and obtain 6 linear equations with e1, y, e6
 2
m1 EI 1 as variables. Non-trivial solution is given if the
¼ ð83Þ determinant of the 6  6 matrix equals zero. By
m 1 þ ðEI 1 =EI 0  1Þ=½1 þ ðm=pÞ2 ðaLÞ2 
Gaussian elimination the matrix is transformed into
and where the characteristic value is expressed as echelon form and the determinant is determined;
(d) if the determinant is not equal to zero, adjust P and
p repeat steps (a)–(c); and
y2;cr ¼ ,
mL (e) for the correct value of P, y2 is given by Eq. (81) and
p mexact by Eq. (84a).
y2;cr;1 ¼ . ð84a; bÞ
m1 L
(1) Euler Case 1—Clamped-free: For a composite column
The buckling length coefficient is denoted by m for of length L with one end clamped and the other free,
composite columns, by mN for fully composite columns. the boundary conditions are given by Eqs. (76) and
The length of the column is denoted by L. The exact (77), respectively. These conditions give the six
eigenvalues y2,cr and hence the buckling length coefficients equations for determining the constants in Eq. (78).
m for composite beam-columns with interlayer slip depend The non-trivial solution of these six equations is given
on the conditions of support. Here they will be evaluated by the determinant of the coefficients e1 y e6 equal to
for the four Euler cases. zero. According to Appendix B, the exact fundamental
Before knowing the exact solution for the buckling buckling length coefficient is obtained as
loads for composite beam-columns with interlayer slip,
mexact ¼ 2. (87)
Girhammar and Gopu [13] proposed that approximate or
tentative solutions could be obtained by using the This exact solution for a partially composite column
characteristic values of y2,cr given for solid columns (i.e. with boundary conditions according to Euler case 1 is
for columns with full composite action or non-composite the same as the solution (86a) for a solid column
action), i.e. or fully composite column. Thus, the fundamental
buckling length coefficient, mexact ¼ 2, is independent of
p p the composite action and relative bending stiffness
y2;cr;approx  y2;cr;1 ¼ ¼ , (85)
m1 L msolid L parameters.
 ARTICLE IN PRESS
U.A. Girhammar, D.H. Pan / International Journal of Mechanical Sciences 49 (2007) 239–255 251

(2) Euler Case 2—Pinned–pinned: For a composite column an engineering point of view, negligible compared to
of length L with both ends pinned, the boundary the value for solid or fully composite columns ðmsolid ¼
conditions are given by Eq. (75). According to 0:699Þ (maximum error less than 2% for aL  7).
Appendix B, the exact fundamental buckling length The fundamental buckling load is found from Eq.
coefficient is obtained as (82). Exact values are found by using buckling length
mexact ¼ 1. (88) coefficients according to Fig. 8 and approximate
values by using the corresponding coefficient from
Again, this exact solution for a partially composite Eq. (86c) ðmsolid ¼ 0:699Þ. Obviously, the approximate
column with boundary conditions according to Euler values deviate from the exact ones in a similar fashion
case 2 is the same as the solution (86b) for a solid as shown for the buckling length coefficients in
column or fully composite column. Thus, the funda- Fig. 8. The approximate fundamental buckling load
mental buckling length coefficient, m ¼ 1, is indepen- deviate from the exact ones to a maximum amount
dent of the composite action and relative bending of 2.5% for EI 0 =EI 1 ¼ 0:25 and aL  8. Thus, from
stiffness parameters. an engineering point of view, the approximate values
(3) Euler case 3—Clamped–pinned: For a composite are acceptable in most practical applications. In
column of length L with one end clamped and the addition, these approximate buckling loads are on the
other end pinned, the boundary conditions are given by safe side.
Eqs. (76) and (75), respectively. The determining (4) Euler case 4—Clamped–clamped: For a composite
transcendental equation becomes (if y1 ¼ 0, the deter- column of length L with both ends clamped, the
mining equation becomes identically equal to zero) boundary conditions are given by Eq. (76). According
tanðy2 LÞ ¼ y2 L to Appendix B, the exact fundamental buckling length
coefficient is obtained as
y32 ðy2 LÞ2 þ ðaLÞ2 ½1  EI 0 =EI 1  P=ða2 EI 0 Þ
þ  mexact ¼ 0:5. (90)
y31 ðy1 LÞ2  ðaLÞ2 ½1  EI 0 =EI 1  P=ða2 EI 0 Þ
 ½tanhðy1 LÞ  y1 L ð89Þ Again, this exact solution for a partially composite
column with boundary conditions according to Euler
It is evident from Eq. (89) that the second term on case 4 is the same as solution (86d) for a solid column
the right-hand side is the term that differs from the or fully composite column. Thus, the fundamental
transcendental equation valid for a conventional solid buckling length coefficient, m ¼ 0.5, is independent of
(fully composite) column, [39]. This term only gives the composite action and relative bending stiffness
small deviations in the conventional buckling length parameters.
coefficient value for Euler case 3. The exact m-values for
the fundamental buckling mode vs. the composite
action parameter and the relative bending stiffness 5.5. Application of second-order analysis procedures—
parameter are shown in Fig. 8. magnification factors
As is evident from Fig. 8 , the difference between the
exact values of the buckling length coefficient for The partial composite action theory developed in the
partially composite columns (that depend on the degree preceding sections is applied to composite beam-columns
of partial composite action and the relative bending subjected to an axial compression load P and a uniformly
stiffness between the two constituent materials) is, from distributed load, q0, as shown in Fig. 6. The first- and

0.713 EI 0
Buckling Length Coefficient 

(1) = 0.25
0.711 EI ∞
(1) Max. error 1.8 %
0.709
for L 7. (2)
EI 0
= 0.50
0.707 EI ∞
(2)
0.705 EI 0
(3) = 0.75
0.703 EI ∞
(3)
0.701 EI 0
(4) = 1; μ = 0.699
0.699 EI ∞
1 10 100
Composite Action Parameter  L

Fig. 8. Exact buckling length coefficient, m, vs. the partial composite action parameter, aL, for different values of the relative bending stiffness parameter,
EI0/EIN for Euler case 3. The coefficient m ¼ 0.699 corresponds to the classical buckling length coefficient for ordinary fully composite (solid) beams. This
value is also used as an approximate value for a partially composite beam, i.e. mapprox ¼ 0.699. The deviation between the exact value mexact for the limit
case, EI0/EIN ¼ 0.25, and the approximate value mapprox ¼ 0.699 is less than 1.8% (maximum for aLE7).
 ARTICLE IN PRESS
252 U.A. Girhammar, D.H. Pan / International Journal of Mechanical Sciences 49 (2007) 239–255

second-order analysis methods are applied to determine the effective bending stiffness can be written as
deflection and internal actions and to evaluate the second-
order magnification in the beam-column example with EI 1
EI eff ¼ . (94)
boundary conditions according to Euler case 2. The first- 1 þ ðEI 1 =EI 0  1Þ=½1 þ ðm=pÞ2 ðaLÞ2 
order solution is given by Eq. (63).
The homogeneous and particular solution for the second McCutcheon [42,43] proposed an approximate expres-
order case for the beam-column are obtained by solving sion for the bending stiffness of simply supported partial
Eqs. (78) and (79) to give the following general solution: composite beams subjected to various loading conditions.
Recently, Bulleit et al. [41] applied this approximate value
q0 L2 ðaLÞ2 =ðy1 LÞ2  1 cosh½y1 ðx  L=2Þ for the partial bending stiffness. McCutcheon numerically
w~ ¼
P ðy1 LÞ2 þ ðy2 LÞ2 coshðy1 L=2Þ adapted the various exact expressions to one common
approximate expression, which reads
ðaLÞ2 =ðy2 LÞ2 þ 1 cos½y2 ðx  L=2Þ
þ
ðy1 LÞ2 þ ðy2 LÞ2 cosðy2 L=2Þ

EI 1
,
EI eff ¼ (95)
1 x   2 x ðaLÞ2 1 þ f D EI 1 =EI 0  1
þ   . ð91Þ
2 L L ðy1 LÞ2 ðy2 LÞ2 where
Accordingly, knowing the solution in terms of the
10
displacement according to Eq. (91), the internal actions, fD ¼ . (96)
~ V~ ; N~ 1 ; V~ s , are given by Eqs. (72)–(74), (7), and
M; ðaLÞ2 þ 10
~
M 1; M~ 2 ; V~ 1 ; V~ 2 ; N~ 2 by (6), (8), (48), and the strain and
UsingEq. (94), the corresponding expression is given by
~
slip, u1 ; u~ 2 ; Du, by (68), (18), (7).
f D ¼ p2 ½ðaLÞ2 þ p2 . Thus, the McCutcheon value is a
The results obtained from the second-order analysis
special value for simply supported beams (Euler 2),
applied to the partially composite beam-column example
reduced from the ‘‘general’’ expression for the effective
according to Fig. 6 are presented in Table 1. The first- and
bending stiffness.
second-order results are also compared in the table and the
The effective bending stiffness can be used to obtain
results clearly show that the second-order effects are
approximate values for the deflection and the internal
significant for both the deflection and the internal actions.
actions in partially composite beams in a way illustrated
Note the difference between the magnification of the
below. For a detailed study in this respect, see Girhammar
internal axial forces and the magnifications obtained for
[44]. An approximate value of the deflection (weff) can be
the other internal actions. This is due to the fact that
obtained as
only that portion of the internal axial force that is
induced by bending is magnified by the second-order EI 1
effect. For further details, see Girhammar and Gopu [40]. weff ¼ w1 , (97)
EI eff
For comparison, the approximate magnification factor for
ordinary beam-columns is given by ([39], cf. also [41]) where wN is the exact deflection in a fully composite member.
1 Knowing the expressions for the internal actions for the limit
n¼ . (92) case of fully composite action ðDu ¼ 0 ; K ! 1Þ obtained
1  P=Pcr
from Eqs. (5) to (7), cf. also [13], the internal normal forces,
This approximate magnification factor applied to moments, and slip forces in beams having partial interaction
partially composite beam-columns is also given in Table 1. can approximately be expressed as
 
EI 0 M
5.6. Application of effective bending stiffness—approximate N i;eff ¼ 1 , (98)
deflection and internal actions EI eff r

According to Eq. (83), the effective bending stiffness for Ei I i

M i;eff ¼ M, (99)
partially composite elements can be written as EI eff
 2
m1 EI 1  
EI eff ¼ EI 0 V
m 1 þ ðEI 1 =EI 0  1Þ=½1 þ ðm=pÞ2 ðaLÞ2  V s;eff ¼ 1 , (100)
(93) EI eff r

The previous results in Section 5.4 justify that buckling where the bending stiffness of the fully composite section,
length coefficients applicable to ordinary solid beam- EIN, is replaced by the effective bending stiffness, EIeff.
columns are also used for partially composite beam- In Table 2, a comparison between exact and approx-
columns, possibly with the exception of composite elements imate values is presented by using the exact expression (63)
with the parameters, EI 0 =EI 1 ¼ 0:25, and, aL  7 in case and the effective bending stiffness according to Eq. (93).
of boundary conditions according to Euler case 3. Thus the For further details, see [44].
 ARTICLE IN PRESS
U.A. Girhammar, D.H. Pan / International Journal of Mechanical Sciences 49 (2007) 239–255 253

Table 2 Also, the maximum theoretical range for the relative

Comparison of exact and approximate results for partially composite bending stiffness for partially composite beams and
beams according to Fig. 6 (without axial load, P ¼ 0)
beam-columns is derived (the relative bending stiffness is
Displacement/ Exact analysis Approximate analysis Error (%) the ratio between the bending stiffness for a corresponding
internal action (using EIeff) non-composite and fully composite cross-section). An
effective bending stiffness, valuable in the determination
wmax 7.560 mm 7.590 mm +0.4
of the critical buckling load for partially composite
M1,max 165.9 Nm 170.7 Nm +2.9
M2,max 497.7 Nm 512.1 Nm +2.9 members, is derived. This effective bending stiffness is also
N1,max 13.363 kN 13.169 kN 1.5 suitable for analysing approximate deflections and internal
N2,max +13.363 kN +13.169 kN 1.5 actions or stresses in composite beams with flexible shear
Vs,max 11.444 kN/m 13.169 kN/m +15.1a connection.
a
The effective bending stiffness parameter will not render as accurate The beam-column analysis is applied to a specific case.
values for slip forces as for deflection, moments and normal forces, since The difference in the approaches to the first- and second-
the slip forces are obtained by differentiation of the normal forces. For order analysis is illustrated by applying the solutions to the
further details, see [44]. given beam-column problem. The results clearly illustrate
the magnification in the actions and displacements due to
the second-order effect. It is demonstrated that the
6. Conclusions application of these magnifiers to the composite beam-
columns is not as straightforward as for the conventional
The ordinary differential equations and general solutions case; consideration must be given to the difference between
for the deflection and internal actions and, especially, the the direct axial load effect and the magnifying moment
pertaining consistent boundary conditions, which are not effect of the applied axial load. The magnification of the
always self-evident by inspection, for partially composite internal axial forces is different from magnifications
Euler–Bernoulli beams and beam-columns are presented. obtained for the other internal actions, since only that
Static loading conditions, including transverse and axial portion of an internal axial force that is induced by
loading and first- and second-order analysis are considered. bending is magnified by the second-order effect.
The theoretical procedure is applicable to general loading
and boundary conditions for uniform composite beams Acknowledgements
and beam-columns with interlayer slip. Specifically, the
admissible boundary conditions of basic engineering The authors express sincere appreciation for the financial
interest are discussed. For example, it is shown that for a support from the County Construction Federation in
clamped end either the internal axial force must equal zero Västerbotten, the Foundation for Technology Transfer in
(i.e., corresponding to an axially rolling or sliding clamped Umeå, and the County Administrative Board in Väster-
support) or the slip must equal zero (i.e., corresponding to botten. We also would like to thank the reviewers for their
an axially non-displaceable or fixed clamped end). constructive comments and suggestions.
Further, the exact closed form characteristic equations
and their associated exact buckling length coefficients for
composite columns with interlayer slip are derived for the Appendix A. Minimum value of relative bending stiffness
four Euler boundary conditions. It is shown that these parameter
coefficients are the same as those for ordinary fully
composite (solid) columns, except for the Euler clamped– The relative bending stiffness parameter is given by
pinned case. For the clamped–pinned case, the difference Eq. (16). Then, we have
between the exact buckling length coefficient and the br r2
corresponding value for solid columns is less than 1.8% 2
¼ 2 . (A.1)
a r þ EA0 EI 0 =EAp
depending on the so-called composite action parameter and
relative bending stiffness parameter. Correspondingly, the For rectangular cross-sections we arrive at

br 3 þ 6ðh2 =h1 Þ þ 3ðh22 =h21 Þ

¼            . (A.2)
a2 4 þ 6 h2 =h1 þ 4 h22 =h21 þ E 1 =E 2 b1 =b2 h1 =h2 þ E 2 =E 1 b2 =b1 h32 =h31

maximum deviation between the exact and approximate The maximum value is given by
buckling load is at most 2.5% (with approximate buckling
load is meant the buckling load for partially composite      
q br q br q br
columns obtained by using the buckling length coefficient ¼ ¼ ¼ 0.
for solid columns). These small differences can in most qðE 1 =E 2 Þ a2 qðb1 =b2 Þ a2 qðh1 =h2 Þ a2
practical cases be neglected. (A.3)
 ARTICLE IN PRESS
254 U.A. Girhammar, D.H. Pan / International Journal of Mechanical Sciences 49 (2007) 239–255

The first two conditions give determining equation becomes

2
E 1 b1 h22 ðy1 LÞ2 þ ðy2 LÞ2 coshðy1 LÞ sinðy2 LÞ ¼ 0, (B.5)
¼ . (A.4)
E 2 b2 h21
i.e.
Introducing Eq. (A.4) in Eq. (A.2), we arrive at np
sinðy2 LÞ ¼ 0 or y2;cr ¼ ;
    L
br EI 0 3 þ 6 h2 =h1 þ 3 h22 =h21 3
¼1 ¼    2 2 ¼ . (A.5) n ¼ 1; 2; 3 . . . or
a 2 EI 1 4 þ 8 h2 =h1 þ 4 h2 =h1 4
m ¼ 1 ðn ¼ 1Þ. ðB:6a; b; cÞ
Eq. (A.5) also satisfies Eq. (A.3).

Appendix B. Exact closed form characteristic equations for B.3. Euler Case 4—Clamped–clamped
partial composite columns
For a composite column of length L with both
In this Appendix, the exact closed form characteristics ends clamped, the boundary conditions are given by
equations for composite columns with interlayer slip for Eq. (76). Noting that sinðy2 LÞ ¼ 2 sinðy2 L=2Þ cosðy2 L=2Þ
the various Euler cases are derived. The roots are given by and cosðy2 LÞ ¼ 1  2 sinðy2 L=2Þ sinðy2 L=2Þ, the determin-
the following characteristic equation ing equation becomes
 2      
D  y21 D2 þ y22 ¼ D4 þ D2 y22  y21  y21 y22 , (B.1) 3 2 2 EI 0 P
4ðy2 LÞ ðy2 LÞ þ ðaLÞ 1  
EI 1 a2 EI 0
where D is a constant and  
1
P a2 P ½1  coshðy1 LÞ cos y2 L þ 4ðy1 LÞ3 ðy1 LÞ2  ðaLÞ2
y21  y22 ¼ a2  ; y21 y22 ¼ . (B.2) 2
EI 0 EI 1  
EI 0 P
The roots y21 and y22 are real and unequal since the  1  sinhðy1 LÞ sinð12y2 LÞ
EI 1 a2 EI 0
expression within the second square root sign in Eqs. (80) EI 0 P
and (81), respectively, is positive. In addition, in order to þ 2 ðy1 LÞ3 ½ðy1 LÞ2  ðaLÞ2 ð1   Þ y2 L
EI 1 a2 EI 0
satisfy Eq. (B.2) these roots must be positive and negative,  
respectively. The negative root implies that y22 has an EI 0 P
 ðy2 LÞ3 ðy2 LÞ2 þ ðaLÞ2 1   y1 L
imaginary value, which is related to trigonometric func- EI 1 a2 EI 0
tions and eigenvalues, i.e., related to critical buckling loads
y22;cr (Pcr).  sinhðy1 LÞ cosð12y2 LÞ sinð12y2 LÞ ¼ 0, ðB:7Þ

i.e.
B.1. Euler Case 1—Clamped-free
  2np
sin y2 L=2 ¼ 0 or y2;cr ¼ ;
For a composite column of length L with one end L
clamped and the other free, the boundary conditions are n ¼ 1; 2; 3 . . . or m ¼ 0:5 ðn ¼ 1Þ. ðB:8a; b; cÞ
given by Eqs. (76) and (77), respectively. The exact
characteristic equation is given by
 
EI 0 P References
ðy1 LÞ2 ðy1 LÞ2  ðaLÞ2 1   2
EI 1 a EI 0
  [1] Pan DH, Girhammar UA, Gustafsson A. Exact dynamic analysis of
EI 0 P
ðy2 LÞ2 ðy2 LÞ2 þ ðaLÞ2 1   2 composite beams with partial interaction. International Journal of
EI 1 a EI 0 Solids and Structures [tentatively accepted].
ðy1 LÞ2 ðy2 LÞ2 coshðy1 LÞ cosðy2 LÞ ¼ 0. ðB:3Þ [2] Stüssi F. Zusammengesetze vollwandträger. International Associa-
tion for Bridge and Structural Engineering (IABSE) 1947;8:249–69.
i.e. [3] Granholm H. On composite beams and columns with special regard
to nailed timber structures. Technical Report 88. Sweden: Chalmers
ð2n  1Þp Univeristy of Technology; 1949 [in Swedish].
cosðy2 LÞ ¼ 0 or y2;cr ¼ ;
2L [4] Newmark NM, Siess CD, Viest IM. Tests and analysis of composite
n ¼ 1; 2; 3 . . . or m ¼ 2 ðn ¼ 1Þ. ðB:4a; b; cÞ beams with incomplete interaction. Proceedings of the Society for
Experimental Stress Analysis 1951;9:75–92.
[5] Pleshkov PF. Theoretical studies of composite wood structures,
Moscow, 1952 [in Russian].
B.2. Euler Case 2—Pinned–pinned [6] Goodman JR. Layered wood systems with interlayer slip. Doctoral
dissertation. Berkley: University of California; 1967.
[7] Amana EJ, Booth LG. Theoretical and experimental studies of nailed
For a composite column of length L with both ends and glued stressed-skin components. Journal of Wood Science
pinned, the boundary conditions are given by Eq. (75). The 1967;4:43–69.
 ARTICLE IN PRESS
U.A. Girhammar, D.H. Pan / International Journal of Mechanical Sciences 49 (2007) 239–255 255

[8] Henghold WM. Layered beam vibrations including slip. Doctoral tional Journal for Numerical Methods in Engineering 2006;65:
dissertation. Fort Collins: Civil Engineering Department, Colorado 1197–220.
State University; 1972. [25] Dall’Asta A. Composite beams with weak shear connection.
[9] Girhammar UA. Rabo wall element—Theoretical and experimental International Journal of Solids and Structures 2001;38:5605–24.
study of composite element of timber and concrete. Research Report [26] Kerr AD. On the derivation of well posed boundary value problems
TULEA 1980:30. Sweden: Division of Structural Engineering, Luleå in structural mechanics. International Journal of Solids and
University of Technology; 1980. Structures 1976;12:1–11.
[10] Girhammar UA. Dynamic analysis of composite structures of [27] Askes H, Metrikine AV. Higher-order continua derived from discrete
concrete, steel and wood. Report A5:85, Sweden: Royal Swedish media: continualisation aspects and boundary conditions. Interna-
Fortifications Administration; 1985 [in Swedish]. tional Journal of Solids and Structures 2005;42:187–202.
[11] Girhammar UA. Design of composite beam-columns with incomplete [28] Landis CM. Energetically consistent boundary conditions for
interaction. Report A6:87. Sweden: Royal Swedish Fortifications electromechanical fracture. International Journal of Solids and
Administration; 1987. Structures 2004;41:6291–315.
[12] Girhammar UA, Gopu VKA. Analysis of P-D effect in composite [29] Diago JC. Second order effects in composite beam-columns, Masters
concrete/timber beam-columns. Proceedings of the Institution of thesis. Baton Rouge: Department of Civil Engineering, Louisiana
Civil Engineers, Part 2: Research and Theory 1991;91:39–54. State University; 1987.
[13] Girhammar UA, Gopu VKA. Composite beam-columns with [30] Washizu K. Variational methods in elasticity and plasticity. Oxford:
interlayer slip—exact analysis. Journal of Structural Engineering, Pergamon Press; 1968.
ASCE 1993;119:1265–82. [31] Oden JT, Reddy JN. Variational methods in theoretical mechanics.
[14] Girhammar UA, Pan D. Dynamic analysis of composite members Berlin: Springer; 1976.
with interlayer slip. International Journal of Solids and Structures [32] Plantema FJ. Sandwich construction. New York: Wiley; 1966.
1993;30:797–823. [33] Hashemi SH, Arsanjami M. Exact characteristic equations for some
[15] Pan DH. Analysis of built-up structural systems—specifically classical boundary conditions of vibrating moderately thick rectan-
reticulated and composite structures of timber. Doctoral dissertation. gular plates. International Journal of Solids and Structures 2005;
Umeå, Sweden: Civil Engineering, Department of Applied Physics 42:819–53.
and Electronics, Umeå University; 2002. [34] Papargyri-Beskou S, Tsepoura KG, Polyzos D, Beskos DE. Bending
[16] Ranzi G, Bradford M, Uy B. A general method of analysis of and stability analysis of gradient elastic beams. International Journal
composite beams with partial interaction. Steel and Composite of Solids and Structures 2003;40:385–400.
Structures 2003;3:169–84. [35] Iwinski T. Theory of beams. London: Pergamon Press Ltd; 1958.
[17] Faella C, Martinelli E, Nigro E. Steel and concrete composite beams [36] Pan HH. Some applications of symbolic functions on beam problems.
with flexible shear connection: exact analytical expression of the Journal of Franklin Institute 1963;275:303–13.
stiffness matrix and applications. Computers and Structures 2002; [37] Biondi B, Caddemi S. Closed form solutions of Euler–Bernoulli
80:1001–9. beams with singularities. International Journal of Solids and
[18] Ranzi G, Bradford M. Analytical solutions for the time-dependent Structures 2005;42:3027–44.
behaviour of composite beams with partial interaction. International [38] Fung YC. Foundations of solid mechanics. Englewood Cliffs, NJ:
Journal of Solids and Structures 2006;43:3770–93. Prentice-Hall Inc.; 1965.
[19] Jurkiewiez B, Buzon S, Sieffert J. Incremental viscoelastic analysis of [39] Timoshenko SP, Gere JM. Theory of elastic stability, International
composite beams with partial interaction. Computers and Structures Student Edition. 2nd ed. Tokyo: McGraw-Hill Kogakusha; 1961.
2005;83:1780–91. [40] Girhammar UA, Gopu VKA. Composite beam-columns with
[20] Ayoub A. A two-fold mixed variational principle for partially interlayer slip—approximate analysis. Journal of Structural Engi-
connected composite beams. Finite Elements in Analysis and Design neering, ASCE [to be published].
2001;37:929–59. [41] Bulleit WM, Pang W-C, Rosowski DV. Modeling wood walls
[21] Ayoub A. A force-based model for composite steel-concrete beams subjected to combined transverse and axial loads. Journal of
with partial interaction. Journal of Constructional Steel Research Structural Engineering, ASCE 2004;131:781–93.
2005;61:387–414. [42] McCutcheon WJ. Method for predicting the stiffness of wood-joist
[22] Ranzi G, Bradford MA, Uy B. A direct stiffness analysis of a floor systems with partial composite action. Research Paper FPL 289,
composite beam with partial interaction. International Journal for USDA, Madison: Forest Service, Forest Products Laboratory; 1977.
Numerical Methods in Engineering 2004;61:657–72. [43] McCutcheon WJ. Stiffness of framing members with partial
[23] Cas B, Saje M, Planinc I. Non-linear finite element analysis of composite action. Journal of Structural Engineering, ASCE 1986;
composite planar frames with an interlayer slip. Computers and 112:1623–37.
Structures 2004;82:1901–12. [44] Girhammar UA. Simplified analysis methods for composite members
[24] Gara F, Ranzi G, Leoni G. Displacement-based formulations for with interlayer slip. Journal of Structural Engineering, ASCE,
composite beams with longitudinal slip and vertical uplift. Interna- submitted for publication.