Sie sind auf Seite 1von 13

Large deections of tapered functionally graded beams

subjected to end forces


Nguyen Dinh Kien
a,
, Buntara Sthenly Gan
b
a
Department of Solid Mechanics, Institute of Mechanics, Vietnam Academy of Science and Technology, 18 Hoang Quoc Viet, Hanoi, Viet Nam
b
Department of Architecture, College of Engineering, Nihon University, Koriyama, Fukushima-ken 963-8642, Japan
a r t i c l e i n f o
Article history:
Received 5 June 2012
Received in revised form24 September 2013
Accepted 22 November 2013
Available online 15 December 2013
Keywords:
Functionally graded material
Tapered beam
Finite element method
Large deection
a b s t r a c t
The large deections of tapered functionally graded beams subjected to end forces are
studied by using the nite element method. The material properties of the beams are
assumed to vary through the thickness direction according to a power law distribution.
A rst order shear deformable beam element employed the exact polynomials to
interpolate the transverse displacement and rotation, is formulated in the context of the
co-rotational approach. The large deection response of the beams is computed by using
the arc-length control algorithm in combination with the NewtonRaphson iterative
method. The numerical results show that the formulated element is capable to assess
accurately the response of the beams by using just several elements. A parametric study
is given to examine the inuence of the material non-homogeneity, taper ratio as well as
the aspect ratio on the large deection behaviour of the beams.
2013 Elsevier Inc. All rights reserved.
1. Introduction
Functionally graded materials (FGMs), which were invented by Japanese scientists in Sendai in 1984 [1], have received
much attention from researchers. The FGMs are formed by varying percentage of constituents in any desired spacial direc-
tion, and as a result the specic physical and mechanical properties of the formed material can be obtained. FGMs are being
used widely as a structural material, and analysis of FGM structures has become an important topic in the eld of structural
mechanics. A comprehensive list of publications on the analysis of FGM structures subjected to different loadings is given in
a review paper by Birman and Byrd [2], contributions that are most relevant to the present work are discussed below.
Sankar [3] proposed an elasticity solution for FGM beams under static transverse loads by assuming the material prop-
erties to vary in the thickness direction by an exponential law.Based on the rst order shear deformation beam theory, Cha-
kraborty et al. [4] formulated a beam element for analyzing the thermoelastic behaviour of FGM beams by using the exact
solution of the governing differential equations of an FGM Timoshenko beam segment to interpolate the displacements and
rotation. Using the spectral nite element method, Chakraborty and Gopalakrishnan [5] studied the wave propagation
behaviour of FGM beams under high frequency impulse loading. Kadoli et al. [6] formulated a beam element to investigate
the static behaviour of metal ceramic beams under ambient temperature by adopting the third order shear deformation
beam theory. Taking the warping effect into consideration, Benatta et al. [7] derived an analytical solution to the bending
problem of a FGM beam. Singh and Li [8] presented a mathematical model for computing the buckling loads of uniform
and non-uniform axially FGM columns. Kang and Li [9,10] proposed closed-form solutions for a nonlinear FGM cantilever
beam with the elastic modulus variation in thickness direction under a tip load or a tip moment by deriving an expression
0307-904X/$ - see front matter 2013 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.apm.2013.11.032

Corresponding author. Tel.: +84 4 3762 8006; fax: +84 4 3762 2039.
E-mail addresses: ndkien@imech.ac.vn, ndkien@yahoo.com (D.K. Nguyen).
Applied Mathematical Modelling 38 (2014) 30543066
Contents lists available at ScienceDirect
Applied Mathematical Modelling
j our nal homepage: www. el sevi er . com/ l ocat e/ apm
for the effective bending rigidity. Lee et al. [11] presented a nite element procedure for computing the post-buckling
response of FGM plates under compressive and thermal loads. Huang and Li [12] studied the free vibration of non-uniform
cross-section beams made of an axially FGM. Employing the multilayering method, Kuti et al. [13] presented a nite
element procedure for modelling a FGM beam with spatial variation of material properties. Also using the nite element
method, Alshorbagy et al. [14] investigated the free vibration characteristics of EulerBernoulli beams with material
graduation in both axially and transversally through the beam thickness. Adopting the Ritz method, Wattanasakulpong
et al. [15] investigated the thermal buckling and elastic vibration of the third-order shear deformable FGM beams. Shahba
et al. [16] computed the natural frequencies and buckling loads of tapered Timoshenko beams composed of axially FGM by
using the exact shape functions of a uniformed homogeneous Timoshenko beam segment to derive the mass and stiffness
matrices. Aminbaghai et al. [17] studied the free vibration of FGM beams with continuous spatial polynomial variation of
material properties by a fourth order differential equation of the second order beam theory. Using the nite element method,
Nguyen et al. [18] studied the dynamic behaviour of non-uniform FGM Timoshenko beams subjected to a variable speed
moving load.
Non-prismatic beams with variable cross section are of great important in engineering because of their ability in optimiz-
ing the weight and strength of structures. Analytical methods are often encountered difculties in analyzing the non-pris-
matic beams due to the presence of variable coefcients in the governing differential equations, and a numerical methods is
necessarily employed instead of. In this line of work, Wood and Zienkiewicz [19] computed the large displacement response
of a non-uniform column subjected to an eccentric axial compressive force by using the nite element method. In [20], Cleg-
horn and Tabarrok employed the homogeneous solution of a tapered Timoshenko beam segment to formulate the mass and
stiffness matrices for computing the vibration characteristics of the beams. Baker [21] used the weight residual method in
solving the governing differential equation of a slender tapered cantilever beam under arbitrarily distributed loads. Lee et al.
[22] presented a RungeKutta based numerical method to solve the governing differential equations of tapered cantilever
beams under large displacements, and then veried the computed results by performing an experiment on a width tapered
steel beam. Brojan et al. [23] developed an exact momentcurvature formula for determining deformed shape of non-pris-
matic cantilever beams obeying the generalized Ludwick law under a tip moment. Attarnejad et al. [24] derived the displace-
ment functions for studying the free vibration of non-prismatic beams by solving the governing equations of motion of a
tapered Timoshenko beam element. Shahba et al. [25] introduced the basic displacement functions and then constructed
the shape functions for derivation of an efcient 2D beam element for the free vibration analysis of rotating tapered Timo-
shenko beams.
To the authors best knowledge, the large deection of tapered FGMbeams subjected to end forces has not been studied in
the literature and this topic will be the subject of investigation by using the nite element method in the present work. The
material properties of the beams are assumed to be described by a power law distribution through the beam thickness. A
nonlinear beam element based on the rst order shear deformation beam theory, employing the polynomials obtained from
the solution of the governing differential equations of a uniform homogeneous Timoshenko beam segment to interpolate the
transverse displacement and rotation, is formulated in the context of the co-rotational approach. Using the formulated ele-
ment, the large deection response of a cantilever FGM beam with different taper cases is computed with the aid of the arc-
length control algorithm in combination with the iterative NewtonRaphson method. Numerical examples are presented to
show the accuracy and efciency of the proposed element. The inuence of the material distribution, taper ratio as well as
the aspect ratio on the large deections of the beams is numerically investigated in detail.
2. Tapered FGM beams
Consider a tapered FGM beam with length of L. Three following taper cases are considered in the present work
Case A: A A
0
1 a
x
L
_ _
, I I
0
1 a
x
L
_ _
Case B: A A
0
1 a
x
L
_ _
, I I
0
1 a
x
L
_ _
3
Case C: A A
0
1 a
x
L
_ _
2
, I I
0
1 a
x
L
_ _
4
where A
0
; I
0
denote the area and moment of inertia of the section at x 0, respectively; 0 6 a < 1 is the taper ratio. Fig. 1
shows the geometry of the three considered cases of the beam in a co-ordinate system, x; y; z, where the z axis directs along
the thickness direction of the beam.
The beam is assumed to be formed from two different materials with volume fractions to be varied according to a power
law as
V
1

z
h

1
2
_ _
k
; V
2
1 V
1
; 1
where the subscript 1 and 2 indicate the material 1 and material 2 the constituents of the FGM, respectively; k is the non-
negative power law exponent, dictating the material variation through the beam thickness; h is the section height, which
longitudinally varies for the case B and case C. According to the rule of mixture, the effective Youngs modulus E, and the
effective shear modulus G of the FGM beam are given by
D.K. Nguyen, B.S. Gan/ Applied Mathematical Modelling 38 (2014) 30543066 3055
Ez E
1
E
2

z
h

1
2
_ _
k
E
2
;
Gz G
1
G
2

z
h

1
2
_ _
k
G
2
;
2
where E
1
; E
2
and G
1
; G
2
denote the Youngs and shear moduli of the constituents. It is evident from Eq. (2) that when
z h=2; E E
2
; G G
2
, and thus the bottom surface contains purely material 2; when z h=2; E E
1
; G G
1
, and the
top surface contains only material 1. Fig. 2 shows the variation of the effective Youngs modulus and the shear modulus
through the thickness of a FGM beam formed from steel and alumina. The Youngs and shear moduli of the alumina and steel
are as follows [4]: E
a
390 GPa, E
s
210 GPa, G
a
137 GPa, G
s
80 GPa, where the subscript a and s stand for alumina
and steel, respectively.
3. Finite element formulation
3.1. Large deection analysis
A convenient way to formulate a geometrically nonlinear element is to introduce a local coordinate system which con-
tinuously moves and rotates during the element deformation process. The element formulation is rstly derived in the local
system then transferred into the global one with the aid of transformation matrices dened from the relationship between
the local and global nodal displacements. The method is known in the literature as the co-rotational method [26,27], and will
be adopted herein.
Fig. 1. FGM beam with three considered taper cases.
200 250 300 350 400
0.5
0.25
0
0.25
0.5
E (GPa)
z
/
h
k = 1
k = 2
k = 3
k = 5
k = 0.2
k = 0.5
80 90 100 110 120 130 140
0.5
0.25
0
0.25
0.5
G (GPa)
z
/
h
k = 0.5
k = 0.2
k = 1
k = 2
k = 3
k = 5
Fig. 2. Variation of Youngs and shear moduli through the thickness of FGM beam formed from steel and alumina.
3056 D.K. Nguyen, B.S. Gan/ Applied Mathematical Modelling 38 (2014) 30543066
Fig. 3 shows a two-node beam element and its kinematics, where x;

z and x; z denotes the local and global coordinate


systems, respectively. The local system is chosen as that its original is always at node 1, and the x axis directs to node 2. By
choosing such a local system, the local axial displacement at node 1 and the local transverse displacements at both the two
nodes are always zeros. Thus, the vector of nodal displacements in the local system contains only three components as

d f

u

h
1

h
2
g
T
; 3
where and hereafter the superscript T denotes a transpose of a vector or a matrix; a bar sufx is used to distinguish a local
quantity from its corresponding global one. The global nodal displacements are general non-zeros, and the vector of the
nodal displacements in the global system, in general contains six components
d fu
1
w
1
h
1
u
2
w
2
h
2
g
T
: 4
The local displacement and rotations in Eq. (3) can be computed as [27]

u l
n
l;

h
1
h
1
h
r
;

h
2
h
2
h
r
: 5
where l; l
n
and h
r
, respectively are the initial length, current length and rigid rotation of the element, which can be obtained
from geometric consideration of Fig. 3 as
l x
2
x
1
; l
n

x
2
u
2
x
1
u
1

2
w
2
w
1

2
_
;
h
r
arctan
w
2
w
1
x
2
u
2
x
1
u
1
_ _
:
6
In Eq. (6), x
1
and x
2
are the abscissas of the node 1 and node 2 with respect to the global system, respectively.
The internal virtual work for the element can be written in both the local and global systems as
dU dd
T
f
in
d

d
T
f
in
dd
T
T
T
1

f
in
; 7
where

f
in
and f
in
are the local and global element nodal force vectors associated with the nodal displacements given by Eqs.
(3) and (4), respectively; T
1
@

d=@d is the transformation matrix, which can be computed from Eqs. (5) and (6). Eq. (7) must
apply for any arbitrary virtual displacement dd, hence the global element internal nodal force vector f
in
is given by
f
in
T
T
1

f
in
: 8
The element tangent stiffness matrix, k
t
is obtained by differentiating the global element nodal force vector with respect to
the global nodal displacements as
k
t

@f
in
@d
T
T
1

k
t
T
1

f
u
T
2

f
h1

f
h2
T
3
; 9
where

f
u
;

f
h1
;

f
h2
are the nodal forces corresponding to the nodal displacements u;

h
1
;

h
2
, respectively;

k
t
@

f
in
=@

d is the
local tangent stiffness matrix; T
2
; T
3
are the transformation matrices dened as
T
2

@
2

u
@d
2
; T
3

@
2

h
1
@d
2

@
2

h
2
@d
2

@
2
h
r
@d
2
: 10
Eqs. (8)(10) completely dene the global element nodal force vector f
in
and tangent stiffness matrix k
t
when the local
element nodal force vector

f
in
and tangent stiffness matrix

k
t
are known.
Fig. 3. Beam element and its kinematics in local and global coordinate systems.
D.K. Nguyen, B.S. Gan/ Applied Mathematical Modelling 38 (2014) 30543066 3057
3.2. Local formulation
Adopting the rst order shear deformation beam theory, the axial and transverse displacements at any point of the beam
can be expressed in the local system as
u

x;

z u
0

zh

x; w

x;

z w
0

x; 11
where u
0
and w
0
, respectively are the axial and transverse displacements of the point on the mid-plane, chosen as the ref-
erence plane; hx is the cross-sectional rotation. The axial strain, , and the shear strain, c, for the large deection analysis
are deduced from Eq. (11) as
u
0;x

1
2
w
2
0;x
_ _

zh
;x

m


zv;
c w
0;x
h;
12
where a coma denotes the derivative with respect to the followed variable;
m
is the membrane strain, and v h
;x
is the
beam curvature. It is noted that in the co-rotational method, a linear denition can be used for the local axial strain [26,27],
but as the rst author has shown in [28] that by adding the nonlinear term
1
2
w
2
0;x
, the convergency of the element can be
improved considerably.
Assuming the material of the FGM beam obeys the Hooks law, the axial stress, r, and shear stress, s, associated with the
axial and shear strains in Eq. (12) are given by
r E

z; s wG

zc; 13
where E

z; G

z are the effective moduli, dened by Eq. (2) with z is replaced by



z; w denotes the shear correction factor,
equals to 5/6 for a rectangular section. The virtual work of the internal force for the element can be expressed as
dU
_
l
0
_
A
rd sdc dAd

x; 14
where d and dc are the small arbitrary virtual strains; A denotes the cross section area, which longitudinally varies herein. In
order to express the virtual work in term of the local nodal displacements, interpolation schemes are necessary to introduce
for the kinematic variables. Noting that u
1
w
1
w
2
0, one can write
u
0
N
u

u; w
0
N
T
w

h; h N
T
h

h; 15
where

h f

h
1

h
2
g
T
; N
w
and N
h
are the matrices of interpolating functions for w
0
and h, respectively. Linear and cubic Hermite
polynomials derived purely from mathematical consideration can be employed to interpolate the displacement eld of a
FGM beam element [14]. The use of exact shape functions for uniform homogeneous beam elements to interpolate the trans-
verse displacement and rotation of the tapered FGM Timoshenko beam leads to a beam element with high accuracy [16]. In
the present work, linear function is adopted for the axial displacement, N
u
x=l, and the exact polynomials derived by
Kosmatka [29] as solutions of the governing differential equations for a uniform homogeneous Timoshenko beam element
are used for the transverse displacement and rotation as
N
w

l
1 /
x
l
_ _
3
2
/
2
_ _
x
l
_ _
2
1
/
2
_ _
x
l
_ _
x
l
_ _
3
1
/
2
_ _
x
l
_ _
2

/
2
x
l
_ _
_ _
16
and
N
h

1
1 /
3
x
l
_ _
2
4 /
x
l
_ _
1 /
3
x
l
_ _
2
2 /
x
l
_ _
_ _
: 17
In Eqs. (16) and (17), / is the ratio of the bending stiffness to the shear stiffness as [29]
/
12
l
2
E
h
I
h
wG
h
A
h
_ _
18
where the subscript h designates the value of the parameters in the homogeneous and isotropic uniform beam. It should be
noted that if the notation / E
h
I
h
=l
2
wG
h
A
h
is used instead of (18), the interpolation functions in Eqs. (16) and (17) resume
the same forms as that formulated by Reddy in [30], and used by Shahba et al. in [16]. Present work uses the area and
moment of inertia of the cross section at the left node of the element as A
h
and I
h
, and the Youngs and shear moduli of
the material at the bottom surface of the beam as E
h
and G
h
, respectively.
Due to the unbalance interpolations adopted for the axial and transverse displacements, the formulated element might
suffer from the membrane locking [26]. To overcome this problem, it is necessary to replace the membrane strain in Eq. (12)
by the effective strain, dened as [26]
3058 D.K. Nguyen, B.S. Gan/ Applied Mathematical Modelling 38 (2014) 30543066

eff:

1
l
_
l
0
u
0;x

1
2
w
2
0;x
_ _
d

x: 19
Using Eq. (16), we can write the effective strain (19) in the form

eff:
b
u

u
1
2l

h
T
_
l
0
b
w
b
T
w
d

x

h; 20
where b
u
N
u;x
1=l, and
b
w
N
w;x

1
1 /
3
x
l
_ _
2
4 /
x
l
_ _
1
/
2
_ _
3
x
l
_ _
2
2 /
x
l
_ _

/
2
_ _
: 21
With the introduction of the effective strain, the virtual work given by Eq. (14) reads
dU
_
l
0
Nd
eff:
Mdv Qdcd

x; 22
where, with the aid of Eqs. (15)(17), the resultants N; M and Q are given by
N A
11

eff:
A
12
v A
11

eff:
A
12
b
T
h

h;
M A
12

eff:
A
22
v A
12

eff:
A
22
b
T
h

h;
Q wA
33
c wA
33
b
T
w
N
T
h

h:
23
In Eq. (23), b
h
N
h;x
; A
11
; A
12
; A
22
and A
33
denote the extensional, coupling, bending and shear rigidities, respectively and
dened as follows
A
11
; A
12
; A
22

_
A
E

z 1;

z;

z
2
dA; A
33

_
A
G

zdA: 24
The virtual strains and virtual curvature in Eq. (22) can be computed from Eqs. (15) and (20) as
d
eff:
b
u
d

u c
T
w
d

h; dv b
T
h
d

h; dc b
T
w
N
T
h
d

h; 25
where
c
w

eff:;

h

1
601 /
2

5/2 /

h
1


h
2
24

h
1


h
2

5/2 /

h
1


h
2
2

h
1
4

h
2

_ _
: 26
Substituting Eq. (25) into Eq. (22), one gets
dU
_
l
0
Nb
u
d

u Nc
T
w
Mb
T
h
Qb
T
w
N
T
h

_ _
d

h
_ _
d

x 27
So that the components of the local element nodal force vector are given by

f
u

_
l
0
Nb
u
d

x;

f
h


f
h
1

f
h
2
_ _
T

_
l
0
Nc
w
Mb
h
Qb
w
N
h
d

x:
28
The local tangent stiffness matrix can be divided into sub-matrices as

k
t

k
uu

k
uh

k
T
uh

k
hh
_ _
; 29
where, the sub-matrices can be computed from Eq. (28) with the aid of Eqs. (23)(26) as follows

k
uu


f
u; u

_
l
0
A
11
b
2
u
dx
1
l
2
_
l
0
A
11
d

x;

k
uh


f
u;

h

_
l
0
b
u
A
11
c
T
w
A
12
b
T
h
d

x;

k
hh


f
h;

h

_
l
0
A
11
c
w
c
T
w
ND 2A
12
b
h
c
T
w
A
22
b
h
b
T
h
wA
33
b
w
N
h
b
T
w
N
T
h

_ _
d

x;
30
D.K. Nguyen, B.S. Gan/ Applied Mathematical Modelling 38 (2014) 30543066 3059
where D is a symmetric matrix with the following form
D c
T
w;

h

1
601 /
2

8 10/ 5/
2
2 10/ 5/
2

2 10/ 5/
2
8 10/ 5/
2
_ _
: 31
The terms containing c
w
and N in Eq. (30) stemming from the nonlinear part of the local strain,
1
2
w
2
0;x
, and they are functions
of the current local nodal displacements. Eqs. (28) and (30) together with Eqs. (8)(10) completely dene the element
formulation. In order to improve convergency of the numerical result, the rigidities A
11
; A
12
; A
22
and A
33
in the local element
nodal force vector and tangent stiffness matrix, and previously dened by Eq. (24), are computed by using the exact variation
of the section prole herein.
In case of homogeneous beams, E
1
E
2
E
h
and G
1
G
2
G
h
, where E
h
and G
h
are the Youngs and shear moduli of the
homogeneous beam, respectively, and thus Eq. (2) gives constant effective moduli. As a result, the coupling rigidity A
12
dened in Eq. (24) is vanished, and the formulated element can be employed to analyze the tapered homogeneous beams
just by assigning E
1
E
2
E
h
and G
1
G
2
G
h
.
3.3. Discrete system equations
The derived formulation is assembled into structural nodal force vector and tangent stiffness matrix to construct the equi-
librium equations, which can be written in the form [26]
gp; k q
in
p kf
ef
0; 32
where the residual force vector g is a function of the current structural nodal displacements p, and the load-level parameter
k; q
in
is the structural nodal force vector, assembled from the formulated element force vector f
in
; f
ex
is the xed external
loading vector. For the beam under the end forces considered in the present work, the loading vector contains zero compo-
nents except for the ones corresponding to the end forces. The boundary conditions for a cantilever beam used in the numer-
ical examples below constrain the displacements and rotation at the left end to zeros.
The system of nonlinear equations (32) can be solved by an incremental/iterative method. In the method, a new iterative
displacement vector dp can be obtained from a truncated Taylor expansion of gp; k around an equilibrium point p
0
; k
0
f
ef

as
gp; k g
0
K
t0
dp 0; so that dp K
1
t0
g
0
; 33
where K
t
is the structural tangent stiffness matrix assembled from the element tangent stiffness matrix k
t
; K
t0
and g
0
are the
values of the matrix K
t
and vector g evaluated at the equilibrium point, p
0
; k
0
f
ef
. The arc-length control algorithm in com-
bination with the NewtonRaphson method described by Criseld in [26] is adopted in the present work.
4. Numerical result and discussion
The large deection of a cantilever FGM beam is investigated by using the formulated element and described numerical
algorithm in this Section. Except for Sub-section 4.6, the beam is assumed to be formed from the steel and alumina with the
Youngs and shear moduli given in Section 2. Otherwise stated, the beam is assumed having an aspect ratio L=h
0
50, where
h
0
is the height of the clamped section. A shear correction factor w 5=6 is used in all analyses. To facilitate the presentation
of numerical results, the following dimensionless parameters are introduced
u
u
L
L
; w
w
L
L
; h
h
L
p=2
; P
P
L
L
2
E
s
I
0
; M
M
L
L
E
s
I
0
; 34
where u
L
; w
L
; h
L
are the displacements and rotation at the free end; P
L
and M
L
are the external load and moment, respectively.
4.1. Formulation validation
The large deections of a homogeneous cantilever beam with linearly varying bending stiffness (case A) subjected to an
end transverse or/and an end moment are investigated. The numerical result of the problem obtained by Lee et al. [22] for
Table 1
Absolute values of tip response for case A homogenous beam under a transverse load or/and a moment (I
0
=I
L
3).
P

Present Ref. [22]


u w h u w h
5 0 0.1671 0.4919 0.5407 0.1678 0.4926 0.5428
0 2 0.1414 0.4135 0.6993 0.1411 0.4136 0.6994
5 2 0.3578 0.6402 0.9962 0.3640 0.6433 0.9969
3060 D.K. Nguyen, B.S. Gan/ Applied Mathematical Modelling 38 (2014) 30543066
the case I
0
=I
L
3 is used to verify the accuracy of the formulated element. Table 1 lists the absolute values of the free end
displacements and rotation of the beam computed by four elements formulated in the present work. In the table,
P

P
L
L
2
=E
s
I
L
and M

M
L
L=E
s
I
L
are introduced for consistent with the work in [22]. The table shows a good agreement
between the numerical result obtained in the present work with that of Ref. [22].
The next example is a uniform FGM beam subjected to a tip moment M
L
. The relation between the displacements and the
applied moment of the beam computed by one and four elements is shown in Fig. 4, where the analytical solutions derived
by Kang and Li [10] are depicted by small circles. The expressions for the axial displacement u
L
and the transverse displace-
ment w
L
at the free end derived by Kang and Li in [10] are as follow
u
L

D
M
L
sin
M
L
D
L
_ _
L; w
L

D
M
L
1 cos
M
L
D
L
_ _ _ _
; 35
where D is the effective bending rigidity of the FGM beam, which can be computed when the geometry and material data as
well as the power law index k are given (cf. Eq. (14) of Ref. [10]). With a section area A 0:006 and a moment of inertia
I 1:8 10
6
used herewith, the effective bending rigidity D 6:3294 10
5
; 5:2380 10
5
; 4:3701 10
5
is obtained for
k 0:2; 1; 10, respectively. Excellent agreement between the numerical result obtained by four elements in the present work
with that of Ref. [10] is seen from the gure. The obtained numerical results show the good accuracy and convergency of the
formulated element. Since four elements can give good accuracy, otherwise stated a mesh of four elements is used in the
computation reported below. It has to be noted that the computation in this sub-section has been performed with / dened
by Eq. (18) using section property at the right node of the element and Youngs and shear moduli of the material on the top
surface, but the same numerical result is obtained.
1.2 1 0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 1
0
1
2
3
4
Free end displacements, u & w
T
i
p

m
o
m
e
n
t
,

M
Kang & Li [10]
present 1 element
present 4 elements
u
w
k:
10
1
0.2
k:
0.2
1
10
Fig. 4. Relation between tip displacements and tip moment of uniform FGM beam.
0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 1
0
2
4
6
8
10
Free end displacements, u & w
T
r
a
n
s
v
e
r
s
e

l
o
a
d
,

P
pure steel pure steel
pure alumina
k=0.5
k=1
k=5
P
case B
u
w
u
w
Fig. 5. Loaddisplacement curves for case B beam under a transverse load (a 0:5).
D.K. Nguyen, B.S. Gan/ Applied Mathematical Modelling 38 (2014) 30543066 3061
4.2. Effect of material non-homogeneity
The material non-homogeneity of the beams is governed by the power law index k in Eq. (2). The large deection re-
sponse of the case B beam subjected to a transverse load or an eccentric axial force is computed with different values of
the index k and with a taper ratio a 0:5. The eccentric load is assumed acting at the highest point of the free end section
as considered in [19]. The loaddisplacement curves of the beam under the transverse load and the eccentric axial force are
shown in Figs. 5 and 6, respectively. The effect of the index k is clearly observed from the gures, where the lower the index k
is, the smaller displacements are, regardless of the loading type. This due to the fact that, as seen from Eqs. (1) and (2), the
beam with a smaller index k contains more alumina, and thus it is stiffer. The inuence of the power law index k on the re-
sponse of the beam is also clearly seen from the deformed congurations depicted in Fig. 7, where the higher index k is, the
more strongly the beam deforms. The congurations have been computed by using a mesh of 16 elements in order to ensure
the smoothness of the curves.
4.3. Effect of taper ratio
Table 2 lists the tip response of the FGM beam subjected to a transverse load with different values of the taper ratios for
the case B beam with an index k 3. The table shows an increase in the tip displacements and rotation when increasing the
taper ratio, regardless of the external load level. A careful examination of the table shows that the tip axial displacement is
the most sensitive to the taper ratio, while the transverse displacement is least affected by this parameter. In addition, the
effect of the taper ratio on the tip response of the beam tends to lessen at the higher external load.
4.4. Effect of taper case
Fig. 8 shows the loaddisplacement curves of the beam with different taper cases under a transverse load for an index
k 3 and a taper ratio a 0:5. As seen from the gure, at a given value of the transverse load, the tip displacements of
1.2 0.8 0.4 0 0.4 0.8
0
2
4
6
8
10
Free end displacements, u & w
E
c
c
e
n
t
r
i
c

a
x
i
a
l

l
o
a
d
,

P
pure alumina
pure steel
pure steel
P
u
w
u w
k=0.5
k=1
k=5
case B
Fig. 6. Loaddisplacement curves for case B beam under an eccentric axial force (a 0:5).
0 0.2 0.4 0.6 0.8
0
0.2
0.4
0.6
0.8
1+u
w
k: 5
1
0.5
case B, =0.5
0.1 0 0.2 0.4
0
0.2
0.4
0.6
0.8
1+u
w
k:
0.5
1
5
case B, =0.5
Fig. 7. Deformed congurations corresponding to P 5 for case B beam with different values of index k.
3062 D.K. Nguyen, B.S. Gan/ Applied Mathematical Modelling 38 (2014) 30543066
the case C beam, which has been tapered in both the width and height directions, are considerably higher than that of the
case A and case B beams. The case A beam, which has been tapered in the width direction only, is the most stiff under the
load, and its tip displacements are the smallest among the three considered cases. The deformed congurations correspond-
ing to a load parameter P 5 for the beam with different taper types are shown in Fig. 9. As expected, the case C beam is
weakest in bearing the load, and at the given external load its conguration deforms much larger than that of the case A
and case B beams, regardless of the loading type.
Table 2
Tip response of case B beam under a transverse load with different taper ratios (k 3).
P Response a
0.2 0.4 0.6 0.8
2 juj 0.1483 0.1898 0.2448 0.3118
w 0.4680 0.5120 0.5544 0.5865
h 0.4971 0.5950 0.7295 0.9056
4 juj 0.3083 0.3522 0.3993 0.4459
w 0.6427 0.6651 0.6820 0.6930
h 0.7156 0.7989 0.8929 0.9778
6 juj 0.4108 0.4478 0.4846 0.5193
w 0.7191 0.7305 0.7384 0.7442
h 0.8199 0.8843 0.9478 0.9916
8 juj 0.4799 0.5107 0.5403 0.5679
w 0.7614 0.7678 0.7723 0.7761
h 0.8777 0.9274 0.9712 0.9952
0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 1
0
2
4
6
8
10
Free end displacements, u & w
T
r
a
n
s
v
e
r
s
e

l
o
a
d
,

P
case A
case B
case C
P
u
w
w
u
k=3, =0.5
Fig. 8. Loaddisplacement curves for FGM beam under a transverse load with different taper cases.
0 0.2 0.4 0.6 0.8
0
0.2
0.4
0.6
0.8
1+u
w
taper case:
A
B
C
k=3, =0.5
0.3 0.2 0.1 0 0.1 0.2 0.3 0.4
0
0.2
0.4
0.6
0.8
1+u
w
taper case: A
B
k=3, =0.5
C
Fig. 9. Deformed congurations corresponding to P 5 for FGM beam with different taper cases.
D.K. Nguyen, B.S. Gan/ Applied Mathematical Modelling 38 (2014) 30543066 3063
4.5. Effect of aspect ratio
The tip responses of the case A and case C beams with different values of the aspect ratio, L=h
0
, subjected to a transverse
load are given in Tablse 3 and 4 for the case k 3 and a 0:5, respectively. While both the tip axial and transverse displace-
ments increase when reducing the aspect ratio, the tip rotation reduces, regardless of the external load level and the taper
case. Examining the tables in more detail, one can see that while the tip axial displacement of the case A beam is more sen-
sitive to the aspect ratio comparing to the case C beam, the tip transverse and rotation of the case A beam are much less
sensitive to the aspect ratio than that of the case C beam.
4.6. Beam with different materials
To study the large deection behaviour of the FGM beam formed from different materials, in addition to the beam formed
from steel and alumina (SA) examined above, two other cantilever FGM beams, one formed from aluminum and steel (AS),
and the other from aluminum and alumina (AA) are considered in this Sub-section. The Youngs modulus and Poissons ratio
of aluminum are 70 GPa and 0.3, respectively [31].
Table 3
Tip response of case A beam under a transverse load with different aspect ratios (k 3; a 0:5).
P Response L=h
0
50 25 20 15 10
4 juj 0.3042 0.3047 0.3050 0.3056 0.3071
w 0.6393 0.6394 0.6394 0.6398 0.6417
h 0.7099 0.7082 0.7073 0.7056 0.7020
8 juj 0.4766 0.4769 0.4772 0.4777 0.4791
w 0.7601 0.7603 0.7607 0.7618 0.7659
h 0.8737 0.8711 0.8697 0.8673 0.8618
Table 4
Tip response of case C beam under a transverse load with different aspect ratios (k 3; a 0:5).
P Response L=h
0
50 25 20 15 10
4 juj 0.5152 0.5154 0.5155 0.5159 0.5170
w 0.7547 0.7552 0.7558 0.7574 0.7626
h 0.9626 0.9592 0.9575 0.9544 0.9479
8 juj 0.6346 0.6348 0.6350 0.6355 0.6369
w 0.8242 0.8259 0.8274 0.8310 0.8420
h 0.9903 0.9858 0.9834 0.9795 0.9710
0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 1
0
2
4
6
8
10
Free end displacements, u & w
T
r
a
n
s
v
e
r
s
e

l
o
a
d
,

P
Beam material:
AS
AA
SA
w
u
P
Case C, k=3, =0.5
Fig. 10. Loaddisplacement curves for case C beam formed from different materials under a transverse load.
3064 D.K. Nguyen, B.S. Gan/ Applied Mathematical Modelling 38 (2014) 30543066
Fig. 10 shows the loaddisplacement curves for the case C beam formed from the different materials under a transverse
load for the case k 3 and a 0:5. It is necessary to note that for the purpose of comparison, the load P of all the curves in
the gure is normalized by the Youngs modulus of steel according to Eq. (34). At a given value of the external load, the AA
beam with the Youngs modulus of the bottom surface material is just one third of that of the SA beam, deforms much larger
than the SA beam. The AS beam with the Young moduli of the materials on both the top and bottom surfaces are lower than
that of the SA beam is the most exible. The deformed congurations of the beams at two values of the load parameter,
namely P 5 and P 10, as depicted in Fig. 11 also conrm the inuence the materials on the large deection behaviour
of the beam.
5. Conclusions
The large deections of tapered FGM beams subjected to the end forces have been studied by the nite element method. A
nonlinear rst order shear deformable beam element, using the polynomials as solutions of a uniform homogeneous Timo-
shenko beam segment to interpolate the transverse displacement and rotation, has been formulated in the context of the co-
rotational formulation. The large deection response of a cantilever FGM beam with various taper cases has been computed
by using the arc-length control algorithm in combination with the NewtonRaphson method. The numerical examples have
shown that the formulated element is capable to give accurate response by using just several elements. A parametric study
has been given to examine the effect of the non-homogeneity material, the taper ratio as well as the aspect ratio on the large
deection behaviour of the beams.
Though the numerical examples in the present work were presented for the cantilever beam, the element formulated in
the present work is independent of the boundary conditions, and thus the nite element procedure described in this paper
can be employed to analyze the beams with other boundary conditions.
Acknowledgement
The nancial support from Vietnam NAFOSTED to the rst author is gratefully acknowledged.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/
j.apm.2013.11.032.
References
[1] M. Koizumi, FGM activities in Japan, Compos. Part B Eng. 28 (1997) 14.
[2] V. Birman, L.W. Byrd, Modeling and analysis of functionally graded materials and structures, Appl. Mech. Rev. 60 (2007) 195216.
[3] B.V. Sankar, An elasticity solution for functionally graded beams, Compos. Sci. Technol. 61 (2001) 689696.
[4] A. Chakraborty, S. Gopalakrishnan, J.R. Reddy, A new beam nite element for the analysis of functionally graded materials, Int. J. Mech. Sci. 45 (2003)
519539.
[5] A. Chakraborty, S. Gopalakrishnan, A spectrally formulated nite element for wave propagation analysis in functionally graded beams, Int. J. Solids
Struct. 40 (2003) 24212448.
[6] R. Kadoli, K. Akhtar, N. Ganesan, Static analysis of functionally graded beams using higher order shear deformation beam theory, Appl. Math. Model. 32
(2008) 25092525.
[7] M.A. Benatta, I. Mechab, A. Touns, E.A. Adda Bedia, Static analysis of functionally graded short beams including warping and shear deformation effects,
Comput. Mater. Sci. 44 (2008) 765773.
[8] K.V. Singh, G. Li, Buckling of functionally graded and elastically restrained nonuniform column, Compos. Part B Eng. 40 (2009) 393403.
0 0.1 0.2 0.3 0.4 0.5
0
0.2
0.4
0.6
0.8
1
1+u
AS
AA
SA
case C, k=3, =0.5
Beam material:
w
0 0.1 0.2 0.3 0.4 0.5
0
0.2
0.4
0.6
0.8
1
1+u
AS
AA
SA
case C, k=3, =0.5
Beam material:
w
Fig. 11. Deformed congurations for case C beam formed from different materials under a transverse load.
D.K. Nguyen, B.S. Gan/ Applied Mathematical Modelling 38 (2014) 30543066 3065
[9] Y.A. Kang, X.F. Li, Bending of functionally graded cantilever beam with power-law nonlinearity subjected to an end force, Int. J. Non-linear Mech. 44
(2009) 696703.
[10] Y.A. Kang, X.F. Li, Large deection of a non-linear cantilever functionally graded beam, J. Reinf. Plast. Compos. 29 (2010) 17611774.
[11] Y.Y. Lee, X. Zhao, J.N. Reddy, Postbuckling analysis of functionally graded plates subjected to compressive and thermal loads, Comput. Methods Appl.
Mech. Eng. 199 (2010) 16451653.
[12] Y. Huang, X.F. Lin, A new approach for free vibration of axially functionally graded beams with non-uniform cross-section, J. Sound Vib. 329 (2010)
22912303.
[13] V. Kuti, J. Murn, R. Belk, J. Paulech, Beam element with spatial variation of material properties for multiphysics analysis of functionally graded
materials, Comput. Struct. 89 (2011) 11921205.
[14] A.E. Alshorbagy, M.A. Eltaher, F.F. Mahmoud, Free vibration characteristics of a functionally graded beam by nite element method, Appl. Math. Model.
35 (2011) 412425.
[15] N. Wattanasakulpong, B.G. Gangadhara, D.W. Kelly, Thermal buckling and elastic vibration of third-order shear deformable functionally graded beams,
Int. J. Mech. Sci. 53 (2011) 734743.
[16] A. Shahba, R. Attarnejad, M.T. Marvi, S. Hajila, Free vibration and stability analysis of axially functionally graded tapered Timoshenko beams with
classical and non-classical boundary conditions, Compos. Part B Eng. 42 (2011) 801808.
[17] M. Aminbaghai, J. Murn, V. Kuti, Modal analysis of the FGM-beams with continuous transversal symmetric and longitudinal vibration of material
properties with effect of large axial force, Eng. Struct. 34 (2012) 314329.
[18] D.K. Nguyen, B.S. Gan, T.H. Le, Dynamic response of non-uniform functionally graded beams subjected to a variable speed moving load, J. Comput. Sci.
Technol. (JSME) 7 (2013) 1227.
[19] R.D. Wood, O.C. Zienkiewicz, Geometrically nonlinear nite element analysis of beams, frames, arches and axisymmetric shells, Comput. Struct. 7
(1977) 725735.
[20] W.L. Cleghorn, B. Tabarrok, Finite element formulation of a tapered Timoshenko beams for free lateral vibration analysis, J. Sound Vib. 152 (1992) 461
470.
[21] G. Baker, On the large deections of non-prismatic cantilevers with a nite depth, Comput. Struct. 46 (1993) 365370.
[22] B.K. Lee, J.F. Wilson, S.J. Oh, Elastica of cantilevered beams with variable cross section, Int. J. Non-linear Mech. 28 (1993) 579589.
[23] M. Brojan, T. Videnic, F. Kosel, Large deections of nonlinearly elastic non-prismatic cantilever beams made from materials obeying the generalized
Ludwick law, Meccanica 44 (2009) 733739.
[24] R. Attarnejad, S.J. Semnani, A. Shahba, Basic displacement functions for free vibration analysis of non-prismatic Timoshenko beams, Finite Elem. Anal.
Des. 46 (2010) 916929.
[25] A. Shahba, R. Attarnejad, S.J. Semnani, V. Shahriari, A.A. Dormohamamadi, Derivation of an efcient element for free vibration analysis of rotating
tapered Timoshenko beams using basic displacement functions, Proc. IMechE Part G J. Aerosp. Eng. 226 (2012) 14551469.
[26] M.A. Criseld, Non-linear Finite Element Analysis of Solids and Structures, Volume 1: Essentials, John Wiley & Sons, Chichester, 1991.
[27] C. Pacoste, A. Eriksson, Beam elements in instability problems, Comput. Methods Appl. Mech. Eng. 44 (1997) 163197.
[28] D.K. Nguyen, A Timoshenko beam element for large displacement analysis of planar beams and frames, Int. J. Struct. Stab. Dyn. 12 (2012), http://
dx.doi.org/10.1142/S0219455412500484.
[29] J.B. Kosmatka, An improve two-node nite element for stability and natural frequencies of axial-loaded Timoshenko beams, Comput. Struct. 57 (1995)
141149.
[30] J.R. Reddy, On locking-free shear deformable beam nite elements, Comput. Methods Appl. Mech. Eng. 149 (1997) 113132.
[31] G.N. Praveen, J.N. Reddy, Nonlinear transient thermoelastic analysis of functionally graded ceramic-metal plates, Int. J. Solids Struct. 35 (1998) 4457
4476.
3066 D.K. Nguyen, B.S. Gan/ Applied Mathematical Modelling 38 (2014) 30543066