Journal of Constructional Steel Research 67 (2011) 585592

Contents lists available at ScienceDirect

Journal of Constructional Steel Research
journal homepage: www.elsevier.com/locate/jcsr
Nonlinear inelastic analysis of space frames
Huu-Tai Thai, Seung-Eock Kim

Department of Civil and Environmental Engineering, Sejong University, 98 Gunja Dong Gwangjin Gu, Seoul 143-747, Republic of Korea
a r t i c l e i n f o
Article history:
Received 20 September 2010
Accepted 4 December 2010
Keywords:
Nonlinear analysis
Stability function
Space frame
Geometric nonlinearity
Material nonlinearity
a b s t r a c t
In this paper, a fiber beamcolumn element which considers both geometric and material nonlinearities
is presented. The geometric nonlinearities are captured using stability functions obtained from the
exact stability solution of a beamcolumn subjected to axial force and bending moments. The material
nonlinearities are included by tracing the uniaxial stressstrain relationship of each fiber on the cross
sections. The nonlinear equilibriumequations are solved using an incremental iterative scheme based on
the generalized displacement control method. Using only one element per member in structure modeling,
the nonlinear responses predictedby the proposedelement compare well withthose givenby commercial
finite element packages and other available results. Numerical examples are presented to verify the
accuracy and efficiency of the proposed element.
2010 Elsevier Ltd. All rights reserved.
1. Introduction
In the past few decades, there have been numerous studies
to improve the accuracy of the beamcolumn element for the
nonlinear analysis of steel frames. In general, the nonlinear
response of steel frames can be predicted by using either the
finite element method or the beamcolumn approach. The finite
element approach is often based on a stiffness or displacement
formulation in which cubic and linear interpolation functions are
used for the transverse and axial displacements, respectively [15].
Since this method is based commonly on an assumed cubic
polynomial variationof transverse displacement along the element
length, it is unable to capture accurately the effect of axial force
acting through the lateral displacement of the element (P
effect) when one element per member is used [6]. Hence, it
overestimates the strength of a member under significant axial
force. Although the accuracy of this method can be improved
by using several elements per member in the modeling, it is
generally recognized to be computationally intensive because of
a very refined discretisation of the structures. The beamcolumn
approach is based on the stability functions which are derived from
the exact stability function of a beamcolumn subjected to axial
force and bending moments [712]. This approach can capture
accurately the P effect of a beamcolumn member by using only
one or two elements per member in the modeling, hence, to save
computational time.
In parallel with the above developments, different beam
column models have been proposed to represent inelastic material

Corresponding author. Tel.: +82 2 3408 3291; fax: +82 2 3408 3332.
E-mail addresses: taispkt@yahoo.com (H.-T. Thai), sekim@sejong.ac.kr
(S.-E. Kim).
behavior. These models can be grouped into two categories:
lumped plasticity [9,10,13] model and distributed plasticity model
[5,1418]. In the lumped plasticity model, the inelastic behavior
of material is assumed to be concentrated at point hinges that are
usually located at the ends of the member. The forcedeformation
relation at these hinges is based on force resultants. The advantage
of this model is that it is simple in formulation as well as
implementation. However, the disadvantage of this model is that
the forcedeformationrelationat the hinges is not always available
and accurate for every section. In the distributed plasticity model,
the inelastic behavior of material is distributed along the member
length since the element behavior is monitored through numerical
integration of constitutive behavior at a finite number of control
sections. The nonlinear constitutive behavior at these sections
is derived using one of the following methods: (1) moment
curvature relations; (2) force and deformation resultants; and
(3) uniaxial stressstrain relations of fibers on the cross sections.
Althoughfiber model is the most computationally intensive among
others, it represents the inelastic behavior of material more
accurately and rationally than concentrated plasticity model.
This paper proposes a fiber beamcolumn element for the
nonlinear inelastic analysis of space steel frames. The spread of
plasticity over the cross section and along the member length is
captured by tracing the uniaxial stressstrain relations of each
fiber on the cross sections located at the selected integration
points along the member length. The GaussLobatto integration
rule is adopted herein for evaluating numerically element stiffness
matrix instead of the classical Gauss integration rule because
it always includes the end sections of the integration field.
Since inelastic behavior in beam elements often concentrates at
the ends of the member, monitoring the end sections of the
element results in improved accuracy and numerical stability [19].
0143-974X/$ see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jcsr.2010.12.003
586 H.-T. Thai, S.-E. Kim / Journal of Constructional Steel Research 67 (2011) 585592
Although the fiber model is included in DRAIN-3DX [20] and
OpenSees [21] programs to represent the material nonlinearity,
the geometric nonlinearity caused by the interaction between the
axial force and bending moments (P effect) was not considered.
Therefore, these methods overestimate the strength of a member
subjected to significant axial force if only one or few elements
per member are used in the modeling. In this research, the
stability functions obtained from the closed-form solution of a
beamcolumn subjected to end forces are used to accurately
capture the P effect. Numerical examples are presented to verify
the accuracy and efficiency of the proposed element in predicting
nonlinear inelastic response of space steel frames.
2. Element formulations
2.1. Geometric nonlinear P effect
To capture the effect of axial force acting through the lateral
displacement of the beamcolumn element (P effect), the
stability functions reported by Chen and Lui [22] are used to
minimize modeling and solution time. Generally only one element
per member is needed to accurately capture the P effect. From
Kim et al. [10], the incremental forcedisplacement equation of
space beamcolumn element which accounts for transverse shear
deformation effects can be expressed as
_

_
P
M
yA
M
yB
M
zA
M
zB
T
_

_
=
_
_
_
_
_
_
_
_
_
EA
L
0 0 0 0 0
0 C
1y
C
2y
0 0 0
0 C
2y
C
1y
0 0 0
0 0 0 C
1z
C
2z
0
0 0 0 C
2z
C
1z
0
0 0 0 0 0
GJ
L
_

yA

yB

zA

zB

_
(1)
where P, M
yA
, M
yB
, M
zA
, M
zB
, and T are incremental
axial force, end moments with respect to y and z axes, and
torsion respectively; ,
yA
,
yB
,
zA
,
zB
, and are the
incremental axial displacement, joint rotations, and angle of twist;
C
1y
, C
2y
, C
1z
, and C
2z
are bending stiffness coefficients accounting
for the transverse shear deformation effects, and are expressed as
C
1y
=
k
2
1y
k
2
2y
+ k
1y
A
sz
GL
2k
1y
+ 2k
2y
+ A
sz
GL
(2a)
C
2y
=
k
2
1y
+ k
2
2y
+ k
2y
A
sz
GL
2k
1y
+ 2k
2y
+ A
sz
GL
(2b)
C
1z
=
k
2
1z
k
2
2z
+ k
1z
A
sy
GL
2k
1z
+ 2k
2z
+ A
sy
GL
(2c)
C
2z
=
k
2
1z
+ k
2
2z
+ k
2y
A
sy
GL
2k
1z
+ 2k
2z
+ A
sy
GL
(2d)
where k
1n
= S
1n
(EI
n
/L) and k
2n
= S
2n
(EI
n
/L); S
1n
and S
2n
are
stability functions with respect to n axis (n = y, z), and are
expressed as
S
1n
=
_

_
k
n
L sin(k
n
L) (k
n
L)
2
cos(k
n
L)
2 2 cos(k
n
L) k
n
L sin(k
n
L)
if P < 0
(k
n
L)
2
cosh(k
n
L) k
n
L sinh(k
n
L)
2 2 cosh(k
n
L) + k
n
L sinh(k
n
L)
if P > 0
(3a)
S
2n
=
_

_
(k
n
L)
2
k
n
L sin(k
n
L)
2 2 cos(k
n
L) k
n
L sin(k
n
L)
if P < 0
k
n
L sin(k
n
L) (k
n
L)
2
2 2 cosh(k
n
L) + k
n
L sinh(k
n
L)
if P > 0
(3b)
where k
2
n
= |P|/EI
n
. EA, EI
n
, and GJ denote the axial, bending and
torsional stiffness of the beamcolumn element, and are defined as
EA =
h

j=1
w
j
_
m

i=1
E
i
A
i
_
j
(4)
EI
y
=
h

j=1
w
j
_
m

i=1
E
i
A
i
z
2
i
_
j
(5)
EI
z
=
h

j=1
w
j
_
m

i=1
E
i
A
i
y
2
i
_
j
(6)
GJ =
h

j=1
Gw
j
_
m

i=1
(y
2
i
+ z
2
i
)A
i
_
j
(7)
in which h is the total number of monitored sections along an
element; m is the total number of fibers divided on the monitored
cross section; w
j
is the weighting factor of the jth section; E
i
and A
i
are the tangent modulus of the material and the area of ith fiber,
respectively; y
i
and z
i
are the coordinates of ith fiber in the cross
section. The element forcedeformation relationship of Eq. (1) can
be expressed in symbolic form as
{F} = [K
e
]{d} (8)
where
{F} =
_
P M
yA
M
yB
M
zA
M
zB
T
_
T
(9)
{d} =
_

yA

yB

zA

zB

_
T
. (10)
The element stiffness matrix is evaluated numerically by the
GaussLobatto integration scheme since this method allows for
two integration points to coincide with the end sections of the
elements [23]. Since inelastic behavior in beam elements often
concentrates at the ends of the member, the monitoring of the
end sections of the element is advantageous from the standpoint
of accuracy and numerical stability. By contrast, the outermost
integration points of the classical Gauss integration method only
approach the end sections with increasing order of integration,
but never coincide with the end sections and, hence, result in
overestimation of the member strength [24].
2.2. Material nonlinear effect
In order to capture the gradual plastification throughout the
members cross section, a fiber model as shown in Fig. 1 is used.
The fiber beamcolumn element is divided into a discrete number
of monitored sections represented by the integration points. Each
monitored section is divided into m fibers and each fiber is
represented by its area A
i
and coordinate location corresponding to
its centroid (y
i
, z
i
). Section deformations are represented by three
strain resultants: the axial strain along the longitudinal axis and
two curvatures
z
and
y
with respect to z and y axes, respectively.
The corresponding force resultants are the axial force N and two
bending moments M
z
and M
y
. The section forces and deformations
are grouped in the following vectors:
Section force vector {Q} =
_
M
z
M
y
N
_
T
(11)
Section deformation vector {q} =
_

z

y

_
T
. (12)
H.-T. Thai, S.-E. Kim / Journal of Constructional Steel Research 67 (2011) 585592 587
Fig. 1. Fiber concept.
The incremental section force vector at each integration points is
determined based on the incremental element force vector {F}
as
{Q} = [B(x)] {F} (13)
where [B(x)] is the force interpolation function matrix given as
[B(x)] =
_

y
(x) 0 0 (x/L 1) x/L 0

z
(x) (x/L 1) x/L 0 0 0
1 0 0 0 0 0
_
(14)
where
y
(x) and
z
(x) are the lateral displacements for the local y
and z axes, respectively. Since the curvature can be approximated
by the secondderivative of the lateral displacement,
y
(x) and
z
(x)
are obtained from solving the differential equations as [22]

y
(x) =
M
zA
EI
z
k
2
z
_
sin(k
z
x)
tan(k
z
L)
cos(k
z
x)
x
L
+ 1
_

M
zB
EI
z
k
2
z
_
sin(k
z
x)
sin(k
z
L)

x
L
_
(15)

z
(x) =
M
yA
EI
y
k
2
y
_
sin(k
y
x)
tan(k
y
L)
cos(k
y
x)
x
L
+ 1
_
+
M
yB
EI
y
k
2
y
_
sin(k
y
x)
sin(k
y
L)

x
L
_
. (16)
The section deformation vector is determined based on the section
force vector as
{q} = [k
sec
]
1
{Q} (17)
where [k
sec
] is the section stiffness matrix given as
[k
sec
] =
_
_
_
_
_
_
_
_
_
m

i=1
E
i
A
i
y
2
i
m

i=1
E
i
A
i
y
i
z
i
m

i=1
E
i
A
i
(y
i
)
m

i=1
E
i
A
i
y
i
z
i
m

i=1
E
i
A
i
z
2
i
m

i=1
E
i
A
i
z
i
m

i=1
E
i
A
i
(y
i
)
m

i=1
E
i
A
i
z
i
m

i=1
E
i
A
i
_

_
. (18)
Following the hypothesis that plane sections remainplane andnor-
mal to the longitudinal axis, the incremental uniaxial fiber strain
vector is computed based on the incremental section deformation
vector as
{e} = [l]{q} (19)
where [l] is the linear geometric matrix given as follows
[l] =
_
_
_
y
1
z
1
1
y
2
z
2
1

y
m
z
m
1
_

_. (20)
Once the incremental fiber strain is evaluated, the incremental
fiber stress is computed based on the stressstrain relationship of
material model. The tangent modulus of each fiber is updated from
the incremental fiber stress and incremental fiber strain as
E
i
=

i
e
i
. (21)
Eq. (21) leads to updating of the element stiffness matrix [K
e
] in
Eq. (8) and section stiffness matrix [k
sec
] in Eq. (18) during the
iteration process. Based on the new tangent modulus of Eq. (21),
the location of the section centroid is also updated during the
incremental load steps to take into account the distribution of
section plasticity. The section resisting forces are computed by
summation of the axial force and biaxial bending moment contri-
butions of all fibers as
{Q
R
} =
_
M
z
M
y
N
_
=
_

_
m

i=1

i
A
i
(y
i
)
m

i=1

i
A
i
z
i
m

i=1

i
A
i
_

_
. (22)
2.3. Element stiffness matrix accounting for P effect
The P effect is the effect of axial force P acting through the
relative transverse displacement of the member ends . This effect
can be considered by using the geometric stiffness matrix [K
g
] as
[K
g
]
1212
=
_
[K
s
] [K
s
]
[K
s
]
T
[K
s
]
_
(23)
where
[K
s
] =
_
_
_
_
_
_
0 a b 0 0 0
a c 0 0 0 0
b 0 c 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
_

_
(24)
and
a =
M
zA
+ M
zB
L
2
, b =
M
yA
+ M
yB
L
2
, c =
P
L
. (25)
The displacement of a beamcolumn element can be decomposed
into two parts: the element deformation and rigid displacement.
The element deformation increment {d} in Eq. (10) can be
obtained from the element displacement increment {D} as
{d} = [T]
612
{D} (26)
where
[T]
612
=
_
_
_
_
_
_
_
_
1 0 0 0 0 0 1 0 0 0 0 0
0 0 1/L 0 1 0 0 0 1/L 0 0 0
0 0 1/L 0 0 0 0 0 1/L 0 1 0
0 1/L 0 0 0 1 0 1/L 0 0 0 0
0 1/L 0 0 0 0 0 1/L 0 0 0 1
0 0 0 1 0 0 0 0 0 1 0 0
_

_
. (27)
588 H.-T. Thai, S.-E. Kim / Journal of Constructional Steel Research 67 (2011) 585592
The tangent stiffness matrix of a beamcolumnelement is obtained
as follows
[K]
1212
= [T]
T
612
[K
e
]
66
[T]
612
+ [K
g
]
1212
. (28)
3. Nonlinear solution procedure
This section presents a numerical method for solving the
nonlinear equations. Among several numerical methods, the
GDC method proposed by Yang and Shieh [25] appears to be
one of the most robust and effective methods for solving the
nonlinear problems with multiple critical points because of its
general numerical stability and efficiency. The incremental form
of equilibriumequation can be rewritten for the jth iteration of the
ith incremental step as
[K
i
j1
]{D
i
j
} =
i
j
{

P} + {R
i
j1
} (29)
where [K
i
j1
] is the tangent stiffness matrix, {D
i
j
} is the
displacement increment vector, {

P} is the reference load vector,

{R
i
j1
} is the unbalanced force vector, and
i
j
is the load increment
parameter.
Eq. (29) can be decomposed into the following equations
[K
i
j1
]{

D
i
j
} = {

P} (30)
[K
i
j1
]{

D
i
j
} = {R
i
j1
} (31)
{D
i
j
} =
i
j
{

D
i
j
} + {

D
i
j
}. (32)
The loadincrement parameter
i
j
is unknown. It is determinedfrom
a constraint condition. For the first iterative step (j = 1), the load
increment parameter
i
j
is determined based on the Generalized
Stiffness Parameter (GSP) as

i
1
=
1
1
_
|GSP| (33)
where
1
1
is an initial value of a load increment parameter, and the
GSP is defined as
GSP =
{

D
1
1
}
T
{

D
1
1
}
{

D
i1
1
}
T
{

D
i
1
}
. (34)
For the following iteration (j 2), the load increment parameter

i
j
is computed as

i
j
=
{

D
i1
1
}
T
{

D
i
j
}
{

D
i1
1
}
T
{

D
i
j
}
(35)
where {

D
i1
1
} is the displacement increment generated by the
reference load {

P} at the first iteration of the previous (i 1)

incremental step, and {

D
i
j
} and {

D
i
j
} denote the displacement
increments generated by the reference load and unbalanced force
vectors, respectively, at the jth iteration of the ith incremental step,
as defined in Eqs. (30) and (31).
The following is a step-by-step summary of solution algorithm
focused on the element state determination process of a single
iteration.
1. Solve the global equation and update the element displace-
ment increment {D}.
2. Compute the element deformation increment {d} using
Eq. (26).
3. Compute the element force increment {F} using Eq. (8) based
on the element stiffness matrix [K
e
]
66
of the previous step.
4. Compute the section force increment {Q} using Eq. (13).
5. Compute the section stiffness {k
sec
} using Eq. (18).
6. Compute the section deformation increment {q} using
Eq. (17).
(a) Cantilever column. (b) Simply supported column.
Fig. 2. Steel columns.
7. Compute the fiber strain increment {e} using Eq. (19) and
update fiber strain {e}.
8. Compute the fiber stress {} based on the constitutive model.
9. Update the fiber tangent modulus {E} using Eq. (21).
10. Compute the section resisting force {Q
R
} using Eq. (22).
11. Update the element stiffness [K
e
]
66
and the element tangent
stiffness [K]
1212
.
12. Assemble the structure resisting force and structure stiffness
matrix.
13. Compute structure unbalanced forces.
14. Check for the structure convergence: If the structure unbal-
anced forces satisfy the specified tolerance (i.e., convergence
is achieved), go to the next increment load step. Otherwise,
return to step 1 for the next iteration to eliminate the struc-
ture unbalanced forces.
4. Numerical examples
A computer program is developed based on the above-
mentioned formulations to predict the strength and behavior of
framed structures. It is verified for accuracy and efficiency by
comparing the predictions with those generated by commercial
finite element packages and other available results through
four numerical examples. The first example is to show how
the proposed element captures geometric nonlinearity effects
accurately and efficiently. The remains are to verify how well the
proposed element predicts the ultimate strength and behavior of
framed structures. In addition, a case study of large-scale twenty-
story space frame is studied to demonstrate the capability of
proposed program. An elasticperfectly plastic material model is
used for all numerical examples. The B23 beam element with
13 numerical integration points (five points in web, five in each
flange) of ABAQUS is used to model the framed structures herein.
4.1. Flexural buckling of columns
The aim of this example is to verify the accuracy and efficiency
of the proposedelement incapturing the buckling loads of columns
with different end conditions. Fig. 2 shows simply supported and
cantilever columns. The section of columns is W8 31. Youngs
modulus and Poissons ratio of the material are E = 200,000 MPa
and = 0.3, respectively.
The loaddeflection curves of the columns obtained by the
proposed element and SAP2000 are compared in Fig. 3. Since
the proposed element is based on the stability functions, which
satisfy the equilibrium equation of the beamcolumn, to derive
the closed-form terms of tangent stiffness matrix, it is capable of
accurately predicting the buckling load of the column by using one
element per member. Meanwhile, the frame element of SAP2000
is based on the cubic interpolation functions to approximate the
transverse displacements. Therefore, it cannot predict accurately
H.-T. Thai, S.-E. Kim / Journal of Constructional Steel Research 67 (2011) 585592 589
(a) Cantilever column. (b) Simply supported column.
Fig. 3. Loaddeflection curve of steel columns.
Fig. 4. Fixed-ended beam.
the buckling load of the columns unless the column is modeled
by a number of elements. The loaddeflection curves shown in
Fig. 3 indicate that SAP2000 requires more thanfive cubic elements
per member in modeling to match the results predicted by the
proposed element.
4.2. Fixed-ended beam
The fixed-endedbeamwitha concentratedloadat the one-third
length of the beam is shown in Fig. 4 with its associated data. This
beam was analyzed by Jiang [26] using the plastic zone method.
Since the inelastic behavior of this beam concentrates at the two
end sections of member, two integration points along the member
length corresponding to two end sections are used in the proposed
element. The cross section is discretized into sixteen fibers (eight
at both flanges, eight at the web).
The analysis results of the beamobtained by using the proposed
element are compared with those predicted by Jiang [26] and
ABAQUS as illustrated in Fig. 5 and Table 1. Since the proposed
element adopts the GaussLobatto integration scheme, which
allows for two integration points to coincide with the end sections
of the elements, in evaluating the stiffness matrix, it can predict
accurately the inelastic behavior of the beam by using only one
element per member. Whereas, outermost integration points of
the classical Gauss integration scheme adopted in ABAQUS, only
approach the end sections with increasing order of integration,
but never coincide with the end sections and, hence, result in
overestimation of the member strength. It can be seen that both
beamelement providedby ABAQUS andplastic zone element given
by Jiang cannot predict accurately the inelastic response the beam
unless several elements per member are used.
Fig. 5. Loaddeflection curve of fixed-ended beam.
Table 1
Comparison of ultimate load factor of fixed-ended beam.
Method Ultimate load factor Difference (%)
Plastic zone (108 elements),
Jiang [26]
8.996
ABAQUS (2 elements) 11.611 29.07
ABAQUS (20 elements) 9.534 5.98
ABAQUS (40 elements) 9.242 2.73
ABAQUS (108 elements) 9.079 0.92
Proposed (2 elements) 9.003 0.08
Table 2
Comparison of ultimate load factor of portal frame.
Method Ultimate load factor Difference (%)
ABAQUS, 1 ele/member 1.206 46.18
ABAQUS, 5 ele/member 0.860 4.24
ABAQUS, 20 ele/member 0.826 0.12
Proposed, 1 ele/member 0.825
4.3. Portal frame
A portal frame subjected to the combined action of gravity load
and lateral load is shown in Fig. 6. The aim of this example is to
demonstrate the capability of the proposed element in capturing
the effects of both geometric and material nonlinearities. Since
the inelastic behavior of this frame distributes along the length
of member due to gravity load, five integration points along the
member length with eighty fibers on the cross section are used in
the proposed element.
Table 2 and Fig. 7 show the comparison of ultimate load factors
and loaddeflection curves of the frame predicted by using the
proposed element and the B23 beam element of ABAQUS. It can
be seen that the ABAQUS needs more than twenty elements per
590 H.-T. Thai, S.-E. Kim / Journal of Constructional Steel Research 67 (2011) 585592
Fig. 6. Portal frame.
Fig. 7. Loaddeflection curve of portal frame.
member to match the results predicted by the proposed element
using only one element per member. To obtain the accurate results
of this portal frame, ABAQUS needs fewer elements per member
compared to the beam structure in the previous example. This is
due to the existence of gravity load applied in the frame. With
the increase of gravity load, the second-order effect contributes
more to the inelastic buckling of the frame, and the spreading of
plasticity inthe members is less severe. This example indicates that
the proposed element can capture accurately both geometric and
material nonlinear effects of framed structures.
4.4. Two-story space frame
The two-story space depicted in Fig. 8 was previously analyzed
by De Souza [27] using the force-based method with fiber model,
and recently by Thai and Kim [13] using beamcolumn method
with refined plastic hinge model. Youngs modulus, Poissons ratio,
and the yield stress of material are E = 19,613 MPa, = 0.3, and

y
= 98 MPa, respectively. Both De Souza and Thai and Kim used
one element per member in the modeling. In the present work,
each member is also modeled by one element with five integration
points along the member length, and eighty fibers on the cross
section.
Fig. 9 shows the comparison of loaddeflection response of the
frame generated by using the proposed element, De Souza [27],
and Thai and Kim [13]. The ultimate loads of the frame are also
compared in Table 3. It can be seen that the results of the proposed
element are well compared with those of force-based method and
beamcolumn method with refined plastic hinge model. It is noted
that for this particular frame, the use of only one element per
member in the plastic hinge analysis by Thai and Kimis sufficiently
accurate. This is due to the fact that the plastic hinge analysis is
Fig. 8. Two-story space frame.
Fig. 9. Loaddeflection curve of two-story space frame.
Table 3
Comparison of ultimate load of two-story space frame.
Method Ultimate load (kN) Difference (%)
De Souza [27] 128.05
Thai and Kim [13] 128.50 0.35
Proposed 128.82 0.60
sufficient for frames with plasticity concentrated at the ends of the
members.
4.5. Case study
The twenty-story space steel frame shown in Fig. 10 was
analyzed by Jiang et al. [5] and Chiorean and Barsan [28] using the
mixed element and distributed plasticity methods, respectively.
A50 steel with a yield stress of 344.8 MPa and Youngs modulus
of 200,000 MPa is used for all sections. The load applied to the
structure consists of gravity loads of 4.8 kN/m
2
and wind loads
of 0.96 kN/m
2
acting in the Y-direction. These loads are converted
into concentrated loads applied at the beamcolumn joints.
Jiang used both the plastic hinge and spread-of-plasticity ele-
ments to model this structure to shorten the computational time
because the use of a full spread-of-plasticity analysis is very com-
putationally intensive. Chiorean and Barsan employed the inelastic
RambergOsgood forcestrain relationships to model the gradual
yielding of cross section. In their analysis, one element with
seven integration points was used to model each beamcolumn
H.-T. Thai, S.-E. Kim / Journal of Constructional Steel Research 67 (2011) 585592 591
(a) Plan view.
(b) Perspective view.
Fig. 10. Twenty-story space frame.
Fig. 11. Loaddeflection curve of twenty-story space frame.
member and two values of n = 30 (corresponding to plastic zone
approach) and n = 300 (corresponding to plastic hinge approach)
for RambergOsgood shape parameters were considered. In this
study, each member is modeled by one element with two integra-
tion points along the member length, and eight fibers on the cross
section (six at both flanges, two at the web).
The loaddeflection curves of node A at the roof of the frame
obtained by the present study and the others are compared in
Fig. 11. The ultimate load factor of the frame is also given in Table 4.
Table 4
Comparison of ultimate load factor of twenty-story space frame.
Method Ultimate load factor Difference (%)
Jiang et al. [5] 1.000
Chiorean and Barsan [28], n = 30 1.069 6.90
Chiorean and Barsan [28],
n = 300
1.006 0.60
Proposed 1.0002 0.02
It can be seen that the ultimate load factor predicted by using
the present software (1.0002) is close to those of Jiang (1.0)
and Chiorean and Barsan using n = 300 (1.006), and the load
deflection curves of the present software are similar to those
of Chiorean and Barsan using n = 300. This example is a good
illustration on the accuracy and efficiency of the proposed element
in predicting the nonlinear inelastic behavior of the large-scale
space frames.
5. Conclusions
A fiber beamcolumn element was successfully developed
for the nonlinear inelastic analysis of framed structures. The
stability functions derived from the exact stability solution of a
beamcolumn subjected to axial force and bending moments are
used to capture the second-order effects. The fiber model based
on tracing the uniaxial stressstrain relations of each fiber on
the cross sections is employed to capture the inelastic effects.
The computer program developed for this work is verified for
accuracy and efficiency through numerical examples and a case
study of large-scale twenty-story space frame. As shown in some
numerical examples, the proposed element is capable of reliably
predicting the nonlinear inelastic behavior of frames by using only
one element per member in the modeling. It can be concluded that
the proposed method proves to be a reliable and efficient tool for
daily use in engineering design of framed structures.
Acknowledgement
This work was supported by the Human Resources Develop-
ment of the Korea Institute of Energy Technology Evaluation and
Planning (KETEP) grant funded by the Korea government Ministry
of Knowledge Economy (No. 20104010100520).
References
[1] Yang YB, Kuo SR. Theory &analysis of nonlinear framed structures. Englewood
Cliffs (NJ): Prentice-Hall; 1994.
[2] Torkamani M, Sonmez M, Cao J. Second-order elastic plane-frame analysis
using finite-element method. Journal of Structural Engineering 1997;123(9):
122535.
[3] Meek JL, Xue Q. A study on the instability problem for 3D frames. Computer
Methods in Applied Mechanics and Engineering 1998;158(34):23554.
[4] Neuenhofer A, Filippou F. Evaluation of nonlinear frame finite-element
models. Journal of Structural Engineering 1999;123(7):958.
[5] Jiang XM, Chen H, LiewJYR. Spread-of-plasticity analysis of three-dimensional
steel frames. Journal of Constructional Steel Research 2002;58(2):193212.
[6] Teh LH. Cubic beam elements in practical analysis and design of steel frames.
Engineering Structures 2001;23(10):124355.
[7] Goto Y, Chen WF. Second order elastic analysis for frame design. Journal of
Structural Engineering 1987;113(7):150119.
[8] Ekhande SG, Selvappalam M, Madugula MKS. Stability functions for three
dimensional beam columns. Journal of Structural Engineering 1989;115(2):
46779.
[9] Liew JYR, Chen H, Shanmugam NE, Chen WF. Improved nonlinear plastic
hinge analysis of space frame structures. Engineering Structures 2000;22(10):
132438.
[10] Kim SE, Park MH, Choi SH. Direct design of three-dimensional frames using
practical advanced analysis. Engineering Structures 2001;23(11):1491502.
[11] Ngo-Huu C, Kim SE, Oh JR. Nonlinear analysis of space steel frames using fiber
plastic hinge concept. Engineering Structures 2007;29(4):64957.
[12] Ngo-Huu C, Kim SE. Practical advanced analysis of space steel frames using
fiber hinge method. Thin-Walled Structures 2009;47(4):42130.
592 H.-T. Thai, S.-E. Kim / Journal of Constructional Steel Research 67 (2011) 585592
[13] Thai HT, Kim SE. Practical advanced analysis software for nonlinear inelastic
analysis of space steel structures. Advances in Engineering Software 2009;
40(9):78697.
[14] Del Coz Diaz J, Nieto P, Fresno D, Fernandez E. Non-linear analysis of
cable networks by FEM and experimental validation. International Journal of
Computer Mathematics 2009;86(2):30113.
[15] Del Coz Diaz J, Garcia Nieto P, Vilan Vilan J, Suarez Sierra J. Non-linear buckling
analysis of a self-weighted metallic roof by FEM. Mathematical and Computer
Modelling 2010;51(34):21628.
[16] Del Coz Diaz J, Garcia Nieto P, Vilan Vilan J, Martin Rodriguez A, Prado
Tamargo J, Lozano M. Non-linear analysis and warping of tubular pipe
conveyors by the finite element method. Mathematical and Computer
Modelling 2007;46(12):95108.
[17] Del Coz Diaz J, Garcia Nieto P, Fernandez Rico M, Suarez Sierra J. Non-linear
analysis of the tubular heart joint by FEM and experimental validation.
Journal of Constructional Steel Research 2007;63(8):107790.
[18] Del Coz Diaz J, Garcia Nieto P, Betegon Biempica C, Fernandez Rougeot G. Non-
linear analysis of unboltedbase plates by the FEMandexperimental validation.
Thin-Walled Structures 2006;44(5):52941.
[19] Spacone E, Ciampi V, Filippou FC. Mixed formulation of nonlinear beam finite
element. Computers and Structures 1996;58(1):7183.
[20] Prakash V, Powell GH. DRAIN-3DX: base program user guide, version 1.10. A
computer program distributed by NISEE/Computer applications. Department
of Civil Engineering, University of California at Berkeley; 1993.
[21] OpenSees. Open system for earthquake engineering simulation. Pacific
Earthquake Engineering Research Center, University of California at Berkeley;
2009.
[22] Chen W, Lui E. Structural stability: theory and implementation. Amsterdam:
Elsevier; 1987.
[23] Stroud A, Secrest D. Gaussian quadrature formulas. Englewood Cliffs (NJ):
Prentice-Hall; 1966.
[24] Spacone E, Filippou F, Taucer F. Fibre beamcolumn model for non-linear
analysis of R/C frames: part I. formulation. Earthquake Engineering and
Structural Dynamics 1996;25(7):71125.
[25] Yang YB, Shieh MS. Solution method for nonlinear problems with multiple
critical points. AIAA Journal 1990;28(12):21106.
[26] Jiang XM. Second-order spread-of-plasticity analysis of spatial steel frames.
Master thesis. Department of Civil Engineering, National University of
Singapore; 2000.
[27] De Souza R. Force-based finite element for large displacement inelastic
analysis of frames. Ph.D. dissertation. Department of Civil and Environmental
Engineering, University of California at Berkeley; 2000.
[28] Chiorean CG, Barsan GM. Large deflection distributed plasticity analysis of 3D
steel frameworks. Computers & Structures 2005;83(1920):155571.