Engineering Structures 23 (2001) 14911502

www.elsevier.com/locate/engstruct

Direct design of three-dimensional frames using practical advanced

analysis
a,*
Seung-Eock Kim , Moon-Ho Park b, Se-Hyu Choi c

a
Department of Civil and Environmental Engineering, Sejong University, 98 Koonja-dong, Kwangjin-ku, Seoul 143-747, South Korea
b
Kyungpook National University, Taegu, South Korea
c
Construction Tech. Research Institute, Sejong University, Seoul, South Korea

Received 19 October 2000; received in revised form 6 March 2001; accepted 9 April 2001

Abstract

A new design method of three-dimensional frames using practical advanced analysis is presented. In this method separate member
capacity checks encompassed by the code specifications are not required, because the stability of separate members and the structure
as a whole can be rigorously treated in determining the maximum strength of the structures. To capture second-order effects
associated with P-d and P- effects, stability functions are used to minimize modeling and solution time. Generally, only one or
two elements are needed per member. The Column Research Council (CRC) tangent modulus concept is used to account for gradual
yielding due to residual stresses. A softening plastic hinge model is used to represent the transition from elastic to zero stiffness
associated with a developing hinge. The load-displacements predicted by the proposed analysis compare well with those given by
other approaches. A design example has been presented for a 22-story frame. The analysis results show that the proposed method
is suitable for adoption in practice. 2001 Published by Elsevier Science Ltd.

Keywords: Advanced analysis; Geometric nonlinearity; Material nonlinearity; Stability function; Steel design; Steel frames

1. Introduction The second and perhaps the most serious limitation is

probably the rationale of the current two-stage process
In the current engineering practice, the interaction in design: elastic analysis is used to determine the forces
between the structural system and its members is rep- acting on each member of a structural system, whereas
resented by the effective length factor. The effective inelastic analysis is used to determine the strength of
length method generally provides a good design of each member treated as an isolated member. There is
framed structures. However, despite its popular use in no verification of the compatibility between the isolated
the past and present as a basis for design, the approach member and the member as part of a frame. The individ-
has its major limitations. The first of these is that it does ual member strength equations as specified in specifi-
not give an accurate indication of the factor against fail- cations are unconcerned with system compatibility. As
ure, because it does not consider the interaction of a result, there is no explicit guarantee that all members
strength and stability between the member and structural will sustain their design loads under the geometric con-
system in a direct manner. It is a well-recognized fact figuration imposed by the frame work.
that the actual failure mode of the structural system often With the development of computer technology, two
does not have any resemblance whatsoever to the elastic aspects, the stability of separate members, and the stab-
buckling mode of the structural system that is the basis ility of the structure as a whole, can be treated rigorously
for the determination of the effective length factor K. for the determination of the maximum strength of the
structures. This design approach is marked in Fig. 1 as
the direct analysis and design method. The development
* Corresponding author. Tel.: +822-3408-3291; fax: +822-3408- of the direct approach to design is called Advanced
3332. Analysis or more specifically, Nonlinear Inelastic
E-mail address: [email protected] (S.-E. Kim). Analysis. In this direct approach, there is no need to

0141-0296/01/$ - see front matter 2001 Published by Elsevier Science Ltd.

PII: S 0 1 4 1 - 0 2 9 6 ( 0 1 ) 0 0 0 4 1 - 4
1492 S.-E. Kim et al. / Engineering Structures 23 (2001) 14911502

ness associated with a developing hinge. The plastic sur-

faces proposed by LRFD [14] and Orbison [6] are used.
The proposed analysis is limited to steel frames sub-
jected to static loads. Bowing effect is not considered,
and lateral torsional buckling is assumed to be prevented
by adequate lateral bracing. The section of members is
assumed to be compact which can develop full plastic
moment capacity without local buckling. Warping tor-
sion is ignored. Unloading effects associated with dis-
tributed loads is not considered.

Fig. 1. Direct approach for steel design.

2. Practical advanced analysis

compute the effective length factor, since separate mem- 2.1. Stability functions accounting for second-order
ber capacity checks encompassed by the specification effect
equations are not required. With the current available
computing technology with advancement in computer To capture second-order (large displacement) effects,
hardware and software, it is feasible to employ advanced stability functions are used to minimize modeling and
analysis techniques for direct frame design. This method solution time. Generally only one or two elements are
has been considered impractical for design office use in needed per member. The simplified stability functions
the past. Fig. 1 compares the conventional approach and reported by Chen and Lui [15] are used here. Consider-
the direct approach [1]. ing the prismatic beam-column element in Fig. 2, the
In order to overcome the difficulties of the conven- incremental force-displacement relationship of this
tional approach, advanced analysis (nonlinear inelastic element may be written as
analysis) should be directly performed. One of the
advanced analyses called the plastic-zone method
discretizes frame members into several finite elements.

S1 S2 0

Also the cross-section of each finite element is further MA qA
subdivided into many fibers [25]. Although the plastic- EI S2 S1 0
MB qB (1)
zone solution is known as an exact solution, it is yet L A
to be used for practical design purposes. The applica- P 0 0 e
I
bility of the method is limited by its complexity requir-
ing intensive computational time and cost. The real chal- where: S1, S2 are stability functions; MA, MB are
lenge in our endeavor is to make this type of analysis incremental end moments; P is the incremental axial
competitive in present construction engineering prac- force; qA, qB are incremental joint rotations; e is the
tices. incremental axial displacement; A, I, L are area, moment
The finite element approaches using shape functions
based on the assumed displacement shape were
developed by Orbison [6], Yang and Kuo [7]. Two-
dimensional beam-column approaches using stability
function based on the basic member force and defor-
mation relationship were developed by Oran [8], Kassi-
mali and Abbasnia [9], and Chan and Zhou [10]. Various
nonlinear inelastic analyses for space structures were
developed by Ziemian et al [11], Prakash and Powell
[12], and Liew and Tang [13].
The purpose of this paper is to present a practical and
direct design method of space frames using stability
function based on the basic member force and defor-
mation relationship. Stability functions are used to cap-
ture second-order effects associated with P-d and P-
effects. The Column Research Council (CRC) tangent
modulus concept is used to account for gradual yielding
due to residual stresses. A parabolic function model is
used to represent the transition from elastic to zero stiff- Fig. 2. Beam-column subjected to double-curvature bending.
 S.-E. Kim et al. / Engineering Structures 23 (2001) 14911502 1493

The force-displacement equation may be extended for

three-dimensional beam-column element as

P
MyA
MyB
(4)
MzA
MzB
T
EA

0 0 0 0 0
Fig. 3. Full plastification surface of AISC-LRFD. L
ELy EIy
0 S1 S2 0 0 0

L L d
of inertia, and length of beam-column element, respect-
ively; and, E is the modulus of elasticity (Figs. 3 and 4). EIy EIy qyA
0 S2 S1 0 0 0
The stability functions given by Eq. (1) may be writ- L L qyB

ten as
EIz EIz qzA
0 0 0 S3 S 0

L 4L qzB
prsin(pr)p rcos(pr)
2 EIz EIz f
0 0 0 S4 S 0
22cos(pr)prsin(pr)
if P0 L 3L
S1 (2) GJ
p2rcosh(pr)prsinh(pr) 0

0 0 0 0
L
22cosh(pr)+prsinh(pr)
if P0
where P, Mya, MyB, MzA, MzB, and T are axial force,
end moments with respect to y and z axes and torsion
respectively. d, qyA, qyB, qzA, qzB, and f are the axial

displacement, the joint rotations, and the angle of twist.
p2rprsin(pr)
S1, S2, S3, and S4 are the stability functions with respect
to y and z axes, respectively.
22cos(pr)prsin(pr)
if P0
S2 (3) 2.2. CRC tangent modulus model associated with
prsinh(pr)p2r residual stresses
22cosh(pr)+prsin(pr)
if P0
The CRC tangent modulus concept is used to account
for gradual yielding (due to residual stresses) along the
length of axially loaded members between plastic hinges.
where r=P/(p2EI/L2), P is positive in tension. The elastic modulus E (instead of moment of inertia I)
is reduced to account for the reduction of the elastic por-
tion of the cross-section since the reduction of the elastic
modulus is easier to implement than a new moment of
inertia for every different section. The rate of reduction
in stiffness is different in the weak- and strong-direc-
tions, but this is not considered since the dramatic degra-
dation of weak-axis stiffness is compensated for by the
substantial weak-axis plastic strength [16]. This simpli-
fication makes the present methods practical. From Chen
and Lui [17], the CRC Et is written as
Et1.0E for P0.5Py (5a)
P
Et4 E 1
Py
P
Py
for P0.5Py (5b)

Fig. 4. Full plastification surface of Orbison.

1494 S.-E. Kim et al. / Engineering Structures 23 (2001) 14911502

2.3. Parabolic function for gradual yielding due to

flexure

The tangent modulus model is suitable for the member

subjected to axial force, but not adequate for cases of
both axial force and bending moment. A gradual stiff-
ness degradation model for a plastic hinge is required to
represent the partial plastification effects associated with
bending. We shall introduce the softening plastic hinge
model to represent the transition from elastic to zero
stiffness associated with a developing hinge. When
softening plastic hinges are active at both ends of an
element, the slope-deflection equation may be
expressed as

EtA
0 0 0 0 0
P L d
MyA 0 kiiy kijy 0 0 0 qyA
MyB 0 kijy kjjy 0 0 0 qyB
(6)
MzA 0 0 0 kiiz kijz 0 qzA
MzB 0 0 0 kijz kjjz 0 qzB
T GJ f
0 0 0 0 0
L
where

S 22
kiiyhA S1 (1hB)
S1
EtIy
L (7a)

EtIy
kijyhAhBS2 (7b)
L

S 22
kjjyhB S1 (1hA)
S1
EtIy
L (7c)

S 24
kiizhA S3 (1hB)
S3
EtIz
L (7d)

EtIz
kijzhAhBS4 (7e)
L

S 24
kjjzhB S3 (1hA)
S3
EtIz
L
. (7f)

The terms hA and hB is a scalar parameter that allows

for gradual inelastic stiffness reduction of the element
associated with plastification at end A and B. This term
is equal to 1.0 when the element is elastic, and zero when
a plastic hinge is formed. The parameter h is assumed
to vary according to the parabolic function:
h
 S.-E. Kim et al. / Engineering Structures 23 (2001) 14911502 1495

kiiykjjy+k 2ijy+kijyAszGL {fe} and {de} are the end force and displacement vectors
Cijy (13b) in Eq. (11). [T]612 is a transformation matrix written as
kiiy+kjjy+2kijy+AszGL
1 0 0 0 0 0 1 0 0 0 0 0
kiiykjjyk 2ijy+kjjyAszGL

Cjjy (13c)
kiiy+kjjy+2kijy+AszGL 1 1
0 0 0 1 0 0 0 0 0 0
kiizkjjzk 2ijz+kiizAsyGL L L
Ciiz (13d) 1 1
kiiz+kjjz+2kijz+AsyGL 0 0 0 0 0 0 0 0 1 0
L L
kiizkjjz+k +kijzAsyGL
2
[T]612 (16)
Cijz
ijz

(13e) 1 1
kiiz+kjjz+2kijz+AsyGL 0 0 0 0 1 0 0 0 0 0
L L
kiizkjjzk +kjjzAsyGL
2
cjjz
ijz
(13f) 1 1
kiiz+kjjz+2kijz+AsyGL 0 0 0 0 0 0 0 0 0 1
L L
0 0 0 1 0 0 0 0 0 1 0 0
2.4. Element stiffness matrix
Using the transformation matrix by equilibrium and
The end forces and end displacements used in Eq. (12) kinematic relations, the force-displacement relationship
are shown in Fig. 5(a). The sign convention for the posi- of a frame member may be written as
tive directions of element end forces and end displace- {fn}[Kn]{dL}. (17)
ments of a frame member is shown in Fig. 5(b). By com-
paring the two figures, we can express the equilibrium [Kn] is the element stiffness matrix expressed as
and kinematic relationships in symbolic form as [Kn]1212[T]T612[Ke]66[T]612. (18)
T
{fn}[T] 612 {fe} (14a)
{de}[T]612{dL} (14b) Eq. (18) can be subgrouped as

{fn} and {dL} are the end force and displacement vec-
[Kn]1212 [Kn]1 [Kn]2
[Kn]T2 [Kn]3
(19)
tors of a frame member expressed as where
T
{fn} (15a)

a 0 0 0 0 0
{rn1 rn2 rn3 rn4 rn5 rn6 rn7 rn8 rn9 rn10 rn11 rn12} 0 b 0 0 0 c
{dL} {d1 d2 d3 d4 d5 d6 d7 d8 d9 d10 d11 d12}
T
(15b) 0 0 d 0 e 0
[Kn]1 (20a)
0 0 0 f 0 0
0 0 e 0 g 0
0 c 0 0 0 h

a 0 0 0 0 0
0 b 0 0 0 c
0 0 d 0 e 0
[Kn]2 (20b)
0 0 0 f 0 0
0 0 e 0 i 0
0 c 0 0 0 j

a 0 0 0 0 0
0 b 0 0 0 c
0 0 d 0 e 0
[Kn]3 (20c)
0 0 0 f 0 0
0 0 e 0 m 0
Fig. 5. Element end forces and displacements notation. 0 c 0 0 0 n
1496 S.-E. Kim et al. / Engineering Structures 23 (2001) 14911502

where 2.5. Geometric imperfection modeling

EtA Ciiz2CijzCjjz CiizCijz
a b c 2.5.1. Braced frame
L L2 L
The proposed analysis implicitly accounts for the
Ciiy2CijyCjjy CiiyCijy GJ effects of both residual stresses and spread of yielded
d e f gCiiy (21)
L2 L L zones. To this end, proposed analysis may be regarded
hCiiz iCijy jCijz mCjjy nCjjz as equivalent to the plastic-zone analysis. As a result,
geometric imperfections are necessary only to consider
fabrication error. For braced frames, member out-of-
Eq. (19) is used to enforce no sidesway in the mem- straightness, rather than frame out-of-plumbness, needs
ber. If the member is permitted to sway, an additional to be used for geometric imperfections. This is because
axial and shear forces will be induced in the member. the P- effect due to the frame out-of-plumbness is
We can relate this additional axial and shear forces due diminished by braces. The ECCS [18,19], AS [20], and
to a member sway to the member end displacements as CSA [21,22] Specifications recommend an initial crook-
edness of column equal to 1/1000 times the column
{fs}[Ks]{dL}. (22) length. The AISC Code recommends the same maximum
where {fs}, {dL}, and [Ks] are end force vector, end dis- fabrication tolerance of Lc/1000 for member out-of-
placement vector, and the element stiffness matrix. They straightness. In this study, a geometric imperfection of
may be written as Lc/1000 is adopted.
The ECCS [18,19], AS [20], and CSA [21,22] Speci-
{fs}T (23a) fications recommend the out-of-straightness varying
parabolically with a maximum in-plane deflection at the
{rs1 rs2 rs3 rs4 rs5 rs6 rs7 rs8 rs9 rs10 rs11 rs12} mid-height. They do not, however, describe how the
{dL}T{d1 d2 d3 d4 d5 d6 d7 d8 d9 d10 d11 d12} (23b) parabolic imperfection should be modeled in analysis.
Ideally, many elements are needed to model the para-
[Ks]1212
[Ks]
[Ks] [Ks]
T
[Ks]
(23c)
bolic out-of-straightness of a beam-column member, but
it is not practical. In this study, two elements with a
maximum initial deflection at the mid-height of a mem-
where ber are found adequate for capturing the imperfection.
Fig. 6 shows the out-of-straightness modeling for a

0 a b 0 0 0 braced beam-column member. It may be observed that
a c 0 0 0 0 the out-of-plumbness is equal to 1/500 when the half
segment of the member is considered. This value is
b 0 c 0 0 0 identical to that of sway frames as discussed in recent
[Ks] (24)
0 0 0 0 0 0 papers by Kim and Chen [2325]. Thus, it may be stated
0 0 0 0 0 0 that the imperfection values are essentially identical for
both sway and braced frames.
0 0 0 0 0 0

and
MzA+MzB MyA+MyB P
a , b , c . (25)
L2 L2 L

By combining Eqs. (17) and (22), we obtain the gen-

eral beam-column element force-displacement relation-
ship as
{fL}[K]local{dL} (26)
where
{fL}{fn}{fs} (27)
[K]local[Kn][Ks]. (28)
Fig. 6. Explicit imperfection modeling of braced member.
 S.-E. Kim et al. / Engineering Structures 23 (2001) 14911502 1497

2.5.2. Unbraced frame regardless of the value of the column parameter, so that
Referring to the European Convention for Construc- the same ultimate strength can be predicted for math-
tional Steelwork [18,19], an out-of-plumbness of a col- ematically identical braced and unbraced members. This
umn equal to 1/200 times the column height is rec- simplification enables us to use the proposed methods
ommended for the elastic plastic-hinge analysis. For easily with consistent imperfection modeling. The Can-
multi-story and multi-bay frames, the geometric imper- adian Standard [21,22] and the AISC Code of Standard
fections may be reduced to 1.200kcks since all columns Practice [14] set the limit of erection out-of-plumbness
in buildings may not lean in the same direction. The Lc/500. The maximum erection tolerances in the AISC
coefficients, kc and ks account for the situation where a are limited to 1 toward the exterior of buildings and
large number of columns in a story and stories in a frame 2 toward the interior of buildings less than 20 stories.
would reduce the total magnitude of geometric imperfec- Considering the maximum permitted average lean of 1.5
tions. According to the ECCS [18,19], the member initial in the same direction of a story, the geometric imperfec-
out-of-straightness should be modeled at the same time tion of Lc/500 can be used for buildings up to 6-stories
with the initial out-of-plumbness if the column para- with each story approximately 10 feet high. For taller
meter LPu/EI is larger than 1.6. This may be necessary buildings, this imperfection value of Lc/500 is conserva-
to consider residual stresses and possible member insta- tive since the accumulated geometric imperfection calcu-
bility effects for highly compressed slender columns, lated by 1/500 times building height is greater than the
however, the magnitude of the imperfection is not speci- maximum permitted erection tolerance.
fied in the ECCS [18,19]. In this study, we shall use Lc/500 for the out-of-
Since plastic-zone analysis accounts for both residual plumbness without any modification because the system
stresses and the spread of yielding, only geometric strength is often governed by a weak story which has
imperfections for erection tolerances need be included an out-of-plumbness equal to Lc/500 [26] and a constant
in the analysis. The ECCS recommends the out-of- imperfection has the benefit of simplicity in practical
plumbness of columns equal to 1/300r1r2 times the col- design. The explicit geometric imperfection modeling
umn height shown in Fig. 7 where r1 and r2 are factors for an unbraced frame is illustrated in Fig. 8.
which account for the length and number of columns,
respectively. For the plastic-zone analysis, the ECCS 2.6. Numerical implementation
does not specify the requirement of the initial out-of-
straightness to be modeled in addition to the out-of- Both the simple incremental and the incremental-iter-
plumbness when the column parameter LPu/EI is larger ation method are available in the analysis. In the simple
than 1.6, since the plastic-zone analysis already includes incremental method, the applied load increment is auto-
residual stresses and spread of yielding in its formu- matically reduced to minimize the error when the change
lation. in the element stiffness parameter (h) exceeds a
Since proposed analysis implicitly accounts for both defined tolerance. In the incremental-iteration load
residual stresses and the spread of yielding, it may be approach, the structure is assumed to behavior linearly
considered equivalent to the plastic-zone analysis. Thus, at a particular cycle of calculation. Because of the lin-
modeling the out-of-plumbness for erection tolerances is earization process, equilibrium may be violated and the
used here without the out-of-straightness for the column, external force may not always balance the internal force.
This unbalance force must be reapplied to the structure.

Fig. 7. Geometric imperfection for plastic-zone analysis rec-

ommended by ECCS. Fig. 8. Explicit imperfection modeling of unbraced frame.
1498 S.-E. Kim et al. / Engineering Structures 23 (2001) 14911502

Table 2
Result of analysis ignoring shear deformation

Method Proposed Orbisons

Plastic strength surface LRFD Orbison Orbison

Ultimate load factor 1.997 2.066 2.059
Displacement at A in Y- 199 mm 208 mm 247 mm
direction

Fig. 9. Space frame of six-story.

Then, the solution is obtained by iteration process until

equilibrium is satisfied. As the stability limit point is
approached in the analysis, convergence of the solution
may be slow. To facilitate convergence, the applied load
increment is automatically reduced.

3. Verifications Fig. 10. Comparison of load-displacement of six-story space frame.

Fig. 9 shows Orbisons six-story space frame [6,13].

The yield strength of all members is 250 MPa (36 ksi) 4. Design principles
and Youngs modulus is 206 850 MPa (30 000 ksi). Uni-
form floor pressure of 4.8 KN/m2 (100 psf) is converted 4.1. Design format
into equivalent concentrated loads on the top of the col-
umns. Wind loads are simulated by point loads of 26.7 Advanced analysis follows the format of Load and
KN (6 kips) in the Y-direction at every beam-column Resistance Factor Design. In AISC-LRFD [14], the fac-
joints. tored load effect does not exceed the factored nominal
The load-displacement results calculated by the pro- resistance of structure. Two kinds of factors are used:
posed analysis compare well with those of Liew and one is applied to loads, the other to resistances. The load
Tangs (considering shear deformations) and Orbisons and resistance factor design has the format
(ignoring shear deformations) results (Tables 1 and 2 and
Fig. 10). The ultimate load factors calculated from the h giQifRn (29)
proposed analysis are 2.057 and 2.066. These values are where: Rn is the nominal resistance of the structural
nearly equivalent to 2.062 and 2.059 calculated by Liew member; Qi is the force effect; f is the resistance factor;
and Orbison, respectively. gi is the load factor corresponding to Qi, h is a factor
relating to ductility, redundancy, and operational impor-
Table 1
Result of analysis considering shear deformation tance.
The main difference between current LRFD method
Method Proposed Liews and advanced analysis method is that the right side of
Eq. (29), (fRn) in the LRFD method is the resistance or
Plastic strength surface LRFD Orbison Orbison strength of the component of a structural system, but in
Ultimate load factor 1.990 2.057 2.062
Displacement at A in Y- 208 mm 219 mm 250 mm the advanced analysis method, it represents the resist-
direction ance or the load-carrying capacity of the whole structural
system. In the advanced analysis method, the load-carry-
 S.-E. Kim et al. / Engineering Structures 23 (2001) 14911502 1499

ing capacity is obtained from applying incremental loads As a result, the method under-predicts the strength of
until a structural system reaches its strength limit state frames subjected to sequential loads, large gravity loads

such as yielding or buckling. The left-hand side of Eq. first and then lateral loads. It is, however, justified for
(29), (h giQi) represents the member forces in the the practical design since the development of the LRFD
interaction equations was also based on strength curves
LRFD method, but the applied load on the structural sys- subjected to simultaneous loading [17] and the current
tem in the advanced analysis method. LRFD elastic analysis uses the proportional loading
rather than the sequential loading.
4.2. Modeling consideration (2) Incremental loading. It is necessary, in an
advanced analysis, to input each increment load (not the
4.2.1. Sections total loads) to trace nonlinear load-displacement
The AISC-LRFD Specification uses only one column behavior. The incremental loading process can be achi-
curve for rolled and welded sections of W, WT, and HP eved by scaling down the combined factored loads by a
shapes, pipe, and structural tubing. The Specification number between 20 and 50. For a highly redundant
also uses same interaction equations for doubly and structure, dividing by about 20 is recommended and for
singly symmetric members including W, WT, and HP a nearly statically determinate structure, the incremental
shapes, pipe and structural tubing, even though the inter- load may be factored down by 50. One may choose a
action equations were developed on the basis of W number between 20 and 50 to reflect the redundancy of
shapes by Kanchanalai [27]. a particular structure. Since a highly redundant structure
The proposed analysis was developed by calibration has the potential to form many plastic hinges and the
with the LRFD column curve. To this end, it is con- applied load (i.e. the smaller scaling number) may be
cluded that the proposed methods can be used for various used.
rolled and welded sections including W, WT, and HP
shapes, pipe, and structural tubing without further modi- 4.3. Design consideration
fications.
4.3.1. Load-carrying capacity
4.2.2. Structural members The elastic analysis method does not capture the
An important consideration in making this advanced inelastic redistribution of internal forces throughout a
analysis practical is the required number of elements for structural system, since the first-order forces, even with
a member in order to predict realistically the behavior the B1 and B2 factors, account for the second-order geo-
of frames. A sensitivity study of advanced analysis for metric effect but not the inelastic redistributions of
two-dimensional frames was performed on the required internal forces. The method may provide a conservative
number of elements [15]. The two-element model estimation of the ultimate load-carrying capacity.
adequately predicted the strength of a two-dimensional Advanced analysis, however, directly considers force
member. This rule may be used for modeling a three- redistribution due to material yielding and thus allows
dimensional member. smaller member sizes to be selected. This is particularly
beneficial in highly indeterminate steel frames. Because
4.2.3. Geometric imperfection consideration at force redistribution may not always be
The magnitudes of geometric imperfections are selec- desirable, the two approaches (including and excluding
ted as y=2/1000 for unbraced frames and y=1/1000 for inelastic force redistribution) can be used. First, the load-
braced frames. To model a parabolic out-of-straightness carrying capacity, including the effect of inelastic force
in the member, two-element model with maximum initial redistribution, is obtained from the final loading step
deflection at the mid-height of a member adequately cap- (limit state) given by the computer program. Secondly,
tures imperfection effects. It is concluded that practical the load-carrying capacity without the inelastic force
advanced analysis is computationally efficient. The pat- redistribution is obtained by extracting that force sus-
tern of geometric imperfections is assumed to be the tained when the first member yield or buckled. Gener-
same as the elastic first order deflected shape. Based on ally, advanced analysis predicts the same member size
the authors experiences, the buckled and first order as the LRFD method when force redistribution is not
deflected shape are usually same. If they are different, considered.
the pattern leading the lower load carrying capacity of
the structural system should be used. 4.3.2. Resistance factor
AISC-LRFD specifies the resistance factors of 0.85
4.2.4. Load and 0.9 for axial and flexural strength of a member,
(1) Proportional loading. In the proposed advanced respectively. The proposed method uses a system-level
analysis, the gravity and lateral loads should be applied resistance which is different from AISC-LRFD specifi-
simultaneously, since it does not account for unloading. cation using member level resistance factors. When a
1500 S.-E. Kim et al. / Engineering Structures 23 (2001) 14911502

structural system collapses by forming plastic mech- eved when members are adequately braced and their
anism, the resistance factor of 0.9 is used since the dom- cross-sections are compact. The limits for lateral
inent behavior is flexure. When a structural system col- unbraced lengths and compact sections are explicitly
lapses by member buckling, the resistance factor of 0.85 defined in AISC-LRFD [14].
is used since the dominent behavior is compression.

4.3.3. Serviceability limit 5. Design example

According to the ASCE Ad Hoc Committee on Ser-
viceability report [28], the normally accepted range of Fig. 11 shows a twenty-two story office building
overall drift limits for building is 1/750 to 1/250 times frame [11]. Its plane and elevation are shown in Fig. 12
the building height, H, with a typical value of H/400. and member sizes are shown in Fig. 13. The yield stress
The general limits on the interstory drift are 1/500 to used was 250 MPa (36 ksi) and Youngs modulus was
1/200 times the story height. Based on the studies by the 206 850 MPa (30 000 ksi). Factored uniform floor press-
Ad Hoc Committee [28], and by Ellingwood [29], the ure of 5.5 KN/m2 (115 psf) was converted into equival-
deflection limits for girder and story are selected as ent concentrated loads on the top of columns. Factored
wind loads were simulated by point loads of 34.7 KN
Floor girder live load deflection: H/360
Roof girder deflection: H/240
Lateral drift: H/400 for wind load
Interstory drift: H/300 for wind load

At service load levels, no plastic hinges are allowed to

occur in order to avoid permanent deformations under
service loads.

4.3.4. Ductility requirement

Adequate rotation capacity is required for members to
develop their full plastic moment capacity. This is achi-

Fig. 11. Twenty-two story frame. Fig. 12. Structural description of twenty-two story frame.
 S.-E. Kim et al. / Engineering Structures 23 (2001) 14911502 1501

Fig. 14. Load-displacement of twenty-two story frame.

6. Conclusions

A new design method of space frames using advanced

analysis is first summarized, and some concluding
remarks are then made.

1. The proposed method can practically account for all

key factors influencing behavior of a space frame:
gradual yielding associated with flexure; residual
stresses; geometric nonlinearity; and geometric
imperfections.
2. The proposed analysis is adequate in assessing the
strengths when compared with the other approaches.
3. The proposed method overcomes the difficulties due
to incompatibility between the elastic global analysis
and the limit state member design in the conventional
LRFD method.
4. The proposed method does not require tedious separ-
Fig. 13. Member sizes for twenty-two story frame.
ate member capacity checks, including the calcu-
lations of K-factor, and thus it is time-effective in
design of space frames.
(7.8 kips) in the Y-direction at exterior beam-column 5. The proposed method can account for an inelastic
joints. The out of plumbness was assumed to be moment redistribution and thus may allow some
y=2/1000. reduction of steel weight, especially for highly inde-
The load-displacement curve of proposed analysis at terminate space frames
the roof of the frame is shown in Fig. 14. The frame
encountered ultimate state when the applied load ratio Since the proposed method strikes a balance between the
reached 1.36. requirement for realistic representation of actual
The system resistance factor of 0.9 was used since behavior of a space frame and the requirement for sim-
the frame collapsed by forming plastic mechanism. The plicity in use, it is recommended for general use.
ultimate load ratio l resulted in 1.22 (=1.360.9), which
was greater than 1.0 and the member sizes of the system
were adequate. Acknowledgements
The overall drift of the first-order analysis is calcu-
lated as H/416 for the wind loads of 1.0W. This does The work presented in this paper was supported by
not meet the drift limit of H/400. funds from the National Research Laboratory Program
1502 S.-E. Kim et al. / Engineering Structures 23 (2001) 14911502

(2000-N-NL-01-C-162) and the Ministry of Science and [13] Liew JYR, Tang LK. Nonlinear refined plastic hinge analysis of
Technology in Korea. Authors wish to appreciate the space frame structures, Research Report No. CE027/98, Depart-
ment of Civil Engineering, National University of Singapore, Sin-
financial support. gapore, 1998.
[14] AISC. Load and resistance factor design specification. 2nd ed.
Chicago: AISC, 1994.
[15] Chen WF, Lui EM. Stability design of steel frames. Boca Raton,
References Florida: CRC Press, 1992.
[16] Kim SE, Chen WF. Further studies of practical advanced analysis
[1] Chen WF, Kim SE. LRFD steel design using advanced analysis. for weak-axis bending. Eng Structures 1997;19(6):40716.
Boca Raton, Florida: CRC Press, 1997. [17] Chen WF, Lui EM. Structural stabilitytheory and implemen-
[2] Clarke MJ, Bridge RQ, Hancock GJ, Trahair NS. Benchmarking tation. New York: Elsevier Science Publication Co, 1986.
and verification of second-order elastic and inelastic frame analy- [18] ECCS. Ultimate limit state calculation of sway frames with rigid
sis programs. In: White DW, Chen WF, editors. SSRC TG 29 joints, Technical Committee 8Structural stability technical
Workshop and Nomograph on Plastic Hinge Based Methods for working group 8.2System publication No. 33, 1984:20.
Advanced Analysis and Design of Steel Frames. Bethlehem, PA: [19] ECCS. Essentials of Eurocode 3 design manual for steel struc-
SSRC, Lehigh University, 1992. tures in building, ECCS-Advisory Committee 5, No. 65, 1991:60.
[3] El-Zanaty M, Murray D, Bjorhovde R. Inelastic behavior of [20] Standards Australia. AS4100-1990, Steel structures, Sydney,
multistory steel frames. In: Structural Engineering Report No. 83. Australia, 1990.
Alberta, Canada: University of Alberta, 1980. [21] CSA. Limit states design of steel structures, CAN/CAS-S16.1-
[4] Vogel U. Calibrating frames. Stahlbau 1985;10:17. M89, Canadian Standards Association, 1989.
[5] White DW. Material and geometric nonlinear analysis of local [22] CSA. Limit states design of steel structures, CAN/CAS-S16.1-
planar behavior in steel frames using iterative computer graphics, M94, Canadian Standards Association, 1994.
M.S. Thesis, Cornell University, Ithaca, NY, 1985:281. [23] Kim SE, Chen WF. Practical advanced analysis for braced steel
[6] Orbison JG. Nonlinear static analysis of three-dimensional steel frame design. ASCE J Struct Eng 1996;122(11):126674.
frames. In: Report No. 82-6. Ithaca, New York: Department of [24] Kim SE, Chen WF. Practical advanced analysis for unbraced steel
Structural Engineering, Cornell University, 1982. frame design. ASCE J Struct Eng 1996;122(11):125965.
[7] Yang YB, Kuo SR. Theory and analysis of nonlinear framed [25] Kim SE, Chen WF. Practical advanced analysis for steel frame
structures. Singapore: Practic Hall, 1994. design, ASCE Structural Congress XIV, April 1996, Chicago,
[8] Oran C. Tangent stiffness in space frames. ASCE, J Struct Div Special Proceeding Volume on Analysis and Computation,
1973;99(ST6):9871001. 1996:1930.
[9] Kassimali A, Abbasnia R. Large deformation analysis of elastic [26] Maleck AE, White DW, Chen WF. Practical application of
space frames. ASCE, J Struct Eng 1991;117(7):206987. advanced analysis in steel design, Proceedings of the 4th Pacific
[10] Chan SL, Zhou ZH. Second-order elastic analysis of frames using Structural Steel Conference, vol. 1, Steel Structure, 1995:11926.
single imperfection element per member. ASCE, J Struct Eng [27] Kanchanalai T. The design and behavior of beam-columns in
1995;121(6):93945. unbraced steel frames, AISI Project No. 189, Report No. 2, Civil
[11] Ziemian RD, McGuire W, Dierlein GG. Inelastic limit states Engineering/structures Research Lab., University of Texas, Aus-
design part II: three-dimensional frame study. ASCE J Struct Eng tin, Texas, 1977:300.
1992;118(9):255068. [28] Ad Hoc Committee on Serviceability. Structural serviceability: a
[12] Prakash V, Powell GH. DRAIN-3DX: Base program user guide, critical appraisal and research needs. ASCE J Struct Eng
version 1.10, A Computer Program Distributed by 1986;112(12):264664.
NISEE/Computer Applications, Department of Civil Engineering, [29] Ellingwood B. Serviceability guidelines for steel structures, AISC
University of California, Berkeley, 1993. Engineering Journal 1989;26 1st Quarter:18.