Anda di halaman 1dari 7

Stiffness optimization technique

for 3D tall steel building


frameworks under multiple lateral
loadings*
C.-M. Chan
Department of Civil and Structural Engineering, The Hong Kong University of Science and
Technology, Kowloon, Hong Kong

A. N. S h e r b o u r n e a n d D. E. G r i e r s o n
Department of Civil Engineering, University of Waterloo, Ontario, Canada N2L 3G1
(Received October 1992; revised version accepted September 1993)

Although today's engineering computer technology allows for precise analysis of the structural response of a building, it does not
readily provide insight for economical design. Due to the complex
nature of a modern tall building consisting of thousands of structural members, the traditional design method is generally highly
iterative and time consuming. This paper describes an efficient
computer-based technique for least-weight design of three-dimensional (3D) tall steel building frameworks under multiple lateral
loading conditions. Stiffness constraints in terms of interstorey
drifts are considered and optimum discrete member sizes are automatically selected from databases of commercial standard steel
sections. The technique is remarkably efficient and the optimum
design generally converges in a few cycles. The designs of two 3D
lateral-load resisting building frameworks are presented as illustrations. The effectiveness and suitability of the technique for the
design of large-scale tall steel building frameworks are discussed.

Keywords: stiffness, drift, optimization, tall steel building frameworks

In the design of a tall slender building, the response due to


the action of lateral loadings is of prime importance in sizing the structural members. In most cases, the design of
these members is governed by lateral drift criteria rather
than by strength criteria. In particular, as the slenderness
ratio of a high-rise building increases, a significant amount
of material may need to be added to the lateral-load
resisting framework in order to maintain the lateral drift
within acceptable limits. Since modern tall buildings
involve complex geometry and consist of thousands of
members, adding efficient and economical stiffness to such
structures to satisfy the lateral drift criteria is usually a difficult task. Though today's engineering computer software

* This paper was presented in part at the IASS-CSCE International Congress 1992 on Innovative Large Span Structures, 13-17 July 1992,
Toronto, Canada
0141-0296/94/080570-07
1994 Butterworth-Heinemann Ltd

571) Engng Struct. 1994, Volume 16, Number 8

allows for precise analysis of the structural response of a


tall building, it does not readily provide insight for minimum cost or weight design. In the absence of an efficient
optimization tool, the design of a tall building is usually
carried out by engineers based on intuition and educated
trial and error. In fact, the traditional design method for a
tall building is generally highly iterative and time consuming.
Unlike member strength constraints which are local-level
constraints, lateral drift constraints are system-level constraints which require simultaneous consideration of the sizing of all members. With the complexity and scale of practical tall building frameworks, the lateral drift design under
multiple lateral loading conditions is generally very difficult and thus is the main focus of this paper. Although
member strength requirements safeguarding against yielding and instability of members are not considered herein
because they do not usually control the design, they can

Stiffness optimization o f tall steel frameworks: C.-M. Chan et al.


easily be accounted for by establishing strength-based sizes
as the minimum sizes for the stiffness-based design process.
In recent years, resizing techniques based on the concept
of uniform strain energy density have been developed for
the design of lateral-load resisting steel building frameworks subject to one displacement constraint '-3. However,
to design a complex large-scale building structure to satisfy
deflection at only one point is certainly not sufficient. Earlier work by the authors developed an efficient resizing technique for the design of tall steel buildings subject to multiple drift constraints and accounting for discrete standard
sections 4~. The present paper extends the previously cited
work to the discrete optimal design of 3D frameworks
under multiple lateral loading conditions. While multiple
interstorey drift constraints for a building framework are
simultaneously satisfied for different lateral-load cases, discrete standard sections are automatically selected from
databases of standard steel sections to result in a minimum
weight design.
The optimization technique is based on an optimality criteria method (OC). The explicit stiffness optimization problem is first formulated and then the details of the OC
method are developed. The merits of the automatic resizing
technique are discussed through two practical 3D design
examples. The first example is a 12-storey braced rigid
frame used to illustrate the effectiveness of the technique.
The second example is a 60-storey tube building with 6720
structural members used to illustrate the practical applicability of the design method. For both design examples, the
floor diaphragms are assumed to be rigid in their own
planes with negligible flexural rigidity. Such an idealization
is commonly used to facilitate the analysis of large-scale
3D building structures so as to reduce the computational
effort. By exploiting the fact that member force distribution
is relatively insensitive to changes in member sizes for typical building frameworks, the design optimization technique
generally converges rapidly in a few cycles.
Design problem formulation
Once the topology and support conditions of the lateralload resisting framework of a tall building is established,
the major design effort is then to economically size the
individual members such that the stiffness requirements are
satisfied. Consider a slender steel framework loaded with
l = 1,2 ..... NC loading cases and having i = 1,2 ..... n members (or member fabrication groups) and j = 1,2..... m storeys. The problem to find the minimum weight design of
the framework under lateral loads can be generally stated as

ditions, where 8it and 6j_1~are the lateral deflections at two


adjacent storey levels, h~ is the corresponding storey height
and dff is the allowable interstorey drift. Equation (lc)
requires each cross-sectional area a i to belong to the set of
areas Ai = {a,, a2. . . . }~prevailing for the standard steel section profile specified for member i.
In order to solve the design optimization problem posed
by equation (1), the implicit drift constraints have to be
explicitly expressed. Employing the principle of virtual
work the lateral deflection at the jth floor level of a general
3D building framework under the /th lateral loading condition can be expressed as

\ Ea

i=1

+~+G~z + G~

M~,m~ M~,m~Elz }i
where: L i
length of member i, E, G are the axial and
shear elastic material moduli, respectively; a, At, Az are the
axial and shear areas for the cross-section; Ix, lr, Iz are
the torsional and flexural moments of inertia for the crosssection; Fxt, Fvt, Fzt are the actual, axial and shear forces,
fx, fY, fz are the virtual, axial and shear forces; Mxt, Mvt, Mz~
are the actual, torsional and flexural moments; mx, my, mz
are the virtual, torsional and flexural moments, respectively. As commercial-standard steel sections are used for
the building framework, section properties such as At, Az,
Ix, It, Iz can be related to the cross-sectional area a by
certain functions through regressional analysis. In References 4 and 7, linear relationships between the reciprocal
section properties are proposed. For example, the strong
moment of inertia Iz and area a for a member can be
expressed as
=

1 C
- C"
Iz a

(3)

where C and C' are regressional constants derived under the


assumption that the cross-section remains within a constant
shape group as it changes size.
Adopting linear regression relationships (equation (3))
between reciprocal section properties for standard steel sections 4,7, the interstorey drift constraints can be formulated
as explicit functions of the sizing variables ai alone as
l(ai) =

+ col
i=l \ ai

(j=l,2 ..... m); (I=1,2 ..... NC)

Minimize
n

W(ai) = ~ w i ai

(la)

i=1

(2)

(4)

where c0~ is a, so-called, virtual strain energy coefficient,


i
and cijt is a correction coefficient which equals zero for
truss structures and is nonzero for flexural structures.

subject to

d/t(a,) = (6j, - 6j_,+) / h/<- dff


ai ~ Ai

Design o p t i m i z a t i o n t e c h n i q u e

(j=l,2 ..... m); (l=1,2 ..... NC)

(lb)

(i=1,2 ..... n)

(lc)

Equation (la) defines the weight for the structure, where


a~ is the axial cross-section area for member i and wi is the
corresponding weight coefficient per unit cross-section
area. Equation (lb) defines the multiple m interstorey drift
constraints for the structure for NC lateral loading con-

With the establishment of the explicit drift constraints


(equation (4)), one can use the optimality criteria method 8.9
to derive first the necessary stationary optimality criterion
at the optimum and then use the criterion to obtain a recursive relation for the sizing variables, ai. Temporarily ignoring the discrete sizing constraints (equation (lc)), the stiffness optimization problem can be reformulated as the
minimization of the Lagrangian function

Engng Struct. 1994, Volume 16, Number 8

571

Stiffness optimization o f tall steel frameworks: C.-M. Chan et al.


wia, + ~

L(a,A,) =

c' + c~j, -

Ajl

/=1

i=l

j=l

i=l

d/:

\ ai

(5)
where Ajt is the Lagrange multiplier for the jth drift constraints under the lth lateral loading condition.
Assuming temporarily that all drift constraints are active,
differentiate equation (5) with respect to the sizing variables ai and rearrange terms to obtain the necessary conditions at the optimum as

/=1

j=l

t~jl E
i=l

CiklCij'
\ wiai 3 / v

i=l

Cik~l--T] (dkU-- dk:" )


air

(k=l,2 ..... m); (/=1,2 ..... NC)

(11)

Equations (7) and (1 1) together form an iterative solution


strategy for the continuous design optimization problem
posed by equation (1). The optimum continuous section
sizes are obtained by successively solving the linear simultaneous equations (equation (11)) for the Aj: and the recursive relations (equation (7)) for the ai until convergence
occurs.

NC m
l=l

j=l

w ia i2

= 1

(6)

To ensure that the optimum design satisfying the optimality


conditions (equation (5)) corresponds to a minimum (rather
than a maximum) weight structure, the Lagrange multipliers must be non-negative such that Aj: > 0 if the constraint jl is active (i.e. it reaches its drift limit), or Aj~ -- 0
if the constraint jl is inactive (i.e., less than its drift limit).
Equation (6) represents stationary conditions, equal to unity
at the optimum, that can be used to derive a linear recursive
relation for the active continuous sizing variables ai, as follows 5 - 7

ai ~-l = ai v

1+

,'

j=l

,t]w.

(i=1,2 ..... n - ~

(7)

where 77 is a step-size parameter that controls the convergence of the recursive process, v+l and u indicate successive iterations, and ~ is the number of (inactive) members
which have been previously assigned fixed discrete sections.
In order to apply equation (4) to find the new sizing
variables a7 ~the current values of the Lagrange multipliers
Ajt v m u s t first be determined. To this end, consider the
change ( d k ~ - d u v) in the klth drift constraint due to
changes ( a y L a ~ ~) in the ( n - ~ active sizing variables, i.e.
-'

dS +' - dS = ~ \ Oai/v

(air+l--air)

(8)

i=l

where, from equation (4) for the drift constraint kl,

Odkt) = _ (cik,t
Oai / ,
\ ai e / v

(i:1,2 . . . . n-~)

(9)

and, from equation (7) for the active sizing variables

ai ~+' - aT = q

l=1 j = l

t~li2

,,'

.12 .....
u

(lO)
Suppose now that the drift constraint kl becomes active
after the (v+l)th iteration so that du ~ = dj u, and then substitute from equations (9) and (10) into equation (8) and
rearrange the terms to obtain the following set of (m*NC)
linear simultaneous equations in terms of (m*NC) Lagrange
multipliers 5-7

572

Engng Struct. 1994, V o l u m e 16, N u m b e r 8

Design procedure
Because the explicit drift constraints are only approximately expressed for statically indeterminate structures, the
design process is generally iterative in nature. Indeed, the
design process includes the coordinated application of
structural analysis, design optimization and section selection processes, as follows.

Analysis phase
Prior to solving the design optimization problem, the structure is first analysed with current member sizes under the
NC lateral loading conditions as well as the m virtual pointload conditions to establish the explicit drift constraints
(equation (4)). Although the optimization technique does
not impose any requirement on the initial member sizes,
members are initially taken to have maximum available
section sizes to commence the design process. If the
response of such an initial design indicates a violation of
any one of the drift constraints, then the design process is
terminated immediately since no better result can possibly
be achieved from among the given set of available sections;
otherwise, the next phase of design is invoked to commence
the application of the optimization resizing algorithm.
Design optimization phase
Assuming initially that member forces do not redistribute
for changes in member sizes, a Gauss-Seidel iterative technique is applied to solve the set of simultaneous equations
(equation ( 11)) for Ajt. At any stage of the solution scheme,
any hj~ found to be negative is set to zero value to reflect
the fact that the corresponding drift constraint is presently
inactive. Upon establishing all hit, new member sizes are
then obtained from equation (7). If a new member size is
found to exceed its sizing limits, it is fixed at its limiting
size and is thereafter treated as being inactive. The structure
is re-analysed to allow for member force redistribution and
equations (11) for Aj: and (7) for ai are recursively applied
again. The analysis and design optimization processes are
repeated until the behaviour of the structure is stabilized
and there is no change in structural weight from one design
stage to the next.
Discrete section selection phase
For practical building design, member sections cannot be
finalized with the continuous optimum sizes because custom fabrication is usually very expensive. An effective
pseudodiscrete section selection technique is developed to
achieve a smooth progressive transition from the continuous variable design to the optimum final design using discrete standard steel sections 5 7. Recognizing that the most

Stiffness optimization of tall steel frameworks: C.-M. Chan et al.


economical design solution occurs when the discrete section sizes are exactly the same as the continuous sizes, the
least possible disturbance from the continuous variable
results is most desirable. In order to minimize such deviation, one strategy is to select member(s) involving the least
change in material weight when the continuous size is
increased to its next available discrete size. Once the currently most economical member is identified and then
treated as inactive by being fixed at the assigned discrete
section (i.e., ~-gq-1), other active members may possibly
reduce in size by re-evaluating equation (11) and then equation (7). This section selection process is progressively
applied until each member is assigned a final discrete standard section. Having the optimum discrete results, a last
analysis is conducted to obtain the final response of the
structure.
Previously cited work 5.6 has applied the foregoing
pseudodiscrete section selection strategy immediately after
each continuous optimization phase before conducting the
next analysis. With a view to computational efficiency,
especially for very large-scale structures, this strategy is
presently applied only at the final stage of the continuous
design process. Little or insignificant difference between
the previous and the current method has been found.

Illustrative design examples


Example 1
A 12-storey, 1-bay by 2-bay braced frame is presented to
illustrate the effectiveness of the design optimization technique. The structural geometry, material properties and
loading information for the framework are shown in Figure
1. All beams and columns are rigidly connected while the
diagonals are simply connected. For a bay width of 20 ft
in the X-direction, 30 ft in the Y-direction and a storey
height of 12 ft, the framework has height-to-width aspect
ratios of 7.2 and 2.4, respectively. Two lateral loading cases
are applied as horizontal point-loads at the centroid of each
floor level. Assuming that the floors are rigid in their own
planes, the lateral loads are distributed internally to each

,~o ft=,
2@30
f ~

''~n ~

vl

.. J c e n t r o i d
15 k -D"

,,f 5 k
~10

,~,10

30 k - ~ "
30 k - ~ l ~

oo,,,,.o,9

'II

T/C.I

11

30 k - - I ~

300

Final Discrete Solution


o Continuous Solution

280

ff_260
.E
._D3

~240

2 3 8 . ~

200

231~?.08
228.25

220
'
1

228.15

'
2
Design Cycle

'
3

Figure2 Design history for example 1 framework


individual flame in accordance with its respective lateral
stiffness.
The framework is to be designed such that each interstorey drift ratio is within a typical limit of 1/400. American
AISC ~ standard wide-flange W steel sections are used to
size the members. The columns and diagonals are to be
selected from the range of sections W1422 to W14730,
while the beams are to be chosen from the range of sections
W2455 to W24x492. To satisfy practical construction
requirements, beams of each of the four faces of the flamework are grouped together as having a common section for
every two storeys, as are diagonals, and the exterior comer
columns are grouped together over two adjacent storeys, as
are interior columns.

Results and discussion. To commence the optimization


process, the framework is first analysed with members
taken to have their corresponding maximum available section sizes as the initial trial design. The iterative analysisdesign optimization process converges rapidly after the
third cycle to a continuous section solution having a structure weight of 228.15 kips, as shown in Figure 2. At a cost
of a 1.3% increase in weight for the pseudodiscrete section
selection process, the final design with discrete standard
sections is achieved at a weight of 231.08 kips. The final
discrete member sizes are listed in Table 1. Figure 3 shows
the initial and final deflected shapes of the building in both

,o,ouft

onoMo,o,o,E-2ooook,,
,,.rMoou,usO-,,2oo,,,

30 k - ' - ' ~

2391.95

X-Direction Drift
Final
12 Initial

[ Y-Direction Drift
12 Initial
Final

10

10

Co,u ns,D,.ooa,s:
w

30 k - - - ~

x.wx.o

t/

30 k - - ~ "
30 k " - ' ~
Interstorey drift limit = h/400
30 k - - - I ~

30 k ' ~ "

,.R-

30 k - ' ~ '
30 k " - - ~

I-.Xq
0

0 1 2 3 4 5
Lateral Deflection (in)

Figure 1 Example 1 framework

0 1 2 3 4 5
Lateral Deflection (in)

Figure3 Drift results for example 1 framework

E n g n g Struct. 1994, V o l u m e

16, N u m b e r

573

Stiffness optimization o f tall steel frameworks: C.-M. Chan e t al.


Table I Example 1, final discrete member sizes

11-12
9-10
7-8
5-6
3-4
1-2

Corner
columns

Middle
columns

Beams in
X-direction
braced frame

Diagonals in
X-direction
braced frame

Beams in
Y-direction
rigid frame

W14x22
W14x48
W14x99
W14x132
W14x159
W14x211

W14x22
W14x22
W14x68
W14x109
W14x132
W14x99

W24x55
W24x55
W24x55
W24x55
W24x55
W24x55

W14x22
W14x26
W14x22
W14x22
W14x22
W14x22

W24x55
W24x55
W24x55
W24x62
W24x76
W24x84

the X and Y directions. (Note that the optimum structure


may not necessarily be fully constrained, and that the final
deflected shape corresponding to the least-weight design is
automatically identified through the progressive evaluation
of the Lagrange multipliers associated with the drift
constraints). As the braced flame in the X-direction exhibits
cantilever behaviour, active drift ratios are found near the
top of the frame. For the rigid flame in the Y-direction,
active drift ratios are found near the lower floor levels as
the flame behaves as a shear beam. Such a result seems to
be reasonable and closely matches engineering intuition.
Figure 4 shows member force distribution at a corner
section of the bottom storey of the framework for the initial
design, for the design after the first cycle and for the final
discrete design. Although the structural weight drops dramatically from the initial value of 2391.95kips to
238.53 kips after the first design cycle, relatively small
changes in the dominating column axial forces for the Xdirection load case and the bending moments for Y-direction load case are found for members at the lower right
bottom comer of the framework. Comparing the force distribution of the final design and that after the first design
cycle, a moderate difference is found due to minor changes
in the member sizes. Indeed, such behaviour should not be
surprising because a building is basically a vertical cantilever, in a global sense, and behaves similarly to a statically
determinate structure. Such behaviour for building flameworks generally results in good approximations of the
145

283

25~ 6 6

,97 6

~
41 0

693
1
86 1

28 5
38 9

explicit drift constraints using the principle of virtual work.


As a result, rapid convergence to the optimum design
occurs when using the proposed optimization technique.
The design process for this example was carried out using
a SUN Sparc 1 station and required 65 s of CPU time to
complete.

Example 2
The framed tube system has been one of the most efficient
lateral-load resisting frameworks for super-tall buildings.
Examples include the Sears Tower in Chicago, the world's
tallest building; and the World Trade Tower in New York,
the world's second tallest building. Closely spaced columns
and relatively deep beams around the perimeter of a building allow tubular behaviour to resist lateral Ioadings on
both the windward and leeward faces.
The 60-storey, 14-bay by 14-bay rigidly framed tube
building with 6720 members shown in Figure 5 is now
considered to illustrate the practical applicability of the proposed design optimization technique for large-scale tall
building frameworks. For a bay width of 10 ft and a storey
height of 12 ft, the tube framework has a height-to-width
aspect ratio of 5.14. The framed tube is to be designed such
that the interstorey drift ratio does not exceed 1:400 under
Column

Orientations

Storey h e i g h t = 12 ft

6 7 ~

Bay width = 10 f t

-87 3

Density - 0.49 kip ! cu.ff.


Y o u n g m o d u l u s , E = 29000 ksi
22 0

go 1

Lnltial Des:gr,

82 9

After lsl Cycle

Shear modulus, G = 11200 ksi

Fina~ Design

Columns:
Corner - 2 W14X22 - 2 W14X730
Others - W14x22 - W14X730

1 column / 2 storey
Beams:

~_~40 1
51 ~ I

1049

W24X55 - W24X492

131 4

26 5

1 corner span / storey

1 middle span ! storey

385

Interstorey

156

218
Initial Design

207
After 1st Cycle

Final Design

b
Figure4 Internal force distributions in members at a bottom
corner of example 1 framework. (a), X-direction braced frame;
(b) Y-direction rigid frame

574

E n g n g S t r u c t . 1994, V o l u m e

16, N u m b e r

J
Corner I

I.J
IMiddlel

beams

beams

II(~orner

beams

Figure 5 Example 2 framework

drift

limit

h/400

Stiffness optimization of tall steel frameworks: C-M. Chan e t al.


lateral wind loading conditions in both the X and Y directions (i.e. 20psi over 0-100 ft, 25 psi over 100-300ft,
30 psi over 300-600 ft, and 35 psi over 600-720 ft). Rigid
floor diaphragms are assumed and lateral loads are applied
as horizontal point-loads, as in example 1, at the centroid
of each floor level. Members are to be designed using AISC
standard sections as follows: beams to be W24 shapes; columns to be W14 shapes except that the comer columns use
pairs of two W14 shapes oriented perpendicularly. For ease
of fabrication, the columns on each line are grouped
together as having a common section over two adjacent
storeys, while the beams of the exterior first four spans of
each tube's frame near the comer column are grouped to
have the same section on each floor level, as are the beams
near the centre of each tube's face.

Results and discussion. Due to the symmetry of the


square framed tube, only half of the members of one of the
tube's frames are listed with their final discrete standard
sizes in Tables 2 and 3. The history of the design process
is shown in Figure 6. The iterative design process is terminated after the 7th design cycle and the pseudodiscrete section selection strategy finalizes the design to a structure
weight of 15849.1 kips, resulting in 13.48 lb of steel per ft2
floor area.
Since a tight convergence criterion is used, the iterative
design process takes relatively more cycles to converge. If
a looser convergence criterion is chosen the design may be
stopped after the third cycle and thus the computational
time can be substantially reduced. The drift results for the
framework are shown in Figure 7. With the exception of
the bottom first storey and the storeys near the top, all other
storeys are found to have active or almost active drift ratios.

Table3 Example 2, final discrete beam sizes


Storey

Corner span
beams

Storey

Middle span
beams

53-60
52
51
50
48-49
46-47
45
44
40-43
38-39
33-37
29-32
25-28
21-24
17-20
13-16
9-12
8
7
6
5
4
3
2
1

W2455
W24x62
W24x68
W2476
W24x84
W2494
W24x103
W24x164
W24117
W24131
W24x146
W24162
W24x176
W24192
W24207
W24x229
W24x250
W24279
W24x250
W24x306
W24x279
W24x335
W24x306
W24x335
W24279

54-60
53
52
51
50
49
48
46-47
45
42-44
40-41

W24x55
W24x68
W24x76
W24x84
W24x94
W24x103
W24104
W24x117
W24x131
W24x146
W24x162
W24x176
W24x192
W24x207
W24x229
W24x207
W24x229
W24x207
W24x192
W24x176
W24x146
W24x131
W24x117
W24x84

36-39
33-35
29-32
18-28
17
16
10-15
8-9
6-7
4-5
3
2
1

Although the analysis can be simplified to model only a


quarter of the structure due to symmetry, a full-scale 3D
framework is deliberately considered to verify the feasibility of the proposed design optimization technique for

Table2 Example 2, final discrete column sizes


Storey

Corner
column

1st int.
column

2nd int
column

3rd int.
column

4th int.
column

5th int.
column

6th int.
column

7th int.
column

59-60
57-58
55--56
53--54
51-52
49-50
47-48
45-46
43-44
41-42
39-40
37-38
35-36
33-34
31-32
29-30
27-28
26-26
23-24
21-22
19-20
17-18
15-16
13-14
11-12
9-10
7-8
5-6
3-4
1-2

2W14x22
2W14x22
2W14x34
2W14x38
2W14x43
2W14x53
2W14x61
2W14x61
2W14x68
2W14x82
2W14x82
2W14x90
2W14x99
2w14109
2w14109
2W14x120
2W14x132
2w14145
2W14x159
2w14176
2W14x193
2W14x233
2W14x257
2W14x283
2w14342
2W14x398
2W14x455
2W14x550
2W14x605
2w14730

W14x22
w1438
W14x53
W14x82
W14x90
w14109
W14x132
W14x145
w14159
w14x176
W14x176
W14x193
W14x211
W14x233
W14x233
W14x257
W14x283
W14x283
w14311
W14x342
W14x342
W14x370
w14398
w14426
W14x426
w14x455
W14x500
W14x550
W14x550
W14x550

w1422
W14x38
w1461
W14x82
W14x99
W14x120
W14x132
w14159
W14x176
w14176
W14x193
W14x211
W14x233
w14257
W14x257
w14257
W14x283
w14311
W14x311
W14x311
W14x342
w14370
W14x370
W14x398
W14x398
W14x426
w14426
w14426
w14426
W14x426

W14x22
W14x38
W14x68
w1490
W14x109
w14132
w14145
W14x159
W14x176
w14193
W14x211
w14233
W14x257
w14257
w14257
w14283
w14283
w14311
w14311
W14x342
w14342
W14x342
W14x342
w14342
w14370
w14370
w14370
W14x370
W14x370
W14x342

w1422
w1438
W14x68
W14x99
W14x120
w14145
w14159
W14x176
W14x193
w14211
w14233
W14x257
W14x257
W14x283
W14x283
w14311
w14311
W14x311
w14311
w14342
W14x342
W14x342
W14x342
w14342
w14342
W14x342
W14x311
w14283
w14283
W14x257

W14x22
W14x30
W14x61
W14x99
W14x132
W14145
W14x176
W14x193
W14x211
w14233
W14x257
W14x257
W14257
w14283
w14311
W14x311
W14x311
W14x311
W14x311
W14x311
w14311
W14x342
W14x342
W14x311
W14x311
W14x283
w14257
W14x233
W14x211
W14x159

W14x22
W14x30
w1468
W14x99
W14x132
W14x159
W14x176
W14x193
W14x211
W14x233
W14x257
W14x257
W14x283
w14283
W14x311
W14x311
W14x311
W14x342
W14x342
W14x311
W14x311
W14x311
W14x311
W14x311
W14x283
w14257
W14x257
W14x233
W14x193
w14159

w1422
w1430
w1468
w14109
W14x145
w14159
w14176
w14211
w14233
w14257
w14257
w14283
w14x283
W14x311
W14x311
w14342
w14342
w14342
w14342
w14342
W14x342
W14x311
W14x311
w14311
W14x283
w14257
w14257
w14233
w14193
w14159

E n g n g S t r u c t . 1994, V o l u m e

16, N u m b e r

575

Stiffness optimization of tall steel frameworks: C.-M. Chan et al.


19

Conclusions

48.069
D Final Discrete Solution
o Continuous Solution

18

17

16.199
~ 1 5 . 8 0 9

_E
._~ 16

o 15.767 15~849

15.838

15.787

15.755

15

The application of an effective stiffness optimization technique for the design of tall steel building frameworks subject to multiple drift constraints and standard section fabrication requirements under multiple loading conditions has
been presented. Rapid convergence to the optimum design
of building frameworks is generally achieved in a few
design cycles. Results have shown that the technique is well
suited for tall buildings with large numbers of sizing variables and relatively few drift constraints.
Acknowledgments

14

4
Design Cycle

Figure 6 Design history for example 2 framework

This work was sponsored by the Natural Sciences and


Engineering Research Council of Canada, and is based
upon research conducted by the first author under the supervision of the second and the third authors for the degree of
Doctor of Philosophy in Civil Engineering at the University
of Waterloo, Ontario, Canada.
References
I

Initial Design

60

Final Design

2
50

3
4o

rift Ratio

,-r 3o

20

10

0
0

5
10
15
20
Lateral Deflection (in)

25

Figure 7 Drift results for example 2 framework


9

practical large-scale building frameworks. The optimum


design was completed in 24 h of CPU time on a SUN
SPARC 1 station.

,576

Engng Struct. 1994, Volume 16, Number 8

10

Velivasakis, E. and DeScenza, R. 'Design optimization of lateral load


resisting frameworks', Proc. Eighth Conference on Electric Computation (J. K. Nelson Ed.), ASCE, New York, 1983, pp 130-143
Baker, W. 'Sizing techniques for lateral systems in multi-story steel
buildings', Proc. 4th world congress on tall buildings: 2000 and
beyond, Council on Tall Buildings and Urban Habitat, Hong Kong,
1990, pp 875-868
Gilsanz, R. and Carlson A. 'Optimization in building design', Proc.
2nd Int. Conf. on Computer Aided Optimum Design of Structures 91
(eds S. Hernandez and C. A. Brebbia), Boston, USA, 1991,
pp 481-492
Grierson, D. E. and Chan C.-M. 'Design optimization of tall steel
buildings', Proc. 2nd Int. Conf. on Computer Aided Optimum Design
of Structures 91 (eds S. Hernandez and C. A. Brebbia), Boston, USA,
1991, pp 541-551
Chart, C.-M. and Grierson, D. E., 'An efficient resizing technique for
the design of tall steel building subject to multiple drift constraints',
Int. J. Struct. Design Tall Buildings, 1993, 2, 17-32
Chan, C.-M. 'An optimality criteria algorithm for tall steel building
design using commercial standard sections', Struct. 017l., 1993, 5,
26-29
Chan, C.-M. "An automatic resizing technique for the design of tall
steel building frameworks', PhD thesis, University of Waterloo, Canada, 1993
Khot, N. S., Berke, L. and Venkayya, V. B. 'Comparison of optimality criteria algorithms for minimum weight design of structures',
AIAA J., 1979, 17 (2), 182-190
Berke, L. and Khot, N. S. 'Performance characteristics of optimality
critera methods', Proc. IUTAM Symposium on Structural Optimization (eds G. I. N. Rozvany and B. L. Karihaloo), Melbourne, Australia, 1979, pp 39-46
American Institute of Steel Construction, Manual of steel construction - load and resistance factor design Chicago, Illinois, USA 1986

Anda mungkin juga menyukai