AN OPTIMALITY CRITERION METHOD FOR
STRUCTURAL OPTIMIZATION PROBLEMS
EDWARD A. S.+DEK*
Faculty of Engineering. Cairo University. Giza, Egypt
;ibstract-This work presents the application of an optimality criterion method uhich exploits the
concept of one most critical constraint. The method eliminates the need to calculate a large set of
Lagrange multipliers for the active constraints. and also eliminates the need for a decision as to whether
or not a particular constraint should be considered active. The method can treat multiple load conditions
under both stress and displacement constraints. A study of the effect of using different starting designs
and how to deal with oassive members has been made. Also some suggestions which have been found
to improve the results obtained are presented.
INTRODUCTION
The development of efficient structural analysis and
design methods in the last decade is unprecedented
and gratifying. Much of this progress was made pos-
sible by even more impressive developments in the
field of digital computer technology. Structural op-
timization has become of great interest to many re-
searchers during the past few years. The primary
goal of this recent work has been to obtain a min-
imum weight structure subject to various con-
straints in minimal computational time and with
minimal computer storage.
Structural optimization problems are generally
characterized by (a) a large number of design var-
iables, (b) a simple objective function and, (c) in-
direct but well-behaved constraint functions. The
constraint functions are indirect in the sense that
they cannot be expressed explicity as functions of
the design variables. Attempts to optimize struc-
tures by nonlinear programming methods have met
with varying degree of success. These methods are
extremely useful in defining the design problem in
proper mathematical terms. When the design var-
iables are few (say less than 50), the nonlinear pro-
gramming methods can be used quite effectively for
optimization. However. in the presence of a large
number of variables, these methods are painfully
slow and erratic in obtaining a solution.
Physical optimality criterion techniques to effi-
ciently design large scale structures have been de-
veloped[l-31. This work is a continuation of the
effort in this direction. Recursion formulas for
stress and displacement constraints, which result
from the Kuhn-Tucker necessity conditions for
each type of constraints, are incorporated into a
design algorithm which exploits the concept of a
single, most critical displacement constraint. The
algorithm requires only one analysis ofthe structure
per design cycle. Redesign of each member is
? Associate Professor.
achieved by means of one of two recursion for-
mulas. No set of Lagrange multipliers need be cal-
culated, no subsidiary LP or NLP must be solved,
no decision as to active or potentially active con-
straints must be made. and no move limits need to
be used. The method is applicable to problems of
fixed geometry under multiple load conditions and
stress and displacement constraints.
THEORI
The design problem to be solved here can be
stated as follows: Find the vector of design varia-
bles A = (A,, A?, A3, . A,V) such that the vol-
ume of the structure
V = 2 A,L;-, min.
i=l
(1)
while
Uik S 5; i = 1.2..
. iv. A- = I, 2., . . . K,
lljk S II,, j = I, 2. . . . J.
(2)
where Ai and Li are the cross-sectional area and
length of the ith member, &’ is the number of mem-
bers, uih is the stress in the ith member of the X-th
load condition, K is the number of load conditions,
and Zi is the limiting stress in the ith member. Also
I(,~ is the displacement in the jth constrained degree
of freedom, lj is the limiting value of the displace-
ment in the jth constrained degree of freedom, and
J is the number of displacement constrained de-
grees of freedom.
Stress corlsIrnints
Considering stress constraints alone. the Kuhn-
Tucker conditions for the design problem of eqns
(I) and (2) results in the vvell-known stress ratio
823