You are on page 1of 4

Chapter 4 OPTIMIZATION PROBLEM FORMULATION 4.

1 General In the majority of structural optimization problems, the equality constraints constitute a part of or the whole equilibrium equation and, so, are excluded from the above formulation and hence are treated implicitly.

Solving constrained optimization problems is generally cumbersome. While there are methods such as the feasible directions method [Kirsch, U.,
Optimum Structural Design, McGraw-Hill, New York (1981)., Rao, S.S., Optimization: Theory and Applications, Wiley Eastern Limited (1984).

] that solves the constrained problems directly, the main

strategy behind the penalty methods is to change the form of the above constrained optimization problem to an unconstrained form so that unconstrained optimization techniques can be applied. To this end, a global objective function, , called the penalty function, is defined, which includes both the main objective function and all the constraints. is so defined that, through its optimization by unconstrained optimization techniques, both the constraints are satisfied and also the objective function is minimized.

Penalty function methods are the most popular constraint handling methods among users [Yeniay, O. \Penalty function methods for constrained optimization with genetic algorithms",
Mathematical and Computational Application, 10(1), pp. 45-46(2005).

]. Interior penalty function method has

been proposed in this research work. The statement of optimization

problem and algorithm development is briefly described in the following sections.

4.2 Problem Statement The general form of a minimization problem with inequality and equality constraints is as follows:
Find

(4.1a) (4.1b)

to minimize f(x) subject to: gj(x) 0; and: hk(x) = 0; j = 1, 2, 3,. .. . m j = 1; 2; ___ ; p;

(4.1c) (4.1d)

where x = n-dimensional vector of design variables, f(x) = objective function to be minimized, and g and h are the inequality and equality constraints respectively; hence there are m inequality and p equality constraints.

4.3 Algorithm Development The original form of the interior penalty function is as follows [Kirsch, U., Optimum Structural Design, McGraw-Hill,New York (1981).

]: (4.2a)

(x,r)=f(x)-r [1/gj(x) - 1/hk(x)] r 0

where r reduces from a high value to 0 gradually.

Rao [Rao,

S.S., Optimization: Theory and Applications, Wiley Eastern Limited (1984).]

has

proposed the following function for selecting r at the start of the optimization procedure: r1=(0.1~1.0)[ f(x1)/-1/g(x1)] (4.2b)

where x1 is the initial point in the feasible region. The optimization procedure is an iterative method in which r reduces to 0 gradually. Figure 4.1 summarizes the optimization procedure of the interior penalty method. Here, the reduction follows [Rao,
tions, Wiley Eastern Limited (1984).]: S.S., Optimization: Theory and Applica-

rn+1=c rn where c is a coe_cient less than 1.

(4.2c)

Figure 4.1 Flow chart for interior penalty function

4.4 Application of the Algorithm The algorithm developed for minimum weight of structural members is applied to steel MRFs and BRBFs respectively. The steel frames considered, MRFs and BRBFs are formed by connecting structural members; columns beams with rigid joints. BRBs are connected at joints by bolts. Therefore all the members have subjected to shear forces, bending moments and axial forces. The shear forces in columns, axial forces in beams and bending moments, and shear forces in BRBs are insignificant (not worthy of notice); therefore discarded in optimization process. The application of the method for the columns beams and BRBs are briefly described as follows;

4.4.1 Columns Members

4.4.2 Beam Members 4.4.3.Buckling Restrained Brace

You might also like