ABSTRACT
A good design is certainly optimal in some sense, but it may not always be possible to express the attributes of such a design within the context of a single objective, nonlinear programming problem. The traditional formulation of a nonlinear programming problem requires the specification of a scalar objective function with all other factors being included as constraints. Nonlinear pro gramming methods also force the computer to operate on a mathematical abstraction of the 'real' design problem with no knowledge of the trade-offs which are being made during the optimization process. The human mind can operate easily in an abstract mode and as such can consider a wide range of design alternatives. In contrast, the use of conventional computer-aided design and analysis tools forces the user to deal with design issues at a very specific level. If one considers design as a process which proceeds from the general to the specific, then the computer cannot have the desired level of impact early in the process where the benefits are potentially the greatest. The result is generally a solution which is optimal with respect to the mathematical formulation applied but far from optimal in a practical design sense. This deficiency has been a major factor in limiting the number of practical applications of design optimization which is unfortunate as the potential of the concept is significant.
