In a conventional design approach, the engineer cycles through trial designs until a satisfactory (feasible) design is found. In the unlikely case that there is still some budget and schedule available, the engineer may continue to run parametric studies to improve (optimize) the design. This is a trial-and-error effort requiring intuition and insight. If there are a significant number of design variables, the process is complex and time consuming.
Optimization theory offers some tools to improve the design process, including design sensitivity and nonlinear programming (NLP) techniques. When incorporated into a general purpose FEA program, optimization methods offer new opportunities for design improvement. In current software, the automated procedure will sequentially improve a starting design to obtain the âbestâ design. This âbestâ design is limited to the class of structures defined by the starting design and the choice of design variables. Because of the sequential nature of NLP, the âbestâ design may be a local minimum rather than a global minimum. Even with these caveats, optimization is a powerful tool in the hands of a knowledgeable user.
This chapter includes a brief overview of optimization theory; however, the main emphasis will be on the application of the tools to optomechanical systems. In a typical optical structure, many response quantities are calculated. These responses may have performance limits (constraints) applied, or may be optimized (objective).
Online access to SPIE eBooks is limited to subscribing institutions.