Extensibility of a linear rapid robust design methodology

Abstract

The extensibility of a linear rapid robust design methodology is examined. This analysis is approached from a computational cost and accuracy perspective. The sensitivity of the solution's computational cost is examined by analysing effects such as the number of design variables, nonlinearity of the CAs, and nonlinearity of the response in addition to several potential complexity metrics. Relative to traditional robust design methods, the linear rapid robust design methodology scaled better with the size of the problem and had performance that exceeded the traditional techniques examined. The accuracy of applying a method with linear fundamentals to nonlinear problems was examined. It is observed that if the magnitude of nonlinearity is less than 1000 times that of the nominal linear response, the error associated with applying successive linearization will result in errors in the response less than 10% compared to the full nonlinear error.

Keywords:

• multidisciplinary design optimization
• robust design
• design complexity
• perturbation analysis

Disclosure statement

No potential conflict of interest was reported by the authors.