Combined Parameter and Tolerance Design for Quality via Computer Experiment: A Design for Thermoelectric Microactuator

Abstract

A thermoelectric microactuator is examined with a computer experimental approach based on the asymmetrical thermal expansion of the microstructure with two beams of different widths. Because a typical U-shaped lateral thermoelectric actuator is used, the remaining concerns are the associated parameters and tolerance values that need to be determined. Conventional approaches consider that parameter and tolerance variables are two unconnected controllable variables, with the condition that the analytical functions representing the design of interest are known. A belief that the two variables are unconnected, usually fails to find truly optimal solutions, particularly a belief that the design functions are nonlinear in a complicated design. Assuming that design functions are normally known is a mistake in most cases, especially in the early stage of design. In this regard, a simultaneous parameter and tolerance design accompanying the computer experiment is developed to ensure that true optimization is achieved. The computer experimental approach consists of both a computer simulation and a statistical method. In this paper, the computer simulation is performed with existing computer-aided engineering (CAE) software, and ANSYS, the finite-element method for the solution of coupled 3-D mechanical problems. The statistical method is one of a design experimental approach such as response surface methodology (RSM). The response value is the conformation rate used in measuring product quality for uncertain conditions. This approach provides designers not only with optimal parameter and tolerance values, but also with the importance ranking for the controllable variables of a product. This is particularly important for the evolutionary design processes in an uncertain environment.