A total CAD-integrated design system for MEMS has been developed which can perform analysis and design considering mechanical performance of MEMS structures. Commercial finite element codes such as ANSYS and pre/post processor for finite element modeling and postprocessing has been integrated into a parametric CAD platform using API(Application programming interface) provided by the CAD software. For optimization, commercial optimization engine has also been integrated to the program.
The design methodology based on a parametric CAD has benefits compared with the conventional methodology where finite element model is used for design variable selection. Users can perform design optimization of various and complex structures with the smallest efforts by utilizing the parametric capability of CAD. Here, parameters of a CAD model can be directly selected as design variables.
Finite element model is made by an automatic mesh generator. Intensive labor often required in model preparation is reduced and large design parameter changes can be tackled without mesh distortion since the whole model can be re-meshed after every design perturbation.
Analysis results such as natural frequency, stress, strain and displacements are used as objective function or constraints in optimization formulation and arranged easily on the window setting.
Two modules are incorporated for optimization. One is using commercial optimization engine DOT and the other by using design sensitivity information calculated by finite difference method(FDM). This latter approach can be applied to many practical structural designs and is very robust.
To test the performance of the developed FDM-Optimization software, several examples such as the comb type structure and a beam type structure used in micro-gyroscope are used. To maximize the performance, two design optimization problems are defined considering the natural frequencies of the system and damping coefficient. The first is to minimize the difference of the sensing frequency(second natural frequency) and the driving frequency(first natural frequency). The third natural frequency is also constrained to be far from the first and second natural frequency. Additional constraint on overall stiffness of the system is imposed for structural integrity. The lengths and width of the beams supporting main plate are selected as design variables. The optimum design obtained has been satisfactory with the objective function reduced dramatically more than 100 percent.
The other example is about a micro accelerometer used in the automobile air-bag sensor. To maximize flexibility of the structure, the strain at the beam support in the model is to be maximized under several constraints on frequency, displacement and stress. The results are found satisfactory compared with those of a previous study.
The approach taken here for integrated optimal design of MEMS structures is shown working well with several practical problems. The capability of a trade-off study included in the development contributes to the attractiveness of the software, enhancing user friendliness and reliability.
