Application of adaptive multilevel substructuring technique to model CMOS micromachined thermistor gas sensor, part (I): A feasibility study

A study has been conducted to investigate the feasibility of using the multilevel substructuring method to perform parametric analysis for MEMS devices. The feasibility study was conducted on a CMOS micromachined thermistor gas sensor. Two multilevel substructuring methods were used, mainly the cumulative and nested superelements methods. The interface problem was found to increase rapidly with the increase in the number of superelements for the cumulative technique. On the other hand, the nested superelements technique was found to provide an almost fixed and much more reduced interface problem. The results show that, for the same number of reduced elements, the nested superelements method provides a better speedup factor (2.36-4.61) compared to the cumulative method. In the current study, two strategies were used to deal with the nonlinear thermal analysis of the microheater. In the first strategy the substructuring was only limited to the linear portions of the model. In the second strategy the substructuring was extended to include portions of the model with a reduced nonlinearity. The second strategy increased the computational savings by a percentage of 20%, compared to the first strategy with a reasonable loss of accuracy of only about 3%.