Case study of circuit element position optimization

Michihiro Takeshima -IDAJ CO.

CHALLENGE - As circuit elements reach high power and high density, temperature becomes problematic. Among the many ways to lower temperature, one is circuit element position optimization. However, it has strong constraints, interaction effect and multi-peak, which makes it difficult to find an optimal solution only by using optimization algorithm in limited time.

SOLUTION - In this case study, an optimal solution is attempted from a combination of optimization algorithm result and data analysis.​The objective is to minimize maximum temperature of 17 elements using the integration of thermal analysis software FloTherm and the optimizer modeFRONTIER. Calculation time took about 42 hours with modeFRONTIER grid computing. The genetic algorithm MOSA was used in the calculations.

BENEFITS - The results of the study show that the multi-objective optimization procedure using the optimization software modeFRONTIER is valid. Using the design guideline, MAX_ICTEMP fell 0.17 ℃. (From 74.45℃ to 74.28℃).