An integrated methodology for multiobjective optimal component placement and heat sink sizing

The advances in power electronics packaging resulting from the integration of various high power devices into single programmable blocks are projected to significantly increase heat dissipation rates up to 2-3 W/mm² within the next decade. Conventional thermal management schemes such as air-cooled heat sinks are inadequate at these high power levels, requiring the use of forced liquid cooled heat sinks, or cold plates. A primary issue concerning the thermal-packaging designers today is the selection of a suitable cold plate and the optimum placement of components on it. This paper focuses on a methodology for selection of cold plates and the optimal placement of power modules on them. The basis of the methodology is the integration of computationally efficient reduced or "compact" thermal models within a genetic algorithm based multiobjective optimization framework. This methodology is used to optimize the placement of multiple power modules on a cold plate. It is also extended to perform a tradeoff analysis between multiple competing objectives for cold plate selection. Compact thermal models for power electronics and heat sinks are developed, validated, and integrated within the optimization framework. The effectiveness of the methodology as a rapid scouting tool to arrive at an optimum tradeoff solution set for further analysis by detailed numerical simulation is discussed. The methodology is general in the sense that it allows for incorporation of changes in optimization criteria or improvements in the heat transfer solver.