Multi-objective optimization of thermal modules in high heat flux laptops

Abstract

The thermal module, as an important cooling pathway for chips, is a major factor affecting the performance of laptops. However, thermal analysis and optimization of the whole thermal module of a laptop is absent. Moreover, cost is also a major consideration in the design of a laptop’s thermal module. In this study, based on the thermal module of a high heat flux prototype laptop, we constructed an efficient thermal resistance network model and a cost model to evaluate the impact of different components. Furthermore, we employed NSGA-II (non-dominated sorting genetic algorithm II) to enhance the thermal module's heat dissipation capacity while minimizing cost. By obtaining Pareto solutions that provided optimal thermal resistances for the components in the thermal module, we then utilized LINMAP (linear programming technique for multidimensional analysis of preference) as a decision-making method to determine the best compromise solution. The best compromise solution focuses on reducing the thermal resistances of the thermal interface material on the high heat flux APU side, the shared heat pipe, and heatsinks to enhance the heat dissipation capacity of the thermal module. To control cost, sacrifices were made in the heat transfer performances of the thermal interface material on the GPU side and the independent heat pipes. Comparing the best compromise solution with the prototype, the cooling capacity is improved by 24% and cost is reduced by 9%. The efficient model coupled with optimization method may provide a new theoretical guideline for the laptop designers.