Topology Optimization of Cold Plates using a Particle Swarm Optimization Algorithm

As modern electronic devices become smaller and more power-dense, the excessive heat and hot spots generated degrades performance and causes failure. While cold plates are used to manage this heat, they often require high pumping power and add weight to the system. One method to improve cold plate designs is topology optimization. Topology optimization optimizes material distribution to minimize an objective function. This study used topology optimization to create an optimized cold plate design. A Particle Swarm Optimization (PSO) algorithm, implemented in MATLAB, was coupled with COMSOL Multiphysics via its LiveLink for MATLAB interface to optimize a design for improved thermal and hydraulic performance, as well as reduced weight. This new design was compared against a standard S-shaped channel cold plate. In simulations with inlet velocities ranging from 0.2 to 0.22 m/s, both designs showed similar thermal performance, the pressure drop started out similar but diverged as the velocity increased with the S shaped channel having the higher pressure drop with a difference of 642 Pa at a velocity of 0.22 m/s, the weight of the optimized design was also reduced by 7%. These results show that topology optimization using PSO can match or surpass the performance of a baseline design.