Liquid-cooled heat sink design methodology with technical and commercial viability considerations: Case study of a partially 3-D printed prototype

Abstract

A two-step methodology is proposed for the design and optimization of liquid-cooled heat sinks, where manufacturing costs are included as a commercial viability assessment metric. First, manifold modifications were made to reduce pressure losses in the baseline design. Second, a full-factorial parametric computational fluid dynamics (CFD) study fed an artificial neural network (ANN) model for further optimization. The aforementioned methodology was applied to the optimization of liquid-cooled heat sinks for CPU cooling. The baseline design consisted of a 3-D printed resin plenum mounted on a metallic heat spreader with parallel fin channels. The plenum delivers a slot jet along the center of the heat sink into a central passage that bifurcates the fin channels. An improved plenum design with ∼11% lower pressure losses than the baseline model was generated after the first optimization step. The designs generated in the parametric CFD study achieved average thermal resistances as low as 0.01 K/W for a theoretical pumping power range of 50–350 mW. A representative design from this study was manufactured and experimentally tested to assess the simulation approach, providing reasonable agreement with the numerical results. Consecutively, the ANN allowed for a higher resolution sampling of the design space, where optimal configurations were found in terms of an overall thermo-hydraulic performance parameter and an additional metric accounting for manufacturing costs. The methodology reported herein caters to heat sink designers seeking commercialization of their prototypes as the contrast between technical performance and manufacturing costs is addressed. Additionally, the mathematical tools utilized allow for cost-effective topology optimizations.