Subtractive design: A novel approach to heatsink improvement

Abstract

Typical Heatsink design includes base and fin thickness, fin height, and fin gap optimization. In situations where material cost or mass of the heat sink are also a design priority further optimization with respect to mass removal can be significant.This paper discusses a method to further improve a heat sink topology by the systematic removal of heat sink mass where the Thermal BottleNeck (BN) Number was found to be lowest. NomenclatureRth -Thermal Resistance (DegC/W) = (Tb -Ta) / P Tb -Heatsink base temperature (above center of heat source) Ta -Ambient temperature P -Dissipated power BN -Thermal BottleNeck Number IntroductionTypical heat sink design involves balancing trade-offs between parameters such as heat spreading, heat sink mass, airflow bypass, and manufacturability. A typical optimization would include varying a small number of dimensional parameters of the heat sink and a temperature rise used as the objective cost function. A Computational Fluid Dynamics (CFD) simulation offers the opportunity to consider the spatial distribution of heat transfer effectiveness and, based on that, changes made to the geometry topology with the intention to further improve the design. Previous studies have considered elements of this approach [1].An earlier study [2] explored an additive methodology to heatsink design. In that study the heatsink fins were allowed to grow where the surface temperature was the highest, as part of an iterative design process. The heatsink topology evolved over a number of cycles until no further performance gains were achieved. It was observed that the heatsink grew as a tree-like structure from the center of the heat source. This process was also applied to grow the heatsink where the Thermal BottleNeck was largest [3], producing slightly improved behaviour compared to using surface temperature to drive the additive design process.Thermal BottleNeck distribution was also used to directly modify fin thickness of a pre-optimized heatsink [4]. The Thermal BottleNeck values in each fin were noted and the fin thicknesses modified in an attempt to equalize the BN in each fin, leading to a marked improvement in thermal performance.

Freeform-Optimized Shapes for Natural-Convection Cooling

“…Next to CFD, also other approaches have been developed such as the Particle Swarm Optimization (PSO) algorithm [15]. [16] and subtractive [17] optimization approaches in which material was either added or removed from a location where the surface temperature has the highest or the contribution to the thermal resistance was the least, respectively.…”

Multi-Material Heatsink Design Using Level-Set Topology Optimization

“…An alternate topological design approach for heatsink optimization has been presented by Bornoff et al The method is based on Bejan's constructal theory [27], which explains the underlying principle behind all naturally existing designs or configurations. Bornoff utilizes the approach as both an additive design method [28] and as a subtractive design method [29] for heatsink designs. In the former study, material is sequentially added at the maximum temperature region and in the later from a baseline heat sink, material is sequentially removed where the bottle neck number is lowest.…”