Modern electronics are continually evolving toward miniaturization and high performance, posing significant challenges for chip-level thermal management under ultra-high heat flux (>1000 W/cm2). Conventional heat sinks are inadequate for these demands. Diamond microchannel heat sinks, leveraging diamond's exceptional thermal conductivity, offer a promising solution. However, the high hardness and cost of diamond lead to elevated manufacturing costs for such heat sinks. Consequently, the co-optimization of thermal-hydraulic performance and manufacturing costs presents a critical challenge. This study employed a machine-learning-based multi-objective optimization approach to design diamond microchannel heat sinks, simultaneously considering thermal-hydraulic performance and manufacturing objectives. An artificial neural network predicted the thermal-hydraulic performance, and a genetic algorithm then identified Pareto-optimal solutions. First, a thermal-hydraulic dual-objective optimization was conducted to analyze the trade-off between the maximum temperature and pressure drop. Subsequently, two manufacturing objectives (aspect ratio and cross-sectional area) were introduced, thereby formulating a manufacturing-constrained multi-objective optimization problem. The results demonstrated clear trade-offs among these four objectives on the Pareto front. One notable optimal solution achieves a 61.6 % reduction in material cost and an estimated 58 % decrease in fabrication difficulty with only a 20 % compromise in thermal-hydraulic performance. Thus, this work provides a systematic design methodology that successfully balances performance with manufacturability, paving the way for the scalable industrial adoption of diamond microchannel heat sinks.


扫码关注我们
求助内容:
应助结果提醒方式:

