Three-objective shape optimization and parametric study of a micro-channel heat sink with discrete non-uniform heat flux boundary conditions
Three-objective shape optimization and parametric study of a micro-channel heat sink with discrete non-uniform heat flux boundary conditions
复制标题
DOI：
10.1016/j.applthermaleng.2018.12.128
复制
发表时间：
2019-03
影响因子：
6.4
通讯作者：
Yaser Hadad;Bharath Ramakrishnan;Reza Pejman;S. Rangarajan;P. Chiarot;A. Pattamatta;B. Sammakia
中科院分区：
文献类型：
--
作者：
Yaser Hadad;Bharath Ramakrishnan;Reza Pejman;S. Rangarajan;P. Chiarot;A. Pattamatta;B. Sammakia
为满足多芯片处理器的工作要求，设计了一种具有平行矩形微通道的水冷多芯片热沉。基于响应面法（RSM）的形状优化策略被用来最小化压力降和模具最高外壳温度。通过数值模拟，分析了芯片与热沉之间的热界面材料和散热器对器件性能的影响。在冷却剂流速和入口温度以及模具的功率、位置和表面积的恒定值下进行优化。在RSM设计点使用CFD（计算流体动力学）模拟确定了通道水力直径、雷诺数、热入口长度和总传热表面积对散热器水力和热性能的影响。进行了灵敏度分析，以评估设计参数对响应参数的影响。采用JAYA算法，通过极小化基于响应参数定义的加权目标函数，实现最优设计。将加权和法的结果与基于Pareto的三目标优化方法和非支配排序遗传算法进行了比较。最后，进行了参数研究，以了解设计参数对响应参数的影响。
A water-cooled multi-die heat sink with parallel rectangular micro-channels was designed to satisfy the operational requirements of a multi-die processor. A shape optimization strategy based on the RSM (response surface method) was used to minimize pressure drop and die maximum case temperatures. The effects of the thermal interface materials and heat spreader between the dies and heat sink were captured by the numerical simulation. The optimization was performed for constant values of coolant flow rate and inlet temperature, as well as the power, location, and surface area of the dies. The influence of channel hydraulic diameter, Reynolds number, thermal entrance length, and total heat transfer surface area on the hydraulic and thermal performance of the heat sink was determined using CFD (computational fluid dynamics) simulations at RSM design points. A sensitivity analysis was performed to evaluate the effect of the design parameters on the response parameters. The optimum designs were achieved by minimizing a weighted objective function defined based on response parameters using JAYA algorithm. The results of weighted sum method were compared with Pareto based three-objective optimization with a NSGA-II (non-dominated sorting GENETIC algorithm). Finally, a parametric study was performed to see the effect of the design parameters on the response parameters.