Developing an Optimum Design of the Double Layer MicroChannel Heat Sink for High Speed CPUs

摘要

High speed CPUs and electronic chips usually dissipate relatively large quantities of energy in the form of heat. The limited available cooling space requires innovative and efficient thermal management techniques. These techniques must accomplish low operation temperatures along with uniform temperature distribution over the chip surface. Recently, several researches focused on developing different microchannel heat sink (MCHS) designs for this purpose. Among these designs is the double layer microchannel heat sink (DL-MCHS), which can be operated in either parallel flow (PF) or counter flow (CF) operation modes. The thermo-hydraulic characteristics of this heat sink was comprehensively investigated in the literature using numerical methods. However, based on the authors literature survey, all the previous numerical investigations considered the computational domain as a straight section with multichannels in each layer. This approach assumes a uniform velocity for all the flow channels in each layer and neglect the effect of the inlet and outlet headers. In addition, the heat interaction between the coolant in both layers through the header section was not considered. These assumptions cause a considerable discrepancy between the numerical results in the literature and the realistic conditions. Therefore, in this work, a detailed 3D conjugate heat transfer model is developed. In this 3D model, DL-MCHS is designed with and without headers. The designed heat sinks are operated under PF and CF conditions. For a specific electronic chip with dimensions of 13 mm×40 mm at a heat flux of 5 kW/m~2. The model is validated with the experimental results in the literature. It is found that, including the header design in the simulation of the DL-MCHS must be considered in the simulation to predict the accurate thermo-hydraulic performance of the DL-MCHS specially in the CF. Further, using the DL-MCHS in PF operation accomplished a lower average chip temperature compared with the CF. Furthermore, at high coolant flowrate, neglecting the effect of the header in the CFD calculations can be approximated to predict the chip surface temperature. And finally, neglecting the header in the CFD calculations significantly affect the calculated pumping power for both PF and CF operated DL-MCHS.