Numerical simulation of subcooled flow boiling in a manifold microchannel heat sink

Abstract

Recently, various miniaturized electronic systems and the rapid increase in power density of microelectronic devices have created additional requirements for micro-scale heat dissipation technology. To dissipate heat in a micro-scale region efficiently, several creative cooling methods for high heat flux dissipation have been designed on the order of 1000 W/cm2. Due to its high surface-to-volume ratio, the microchannel heat sink has become the most popular one among emerging technologies and an effective tool for high heat flux thermal management. Nevertheless, in common electronic microchannel cooling systems, the traditional parallel microchannel causes a dramatic pressure loss along channel length. In this regard, the manifold microchannel heat sink cooling technology becomes promising with its high heat dissipation potential and moderate pressure drop penalty. The shortened effective flow length compared with the traditional microchannel provides a smaller pressure drop and delays the onset of the critical-heat-flux point. However, while a number of numerical and experimental studies have been conducted for the single-phase flow in the manifold microchannel, two-phase flow boiling in this kind of channel is not widely employed because of its issues with flow pattern prediction. This work develops a new solver based on open source software OpenFOAM to perform the transient process of subcooled flow boiling in the manifold microchannel heat sink. Additionally, heat transfer in the silicon substrate is considered to improve the accuracy of simulation. Similar to single-phase flow simulation for the manifold microchannel, a unit cell model extracted from the whole cooling plate is employed for the two-phase boiling flow to avoid huge computational costs. Through several reasonable assumptions, the symmetry boundary condition is applied on four lateral surfaces in the unit cell, while uniform heat flux is adopted on the bottom wall. A uniform inlet velocity boundary estimated from the total volume flow rate condition is coupled with a fixed temperature value of 7°C lower than saturation temperature. After the mesh independence test, the self-programming solver and numerical methods are validated by comparing numerical results with experimental data. It is determined that the boiling curve obtained by the empirical rate phase-change model with parameters r l= r v=100 agrees well with experimental results. Furthermore, heat transfer and pressure drop performance differences are caused by microchannel width w c and fin width w f. In comparison to sample A ( w c= w f=15 μm), decreasing w c and w f simultaneously leads to lower average wall temperature of the heated surface; the channel pressure drop between inlet and outlet surfaces rises dramatically when the widths decrease. When the total number of microchannels in the manifold microchannel cooling plate remains unchanged, increasing w c results in a decreased w f. Based on experimental results, heat sink sample C shows that the pressure loss can be considerably reduced under the premise of sacrificing a small degree of heat transfer performance, which can be attributed to the increased microchannel hydraulic diameter. The manifold microchannel heat sink with w c> w f is recommended for practical application due to lower pressure loss and average temperature variation.