Multidisciplinary optimization of liquid cooled heat sinks with compound jet/channel structures arranged in a multipass configuration

In this paper, we present the multidisciplinary optimization of microchannel layout in a multipass heat sink by topology optimization. Unlike previous approaches that have focused on designing microchannels in laminar region, the current work focuses on manipulating fin geometries of heat sinks in turbulence region, which allows more complex and higher velocity flow and therefore results in better heat sink performance. To save the computational cost in topology optimization, a simplified Darcy-flow based model is utilized to cope with turbulence calculation, of which the permeability parameter is carefully calibrated in order to make the output physical fields ideally be close to those from full Navier Stokes-based model. As an iterative process, topology optimization gradually evolves channel layout from simple to complex. The selection of objective function and the setting of fluid volume constraints are studied in the optimization process. Besides the SIMP-based optimization, an integrated optimization strategy is developed, which comprises two different topology optimizers: a moving morphable components based optimizer (MMC) for the initial topology prediction and a density based optimizer (SIMP) for subsequent topology elaboration. Detailed 3D computational fluid dynamics (CFD) analysis is then implemented to examine the thermohydraulic characteristics of the optimized heat sinks and a specially designed experimental set up is built to validate the 3D simulation results. The numerical predictions about the trend of the pressure drop and unit thermal resistance change with the coolant inlet flow are in good agreement with the experimental results. Compared with SIMP-based optimization, the design based on hybrid optimization strategy has lower unit thermal resistance at the same inlet flow rate without large pressure drop increasing.