Abas Abdoli Department of Mechanical and Materials Engineering, Florida International University, MAIDROC Laboratory, EC2960, 10555 West Flagler Street, EC3462, Miami, FL 33174 e-mail: aabdo004@fiu.edu George S. Dulikravich Professor and Director of MAIDROC Fellow ASME Department of Mechanical and Materials Engineering, Florida International University, MAIDROC Laboratory, EC2960, 10555 West Flagler Street, EC3462, Miami, FL 33174 e-mail: dulikrav@fiu.edu Multi-objective Design Optimization of Branching, Multifloor, Counterflow Microheat Exchangers Heat removal capacity, coolant pumping power requirement, and surface temperature nonuniformity are three major challenges facing single-phase flow microchannel com- pact heat exchangers. In this paper multi-objective optimization has been performed to increase heat removal capacity, and decrease pumping power and temperature nonuni- formity in complex networks of microchannels. Three-dimensional (3D) four-floor config- urations of counterflow branching networks of microchannels were optimized to increase heat removal capacity from surrounding silicon substrate (15  15  2 mm). Each floor has four different branching subnetworks with opposite flow direction with respect to the next one. Each branching subnetwork has four inlets and one outlet. Branching patterns of each of these subnetworks could be different from the others. Quasi-3D conjugate heat transfer analysis has been performed by developing a software package which uses quasi-1D thermofluid analysis and a 3D steady heat conduction analysis. These two solv- ers were coupled through their common boundaries representing surfaces of the cooling microchannels. Using quasi-3D conjugate analysis was found to require one order of magnitude less computing time than a fully 3D conjugate heat transfer analysis while offering comparable accuracy for these types of application. The analysis package is ca- pable of generating 3D branching networks with random topologies. Multi-objective opti- mization using modeFRONTIER software was performed using response surface approximation and genetic algorithm. Diameters and branching pattern of each subnet- work and coolant flow direction on each floor were design variables of multi-objective optimization. Maximizing heat removal capacity, while minimizing coolant pumping power requirement and temperature nonuniformity on the hot surface, were three simul- taneous objectives of the optimization. Pareto-optimal solutions demonstrate that thermal loads of up to 500 W/cm 2 can be managed with four-floor microchannel cooling networks. A fully 3D thermofluid analysis was performed for one of the optimal designs to confirm the accuracy of results obtained by the quasi-3D simulation package used in this paper. [DOI: 10.1115/1.4027911] Keywords: electronics cooling, microheat exchanger, multi-objective optimization, con- jugate heat transfer, single-phase flow microchannel 1 Introduction Cooling systems for new generation portable electronic devices with higher capacity of heat removal, higher efficiency and smaller size is one of the challenges in the heat transfer field. The heat dissipation of microprocessors has delineated an exponential increase over the past decade and up to 10 times larger heat fluxes, with respect to current devices, are expected in next-generation microelectronics [1]. One of the cooling system technologies is the cooling micro- channel based compact heat sink. Significantly smaller sizes of the microchannels offer major advantage of this method which allows multichip integration. The main challenges of this method are high pressure drop which require higher pumping power, sur- face temperature nonuniformity, liquid maldistribution, and cool- ant leaks [2]. Microchannel heat sinks have been investigated both experimentally and numerically [1–6]. Single-phase flow heat transfer in microchannels has been studied by many investigators. Heat transfer coefficients and friction factors in microchannels have been experimentally investigated by Kosar and Peles [7] for heat fluxes ranging from 3.8 to 167 W/cm 2 . Colgan et al. [8] investigated practical implementation of a single-phase micro- channel flow in silicon substrates to enhance removal of heat load up to 300 W/cm 2 using water as coolant. Walchli et al. [9] applied oscillating flow method on water cooling system for thin form fac- tor high performance electronics with 180 W/cm 2 heat flux load. A computational and experimental investigation of pressure losses and heat transfer in microchannel networks containing T-type junctions have been performed by Haller et al. [10]. Kim et al. [11] numerically studied the thermal and hydraulic perform- ance of single-phase microchannel flows versus phase change flows for different coolants. One of the first vestiges of the application of optimization methods to improve channel geometries was in the design of gas turbine blades. Martin and Dulikravich [12] presented a fully automated program for inverse design and optimization of cooling passages in internally cooled turbine blades, which was validated against experimental results from Pratt & Whitney Aircraft Com- pany. A few years later, Jelisavcic et al. [13] applied hybrid evolu- tionary optimization to the same concept of channel network optimization for turbomachinery applications. Hong et al. [14] presented a great effort to enhance the cooling uniformity of microchannel heat exchangers through the design of fractal tree- like networks, attempting to reduce coolant pumping power. Contributed by the Heat Transfer Division of ASME for publication in the JOURNAL OF HEAT TRANSFER. Manuscript received September 14, 2013; final manuscript received June 20, 2014; published online July 15, 2014. Assoc. Editor: Oronzio Manca. Journal of Heat Transfer OCTOBER 2014, Vol. 136 / 101801-1 Copyright V C 2014 by ASME Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 07/30/2014 Terms of Use: http://asme.org/terms