Rapid developments in the design of chips and electronic devices for high-performance computers have led to a need for new and more effective methods of chip cooling. The purpose of this study was to investigate the heat transfer development and characteristics of the aluminum foam heat sink for Intel core i7 processor. The aluminum foam heat sink was subjected to a steady flow of water covering the non-Darcy flow regime (297–1353 Reynolds numbers). The bottom side of the heat sink was heated with a heat flux between 8.5 and 13.8 W/cm2. The heat development and thermal entry length of the aluminum foam heat sink were examined and presented. The distributions of the local surface temperature and the local Nusselt number were measured and compared with the numerical data obtained using finite element method. The average Nusselt number was obtained for the range of Reynolds numbers and an empirical correlation of the average Nusselt number as a function of the Reynolds number was derived. A comparison of previous studies using water as a coolant through aluminum foam was conducted in order to calibrate the empirical correlation of the average Nusselt number. The pressure drop across the foam was measured. The thermal performance of aluminum foam heat sink was evaluated based on the average Nusselt number and the required pumping power. The experimental results revealed that the thermal entry length increases along with increases in the Reynolds number. The results also revealed that local surface temperature increases as the heat flux increases, decreasing the Reynolds number and increasing the flow direction axis. In the fully developed region, the local Nusselt number is strongly dependent on the Reynolds number. The thermal efficiency index was defined in the study in order to evaluate the thermal performance for the aluminum foam heat sink. The results were also compared with those of previous experimental studies that used air as a coolant. The results indicated that, as a coolant, water provides lower average surface temperatures and a more uniform temperature profile. The numerical results were in good agreement with the local Nusselt number and the local experimental temperature with a maximum relative error of 0.83% and 0.43%, respectively.
