Air-Cooled Closed Loop Thermosyphon Cooling System Experimental Campaign: Effects of Working Fluid, Heat Load and Air Flow Rate

Abstract

This study presents the experimental results of a test campaign performed on a closed loop thermosyphon having a footprint cooling area of 50 mm x 50 mm designed specifically for cooling of high heat flux microprocessors for 2U servers by JJ Cooling Innovation and Novark Technologies. Importantly, the cooling load to fan power load performance ratios determined by measuring the fan’s consumption showed ratios up to 60-to-1 were obtained using this compact thermosyphon cooling technology. In particular, the effects of working fluid charge and type, heat load and air flow rate have been investigated. In the first set of tests, the main parameter of interest was the filling ratio. The results show that for the three low global warming potential refrigerants R1233zd(E), R1234ze(E) and R1336mzz(Z), the thermal resistance increases with the filling ratio. To ensure safe operation, the optimal filling ratio was chosen as 50% for R1233zd(E), 42% for R1234ze(E) and 50% for R1336mzz(Z). A second set of tests was performed at these optimal filling ratios at different heat loads (from 50 W to 750 W) and air volumetric flow rates (50, 100 and 150 CFM). For the range of heat loads tested, when the heat load was increased, the thermal resistance decreased until a lower plateau limit value was reached. The smallest thermal resistance was reached with R1234ze(E) at 750 W and 150 CFM, equal to 0.065 K/W. Finally, to more closely emulate a real CPU cooling application, the heated copper block used in the previous tests was replaced by a power resistor in a last set of tests but only for refrigerant R1233zd(E). Furthermore, no temperature overshoots were observed for cold startup tests; however, the maximal heat load without reaching dry out was 250 W lower than with the copper block due to non-uniform heat generation of this type of heater.