Thermal analysis on the impact of different velocity calculation methods for microchannel liquid cooling system in GPUs

Recently, computational tasks are carried out by GPUs which consists of hundreds of cores that compute at higher speeds to provide systematic results. The computational power of GPUs suffers significant disadvantage in the form of heat emission. Traditional liquid cooling methods circulate the working fluid at constant velocity irrespective of the temperature raised. To mitigate the effects of improper cooling, the research focuses on analysing liquid cooling system for GPUs where Ethylene Glycol 40 was circulated through 20 microchannels at different velocities and thermal behaviour was analysed using a CFD analysis software. The paper aims to provide a clarity on the methods to calculate velocity for simulation approach. The velocity was calculated using two distinct approaches, the first approach uses the mass flow rate equation and second approach uses Reynolds number equation considering laminar flow. On assuming three different values for mass flow rate and Reynolds number, the velocity of the fluid is calculated. For Mass Flow rate of 0.1 g s-1, 0.2 g s-1 and 0.3 g s-1, the obtained velocities were 0.01886 m s-1, 0.03773 m s-1 and 0.05659 m s-1 respectively. For Reynolds number of 10, 20 and 30, the obtained velocities were 0.04 m s-1, 0.09 m s-1 and 0.1422 m s-1 respectively. The analysis proved that at lower velocities, the working fluid was not effective in carrying the heat but at higher velocities the temperature reduction was higher; displaying the ability of the coolant to absorb heat depending on the velocity. With the temperature data from the simulation, the heat transfer rates were calculated as 10352.4049 J, 7948.3535 J, 7711.2185 J, 6719.9088 J, 5318.3105 J, 3976.2213 J for 0.01886 m s-1, 0.03773 m s-1, 0.04 m s-1, 0.05659 m s-1, 0.09 m s-1 and 0.1422 m s-1 respectively. The analysis concludes that by changing the velocity of the working fluid, the thermal management system can work effectively under different scenarios.