Anthony James Robinson

The present study examines the application of miniature Vapour-Compression Refrigeration (VCR) technology for cooling of high-power Central Processing Units (CPUs). Specifically, VCR-assisted liquid-cooling, which enables sub-ambient liquid coolant temperature, is investigated in terms of cooling and energy performance. A steady state mathematical model was developed that combines heat transfer, fluid mechanic, and thermodynamic sub-models of the compressor, condenser, expansion valve, evaporator, and liquid cooling circuit to predict CPU die temperatures under varying heat loads. The model was validated against experimental data across a wide range of operating conditions: CPU heat fluxes of 25 - 75 W/cm², compressor speeds of 2540 - 6502 rpm, air velocities of 3.3 - 4.2 m/s, and liquid coolant flow rates of 1 - 6 L/min. Subsequently, the model was integrated within a Design of Experiments (DOE) framework to investigate parameter sensitivities and perform system-level design optimization, primarily focussing on achieving target CPU die temperatures with maximum Coefficient of Performance (COP). The results and analysis indicate that the compressor speed exerts the dominant influence on cooling performance, though at the highest energy consumption cost. However, the COP can be increased significantly via design optimization which balances the cooling and power consumption trade-offs. In particular, increasing either or both the liquid coolant flow rate and evaporator size have a disproportionately high influence on the cooling performance compared with the additional energy consumption penalty, allowing CPU cooling targets to be achieved with maximum COP.

This study investigates optimisation of the thermal-hydraulic performance of liquid impinging jet arrays for cooling of concentrated heat sources with an integrated heat spreader, mimicking a typical high-powered CPU package. Three heat source sizes were investigated, namely 10 x 10 mm ² , 15 x 15 mm ² , and 20 x 20 mm ² , all with a 1.3 mm thick copper heat spreader. An automated optimisation workflow was developed in a CFD-based environment to determine the ideal geometric design parameters of the impinging jet array heat exchangers, with the objective functions chosen to minimise hydraulic power consumption and/or minimise the overall thermal resistance, representing both single and multi-objective optimisation strategies. The analysis showed that for maximising the heat transfer performance only, the design tended to favour fewer closely spaced and higher velocity jets that targeted a concentrated region within the footprint on the heat spreader that was commensurate with the heat source. This focussed the high convective intensity of the impinging jets on the central region of the heat spreader where the highest heat flux levels exist. However, the high jet velocities incurred a comparatively high hydraulic pumping power penalty. Conversely, when minimising both thermal resistance and hydraulic power consumption simultaneously, the design favoured a much higher population of lower velocity jets in order to mitigate the pressure drop and reduce the hydraulic power. Here, despite the lower convective cooling intensity of the lower velocity jets, the designs leveraged a much larger surface area of the heat spreader, which tended to maintain high heat transfer performance with disproportionately lower hydraulic penalties.

Sessile water droplet evaporation in varied gravity and electric fields has been experimentally studied. Specifically, the influences of gravity and electric fields are investigated in the context of the heat flux distribution beneath the droplets, as well as the droplet mechanics and resulting shapes. Experimental testing was carried out during a European Space Agency (ESA) Parabolic Flight Campaign (PFC 66). The droplets tested evaporated with a pinned contact line, a single wettability condition, and varied droplet volume and substrate heat flux. The peak heat transfer was located at the contact line for all cases. The peak heat flux, average heat flux, and droplet evaporation rate were shown to vary strongly with gravity, with higher values noted for hypergravity conditions and lower values in microgravity conditions. The droplet thermal inertia was shown to play a significant role, with larger droplets taking more time to reach thermal equilibrium during the parabolic testing period. No significant impact of the electric field on the droplet evaporation was noted for these test conditions.

Growth rate is an important consideration in solidification as it influences features of the micro and macrostructure such as morphology and the columnar to equiaxed transition. Thus, accurately quantifying the growth rate is key to understanding the solidification process as well as verifying numerical simulations. In this work, an experimental apparatus has been developed to investigate in-situ directional solidification using a transparent analogue material, neopentyl glycol-35 wt%(D)camphor. The facility enables the control of the mushy zone growth rate as well as the imposed temperature gradient during solidification, similar to Bridgman type furnaces. Image processing techniques were developed for automated tracking of position and growth rate of the solidification front (SF) using background subtraction on photographic image sequences captured in experiments. Growth rate measurements using this technique and the traditional method (liquidus isotherm speed) have been benchmarked against manual measurements of position and growth rate. Results show that automatically tracked growth rate compares well throughout experiments with errors less than 2.7%. Contrastingly, isotherm speed was in error up to 19%. Three columnar and two equiaxed morphologies were investigated and SFs were characterised for their growth rates and thermal conditions.