Parametric Design of a Manifold for a Hollow Micropillar Evaporator for use in High Heat Flux Applications
Date of Award
Master of Science (MS)
With the failure of Dennard Scaling and changes in electronics packaging leading toward 2.5 and 3D packages, the current methods of thermal management are failing to keep up with rising heat fluxes. In order to manage these rising heat fluxes, engineers and scientists are looking towards two phase cooling and microchannel flows.
Many forms of novel thermal management are currently being investigated as means to address the thermal challenges present in modern microelectronics, including flow boiling, pool boiling, and direct liquid cooling. One technology of interest is evaporative cooling using continuously fed micro-droplets. High evaporation rates can be achieved by using engineered hollow micropillar structures, and these devices have been shown to reject upwards of 280 watts/cm$^2$ with water as a working fluid. Current systems only address the management of high heat fluxes, and do not consider the packaging, liquid delivery mechanisms, and thermal resistance constraints that are present in actual electronic systems.
This paper seeks to develop a manifold and enclosure to allow a hollow micropillar evaporator to be implemented on a physical device. Geometric, thermal, and hydraulic constraints are created to define the geometry of a manifold to deliver liquid to the evaporator. The main challenge present in the design of the manifold and enclosure is to ensure uniform delivery of the working fluid to each pillar in the array while minimizing the thermal resistance between the bottom of the manifold and the evaporator.
Two different layouts are presented to achieve this balance between uniform flow rate and minimal thermal resistance. The first architecture flows the working fluid laterally underneath the evaporator, and the second incorporates a 180 degree turn of the working fluid to flow vertically before reaching the evaporator. 2D parametric CAD models are created of both layouts to simulate a range of design variables. First, an analytical resistance model is developed to analyse the lateral flow geometry. Then a computational fluid dynamics (CFD) model is created and used to run a multi-factor design of experiments (DOE) which allows for the comparison of several geometric variables along with different viscosities, densities, and fluid velocities on the flow rate through each pillar in the system. Since there is not a specific device or implementation currently in mind, this work seeks to develop an understanding of the fundamental physics present in the single phase fluid flow leading to the evaporator. Interactions and main effects are identified from the DOE, with only the height of the liquid channel dictating the performance of the system. The fluid density, viscosity, and flowrate does not significantly impact the distribution of liquid through each pillar at the values under study.
Lastly, a physical prototype is developed for use with R245-fa refrigerant. The final prototype incorporates a 3D manifold, liquid overflow channels, and mating features to aid in the assembly of the evaporator, copper heat spreader, and manifold. Material properties are assessed to ensure compatibility with R245-fa, along with feature size requirements. Several prototypes are assembled and are tested to seal up to 15 bar. A loop thermosyphon utilizing the evaporator was developed and shown to evaporate the refrigerant and create a two phase cycle, however full data was not able to be collected on the experiment.
The developments presented in this thesis provides the foundation for future implementation of hollow micropillar evaporators. Additional work can be done using the DOE to create a reduced order model (ROM), which would allow for rapid evaluation of future potential geometries, flow rate, and working fluids. Manufacturing limits and materials are suggested which allows for faster development of physical prototypes. Future work on the upper manifold, specifically the bifurcation from one inlet to a square outlet needs to be addressed. Lastly, a multi-phase numerical model needs to be developed to understand the change in evaporation rate as the droplets grow or shrink, as well as transient effects that might arise due to startup or variation in thermal loading.
Dave Peters Jackson Potter