Exploring Microchannel Cooling Designs through the uses of Numerical Optimization and Additive Manufacturing

Kirsch, Kathryn Louise
Graduate Program:
Mechanical Engineering
Doctor of Philosophy
Document Type:
Date of Defense:
October 09, 2017
Committee Members:
  • Karen Ann Thole, Dissertation Advisor
  • Karen Ann Thole, Committee Chair
  • Stephen P Lynch, Committee Member
  • Robert Francis Kunz, Committee Member
  • Edward William Reutzel, Outside Member
  • Reid Adam Berdanier, Special Member
  • heat transfer
  • optimization
  • additive manufacturing
  • internal cooling
  • adjoint method
Advancing the performance of heat exchangers beyond their current capabilities requires a fresh look into the relationship between design methodologies and manufacturing techniques. Progress in design tools can lead to the efficient generation of unique geometries optimized for a given application; progress in manufacturing capabilities can turn those unique geometries into reality. Currently, however, the translation from computational design to as-built manufactured part is rarely one to one. This work seeks to highlight the effects of both nontraditional and traditional design methodologies on microchannels manufactured using Laser Powder Bed Fusion. The nontraditional design approach made use of an adjoint-based sensitivity analysis to optimize the shape of four baseline microchannel designs for three different objective functions. The sensitivity analysis was coupled with a conventional flow solver, and identified regions of the baseline designs that exerted the strongest influence on the objective function. The shape of those most sensitive regions were then morphed to achieve the optimization goal. The resultant microchannel geometries were largely reproduced in the Laser Powder Bed Fusion process, though contained a much higher surface roughness than was modeled. Even still, some of the objective functions were achieved experimentally. A follow-up study was conceived to delve deeper into the efficacy of the design tool, and to isolate the effects of the optimized features from the effects of high surface roughness in the as-built metal parts. A stereolithography process was used to build the optimized channels in plastic, and allowed for experimentation on the true, intended optimized designs. In general, the adjoint-based sensitivity analysis was successful: experiments on the plastic optimized channels yielded analogous performance to the numerical simulation. The influence of surface roughness in the metal parts varied depending on the shape changes. Another study was performed that applied a traditional design approach to microchannels containing pin fins. These arrays were also built using Laser Powder Bed Fusion. Results from this study highlighted the differences in performance between conventionally and additively manufactured pin fin arrays. Additionally, the study identified in the additive process a dependence on surface roughness that was based on pin density in the channels; such a manufacturing consideration need not be taken into account with a conventional method.