Heat transfer enhancement in microchannel heat sinks with dual split-cylinder and its intelligent algorithm based fast optimization

Microchannel heat sink (MCHS) with dual split-cylinder (initial model) is proposed in this work, to obtain better thermal performance (TP) and eliminate local heat transfer deterioration. Comparing with the MCHS with cylinder (prototype), the results indicate that the maximum relative increment of TP of initial model is 34.63%, obtained at Re = 200. And, the temperature uniformity is also improved at this case as well. However, this superiority is weakened as Re increases or decreases. Therefore, intelligent algorithm based fast optimization method, combing three dimensional geometry reconstruction, spatial discretization, and simulation, is further established and applied in the initial model. The generalized pattern search algorithm (GPS) is utilized to control the direct search of width (S_width) and location (S_location) of dual split-cylinder channel for optimum TP. In the optimization, the interval size of search is adjusted to improve optimization precision and avoid the duplication of search points. After optimization, the optimum S_width for each Re differs, which is decreased with the increase of Re. While, for all Re, the optimum S_location are decreased to the boundary of design variables range. Accordingly, comparing with initial model, the TP of optimized model is increased a lot at each Re. And, the largest TP = 3.10 is obtained at Re = 300. And that, the maximum relative increment of TP from prototype model to optimized model is obtained at Re = 250, which is 63.41%. Besides, the temperature uniformity of optimized model is also improved effectively, which is more obvious as Re increases.