Optimal design of wavy microchannel heat sinks based on prediction and multi-objective optimization algorithm

Abstract

In this paper, the multi-objective optimization of wavy microchannel heat sinks is performed by combining numerical calculation, prediction algorithm and genetic algorithm. In numerical calculation, the fluid-solid conjugate heat transfer of heat sinks with different parameters are simulated in Fluent. On this basis, the variable parameters and objective parameters are used to complete the training of neural network model, which aims to achieve accurate prediction of objective parameters. Finally, the multi-objective genetic algorithm is applied to find the Pareto front according to different requirements on the foundation of the prediction model. Results show that the coefficient of determination of the neural network models are all greater than 0.85, which proves that the prediction model has a high accuracy. The Pareto fronts are obtained by non-dominated sorting genetic algorithm (NSGA-II) with different objective parameters and they reveal that the channel with the optimal performance corresponds to a larger channel width or Reynolds number. In addition, it is also found the dimensionless temperature difference is correlated with Nusselt number.

Keywords

• Wavy microchannel heat sinks
• Latin hypercube sampling
• Feedforward neural network
• Multi-objective optimization
• Pareto front