Hydrodynamic design of a circulating water channel based on a fractional-step multi-objective optimization

Abstract

In this study, a multi-objective optimization of a circulating water channel (CWC) circuit was conducted to reduce the circuit's total pressure loss and limit the construction costs. The numerical optimization system was composed of an optimal Latin hypercube design (Opt LHD), a fast flexible space-filling design (FFF design), the Design Program for Wind Tunnels and Circulating Water Channels (DPWC), a support vector machine (SVM), a support vector regression (SVR), and a particle swarm optimization (PSO). To save simulation time, a simplified CWC model was proposed and verified to replace the complete CWC model. The optimization process was divided into two steps. In the first step, four design variables, including the casing's diameter, the diffuser's width ratio, the diffuser's height ratio, and the nozzle's area ratio, were selected to reshape the circuit, and an initial sample set for classification was constructed through CFD simulation. Then, a design subspace was extracted by an SVM to avoid flow separation in derivative CWCs. In the second step, a new sample set was constructed in the design subspace, and an SVR-based surrogate model was constructed based on it. Finally, the multi-objective optimization was performed within the design subspace. Optimization results showed that the CWC's total pressure loss was negatively correlated with the CWC's surface area. The optimal CWC circuit was selected from the Pareto front of two-objective optimization. Numerical results of the optimal CWC circuit showed that the total surface area increased by 7.5% while the total pressure loss significantly decreased by 19.3%, and that the residual hydrodynamic performance met the design requirements.