Multi-stage optimization of spray flow characteristics and structural performance for a pressurizer spray valve in nuclear power plants

Abstract

To address abnormal flow increase and internal wear in the pressurizer spray valve (PSV) under small openings, this study conducts flow analysis and structural optimization. A CPO-SVR surrogate model combined with the hippopotamus optimization algorithm is proposed. The model achieves high accuracy with an R 2 of 0.95 for stress prediction and 0.99 for flow coefficient prediction, with prediction errors of 0.23% and 1.4%, respectively, while reducing computational cost by approximately 86%. After optimization, valve seat stress drops by 49.4% to 150.58 MPa, maximum flow velocity is reduced by 7.9%, and the abnormal flow plateau in the 0%–20% opening range is eliminated, yielding a smooth characteristic that closely follows the ideal equal-percentage curve. This study provides an effective method for the design of nuclear-grade valves.