Reliability assessment and optimization for large-scale natural draft wet cooling system in nuclear power plants

Large-scale natural draft wet cooling systems (NDWCs) have become the primary cold-end system for nuclear power plants (NPPs). However, it suffers from reduced safety, reliability and economy due to strong random disturbance issues from both power grid and climate. Accordingly, evaluating and strengthening the stability of NDWCs is essential for enhancing the overall reliability of NPPs. This study develops an environmental boundary quantification model to characterize the operating state space of the power plant cooling system, thereby enabling effective capture of random reliability disturbances in nuclear power plants caused by climate fluctuations. Furthermore, a SHAP-weighted GA optimization (Shapley additive explanations) is used to analyze interpretable trade-offs between reliability and economics, thereby achieving synergistic improvements in stability and cost-effectiveness. Ultimately, system robustness is co-optimized through the thermo-flow synergistic effects of the cooling tower structure, where enhanced draft and ventilation performance contribute to stable operation under environmental conditions. The optimized design results in a net power increase of 977 million kWh per year, a 3.95% increase in annual revenue, approximately $1.163 million, and an average backpressure reduced by 0.09 kPa compared to the original design. These improvements significantly enhance the economic performance and operational reliability of the nuclear power system. This approach provides theoretical support for the design of nuclear power cooling systems to address environmental uncertainties.