关键词: 塔式光热电站, 布雷敦循环, 基于S-CO2的混合工质, 智能进化算法, 热力学性能

Abstract: For tower solar thermal power plants, a genetic-sequential quadratic programming (SQP) dual-layer intelligent evolutionary algorithm was employed. Based on REFPROP 9.1, four groups comprising forty CO₂-based binary mixtures (H₂S, He, butane, and Kr) were constructed, and their performance was analysed and compared. Using four typical seasonal days in Delingha as representative operating conditions, three representative working fluids (S-CO₂, S-CO₂-50%He, and S-CO₂-50%H₂S) were selected for performance comparison. The algorithm was validated and further applied to a literature case, achieving an increase of 24.6% in specific work and 2.12% in efficiency, with the optimal configuration identified as reheating combined with intercooling and without flow splitting. Results of pure substances indicate that, relative to S-CO₂, He and H₂S increased the specific work by approximately 324.8% and 28.5%, respectively, while butane and Kr decreased it by 35.9% and 71.5%, respectively. Within the series of He- and H₂S-mixed fluids, S-CO₂-50%He and S-CO₂-50%H₂S were selected for detailed analysis under Delingha typical-day conditions. The duration of high specific work was found to increase progressively from the spring equinox to the winter solstice due to meteorological conditions. During high-temperature periods, the specific work of S-CO₂-50%He (286.1 kJ/kg) was significantly higher than that of S-CO₂-50%H₂S (138.2 kJ/kg) and S-CO₂ (119.4 kJ/kg), whereas at low temperatures, S-CO₂-50%He fell below S-CO₂, while S-CO₂-50%H₂S consistently provided gains across all periods.

Key words: Solar tower power plant, Brayton cycle, S-CO₂-based mixed working fluid, Intelligent evolutionary algorithm, Thermodynamic Performance