A flexible retrofitting method for thermal-energy-storage-coupled thermal power units is proposed. The exergy flow Sankey diagram and efficiency of the three charging methods was analyzed in detail, and comparative data were provided. The new thermal energy storage mode coupled with steam ejectors was the optimal solution, with a significantly higher round-trip efficiency than the published results. Further study of the mode revealed slight temperature variations in the molten salt did not affect system efficiency. As the extraction of reheated steam decreases and the extraction of main steam increases, the minimum generation capacity of the unit decreases continuously, reaching a minimum of 40 MW (6.67%). In contrast, the maximum output of the unit at 100% THA release can reach 682 MW, which caused the unit load to change from 26-100% to 6.67–113.67% before and after the retrofit. As the ejection coefficient increased, the round-trip efficiency increased significantly, corresponding to 73%, 77%, and 80% at the study points. The mathematical model based on principal component analysis can effectively characterize the relationship between the ejection coefficients, extraction of main steam, extraction of reheat steam, low valley peaking rate, peak modulation rate, and round-trip efficiency, providing a signal conversion basis for automatically controlling peaking commands during unit operation. Finally, the benefits before and after retrofitting were analyzed in terms of thermal plant operation, proving that the payback period of the system was 4.09 years. Peak tariff and auxiliary service revenue are the most critical parameters affecting the payback period, and a 10% reduction in the peak tariff corresponds to an increase in the payback period of one ∼ two years. The results of this study provide substantial theoretical and practical guidance for the flexible retrofitting of thermal power units.