Investigation of off-design performance of supercritical carbon dioxide recompression cycle using a deep learning-based turbine with variable inlet guide vanes

Many control strategies have been proposed to improve the off-design performance of supercritical carbon dioxide (sCO₂) cycles caused by changes in cycle parameters. However, the component-level control scheme solely utilizes the compressor with variable inlet guide vanes (VIGV) and not the VIGV-equipped turbine for the sCO₂ cycle. In this study, we examine the potential of using a VIGV-equipped turbine in a supercritical carbon dioxide recompression cycle (sCO₂-RC) heated by molten salt to enhance the system's off-design thermal efficiency under low-load conditions. To accurately model the system's behavior, one-dimensional design and off-design models of the sCO₂-RC are developed, as simplified component models are inadequate. Following the completion of the preliminary design, we present the optimal off-design performance of the inventory-controlled system as the load changes. We then analyse the impact of the VIGV-equipped turbine on the system's performance at low load. The results show that when the load is decreased from 100% to 10%, both the system and turbine efficiencies decrease from 41.55% and 90% to 24.31% and 66.62%, respectively, and the surge margin of the compressor is significantly reduced. While the VIGV-equipped turbine improves the efficiencies of both the system and turbine at 10% load, the nozzle opening is constrained by the surge margin of the compressor. We propose a hybrid control strategy, comprising of inventory and turbine bypass, to facilitate the utilization of the VIGV-equipped turbine and enhance its efficiency while maintaining the system's performance no lower than the initial value.