Multi-objective optimization of the RBC-PTES system using s-CO2 mixtures as the working fluid

• A multi-objective optimization (NSGA-II) was performed on an RBC-PTES system. • The CO 2 /Xe mixture exhibited the lowest LCOS of 133.6 $/MWh, while the CO 2 /Kr mixture achieved the highest round-trip efficiency η rt of 61.2 %. • Exergy analysis revealed irreversibilities primarily in the turbomachine, while recuperators (LTR and HTR) presented the highest efficiencies. The increasing penetration of renewable energy sources requires the development of efficient and economically viable storage technologies to ensure grid stability and operational flexibility. Pumped Thermal Energy Storage (PTES) based on the Recompression Brayton Cycle (RBC) has emerged as a promising solution, particularly when employing supercritical carbon dioxide (s-CO 2 ) and its mixtures as working fluids. This study proposes a multi-objective optimisation of an RBC-PTES system, employing the Non-Dominated Sorting Genetic Algorithm II (NSGA-II), with the dual objectives of maximising round-trip efficiency and minimising the Levelized Cost of Storage (LCOS). An evaluation was conducted in which two s-CO 2 mixtures, CO 2 /Kr and CO 2 /Xe as working fluid, were assessed against pure CO 2 . The findings reveal mixtures improve the behaviour of the system over pure CO 2 : the CO 2 /Xe mixture attains the lowest level of LCOS (133.6 $/MWh), while CO 2 /Kr attains the highest round-trip efficiency (61.2 %), thereby highlighting a discernible trade-off between cost-effectiveness and efficiency. With this results it is demonstrated that the utilisation of mixtures based on s-CO 2 as a working fluid enhances the technical–economic and exergetic performance of RBC-PTES systems. A detailed exergetic analysis revealed that irreversibilities are mainly concentrated in the high-temperature recuperator and compression stages, suggesting opportunities for targeted design improvements. Furthermore, an electricity price forecast was developed using a GA-optimized bidirectional LSTM model and successfully captured a sustained upward trend in last year energy prices. A relevant finding of this work is that the incorporation of this forecast into the economic assessment demonstrated the strong dependency of LCOS on electricity market variability, thereby confirming that dynamic price conditions have the capacity to significantly alter system competitiveness.