Quasi-dynamic thermo-economic optimization and stability analysis of geothermal organic rankine cycle systems using multi-objective particle swarm optimization: an adaptive off-design approach

The efficient exploitation of low-enthalpy geothermal resources requires Organic Rankine Cycle (ORC) systems that can adapt to unavoidable fluctuations in the source. While previous studies have focused on static multi-objective optimization using Genetic Algorithms (NSGA-II), these often lack the computational speed required for real-time operational adjustments. This study investigates a custom Multi-Objective Particle Swarm Optimization (MOPSO) algorithm to optimize a low-GWP zeotropic mixture (Pentane/Isobutane) under quasi-dynamic geothermal conditions. Net electrical power output and cycle thermodynamic efficiency were simultaneously optimized, with the Levelized Cost of Energy (LCOE) mapped as a post-processed techno-economic metric along the Pareto frontier, incorporating the degree of superheating Δ T sh as a critical decision variable for turbine protection. The results indicate that the MOPSO algorithm identified a high-performing configuration with a peak net power output of 4,432.68 kW and a minimum LCOE of 0.0472 €/kWh. Under the evaluated nominal conditions, the optimization converged toward a 0 K superheating degree. Furthermore, MOPSO generated a denser Pareto solution set while reducing the number of objective-function evaluations by 75% relative to NSGA-II. Under a simulated 30 °C reduction in geothermal source temperature (from 150 °C to 120 °C), the swarm-based approach enabled adaptive Power Point Tracking and maintained a 7.9% higher power output than the corresponding static operating strategy. These findings suggest that MOPSO-based frameworks can enhance operational adaptability and support thermo-economic performance under varying geothermal resource conditions.