Seamless superstructure-free topology exploration for supercritical CO2 cycle optimization in waste heat recovery

Supercritical carbon dioxide (sCO2) cycles are promising for waste heat recovery, yet their performance depends critically on the structural layout of components. Superstructure-free methods can generate candidate topologies from first principles, but existing approaches have several limitations. To address these, this study proposes an enhanced superstructure-free optimization framework that introduces four key improvements: ⅰ) the levelized cost of electricity (LCOE) is adopted as an economic objective for project-level assessment; ⅱ) a topology control parameter λt enables seamless modulation among various cycle types, substantially broadening the search space; ⅲ) the heat exchanger network (HEN) optimization incorporates both nonlinear cost functions and three stream merge strategies to improve cost estimation fidelity; and ⅳ) the entropy-weighted TOPSIS method is employed for systematic multi-criteria decision making. A bi-level optimization architecture couples NSGA-II at the outer level with a global MINLP solver at the inner level. Applied to waste heat recovery from an internal combustion engine exhaust, the framework identifies solutions spanning 24.9–53.0 kW net power output and 0.119–0.277 $/kWh LCOE across nine pressure-level combinations. The TOPSIS-selected design achieves 35.7 kW at 0.135 $/kWh. The proposed framework provides a systematic and extensible tool for the automated synthesis and thermo-economic optimization of sCO2 cycles.