Geothermal-driven polygeneration plant integrating air separation, LNG cooling, ORC, and PEME: Thermo-Enviro-Economic analysis and ML-based optimization

In this study, a novel thermally-integrated polygeneration plant is proposed that combines an integrated air separation unit, proton exchange membrane electrolyzer, liquefied natural gas (LNG) cooling cycle, geothermal binary power cycle, and organic Rankine cycle. The plant can generate electricity, hydrogen, liquid oxygen, cold and hot water all at once through extensive thermal integration between LNG cold energy and geothermal heat resources. Thorough thermo-enviro-economic analyses are performed, and a hybrid machine learning–multi-objective optimization framework combining CatBoost and the Jellyfish Search Algorithm (JSA) is employed to determine optimal operating conditions. As a result, the base plant generates 9706 kW of net power, 3196 kg/h of liquid oxygen, 16.13 kg/h of hydrogen, 3386 kW of heating, and 2066 kW of cooling. The plant achieves an energy efficiency of 18.88% and an exergy efficiency of 47.11%, with a sustainability index of 1.891 and an exergo-environmental impact improvement of 1.123. Economic evaluation indicates a net present value (NPV) of 61.595 M$, levelized cost of energy of 0.019 $/kWh, total unit cost of products of 7.21 $/GJ, and a payback period of 2.25 years. The best two-objective optimization results demonstrate further performance improvements, increasing the exergy efficiency to 49.75% while reducing the NPV to 78.12 M$. Sensitivity analysis reveals that stream pressure P32 is the most influential parameter affecting the thermo-enviro-economic performance of the system. The proposed configuration demonstrates strong potential for efficient and sustainable multi-energy production in regions with accessible LNG and geothermal resources, and it has significant potential for industrial zones and urban energy infrastructures.