Machine learning-based optimization and 4E analysis of renewable-based polygeneration system by integration of GT-SRC-ORC-SOFC-PEME-MED-RO using multi-objective grey wolf optimization algorithm and neural networks

Abstract

This study introduces a new solar-methane-driven setup that recovers waste heat from flue gases to produce electricity, hydrogen, oxygen, fresh water (FW), and domestic hot water (DHW) while reducing the environmental impact of gas turbines. This integration includes the Brayton cycle and solid oxide fuel cell as motive cycles, the steam and organic Rankine cycle, a parabolic trough solar collector (PTC), a proton exchange membrane electrolyzer, multi-effect distillation, and reverse osmosis. Using Engineering Equation Solver software, analyses in energy, exergy, exergoeconomic, and exergoenvironment (4E) are conducted to identify critical system evaluation parameters in thermodynamics, economics, and the environment. Additionally, the proposed system is optimized using an artificial neural network and the multi-objective grey wolf optimizer algorithm, utilizing decision variables identified through sensitivity analysis to enhance system performance. The results of six-objective optimization show energy and exergy efficiencies at 49.48 % and 47.21 %, with a net power production of 133 MW and a total cost rate of $7903/h. Among all the components, the combustion chamber and PTC have the maximum exergy destruction, with values of 58.87 MW and 17.29 MW, respectively. Also, hydrogen, FW, and DHW production rates are 201.6 kg/h, 43.88 kg/s, and 40.98 kg/s, respectively. Moreover, the costs of hydrogen, FW, and DHW are $953.64/h, $57.35/h, and $82.48/h, respectively. Finally, Bloemfontein, Las Vegas, and Tunis are the most favorable regions economically, with a low levelized cost of electricity of $0.05931, $0.06006, and $0.05932 per kWh, along with payback periods of 0.6144, 0.5524, and 1.069 years, respectively.