Recurrent machine learning based optimization of an enhanced fuel cell in an efficient energy system: Proposal, and techno-environmental analysis

Abstract

Real hybrid energy systems based on developing fuel cell (FC) technology are promising in addressing energy and environmental problems. Before looking at the research investigations, the earlier review studies are briefly examined to spot any gaps in the literature. In the present study, a novel scheme of fuel cells is introduced with higher efficiency and compatibility. The exhaust gases are then introduced to the transcritical carbon dioxide cycle for the production of more power and to extract heat from the intercoolers. The system is analyzed from the multiple objective functions aspect, and the net present value method is occupied for determining the system's payback period. The important design conditions are then put to the test to seek their effect on the overall system. The genetic algorithm is applied for the sake of optimization, and the efficiency is maximized while minimization of environmental impact and the cost of the system. The results indicate that maximum ηII and minimum of LCOP and ζ at the ideal point which equals to 73.8%, 0.24 $/MWh and 0.07 kg/kWh, can be reached correspondingly. Also, by changing the selling price of electricity from 0.19 $/kWh to 0.22 $/kWh, the net present value and the payback period reach 1329000 $ and 6.24 years, respectively.