A novel hybrid biogas–solar-driven energy system integrated with carbon capture for multi-generation: Machine learning-based technical, economic, and environmental optimization
Аннотация
• A novel hybrid biogas-solar multi-energy system with carbon capture is proposed. • A comprehensive thermodynamic, economic, and environmental analysis is performed. • Machine learning optimization significantly improves performance and cost-efficiency. • The optimization reduces LCOE to 9.31 cents/kWh and PP to 4.65 years. The growing global demand for clean energy, freshwater, and hydrogen—alongside the urgent need to reduce carbon emissions—necessitates the development of efficient multi-generation systems. However, existing solutions often suffer from limited resource coupling, inefficient energy recovery, and high environmental and operational costs. This research introduces an innovative hybrid biogas–solar-driven multi-energy system capable of simultaneously generating electricity, freshwater, hydrogen, and cooling, while integrating post-combustion carbon capture (CC) to mitigate CO 2 emissions. The system combines an organic Rankine cycle (ORC) and a Kalina cycle (KC) to enhance power generation from biogas, while utilizing waste heat through an ejector refrigeration cycle (ERC) and a humidification–dehumidification (HDH) unit. To enhance environmental sustainability, the system integrates a solar-assisted monoethanolamine (MEA)-based CC unit, effectively mitigating CO 2 emissions and minimizing the system’s carbon footprint. Thermodynamic, economic, and environmental analyses are carried out, supported by sensitivity and parametric studies. The baseline results indicate that the system produces 2,350 kW of electricity with a levelized cost of electricity (LCOE) of 10.74 cents/kWh. Integrating parabolic trough solar collectors (PTSCs) into the CC process provides renewable heat for solvent regeneration, reducing fossil fuel dependency and enhancing cost-effectiveness. Additionally, decreasing the lean/rich heat exchanger temperature difference from 15 K to 5 K lowers reboiler duty from 3.6 to 2.95 MJ/kg CO 2 and reduces LCOE. Given the complexity of system interactions, a machine learning-based optimization method—integrating genetic algorithm (GA) with artificial neural networks (ANNs)—is employed to determine optimal design parameters. The optimized system reduces the payback period from 6.09 to 4.65 years and increases the net present value (NPV) from $12.55 million to $15.09 million over a 20-year lifetime.