Thermodynamics modelling and optimisation of a biogas fueled decentralised poly-generation system using machine learning techniques

Abstract

In the forthcoming era of smart energy systems, decentralised solutions are gaining increasing prominence due to their superior adaptability for interconnecting sectors, reduced inefficiencies, and environmentally friendly operation. This study introduces a new medium-scale biogas-based power plant that utilises a gas turbine to meet the energy needs of a specific locality, encompassing electricity, heating, cooling, and water supply, all whilst considering the system's environmental impact. To optimise the plant's performance, three different multi-objective optimisation scenarios employing machine learning methodologies and Greywolf algorithms with distinct objective functions are analysed. Under the base conditions, the proposed plant showcases impressive capabilities, delivering 1372 kW of electricity, 246.2 kW of heating, 293.3 kW of cooling, and 4.1 kg/s of distilled water. It operates with first and second law thermodynamics efficiencies of 72.3 % and 41.4 %, respectively, while maintaining a CO2 emission index of 0.778 kgCO2/kWh. Furthermore, the net present value and investment return period for the investment are estimated to be approximately 4.4 million USD and 4 years, respectively. Through optimisation (scenario 1) that prioritises maximising efficiency while minimising product costs and environmental impact, the following parameters are achieved: an exergy efficiency of 42.7 %, a cost of products at 28.8 $/GJ, and a reduced CO2 emission index of 0.762 kgCO2/kWh. The results reveal that the proposed system not only excels in efficiency but also proves to be economically viable and environmentally beneficial.

Keywords

• Biogas
• Economic analysis
• CO2 emissions
• Multi-objective optimisation
• Machine learning