Experimental validation and support vector machine optimization of rice husk gasification for sustainable syngas production and dual-fuel engine application

Waste agricultural residues can be turned into energy-rich syngas through biomass gasification, a sustainable way to generate energy. Despite progress, challenges remain in optimizing the gasification process and predicting how much syngas will be produced. This study combines a Support Vector Machine (SVM) machine learning approach with experimental analysis of rice husk gasification. In the laboratory, a custom-built fluidized bed gasifier operated at feed rates of 10 to 18 kg/h, reaching a maximum cold gas efficiency of 64 % and a low heating value (LHV) of 5.2 MJ/Nm³ at 740 °C. The syngas produced powered a 7.5 kW diesel engine in dual-fuel mode, replacing up to 68 % of diesel fuel without significant loss of efficiency. The SVM model proved highly accurate, with a mean absolute error of 0.15 MJ/m³ and an R-squared value of 0.93. This paper offers the first comprehensive validation of an integrated rice husk gasification system with machine learning as a control method, implemented on an operating dual-fuel engine. It was established that machine learning methods could enhance biomass power plants considerably, and that the utilization of syngas from rice husk gasification could reduce CO₂ emissions by approx 12.76 metric tons per year.