Process optimization of biomass gasification with a Monte Carlo approach and random forest algorithm

Gasification technologies have been extensively studied for their potential to convert biomass feedstocks into syngas (a mixture of CH4, H2, and CO mainly) that can be further turned heat or electricity upon combustion. It is crucial understand optimal gasification process parameters practical design operation maximizing the potential. This study combined Monte Carlo simulation approach, kinetic modeling, random forest algorithm predict (i.e. water content, particle size, porosity, thermal conductivity, emissivity, shape, reaction temperature) towards a maximum yield. The approach randomly generated data pool following either normal uniform distribution, which was then fed validated model create 2,000 datasets (process yields). For model, mean decrease accuracy Gini were used assess importance on yields. optimization method evaluated using coefficient determination (R2), root means square error (RMSE), absolute (MAE). Generally, predictions distribution case closer experimental obtained from existing literature than case. yield wood it shown generally in good agreement (<12% difference distribution) with results. serves as useful tool determining design.