Synergistic Production of Fuel-Grade Bio-Oil and Carbon Rich Biochar from Co-Pyrolysis of Spent Coffee Grounds and HDPE: Process Optimization Using Response Surface Methodology and Machine Learning

Abstract

This study investigates the modeling and optimization of copyrolysis of Spent Coffee Grounds and HDPE for sustainable biofuel production, emphasizing the influence of key process parameters, including heating rate, temperature and inert flow rate. A framework coupling Response Surface Methodology (RSM) and XGBoost algorithm achieved high predictive accuracy with minimal error. Interpretation via explainable artificial intelligence (XAI) identified temperature as the most influential parameter, followed by heating rate. At optimized conditions (A heating rate of 48.12 °C/min, temperature of 544.45 °C and inert flow rate of 183.59 mL/min.), the actual yield of bio-oil was 53.11 wt %, consistent with RSM (52.12 wt %) and XGBoost (53.04 wt %) predictions, reflecting absolute errors within 0–1 wt %. GC–MS analysis confirmed that the bio-oil contained hydrocarbons exceeding 80%. Comprehensive physicochemical analysis of the noncondensable gases (NCGs) and solid residue confirms their viability as energy vectors and highlights their prospective utilization in various industrial and environmental applications. This study demonstrates the efficacy of the copyrolysis process for the production of biofuels from SCG and HDPE. Further research is necessary to optimize process parameters, conduct a comprehensive life-cycle assessment, and evaluate the techno-economic viability of scaling up the process.