Jian Li

Growing demand for green methanol as a low-carbon marine fuel calls for efficient conversion of high-moisture biomass residues. In this study, a semi-continuous anaerobic digestion (AD)–digestate gasification system was investigated using distillers’ grains (DGs), with AD extent regulated by total solids (TS) content and organic loading rate (OLR) coordinate biogas production and residual digestate gasification. AD was conducted at TS contents of 5%, 10%, and 15% and OLRs of 2.0, 3.5, and 5.0 g VS·L-1·d-1, followed by digestate characterization and gasification. The resulting biogas and digestate-derived gasification gas were further integrated through methane reforming and syngas conditioning for methanol synthesis. A TS content of 5% and an OLR of 5.0 g VS·L-1·d-1 provided the optimal balance between anaerobic conversion and subsequent gasification. AD extent was interpreted as a carbon redistribution process, in which carbon allocation between biogas and digestate-derived gasification gas determined the suitability of the combined gaseous feedstock for methanol synthesis. Based on process simulations constrained by the experimental results, 1 t of fresh DGs could produce approximately 0.36 t of methanol, corresponding to a biomass-to-methanol conversion efficiency of 36.41%. The dry reforming of methane based conditioning route produced 910.30 Nm3 of conditioned syngas per ton of fresh DGs, with carbon and hydrogen utilization efficiencies of 74.71% and 43.20%, respectively and achieved better methane yield, carbon–hydrogen coordination and element utilization than steam reforming. These results demonstrate that integrating controlled AD with digestate gasification provides an effective strategy for optimizing green methanol production from high-moisture biomass residues.

Aerobic composting of biogas residue is critical for sustainable resource recovery. This study established a machine learning model to predict composting indexes (germination index, C/N_Final) and identified optimal operational conditions. The Random Forest algorithm was used to construct the machine learning model. The model was optimized through hyperparameter tuning and k-fold cross-validation, it achieved >94% accuracy across five performance indexes. Dual-method feature importance analysis identified feedstock composition (C/N_Inital, moisture content, organic matter) as the dominant factor, contributing 45.76% to composting efficiency. Experimental validation under model-predicted optimal conditions confirmed robust performance: high-benefit compost exhibited >95% agreement between predicted and actual values (germination index, C/N_Final), with compost maturity meeting international standards. This work provides a validated, data-driven strategy to maximize biogas residue valorization through precision composting.