Can wood waste be a feedstock for anaerobic digestion? A machine learning assisted meta-analysis

Gao, Zhenghui, Cui, Tianyi, Qian, Hang, Sapsford, Devin J. ORCID: https://orcid.org/0000-0002-6763-7909, Cleall, Peter J. ORCID: https://orcid.org/0000-0002-4005-5319 and Harbottle, Michael J. ORCID: https://orcid.org/0000-0002-6443-5340 2024. Can wood waste be a feedstock for anaerobic digestion? A machine learning assisted meta-analysis. Chemical Engineering Journal 487 , 150496. 10.1016/j.cej.2024.150496

PDF - Published Version Available under License Creative Commons Attribution. Download (2MB)

Abstract

Anaerobic digestion is widely employed to process various organic wastes while generating renewable energy and nutrient-rich digestate. However, lignocellulosic wastes, especially wood waste, suffer from the recalcitrance associated with high lignin content, thereby adversely impacting on biogas production. It remains unclear whether wood waste is suitable as a feedstock for anaerobic digestion and to what extent pretreatment techniques could affect its biochemical methane potential. In this paper, 769 datasets on methane production from wood waste were collected for meta-analysis. The results showed an average 146 % increase in methane production for other organic wastes compared to wood waste when pretreatment techniques were not applied, but this gap could be mitigated to 99 % when pretreatment techniques were considered, indicating that pretreatment techniques could be more effective for wood waste. A further analysis of different pretreatment techniques showed that pretreatment significantly increased the methane production of wood waste by 113 % and that a combination of pretreatment techniques was more effective than a single method. Finally, three machine learning algorithms were applied to explore the relationship between methane production and selected variables. The results showed that the random forest method yielded better predictive performance for methane production (R2 = 0.9643) than artificial neural networks and support vector regression. Feature importance analysis found that particle size had a higher influence than temperature or feedstock composition. Overall, this study gives insight into the potential of utilizing wood waste as a feedstock for anaerobic digestion and the importance of employing suitable pretreatment methods. This work also reveals correlations between methane production and critical variables, which could serve as a guide for optimizing operational adjustments during anaerobic digestion.

Item Type:	Article
Date Type:	Publication
Status:	Published
Schools:	Engineering
Publisher:	Elsevier
ISSN:	1385-8947
Date of First Compliant Deposit:	18 March 2024
Date of Acceptance:	16 March 2024
Last Modified:	03 Apr 2024 11:15
URI:	https://orca.cardiff.ac.uk/id/eprint/167338

Actions (repository staff only)

Edit Item