Simulation models of microbial community, pH, and volatile fatty acids of anaerobic digestion developed by machine learning
Published in Journal of Water Process Engineering, 2024
Recommended citation: Yu, S. I., Jeong, H., Shin, J., Shin, S. G., Abbas, A., Yun, D., Bae, H., & Cho, K. H. (2024). Simulation models of microbial community, pH, and volatile fatty acids of anaerobic digestion developed by machine learning. Journal of Water Process Engineering, 60, 105225. https://doi.org/10.1016/j.jwpe.2024.105225
Anaerobic digestion turns organic waste into biogas through communities of microbes working without oxygen, but the process can turn sour and stall. Machine learning was used here to simulate three things at once — which microbes are present, the acidity, and the acids that pile up when digestion goes wrong — and to establish how the three are linked. Comparing five modelling approaches, the key organisms and chemical indicators were reproduced with better than 80% of their variation explained. Running the resulting simulator forward showed that shifting the microbial mix alone swings acetate accumulation between 500 and 1,500 mg/L. From that the study derives a concrete operating target: holding two families of methane-producing microbes at 15–37% and 3–15% of the community keeps acetic acid below 1,000 mg/L.