Optimizing Ash Content in Biomass Hydrothermal Liquefaction through Machine Learning and Sustainability Assessment

Abstract

Hydrothermal liquefaction (HTL) is a highly promising thermochemical technology for converting wet biomass into bioliquid fuels. However, the effects of ash content in process performance and sustainability remain poorly understood. This study examined acid washing, alkali washing, and feedstock blending as methods for altering the ash content in a given biomass feedstock. Machine learning tools were employed in sustainability assessment for a 200 t/day plant, using life cycle assessment and techno-economic analysis. Feedstock blending best reduced ash content (ash from 34.4 to 6.40 wt %). Ash content had opposing effects on product yield and oil quality. Across ash contents of 3.60–34.4 wt %, the oil higher heating value consistently rose, whereas oil yield initially increased then declined. The char yield followed an opposite trend. Integrated environmental and economic assessment identified 6.40 wt % ash as the optimal HTL condition for sludge and Chlorella. Under a boundary excluding biomass cultivation and harvesting, this condition gave an energy return on investment of 3.21 and a global warming potential of −317 kg CO2 eq/t dry biomass. The minimum fuel selling price was 0.517 $/L. These results show that reducing biomass ash content is not always advantageous. Feedstock blending demonstrated substantial benefits for large-scale oil production from high-ash biomass.