Identification of heavy metal leaching patterns in municipal solid waste incineration fly ash based on an explainable machine learning approach

Abstract

The leaching risk of heavy metal (HM) in municipal solid waste incineration fly ash (MSWI-FA) leads to a big challenge for FA landfilling. In this work, the HM leaching patterns were identified via 6 highly available indices of FA, in which 160 stabilized FA samples were collected from 4 incineration plants in eastern China and an explainable machine learning approach based on boosting and game analysis was conducted to assess the leaching potentials of 6 HMs (Cr, Cd, Cu, Ni, Pb and Zn). We found that, there remained high exceeding risks of Cd and Pb in stabilized FA. In addition, the S-Cl (soluble chlorine) content was the most influential factor of the leaching behaviors of Cd, Cu, Pb and Zn, more important than pH in regard to Cu, Pb and Zn. We quantified the influence of S-Cl on the HM leaching of Cd, Cu, Pb and Zn, whereby their leaching concentrations would increase by 223.5%, 215.2%, 216.5% and 222.6%, respectively, for every 0.5 mol/L order increase in S-Cl concentration. Finally, a fast determination criterion for the FA landfill was proposed, that is, FA of which S-Cl was less than 0.412 mol/L was acceptable.

Keywords: Fly ash; Heavy metal; Leaching potential; Machine learning; Municipal solid waste.