Predicting the environmental risks of potentially toxic metal(loid)s in mechanochemically treated fly ash using...

Area of Science:

• Environmental Science
• Materials Science
• Data Science

Background:

• Municipal solid waste incineration fly ash (MSWIFA) poses environmental risks due to potentially toxic metal(loid)s (PTMs).
• Mechanochemical (MC) treatment offers a green and efficient method to mitigate these risks.
• Accurate prediction of PTM environmental risks and optimization of MC treatment conditions are crucial but challenging.

Purpose of the Study:

• To develop and validate machine learning (ML) models for predicting the environmental risks of PTMs in MC-treated fly ash.
• To identify key factors influencing PTM environmental risks and optimize MC treatment parameters.
• To create a practical tool for rapid risk assessment and process optimization in engineering applications.

Main Methods:

• Six machine learning (ML) models were employed to predict the Overall Pollution Toxicity Index (OPTI) of PTMs in MC-treated fly ash.
• The eXtreme Gradient Boosting (XGB) model was selected for its superior predictive accuracy (R²=0.986 training, R²=0.921 test).
• Feature importance analysis identified key predictors, and a graphical user interface (GUI) was developed.

Main Results:

• The XGB model demonstrated excellent predictive accuracy and generalization for environmental risks of PTMs.
• Additives (48.2%), MC conditions (28.7%), PTM properties (21.3%), and fly ash composition (1.8%) were ranked by influence.
• Key features included MC treatment time, initial concentration, specific additives (Ca, P, Ca-P, Si-Al based), leaching pH, and Cl.

Conclusions:

• A data-driven framework using ML effectively predicts environmental risks in MC-treated fly ash.
• The developed GUI provides practical tools for risk assessment and process optimization.
• Developing green and efficient additives is the most effective strategy for managing PTMs in MSWIFA.