Simulation and optimization of multiple permeable reactive barriers (multi-PRBs) for acid mine drainage (AMD) based...

Area of Science:

• Environmental Engineering
• Water Treatment Technologies
• Geochemistry

Background:

• Acid mine drainage (AMD) poses significant environmental challenges.
• Permeable reactive barriers (PRBs) are a promising in-situ treatment technology for AMD.
• Current implementation of multi-PRBs is limited by a lack of mechanistic understanding and decision-making tools.

Purpose of the Study:

• To develop a coupled processes numerical model for simulating synergistic TFe and SO₄2⁻ removal in multi-PRBs.
• To propose a machine learning-integrated optimization approach (ML-NSGAII) for efficient multi-PRB design.
• To identify key design parameters and establish a practical design strategy for AMD treatment.

Main Methods:

• Development of a coupled processes numerical model for multi-PRB simulation.
• Integration of a Backpropagation Neural Network (BPNN) as a surrogate model within the Non-dominated Sorting Genetic Algorithm (ML-NSGAII).
• Application of Spearman correlation, SHAP analysis, and TOPSIS-entropy weight method for parameter identification and optimal design selection.

Main Results:

• The ML-NSGAII approach reduced computational load by 99.7% with a highly accurate surrogate model (R² > 0.99).
• Hydraulic load and filler size were identified as the most influential design parameters.
• The optimized multi-PRB design, using limestone, biochar, and D201 resin, significantly improved service life and treatment capacity while reducing costs.

Conclusions:

• A computationally efficient and practical strategy for designing multi-PRBs for AMD treatment has been established.
• The developed ML-NSGAII framework offers a robust tool for optimizing complex environmental remediation systems.
• This research provides valuable insights for the effective implementation of multi-PRBs in acid mine drainage remediation.