Modeling optimization and application of modified drinking water sludge for high-efficiency phosphorus removal

Area of Science:

• Environmental Science
• Materials Science
• Data Science

Background:

• Phosphorus pollution drives water eutrophication, necessitating advanced removal techniques.
• Adsorption is a key method for phosphorus removal, avoiding secondary pollution.
• Drinking water treatment sludge (DWS) is a low-cost material for adsorbents, but its modification is inefficient.

Purpose of the Study:

• To develop a data-driven approach for optimizing DWS modification for enhanced phosphorus removal.
• To evaluate machine learning models for predicting the performance of modified DWS.
• To design a novel adsorbent using optimized modification parameters.

Main Methods:

• Utilized 657 datasets from 14 sources to train and evaluate six machine learning models.
• Developed an improved Hybrid Encoding Genetic Algorithm (HEGA) for inverse optimization of modification parameters.
• Prepared calcium-aluminum layered double oxides (Ca-Al LDOs) using Gradient Boosting Decision Tree (GBDT)-HEGA optimization.

Main Results:

• GBDT model achieved the highest prediction accuracy (R² = 0.992).
• Optimized Ca-Al LDOs demonstrated 96.81% phosphate removal, achieving effluent concentration below China's Class III Surface Water Standard.
• High removal efficiency (87.09%) was maintained in actual lake water tests, with a low cost of $0.8 per kg phosphorus removed.

Conclusions:

• The GBDT-HEGA framework provides an efficient, data-driven method for optimizing environmental processes.
• The developed Ca-Al LDOs offer a feasible, low-cost solution for advanced phosphorus removal in water treatment.
• This approach overcomes traditional trial-and-error inefficiencies, showing significant practical value for controlling water eutrophication.