Decoupling ion concentrations from effluent conductivity profiles in capacitive and battery electrode deionizations...

Area of Science:

• Water treatment technologies
• Environmental science
• Materials science

Background:

• Effluent conductivity is a common but often inaccurate metric for evaluating desalination performance in processes like capacitive deionization (CDI) and battery electrode deionization (BDI).
• Accurate real-time measurement of ion concentrations is crucial for reliable desalination performance assessment.

Purpose of the Study:

• To develop and validate an artificial intelligence (AI) model for accurate real-time prediction of ion concentrations from effluent conductivity in CDI and BDI systems.
• To overcome the limitations of using direct effluent conductivity for ion concentration estimation.

Main Methods:

• Development of a random forest (RF)-based AI model.
• Validation of the RF model for predicting effluent conductivity in CDI and BDI.
• Application of the RF model to predict individual ion concentrations (Na⁺, K⁺, Ca²⁺, Cl⁻) from conductivity data.
• Evaluation of the impact of sampling intervals on prediction accuracy.

Main Results:

• The RF model demonstrated high accuracy in predicting effluent conductivity for CDI (R² = 0.86) and BDI (R² = 0.95).
• The model successfully predicted individual ion concentrations with accuracy potentially exceeding conductivity prediction.
• Prediction accuracy remained stable for sampling intervals up to 80 seconds.

Conclusions:

• An RF-based AI model provides a reliable method for predicting ion concentrations in CDI/BDI systems.
• This AI approach offers a more congruent method for real-time desalination performance evaluation than effluent conductivity alone.
• The findings support the use of predicted ion concentrations as key indicators for assessing desalination efficiency.