Knowledge-infused hierarchical causal inference framework for decoding sludge settleability to quantitatively...

Area of Science:

• Environmental Engineering
• Wastewater Treatment Technologies
• Causal Inference

Background:

• Sludge settleability, measured by sludge volume index (SVI), is critical for wastewater treatment plant (WWTP) operational stability.
• Existing research often examines SVI correlations within single processes, neglecting causal relationships across multi-process systems.

Purpose of the Study:

• To develop a framework for identifying causal relationships influencing SVI across different wastewater treatment processes.
• To provide data-driven, causality-aware operational strategies for enhancing WWTP stability and sustainability.

Main Methods:

• A knowledge-infused feature engineering and hierarchical directed causal informatics framework was applied.
• Machine learning (ML) and structural causal models (SCM) were used with data from three parallel WWTP processes: Oxidation Ditch (OD), Anaerobic-Anoxic-Oxic (A2O), and A2O-Membrane Bioreactor (A2O_MBR).

Main Results:

• The integrated ML model achieved an average R-squared of 0.83, demonstrating excellent fitting and generalization.
• Temperature significantly impacts SVI across processes, with the strongest negative average treatment effect in A2O (-0.4372).
• A2O_MBR showed superior disturbance resistance, suitable for variable climates.

Conclusions:

• Causality-aware operational strategies were developed for different processes and temperature conditions.
• Specific strategies include managing biochemical oxygen demand loading and sludge recirculation in OD at low temperatures, and maintaining mixed liquor suspended solids and pH in A2O at high temperatures to prevent issues like bulking and aging.