WASTEWATER PIPE RATING CLASSIFICATION USING PHYSICS-BASED K-NEAREST NEIGHBORS: A DATA-DRIVEN APPROACH FOR RELIABLE INFRASTRUCTURE ASSESSMENT

Aging wastewater infrastructure poses considerable challenges for municipal agencies worldwide, as pipe failures can lead to environmental contamination, public health issues, and high repair costs. Traditional rating systems for wastewater pipes often rely on empirical rules or subjective visual inspections. This study proposes an innovative physics-based K-nearest neighbors (K-NN) classification framework that integrates domain-specific fluid and structural mechanics into a data-driven pipeline. We introduce physically derived features—such as hoop stress and material stiffness—alongside corrosion and hydraulic factors. These features are weighted in the K-NN distance metric, ensuring that critical physical attributes have a proportionally greater influence on the classification outcome. Empirical results on a curated wastewater pipe dataset show that the physics-based K-NN model achieves a 92.5% classification accuracy, outperforming standard K-NN, logistic regression, and random forest baselines. This methodology offers a robust, interpretable, and scalable approach for wastewater pipe rating, guiding proactive maintenance and minimizing failures.