Addressing Drinking Water Contamination: A Case Study Comparing Traditional with Model-Based Approaches

Rapid and effective decision-making is crucial during drinking water contamination events to ensure public safety. This paper examines a case study where a water utility, responding to customer complaints, suspected wastewater contamination in its network. We compare the traditional expert judgement approach to a model-based approach using the PathoINVEST tool. The tool performs simulations of contamination events informed by sensor measurements, identifies contamination sources using sampling results, and suggests optimal valve closures for mitigation. Our findings show that the model-based approach significantly enhances response efficiency and accuracy. It identified the contamination source with four samples in 1.3 h, compared to 11 samples in 3.7 h for the traditional approach, and resulted in a lower infection risk (12% versus 20%) at the time of source identification. Regarding valve closure, the model-based approach performed better, resulting in a 3%-point reduction in infection risk compared to the traditional approach. Modeling uncertainty is addressed by considering valve settings uncertainty; despite a 0.7% discrepancy in valve settings compared to the model, the tool accurately pinpointed the contamination vicinity 75% of the time. These findings support the claim that integrating modeling and sensor tools into emergency response protocols for drinking water contamination events can improve early identification and mitigation, potentially safeguarding public health in urban water supply systems.