A Novel Typology for Categorizing Public Water Systems in the United States

<p dir="ltr">More than 90 percent of Americans receive their drinking water through a public water system (PWS). A PWS provides water for human consumption through pipes to at least 15 service connections or serves an average of at least 25 people for at least 60 days a year the United States Environmental Protection Agency (USEPA) classifies each PWS according to the source of their water, the number of people they serve, the duration of service and if they consistently serve the same population. [1] To maintain public health and drinking water safety, all PWS are expected to comply with the standards set by The Safe Drinking Water Act (SDWA), which authorizes the EPA to nationally regulate the maximum allowable contaminant levels for drinking water and delegates the primacy to implement and enforce them to the states and tribal authorities. To monitor compliance, states are required to periodically report system information to the EPA, which is then stored in a federally managed database, Safe Drinking Water Information System (SDWIS). [2], [3] Further, the compliance thresholds are based on the same or similar criteria used to define the system classifications, such as population served. While perhaps convenient for larger discussions, existing classifications are often too broad to make any meaningful generalizations about the systems within each group beyond the classifying criteria. Using system information from the SDWIS database, we propose an alternative classification, that group PWS into types based on a commonality in characteristic system features. To develop this classification or typology, we conducted exploratory factor analysis, on a cleaned modified version of the dataset, to extract the latent attributes within reduce the overall dimensionality. The latent attributes, were then clustered, generating characteristic basic groupings that will ultimately comprise the resulting typology.</p><p dir="ltr">This paper was presented at the 21st Computing and Control in the Water Industry Conference (CCWI 2025) at the University of Sheffield (1st - 3rd September 2025).</p>