Sentinel pipe racks quantify orthophosphate’s effect on lead release into drinking water

Abstract

Orthophosphate is used widely to control lead release from plumbing into tap water. Its effect can be difficult to quantify, though, since tap water lead concentrations are site-specific. Sentinel homes with lead service lines are ideal for evaluating orthophosphate corrosion control programs, but best practices dictate the removal of lead service lines once they are identified. Sentinel homes, then, often have too short a useful life to be used effectively. Here we explore an alternative: sentinel pipe racks constructed with recovered lead pipes and supplied with water directly from the distribution system. We also propose a strategy for analyzing pipe rack data based on the generalized additive model, which approximates time series as a sum of smooth functions. In this study, geometric mean lead release from pipe racks exhibited a pronounced dose response, falling by 54% (95% credible interval: 14–77%) after an increase from 1 to 2 mg PO4 L-1, and then climbing by 55% (95% credible interval: 5–143%) after a decrease to 1.5 mg PO4 L-1. Data from the sentinel homes were largely consistent with those from pipe racks: geometric mean lead levels at the high orthophosphate dose (2 mg L-1) were 60% of those at the low dose (1 mg L-1, 95% credible interval: 50–76%). Our results demonstrate sentinel pipe racks as a viable alternative to at-the-tap sampling for non-regulatory corrosion control monitoring. They also provide a fully Bayesian framework for quantifying orthophosphate’s effect on lead release that is well-suited to incorporating information from multiple sources.

Evaluating Sentinel Pipe Racks for Monitoring Lead Release and Optimizing Corrosion Control

“…Materials (R code and data) necessary to reproduce the main results of the paper are available online. 48 Sentinel Pipe Racks. Lead in pipe rack effluent, y t , was modeled using a robust hierarchical Bayesian generalized additive model (GAM) with continuous-time first-order autoregressive errors.…”