Accuracy of Real-Time Distribution System Models for Predicting Field Scale Tracer and Chlorine Transport AWWA Water Quality Technology Conference

摘要

The Greater Cincinnati Water Works (GCWW) has been pushing the limits of distribution systemrnhydraulic and water quality monitoring and modeling over the past decade. Recent work hasrnincluded the development of a complex all-pipes distribution system model, hydraulic andrnwater quality model calibration activities, and on-line water quality monitoring at facilitiesrnthroughout the service area.rnOne practical problem has, however, plagued network water quality models ever since theirrndevelopment more than 20 years ago. Unlike hydraulic models, which can usefully representrncritical or average steady-state network conditions for master planning, water quality modelsrnare inherently dynamic, because quality evolves in response to hourly scale operationalrndecisions and consumer demands. As a result, water quality models have been developed andrncalibrated to particular time ranges, say several weeks, during which field data are collected.rnThe “calibrated” water quality network models are rarely verified during different operationalrnperiods, and their use for predictions — during different seasons, or as a result of differentrnoperational protocols — has been limited.rnOver the past year, GCWW has invested in modeling and software technology to leveragernprevious work on network models, and explicitly link the model inputs to SCADA data streams.rnThese “real-time” network models promise to circumvent the problems summarized above,rnwhereby network water quality models are restricted to the data and conditions used torndevelop them. By providing a direct, persistent, and fully automated connection to SCADA datarnstreams, the GCWW real-time models are continually updated to reflect changes in waterrndemand and actual operational practices. Such models, with proper software, can be used to provide predictions of both historic and current conditions, without the additional manual workrnrequired to import and “clean” SCADA data streams, calculate demands using spreadsheets,rnand manually convert the operational record into operational controls.rnThe current status of the GCWW real-time modeling project is that software has beenrnconfigured for the current distribution system infrastructure and SCADA data streams. Thisrnsoftware is currently undergoing an evaluation for hydraulic model accuracy, by comparingrnhistorical SCADA data to associated modeled values, over significant time periods. The next steprnis to leverage the real-time hydraulic model to provide automated predictions of tracerrnmovement for field tracer tests conducted at six different distribution system locations duringrnSpring 2014, and also to simulate chlorine residual for comparison to measured values at onlinernmonitors in the distribution system. These comparisons will provide a crucial benchmark as tornthe accuracy of real-time model predictions using current technology. We describe the keyrnsetup and configuration steps of the real-time model below, and will report objectively on thernaccuracy metrics from hydraulic and water quality comparisons, at the WQTC presentation.