Using Sensitivity Analysis and SCADA/Model Comparisons to Diagnose Model and System Issues

Water distribution system computer modeling has long been a vital tool for system operations, capital planning and engineering design. Water distribution models typically rely on data collection from SCADA systems to provide reference data for model calibration. However, by loading snapshots of SCADA system data directly to the hydraulic model, both operational and model calibration issues can be diagnosed. A comparison of model predictions based on real-time operating conditions and SCADA flows and pressures allows utility personnel to immediately identify situations where the model and SCADA data diverge. Where the SCADA data and model prediction diverge, the problem may be related to an unknown operational condition or change, a dysfunctional SCADA device or inappropriate model parameters. These situations can be quickly diagnosed by observing the historical to present trends in the data and performing model sensitivity analysis to determine the relative impact of changing various model parameters, including operational constraints. The resulting diagnoses may improve the distribution system model, guide SCADA equipment repair or calibration work orders or suggest potential changes for more effective distribution system facility operation. This paper investigates case studies where SCADA snapshots were uploaded to the hydraulic model to set the boundary conditions for operating parameters, such as valve positions, tank levels and pump operational statuses. Both SCADA and model historical values were archived for system points where pressures and flows are recorded in the SCADA system, making the data trend for each viewable within the application. These trends were utilized to evaluate the model predictions versus SCADA records. Cases are presented where these trends diverge along with appropriate conclusions regarding system operation, the current state of the SCADA device and/or the hydraulic parameters of the model.