Real-Time Forecasting with EPANET

Automatic on-line simulation provides the system operator with the capability to model the water distribution system in real-time, including predictive estimation of the system behavior. This is essential when performing emergency response and can greatly assist in confirming normal system performance, system trouble-shooting, improvement of system operations, and projection of the current operating scenario. This approach also allows the operator to see how the entire system is operating, instead of relying on the feedback from a few SCADA sensors placed in the network, thus allowing for any breakdown in the network to be readily identified, evaluated, and rectified. In order to implement the automatic hydraulic simulation decision support tool, which can be used by the water utility operators on real-time basis, the coupling between EPANET 2.0 toolkit and SCADA system has been developed. This coupling operates in the real-time mode and can be used for automatic model update based on the measured data, running periodical steady state simulations, and for forecasting system hydraulics with the next time interval i.e. 3 hours, 1 day, 3 days. Forecasting mode requires proper initial and boundary conditions. The system initial conditions are retrieved from the SCADA real-time database and the system boundary conditions are developed based on predefined control strategies, historical data, and estimates. The state estimate is needed for node demands and diurnal profiles, water sources, availability of pumps and valves based on the operation and control mode. Additionally, scheduled maintenance and equipment availability needs to be obtained from the SCADA system and included in the automatic setup of the hydraulic model. Forecasted results are transferred to the validation tool, which is used by the system operators, and which allows running corrective simulations. Once validated, the estimated system control variables (set-points) are uploaded back to the SCADA system and used for automatic scheduling. Similarly, alternate water supply strategies can be quickly modeled and evaluated, guiding the operator until the situation is remedied or stabilizes. The experience from development and implementation of a system, which performs real-time forecasting of hydraulic behavior of water supply and water distribution systems, is discussed in this paper. The implementation of such system is illustrated on one of the recent case studies.