Using a Digital Twin to Control Sewer Overflows - Sweden

With five municipalities Burlov, Eslov, Lomma, Lund and Malmo and half a million customers, VA SYD is one of Sweden's largest agencies for water distribution and sewage collection and treatment. VA SYD need a digital solution to minimize the impact of flow variations due to extreme weather events, improve water quality in receiving water bodies, attenuate flows to the wastewater treatment plant, and reduce sewer overflows. Also, if sewer overflows could not be avoided, the agency needed to control where the sewer overflows occur in order to minimize environmental impacts. VA SYD implemented a digital twin hydrologic and hydraulic model in the city of Lund, covering the entire service area draining to the Källby wastewater treatment plant. The digital twin was developed to allow for on-line simulation and forecasting capabilities with weather adaptive control to meet the goals and objectives of VA SYD. Phase 1 of the project included implementation of the digital twin with a data connection to existing meters and gauges, SCADA system and weather forecasts. The existing collection systems model was calibrated against flow measurements both in terms of inflow components in connection to rainfall events and infiltration from groundwater. Phase 1 included forecast calculations for the upcoming 24 hours. The forecasts were carried out automatically once an hour. The digital twin system displayed measured flows and levels as well as calculated flows and levels at selected points along the collection system. The system displayed actual and calculated sewer overflow volumes, as well as water volume in the network and basins All displays were both hindcast and forecast. Finally the digital twin user interface also displayed estimated run-off from each sub-area from each type of flow components, i.e. inflow versus infiltration. Data assimilation, detailed weather forecasts and control strategies were developed and implemented during Phase 2 of the project. Data assimilation was applied to the calibrated collection systems model to deal with residual anomalies. This means the digital twin continuously analyses the deviations between measured and calculated values and compensates for them up to the time of the forecast. The deviations are interpreted and can then be used during the forecast calculation via an 'error forecast'. In this way, the digital twin model calculation is corrected both in the historical part and in the forecast. The digital twin model's weather forecasts were upgraded to include a hybrid between C-band radar and several weather models, where the most likely forecast is selected. The forecast horizon is 36 hours with hourly resolution and spatial resolution of 2 km. A new forecast is produced every hour. During the first few hours, primary radar data is used while the weather model gradually takes over. The data grid over the City of Lund was interpolated into each sub-catchment of the collection system model. This allows each catchment to get its own unique rain series in the forecast horizon, which is combined with measured rainfall data. Within the City of Lund, there are two major basins and a basin at the pumping station in Dalby. How and when to use these basins most efficiently was evaluated based on what to achieve. Based on these evaluations, decisions were made on what type of control strategies VA SYD wanted to deploy and what conditions that were available with the current control measures in place. The selected control strategies and conditions for automated selection between were programmed into Future City Flow as a scheduled separate forecast calculation. The results of these calculations were then used as setpoints in the operative control of available control equipment. Procedures for extracting setpoints and transmitting them to both VA SYD's SCADA system and directly to the PLC via OPC UA was implemented.