The management and operation of large raw water conveyance systems (Demo Case #1) will be improved developing a water quantity routing application providing (offline) what-if hydraulic scenario assessment for optimal water conveyance using real time data. The application will combine physically based hydraulic models
with data driven technologies (incl. surrogate-based optimisation models) so that the hydraulic solutions to what-if scenarios requested by operators in real time, take advantage of (almost) current information from the flow sensors deployed in the conveyance system, while performing the necessary calculations fast enough to be of use to operators. Scenarios will include advice towards alternative routes for conveyance when parts of the network are to be closed for maintenance (or are malfunctioning) and operational rules for actuators at sluice gates to enforce this advice if the operator approves.
In addition, the early detection of and optimisation of the reaction to water contamination events in raw water aqueducts (Demo Case #1) will be improved by the development of a real time water quality early warning application providing advance warning of high turbidity water in raw water supply aqueducts before it reaches the treatment plants. This application will fuse information from different sources including: hydro-meteorological stations at the raw water reservoir areas, satellite data, drone observations and sensor data from water quality sensors deployed in the conveyance system with hydraulic information from the flow sensors and the water quantity application developed in the same sub- task (above) to provide initially a medium-term (1-2 days) forecast of the probability for high turbidity water to be present at the raw water sources, and then subsequently an alert (3-5h in advance) if high turbidity water is actually observed in the conveyance system with an estimate of the time to treatment plant. Particular attention will be given to quantifying the related belief in the prediction related to uncertainties in the data and models using Evidence Theory.
