Proactive control of water distribution systems
Rising temperatures, changes in precipitation patterns and the increase in dry periods and droughts are influencing the water balance and the available water supply. As a result, flood and low water events and the associated weakening of ecosystems will occur more and more frequently. At the same time, the predictability of such events is becoming increasingly difficult.
As part of the water infrastructure, dams fulfil a variety of objectives such as low water elevation and flood protection. During dry periods, they ensure the release of water into the lower reaches and reduce the risk of flooding during heavy rainfall. The retained precipitation water can also be treated to produce drinking water and thus potentially reduce groundwater extraction for drinking and process water supply at times. This protects the landscape water balance of a particular location. In this way, they safeguard ecosystem services and the population from the effects of extreme events, while at the same time renewable energy can be generated using hydropower. With regard to climate change adaptation, there are a large number of synergy options between flood and low water risk management in reservoir management. Efficient control approaches for dams and water distribution networks are therefore essential for climate impact-orientated operation.
Previous (semi-)rigid control rules, in which the current system is detected via a network of sensors and measurements and then brought into the desired system state, only enable proactive measures to a limited extent. However, a prediction-based control system that reacts to future events is dynamic climate impact adaptation.
The aim of the project is the model-based implementation of a demonstrator for prediction-based quantity- and quality-weighted (dam) control based on data-driven models. Predictions of water demand and supply are generated using ML methods. These predictions are then integrated into a hydrodynamic optimisation model, which is created as part of the project. The output of the model is a decision support system for controlling the dam-water distribution network based on the predictions and the various prioritised multiple objectives (flood protection, security of supply, ecosystem services and landscape water balance, energy generation). The focus here is on the uncertainties resulting from the modelling process and external influencing factors as well as on various forecast horizons that are relevant to the water management objectives. With the help of the demonstrator, the resulting advantages over conventional control can be emphasised by means of a multi-criteria evaluation.
