- Energy trading
- Energy solutionsAsset managementEnergy management
- Power generation
‘Water for Alpiq is like flour for the baker,’ says Jonathan Fauriel, Head Civil Engineering and Environment. As Switzerland’s second largest producer of electricity from renewable energies, we have a great responsibility to handle water sustainably and to know how to use it most effectively. To do this, reliable forecasts of inflows into the reservoirs are crucial. ‘We need to know what amount of water is available in order to avoid wasting it and to be able to use it fully,’ says Jonathan. This knowledge gives an electricity producer like Alpiq planning security while also enabling a secure supply.
To avoid water loss. If there is unexpected excess water inflow due to high precipitation or high temperatures (ice melting), reservoirs and water intake facilities can overflow and up to 40 GWh of electricity can be lost annually for Alpiq.
To optimise turbination times. For seasonal planning, it is important to know how much water is available for power generation.
To reduce the cost of pumping energy. The time when water is pumped back into the reservoirs as well as how much can be better planned in advance.
To optimise ancillary services. The power plants can be better planned for the use of ancillary services for Swissgrid and can stabilise the grid more reliably.
To minimise the need for balancing energy. This reduces the gap between the predicted amount of electricity already sold on the market and the amount of electricity available at the time of delivery. The financial and physical risk is minimised.
To enhance safety below the dam. Floods can be prevented or the population can be warned in time.
‘Qualitative predictions are relatively easy. Water just comes,’ Jonathan jokes and immediately adds:
For a forecast of water inflows, vast amounts of raw data must be collected, processed, prepared and integrated into the existing control systems and modelling. This includes weather forecasts, data from the weather and radar stations, data about the flow rates of individual power plants and current water levels of the lakes. With all this data, we try to determine the water quantities as precisely as possible in order to use them optimally for power generation and marketing. ‘To do this, we make use of all available tools and resources,’ says Jonathan resolutely, adding that models based on artificial intelligence are also included. Alpiq is currently pursuing two projects, Radar4Infra and Defrost for Hydropower, which are based on radar measurements and satellite data.
The Radar4Infra project, supported by Innosuisse, aims to improve all components of the forecast and was actually developed to provide more reliable flood risk warnings and quantitative flow estimations.¬ The forecasts are based on a combination of radar signals, weather forecast models and development scenarios. The radar measures the drop size of the rain, the types of snow crystals or the presence of hailstones in the clouds and derives the precipitation intensity from this. Machine learning is to be used to achieve ever greater accuracy in the modelling. The Radar4Infra project focuses on smaller geographical units and catchment areas and aims to provide a response time of just a few minutes – ideal for short-term forecasting of reservoir inflows. The first results of the project should be available in mid-2022.
The Defrost for Hydropower project is based on satellite data. It combines the SLF’s snow expertise with satellite imagery and geodata technology from Wegaw and modelling from Hydrique Engineers to predict water inflow up to four months in advance. In the process, satellite information on snow water equivalent (SWE) is converted and integrated into a seasonal hydrological inflow forecast. Partners are the Swiss Federal Office of Energy and the electricity producers FMV SA, SIG, Groupe E, Romande Energie, Alpiq, EnBAG AG, OIKEN, EnAlpin and the Etat de Genève. The methodology was successfully tested in two watersheds at the beginning of 2021. By mid-2022, the simulated inflow data will be compared with information from all participating electricity producers in order to identify opportunities for improvement in production and trading.
In both projects, the aim is to integrate the data optimally into our existing systems and to advance the modelling so that not only water as a resource is used in the best possible way, but the electricity is also marketed profitably. Despite all the promising attempts to perfect the forecasts, Jonathan retains a healthy pragmatism: ‘We can optimise the forecasts, but ultimately we can’t influence the weather. We can only be better prepared for it.’