When energy meteorologists make false predictions, it can have an expensive impact on short-term trading on the energy exchange, for instance in cases when much less wind energy is available than was forecast and demand exceeds the supply. The expansion of volatile renewable energies means increased demand for electricity for e-mobility and heat pumps, but also increased uncertainty of electricity supply. Thus, the energy transition can only work if we succeed in synchronizing energy production in terms of both time and place. In order for the electricity market to take energy flows into account and react to changing environmental influences, smart grids must know exactly how much energy is being consumed where as well as how much is being produced, and must be able to translate this information into flexibility signals. That’s why weather forecasts and yield forecasts for photovoltaics and other power generation systems, along with load forecasts on the consumption side, are so important. Thanks to digitalization and artificial intelligence (AI), these forecasts are becoming more and more accurate.
Calculating disruptions
meteocontrol GmbH has turned to machine learning to improve their solar power forecasts when photovoltaic installations are covered in snow. When this is the case, the installations cannot produce any electricity, even if the sun is shining. It is also impossible to predict when the snow will melt or thaw out and slide off a given module. The company has developed a correction model for snowy conditions and trained it with monitoring data from 50,000 photovoltaic installations.
Even ash and dust particles or grains of sand present in the atmosphere can unravel the best forecasts, as they decrease solar irradiation and thus photovoltaic yields. A team of researchers from the Karlsruhe Institute of Technology, Germany’s National Meteorological Service – Deutscher Wetterdienst – and meteocontrol have developed a joint project for investigating the effect minuscule particle matter has on power generation using measurement data from weather stations as well as satellite data. In the future, power grid operators are set to be able to use the research findings in the form of new forecast models.
Using grid analysis to identify bottlenecks early on
Better forecasts also help grid operators meet the new legal requirements for managing grid congestion (Redispatch 2.0). Operators are required to use predicted load profiles for generation plants and electrical devices to prepare grid analyses that can help identify congestion early on, allowing for preemptive measures to be taken. Such an approach requires uniform standards for data exchange as well as automated, digital data delivery. In order to develop solutions such as a common software program, German grid operators are collaborating as part of the nationwide project “Connect+.”
Visitors to EM-Power Europe in Munich from May 11 to 13, 2022, will enjoy access to specialist information, products and services relating to weather and yield forecasts as well as for monitoring energy plants.