Elia and Tangent Works collaborate to make system imbalance forecasting more flexible, accurate and cost efficient.
Elia is not just Belgium’s transmission system operator – it is also a key player at European level.
Elia is Belgium’s high-voltage transmission system operator (30 kV to 380 kV), operating over 8,600 km of lines and underground cables throughout Belgium.
Elia plays a crucial role in the community by transmitting electricity from generators to distribution systems, which in turn deliver it to the consumer. Elia also plays an essential part in the economy, as a supplier off power directly to major companies connected to the grid.
Boasting a pivotal location in Europe, Elia is also a key player in the energy market and the interconnected electricity system. The company has set up multiple initiatives promoting the development of an efficient, transparent and fair electricity market for the benefit of consumers.
Introducing new technologies to better run its operations and to prepare the Belgian energy market for the future where renewables and decarbonization will demand more flexible grid operations is a key goal of Elia.
The Elia innovation group has AI and machine learning initiatives to explore its potential. Tangent Works has done several innovation explorations with its product TIM™, Tangent Information Modeler in the context of Elia operations.
DSO process forecasting, load forecasting and system imbalance forecasting were tested with good results by TIM™.
This year, Elia will test the potential roll out of TIM™ in its Network Control Centre to automatically generate the predictive models for short terms system imbalance forecasting. This is a crucial element in the Elia operations as it defines if and the nature of the strategic regulation reserves that need to be activated. The impact of these decisions on the grid operations and even market prices can be important.
TIM™ will generate the models automatic, adapt them to new dynamics in the data as they emerge and will generate every 15 minutes a new forecast of the expected system imbalances. This process is completely data driven and should allow for a more trust worthy system imbalance forecast.
Looking forward, the same technology can be applied for other forecasting scenario’s, not requiring any involvement of data scientists to generate and tune the predictive models.
Building and refining accurate predictive models is a laborious, iterative task that requires a combination of domain and data science expertise and weeks/months of effort.