COVID-19’s devastating impact on health and the global economy also has a silver lining: an opportunity to tackle climate change.

Enter wind and solar – rapidly growing sources of renewable, affordable and available energy in Europe. As the EU mobilizes to support green energy projects as part of its economic recovery strategy, machine learning and AI are well positioned to help the continent’s energy and utility companies adapt and evolve their existing asset infrastructure and operational practices to meet increasing demand.

Wind farms and solar arrays are formidable yet fragile feats of engineering – prone to wide and varied environmental forces – from the sun- and wind-scorched Canary Islands to the blustery North Sea to the bone-chilling Arctic circle. Managing peak loads and distribution amidst unpredictable weather while keeping systems running poses tough challenges to maintaining a balanced and resilient electric grid.

In the last two years, IBM has applied data and AI solutions to renewable energy management projects at Spain’s Red Electrica, Nukkisiorfiit in Greenland and James Fisher in the UK. Recently, IBM was able to demonstrate to Denmark’s Electric Transmission System Operator (TSO) how machine learning capabilities in IBM Cloud Pak for Data could accelerate a faster transition to green energy –  meeting the need for utility asset performance management, reliability and operational excellence.

Accelerate your journey to AI.

Watch how it happened:

Energinet operates and develops large transmission grids that form the backbone of the country’s electrical supply for 5.8 million citizens – with interconnectors that transmit power between Denmark and surrounding countries Sweden, Norway, the UK, Germany and the Netherlands.

The country aims to rely 100% on renewables by 2030 –  so for Energinet, the challenge has been three-fold: provide citizens with increasing levels of green electric power resilience and security of supply – at a price point that all can afford.

Energinet knew it needed a fresh approach and new thinking to re-write its energy future –  so it engaged IBM Garage on a three-month pilot project to design a “virtual operator” that could estimate risks to the grid based on large simulation data amounting to 400 terabytes.

The team’s goal was to deliver an easy-to-use interface capable of modeling different scenarios – both real and hypothetical – such as the impact on the system of taking equipment out of service during a certain period of time.

The solution was deployed using IBM Cloud Pak for Data, handling terabytes of simulation data within Watson Studio’s Machine Learning capabilities to evaluate the transmission system’s ability to withstand shocks under fluctuating power flows coming in and out of the country.

After implementing a hybrid cloud platform architecture, the joint team layered machine learning and artificial intelligence on top of terabytes of “N-1” data, consisting of historical facts about energy flow and overload situations.

The user interface reveals detailed risk probability instances, displaying the chances of an operational limit violation. Informed by years and combinations of past operational and environmental conditions encountered by the Energinet transmission system, the trained model risk profile allows the solution to provide robust decision support capability for the Operations Center, with “look-ahead” scenario generation.

Users can see where operational limits might occur, accept risk or initiate interventions to increase maintenance efficiency and identify and validate the most critical infrastructure needs.

Learn how IBM Garage combines startup speed and enterprise scale to tackle tough problems, ignite innovation and spark creativity.

Accelerate your journey to AI.