Estonia sets its sights on energy abundance and zero-emission generation, with a lofty goal to balance greenhouse gas emissions and removals by 2050 at the latest. To get there, the work starts today, and Estonian energy companies are already fielding cutting edge technologies and analytical tools for future transformation.
Estonia’s top public energy company, Enefit, has taken a creative approach to tackle an urgent technology issue. While it’s undoubtedly a good problem to solve, the increasing number of home energy producers underscores the need to address energy imbalance and costs. As machine learning and data science quickly evolve, instead of collaborating with a traditional IT development firm, Enefit launched a competition on Kaggle, the global data science and ML community platform now owned by Google.
The competition gained widespread attention for its challenging real-world task, attracting top data scientists globally, with a record 2,715 teams participating from Japan, the US, China, and other countries.
The competition addressed the growing energy imbalance issue due to the rise in households installing solar panels. In Estonia, small electricity generators have increased from 3,000 to 21,000 in the last five years. Prosumer households, who both consume and generate energy, face challenges with inaccurate energy usage and production forecasts. These inaccuracies result in significant fines for energy companies, impacting consumer prices.
The financial impact on balance managers can vary, but may reach several million euros monthly in some markets. Germany and Italy, for example, already face such challenges, as noted by Lion Hirth, a notable professor in power systems and energy markets.
The Kaggle competition tasked participants with creating a forecasting model to improve energy predictions. Enefit provided detailed datasets, including weather conditions, energy prices, and solar capacities. Teams had three months to build their models and compete for a $50,000 prize. Using Kaggle’s time-series API to generate predictions, teams were ranked based on prediction accuracy. The top six teams, including leading Kaggle member Yide Huang, were awarded portions of the prize fund for their precise models.
And all of this is not just a test drill, as Enefit is prepared to integrate the new code into their system. The smart energy optimizer will manage energy usage efficiently, potentially encouraging more people to become prosumers and support renewable energy adoption.
Harnessing technologies for better energy usage
“We were overwhelmed by the level of interest in our data challenge. This reflects the growing appeal of the energy sector among technical professionals. Our positive experience with Kaggle certainly inspires us to continue leveraging innovative, on-demand methods to engage with the world’s top talent,” remarked Kristjan Kuhi, a Board Member at Enefit.
According to Kuhi, its not the only way Enefit uses new technologies to help provide customers with smart, money-saving, and eco-friendly solutions. ”One such example is the energy use optimizer being developed by Enefit, which monitors consumption, production, and energy market prices to make choices that maximize customer benefits. Based on a person’s home consumption profile, it controls solar panels, energy storage, and electric car chargers in the most profitable way, taking into account both electricity exchange prices and the network tariff,” he noted.
As another example, Enefit recently implemented a new solar energy solution at the Estonia dairy farm in Järva County. It includes a 150 kW storage unit, effectively doubling the solar power plant capacity to 300 kW without requiring a larger connection point. The storage unit also participates in flexibility markets to ensure profitability throughout the year. Customers generating their own electricity contribute significantly to the national electricity system by reducing peak loads and grid losses.
And AI also now makes a mark in company’s operations. “Enefit also uses AI-based solutions to trade on energy markets, forecast production and consumption, and reduce emissions from power plants. For example, the company has developed a power plant simulator, which considers the price and quantity of each fuel component, the quantities of CO2 and NOx emitted (and the corresponding environmental taxes), and finds the most optimal fuel mix. As fuel prices and emissions vary constantly, continuous calculations and evaluations are necessary to maximize efficiency. Within the company, Enefit has also created a GPT assistant, which helps save approximately 90-95% of the time spent searching for documents,” shares Kuhi.
Interested in investing in Estonia? Find out more about the country’s energy sector here and book a time via e-Consulting to speak to your personal advisor.