How to reduce imbalance costs with automatic forecasting

“Why the hell do our imbalance costs keep growing every year?”. – CEO of an energy company

 

Does this question haunt you too? The costs resulting from a difference in energy consumption forecast and reality are bleeding the energy industry dry. But while established players are struggling, young companies with new business models are generating profits.

 

 

The opportunity hidden in imbalance costs

Why are some companies generating profit when others are losing money? Do they know more about physics or meteorology? Yes and no. They don’t need to be physics or math geniuses! They need automatic forecasting, forecasting that is more accurate, more granular, generated faster and an integral part of their systems.

 

Hourly and day-ahead forecasts play a huge part in controlling imbalance costs for large grid operators, distributors and generators. These established players are having a hard time generating sufficiently accurate forecasts. This allows companies, that are better at predicting, to trade electricity and generate a profit off imbalances. Others can save money by going directly into the energy market.

 

Flexibility service providers are examples of companies using automatic forecasting. Their platforms allow businesses to circumvent an energy distributor and trade directly in the energy market. Thanks to their forecasting, companies can save money on their electricity bill.

 

How can smaller companies have forecasts that are superior to those of established players in the energy sector, even without a team of specialised data scientists? One recipe for automatic forecasting is TIM!

 

TIM, Tangent Information Modeller, is a tool that allows one person to do the work of multiple Data Scientists. Thanks to machine learning and informationgeometry, it can come up with forecasting models automatically and in seconds instead of days.

 

Being able to generate models at this rate allows companies using TIM to predict energy consumption and generation on the level of individual assets. It also allows for automatic recalculation of models if conditions change. And best of all, TIM can feed the forecasts directly into production systems through an API.

 

What does TIM need to generate forecasts automatically?

The most important factor in forecasting is quality of data. Our forecasts are as good as data allows. If you have clean data, TIM can automatically generate models and forecasts with high accuracy. If you don’t have good data storage practices and clean historical data, you need to fix that first.

 

Is automatic forecasting for real businesses or is it academic?

Powerhouse is one of our largest clients. Their Easyflex platform is penetrating the market with forecasts from TIM. TIM was the only solution that fit their needs for automatic model and forecast generation. Within 3 months since our first contact, it was an integral part of their systems.

 

“Without TIM, the autobid part of our product could not exist. No other solution we have found could do this.” – Product Owner, Powerhouse

 

During a testing phase with another client, we did parallel forecasting to show how much money we can save with TIM’s automatic forecasts. For this energy distributor, the final figure was 40%, which comes to millions of Euros annually.

 

Are automatic forecasts accurate?

To prove TIM’s accuracy once and for all, we entered the Global Energy Forecasting Competition 2017 (GEFCOM). Our automated models competed against 177 academic teams from around the world. TIM was the overall winner in this competition, showing that its automatically generated forecasting models are on par with the best Data Scientists in the world.

 

Do you want to know how TIM can help you? Sign up for a webinar!