FeaturesScience & Digital

AI in Energy: Magical Puzzle Solver or Not Just Yet?

Everywhere you look today, you see “Artificial Intelligence is the new ‘this’” and “AI transforms the ‘insert industry name’”. And the energy sector is no exception

With the need to accurately predict countless variables both day and night, AI might just be the latest solution. But can this new tech magically solve these increasingly complex puzzles, or is it just a helpful tool with loads of potential?

Big Changes with New Problems to Solve

The already complicated system present in the energy sector has only increased with the advent of renewables, climate change, decarbonisation, electromobility, and production decentralisation. These factors also bring uncertainty. So far, all evidence has shown that AI isn’t a miracle solution to these issues but rather an advantageous helper that can result in tens to even hundreds of millions of dollars in cost savings.

Only a Warmup for the Mathematician but a Rubik’s Cube for the AI

Reducing the balance interval to 15 minutes significantly limits traders’ ability to work with the position. That’s because it’s technically much harder to get to larger extremes that have the quarter-hourly assessments resulting in a balance that follows the contracted state. Not to mention the hefty penalties for any deviations.

In practice, this presents an array of changes to market participants’ operations and how they run their systems. Up until now, traders have predicted consumption by the hour – they now have to move to a period 4x shorter. Of course, generation will also have to be forecasted with much greater frequency, which would already be challenging enough without throwing in the renewables mentioned above.

Big changes can also be seen in the principle of trading. For example, in the daily market of the average electricity market operator (when dealing with the following day), participants have so far been able to offer 24 volume and price values. Now with the new 15-minute period, that number jumps up to 96 values. Plus there’s a different way they need to be entered, more complex evaluations, and a slightly new message format for communicating with the market operator.

The move to 15-minute intervals also affects performance, again, because the shorter interval equals more data to process. For example, the annual chart will jump from a manageable 8,760 points to an astounding 35,040 points.

And this shift meant big changes for us as well. As a result, we’ve modified our entire energy platform and its individual products. One might be tempted to say we made these adjustments from the ground up, but in reality, the high level of standardisation helped out a lot. But more importantly, we always stick to the same principle for all Lancelot products, whichaim to “reflect market changes in a high-quality and timely manner”.

Image: Unicorn Systems

AI Lightens the Load for Human Labour

The magnitude of these changes that see a reduction in the balance sheet period can’t be overlooked. We realised this when preparing our energy trading products within the Lancelot range for these legislative changes. These products include Lancelot FMS, which provides traders with various tools to calculate forecasts of their portfolio’s consumption and production as well as the energy trading risk management system Lancelot ETRM.

Every component of this software needed to be adjusted to match the new 15-minute intervals, thus ensuring their outputs meet the new requirements.

Then there was the visualisation aspect, i.e., all the tables and graphs – in other words, the entire UI had to undergo a major overhaul to work efficiently within the shortened time slots. 

Here’s an example: the most common display used to be a 2D table with the days on the Y-axis and the hours on the X-axis. Once, this was the perfect fit, given that the 24 values could be displayed on a single screen and still be legible. But now that number is multiplied by 4… See where we’re going with this?

Image: Unicorn Systems

1% = Hundreds of Millions of Dollars

At Unicorn, we’re testing what can be achieved by implementing AI. We’re performing these tests at our Unicorn University and with help from the Institute of Computer Science of the Academy of Sciences. So far, we’ve discovered that AI can improve the accuracy of our consumption predictions by a percentage point or two.

Now, 1-2% may not sound like a lot, but for an electricity trader, it’s a total game-changer. Considering today’s penalties for deviations from the balance of production and consumption, AI’s marginal help can actually save the market’s larger players tens to even hundreds of millions of dollars year on year. 1-2% doesn’t really seem all that small anymore…

Even the Best AI Still Needs Data

For now, we can safely say that AI isn’t some magic code or a silver bullet. It can’t mystically solve every energy puzzle for you, but it can act as a powerful tool and potentially save companies millions along the way.

Although AI-driven forecasting models are improving, their overall success still crucially depends on the quality of the input data. That’s why one of the key products in our energy-focused Lancelot range is Lancelot Hub, which concentrates and cleanses data from the entire energy market.

At Unicorn, we have a popular motto: “data at the centre”. Even as the energy industry evolves and artificial intelligence starts to play a bigger and bigger role, this motto will only grow ever increasingly true.

About the Author

Luboš Zelinka works at Unicorn Solutions as the Lancelot Family Business Architect. In this role, he’s responsible for implementing Lancelot products. He also serves as Lancelot’s Chief Evangelist and is responsible for the long-term development of the Lancelot energy platform and the products built upon it. Luboš has worked in the energy industry for over 18 years. During this time, he’s been involved in delivering many Lancelot implementations to leading Czech and foreign energy companies.

This is a featured article.

Related Posts