Artificial intelligence, a valuable ally for the energy market
Renewable sources, distributed generation, digitalized grids, storage devices, electrification of consumption, customers who are more empowered and increasingly are also producers, free markets, climate and weather variability: if there’s one word that describes today's energy world, it’s complexity. The ongoing energy transition involves a system that’s increasingly sustainable and efficient, but also more complex.
Global Trading: what’s it all about?
Energy markets are also increasingly complex. The Enel Group, in the countries where we’re present, operates in markets in two different but related areas. One is energy management, including the energy market on which we sell the electricity produced by our Group in the various market sessions, trying to sell as much of it as possible so as to avoid waste and maximize revenue.
The other is the wholesale energy commodity market, which in our case is mainly gas. Here the goal is to get the quantities we need at the best possible price.
These two areas have different characteristics, but they have one thing in common: a high degree of complexity. They’re mixed markets, which combine the intangible, financial aspect of stock exchanges with the concrete, physical aspect of transportation infrastructure (transmission and distribution grids in the case of electricity, tankers or gas pipelines in the case of commodities). To manage this complexity, Enel has at its disposal a dedicated and specialized company: Enel Global Trading.
Today, this company is operating in a context in which, as in all financial markets, times are getting faster and there’s more and more information; so it’s more difficult to analyze and interpret. Given all the variables involved, it is essential to have reliable forecasts as far in advance as possible. The solution is the intervention of a valuable ally: artificial intelligence, with its ability to process big data very quickly.
Artificial intelligence: applications in the energy market
In the energy market, sessions are becoming increasingly intense and fast-paced and begin the day before trading. The use of artificial intelligence models allows estimates of several market variables to be made quickly and accurately. Models are trained on historical data and other relevant parameters, such as weather conditions, to estimate trends in prices, demand and supply volumes. In addition, again by using historical data, the models identify recurring patterns of behavior by competitors.
Finally, on the basis of the results thus obtained – as well as constant analysis of grid conditions – it’s possible to carry out simulations to understand if and when interventions will be needed to maintain system stability: this is very useful information, which enables us to take advantageous positions in advance when it comes to the market for ancillary services, i.e., the grid services needed to keep the system in balance.
As for commodity markets, applications of artificial intelligence bring a similar benefit, with the development of more effective and faster predictive models. Of course, given the size of the numbers involved, models must be thoroughly tested – and possibly fine-tuned – before they’re adopted.
AI and weather forecasting
Another area, perhaps unexpected but particularly crucial today, is weather forecasting. The stars of the energy transition are the new renewable sources, namely wind and solar, which inevitably involve a margin of uncertainty due to the variability of weather conditions.
Traditional weather forecasts were based on mathematical physics techniques, which are able to provide estimates on a large scale, across geographic areas, but are not as precise for individual locations. There was then a shift to statistical models, which are more reliable at short range.
Artificial intelligence has been introduced to further – and considerably – improve the reliability of forecasting at the local level: it has been shown to be efficient in predicting the output of a single photovoltaic installation or wind farm, or even of a single turbine.
Natural Language Processing
One of the less obvious applications of artificial intelligence to energy markets is related to its ability to process written language (Natural Language Processing, NLP). Financial markets, and commodity markets in particular, are sometimes affected by rumors from the world of economics, finance, and politics: analyzing the content of news helps us predict its possible impacts on prices, and thus helps us move in the most advantageous way in the market.
Language analysis can also be providential in another area. An increasingly popular energy market product, especially for large customers, is Power Purchase Agreements (PPAs), medium- to long-term contracts for the supply of renewable energy – a tool that allows customers to hedge against possible price fluctuations, allows suppliers to have secure estimates of electricity sales, and, most importantly, contributes to the sustainable energy transition.
The human factor in global trading choices
PPA contracts, however, are often packed with clauses, so much so that their potential risky consequences can escape even an experienced analyst. Artificial intelligence can read the contracts and immediately highlight if any clauses are out of line with the standard, in which case they can be reviewed and, if necessary, corrected accordingly.
However, it’s people who make the changes. Just as it’s always people who make the markets: artificial intelligence has the great advantage of being able to process big data in very little time, but it cannot replace human expertise. Its role is to give analysts and market participants all the tools they need to make the best decisions.