Improving Hydropower Production with Artificial Intelligence
Category Machine Learning Sunday - May 7 2023, 07:45 UTC - 1 year ago Hossen Farahmand and his team of researcher from the Western Norway University of Applied Sciences (HVL), NTNU and other institutes are combining advanced artificial intelligence with data from the electricity market to create a model predicting river inflow into hydro reservoirs. The model helps to improve hydropower producers decisions and help mitigate energy crisis.
Reza Arghandeh from the Western Norway University of Applied Sciences (HVL), Hossein Farahmand (NTNU) and their team are studying how hydropower producers can make better use of natural resources and flex with the market at any given time. The researchers have developed concrete methods within artificial intelligence to calculate how producers should regulate the degree of filling in water reservoirs.
The leader of the research project on hydropower production is Hossen Farahmand at NTNU, a professor and head of the research group for electricity markets and energy system planning. The team also includes Mojtaba Yousefi from HVL and Jayaprakash Rajasekharan and Jinghao Wang, both from NTNU."If hydropower producers could make decisions that were just one percent better than before, it would amount to billions of kroner in difference and help to mitigate the energy crisis," says Arghandeh.
The future European power system—based primarily on renewable energy sources—will be much more weather dependent than the power system today. The two researchers believe that consumption patterns will also change. All these factors contribute to creating uncertainty around the energy supply, causing decision-making to be far more complicated. The researchers will help to reduce this uncertainty, so that it will be easier to secure access to energy.
Completely dependent on electricity, Arghandeh has found a way to monitor meteorological data, hydrological data (how much water is supplied to the reservoirs) and topographical data (shape of the landscape) and interpret them with the help of advanced artificial intelligence. Farahmand further combines the use of these AI models with data on the electricity market. Together, they have created a model that accounts for uncertainty in the market and in the weather and wind.
"Electricity isn't an ordinary product, but one that keeps our society afloat, almost like oxygen. That's why it's incredibly important to create affordable, reliable and sustainable access to this basic good," says Arghandeh.
Accurate and reliable inflow forecasting has always been a challenge in the production of hydropower. Around Easter time, the reservoirs fill up. Then the snow melts on the mountain tops, and water begins to flow from the heights into the reservoirs, regulated waters and rivers around the country. Between Easter and until winter returns around November, producers release water at regular intervals to supply the market. They also need to make sure to have enough water left in the reservoirs to last through the winter season, when the supply dries up (or freezes).
But when and how much water should be released? That depends on many different factors, including weather conditions, landscapes, rainfall, winter temperatures, the electricity market and the political situation in Europe. Since the power system in Norway is connected to Europe, making good decisions becomes all the more difficult. This was powerfully demonstrated by the events that followed the Russian invasion of Ukraine. When Russia stopped the sale of gas to Europe as a result of European support for Ukraine, the effect on both energy access and energy prices was dramatic. Calculations become extremely complicated with so many factors. The predictions of how much water can be released and when must take into account the changing weather, seasonal variations and even the geopolitical situation.
Share