Data Science @ Statnett

In 2019, Statnett installed its first sensors for dynamic line rating. Now, in 2023, we have finally reached the point where we can use these sensors to increase the capacity of power lines with up to 20%. This post will explain how data science has played an important role in making this possible.

The Norwegian grid is close to its maximum capacity

European industry and consumers are moving towards greener energy sources, often supplied as electricity, to reduce carbon emissions. This has created an increase in demand for electricity. The Norwegian power grid is already close to its maximum capacity, which means that new consumers have to wait for several years before being connected to the grid. In Statnett’s systemutviklingsplan 2023, we can read that Statnett expects the consumption to double within 2050.

Statnett is taking a number of different steps to increase the capacity in the grid. New power lines are being planned, and old ones are being upgraded to higher voltages, but these measures take several years to complete. To increase the allowable current of existing lines in a much shorter time, we therefore turn to dynamic line rating. This has been presented previously on this blog: Smarter Transmission Grid Capacities with Weather Data – Data Science @ Statnett

Increasing capacity with dynamic line rating

The capacity of power lines is often limited by thermal limits, i.e. how much current the line can handle before reaching a temperature limit. High temperatures are problematic for two main reasons: They can damage the line material, and they may cause the line to expand too close to the ground, which could be a safety hazard.

Dynamic line rating sensors allow us to measure the temperature and current of the line in different weather conditions, thus giving us better control over how much current the line can handle. By using a weather forecast, we can then predict how much current the line can handle in the near future as well.

What have we learned so far?

This current limit forecast has now been used in operation on three power lines since June 2023. What have we learned so far?

This figure shows the average capacity increase for each week since we started using the DLR forecast to set the capacity. In this plot, 100% represents the “normal” capacity before DLR. We notice that we can get almost 20% extra capacity in some weeks, but there is a large variation between weeks. This is especially noticeable for Line 3, where the capacity increase varies from 3% in August to 16% in October. We expect that this large variation can be explained by variation in wind, since wind is the most important contributor to increased capacity in DLR.

In this figure, we see the capacity on Line 3 together with the average measured wind at a weather station located close to the line. We notice that there is a clear correlation between the measured wind and the capacity. The linear correlation coefficient is 0.81, which means that 81% of the variation in capacity can be (linearly) explained by wind speed. Here, we have not considered the effect of wind direction. The wind direction also plays an important role, since wind perpendicular to the line has a larger cooling effect than wind parallel to the line.

The future of dynamic line rating

The last 5 months have given us valuable experience on how to use DLR in an operational setting, and has confirmed that it can give significantly higher line capacities. During the next year, we plan to install DLR sensors on even more lines, increasing capacity on constrained lines that connect bidding areas. In the long term, this can hopefully lead to lower price differences between areas in Norway.

Looking even further ahead, we hope to be able to predict weather along the lines without the need for DLR sensors. This can be achieved with a combination of weather stations and more fine-grained weather forecast models. We will develop improved forecast models as part of the research project Kortprog, together with our partners Norconsult, Telenor and Meteorologisk institutt.