Alliander improves load estimation with STORM
Image courtesy Alliander
Dutch network operator Alliander and the Radboud University have developed the STORM algorithm to help load estimation on a packed electricity network.
The STORM algorithm, with the aim to improve load estimation in networks, automatically filters measurement data to remove disturbances such as measurement errors and temporary switching events.
This saves technical experts who traditionally have filtered the data by hand, some 75% of their time, while also significantly improving the quality of the data, Alliander reports in a statement.
With this filtered data, it is possible to better estimate how much capacity is really available for new connections and how the network can be used smarter, the utility adds.
Have you read?
Congestion management studies yield 880MW for Dutch grid infeed
Unlocking flexibility in Europe’s evolving energy systems
The Dutch electricity grid is under great pressure due to the rapid growth in electricity demand and the transition to renewable energy.
While expanding the network is a necessity, it takes time and in the meantime Alliander’s intent is to make optimal use of the existing capacity.
Underlying the development of STORM is the concept of ‘explainability’ to better substantiate why an application is or is not possible.
Drawing on primary substation data on Alliander’s grid, and optimised for anomaly and switch event detection, the algorithm combines binary segmentation for change point detection and statistical process control for anomaly detection as the most effective strategy.
In the modelling approximately 90% of the automatic load estimates fell within a 10% error margin, with only a single significant failure in both the minimum and maximum load estimates across 60 measurements in the test set.
“Our methodology’s interpretability makes it particularly suitable for critical infrastructure planning, thereby enhancing decision-making processes,” Alliander’s and the Radboud University’s report reads.
Open data
The STORM substation dataset now forms part of Alliander’s open data offering.
Other open datasets available cover location data on electricity grids, generation data for small-scale connections, feed-in data for small-scale connections, transport forecasts, consumption data for small scale connections, smart meter consumption data 2012-2014, consumption profiles for large-scale electricity connections and consumption profiles for large-scale gas connections.
By making such data open and anonymised so that it cannot be traced back to an individual or connection, Alliander’s intent is to provide insights and enable smart solutions.
Alliander intend to continue to work on further improvements to STORM to make it even more time effective and accurate.
As part of their collaboration Alliander and Radboud University also have launched a new course programme, CHARGE, focussed on the use of data to address grid congestion.
