Routing on electricity grids: Net node allocation for Germany’s electrical neighbours
In order to be able to plan the energy transition efficiently and sustainably, the complex interplay between renewable electricity generation, electricity demand, storage and power-to-X technologies with the electricity grid is mapped in models. Because data availability is limited and computing capacity is constrained, simplified assumptions must be made when modelling the power grid at all voltage levels.
The 220 kV and 380 kV extra-high voltage levels are modelled in the FfE energy system model ISAaR by a linearizing load flow calculation. Through a routing approach that goes beyond a simple “Nearest Neighbour” allocation and also takes the network topology into account, the underlying extra-high voltage level can also be modelled for Germany. This allows a higher accuracy of the allocation between electricity load and generation to the network nodes and thus improves the quality of the energy system model. Since the results of the routing approach proved to be promising for Germany, the approach was also extended to surrounding countries within the scope of this work. Thus, the routing approach can also improve the energy system modelling in these countries. The approach has shown promise for 6 of the 10 European countries considered.
How the routing approach works
The routing method makes it possible to represent the electrotechnical conditions of the high-voltage grid in a grid modelling that only considers the extra-high voltage level in detail. For the grid calculation, it is decisive how much electrical load or generation is assigned to a network node. In the course of this, NUTS-3 regions (a European administrative division) are assigned to the network nodes of the transmission grid. A common method for this allocation is a simple geo-allocation: if exactly one network node is located within a region, the entire load or generation is allocated to this node. If there are several network nodes within a region, the load or generation is distributed proportionally according to the area that the network nodes occupy depending on their location to each other.
The routing approach improves this allocation of electricity demand and generation to the network nodes at the extra-high-voltage (EHV) level by taking into account the underlying high-voltage grid topology. For this purpose, it uses a search algorithm from navigation (Dijkstra algorithm). This uses the feed-in or offtake points of the voltage levels below the high voltage as starting points and the network nodes with transformer at EHV level as end points of the routing. The algorithm routes from the start nodes to the next 5 end nodes and uses the distance as an inverse weight for the distribution of load and generation. In this way, a mapping of the load and generation distribution of a county to the network nodes can be generated. Network nodes that are further than 25 km away from starting points are not taken into account.
In order to extend this routing approach, which has already been implemented for Germany, to surrounding countries, research first had to be carried out on the transmission and distribution grids to determine how the high-voltage level can be defined in the countries. Here, mainly the official information from transmission system operators and distribution system operators was used.
In order to test and apply the routing approach, so-called Voronoi polygons were formed around the feed-in and offtake points – areas that extend in all directions to exactly half the Euclidean distance to the nearest feed-in or offtake point. These areas approximate the catchment area of the point and are intersected with a power consumption grid. The grid has an edge length of 250 m, the pixels contain information about the electricity consumption of private households (PHH) as well as trade, commerce, services and industry (GHDI) according to IEA data [1][2][4] from the year 2020.
For the implementation, all steps were carried out in FREM [3], the Regionalised Energy System Model of FfE, analogously to the routing approach for Germany. Open data from the OpenStreetMap project [4], in particular the GridKit tool, which was developed in the course of the SciGRID project [5], served as the data basis for the network nodes and the high-voltage grid. The tool transforms raw OSM data with electrical tags into an electrical network model using spatial and topological methods.
The data obtained from the OSM project was visually aligned with ENTSO-E’s GridMap [6]. Except for Switzerland, where some lines are missing, this showed that the data from the OSM project is almost complete (as of 2019).
Results
Whether the routing approach can be usefully applied in a country depends mainly on whether there is a high-voltage grid in the country for which sufficient grid data can be obtained. The routing approach has proven useful for the Netherlands, Belgium, Austria, Italy, the Czech Republic and Poland. These countries have a well-developed high-voltage grid for which sufficient grid data is available to build a routable network. The routing approach is particularly useful in countries with large NUTS 3 regions such as the Czech Republic. Since the simplified geo-allocation assigns NUTS-3 regions to network nodes without taking the high-voltage network topology into account, this often results in a very inaccurate allocation.
The usefulness of the routing procedure for large NUTS 3 regions can be seen in Figure 1, for example.
This NUTS 3 region contains three network nodes. The simple procedure adds up the total consumption load of the region and distributes it relatively evenly among the network nodes. In reality, however, a large part of the load is located in the south of the region and should thus be allocated to the lowest network node. This fact is also taken into account in the routing approach, but not in the simple geo-allocation.
Another advantage of the routing method is that it often results in a more realistic homogenisation of the load. This can be seen in an example in Belgium in Figure 2:
Here, in the simplified geo-allocation, the entire load of the NUTS 3 region marked in red is allocated to a single network node. However, since several network nodes are located in the immediate vicinity of (but not within) the NUTS 3 region, it would be more realistic to allocate part of the load to these network nodes as well. This is implemented in the routing approach.
However, the routing approach is currently of very limited use for France, Denmark and Luxembourg. The grid structure of these countries is not suitable for the routing methodology, as the routing procedure relies on a separation of the extra-high-voltage and high-voltage levels, which does not exist in these countries. For Luxembourg, it is also not necessary to apply the routing approach due to the small size of the country.
For Switzerland, the routing approach is currently only partially useful, as there is a less developed high-voltage grid here compared to the other countries considered.
Validation of the maximum distance of 25 km
In the routing approach, network nodes that are further than 25 km away from starting points are not considered. This choice of 25 km has not been validated so far. As part of this work, a validation of the distance parameter was carried out using the German power grid. The network node loads resulting from the routing approach were compared with real network bottlenecks, which are presented in the Monitoringbericht 2022 von Bundesnetzagentur und Bundeskartellamt [7].
Two groups of network nodes were considered: Network nodes that are in the immediate vicinity of real network bottlenecks and network nodes that are not in the immediate vicinity of a bottleneck. From a model that is a good approximation of reality, network nodes close to real bottlenecks can be expected to have high loads. Network nodes that are far away from bottlenecks should have a rather low load on average.
An analysis with different values for the maximum distance showed that 25 km represents the real bottlenecks more realistically than other parameter values.
Conclusion
The implementation of the routing approach was successful for 6 of the 10 electrical neighbours and will be available in the future as a useful alternative to simplified geo-allocation. In addition, the maximum distance parameter for routing was validated and set to 25 km.
Weitere Informationen
Literature
[1] International Energy Agency – Country Statistics in: http://www.iea.org/. Paris: International Energy
Agency (IEA), 2020.
[2] GHSL – Global Human Settlement Layer in: https://ghsl.jrc.ec.europa.eu/. Brüssel: Global Human Settlement Layer (GHSL), 2020.
[3] Corradini, R.; Konetschny, C.; Schmid, T.: FREM – Ein regionalisiertes Energiesystemmodell in: et –
Energiewirtschaftliche Tagesfragen Heft 1/2 2017. München: Forschungsstelle für Energiewirtschaft, 2017.
[4] OpenStreetMap (OSM) – OpenStreetMap und Mitwirkende: http://www.openstreetmap.org/;
Cambridge: OpenStreetMap Foundation, 2004 (überarbeitet: 2019).
[5] SciGRID: Power Relations in OpenStreetMap in http://scigrid.de/posts/2015-Jul-02_power-relationsin-
openstreetmap.html (besucht am 18.04.2017). (Archived by WebCite® at http://www.webcitation.org/6poPhZ7fA). Oldenburg: SciGRID, 2015.
[6] Grid Map in https://www.entsoe.eu/data/map/ (besucht am 18.04.2023). European Network of Transmission System Operators for Electricity, 2023.
[7] Monitoringbericht 2022 von Bundesnetzagentur und Bundeskartellamt. Bonn: Bundesnetzagentur für Elektrizität, Gas, Telekommunikation, Post und Eisenbahnen, Bundeskartellamt, 2022.
