Series of articles on the characterisation of low-voltage grids and grid representatives: Identification of reference grids for the clusters

Many research projects in the context of the electrification of mobility, heat and industry are investigating how new electrical consumers in combination with the ramp-up of renewable energies will affect electricity grids. A large share of the added electrical load is connected in low-voltage grids (LV grids). One approach for their integration is to make consumption behavior more flexible in order to avoid or eliminate congestions in the grid. To evaluate the concepts in terms of their impact on the grids, they are simulated with future load scenarios and grid operation strategies.

In total, there are over 500,000 LV grids in Germany, which are operated by around 800 distribution grid operators [1]. In total, these result in a line length of over 1,200,000 km [2]. It is hardly possible to simulate all grids individually. On the one hand, they are not available in a simulatable form, and on the other hand, this would require an enormous amount of computation capacity. In order to be able to make statements for different grid structures in Germany, reference grids are therefore used for simulations. These allow to draw conclusions about the grid infrastructure from geographical and structural parameters.

This series of articles shows which parameters and methods are used to create reference grids. Furthermore, data from literature is merged and a set of reference grids is created. Specifically, the following topics are addressed:

1. Low-voltage grids in Germany
2. Characterisation of low-voltage grids
3. Identification of grid clusters in the low voltage level
4. Identification of reference grids for the clusters
5. Distribution grids in the Climate-Neutral Energy System – Scenarios for Consumption, Generation and Grid Operation

Modelling of characteristic low-voltage grids

In the context of the energy transition, low-voltage distribution grids (LV grids) must ensure the integration of numerous new generation facilities as well as the supply of new consumers, such as heat pumps and electric vehicles. Grid models and simulations are a tool for investigating the stress on the grids in forecast energy systems. In this way, the suitability of the grids for a future climate-neutral energy system can be assessed. As the first three articles in this series show, the LV grids in Germany have a very heterogeneous character, the characterisation of which is a complex undertaking. The second article in this series described parameters and criteria that characterise LV grids according to the literature. The meta-clusters identified in the third part of this series build on these parameters and thus enable a delineation of different grid types. It can be seen that the topology and topography in the grid areas are influenced by various factors, such as regionality, settlement structure or the primary supply task. These influences are ultimately also reflected in the low-voltage grids included in each case.

The distribution grid analyses in the literature also illustrate that the representative mapping of LV grids by representatives is a complex undertaking. In the scientific analysis of LV distribution grids, there is often a conflict of objectives between a representative versus a detailed representation of the grid load. The public availability of real topology data of LV grids is severely limited, also for scientific analyses. This is primarily because LV grids are classified as “critical infrastructure” in the broader sense and there are also data protection concerns (protection of end consumers). So-called reference grid topologies, which correspond to a characteristic representation of real topologies, provide a remedy here. Compared to real topologies, they have the disadvantage that the degree of detail and the informative value concerning real grid load conditions are significantly reduced by statistical simplification. On the contrary, these topologies are particularly suitable in science for comparing different scenarios with each other. Furthermore, the statistical representativeness of the type networks makes them more informative about supra-regional effects. This makes it possible, for example, to extrapolate to municipalities or city districts. [3] In the context of the identified meta-clusters (cf. third part of the series of contributions), characteristic reference grid topologies were modelled in the following.

Methodical procedure for the modelling of reference grid topologies

Modelling reference grids aim to represent the identified characteristics of a specific group of distribution grids as representatively as possible. For example, the grids of a selected region or settlement structure. To map the characteristics of the identified meta-clusters as representatively as possible, various characteristic values were calculated and evaluated during the modelling, which describes the topology. From the topologies and clusters in the literature and real topologies used for verification, the following characteristic values were taken into account for each grid area:

• Average number of branches.
• Minimum, average and maximum branch lengths
• Average length of house connection lines
• Average number of grid connection points (GCPs) per grid area and line
• Average distances between GCPs
• Degree of meshing (number of closed rings/meshes)
• Average transformer rating (standard power classes)

Most of the distribution grids identified from the literature already originate from an upstream cluster process. Therefore, an individual weighting factor was assigned to them, depending on the level of detail of information on the topology and the number of topologies represented in each case. From the calculated parameters per distribution grid and the defined weighting factors, the mean values of the parameters were calculated, which were used to model the reference grid topologies.

For the modelling, the goal was set to represent the average distribution grid with a specific characteristic on the one hand, but on the other hand also to consider topologically more extreme characteristics or potential grid overload states in the reference grids. To achieve this goal, the entire calculated bandwidth of the reference grids was considered in the dimensioning and allocation of the branches of the reference grid topologies. The branches were modelled according to the following pattern:

• Branch 1: Maximum branch length with a maximum number of GCPs per branch.
• Branch 2: Minimum branch length with a minimum number of GCPs per branch
• Branch 3: Average line length with an average number of GCPs per line.
• Branch 4-n: Randomly normally distributed branch length, around a medium branch length with an equally randomly normally distributed number of network connection points per branch.

This distribution of the branches makes it possible to represent the average character of a reference grid, whereby the bandwidth of the reference grids was also considered by integrating one circuit each with a minimum and maximum branch length. The total grid length and the average number of GCPs were considered as boundary conditions in the modelling to represent the average grid about these parameters.

The grids were modelled as radial grids, as the meshes in the reference topologies depicted situations that were too individual to be standardised. Standard aluminium cables were modelled as cable types (cross-section: 4×150 mm² in the distribution 4×50 mm² at the GCP). The calculated number of GCP per string was distributed equally along the strings, thus standardising the average distances. The transformer rating was rounded from the mean value to the nearest standard transformer size. Test simulations with a scenario representing the status quo of the components were carried out to check whether the transformers were sufficiently large. In the case of a disproportionately high load, a transformer from the next higher power class was integrated. These methodological assumptions resulted in a reference grid topology for each of the defined meta-clusters, which are structured similarly in their basic topology. Figure 1 below shows an example of one of these modelled topologies.

By modelling the reference grids, the characteristics of the distribution grid types identified in the meta-clusters could be identified, which ultimately enables the simulation and evaluation of grid load conditions. It must be taken into account that the reference grids cannot represent the diversity of German low-voltage grids, but the analyses carried out and the modelling represent a further step towards typification. In the download area, the modelled topologies, equivalent to the meta clusters, are presented in individual profiles. In addition, the modelled reference grids are made available as node-edge models (in the OpenDSS standard) on GitLab. In the application/use of these topologies for further analyses/studies, it should be taken into account that these topologies are not suitable for mapping real grid load conditions in individual LV grids. Rather, they enable the comparison of regions/scenarios, such as the assessment of regional impacts of the ramp-up of electric mobility, or the extrapolation of impacts, such as grid expansion in rural regions, preferably using a Monte Carlo simulation. Further information on the advantages/disadvantages and the sensible application of reference grids is published in [3].