Development of Grid Load at the Municipal Level – Estimating the Need for Grid Expansion and Grid-Oriented Flexibility

In the previous article, we introduced a novel method for estimating low-voltage distribution grids at the municipal level. In addition to the previous analysis, we used these synthesized grids to estimate the expected increase in grid load in the low-voltage distribution grids due to the addition of new decentralized generation plants and electric consumers. The results are presented in the form of map diagrams at the municipal level.

Motivation and Objectives

The energy transition most noticeably takes place in low-voltage distribution grids, through the addition of renewable generators but also through new electricity consumption habits. However, there is very little reliable information on the effects of the energy transition on the low-voltage grids, as these grids have hardly been digitalized to date. Further, there are hardly any concepts in German regulation that address the resulting challenges at the local level. With the provision of § 14a EnWG, the BNetzA has introduced two tools to ensure grids stability in critical situations through the flexibility of controllable consumers. On the one hand, grid operators receive an instrument with grid-oriented control to limit the power consumption of controllable consumers to up to 4.2 kW, thus solving acute consumption-driven bottlenecks. In addition, the so-called incentive module is intended to create the possibility to reflect the grid load via time-variable grid fees, thus achieving a proactive shift in consumption to non-critical times. However, both grid-oriented control and the dynamization of the grid fees are instruments that aim to reduce consumption. A tool for making generation more flexible or for the efficient use of local RE surpluses has not yet been anchored in regulation.

A promising possibility to coordinate consumption and generation locally, relieve the grids and involve citizens in the energy transition, are local energy communities [2]. The research project PEAK (funding code: 03E16035F) has set itself the goal of developing and testing such a platform. Similarly, platforms on which flexibility products – adjusting generation and consumption – are traded, could address the aforementioned challenges. The FfE is specifically investigating the regionalized potential for these concepts within the framework of the project. For this purpose, we want to analyze the effects of the addition of decentralized assets and at the same time provide an approach for classifying what this addition means in terms of grid load. In this article, we specifically address the following questions:

  1. According to which guidelines are low-voltage grids dimensioned today? Can their power reserves be estimated in order to cope with the ramp-up of renewable generation and electrical consumers?
  2. How will the load peaks in the synthesized low-voltage grids develop as a result of the assumed ramp-up?
  3. Will the need for grid expansion in the synthesized grids be driven to a greater extent by the addition of decentralized RE or new electrical consumers such as heat pumps or electric vehicles?
  4. To what extent are energy communities a suitable tool to leverage flexibilization potential at the low-voltage level?
  5. Can we identify and explain regional differences in the need for network expansion?

Simulation Environment & Scenarios

To consider the future increasing grid load and the associated flexibility potentials, the grid load is determined at the level of the low-voltage grids in two scenarios. In the simplified “NoFlex” scenario, it is assumed that there are no home storage systems (HSS), household heat pumps (HP) or electric vehicles (EV) in the system. For photovoltaic rooftop systems in the household sector, the stock from 2019 is used in this scenario. In the Flex 2035 scenario, a certain addition of these assets and their non-flexible use is assumed. The differences between the two scenarios in terms of (peak) load enable a quantitative consideration of the increase in grid load. Thus, the resulting potential for flexible operation of the added technologies can be demonstrated. A detailed description of the underlying simulation framework, the scenarios, and the clustering approach used to synthesize the low-voltage grids can be found in the previous article.

Status Quo & Estimation Method

The synthesized low-voltage grids are used to determine a total load from the load profiles of the individual grid connections per grid. This metric is crucial, as we want to assess the strain put on middle to low-voltage transformers. The maps derived from this in the following sections can provide grid operators with an estimation of regions where the addition of new decentralized generation plants or electric consumers makes grid expansion a relevant problematic. In our maps, we therefore always refer to the grid of a municipality that records the highest peak load (worst-case estimate). However, for the question of the necessity of grid expansion, the peak load alone is not a meaningful metric to determine the criticality. The capacity of the installed transformers is also decisive. Unfortunately, the information on the components installed in the field is only very limited, so that an evaluation of the grid load at the distribution grid level for all of Germany is not possible. In general, it can be stated that the data situation in medium and low voltage remains poor. Measured values, e.g. of current and voltage (e.g. from intelligent local network stations, smart meters, or sensors) are still a rare exception in the low-voltage sector.

The currently installed transformers usually do not follow a strict dimensioning specification but were installed based on the currently available models. This can be seen from the clear scattering in Figure 1, whereby the representation also reveals that at least in the present (geo) data set, which comprises several thousand real low-voltage grids, all grids are currently over-dimensioned to a certain extent and still have power reserves at the transformer. The dimensioning guideline drawn in the figure is based on the methodology from [2] and only considers private households. In addition, the grids contained in the data set come from a spatially limited area and thus do not represent a representative sample, which is why no generally valid assumptions can be derived about the existing power reserves.

Figure 1: Real-world transformer capacities for a set of low-voltage grids and comparison to the dimensioning prescription

Development of Peak Loads

As can be seen in Figure 2, the annual peak loads that act on the transformers of the low-voltage grids will increase significantly throughout Germany by 2035 due to the addition of production plants and consumers, although strong regional differences can be seen. The map provides a first insight into which regions are likely to come under the most pressure and where grid expansion should be evaluated in the near future. However, as part of the project, we also want to estimate whether grid expansion can possibly be reduced or even avoided by making consumption more flexible.

Figure 2: Relative increase in maximum peaks between the NoFlex 2019 and Flex 2035 scenarios. Peaks will be 2, 3, or even 5 times as high as in 2019 in many grids by 2035, which could be a challenge for many grids.

To answer this question, we consider the following aspects:

  • Depending on whether the peaks are caused by consumption or generation, there exist different approaches for flexibilization (see introduction).
  • If the peaks only last a few hours, they can probably be smoothed out by flexibilization of supply and demand, thus avoiding, or at least delaying grid expansion. However, if the peak loads last for several hours or even several days, there is very likely no possibility of avoiding grid expansion.

Influence of decentral renewable energies on peak loads

In all grids, we see a simultaneous increase in decentralized generation and consumption. However, in the future, there will be an imbalance between generation and consumption (so-called Supply-Demand-Ratio; SDR) in most grids. If we look at the SDR over the entire year (not shown), we can first see that there are strong regional differences. Currently, most grids consume more than they feed in. In the scenario “Flex 2035”, the situation looks significantly different: many distribution grids are closer to balance and generate about as much as they consume over the entire year. This balance over the entire year, however, does not say anything about the utilization of the grid. Looking at individual hours, the SDRs are significantly more extreme, as the local simultaneity of renewable generation (especially wind and PV) in particular leads to large fluctuations here. Alternating high generation and consumption peaks can indeed occur, even with a balanced SDR. In Figure 3, we therefore show the ratio between the highest feed-in and the highest consumption peaks for both scenarios for all municipalities.

Due to the focus of our investigations on the potential for energy communities, currently only renewable energies and private households are considered. Commerce and industry as well as conventional generation but also large RE generators (because not connected to the low voltage) play no role here.

Figure 3: Ratio between the highest generation peak and the highest consumption peak in the NoFlex 2019 scenario (left) and Flex 2035 (right). It becomes clear that the grids were predominantly burdened by consumption in 2019, while in 2035 they will be predominantly burdened by generation peaks.

It becomes clear that in the NoFlex 2019 scenario, the highest peaks were caused by consumption (Ratio<1) in most municipalities. In stark contrast, in our Flex 2035 scenario, the peaks in most municipalities are caused by generation (Ratio>1). The expected ratios between the peak heights for Flex 2035 often lie between a factor of 1.5 and a factor of 3, and in very rare cases even higher, which underlines the strong influence of decentralized renewable energies on grid load. This is also in line with the statements from [3], according to which the need for grid expansion in the distribution grids is driven in particular by PV expansion. For this reason, distribution grid operators are currently allowed to take into account a curtailment of 3% of the annual generation when planning expansion, which should significantly reduce the need for expansion [4]. However, our analysis did not consider the effects of dynamic tariffs, for example, which can also lead to higher simultaneity on the consumption side and therefore higher peak loads.

Potential Analysis of Grid-Serving Flexibility

The question of whether grid expansion can possibly be avoided by making consumption more flexible is particularly interesting. An important part of the answer lies in the duration of the peak loads, as the consumption of electric vehicles, heat pumps, or home storage systems can be relatively easily shifted over periods of a few minutes to a few hours. For longer periods, however, this shift potential is limited, as prosumers usually do not have large storage facilities, cannot do without their car for several days, and buildings cool down after a certain time. In addition, there are catch-up effects, especially with heat pumps, so that the peak load can increase quickly at the end of a power reduction or an expensive price signal.

As mentioned at the beginning of the article, we must make assumptions about transformer capacity when estimating grid load. In Figure 4, we present the potential for saving grid expansion through the generalization of flexible consumption under various assumptions about transformer capacity. For the maps in Figure 4, we differentiate between generation (yellow) and consumption peaks (blue), as the measures for peak shaving also differ. We define all events as “peaks” in which the total grid load exceeds the assumed transformer capacity.

Additionally, we use the following rules to determine the severity of a load peak: (The assumed time periods are based on the scenarios from [5])

  • If the peak lasts less than 2 hours, peak-shaving is probably relatively easy
  • If the peak lasts between 2 and 6 hours, it is probably hard to shave
  • If the peak lasts more than 6 hours, it can probably not be shaved with simple measures, and grid expansion probably can’t be avoided
Figure 4: Potential for Peak-Shaving in the scenario Flex2035 through the use of flexible loads, assuming different transformer capacities in relation to peak load in the NoFlex 2019 scenario (1x to 5x as high)

If many grids were already close to their nominal capacity in the NoFlex scenario in 2019, they will quickly reach their limits due to the addition of decentralized plants: under this assumption, peaks of over 6 hours occur in most municipalities and grid expansion seems indispensable. However, if there are currently larger power reserves at the transformers, e.g. still three times as much power as the highest peak in 2019, then in Flex 2035 most municipalities only experience short peaks, which can very likely be remedied by making consumption more flexible, suggesting potential savings in grid expansion.

The map is intended to provide grid operators who know the power reserves of their transformers with an initial estimate of the expected expansion/flexibilization requirements. The maps in the figure also make it clear that, depending on the available power reserve, generation peaks in particular offer great potential for peak shaving, provided that the appropriate tools for the flexibilization of consumption are implemented. As described in the introduction, existing flexibility mechanisms in the low voltage grid are not able to address this problem. And although home storage systems, for example, offer great potential for flexibilization, the currently widespread surplus charging rarely contributes to grid relief.

On sunny days, the storage systems are usually full after just a few hours, meaning that the PV systems feed full power into the public grid at midday, which leads to overloads [6]. To date, there has been no financial incentive to cover this midday peak by adjusting consumption patterns. However, energy communities, which form a price signal based on the local residual load, could incentivize this flexibilization. If the price on the local market falls during midday hours due to high PV feed-in, producers would have an incentive to fill their batteries at these times and to feed back into the grid in times of a low SDR, thus contributing to smoothing the production peaks.

Discussion: Regionalized statements and need for further research

As can already be seen in Figures 2-4, the three metrics for estimating grid utilization in 2035 – peak increase, SDR of peaks and duration of peaks – appear to be strongly correlated. There are no significant differences between the federal states. In particular, and somewhat surprisingly, the wind-dominated north of Germany and the PV-strong south differ little from each other in terms of those metrics.

Based on the clustering method of [7], the German municipalities can be divided into four clusters.

  • Cities (or zones with high population density and comparatively low RE potential)
  • Municipalities with a low population and low RE potential
  • Municipalities with a low population and high wind power potential (typically in northern Germany)
  • Municipalities with a low population and high PV potential (typically in southern Germany)

We are particularly interested in the question of whether the expected future grid load differs between the two latter categories. To answer this question, we analyze the relationship between the SDR in 2035 and the longest peak load in 2035, assuming that the transformers’ maximal capacity is two times their respective maximum utilization in 2019 (Figure 5). The same findings also hold for other assumptions regarding the installed transformer capacity (results not shown here).

Figure 5 clearly shows that the SDR 2035, as a measure for the installed RE capacity, also strongly defines the length of the load peak (R²~0.8). As a second finding, Figure 5 shows no significant difference between PV-dominated and wind-dominated rural municipalities in terms of their peak length development at increased RE roll-out rates.

However, it can be assumed that the strategies for peak smoothing will differ between these municipalities. For example, generation peaks for PV systems are always expected to occur around midday, whereas they can occur more variably in time for wind power plants. In order to properly distinguish these differences and formulate recommendations for an optimal peak smoothing strategy, the load profiles over time of the different municipality types would have to be compared. The simulation environment ([8]) developed as part of the project InDEED enables the investigation of this research question, which would be a promising starting point for future research projects.

Figure 5: Comparison of wind-power-dominated vs. PV-dominated rural municipalities with regard to their longest peak load in 2035, assuming that the transformers were only utilized at <= 50% of their maximum capacity in 2019. The same findings remain valid for other assumptions on the installed transformer capacity (not shown). The relationship between PDR ratio and longest load peak is approximately linear (coefficient of determination of the linear fit R² ~ 0.8), with marginal differences in slope between PV and wind power communities.

Discussion and Recommended Action

It is important to point out once again that the maps shown do not represent the exact grid expansion requirement, as uniform assumptions had to be made about transformer performance. In reality, the transformers installed today are tendentially oversized, with the power reserves showing a strong scatter. We hope to be able to enrich our method in this regard successively with data on the real installed transformer capacities, which, however, first requires a comprehensive digitalization of the low-voltage grids.

Under the uniform assumptions made about existing power reserves, the load profiles generated by the InDEED simulation environment were analyzed. The German municipalities were compared regarding potential overloads of their local grid transformers during the RE-roll-out until 2035. The results show that, regardless of the type of additional RE capacity (PV/wind), the need for expansion of a large part of the grids is driven by increased renewable generation and not by additional electrified loads.

We therefore recommend promoting local energy communities in Germany more strongly, as they not only represent an effective instrument for strengthening the participation of prosumers in the energy transition but can also incentivize the consumption of RE surpluses locally and thus reduce peak loads.



[1] Wiesenthal, Jan: Energy Sharing: Eine Potenzialanalyse. Berlin: Institut für ökologische Wirtschaftsforschung (IÖW), 2022.

[2] Müller, M.; Biedenbach, F.; Reinhard, J.: Development of an Integrated Simulation Model for Load and Mobility Profiles of Private Households. Energies 2020, 13, 3843.

[3] Flechtner, Jakob: Faktenpapier Ausbau der Stromnetze – Grundlagen – Planungen – Alternativen. Berlin: Deutscher Industrie- und Handelskammertag e.V., 2017

[4] Spitzenkappung – ein neuer planerischer Freiheitsgrad – Möglichkeiten zur Berücksichtigung der Spitzenkappung bei der Netzplanung in Verteilnetzen – FNN-Hinweis. Berlin: Forum Netztechnik / Netzbetrieb im VDE (FNN), 2017.

[5] Elberg, Christina, Frings, Cordelia, Jeddi, Samir, Sitzmann, Amelie: Kurzstudie: Flexibilitätspotenzial von Haushalten zur netzdienlichen Reduktion von Nachfragespitzen. Köln: ewi Energy Research and Scenarios gGmbH, 2018.

[6] Agora Energiewende und Forschungsstelle für Energiewirtschaft e. V. (2023): Haushaltsnahe Flexibilitäten nutzen. Wie Elektrofahrzeuge, Wärmepumpen und Co. die Stromkosten für alle senken können.

[7] Bogensperger, Alexander Johannes: Assessing the Potential of Multiple Use Cases for German Energy Communities via Integration of Machine Learning in the Energy-Economic Modeling Process. Dissertation. Herausgegeben durch Technische Universität München – TUM School of Engineering and Design der Technischen Universität München: München, 2023.

[8] Bogensperger, Alexander et al.: Comparison of Pricing Mechanisms in Peer-to-Peer Energy Communities. In: 12. Internationale Energiewirtschaftstagung (IEWT) 2021. Wien: Technische Universität Wien, 2021.