Probabilistic Load Profile Model for Public Charging Infrastructure to Evaluate the Grid Load
In the project “Munich electrified”, modelling of the load of public charging infrastructure represents a central component for assessing the future grid load resulting from electric mobility (in addition to “charging at home”  and “charging at work” ). Within the scope of a publication, the grid load of public charging processes was modelled and evaluated. This publication is fully available as open-access-paper at the following Link. The contribution was peer-reviewed, with the review also being published in a fully transparent format (open-review). In addition to the publication, a model was implemented and published under public license.
The development of electric vehicle registrations over the past years associated with transformation of traffic sector requires expansion of charging infrastructure in Germany. This expansion is associated with challenges. Especially in densely populated areas, conditions for installation of charging points on private property or at the workplace are often not given, e.g. for reasons of limited space. One solution to this challenge is the installation of charging infrastructure at available public places, which can thus serve as an enabler for electric mobility. Between 440,000 and 843,000 charging points are the predicted growth of public charging infrastructure in Germany by 2030 . However, public charging infrastructure can currently only be operated economically if there is a high frequency of charging electric vehicles. Another challenge that arises is the integration of charging points and associated charging processes into the infrastructure and operation processes of distribution grids. In order to ensure reliability and stability of the future energy system, a continuous investigation of grid load caused by charging processes at public charging points is necessary. Modelling and simulation of public charging infrastructure can contribute as important predictors to address these challenges. To assess the distribution grid perspective, the presented publication addresses following research questions:
- What concepts are available for modelling grid load of public charging infrastructure in distribution grids and how do they distinguish themselves?
- Which approach is suitable for modelling load profiles in a “generally valid” way, and what are constraints?
- How significant is the additional load at the distribution grid level when integrating an “electric avenue” of public charging points?
Modelling charging processes of public charging infrastructure
Content of the contribution includes the topics incorporation into the research context as well as into the state of the art in science and technology, followed by a comprehensive description of applied methods to structure data set and model. This includes analysis and preparation of input data for calculation of probabilities, as well as creation of charging profiles by public charging infrastructure. This is followed by the application of the model in a case study, as well as a data-driven excursus on the influence of the COVID-19 pandemic on charging behavior. Finally, the publication and methods are critically reviewed and the published model is described and referenced.
According to literature, a stochastic approach was chosen for the basic structure of the models design. This choice was mainly supported by an available dataset of public charging data from all over Germany, which served as input data to the model . It was implemented in a modular way in order to enable it to be utilized both as a sub-module in the GridSim framework and to provide it to public in a stand-alone variation. The requirements lead to the model structure shown in Figure 1, divided into two interconnected, main modules. The core of the model is the “module for generating the probabilities”, which is used to generate electrical load profiles in the superordinate “module for modeling the electrical loads”.
In the “module for generating probabilities”, different probabilities are created starting from the basic data set, whereby individual loading processes are created via random variables. Fundamental to modelling these charging processes is probability of the start of a charging process, which depends on different factors. Various criteria (day of the week, seasonality, location of the charging station, recording period) were examined and corresponding probabilities for the start of a charging process were calculated. In the next instance, plug-in duration was analyzed and probabilities were calculated. This calculation of probabilities for a specific plug-in duration is performed as a function of the time of the day. Figure 2 presents the probability for plug-in duration of a charging process as a function of the time of the day.
Figure 2 shows an increase in probability of a short plug-in duration from the morning onwards, which decreases again towards the evening. This characteristic represents the expected pattern of short-term intermediate charging in public spaces during the day. At the same time, an increased probability of a long plug-in duration starting at midday and lasting into the night is identified, whereby in this pattern the duration of plug-in decreases the later it becomes in the day. This represents the expected pattern of vehicles occupying a public charging point throughout the night. In the simulation, a plug-in duration is calculated according to these probabilities, selected time intervals and the starting timestep of a charging process. Depending on calculated plug-in durations, charged energy is then determined and charging load curves are created.
Published model for the calculation of load curves
Within the scope of literature work, a “gap” of available models enabling calculation and consideration of public charging infrastructure from an electrical/energy engineering perspective was identified. In order to alleviate this deficit, the tool created, including calculated probabilities for start of charging, plug-in duration and charged energy, was published under a public license (FfE Munich). The tool thus enables probability-based calculation of charging processes and resulting charging load profiles.
As the publication clarifies, frequency and charging characteristics at public charging infrastructures are influenced by many factors. It is important to consider that restrictions associated with preparation of the data set also shape the calculated probabilities. Therefore, these only serve as an initial indicator for calculation of charging load profiles, but do not reflect local characteristics, such as spatial distinctions (charging point in municipality vs. charging point in large city). If alternative, more specific data on charging processes are available, individual probabilities can be calculated according to the methods described in the publication and then converted into charging load profiles in the provided MATLAB framework.
 Müller, M.; Biedenbach, F.; Reinhard, J. Development of an integrated simulation model for load and mobility profiles of private households. Energies 2020, 13, 3843.
 Weiß, A.; Müller, M.; Franz, S. Spitzenlastkappung durch uni- und bidirektionales Laden von Elektrofahrzeugen und Analyse der resultierenden Netzbelastung in Verteilnetzen. Forsch. Ing. 2021, 85, 469–476.
 Windt A.; Arnhold, O. Ladeinfrastruktur nach 2025/2030: Szenarien für den Markthochlauf; Study Commissioned by Federal Ministry of Transport and Digital Infrastructure; Nationale Leitstelle Ladeinfrastruktur: Berlin, Germany, 2020; p. 5.
 Zentrales Datenmonitoring des Förderprogramms Elektromobilität vor Ort. Available online: www.now-gmbh.de/aktuelles/pressemitteilungen/zentrales-datenmonitoring-legt-ergebnisse-vor/ (accessed on 4 January 2022).