This paper provides empirical data and a profitability estimation of public charging infrastructure usage in Germany. Given that, in Germany, there are now 2.5 times as many vehicles per charging station compared with 2017, the system needs to allocate charging points efficiently. To this end, this paper presents representative data on energy consumption, arrival times, occupation, and estimated profitability of 22,200 charging stations in Germany. The observed patterns are translated into compact empirical models that allow working with the results without the burden of the large-scale datasets. Charging happens mainly during the day and on weekdays for AC charging stations, whereas DC fast-charging stations are more popular on the weekend. Fast-chargers service approximately three times as many vehicles per charge point at higher profits because of better margins. For AC chargers, up to 20 kWh of energy are charged in an average charge event, whereas fast-chargers supply approximately 40 kWh. Energy transfer typically terminates after 4 h for AC chargers and 45 min for fast-chargers. The power rates are significantly below the rated station power and rarely exceed 11 kW for AC charging. This paper allows fellow researchers to build simulation and test scenarios using presented data or to verify models.

Home storage systems play an important role in the integration of residential photovoltaic systems and have recently experienced strong market growth worldwide. However, standardized methods for quantifying capacity fade during field operation are lacking, and therefore the European batteries regulation demands the development of reliable and transparent state of health estimations. Here we present real-world data from 21 privately operated lithium-ion systems in Germany, based on up to 8 years of high-resolution field measurements. We develop a scalable capacity estimation method based on the operational data and validate it through regular field capacity tests. The results show that systems lose about two to three percentage points of usable capacity per year on average. Our contribution includes the publication of an impactful dataset comprising approximately 106 system years, 14 billion data points and 146 gigabytes, aiming to address the shortage of public datasets in this field.

The current increase of electric vehicles in Germany requires an adequately developed charging infrastructure. Large numbers of public and semi-public charging stations are necessary to ensure sufficient coverage. To make the installation worthwhile for the mostly private operators as well as public ones, a sufficient utilization is decisive. An essential factor for the degree of utilization is the placement of a charging station. Therefore, the initial site selection plays a critical role in the planning process. This paper proposes a charging station placement procedure based on real-world data on charging station utilization and places of common interest. In the first step, we correlate utilization rates of existing charging infrastructure with places of common interest such as restaurants, shops, bars and sports facilities. This allows us to estimate the untapped potential of unexploited areas across Germany in a second step. In the last step, we employ the resulting geographical extrapolation to derive two optimized expansion strategies based on the attractiveness of locations for electric vehicle charging.

Electric vehicles may reduce greenhouse gas emissions from individual mobility. Due to the long charging times, accurate planning is necessary, for which the availability of charging infrastructure must be known. In this paper, we show how the occupation status of charging infrastructure can be predicted for the next day using machine learning models— Gradient Boosting Classifier and Random Forest Classifier. Since both are ensemble models, binary training data (occupied vs. available) can be used to provide a certainty measure for predictions. The prediction may be used to adapt prices in a high-load scenario, predict grid stress, or forecast available power for smart or bidirectional charging. The models were chosen based on an evaluation of 13 different, typically used machine learning models. We show that it is necessary to know past charging station usage in order to predict future usage. Other features such as traffic density or weather have a limited effect. We show that a Gradient Boosting Classifier achieves 94.8% accuracy and a Matthews correlation coefficient of 0.838, making ensemble models a suitable tool. We further demonstrate how a model trained on binary data can perform non-binary predictions to give predictions in the categories “low likelihood” to “high likelihood”.