
With the accelerating electrification of the transportation system, it becomes increasingly important to understand charging behaviour and how it can align with the needs of the electricity grid. Real-world data from logged EVs shows that there is significant potential for charging flexibility.
The SEC project “Logging of Electric Vehicles – Characterization of Charging Patterns and Grid Impacts” aims to estimate the impact of electric vehicles on the electricity system in Sweden. If charging is not managed, it risks creating additional peaks in electricity demand. However, with smarter charging and vehicle to grid solutions, EVs can help to balance the grid.
Yuki Kobayashi, PhD student in the project, has studied charging behaviour using logged real-world data from about 400 EVs over the period of one year.
“We often rely on assumptions in energy system modelling,” Kobayashi explains. “But without real-world data, we do not know how people actually behave. Data from charging stations alone only shows when charging happens, not why. By combining charging data with driving patterns and user characteristics, we can better understand the underlying behaviour and improve energy system modelling.”
“There are clear differences between different types of users. For example, EV owners living in houses often have hourly electricity contracts and actively schedule charging when prices are low. In contrast, people living in apartments typically do not have direct control over charging costs and tend not to adjust charging behaviour based on electricity prices,” says Kobayashi.
EV batteries offer room for smarter charging
Charging flexibility is the possibility to shift charging in time to better align with the needs of the electricity system. The flexibility is closely linked to how much energy that is left in the battery, which is called state of charge (SOC).
In Kobayashi’s study, the aggregated SOC of all logged EVs generally remained between 60% and 80%, while the average flexible battery capacity range of individual vehicles was estimated to be 59% of battery capacity. This suggests that many EVs have sufficient energy available to allow charging to be delayed or shifted to more suitable times without affecting everyday driving needs.
“The data shows significant potential for flexibility. Around 90% of EVs are parked at home overnight, but only about 40% are actually charged during those periods. This means that many vehicles are not plugged in, which limits flexibility. If more vehicles were connected to chargers during parking, the potential for flexible charging would increase substantially,” says Kobayashi.
Charging can often shift without affecting users
“Another important finding is that in about 70% of weeks, EVs only need around three or less charging events. This indicates that charging can often be shifted without affecting users,” Kobayashi further explains.
The vehicles in Kobayashi’s study were selected across all of Sweden with the help of Statistics Sweden, and represent different types of EV users. These included EV owners living in apartments and houses, as well as in large cities, small towns and rural areas.
“The data can be used in many types of energy system models, from household-level analysis to national-scale simulations. It can also be applied to studies of frequency regulation and battery degradation. The dataset is representative at a national level and remains useful for regional analysis. However, sample sizes are smaller in regions with smaller car populations, which may reduce the precision of regional estimates,” Kobayashi concludes.
About Yuki Kobayashi
Before joining Chalmers University of Technology as a PhD student, Yuki Kobayashi worked as a researcher at Nissan in Japan and also spent time in Sweden during an earlier exchange at KTH.
“Sweden is further ahead in terms of electric vehicle adoption, especially for private cars. It provides a valuable environment for studying how EVs interact with the electricity system.”
Yuki Kobayashi’s licentiate thesis is available to download from here: https://research.chalmers.se/en/publication/552119

