The electrification of road transport has the potential to mitigate local traffic emissions, reduce dependency on fossil fuels and support the grid integration of renewable energies. With large market penetration of plug-in electric vehicles (PEVs), however, there will be a significant additional demand for electricity which may have detrimental effects on power grid stability (Dharmakeerthi et al., 2014). These could be avoided either by expanding the infrastructure or by implementing smarter charging strategies.
In a previous post we described the framework we are developing to investigate the impact of electric vehicle charging on the power system. This approach consists of coupling an agent-based traffic simulation with a time-stepped power system simulation through the IEEE Standard 1516-2010 (High Level Architecture)
Using this framework we conducted a first study for the case of Singapore. A realistic investigation is difficult due to the fact that detailed information on the power system could not be obtained. Therefore, we artificially generated a power grid based on tempo-spatially resolved power demand data. This power grid exhibits the typical topological characteristics of power systems and can therefore be considered a rough approximation to the real world. While this does not allow definite conclusions on the particular issues arising from electric vehicle charging in Singapore, it serves as a preliminary proof-of-concept of the developed approach and can furthermore give a few hints on possible bottlenecks in the power system.
The study was conducted for a few different scenarios considering different market penetrations of electric vehicles as well as different charging strategies. The scenarios considered comprise market penetrations of 0%, 5% and 100% of the private vehicle fleet. Furthermore, scenarios with 3.6 kW household charging, 120 kW superfast charging and variable power mean charging are also considered. The latter one distributes the charging process over the entire duration of a stop allowing completely recharging the battery at a minimum peak power.
As shown in Figure 1 (left), the results indicate that the peak power demand is virtually unaffected by electric vehicle charging. This indicates that no supply bottlenecks need to be expected even if the entire pool of private vehicles is electrified. The same conclusion applies to the high voltage and low voltage grid which do not show any capacity bottlenecks regardless of the number of electric vehicles. In contrast, the LV layer may be negatively affected by uncoordinated charging of large numbers of PEVs resulting in grid congestion and voltage drops. This can be seen in Figure 1 (middle) which shows that a certain share of branches is overloaded in case of 100% market penetration of electric vehicles. The comparison of the different charging strategies in Figure 1 (right), however, indicates that an even larger numbers of PEVs can be integrated into the power grid if appropriate measures for smart charging were taken. Thus through the implementation of coordinated charging strategies it is possible to handle much large numbers of electric vehicles.
When interpreting these results, it, however, needs to be considered that the artificial power network is not an exact representation of the real world. More accurate quantitative conclusions could therefore be drawn with access to greater amounts of real-world data. This should include data on the power grid, especially exact information on actually laid power lines and their overcapacities in the distribution network.
 D. Zehe, A. Knoll, W. Cai, and H. Aydt, “SEMSim cloud service: Large-scale urban systems simulation in the cloud,” Simulation Modelling Practice and Theory, 2015.
 D. Ciechanowicz, H. Aydt, and A. Knoll, “SEMSim power as an application of USES,” in Proc. IASTED Int. Symp. Power and Energy 2013, 2013.
 H. Seifi, M.S. Sepasian, “Electric power system planning: issues, algorithms and solutions” Springer Science & Business Media, 2011.