Abstract
The electrical vehicle (EV) fleet is increasing in Norway. To plan and operate the long-term power system and evaluate EVs’ effect on the power grid, accurate load-profile generation models are needed. Such models are also needed to analyse optimal EV charging strategies. The purpose of this thesis is to develop a model to simulate realistic hourly load profiles for dumb private home charging, based on real-world EV-charging data. The data is provided by charging reports from charging point operators (CPOs), and gives information on the date, user type, user ID, plug-in and plug-out time, connection time, and charged energy for every measured charging session. Analysis of the data reveals that the factors EV type and day type impacts the EV user charging habits, as such these factors are considered in the model. The model is a stochastic bottom-up model, providing single load profiles for EVs being charged at home in Norway for a year. The load profiles depend on two types of EVs defined as “large EV” and “small EV”, referring to the battery size of the car. It is possible to simulate any number of EVs and composition of EV stock. In addition, information for plug-in and plug-out time, charged energy, charging frequency and idle hours for each EV user is extracted when running the model. Three different cases simulating load profiles for 1000 EVs are used to analyse and evaluate the model: BASE, LOW, and HIGH. In LOW, the EVs are assumed to be “small EVs” with a maximum charging power of 3.6 kW. In HIGH, EVs are assumed to be “large EVs” with a maximum charging power of 7.2 kW. In BASE, the battery sizes and maximum charging powers reflect the composition of the EV stock of the data set and combines the two other cases. The simulation results show that the aggregate load profiles have the same shape in all three cases, and the daily average peak power occurs at the same time for the different day types: between hour 17 and 18 on weekdays, between hour 18 and 19 on Saturdays, and between hour 18 and 19 on Sundays. As the load profiles presumes dumb charging, they reflect the distribution of the plug-in time for the different day types used in the model. The power peak and annual energy need are largest in HIGH and smallest in LOW, while BASE is between the two. The results validate that the model can account for factors such as charging frequencies and energy need being dependent on the EV type. This is also seen in the idle hours and shiftable energy levels. Even though the idle hours are higher in LOW, the shiftable energy level is higher in HIGH. To further study the model, load profiles are simulated for the same cases assuming flexible charging. In this thesis, flexible charging means distributing the charged energy equally over the connection time. Compared to the load profiles for dumb charging, the peak powers are reduced by 35-38%. In addition, they are moved to occurring at night in all three cases. It is a perception that EVs with large EV batteries and high maximum charging powers are preferred if using EVs as a flexible source in the grid. From this work, it is seen that EVs with these characteristics have fewer idle hours and are therefore a less flexible resource. When planning to use EVs as a flexible source, it is important to be aware of this trend. All in all, the model generates realistic results for the aggregate load profile. However, to make it more robust, more charging data should be analysed and included.