| Summary: | The market for plug-in electric vehicles is expected to grow significantly over the next few years as a number of automobile manufacturers have released electric vehicle models onto the market. The charging demand of wide-scale use of EVs may have a significant impact on domestic electricity loads and could risk overloading the power distribution system unless appropriate charging strategies are applied to prevent this. In order to quantify the future electric vehicle charging demand, it is necessary to gain a good understanding of current privately owned car use. In this thesis, domestic car use patterns have been studied in detail by analysing the United Kingdom Time of Use Survey 2000 data. The key findings show that weekday car use patterns are rather different than weekend ones. The majority of domestic cars are used for commuting to work during week days. Car activities, such as depart from home and arrive home are highly correlated and dependent on time of the day. Cumulative driving times are significantly dependent on the car arrival time. In most research, the relationship among these types of events are often ignored, which leads to errors in the calculation of charging demand. Three high resolution Monte Carlo simulation models are structured based on these domestic car use statistics in order to represent the weekday car use patterns; they represent three different approaches to trying to capture the complex dependencies associated with car use. The return time dependent Monte Carlo model utilise car returning home probabilities and the cumulative driving period dependent on arrival home time statistics. The single time increment Monte Carlo model uses two-state probability distributions of car departure and arrival to reproduce the weekday car location status. Although the correlation between car departure and arrival home events are not explicitly captured in this model, the multiple time increments Monte Carlo model captures this relationship by sampling from car away and parking period time dependent probability distributions. xviii Validation of the simulation results shows that all three models generate acceptably accurate car use patterns with home as the primary parking location. In the later part of the thesis, assessments the impact of electric vehicle home charging on the distribution network have been performed for two case studies; one focuses on the peak load impact on substation transformers, and the other one examines individual household voltages (at 230V low voltage level). For the specific network considered, it is shown that distribution substation transformers (i.e. primary and secondary) will face increasing peak load due to electric vehicle charging in the case that householders start charging as soon as they arrive home. It is recognised for the first time that domestic car use behaviour has effects on the household electricity consumption model and to reflect this the household electricity model has been modified to account for the changes in occupancy associated with car movements. In the low voltage case study, the household voltage issue has been investigated for this specific network by performing power flow calculations, and shows that a household, located at the end of a long service cable, suffers under voltage before the substation feeder reaches its thermal limit. In the last part of the thesis, several vehicle charging strategies have been developed to mitigate the problem of overloading the substation transformer. It is shown that a simple time delay to charging strategy creates an additional peak load on the substation loading profile; however, with a random time delay, overloading of the substation can be avoided. The potential role of EVs as responsive demand has been explored with the aim of utilising vehicle charging to support local power system operation with surplus wind generation. An algorithm is proposed that can effectively shift vehicle charging by implementing a linear cost function to track the surplus wind generation.
|
|---|
