The main focus of this thesis is to minimise power loss in the Distribution Network (DN) caused by Electric Vehicle (EV) penetration. When large numbers of EVs are connected to the DN, the power loss in the DN increases dramatically. In order to reduce this power loss, optimal active and reactive power dispatch methods of Energy Storage System (ESS) and charging stations are proposed in this study.
This study develops a new active and reactive optimal power dispatch method using ESS to reduce the power loss caused by EV penetration. A power flow analysis model of two ESSs was built in order to minimise the power loss. Two sub-methods based on this optimal power dispatch method are presented. These are uncoordinated optimal active-reactive power flow of the ESS and coordinated optimal active-reactive power flow of ESS. These two methods were tested in an IEEE 33-bus DN. The power loss was compared with and without optimisation methods: meanwhile, the power loss caused by the EVs was quantified. The simulation results show that by using the proposed method in this study, 1.43 MW of total power loss can be reduced and 1.64 MW of active power does not need to be imported from the DN.
This study also develops a novel analytical location choosing method based on optimal active and reactive power dispatch and power flow analysis for optimal placement of charging stations. A concept of charging stations combined with Battery Energy Storage Systems (BESSs) is given. The analytical method is compared with the current density method used in other papers. The results show the analytical location method was more accurate than the current density method for finding the optimal location of charging station two’s location. This analytical method was tested in an 11-bus distribution line, the IEEE 33-bus DN and a 36-bus DN. The simulation results proved the accuracy of analytical method used in this study. Moreover, 27% of the average active power loss was saved by installing two charging stations rather than no charging stations in the test-line. It is also shown that a 2.6% annual yield above inflation can be obtained by investing in installing and running such charging stations.
In order to analyse how the optimal location of charging station two changes with different impact factors, such as different EV charging locations, line resistance, reactance and different loads, a quantitative analysis was carried out by using the proposed active and reactive method. The 36-bus DN was used as the test network. The results showed that the optimal location only changed when all the impact factors were changed simultaneously.
In addition, to reduce the calculation time and find more optimal charging station locations, a Genetic Algorithm (GA) was developed to find multiple (>2) charging stations’ optimal locations for power loss minimisation. The GA was tested in the 36-bus DN, which found that 6 charging stations were optimised. Meanwhile, the GA’s different settings, such as population size, cross over probability, mutation probability and the stopping criteria were changed in order to analyse how these settings influenced the GA’s performance. Moreover, the calculation time of traditional quadratic optimisation method and GA calculation time was compared. From the comparison, the GA was 22 times faster than the quadratic optimisation method for finding the six charging stations’ optimal locations in the 36-bus DN case.
|Date of Award||24 May 2017|
|Supervisor||Roderick Dunn (Supervisor) & Weijia Yuan (Supervisor)|