Search

Electric Vehicles (EVs) are becoming a promising solution to the environmental problems caused by the gasoline powered internal combustion engine automobiles. The major challenges to the deployment of EVs are high initial cost, limited range, increased peak demand, etc. But with good charging infrastructure which is integrated into smart grid, most of the disadvantages can be eliminated. Moreover, the EV batteries can be considered as distributed storage and used for various smart grid applications. For integration of EVs into smart grid, the requirements and feasibility of different charging methodologies for EVs are to be investigated. Based on the investigation, economic and efficient models for EV charging along with the associated charging strategies are to be derived for efficient integration. The impacts of different charging methods on battery life and distribution system peak demand are the two most important factors to be considered while deriving the models. When the EVs are plugged into the grid for charging, all the operations including charging of their batteries and auxiliary functions such as grid support using Vehicle to Grid (V2G) and demand response management (DRM) are to be coordinated. This thesis presents a dynamic scheduling algorithm using novel priority criteria for integrating EVs into smart grid. The proposed dynamic scheduling algorithm is used for various other applications such as V2G capacity estimation and DRM. The scheduling algorithm is also used for verifying the methodology for estimating the number of EVs that can be supported by typical smart grid clusters with distributed energy resources. A novel data driven load model is used for determining the charging profiles of the EVs. Matlab based simulation models are developed to implement the proposed methods and illustrative case studies are used to evaluate the performance of the proposed system under a given set of conditions. The proposed dynamic scheduling algorithm uses model based scheduling approach to overcome the disadvantages of existing scheduling methods. The load models used are stochastic building load demand (without EVs) and predicted EV charging profiles. Predicting the charging profiles of EVs connected to a building incorporated with a Building Energy Management System (BEMS) will improve the energy efficiency of the building. The predicted charging profiles along with the stochastic load data can be used for calculating V2G capacity and for performing load/source scheduling. Data driven modeling has significant advantages in predicting the charging profiles of EVs, hence an Artificial Neural Network (ANN) based model is proposed for predicting the charging profiles of EVs connected to a building. The ANN model considers the previous charging profiles, initial State of Charge (SOC) and final SOC for predicting the charging profiles of EVs. Appropriate dynamic priority criteria required specifically for EV scheduling is also proposed. The algorithm is applied for both time coordinated EV charging and power coordinated EV charging. The objective function of the algorithm is to minimize the variance in priority values of the connected EVs. Minimizing the variance in priority values of the EVs will reduce the variations in fairness given to the EVs. The impact of different priority criteria and different combination of priority criteria on the fairness and chargeability of the EVs are also studied. Furthermore, the need for using weighted priority criteria is demonstrated. The proposed dynamic scheduling algorithm is also applied to the real-time V2G capacity estimation for a group of vehicles. Using scheduling for V2G capacity estimation is a novel approach and has significant advantages. The V2G capacity that a group of vehicles can provide is estimated by evaluating the chargeability of the EVs to desired final SOC. Using scheduling for determining the V2G capacity will have a greater significance as the accuracy of the estimation is not affected by the time at which the estimation is carried out. A comparison of the proposed method with other methods which do not employ scheduling for the V2G capacity estimation is presented to demonstrate the advantages. The proposed dynamic scheduling algorithm is also applied for novel pre-emptive DRM of EVs in a building. The load shedding of the EVs is carried based on the priority values given by the proposed dynamic priority criteria. The chargeability of the EVs is ensured before deploying the EVs for DRM. The dynamic priority based pre-emptive DRM of EVs is implemented as part of intelligent pre-emptive DRM in a building which ensures that contracted capacity or demand limit (CC/DL) is not exceeded and at the same time reduction of energy consumption in the building is achieved. A methodology which uses stochastic calculations (Monte-Carlo simulations) and linear programming for the estimating the optimal number of EVs that can be deployed under a high-rise building in Singapore is presented in this thesis. The proposed dynamic scheduling algorithm is used to validate the accuracy of the estimation. Load demand data for five years and load model of EVs are used for the validation. Although there are various obstacles for the deployment of EVs, unavailability of charging infrastructure is the biggest obstacle. Hence, the cost-benefit analysis for installation of charging stations in buildings is also discussed by varying different parameters. It is important to ensure that the number of EVs integrated into the smart grid does not bring any adverse effect. A novel method for determining the limits based on the hot-spot temperature of distribution transformer is also proposed. DOCTOR OF PHILOSOPHY (EEE)

Large-scale penetration of electric vehicles (EVs) would significantly increase the load requirements of buildings in highly urbanized cities. EVs exhibit higher degree of charging flexibility when compared to other interruptible loads in buildings. Hence, EVs can be assigned lower priority and interrupted before interrupting any other loads. Any temporary interruption will have minimum impact on EV owner's satisfaction/comfort. However, it should be ensured that the EVs could be charged to the owner's required state of charge (SOC) by the time of departure. The scheduling algorithms that are used to manage the EV charging process ensure that the charging requirements are fulfilled even when there are temporary interruptions. The capability of the scheduling algorithms to manage mismatches decreases with the decrease in time available for charging. In this paper, the impact of demand response management (DRM) on the chargeability of the EVs while using different priority criteria is examined. Subsequently, the proportion of interruption for each EV with different priority criteria and the need for determining the chargeability of EVs before shedding them are studied. A scheduling driven algorithm is proposed which can be used for determining the chargeability of EVs and can be used in combination with DRM. Published version

Coordinated charging is the superior charging method for integrating electric vehicles (EVs) smoothly and efficiently into the existing power system. Coordinated EV charging can be further classified into two types, namely time coordinated charging (TCC) and power coordinated charging (PCC). Scheduling of EV charging is the vital component in both types of coordinated EV charging methods. Generally, priority criteria such as battery state of charge (SOC) and slack time are used for scheduling EV charging. In this paper, the impact of different priority criteria on the chargeability of EVs and charging fairness is studied for both types of coordinated EV charging methods. The overall performance of the coordinated EV charging methods employing various combinations of different priority criteria is evaluated using data-driven simulations. In addition, this paper evaluates the need for using weighted priority criteria for scheduling EV charging along with appropriate weights required for different cases.

Prolific integration of renewable energy sources (RESs) such as solar photovoltaic systems into the distribution network will result in various issues associated with their intermittent nature. Energy storage is a vital component for overcoming issues associated with the intermittent nature of such RES. Though stationary battery systems are used as energy storage for such applications, smart energy storage (SES) systems are also becoming popular owing to various advantages and advent of smart grid systems. SES can be achieved by aggregating electric vehicles (EVs) or by using demand response management for loads with large time constants. Aggregated residential refrigerators are potential candidates for creating SES which has virtual storage capacity, unlike EVs. In this study, residential refrigerators are modelled analogously to energy capacity and self-discharge of electro-chemical batteries using the artificial neural network based kWh modelling. The model is further extended to estimate the virtual energy storage (VES) capacity with aggregated residential refrigerators; particularly in high-rise residential buildings. Simulation results are presented for scenarios covering the complete range of thermal capacity of typical refrigerators applicable in Singapore's climatic condition. Furthermore, a brief description of the possible applications for the estimated VES, pertaining to smart grid architecture and cyber-attack is also presented