Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.785400
Title: Cost-optimal charging of electric vehicles using real-time pricing
Author: Mody, Sagar M.
ISNI:       0000 0004 7970 9193
Awarding Body: Loughborough University
Current Institution: Loughborough University
Date of Award: 2018
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
The large-scale adoption of EVs presents both potential benefits and difficult challenges. The already stressed electricity grids will have to manage the influx of EV charging requirements, which is especially difficult at peak times. This calls for smart solutions to optimally charge EVs in a grid-friendly way, using demand response where possible. In line with the demand, the electricity prices at peak times can be very high and it would also be advantageous for the user to avoid charging at these times. Therefore, the goal of grid friendly charging is twofold: to avoid putting additional load on the electricity grid when it is heavily loaded already, and to reduce the cost of charging to the consumer. Along with the technological progress in the EV field, the electricity grid is evolving toward a smart-grid. One of the changes a smart-grid will bring is smart-metering. In such a system, Day Ahead tariff (DA) prices are announced in advance for the next day. However, the balance of supply and demand is notfully known in advance and therefore, the Real-Time Prices (RTP) are more reflecting of the actual grid situation, but unknown in advance. This thesis presents control strategies for Cost Optimal Charging of Electric Vehicles, from the point of an EV user connected to a real time pricing tariff system. Firstly, since there are differences in the DAP and RTP, the thesis proposes a predictor to create an unbiased estimate of the RTP tariff based on the available factors in the pricing data. It uses a linear regression on historical data to find the best prediction of the expected price. The results find that the predictor achieves a slight reduction in prediction uncertainty with the used data set and has a negligible effect on overall cost. It means that the DAP can be used as a fair prediction of RTP. The first charging strategy proposed, uses the available DAP (price-prediction) for optimisation and follows a deterministic approach, to achieve the lowest charging cost. It achieves a sub optimal solution in which the controller successfully picks the times of lowest electricity cost from the prediction and provides a full charge to the vehicle by the time the user requires it. Since the electricity prices are affected by random disturbances and therefore the RTP can be different, it makes the charging process less predictable and introduces a stochastic element into the problem. A second optimal controller is presented which takes this problem into account by following a stochastic optimisation approach, specifically based on a stochastic dynamic program (SDPM). It uses a stochastic optimisation algorithm to minimise total cost of charging over a given time-period, whilst still providing required state of charge (SoC) in the EV battery. The controller does this by predicting future prices changes from available data, based on a probability. It takes into account price variability via a simple grid model that allows for unexpected price rises and a gradual return to a normal grid price. Finally, a case study is presented based on the price data available from the Illinois Electricity Grid (USA), to validate the optimal controllers over a year. The Stochastic Dynamic Programming optimal controller, can save up to (US) $112.88 over a year, compared to charging directly. This is very close to the theoretical optimum (full knowledge of real prices in advance) of $119.76. The controller uses the DAP and RTP prices effectively in simulation and optimisation stages, to avoid times of high price and price spikes. The result is, lower charging cost over the year which is achieved by shifting the charging over to the off-peak hours. Both strategies demonstrate significant advantages against conventional charging. The simple optimisation can realise most of the benefits and may therefore be the preferred strategy in practical terms, while the stochastic optimisation does offer slight further benefits at more significant complexity. This may change as the smart grid matures, the billing periods become shorter, and the processing capability of chargers increases.
Supervisor: Not available Sponsor: Loughborough University ; EPSRC
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.785400  DOI: Not available
Keywords: Engineering not elsewhere classified ; Electric vehicles ; Charging ; Charging control ; Optimal charging ; Cost saving ; Stochastic optimisation ; Deterministic optimisation ; Dynamic programming ; Stochastic dynamic programming ; Linear programming ; EV ; EV charging control ; Automatic EV charging ; Demand response ; Real-time pricing
Share: