Electric Vehicle (EV)-Assisted Demand-Side Management in Smart Grid

Abstract

While relieving the dependency on diminishing fossil fuels, Electric Vehicles (EVs) provide a promising opportunity to realise an eco-friendly and cost-effective means of transportation. However, the enormous electricity demand imposed by the wide-scale deployment of EVs can put power infrastructure under critical strain, potentially impacting the efficiency, resilience, and safety of the electric power supply. Interestingly, EVs are deferrable loads with flexible charging requirements, making them an ideal prospect for the optimisation of consumer demand for energy, referred to as demand-side management. Furthermore, with the recent introduction of Vehicle-to-Grid (V2G) technology, EVs are now able to act as residential battery systems, enabling EV customers to store energy and use them as backup power for homes or deliver back to the grid when required.

Hence, this thesis studies Electric Vehicle (EV)-assisted demand-side management strategies to manage peak electricity demand, with the long-term objective of transforming to a fully EV-based transportation system without requiring major upgrades in existing grid infrastructure. Specifically, we look at ways to optimise residential EV charging and discharging for smart grid, while addressing numerous requirements from EV customer's perspective and power system's perspective.

First, we develop an EV charge scheduling algorithm with the objective of tracking an arbitrary power profile. The design of the algorithm is inspired by water-filling theory in communication systems design, and the algorithm is applied to schedule EV charging following a day-ahead renewable power generation profile. Then we extend that algorithm by incorporating V2G operation to shape the load curve in residential communities via valley-filling and peak-shaving. In the proposed EV charge-discharge algorithm, EVs are distributedly coordinated by implementing a non-cooperative game. Our numerical simulation results demonstrate that the proposed algorithm is effective in flattening the load curve while satisfying all heterogeneous charge requirements across EVs.

Next, we propose an algorithm for network-aware EV charging and discharging, with an emphasis on both EV customer economics and distribution network aspects. The core of the algorithm is a Quadratic Program (QP) that is formulated to minimise the operational costs accrued to EV customers while maintaining distribution feeder nodal voltage magnitudes within prescribed thresholds. By means of a receding horizon control approach, the algorithm implements the respective QP-based EV charge-discharge control sequences in near-real-time. Our simulation results demonstrate that the proposed algorithm offers significant reductions in operational costs associated with EV charging and discharging, while also mitigating under-voltage and over-voltage conditions arising from peak power flows and reverse power flows in the distribution network. Moreover, the proposed algorithm is shown to be robust to non-deterministic EV arrivals and departures.

While the previous algorithm ensures a stable voltage profile across the entire distribution feeder, it is limited to balanced power distribution networks. Therefore, we next extend that algorithm to facilitate EV charging and discharging in unbalanced distribution networks. The proposed algorithm also supports distributed EV charging and discharging coordination, where EVs determine their charge-discharge profiles in parallel, using an Alternating Direction Method of Multipliers (ADMM)-based approach driven by peer-to-peer EV communication. Our simulation results confirm that the proposed distributed algorithm is computationally efficient when compared to its centralised counterpart. Moreover, the proposed algorithm is shown to be successful in terms of correcting any voltage violations stemming from non-EV load, as well as, satisfying all EV charge requirements without causing any voltage violations.