Coordinated Deliverable Energy Flexibility from EV Aggregators in Distribution Networks

TL;DR

The paper addresses coordinating EV aggregators with a distribution system operator to maximize deliverable energy flexibility from EV charging under load and PV uncertainty. It introduces a two-level framework: the DSO solves a grid model to determine $P^{\sf gf}_{n,t}$ and $Q^{\sf gf}_{n,t}$ via a hybrid stochastic-robust transformation, and each aggregator then optimizes its fleet’s charging to minimize costs given that flexibility. A key finding is that allowing EVs to inject reactive power ($Q^{\sf EV}$) increases allowable flexible loads and improves voltage support, while incorporating uncertainty reduces the feasible EV flexibility; numerical tests on a modified IEEE 33-bus network illustrate these effects. The approach balances robustness and computational tractability and has practical implications for integrating EVs and distributed renewables into distribution networks.

Abstract

This paper presents a coordinated framework to optimize electric vehicle (EV) charging considering grid constraints and system uncertainties. The proposed framework consists of two optimization models. In particular, the distribution system operator (DSO) solves the first model to optimize the amount of deliverable energy flexibility that can be obtained from EV aggregators. To address the uncertainties of loads and solar energy generation, a hybrid robust/stochastic approach is employed, enabling the transformation of uncertainty-related constraints into a set of equivalent deterministic constraints. Once the DSO has computed the optimal energy flexibility, each aggregator utilizes the second optimization model to optimize the charging schedule for its respective fleet of EVs. Numerical simulations are performed on a modified IEEE 33-bus distribution network to illustrate the efficiency of the proposed framework.

Figures (7)

• Figure 1: System Model
• Figure 2: Allowable flexible active power at nodes 25 and 33 without uncertainty at unity and non-unity power factors
• Figure 3: Consumed active power by EVs at nodes 25 and 33 with and without Q without uncertainty
• Figure 4: Allowable flexible active power at nodes 25 and 33 with uncertainty at unity and non-unity power factor
• Figure 5: Voltage at node 17 with unity and non-unity power factor considering uncertainty