Modelling the aggregated dynamic response of electric vehicles

Abstract

There is an increasing interest in the use of electric vehicles (EVs) for providing fast frequency reserves due to their large installed capacity and their very fast response. Most works focus on scheduling and optimization and usually neglect their aggregated dynamic response, which is particularly important from the power system perspective when EVs offer significant shares of such services. We present a literature review on the aggregated modelling of EVs and derive analytical expressions for the representation of EV populations based on the probability distributions of their parameters. Such approximations can be used in power system studies, in order to capture the dynamics of an EV population more accurately. Finally, we compare our approach to the most widely used in the literature, i.e. the averaging method where all EVs are represented with the population's average values, and discuss the key differences of the two approaches. Index Terms-Aggregated dynamic response, electric vehicles aggregation, electric vehicles dynamics. C. Ziras, Y. Shi and H.W. Bindner would like to acknowledge the financial support of the EUDP project, Ecogrid 2.0, No 64015-0082.

Multi‐objective optimal provision of fast frequency response from EV clusters

“…This paper proposes an enhanced frequency-responsive aggregate model (EFRAM) to represent a cluster of EVs providing FFR. The proposed EFRAM leverages the existing research in aggregate modelling of EV clusters [14], [21] and combines it with a frequency-responsive control for the provision of FFR, as shown in Fig. 1.…”

“…A first-order transfer function is used to represent the EV charger dynamic with the time constant (TEV). The EFRAM uses the Laplace transformation of the cumulative distribution function (CDF) of the time delay as suggested in [14] to reduce the error that appears when the time delay of the EVs in the cluster is widely distributed. The EFRAM considers that the change in the output power from the EV cluster (ΔPEV), as a consequence of a frequency deviation, can be described by the following equation:…”

“…The parameters τa and τb make up the bounds of the delay distribution, and their value depends on the probability distribution function (PDF) of the EV cluster measurement time delay. Table I shows two useful approximations of the aggregation bounds for the case when the time delay follows a uniform (U) or a normal (N) distribution [14]. The effects from the number of V2G enabled EVs in the cluster, the rated power of the charging stations and the power base of the system for per unit calculations, are summarised in the EV cluster constant (kcluster), defined as follows:…”

“…The minimisation problem formulated in (13) and (14) includes two network security inequality constraints imposed by the TSO. These constraints are divided into two major groups.…”

Large-scale provision of frequency control via V2G: The Bornholm power system case

“…By modulating their battery charging/discharging process, EVs can perform vehicle-to-grid (V2G) services such as primary frequency control (PFC). The research emphasis in the field is put, among others, on EV fleet modelling for V2G services [1]- [3], combined smart charging and frequency regulation [4]- [6], and impact on the distribution level [7]- [9]. Studies on improvements in islanded power systems dynamics with high shares of renewable generations are also found in the literature [10]- [14].…”