1. Introduction
Electric Vehicles (EVs) are, slowly but undoubtedly, becoming a real alternative to traditional combustion engine cars. In recent years, several European countries have set national electric car deployment targets as a key measure to diminish pollutant and greenhouse gas emissions in urban areas [
1]. Financial incentives, together with an increased availability of charging infrastructure and certain benefits when accessing and parking in some urban areas in many cities, are decisive factors for the expected growth of electric vehicle market shares in the coming years [
2,
3].
Due to these predicted changes in transportation markets, the increase of EVs penetration in Low-Voltage (LV) distribution networks is likely to become a challenge for Distribution System Operators (DSOs), since it can lead to power quality issues. A typical topology for European LV distribution networks is the three-phase four-wire network, where loads and EVs are not equally distributed across the three phases. The combined effect of both EVs and household loads highly contributes to the unbalanced operation of the network, which results in voltage quality concerns such as under-voltage conditions and voltage unbalance [
4,
5,
6].
EVs’ smart charging impacts on LV networks’ power quality has been studied from a wide range of perspectives, including the capability of EVs to mitigate the aforementioned voltage quality concerns. Among the specialized literature, smart charging controls can be divided into centralized and decentralized architectures, whose advantages and disadvantages have been discussed in [
7,
8,
9,
10]. Centralized smart charging requires a central aggregator that controls the charging of all vehicles and optimizes their charging load profiles based on a specific control algorithm. It requires a huge interconnected communication network, so charging profile information can be exchanged between each EV and the aggregator. The centralized architecture allows smoothing the aggregated electric load profile and improves voltage and power quality in a region, but may cause individual vehicles to have voltage peaks in their charging profiles.
On the other hand, the decentralized architecture manages the charging process of the EVs locally. Decentralized controls present several advantages such as scalability, constant computational effort, and reduced communication requirements [
11]. Besides, it allows each individual EV to minimize their respective charging costs, but, collectively, this may not be the optimal solution for the region, since there could be instabilities when all controllers react simultaneously to the measurements. Considering that cost and robustness are similar in both architectures, the decentralized control tends to be a more practical solution, since it is based on local measurements and does not need additional communication infrastructure [
10,
12].
In addition to the classification between centralized and decentralized architectures, EV smart charging controls can be divided into those controls suitable for being implemented in a single-phase charger and those applicable to three-phase chargers. Historically, in European countries, the three-phase connection has been reserved for industrial consumers, but, in recent years, the number of residential houses with three-phase connection has increased considerably in Northern and Central Europe [
12]. Regarding single-phase connection, EVs’ smart charging impact on LV networks has been investigated mainly by modulating the active power consumed by the EVs through a droop control [
13,
14,
15]. This control highly influences the time needed for a full charge and could potentially result in inconveniences for the user. Besides, the power electronic converter in the charger might introduce a reactive power capability, which can contribute to ancillary services and the correction of voltage issues, such as [
16,
17]. Therefore, taking advantage of the available hardware, the reactive power provision has been studied in [
18,
19] for a balanced system, and the work in [
12] proposed a new control for an unbalanced system based on a droop control.
In relation to three-phase connection, the work in [
20] investigated the capability of providing a negative and zero sequence in order to diminish voltage unbalance. Moreover, research presented in [
21,
22] demonstrated that it is possible to achieve substantial improvement in voltage unbalance by balancing the LV network loads. In this case, the strategy also requires a reactive power capability, which is always limited by the charger’s rated power. Thus, the more reactive power the charger is dealing with, the less reactive power is available.
This paper analyses the most relevant Electric Vehicle (EV) charging strategies to mitigate voltage unbalance and under-voltage conditions in Low-Voltage (LV) distribution networks. Besides, a comparison is made between the results obtained with three-phase charging strategies, single-phase charging strategies, and a combination of both.
This paper is organized as follows:
Section 2 presents the test LV network, together with the household consumptions and EV charging power demand.
Section 3 describes the voltage quality indicators and standards and introduces the EV smart charging controls selected for the analysis. Finally, the scenarios conducted and results are discussed in
Section 4, while the conclusion is given in
Section 5.
5. Conclusions
EVs’ increased penetration will have a great impact on Low-Voltage (LV) distribution networks, since it leads to power and voltage quality concerns. In order to mitigate the undesirable effects originated by uncontrolled EV charging, several decentralized smart charging controls have been proposed in recent years, both for three-phase and single-phase EV chargers. Trying to clarify which of the different solutions acts in a better way regarding a specific objective, this paper analyses and compares four of the most relevant smart charging controls, with the aim of reducing the VUF and avoiding under-voltage conditions. The selected controls are active power droop control (Droop P) and reactive power droop control (Droop Q) for single-phase EV chargers and Load Balancing control (LB) and Sequence Compensation control (SC) for three-phase EV chargers. Comparison and analysis were done between the two single-phase controls, between the two three-phase controls, and between a combination of Droop P-LB, Droop P-SC, Droop Q-LB, and Droop Q-SC controls. The selected controls were tested on the standardized European Low Voltage Distribution Network developed by CIGRE for the worst case scenario, i.e., 100% EV penetration during a highly-loaded and unbalanced time period.
Single-phase controls provide phase voltage regulation, since they are not able to mitigate VUF due to the lack of information about the three phases. On the other hand, three-phase controls aim to mitigate VUF, but they do not provide voltage level regulation. From the results, it was concluded that Droop P control provides a huge voltage level rise and reduces the voltage variability, although it only acts on nodes with low voltages. Furthermore, as a side effect, Droop P control balances the phase voltages and is able to reduce VUF. On the other hand, Droop Q control’s influence on voltage levels is not as noticeable as Droop P control, but it affects every node.
Regarding three-phase controls, both LB and SC controls are able to reduce VUF when every EV charger is equipped with the same charging control. Nevertheless, SC control loses effectiveness when combined with single-phase charging controls, while LB control is able to contribute to VUF mitigation in that situation.
In conclusion, the implementation of a decentralized EV charging control is an adequate solution for DSOs since it improves the reliability and security of the network. Moreover, Droop P and LB controls have been proven to be the most suitable solution for DSOs when dealing with under-voltages, due to the high improvement in voltage level at critical nodes provided by Droop P control and the better association with and contribution to VUF mitigation provided by LB control with respect to SC control. If the network experience over-voltages, Droop P control is unable to regulate voltage levels in order to be compliant with the standards. However, Droop Q control is able to regulate both over- and under-voltages due to its versatility. Hence, the control strategies selected by the DSO should take into consideration the aforementioned issues.
References