1. Introduction
The energy and environmental crises have become increasingly serious with the advancement of human society: an escalating energy demand, the exhaustive nature of fossil fuels, and CO
2 emissions are among the major threats [
1]. There is a universal consensus that in order to achieve the goals of sustainable development and a low-carbon economy in the transportation field, transport electrification, exemplified by electric vehicles (EVs), must be implemented worldwide [
2]. Replacing fossil fuel vehicles with EVs can also provide health benefits to people by reducing air pollution [
3].
With the rapid development of battery technology, a recharge running range of greater than 100 km on a battery charged for less than 10 min in the fast charging mode is easily achievable [
4]. Additionally, high-power direct current (DC) fast charging has become available and popular. Compared with normal alternating current (AC) charging and battery-swapping charging, DC fast charging has a shorter recharging duration and lower investment for different types of spare battery packs or vehicles [
5,
6]. The pure electric bus (PEB) has characteristics that are distinct from those of private EVs: namely, the PEB has a larger onboard battery, operates in strict accordance with a route schedule, and is charged at the driver’s will at a fixed station with limited intervals. All of these factors lead to a greater distribution capacity demand and more severe charging load fluctuations [
7]. Moreover, the investment needed to expand the distribution capacity is huge, particularly in big cities; for example, expansion of the power grid capacity in Beijing is estimated to cost 1500 USD/kVA. With the city distribution network operating under heavy load, the plan to implement public transport electrification may face considerable challenges [
8].
To promote the collaborative development between EVs and the distribution network, some research has focused on the optimization of charging EV to best use the existing power sources or reduce the charging electricity cost. A model of the impacts of EV charging station load on the distribution network was established in [
9], and it used several influence factors: voltage stability, reliability, power loss, and economic loss by voltage deviation. The results of a detailed simulation with the IEEE 33 bus test system illustrated the influence of EV charging station load on the distribution network. The research was on the distribution network level, the suppression of distribution capacity demand of fast-charging station was not mentioned. The authors of [
10] proposed a charging schedule planning method that was based on an actual electric bus system, and they effectively reduced the electricity cost with three-level time-of-use (TOU) tariffs. The charging power was 50 kW for a single route of ten electric buses, but the influence of the distribution capacity was not considered. In another study, a daytime charging strategy with a charging price for private EVs was proposed, and the charging cost was practically reduced by achieving maximum utilization of photovoltaic power combined with distribution capacity constraints [
11]. The authors in [
12] presented a coordinated charging strategy for electric taxis in a temporal and spatial domain, and a Particle Swarm Optimization (PSO) algorithm was used to balance the charging load dispatch among different stations, as well as the charging times for the electric taxis. These studies have concentrated on private EVs or electric taxis with distribution capacity constraints, and their aim has been to improve charging costs. On the other hand, the distribution capacity demand of a PEB fast charging station, as well as the charging power which is as high as several hundred kilowatts, will continue to grow with the increasing number of PEBs and the accumulation of many necessary charging loads. Thus, more effective and direct strategies need to be developed to handle this issue.
Distributed energy resources (DERs) have been developed quickly and can be used effectively to meet the charging power demand in distribution networks and reduce EV owners’ basic electricity costs [
13]. The charging load of a PEB fast charging station mostly occurs in the daytime and can be as high as several hundred kilowatts; this is different from the private EV, whose load primarily occurs at night and is only several kilowatts [
14]. Photovoltaic power generation needs more space to generate more power, and wind power generation has anti-peak power and an allocation characteristic; therefore, neither of them is suitable for stations in cities. A gas turbine based combined cooling, heating and power system is economic and highly efficient, but it requires a reliable and sufficient gas supply [
15]. The battery energy storage system (BESS) is widely used for peak power shaving in many scenarios; for example, it has been used as a buffer to reduce charging load fluctuations and to shave peak power [
16]. For a fast charging station, the ability to suppress peak charging power primarily depends on the capacity of the BESS [
17], and the economics of the BESS application determine the deployment scale. Besides that, charging topologies affect the configuration strategy. The authors in [
18] proposed a DC bus concept for plug-in EVs and established a charging topology was established with a bipolar DC bus. The later was based on a central neutral-point clamped converter and provided a flexible connection to loads and electrical storage systems (ESSs) with higher voltage and power.
The authors of [
19]—a study on fast charging stations for plug-in hybrid EVs—applied electrical storage systems to reduce the station’s operational costs and alleviate the negative impacts of its operation on the power grid. A BESS configuration model that was based on a general charging topology was proposed in [
20], and it considered the investment costs and benefits of a BESS. As a result, the peak charging load of the PEB fast charging station significantly decreased. The authors in [
21] established a business model of a private EV charging station with a stationary Li-ion battery pack. The model, which considered the battery lifetime, power charges, and electricity tariffs, was used to alleviate the high costs of power charges and grid investment, but the detailed configuration method and results were not given. An integer nonlinear programming model that incorporated the investment cost, lifespan, and time-of-use electricity price was proposed to estimate the value of an energy storage system (ESS) for the electric bus fast charging station, and the effectiveness of two kinds of Li-ion battery was compared [
22]. However, only one kind of BESS configuration was researched, and energy loss was not included in the model. The authors in [
23] explored the technical and economic suitability of coupling a ground ESS to a DC fast charge unit for mitigating the demand charges and lessening the impact on the local electricity network. However, neither the numerical value of the ESS configuration nor the influence factors were taken into account.
The published studies on the ESSs of charging stations have been mostly concerned with the aspects of optimized control strategies or economic estimation. To develop a BESS configuration of a high power PEB fast charging station, the following points needs to be established:
Charging topology: the possible configuration allocations and the size of a single BESS must be determined.
Integration points of the BESS: the charge-discharge power and energy loss are dependent on this factor.
Related influence factors: the model will be more accurate if more aspects are considered.
This work focuses on BESS configuration methods for a newly high power PEB fast-charging station in Beijing, and aims to establish an economical method of BESS configuration to suppress the distribution capacity demand. The present and improved charging topologies of the station were presented based on actual applications. Considering the charging topologies and load characteristics analyzed, three practical BESS configuration strategies are proposed, i.e., integrating the BESS with the AC bus, each charging pile, and the DC bus. In order to compare the investment costs and benefits of the three methods with time-of-use tariffs, a novel multi-objective cost model was established according to Linear Programming theory. Compare with the existing literature, the model includes the cost factors such as basic electricity fees, electricity cost, cost of the energy storage system, costs of the transformer and converter equipment, and electric energy loss. The electric energy loss of the transformers, converters, and battery by power flow was newly modeled, and the accumulated charge-discharge number was adopted in the model to evaluate the battery lifetime. A case simulation was carried out according to a realistic operation load and practical parameters. The most economical BESS capacity was determined, and the case study result was analyzed and verified through simulations for each influence factor.
The case results show that integrating the BESS with the DC bus is the most economical method at the given charging load. Relative to the station cost without a BESS, the BESS/DC strategy yields a reduction of 160.92 USD per day during the lifetime of the battery (8 years), and the peak charging power is reduced by 67.03%. Although the cost of transformers and converters is higher, integrating the BESS with the AC bus is the most effective method to suppress peak charging power. This research can be applied widely to evaluate the economics for the high-power PEB fast charging stations to suppress distribution capacity demand using a BESS. Furthermore, the study may be used to facilitate the application of PEBs in areas with limited distribution network resources and thus alleviate environmental problems.
The rest of this paper is organized as follows: charging topologies and BESS configuration strategies are presented in
Section 2. In
Section 3, novel models of BESS configuration and a solution method are developed in detail. The case simulation and analysis of each influence factor are detailed in
Section 4. Then,
Section 5 presents the discussion. Finally, conclusions are put forth in
Section 6.