**3. Charging Facility**

#### *3.1. Specifications and Configuration of the Charging Facility*

Fast-charging electric vehicles require a sufficiently powerful connection to the electrical grid, which may require connecting directly to the high-voltage transmission grid. It is quite expensive to connect it to the high-voltage mains because of the switchgear and the space required to build a substation. There are service areas distributed every 30 km on average in the Italian motorway system and, according to our survey, each service area has 13 fuel pumps on average in each direction. The survey data come from Google Earth and Google Street View imaging service; we counted the number of refuelling bays in some service stations on A4 Milan-Turin, A1 Milan-Bologna and A1 Rome-Naples motorway and then we calculated an average. If the refuelling process lasts 5 min and all the available fuel pumps are being used, it is possible to refuel 78 cars in half an hour, which means that 78 charging bays are needed to recharge the same number of vehicles in the same half-hour. This means that each direction needs 7.8 MW of power to recharge each vehicle with a 100 kW rating. The Italian high-speed rail network has been built near motorways (when possible) and is able to deliver high power at a relatively low voltage, so it makes sense to study the effects of such a solution on the 2 × 25 kV railway supply system to evaluate the possibility of connecting the motorway charging points to the nearby railway. This solution can be particularly advantageous in the countryside, where the high-voltage network node is relatively far away and the high-speed line is quite near the service station because it avoids the construction of high-voltage lines. It is important to note that the railway infrastructure owner has the right to disconnect the motorway car charging facility in case that power is needed to maintain a set quality of service on the railway line itself.

The hypothesis of simultaneous use of all the refuelling bays is not quite true, as usually only some are actually used simultaneously in real life. This can be expressed through a coefficient that indicates the percentage of bays used. This is actually very useful, because it allows for three options: reducing the number of available charging bays; or sharing the total 100 kW power rating between two bays; or both. If the power-sharing option is chosen, then it would be better to share the power dynamically between the two cars in order to give more power to the car with the lower state of charge rather than sharing the power fifty-fifty between the two users.

#### *3.2. Simulation Model of the Load*

Each charging station has a variable load with 100 kW maximum power. The load model has to potentially include each main rectifier topology for harmonics injection in the network and reactive power consumption.

The main topologies are Graetz bridges with both diodes and thyristors and the PWM (Pulse-width modulation) controlled switched AC/DC converter; these are very different in both harmonics injected and reactive power consumption. The Graetz bridge configuration is characterised by high harmonic currents at low frequency, i.e., bridge rectifiers inject high third, fifth and seventh harmonic order currents that normally have to be filtered out. On the other hand, they are easy to maintain because each valve is turned on and naturally, but the output voltage cannot be regulated without an additional switching DC/DC converter. Thyristors bridges are more complicated indeed requiring a controller, but they allow regulating the output voltage. Voltage regulation is possible by changing the ring angle, which causes the output voltage to drop; on the other hand, the ring angle is directly linked to the reactive power absorbed, which means that an appropriate power factor correction device is needed.

Switched AC/DC power converters constitute the most modern approach to rectification. Their main feature is that they achieve unity power factor and limit low order-harmonic-current injection through their high switching frequency, which, depending on the type of semiconductor used (MOSFETs, IGBTs, GTOs), varies from some kilohertz to over 20 kHz. Being directly linked to the switching frequency, the frequency of the current harmonics is high enough to make it possible for the current amplitude to be naturally dumped by cables and transformers inductances resulting in low THD without the need of expensive filters. What is more, if converters are grouped and controlled with a technique called interlacing, the net result is that each group of harmonic emissions is equivalent to that of a single converter operating at a frequency multiple of the converters number in each group and each converter's switching frequency.

Most simulations feature a switching AC/DC converter. The load itself is made of the battery, the converters and the interfacing transformer. The battery has been modelled using the battery block available in SimPower systems within MATLAB-Simulink. It is able to simulate every kind of battery form, from lead-acid to lithium-ion ones. The battery modelled in this case is the one available in the recent electric cars belonging to the category of Battery Electric vehicles (BEV) [40,41]. The battery's main features are summarised in Table 3.


Rated pack capacity (Ah) 66.2

**Table 3.** Main characteristics of the electric vehicle battery.

The four-quadrant (4Q) power converter used to rectify the AC voltage uses IGBTs as semiconductors and is controlled through a dedicated PWM controller. The main modifications needed to make it run consisted in disabling the maximum power point tracking (MPPT) system used to extract the maximum power available from the PV panel and adjusting the regulator parameters. This second step has been performed through a trial-and-error method until a functioning device was obtained. A further aspect that has been modified is the value of the DC bus capacitor, which has been changed to 40 mF in order to maintain the voltage ripple in a 5% band. The DC bus nominal voltage has been changed to 500 V from the previous value of 425 V. The new nominal value has been chosen because, actually, it is the maximum voltage that has CHAdeMO and CCS (Combined Charging System).

The converter and its control system are shown in Figure 5. It is necessary to install a DC/DC converter to adjust the current flow to the battery because of the difference between the battery voltage (about 400 V when fully charged) and the main DC bus (500 V). This converter is a two-quadrant converter that allows the power to flow in both directions, i.e., from the battery to the grid or vice-versa.

In this case, the battery charging function is more interesting. The control is done through a PI controller in order to obtain a voltage reference that is able to drive the PWM signal generator. The controller is again tuned through a trial-and-error process. The current reference signal has been made variable so that the regulator is enabled with a current reference equal to zero. The reference

signal increases linearly to the maximum value of 250 A in 2 s. Battery, converter and filters are shown in Figure 6.

**Figure 5.** Model of the four-quadrant (4Q) converter and its control system.

**Figure 6.** Model of the car battery and the converter.

The connection between the low-voltage systems that supply the power electronics and the medium voltage from the railway line is achieved through a short cable and a power transformer rated at 150 kVA. We decided to maintain the original topology of a centre-tapped two windings transformer used for the 120 V distribution system in the US for a couple of reasons; the most important of those is safety. As both lines are live, but their potential to earth is only 120 V, it is safer for people in case of an insulation failure. We hypothesised a cable of about 200 m length with a cross section of 180 mm<sup>2</sup> and a resistance of 0.106 km<sup>−</sup>1. The transformer and line model is depicted in Figure 7. The resistor *Rg* represents the Earth system resistance.

**Figure 7.** Model of the power transformer and the short low voltage line.

The 2 × 25 kV system is modelled as shown in Section 2.2. Four cases have been examined. These are: (a) the absence of trains; (b) the presence of one train; (c) the presence of two trains in the same cells; or (d) the presence of two trains in different cells. This is particularly important, as the main load continues to be the train traffic and the system is viable only if it is possible to power both the railway traffic and the charging facility at the same time. One of the possible scenarios simulated is the presence of two trains in the same cell to evaluate the voltage drop in the system. The considered trains are two modern High Speed Trains with power consumption of 8.8 MW and 9.8 MW, respectively. It is assumed that the two trains absorb a sinusoidal current with a unity power factor. Figure 8 shows the absorbed current by both trains without charging stations connected to the line. It can be seen that the system is capable of supplying both trains with no problem and maintains keeping the line voltage close to its nominal value.

**Figure 8.** Voltage and current drawn by a High Speed Train (HST) on a high-speed line with no charging stations.

In order to verify that the actual current distribution is comparable to the ideal one, the current at the main transformer and at the autotransformer terminals have been measured (Figure 9). As can be seen, the actual current distribution is close to the ideal one even if some differences due to the real impedances of the line can be noted. In particular, it is possible to observe a small current in the rails and an imbalance between feeder and contact line (catenary).

**Figure 9.** Currents measured on the (**a**) feeding transformer connections and (**b**) autotransformer connections with trains as the only load.

#### **4. Simulation of the Charging System**

The aim of these simulations is to assess how the quality of power of the railway systems is affected by the charging stations for road vehicles. Therefore, the analysis considers simulations of a very detailed model for a short period (max 3.5 s).

There are two simulation steps. The first step is about making the charging system work by itself. It is very important because it allows us to sort out the system problems before the two systems are made to work together. When both systems work as expected, it is then possible to connect them together and have the complete simulation of all the effects.

The first simulation phase has given the following results: the model was working with the following regulator values, 3000 and 1000 as proportional and integrator coefficients for the voltage regulator and 500 and 1000 as proportional and integrator coefficients for the current regulator. The current absorbed was not constant when the device was on full load as the DC bus voltage was still floating. The battery status is shown in Figure 10. It can be seen that the battery is actually recharged from the initial state of charge set to 10%. It shows current and voltage of the battery; current is represented with a negative value because it is charging. On the other end, the battery current would be positive if the battery was discharging. The battery is charged with a constant current using a soft inrush defended by ramp. As the current is constant, the state of charge increases linearly, instead the internal voltage drop is quite proportional to the current in a short time windows; it seems in this period, it is mainly due to the homic voltage drop. Moreover, it is possible to observe the effect of the switching of the power converters in contactors that produce a high-frequency ripple overlap to the main current.

**Figure 10.** Battery main variables: state of charge, current and voltage.

The system is then tested inside the 2 × 25 kV railway line connected between the feeder and the grounding conductor. In this case, the model does not consider only one 100 kW charging spot, but twenty 100 kW charging bays for each direction that cause total power absorption of 4 MW from the railway systems. It is possible to replace the single bay with double bays sharing the same amount of power, thus allowing more people to charge simultaneously even if at a lower rate. The charging area is modelled with one charging bay and a current generator. The idea is to model one charging bay in order to see the current shape it absorbs and, at the same time, inject in the railway system the same current multiplied for the remaining charging bays. The assumption is that all charging bays are working at the same rate. The charging stations are connected from feeder to earth, rather than from catenary to earth, as trains are (Figure 11).

**Figure 11.** Section of a real High Voltage (HV) line with indication of the feeder and earth wires.

It has been decided to connect the charging facility in the middle of the cell. The first scenario is simulated with no train on the line and the results show that the system works quite well. The simulated time lapse is 3.5 s (Figure 12). In particular, the results are shown towards the end of the simulation because the charger is working at full load there. Figure 9 shows also the impact of all the charging infrastructures installed in two charging areas on the railway voltage. It is possible to remark

that the voltage value near the charging area is still inside the limits set for the system on both tracks, despite the fact that the current is not always sinusoidal.

**Figure 12.** Feeder and catenary voltage due to the charging station presence.

Despite the shape of the current, Figure 13 shows that the autotransformers work as expected by shifting the earth current to the feeder and catenary system. This means that the feeding transformers see the most current coming from the feeder and the catenary with only a small current in the rail as Figure 14 shows. Actually, the two charging areas behave as a low-power HST, and it seems the power requested by the charging infrastructure is about one half of the power requested by a HST during acceleration. In fact, allowed connection between feeder and ground has the same behaviour of the train supplied from contact line and the rails.

The fact that the system works with no trains does not imply that it works while performing its main duty, i.e., powering high-speed trains. In order to verify this, a simulation has been produced in which the system has one or more trains on the tracks. The simulation represents the worst-case scenario with all the loads fed from the two autotransformers.

The first simulation involves the presence of one train on the line. The modelled train is an ETR 400 high-speed train with a maximum power of 9.8 MW located 3 km away from the charging stations. It has been computed that in 3.5 s the train travels less than 500 m, so it has been modelled to assume that the train is still. The total power of the three loads is about 14 MW, so it is expected that the system is able to power all the loads without an excessive voltage drop. The simulation results demonstrate exactly that the three loads can coexist. Figures 15 and 16 show that the voltage measured near the loads and the train is within the limits set by international standards with voltage value being around 23 kV.

**Figure 13.** Current measured on the autotransformers connections due to the charging stations.

**Figure 14.** Current measured on the feeding transformer with the charging stations and no trains on the track.

**Figure 15.** Catenary and feeder voltages on the charging facility connection point.

**Figure 16.** Voltage on the train position and current drawn by the train.

There are some differences between this and the preceding scenario; in particular, the currents have a different shape because of the sinusoidal train current superposed to the charger current. Figures 17 and 18 show the currents flowing in the autotransformers and in the main transformer.

The connection of the charging infrastructure between the feeder and the ground attempts to improve the balancing of the system as the current flows directly between the contact line and the feeder without involving the autotransformers and the sells not occupied by the train.

**Figure 17.** Current drawn by the balancing autotransformer with the charging stations and a train on the tracks.

**Figure 18.** Current measured on the feeding transformer low voltage winding with the charging stations and a train on the tracks.

The last scenario evaluates the ability of the system to power two HST at 9.8 MW, each travelling in opposite directions and the same 40 charging bays. This case is very demanding and the system struggles to power the load. This is illustrated by the voltage value, which dropped to about 19 kV as shown in Figures 19 and 20.

The main similarity with the preceding scenario is that the currents in the feeding transformer and in the autotransformer are significantly influenced by the train rather than the relatively small load given by the charging bays. Figures 21 and 22 depict the current drawn by the autotransformers and the feeding transformer, respectively. The cause of this problem can be traced to two sources: the power converter of battery charging facility or the connection of such a big load between feeder and catenary. It is possible to speculate about the cause of this kind of problem being the connection of

such a big load between feeder and earth. The system can cope with the load due to the line technical systems, such as signalling, switches and line diagnostics, but these loads have a maximum rating of 250 kVA each and each substation powers only a few of them.

**Figure 19.** Voltage on the train position and current drawn by the train with two trains on the track.

**Figure 20.** Voltage at the train pantograph and current drawn.

**Figure 21.** Current drawn by the autotransformer in the two-train scenario.

**Figure 22.** Current drawn by the feeding transformer in the two-train scenario.

#### **5. Conclusions**

The need to limit pollution enforces stricter emission standards that will increase the cost of producing traditional cars; moreover, increases in oil prices will result in fuel that is more expensive. In the meantime, a reduction in battery costs and government subsidies will lead to an increase in electric vehicle purchases. Another incentive to use electric vehicles comes from both battery manufacturers and the car industry: as battery technology develops further, it will be important to decrease the charging time for the battery pack while maintaining its performance for its entire life expectancy.

As electric vehicles are going to become even more popular, it is necessary to build fast charging infrastructures, especially on the highway network. One of the problems encountered is finding a suitable power source to charge many cars quickly despite the increase in battery capacity and in the number of vehicles in stock. Since the high-voltage grid is not always easily accessible, as the highway lines are usually far from the urban area and from the electricity grid but they are usually close to new high-speed railway lines, the authors have investigated the possibility to supply charging areas from the electric distribution system used for the railway service. These are promising solutions because the power absorbed by the charging areas is of the same order of the magnitude of the power absorbed by the train.

This solution has to compromise the needs of the rail operators, who want their trains adequately powered, and the service areas, who want to offer a service to car drivers.

To demonstrate the feasibility of the concept, a model of a 2 × 25 kV system to feed the railway has been developed. This latter has been implemented in MATLAB/Simulink/SimPower systems to simulate the railway. Then it has been applied to simulate the battery charger and the system.

The results through modelling and simulation disclosed that a compromise is possible but the charging station power has to be limited to allow trains to be properly powered. The simulations reveal that particular attention has to be paid to the quality of the power of the railway system in order to not be compromised by the high-power charging infrastructures. This means that it is not possible to achieve the same refuelling rate of traditional cars because the charging time is much greater than a traditional car refuelling time and the high-speed rail is not able to provide the extra power a larger charging facility needs. Nevertheless, it can be a good solution to begin building the fast charging infrastructure where the high-speed rail is easily accessible.

**Author Contributions:** Morris Brenna and Michela Longo proposed the core idea, developed the models. They performed the simulations, exported the results and analysed the data. Wahiba Yaïci revised the paper. Morris Brenna, Michela Longo and Wahiba Yaici contributed to the design of the models and the writing of this manuscript.

**Conflicts of Interest:** The authors declare no conflict of interest.
