Infrastructure Issues regarding the Ultrafast Charging of Electric Vehicles

This paper is devoted to the infrastructure issues arising from the ultrafast charging of electric vehicles in the timeframe of ~5 minutes. To mitigate variances and peaks caused by high energy transfer rate and pulse-like load, an ultrafast charging station must be partially decoupled from the utility grid by the usage of intermediate energy buffers. The perspective charging station load and its management by intermediate buffering are analysed and discussed with a tentative design proposal. Introduction The state-of-the-art electric vehicle (EV) charging is limited to the rated current and voltage of conventional household sockets. For continental Europe, where 230 V/400 V phase-to-neutral and phase-to-phase voltages are used with 16 A sockets as standard, recharging an average EV battery takes at least 6 hours from one-phase connection and 2.5 hours if a three-phase connection is available. There exists already a quick-charging method promoted by the CHAdeMO consortium, allowing recharging an EV battery to 80 % of its rated capacity within 20...30 minutes and based on the IEC 61851-23 standard On a highway, this yields driving/charging time ratio in the range of 3:1, which is far away from making EV a serious alternative for long distance driving. Transferring energy to an electric vehicle traction battery in as short timeframe as possible requires high power, determined not only by the battery’s capacity and charging time, but also by the inherent losses due to the electrochemistry. From the grid operator’s viewpoint such peaks are undesirable, because they necessitate overdimensioning of cables, power transformers, protection devices etc. The situation becomes even more aggravated if multiple vehicles are charging simultaneously, which brings along the need for a medium voltage connection. A possibility to alleviate the grid impact of the ultrafast charging lies in decoupling load from the grid. This can be done with the implementation of energy storage elements, which act as a buffer between the grid and the charging terminal. A similar approach has been recently implemented in the fast charge of compressed air propelled vehicles. Several energy storage media are further evaluated in terms of performance and costs and an optimal solution proposed. Finally, a buffered ultrafast EV charging station structure is proposed. Such a station is composed of several modules, comprising in connection ports for the utility grid, EV, storage medium and a common power bus. The modular architecture ensures extensibility if the station’s utilisation grows, i.e. the EV market share increases. To point out the necessity of deepened research in the selected field, available charging methods should be described based on the state-of-the-art and market analysis. The IEC 60851 standard, applicable for conductive charging systems, defines four charging modes, as given in Table 1. Table 1. Charging modes according to IEC 61851-1, 230 V / 400 V voltage system Mode Max current per phase Max charging power per phase Charger location 1 16 A 3.6 kW On-board 2 32 A 7.3 kW 3 63 A 14.7 kW 4 dc 400 A dc 150 kW Off-board However, a discrepancy between the above mentioned standard and market situation exists. Usually the EV manufacturers prefer to sell their products in set with a Mode 1 one-phase on-board charger; as standard household 16 A sockets are used, de-rated to 10 A ... 12 A at constant load, the charging power is even more limited, to 2.3 kW ... 2.8 kW. Thus a small-sized EV with a 16 kW·h traction battery would need at least 6 hours to fully recharge. The Mode 4 utilising off-board chargers has been implemented by the CHAdeMO consortium. As for today, the charging current is limited to 120 A by the used connector, which enables to recharge a commercial EV within 20 min ... 30 min depending on the battery capacity. As for Mode 3, the manufacturers have not reached an agreement on the standard connector, the increased charging rate is achieved by reversing the power flow in the traction inverter and using motor windings as smoothing reactors, thus the charging power can be nearly equal to the rated driveline power. A comparison between commercially available charging methods is given in Table 2. Table 2. Commercially available EV charging methods Charging type Mode Min charging time Autonomy flowrate Domestic one-phase charging 1 6 h ... 8 h 0.3 km / min Three-phase semi-quick charging 3 20 min ... 30 min 4 km / min CHAdeMO semi-quick charging 4 Diesel tanking for a family car N/A 1 min 30 s 600 km / min To make an EV attractive for distances beyond single charging autonomy, an optimal relationship between the battery capacity and autonomy flowrate is the key objective: to increase the average speed, there must be less charging stops and shorter charging times. The transfer of the same amount of energy, in turn, means higher charging power with increased requirements both for the EV battery system and the utility grid connection. Materials and Methods To estimate the load curve of a perspective ultrafast EV charging station, its frequentation must be determined at first. As there are no data on real stations available, analogies may be drawn with conventional fuel stations or other relevant existing statistics applied. It may be assumed, that the frequentation of an ultrafast EV charging station is distributed in time similarly to traffic density, the latter statistics is provided in Switzerland by the Federal Office of Statistics. In Fig. 1, the typical hourly traffic densities are given for highway (counting point near Yverdon) and urban conditions (Chauderon in Lausanne). For charging station load profile generation, averaged urban-extra urban distribution is selected. It should be mentioned, that the frequentation distribution is close to the overall energy consumption distribution with daytime maximums and nighttime minimums. Fig. 1. Typical daily traffic density distributions The actual load curve, imposed to the utility grid by the ultrafast EV charging depends on the following parameters: 1) objective charging time: taken equal to 5 minutes; 2) EV battery capacity (Ebat): ranging from 16 kW·h for small vehicles to 55 kW·h for sport cars; 3) initial battery state-of-charge (SoCi), ranging from 0 % to 50 %; 4) EV arrival times to the station, subjected to hourly distribution and daily frequentation. In following calculations, the Ebat values are subjected to left-truncated normal distribution (Fig. 2) based on market analysis and SoCi to normal distribution (Fig. 3). 0% 20% 40% 60% 80% 100% 0% 1% 2% 3% 4% 5% 6% 7% 8% 9% 01 .0 0 h 02 .0 0 h 03 .0 0 h 04 .0 0 h 05 .0 0 h 06 .0 0 h 07 .0 0 h 08 .0 0 h 09 .0 0 h 10 .0 0 h 11 .0 0 h 12 .0 0 h 13 .0 0 h 14 .0 0 h 15 .0 0 h 16 .0 0 h 17 .0 0 h 18 .0 0 h 19 .0 0 h 20 .0 0 h 21 .0 0 h 22 .0 0 h 23 .0 0 h 00 .0 0 h Cu m ul at iv e pe rc en ta ge Ho ur ly p er ce nt ag e Yverdon Chauderon Averaged Cumulative percentage Fig. 2. Presumed EV battery capacity (Ebat) distribution Fig. 3. Presumed battery initial state-of-charge (SoCi) distribution The load curve simulations are carried out for three frequentation cases: 50 EV/day, 100 EV/day and 200 EV/day. It is further assumed, that an EV can arrive at a charging station in any minute of the given hour, depending on the frequentation and daily distribution (Fig. 1). As the vehicles show variable Ebat and SoCi values, stochastic approach on Monte Carlo method with 10’000 iterations is utilised; the same method already been proposed by some authors. As the objective charging time was fixed to 5 minutes, the instantaneous charging power varies according to the transferred energy, defined by Ebat, SoCi and inherent losses. The Monte Carlo simulation returns following statistics: 1) instantaneous charging power: median, 3 quartile and maximum values; 2) daily transferred energy: median, 3 quartile and maximum values; 3) number of simultaneously charging vehicles: 3 quartile and maximum values. As it is known from the descriptive statistics theory, the median value represents the middle of a data set, of which 50 % are smaller and 50 % greater. The 3 quartile (further referred to as 3Q), in turn, refers to a value of which 75 % are smaller and 25 % greater in a studied data set. So, the median is valid for 50 % of the modelled cases and the 3 quartile to 75 %. Further, to suppress unwanted peaks in the utility grid, an energy storage buffer must be installed for power flow management. Here, two main partial decoupling strategies are observed: 1) load levelling – an ultrafast EV charging station is supposed to draw moving average charging power from the grid, the average is in the studied case taken over an hour and the strategy itself is based on the discrete low-pass filter analogy; 2) load shifting – to an ultrafast EV charging station, more power is allocated during nighttime and less power during the grid peak hours, so the buffer absorbs energy when the grid overall load is minimal and releases energy for EV charging when the grid is more heavily loaded. A buffered EV charging station can be described as a three port entity with connection ports for the utility grid, electric vehicle and energy storage buffer (Fig. 4). This general conception is based upon the dc architecture with a common dc bus and each port characterised by power Pgr, PEV, Pst as well as energy conversion and transmission efficiency ηgr, ηEV, ηst, respectively. 0% 20% 40% 60% 80% 100% 0% 2% 4% 6% 8% 10% 12% 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 Cu m ul at iv e pe rc en ta ge