Ultra-Fast DC-Charge Infrastructures For

1

Ultra-Fast DC-Charge Infrastructures for EV-Mobility and Future Smart Grids D. Aggeler, F. Canales, H. Zelaya - De La Parra, A. Coccia, N. Butcher, and O. Apeldoorn Email: [email protected], [email protected], [email protected]

Index Terms—dc fast charger, e-mobility, electric vehicle, charging infrastructures, smart grid.

I. I NTRODUCTION As leading power and automation technology company, ABB feels a strong commitment to e-mobility. A commitment towards utilities and infrastructure providers in order to offer integrated smart charging solutions and to prepare the grid for the challenges coming along with e-mobility; a commitment towards the EV (Electric Vehicle) driver to enable safe recharging at the required speed and to prevent battery depletion; a commitment towards the environment to make individual mobility more sustainable and allow for e-mobility powered by renewables, all integrated in a reliable smart grid. For a sustainable transportation, ABB provides different charging solutions for EV batteries which are organized into three main categories: residential, public and fast/ultra-fast infrastructures. With regards to residential charging infrastructures and home chargers, typically providing single phase ac line at 110/220 Vrms and 50/60 Hz, deliver efficient, low power vehicle charging sequences which can refill a battery during the D. Aggeler, and F. Canales are with the ABB Corporate Research, Segelhofstrasse 1K 5405 Baden-Daettwil, Switerland. H. Zelaya - De La Parra is with the ABB Corporate Research, V¨ aster˚ as, Sweden. A. Coccia was formerly with the ABB Corporate Research, Baden-Daettwil, Switzerland. N. Butcher, and O. Apeldoorn are with ABB Industrie, Turgi, Switzerland.

night and reaching full capacity before morning. Charging overnight ensures that the load on the grid is low, and the car is refilled economically using low cost night-rate power. A range of home chargers to suit the needs of different homes are available - indoor, outdoor, wall mounted - and all incorporate the safety systems you can expect of any home appliance plus those requirements in the available standards (IEC 61851, SAE J1772). With respect to public charging stations, these are typically supplied from three phase ac mains at 50/60 Hz. One can classify them semi-fast charging solutions as they can charge a battery in few hours while the driver is at work or be used to keep the car charged up during everyday activities such as shopping or dining out. These charging poles will be found throughout the town or city at company parking, public buildings, stores and large car parks. These charging poles are strong and safe to fit the requirements of a public space. In most cases the consumer will pay for the electricity used in the charging, as a service so the charging pole will include an authentication and/or payment system. Finally, ultra-fast charging stations technologies allow a ‘fuel stop’-equivalent for EVs, by feeding direct currents into the battery sets at variable voltage levels in the range between 50 -700 Vdc in order to satisfy the different vehicles and battery

60

50 Minutes to charge 80% capacity

Abstract—Nowadays Power Electronics (PE) is entering more and more in technology which traditionally belongs to different engineering disciplines. E-mobility is one of these. Power Electronics in fact imposes itself as an emerging technology to enhance sustainable mobility, addressing all the engineering aspects starting from energy distribution for charging purposes until energy transformation on board of the traction related vehicles. This paper in particular focuses on newly developed PE infrastructure technologies enabling fast battery charging processes. Depending on the battery and vehicle type, a recharge sufficient for a travel range of more than 100 km in less than 10 min is readily achievable. As battery technologies continuously advance, recharging will become available with the speed and simplicity of a today’s fuel stop. Two PE converter architectures for recharging infrastructure applications will be presented and discussed based on both lowfrequency (LF) and high-frequency (HF) isolation requirements. Technical evaluation of the two different technologies will be addressed and presented, including a pro- and contra analysis. The impact on the grid is studied by means of simulation with the assumption of a dc fast charging station placed in a rural area in Sweden.

40

30

Heavy Vehicle Sedan/SUV City EV

48 kWh 24 kWh

20 16 kWh 10

0 50 Fig. 1.

100 150 200 Charging power in kW

250

Typical 80 % charging time versus dc charger available power.

2

II. DC CHARGE PE SYSTEM ARCHITECTURE THE REQUIREMENTS

As discussed above, this paper describes two solutions for the implementation of the dc fast charge processes for EVs. The topologies must comply with several basic requirements in order to enable a safe, secure, fast and efficient charging. As a first absolute requirement, the PE system architecture should be able to guarantee an universal supply, to cope with the different specifications of several types of vehicles: from small cars (usually characterized by a few kWh at nominal energy stored in the battery sets), untill trucks (usually characterized by several tens of kWh at nominal energy stored).

350

Critical region – key performance target for optimization

300 DC output current Iout [A]

types. This allows to the car to be charged in the shortest possible time. Combined with the latest battery technologies, this could allow a full re-charge in less than five minutes. These chargers will be installed in highway rest areas and convenient city refueling points. Grid energy management and power quality are provided by state of the art power electronics; Smart grid interfacing and integrated energy storage to manage the variations in power production. The vehicle connections will be based on industry standards, ensuring compatibility with all vehicles. In particular, the time to charge an electric vehicle depends on the power available from the charger to fill the battery pack. Short charging times, especially for larger vehicles, require high power. The newly developed dc fast chargers support very fast charging sequences even for heavy vehicles, and as new battery types capable to support these rates of charge become widespread the traditional range limitation of electric vehicles will fade away, enabling long distance journeys and high use fleet vehicles. Figure 1 shows typical charging time indicators versus power available from the dc-fast charging unit. There has been several publications over the last years that describe the effects of the EV in the network. The great majority concentrates on the overall national grid but only very few look at the consequences at the distribution grid level. The EV involvement in a smart grid context is also referred to as e-mobility. ABB sees its commitment towards e-mobility in several areas: • enabling safer charging in a required time • preventing unnecessary battery depletion • providing integrated and smart charging solutions • preparing the grid to the challenges of e-mobility • providing storage devices As previously mentioned, this paper focuses on newly developed PE infrastructure technologies enabling fast battery charging processes. Therefore, first of all the dc charging stations requirements are discussed in Section II. In the Section III and Section IV the two PE converter architectures for charger infrastructure applications will be presented and discussed based on both low-frequency (LF) and highfrequency (HF) isolation requirements. Technical evaluation of the two different technologies will be addressed and presented, including a pro- and contra analysis in Section V. Finally, in Section VI, the integration into a smart grid is discussed and the most important issues re highlighted.

250 200 150 Front end power constrained

100

This line has been defined based on constant peak current at reducing duty cycle

50 0 0

100

200

300 400 500 600 DC Output Voltage Vout [V]

700

800

Fig. 2. Output current profile for non-isolating and isolating 125 kW converter dc charging station units.

This first basic requirement implies that the PE architecture must be able to supply any dc output voltage in the range between 100-600 V, according to the design voltages of the battery packs installed on board of different types of EVs. As a second requirement, the maximum charging time for a high specific-energy battery set (around 30-40 kWh) should not exceed 10 minutes duration to bring the battery set to the 80 % of its nominal SOC (State of Charge). The final goal is indeed to reduce the driver waiting time at the ‘re-filling’ station. This second requirement therefore sets the minimum output power that the PE system architecture has to provide, (in this specific case, the system is rated at 125 kW), and the charging current profile which should be guaranteed in the whole output voltage range. Figure 2 represents the battery charging current profile versus the converter output voltages. In order to guarantee secure operation for the battery set, the electrical quantities (voltages and current) ripple at the output of the converter charging station is set to very tight values. In particular the maximum current ripple is set to be 1 % of the minimal value from the output current profile, while the voltage ripple is specified to be smaller than 5 % of the maximum output voltage (respectively 50-300 A depending on the output current profile and 600 V as given in specification sets for future charging stations). Last but not least, according to the safety standards, the charging station must guarantee a fully galvanic isolation towards the main distribution network. Depending then on the other main product requirement specifications (efficiency, power density, cost, reliability), it is possible to attack this last requirement in two different approaches. The first one addresses the galvanic isolation issue by means of a traditional LF transformer connected to the grid, while the second approach, implements the galvanic isolation requirement at the level of the dc-dc converter which properly shapes the charging current profile. It is clear that in this second case an input PE converter stage connects directly to the main distribution network. The next chapters will technically describe the two different approaches whereas the main focus is put on the dcdc converter stage, identifying the different key issues while designing the converter architectures. Finally a comparison will highlight the differences between the two different so-

3

lutions in terms of strengths and drawbacks. on

III. LF CONVERSION ARCHITECTURE FOR

S1

DC CHARGING STATIONS

Active front end AC

DC DC

Line-frequency a) transformer

S1

Vin

3-ph buck converter

LCL-filter

S2

S3

Cin S4

DC

S5

S6

ILj

Lj

IL Iout Cout

Vout

b) Fig. 3. (a) LF converter solution for dc fast charging stations and (b) the three phase interleaved buck converter based on the standard ABB product.

on S2 off on S3 off 0

0.5 1.0 Time [ms]

a)

iLj [A]

100 90 iL1

80

Vout [V]

Figure 3 represents the circuit layout utilizing a LF galvanic isolation concept. The system architecture comprises: • a 1:1 turns ratio input line-frequency transformer to isolate the system from the main distribution network galvanically • an input LCL filter to be compliant to the international IEC harmonic injection standards • a three phase active rectifier unit which guarantees active power factor control for the line side electrical quantities and which produces a constant dc-link voltage set at 750 Vdc for the secondary conversion stage • a three phase interleaved buck conversion unit to allow appropriate control of the battery currents during the charging cycle. Both of the conversion units are implemented with two standard ABB modules, where in particular the output of the second conversion stage is connected in star point configuration through three different inductors. By means of this configuration it is in fact possible to share the output power separately over three different converter branches and in this way reducing the overall semiconductor stress and total converter losses. Furthermore, the interleaved converter topology [1] offers advantages regarding the filter reduction due to the fundamental frequency been effectively multiplied by the number of phases. Figure 4 shows the characteristic waveforms of the interleaved buck converter as well as the simple digital control signals of the switches which is a further benefit of this power electronics architecture. Finally, the bidirectional configuration of the system enables not only power to be transferred from the grid to the vehicle (G2V) but also from the vehicle to the grid (V2G). During the converter design process special attention must be paid to the system optimization in order to meet the proper product requirement specifications in terms of efficiency, cost, size, volume and weight. All the components from semiconductors, through heat-sinks up to passive filter components

Gate Signals

off

iL2

iL3

410 Output voltage 400 390 0

b)

0.5 1.0 Time [ms]

Fig. 4. Characteristic waveforms of the three phase interleaved buck topology: (a) Gate signals for the three interleaved buck branches phaseshifted by 120 ◦ C to reduce the output current ripple. (b) Converter output quantities with the interleaved inductor current and the output voltage.

will have to be designed accordingly to develop an optimized system. In particular, based on the existing standard ABB modules, first a suitable operating frequency of the three phase interleaved buck converter has to be investigated. Therefore, the thermal behavior of the used 1200 V semiconductors as well as the output charging profile which is required (cf. Figure 2) have to be taken into account. Regarding the three phase interleaved buck converter, the reduction in volume of the passive elements (inductor and capacitor) preferably desire a high switching frequency. However, the analysis of the semiconductor power losses and the thermal behavior, with assumed thermal specifications of Tamb =50 ◦ C and Rth,hs =0.05 K/W (air inlet speed of 4.0 m/s) results in a applicable switching frequency of fs =2 kHz. Figure 5 illustrates the maximal possible output current profile depending on the switching frequencies compared with the target profile characteristic. Out of this it is clearly shown that fs =2 kHz meets best the target output current profile. According to the suitable operating switching frequency per phase buck converter of fs =2 kHz, the required inductance value can be determined. Based on the interleaved buck topology of three single phase (N = 3) buck converters [2], [3]

4

OP 3 OP 2

300 250 200

OP 1

150

fs = 1 kHz fs = 2 kHz fs = 3 kHz

100 50 0 0

V/I target characteristic 100

200 300 400 500 Output Voltage Vout [V]

a)

600 b)

45 40 35 30 25 20 15 10 5 0

100

V/I target characteristic

98

OP 2

Efficiency [%]

Output Current Iout [A]

350

Phase inductance Lj [mH]

400

96 94 92 90

0 100 200 300 400 500 600 Output Voltage Vout [V]

88 86

Fig. 5. (a) Maximal output current profiles (OP1, OP2 and OP3) calculated at a maximal IGBT junction temperature of Tj =110 ◦ C and for the set of switching frequencies fs ={1,2,3 }kHz. (b) Design of the minimal phase inductor value at a switching frequency of fs =2 kHz for the target characteristic and the OP2 to guarantee a output current ripple of maximal iL,pp =1% · min(Iout ).

the minimal inductance to guarantee an output current ripple of iL,pp =1% · min(Iout (Vout )) is calculated as
v   m  out · 1− Lj = max · [1 + m − N D] , (1) iL,pp · fs ND whereas m denotes the function f loor(N D), D=Vout/Vin is the duty cycle per phase buck converter and j ∈ {1,...,N } implies the j-th buck phase. Regarding Eq. (1) the required inductance value per phase Lj is strongly depending on the output current profile iout (vout ). Therefore, the maximal inductance for the target characteristic is calculated at Vout =50 V to Lj,50 V =40 mH as illustrated in Figure 5 (b). With the specified output profile OP 2 based on the thermal behavior of the applied IGBT semiconductors, the inductance value can be reduced. As presented in Figure 5 (b), the maximal inductance appears at Vout =125 V and is calculated to Lj,125 V =11.63 mH. To realize the calculated inductance value of OP 2, the standard available core materials nanocrystalline [4] and powder ironsilicon cores (93.5 % Fe and 6.5 % Si) [5] have been investigated. In case of the nanocrystalline material the saturation of the core, due to the high dc current (up to 100 A per phase), establishes as the critical part for the design. Thus, no suitable design could be made with the conventional available cores nor an adequate stacked solution. The low inductance factor AL compared to the nanocrystalline material and the large effective cross section area of the iron-silicon material shows advantages related to saturation of the core material. Nevertheless, a large number of cores have to be stacked on TABLE I I NTERLEAVED

PHASE INDUCTOR Lj CHARACTERISTICS FOR THE OUTPUT PROFILE OP 2.

Units

Iron-Silicon (Fe-Si)

Core type

-

toroidal

Max. core flux density

T

1.5

Turns

-

57

A/cm2

200

Current density Stacked cores AC core flux density Peak core flux density Volume

-

3x 7

mT

0.058

T

1.3

dm3

3x 3.5

0

100

200 300 400 Output voltage Vout [V]

500

600

Fig. 6. Efficiency of the three phase interleaved buck converter as a function of the output voltage. Note: DC-link capacitor losses on the input and output side of the interleaved buck converter are not considered in the loss calculation.

each other to guarantee 1 % current ripple at the worst case. In Table I the design parameters for the stacked phase inductor solution based on the iron-silicon material is summarized. Due to the large number of stacked cores per phase inductor prefereably the number is splitted into three smaller inductors which are connected in series to minimie the influence of the parasitic inductor capacitances. Beside the phase inductor also the single output capacitor combined for all the buck phases has to be designed. To guarantee an output voltage ripple less than 5 % of the maximal output voltage, the output capacitor is calculated as
1 1 iL,pp Ts  · · · , (2) Cout = max ∆vout 2 2 2N which results in a conservative designed output capacitance of Cout =1.86 µF. Finally, the efficiency of the three phase interleaved buck converter is analyzed and presented in Figure 6. The efficiency value drops down for low duty cycle or low output voltage values respectively as well know for buck converter topologies. An efficiency of 98 % can be achieved at operation above an output voltage of 350 V. IV. HF CONVERSION ARCHITECTURE FOR DC CHARGING STATIONS

Figure 7 (a) represents the circuit layout implementing a HF galvanic isolation concept. The system architecture comprises: • an input LCL filter to be compliant to the international IEC harmonic injection standards • a three phase active rectifier unit which guarantees active power factor control for the line side electrical quantities and which produces a constant dc-link voltage set at 750 Vdc for the secondary conversion stage • two in parallel dc/dc isolated converter stages, implementing the high frequency isolation. As can also be seen from Figure 7 (b), the dc-dc isolated converter consists of: • a single phase input inverter unit, operated at a fixed switching frequency of fs =8 kHz and modulated with

5

Galvanic isolated dc-dc

Active front end

on AC

DC

S1

off DC

on

DC LCL-filter

a)

S3

Gate signals

DC

DC

off on S2 off

HF dc/dc converter S1 Vin

S2

IL Lout

ILV C 7

Cin S3

LV

S4

CS

on

Iout

Cout

S4

off Vout

0

N1:N2

62.5

Iof f,S1 > Icrit   2
4 1 2. · CIGBT + · (CT + CS ) · Vin = · Lσ 3 2

(3)

In addition, the output current ripple has to be considered which reflects to the primary side and therefore influences the ZVS operating range. Due to the constant high output current of 100 A per dc-dc converter stage and the suitable transformer design, including the leakage inductance Lσ and parasitic capacitance CT , ZVS condition is achieved over the whole load range at nominal output current condition. If the dc-dc converter is operated in partial load, this is depending on the EV type and battery charging capability, ZVS condition is not any more guaranteed and high turn on losses occur. Moreover, due to the phase-shift control, where the commutation instants of the two inverter branches are phase-shifted continuously in the interval [0 ◦ C-180 ◦ C] to properly adapt the average output voltage level, the turn off commutation

Current [A] Current [A]

phase-shift control to vary the output dc voltage applied to the battery sets • a medium frequency transformer • an output rectifier Also with the HF conversion architecture, it is possible to achieve not only G2V operation, but also V2G operation mode if the output rectification stage is implemented by active devices. The complete high frequency conversion stage is implemented with a PWM concept as illustrated with the gate signals in Figure 8 (a), in order to reduce the output voltage and output current ripple within the values defined in the specification set. To achieve high efficiency, the HF converter stage is operated under soft switching approach of zero voltage switching (ZVS). To achieve proper ZVS condition for the primary switches, the load depending turn off current Iof f,S1 has to be larger than the critical current Icrit [6] which can be calculated as

Voltage [V]

Fig. 7. (a) HF converter solution for dc fast charging stations with two in parallel connected galvanic isolated dc-dc stages and (b) the corresponding HF dc-dc isolated converter topology.

187.5

250

Time [ms]

a) b)

125

100 50 0 -50 -100 105

Ioff,S1

iLV

iout

100 95 800 400 0 -400 -800 b)

vp vs 0

62.5

125

187.5

250

Time [Ps]

Fig. 8. Characteristic waveforms of the PWM dc-dc converter: (a) Gate signals for all switches of the full bridge implemented with phase-shift modulation. (b) Illustrated are the leakage inductor current (top), the output current with ≤ 1 % current ripple (middle) and the primary and secondary transformer voltages (bottom).

occurs at hard switching. Therefore, to reduce the switching losses of the primary semiconductors, some capacitive snubbers [7] have been place in parallel to the active devices in order to reduce the overlapping region between devices voltages and currents during the commutations. Finally to reduce the over-voltages across the secondary side switches/diodes, some partly regenerative RCD (resistor, capacitor and diode) snubber circuits [8] have been placed in front of the secondary side dc-link output capacitor Cout . Using 1200 V/400 A IGBT devices for the full bridge and 1200 V diode devices for the passive rectifier stage and considering same thermal condition as for the three phase interleaved converter, a constant current output profile as presented in Figure 9 (a) is resulting. Comparing the achievable current profile with the target characteristics shows that several dc-dc stages, in fact three of them (Np =3 ), have to be connected in parallel to meet the target current profile and also the selected current profile of the LF charger stations, respectively.

6

300

Output inductance Lout [mH]

Output current Iout [A]

350 3x dc-dc stage

250 2x dc-dc Stage

200 150

1x dc-dc Stage

100 50 0 0

V/I target characteristic 100

a)

200 300 400 Output voltage Vout [V]

500

100

45 40 35 30 25 20 15 10 5 0

98 V/I target characteristic 1x dc-dc Stage 3x dc-dc stage

0 100 200 300 400 500 600 Output voltage Vout [V]

600 b)

Fig. 9. (a) Maximal output current profiles calculated at a maximal IGBT junction temperature of Tj =110 ◦ C and for a fixed switching frequency of fs =8 kHz (assumed are 30 % reduction of switching losses due to the additional snubber circuit). (b) Design of the minimal output inductor per isolated dc-dc converter to guarantee the required current ripple (current interleaving of the parallel connected dc-dc stages is considered).

With respect to the practically achievable output current profile, the output inductance value can be equally determined as in Eq. (1). Based on the 1 % current output ripple requirements and considering the transformers turns ratio as also the doubled output frequency of the passive rectifier as illustrated in Figure 8 (b). Therefore, the output inductance is calculated as
v  m  out · 1− Lout = max iL,pp · 2fs Np D  (4) · [1 + m − Np D] , whereas fs denotes the switching frequency of the primary side switches. To calculate the duty cycle, the turns ratio of the transformer n = N1 : N2 has to be taken into account and therefore the duty cycle results in D

=

Vout . Vin · 1/n

(5)

The required parallel connection of Np = 3 isolated dcdc stages reduces the inductance value per dc-dc stage if the implementation of the modulation method considers the current interleaving at the output. Out of this, a minimal output inductance value is calculated at a duty cycle of 0.5 and resulted in Lout =1.1 mH. The Table II summarizes the characteristic parameters of the output inductor design. As shown in Figure 7 (a), the output capacitor Cout can either be combined for all the parallel connected isolated dcTABLE II CHARACTERISTICS OF A SINGLE DC - DC STAGE AND CALCULATED FOR THE OUTPUT PROFILE OF THREE ISOLATED DC - DC STAGES CONNECTED IN PARALLEL .

O UTPUT INDUCTOR Lout

Units

Iron-Silicon (Fe-Si)

Core type

-

toroidal

Max. core flux density

T

1.5

Turns

-

49

A/cm2

200

Current density Stacked cores AC core flux density Peak core flux density Volume

-

3

mT

0.051

T

1.19

dm3

2.2

Efficiency [%]

400

96 94 92 90 88 86 84 82 80 0

100

200

300 400 Output voltage [V]

500

600

Fig. 10. Efficiency of the galvanic isolated dc-dc converter as a function of the output voltage. Note: The RCD-snubber losses of the passive rectifier bridge as well as the dc-link capacitor losses on the input and output side of the PWM dc-dc converter are not considered in the efficiency considerations.

dc converters or each converter has its own output capacitor. Assuming one output capacitor for all the dc-dc stages the output capacitor results to Cout =1.86 µF . Therefore the same value calculated for the three phase interleaved buck approach resulted because of the same current ripple requirement and current output profile. Finally, the core element of the isolated dc-dc architecture approach, the HF transformer which provides the galanvic isolation to the charging stations has to be designed. The standard ABB module which also provides galvanic isolation shows that such a transformer is realizable with small leakage inductance and for the required output current profile. Using nanocrystalline material for the transformer cores resulted in a power density of 8.5 kW/dm3 . The galvanic isolated dc-dc converter stage efficiency under ZVS condition (nominal load/current condition) is presented in Figure 10. Obviously, there is the same efficiency drop for low output voltages and low duty cycle values resulting from the buck converter characteristics as for the first architecture approach. V. LF VERSUS HF ISOLATION ARCHITECTURE T HE C OMPARISON In this section, the technical evaluation of the two different investigated PE architectures is addressed and advantages and disadvantages for the practical realization are discussed. Both, the LF and HF isolation approach, use the same active front end PE converter which is based on standard ABB modules. Therefore, the main architecture difference is in the dc-dc converter stage. Obviously, the dc-dc converter stage of the LF approach provides no galvanic isolation. Thus, a linefrequency transformer is needed which galvanically isolates the batteries from the mains. In the HF isolation architecture, the galvanic isolation is provide in the dc-dc converter stage with a HF transformer. The main advantages of the LF approach with the three interleaved buck converters are the well known and simple topology with only three switches and three diodes. Furthermore, the interleaved approach results in smaller output ripple current due to current ripple cancellation. Futhermore

7

A to B >100 km

Fig. 11.

Rural charging station.

the implementation of the 120 ◦ C phase-shifted gate signal needs low effort. The drawbacks of this PE architecture are the relatively high switching losses and reverse recovery losses which limit the operating switching frequency fs . This results in a large filter size (Lj and Cout ). In combination with the line-frequency transformer this charging station approach results in a large size and high cost system mainly influenced by the required magnetic materials. To reduce the amount of magnetic material and to decrease the total volume requirements of the charging station the operating switching frequency has to be increased and the galvanic isolation needs to be integrated into the dc-dc stage (HF approach cf. Section IV). There, the HF transformer design is performed with a high switching frequency of 8 kHz (instead of 50 Hz transformer) and the equivalent frequency for the inductor design will be 3 · 16 kHz (instead of 3 · 2 kHz). Resulting is a filter with a higher power density. Moreover the volume and weight are reduced by a factor which can be expressed as V [dm3 ] ∼ m[kg] ∼

1 . f 0.75

(6)

It should be noted that Eq. (6) is only valid in the switching frequency range proposed in this paper. The main disadvantage of the HF PE architecture are the high snubber losses to avoid over-voltages across the passive rectifier devices. Additionally, it has to be considered that for this kind of topology, the transformer design/layout mainly influences the soft switching condition especially in partial load condition.

Fig. 13.

Simulation results of rural charging stations.

VI. T OWARDS INTEGRATION INTO THE SMART GRID The future Smart Grid system will have EV charging in any of the forms already mentioned earlier in the paper: slow, semi-fast, fast or ultra fast charging. The particular case of interest in this paper is the ultra fast dc charging, which as stated in the previous section can be in the range of 125 kW to 300 kW. The concept of dc fast charging station has been studied with different possible scenarios, which include various degrees of EV penetration as well as the location of the charging station in the city or in a rural location. Based on the existing infrastructure in a rural area in Sweden, the impact of a fast charging station has been investigated. Figure 11 shows the location of the rural charging station. A simulation has been run with a test scenario as shown in Figure 12 which contemplates a charging station with 8x300 kW chargers. The results of the case study show that voltage variations can be expected between 3 to 8 % on buses on the vicinity of the charging station. This represents a worst case scenario, when the charging poles are connected via a 20-25 km long high voltage cable. This is summarized in Figure 13. The voltage variations can be avoided by means of an energy reservoir such as batteries or kinetic energy (flywheel). Since both types of battery chargers mentioned in this papers have active power factor correction on the front end, the effects of harmonics into the system is minimized.

VII. C ONCLUSION

Fig. 12.

Diagram for rural charging stations.

The paper focuses on two different conversion structures to implement ultra fast dc charging stations for EV. The LF isolation, with non-isolated three phase interleaved buck converter, and the HF isolation, with the galvanic isolated dc-dc converter topology, approach have been discussed and benchmarked. Finally, the strengths and weaknesses of both PE architecture approaches have been analyzed and discussed related to the most important product requirement specifications. The impact of the dc fast charging stations have been evaluated based on simulation of rural charging stations.

8

R EFERENCES [1] O. Garcia, P. Zumel, A. de Castro, and A. Cobos, “Automotive dc-dc bidirectional converter made with many interleaved buck stages,” Power Electronics, IEEE Transactions on, vol. 21, no. 3, pp. 578 –586, May. 2006. [2] T. Hegarty, “Benefits of multi-phasing buck converters,” 2007. [3] C. Chang and M. Knights, “Interleaving technique in distributed power conversion systems,” Circuits and Systems I: Fundamental Theory and Applications, IEEE Transactions on, vol. 42, no. 5, pp. 245 –251, may. 1995. [4] “http://www.vacuumschmelze.de.” [5] “http://www.micrometalsarnoldpowdercores.com.” [6] J. Sabate, V. Vlatkovic, R. Ridley, F. Lee, and B. Cho, “Design considerations for high-voltage high-power full-bridge zero-voltageswitched PWM converter,” in Applied Power Electronics Conference and Exposition, 1990. APEC ’90, Conference Proceedings 1990., Fifth Annual, mar 1990, pp. 275 –284. [7] A. Petterteig, J. Lode, and T. Undeland, “IGBT turn-off losses for hard switching and with capacitive snubbers,” in Industry Applications Society Annual Meeting, 1991., Conference Record of the 1991 IEEE, sep-4 oct 1991, pp. 1501 –1507 vol.2. [8] L. Mweene, C. Wright, and M. Schlecht, “A 1 kW 500 kHz front-end converter for a distributed power supply system,” Power Electronics, IEEE Transactions on, vol. 6, no. 3, pp. 398 –407, jul 1991.