Advantages of Using Supercapacitors and Silicon Carbide on Hybrid Vehicle Series Architecture

In recent years enormous growth has taken place in the hybrid vehicle sector; parallel architecture is the most widespread configuration regarding medium size cars. At the same time, storage systems and power electronics have experienced some important innovations. The development of supercapacitors has permitted management of high power with elevated efficiency. Moreover, the availability on the market of silicon carbide components has allowed a significant reduction of power electronic losses. These improvements may challenge the hybrid architecture used in medium size cars nowadays. On one hand, series architecture would relevantly benefit from an electric powertrain efficiency increase, on the other hand, these innovations would generate low benefits in parallel architectures. The aim of this paper is to evaluate electric component average efficiency over different road missions, in order to estimate fuel economy over various working conditions and finally to establish which hybrid configuration is most efficient in vehicle applications.


Introduction
The spread on the market of medium size cars with hybrid parallel architecture has led to increased studies in the efficiency of the electric and electronic components necessary to realize this configuration.Often this architecture is realized with a planetary gearbox and two electric machines (power-split transmission).In [1] the component efficiency of a 2007 Toyota Camry is analyzed, whereas in [2,3] the same study is carried out, respectively, for a 2008 Toyota Lexus and a 2010 Toyota Prius.In [4,5] different studies are performed in order to synthesize optimal control logics for power split transmissions.
Whether the hybrid architecture is realized with a traditional transmission or a planetary gearbox, the term parallel is related to the possibility of providing power to the wheel shaft both with the combustion engine and the electric machine.Parallel architecture is sketched in Figure 1 with a motor generator unit (MGU); regenerative braking is allowed, whereas during combustion engine operation mode no electric losses are involved, the engine being directly connected to the transmission gearbox.
With regards to series architecture, there have been applications and studies on the use of such structures on urban buses.In [6] an analysis of potential fuel consumption reduction, using a series architecture on urban buses, is performed.The use of supercapacitors as the storage system in this architecture seems to offer potential benefits [7].Supercapacitors seem to offer benefits on urban bus applications also using other hybrid architectures [8].In [9] different storage systems (including supercapacitors) are evaluated, and the series architecture improvement on urban buses equipped with turbo-compound technology is shown; this technology consists of a combined cycle made of a combustion engine plus a gas turbine that allows exhaust gas enthalpy recovery.With regards to series architecture, there have been applications and studies on the use of such structures on urban buses.In [6] an analysis of potential fuel consumption reduction, using a series architecture on urban buses, is performed.The use of supercapacitors as the storage system in this architecture seems to offer potential benefits [7].Supercapacitors seem to offer benefits on urban bus applications also using other hybrid architectures [8].In [9] different storage systems (including supercapacitors) are evaluated, and the series architecture improvement on urban buses equipped with turbo-compound technology is shown; this technology consists of a combined cycle made of a combustion engine plus a gas turbine that allows exhaust gas enthalpy recovery.
In [10] different plug-in hybrid vehicle configurations are compared.The evaluation of fuel economy improvement with different hybrid powertrains is a current issue, in fact it is strictly related to component efficiency.
Electric and electronic components have greatly improved their performance in recent years.As far as supercapacitors are concerned, the nominal cell voltage has continuously increased up to an available 3.0 V (BCAP3000 P300 K04 [11]) on today's market; this increase guarantees a great increase in energy density.Moreover, silicon carbide power electronics have permitted a great reduction in electric drive losses [12].
In this paper, the efficiency of the components necessary to realize a series powertrain will be analyzed in order to evaluate the performance achievable with the technologies available on the market.Both silicon and silicon carbide power electronic solutions are investigated, as well as battery and supercapacitor storage.Using an appropriate model created in MATLAB Simulink environment, the efficiency of such components is simulated on the Highway Fuel Economy Test (HWFET), the Urban Dynamometer Driving Schedule (UDDS), the US06 [13] and on some experimentally measured road missions.
Finally, combustion engine losses are introduced, in order to calculate the overall powertrain efficiency and fuel economy on the above-mentioned road missions.

Control Logic
The series architecture is sketched in Figure 2, where the generator (G) and the traction machine (M) are shown.In [10] different plug-in hybrid vehicle configurations are compared.The evaluation of fuel economy improvement with different hybrid powertrains is a current issue, in fact it is strictly related to component efficiency.
Electric and electronic components have greatly improved their performance in recent years.As far as supercapacitors are concerned, the nominal cell voltage has continuously increased up to an available 3.0 V (BCAP3000 P300 K04 [11]) on today's market; this increase guarantees a great increase in energy density.Moreover, silicon carbide power electronics have permitted a great reduction in electric drive losses [12].
In this paper, the efficiency of the components necessary to realize a series powertrain will be analyzed in order to evaluate the performance achievable with the technologies available on the market.Both silicon and silicon carbide power electronic solutions are investigated, as well as battery and supercapacitor storage.Using an appropriate model created in MATLAB Simulink environment, the efficiency of such components is simulated on the Highway Fuel Economy Test (HWFET), the Urban Dynamometer Driving Schedule (UDDS), the US06 [13] and on some experimentally measured road missions.
Finally, combustion engine losses are introduced, in order to calculate the overall powertrain efficiency and fuel economy on the above-mentioned road missions.

Control Logic
The series architecture is sketched in Figure 2, where the generator (G) and the traction machine (M) are shown.
The advantage of this architecture is that the combustion engine can be controlled independently from the road mission profile.The totality of the energy supplied by the engine is subject to losses in the generator, in the traction electric machine and in the power converters.The storage balances the instantaneous difference between the power provided by the engine and that required by the road mission.Only the stored energy is subject to DC-DC converter and storage losses.
It is therefore clear that the higher the efficiency of the storage system (with its related converter), the higher the possibility to control the combustion engine.Indeed, if the storage losses are low, the engine can operate at maximum efficiency working conditions (approximately at 50% of the maximum power) and the exceeding energy can be stored; when the storage system reaches full charge, it is possible to shut down the engine until the stored energy is no longer sufficient to guarantee total electric traction.The advantage of this architecture is that the combustion engine can be controlled independently from the road mission profile.The totality of the energy supplied by the engine is subject to losses in the generator, in the traction electric machine and in the power converters.The storage balances the instantaneous difference between the power provided by the engine and that required by the road mission.Only the stored energy is subject to DC-DC converter and storage losses.
It is therefore clear that the higher the efficiency of the storage system (with its related converter), the higher the possibility to control the combustion engine.Indeed, if the storage losses are low, the engine can operate at maximum efficiency working conditions (approximately at 50% of the maximum power) and the exceeding energy can be stored; when the storage system reaches full charge, it is possible to shut down the engine until the stored energy is no longer sufficient to guarantee total electric traction.
The threshold to shut down the engine cannot be 100% of the storage state of charge (SOC); it is obviously necessary to guarantee an appropriate margin to recover backward energy caused by possible prolonged braking (long downhill roads).
In this study supercapacitors (SCs) were used as storage system, this choice was made for different reasons.A large quantity of power flows in the storage in the series architecture; therefore, long life in terms of number of cycles, is an important characteristic.Moreover, SOC measurement is precise and reliable.Finally, SCs can charge and discharge with high efficiency.
A heuristic management of power generation was developed.The internal combustion engine (ICE) charges the SCs until 75% of their maximum voltage (in such a way leaving a suitable margin for long braking); when the voltage goes below 55%, ICE turns on again.SCs can work until 50% of their maximum voltage; the lower limit margin (between 50% and 55%) is used to avoid frequent ICE ignitions when the speed is lower than 30 km/h (for comfort reasons).

Internal Combustion Engine
The spark-ignition engine presented in [14] was used as reference.In this study a 40 kW ICE was used (as reported in Section 2.3), therefore ICE [14] torque and speed were normalized in per units and the related efficiency (plotted in Figure 3) was considered in this study.The threshold to shut down the engine cannot be 100% of the storage state of charge (SOC); it is obviously necessary to guarantee an appropriate margin to recover backward energy caused by possible prolonged braking (long downhill roads).
In this study supercapacitors (SCs) were used as storage system, this choice was made for different reasons.A large quantity of power flows in the storage in the series architecture; therefore, long life in terms of number of cycles, is an important characteristic.Moreover, SOC measurement is precise and reliable.Finally, SCs can charge and discharge with high efficiency.
A heuristic management of power generation was developed.The internal combustion engine (ICE) charges the SCs until 75% of their maximum voltage (in such a way leaving a suitable margin for long braking); when the voltage goes below 55%, ICE turns on again.SCs can work until 50% of their maximum voltage; the lower limit margin (between 50% and 55%) is used to avoid frequent ICE ignitions when the speed is lower than 30 km/h (for comfort reasons).

Internal Combustion Engine
The spark-ignition engine presented in [14] was used as reference.In this study a 40 kW ICE was used (as reported in Section 2.3), therefore ICE [14] torque and speed were normalized in per units and the related efficiency (plotted in Figure 3) was considered in this study.

Vehicle Features
The vehicle features used for the simulations are shown in Table 1.

Vehicle Features
The vehicle features used for the simulations are shown in Table 1.

Vehicle Features Values
Vehicle mass 1   3 [kW] 20 ICE rated speed 4 [rpm] 2500 1 The vehicle mass includes the electric machines (about 80 kg), the storage system (85 kg) and the electric drives (about 15 kg). 2 The vehicle performance is related to the electric motor and it is independent of the ICE sizing; as a consequence, ICE downsizing about 50%. 3 Power in the maximum efficiency working condition. 4Speed in the maximum efficiency working condition.

Road Missions
The reference cycles (US06, UDDS, and HWFET) allow a comparison between the results in literature and the results obtained in this study; the disadvantage of these road missions is that they are at constant slope.This is not usually relevant, but since long downhill missions are particularly critical in this study as SCs have lower specific energy compared to batteries, they may fully charge and dissipative braking might be necessary.
For this reason, in addition to the three standard cycles, two experimentally measured missions in Genoa city (urban and fast-urban) and another mission measured in Genoa's hinterland (extra-urban), were simulated.The main features of the six road missions are presented in Table 2.The extra-urban mission is particularly critical for the storage system due to a long downhill road.In particular, this mission is characterized by a significant difference in altitude, as shown in Figure 4.The extra-urban mission is particularly critical for the storage system due to a long downhill road.In particular, this mission is characterized by a significant difference in altitude, as shown in Figure 4.

Electric Machines
In Table 3 the main features of the electric machines are shown.

Electric Machines
In Table 3 the main features of the electric machines are shown.The traction machine weight, as well as the generator and the converters weight, were estimated according to the values shown in [3]. 2 The continuative-use torque is about 55% of the maximum torque (shown in the table), according to what is presented in [15]. 3The generator is directly connected at the ICE shaft, so the torque-speed characteristic of the generator should respect the ICE speed; for that reason the maximum torque and the base speed of generator and traction machine are different.
The electric machine losses were modelled according to the results illustrated in [15]; these losses, together with the related efficiencies, are plotted in Figure 5.In order to evaluate the losses of machines with different characteristics, torque, speed and losses were converted in per unit (p.u.) and identical losses were supposed to derive from identical machine typology according to what is proposed in [16].
An equivalent per unit table was obtained and used as a look-up table in the MATLAB Simulink model.
With the above-mentioned hypothesis, the look-up table can be used both for the traction machine and for the generator, despite the different values of nominal torque and speed.

Inverters
Inverter losses using traditional silicon (Si) components and innovative silicon carbide (SiC) components were both evaluated.
Starting from the data shown in [12], available for a 350 V and a 800 V DC-link and by interpolating these values, losses for a 650 V DC-link converter were calculated.
Losses for Si are shown in Equations ( 1) and ( 2) and plotted, together with the related efficiency, in Figure 6; in the same way , losses for SiC are shown in Equations ( 3) and ( 4) and plotted in Figure 7. .
speed [ ] 6000 In order to evaluate the losses of machines with different characteristics, torque, speed and losses were converted in per unit (p.u.) and identical losses were supposed to derive from identical machine typology according to what is proposed in [16].
An equivalent per unit table was obtained and used as a look-up table in the MATLAB Simulink model.
With the above-mentioned hypothesis, the look-up table can be used both for the traction machine and for the generator, despite the different values of nominal torque and speed.

Inverters
Inverter losses using traditional silicon (Si) components and innovative silicon carbide (SiC) components were both evaluated.
Starting from the data shown in [12], available for a 350 V and a 800 V DC-link and by interpolating these values, losses for a 650 V DC-link converter were calculated.
Losses for Si are shown in Equations ( 1) and ( 2) and plotted, together with the related efficiency, in Figure 6; in the same way , losses for SiC are shown in Equations ( 3) and ( 4) and plotted in Figure 7.

Supercapacitor Storage
The weight of supercapacitor storage was limited to 85 kg, according to the current applications [3] and the assumptive sizing available in literature [17,18].
Using the BCAP3000 P300 K04 [11] elementary cells, connected in series up to 85 kg in weight, a

Supercapacitor Storage
The weight of supercapacitor storage was limited to 85 kg, according to the current applications [3] and the assumptive sizing available in literature [17,18].
Using the BCAP3000 P300 K04 [11] elementary cells, connected in series up to 85 kg in weight, a

Supercapacitor Storage
The weight of supercapacitor storage was limited to 85 kg, according to the current applications [3] and the assumptive sizing available in literature [17,18].
Using the BCAP3000 P300 K04 [11] elementary cells, connected in series up to 85 kg in weight, a storage characterized by the parameters shown in Table 4 was obtained.
Supercapacitor storage was modelled in the MATLAB Simulink environment as an ideal capacitor with a series resistance (ESR).More complicated models, like the one in [19], simulate the supercapacitor behavior for high frequency; this behavior becomes significant above 0.2 Hz frequency; however, considering the road missions low frequency dynamic, those effects were neglected.The self-discharge phenomenon (modelled with the parallel resistance) was also neglected; although this effect is of primary importance for the storage management, these results are negligible from an energy point of view. 1 Equivalent series resistance; 2 The storage nominal power is one order of magnitude higher than the converter nominal power, therefore this is not a critical parameter for the storage.In fact, the sizing criterion for supercapacitors is the stored energy; 3 Usable energy, calculated considering a working condition between 50% and 100% of the nominal voltage.This energy is calculated as E = 0.5 × 0.75 × C × V 2 max .

DC-DC Converter
The supercapacitor voltage changes significantly because it depends on the stored energy; therefore, a DC-DC converter, connecting the storage to the DC-link, is necessary.All the energy which flows in the storage is affected by the converter losses both in the charging and discharging phases; for this reason, the efficiency evaluation of this component is of primary importance.
The DC-link voltage was chosen equal to 650 V, according to the available applications [3], the storage maximum voltage is 495 V, hence a half-bridge converter is suitable for this application because the storage voltage is always lower than the DC-link voltage.
The losses in this component are the sum of switching losses and conduction losses.The energy losses in each switching are given by Equation ( 5) [12]: where I c is the storage current and V ce is the DC-link voltage.Generally the switching power loss is: where L ON is the switching on loss, L OFF the switching off loss, L R is the diode recovery loss, and f C is the switching frequency.Replacing ( 5) in ( 6) one obtains: The three-switching-loss contributions can be given by the constructors both separately or jointly.
From the IGBT F300R12KE3 [20] data-sheet one obtains: For SiC technology, considering the MOSFET CAS300M12BM2 [21], for which the total switching losses are given, from its data-sheet one obtains: Thus, the switching power loss is given by the Equation ( 10): The conduction losses were simulated using an averaged switch model (ASM), which also evaluates the losses related to the current ripple; the model presented in [22] was used.
From the data-sheets of the above-mentioned components, one obtains the parameters shown in Table 5, which are necessary to calculate the conduction losses.

ASM Parameters
Si SiC Concerning inductance, DC link capacitance and switching frequency, according to the values available in [22,23], the following values were used: A model in the MATLAB Simulink Plecs environment was created to evaluate the converter losses, according to that explained above.
The converter efficiency, over the storage working conditions, is reported in Figure 8 for the Si technology and in Figure 9 for the SiC technology.
Energies 2017, 10, 920 9 of 15 The conduction losses were simulated using an averaged switch model (ASM), which also evaluates the losses related to the current ripple; the model presented in [22] was used.
From the data-sheets of the above-mentioned components, one obtains the parameters shown in Table 5, which are necessary to calculate the conduction losses.Concerning inductance, DC link capacitance and switching frequency, according to the values available in [22,23], the following values were used: A model in the MATLAB Simulink Plecs environment was created to evaluate the converter losses, according to that explained above.
The converter efficiency, over the storage working conditions, is reported in Figure 8 for the Si technology and in Figure 9 for the SiC technology.

Electric Powertrain Efficiency
In order to evaluate component efficiency and overall powertrain efficiency in different working conditions, a MATLAB Simulink model was created (Figure 10).

Electric Powertrain Efficiency
In order to evaluate component efficiency and overall powertrain efficiency in different working conditions, a MATLAB Simulink model was created (Figure 10).The efficiency (η) of the electric components necessary to realize a series architecture, obtained by simulations with the model sketched in Figure 10, are presented in Table 6.
It is interesting to notice that the traction machine inverter efficiency is comparable to the one obtained in [12] over the three different road missions.
Regarding DC-DC converter and SC storage, with the above mentioned control logic, the major quantity of generated energy is affected by both charging and discharging losses; for this reason both efficiencies are evaluated.
The electric machine efficiency, as well as its inverter, instead, is evaluated only during motor working condition and the regenerative braking is not considered.The efficiency in regenerative deceleration, in fact, would be the same as a parallel architecture, so it is not useful to show this parameter here, where the aim is to evaluate the additional electric losses introduced by a series architecture.
An important result was for the storage system not to reach the full charge state (also in the extra-urban mission, despite the long downhill road).For this reason, the use of SCs is suitable for this application, as well as the sizing and the control logics.
In Table 7, the overall electric powertrain efficiency is shown (obtained as a product of Table 6 efficiencies).The efficiency (η) of the electric components necessary to realize a series architecture, obtained by simulations with the model sketched in Figure 10, are presented in Table 6.
It is interesting to notice that the traction machine inverter efficiency is comparable to the one obtained in [12] over the three different road missions.
Regarding DC-DC converter and SC storage, with the above mentioned control logic, the major quantity of generated energy is affected by both charging and discharging losses; for this reason both efficiencies are evaluated.
The electric machine efficiency, as well as its inverter, instead, is evaluated only during motor working condition and the regenerative braking is not considered.The efficiency in regenerative deceleration, in fact, would be the same as a parallel architecture, so it is not useful to show this parameter here, where the aim is to evaluate the additional electric losses introduced by a series architecture.
An important result was for the storage system not to reach the full charge state (also in the extra-urban mission, despite the long downhill road).For this reason, the use of SCs is suitable for this application, as well as the sizing and the control logics.
In Table 7, the overall electric powertrain efficiency is shown (obtained as a product of Table 6 efficiencies).Considering the very high electric powertrain efficiency, especially using SiC components, the series architecture could be a valid alternative to the parallel.

Powertrain Fuel Economy
To compare the two structures, series and parallel, simulations of the entire series powertrain efficiency were carried out: the ICE efficiency, the differential gear efficiency and the backward energy deriving from regenerative slowdowns were evaluated.In series architecture, the ICE can operate at a maximum efficiency condition, therefore its efficiency is about 34%.
To evaluate the energy virtuosity of series architecture, a comparison with the fuel economy of a parallel architecture is of primary importance.
In Table 8, different configurations of parallel and series architectures are compared.A parallel structure like the one sketched in Figure 1 was used; in fact, despite power-split architecture it is the most wide spread on the market, the first is more energy saving [24].An 80% charge-discharge battery efficiency was considered for both architectures; this efficiency value can be considered both an optimistic evaluation for a NiMH battery and a real working condition efficiency for a Li-ion battery [25,26]; in any case battery efficiency has a low impact on the overall powertrain efficiency as far as parallel architecture is concerned [27].As for series architecture, a simple rule-based energy management system was used in parallel architecture.Battery recharging is almost guaranteed only with regenerative braking and electric traction is preferred at low speed; the engine turn-on threshold depends on SOC measurement.SCs and SiC component technology in parallel architecture would lead to low benefits.In fact, as shown in [27], parallel architecture is scarcely sensitive to battery efficiency variation.Moreover, in parallel architecture converters are used to interface the electric machine with the storage only during regenerative braking and electric traction.The only backward energy is affected by electric losses (storage and converter losses), whereas in series architecture all the generated energy is affected by these losses.

Table 2 .
Simulated road mission features.

Table 2 .
Simulated road mission features.

Table 5 .
Parameters to evaluate conduction losses.

Table 5 .
Parameters to evaluate conduction losses.

Table 7 .
Electric powertrain efficiency over different road missions.