A Battery-Free Sustainable Powertrain Solution for Hydrogen Fuel Cell City Transit Bus Application

Abstract

The paper presents a sustainable electric powertrain for a transit city bus featuring an electrochemical battery-free power unit consisting of a hydrogen fuel cell stack and a kinetic energy storage system based on high-speed flywheels. A rare-earth free high-efficiency motor technology is adopted to pursue a more sustainable vehicle architecture by limiting the use of critical raw materials. A suitable dynamic energetic model of the full vehicle powertrain has been developed to investigate the feasibility of the traction system and the related energy management control strategy. The model includes losses characterisation, as a function of the load, of the main components of the powertrain by using experimental tests and literature data. The performance of the proposed solution is evaluated by simulating a vehicle mission on an urban path in real traffic conditions. Considerations about the effectiveness of the traction system are discussed.

1. Introduction

The EU’s Energy Transition and Sustainable Development policies aimed at mitigating global warming have set the goal of zero greenhouse gas emissions by 2050. The resulting objective for the transport sector is a reduction of greenhouse gas (GHG) emissions by 33% (compared to 2005) by 2030 and by 90% (compared to 1990) [1,2].

One of the main actions in the transport sector of EU strategy aims to reduce CO₂ and other GHG emissions by replacing the conventional internal combustion engine (ICE) based vehicles with electric vehicles (EV) [3,4]. EVs represent a viable alternative to traditional ICE because of their low emissions, and efficient reduction of CO₂ also achieved by energy recovery during braking operations and zero local emission on the exhaust pipe [5].

Undoubtedly, the transition to electric mobility represents a fundamental step towards eliminating direct environmental emissions, while indirect emissions depend on the nature of the energy sources used to produce electricity. As for the UE28, forecasts are encouraging, being expected that the EU grid mix will come down from 300 g CO₂ eq/km in 2015 to 200 g CO₂ eq/km in 2030 and 80 g CO₂ eq/km in 2050 [6]. Therefore, EVs powered by renewable energies represent a sustainable solution since they allow for zeroing and/or minimising both direct and indirect emissions during the entire energy cycle.

EVs use different alternative vehicle powertrain technologies [7], including hybrid configurations (using a combination of onboard energy sources) and/or alternative fuels, both based on renewable or fossil sources.

EVs are highly dependent on available energy storage technologies, such as electrochemical battery cells, fuel cells (FC) and ultra capacitors for power [8]. Each of them is characterized by different values of energy density and power density that, depending on the type of application, must meet the performance requirements of the vehicle.

Currently, both hybrid electric vehicles (HEV) and battery electric vehicles (BEV) are based on the use of electrochemical batteries to store electric energy onboard. In the specific calculation of the vehicle GHG emissions, it is also necessary to include the environmental impact associated with the entire life cycle (LCA—“Life Cycle Assessment”) of electrochemical storage systems. Recent LCA studies carried out on different types of lithium-ion batteries used in electric vehicles show consistent GHG emission values ranging from 70–175 kg CO₂ eq per kWh of storable energy. Assuming proper recycling of materials or reuse of the same batteries for other applications, an average emission reduction of 20 kg CO₂ eq/kWh can be achieved [8].

The hydrogen-powered (by fuel cell) vehicles (HFCVs) are another technological option trying to overcome BEV limitations still having zero local emissions. As fuel cells are inadequate to follow highly transient road power demands, for automotive applications, they are usually hybridized with electrochemical batteries [9,10,11,12]. Over the last decade, more than 80 hybrid transit buses using hydrogen FC and batteries have been operated on about 8 million kilometres in several European cities, demonstrating the suitability of the technology [13].

As an alternative to electrochemical batteries, the energy can be stored as kinetic energy in flywheel energy storage systems (FESS), based on a flywheel connected to an electric motor–generator, and on-demand exchanged as electric energy with the other powertrain components. High-speed FESSs use a high-performance rotor material to maximize the specific energy density, magnetic bearings and vacuum vessel to decrease friction and windage losses. High-speed FESS is an emerging, environmentally friendly technology that, thanks partly to its use in Formula 1 cars, is rapidly reaching the level of maturity (TRL) needed for mass automotive practical applications.

Compared with electrochemical batteries, the advantages of FESS can be synthesized as follows: long lifetime, high power density, high efficiency (90%); typical storage capacities vary between 3 kWh and 133 kWh [14,15].

Examples of flywheels optimized for vehicular applications are developed with a specific power of 5.5 kW/kg and specific energy of 3.5 Wh/kg; other flywheel systems (3.15 kW/kg and 6.4 Wh/kg) can be compared to a state-of-the-art supercapacitor vehicular system [16].

The disadvantages of FESS are the lower energy density, the lower TRL and higher costs, although these are expected to be improved in the near future.

Beyond the GHG emissions, aspects related to critical raw materials need to be considered when large mass production of EVs is envisioned in the coming years, Figure 1 synthesizes the recognized EU critical raw materials needed in the production of the relevant key technologies [17]. Considering the EV production, lithium-ion batteries are based on critical materials such as lithium (low risk) and natural graphite and cobalt (medium risk) raising concerns in the supply chain. Even more attention is needed on rare earths (RE) since they are key materials in the production of permanent magnets (PM) for high-efficiency EV traction motors and windmills generators in radial and axial flux configurations [18,19]. The RE’s very high supply risk is affected by their price volatility associated with their localized geopolitical provenience; moreover, their sustainability is poor due to the high pollutant production process related to their extraction and refining [20]. Therefore, there is growing attention to alternative solutions that include RE free or reduced RE motor technologies [21,22], where relevant advances were made on synchronous reluctance motor [23,24,25] and induction motor [26,27] technologies.

With the aim to propose a sustainable battery-free powertrain architecture a novel hybrid power unit consisting of a hydrogen FC stack and a set of high-speed FESSs was already proposed by the authors for small-sized city buses [28,29,30,31,32,33]. A power control strategy was also proposed based on a partitioning of the FC stack in a number of FC units: each of those is on–off controlled in order to work around its best-efficiency operating point; given that the FC power is step variating, the mismatch between the required and the FC power is managed by the FESS bench [34,35].

In this paper, a more sustainable architecture is proposed, where battery-free powertrain architecture is considered along with the adoption of RE free motor technologies, in both the traction motor and the FESS motors.

Within this context, the paper gives a further contribution to the improvement of the system modelling by including a detailed representation of the electrical motors and converters’ efficiency, now depending on the actual load. To this aim, the efficiency map of each powertrain component has been adopted. Moreover, the impact of vehicle auxiliary loads on energy consumption has been analysed.

Hence, the validity of the proposed vehicle architecture is proven by simulation analysis based on the developed energetic model; moreover, the input data (vehicle speed and path grade) have been experimentally logged on a city bus operating in real traffic conditions over an urban bus line.

This paper is organized as follows: Section 2 illustrates the system architecture, the proposed hybrid power unit and the modelling approach. Then, in Section 3, the case study is presented including vehicle definition, drive cycle and path topography. Section 4 reports simulation results and discussion. Finally, conclusions are drawn in Section 5.

2. System Architecture and Modelling

In this section, the hybrid vehicle configuration is illustrated; a novel powertrain architecture and power strategy control are proposed; and models and methods for the sizing of the main powertrain components are proposed.

2.1. System Architecture

The conventional FC management solutions are mainly based on two approaches as follows:

• A dynamically controlled FC stack, where the stack power setting is regulated as a function of the vehicle demanded power, with the drawback that the stack needs to be sized at the peak vehicle power. Moreover, considering that FC efficiency strongly depends on its operating point, the FC works at low efficiency where the required power load is fluctuating.
• A static controlled FC stack, where the FC is kept at a constant operating point close to the maximum efficiency one. The transient power is managed by the energy storage system (ESS), which requires significant storage capability.

On the contrary, the proposed solution is based on the sectioning of FC stack into a series of units operating independently in an on–off mode always at the maximum efficiency operating point. In this case, the transient power is shared between the FC stack and the energy storage system (ESS).

This innovative approach makes it possible to create a hybrid power unit (FC + EES) able to combine high-efficiency FC operation and to reduce the ESS sizing. Therefore, large capability ESS such as lithium-ion batteries can be replaced by more environmentally-friendly and efficient FESS batteries.

The proposed powertrain architecture is detailed in Figure 2, it uses an electric drivetrain (ED) including the electric traction motor (EM) and its power electronics and gearing. The ED is fed by a hybrid power unit consisting of an FC stack and a bench of FESSs. The ED and the FESSs are connected to the DC power bus (continuous red line) by means of converters (CM and CFESS) to manage the power flows as required by the master control system (CS) via the communication bus (green line).

The control strategy of the hybrid power unit is based on the state of charge of the FESS. It may activate a number of stacks independently resulting in different available power levels for the FC power unit, all obtainable with its maximum efficiency performances. Due to the FC’s slow dynamic response, the master controller imposes a constant operating point for each FC stack so that the FESS handles the load variations with the aim of:

• providing power when the load power is higher than the FC power.
• recovering when the FC power is higher than the load power and during the regenerative electrical braking. The schematic of the hybrid powertrain and the reference directions of the power flows are shown in Figure 2.

The FC stack and FESS bench are conceived as modular in the proposed architecture, their sizing can be carried out on the base of the vehicle ED characteristics. In detail, the FC stack rated power $\left({\overline{P}}_{FC,s}\right)$ is evaluated considering the continuous power of the ED $({\overline{P}}_{ED,s}$), while the power of the FESS bench $(P_{FESS})$ is evaluated considering the peak power of the ED $({\widehat{P}}_{ED})$. About the energy sizing of the FESS bench, its minimum energy storage capability $(E_{FESS})$ is evaluated considering the kinetic energy of the vehicle $\left(E_k\right)$ at cruising speed. Moreover, to enhance the energy recovery in the braking condition the FESS bench energy can be increased accounting for the variation of the gravitational potential energy $\left(E_g\right)$ related to the vehicle driving on an envisioned urban path.

$${\overline{P}}_{FC,s}={\overline{P}}_{ED,s} P_{FESS}={\widehat{P}}_{ED}$$

(1)

$$E_k\le E_{FESS}\le E_g$$

(2)

Concerning the FC stack partitioning, different approaches can be used; the number of the modules inside the FC stack is selected considering the power control dynamic response (including FC rise time, vehicle dynamics, FESS sizing) and off-the-shelf available FC power ratings. The most suitable number of modules is evaluated by the analysis of the traction power profile.

Regarding the composition of the FESS bench, the off-the-shelf devices have been considered to achieve the conceived power and energy needs.

The control strategy of the hybrid power unit is based on the state of charge (SOC) of the FESS. The number of active FC modules depends on the difference between the SOC set point and the actual SOC value. The schematic of the hybrid powertrain and the reference directions of the power flows are shown in Figure 3.

The master control system (CS) demands the FC controller (CFC) to power on a suitable number of FC units to produce the required power. The CFC enables each FC unit by imposing a constant output power setting to maximise the FC efficiency. The FESS controller (CFESS) monitors the voltage of the DC bus; when the voltage increases because of the excessive power incoming into the DC bus (due to FC or regenerative braking) the CFESS requires a flywheel acceleration to store kinetic energy. Otherwise, when there is a lack of power in the DC bus, the voltage decreases and the CFESS decelerates the flywheel to inject fuel into the DC bus.

2.2. Model-Based Approach

The powertrain is described by using a model-based approach. A parametric dynamic simulator has been developed in the MATLAB-Simulink environment, consisting of interconnected analytical-numerical sub-models, to simulate the travel of a given vehicle along a path and to evaluate the performance of each component. The submodules interconnection is illustrated in the block diagram in Figure 4; due to the presence of restricted data and IP content, the authors prefer not to add the MATLAB-Simulink model to the paper.

The implemented software calculates the power flows and the energy consumption starting from the definition of the inputs concerning the following three blocks:

• “Vehicle”, addressing the vehicle (mass, dimensions, mechanics, efficiency, payload, etc.);
• “Road”, concerning the topography of the path and the road surface characteristics;
• “Drive Cycle”, describing the demanded driving profile by specifying the velocity vs. time;
• The “Traction power” block computes the mechanical power needed to drive over the input path with the selected vehicle at the specified speed profile.

The “Electric Drivetrain” block computes the electrical power needed for the traction, accounting for the energy losses of the gear, motor and power converter. The computation of the drivetrain losses is based on look-up tables validated by means of the vehicle’s characteristics, experimental data and literature. The torque-speed working point of the electric drivetrain is the input of the look-up table; power losses are the output.

The vehicle simulation software developed by the authors was used for the investigations presented in this paper. The simulator solves the following vehicle equations of motion (3)–(8) through numeric integration:

$$F\left(v\left(t\right)\right)=m\cdot \left(1+k\right)\cdot \frac{dv}{dt}$$

(3)

$$\mathrm{F}\left(v\left(t\right)\right)=T\left(v\left(t\right)\right)-\left[mgsen\beta +C{v}^2\left(t\right)+Bv\left(t\right)+A\right]$$

(4)

From the above equations it yields:

$$m\cdot \left(1+k\right)\cdot \frac{dv}{dt}=T\left(v\left(t\right)\right)-\left[mgsen\left(\beta \right)+C{v}^2\left(t\right)+Bv\left(t\right)+A\right]$$

(5)

where v is the vehicle speed, $\frac{dv}{dt}$ is the acceleration, A, B and C are the constant terms of resistance in the Davis equation [36], g is the gravitational acceleration; m is vehicle mass; t is the time; T is traction force; $\beta$ is the angle of the road slope; and k takes into account the rotational mass.

Considering regenerative electrical braking, the electrical traction power P_u(t) is calculated as:

$$\{\begin{array}{l}{P}_u\left(t\right)=\frac{T\left(v\left(t\right)\right)v\left(t\right)}{\left({\eta }_t{\eta }_D\right)} \mathrm{if} T\left(v\left(t\right)\right)v\left(t\right)>0\\ P_u\left(t\right)=T\left(v\left(t\right)\right)v\left(t\right){\eta }_t{\eta }_D \mathrm{if} T\left(v\left(t\right)\right)v\left(t\right)<0 \end{array}$$

(6)

where η_t is the transmission efficiency and η_D is the efficiency of the electrical traction drive, i.e., the converter and the motor.

The energy recovered during braking is stored according to the SOC of the FESS bench, in the ways and within the limits explained in the previous paragraph.

In addition, the use of on-board auxiliary devices are considered, whose operation requires energy $E_{aux}\left(t\right)$, which is evaluated by considering the absorbed power $P_{aux}\left(t\right)$.

With reference to the FC and FESS components of the power unit, the following simplified models were used.

As for the FC, a proton exchange membrane (PEM) type was selected for this application. The FC operation was equated with that of a voltage generator modelled by means of its static voltage-current characteristic, thus neglecting its dynamic behaviour and any delay.

The vehicle consumption for traction, in terms of hydrogen mass, is calculated as follows:

$$m_{H2}=\frac{E_e}{{\eta }_{FC} H_i}$$

(7)

where E_e is the FC electrical energy output during a vehicle mission, ${\eta }_{FC}$ is the FC efficiency and H_i is the lower heating hydrogen value (119.9 MJ/kg). Since the FC operates at constant power, a constant efficiency (${\eta }_{FC}$) is assumed in the model.

The FESS model is based on the usable kinetic energy in a flywheel in the speed range over which it is allowed to operate:

$$E_{K,u}=\frac{{\omega }_{max-}^2{\omega }_{min}^2}{2}{\displaystyle \int }{r}^{2}dm=\frac{1}{2}I\Delta {\omega }^2$$

(8)

where I is the moment of inertia about the axis of rotation, ω_max and ω_min are the maximum and minimum values of the rotor angular speed, respectively, and dm is a small mass element at a distance r from the axis of rotation.

3. Application Case

With the aim of validating the proposed more sustainable vehicle architecture, a real urban path covered by the public transportation system was identified. An urban transit bus with a 20-passengers capacity at full load was considered to run over the selected path.

3.1. Path Characteristics

The actual public transport line #143 of the Rieti city (Italy) transport network was considered as a case study. As illustrated in Figure 5, the line has partially different paths on the outward and return journeys.

The mission data on the line route have been recorded in real traffic conditions by means of a suitable georeferencing system based on high precision GPS installed onboard the bus. In detail, vehicle position, speed, acceleration and altitude have been logged at a fixed sample time (0.2 s).

The position, speed and acceleration vs. mission time are shown in Figure 6, Figure 7 and Figure 8, respectively; while Figure 9 illustrates the path grade vs. mission time.

Table 1. Round trip main data.

Parameter	Unit	Value
Total cycle length	km	26
Total cycle time	s	3300
Maximum grade	%	−4.8; +6.8
Maximum acceleration	m/s2	+1.21; −1.40
Average speed	km/h	28.36
Maximum speed	km/h	64

reports the key characteristics of the vehicle mission, requiring a total time of 55 min, dead times included, to travel over a 26 km distance. A maximum speed of 64 km/h is reached during the mission with peaks of positive and negative acceleration of 1.21 m/s² and 1.4 m/s², respectively. Due to the partly different outward and return paths, the logged positive and negative peaks of the grade are different, +6.8 and −4.8, respectively.

3.2. Vehicle Definition

A transit bus topology, with a 20-passengers carrying capacity at full load was considered to run over the selected line illustrated above; the main characteristics of the baseline vehicle are listed in

Table 2. Main vehicle data.

Parameter	Symbol	Unit	Value
Length		mm	7168
Width		mm	2052
Height		mm	2900
Passengers sits			20
Full load vehicle mass	m	kg	5600
Time from 0 to 100 km/h	ta	s	15
Frontal section		m2	5.5
Aerodynamic coefficient	cx		0.316
Tires			195/75 R 16 110 R

.

The main auxiliary loads (power steering, air conditioning and alternator) have also been considered in the energetic model; given that the corresponding powers change with the time, a duty cycle has been used for each of them as shown in

Table 3. Auxiliary loads installed power and envisioned duty cycles.

Auxiliary	Power	Duty Cycle
Power steering	2 kW	7 s @ 70%
		14 s @ 30%
Air conditioning	10 kW	21 s @ 100%
		21 s @ 20%
Alternator	2.5 kW	10 s @ 30%
		10 s @ 50%
		20 s @ 80%

.

The characteristics of the conceived powertrain are shown in

Table 4. Electric powertrain main data.

Parameter	Unit	Value
Motor type		RE Free Synchronous Reluctance motor
Motor speed	rpm	14,000
Drive type	m	SiC Mosfet based three-phase inverter
Drivetrain peak power	m	200 kW
Drivetrain continuous power	m	70 kW
Electrical drive peak efficiency	%	94
Transmission ratio		25
Transmission efficiency (average)	%	93
FC power (at max efficiency)	kW	10
FC max efficiency	%	55
Number of FC in the stack		5
FESS max speed	rpm	40,000
FESS bench power	kW	200
FESS bench energy	kWh	1.5
Number of FESS in the bench		4
FESS electrical drive peak efficiency	%	94

. Concerning the electric drivetrain, an RE-free solution has been chosen; it is based on high-speed synchronous reluctance motor technology featuring high torque, significant power density and a peak efficiency of 96% [16]. The peak power of the motor (200 kW) is provided by means of the FESS bench, while the continuous power (50 kW) is provided by the FC.

According to the motor power (peak and continuous), the FC stack power (50 kW) is obtained by using five modules of 10 kW each, while the bench FESS consists of four units of 50 kW power, 375 Wh net energy each.

4. Simulation Results

The power flows of each single block of the whole system are evaluated by using the developed dynamic model to simulate the vehicle mission in the application case.

The motion resistance (Figure 10) and the vehicle acceleration give the total traction effort to compute the mechanical power at the motor shaft that is provided by the electric drivetrain and then the required electric power (Figure 11).

In detail, the electric drivetrain input power is computed from the mechanical power by considering its efficiency map reported in Figure 12, where the actual torque-speed operating points of the traction motor are superimposed to evaluate the effectiveness of the electric drivetrain sizing. The resulting drivetrain losses are reported in Figure 13.

To highlight the relevance of adopting detailed efficiency maps instead of constant efficiency of the traction drivetrain, the energy consumption in both cases is reported in Figure 14. Up to 20% more energy consumption is evaluated for the vehicle mission when a constant efficiency of the electric drivetrain of 85% is considered.

The electrical power is provided by the FC and FESS, managed by the control strategy based on the FESS SOC reference value; the behaviour of the FC in terms of output power and of the FESS SOC over the vehicle mission is reported in Figure 15. It can be noticed that the output power of the FC is strictly related to the error between the SOC and the SOC reference. A turn-on delay at power-on has been considered in the FC, moreover, a last-on last-off policy is adopted to mitigate very short power requests on a single unit.

Although the FC units are on–off controlled, the total electrical power is managed with the FESS contribution as shown in Figure 16. The presence of power peaks is related to the acceleration profile recorded in real driving conditions. While the vehicle is not running, the auxiliary loads are still powered by the FC and FESS or only by the FESS, depending on the SOC status.

Figure 17 shows the most significant energy values during the mission cycle. Given that the FESS SOC value at the end of the cycle is equal to the initial one, such a value can be neglected, and the following considerations can be drawn from an energy balance point of view. The electric recovery braking energy is almost 10 kWh, but this is an optimistic value because in some cases the mechanical brake, which is not considered in the model, must intervene. Moreover, the useful energy recovered is about 8 kWh because of the FESS losses (almost 2 kWh); on the other hand, no energy recovered could be stored without the FESS system. Without the storage system and the recovery braking, the energy generated by the FC would arise from 19.2 to 27.2 kWh and then 29.4% of energy saving has been obtained.

Finally, the hydrogen consumption computed by means of Equation (7), is plotted in Figure 18.

5. Conclusions

The paper proposes a more sustainable battery-free powertrain architecture for electric transit city bus. The two main features of the proposed powertrain are the adoption of a flywheel energy storage systems instead of chemical batteries and the adoption of rare earth free electric motors in both the electric drivetrain and the FESS bench.

To evaluate the effectiveness of the proposed architecture from the energetic point of view, a dedicated efficiency model of the RE free motor-drives has been adopted and a proper control strategy of the fuel cell stack and the flywheel bench has been developed.

A set of electric auxiliary loads has been considered to complete the energetic representation of the system.

With the aim of validating the proposed more sustainable vehicle architecture, data of a vehicle travelling in a real urban path covered by the public transportation system have been recorded as a reference.

By using the actual vehicle travelling data, the proposed vehicle architecture has been simulated to evaluate the energy flow among the powertrain single systems revealing the effectiveness of the adopted architecture. The main results demonstrate significant braking energy recovery capabilities of the FESS bench, while the control of the FC seems suitable to respond to the output energy variations occurring during the driving.

The effect of the auxiliary loads is relevant in terms of energy consumption; nevertheless, their energy absorption during braking operation or driving downhill helps to keep the FESS SOC within its boundaries.

Despite the adopted RE free motors being still less efficient than RE based ones, they seem to be adequate for the target vehicle application from an energetic point of view.

Further developments in the model are related to the refinement of the modelling of the fuel cells and to the powertrain sizing. Moreover, the impact of the traffic conditions on the vehicle efficiency estimation needs to be investigated.