Development of energy management system based on a power sharing strategy for a fuel cell-battery-supercapacitor hybrid tramway

Accepted Manuscript Development of Energy Management System based on a Power Sharing Strategy for a Fuel Cell-Battery-Supercapacitor Hybrid Tramway Qi Li, Weirong Chen, Zhixiang Liu, Ming Li, Lei Ma PII:

S0378-7753(14)02070-9

DOI:

10.1016/j.jpowsour.2014.12.042

Reference:

POWER 20321

To appear in:

Journal of Power Sources

Received Date: 7 October 2014 Revised Date:

26 November 2014

Accepted Date: 12 December 2014

Please cite this article as: Q. Li, W. Chen, Z. Liu, M. Li, L. Ma, Development of Energy Management System based on a Power Sharing Strategy for a Fuel Cell-Battery-Supercapacitor Hybrid Tramway, Journal of Power Sources (2015), doi: 10.1016/j.jpowsour.2014.12.042. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Development of Energy Management System based on a Power Sharing Strategy for a Fuel

RI PT

Cell-Battery-Supercapacitor Hybrid Tramway Qi Li*a, Weirong Chena, Zhixiang Liua, Ming Lib, Lei Maa

a. School of Electrical Engineering, Southwest Jiaotong University, Chengdu 610031, Sichuan Province, China b. Tangshan Railway Vehicle Co. Ltd, Tangshan 063000, Hebei Province, China

SC

*Corresponding author. Address: School of Electrical Engineering, Southwest Jiaotong University, Chengdu 610031, Sichuan Province, China. Tel.: +862887603332; fax: +862887605114. E-mail address: [email protected] (Qi Li)

Abstract: A hybrid powertrain configuration based on a proton exchange membrane (PEMFC), a battery and a

M AN U

supercapacitor (SC) is designed without grid connection for the LF-LRV tramway. In order to avoid rapid changes of power demand and achieve high efficiency without degrading the mechanism performance, a power sharing strategy based on a combination of fuzzy logic control (FLC) and Haar wavelet transform (Haar-WT) is proposed for an energy management system of the hybrid tramway. The results demonstrate that the proposed energy management system is able to ensure the major positive portion of the low frequency components of power demand can be deals with the

TE D

PEMFC. The battery can help provide a portion of the positive low frequency components of power demand to reduce the PEMFC burden while the SC bank can supply all the high frequency components which could damage the PEMFC membrane. Therefore, the energy management system of high-power hybrid tramway is able to guarantee a safe operating condition with transient free for the PEMFC and extend the lifetime of each power source. Finally, the

EP

comparisons with other control strategies verify that the proposed energy management system can achieve better energy efficiency of the overall hybrid tramway.

AC C

Keywords: proton exchange membrane fuel cell, hybrid tramway, energy management system, power sharing strategy, fuzzy logic control, Haar wavelet transform.

1. Introduction

As a promising technology fuel cells that convert chemical energy of the fuel into electricity without combustion are expected to become a viable solution for transportation applications. They could be used as predominantly renewable energy supplies instead of imported oil to help meet one of our most pressing energy needs. Although there are various fuel cell technologies available for use in vehicular systems, a proton exchange membrane fuel cell (PEMFC) has been found to be a prime candidate, since the PEMFC has lower operating temperatures and higher power density while compared to the other types of fuel cells 1

ACCEPTED MANUSCRIPT [1-3]. As energy converters, the PEMFC is also more efficient and environmental than internal combustion engine (ICE) in transportation applications [4-6]. In order to encourage the development of hydrogen economy and reduce dependence on fossil fuels, large research efforts have been underway to develop the locomotives and the tramways powered by the PEMFC in recent years [6-15]. Compared

RI PT

with the catenary-electric and diesel-electric types, the locomotives and the tramways powered by the PEMFC have many advantages which show great extensive application potential [11-17]. However, a stand-alone PEMFC integrated into a powertrain is not always sufficient to satisfy the load demands of the high-power propulsion system [18,19]. Although the PEMFC exhibits good power capability during steady-state operation, the slow response during transient peak power demands

SC

has restrained the PEMFC from being used in large-scale and high-power transportation applications, such as the locomotives and the tramways. Due to the unidirectional power flow characteristics of the PEMFC, the energy from regenerative braking of a

M AN U

vehicle cannot be handled by the PEMFC, thus the system efficiency enhancement is not possible through the PEMFC alone. Furthermore, the lifetime of the PEMFC may be dramatically impacted by the rapid power demand variations [20]. Nevertheless, the effect of these drawbacks can be reduced by hybridizing the PEMFC with an energy store system (ESS), such as battery, supercapacitor (SC), or a combination of both, in order to meet the total power demand of a hybrid system. In general, the battery has higher specific energy than the supercapacitor and thus can provide extra power for a longer period of time. The supercapacitor which has a higher specific power than the battery is more efficient, and has a longer lifetime in terms

TE D

of number of charge/discharge cycles. Currently, a hybrid propulsion system based on the PEMFC and battery is the option chosen by the few projects which have used to operate high-power vehicles, because it is the most economical choice for the powertrain [5,11,14]. In addition, the incorporation of supercapacitor as secondary ESS in a hybrid electric vehicle with the

EP

PEMFC and the battery improves the overall vehicle dynamic response [4,12,18]. In order to fulfill the power balance between the load power and the power sources, the energy flows are controlled by the

AC C

energy management system, which determines at each sampling time the power generation split between the PEMFC and the energy storage system (ESS), such as battery and supercapacitor. Hence, the energy management strategies of hybrid propulsion system that decide the power assignment are an important technique. In recent years, the development of energy management strategies for hybrid propulsion system has become a topic of interest for researchers. Kim et al. [20] have proposed a fuzzy controller to optimally distribute the relative power between the fuel cell and the battery. Liangfei Xu et al. [5] have proposed an adaptive supervisory control strategy for a fuel cell/battery-powered city bus which consists of two fuel cell stacks with a rated power of 40 kW and a Ni-MH battery to fulfill the complex road conditions in Beijing bus routes. Erdinc et al. [6,7] have designed a FC/UC hybrid vehicular power system by using a wavelet based load sharing and fuzzy logic based control algorithm, and also have presented a wavelet and fuzzy logic 2

ACCEPTED MANUSCRIPT based energy management strategy for a fuel cell/battery/ultra-capacitor hybrid vehicular power system. Vural et al. [8] have developed a test bench of FC/UC hybrid configuration that can emulate the dynamics of vehicular systems to verify wavelet transform and fuzzy logic based energy management strategy. Eren et al. [9] have proposed a fuzzy logic supervisory controller based power management strategy that secures the power balance in hybrid structure, enhances the FC performance and

RI PT

minimizes the power losses for an FC/UC hybrid vehicular power system. Thounthong et al. [10] have used an innovative control law based on flatness properties for fuel cell/supercapacitor hybrid power source. Jia et al. [4] have described the electrical characteristic of a hybrid power supply system combining PEMFC and SC, and have investigated on the platform of an electric bicycle to effectively improve the system efficiency and prolong the cruise mileage of the vehicle. Pablo García et al.

SC

[12] have evaluated the option of using a new powertrain based on fuel cell, battery, and SC for the tramway, and the energy management system used for controlling the components of the new hybrid system has allowed optimizing the fuel consumption

M AN U

by applying an equivalent consumption minimization strategy. Torreglosa et al. [13] have presented an equivalent consumption minimization strategy to minimize the hydrogen consumption for a real tramway powered by a hybrid system based on fuel cell and battery. Pablo García et al. [14] have achieved a comparative study performed in order to select the most suitable control strategy for electric vehicles powered by FC, battery and SC, and five different control strategies have been adopted for this kind of hybrid vehicles. Luis M. Fernandez et al. [15] have designed a hybrid system which consists of the fuel cell and Ni-MH battery integrating a single DC/DC converter to provide the power demand by the tramway loads and a state machine control has

TE D

been used to provide power demand by the driving cycle. Qi Li et al. [18] have presented an energy management strategy based on the fuzzy logic control method for a fuel cell/ultra-capacitor/battery combined electric vehicle by electric vehicle simulation software ADVISOR. Liangfei Xu et al. [20] have developed an optimal real-time energy management strategy based on the

EP

pontryagin’s minimal principle targeting at minimizing operation cost for a plug in fuel cell city bus. Due to different power levels present various output dynamic characteristics, high-power PEMFC model based on dynamic behaviors should be

AC C

proposed by the experiment testing. However, most of the proposed hybrid propulsion system models focus on the modeling of static, simplified, low-power PEMFC stack for vehicular systems, and no published papers have developed the dynamic and high-power PEMFC control system model based on the commercial PEMFC power unit for the large-scale and high-power transportation applications, such as the locomotives and the tramways. These energy management strategies mentioned above have been proven effective in dealing with power distribution by a fuzzy logic control (FLC) [18,21]. The fuzzy logic based methodology offers a remarkably simple way to draw definite conclusions from vague, ambiguous or imprecise information, which allows modeling of complex systems using a higher level of abstraction originating from our knowledge and experience. The FLC provides a quite suitable structure compared to conventional control methods especially for the systems composed of nonlinear behaviors where the whole exact mathematical 3

ACCEPTED MANUSCRIPT model is difficult to obtain. Thus, due to the nonlinear of multivariable hybrid vehicular system, the FLC strategy which can manage power sharing between various components and to guarantee the power sources performance is more suitable for the energy management. However, in above studies the power demand of a vehicular system consists of many sharp and instantaneous changes, the proposed FLC strategies have not adequately considered the balance between energy efficiency and

RI PT

dynamic property for high-power transport and the lifetime of the PEMFC and ESS due to the dramatically negative impact of the frequent and rapid power demand fluctuations in real driving cycles. The lifetime and efficiency of the PEMFC and ESS are critical factors for high-power transport economy [4-6,12,20]. Therefore, in order to avoid the rapid variation and large differential power demand for the PEMFC, distribute the power demands to power sources appropriately according to their

SC

natural characteristics and achieve high efficiency without degrading the mechanism performance, an appropriate power sharing strategy which should isolate the base power demand of a given signal from its transients (low and high frequency power

M AN U

demand signals) is quite essential for the energy management of a high-power transport. For the development of a power sharing strategy, a wavelet transform (WT) which can be embedded into the structure of FLC strategy is utilized since the WT has shown excellent performance in analyzing the transients.

The WT is a new signal processing approach that has proven its usefulness in analysis of various types of signals and has recently been applied to variety of applications [7,22-25]. In the WT, an original signal can be decomposed into localized contributions characterized by a scalable modulated time window of varying size [22]. Each contribution represents a portion of

TE D

the signal with a different frequency. By employing the long windows at low frequencies and short windows at high frequencies, the WT is capable of comprehending the time and frequency information simultaneously [24]. Thus, different subcomponents with desired frequencies can be acquired using a filtering implementation. The WT is capable of providing a filter to extract

EP

characteristics of transient signals and sharp changes, and the loss of important edge information is minimized compared to conventional filtering techniques [25]. Among different kinds of wavelets, Haar wavelet has the shortest filter length in the time

AC C

domain. Besides, Haar wavelet is the simplest and most popular type of WT [23]. The multi-level Haar-WT can be employed as a very effective tool for separating the high frequency and low frequency components of the load profile. Thus, the attractive features of multi-level Haar-WT in analyzing instantaneous variations make it possible to accurately capture and localize transient features in driving cycle profile. Although the similar method of combination of FLC and WT has been utilized for low-power hybrid vehicular systems [6,7], the study of energy management strategies based on dynamic and high-power PEMFC system have not been carried out for large-scale and high-power transportation applications. Hence, in order to rationally assign the demand power to the PEMFC, battery and SC, a power sharing algorithm is proposed based on the Haar-WT algorithm and the FLC approach. In the fuzzy inference system, two decentralized FLC (2-FLC) are designed according to the power sharing algorithm and the control 4

ACCEPTED MANUSCRIPT objective would be achieved by improving the efficiencies of high-power transportation and PEMFC system on condition that dynamic property of high-power transportation is satisfied. The combination of 2-FLC and WT in this work ensures increment in both lifetime and energy efficiency. Wavelet guarantees a safe operating condition with transient free for the PEMFC and extends the lifetime of each power source, and the 2-FLC targets to achieve better energy efficiency as a master of the overall

RI PT

high-power hybrid system power flow. Besides, the regulation of state of charge (SOC) of ESS, such battery and SC bank, is common objective of both methods.

LF-LRV developed by the Chinese manufacturer of Tangshan Railway Vehicle Co. Ltd. is a tramway, which a line links “Gar” station and “University” station, operating in Samsun, Turkey. It consists of 21 stops, separated approximately 779 m and

SC

this line is 15.58 km long. The actual tramway presents a capacity of 283 passengers, and reaches a maximum speed of 77 km/h. The tramway is composed of two motor units and one trailer unit. The two motor units are supplied by the tramway traction

M AN U

system via an inverter box. It operates as a catenary-powered tramway, which uses an overhead line and pantograph to transmit electrical energy to the tramway. Currently, the hybrid LF-LRV tramway without grid connection is being developed by Tangshan Railway Vehicle Co. Ltd and Southwest Jiaotong University as shown in Fig. 1.

Fig. 1. The developing hybrid LF-LRV tramway without grid connection

TE D

In this paper, a hybrid propulsion system based on the PEMFC, the battery and the SC is developed for LF-LRV. The PEMFC acts as main energy source of tramway, and the Li-ion battery and the SC bank are utilized as energy store system (ESS) to supplement the PEMFC output power the during tramway acceleration and cruise and are also used for energy recovery during

EP

braking. Specially, the SC bank is able to consume the peak power that neither the PEMFC nor the battery can store because of its high dynamic response and high specific power. Furthermore, a power sharing strategy based on the 2-FLC and the Haar-WT

AC C

is proposed for the energy management system of the hybrid tramway to avoid the transients and rapid changes of power demand, achieve high efficiency and distribute the power demand to each power source appropriately. According to the hybrid system model of tramway which is developed with commercially available devices, the energy management system based on the proposed power sharing strategy is evaluated according to the real drive cycle of the tramway and the comparisons with other control strategies are also carried to verify the rationality and validity in term of energy efficiency of the overall hybrid tramway.

2. Energy management system for hybrid LF-LRV tramway The proposed configuration of PEMFC-battery-SC powered hybrid system for the tramway is shown in Fig. 2. The hybrid system is composed of the PEMFC, the battery, the SC, the unidirectional DC/DC converter, the bidirectional DC/DC converters, the energy management system (EMS), the traction motors, auxiliary service module and braking resistor. 5

ACCEPTED MANUSCRIPT The PEMFC is the primary energy source of tramway. It is connected to the boost-type unidirectional DC/DC converter which raises the low DC voltage delivered by the PEMFC to 800V DC traction bus. On the other hand, the rechargeable Li-ion battery and SC are utilized as ESS to supplement the output power of the PEMFC during tramway acceleration and cruise and are used for energy recovery during braking. Specially, the SC is able to consume the peak power that neither the PEMFC nor

RI PT

the battery can store because of its high dynamic response and high specific power. The battery and SC are connected to the traction DC bus through the bidirectional DC/DC converters respectively, which allow the charge and discharge of the battery and SC. The tramway loads supplied from the ESS are the auxiliary services and the traction system. The auxiliary service module represents the power consumption due to the tramway auxiliary equipment, such as lighting, fans, air conditioning

SC

systems, etc.

In order to distribute the demand power, the EMS determines the reference power signals for the PEMFC, the battery and the

M AN U

SC through regulating the unidirectional and bidirectional DC/DC converters. In addition, it also determines the reference signal for energy dissipation via the braking resistor if required during regenerative braking.

Fig. 2. Configuration of PEMFC-battery-SC powered hybrid system for the tramway

2.1 Modeling of 150 kW PEMFC Power Unit

TE D

The heart of the hybrid LF-LRV tramway is the PEMFC power unit which consists of the power module subsystem, the hydrogen storage subsystem and a control subsystem. The power module includes a 150kW PEMFC stack module, an air delivery module, and a cooling module [11,26]. The schematic diagram of the PEMFC power module is shown in Fig. 3. The working process of PEMFC stack is mixed flow transportation, heat conduction and electrochemical dynamic reaction. The

EP

PEMFC consists of a membrane-electrode assembly (MEA) where a solid polymeric membrane, acting as an electrolyte, is pressed between two electrodes (anode and cathode). At the anode, hydrogen flows in the channels and diffuses through the

AC C

porous electrode into the three phase interface. Hydrogen ions are generated and carried by the electrolyte membrane towards the cathode interface. Liberated electrons are carried through the load towards the cathode and provide electric power along the way. On the other hand, oxygen at the cathode interface combines with protons and electrons to produce water molecules [27-29].

Fig. 3. Schematic of the 150kW PEMFC power module

The PEMFC power unit is setup in the clean energy laboratory of Southwest Jiaotong University as shown in Fig. 3. In the PEMFC power unit, the PEMFC stack module is a Ballard Stack Modules-FCvelocity™ HD6. This stack module is rated at 150kW gross power, which includes the auxiliary components for air humidification, water recovery, hydrogen recirculation and purge. Hydrogen is supplied to the stack module and is pressure-regulated and re-circulated inside the stack module. The turbo 6

ACCEPTED MANUSCRIPT charger manufactured by ROTREX ™ Corporation is used as the air compressor. Cooling for the power module is achieved with two separate cooling loops. The primary cooling loop provides heat rejection for the HD6 Module. Additionally, it maintains de-ionization of the de-ionized water coolant through the utilization of a mixed bed ion-exchange resin. The second cooling subsystem provides heat rejection for the PEMFC stack module condenser to ensure enough process water available at

RI PT

all times for air humidification. In addition, the operating states of all the subsystems must be monitored and coordinated by a central control system namely XBO that provides CAN interface to communicate with a grid-connected system controller of the cascaded power electronics [11,26].

In this paper, a model of PEMFC power unit is proposed based on the Ballard 150 kW FCvelocity™ HD6 Module testing

SC

data. In order to focus on the dynamic response of the HD6 Module requested current and output voltage analysis, a set of assumptions, commonly undertaken in similar studies, is considered here [30-34]: All gases obey the ideal gas law and are equably distributed.

2)

The temperatures of hydrogen inside the anode and oxygen inside the cathode are equal to the cell temperature.

3)

The total pressure inside the cell is uniform.

4)

The reactants are saturated with vapors.

5)

The concentration losses are neglected under practical working condition.

M AN U

1)

The PEMFC potential is decreased from its equilibrium potential because of irreversible losses. The PEMFC equivalent

Vfc = Eoc − Vohmic − Vact

TE D

circuit model is shown in Fig. 4 and the output voltage equation is given by Eq. (1), (1)

overvoltage.

EP

where, Vfc is the stack output voltage, Eoc is the open circuit voltage, Vohmic is the ohmic overvoltage, and Vact is the activation

AC C

Fig. 4. PEMFC equivalent circuit model

Considering the effect of temperature and gas pressure, the open circuit potential of PEMFC is expressed as −44.43 RT   Eoc = Kc  E 0 + (T − 298) + ln PH2 PO1/2  2 zF zF  

(

)

(2)

where, Kc rated voltage constant, T is the Operating temperature, E0 is the electromotive forces under standard pressure, z is transfer electron number, F is Faraday constant, R is the gas constant, PH2 and PO2 are the gas pressures. The activation overvoltage and the ohmic overvoltage can be represented as

7

ACCEPTED MANUSCRIPT   i fc  1 ⋅ NAln   Vact = τ s +1   i0   V ohmic = Rinternal ⋅ i fc 

(3)

where, Rinternal is inner resistance of a stack, ifc is the cell output current, τ is the dynamic response time constant, N is the number of cells. The Tafel slope A and the exchange current i0 can be expressed as follows

(

)

RI PT

 zFk PH2 + PO2 −∆G i0 = e RT  Rh  RT  A=  zα F

(4)

SC

where, k is Boltzmann's constant, h is Planck's constant, ∆G is the Gibbs energy. According to nominal operating point [Inom,Vnom] and maximum operating point [Iend, Vend] which are obtained from the HD6 Module Manual, i0, Anom, and Rinternal can

(5)

TE D

 U (1) − Eoc + Rinternal  i0 = e N ⋅ Anom  U − U end  U (1) − U nom + (1 − I nom ) ⋅ nom  I nom − I end  Anom =  ln ( I end I nom )   N  ln ( I nom ) − (1 − I nom ) ⋅   I nom − I end     N ⋅ Anom ⋅ ln ( I end I nom ) − (U nom − U end )  Rinternal = I nom − I end 

M AN U

be represented as

According to the above equations, a model of 150 kW PEMFC power unit is developed based on the HD6 Module and a schematic of the power unit simulation is shown in Fig. 5. The performance parameters of 150kW HD6 Module are shown in

EP

Table. 1 and the relations of testing data of air mass flow, air input pressure and output power from PEMFC power unit are also presented in Fig. 6. In order to prove the validity of the PEMFC power unit model, the comparisons between the experimental data and the characteristics curve of polarization by simulation are carried out as shown in Fig. 7. Although there are some

AC C

smaller deviations (the average error is 5.74e-3), the characteristics of polarization and the experimental data are in good match. Furthermore, a dynamic step simulation results compared with the experiments are shown in Fig. 8. As the step of the requested current I_REQ is performed from 20A to 240A, the voltage and current of PEMFC power unit changed correspondingly have a delay time to reach the steady state. The delay time caused by air supply subsystem must be considered in the model. Hence, in order to describe the delay, a transfer function obtained by the identification method is expressed as GVI ( s ) =

V fc ( s )

I _ REQ ( s )

=

3.628

( s + 4.72 )

(6)

2

Fig. 5. Schematic of 150 kW PEMFC power unit simulation

8

ACCEPTED MANUSCRIPT

Table. 1 Performance parameters of 150kW HD6 Module

Fig. 6. Testing data of air mass flow, air input pressure and output power from PEMFC power unit

RI PT

Fig. 7. Comparisons between the experimental data and the polarization curve by simulation

Fig. 8. (a) Dynamic step testing data and simulation results of PEMFC power unit current. (b) Dynamic step testing data and simulation results of PEMFC power unit voltage

SC

2.2 Modeling of energy store system

The battery utility as an energy storage device has been proved in several applications [5,12-15,18]. Li-ion battery is an

M AN U

attractive proposition for use in high-performance electric vehicles [12,13,18]. In comparison with other rechargeable battery, Li-ion battery provides very high specific energy and a large number of charge/discharge cycles and can be charged very rapidly which makes it ideal for use in high-power applications, such as hybrid tramways, which can be recharged during frequent stops. In this work, a Li-ion battery pack is utilized both for supplying a portion of the base load together with PEMFC and capturing the braking energy together with the SC. The behavior of this battery is represented by the model presented in [12,13,35,36],

TE D

which is available in the SimPowerSystems Toolbox of Simulink. The validity of this model has been verified by experimental studies. The detailed description about the modeling of the battery model can be found in [12,13,35,36]. In this model, the battery is developed by its circuit equivalent, which is composed of a variable voltage source in series with a resistor. The battery output voltage can be calculated due to the battery open circuit voltage and voltage drop resulting from the battery

Vbat = Ebat − Ri ibat

EP

equivalent internal impedance. Accordingly, the output voltage of battery can be expressed as (7)

AC C

where, Vbat is the battery voltage, Ebat is the open circuit voltage (OCV), ibat is the battery current, and Ri is the internal resistance, which is assumed to be constant during the charge and the discharge cycles. The proposed discharge and charge model can represent accurately the battery voltage dynamics if the battery current varies. It can be calculated as follows

Q Q   Ebat _ dis = E0 − K Q − C id − K Q − C Ci + M exp(−N ⋅ Ci )  i i  Q Q E = E −K id − K Ci + M exp(−N ⋅ Ci )  bat _ch 0 Ci + 0.1Q Q − Ci

(8)

where, E0 is the constant voltage, id is the filtered current which is the result of applying a low-pass filter to the battery current, K is the polarization constant, Ci is the extracted capacity, Q is the maximum battery capacity, M is the exponential voltage and 9

ACCEPTED MANUSCRIPT N is the exponential capacity. The battery stage of charge (SOC) must be kept between 40% and 80% of capacity in order to achieve high charge efficiency [12,36]. The energy management strategy is designed to make to work the battery in this operating range. The battery SOC is calculated as

 1 t  SOC = 100 ⋅ 1 − ∫ ibat (t )dt   Q0 

RI PT

(9)

The characteristics of the charge and discharge pulses are high power levels and the duration is from tens of milliseconds to tens of seconds in the tramway driving cycle [12,14]. An analysis of hybrid vehicles power requirements and related literature

appropriate to meet transient and instantaneous peak power demands.

SC

show that the SC bank is able to deal with the frequent charge and discharge pulses from the expected power profiles, and is

M AN U

In this work, the behavior of the SC bank is represented by the model presented in [4,12,14,18], which is available in the SimPowerSystems Toolbox of Simulink. This model consists of a capacitance representing the SC performance during the charging and discharging process, an equivalent series resistance representing the charging and discharging resistance, and an equivalent parallel resistance representing the self-discharging losses.

In order to validate the SC model, a comparison test with commercial Maxwell BMOD SC which is specifically designed for high-power transport applications such as locomotive and the tramway has been carried out in [12,14]. The validity of this

[4,12,14,18]. 2.3 Modeling of DC/DC converter

TE D

model has been verified by experimental studies. The detailed description about the modeling of the SC model can be found in

EP

The voltage of the hybrid system energy sources varies depending on the demand current. Hence, a power electronic device is needed to process the output power of the energy sources, providing the demand power via the device at a constant voltage in

AC C

the DC bus. Specifically, the power electronic device is composed of a PWM-based DC/DC converter for each energy source, which connects the sources with the DC bus, where the auxiliary services and the inverter module are supplied. The PEMFC, battery and SC present a lower terminal voltage than the DC voltage necessary to feed the traction inverter. In this study, a unidirectional boost DC/DC converter connects the PEMFC with the 800VDC bus maintaining the PEMFC system stable despite variation s in load. The PEMFC is also protected from high frequency and surge load power demand so that the life time could be extended. The converter which is shown in Fig. 9(a) is composed of a high frequency inductor L1, a filtering capacitor C1, a diode D1, and a main switch S1. This converter has been developed by utilizing the two-quadrant chopper model included in SimPowerSystems of Simulink [37,38]. In this model, the converter is represented by its average value equivalent, which is composed of a controlled current source at the DC bus side and a controlled voltage source at the 10

ACCEPTED MANUSCRIPT PEMFC side (Vlow). The current and voltage sources are used to represent the power electronic switches. This equivalent scheme has been used in the simulations performed, because it enables larger sample times while preserving the average voltage dynamics [38]. In addition, this converter is adjusted by a current control loop. By acting on the switch S1 gate signal, it is therefore possible to determine the power generated by the PEMFC.

RI PT

Two bidirectional converters are respectively used to the battery and SC with boost operation if discharging and buck operation if charging. The battery and SC are located on the low voltage side (Vlow), and the high voltage side is connected to the 800VDC bus. Hence, the voltage from the energy sources is boosted on the DC bus with the aim of voltage regulation. This converter consists of a high frequency inductor L2, a filtering capacitor C2, and two switches S3 and S4, as shown in Fig.

SC

9(b). In the charging mode, the S3 switch and S4 diode act as a unidirectional buck converter that uses the DC bus energy to charge the battery or SC. In the discharging mode, the S4 switch and S3 diode act as a unidirectional boost converter that delivers

M AN U

the energy to the DC bus. It has also been developed by using the two-quadrant chopper model included in SimPowerSystems [37,38]. This converter is controlled by current and voltage loop. In fact, effective control of the duty cycle of the bidirectional converter assures the rated voltage of the DC bus and the charge and discharge safety of the battery and the SC.

Fig. 9. (a) Unidirectional DC/DC converter topology for PEMFC

(b) Bidirectional DC/DC converter topology for battery and SC

TE D

3. Power sharing algorithm based on Harr-WD and FLC

A given signal can be decomposed into transients, and desired characteristics can be extracted simultaneously in both the time and frequency domain by using the WT [22-24]. The discrete wavelet transform (DWT) is used to decompose a discretized signal into different resolution levels. The DWT can be expressed as

1

λ

t −u  ,  dt  λ 

EP

W ( λ , u ) = ∫ s (t )

ψ

λ = 2 j,

u = k2j

(10)

AC C

where, the wavelet coefficients W are obtained as a function of scale λ and position u. The scale parameter λ controls the length of the frequency band and the position parameter u controls the size of the time window. The original signal s(t) is decomposed to obtain the coefficients W(λ, u). The signal s(t) can be reconstructed from the coefficients W(λ, u), which is called the WT synthesis. ψ(t) is a mother function in both time domain and frequency domain called the mother wavelet. This mother wavelet is for the purpose of providing a source function to generate the daughter wavelets which are simply the translated and scaled versions of the mother wavelet. The inverse DWT is given by

s (t ) = ∑∑ W ( j , k )ψ j , k (t )

(11)

j∈Z k ∈Z

11

ACCEPTED MANUSCRIPT As the most popular mother wavelet, the Haar wavelet which has the shortest filter length in the time domain compared to other wavelets makes the decomposition calculation much simpler do for realization of the WT strategy in a real-time system. Meanwhile, the function of extracting transients can still be implemented well without degradation. The detailed description about the mathematical basis of the Haar wavelet can be found in [6,7,25]. Hence, the Haar-WT is utilized in this work for load

which are assigned for the ESS. The Haar wavelet is expressed as

 1,  ψ (t ) = −1,  0, 

if if

0 ≤ t < 0.5 0.5 ≤ t < 1 otherwise.

(12)

RI PT

profile decomposition to obtain the transients (low and high frequency signals) corresponding to sharp peak power demand

SC

In this work, three-level Haar wavelet decomposition and reconstruction are applied for the original signal s(n) as shown in Fig. 10. The Haar-WT algorithm uses a low pass filter and a high-pass filter in the wavelet decomposition and reconstruction

M AN U

structure. This structure consists of a filter bank to decompose the signal into low and high frequency components. In the filter bank, l0(z) represents the low pass analysis filter, and h0(z) represents the high pass analysis filter. The down-sampling and up-sampling methods are employed in the decomposition and reconstruction processes respectively. The data size reduces by half in down-sampling operations while it doubles in up-sampling operations. The low pass and high pass filters are combined with down-sampling operations, that is, steps that throw away every other sample at each process, reducing the data size by 50%

TE D

each time [6,7,25]. Similarly, the synthesis filter bank consists of the low pass synthesis filter l1(z) and the high pass synthesis filter h1(z) with up-sampling operations. The signal s(n) is the power demand profile of tramway driving cycle. After the original signal reconstructed from the Haar wavelet decomposition structure, the maximum error is calculated as 4.6566e-10. This error value shows the accuracy and reliability of the load sharing algorithm and achieves the perfect reconstruction of the original

EP

signal as well. The principle depicted in Fig. 10 is considered as a basis for the load sharing strategy. However, only adopting a load sharing algorithm may not be sufficient for regulating all the system dynamics in desired

AC C

range. The battery and the SC bank should have enough charge to supply the required load demand if the tramway is accelerating and it should have enough capacity to recuperate energy during braking, which is rather important for increased energy savings. Hence, in order to raise the efficacy of the hybrid system, a control system based on a power sharing strategy must be developed for keeping the output power and power changing rate of the PEMFC system in a suitable range while maintaining the SOC of the battery and the SC in predefined limits.

Fig. 10. Three-level Haar decomposition and reconstruction diagram

As the hybrid system is nonlinear and multivariable, the fuzzy logic controller (FLC) is more suitable for the energy management. The controller relates its output to inputs using a list of IF-THEN rules. The IF part of a rule that describes regions 12

ACCEPTED MANUSCRIPT of input variables specifies the condition for which a rule holds. A particular input value belongs to these regions to a certain degree, represented by a degree of membership function which is assigned to the variables according to the membership functions definition. The THEN part of a rule refers to values of the output variable to obtain the output of the controller. The membership degree of the IF part of all rules are evaluated and all rules of the THEN part are averaged and weighted by these

RI PT

membership degrees [18]. In order to rationally assign the demand power to the PEMFC, battery and SC, a power sharing algorithm is proposed based on the Haar-WT algorithm and the FLC approach. In the fuzzy inference system, two decentralized FLC (2-FLC) are designed according to the power sharing algorithm. The control objective with the 2-FLC would be achieved by improving the

SC

efficiencies of tramway and PEMFC system on condition that dynamic property of tramway is satisfied, which is major advantages for utilization of the Haar-WT and 2-FLC in the tramway energy management system.

M AN U

The No.1 FLC (FLC1) has three input variables and one output variable. The input variables include the power signal Pm1=s0 generated by the Haar-WT, the battery SOC and the SC bank SOC (CSOC). The output variable is the PEMFC reference power Pref1. For FLC1, the fuzzy field scope of the power signal Pm1 is set as [-1, 1], the fuzzy field scope of SOC is set as [0, 1], the fuzzy field scope of CSOC is set as [0, 1], and the fuzzy field scope of PEMFC reference power Pref1 is set as [0, 1]. The fuzzy subset of Pm1 is divided into {NH, NL, Z, PL, PH}, the fuzzy subsets of SOC and CSOC are divided into {L, M, H}, the fuzzy subset of Pref1 is divided into {L1, L2, L3}.

TE D

The No.2 FLC (FLC2) has three input variables and one output variable. The input variables include the power signal Pm2=s1 + s2 + s3+ (s0- Pref1) generated by the Haar-WT, and the battery SOC and the SC CSOC. The output variable is the SC reference power Pref3. For FLC2, the fuzzy field scope of the power signal Pm2 is set as [-1, 1], the fuzzy field scope of SOC is set as [0, 1],

EP

the fuzzy field scope of CSOC is set as [0, 1], and the fuzzy field scope of the SC reference power Pref3 is set as [-1, 1]. The fuzzy subset of Pm2 is divided into {NH, NM, NL, PL, PM, PH}, the fuzzy subsets of SOC and CSOC are divided into {L, M,

AC C

H}, the fuzzy subset of Pref3 is divided into {NH, NM, NL, Z, PL, PM, PH}. The membership functions are achieved by inhomogeneous distribution as shown in Fig. 11 and Fig. 12. The Trapezoid Membership Functions are employed for Pm1 and Pm2, the S-type Membership Function, the Z-type Membership Function and the Triangle Membership Function are employed together for SOC and CSOC.

Furthermore, the fuzzy reasoning rules with 54 items are respectively established for two FLC in order to achieve the control objective. The part rules are listed in Table. 2 and Table. 3. The form of fuzzy reasoning is “IF Pm1 is Ai AND SOC is Bi AND CSOC is Ci, Then Pref1 is Di”. All rules are on the basis of practical experience of energy management. Mamdani Inference

13

ACCEPTED MANUSCRIPT Method is utilized to carry out the fuzzy reasoning. Finally, the method of weighted mean (centroid method) is used to convert the vector into a single value and then the defuzzification is achieved. u0 =

∫ x ⋅ µ ( x )dx ∫ µ ( x )dx i

i

i

i

i

(13)

RI PT

where xi is the element of a set in the field, µ(xi) is the membership degree, u0 is the judgment result of fuzzy output.

Fig. 11. Membership functions of input and output variables for No.1 FLC

SC

Fig. 12. Membership functions of input and output variables for No.2 FLC

M AN U

Table. 2 Part fuzzy reasoning rules of No.1 FLC

Table. 3 Part fuzzy reasoning rules of No.2 FLC

An energy management system based on the proposed power sharing algorithm is shown in Fig. 2. By using the power sharing strategy, the power demand of tramway driving cycle is decomposed into high frequency and low frequency components

TE D

in three-levels for power distribution, which can take advantage of the three power sources. In this proposed system, the three power sources and corresponding power control are independent. The required power demand suitable for the characteristics of an individual power source can be met precisely. The PEMFC system deals with the majority of the positive portion of the low frequency components of power demand derived from the Haar-WT. The power flow from the PEMFC is determined by switch

EP

duty cycle of the unidirectional boost DC/DC converter. The battery absorbs the slow-variation negative portion with the direction from the load to power sources. The battery also helps provide a portion of the positive low frequency components of

AC C

power demand derived from the Haar-WT to reduce the burden of the PEMFC. All the high frequency components of the power demand are satisfied by the SC bank, which can protect the PEMFC from the membrane damage resulting from the high frequency power demand variations. Furthermore, according to the fuzzy reasoning rules designed above, the SOC and CSOC must be controlled within a suitable range while PEMFC delivers power in an appropriate power changing rate provided by the power sharing strategy. The required power from the battery and SC are determined by properly adjusting the switches existing in the bidirectional DC/DC converter. Hence, the high efficiency of the PEMFC is achieved and the lifetime of the PEMFC is extended consequently under the assistance of the battery and the SC. Hence, the reference power demands of the PEMFC Pref1, the battery Pref2 and the SC Pref3 can be expressed as

14

ACCEPTED MANUSCRIPT if 0 < s0 (n) < V fc I fc max otherwise

(14)

Pref 2 = s1 ( n) + s2 ( n) + s3 (n ) + s0 ( n) − Pref 1 − Pref 3

(15)

Pref 3 = PFLC 2

(16)

4. Results and Discussions

RI PT

P Pref 1 =  FLC1  0

In this study, in order to evaluate the performance of the proposed power sharing strategy for the hybrid system of tramway, the real driving cycle of the LF-LRV tramway in Turkey is utilized. A round-trip route (278 s) which consists of a symmetrical route of going and return has been used as shown in Fig. 13. The hybrid propulsion system of LF-LRV tramway consists of

SC

Ballard FCvelocity™ HD6, Phylion Li-ion Battery XH61, and Maxwell BMOD0615.

Several premises must be considered carefully for a hybrid system sizing. In Fig.13, the rates of power change for the driving

M AN U

cycle are 30.02 kW/s, 62.37 kW/s and 58.79 kW/s respectively during three acceleration processes. However, the rate of power change for 150 kW PEMFC power unit is 24.20 kW/s in Fig.8, and it’s not enough to satisfy the driving cycle dynamic. Hence, the sole PEMFC power unit can not satisfy the rate of power change demand for the driving cycle and may damage the PEMFC stack. In addition, the PEMFC maximum power should be higher than the average power demanded during the driving cycle in order to avoid an excessive drop of the state of charge (SOC) of battery and SC. In Fig. 13, the average power of driving

TE D

requirement is 82.2 kW according to the driving cycle of LF-LRV tramway. Thus, the PEMFC rated power is selected the value of 150 kW, in order to avoid great requirements for the ESS. Due to the peak power of driving requirement is 532.6 kW, it gives rise to a 382.6 kW peak power from the ESS. Furthermore, considering the average power and the peak power of regenerative braking is 113.9 kW and 413.1kW, the SC which has a higher specific power and a longer lifetime in terms of number of

EP

charge/discharge cycles is utilized to improve the dynamic response. Therefore, at present work, a hybrid propulsion system considered for the LF-LRV tramway presents a 150 kW PEMFC, a 50 Ah Li-ion battery and a SC bank with a total capacity of

AC C

30 F. These devices have been selected from commercially available components as Table. 4 shown.

Fig. 13. Driving cycle of the LF-LRV tramway in Turkey

Table. 4 Parameters of the commercial devices

As the results of the hybrid propulsion system of LF-LRV tramway based on the proposed power sharing strategy, the PEMFC system output power, voltage, current and efficiency are given in Figs. 14, the battery power, voltage, current and SOC are presented in Figs. 15, the SC bank power, voltage, current and SOC are illustrated in Figs. 16, respectively.

15

ACCEPTED MANUSCRIPT Fig. 14. Output power, voltage, current and efficiency of PEMFC system

Fig. 15. Output power, voltage, current and SOC of battery pack

Fig. 17. DC bus voltage and current

RI PT

Fig. 16. Output power, voltage, current and efficiency of SC bank

SC

The PEMFC, battery and SC output power curves are depicted in Figs. 14-16 during the real driving cycle of the LF-LRV tramway. With regards to the output power of PEMFC system, it can be observed that the proposed power sharing strategy is

M AN U

able to guarantee the stable operation of the PEMFC system during the most of the drive cycle. The PEMFC system only increases or decreases the power during high accelerations or brakings. In addition, the PEMFC system almost operates at high efficiency region under the proposed power sharing strategy. Due to the dynamic limitations of the PEMFC, the SC bank supplies the transient power demand successfully during sudden acceleration or braking as illustrated in Fig. 16. Its fast response fulfills the power demand, increases the hybrid system power density and has to generate or absorb the power that the either the

TE D

PEMFC or the battery are not able to generate or absorb because of its high dynamic response and high specific power. In fact, the SC output power is zero or close to zero at most of the time. Furthermore, the battery output power alternates between negative and positive according to charging or discharging as shown in Fig. 15. It helps provide a portion of the positive low frequency components of power demand to reduce the burden of the PEMFC, and absorbs the slow-variation negative portion.

EP

Specially, the SOCs of battery and SC bank are maintained at reasonable levels under the proposed power sharing strategy. The DC bus voltage and current during the drive cycle are shown in Fig. 17. It is easily observed that the DC bus voltage is

AC C

maintained around the desirable value (800V) under the proposed power sharing strategy. Furthermore, the comparisons with the sole FLC, the sole Harr-WD strategy, the proposed power sharing strategy (Harr-WD FLC) are also carried out under the same the driving cycle and hybrid topology. The energy efficiencies of the hybrid tramway and the PEMFC system are calculated as follows

∫P

HT

η HT =

dt (17)

t

∫P

fcs

t

dt + ∫ Pbat dt + ∫ Psc dt + ∫ Pbr dt t

t

t

16

ACCEPTED MANUSCRIPT

∫ P dt fc

η fcs =

(18)

t

∫ Pfcs dt t

where, PHT is the traction power of the hybrid tramway, Pfcs is total output power of the PEMFC system including the power

braking resistor, Pbat is the battery output power, Psc is the SC output power.

RI PT

demanded by the auxiliary components of the PEMFC system, Pfc is the PEMFC output power, Pbr is the power dissipated in the

Table. 5 Results of different strategies during the drive cycle

SC

Table. 5 shows the results obtained by FLC, Harr-WD and Harr-WD FLC. The indexes used in the comparative research are: tramway average efficiency, PEMFC average efficiency, average value of the battery SOC and SC CSOC. In Table. 5, the best

M AN U

value achieved is denoted in italic and bold. In addition, the difference of each index and strategy with respect to the best value of the index is shown in percent.

Comparing the tramway and the PEMFC average efficiencies, the Harr-WD FLC strategy achieves a greater value than others during all of the drive cycle. In fact, the tramway average efficiency is 50.86% in the case of the Harr-WD FLC, whereas it is 48.63% for the FLC and it is 45.03% for the Harr-WD. The maximum difference obtained between the best and the worst tramway average efficiency is 11.4%. The same results occur for the PEMFC average efficiency and the average SOC of battery

TE D

as shown in Table.5. However, the best results related to the average CSOC of SC is obtained by the FLC. In fact, although the average CSOC value is better than others achieved by the FLC during all of the drive cycle, the tramway and the PEMFC average efficiencies which are more significant for the hybrid tramway is not the optimal. The reason is that the Harr-WD FLC

EP

is able to precisely satisfy the required power demand according to the characteristics of each power source and ensures the majority of the positive portion of the low frequency components of power demand can be deals with the PEMFC system.

AC C

Furthermore, the battery helps provide a portion of the positive low frequency components of power demand to reduce the burden of the PEMFC by the Harr-WD FLC and the SC bank supplies all the high frequency components of the power demand which could damage the PEMFC membrane and reduce the PEMFC lifetime. Hence, the higher average efficiencies of the tramway and the PEMFC are achieved and the lifetime of the PEMFC system is extended consequently by the Harr-WD FLC.

5. Conclusions

Hydrogen-powered hybrid propulsion system could play a central role in future transportation. In this paper, the hybrid system based on the PEMFC, the battery and the SC is designed without grid connection for the LF-LRV tramway in Samsun, Turkey. The PEMFC acts as main energy source of tramway, the Li-ion battery and the SC bank are utilized as the ESS to supplement the PEMFC output power the during tramway acceleration and cruise, and they are also used for energy recovery 17

ACCEPTED MANUSCRIPT during braking. Specially, the SC bank is able to consume the peak power that neither the PEMFC nor the battery can store because of its high dynamic response and high specific power. The lifetime and efficiency of the PEMFC and ESS are critical factors for high-power transport economy. Hence in order to avoid the rapid variation and large differential power demand for the PEMFC, and distribute the power demands to power sources appropriately, an appropriate power sharing strategy which

RI PT

should isolate the base power demand of a given signal from its transients is quite essential for the energy management of the high-power tramway. Duo to its excellent performance in analyzing the transients, the multi-level Haar-WT which can be embedded into the structure of 2-FLC strategy is utilized to separate the high frequency and low frequency components of the load profile. Therefore, the power sharing strategy based on the combination of FLC and Haar-WT is proposed to avoid the

SC

rapid changes of power demand, achieve high efficiency without degrading the mechanism performance and appropriately distribute the power demand to each power source.

M AN U

In this work, the hybrid propulsion system model of LF-LRV tramway which consists of Ballard FCvelocity™ HD6, Phylion Li-ion Battery XH61, and Maxwell BMOD0615 is developed with commercially available devices to evaluate the performance of the proposed energy management system. According to the real drive cycle of the tramway in Turkey, the testing of this energy management system based on the proposed power sharing strategy is carried out. The results demonstrate that the PEMFC can deal with the majority of the positive portion of the low frequency components of power demand derived from the Haar-WT. Simultaneously, the battery is able to absorb the slow-variation negative portion with the direction from the load to

TE D

power sources. And it also helps provide a portion of the positive low frequency components of power demand derived from the Haar-WT to reduce the burden of the PEMFC. All the high frequency components of the power demand are satisfied by the SC bank, which can protect the PEMFC from the membrane damage resulting from the high frequency power demand variations.

EP

Therefore, the proposed energy management system is able to guarantee a safe operating condition with transient free for the PEMFC and extend the lifetime of each power source. Furthermore, the comparisons with the sole FLC and the sole Harr-WD

AC C

strategies verify that the higher average efficiencies of the tramway and the PEMFC are achieved and the lifetime of the PEMFC system is extended consequently with the proposed strategy. Therefore, the proposed power sharing strategy will provide a novel approach for the advanced energy management system of high-power hybrid tramway.

Acknowledgments The authors would like to thank the reviewers for their helpful suggestions. This work was supported by National Natural Science Foundation of China (51177138, 61473238, 51407146), National Key Technology R&D Program (2014BAG08B01), Sichuan Provincial Youth Science and Technology Fund (2013JQ0033), Science and Technology Development Plan of Ministry of Railways (2012J012-D), and Specialized Research Fund for the Doctoral Program of Higher Education (20120184120011). 18

ACCEPTED MANUSCRIPT

References [1]

Cheng Bao, Minggao Ouyang, Baolian Yi, “Modeling and control of air stream and hydrogen flow with recirculation in a PEM fuel cell system-II. Linear and adaptive nonlinear control”, Int. J. Hydrogen Energy, 31 (2006) 1897-1913.

[2]

Liu ZX, Zhang HY, Wang C, Mao ZQ, “Numerical simulation for rib and channel position effect on PEMFC performances,” Int. J. Hydrogen Energy, vol.

RI PT

35, pp. 2802-2806, 2010. [3]

Liu ZX, Mao ZQ, Wang C, Zhuge WL, Zhang YJ, “Numerical simulation of a mini PEMFC stack”, J. Power Sources. 160 (2006) 1111-1121.

[4]

J. Jia, G. Wang, Y. T. Cham, Y. Wang, M. Han, “Electrical Characteristic Study of a Hybrid PEMFC and Ultracapacitor System”, IEEE Trans. Industrial Electronics, 57 (2010) 1945-1953.

Liangfei Xu, Jianqiu Li, Jianfeng Hua, Xiangjun Li, Minggao Ouyang, “Adaptive supervisory control strategy of a fuel cell/battery-powered city bus”, J. Power Sources, 194 (2009) 360-368.

[6]

O. Erdinc, B. Vural, M. Uzunoglu, Y. Ates, “Modeling and analysis of an FC/UC hybrid vehicular power system using a wavelet-fuzzy logic based load

O. Erdinc, B. Vural, M. Uzunoglu, “A wavelet-fuzzy logic based energy management strategy for a fuel cell/battery/ultra-capacitor hybrid vehicular power system”, J. Power Sources, 194 (2009) 369-380.

[8]

M AN U

sharing and control algorithm”, Int. J. Hydrogen Energy, 34 (2009) 5223-5233. [7]

SC

[5]

B. Vural, A. R. Boynuegri, I. Nakir, O. Erdinc, A. Balikci, M. Uzunoglu, H. Gorgun, S. Dusmez, “Fuel cell and ultra-capacitor hybridization: A prototype test bench based analysis of different energy management strategies for vehicular applications”, Int. J. Hydrogen Energy, 35(2010) 11161-11171.

[9]

Y. Eren, O. Erdinc, H. Gorgun, M. Uzunoglu, B. Vural, “A fuzzy logic based supervisory controller for an FC/UC hybrid vehicular power system”, Int. J. Hydrogen Energy, 34(2009) 8681-8694.

TE D

[10] P. Thounthong, S. Pierfederici, J.-P. Martin, M. Hinaje, B Davat, “Modeling and Control of Fuel Cell/Supercapacitor Hybrid Source Based on Differential Flatness-Control”, IEEE Trans. on Vehicular Technology, 59 (2010) 2700-2710. [11] Fei Peng, WeiRong Chen, Zhixiang Liu, Qi Li, Chaohua Dai, “System Development of 150kW Electric Locomotive Powered by Individual PEM Fuel Cell Stack”, Int. J. Hydrogen Energy, 39 (2014) 13886–13893.

EP

[12] Pablo García, Juan P.Torreglosa, Luis M.Fernández, Francisco Jurado, “Viability study of a FC-battery-SC tramway controlled by equivalent consumption minimization strategy”, Int. J. Hydrogen Energy, 37 (2012) 9368–9382. [13] J.P. Torreglosa, F. Jurado, P. Garcı´a , L.M. Ferna´ndez, “Hybrid fuel cell and battery tramway control based on an equivalent consumption minimization

AC C

strategy,” Control Engineering Practice, 19 (2011) 1182–1194. [14] Pablo García, Juan P.Torreglosa, Luis M.Fernández, Francisco Jurado, “Control strategies for high-power electric vehicles powered by hydrogen fuel cell, battery and supercapacitor”, Expert Systems with Applications, 40 (2013) 4791–4804. [15] Luis M. Fernandez, Pablo Garcia, Carlos Andrés Garcia, Francisco Jurado, “Hybrid electric system based on fuel cell and battery and integrating a single dc/dc converter for a tramway,” Energy Conversion and Management, 52 (2011) 2183-2192. [16] A.R. Miller, K.S. Hess, D.L. Barnes, T.L. Erickson. “System design of a large fuel cell hybrid locomotive,” J. Power Sources, 173 (2007) 935-942. [17] Oosawa, Mitsuyuki Fujii, Takehito. “Development of a New Energy (NE) Train,” Japanese Railway Engineering, 156 (2006) 62-70. [18] Qi Li, Weirong Chen, Yankun Li, Shukui Liu, Jin Huang, “Energy management strategy for fuel cell/battery/ ultracapacitor hybrid vehicle based on fuzzy logic”, J. Electrical Power and Energy Systems, 43 (2012) 514-525. [19] Phatiphat Thounthong, Pietro Tricoli, Bernard Davat, “Performance investigation of linear and nonlinear controls for a fuel cell/supercapacitor hybrid power plant,” J. Electrical Power and Energy Systems, 54 (2014) 454-464.

19

ACCEPTED MANUSCRIPT [20] Liangfei Xu, Minggao Ouyang, Jianqiu Li, Fuyuan Yang, Languang Lu, Jianfeng Hua, “Application of Pontryagin’s Minimal Principle to the energy management strategy of plugin fuel cell electric vehicles,” Int. J. Hydrogen Energy, 38 (2013) 10104-10115. [21] Kim M, Sohn YJ, Lee WY, Kim CS, “Fuzzy control based engine sizing optimization for a fuel cell/battery hybrid mini-bus”, J. Power Sources, 178 (2008) 706-710. [22] P. L. Mao and R. K. Aggarwal, “A novel approach to the classification of the transient phenomena in power transformers using combined wavelet transform and neural network,” IEEE Trans. Power Del., vol. 16, no. 4, pp. 654–660, Oct. 2001.

RI PT

[23] C. Capilla, “Application of the Haar wavelet transform to detect microseismic signal arrivals,” J. Appl. Geophys., vol. 59, no. 1, pp. 36–46, May 2006. [24] L. Joseph and T. Minh-Nghi, “A wavelet-based approach for the identification of damping in nonlinear oscillators,” Int. J. Mech. Sci., vol. 47, no. 8, pp. 1262–1281, 2005.

[25] Xi Zhang, Chris Chunting Mi, Abul Masrur, David Daniszewski, “Wavelet-transform-based power management of hybrid vehicles with multiple on-board

SC

energy sources including fuel cell, battery and ultracapacitor”, J. Power Sources, 185 (2008) 1533-1543.

[26] Qi Li, Weirong Chen, Zhixiang Liu, Ai Guo, Jin Huang, “Nonlinear Multivariable Modeling of Locomotive Proton Exchange Membrane Fuel Cell System ”, Int. J. Hydrogen Energy, 39 (2014) 13777-13786.

1005-1014.

M AN U

[27] M. Han, S. H. Chan, S. P. Jiang, “Development of carbon-filled gas diffusion layer for polymer electrolyte fuel cells”, J. Power Sources, 159 (2006)

[28] Meiying Ye, Xiaodong Wang and Yousheng Xu, Parameter identification for proton exchange membrane fuel cell model using particle swarm optimization, Int. J. Hydrogen Energy, 34 (2009) 981-989.

[29] Lin Wang, Attila Husar, Tianhong Zhou, et al., “A parametric study of PEM fuel cell performances”, Int. J. Hydrogen Energy, 28 (2003) 1263-1272. [30] R. F. Mann, J. C. Amphlett, M. A. Hooper, et al., “Development and application of a generalized steady-state electrochemical model for a PEM fuel cell”, J. Power Sources, 80 (2000) 173-180.

TE D

[31] Q. Li, W. Chen, Y. Wang, S. Liu, J. Jia, “Parameter identification for PEM Fuel Cell Mechanism Model Based on Effective Informed Adaptive Particle Swarm Optimization”, IEEE Trans. Industrial Electronics, 58 (2011) 2410-2419. [32] Qi Li, Weirong Chen, Zhixiang Liu, Ai Guo, Shukui Liu, “Control of proton exchange membrane fuel cell system breathing based on maximum net power control strategy,” J. Power Sources, 214 (2013) 212-218.

EP

[33] J.T. Pukrushpan, A.G. Stefanopoulou, H. Peng, “Control of fuel cell breathing,” IEEE Trans. on Control Systems, 24 (2004) 30-46. [34] Arce A., del Real A.J., Bordons C., Ramirez D.R., “Real-Time Implementation of a Constrained MPC for Efficient Airflow Control in a PEM Fuel Cell,” IEEE Trans. Industrial Electronics, 57 (2010) 1892-1905.

AC C

[35] Yang TC, “Initial study of using rechargeable batteries in wind power generation with variable speed induction generators,” IET Renewable Power Generation, 2(2008) 89-101.

[36] Sikha G, White RE, Popov BN. A mathematical model for a lithium-ion battery/electrochemical capacitor hybrid system. J Electrochem Soc 2005;152(8):1682–93.

[37] Majid Delshad, Hosein Farzanehfard, “A new soft switched push pull current fed converter for fuel cell applications”, Energy Conversion and Management, 52 (2011) 232-243. [38] Marquezini DD, Ramos DB, Machado RQ, Farret FA, “Interaction between proton exchange membrane fuel cells and power converters for AC integration,” IET Renewable Power Generation, 2(2008) 151-161.

20

ACCEPTED MANUSCRIPT Table. 1 Performance parameters of 150kW HD6 Module

Values

Performance Parameters

Values

Nominal Operating Point

[568V,267A]

Operating Temperature

330K

Maximum Operating Point

[550V, 300A]

Nominal air mass flow

3653LPM

Cell Number

762

Nominal Hydrogen Pressure

2.24bar

Nominal Efficiency

55%

Nominal Air Pressure

2.06bar

SC

RI PT

Performance Parameters

Reactant Concentration

[99.999(H2), 21(O2)]

0.1s

AC C

EP

TE D

M AN U

Dynamic Response Time Constant

ACCEPTED MANUSCRIPT Table. 2 Part fuzzy reasoning rules of No.1 FLC

1. If (Pm1 is NH) and (SOC is H) and (CSOC is M) then (Pref1 is L1) 2. If (Pm1 is NL) and (SOC is M) and (CSOC is L) then (P ref1 is L1) 3. If (Pm1 is Z) and (SOC is L) and (CSOC is L) then (P ref1 is L2)

AC C

EP

TE D

M AN U

SC

5. If (Pm1 is PH) and (SOC is L) and (CSOC is L) then (P ref1 is L3)

RI PT

4. If (Pm1 is PL) and (SOC is L) and (CSOC is M) then (P ref1 is L2)

ACCEPTED MANUSCRIPT Table. 3 Part fuzzy reasoning rules of No.2 FLC

1. If (P m2 is NH) and (SOC is L) and (CSOC is M) then (P ref3 is NM) 2. If (P m2 is NM) and (SOC is L) and (CSOC is H) then (Pref3 is Z) 3. If (P m2 is NL) and (SOC is M) and (CSOC is L) then (Pref3 is NL)

AC C

EP

TE D

M AN U

SC

5. If (P m2 is PH) and (SOC is L) and (CSOC is H) then (Pref3 is PH)

RI PT

4. If (P m2 is PL) and (SOC is H) and (CSOC is M) then (P ref3 is PL)

ACCEPTED MANUSCRIPT Table. 4 Parameters of the commercial devices PEMFC Manufacturer

Ballard Stack Modules-FCvelocity™ HD6

Rated Power (kW)

150

Rated Voltage Range (V)

570-650

Operating Temperature (°C)

63

Mass (kg)

355

Li-ion Battery Phylion Battery XH61

Capacity (Ah)

50

Rated Voltage (V)

Maximum Discharging Rate (C)

5

Set Numbers

Maximum Discharging Current (A)

240

Mass (kg)

Supercapacitor (SC) Manufacturer

Maxwell BMOD0615

2 parallel 160

Rated Voltage (V)

Absolute Maximum Current (A)

1900

Specific Power (W/kg)

3300

Set Numbers

11 series*2 parallel

Mass (kg)

13.9

TE D EP AC C

528

SC

30

M AN U

Capacity (F)

518

RI PT

Manufacturer

ACCEPTED MANUSCRIPT Table. 5 Results of different strategies during the drive cycle FLC

Harr-WD

Harr-WD FLC

Tramway Average Efficiency (%)

48.63 (4.4%)

45.03 (11.4%)

50.86 (0.0%)

PEMFC Average Efficiency (%)

49.88 (5.2%)

46.80 (11.0%)

52.60 (0.0%)

Average SOC (%)

73.72 (0.9%)

73.52 (1.2%)

Average CSOC (%)

79.94 (0.0%)

67.27 (15.8%)

RI PT

Indexes

74.45 (0.0%)

AC C

EP

TE D

M AN U

SC

78.23 (2.1%)

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

Fig. 1. The developing hybrid LF-LRV tramway without grid connection

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

Fig. 2. Configuration of PEMFC-battery-SC powered hybrid system for the tramway

ACCEPTED MANUSCRIPT

VENT OUT

HD6 Block Diagram Particulate Filter

Pressure Regulator

Hydrogen Blower Exhaust Gas

Solenoid Valve

FUEL IN Purge Valve

VENT IN

DI Filter

Coolant Circulation Radiator

Particulate Filter

DI-WEG OUT

Condensor Pump

AIR IN Particulate Motor Filter

RI PT

Coolant Pump

DI-WEG IN

COND IN Spray Pump

Air Compressor

Condenser Circulation Radiator

Water Separator Humidifier

COND OUT

AIR OUT

SC

Water Tank

AC C

EP

TE D

M AN U

Fig. 3. Schematic of the 150kW PEMFC power module

AC C

EP

TE D

M AN U

SC

Fig. 4. PEMFC equivalent circuit model

RI PT

ACCEPTED MANUSCRIPT

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

Fig. 5. Schematic of 150 kW PEMFC power unit simulation

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

Fig. 6. Testing data of air mass flow, air input pressure and output power from PEMFC power unit

ACCEPTED MANUSCRIPT

Stack voltage vs current 800 simulated measured

700 650 600 550 0

50

100

150 Current(A)

200

Stack power vs current 200

100

50

100

M AN U

50

0 0

250

300

SC

Power(kW)

150

RI PT

Voltage(V)

750

150 Current(A)

200

250

AC C

EP

TE D

Fig. 7. Comparisons between the experimental data and the polarization curve by simulation

300

ACCEPTED MANUSCRIPT

250 Measured Current Simulated Current Requested Current

150

100

50

2

4

6

8 Time (s)

10

12

14

SC

0

RI PT

Current (A)

200

(a)

M AN U

800

Simulated Voltage Measured Voltage

Voltage (V)

750

700

650

TE D

600

550

2

4

6

8 Time (s)

10

12

14

(b)

EP

Fig. 8. (a) Dynamic step testing data and simulation results of PEMFC power unit current. (b) Dynamic step testing data and

AC C

simulation results of PEMFC power unit voltage

RI PT

ACCEPTED MANUSCRIPT

(a)

(b)

AC C

EP

TE D

M AN U

SC

Fig. 9. (a) Unidirectional DC/DC converter topology for PEMFC (b) Bidirectional DC/DC converter topology for battery and SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

Fig. 10. Three-level Haar decomposition and reconstruction diagram

ACCEPTED MANUSCRIPT PL

L

0.8

0.8

Degree of membership

1

0.6

0.4

0

0

-0.6

-0.4

-0.2

0 Pm1

0.2

0.4

L

0.6

0.8

M

0

1

0.1

0.8

0.8

Degree of membership

1

0.4

0.2

0.2

0.3

L1

H

1

0.6

H

0.4

0.2

-0.8

M

0.6

0.2

-1

0.4

0.5 SOC

0.6

0.7

0.8

L2

0.9

1

0.9

1

L3

0.6

0.4

0.2

0

0

0

0.1

0.2

0.3

0.4

0.5 CSOC

0.6

0.7

0.8

0.9

1

0

0.1

0.2

0.3

0.4

0.5 Pref1

EP

TE D

Fig. 11. Membership functions of input and output variables for No.1 FLC

AC C

Degree of membership

PH

RI PT

Z

SC

NL

M AN U

Degree of membership

NH 1

0.6

0.7

0.8

ACCEPTED MANUSCRIPT

NL

PL

PM

PH

L 1

0.8

0.8

0.6

0.4

0.2

-0.4

-0.2

0 Pm2

0.2

0.4

0.6

0.8

M

1

0

NH

H

1

0.8

0.8

0.4

0.2

0.3

0.4

NM

NL

M AN U

Degree of membership

1

0.6

0.1

0.5 SOC

0.6

0.7

SC

-0.6

L

Z

PL

0.8

0.9

PM

1

PH

0.6

0.4

0.2

0.2

0

0 0

0.1

0.2

0.3

0.4

0.5 CSOC

0.6

0.7

0.8

0.9

1

-1

-0.8

-0.6

-0.4

-0.2

0 Pref3

EP

TE D

Fig. 12. Membership functions of input and output variables for No.2 FLC

AC C

Degree of membership

0.4

0 -0.8

H

0.6

0.2

0 -1

M

RI PT

NM

Degree of membership

Degree of membership

NH 1

0.2

0.4

0.6

0.8

1

ACCEPTED MANUSCRIPT 600

400

Power (kW)

200

0

-400

-600

0

50

100

150 Time (s)

200

RI PT

-200

250

AC C

EP

TE D

M AN U

SC

Fig. 13. Driving cycle of the LF-LRV tramway in Turkey

Power (kW)

200

Voltage (V)

800

Efficiency (%) Current (A)

ACCEPTED MANUSCRIPT

300

100 0

0

50

100

150

200

0

50

100

150

200

0

50

100

150

200

0

50

100

150 Time (s)

250

600 500

200 0 100

0

200

M AN U

50

EP

TE D

Fig. 14. Output power, voltage, current and efficiency of PEMFC system

AC C

250

250

SC

100

RI PT

700

250

100 0 -100

50

100

150

200

0

50

100

150

200

0

50

100

150

200

0

50

250

400

80

70

M AN U

75 100

150 Time (s)

200

EP

TE D

Fig. 15. Output power, voltage, current and SOC of battery pack

AC C

250

250

SC

500 0 -500

RI PT

600

SOC (%)

Current (A)

0

800

Voltage (V)

Power (kW)

ACCEPTED MANUSCRIPT

250

ACCEPTED MANUSCRIPT

0

50

100

150

200

0

50

100

150

200

0 100

50

100

150

200

50

100

150 Time (s)

400 300

80 60

200

M AN U

0

EP

TE D

Fig. 16. Output power, voltage, current and efficiency of SC bank

AC C

250

250

SC

Current (A)

1000 500 0 -500

SOC (%)

250

500

RI PT

Voltage (V)

Power (kW)

400 200 0 -200 -400

250

ACCEPTED MANUSCRIPT

900 800 700 600

0

50

100

150

200

0

50

100

150 Time (s)

200

Current (A)

500 0

EP

TE D

M AN U

Fig. 17. DC bus voltage and current

AC C

250

SC

-500

250

RI PT

Voltage (V)

1000

ACCEPTED MANUSCRIPT Highlights A hybrid system model of tramway is developed with commercial devices.




A power sharing strategy based on 2-FLC and Haar-WT is proposed.




An energy management system is evaluated according to real driving cycle.




The results verify the validity of proposed energy management system.

AC C

EP

TE D

M AN U

SC

RI PT