- Email: [email protected]
Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle
international journal of hydrogen energy xxx (xxxx) xxx
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/he
Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle Hanqing Wang a,b,c,*, Arnaud Gaillard a,b,c, Daniel Hissel a,c a
Univ. Bourgogne Franche-Comte, 90010, Belfort Cedex, France University of Technology of Belfort-Montbeliard, 90010, Belfort Cedex, France c FCLAB Research Federation, FR CNRS 3539, FEMTO-ST/Energy Department, UMR CNRS, 6174, France b
article info
abstract
Article history:
Declining reserves of the crude oil and increasingly serious environmental pollution have
Received 27 August 2018
emphasized the requirement of a suitable substitute to our actual petroleum-based auto-
Accepted 31 October 2018
mobile market. An environmentally-friendly and efficient power generation device based
Available online xxx
on a sustainable energy source is attractive to settle this issue and realize cleaner production. Proton Exchange Membrane Fuel Cell (PEMFC), which achieves zero emission,
Keywords:
modular construction, high energy conversion ratio and etc., has been treated as one of the
Online EIS
most promising solution for automobile applications. Nevertheless, many technical re-
PEMFC
strictions such as relatively short life cycle have still to be conquered before satisfying the
DC/DC converter
requirements of large-scale commercialization.
Fault detection Fuel cell electric vehicle
Electrochemical Impedance Spectroscopy (EIS) is an effective technique for fault detection of electrochemical system. This paper presents an on-line EIS detection strategy based on the proposed fuel cell stack connected step-up converter. No additional equipment is required compared with conventional detection process. Furthermore, the proposed 6-phase Interleaved Boost Converter (IBC) based on Silicon Carbide (SiC) semiconductors and inverse coupled inductors has achieved low input current ripple, high efficiency, high voltage gain ratio, high compactness and high redundancy. Benefiting from these advantages, the lifespan of fuel cell stack can be extended. The proposed online EIS detection has been realized and the results have been compared with theoretical analysis. © 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
Introduction Greenhouse gas emission caused by transportation is becoming one of the most serious contaminations for our daily life. Hydrogen energy-based technology is regarding as a promising solution which can not only increase engine efficiency but also decrease pollutant emissions [1]. Fuel Cell
Electric Vehicle (FCEV) is treated as an extremely promising and environmentally friendly technology to change the current transportation mode [2e4]. Proton Exchange Membrane Fuel Cell (PEMFC) is the most suitable type of fuel cells for automotive applications because of their operating temperature range and quick response compared with other types of fuel cells. Some automotive corporations, such as Nissan,
* Corresponding author. FCLAB Research Federation, FR CNRS 3539, FEMTO-ST/Energy Department, UMR CNRS, 6174, France. E-mail address: [email protected] (H. Wang). https://doi.org/10.1016/j.ijhydene.2018.10.242 0360-3199/© 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved. Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242
2
international journal of hydrogen energy xxx (xxxx) xxx
Toyota and Ford, have already proposed several models of FCEV based on PEMFC [5]. So far, general lifespan of PEMFC cannot reach the expectation of commercial application. For example, a typical life expectancy of the PEMFC under actual transportation constraints is around 3000 h, whereas transportation applications require at least 5000 h [6]. Hence, the primary assignment to be solved for universal application of PEMFC is to increase its reliability and durability. Indeed, the manufacture of PEMFC is mainly depending on high cost noble materials (platinum catalyst, acid membrane, etc.) with limited durability especially in unstable operations and cyclic stress operation. Concerning these limitations, various faults can occur to PEMFC during operating condition. Short-circuit, whose response time is relatively short (106~102 s), can lead to membrane and catalyst layer degradation [7]. Starvation, which can cause catalytic layer degradation, has been defined by Yousfi-Steiner et al. [8] as an undersupply of reactants that could occur either at local or global level. CO poisoning owns a relative long response time (101~105 s) and can issue in performance losses and then starvation of PEMFC [9]. Actually, presence of liquid water inside PEMFC can impede performance and durability of the system. Thus, the water management inside cells is one of the key problems needed to be better understood [10]. Under normal operation, the membrane of PEMFC is required to be wet. However, improper operating temperatures, air flow rates, and reactant gases pressure can lead to too little or too much water being injected into the fuel cell [11]. These variations separately result in membrane drying or cathode flooding. Longstanding operation during these two states decreases the output power of the fuel cell. For membrane drying to occur, insufficient water in injected into the cell causing it to operate at a low current density [12]. As the membrane of the cell dries out, the voltage gradually drops. Eventually, the cell dries out and the voltage drops suddenly to zero in a similar manner to a concentration drop [13]. Concerning cathode flooding, it is the major cause of the oxygen mass transport limitation in a PEMFC, especially at high current densities [14]. It is often difficult to remove the product water from the cathode side, which leads to the compromised transfer of oxygen to the reaction sites at the cathode electrode. The operating conditions of PEMFC are essential to be monitored and controlled for the purpose of extending lifespan of power source. Electrochemical impedance spectroscopy (EIS) is established as a powerful characterization tool to detect different failure mechanisms occurring in a fuel cell. Impedance spectra can help to characterize a cell in much more complicated mode than just analyzing the polarization curve [15,16]. In view of its low-voltage high-current characteristic, a DC/ DC converter is essential to be connected to PEMFC system in order to increase the output voltage of fuel cell stack to approximately a few hundred volts to satisfy the DC bus voltage requirement. The ripple frequency of a DC/DC converter is just the same as the switching frequency of the power switching semiconductors such as power MOSFET. This method provides a good demonstration for fuel cell system diagnosis without any other additional equipment to respect the limited space in a FCEV. Narjiss et al. [17] and Depernet et al. [18] have proposed an approach to realize on-line EIS detection functionality based
on the conventional Full-Bridge converter connected to the fuel cell stack. These measurements were based on the current perturbation injection technique. However, only DC bus voltage was controlled by PI controller in these two studies. What is worse, close loop control was impossible during the detection process, which means the stability of DC bus voltage cannot be ensured. Hong et al. [19], Hinaje et al. [20] and Bethoux et al. [21] have also separately realized this objective based on the conventional DC/DC Boost converters. The dualloop PI controllers have been selected by Hong [19] and Bethoux [21] while the PID controller has been selected by Hinaje [20]. According to the study of Bethoux [21], close loop control was realized during the detection process of EIS; the perturbation signal should be injected to the DC bus reference or the input current reference was depending on the frequency of perturbation signal. Nonetheless, towards the application of FCEV, only conventional DC/DC converters such as Boost and FullBridge can be found depending on these references. In light of these mentioned previously, a 6-Phase Interleaved Boost Converter (IBC) is selected in this paper as the power conversion interface of PEMFC for FCEV application. Silicon Carbide (SiC) semiconductors are selected for the purpose to operate at high switching frequency and good thermal performance. Inverse coupled inductors are introduced to achieve a more compact system, which is an important point for automotive applications, and meanwhile, the core losses can be decreased compared with uncoupled inductor. Furthermore, Sliding Mode Control (SMC) is selected to well control the highly non-linear currents of inverse coupled inductors. Meanwhile, DC bus voltage can be well controlled either under nominal condition or diagnosis condition. Small sinusoidal perturbation is superposed with the current reference signal and in this way AC perturbation can be added to fuel cell voltage and current. The detected AC components of fuel cell voltage and current are analyzed by Discrete Fourier Transform (DFT) and the impedance spectrums are obtained. Therefore, EIS has been achieved by the proposed DC/DC converter without additional hardware, cost and volume. The proposed process allows real time using of EIS results for embedded diagnosis of PEMFC. The rest of this paper is organized as follows. In Section EIS fundamental, the equivalent circuit model of fuel cell utilized in our study is described. The illustration of the selected converter and the volume analysis of inverse coupled inductor are separately explained in Section The Proposed DC/DC Boost Converter. In Section The control strategy design process, firstly, the design process of SMC is analyzed; secondly, the integration between EIS detection and control strategy is demonstrated. Some typical simulation results are illustrated in Section The simulation results, and several conclusions are drawn in the last section.
EIS fundamental EIS validity conditions According to the literature, linearity, causality, stability and finiteness are essential qualifications for an effective EIS
Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242
international journal of hydrogen energy xxx (xxxx) xxx
measurement. Herein, linearity and stability are emphatically considered for EIS implementation [22]. A linear system should satisfy the properties of superposition and scaling. To guarantee the linearity, the FC stack should work in a linear or quasi-linear region. In terms of the impedance, the acquired impedance should be irreverent to the amplitude of the perturbation signal when the linearity condition is met. However, certain cautions should be noted: firstly, in order to guarantee a good signal-to-noise ratio, the amplitude of the perturbation should be larger than the system noises; secondly, the amplitude should not be too large not to disturb the stack from stabilization. A suggested range for the AC current's amplitude in Galvanostatic mode is 5% ~10% of the stack DC current. The stability condition requires that the system works in the same stable state before and after the perturbation. The stable operating state is usually characterized by a steadystate stack voltage and stack temperature. The EIS measurements should also be stationary, which means it is timeindependent. This condition can be easily verified by repeating the measurements and check the consistency of the Nyquist plot.
3
transfer through the membrane and water content in the membrane has much influence on proton transfer property. As for the double layer capacitor Cdl, it stands for the capacitance property in the catalyst layer where electrochemical reactions happen. Of course, the resistance Rc stands for the resistance against the charge transfer around the catalyst layer. In addition, the Nernst voltage of the fuel cell is defined as [24]. Eo, PH2, PH2O, and PO2 are the Nernst voltage in standard state, hydrogen partial pressure, water vapor partial pressure and oxygen partial pressure against the standard atmosphere pressure respectively. PH2 P0:5 RT O2 ln En ¼ Eo þ 2F PH2 O
! (1)
For this equivalent circuit, the total impedance can be defined as Zu ¼ Rm þ
Rc jCdl Rc u þ 1
(2)
where u stands for the angular frequency. According to the Eq. (2), the resistance Zu is equal to (Rm þ Rc) approximately when the frequency f is close to zero.
Equivalent circuit model of the fuel cell stack Electrical characteristic of fuel cell can be represented by the equivalent circuit model. The most common equivalent circuit is the Randles circuit proposed by Randles [23]. As presented in Fig. 1, the electrolyte resistance Rm is connected in series with a parallel combination of the double layer capacitor Cdl and the charge transfer resistance Rc. According to principles described in many references, the resistance Rm stands for the resistance against the proton
Fig. 1 e Randles cell.
The proposed DC/DC boost converter According to our previous work [26], a six-phase Interleaved Boost Converter (IBC) based on Silicon Carbide (SiC) semiconductors has been proposed. Meanwhile, inverse coupled inductors based on the cyclic cascade configuration [27] are introduced in order to reduce the total volume and weight of proposed converter; and the output current ripple of the FC can be reduced dramatically, thus, the fuel cell stack's lifespan can be extended [25]. The proposed topology is shown in Fig. 2. Each two windings in the range of yellow block stand for one inverse coupled inductor. Some important specifications of our converter are presented in Table.1. In our study, the input power source is a 21 kW PEM FCvelocity-9SSL module from Ballard® Company [28]. To simplify the study of proposed converter, the Randles model is chosen. The required voltage ratio has been met and high power transformer is not needed. What is more, the current ripple of
Fig. 2 e Six-phase IBC based on inverse coupled inductor for each phase. Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242
4
international journal of hydrogen energy xxx (xxxx) xxx
Table 1 e Specifications of power converter. Symbol Vin PN Vout fs
Parameter
Value
Unit
Input voltage Nominal power Output voltage Switching frequency
70 21 350 100
V kW V kHz
power sources has been reduced significantly by this topology. This is one of the most interesting factors to extend power source lifespan. SiC semiconductors are introduced to satisfy the high switching frequency, and to obtain a preferable thermal performance. Low switching losses and negligible reverse recovery losses are separately achieved by SiC MOSFET and SiC Schottky diode. Thus, the performance and power density of the system can be optimized and the efficiency analysis can be achieved by the previous study [26]. In this study, the inductors are inversely coupled and constructed by a cyclic cascade structure. In other words, each phase is just coupled with the adjacent phase and on the whole six coupled inductors are needed for our application. Taking the volume, weight, complexity of installation and other factors into consideration, EE magnetic cores are selected. The physical structure and electrical structure of inverse coupled magnetic core are separately presented in Fig. 3(a)e(b). Due to the inverse coupled structure, DC magnetic fluxes caused by the two windings flow reversely and total DC magnetic flux is thus almost zero inside the core. The core losses mainly depend on the AC magnetic flux, which is a relatively small value due to the low current ripple, thus the volume of inductor can be decreased. The inductor volume is closely related to inductor value, current RMS value, current ripple level and maximum flux density. According to [29], the effective electrical size of magnetic cores can be described by the core geometrical constant Kgmin. Eq. (3) is utilized to determine the minimum core geometrical constant Kgmin of the inverse coupled inductor core; and the minimum core geometrical constant Kgmin_uncoupled is obtained by Eq. (4). 2
Kgmin
Kgmin
coupled
¼
uncoupled
rcu L2M iM;sat ðIL1 þ IL2 Þ2 B2sat ¼
Ku Pcu
(3)
tot
rcu L2 I2Lmax B2sat Ku RL
Fig. 3 e The geometric structure of inverse coupled inductor.
(4)
The wire effective resistivity rcu is 1.724 108 U,m. The magnetizing inductance LM is 80 106 H. The maximum magnetizing current iM,sat is 1.75A. The DC component of each winding's current, IL1 and IL2, is 50A. The saturation flux density Bsat is 0.4T. Total copper losses Pcu_tot are 50W. The winding fill factor Ku is 0.6. Fig. 4 presents the relationship between number of interleaved phases (N) and minimum magnetic core geometrical constant (Kgmin) separately for uncoupled and inverse coupled inductors. Obviously, Kgmin_coupled is much smaller than Kgmin_uncoupled. Additionally, the minimum core geometric constant decreases while the number of interleaved phases increases with both designs.
The control strategy design process The design of sliding-mode control strategy The control objectives are necessary to be established for the purpose of defining the control strategy: 1) Output voltage regulation under load uncertainty. To avoid load damages, it is necessary to regulate the DC bus voltage and maintain it constantly. 2) Equal current sharing between each phase of coupled inductors. The current waveforms should be equal in order to avoid overloading one of the phases, especially when supplying heavy loads. Also the currents must be interleaved in order to reduce the current ripple which is undesirable in fuel cell stack. 3) Asymptotic stability of the closed loop system. Global asymptotic stability is required to avoid imposing restrictions on the allowed initial conditions. A Proportional-Integral (PI) controller is well known and widely applied in industrial application fields to optimize the steady-state performance. The proportional term contributes to reduce the current error value while the integral term accelerates the movement of the process towards set point and eliminates the residual steady-state error that occurs with a pure proportional controller [30]. Nonetheless, an obvious drawback faced with PI controller is that it is linear. Hence, the performance of PI controller in non-linear system is variable. When we design a PI compensator for a Boost converter, it relies on the small signal model of the converter. In order to utilize small signal model, the premise is that the converter operates under static work point. In such a condition, AC small signal analysis can be preceded. Unfortunately, the classical PI control strategy cannot satisfy the highly non-linear characteristic of inverse coupled inductor, especially when dynamic change occurs. The unifications of inductor currents are strictly required in our study to guarantee the magnetic components operating under nonsaturation zone. This requirement is needed to be satisfied whether under static work point or dynamic processes. Further, the operating processes of coupled inductor are strongly related to duty cycle and number of interleaved phases. More interleaved branches can lead to complicated operating process. Meanwhile, duty cycle, which is closely
Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242
5
international journal of hydrogen energy xxx (xxxx) xxx
Fig. 4 e (a) Stands for the relationship between N and Kg for both uncoupled and inverse coupled inductors; (b) stands for the zoom in on the red line of (a) for inverse coupled inductor.
related to load level, can influence the regulating process of inductor currents significantly. Supposing significant differences occurring between inductor currents, magnetic saturation could be introduced. In other words, well controlled inductor currents can lead to desirable performance of magnetic components. Hence, a control strategy which can achieve a zero steady-state error and be insensitive to disturbance is attractive to this study. Sliding Mode Control (SMC), which has been proposed by [31], is theoretically robust to the plant parameter variations and quite suitable for our requirement. More importantly, the design process of SMC depends on the large signal model of DC/DC converter. Its stability is not restricted by variations around operating point, which contributes to an overall improved controller performance. The sliding surface in our study is defined by the following expression. SLi ¼ ILi ILi
ref
Zt þ KLi
ILi ILi
ref
dt
(5)
0
In Eq. (5), i is for the sequence number, ILi stands for the average current of each inductor and ILi_ref represents the reference current. KLi, which is positive, defines the dynamic of convergence to zero of the static error. The Control-Lyapunov function is used to test whether a system is feedback stabilizable [32]. A characteristic method is applying a Lyapunov candidate function to test the stability of dynamic system. Hence, ε, which equals to ILi minus ILi_ref, is defined as the error between actual and desired inductor current. Thus, Eq. (5) can be re-written as Zt SLi ¼ ε þ KLi
εdt
(6)
A Control-Lyapunov candidate is presented as Eq. (7), which is positive definite for all ILi_refs0, d(ILi_ref)/dts0. 1 V ¼ S2 2
(7)
The purpose is to achieve the time derivative to be ·
V ¼ lV; l > 0
(8)
which is globally exponentially stable if V is globally definite. The convergence dynamic of the sliding surfaces to zero is defined as Eq. (9), which can be proved by Eqs. (6)e(8). This control law will guarantee global exponential stability. S0Li ¼ lLi SLi
(9)
The convergence factors (lLi) are positive real numbers. A larger convergence factor can lead the system to reach its steady state faster. However, it can only be selected in a certain range due to the limitations of the system. The state space model of proposed converter is essential to design the control strategy, and it can be obtained by Eq. (10) according to the operating process. Depending on the equations mentioned previously, Eq. (11) can be achieved which represents the duty cycle of each phase as a function of time. d ½ILi ¼ ½L1 ð½Vin ½1 Di Vo ½iLi rL Þ dt ½Di ðtÞ ¼ 1
h 1 ½Vin þ rL ½ILi þ ½LA lLi Si þ I0Li Vo i
(10)
ref
KLi
εi
(11) The inverse coupled inductor matrix and the similar matrix of its inverse matrix are separately presented by [L] and [LA] as follow. Depending on Eq. (11), the control inputs are independent and unaffected by the disturbance of loads.
0
Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242
6
international journal of hydrogen energy xxx (xxxx) xxx
2
2L 6 M 6 6 0 ½L ¼ 6 6 0 6 4 0 M 2 1
M 2L M 0 0 0
6 2ðL MÞ 6 6 6 0 6 6 6 6 6 6 0 6 ¼6 6 6 0 6 6 6 6 6 0 6 6 4 0
0 0 M 0 2L M M 2L 0 M 0 0
0 0 0 M 2L M
3 M 0 7 7 0 7 7 ½LA 0 7 7 M 5 2L 66
3
0
0
0
0
0
1 2ðL þ MÞ
0
0
0
0
0
1 2L M
0
0
0
0
0
1 2L M
0
0
0
0
0
1 2L þ M
0
0
0
0
0
1 2L þ M
7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 5
Combining Eqs. (5),(9) and (11) together, we can obtain: ε0 þ ε ðlLi þ KLi Þ þ lLi KLi
Zt εdt ¼ 0
(12)
0
The time derivative of Eq. (12) is 00
ε þ ε0 ðlLi þ KLi Þ þ lLi KLi ε ¼ 0
(13)
A high robustness for the controller can be achieved because this equation is irrelevant with the topology parameters. In order to achieve the desired performance, the coefficients KLi and the convergence factors lLi can be determined by the classical tuning method for second-order systems.
The integration of EIS with SMC In consideration of the limited effective volume of FCEV, the integration of EIS functionality in control strategy of converter is a promising technique in purpose of realizing on-line monitoring the state of health of FC [22]. The strategy of EIS integration in SMC consists in injecting the current perturbation Isin around the polarization current IFC as presented in Fig. 5(b). In Fig. 5, a delay (Tdelay) existed between neighbor phases drive signals (PWM Phase N). This value depends on number of phases (N) and cycle period (Ts). In our case, N and
Ts separately equal to 6 and 10 ms. Thus, Tdelay can be achieved as (10/6) us. At nominal condition, DC bus voltage (Vo) of converter will be compared with the reference voltage (Vref), and then the difference will be transferred to a PI controller (out loop controller) to obtain the current reference (IL_ref) of each phase. After that, IL_ref, fuel cell voltage (VFC), fuel cell current (IFC) and inductor current (IL) of each phase will be sent into the SMC (inner loop controller). After the calculation process, PWM signals can be obtained and will be sent to corresponding power MOSFET as drive signal. On-line detection of EIS based on the proposed DC/DC converter is presented by Fig. 6. The process can only be handled when fuel cell stack operates at stable state, which is usually characterized by a steady-state stack voltage and stack temperature [22]. When the EIS monitoring is required, a sinusoidal perturbation will be superimposed to the output signal of PI controller (IL_ref). According to the process of SMC, the current of each phase will track this reference value. Thus, the sinusoidal perturbation can be superimposed successfully to the fuel cell current. Owing to operate at linear range, the fuel cell voltage will also be perturbed by a sinusoidal signal combined with the same frequency as the current perturbation. Under this condition, the mean values of the fuel cell stack output voltage and current can be obtained separately. These two parameters are essential for the measurement of AC impedance. Meanwhile, benefiting from the high robustness of this control strategy, the output voltage of DC/DC boost converter can still be maintained relatively stable at the required value (<7%). In order to illustrate the principle of AC impedance measurement, hence, the sampling process of fuel cell stack current is taken as an example. The sampling processes both of VFC and IFC are shown in Fig. 7. The difference between fuel cell current mean value and perturbative fuel cell current is an AC component. The AC component is composed by the fuel cell current ripple under steady state and a sinusoidal perturbation. The frequency of the fuel cell current ripple equals to switching frequency. Thus, the mean value of the AC component in each switching period can be calculated. A sinusoidal wave based on the same frequency as the perturbation signal, can be achieved. This signal is the exact extracted signal, iFC_AC, which is caused by the injected sinusoidal perturbation, Isin. Perturbations combined with different frequencies can let different responses of fuel cell current and corresponding responses of
Fig. 5 e The strategy of EIS integration in SMC.
Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242
international journal of hydrogen energy xxx (xxxx) xxx
8 N1 X > 2p > > ReðI nk Þ ¼ I ðnÞ cos > FC FC > < N n¼0
; k ¼ 0; 1; //; N 1 > N1 X > 2p > > IFC ðnÞ sin nk > : ImðIFC Þ ¼ N
7
(15)
n¼0
8 VFC ReðVFC Þ þ j$ImðVFC Þ > > > ZFC ¼ I ¼ ReðI Þ þ j,ImðI Þ ¼ Re ZFC þ j,Im ZFC > FC FC FC > > > > < ReðV Þ ReðI Þ þ ImðV Þ ImðI Þ FC FC FC FC Re ZFC ¼ > Re2 ðIFC Þ þ Im2 ðIFC Þ > > > > > > > ImZ ¼ ImðVFC Þ ReðIFC Þ ReðVFC Þ ImðIFC Þ : FC Re2 ðIFC Þ þ Im2 ðIFC Þ
(16)
Normally, there is no limitation for the minimum frequency of the perturbation signal, but the maximum frequency shall be smaller than fS/N. N stands for the number of samples and it is selected as 10 in our case in consideration with the sampling precision [18]. An impedance spectrum can be formed by a series of the stack impedances at discrete frequency points within a wide range of frequencies. The proposed converter, which is based on SiC semiconductors, makes it possible to obtain impedances at high frequency due to the ability of operating at high switching frequency. The frequency of stimulus signal is here designed from 1 Hz to 10 kHz.
The simulation results
Fig. 6 e The process flow diagram of on-line detection of EIS based on the proposed DC/DC converter.
PLECS, which is a software tool and especially designed for system-level simulations of power electronics, is utilized in our study to verify the proposed EIS monitoring functionality. The simulation results of proposed DC/DC converter and EIS achievement are separately presented in the following sections.
Electrical part simulation results fuel cell voltage. These independent responses can be utilized to plot impedance spectrums. Discrete Fourier Transform (DFT) is utilized to separate the real part and imaginary part of iFC_AC. The sampling frequency (fe) is set at twenty times the value of the perturbation signal frequency. The number of samples (N) is chosen as 10. In this study, the highest order harmonic is selected as ten-time harmonic which is enough for our calculation and analysis. The real part and imaginary part will be utilized to calculate the AC impedance of the fuel cell stack. With regards to the sampling process of the fuel cell stack voltage under perturbed conditions, the principle is the same as illustrated in foregoing paragraphs. Combining DFT and Euler's formula, the impedance of FC stack can be expressed as Eqs. (14)e(16). 8 N1 X > 2p > > VFC ðnÞ cos nk > ReðVFC Þ ¼ > < N n¼0
; k ¼ 0; 1; //; N 1 > N1 X > 2p > > VFC ðnÞ sin nk > : ImðVFC Þ ¼ N n¼0
(14)
The required specifications are obligatory to be achieved both under nominal condition and disturbed condition. Naturally, nominal condition is tested firstly to verify the stability of the control strategy. Fig. 8 presents the simulation results of the proposed converter under nominal condition. Visibly, both loading and unloading tests are realized. Due to the high robust characteristics of SMC, the proposed DC/DC converter operates at steady state. DC bus voltage is stabilized at 350V. DC components of inductor current are totally maintained to identical values. AC components are successfully regulated in a small range. The saturation of inductors can indeed be avoided. Meanwhile, the fuel cell current is likewise well controlled at the linearity range. Especially, due to the interleaved structure of the power converter and the use of coupled inductor, the current ripple of fuel cell stack is rather small, which is effective for the purpose of extending fuel cell lifespan. Hereby, in purpose of verifying the proposed detection strategy, the perturbation signal which possess the frequency of 10 kHz was injected to the inductor current reference and the results were presented by Fig. 9. Obviously, the sinusoid
Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242
8
international journal of hydrogen energy xxx (xxxx) xxx
Fig. 7 e The processes of fuel cell voltage and fuel cell current sampling in purpose of the realization of EIS functionality. From (b) to (g) are the zooms of typical signals, either fuel cell voltage or current, which are essential during the procedures of the proposed strategy. A sinusoidal perturbation combined with frequency of 100Hz and amplitude of ±1A is taken as an example in these figures.
perturbation has been successfully integrated with the fuel cell current and voltage. To be emphasized, close loop control has been achieved even during the process of perturbation injection. DC bus voltage has been well controlled and stabilized at the required value (Vbus_DC ¼ 350V, △Vbus<4%).
EIS detection using the proposed DC/DC converter As analyzed previously, when an AC signal of known amplitude and frequency is sent to the inductor current reference of proposed DC/DC converter, the amplitude responses of the fuel cell stack's voltage and current can be recorded separately. This process can be repeated at different frequencies, thus, a Nyquist impedance spectrum can be formed by a series
of the stack impedances at discrete frequency points to achieve more information about the cell parameters. Under nominal condition, Nyquist plot is obtained as presented by Fig. 10. Due to the characteristics of Randles model, only one arc can be exhibited in Nyquist plot. Hence, three feature points, Rm, Rc and 4, which reflect the operating condition of fuel cell stack, have been illustrated as follow [22]. Rm, which is the left intersection of the curve and real axis, indicates the total Ohmic resistances of fuel cell stack. The humidification degree of the membrane can be reflected by the parameter. Rc, which can be achieved when the perturbation frequency approaches zero and it is actually the low frequency
Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242
international journal of hydrogen energy xxx (xxxx) xxx
9
Fig. 8 e Normal condition without perturbation signal.
Fig. 9 e Injected small disturbance small signal combined with frequency of 10 kHz and amplitude of 1A.
intercept of the impedance arc. The global performance of fuel cell stack can be represented by it. 4, which corresponds to the maximal phase in the Phase plot and indicate the size of low frequency diffusion arc. A bigger 4 represents a large arc diameter and reflects a difficult mass transport process. In Fig.11.(a), the red curve and the yellow curve separately represent simulation results and theoretical calculation results of impedance spectrum at nominal condition. The simulation results are identical with theoretical analyses. The
proposed strategy is verified under nominal operating conditions based on Randles model. In order to verify the proposed detection method, some parameters of fuel cell model are modified to simulate different faults occurring on the fuel cell stack. Water management inside fuel cell stack is essential to avoid breakdown or damage occurred to power source. Different types of fault (flooding or drying) can be indicated by different impedance spectrums as a result of variations of fuel cell model parameters [33]. Therefore, according to visible variation introduced
Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242
10
international journal of hydrogen energy xxx (xxxx) xxx
to impedance curves when these faults are brought into existence, EIS is an effective diagnosis tool for these situations. Table.2 presents different fuel cell parameters utilized for different operating states. These parameters are man-set values and just utilized in purpose of verifying the validity of proposed strategy. 1) Simulating membrane drying (Rm increased)
Fig. 10 e Typical EIS of Randles model under normal operating condition.
Drying can be caused if the membrane is not sufficiently hydrated, or the inlet gases are insufficiently humidified. The inlet gases’ temperature is below the fuel cell operating one can be another reason. Drying out can lead to an increase of membrane resistance. Hence, this fault can be detectable by any increase of cell resistance [34]. Rm is replaced by a bigger value to simulate drying fault occurring to the fuel cell stack. Both theoretical analyses and simulation results are presented in Fig. 11.(b). Distinctly, the impedance spectrums achieved by simulation are visibly different from nominal condition. In Nyquist plot, the left intersection of the curve and real axis moves to the right of positive axis which indicates the increase of Rm, or the occurring of drying fault. The waveform doesn't change, which means the biggest imaginary value is the same as in nominal condition. 2) Simulating cell flooding (Rc increased) Flooding, which is one of the most recurrent PEMFC's faults, is defined as an accumulation of liquid water in the gas channels or electrodes, impeding the access of reactive gases to the active layers, and therefore decreasing the reaction rate [14]. The flooding probability increases the difficulty of mass transport process. The lower the O2 diffusion rate is, the more severe the flooding will be. Difficult mass transport leads to an arc owing a bigger diameter in Nyquist plot. Rc is replaced by a bigger value to imitate flooding fault occurred to fuel cell stack. Both theoretical analyses and simulation results are presented in Fig.11.(c). Obviously, the impedance spectrums obtained by simulation are visibly different from nominal condition. In Nyquist plot, due to the untouched membrane resistance, the left intersections of the curves and real axis are the same. An arc combined with a bigger diameter is achieved by the flooding conditions. Hence, according to the comparison analyses between theoretical and simulated results, the proposed strategy,
Fig. 11 e Electrochemical Impedance Spectrums of PEMFC under normal operating condition, drying faults and flooding faults are separately by (a), (b) and (c). EIS, one obtained by simulation results from PLECS, the other obtained by calculation results based on Randles model and parameters in Table.2, are presented to be compared.
Table 2 e PEMFC parameters for the assumption of different man-set faults to verify proposed EIS detection functionality. State of Health (SOH) Normal Drying Flooding
Rm/mOhms
Rc/mOhms
Cdl/F
5.58 8 5.58
15.46 15.46 50
1.37 1.37 1.37
Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242
international journal of hydrogen energy xxx (xxxx) xxx
which focuses on integration of EIS functionality with slidingmode control on the considered DC/DC power converter, is effective to be applied for monitoring the state of health of a fuel cell stack. Both drying and flooding modes are detectable from the achieved impedance spectrums. On-line diagnosis of the fuel cell stack is realized; meanwhile DC bus voltage is well regulated even during detection procedure. To be emphasized, the proposed on-line diagnosis approach is accomplished without any additional equipment which is highly attractive for practical FCEV applications due to the limited space inside a vehicle.
Conclusion This work concentrates on the realization of on-line detection of impedance spectroscopy for PEMFC application based on connected electric power converter. On-line EIS can present real-time frequency-impedance information regarding the state of health that can be used in the diagnosis or control purposes. The proposed converter based on high switching frequency, SiC semiconductors and inverse coupled inductors is an innovative solution to settle the problem of regulating PEMFC voltage to satisfy the voltage requirement of the fuel cell electric vehicle DC bus. The selection of interleaved structure is meaningful for the reduction of the FC current ripple and the extension of the FC lifespan. Thanks to the selection of SiC semiconductors and coupled inductor technique, the total volume and weight of the power converter have been reduced significantly. The main advantages of this work are the full integration of EIS detection algorithms in the power converter control strategy. No additional equipment required and the cost of EIS implementation is negligible, thus, the integration constraints are minimized. Compared with the existing studies, the proposed strategy has been verified by FC stack Randles model in a wide range of frequencies (maximum 10 kHz). Besides, the selected SlidingMode Control can well regulate the fuel cell current and DC bus voltage, and realize close loop control either under nominal operating conditions or disturbed conditions. In summary, on-line fuel cell impedance acquisition and power conversion have been realized by the proposed detection strategy and power converter. Hence, the efficiency, reliability and lifespan of PEMFC can be guaranteed further according to our study which is significant in practical FCEV application.
references
[1] Hissel D, Pera MC. Diagnostic & health management of fuel cell systems: issues and solutions. Annu Rev Contr 2016;42:201e11. € gel P, Oltra C, Sala R, Lores M, Upham P, Du¨tschke E,, [2] Bo Wiemann P. The role of attitudes in technology acceptance management: reflections on the case of hydrogen fuel cells in Europe. J Clean Prod 2018;188:125e35. [3] Cherubini S, Iasevoli G, Michelini L. Product-service systems in the electric car industry: critical success factors in marketing. J Clean Prod 2015;97:40e9.
11
[4] Dalla Chiara B, Pellicelli M. Sustainable road transport from the energy and modern society points of view: perspectives for the automotive industry and production. J Clean Prod 2016;133:1283e301. [5] Yoshida T, Kojima K. Toyota MIRAI fuel cell vehicle and progress toward a future hydrogen society. The Electrochemical Society Interface 2015;24(2):45e9. [6] Sutharssan T, Montalvao D, Chen YK, Wang WC, Pisac C, Elemara H. A review on prognostics and health monitoring of proton exchange membrane fuel cell. Renew Sustain Energy Rev 2017;75:440e50. [7] Silva RE, Harel F, Jemei S, Gouriveau R, Hissel D, Boulon L, et al. Proton exchange membrane fuel cell operation and degradation in short-circuit. Fuel Cell 2014;14(6):894e905. [8] Jia F, Guo L, Liu H. Mitigation strategies for hydrogen starvation under dynamic loading in proton exchange membrane fuel cells. Energy Convers Manag 2017;139:175e81. [9] Zhang C, Zhou W, Ehteshami MM, Wang Y, Chan SH. Determination of the optimal operating temperature range for high temperature PEM fuel cell considering its performance, CO tolerance and degradation. Energy Convers Manag 2015;105:433e41. [10] Jian-hua T, Zhao-yuan S, Jin-song S, Zhong-qiang S. Preparation of water management layer and effects of its composition on performance of PEMFCs. Energy Convers Manag 2008;49(6):1500e5. [11] Shyu JC, Hsueh KL, Tsau F. Performance of proton exchange membrane fuel cells at elevated temperature. Energy Convers Manag 2011;52(12):3415e24. [12] Barbir F, Gorgun H, Wang X. Relationship between pressure drop and cell resistance as a diagnostic tool for PEM fuel cells. J Power Sources 2005;141(1):96e101. [13] Gerard M, Poirot-Crouvezier JP, Hissel D, Pera MC. Oxygen starvation analysis during air feeding faults in PEMFC. Int J Hydrogen Energy 2010;35(22):12295e307. [14] Santarelli MG, Torchio MF. Experimental analysis of the effects of the operating variables on the performance of a single PEMFC. Energy Convers Manag 2007;48(1):40e51. [15] Becherif M, Pera MC, Hissel D, Zheng Z. Determination of the health state of fuel cell vehicle for a clean transportation. J Clean Prod 2018;171:1510e9. [16] Swierczynski M, Stroe DI, Stanciu T, Kær SK. Electrothermal impedance spectroscopy as a cost efficient method for determining thermal parameters of lithium ion batteries: prospects, measurement methods and the state of knowledge. J Clean Prod 2017;155:63e71. [17] Narjiss A, Depernet D, Candusso D, Gustin F, Hissel D. Online diagnosis of PEM fuel cell. In: Power Electronics and Motion Control Conference, 2008. EPE-PEMC 2008. 13th; 2008, September. p. 734e9. IEEE. ra MC. [18] Depernet D, Narjiss A, Gustin F, Hissel D, Pe Integration of electrochemical impedance spectroscopy functionality in proton exchange membrane fuel cell power converter. Int J Hydrogen Energy 2016;41(11):5378e88. [19] Hong P, Xu L, Jiang H, Li J, Ouyang M. A new approach to online AC impedance measurement at high frequency of PEM fuel cell stack. Int J Hydrogen Energy 2017;42(30):19156e69. €l S, Davat B. [20] Hinaje M, Sadli I, Martin JP, Thounthong P, Rae Online humidification diagnosis of a PEMFC using a static DCeDC converter. Int J Hydrogen Energy 2009;34(6):2718e23. [21] Bethoux O, Hilairet M, Azib T. A new on-line state-of-health monitoring technique dedicated to PEM fuel cell. In: Industrial Electronics, 2009. IECON'09. 35th Annual conference of IEEE. IEEE; 2009, November. p. 2745e50. [22] Zheng Z. On line fault diagnosis of PEMFC stacks via on nonmodel based methods and EIS measurements. 2014.
Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242
12
international journal of hydrogen energy xxx (xxxx) xxx
[23] Fouquet N, Doulet C, Nouillant C, Dauphin-Tanguy G, OuldBouamama B. Model based PEM fuel cell state-of-health monitoring via ac impedance measurements. J Power Sources 2006;159(2):905e13. [24] Larminie J, Dicks A, McDonald MS. Fuel cell systems explained, vol. 2. Chichester, UK: J. Wiley; 2003. [25] Wahdame B, Girardot L, Hissel D, Harel F, Franc¸ois X, Candusso D,, Dumercy L. Impact of power converter current ripple on the durability of a fuel cell stack. In: Industrial Electronics, 2008. ISIE 2008. IEEE International Symposium on. IEEE; 2008, June. p. 1495e500. [26] WANG, H., GAILLARD, A., & HISSEL, D. Six-phase softswitching interleaved boost converter based on SiC semiconductor and coupled inductor for fuel cell vehicles. [27] Park IG, Kim SI. Modeling and analysis of multi-interphase transformers for connecting power converters in parallel. Power electronics specialists conference, 1997. PESC'97 record., 28th annual IEEE, vol. 2. IEEE; 1997, June. p. 1164e70. [28] De Bernardinis A. Current-fed inverter topologies and control strategy applied to modular power fuel cells in transportation applications. In: Transportation
[29] [30]
[31]
[32]
[33]
[34]
electrification conference and expo (ITEC), 2015 IEEE. IEEE; 2015, June. p. 1e6. Erickson RW, Maksimovic D. Fundamentals of power electronics. Springer Science & Business Media; 2007. Ang KH, Chong G, Li Y. PID control system analysis, design, and technology. IEEE Trans Contr Syst Technol 2005;13(4):559e76. Chen TL, Wu YC. An optimal variable structure control with integral compensation for electrohydraulic position servo control systems. IEEE Trans Ind Electron 1992;39(5):460e3. Freeman R, Kokotovic PV. Robust nonlinear control design: state-space and Lyapunov techniques. Springer Science & Business Media; 2008. ra MC, Pe rez BG. A review Dijoux E, Steiner NY, Benne M, Pe of fault tolerant control strategies applied to proton exchange membrane fuel cell systems. J Power Sources 2017;359:119e33. guy P, Candusso D. Diagnosis of Steiner NY, Hissel D, Moc¸ote polymer electrolyte fuel cells failure modes (flooding & drying out) by neural networks modeling. Int J Hydrogen Energy 2011;36(4):3067e75.
Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/he
Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle Hanqing Wang a,b,c,*, Arnaud Gaillard a,b,c, Daniel Hissel a,c a
Univ. Bourgogne Franche-Comte, 90010, Belfort Cedex, France University of Technology of Belfort-Montbeliard, 90010, Belfort Cedex, France c FCLAB Research Federation, FR CNRS 3539, FEMTO-ST/Energy Department, UMR CNRS, 6174, France b
article info
abstract
Article history:
Declining reserves of the crude oil and increasingly serious environmental pollution have
Received 27 August 2018
emphasized the requirement of a suitable substitute to our actual petroleum-based auto-
Accepted 31 October 2018
mobile market. An environmentally-friendly and efficient power generation device based
Available online xxx
on a sustainable energy source is attractive to settle this issue and realize cleaner production. Proton Exchange Membrane Fuel Cell (PEMFC), which achieves zero emission,
Keywords:
modular construction, high energy conversion ratio and etc., has been treated as one of the
Online EIS
most promising solution for automobile applications. Nevertheless, many technical re-
PEMFC
strictions such as relatively short life cycle have still to be conquered before satisfying the
DC/DC converter
requirements of large-scale commercialization.
Fault detection Fuel cell electric vehicle
Electrochemical Impedance Spectroscopy (EIS) is an effective technique for fault detection of electrochemical system. This paper presents an on-line EIS detection strategy based on the proposed fuel cell stack connected step-up converter. No additional equipment is required compared with conventional detection process. Furthermore, the proposed 6-phase Interleaved Boost Converter (IBC) based on Silicon Carbide (SiC) semiconductors and inverse coupled inductors has achieved low input current ripple, high efficiency, high voltage gain ratio, high compactness and high redundancy. Benefiting from these advantages, the lifespan of fuel cell stack can be extended. The proposed online EIS detection has been realized and the results have been compared with theoretical analysis. © 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
Introduction Greenhouse gas emission caused by transportation is becoming one of the most serious contaminations for our daily life. Hydrogen energy-based technology is regarding as a promising solution which can not only increase engine efficiency but also decrease pollutant emissions [1]. Fuel Cell
Electric Vehicle (FCEV) is treated as an extremely promising and environmentally friendly technology to change the current transportation mode [2e4]. Proton Exchange Membrane Fuel Cell (PEMFC) is the most suitable type of fuel cells for automotive applications because of their operating temperature range and quick response compared with other types of fuel cells. Some automotive corporations, such as Nissan,
* Corresponding author. FCLAB Research Federation, FR CNRS 3539, FEMTO-ST/Energy Department, UMR CNRS, 6174, France. E-mail address: [email protected] (H. Wang). https://doi.org/10.1016/j.ijhydene.2018.10.242 0360-3199/© 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved. Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242
2
international journal of hydrogen energy xxx (xxxx) xxx
Toyota and Ford, have already proposed several models of FCEV based on PEMFC [5]. So far, general lifespan of PEMFC cannot reach the expectation of commercial application. For example, a typical life expectancy of the PEMFC under actual transportation constraints is around 3000 h, whereas transportation applications require at least 5000 h [6]. Hence, the primary assignment to be solved for universal application of PEMFC is to increase its reliability and durability. Indeed, the manufacture of PEMFC is mainly depending on high cost noble materials (platinum catalyst, acid membrane, etc.) with limited durability especially in unstable operations and cyclic stress operation. Concerning these limitations, various faults can occur to PEMFC during operating condition. Short-circuit, whose response time is relatively short (106~102 s), can lead to membrane and catalyst layer degradation [7]. Starvation, which can cause catalytic layer degradation, has been defined by Yousfi-Steiner et al. [8] as an undersupply of reactants that could occur either at local or global level. CO poisoning owns a relative long response time (101~105 s) and can issue in performance losses and then starvation of PEMFC [9]. Actually, presence of liquid water inside PEMFC can impede performance and durability of the system. Thus, the water management inside cells is one of the key problems needed to be better understood [10]. Under normal operation, the membrane of PEMFC is required to be wet. However, improper operating temperatures, air flow rates, and reactant gases pressure can lead to too little or too much water being injected into the fuel cell [11]. These variations separately result in membrane drying or cathode flooding. Longstanding operation during these two states decreases the output power of the fuel cell. For membrane drying to occur, insufficient water in injected into the cell causing it to operate at a low current density [12]. As the membrane of the cell dries out, the voltage gradually drops. Eventually, the cell dries out and the voltage drops suddenly to zero in a similar manner to a concentration drop [13]. Concerning cathode flooding, it is the major cause of the oxygen mass transport limitation in a PEMFC, especially at high current densities [14]. It is often difficult to remove the product water from the cathode side, which leads to the compromised transfer of oxygen to the reaction sites at the cathode electrode. The operating conditions of PEMFC are essential to be monitored and controlled for the purpose of extending lifespan of power source. Electrochemical impedance spectroscopy (EIS) is established as a powerful characterization tool to detect different failure mechanisms occurring in a fuel cell. Impedance spectra can help to characterize a cell in much more complicated mode than just analyzing the polarization curve [15,16]. In view of its low-voltage high-current characteristic, a DC/ DC converter is essential to be connected to PEMFC system in order to increase the output voltage of fuel cell stack to approximately a few hundred volts to satisfy the DC bus voltage requirement. The ripple frequency of a DC/DC converter is just the same as the switching frequency of the power switching semiconductors such as power MOSFET. This method provides a good demonstration for fuel cell system diagnosis without any other additional equipment to respect the limited space in a FCEV. Narjiss et al. [17] and Depernet et al. [18] have proposed an approach to realize on-line EIS detection functionality based
on the conventional Full-Bridge converter connected to the fuel cell stack. These measurements were based on the current perturbation injection technique. However, only DC bus voltage was controlled by PI controller in these two studies. What is worse, close loop control was impossible during the detection process, which means the stability of DC bus voltage cannot be ensured. Hong et al. [19], Hinaje et al. [20] and Bethoux et al. [21] have also separately realized this objective based on the conventional DC/DC Boost converters. The dualloop PI controllers have been selected by Hong [19] and Bethoux [21] while the PID controller has been selected by Hinaje [20]. According to the study of Bethoux [21], close loop control was realized during the detection process of EIS; the perturbation signal should be injected to the DC bus reference or the input current reference was depending on the frequency of perturbation signal. Nonetheless, towards the application of FCEV, only conventional DC/DC converters such as Boost and FullBridge can be found depending on these references. In light of these mentioned previously, a 6-Phase Interleaved Boost Converter (IBC) is selected in this paper as the power conversion interface of PEMFC for FCEV application. Silicon Carbide (SiC) semiconductors are selected for the purpose to operate at high switching frequency and good thermal performance. Inverse coupled inductors are introduced to achieve a more compact system, which is an important point for automotive applications, and meanwhile, the core losses can be decreased compared with uncoupled inductor. Furthermore, Sliding Mode Control (SMC) is selected to well control the highly non-linear currents of inverse coupled inductors. Meanwhile, DC bus voltage can be well controlled either under nominal condition or diagnosis condition. Small sinusoidal perturbation is superposed with the current reference signal and in this way AC perturbation can be added to fuel cell voltage and current. The detected AC components of fuel cell voltage and current are analyzed by Discrete Fourier Transform (DFT) and the impedance spectrums are obtained. Therefore, EIS has been achieved by the proposed DC/DC converter without additional hardware, cost and volume. The proposed process allows real time using of EIS results for embedded diagnosis of PEMFC. The rest of this paper is organized as follows. In Section EIS fundamental, the equivalent circuit model of fuel cell utilized in our study is described. The illustration of the selected converter and the volume analysis of inverse coupled inductor are separately explained in Section The Proposed DC/DC Boost Converter. In Section The control strategy design process, firstly, the design process of SMC is analyzed; secondly, the integration between EIS detection and control strategy is demonstrated. Some typical simulation results are illustrated in Section The simulation results, and several conclusions are drawn in the last section.
EIS fundamental EIS validity conditions According to the literature, linearity, causality, stability and finiteness are essential qualifications for an effective EIS
Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242
international journal of hydrogen energy xxx (xxxx) xxx
measurement. Herein, linearity and stability are emphatically considered for EIS implementation [22]. A linear system should satisfy the properties of superposition and scaling. To guarantee the linearity, the FC stack should work in a linear or quasi-linear region. In terms of the impedance, the acquired impedance should be irreverent to the amplitude of the perturbation signal when the linearity condition is met. However, certain cautions should be noted: firstly, in order to guarantee a good signal-to-noise ratio, the amplitude of the perturbation should be larger than the system noises; secondly, the amplitude should not be too large not to disturb the stack from stabilization. A suggested range for the AC current's amplitude in Galvanostatic mode is 5% ~10% of the stack DC current. The stability condition requires that the system works in the same stable state before and after the perturbation. The stable operating state is usually characterized by a steadystate stack voltage and stack temperature. The EIS measurements should also be stationary, which means it is timeindependent. This condition can be easily verified by repeating the measurements and check the consistency of the Nyquist plot.
3
transfer through the membrane and water content in the membrane has much influence on proton transfer property. As for the double layer capacitor Cdl, it stands for the capacitance property in the catalyst layer where electrochemical reactions happen. Of course, the resistance Rc stands for the resistance against the charge transfer around the catalyst layer. In addition, the Nernst voltage of the fuel cell is defined as [24]. Eo, PH2, PH2O, and PO2 are the Nernst voltage in standard state, hydrogen partial pressure, water vapor partial pressure and oxygen partial pressure against the standard atmosphere pressure respectively. PH2 P0:5 RT O2 ln En ¼ Eo þ 2F PH2 O
! (1)
For this equivalent circuit, the total impedance can be defined as Zu ¼ Rm þ
Rc jCdl Rc u þ 1
(2)
where u stands for the angular frequency. According to the Eq. (2), the resistance Zu is equal to (Rm þ Rc) approximately when the frequency f is close to zero.
Equivalent circuit model of the fuel cell stack Electrical characteristic of fuel cell can be represented by the equivalent circuit model. The most common equivalent circuit is the Randles circuit proposed by Randles [23]. As presented in Fig. 1, the electrolyte resistance Rm is connected in series with a parallel combination of the double layer capacitor Cdl and the charge transfer resistance Rc. According to principles described in many references, the resistance Rm stands for the resistance against the proton
Fig. 1 e Randles cell.
The proposed DC/DC boost converter According to our previous work [26], a six-phase Interleaved Boost Converter (IBC) based on Silicon Carbide (SiC) semiconductors has been proposed. Meanwhile, inverse coupled inductors based on the cyclic cascade configuration [27] are introduced in order to reduce the total volume and weight of proposed converter; and the output current ripple of the FC can be reduced dramatically, thus, the fuel cell stack's lifespan can be extended [25]. The proposed topology is shown in Fig. 2. Each two windings in the range of yellow block stand for one inverse coupled inductor. Some important specifications of our converter are presented in Table.1. In our study, the input power source is a 21 kW PEM FCvelocity-9SSL module from Ballard® Company [28]. To simplify the study of proposed converter, the Randles model is chosen. The required voltage ratio has been met and high power transformer is not needed. What is more, the current ripple of
Fig. 2 e Six-phase IBC based on inverse coupled inductor for each phase. Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242
4
international journal of hydrogen energy xxx (xxxx) xxx
Table 1 e Specifications of power converter. Symbol Vin PN Vout fs
Parameter
Value
Unit
Input voltage Nominal power Output voltage Switching frequency
70 21 350 100
V kW V kHz
power sources has been reduced significantly by this topology. This is one of the most interesting factors to extend power source lifespan. SiC semiconductors are introduced to satisfy the high switching frequency, and to obtain a preferable thermal performance. Low switching losses and negligible reverse recovery losses are separately achieved by SiC MOSFET and SiC Schottky diode. Thus, the performance and power density of the system can be optimized and the efficiency analysis can be achieved by the previous study [26]. In this study, the inductors are inversely coupled and constructed by a cyclic cascade structure. In other words, each phase is just coupled with the adjacent phase and on the whole six coupled inductors are needed for our application. Taking the volume, weight, complexity of installation and other factors into consideration, EE magnetic cores are selected. The physical structure and electrical structure of inverse coupled magnetic core are separately presented in Fig. 3(a)e(b). Due to the inverse coupled structure, DC magnetic fluxes caused by the two windings flow reversely and total DC magnetic flux is thus almost zero inside the core. The core losses mainly depend on the AC magnetic flux, which is a relatively small value due to the low current ripple, thus the volume of inductor can be decreased. The inductor volume is closely related to inductor value, current RMS value, current ripple level and maximum flux density. According to [29], the effective electrical size of magnetic cores can be described by the core geometrical constant Kgmin. Eq. (3) is utilized to determine the minimum core geometrical constant Kgmin of the inverse coupled inductor core; and the minimum core geometrical constant Kgmin_uncoupled is obtained by Eq. (4). 2
Kgmin
Kgmin
coupled
¼
uncoupled
rcu L2M iM;sat ðIL1 þ IL2 Þ2 B2sat ¼
Ku Pcu
(3)
tot
rcu L2 I2Lmax B2sat Ku RL
Fig. 3 e The geometric structure of inverse coupled inductor.
(4)
The wire effective resistivity rcu is 1.724 108 U,m. The magnetizing inductance LM is 80 106 H. The maximum magnetizing current iM,sat is 1.75A. The DC component of each winding's current, IL1 and IL2, is 50A. The saturation flux density Bsat is 0.4T. Total copper losses Pcu_tot are 50W. The winding fill factor Ku is 0.6. Fig. 4 presents the relationship between number of interleaved phases (N) and minimum magnetic core geometrical constant (Kgmin) separately for uncoupled and inverse coupled inductors. Obviously, Kgmin_coupled is much smaller than Kgmin_uncoupled. Additionally, the minimum core geometric constant decreases while the number of interleaved phases increases with both designs.
The control strategy design process The design of sliding-mode control strategy The control objectives are necessary to be established for the purpose of defining the control strategy: 1) Output voltage regulation under load uncertainty. To avoid load damages, it is necessary to regulate the DC bus voltage and maintain it constantly. 2) Equal current sharing between each phase of coupled inductors. The current waveforms should be equal in order to avoid overloading one of the phases, especially when supplying heavy loads. Also the currents must be interleaved in order to reduce the current ripple which is undesirable in fuel cell stack. 3) Asymptotic stability of the closed loop system. Global asymptotic stability is required to avoid imposing restrictions on the allowed initial conditions. A Proportional-Integral (PI) controller is well known and widely applied in industrial application fields to optimize the steady-state performance. The proportional term contributes to reduce the current error value while the integral term accelerates the movement of the process towards set point and eliminates the residual steady-state error that occurs with a pure proportional controller [30]. Nonetheless, an obvious drawback faced with PI controller is that it is linear. Hence, the performance of PI controller in non-linear system is variable. When we design a PI compensator for a Boost converter, it relies on the small signal model of the converter. In order to utilize small signal model, the premise is that the converter operates under static work point. In such a condition, AC small signal analysis can be preceded. Unfortunately, the classical PI control strategy cannot satisfy the highly non-linear characteristic of inverse coupled inductor, especially when dynamic change occurs. The unifications of inductor currents are strictly required in our study to guarantee the magnetic components operating under nonsaturation zone. This requirement is needed to be satisfied whether under static work point or dynamic processes. Further, the operating processes of coupled inductor are strongly related to duty cycle and number of interleaved phases. More interleaved branches can lead to complicated operating process. Meanwhile, duty cycle, which is closely
Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242
5
international journal of hydrogen energy xxx (xxxx) xxx
Fig. 4 e (a) Stands for the relationship between N and Kg for both uncoupled and inverse coupled inductors; (b) stands for the zoom in on the red line of (a) for inverse coupled inductor.
related to load level, can influence the regulating process of inductor currents significantly. Supposing significant differences occurring between inductor currents, magnetic saturation could be introduced. In other words, well controlled inductor currents can lead to desirable performance of magnetic components. Hence, a control strategy which can achieve a zero steady-state error and be insensitive to disturbance is attractive to this study. Sliding Mode Control (SMC), which has been proposed by [31], is theoretically robust to the plant parameter variations and quite suitable for our requirement. More importantly, the design process of SMC depends on the large signal model of DC/DC converter. Its stability is not restricted by variations around operating point, which contributes to an overall improved controller performance. The sliding surface in our study is defined by the following expression. SLi ¼ ILi ILi
ref
Zt þ KLi
ILi ILi
ref
dt
(5)
0
In Eq. (5), i is for the sequence number, ILi stands for the average current of each inductor and ILi_ref represents the reference current. KLi, which is positive, defines the dynamic of convergence to zero of the static error. The Control-Lyapunov function is used to test whether a system is feedback stabilizable [32]. A characteristic method is applying a Lyapunov candidate function to test the stability of dynamic system. Hence, ε, which equals to ILi minus ILi_ref, is defined as the error between actual and desired inductor current. Thus, Eq. (5) can be re-written as Zt SLi ¼ ε þ KLi
εdt
(6)
A Control-Lyapunov candidate is presented as Eq. (7), which is positive definite for all ILi_refs0, d(ILi_ref)/dts0. 1 V ¼ S2 2
(7)
The purpose is to achieve the time derivative to be ·
V ¼ lV; l > 0
(8)
which is globally exponentially stable if V is globally definite. The convergence dynamic of the sliding surfaces to zero is defined as Eq. (9), which can be proved by Eqs. (6)e(8). This control law will guarantee global exponential stability. S0Li ¼ lLi SLi
(9)
The convergence factors (lLi) are positive real numbers. A larger convergence factor can lead the system to reach its steady state faster. However, it can only be selected in a certain range due to the limitations of the system. The state space model of proposed converter is essential to design the control strategy, and it can be obtained by Eq. (10) according to the operating process. Depending on the equations mentioned previously, Eq. (11) can be achieved which represents the duty cycle of each phase as a function of time. d ½ILi ¼ ½L1 ð½Vin ½1 Di Vo ½iLi rL Þ dt ½Di ðtÞ ¼ 1
h 1 ½Vin þ rL ½ILi þ ½LA lLi Si þ I0Li Vo i
(10)
ref
KLi
εi
(11) The inverse coupled inductor matrix and the similar matrix of its inverse matrix are separately presented by [L] and [LA] as follow. Depending on Eq. (11), the control inputs are independent and unaffected by the disturbance of loads.
0
Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242
6
international journal of hydrogen energy xxx (xxxx) xxx
2
2L 6 M 6 6 0 ½L ¼ 6 6 0 6 4 0 M 2 1
M 2L M 0 0 0
6 2ðL MÞ 6 6 6 0 6 6 6 6 6 6 0 6 ¼6 6 6 0 6 6 6 6 6 0 6 6 4 0
0 0 M 0 2L M M 2L 0 M 0 0
0 0 0 M 2L M
3 M 0 7 7 0 7 7 ½LA 0 7 7 M 5 2L 66
3
0
0
0
0
0
1 2ðL þ MÞ
0
0
0
0
0
1 2L M
0
0
0
0
0
1 2L M
0
0
0
0
0
1 2L þ M
0
0
0
0
0
1 2L þ M
7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 5
Combining Eqs. (5),(9) and (11) together, we can obtain: ε0 þ ε ðlLi þ KLi Þ þ lLi KLi
Zt εdt ¼ 0
(12)
0
The time derivative of Eq. (12) is 00
ε þ ε0 ðlLi þ KLi Þ þ lLi KLi ε ¼ 0
(13)
A high robustness for the controller can be achieved because this equation is irrelevant with the topology parameters. In order to achieve the desired performance, the coefficients KLi and the convergence factors lLi can be determined by the classical tuning method for second-order systems.
The integration of EIS with SMC In consideration of the limited effective volume of FCEV, the integration of EIS functionality in control strategy of converter is a promising technique in purpose of realizing on-line monitoring the state of health of FC [22]. The strategy of EIS integration in SMC consists in injecting the current perturbation Isin around the polarization current IFC as presented in Fig. 5(b). In Fig. 5, a delay (Tdelay) existed between neighbor phases drive signals (PWM Phase N). This value depends on number of phases (N) and cycle period (Ts). In our case, N and
Ts separately equal to 6 and 10 ms. Thus, Tdelay can be achieved as (10/6) us. At nominal condition, DC bus voltage (Vo) of converter will be compared with the reference voltage (Vref), and then the difference will be transferred to a PI controller (out loop controller) to obtain the current reference (IL_ref) of each phase. After that, IL_ref, fuel cell voltage (VFC), fuel cell current (IFC) and inductor current (IL) of each phase will be sent into the SMC (inner loop controller). After the calculation process, PWM signals can be obtained and will be sent to corresponding power MOSFET as drive signal. On-line detection of EIS based on the proposed DC/DC converter is presented by Fig. 6. The process can only be handled when fuel cell stack operates at stable state, which is usually characterized by a steady-state stack voltage and stack temperature [22]. When the EIS monitoring is required, a sinusoidal perturbation will be superimposed to the output signal of PI controller (IL_ref). According to the process of SMC, the current of each phase will track this reference value. Thus, the sinusoidal perturbation can be superimposed successfully to the fuel cell current. Owing to operate at linear range, the fuel cell voltage will also be perturbed by a sinusoidal signal combined with the same frequency as the current perturbation. Under this condition, the mean values of the fuel cell stack output voltage and current can be obtained separately. These two parameters are essential for the measurement of AC impedance. Meanwhile, benefiting from the high robustness of this control strategy, the output voltage of DC/DC boost converter can still be maintained relatively stable at the required value (<7%). In order to illustrate the principle of AC impedance measurement, hence, the sampling process of fuel cell stack current is taken as an example. The sampling processes both of VFC and IFC are shown in Fig. 7. The difference between fuel cell current mean value and perturbative fuel cell current is an AC component. The AC component is composed by the fuel cell current ripple under steady state and a sinusoidal perturbation. The frequency of the fuel cell current ripple equals to switching frequency. Thus, the mean value of the AC component in each switching period can be calculated. A sinusoidal wave based on the same frequency as the perturbation signal, can be achieved. This signal is the exact extracted signal, iFC_AC, which is caused by the injected sinusoidal perturbation, Isin. Perturbations combined with different frequencies can let different responses of fuel cell current and corresponding responses of
Fig. 5 e The strategy of EIS integration in SMC.
Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242
international journal of hydrogen energy xxx (xxxx) xxx
8 N1 X > 2p > > ReðI nk Þ ¼ I ðnÞ cos > FC FC > < N n¼0
; k ¼ 0; 1; //; N 1 > N1 X > 2p > > IFC ðnÞ sin nk > : ImðIFC Þ ¼ N
7
(15)
n¼0
8 VFC ReðVFC Þ þ j$ImðVFC Þ > > > ZFC ¼ I ¼ ReðI Þ þ j,ImðI Þ ¼ Re ZFC þ j,Im ZFC > FC FC FC > > > > < ReðV Þ ReðI Þ þ ImðV Þ ImðI Þ FC FC FC FC Re ZFC ¼ > Re2 ðIFC Þ þ Im2 ðIFC Þ > > > > > > > ImZ ¼ ImðVFC Þ ReðIFC Þ ReðVFC Þ ImðIFC Þ : FC Re2 ðIFC Þ þ Im2 ðIFC Þ
(16)
Normally, there is no limitation for the minimum frequency of the perturbation signal, but the maximum frequency shall be smaller than fS/N. N stands for the number of samples and it is selected as 10 in our case in consideration with the sampling precision [18]. An impedance spectrum can be formed by a series of the stack impedances at discrete frequency points within a wide range of frequencies. The proposed converter, which is based on SiC semiconductors, makes it possible to obtain impedances at high frequency due to the ability of operating at high switching frequency. The frequency of stimulus signal is here designed from 1 Hz to 10 kHz.
The simulation results
Fig. 6 e The process flow diagram of on-line detection of EIS based on the proposed DC/DC converter.
PLECS, which is a software tool and especially designed for system-level simulations of power electronics, is utilized in our study to verify the proposed EIS monitoring functionality. The simulation results of proposed DC/DC converter and EIS achievement are separately presented in the following sections.
Electrical part simulation results fuel cell voltage. These independent responses can be utilized to plot impedance spectrums. Discrete Fourier Transform (DFT) is utilized to separate the real part and imaginary part of iFC_AC. The sampling frequency (fe) is set at twenty times the value of the perturbation signal frequency. The number of samples (N) is chosen as 10. In this study, the highest order harmonic is selected as ten-time harmonic which is enough for our calculation and analysis. The real part and imaginary part will be utilized to calculate the AC impedance of the fuel cell stack. With regards to the sampling process of the fuel cell stack voltage under perturbed conditions, the principle is the same as illustrated in foregoing paragraphs. Combining DFT and Euler's formula, the impedance of FC stack can be expressed as Eqs. (14)e(16). 8 N1 X > 2p > > VFC ðnÞ cos nk > ReðVFC Þ ¼ > < N n¼0
; k ¼ 0; 1; //; N 1 > N1 X > 2p > > VFC ðnÞ sin nk > : ImðVFC Þ ¼ N n¼0
(14)
The required specifications are obligatory to be achieved both under nominal condition and disturbed condition. Naturally, nominal condition is tested firstly to verify the stability of the control strategy. Fig. 8 presents the simulation results of the proposed converter under nominal condition. Visibly, both loading and unloading tests are realized. Due to the high robust characteristics of SMC, the proposed DC/DC converter operates at steady state. DC bus voltage is stabilized at 350V. DC components of inductor current are totally maintained to identical values. AC components are successfully regulated in a small range. The saturation of inductors can indeed be avoided. Meanwhile, the fuel cell current is likewise well controlled at the linearity range. Especially, due to the interleaved structure of the power converter and the use of coupled inductor, the current ripple of fuel cell stack is rather small, which is effective for the purpose of extending fuel cell lifespan. Hereby, in purpose of verifying the proposed detection strategy, the perturbation signal which possess the frequency of 10 kHz was injected to the inductor current reference and the results were presented by Fig. 9. Obviously, the sinusoid
Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242
8
international journal of hydrogen energy xxx (xxxx) xxx
Fig. 7 e The processes of fuel cell voltage and fuel cell current sampling in purpose of the realization of EIS functionality. From (b) to (g) are the zooms of typical signals, either fuel cell voltage or current, which are essential during the procedures of the proposed strategy. A sinusoidal perturbation combined with frequency of 100Hz and amplitude of ±1A is taken as an example in these figures.
perturbation has been successfully integrated with the fuel cell current and voltage. To be emphasized, close loop control has been achieved even during the process of perturbation injection. DC bus voltage has been well controlled and stabilized at the required value (Vbus_DC ¼ 350V, △Vbus<4%).
EIS detection using the proposed DC/DC converter As analyzed previously, when an AC signal of known amplitude and frequency is sent to the inductor current reference of proposed DC/DC converter, the amplitude responses of the fuel cell stack's voltage and current can be recorded separately. This process can be repeated at different frequencies, thus, a Nyquist impedance spectrum can be formed by a series
of the stack impedances at discrete frequency points to achieve more information about the cell parameters. Under nominal condition, Nyquist plot is obtained as presented by Fig. 10. Due to the characteristics of Randles model, only one arc can be exhibited in Nyquist plot. Hence, three feature points, Rm, Rc and 4, which reflect the operating condition of fuel cell stack, have been illustrated as follow [22]. Rm, which is the left intersection of the curve and real axis, indicates the total Ohmic resistances of fuel cell stack. The humidification degree of the membrane can be reflected by the parameter. Rc, which can be achieved when the perturbation frequency approaches zero and it is actually the low frequency
Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242
international journal of hydrogen energy xxx (xxxx) xxx
9
Fig. 8 e Normal condition without perturbation signal.
Fig. 9 e Injected small disturbance small signal combined with frequency of 10 kHz and amplitude of 1A.
intercept of the impedance arc. The global performance of fuel cell stack can be represented by it. 4, which corresponds to the maximal phase in the Phase plot and indicate the size of low frequency diffusion arc. A bigger 4 represents a large arc diameter and reflects a difficult mass transport process. In Fig.11.(a), the red curve and the yellow curve separately represent simulation results and theoretical calculation results of impedance spectrum at nominal condition. The simulation results are identical with theoretical analyses. The
proposed strategy is verified under nominal operating conditions based on Randles model. In order to verify the proposed detection method, some parameters of fuel cell model are modified to simulate different faults occurring on the fuel cell stack. Water management inside fuel cell stack is essential to avoid breakdown or damage occurred to power source. Different types of fault (flooding or drying) can be indicated by different impedance spectrums as a result of variations of fuel cell model parameters [33]. Therefore, according to visible variation introduced
Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242
10
international journal of hydrogen energy xxx (xxxx) xxx
to impedance curves when these faults are brought into existence, EIS is an effective diagnosis tool for these situations. Table.2 presents different fuel cell parameters utilized for different operating states. These parameters are man-set values and just utilized in purpose of verifying the validity of proposed strategy. 1) Simulating membrane drying (Rm increased)
Fig. 10 e Typical EIS of Randles model under normal operating condition.
Drying can be caused if the membrane is not sufficiently hydrated, or the inlet gases are insufficiently humidified. The inlet gases’ temperature is below the fuel cell operating one can be another reason. Drying out can lead to an increase of membrane resistance. Hence, this fault can be detectable by any increase of cell resistance [34]. Rm is replaced by a bigger value to simulate drying fault occurring to the fuel cell stack. Both theoretical analyses and simulation results are presented in Fig. 11.(b). Distinctly, the impedance spectrums achieved by simulation are visibly different from nominal condition. In Nyquist plot, the left intersection of the curve and real axis moves to the right of positive axis which indicates the increase of Rm, or the occurring of drying fault. The waveform doesn't change, which means the biggest imaginary value is the same as in nominal condition. 2) Simulating cell flooding (Rc increased) Flooding, which is one of the most recurrent PEMFC's faults, is defined as an accumulation of liquid water in the gas channels or electrodes, impeding the access of reactive gases to the active layers, and therefore decreasing the reaction rate [14]. The flooding probability increases the difficulty of mass transport process. The lower the O2 diffusion rate is, the more severe the flooding will be. Difficult mass transport leads to an arc owing a bigger diameter in Nyquist plot. Rc is replaced by a bigger value to imitate flooding fault occurred to fuel cell stack. Both theoretical analyses and simulation results are presented in Fig.11.(c). Obviously, the impedance spectrums obtained by simulation are visibly different from nominal condition. In Nyquist plot, due to the untouched membrane resistance, the left intersections of the curves and real axis are the same. An arc combined with a bigger diameter is achieved by the flooding conditions. Hence, according to the comparison analyses between theoretical and simulated results, the proposed strategy,
Fig. 11 e Electrochemical Impedance Spectrums of PEMFC under normal operating condition, drying faults and flooding faults are separately by (a), (b) and (c). EIS, one obtained by simulation results from PLECS, the other obtained by calculation results based on Randles model and parameters in Table.2, are presented to be compared.
Table 2 e PEMFC parameters for the assumption of different man-set faults to verify proposed EIS detection functionality. State of Health (SOH) Normal Drying Flooding
Rm/mOhms
Rc/mOhms
Cdl/F
5.58 8 5.58
15.46 15.46 50
1.37 1.37 1.37
Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242
international journal of hydrogen energy xxx (xxxx) xxx
which focuses on integration of EIS functionality with slidingmode control on the considered DC/DC power converter, is effective to be applied for monitoring the state of health of a fuel cell stack. Both drying and flooding modes are detectable from the achieved impedance spectrums. On-line diagnosis of the fuel cell stack is realized; meanwhile DC bus voltage is well regulated even during detection procedure. To be emphasized, the proposed on-line diagnosis approach is accomplished without any additional equipment which is highly attractive for practical FCEV applications due to the limited space inside a vehicle.
Conclusion This work concentrates on the realization of on-line detection of impedance spectroscopy for PEMFC application based on connected electric power converter. On-line EIS can present real-time frequency-impedance information regarding the state of health that can be used in the diagnosis or control purposes. The proposed converter based on high switching frequency, SiC semiconductors and inverse coupled inductors is an innovative solution to settle the problem of regulating PEMFC voltage to satisfy the voltage requirement of the fuel cell electric vehicle DC bus. The selection of interleaved structure is meaningful for the reduction of the FC current ripple and the extension of the FC lifespan. Thanks to the selection of SiC semiconductors and coupled inductor technique, the total volume and weight of the power converter have been reduced significantly. The main advantages of this work are the full integration of EIS detection algorithms in the power converter control strategy. No additional equipment required and the cost of EIS implementation is negligible, thus, the integration constraints are minimized. Compared with the existing studies, the proposed strategy has been verified by FC stack Randles model in a wide range of frequencies (maximum 10 kHz). Besides, the selected SlidingMode Control can well regulate the fuel cell current and DC bus voltage, and realize close loop control either under nominal operating conditions or disturbed conditions. In summary, on-line fuel cell impedance acquisition and power conversion have been realized by the proposed detection strategy and power converter. Hence, the efficiency, reliability and lifespan of PEMFC can be guaranteed further according to our study which is significant in practical FCEV application.
references
[1] Hissel D, Pera MC. Diagnostic & health management of fuel cell systems: issues and solutions. Annu Rev Contr 2016;42:201e11. € gel P, Oltra C, Sala R, Lores M, Upham P, Du¨tschke E,, [2] Bo Wiemann P. The role of attitudes in technology acceptance management: reflections on the case of hydrogen fuel cells in Europe. J Clean Prod 2018;188:125e35. [3] Cherubini S, Iasevoli G, Michelini L. Product-service systems in the electric car industry: critical success factors in marketing. J Clean Prod 2015;97:40e9.
11
[4] Dalla Chiara B, Pellicelli M. Sustainable road transport from the energy and modern society points of view: perspectives for the automotive industry and production. J Clean Prod 2016;133:1283e301. [5] Yoshida T, Kojima K. Toyota MIRAI fuel cell vehicle and progress toward a future hydrogen society. The Electrochemical Society Interface 2015;24(2):45e9. [6] Sutharssan T, Montalvao D, Chen YK, Wang WC, Pisac C, Elemara H. A review on prognostics and health monitoring of proton exchange membrane fuel cell. Renew Sustain Energy Rev 2017;75:440e50. [7] Silva RE, Harel F, Jemei S, Gouriveau R, Hissel D, Boulon L, et al. Proton exchange membrane fuel cell operation and degradation in short-circuit. Fuel Cell 2014;14(6):894e905. [8] Jia F, Guo L, Liu H. Mitigation strategies for hydrogen starvation under dynamic loading in proton exchange membrane fuel cells. Energy Convers Manag 2017;139:175e81. [9] Zhang C, Zhou W, Ehteshami MM, Wang Y, Chan SH. Determination of the optimal operating temperature range for high temperature PEM fuel cell considering its performance, CO tolerance and degradation. Energy Convers Manag 2015;105:433e41. [10] Jian-hua T, Zhao-yuan S, Jin-song S, Zhong-qiang S. Preparation of water management layer and effects of its composition on performance of PEMFCs. Energy Convers Manag 2008;49(6):1500e5. [11] Shyu JC, Hsueh KL, Tsau F. Performance of proton exchange membrane fuel cells at elevated temperature. Energy Convers Manag 2011;52(12):3415e24. [12] Barbir F, Gorgun H, Wang X. Relationship between pressure drop and cell resistance as a diagnostic tool for PEM fuel cells. J Power Sources 2005;141(1):96e101. [13] Gerard M, Poirot-Crouvezier JP, Hissel D, Pera MC. Oxygen starvation analysis during air feeding faults in PEMFC. Int J Hydrogen Energy 2010;35(22):12295e307. [14] Santarelli MG, Torchio MF. Experimental analysis of the effects of the operating variables on the performance of a single PEMFC. Energy Convers Manag 2007;48(1):40e51. [15] Becherif M, Pera MC, Hissel D, Zheng Z. Determination of the health state of fuel cell vehicle for a clean transportation. J Clean Prod 2018;171:1510e9. [16] Swierczynski M, Stroe DI, Stanciu T, Kær SK. Electrothermal impedance spectroscopy as a cost efficient method for determining thermal parameters of lithium ion batteries: prospects, measurement methods and the state of knowledge. J Clean Prod 2017;155:63e71. [17] Narjiss A, Depernet D, Candusso D, Gustin F, Hissel D. Online diagnosis of PEM fuel cell. In: Power Electronics and Motion Control Conference, 2008. EPE-PEMC 2008. 13th; 2008, September. p. 734e9. IEEE. ra MC. [18] Depernet D, Narjiss A, Gustin F, Hissel D, Pe Integration of electrochemical impedance spectroscopy functionality in proton exchange membrane fuel cell power converter. Int J Hydrogen Energy 2016;41(11):5378e88. [19] Hong P, Xu L, Jiang H, Li J, Ouyang M. A new approach to online AC impedance measurement at high frequency of PEM fuel cell stack. Int J Hydrogen Energy 2017;42(30):19156e69. €l S, Davat B. [20] Hinaje M, Sadli I, Martin JP, Thounthong P, Rae Online humidification diagnosis of a PEMFC using a static DCeDC converter. Int J Hydrogen Energy 2009;34(6):2718e23. [21] Bethoux O, Hilairet M, Azib T. A new on-line state-of-health monitoring technique dedicated to PEM fuel cell. In: Industrial Electronics, 2009. IECON'09. 35th Annual conference of IEEE. IEEE; 2009, November. p. 2745e50. [22] Zheng Z. On line fault diagnosis of PEMFC stacks via on nonmodel based methods and EIS measurements. 2014.
Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242
12
international journal of hydrogen energy xxx (xxxx) xxx
[23] Fouquet N, Doulet C, Nouillant C, Dauphin-Tanguy G, OuldBouamama B. Model based PEM fuel cell state-of-health monitoring via ac impedance measurements. J Power Sources 2006;159(2):905e13. [24] Larminie J, Dicks A, McDonald MS. Fuel cell systems explained, vol. 2. Chichester, UK: J. Wiley; 2003. [25] Wahdame B, Girardot L, Hissel D, Harel F, Franc¸ois X, Candusso D,, Dumercy L. Impact of power converter current ripple on the durability of a fuel cell stack. In: Industrial Electronics, 2008. ISIE 2008. IEEE International Symposium on. IEEE; 2008, June. p. 1495e500. [26] WANG, H., GAILLARD, A., & HISSEL, D. Six-phase softswitching interleaved boost converter based on SiC semiconductor and coupled inductor for fuel cell vehicles. [27] Park IG, Kim SI. Modeling and analysis of multi-interphase transformers for connecting power converters in parallel. Power electronics specialists conference, 1997. PESC'97 record., 28th annual IEEE, vol. 2. IEEE; 1997, June. p. 1164e70. [28] De Bernardinis A. Current-fed inverter topologies and control strategy applied to modular power fuel cells in transportation applications. In: Transportation
[29] [30]
[31]
[32]
[33]
[34]
electrification conference and expo (ITEC), 2015 IEEE. IEEE; 2015, June. p. 1e6. Erickson RW, Maksimovic D. Fundamentals of power electronics. Springer Science & Business Media; 2007. Ang KH, Chong G, Li Y. PID control system analysis, design, and technology. IEEE Trans Contr Syst Technol 2005;13(4):559e76. Chen TL, Wu YC. An optimal variable structure control with integral compensation for electrohydraulic position servo control systems. IEEE Trans Ind Electron 1992;39(5):460e3. Freeman R, Kokotovic PV. Robust nonlinear control design: state-space and Lyapunov techniques. Springer Science & Business Media; 2008. ra MC, Pe rez BG. A review Dijoux E, Steiner NY, Benne M, Pe of fault tolerant control strategies applied to proton exchange membrane fuel cell systems. J Power Sources 2017;359:119e33. guy P, Candusso D. Diagnosis of Steiner NY, Hissel D, Moc¸ote polymer electrolyte fuel cells failure modes (flooding & drying out) by neural networks modeling. Int J Hydrogen Energy 2011;36(4):3067e75.
Please cite this article as: Wang H et al., Online electrochemical impedance spectroscopy detection integrated with step-up converter for fuel cell electric vehicle, International Journal of Hydrogen Energy, https://doi.org/10.1016/j.ijhydene.2018.10.242