SMES HPS under unknown load containing pulses

SMES HPS under unknown load containing pulses

Renewable and Sustainable Energy Reviews 105 (2019) 14–37 Contents lists available at ScienceDirect Renewable and Sustainable Energy Reviews journal...

6MB Sizes 0 Downloads 5 Views

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews journal homepage: www.elsevier.com/locate/rser

Hybrid power sources (HPSs) for space applications: Analysis of PEMFC/ Battery/SMES HPS under unknown load containing pulses Nicu Bizona,b, a b

T



University of Pitesti, Faculty of Electronics, Communications and Computers Science, 1 Targu din Vale, 110040 Pitesti, Arges, Romania University Politehnica of Bucharest, Doctoral School, Bucharest, Romania

A R T I C LE I N FO

A B S T R A C T

Keywords: Space technology Solar energy Fuel Cell Battery Superconducting Magnetic Energy Storage Hybridization Optimization

This study presents a brief review of Hybrid Power Sources (HPSs) for space applications to compare the results obtained for a HPS under unknown load containing pulses. The reliable technologies for energy sources and Energy Storage Systems (ESS) that can operate safety in extreme environments (very low temperature, intense radiation environments etc.) and under dynamic load demand (including load pulses) are compared based on the targets for power and energy density, efficiency, and lifetime. The pros and cons for HPS architectures and ESS topologies proposed in the literature are discussed in frame of optimization of the whole system. Two new optimization strategies were proposed to optimally operate the Fuel Cell (FC) system based on two control loops implemented based on the global optimization control of the boost DC-DC converter and the load-following control of the fuel flow rate or of the air flow rate. The comparative study performed (under constant load, dynamic load, and variable PV power) points out the advantages of one of the proposed optimization strategy in all performance indicators. For example, the gaps compared with the reference strategy are of 1.88%, 13.61 W/ lpm, and 293 lpm for FC system efficiency, fuel consumption efficiency, and fuel economy, if the maximum load is considered. Also, different control methods are proposed at the ESS side to mitigate the load pulses (protecting the FC system) and regulate the DC voltage. The results obtained in this paper are discussed related to other ESS hybridizations and control solutions reported in the literature.

1. Introduction It is well known that the requirements and constraints for generating and storing energy are very specific for space applications due to environmental conditions and profile requested for load demand. In general, the load demand is unknown and has a sharp profile containing load steps and pulses. Furthermore, the energy must be available during the eclipse phases in orbits or on surfaces. Thus, an Energy Storage System (ESS) (e.g. energy and power storage devices) and Auxiliary Energy System (AES) (e.g. regenerative Fuel Cells (FC)) are necessary [1]. Also, different power storage technologies, such as ultracapacitors (UCs), Superconducting Magnetic Energy Storage (SMES) devices, and

high-speed Fly-Wheels (FWs), were considered to design the ESS for load pulses [2]. In recent decades, the Proton Exchange Membrane FC (PEMFC) gained the competition with other FC types to be used into AES for terrestrial and space applications [3]. Thus, NASA is considering upgrading the existing alkaline FC units with PEMFC units or Regenerative Fuel Cells (RFCs) [4]. The latter were further developed for high altitude and space missions in the Low Emissions Alternative Power Project [5]. Fuel cells have been used starting with human space missions such as Gemini, Centaur, Apollo, and the space shuttles and lunar vehicles [6], so NASA has gained large expertize in the safe and effective handling of hydrogen for its optimal use in PEMFC systems. Note that

Abbreviations: AES, Auxiliary Energy System; AirFr, Air Flow rate; AFC, Alkaline Fuel Cell; AV, Average Value; BMS, Battery Management System; BTS, Base Transceiver Stations; BoP, Balance-of-Plant; BPF, Band-Pass Filter; DBPFC, Direct borohydride–hydrogen peroxide fuel cell; DOD, Depth of Discharge; DRFC, Discrete Regenerative Fuel Cell; EFSS, Energy Flow Split Strategies; ES, Extremum Seeking; ESS, Energy Storage System; FC, Fuel Cell; FCHPS, Fuel Cell Hybrid Power System; FW, Flywheel; Fueleff, Fuel consumption efficiency; FuelFr, Fuel Flow rate; GES, Global Extremum Seeking; HF, High frequency; HPF, High-Pass Filter; LC, Load Cycle; LSS, Life Support System; LF, Low Frequency; LFW, Load-Following; LPF, Low-Pass Filter; PCS, Power Condition System; PEM, Proton Exchange Membrane; PEMFC, Proton Exchange Membrane Fuel Cell; PMAD, Power Management And Distribution; PV, Photovoltaic; PWM, Pulse Width Modulation; RFC, Regenerative Fuel Cell; RES, Renewable Energy Source; RTO, Real-Time Optimization; SMES, Superconducting Magnetic Energy Storage; SOC, State-of-Charge; SOFC, Solid Oxide Fuel Cell; sFF, Static Feed-Forward; FuelT, Total fuel consumption; UC, ultracapacitors; URFC, Unitized Regenerative Fuel Cell; ηsys, FC system efficiency ⁎ Correspondence address: University of Pitesti, Faculty of Electronics, Communications and Computers Science, 1 Targu din Vale, 110040 Pitesti, Arges, Romania. E-mail address: [email protected] https://doi.org/10.1016/j.rser.2019.01.044

1364-0321/ © 2019 Elsevier Ltd. All rights reserved.

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

N. Bizon

Seeking (GES) control of the DC-DC converter (of boost type) and the load-following (LFW) control of the fuel flow rate (FuelFr) or of the air flow rate (AirFr); (iv) a comparative study of the two proposed optimization strategies for HPS under unknown load is performed in comparison with a reference strategy; (v) two new current-mode control methods are proposed on the SMES side to mitigate the load pulses, protecting the PEMFC system; (vi) a comparative study of the methods for DC voltage regulation is performed by analyzing where is best to correct the reference current; three cases were considered, with voltage regulation implemented on the control of the backup energy source (the PEMFC system), the energy storage device (the battery), or the power storage device. The structure of the paper is presented as follows. Section 2 is dedicated to HPS design requirements and challenging targets (Section 2.1), power generation subsystems (Section 2.2), load estimation (Section 2.3), and hybrid ESS topologies and technologies usually used for space applications (Section 2.4). The Alkaline FC (AFC), Solid Oxide FC (SOFC), PEMFC, and PV technologies (which can operate in low-intensity and low temperature conditions with high efficiency) were briefly presented as potential energy sources based on the targets for power and energy density, efficiency, and lifetime. The ESS hybridization using energy storage devices (such as batteries, RFC, and FW) and power storage devices (such as capacitors, UCs, SMESs, and high speed FW) in active or semi-active topologies is briefly analyzed in Section 2.4. The challenging objectives for Power Management And Distribution (PMAD) are highlighted in Section 2.5. Section 3 is dedicated to RES/PEMFC/ESS Hybrid Power Source proposed here and the first subsection presents the HPS architecture and modeling. The power on-board needed for a space application is arguably one of the key design requirements, so this is carefully approached in Section 3.2 to design appropriately the need of energy from the energy sources and ESS. Control of the battery/ SMES hybrid ESS is performed in Section 3.3 using a dynamic profile for load demand. The optimization of the RES/PEMFC/ESS Hybrid Power Source is analyzed in Section 4 (in first three subsections using constant load, dynamic load, and variable RES power). The performance indicators, such as the total fuel consumption (FuelT), the fuel consumption efficiency (Fueleff ≅ PFCnet / FuelFr), and the FC system efficiency (ηsys), are evaluated for both optimization strategies proposed here and compared with the static Feed-Forward (sFF) control (which is considered as reference). The behavior of the hybrid ESS is analyzed in the subsection 4.4 using different scenarios (with dynamic load, with load pulsed, and with/without RES) and the results obtained are discussed in subsection 4.5. The Conclusion section ends the paper.

these past applications were all nonregenerative and request enough fuel in cryogenic tanks to support the entire spatial mission because the hydrogen production in space is not yet possible. Thus, besides the developing the RFC systems, the components integration durability and reliability of PEMFC system was continuously improved [7], resulting in IEC standards for safety use of stationary [8] and portable [9] FC power systems, and for performance indicators as well [10]. The RFC is a single system that can either work in electrolyser mode or FC mode. Two RFC types were developed: discrete RFC (DRFC) and unitized RFC (URFC) [11]. In a DRFC system, two separate devices are initially used for power generation and electrolysis process and these were integrated in one system for aerospace applications [12]. A URFC is a single device that can operate in FC-mode and electrolyser-mode, ensuring the dual functions of electricity and hydrogen generation [13]. So, URFC systems have a reduced mass and volume in comparison with DRFC systems, having huge potential to be used in ESSs for different terrestrial and space applications [14]. The overview on the URFC technology, including the main technical issues (such as a lack of operating experience at high efficiency of kW scaled URFC stacks) and the solutions to overcome them are shown in [14]. The study presented in [14] have shown comparable efficiency for URFC with specialized devices such as PEM electrolyser and PEMFC, but the URFC efficiency for the FC mode requires improvements by taking into account the Balanceof-Plant (BoP) energy consumption. Space Hybrid Power Sources (HPS) use different converters to convert the energy flow according to load demand [16,17]. So, it is important to develop reliable and effective HPS architectures by modeling and simulation of these HPSs during the design and testing phase [18,19]. HPSs for space applications (such as spacecrafts, spatial vehicles, communication satellites, Life Support Systems (LSSs), extravehicular space suit LSS etc.) are designed to comply with specific mission requirements such as required service lifetime, restrictions imposed by distance and trajectory, dynamic profile of the load demand, the level of reliability and redundancy imposed, the range of operating temperature, the size and weight limits, the overall cost, etc., which all are finally the design parameters of an HPS [15,20–23]. As spatial HPS projects are always faced with size and weight constraints, energy and power density are essential parameters for choosing the ESS devices by integrating different technologies [23,24]. For example, a portable LSS for astronauts combining cooling and power functions with the metal-hydride, fuel cell and absorption chiller is proposed in [24]. The new LSSs provide improved safety by improving the management of residual water and reducing in the same time the hydrogen storage tank [23,24]. Thus, both energy generation and drinking water production processes were improved. If Renewable Energy Sources (RES) such as photovoltaic (PV) or wind energy are available, these will be used to recharge the ESS. Space applications are ultimate test for RES/PEMFC/ESS HPSs because here the load profile is unknown [15,25], containing sharp load levels and pulses, including HPSs for communication satellites [26,27] and their Base Transceiver Stations (BTS) [28]. Thus, in this study, besides a brief review of HPS architecture for space applications, it is proposed a RES/ PEMFC/ESS HPS under unknown load profile with pulses. This study proposes a PV/PEMFC/ESS HPS architecture that could be optimal solution for terrestrial autonomous applications and could meet most of the requirements for space missions. The main objectives and the novelty of this study are mentioned below as follows: (i) a brief review of recent proposals related to HPSs for space applications under unknown load in order to compare the results obtained for HPS proposed in this paper; (ii) a brief review of the potential energy storage devices and advanced power topologies in order to highlight the pros and cons of the ESSs reliable for space applications under load pulses; (iii) two new optimization strategies have been proposed to operate the PEMFC system at the optimal point by using the feedback loops implemented based on the Global Extremum

2. Hybrid power sources (HPSs) for space applications HPSs must be accommodated in the narrow space of spacecraft power system or communication satellite in order to provide required power on the DC bus by using both energy and power storage devices to compensate the power flow balance [29–32]. In first space applications, both primary (one discharge) and secondary (rechargeable) batteries have been used for ESS [15,25]. The power system consisting from power generation sources, ESS, and PMAD unit requires further technological improvements to comply with requirements of a space mission, which are as follows [32]: to produce power systems with considerable mass and volume cuts, to have higher efficiency in large temperature range, and to operate in extreme environments (very low temperature, intense radiation etc.). For example, the architecture of PV/PEMFC/ESS HPS with semiactive ESS topology (where only power storage device is interfaced with the DC bus using a DC-DC bidirectional converter) is shown in Fig. 1. If both power and energy devices are interfaced with the DC bus using a DC-DC bidirectional converter, then the PV/PEMFC/ESS HPS with active ESS topologies is obtained (see Fig. 2) [33]. 15

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

N. Bizon

2.1. HPS design requirements and challenging targets HPSs are necessary for anything that requires power in missions of space exploration, such as spacecraft, launch vehicles, landers, rovers, LSSs, and measurement and communication equipment. For example, the actual HPSs represents up to one third from the spacecraft's mass. So, HPS miniaturization based on new HPS architectures is a design goal that will improve the landing impact, increase the available space for crew, equipment and samples, and enable the large scale use of nanosatellites and small planetary probes as well. Thus, HPS capability must be from tens of watts (W) to MW in order to cover all range of space applications, but with a specific mass per energy up to tens of kg/kWe as a challenging target. This means high power density for actual technologies (higher than 1 kWe/kg or 3600 Wh/kg), but higher for far-term advanced technologies (when the power density could achieving tens of kWe / kg). Because the HPS architecture includes power generation units, ESS, and PMAD unit, it is necessary to use efficient power converters (with efficiency greater than 95%), and optimal energy management based on intelligent and adaptive control loops. The challenging goals for power converters of tens of kW are as follows: power density higher than 500 W/kg and at least 96% efficiency. Furthermore, HPS must operate safety up to 300 °C and tolerate the Jovian radiation levels as challenging goals for safe operation in deep space or difficult environments of planets. The hybridization of all subsystems is mandatory to take the advantages of different available technologies (the mature technologies or in the testing phase), but the HPS cost must be kept within rational limits that will further ensure an affordable life cycle requested by specific mission. Also, safety and reliability of the HPS must be maintained high by integrating new technologies and materials. The selection of them must comply with above mentioned requirements, which are briefly as follows: high power density of used technologies, with increased efficiency, and sufficient durability, safety and reliability, as well as the ability to feed an unknown load profile, even in a wide range of temperature and radiation.

Fig. 1. PV/PEMFC/ESS HPS with semi-active ESS topology.

2.2. Power generation subsystems The subsystem of power generation could include radioisotope power generators, PV arrays, FC systems, and fission and fusion nuclear reactors [32,34], as well as energy scavenging (or energy harvesting) techniques [35] to obtain power from other sources (e.g., waste liquids, waste heat utilization etc.) [23]. The latter is already integrated into the power generation units. The radioisotope, fusion and fission energy sources are quite expensive in comparison with PV array and FC system. Furthermore, the Carnot efficiency limits the specific power of a fission system at maximum operating temperature. Anyway, these energy generation solutions remain good candidates for deep space missions. In this study, the PV arrays and PEMFC system are considered as power generation subsystems. 2.2.1. Fuel Cell (FC) systems FC candidates are the ones who use oxygen (O2) and hydrogen (H2) or methane (CH4) as fuels, such as: Alkaline Fuel Cells (AFCs), Solid Oxide Fuel Cells (SOFCs), and PEMFCs [32]. The fuels will be obtained by electrolysis process using a RFC system. The above FC systems have specific power and power density in range of 50–100 We/kg and 500–1000 Wh/kg, including tank and fuel needed, and ensure a reliable operation for up to 5000 h, but with an efficiency lower than 50%. The target for performance of FC systems are as follows: specific power and power density higher than 130 We/kg and 2000 Wh/kg (including tank and fuel needed), and lifetime and efficiency higher than 10,000 h and 70%.

Fig. 2. PV/PEMFC/ESS HPS with active ESS topology.

16

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

N. Bizon

hydroxide (KOH) electrolyte. Because the alkaline separator requires asbestos, which is now unavailable, other advanced proton and alkaline (OH-) exchange membranes are under test to obtain very high efficiency FC (having efficiency of 80%) [32,38,39].

On the DC bus of the PV/PEMFC/ESS HPS shown in Fig. 1 (or Fig. 2), the power flow balance can be written as below:

CDC udc

dudc = pFC + pESS + pPV − pload dt

(1)

where CDC is the filtering capacitor of the DC voltage (udc), pFC is the FC net power, and pESS = pESS-P + pESS-E, pPV and pLoad are the ESS power, the PV power, and the load power requested on the DC bus. The average value (AV) written for the power flow balance (1) will be given by (2):

0 = PFC (AV ) + PESS (AV ) + PPV (AV ) − Pload (AV )

2.2.1.2. Solid Oxide Fuel Cells (SOFCs). CH4-air solid oxide fuel cell (SOFC) systems are now commercially available for terrestrial applications [40,41], but integrated SOFC systems for space applications are not yet built [32]. These are still in the testing phase, but the results shown that could be potential candidates for missions on planetary surfaces to establish sustainable outposts due to ability to directly process residual propellants from landers and fuels, which are generated from in-situ resource utilization systems. Integrated systems for space applications are not yet built. Furthermore, high-temperature (700–1000 °C) SOFCs enable high compatibility with fuel reformed from such hydrocarbons as natural gas, reliable operation under thermal transients during the on-off cycles, and a better heat rejection per unit mass [32]. The specific power is now of 100 W/kg, the efficiency up to 50% and the lifetime is limited to 3000 h [32,38,39].

(2)

So, since that the load-following (LFW) control of PDC(AV) = PLoad(AV) - PPV(AV) is used, then the AV of the ESS power will be almost zero (3):

PESS (AV ) ≅ 0

(3)

Thus, the size of the ESS could be minimized because the ESS devices will compensate dynamically the sharp changes and pulses from the load profile. FC systems generally ensures the load demand when solar power is unavailable, but here, considering (3) due to the LFW control, the FC power (P`FC = PFC / ηboost) will generate the AV of requested energy on the DC bus (PDC(AV)).

P′FC =

2.2.1.3. Proton Exchange Membrane Fuel Cells (PEMFCs). Due to availability of hydrogen as a fuel, PEMFC systems are intensively used for terrestrial applications (mobile applications [41–43] or stationary applications [44] up to 400 kWe). PEMFC systems use pure hydrogen and air (rather than to use corrosive pure oxygen) to generate energy and expel product water (which is clean and drinkable). The specific power is now of 100 W/kg, the efficiency higher than 50% (but up to 70%) and the lifetime of the PEM membrane is limited to 5000 h due to variable hydration and heat effects [32,38,39]. PEMFC can also be utilized as Uninterrupted Power Supply or AES unit in stationary RES/ESS HPSs. The PEMFC is operated here to use the in-situ fuel production. Obviously, the space applications also need hydrogen and oxygen stored in tanks or fuel that is generated in-situ production units. Hydrogen and oxygen production in space by means of electrolysis as part of a RFC system is likely to become the best method candidate, besides the utilization of nuclear power [32]. Because the methane pyrolysis is currently at low technological readiness, oxygen generation (e.q. for LSSs on the International Space Station) by water electrolysis is a state-of-the-art technology, and hydrogen is a byproduct that can be then stored either physically (compressed hydrogen in tanks, liquid or cryogenic-compressed) or chemically (bonded hydrogen in sorbents, metal hydrides, chemical hydrides or graphite nanofibers) [45–47]. The high-performance cryo-coolers could be good candidate for long-term space missions, but for short space missions of few weeks the HPS may be based on primary (nonregenerative) PEMFC [48,49]. Research on other technologies candidate for FC membrane electrode assembly (MEA) such as MEA based on graphene oxide, which uses nano-filtration for high performance separation processes, is already in testing phase [50]. PEMFC can be supplied with hydrogen and oxygen using as source the sodium borohydride (NaBH4) and the hydrogen peroxide (H2O2), but new developed direct borohydride–hydrogen peroxide FCs (DBPFCs) could solve such problems associated with PEMFCs [51]. However, DBPFCs need to be further improved to become a competitive energy source for space missions [13,32].

PDC (AV ) ηboost

(4)

where ηboost is the AV of energy efficiency of the boost converter (the DC-DC power converter which interfaces the FC system with DC bus in Figs. 1 and 2). PDC(AV) may be obtained by low-pass filtering of the pDC = pload−pPV, the difference of load power and PV power, but other filtering techniques could be used as well [36]. This means that ESS will be used only for dynamic compensation of the power flow balance (1), and the battery's capacity (from the ESS) could be reduced appropriately. This advantage means an increased specific energy of the HPSs used for space application (because the actual batteries have the specific energy of 100 Wh / kg, with target for next years of 200 Wh / kg, but much less than that of the FC system) [32]. It is known that time response of the FC stack is higher than the time response of the boost converter or other power converter used to interface the FC system, as well as the pDC profile [37]. Consequently, high-voltage capacitors on DC bus (CDC) and power storage devices on the ESS, such as ultracapacitors (UCs), SMES, or high-speed FW, will quickly compensate the power flow balance (1). Therefore, the response time of FC system and dynamic of the pDC profile will set the shape and level of the power pulses that will be sustained (charged/discharged) by the power storage device. Besides other advantages, the PEMFC system has the shortest response time comparing with both AFC and SOFC technologies, so PEMFC was used for NASA space applications (conducted between 1960 and 2011) and could be a good candidate for next missions, considering the improvements obtained in meantime [15,25,32]. In next sections, the AFC, SOFC, and PEMFC actual technologies will be briefly presented as potential candidates for space applications, considering the following targets for next products: power higher than 5 kW and 50 kW for mobile and stationary applications, both with specific power higher than 130 W/kg, more than 75% efficiency and a lifetime of 10,000 h at least [32,38,39].

2.2.2. PV arrays Photovoltaic panels for space applications must have a specific power and efficiency higher than 100 W/kg and 50%, lifetime at least one year under acid attack, intense Jovian radiation, and at high temperature [32]. The actual PV panels using triple-junction solar cells can reach conversion efficiency around 31, 34% and 41% under the space spectrum, the standard 1 sun global spectrum, and the concentrated sunlight [52,53]. Simulations show that 70% conversion efficiency

2.2.1.1. Alkaline Fuel Cell (AFC). The AFC was the last FC used on the Space Shuttle Orbiter, generating 2–12 kWe and using pumps and separators to safe manage the water flow [15]. The specific power is now of 50 W/kg, the efficiency up to 70% and the lifetime is limited to 5000 h due to leakage of the caustic potassium 17

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

N. Bizon

could be obtained with multijunction solar cell, allowing for efficiency of the concentrated photovoltaics module to be over 50% [54], but all these technologies are still in testing phase for space application [32]. New technologies for PV system are recently developed that can operate in low-intensity and low temperature conditions with high efficiency, being candidate solutions for space missions farther from the sun (because it is known that the PV power decreases with the square of the distance to the Sun). For example, PV efficiency of carbon nanotube conductive coatings on III-V PV cells and full-spectrum hybrid solarthermal systems could be up to and higher than 50%, but these technologies are under testing phase [32]. Because these technologies are very costly (up to $1000/W), advanced technologies that can reduce the cost are of high interest. The communication satellites and fly space missions usually request PV arrays that can generate up to 10 kWe, but power requested by electric propulsion spacecraft and LSS on the International Space Station is much higher, being up to 100 kWe [32]. So, the estimation of load demand (power on-board needed) for a space vehicle or stationary base is arguably one of the key design requirements [55]. 2.3. Load estimation The HPS enables the safe operating of all equipment for scientific activities, including for propulsion and communication, and LSS if is the case. The load demand has increased from few watts up to hundreds of kW for manned missions. If small power is requested, this can be assured with small PV panels and batteries, but for a higher power demand it is necessary to use multiple PV arrays balanced with PEMFC and ESS during its orbiting period [15]. FC systems with PEMFC and/or RFC units will provide energy for primary shuttle operations and LSS as well as energy for other equipment and spacecraft in the storage bay. PEMFC systems are similar to primary batteries in that the delivered energy is limited to the fuel and oxidant from cryogenic tanks [46,47], but RFC systems open new way to optimally manage the available energy and water considering load demand and existing constrains [14]. Load demands of tens of kW or less will mandatory request the use of FC systems, because the battery technologies will not be effective any longer [38,39]. Also, it is known that there are different telecommunication satellites as size and power, using different orbits and different frequencies, which transmit very different types of signals and operate based on HPSs appropriately designed according to their purpose [26,27]. In the same measure, different BTSs exist as well [28]. A communication satellite or a BTS can use the PV energy excepting the eclipses (periods during which this is shadowed by the earth) or excepting the nights and cloudy days, respectively. In this study, the load demand for PV/PEMFC/ESS HPS was estimated for a medium sized BTS, but this solution can be redesigned for other load demand profile. The advantage of reduced size of ESS remains, being a control feature of FC system that follows PDC(AV). Anyway, the ESS is necessary to compensate the power flow balance (1), but hybridization of FC system has other advantages as follows: the FC system efficiency increase, and stability and reliability of the FC system are sufficient high for space applications by using hybrid FC/ESS systems. Parallel UC hybridization has advantages for pulse load, but the ripple FC current could be higher (for example, the ripple is about 3.6 A / 10 A × 100 = 36% in [56]) than admissible limits (which must to be up to 5% for low frequency (LF) power ripple [16,17]). Thus, appropriate methods must be used to mitigate the LF ripple by spreading it in a large band of frequencies [57,58]. So, hybrid ESS with semi-active or active topologies are usually used, even if this hybrid approach requests a supplementary cost due to use of one bidirectional power converter or two such converters, and appropriate control circuitry of the Power Condition System (PCS) from

Fig. 3. ESS topologies.

PMAD. 2.4. Energy Storage System (ESS) The batteries, regenerative fuel cells, and flywheels are usually used as energy storage devices in ESS for space applications. The hybridization of ESS with power storage devices is a mature storage solution for terrestrial applications such as the smart grids integrating RES [59,60], and other stationary [61] and transport applications [62,63]. 2.4.1. Hybrid ESS topologies The hybrid ESS topologies can efficiently provide power and operate safely in space temperature and intense radiation with sufficient lifetime, and significant size reductions and high specific energy. Furthermore, hybridization of ESS can avoid some of the battery ESS disadvantages related to operation, maintenance, and potential environmental hazards [30,63–65]. In general, hybrid ESS topologies use active or semi-active topology in comparison with passive topology due to their flexibility and performance (Fig. 3). The energy and power storage devices for space applications are in short presented in the following two sections. 18

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

N. Bizon

2.4.3.1. Capacitors. Capacitors are best candidates for space applications (such as communications satellites) where is required to compensate high power pulses at low temperatures [26,46,56]. The main technological challenges for ultracapacitors are to improve the specific energy and power over wide operating temperature (−60 °C to +300 °C) and radiation (higher than 100 Wh/kg and 10,000 W/kg, respectively), to reduce the equivalent series resistance, and to obtain a capacity of hundreds of mF at 1500 V voltage in a small sized volume (less than 1 cm3) [17,32].

2.4.2. Energy storage devices Energy storage devices are needed for stationary (LSS bases) and mobile (rovers) space applications. If RFCs are used in long-term exploration missions and high energy satellites for Earth observation and communications due to the energy densities higher than 500 Wh/kg, the batteries are first candidates for short-term missions even if the energy densities is smaller than 150 Wh/kg for current Li–ion-battery systems [32]. Therefore, PEMFC and RFC systems will not be used for small satellite, even if FC systems have higher energy densities than the batteries [26].

2.4.3.2. High-speed Flywheels (FW). The main technological challenges for high-speed FW are to store energy for LSSs in range of kWh to MWh, with at least 2700 Wh/kg specific energy higher and 20 years lifetime by using carbon nanofiber rotors, advanced generators, and superconducting magnetic bearings [32,76]. Replacing momentum wheels and batteries, the FW reduces the mass of the integrated system [76]. This technology is under testing phase, being reported stationary prototypes (of about 25 Wh/kg specific energy; see for example, the Japan prototype with 100 kWh storage capacity and weight of 4 t, 133 kWh pack of the University of Texas at Austin, NASA ground-tested units etc. [77]) and mobile applications with specific energy in range 3–7 Wh/kg [78]. The main issue in any high-speed FW is the carbon composite [79]. SMES and FW devices have a low self-discharge rate compared with any ESS device and can quickly and repeatedly deliver the energy stored [29,80]. So, the SMES devices were used for high-tech ESS [81] and have recently attracted the attention of the specialists in the field of space applications [82].

2.4.2.1. Batteries. There are many battery technologies from which one can be chosen and evaluated based on terrestrial applications needs and performance requirements such as follows: lead acid (which are used for high pulse applications that requires very low costs, so it is not a potential technology for space applications), nickel metal (which has twice energy density in comparison with lead acid technology, being used for today and past space applications), sodium-based batteries (which are used for very high temperatures applications, having about the same level of the energy density with the nickel-based batteries), and lithium-ion (which has twice energy density in comparison with nickel technology, so it won the competition, being used into ESSs for both terrestrial and space applications) [20,63–66]. The most used cathode chemistries and anode materials are based on graphite and both hard and soft carbons [67–71]. The used energy management strategy sets the performance of large-scale lithium-ion (Li-ion) battery [72–75]. Li-ion battery must operate at extreme temperatures, from −150–450 °C, and depth of discharge (DOD) lower than 30% or up to 90% for outer or inner planet missions. The main technological challenges for Li-ion batteries are obtaining of higher energy density than 1000 Wh/kg, high-voltage and conductive electrolytes with safe capabilities related to the flame retardant and overcharge protection, and high lifetime [32]. For example, 2600 Wh/kg energy density has obtained for Li-sulfur batteries (LSBs), which means that is ten times higher than those of the Li-ion batteries, but the technology is still under testing phase for space applications [32,75]. While the specific energy is major objective for any battery, to comply it with other mission requirements mentioned above, different technology are researched for very low- or high temperature Li-ion batteries and highreliability batteries (with lifetime higher than 15 years) that maintain high energy density required for each specific mission. Because the energy density of the RFCs is three times higher in comparison with Li-ion batteries [14], air-based RFCs are used for terrestrial applications [10–12] and the RFC could be a good candidate for mobile space applications, even if higher complexity is added to the PCS [10,32].

2.4.3.3. Superconducting Magnetic Energy Storage (SMES) devices. The Superconducting Magnetic Energy Storage (SMES) devices were used in terrestrial applications such as energy storage, power energy system stability, uninterruptible power supply, and RES integration in smart grids [83–85]. The SMES can compensate the pulse load in a Shunt Active Power Filter topology, maintain power system stability based on droop control algorithm [86]. The high-temperature SMES in pulsed power supply for high-power electric thruster [82], and for military and high-tech applications is shown in [87,88]. The methodology to design the battery / SMES ESS is proposed in [89,90] and will be used in this study as well, but a new control is proposed here for PCS to sustain the load pulses with unknown levels and shapes, and regenerative load if is the case. Note that among others requirements, the choosing of any of these devices mentioned before will mainly depend on load power profile and mission duration. In general, both are unknown, but both of which can be estimated at least from statistical point of view. Besides the power generation and ESS, the energy management and control from PMAD is a critical unit that must be designed appropriately based on power flow balance (1) and other control objectives and constraints [91].

2.4.2.2. Regenerative Fuel Cells (RFCs). Because the air is not available for space applications [32], the RFC must to recycle oxygen [14]. The RFC systems must be designed for zero gravity up to high gravity levels (during launch operations) and also for optimal thermal and water management in vacuum conditions. Both the PEM and the solid oxide RFC technologies are candidates for space applications and the prototypes tested have demonstrated the viability of these technologies. The targets for high reliability RFCs are 1500 Wh/kg specific energy, 70% efficiency, and lifetime of 10,000 h. So, the research is focused to design such RFCs by developing highefficiency fuel cells and electrolyzers, and improving the thermal and water management subsystems operating in vacuum conditions [32].

2.5. Power Management and Distribution (PMAD) The challenging objectives for PMAD are as follows: design of an effective energy management, and intelligent and adaptive control for autonomous and safe operation of the space system without communication from base station on the Earth, on ground of planet, or in the mother station; use of sensors with wireless connections that are able to operate in space environment (high radiation and range of extreme temperatures); high-fidelity system tests for verification and validation of all units as state of operation compared with the original mission tasks and targets [32]. It is obviously that this requires real-time systems to evaluate the state variables and then to perform advanced control [71]. The sensors and control algorithms developed for terrestrial Smart Grid technologies could be applied for space applications after testing

2.4.3. Power storage devices Considering the PV/PEMFC/ESS HPS with active or semi-active ESS topologies shown in Fig. 3, the hybridization of ESS with power storage devices is necessary in case of pulsed load [32]. 19

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

N. Bizon

by superposition of three rectified sinusoidal signals of frequency f0, 3f0, and 5f0, where f0 = 50 Hz, and 450 W peak amplitude. Besides the LF ripple, in order to test the capability of SMES converter control to mitigate the pulses, two patterns are used for the pulses: one is shown in Fig. 5 and the other will be presented later. The first pattern contains one power pulse (with 0.0225 s width and power level of 800 W), and two more short and small pulses (one positive and other negative with power of 40 W and both with 0.1 s width). The second pattern will also contains one short power pulse, but the power of positive and negative pulses is of 400 W instead of 40 W in order to test the filters used to generate the reference currents for control loops of the power converters from ESS. The dynamic load with stair type profile (step-up and step-down stairs represented in top plot from Fig. 5) has Pload1(AV) = 6 kW, but the average value will be changed in range of 2–6 kW, without overloading the 6 kW / 45 V PEMFC used (with maximum FC power of 8 kW). PEMFC systems are used in last decades for automotive, airplanes, submarines and space vehicles due to recent improvements in performance and safe use [96,97]. Most of PEMFC hybridizations proposed are based on battery ESS with a Battery Management System (BMS) that will keep the battery State-of-Charge (SOC) into imposed range, avoiding overcharging the battery during the charging phases with the extra power generated by the PEMFC and RES (pload < pFC+ pRES) and deep DOD during discharging phases when is lack of power on DC bus (pload > pFC+ pRES) [98–100]. A review of the proposed RES/PEMFC/ESS HPS architectures is presented in [98], including standard control methods. A review of the advanced control methods is made in [99], where an adaptive control is proposed for AirFr. BMS proposed in [100] is quite simple and practical for dynamic load with stair type profile, but it is unpractical for load power pulses because uses switching of two relays controlled by a state diagram

and validation [41]. As it is known, the Smart Grids is a complex system, with many sources, ESS and loads interconnected via power system, which requires reliable, effective and advanced power flow strategies and intelligent control algorithms [31,44,92]. The design of intelligent PMAD is a challenging task which requires research and future developments of equipment for future space missions. The implementation of the artificial intelligence concepts in PMAD and fault detection, isolation, and recovery units will allow systems' autonomy and on-board decision making, reducing the risks of misconduct, lack of communication with base station, and finally the failure of mission [92]. Space applications must be inspired from terrestrial proposals (including biomimetic applications) as preliminary test of operation and then validated based on space standards [93–95]. Therefore, here is proposed a RES/PEMFC/ESS HPSs for a medium-sized Base Transceiver Station, which could supply with energy a communication satellite or space rover as well, after it passed the space specific standards. 3. Proposed RES/PEMFC/ESS hybrid power source The medium-sized BTS considered in this study has peak power (short impulse at the start-up of the air conditioner or other high power equipment) less than 8 kW, and normally the power demand has a stairs profile type, containing power pulses (requested during communication phases and other required operations) and LF ripple (as effect of normal operation of the inverter on DC bus) as well. So, the load demand (pload profile) will be modeled here as in Fig. 4, using (5):

pload = pload1 + pload2 + pload3

(5)

where pload1, pload2, and pload3 are the load demand with stairs profile, pulses, and LF ripple. For example, pload1, pload2, and pload3 profiles are shown in Fig. 5 during 12 s of simulation (including zooms). The LF ripple is obtained

Fig. 4. The diagram of the equivalent load on DC bus. 20

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

N. Bizon

Fig. 5. The load profiles: dynamic load with stair type profile (top), pulses (middle), and LF ripple (bottom).

reducing the battery lifetime. But even if a tight voltage regulation can be achieved with low voltage ripple on DC bus, the bidirectional converter tied to UC has poor utilization factor. Thus, a reconfigurable FC/UC HPS architecture is proposed in [101], which improves the utilization factor of the UC converter and uses a flow chart for energy management algorithm to ensure a controlled power flows from the FC and UC stacks to the dynamic load with stair type profile and voltage regulation on output DC bus. The issues of this proposals are as follows: high capacity are requested for UC stack for power pulsed that must be sustained only by UC stack; high spikes on FC current can be observed in experimental results (at start and end of mode 3), which could damage the FC stack (by starvation phenomena); the state diagram designed could generate different sequences of the operating modes that were not studied for

designed for PV/FC/Battery HPS of unmanned aerial vehicles. Most of BMSs used for RES/PEMFC/ESS HPS architectures proposed in the literature are based on charging / discharging cycles of the battery to sustain the power flow balance (1). So, it is obvious that the battery lifetime will decrease. The BMS proposed here operates the battery in charge sustaining mode based on proposed LFW control for the PEMFC system. This concept of LFW control for the PEMFC system was previously proposed, with RES [33] and without RES [101,102], by using a semi-active battery / UC hybrid ESS topology with feedback loop to stabilize the voltage on DC bus at VDC(ref) based on UC stack power. This approach need to oversize the UC stack in order to regulate DC bus voltage under dynamic load and to maintain the voltage ripple low. If not, the high ripple on DC bus (where the batteries stack is directly connected) means a high ripple of the battery power as well, 21

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

N. Bizon

Fig. 6. The diagram of the RES/PEMFC/ESS HPS using the Fuel-LFW/Boost-GES-RTO strategy.

maximum). The Air-LFW/Boost-GES-RTO strategy is obtained considering both switches SW1 and SW2 in the upper positions, and switch SW3 remains in the same position. In this case, the LWF control can optimally control the both AirFr and FuelFr inputs by the FC current. To compare the results obtained with both Boost-GES-RTO strategies, the sFF control strategy will be used as reference. The sFF-control strategy is obtained considering the switches SW1 and SW2 in position of sFF control, and SW3 in LFW position to follow the load demand. Thus, the sFF-control strategy will not use the GES-based optimization loop to control the AirFr and FuelFr in Boost-GES-RTO strategies proposed in this paper. The PEMFC model used here is the 6 kW / 45 V PEMFC default model from the library of Matlab-Simulink®. This model is a detailed model including the dynamical part, which is used in other studies as well [99,102], being a model with sufficient accuracy, as it is reported in [103]. The FC time constant was set to 0.1 s. A DC-DC unidirectional boost converter boosts the FC voltage to VDC ≅ VDC(ref) = 200 V. The boost DC-DC converter is controlled by a hysteretic controller, which has the inputs IFC and IrefLFW. The last signal (IrefLFW) is generated by LFW control block (see Fig. 7) or by the GES control block (see Fig. 8; the latter being used only in Boost-GES-RTO strategies).

FC/UC HPS operating under standard and unknown load cycles. Anyway, the conclusion of this study is that would be helpful to be used an ESS with more storage devices compared with ESS with single storage device if the requirements for performance and safe are more restrictive. Therefore, an RES/FC/ESS HPS with active Battery / SMES hybrid ESS topology is proposed here. The Real-Time Optimization (RTO) algorithm designed ensures the voltage regulation on DC bus, the operation in charge sustaining mode for battery (so a reduced capacity and an extended lifetime for battery will result), the load demand for any type of load cycle (which includes pulses and LF ripples), and the efficient operation of the PEMFC system. 3.1. The HPS architecture and modeling The architecture of RES/PEMFC/ESS HPS using the Fuel-LF/BoostGES-RTO strategy is presented in Fig. 6 (with the switches SW1, SW2, and SW3 as it is shown in Fig. 6). In this case the LWF control is applied to FuelFr and AirFr is controlled by the FC current, which will find under a RTO algorithm (such as the GES algorithm used here) the optimal value Iref (where the optimization function implemented is

Fig. 7. The diagram of the load-following (LFW) control.

22

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

N. Bizon

Fig. 8. The diagram of the GES control.

The LFW control block implements (4), with PDC(AV) obtained by a low-pass filtering (LPF) of the pDC = pload -pPV. The AV efficiency of the boost DC-DC converter was set at 95% and the saturation block limits the range of FC current. The GES control block will implement (6):

where P′FC and Pcm are the FC power generated and the air compressor power. The Pcm power can be estimated using (9) [104]:

y = f (v1, v2), yN = kNy⋅y

where the coefficients are [104]: a0 = 0.6, a1 = 0.04, a2 = −0.00003231, b0 = 0.9987, and b1 = 46.02. The inputs FuelFr and AirFr are estimated based on (10) using the reference currents Iref(H2) and Iref(O2):



(6a) •

yf = −ωh⋅yf + ωh⋅yN , yHPF = yN − yf , yBPF = −ωl⋅yBPF + ωl⋅yHPF

(6b)

yDM = yBPF ⋅sd,

(6c)

• yInt

Pcm = Icm⋅Vcm = (a2⋅AirFr 2 + a1⋅AirFr + a0)⋅(b1⋅IFC + b0)

sd = sin(ωt ),

FuelFr = = yDM

Gd = yMV , yMV =

(6d)

1 ⋅ Td

∫ yBPF dt

yM = Gd

(6f)

p1 = k1⋅yInt , k1 = γsd⋅ω

(6g)

p2 = k2⋅yM ⋅sd

AirFr =

(6e)

(6i)

IrefGES = kNp⋅(p1 + p2 + p3 ),

(6j)

The switch SW3 is used to select the reference currents: IrefGES in Boost-GES-RTO strategies and IrefLFW in sFF control strategy. The normalization gains are set to kNy = 1/YMax and kNp= IFC(rated) / 2, where IFC(rated) and YMax are the rated FC current and the maximum estimated for the optimization function. The optimization function f used in this study is defined to increase the FC system efficiency and reduce the total fuel consumption [102] by maximizing:

knet⋅PFCnet + kfuel⋅Fueleff = f (x , AirFr , FuelFr , PLoad )

(7a)

subject to:

x ̇ = g (x , AirFr , FuelFr , PLoad ), x ∈ X

(7b)

where the weighting coefficients are knet= 0.5, kfuel= 25, and x, g, and PLoad are the state vector, the dynamics function of the FC system, and the disturbance input. The performance indicators used in this study are the FC net power (PFCnet), the FC system efficiency (ηsys), the fuel consumption efficiency (Fueleff ≅ PFCnet / FuelFr), and the total fuel consumption (FuelT), being defined by (8):

′ − Pcm PFCnet ≅ PFC

(8a)

ηsys =

PFCnet PFC

(8b)

Fuelefft

P = FCnet FuelFr

(8c)

FuelT =

∫ FuelFr (t ) dt

60000⋅R⋅(273 + θ)⋅NC⋅Iref (H 2) 2F ⋅(101325⋅Pf (H 2) )⋅(Uf (H 2)/100)⋅(xH 2 /100)

(10a)

60000⋅R⋅(273 + θ)⋅NC⋅⋅Iref (O2) 4F ⋅(101325⋅Pf (O2) )⋅(Uf (O2)/100)⋅(yO2 /100)

(10b)

where the constants and parameters are all defined in [104]. The FuelFr and AirFr can be selected as control variables of the FC power using the fuel and air regulators (10) and the reference currents Iref(H2) and Iref(O2) selected by switches SW1 and SW2 in order to implement one of the Boost-GES-RTO strategies or the sFF control strategy. The proposed GES control will search the optimum (the maximum value) of the optimization function f [105], optimizing the operation of the PEMFC system: low fuel consumption at high efficiency operation of the PEMFC system. The PEMFC safety operation will be ensured by including a slope limiters in both air and fuel regulators. Furthermore, these regulators will ensure the oxygen and hydrogen stoichiometry within admissible range. An interesting LQR/LQG control of oxygen stoichiometry is proposed in [106]. An optimal control for AirFr is proposed in [107] to maximize the FC net power. The FC net power is maximized using time delay control of AirFr [108]. Control of AirFr proposed in [109] could increase the PEMFC life time. Nonlinear control (such as the second order sliding mode control proposed in [110]), adaptive control [98], and ES control (such as the load governor strategy based on ES control proposed in [111]) are proposed as advanced control variants in the literature. In this study, the GES control is proposed as an advanced control of air flow rate or fuel flow rate. One of the fueling rates is optimized by GES control, the other is LFW controlled. Note that LFW control is different than the methods of intrusive and non-intrusive load monitoring of loads analyzed in [112]. Due to LFW control, the boost DC-DC converter will supply the DC bus with

(6h)

p3 = Am ⋅sd

(9)

′ PDC (AV ) = ηboost PFC

(11)

This value is evaluated as AV value of output power of the boost converter to be compared with the PDC(LPF), which is estimated by the LPF from the LFW block (see Fig. 7), having as input the pDC power (12):

(8d) 23

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

N. Bizon

Fig. 9. The diagram of the Battery/SMES hybrid ESS.

pDC = pload − pPV

(12)

simulation's results show that this value may be lower (less than 0.1F). During the pulse, the SMES current varies from ISMES - ΔISMES to ISMES + ΔISMES, and can store the energy (18):

Thus, considering (3), the AV power exchanged by ESS will be almost zero, which means that the battery operates in charge-sustaining mode if no other control conditions (such as the DC voltage regulation) will be implemented for it:

PBAT (AV ) ≅ 0

2

ΔELsmes =

(13)

The ESS block represented in Fig. 6 implements the diagram of the Battery/SMES hybrid ESS detailed in Fig. 9. The load 2 may contain many individual pulses, which can appear for ncycle(i) – times during a load cycle considered. If the amplitude and duration of each pulse(i) are ΔPload(i) and Δtload(i), then the energy of pulse(i) is given by (14):

ΔEpulse (i) = ΔPload (i)⋅Δtload (i)⋅ncycles (i)

LSMES = 2 ISMES ⋅⎡



(14)

∑ ΔPload (i)⋅Δtload (i)⋅ncycles (i)

(15)

i

The capacitor on DC bus (CDC) will be discharged or charged if the pulse is of consumption type or regenerative type. Limiting the voltage ripple at ± ΔVDC, the charging energy will be (16): 2

2

1 ΔVDC ⎞ ΔVDC ⎞ ⎤ 2 ⎡⎛ ⋅CDC⋅VDC ⋅⎢ 1 + − ⎛1 − 2 VDC ⎠ VDC ⎠ ⎥ ⎝ ⎣⎝ ⎦ ⎜







(16)

If ΔECdc ≅ ΔEload , then the value of the capacitance CDC will be given by (17):

2⋅ΔEload

CDC ≅ 2 ⎡ VDC ⋅



(

1+







(18)

ΔVDC 2 VDC

) (

− 1−

ΔVDC 2 ⎤ VDC

)⎦

(1 +

2⋅ΔEload ΔISMES 2 − ISMES

) (1 −

ΔISMES 2 ⎤ ISMES

)⎦

(19)

LSMES = 112 mH is obtained for the SMES inductance considering in (13) the same values of the load pulse, and ISMES = 500 A (the maximum value) and ΔISMES= 2.5% ISMES = 12.5 A. The value of 100 mH will be used first for the LSMES, but then the inductance was reduced to LSMES = 10 mH to see if the SMES still supplies/stores the power pulses. The SMES is modeled using a 100 mH/0 Ω inductance from Matlab & Simulink ® toolboxes. The shape of the current Ipulse is controled by signal PWMsmes applied to an asymmetric full-bridge DC-DC converter. The voltage on the capacitor CDC is initially set to VDC= 200 V and the battery voltage was chosen of Vbat= 100 V. So, 100 Ah/100 V lithium-ion battery from Matlab & Simulink ® toolboxes was chosen, with 80% initial SOC and the other values set to the default values. The battery's capacity (CBat) was estimated for same energy pulses and ΔVbat = 15%Vbat. If ΔEBat ≅ ΔEload , then the value of the capacity of battery CBat will be given by (20):

So, the energy of load 2 during a load cycle considered is (15)

ΔECdc =



If ΔELsmes ≅ ΔEload , then the value of the SMES inductance will be given by (19):

3.2. ESS design

ΔEload =

2

1 ΔISMES ⎞ ΔI 2 ⎛ − ⎛1 − SMES ⎞ ⎤ ⋅LSMES ⋅ISMES ⋅⎡ ⎢ 1 + ISMES 2 ISMES ⎠ ⎥ ⎠ ⎝ ⎣⎝ ⎦

CBat = (17)

ΔPload⋅ncycles⋅Tcycle (Bat ) ΔVBat

(20)

The battery's power flow to DC bus is controled by signal PWMbat via the half-bridge DC-DC converter. The 10 mH/0.01 Ω inductance (LBat) will operate the half-bridge DC-DC converter in continuous-current mode, considering DBat = VDC / VBat = 0.5 (the duty cycle), ΔIbat = 20%Ibat, Ibat < Ibat(max)= 100 A, and fmin = 250 kHz in (21):

The value obtained for the capacitance is CDC = 1.1667 F considering VDC= 200 V and voltage ripple of ΔVDC= 1,5% VDC = 3 V for only one type of pulse in (15) with the following parameters: ΔPload = 4Pload = 400 W, Δtload = 1% Tcycle = 3,5 ms, and ncycle= 1000. The value of 1F will be used first for the capacitor CDC, but the 24

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

N. Bizon

Fig. 10. The BMS diagram of control for the hybrid ESS.

LBat =

VDC⋅(1 − DBat ) 2⋅fmin ⋅ΔIBat

(HF) noise in the SMES converter control loop. The cut-off frequency of the LPFSMES is set at 1000 Hz in order to not mitigate the LF ripple considered (pload3) and to not distort the pulses too much (pload2). The pulses and LF ripples could be mitigated by appropriate command of the SMES power converter based on the errors ePulse = IPulse(ref) − IPulse(LPF), where IPulse(LPF) is the output of the LPFPulse (with the cutoff frequency of 1000 Hz):

(21)

3.3. ESS control The ESS control proposed in the literature is based on the Energy Flow Split Strategies (EFSS) [113–115] such as rules-based control [116], filtering-based control [117], and AI-based control [118,119]. The rule based control proposed in [116] uses rules tables to calculate the weighting parameters and select the operation modes. The filteringbased control is simple and efficient to separate the power in LPF and HPF frequency bands [117]. Most of proposed hybrid ESS are based on batteries and ultracapacitors as energy and power storage devices. In comparison with different energy storing applications based on Battery/UC ESS, only few papers report the use of SMES as power storage device [29,64,83,89,90,120–122]. In this study a Battery/SMES ESS is proposed to mitigate the load pulses or the LF ripple on DC bus in order to protect the PEMFC system. The BMS diagram of control for the hybrid ESS is presented in Fig. 10. The EFSS filtering-based control is implemented here considering the BPFSMES and LPFBat to generate the reference currents IPulse(ref) and IBat(ref) using as inputs Pload and dP given by (5) and (22):

dP = pESS = pload − pFC − pPV

IPulse (LPF ) = LPFPulse (Ipulse )

IPulse is the current that supply (or is generated by) the SMES power converter (see Fig. 10). If,

ePulse = 0

IPulse ≅

IBat (ref ) = LPFBat (dP /VDC )

(23b)

Pload2 + Pload3 VDC

(26)

Two control methods were implemented: one is based on the Proportional-Integral (PI) controller and the Pulse Width Modulation (PWM) generator, and other is based on hysteresis current-mode controller. It was observed that both control methods can ensure (25) with sufficient precision. So, the control method implemented for battery power converter uses a hysteresis current-mode controller based on the errors eBat = I′Bat(ref) – IBat, where

So, the reference currents IPulse(ref) and IBat(ref) will be given by (23): (23a)

(25)

then, considering (23a) and (24):

(22)

IPulse (ref ) = BPFSMES (Pload/ VDC )

(24)

′ (ref ) = IBat (ref ) − VDC (correction) IBat

(27)

VDC(correction) is the PI correction of the battery reference current (23b) based on the voltage errors eVdc = VDC−VDC(ref), if SW4 is on position shown in Fig. 10, or VDC(correction) = 0 if SW4 is switched to

The BPFSMES is used instead of HPFSMES to filter the high frequency 25

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

N. Bizon

Fig. 11. Searching of the optimal point for FC/Battery/SMES HPS under 6 kW load by using Air-LFW/Boost-GES-RTO strategy.

4. Results

Table 1 The Air-LFW/Boost-GES RTO strategy applied to FC/battery/SMES HPS at different Pload1. Pload1 [kW]

IFC1A [A]

FuelFr1A [lpm]

AirFr1A [lpm]

PFCnet1A [W]

ηsys1A [%]

Fueleff1A [W/lpm]

FuelT1A [l]

2 3 4 5 6 7 8

57.78 82.52 85.37 112.7 130.3 161.8 162.4

22.07 31.58 32.53 42.87 49.04 60.76 60.89

105.1 159.7 192.3 254.5 312.3 388.3 441.8

2513 3494 3928 4612 5206 5876 5861

92.32 90.61 90.3 88.59 86.98 85.05 84.18

113.2 109.6 120.2 111.9 106.7 97.96 97.36

52.49 73.01 74.85 98.5 117.8 136.6 137.4

The simulation results without RESs (PRES = 0) and without DC voltage regulation (by using VDC(correction) = 0) are shown in next two section for constant and dynamic load in order to compare the performance of the Boost-GES-RTO strategies. 4.1. Constant load The pload1 is a constant load if switch SW6 is respective position (constant load) in Fig. 4. The noise added to it is given by pload2, and pload3 shown in Fig. 5 and Fig. 11 (the second plot). Fig. 11 represents the searching of the optimal point for FC/Battery/SMES HPS under a 6 kW load by using Air-LFW/Boost-GES-RTO strategy. The structure of the plots is as follows: first two plots represent the load components; the variation of the DC voltage (VDC) around the reference VDC(ref) = 200 V is represented in the 3rd plot (the voltage drop is about 14 V during startup without DC voltage regulation (by using VDC(correction) = 0), but then the DC voltage fluctuates around reference due to the battery's operation in charge-sustaining mode; the battery's power (pBat) is represented in 4th plot (it is observed that

0 V. In both cases, (13) is ensured during the stationary regimes. Considering [114], the proportional gains of the PI controllers for DC voltage error (eVdc) and pulse current error (ePulse) are 5 and 1, and the integral gains are 1 for both PI controllers. The hysteresis value was chosen 0.1 A for both hysteresis current-mode controllers.

26

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

N. Bizon

Fig. 12. Searching of the optimal point for FC/Battery/SMES HPS under 6 kW load by using Fuel-LFW/Boost-GES-RTO strategy. Table 2 The Fuel-LFW/Boost-GES RTO strategy applied to FC/Battery/SMES HPS at different Pload1. Pload1 [kW]

IFC2A [A]

FuelFr2A [lpm]

AirFr2A [lpm]

PFCnet2A [W]

ηsys2A [%]

Fueleff2A [W/lpm]

FuelT2A [l]

2 3 4 5 6 7 8

44.59 67.12 92.4 120.2 139.6 172.4 208.5

16.11 24.86 34.04 44.15 52.59 65.74 79.91

101.3 156.1 212 272.6 314.3 392.3 473.9

2111 3119 3992 4798 5475 6306 6977

92.36 90.82 90.08 88.39 86.94 84.95 83.14

131.9 124 116.5 108.6 104.8 95.62 87.12

38.6 59 80.83 104.5 124.5 153.3 181.6

Table 3 The sFF control applied to FC/Battery/SMES HPS at different Pload1. Pload1 [kW]

IFC3 [A]

FuelFr3 [lpm]

AirFr3 [lpm]

PFCnet3 [W]

ηsys3 [%]

Fueleff3 [W/lpm]

FuelT3 [l]

2 3 4 5 6 7 8

36.62 58.95 82.62 108.1 138.9 173 220.6

14.58 24.05 31.65 41.45 53.25 67.5 82.8

85.52 144.2 189.2 248.1 318.1 404.6 496.2

1935 2991 3773 4632 5441 6178 6912

93.15 91.55 90.25 88.51 86.57 84.19 82.3

135.7 126.2 119.9 111.5 102.6 91.9 83.75

34.02 56.3 74.88 98.6 125.58 158.34 196

Iref (O2) =

battery ensure the power balance (1) until the PEMFC will generate the requested power by the load, but then PBat(AV) ≅ 0). The FuelFr (represented in the 6th plot) is controlled by the FC current (represented in the 5th plot), so both signals have the same shape. The FC current follows the reference current IrefGES (6j), searching the maximum of the optimization function f (7a). Due to the LFW control (4), the reference current Iref(O2) will follow the load demand on DC bus based on (28) (see the AirFr shown in the 8th plot):

PDC (AV ) ′ PFC = VFC ηboost VFC

(28)

The performance indicators (8) are also shown in Fig. 11: the total fuel consumption (FuelT) in the 7th plot, the fuel consumption efficiency (Fueleff ≅ PFCnet / FuelFr) in the 9th plot, and the FC system efficiency (ηsys) in the last plot. The results of simulations for different levels of constant load Pload1 are summarized in Table 1. 27

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

N. Bizon

Table 4 The performance indicators difference for Air-LFW/Boost-GES RTO strategy compared to sFF control. Pload1 [kW]

ΔPFCnet=PFCnet1A-PFCnet3 [W]

Δηsys= ηsys1A-ηsys3 [%]

ΔFueleff=Fueleff1A-Fueleff3 [W/lpm]

ΔFuelT1A=FuelT1A-FuelT3 [l]

60·ΔFuelT1A/ 12 [lpm]

2 3 4 5 6 7 8

578 503 155 − 20 − 235 − 302 − 1051

− 0.83 − 0.94 0.05 0.08 0.41 0.86 1.88

− 22.5 − 16.6 0.3 0.4 4.1 6.06 13.61

18.47 16.71 − 0.03 − 0.1 − 7.78 − 21.74 − 58.6

92.35 83.55 − 0.15 − 0.5 − 38.9 − 108.7 − 293

Table 5 The performance indicators difference for Fuel-LFW/Boost-GES RTO strategy compared to sFF control. Pload1 [kW]

ΔPFCnet=PFCnet2A-PFCnet3 [W]

Δηsys= ηsys2A-ηsys3 [%]

ΔFueleff=Fueleff2A-Fueleff3 [W/lpm]

ΔFuelT2A=FuelT2A-FuelT3 [l]

60·ΔFuelT2A/12 [lpm]

2 3 4 5 6 7 8

176 128 219 166 34 128 65

− 0.79 − 0.73 − 0.17 − 0.12 0.37 0.76 0.84

− 3.8 − 2.2 − 3.4 − 2.9 2.2 3.72 3.37

4.58 2.7 5.95 5.9 − 1.08 − 5.04 − 14.4

22.9 13.5 29.75 29.5 − 5.4 − 25.2 − 72

Fig. 13. The difference in Fueleff (see Tables 4 and 5) using the Boost-GES-RTO strategy with Air-LF (♦) and Fuel-LF (■) compared to sFF control. Fig. 15. The difference in FuelT (see Tables 4 and 5) using the Boost-GES-RTO strategy with Air-LF (♦) and Fuel-LF (■) compared to sFF control.

Fig. 14. The difference in ηsys (see in Tables 4 and 5) using the Boost-GES-RTO strategy with Air-LF (♦) and Fuel-LF (■) compared to sFF control.

Then, the Fuel-LF/Boost-GES-RTO strategy is analyzed. Fig. 12 represents the searching of the optimal point for FC/Battery/SMES HPS under the same 6 kW load by using Fuel-LF/Boost-GES-RTO strategy in order to compare the results. The structure of the plots of Fig. 12 is the same as of Fig. 11. The DC voltage (VDC) varies around the reference VDC(ref)= 200 V

Fig. 16. The optimal points position for the Boost-GES RTO strategies with AirLF ( ) and Fuel-LF ( ) compared to sFF control ( ).

28

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

N. Bizon

Table 6 Fuel economy for the Air-LFW/Boost-GES-RTO compared to sFF control. Pload1(AV)

sFF-RTO strategy

[kW]

IFC3(AV) [A]

FuelFr3(AV) [lpm]

AirFr3(AV) [lpm]

FuelT3(LC) [l]

IFC2A(AV) [A]

FuelFr2A(AV) [lpm]

AirFr2A(AV) [lpm]

2 3 4 5 6 6.25

26.07 43.92 57.23 73.38 90.1 104.2

10.39 16.39 22.42 28.99 36 40.53

62.86 97.59 134.2 173.3 215.3 244

34.52 54.23 75.96 100.4 129.1 149

46.83 63.55 78.81 79.25 99.16 104.7

17.6 24.05 29.7 30.43 37.28 39.42

79.65 117.5 154.4 179.1 221.5 233.6

ΔFuelT2A(LC)=FuelT2A(LC)-FuelT3(LC)

60·ΔFuelT2A(LC)/12

FuelT2A(LC) [l]

[l]

[lpm]

49.33 64.49 82 96.79 109.9 112.1

14.81 10.26 6.04 − 3.61 − 19.2 − 36.9

74.05 51.3 30.2 − 18.05 − 96 − 184.5

ΔFuelT1A(LC)=FuelT1A(LC)-FuelT3(LC)

60·ΔFuelT1A(LC)/12

Air-LF/Boost-GES RTO strategy

Table 7 Fuel economy for the Fuel-LFW/Boost-GES-RTO compared to sFF control. Pload1(AV)

sFF-RTO strategy

Fuel-LF/Boost-GES RTO strategy

[kW]

IFC3(AV) [A]

FuelFr3(AV) [lpm]

AirFr3(AV) [lpm]

FuelT3(LC) [l]

IFC1A(AV) [A]

FuelFr1A(AV) [lpm]

AirFr1A(AV) [lpm]

FuelT1A(LC) [l]

[l]

[lpm]

2 3 4 5 6 6.25

26.07 43.92 57.23 73.38 90.1 104.2

10.39 16.39 22.42 28.99 36 40.53

62.86 97.59 134.2 173.3 215.3 244

34.52 54.23 75.96 100.4 129.1 149

34.01 48.6 65.62 85.78 107.4 110.8

12.03 18.09 24.38 31.5 39.02 40.81

77.71 113.3 152.1 196 242.6 252

38.43 58.09 79.21 103.2 125.9 132.4

3.91 3.86 3.25 2.8 − 3.2 − 16.6

19.55 19.3 16.25 14 − 16 − 83

searching the maximum of the optimization function f (7a). Thus, the performance indicators (8) are a bit different, being represented in Fig. 11 (the total fuel consumption (FuelT) in the 7th plot, the fuel consumption efficiency (Fueleff ≅ PFCnet / FuelFr) in the 9th plot, and the FC system efficiency (ηsys) in the last plot) and summarized in Table 2 for different levels of constant load Pload1. The sFF control was chosen as reference strategy to control the FC/ Battery/SMES HPS due to its frequent use in different applications reported in the literature [35–41], including the commercial ones [37,38]. The results obtained are summarized in Table 3 for different levels of constant load demand (Pload1). The performance indicators difference using the data from Tables 1 and 2 for Boost-GES RTO strategies compared to sFF control (Table 3) are computed in Tables 4 and 5. The gaps in performance indicators shown in Tables 4 and 5 are represented in Fig. 13 (Fueleff), Fig. 14 (ηsys), and Fig. 15 (FuelT). The following conclusions result from analysis of these results: Fig. 17. Fuel (see Tables 6 and 7) for the Boost-GES RTO strategy with Air-LF (♦) and Fuel-LF (■) compared to sFF control.

– In range of load demand from 3.8 kW up to 8 kW, the Air-LFW/ Boost-GES RTO strategy compared to Fuel-LFW/Boost-GES RTO strategy gives better results in all performance indicators; – For light load (up to 3.8 kW), the sFF control compared to both Boost-GES RTO strategies gives better results in all performance indicators. – In range of load power from 4 kW up to 8 kW, the Air-LFW/BoostGES RTO strategy compared to sFF control gives better results in all performance indicators; furthermore, the differences in all performance indicators increase with the level of the load power.

(see the 3rd plot) in the same manner as shown in Fig. 11, having a voltage drop of about 14 V during startup if the DC voltage regulation by using VDC(correction) is not used. The battery's power (pBat) have the same variation as shown in Fig. 11 (see the 4rd plot) due to the battery's operation in charge-sustaining mode that results from the power balance (1). Minor differences appear in shapes of signals related to ESS behavior because the same ESS control is used in both Boost-GES-RTO strategies. But different control loops are used for each Boost-GES-RTO strategy. In case of the Fuel-LF/Boost-GES-RTO strategy, the LFW control (4) is applied to the reference current Iref(H2):

Iref (H 2) =

PDC (AV ) ′ PFC = VFC ηboost VFC

The position of optimal points for Boost-GES RTO strategies compared to sFF strategy are presented in the phase plane of the control variables FuelFr and AirFr based on data from Tables 1–3 (see Fig. 16). This illustrates that optimization function is a dynamic surface depending by controllable variables FuelFr and AirFr in the control loops of each strategy considered here. Note that even if the positions of optimal points are close one to another, the values obtained for performance indicators may be very different [105] due to dynamic surface that depends by process's inputs and used strategy [123,124].

(29)

so the FuelFr (represented in the 6th plot) will follow the load demand on DC bus based on (29). The AirFr (represented in the 8th plot) is controlled by the FC current (represented in the 5th plot), so both signals have the same shape. The FC current follows the reference current IrefGES (6j), 29

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

N. Bizon

Fig. 18. Searching of optimal points by using Air-LFW/Boost-GES RTO strategy for RES/FC/Battery/SMES HPS under 6 kW AV-LC.

strategy; furthermore, the fuel economy increases with the level of the load power.

4.2. Variable load The pload1 is a variable load if switch SW6 is on position shown in Fig. 4. Also, the noise added to it is given by pload2, and pload3 shown in Fig. 5 and Fig. 11 (the second plot). The findings for constant load must also be verified for a dynamic load with stairs profile (pload1) having the levels and the AV value (Pload1(AV)) mentioned the same as in [33] for further comparison of the results obtained with different RTO strategies. For dynamic load, only FuelT performance indicator can give relevant information about the Boost-GES-RTO strategies compared to sFF strategy. The simulations have been performed and the results are presented in Tables 6 and 7. The gaps in FuelT performance indicator shown in Tables 6 and 7 are represented in Fig. 17. The following conclusions result from analysis of these results:

Therefore, the Air-LFW/Boost-GES RTO strategy will be considered below to be tested with a variable profile of RESs (which can be available), but still without DC voltage regulation. 4.3. Dynamic load and variable RES power The simulation results in case of variable RES power and dynamic profile of load demand is shown in Fig. 18 for the Air-LFW/Boost -GESRTO strategy. The profiles of the RES power and dynamic load (pload1) are represented in first two plots of Fig. 18. The rest of the plots are as follows: the variation of the DC voltage (VDC) around the reference VDC(ref)= 200 V is represented in the 3rd plot (in addition to voltage drop of about 14 V during startup, others voltage variations of maximum 5 V can be observed due to changes in both RES and load power profiles; anyway the DC voltage still fluctuates around reference due to the battery's operation in charge-sustaining mode); the shape of the battery's power (pBat) represented in 4th plot confirm this mode of operation (PBat(AV) ≅ 0). The FuelFr (represented in the 6th plot) follows the FC current (represented in the 5th plot) due to (28), but AirFr (represented in the

– In range of load power from 4.3 kW up 8 kW, the fuel economy is higher for Air-LFW/Boost-GES RTO strategy compared to Fuel-LFW/ Boost-GES RTO strategy; – For light load (up to 4.6 kW), the fuel economy is higher for sFF strategy compared to Air-LFW/Boost-GES RTO strategy. – In range of load power from 4.6 kW up 8 kW, the fuel economy is higher for Air-LFW/Boost-GES RTO strategy compared to sFF 30

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

N. Bizon

a Fig. 19. The behavior of the Battery/SMES hybrid ESS under the two patterns of pulses by using Air-LFW/Boost-GES-RTO strategy, hysteresis current-mode controller for both SMES and battery power converters, and DC voltage regulation. (a) Under the first pattern of pulses considered. (b) Under the second pattern of pulses considered.

next section considering the error ePulse (25)

8th plot) follows the load demand on DC bus based on (27). The shapes of the RES power profile can be identified in the AirFr variation. The performance indicators (8) are also shown in Fig. 18: the total fuel consumption (FuelT) in the 7th plot, the fuel consumption efficiency (Fueleff ≅ PFCnet / FuelFr) in the 9th plot, and the FC system efficiency (ηsys) in the last plot. A fair comparison of the RTO strategies for RES/PEMFC/ESS HPS can be made if the fuel economy is compared under same test conditions (related to load cycle, RES profile etc.) [31,33,125–127]. This comparison was made for Boost-GES RTO strategies proposed here, and the results obtained confirm the superiority of the Air-LF/Boost-GES RTO strategy (but the results are not shown here because is outside of the goal of this study). It will be interesting to compare in next study the Air-LF/Boost-GES RTO strategy proposed here with others RTO strategy proposed in [33,102,105] based on LFW control and GES control (one or two GES control loops) but different applied to controllable inputs. In this study the mitigation of load pulses and LF load ripples will be analyzed considering the hysteresis current-mode control for both battery and SMES power converters. Thus, the behavior of the Battery/ SMES ESS with voltage regulation of the DC voltage, and mitigation control of the load pulses and LF load ripples will be analyzed below. Note that the mitigation effect could be observed on the DC voltage presented in previous Figs. (11 and 12), as only very small voltage sag during the pulse, but the performance of mitigation will be shown in

4.4. Battery/SMES ESS behavior The mitigation control will use the hysteresis current-mode control, but the PI control is also tested here (even if this is not recommended to track a sinusoidal (alternative)) reference due to error in both magnitude and phase parameters [31,35,57,58,125,128], but some improved schemes are reported recently for both controllers [129–131] to avoid their disadvantages [125]. The hysteresis controller generates directly the PWM commands (without a PWM generator that is mandatory for PI controller), but the switching frequency varies in a large band, generating harmonics in frequency band considered (21). The LPF for Ipulse (LPF_Pulse in Fig. 10) is designed to filter these HF harmonics. Besides the LF load ripple pload3 (a superposition of three rectified sinusoidal signals of frequency f0, 3f0, and 5f0, where f0 = 50 Hz), two patterns are used for the load pulses (see Fig. 19a and b) to test the capability of SMES convertor control to mitigate the pulses. Fig. 19 illustrates the behavior of the Battery/SMES hybrid ESS under the pulses shown in the last plot of Fig. 19 by using Air-LF/Boost-GES-RTO strategy, hysteresis current-mode controller for both SMES and battery power converters, and DC voltage regulation. The first pattern (see Fig. 19a) contains one power pulse (with 0.0225 s width and power level of 800 W), and two more short and 31

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

N. Bizon

b Fig. 19. (continued)

SMES can compensate these changes in load demand pDC, but also short load pulses and LF load ripples.

small pulses (one positive and other negative with power of 40 W, but both with 0.1 s width). The second pattern (see Fig. 19a) will also contains the short power pulse, but the power of positive and negative pulses is now of 400 W (instead of 40 W). The structure of plots for Fig. 19a and b are the same and minor differences can be observed between the shapes of same signals: the DC voltage regulation is represented in first plot (the use of VDC(correction) for battery reference current (23b) based on PI correction of the voltage errors, eVdc = VDC - VDC(ref), reduces the voltage sag during startup at less than 0.8 V compared with 14 V without DC voltage regulation); the SMES current is represented in second plot, and the SMES power and remaining SMES energy (30) are represented in 6th and 7th plots; the battery current is represented in 4th plot, and the battery power and the battery SOC are represented in 5th and 8th plot (the battery supply the DC bus with energy to regulate the DC voltage and sustain the power flow balance (1) during changes in load or RES power). In fact, the both energy and power storage devices compensate dynamically the power flow balance on the DC bus during changes in load demand pDC (12). The shape (30) of the SMES energy shows that

ESMES =

∫ pSMES dt

(30)

The mitigation process of short load pulses and LF load ripples will be better shown in Figs. 20–22. Figs. 20 and 21 represent the behavior of ESS for the first and second pattern of the pulses during the startup phase (Fig. 20a and Fig. 21a), the pulse (Fig. 20b and Fig. 21b), and the LF ripple (Fig. 20c and Fig. 21c) by using hysteresis current-mode controller to generate the commands for the SMES power converter. The structure of the plots is the same for both Figs. 20 and 21. Fig. 20a and Fig. 21a represent the signals during the startup phase: the reference current IPulse(ref) (in the first plot), the pulse (IPulse) generated by the SMES converter (in the second plot), and the errors ePulse = IPulse(ref) - IPulse(LPF) (in the 3rd plot), where IPulse(LPF) is given by (24). Fig. 20b and Fig. 21b represent the same signals during the phase of a load pulse and Fig. 20c and Fig. 21c represent the same signals during the phase of a load LF ripple. It is observed that the pulse (IPulse) supplied (or generated) by the SMES converter can follow the reference 32

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

N. Bizon

a

b

c Fig. 20. The generation of the Ipulse considering the first pattern of pulses (using Air-LFW/Boost-GES-RTO strategy and hysteresis current-mode controller for SMES power converter). (a) The startup phase. (b) Zoom of the pulse. (c) Zoom of the LF ripple.

the main problem remains at ESS level [113–119] in choosing the simplest strategy to generate the reference currents IPulse(ref) and IBat(ref) (23), which seems to be one that is based on filtering [117]. The cut-off frequency of the sharing filters LPFFC and HPFSMES was chosen here of 0.1 Hz. The value of this sharing frequency is important in sizing the power storage device without support of the battery in ensuring the power flow balance (1) [117]. A lower value will better mitigate the LF ripple from the reference current for the FC system (ILF(ref)), but the response time of the LPFFC increase as well. So, a faster type for the LPFFC must be used [117]. Other problem is where the DC voltage regulation will be performed as correction (VDC(correction)) of the reference current: on the control side of the PEMFC system, the battery, or the power storage device (UC or

current IPulse(ref) with peak error ePulse less than 0.5 A. Also, the battery current will follow the reference current I′Bat(ref) with peak error eBat = I′Bat(ref) – IBat less than 1 A. This error generates a HF power ripple with level less than 100 W and 200 W, which will be filtered by the capacitor on DC bus (CDC). The DC voltage in Fig. 19 has no HF ripple. To compare the results using the same structure of plots as Figs. 20 and 21, Fig. 22 represents the generation of the Ipulse considering the second pulse pattern and the PI controller to command the SMES power converter. 4.5. Discussion Thus, based on the results obtained and reported in the literature, 33

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

N. Bizon

a

b

c Fig. 21. The generation of the Ipulse considering the second pattern of pulses (using Air-LFW/Boost-GES-RTO strategy and hysteresis current-mode controller for SMES power converter). (a) The startup phase. (b) Zoom of the pulses. (c) Zoom of the LF ripple.

The DC voltage regulation was tested here at the FC control side as well considering the same time response for FC and battery converters. The results obtained are less good due to use of slope limiters for the fueling rate flows.

SMES). For example, in FC/UC HPS analyzed in [117], the DC voltage regulation was performed at both FC control and UC control sides, but the results are commented only from point of view of power sharing between the FC and UC. The results shown in [33,132] for FC/UC HPS with DC voltage regulation at the UC control side have shown that the battery power results more noisy than the shape obtained here, where the DC voltage regulation is performed at the battery control side. Anyway, the control cannot be implemented here at the SMES control side because the SMES control is designed to mitigate the load pulses and the correction of the IPulse(ref)will deteriorate the mitigation control performance.

5. Conclusion The HPS architectures and the ESS topologies proposed in the literature are discussed here in frame of the space applications, where extreme environments (very low temperature, intense radiation environments etc.) and dynamic load demand (including load pulses) are standard test conditions. The potential energy sources (including FC 34

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

N. Bizon

Fig. 22. The generation of the Ipulse considering the second pattern of pulses (and using Air-LFW/Boost-GES-RTO strategy and PI controller for SMES power converter).

The results obtained here have been discussed compared to other HPS architectures, ESS hybridizations and control solutions reported in the literature.

systems and PV array) and the reliable technologies for HPS and hybrid ESS are also discussed here in frame of the performance (power and energy density, efficiency, and lifetime) for space applications and future targets. The optimization strategies proposed to optimally operate the Fuel Cell (FC) system using two control loops implemented based on the global optimization control of the boost DC-DC converter and the LFW control of the FuelFr or the AirFr are analyzed for constant load, dynamic load, and without RES power. The main findings for constant load are: (1) Air-LFW/Boost-GES RTO strategy compared to Fuel-LFW/Boost-GES RTO strategy gives better results for all performance indicators in range of load power from 3.8 kW up 8 kW; (2) Air-LFW/Boost-GES RTO strategy compared to sFF strategy (the reference strategy) gives better results across all performance indicators; (3) the positions of optimal points are close one to another, but the values obtained for performance indicators are very different due to dynamic of the optimization surface; (4) the gaps across all performance indicators increase with the level of the load power. For example, if the maximum load is considered, the gaps compared with the reference strategy are of 1.88%, 13.61 W/lpm, and 293 lpm for FC system efficiency, fuel consumption efficiency, and fuel economy. The main findings for dynamic load are: (1) the fuel economy is higher for Air-LFW/Boost-GES RTO strategy compared to Fuel-LFW/ Boost-GES RTO strategy in range of load power from 4.3 kW up 8 kW; the fuel economy is higher for Air-LFW/Boost-GES RTO strategy compared to sFF strategy in range of load power from 4.6 kW up 8 kW; (3) fuel economy increase with the level of the load power. The robustness of the Air-LFW/Boost-GES RTO strategy is tested for dynamic load and variable RES power. Also, the control of the battery/ SMES hybrid ESS is analyzed in different scenarios. The mitigation control of the load pulses and LF load ripples based on hysteresis current-mode control and PI controller with PWM generator is proposed for the SMES power converter. The pulse (IPulse) supplied (or generated) by the SMES converter follows (with a peak error less than 0.5 A) the reference current IPulse(ref), based on sharing method in frequency band of the load profile. Different load pulses are used to test the performance of mitigation control proposed at the ESS side to protect the PEMFC system. The DC voltage regulation was implemented here on the control side of the PEMFC system and the battery considering the same time response for the PEMFC and battery converters. The best results are obtained in first case because the slope limiters of the fueling rate flows used for safe operation of the PEMFC system increase the time response in the voltage regulation loop. For example, without the DC voltage regulation, the voltage drop is of about 14 V during startup, but this can be reduced at less than 0.8 V if the DC voltage regulation is implemented.

Acknowledgements This work was supported by a grant of the Ministry of National Education and Scientific Research, Romania (167STAR) within RDI Program for Space Technology and Advanced Research - STAR, project number 167/2017: “Concept Development of an Energy Storage Unit Using High Temperature Superconducting Coil for Spacecraft Power Systems (SMESinSpace)”. References [1] Belz S. A synergetic use of hydrogen and fuel cells in human space flight power systems. Acta Astronaut 2016;121:323–31. [2] (a) Summerer L. Thinking tomorrows' space – research trends of the ESA advanced concepts team 2002–2012. Acta Astronaut 2014;95:242–59; (b) Bhogilla SS, Ito H, Kato A, Nakano A. Research and development of a laboratory scale totalized hydrogen energy utilization system. Int J Hydrog Energy 2016;41:1224–36. [3] Frischauf N. Hydrogen-fueled spacecraft and other space applications of hydrogen. In: Subramani Angelo Basile, Veziroglu T Nejat, editors. Compendium of hydrogen energy: hydrogen production and purification. Woodhead Publishing; 2015. p. 87–107 [chapter 5]. [4] NASA. Summary: space applications of hydrogen and fuel cells. 2015〈http://www. nasa.gov/topics/technology/hydrogen/hydrogen_2009.html〉 [Accessed in January 2018]. [5] Gruntman M. Blazing the trail: the early history of spacecraft and rocketry. Reston, USA: AIAA (The American Institute of Aeronautics and Astronautics); 2004. [6] Wang J. System integration, durability and reliability of fuel cells: challenges and solutions. Appl Energy 2017;189:460–79. [7] IEC 62282-3-100. Fuel cell technologies - Part 3-100: Stationary fuel cell power systems – Safety; 2012. [8] IEC 62282-IEC 625. Fuel cell technologies - Part 5-100: Portable fuel cell power systems - Safety; 2015. [9] IEC 62282-3-200. Fuel cell technologies – Part 3-200: Stationary fuel cell power systems – Performance test methods; 2015. [10] Barbir F, Molter T, Dalton L. Efficiency and weight trade-off analysis of regenerative fuel cells as energy storage for aerospace applications. Int J Hydrog Energy 2005;30:351–7. [11] Andrews J, Doddathimmaiah AK. Regenerative fuel cells. In: Gasik M, editor. Fuel cell materials. Cambridge: Woodhead Publishing; 2008. [12] Doddathimmaiah AK, Andrews J. Theory, modelling and performance measurement of unitised regenerative fuel cells. Int J Hydrog Energy 2009;34:8157–70. [13] Oh TH. Design specifications of direct borohydride–hydrogen peroxide fuel cell system for space missions. Aerosp Sci Technol 2016;58:511–7. [14] Paul P, Andrews J. PEM unitised reversible/regenerative hydrogen fuel cell systems: state of the art and technical challenges. Renew Sust Energ Rev 2017;79:585–99. [15] Bizon N. Effective mitigation of the load pulses by controlling the battery/SMES hybrid energy storage system. Appl Energy 2018;229:459–73. [16] Emadi A, Williamson SS. Status review of power electronic converters for fuel cell applications. J Power Electron 2002;1(2):133–44. [17] Bizon N, Mazare AG, Ionescu LM, Enescu FM. Optimization of the proton exchange membrane fuel cell hybrid power system for residential buildings. Energ Convers

35

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

N. Bizon

nanotubes. Energy 2014;76:911–9. [52] Nguyen BM, Hoffman D, Huang EK-wei, Bogdanov S, Delaunay PY, Razeghi M, Tidrow MZ. Current-matched triple-junction solar cell reaching 41.1% conversion efficiency under concentrated sunlight. Appl Phys Lett 2009;94(22):223504. https://doi.org/10.1063/1.3148326. [53] Gutera W, Schöne J, Philipps SP, Steiner M, Siefer G, Wekkeli A, Welser E, Oliva E, Bett AW, Frank F. Current-matched triple-junction solar cell reaching 41.1% conversion efficiency under concentrated sunlight. Appl Phys Lett 2009;94(22). https://doi.org/10.1063/1.3148341. [54] Leite MS, Woo RL, Munday JN, Hong WD, Mesropian S, Law DC, Atwater HA. Towards an optimized all lattice-matched InAlAs/InGaAsP/InGaAs multijunction solar cell with efficiency > 50%. Appl Phys Lett 2013;102(3). https://doi.org/10. 1063/1.4758300. [55] Summerer L, Ongaro F. Advanced space technology for 21st century energy systems: solar power from space. In: Proceedings of 2nd international conference on recent advances in space technologies (RAST), 2005 doi: 10.1109/RAST.2005. 1512527. [56] Shin D, Lee K, Chang N. Fuel economy analysis of fuel cell and supercapacitor hybrid systems. Int J Hydrog Energy 2016;41:1381–90. [57] Bizon N. Nonlinear control of fuel cell hybrid power sources: part II –current control. Appl Energy 2011;88(7):2574–91. [58] Bizon N. Nonlinear control of fuel cell hybrid power sources: part I –voltage control. Appl Energy 2011;88(7):2559–73. [59] Olabi AG. Renewable energy and energy storage systems. Energy 2017;136:1–6. [60] Penthia T, Panda AK, Sarangi SK. Implementing dynamic evolution control approach for DC-link voltage regulation of superconducting magnetic energy storage system. Int J Electr Power Energy Syst 2018;95:275–86. [61] Elsisi M, Soliman M, Aboelela MAS, Mansour W. Optimal design of model predictive control with superconducting magnetic energy storage for load frequency control of nonlinear hydrothermal power system using bat inspired algorithm. J Energy Storage 2017;12:311–8. [62] Hemmati R, Saboori H. Emergence of hybrid energy storage systems in renewable energy and transport applications – a review. Renew Sust Energ Rev 2016;65:11–23. [63] Wang H, Wang Q, Baozan Hu B. A review of developments in energy storage systems for hybrid excavators. Autom Constr 2017;80:1–10. [64] Li J, Gee AM, Zhang M, Yuan W. Analysis of battery lifetime extension in a SMESbattery hybrid energy storage system using a novel battery lifetime model. Energy 2015;86:175–85. [65] Guneya MS, Tepe Y. Classification and assessment of energy storage systems. Renew Sust Energ Rev 2017;75:1187–97. [66] Sarasketa-Zabala E, Martinez-Laserna E, Berecibar M, Gandiaga I, RodriguezMartinez LM, Villarreal I. Realistic lifetime prediction approach for Li-ion batteries. Appl Energy 2016;162:839–52. [67] Reddy T. Linden's handbook of batteries. 4th edition McGraw-Hill Education; 2010. [68] Julien C, Mauger A, Vijh A, Zaghib K. Lithium batteries – science and technology. Springer; 2016. [69] Warner J. The handbook of lithium-ion battery pack design – chemistry, components, types and terminology. Elsevier Science; 2015. [70] Li W, Zeng L, Wu Y, Yu Y. Nanostructured electrode materials for lithium-ion and sodium-ion batteries via electrospinning. Sci China Mater 2016;59(4):287–321. [71] Zhang C, Wei YL, Cao PF, Lin MC. Energy storage system: current studies on batteries and power condition system. Renew Sust Energ Rev 2017. https://doi. org/10.1016/j.rser.2017.10.030. [72] Shah K, Balsara N, Banerjee S, Chintapalli M, Cocco AP, Chiu WKS, Lahiri I, Martha S, Mistry A, Mukherjee PP, Ramadesigan V, Sharma CS, Subramanian VR, Mitra S, Jain A. State of the art and future research needs for multiscale analysis of Li-Ion cells. J Electrochem En Conv Stor 2017;14(2). [73] Sarasketa-Zabala E, Martinez-Laserna E, Berecibar M, Gandiaga I, RodriguezMartinez LM, Villarreal I. Realistic lifetime prediction approach for Li-ion batteries. Appl Energy 2016;162:839–52. [74] Tao L, Ma J, Cheng Y, Noktehdan A, Chong J, Lu C. A review of stochastic battery models and health management. Renew Sust Energ Rev 2017;80:716–32. [75] Mukherjee R, Krishnan R, Lu TM, Koratkar N. Nanostructured electrodes for highpower lithium ion batteries. Nano Energy 2012;1(4):518–33. [76] NASA. Flywheel Program 2015. 〈https://www.grc.nasa.gov/WWW/portal/pdf/ flywheel.pdf〉 [Accessed in January 2018]. [77] Hedlund M, Lundin J, de Santiago J, Abrahamsson J, Bernhoff H. Flywheel energy storage for automotive applications. Energies 2015;8:10636–63. [78] Wang B. Current Flywheels moving to Superconducting flywheels using carbon fiber or carbon nanotubes; 2017. 〈https://www.nextbigfuture.com/2017/01/ukbuilding-38-million-combat-laser.html〉 [Accessed in January 2018]. [79] Ha SK, Kim MH, Han SC, Sung TH. Design and spin test of a hybrid composite flywheel rotor with a split type hub. J Compos Mater 2006;40:2113–30. [80] Jin JX. HTS energy storage techniques for use in distributed generation systems. Phys C: Supercond Appl 2007;460–462(Part 2):1449–50. [81] Gubser DU. Superconductivity research and development: department of defense perspective. Appl Supercond 1995;3(1–3):157–61. [82] Shawyer R. Second generation EmDrive propulsion applied to SSTO launcher and interstellar probe. Acta Astronaut 2015;116:166–74. [83] Yang J, Zhang L, Wang X, Chen L, Chen Y. The impact of SFCL and SMES integration on the distance relay. Phys C: Supercond Appl 2016;530:151–9. [84] Kangarlu MF, Pahlavani MRA. Cascaded multilevel converter based superconducting magnetic energy storage system for frequency control. Energy 2014;70:504–13.

Manag 2018;163:22–37. [18] Bizon N. Energy efficiency for the multiport power converters architectures of series and parallel hybrid power source type used in plug-in/V2G fuel cell vehicles. Appl Energy 2013;102:726–34. [19] Bizon N. Energy efficiency of multiport power converters used in Plug-In/V2G fuel cell vehicles. Appl Energy 2012;96:431–43. [20] Zhou H, Li YZ, Wang SN, Zhou GD. Performance of extravehicular space suit life support system based on cooling-heat-power integration. J Aerosp Power 2014;29(3):541–8. [21] ESA Telecommunications & Integrated Applications. 2017. 〈http://www.esa.int/ Our_Activities/Telecommunications_Integrated_Applications〉, [Accessed in January 2018]. [22] Vasko CA, Adriaensen M, Bretel A, Duvaux-Bechon I, Giannopapa CG. Space assets, technology and services in support of energy policy. Acta Astronaut 2017;138:295–300. [23] Elitzur S, Rosen V, Gany A. Combined energy production and waste management in manned spacecraft utilizing on-demand hydrogen production and fuel cells. Acta Astronaut 2016;128:580–3. [24] Wang S, Li Y, Li YZ, Peng X, Mao Y. Exergy based parametric analysis of a cooling and power co-generation system for the life support system of extravehicular spacesuits. Renew Energ 2018;115:1209–19. [25] Patel MR. Spacecraft power systems. Boca Raton, USA: CRC Press; 2004. [26] Shimizu T, Underwood C. Super-capacitor energy storage for micro-satellites: feasibility and potential mission applications. Acta Astronaut 2013;85:138–54. [27] Cabrières B, Alby F, Cazaux C. Satellite end of life constraints: technical and organisational solutions. Acta Astronaut 2012;73:212–20. [28] Da HS, Tan CW, Yatim AHM, Lau KY. Feasibility analysis of hybrid photovoltaic/ battery/fuel cell energy system for an indigenous residence in East Malaysia. Renew Sust Energ Rev 2017;76:1332–47. [29] Ali MH, Wu B, Dougal RA. An overview of SMES applications in power and energy systems. IEEE Trans Sustain Energy 2010;1(1):38–47. [30] Hannana MA, Hoque MM, Mohamed A, Ayob A. Review of energy storage systems for electric vehicle applications: issues and challenges. Renew Sust Energ Rev 2017;69:771–89. [31] Bizon N, Tabatabaei NM, Shayeghi H, editors. Analysis, Control and Optimal Operations in Hybrid Power Systems - Advanced Techniques and Applications for Linear and Nonlinear Systems. Springer; 2013. [32] NASA. Technology Roadmaps - TA 3: Space Power and Energy Storage, 〈https:// www.nasa.gov/sites/default/files/atoms/files/2015_nasa_technology_roadmaps_ ta_3_space_power_energy_storage_final.pdf〉, [Accessed in January 2018]. [33] Bizon N. Real-time optimization strategy for fuel cell hybrid power sources with load-following control of the fuel or air flow. Energ Convers Manag 2018;157:13–27. [34] Buden D. Space nuclear fission electric power systems. Lakewood, USA: Polaris Books; 2011. [35] Bizon N, Tabatabaei NM, Blaabjerg F, Erol Kurt E, editors. Energy harvesting and energy efficiency: technology, methods and applications. NY, USA: Springer International Publishing; 2017. [36] Erdinç O, Vural B, Uzunoglu M. A wavelet-fuzzy logic based energy management strategy for a fuel cell/battery/ultra-capacitor hybrid vehicular power system. J Power Sources 2009;194:369–80. [37] Gou B, Na WK, Diong B. Fuel Cells: modeling, control, and applications. Boca Raton, USA: CRC Press; 2010. [38] Stolten D, Samsun RC, Garland N. Fuel Cells: data, facts, and figures. Weinheim, Germany: Wiley; 2016. [39] Sasaki K, Li HW, Hayashi A. Hydrogen energy engineering: a Japanese perspective. NY, USA: Springer International Publishing; 2016. [40] Nehrir MH, Wang C. Modeling and control of Fuel Cells: distributed generation applications. Weinheim, Germany: Wiley; 2009. [41] Bizon N, Dascalescu L, Tabatabaei NM, editors. Autonomous vehicles: intelligent transport systems and smart technologies. USA: Nova Science Publishers Inc.; 2014. [42] Frischauf N, Acosta-Iborra B, Harskamp F, Moretto P, Thomas Malkow T, Michel Honselaar M, Steen M, Hovland H, Hufenbach B, Schautz M, Wittig M, Soucek A. The hydrogen value chain: applying the automotive role model of the hydrogen economy in the aerospace sector to increase performance and reduce costs. Acta Astronaut 2013;88:8–24. [43] Thomas CE. Hydrogen-powered fuel cell electric vehicles compared to the alternatives, 2015. 〈http://www.azocleantech.com/article.aspx?ArticleID=214〉 [Accessed in January 2018]. [44] Obara S. Fuel Cell micro-grids. NY, USA: Springer International Publishing; 2009. [45] Hwang HT, Varma A. Hydrogen storage for fuel cell vehicles. Curr Opin Chem Eng 5. 2014. p. 42–8. [46] Eftekhari A, Fang B. Electrochemical hydrogen storage: opportunities for fuel storage, batteries, fuel cells, and supercapacitors. Int J Hydrog Energy 2017;42:25143–65. [47] Klebanoff L. Hydrogen storage technology: materials and applications. Boca Raton, USA: CRC Press; 2012. [48] Sone Y. A 100-W class regenerative fuel cell system for lunar and planetary missions. J Power Sour 2011;196:9076–80. [49] Kim T, Kwon S. Design and development of a fuel cell-powered small un-manned aircraft. Int J Hydrog Energy 2012;37:615–22. [50] Farooqui UR, Ahmad AL, Hamid NA. Graphene oxide: a promising membrane material for fuel cells. Renew Sust Energ Rev 2018;82:714–33. [51] Oh TH, Jang B, Kwon S. Performance evaluation of direct borohydride–hydrogen peroxide fuel cells with electrocatalysts supported on multiwalled carbon

36

Renewable and Sustainable Energy Reviews 105 (2019) 14–37

N. Bizon

[110] Kunusch C, Puleston PF, Mayosky MA, Fridman L. Experimental results applying second order sliding mode control to a PEM fuel cell based system. Control Eng Pract 2013;21(5):719–26. [111] Laghrouche S, Matraji I, Ahmed FS, Jemei S, Wack M. Load governor based on constrained extremum seeking for PEM fuel cell oxygen starvation and compressor surge protection. Int J Hydrog Energy 2013;38(33):14314–22. [112] Abubakar I, Khalid SN, Mustafa MW, Shareef H, Mustapha M. Application of load monitoring in appliances' energy management – a review. Renew Sust Energ Rev 2017;67:235–45. [113] Li Q, Yang H, Han Y, Li M, Chen W. A state machine strategy based on droop control for an energy management system of PEMFC - battery - supercapacitor hybrid tramway. Int J Hydrog Energy 2016;41:16148–59. [114] Bendjedia B, Rizoug N, Boukhnifer M, Bouchafaa F. Hybrid fuel cell/battery source sizing and energy management for automotive applications. IFAC Pap 2017;50(1):4745–50. [115] Santucci A, Sorniotti A, Lekakou C. Power split strategies for hybrid energy storage systems for vehicular applications. J Power Sour 2014;258:395–407. [116] Gokce K, Ozdemir A. A rule based power split strategy for battery/ultracapacitor energy storage systems in hybrid electric vehicles. Int J Electrochem Sci 2016;11:1228–46. [117] Turpin C, Morin B, Bru E, Rallieres O, Roboam X, Sareni B, Arregui GA, Roux N. Power for aircraft emergencies – a hybrid proton-exchange membrane H2/O2 fuel cell and ultracapacitor system. IEEE Electr Mag 2017;5(4):72–85. https://doi.org/ 10.1109/MELE.2017.2758879. [118] Wu J, Wang X, Li L, Qin C, Du Y. Hierarchical control strategy with battery aging consideration for hybrid electric vehicle regenerative braking control. Energy 2018;145:301–12. [119] Chong LW, Wong YW, Rajkumar RK, Isa D. An optimal control strategy for standalone PV system with battery-supercapacitor hybrid energy storage system. J Power Sour 2016;331:553–65. [120] Ise T, Kita M, Taguchi A. A hybrid energy storage with a SMES and secondary battery. IEEE Trans Appl Supercond 2005;15(2):1915. [121] Trevisani L, Morandi A, Negrini F, Ribani PL, Fabbri M. Cryogenic fuel-cooled SMES for hybrid vehicle application. IEEE Trans Appl Supercond 2009;19(3):2008. [122] Cansi A, Faydaci C, Qureshi MT, Usta O, McGuiness DT. Integration of a SMES–battery-based hybrid energy storage system into microgrids. J Supercond Nov Magn 2017. https://doi.org/10.1007/s10948-017-4338-4. [123] Bizon N. Global extremum seeking control of the power generated by a photovoltaic array under partially shaded conditions. Energ Convers Manag 2016;109:71–85. [124] Bizon N. Global Maximum Power Point Tracking (GMPPT) of photovoltaic array using the Extremum Seeking Control (ESC): a review and a new GMPPT ESC scheme. Renew Sust Energy Rev 2016;57:524–39. [125] Monfared M, Golestan S. Control strategies for single-phase grid integration of small-scale renewable energy sources: a review. Renew Sust Energy Rev 2012;16:4982–93. [126] Bizon N, Radut M, Oproescu M. Energy control strategies for the fuel cell hybrid power source under unknown load profile. Energy 2015;86:31–41. [127] Bizon N, Oproescu M, Raceanu M. Efficient energy control strategies for a standalone renewable/fuel cell hybrid power source. Energ Convers Manag 2015;90:93–110. [128] Bizon N. A new topology of fuel cell hybrid power source for efficient operation and high reliability. J Power Sources 2011;96(6):3260–70. [129] Athari H, Niroomand M, Ataei M. Review and classification of control systems in grid-tied inverters. Renew Sust Energ Rev 2017;72:1167–76. [130] Pavković D, Lobrović M, Hrgetić M, Komljenović A. A design of cascade control system and adaptive load compensator for battery/ultracapacitor hybrid energy storage-based direct current microgrid. Energ Convers Manag 2016;114:154–67. [131] Syed AH, Abido MA. New enhanced performance robust control design scheme for grid-connected VSI. Control Eng Pract 2016;53:92–108. [132] Bizon N, Lopez-Guede JM, Erol Kurt, Thounthong P, Mazare AG, Ionescu LM, Iana G. Hydrogen economy of the fuel cell hybrid power system optimized by air flow control to mitigate the effect of the uncertainty about available renewable power and load dynamics. Energy Convers Manag 2019;179:152–65.

[85] Dong L, Xu Q, Lu F, Nie X, He Y, Wang Y, Yan Z. Simulation and experimental investigation of a high-Temperature superconducting inductive pulsed power supply with time delay effect of the secondary side. Phys C: Supercond Appl 2017;541:16–21. [86] Li J, Yang Q, Robinson R, Liang F, Zhang M, Yuan Y. Design and test of a new droop control algorithm for a SMES/battery hybrid energy storage system. Energy 2017;118:1110–22. [87] Mcnab IR. Developments in pulsed power technology. IEEE Trans Magn 2001;37(1):375–8. [88] Akiyama H, Sakugawa T, Namihira T, Takaki K, Minamitani Y, Shimomura N. Industrial applications of pulsed power technology. IEEE Trans Fundam Mater 2007;14(5):1051–64. [89] Chaine S, Tripathy M. Design of an optimal SMES for automatic generation control of two-area thermal power system using Cuckoo search algorithm. J Electr Syst Inform Technol 2016;2(1):1–13. [90] Panda AK, Penthia T. Design and modeling of SMES based SAPF for pulsed power load demands. Int J Electr Power Energy Syst 2017;92:114–24. [91] Tabatabaei NM, Bizon N, Aghbolaghi AJ, Blaabjerg F, editors. Fundamentals and contemporary issues of reactive power control in AC power systems. Springer; 2017. [92] Girimonte D, Izzo D. Artificial intelligence for space applications. In: Schuster AJ, editor. Intelligent computing everywhere. NY, USA: Springer International Publishing; 2007. p. 235–53 [chapter12]. [93] ISO. Space systems and operations. 〈https://www.iso.org/ics/49.140/x/〉, [Accessed in January 2018]. [94] European Cooperation for Space Standardization, ISO/TC 20/SC 14, 〈https:// www.iso.org/committee/46614.html〉, [Accessed in January 2018]. [95] American Institute of Aeronautics and Astronautics. Space Systems and Vehicles, 〈https://arc.aiaa.org/page/standards〉, [Accessed in January 2018]. [96] Zhongfu T, Chen Z, Pingkuo L, Reed B, Jiayao Z. Focus on fuel cell systems in China. Renew Sust Energ Rev 2015;47:912–23. [97] Luo Y, Jiao K. Cold start of proton exchange membrane fuel cell. Prog Energy Combust Sci 2018;64:29–61. [98] Das V, Padmanaban S, Venkitusamy K, Selvamuthukumaran R, Blaabjerg F, Siano P. Recent advances and challenges of fuel cell based power system architectures and control – a review. Renew Sust Energ Rev 2017;73:10–8. [99] Han J, Yu S, Yi S. Adaptive control for robust air flow management in an automotive fuel cell system. Appl Energy 2017;190:73–83. [100] Gang BG, Kim H, Kwon S. Ground simulation of a hybrid power strategy using fuel cells and solar cells for high-endurance unmanned aerial vehicles. Energy 2017;141:1547–54. [101] Latha K, Umamaheswari B, Chaitanya K, Rajalakshmi N, Dhathathreyan KS. A novel reconfigurable hybrid system for fuel cell system. Int J Hydrog Energy 2015;40:14963–77. [102] Bizon N. Optimal operation of fuel cell / wind turbine hybrid power system under turbulent wind and variable load. Appl Energy 2018;212:196–209. [103] Vasilyev A, Andrews J, Jackson LM, Dunnett SJ, Davies B. Component-based modelling of PEM fuel cells with bond graphs. Int J Hydrog Energy 2017;42(49):29406–21. [104] Ramos-Paja CA, Spagnuolo G, Petrone G, Emilio Mamarelis M. A perturbation strategy for fuel consumption minimization in polymer electrolyte membrane fuel cells: analysis, design and FPGA implementation. Appl Energy 2014;119:21–32. [105] Bizon N. Energy optimization of fuel cell system by using global extremum Seeking algorithm. Appl Energy 2017;206:458–74. [106] Nikezhadi A, Fantova MA, Kunusch C, Martinez CO. Design and implementation of LQR/LQG strategies for oxygen stoichiometry control in PEM fuel cells based systems. J Power Sources 2011;196(9):4277–82. [107] Beirami H, Shabestari AZ, Zerafat MM. Optimal PID plus fuzzy controller design for a PEM fuel cell air feed system using the self-adaptive differential evolution algorithm. Int J Hydrog Energy 2015;40:9422–34. [108] Wang YZ, Xuan DJ, Kim YB. Design and experimental implementation of time delay control for air supply in a polymer electrolyte membrane fuel cell system. Int J Hydrog Energy 2013;38:13381–92. [109] Zhou N, Yang C, Tucker D, Pezzini P, Traverso A. Transfer function development for control of cathode airflow transients in fuel cell gas turbine hybrid systems. Int J Hydrog Energy 2015;40(4):1967–79.

37