The standalone Renewable/Fuel Cell Hybrid Power Source system 

Contexts in source publication

Context 1

... = 96485 As/mol; N C represents the number of cells in series (65);  - operating temperature (65 o Celsius) U f(H2) , U f(O2) - nominal utilization of hydrogen (99.56%) and oxygen (59.3%); P f(H2) , P f(O2) - pressure of the fuel (1.5 bar) and air (1 bar); x H2 , y O2 – composition of fuel (99.95%) and oxidant (21%); I ref(H2) , I ref(O2) - reference currents. The reference currents are generated by the SIDO ES ( I ref1 ) and LF ( I ref2 ) control schemes: I ref(H2) = I ref1 and I ref(O2) = I ref2 in Figure 2a, and vice versa in Figure 2b, where I ref(H2) = I ref2 and I ref(O2) = I ref1 . If the sFF control is applied to FuelFr or AirFr regulaters, then I ref(H2) = I FC (Figure 2a) or I ref(O2) = I ref2 (Figure 2a). If both FuelFr and AirFr are regulated through the FC current based on (6) and (7), then the classical sFF control is applied [60]. The surface of the FC net power is shown in Figure 5 through the contour lines obtained for different FC currents. The FC nominal parameters are also mentioned in Figure 5. 4.2. Renewable energy sources RESs have a lot of advantages including sustainability, reduction of carbon dioxide, and economic benefits, but the intermittent nature and low efficiency of many RESs request to be integrated into a RES HPS [10,19,24]. For example, the WT and PV energy resources in a given area can be complementary in a certain period of time [26]. A generic model of any combination of RESs is used in this paper in order to test the control topologies proposed under a RES power profile close to real one (Figure 6). Thus, two power sequences define the PV and WT power profiles without random fluctuations. The variability of the RES power profile is added randomly based a random sequence. A RES power profile is shown in Figure 6 for a sunny day with some clods at noon and afternoon, and moderate wind in the evening. The peak of the RES power profile is of 5 kW, close to the peak of the load demand for a smart home. Nevertheless, because these peaks of the PV and WT power profile are not synchronized, a backup energy source is usually used [30,31]. 4.3. Energy Storage System The hybrid batteries/ultracapacitors ESS is necessary to ensure the power flow balance (1) under unknown RES power profile and random load demand. The hybrid semi-active topology for the ESS is used here to evaluate the impact of adding ESS on the performance of a RES HPS [47]. The ESS is charged during an energy surplus on the DC bus (RES power higher than load demand and FC system operates on standby mode), when the electrolyzer can be started depending of the battery SOC. The ESS is discharged a bit during step-up load until the FC system starts to operate in the LF control loop. Thus, the deep-discharge cycles are avoided if the LF control is implemented to fuel the FC stack. This number of deep-discharge cycles is limited for all types of batteries, but it is practically unlimited for the ultracapacitors stack. Consequently, the voltage regulation on the DC bus is ensured by appropriate control of the bidirectional DC-DC converter that interface the ultracapacitors stack to the DC bus. The batteries stack is connected directly to the DC bus in the hybrid semi-active ESS topology (see Figure 7) used here to operate the batteries stack in CS mode through the LF control proposed. The ultracapacitors stack is connected to the DC bus via a bidirectional DC-DC converter in order to decrease the number of ultracapacitor cells, assures a large range for the ultracapacitor SOC, and achieve an active control of DC bus voltage [23]. Thus, the FC system and ultracapacitor stack operate as active controlled power sources. The batteries stack operates as passive controlled energy sources assuring the power flow balance (1). In this paper, the generic model of the lithium-ion battery included in the SimPowerSystems library of the Matlab - Simulink® is used [57]. The nominal voltage, rated capacity and initial SOC were set at 250 V, 100 Ah and 80%. The preset battery parameters have been computed depending on battery type and rating values. Different battery sizes are also used in simulation, according to the EV33 rule [61]. The power flow balance on the DC bus is dynamically compensated by the ultracapacitor stack. Different ultracapacitor models have been proposed based on electric [62] and non-electric models [63]. The last class of models uses a controlled current source to be connected to the electrical circuit and a computational algorithm to evaluate that current. The electrical models have the disadvantage of direct connection to the bidirectional buck-boost converter (see Figure 7). The classical electric circuit model, which includes the equivalent series and parallel resistors (ESR and EPR), and a capacitor (C), will be used in this paper. The initial voltage, C, ESR and EPR have been set at 100 V, 100 F, 0.1  and 10 k  . 4.4. Power interfaces Power interfaces are needed to effectively adapt the RES voltages to DC voltage based on MPPT controller that uses a current or voltage reference. The reference is used in the PWM control of the respective power interface. For example, an AC-DC and DC-DC power converter is necessary to interface the WT and PV systems. The FC system is interfaced through a DC-DC power converter that is MEPT controlled. If multi-source HPS architecture is need to be designed, then multiport power convertor topology is recommended, instead of implementing several power converters. Furthermore, efficiency and cost are greatly enhanced because the number of passive and switching components is heavily reduced [40]. The DC power bus is the output of the multiport power convertor (see Figure 1). The AC loads are supplied from the DC bus via an inverter system. Multiport power convertor is directly fed by all energy sources available based on the optimized EMU strategy [65]. As it was mentioned in previous section, three DC-DC power converters are used to interface the FC, PV and ultracapacitors stack. The first two DC-DC power converters are of unidirectional type, adapting the FC and PV voltage to the voltage level set on the DC bus (250 V). For example, the unidirectional DC-DC power converter that interfaces the FC system is of boost type (see Figure 2) because the FC voltage is lower than 250 V, being in range of 40-65 V for a FC power variation from rated power to standby power level. The AV value of the FC power flow that supply the DC bus is dependent to energy efficiency of the boost ...

Context 2

... profiles, with direct implications in efficiency of the whole RES/FC HPS. This paper is organized as follows. Section 2 presents briefly the issue related to unit sizing and optimization of the FC/RES HPS. The structure and control loops of the proposed RES/FC HPS unit are also explained here. The four EMU strategies to control the fueling rates are explained in Section 3 based on the AV power flow balance. The models of all units used in simulation of the RES/FC HPS are briefly shown in Section 4. The implementation of the LF and MEPT control loops are detailed in Section 5. The results obtained by comparing the proposed EMU strategies under constant load, random pulsed load, and variable load under different RES power profiles are shown and discussed in Section 6. Section 7 concludes the paper. 2. Standalone Renewable/Fuel Cell Hybrid Power Source As it is known, the RESs have different operating characteristics, but in general these have a global MPP that must be tracked under different environment conditions. Also, the operating characteristic of the PEMFC system has a maximum (for example, the FC net power related to FC current). Thus, the usual method to integrate RES and FC into a RES/FC HPS is based on power interfaces controlled to track the MPP and MEP of the input energy source. Besides this general used connection of the energy sources to the DC bus [40], which define the DC bus configuration, there are other two possible configurations to integrate different RESs, FC and ESS into a RES/FC HPS: AC coupled configuration and Hybrid coupled configuration [27,41, 42]. In this paper, the DC coupled configuration and equivalent DC load concept is used to test the control performances under different RES and load power profiles. 2.1. Unit sizing and optimization As it was mentioned on Introduction, the RESs have a random behavior. Thus, it is difficult to be predicted the available RES power flow in order to size the FC stack and ESS to make face to any load demand. In sunny and windy days the ESS is charged up to higher charge limit (set for the SOC) and electrolyzer refills the hydrogen tanks that are underutilized. Instead of this favorable case, the RES power flow is low during the cloudy days with moderate wind. Consequently, the ESS is discharged under the lower discharge limit, the electrolyzer is stopped, and the hydrogen tank may be empty, if this situation continues on a long period of time, which is difficult to be predicted under the current climate change. So, the size of the FC and batteries stack, the electrolyzer and hydrogen storage tanks must to be designed carefully considering the load demand and RES power flow available in all extreme weather conditions. Furthermore, the load demand of a smart home has also a random power profile. Thus, the unit sizing and optimization of standalone RES/FC HPS is not a trivial design problem, requiring an algorithm to size the HPS components by minimizing the system cost while maintaining system reliability [9-10]. It is obvious that the design objectives such as acceptable cost and reliability level are conflicting with one another [12,31]. For example, over sizing the HPS components will increase the HPS cost while under sizing can lead in lack of power to continuously sustain the load demand. So, the algorithms to size the HPS components must ensure a reasonable tradeoff between the design objectives, optimizing the HPS energy sources to achieve the target levels of cost and reliability. The HPS reliability is determined by estimating the loss of load probability, which is the ratio between estimated power deficit and the load demand during a load cycle, or appropriate other reliability indicators such loss of power probability, loss of power supply probability, and load coverage rate [42]. Besides the HPS cost and reliability, the sizing algorithms optimize the HPS components or other HPS parameters, such as investment cost, output energy cost, fuel consumption or appropriate other parameters such as net present cost, levelized cost of energy and fuel efficiency [43]. The algorithms to size the HPS components obviously depend on the availability of the weather data on short or long term, so two class are defined in the literature: conventional sizing algorithms based on weather data and expert algorithms if the weather data is not available or available only on short term, such is the case of remote isolated sites [19,41,42]. In first case, the concept of energy balance is a simple algorithm of sizing the HPS components [43]. In last case, the algorithms based on artificial intelligence are usually used [44, 45]. The rated power of the FC stack is set here to be higher than the maximum load demand because the sizing objective is outside the scope of this paper. 2.2. The structure of the proposed unit The HPS unit structure based on RES and FC system that has a new fueling control strategy based on LF and ES control loops is proposed here (Figure 1). It is known that the FC stack needs an ESS under dynamic loads to improve the HPSs system performance under random load demand and random RES power profile [46]. The ESS topology used here is of semi-active type: batteries stack direct connected to the DC bus and ultracapacitors stack via a bidirectional power converter [47]. This hybrid ESS topology is usually used due their advantages: the SOC variation of the ultracapacitors stack in large range ensures the dynamically compensation of the power flow balance at reduced cost. Thus, the proposed HPS that is shown in Figure 1 is composed of energy sources (FC, WT, and PV systems), ESS (batteries and ultracapacitors stacks), equivalent load (AC and DC loads), multiport power converter, and EMU (containing the ES and LF controllers). The fuel is classically or efficiently controlled based on static feed-forward (sFF) control or ES control scheme, which means the use of the FC current (I FC ) or the reference current 1 (I ref1 ) for the fuel regulator to control the fuel flow rate (FuelFr). The second ES controller generates the reference current (I ref ) to control the FC power in the LF control loop (see Figure 2). The LF controller generates the reference current 2 (I ref2 ) to control the air flow rate (AirFr) (Figure 1). Note that both ES controllers have as input the same signal, the FC net power (P FCnet ), as it shown in Figure 2. In fact, the unit structure shown in Figure 1 is detailed in Figure 2a, where is given the two simulation diagram implemented in MATLAB – Simulink ® simulation environment to control the AirFr based LF loop and FuelFr based sFF or ESC loop. Other two control configurations to LF control the FuelFr are shown in Figure 2b, where the AirFr is sFF or ESC controlled. All four control configurations will be analyzed here. The FC system must to operate efficiently, close to the maximum of the FC net power (named MEP). Different MEPT algorithm are proposed in the literature such as the ES control scheme [37], Perturb and Observe (P&O) method [48] or other type of searching algorithm [20, 49]. One MEPT algorithm will be implemented in the SIDO block to generate the both current references I ref and I ref1 . The voltage on DC bus could be chosen lower than 350-400 V if a bidirectional Z-Source Inverter (ZSI) is used to connect the AC loads [50]. Furthermore, the boost converter that interface the FC stack to DC bus will have a lower voltage ratio, being here implemented through a basic unidirectional boost topology. Thus, the DC voltage, u dc , was set to 250 V on the DC bus, which is modeled as a capacitor, C DC [51]. Consequently, the rated voltage of the rechargeable Li-ion batteries stack will be 250 V. The rated voltage of the ultracapacitors stack was chosen of 100 V. The bidirectional buck-boost converter interface the ultracapacitors stack to the DC bus, ensuring the dynamic power compensation on the DC bus. Linear and nonlinear control techniques can be used to regulate the voltage on the DC bus [46]. The reference current 2, I ref2 , is computed on LF control block based on the AV value of requested FC power on DC bus, P Load – P RES , considering that the AV value of the ESS power is zero (because the batteries stack operates in CS mode based on the LF control proposed here). It is known that the dynamic FC response is dependent to the fueling rate, stoichiometric ratio, temperature, humidity and pressure [52]. Consequently, the fuel regulators must include a rate limiter to avoid the gas starvation [53] and a saturation block to operate the FC stack in available range of FC power. The minimum level of the FC power is set different to zero, avoiding the gas starvation that may appear during a repetitive start-up. So, the FC stack will operate in standby-mode at low fueling rates during the phases of low power demand, when RES power is higher that load demand. Consequently, the power flow balance on DC bus is a key relationship to design the EMU strategies [15,21,24,54]. 3. The EMU strategies to control the fueling rates Besides the power flow balance, which means acquisition of the all power flows (Figure 1), the other key decision parameters for the EMU strategies are the level of the FC and WT power, the state signals (such as SOC of the batteries and ultracapacitors stacks), the voltage on the FC and DC bus, the user signals (such as personalized prioritization of the loads into a smart home or environmental and internal ambient rules for the thermal comfort, rules to start the electrolyzer or charge the electrical vehicle, and so on), and the protection signals from and outside of the HPS. Note that the main objective for all applied EMU strategies in the integrated HPS is to ensure the load demand. Consequently, the analysis performed in this paper is focused to show this based on the results obtained with the proposed LF and MEPT control loops, without considering the EMU meta-rules based on other key ...

Context 3

... on the availability of the weather data on short or long term, so two class are defined in the literature: conventional sizing algorithms based on weather data and expert algorithms if the weather data is not available or available only on short term, such is the case of remote isolated sites [19,41,42]. In first case, the concept of energy balance is a simple algorithm of sizing the HPS components [43]. In last case, the algorithms based on artificial intelligence are usually used [44, 45]. The rated power of the FC stack is set here to be higher than the maximum load demand because the sizing objective is outside the scope of this paper. 2.2. The structure of the proposed unit The HPS unit structure based on RES and FC system that has a new fueling control strategy based on LF and ES control loops is proposed here (Figure 1). It is known that the FC stack needs an ESS under dynamic loads to improve the HPSs system performance under random load demand and random RES power profile [46]. The ESS topology used here is of semi-active type: batteries stack direct connected to the DC bus and ultracapacitors stack via a bidirectional power converter [47]. This hybrid ESS topology is usually used due their advantages: the SOC variation of the ultracapacitors stack in large range ensures the dynamically compensation of the power flow balance at reduced cost. Thus, the proposed HPS that is shown in Figure 1 is composed of energy sources (FC, WT, and PV systems), ESS (batteries and ultracapacitors stacks), equivalent load (AC and DC loads), multiport power converter, and EMU (containing the ES and LF controllers). The fuel is classically or efficiently controlled based on static feed-forward (sFF) control or ES control scheme, which means the use of the FC current (I FC ) or the reference current 1 (I ref1 ) for the fuel regulator to control the fuel flow rate (FuelFr). The second ES controller generates the reference current (I ref ) to control the FC power in the LF control loop (see Figure 2). The LF controller generates the reference current 2 (I ref2 ) to control the air flow rate (AirFr) (Figure 1). Note that both ES controllers have as input the same signal, the FC net power (P FCnet ), as it shown in Figure 2. In fact, the unit structure shown in Figure 1 is detailed in Figure 2a, where is given the two simulation diagram implemented in MATLAB – Simulink ® simulation environment to control the AirFr based LF loop and FuelFr based sFF or ESC loop. Other two control configurations to LF control the FuelFr are shown in Figure 2b, where the AirFr is sFF or ESC controlled. All four control configurations will be analyzed here. The FC system must to operate efficiently, close to the maximum of the FC net power (named MEP). Different MEPT algorithm are proposed in the literature such as the ES control scheme [37], Perturb and Observe (P&O) method [48] or other type of searching algorithm [20, 49]. One MEPT algorithm will be implemented in the SIDO block to generate the both current references I ref and I ref1 . The voltage on DC bus could be chosen lower than 350-400 V if a bidirectional Z-Source Inverter (ZSI) is used to connect the AC loads [50]. Furthermore, the boost converter that interface the FC stack to DC bus will have a lower voltage ratio, being here implemented through a basic unidirectional boost topology. Thus, the DC voltage, u dc , was set to 250 V on the DC bus, which is modeled as a capacitor, C DC [51]. Consequently, the rated voltage of the rechargeable Li-ion batteries stack will be 250 V. The rated voltage of the ultracapacitors stack was chosen of 100 V. The bidirectional buck-boost converter interface the ultracapacitors stack to the DC bus, ensuring the dynamic power compensation on the DC bus. Linear and nonlinear control techniques can be used to regulate the voltage on the DC bus [46]. The reference current 2, I ref2 , is computed on LF control block based on the AV value of requested FC power on DC bus, P Load – P RES , considering that the AV value of the ESS power is zero (because the batteries stack operates in CS mode based on the LF control proposed here). It is known that the dynamic FC response is dependent to the fueling rate, stoichiometric ratio, temperature, humidity and pressure [52]. Consequently, the fuel regulators must include a rate limiter to avoid the gas starvation [53] and a saturation block to operate the FC stack in available range of FC power. The minimum level of the FC power is set different to zero, avoiding the gas starvation that may appear during a repetitive start-up. So, the FC stack will operate in standby-mode at low fueling rates during the phases of low power demand, when RES power is higher that load demand. Consequently, the power flow balance on DC bus is a key relationship to design the EMU strategies [15,21,24,54]. 3. The EMU strategies to control the fueling rates Besides the power flow balance, which means acquisition of the all power flows (Figure 1), the other key decision parameters for the EMU strategies are the level of the FC and WT power, the state signals (such as SOC of the batteries and ultracapacitors stacks), the voltage on the FC and DC bus, the user signals (such as personalized prioritization of the loads into a smart home or environmental and internal ambient rules for the thermal comfort, rules to start the electrolyzer or charge the electrical vehicle, and so on), and the protection signals from and outside of the HPS. Note that the main objective for all applied EMU strategies in the integrated HPS is to ensure the load demand. Consequently, the analysis performed in this paper is focused to show this based on the results obtained with the proposed LF and MEPT control loops, without considering the EMU meta-rules based on other key decision parameters mentioned above. For example, the operating meta-rule for the hydrogen production via water electrolysis depends on the excess or shortage of power from the RES in comparison with the load demand, the SOC level of the batteries stack, the level of hydrogen in the storage tank of SOC. Implementation of such meta-rule based on the concept of Hierarchical Control Theory [31- 32], which is mainly applied in integrated hybrid systems to schedule the operation of the involved subsystems based on a predefined hierarchy made on the decision flow [27], could be simple if the RES power flow is constant or varying slowly during a load cycle. The huge variability of RES power on the DC bus and random behavior of the load demand in a smart home increase the complexity of the EMU [25-26]. Furthermore, frequent start-up and shut-down actions for the FC stack and electrolyzer may degrade their life cycle [32,55]. Therefore, the EMU strategy based on LF control that continuous operates the FC is proposed here. The capacity of the batteries stack that operates in CS mode will be designed at minimum value needed to compensate the short-term variability of the RES power and sharp peaks of the random load demand. Thus, the batteries’ lifespan is improved and the cost of the batteries stack is substantially reduced, and subsequently these influence the operating and maintenance costs of the entire HPS. The MEPT control is proposed here to improve the energy efficiency of the entire HPS based on control of the fueling rates for the FC stack. Thus, the efficiency of the fuel consumption (or shortly the fuel efficiency) is improved. In brief, the EMU strategy proposed here based on efficient control loops aims to ensure the HPS operation under variable weather conditions and random load demand, maintaining the operating costs at a reasonable level. The power flow balance on the DC bus is given ...

Context 4

... used as backup source because of the advantages of this technology related to other FC technologies [13-15,20]. In addition, the hydrogen is abundant in the nature and can be obtained by reforming the natural gas, ethanol, methanol, biogas, and so on, or by water electrolyzing. Besides the eco-friendly operation of the RES/FC HPS, the economic aspect makes the PEMFC a competitive technology in comparison to diesel generator, offering a reduced maintenance effort and cost [5-6,21]. So, integrating RES, FC and ESS into a RES/FC HPS via a multiport power converter (see Figure 1), the hybrid source will have the energy generation/storage support to make face to energy consumption and variability of the RES power that are not always synchronized with the load demand. Thus, EMU strategies are important in optimizing the energy management of the energy sources [1-4, 9, 11-15, 21]. The main goal of the EMU strategies must to ensure the load demand. Besides this, the other specific goals of the EMU strategies must be related to fuel consumption, energy efficiency of the energy sources that must to be also safe operated, life cycle of the HPS, and cost [11-15, 21-22]. Consequently, some variables of the HPS (named in Figure 1 as state signals; for example the state-of-charge (SOC) level of the ESS devices) and control variables (for example the DC bus voltage, FC current, load demand, and RES power level) must to be inputs of the EMU strategy. Besides these signals, the protection and user signals are used to safe operation of the HPS. The studies above mentioned analyze the energy efficiency of the EMU strategies using dynamic models of the HPS based on short-term simulations with time scale of seconds [23] or minutes [24]. The voltage regulation on the DC bus of the PV/FC HPS is approached by classical linear or non-linear control algorithm (for example, based on the differential flatness principle [25]). Different energy management strategies are proposed in the literature. For example, the fuzzy logic control was used in [26] to monitor the SOC of the ESS devices, improving the utilization costs and lifetime of the battery and hydrogen system as well. Also, the management of power flows between the FC and ESS in HPS grid connected is analyzed based on a fuzzy logic control [22, 27] or ANFIS control [28]. The energy efficiency of a standalone RES/FC HPS based on EMU strategy to make face to RES power variations was investigated in [24,29]. In long-term analysis, the main goal of the EMU strategies is focused on meeting the load demand considering in the energy dispatch other specific goals such as the cost and the energy efficiency, and state signals such as the SOC level of the ESS devices, etc. An adaptive predictive strategy for a RES/FC HPS to meet the above goals is analyzed with a time scale of hours in [30]. Three EMU strategies to meet the load demand and increase the energy efficiency based on long-term simulations throughout one year are proposed in [12,31]. Also, the expected lifetime of a standalone HPS was evaluated in [32]. The energy strategies mentioned above are mainly based on a monitoring of the power flows and states of the ESS devices to decide the energy dispatch between the FC and battery to the load [21,33]. In this paper, the battery operates in CS mode to minimize its size, the load demand being sustained based on the LF control proposed by the RESs and FC system. Also, it can be noted that the dynamics of the energy sources and especially of the power interfaces are frequently neglected [34]. The power converters are modeled in this paper using components from the SimPowerSystems library of the Matlab - Simulink® and not using an AV model that is recommended for a long-term analysis of the HPS behavior and performance. Thus, a short-term analysis will be performed here based on the sample frequency of the 10 kHz, which is more than enough to efficiently use the real profile of the RES power based on advanced Maximum Power Point Tracking (MPPT) control with high search speed and good tracking accuracy [23,35]. The MEPT technique based on the Extremum Seeking (ES) control scheme is used here to control one of the fueling rates [36], but note that any MEPT advanced technique could be used as well. The ES control is perturbed – based scheme, using dithers with different frequencies or orthogonal dithers to implement Single- Input Double-Outputs (SIDO) ES control scheme [37]. Note that only the performances of a PEMFC system are tested in [35-37], while here the performances of a whole RES/FC HPS are shown. Thus, besides the EMU that includes the LF and MEPT control schemes, the main components of the standalone RES/FC HPS includes a PV array, a set of wind generators, a FC system as backup source, a short – term ESS based on li-ion batteries and ultracapacitors, and a long-term ESS that consists of an electrolyzer and a hydrogen storage system with pressurized tanks (only this is shown in Figure 1), DC loads and AC load interfaced with inverters (named the equivalent load), and a multiport power converter or independent power interfaces for each energy sources and load. Summarizing, the research directions in field of the RES/FC HPS that are approached in the studies mentioned above are the following [38]: • The RES need advanced control techniques to harvest all power available, operating the RES close to MPP. This control is mandatory due to poor efficiency of solar PV, which is the main impediment in encouraging its use until the PV technology will be improved. • The power losses in power interfaces have been substantially reduced using advanced topologies, switches and appropriate control scheme, eventually integrated in multiport power converter structure. In general, the energy efficiency of power converters used in HPS is higher than 95%. • The hybrid batteries/ultracapacitors ESS is used in HPS to ensure the power flow balance on the DC bus, but their life-cycle need to be improved through innovative technologies, too. The CS mode proposed here for the ESS can improve the life-cycle of the batteries stack, besides other advantages such as reduced size and low costs of the ESS. • The cost reduction could be an incentive for the producers of HPS to implement such systems that will ensure decreasing of the payback time for the capital invested. • The hydrogen technology (to generate and store hydrogen) is still a costly and unsafe technology. So, alternative technologies based on fuel reformers are developed at low cost. Furthermore, the PEMFC system must be operated at the MEP to increase the energy efficiency of the entire RES/FC HPS, as it is proposed here. • These standalone RES/FC HPSs based on EMU strategies must to predict the RES power available or be adaptable to RES power fluctuations to sustain any unpredictable load demand. The LF control proposed here could be o solution, because it is very simple to be implemented in the commercial RES/FC HPS, requesting only software upgrade and few circuits reconfiguration. • The protection issue, especially for use of PEMFC system and batteries stack from the ESS, are identified and implemented in EMU of the commercial RES/FC HPS. The proposed protection measures are implemented in simulation diagram used here, but this issue was not extensively studied, being outside of the scope of this paper. For example, the recommended rate [39] to limit the FC current is used in all simulation shown in this paper. In this paper, four new control topologies of the RES/FC HPS are introduced. The innovative idea is to use the LF control based on the MEPT control scheme to control efficiently both fueling rates in order to ensure the power flow balance on the DC bus. The main contributions of this paper are the following: (1) the LF control based on the MEPT control scheme is proposed to ensure the load demand and improve the energy efficiency of whole RES/FC HPS (2) the both fuel consumption and fuel efficiency are used as performances indicators; (3) a methodology to compare the four new control topologies of the RES/FC HPS is performed based on the performances indicators proposed; (4) the advantage of the LF control that operates the ESS in CS mode is shown under any load demand, with direct implications in the size and cost of the RES/FC HPS; (5) the advantage of the MEPT control that operates efficiently the FC stack is also shown considering the variability of RES and load power profiles, with direct implications in efficiency of the whole RES/FC HPS. This paper is organized as follows. Section 2 presents briefly the issue related to unit sizing and optimization of the FC/RES HPS. The structure and control loops of the proposed RES/FC HPS unit are also explained here. The four EMU strategies to control the fueling rates are explained in Section 3 based on the AV power flow balance. The models of all units used in simulation of the RES/FC HPS are briefly shown in Section 4. The implementation of the LF and MEPT control loops are detailed in Section 5. The results obtained by comparing the proposed EMU strategies under constant load, random pulsed load, and variable load under different RES power profiles are shown and discussed in Section 6. Section 7 concludes the paper. 2. Standalone Renewable/Fuel Cell Hybrid Power Source As it is known, the RESs have different operating characteristics, but in general these have a global MPP that must be tracked under different environment conditions. Also, the operating characteristic of the PEMFC system has a maximum (for example, the FC net power related to FC current). Thus, the usual method to integrate RES and FC into a RES/FC HPS is based on power interfaces controlled to track the MPP and MEP of the input energy source. Besides this general used connection of the energy sources to the DC bus [40], which define the DC bus configuration, there are other two possible configurations to integrate ...

Context 5

... 5. The results obtained by comparing the proposed EMU strategies under constant load, random pulsed load, and variable load under different RES power profiles are shown and discussed in Section 6. Section 7 concludes the paper. 2. Standalone Renewable/Fuel Cell Hybrid Power Source As it is known, the RESs have different operating characteristics, but in general these have a global MPP that must be tracked under different environment conditions. Also, the operating characteristic of the PEMFC system has a maximum (for example, the FC net power related to FC current). Thus, the usual method to integrate RES and FC into a RES/FC HPS is based on power interfaces controlled to track the MPP and MEP of the input energy source. Besides this general used connection of the energy sources to the DC bus [40], which define the DC bus configuration, there are other two possible configurations to integrate different RESs, FC and ESS into a RES/FC HPS: AC coupled configuration and Hybrid coupled configuration [27,41, 42]. In this paper, the DC coupled configuration and equivalent DC load concept is used to test the control performances under different RES and load power profiles. 2.1. Unit sizing and optimization As it was mentioned on Introduction, the RESs have a random behavior. Thus, it is difficult to be predicted the available RES power flow in order to size the FC stack and ESS to make face to any load demand. In sunny and windy days the ESS is charged up to higher charge limit (set for the SOC) and electrolyzer refills the hydrogen tanks that are underutilized. Instead of this favorable case, the RES power flow is low during the cloudy days with moderate wind. Consequently, the ESS is discharged under the lower discharge limit, the electrolyzer is stopped, and the hydrogen tank may be empty, if this situation continues on a long period of time, which is difficult to be predicted under the current climate change. So, the size of the FC and batteries stack, the electrolyzer and hydrogen storage tanks must to be designed carefully considering the load demand and RES power flow available in all extreme weather conditions. Furthermore, the load demand of a smart home has also a random power profile. Thus, the unit sizing and optimization of standalone RES/FC HPS is not a trivial design problem, requiring an algorithm to size the HPS components by minimizing the system cost while maintaining system reliability [9-10]. It is obvious that the design objectives such as acceptable cost and reliability level are conflicting with one another [12,31]. For example, over sizing the HPS components will increase the HPS cost while under sizing can lead in lack of power to continuously sustain the load demand. So, the algorithms to size the HPS components must ensure a reasonable tradeoff between the design objectives, optimizing the HPS energy sources to achieve the target levels of cost and reliability. The HPS reliability is determined by estimating the loss of load probability, which is the ratio between estimated power deficit and the load demand during a load cycle, or appropriate other reliability indicators such loss of power probability, loss of power supply probability, and load coverage rate [42]. Besides the HPS cost and reliability, the sizing algorithms optimize the HPS components or other HPS parameters, such as investment cost, output energy cost, fuel consumption or appropriate other parameters such as net present cost, levelized cost of energy and fuel efficiency [43]. The algorithms to size the HPS components obviously depend on the availability of the weather data on short or long term, so two class are defined in the literature: conventional sizing algorithms based on weather data and expert algorithms if the weather data is not available or available only on short term, such is the case of remote isolated sites [19,41,42]. In first case, the concept of energy balance is a simple algorithm of sizing the HPS components [43]. In last case, the algorithms based on artificial intelligence are usually used [44, 45]. The rated power of the FC stack is set here to be higher than the maximum load demand because the sizing objective is outside the scope of this paper. 2.2. The structure of the proposed unit The HPS unit structure based on RES and FC system that has a new fueling control strategy based on LF and ES control loops is proposed here (Figure 1). It is known that the FC stack needs an ESS under dynamic loads to improve the HPSs system performance under random load demand and random RES power profile [46]. The ESS topology used here is of semi-active type: batteries stack direct connected to the DC bus and ultracapacitors stack via a bidirectional power converter [47]. This hybrid ESS topology is usually used due their advantages: the SOC variation of the ultracapacitors stack in large range ensures the dynamically compensation of the power flow balance at reduced cost. Thus, the proposed HPS that is shown in Figure 1 is composed of energy sources (FC, WT, and PV systems), ESS (batteries and ultracapacitors stacks), equivalent load (AC and DC loads), multiport power converter, and EMU (containing the ES and LF controllers). The fuel is classically or efficiently controlled based on static feed-forward (sFF) control or ES control scheme, which means the use of the FC current (I FC ) or the reference current 1 (I ref1 ) for the fuel regulator to control the fuel flow rate (FuelFr). The second ES controller generates the reference current (I ref ) to control the FC power in the LF control loop (see Figure 2). The LF controller generates the reference current 2 (I ref2 ) to control the air flow rate (AirFr) (Figure 1). Note that both ES controllers have as input the same signal, the FC net power (P FCnet ), as it shown in Figure 2. In fact, the unit structure shown in Figure 1 is detailed in Figure 2a, where is given the two simulation diagram implemented in MATLAB – Simulink ® simulation environment to control the AirFr based LF loop and FuelFr based sFF or ESC loop. Other two control configurations to LF control the FuelFr are shown in Figure 2b, where the AirFr is sFF or ESC controlled. All four control configurations will be analyzed here. The FC system must to operate efficiently, close to the maximum of the FC net power (named MEP). Different MEPT algorithm are proposed in the literature such as the ES control scheme [37], Perturb and Observe (P&O) method [48] or other type of searching algorithm [20, 49]. One MEPT algorithm will be implemented in the SIDO block to generate the both current references I ref and I ref1 . The voltage on DC bus could be chosen lower than 350-400 V if a bidirectional Z-Source Inverter (ZSI) is used to connect the AC loads [50]. Furthermore, the boost converter that interface the FC stack to DC bus will have a lower voltage ratio, being here implemented through a basic unidirectional boost topology. Thus, the DC voltage, u dc , was set to 250 V on the DC bus, which is modeled as a capacitor, C DC [51]. Consequently, the rated voltage of the rechargeable Li-ion batteries stack will be 250 V. The rated voltage of the ultracapacitors stack was chosen of 100 V. The bidirectional buck-boost converter interface the ultracapacitors stack to the DC bus, ensuring the dynamic power compensation on the DC bus. Linear and nonlinear control techniques can be used to regulate the voltage on the DC bus [46]. The reference current 2, I ref2 , is computed on LF control block based on the AV value of requested FC power on DC bus, P Load – P RES , considering that the AV value of the ESS power is zero (because the batteries stack operates in CS mode based on the LF control proposed here). It is known that the dynamic FC response is dependent to the fueling rate, stoichiometric ratio, temperature, humidity and pressure [52]. Consequently, the fuel regulators must include a rate limiter to avoid the gas starvation [53] and a saturation block to operate the FC stack in available range of FC power. The minimum level of the FC power is set different to zero, avoiding the gas starvation that may appear during a repetitive start-up. So, the FC stack will operate in standby-mode at low fueling rates during the phases of low power demand, when RES power is higher that load demand. Consequently, the power flow balance on DC bus is a key relationship to design the EMU strategies [15,21,24,54]. 3. The EMU strategies to control the fueling rates Besides the power flow balance, which means acquisition of the all power flows (Figure 1), the other key decision parameters for the EMU strategies are the level of the FC and WT power, the state signals (such as SOC of the batteries and ultracapacitors stacks), the voltage on the FC and DC bus, the user signals (such as personalized prioritization of the loads into a smart home or environmental and internal ambient rules for the thermal comfort, rules to start the electrolyzer or charge the electrical vehicle, and so on), and the protection signals from and outside of the HPS. Note that the main objective for all applied EMU strategies in the integrated HPS is to ensure the load demand. Consequently, the analysis performed in this paper is focused to show this based on the results obtained with the proposed LF and MEPT control loops, without considering the EMU meta-rules based on other key decision parameters mentioned above. For example, the operating meta-rule for the hydrogen production via water electrolysis depends on the excess or shortage of power from the RES in comparison with the load demand, the SOC level of the batteries stack, the level of hydrogen in the storage tank of SOC. Implementation of such meta-rule based on the concept of Hierarchical Control Theory [31- 32], which is mainly applied in integrated hybrid systems to schedule the operation of the involved subsystems based on a predefined hierarchy made on the decision flow [27], could be simple ...

Context 6

... configurations to integrate different RESs, FC and ESS into a RES/FC HPS: AC coupled configuration and Hybrid coupled configuration [27,41, 42]. In this paper, the DC coupled configuration and equivalent DC load concept is used to test the control performances under different RES and load power profiles. 2.1. Unit sizing and optimization As it was mentioned on Introduction, the RESs have a random behavior. Thus, it is difficult to be predicted the available RES power flow in order to size the FC stack and ESS to make face to any load demand. In sunny and windy days the ESS is charged up to higher charge limit (set for the SOC) and electrolyzer refills the hydrogen tanks that are underutilized. Instead of this favorable case, the RES power flow is low during the cloudy days with moderate wind. Consequently, the ESS is discharged under the lower discharge limit, the electrolyzer is stopped, and the hydrogen tank may be empty, if this situation continues on a long period of time, which is difficult to be predicted under the current climate change. So, the size of the FC and batteries stack, the electrolyzer and hydrogen storage tanks must to be designed carefully considering the load demand and RES power flow available in all extreme weather conditions. Furthermore, the load demand of a smart home has also a random power profile. Thus, the unit sizing and optimization of standalone RES/FC HPS is not a trivial design problem, requiring an algorithm to size the HPS components by minimizing the system cost while maintaining system reliability [9-10]. It is obvious that the design objectives such as acceptable cost and reliability level are conflicting with one another [12,31]. For example, over sizing the HPS components will increase the HPS cost while under sizing can lead in lack of power to continuously sustain the load demand. So, the algorithms to size the HPS components must ensure a reasonable tradeoff between the design objectives, optimizing the HPS energy sources to achieve the target levels of cost and reliability. The HPS reliability is determined by estimating the loss of load probability, which is the ratio between estimated power deficit and the load demand during a load cycle, or appropriate other reliability indicators such loss of power probability, loss of power supply probability, and load coverage rate [42]. Besides the HPS cost and reliability, the sizing algorithms optimize the HPS components or other HPS parameters, such as investment cost, output energy cost, fuel consumption or appropriate other parameters such as net present cost, levelized cost of energy and fuel efficiency [43]. The algorithms to size the HPS components obviously depend on the availability of the weather data on short or long term, so two class are defined in the literature: conventional sizing algorithms based on weather data and expert algorithms if the weather data is not available or available only on short term, such is the case of remote isolated sites [19,41,42]. In first case, the concept of energy balance is a simple algorithm of sizing the HPS components [43]. In last case, the algorithms based on artificial intelligence are usually used [44, 45]. The rated power of the FC stack is set here to be higher than the maximum load demand because the sizing objective is outside the scope of this paper. 2.2. The structure of the proposed unit The HPS unit structure based on RES and FC system that has a new fueling control strategy based on LF and ES control loops is proposed here (Figure 1). It is known that the FC stack needs an ESS under dynamic loads to improve the HPSs system performance under random load demand and random RES power profile [46]. The ESS topology used here is of semi-active type: batteries stack direct connected to the DC bus and ultracapacitors stack via a bidirectional power converter [47]. This hybrid ESS topology is usually used due their advantages: the SOC variation of the ultracapacitors stack in large range ensures the dynamically compensation of the power flow balance at reduced cost. Thus, the proposed HPS that is shown in Figure 1 is composed of energy sources (FC, WT, and PV systems), ESS (batteries and ultracapacitors stacks), equivalent load (AC and DC loads), multiport power converter, and EMU (containing the ES and LF controllers). The fuel is classically or efficiently controlled based on static feed-forward (sFF) control or ES control scheme, which means the use of the FC current (I FC ) or the reference current 1 (I ref1 ) for the fuel regulator to control the fuel flow rate (FuelFr). The second ES controller generates the reference current (I ref ) to control the FC power in the LF control loop (see Figure 2). The LF controller generates the reference current 2 (I ref2 ) to control the air flow rate (AirFr) (Figure 1). Note that both ES controllers have as input the same signal, the FC net power (P FCnet ), as it shown in Figure 2. In fact, the unit structure shown in Figure 1 is detailed in Figure 2a, where is given the two simulation diagram implemented in MATLAB – Simulink ® simulation environment to control the AirFr based LF loop and FuelFr based sFF or ESC loop. Other two control configurations to LF control the FuelFr are shown in Figure 2b, where the AirFr is sFF or ESC controlled. All four control configurations will be analyzed here. The FC system must to operate efficiently, close to the maximum of the FC net power (named MEP). Different MEPT algorithm are proposed in the literature such as the ES control scheme [37], Perturb and Observe (P&O) method [48] or other type of searching algorithm [20, 49]. One MEPT algorithm will be implemented in the SIDO block to generate the both current references I ref and I ref1 . The voltage on DC bus could be chosen lower than 350-400 V if a bidirectional Z-Source Inverter (ZSI) is used to connect the AC loads [50]. Furthermore, the boost converter that interface the FC stack to DC bus will have a lower voltage ratio, being here implemented through a basic unidirectional boost topology. Thus, the DC voltage, u dc , was set to 250 V on the DC bus, which is modeled as a capacitor, C DC [51]. Consequently, the rated voltage of the rechargeable Li-ion batteries stack will be 250 V. The rated voltage of the ultracapacitors stack was chosen of 100 V. The bidirectional buck-boost converter interface the ultracapacitors stack to the DC bus, ensuring the dynamic power compensation on the DC bus. Linear and nonlinear control techniques can be used to regulate the voltage on the DC bus [46]. The reference current 2, I ref2 , is computed on LF control block based on the AV value of requested FC power on DC bus, P Load – P RES , considering that the AV value of the ESS power is zero (because the batteries stack operates in CS mode based on the LF control proposed here). It is known that the dynamic FC response is dependent to the fueling rate, stoichiometric ratio, temperature, humidity and pressure [52]. Consequently, the fuel regulators must include a rate limiter to avoid the gas starvation [53] and a saturation block to operate the FC stack in available range of FC power. The minimum level of the FC power is set different to zero, avoiding the gas starvation that may appear during a repetitive start-up. So, the FC stack will operate in standby-mode at low fueling rates during the phases of low power demand, when RES power is higher that load demand. Consequently, the power flow balance on DC bus is a key relationship to design the EMU strategies [15,21,24,54]. 3. The EMU strategies to control the fueling rates Besides the power flow balance, which means acquisition of the all power flows (Figure 1), the other key decision parameters for the EMU strategies are the level of the FC and WT power, the state signals (such as SOC of the batteries and ultracapacitors stacks), the voltage on the FC and DC bus, the user signals (such as personalized prioritization of the loads into a smart home or environmental and internal ambient rules for the thermal comfort, rules to start the electrolyzer or charge the electrical vehicle, and so on), and the protection signals from and outside of the HPS. Note that the main objective for all applied EMU strategies in the integrated HPS is to ensure the load demand. Consequently, the analysis performed in this paper is focused to show this based on the results obtained with the proposed LF and MEPT control loops, without considering the EMU meta-rules based on other key decision parameters mentioned above. For example, the operating meta-rule for the hydrogen production via water electrolysis depends on the excess or shortage of power from the RES in comparison with the load demand, the SOC level of the batteries stack, the level of hydrogen in the storage tank of SOC. Implementation of such meta-rule based on the concept of Hierarchical Control Theory [31- 32], which is mainly applied in integrated hybrid systems to schedule the operation of the involved subsystems based on a predefined hierarchy made on the decision flow [27], could be simple if the RES power flow is constant or varying slowly during a load cycle. The huge variability of RES power on the DC bus and random behavior of the load demand in a smart home increase the complexity of the EMU [25-26]. Furthermore, frequent start-up and shut-down actions for the FC stack and electrolyzer may degrade their life cycle [32,55]. Therefore, the EMU strategy based on LF control that continuous operates the FC is proposed here. The capacity of the batteries stack that operates in CS mode will be designed at minimum value needed to compensate the short-term variability of the RES power and sharp peaks of the random load demand. Thus, the batteries’ lifespan is improved and the cost of the batteries stack is substantially reduced, and subsequently these influence the operating and maintenance costs of the entire HPS. The MEPT control is proposed here ...

Context 7

... input energy source. Besides this general used connection of the energy sources to the DC bus [40], which define the DC bus configuration, there are other two possible configurations to integrate different RESs, FC and ESS into a RES/FC HPS: AC coupled configuration and Hybrid coupled configuration [27,41, 42]. In this paper, the DC coupled configuration and equivalent DC load concept is used to test the control performances under different RES and load power profiles. 2.1. Unit sizing and optimization As it was mentioned on Introduction, the RESs have a random behavior. Thus, it is difficult to be predicted the available RES power flow in order to size the FC stack and ESS to make face to any load demand. In sunny and windy days the ESS is charged up to higher charge limit (set for the SOC) and electrolyzer refills the hydrogen tanks that are underutilized. Instead of this favorable case, the RES power flow is low during the cloudy days with moderate wind. Consequently, the ESS is discharged under the lower discharge limit, the electrolyzer is stopped, and the hydrogen tank may be empty, if this situation continues on a long period of time, which is difficult to be predicted under the current climate change. So, the size of the FC and batteries stack, the electrolyzer and hydrogen storage tanks must to be designed carefully considering the load demand and RES power flow available in all extreme weather conditions. Furthermore, the load demand of a smart home has also a random power profile. Thus, the unit sizing and optimization of standalone RES/FC HPS is not a trivial design problem, requiring an algorithm to size the HPS components by minimizing the system cost while maintaining system reliability [9-10]. It is obvious that the design objectives such as acceptable cost and reliability level are conflicting with one another [12,31]. For example, over sizing the HPS components will increase the HPS cost while under sizing can lead in lack of power to continuously sustain the load demand. So, the algorithms to size the HPS components must ensure a reasonable tradeoff between the design objectives, optimizing the HPS energy sources to achieve the target levels of cost and reliability. The HPS reliability is determined by estimating the loss of load probability, which is the ratio between estimated power deficit and the load demand during a load cycle, or appropriate other reliability indicators such loss of power probability, loss of power supply probability, and load coverage rate [42]. Besides the HPS cost and reliability, the sizing algorithms optimize the HPS components or other HPS parameters, such as investment cost, output energy cost, fuel consumption or appropriate other parameters such as net present cost, levelized cost of energy and fuel efficiency [43]. The algorithms to size the HPS components obviously depend on the availability of the weather data on short or long term, so two class are defined in the literature: conventional sizing algorithms based on weather data and expert algorithms if the weather data is not available or available only on short term, such is the case of remote isolated sites [19,41,42]. In first case, the concept of energy balance is a simple algorithm of sizing the HPS components [43]. In last case, the algorithms based on artificial intelligence are usually used [44, 45]. The rated power of the FC stack is set here to be higher than the maximum load demand because the sizing objective is outside the scope of this paper. 2.2. The structure of the proposed unit The HPS unit structure based on RES and FC system that has a new fueling control strategy based on LF and ES control loops is proposed here (Figure 1). It is known that the FC stack needs an ESS under dynamic loads to improve the HPSs system performance under random load demand and random RES power profile [46]. The ESS topology used here is of semi-active type: batteries stack direct connected to the DC bus and ultracapacitors stack via a bidirectional power converter [47]. This hybrid ESS topology is usually used due their advantages: the SOC variation of the ultracapacitors stack in large range ensures the dynamically compensation of the power flow balance at reduced cost. Thus, the proposed HPS that is shown in Figure 1 is composed of energy sources (FC, WT, and PV systems), ESS (batteries and ultracapacitors stacks), equivalent load (AC and DC loads), multiport power converter, and EMU (containing the ES and LF controllers). The fuel is classically or efficiently controlled based on static feed-forward (sFF) control or ES control scheme, which means the use of the FC current (I FC ) or the reference current 1 (I ref1 ) for the fuel regulator to control the fuel flow rate (FuelFr). The second ES controller generates the reference current (I ref ) to control the FC power in the LF control loop (see Figure 2). The LF controller generates the reference current 2 (I ref2 ) to control the air flow rate (AirFr) (Figure 1). Note that both ES controllers have as input the same signal, the FC net power (P FCnet ), as it shown in Figure 2. In fact, the unit structure shown in Figure 1 is detailed in Figure 2a, where is given the two simulation diagram implemented in MATLAB – Simulink ® simulation environment to control the AirFr based LF loop and FuelFr based sFF or ESC loop. Other two control configurations to LF control the FuelFr are shown in Figure 2b, where the AirFr is sFF or ESC controlled. All four control configurations will be analyzed here. The FC system must to operate efficiently, close to the maximum of the FC net power (named MEP). Different MEPT algorithm are proposed in the literature such as the ES control scheme [37], Perturb and Observe (P&O) method [48] or other type of searching algorithm [20, 49]. One MEPT algorithm will be implemented in the SIDO block to generate the both current references I ref and I ref1 . The voltage on DC bus could be chosen lower than 350-400 V if a bidirectional Z-Source Inverter (ZSI) is used to connect the AC loads [50]. Furthermore, the boost converter that interface the FC stack to DC bus will have a lower voltage ratio, being here implemented through a basic unidirectional boost topology. Thus, the DC voltage, u dc , was set to 250 V on the DC bus, which is modeled as a capacitor, C DC [51]. Consequently, the rated voltage of the rechargeable Li-ion batteries stack will be 250 V. The rated voltage of the ultracapacitors stack was chosen of 100 V. The bidirectional buck-boost converter interface the ultracapacitors stack to the DC bus, ensuring the dynamic power compensation on the DC bus. Linear and nonlinear control techniques can be used to regulate the voltage on the DC bus [46]. The reference current 2, I ref2 , is computed on LF control block based on the AV value of requested FC power on DC bus, P Load – P RES , considering that the AV value of the ESS power is zero (because the batteries stack operates in CS mode based on the LF control proposed here). It is known that the dynamic FC response is dependent to the fueling rate, stoichiometric ratio, temperature, humidity and pressure [52]. Consequently, the fuel regulators must include a rate limiter to avoid the gas starvation [53] and a saturation block to operate the FC stack in available range of FC power. The minimum level of the FC power is set different to zero, avoiding the gas starvation that may appear during a repetitive start-up. So, the FC stack will operate in standby-mode at low fueling rates during the phases of low power demand, when RES power is higher that load demand. Consequently, the power flow balance on DC bus is a key relationship to design the EMU strategies [15,21,24,54]. 3. The EMU strategies to control the fueling rates Besides the power flow balance, which means acquisition of the all power flows (Figure 1), the other key decision parameters for the EMU strategies are the level of the FC and WT power, the state signals (such as SOC of the batteries and ultracapacitors stacks), the voltage on the FC and DC bus, the user signals (such as personalized prioritization of the loads into a smart home or environmental and internal ambient rules for the thermal comfort, rules to start the electrolyzer or charge the electrical vehicle, and so on), and the protection signals from and outside of the HPS. Note that the main objective for all applied EMU strategies in the integrated HPS is to ensure the load demand. Consequently, the analysis performed in this paper is focused to show this based on the results obtained with the proposed LF and MEPT control loops, without considering the EMU meta-rules based on other key decision parameters mentioned above. For example, the operating meta-rule for the hydrogen production via water electrolysis depends on the excess or shortage of power from the RES in comparison with the load demand, the SOC level of the batteries stack, the level of hydrogen in the storage tank of SOC. Implementation of such meta-rule based on the concept of Hierarchical Control Theory [31- 32], which is mainly applied in integrated hybrid systems to schedule the operation of the involved subsystems based on a predefined hierarchy made on the decision flow [27], could be simple if the RES power flow is constant or varying slowly during a load cycle. The huge variability of RES power on the DC bus and random behavior of the load demand in a smart home increase the complexity of the EMU [25-26]. Furthermore, frequent start-up and shut-down actions for the FC stack and electrolyzer may degrade their life cycle [32,55]. Therefore, the EMU strategy based on LF control that continuous operates the FC is proposed here. The capacity of the batteries stack that operates in CS mode will be designed at minimum value needed to compensate the short-term variability of the RES power and sharp peaks of the random load demand. Thus, the batteries’ lifespan is improved and the cost of the ...

Context 8

... Introduction The environment could be protected from further deterioration if the RESs will be used more intensive in energy production. Furthermore, the solar and wind energy are free and clean, but available at variable levels related to the local environment parameters (insolation, wind speed and temperature). Besides, the long lifetime and low maintenance requirements recommend these types of RESs [ 1]. The combinations of PV and WT systems with ESS for the surplus of RES energy or lack of power under load peaks have a widespread use [2, 3, 4, 5, 6, 7, 8, 9 – 10]. The main feature of RES HPS is to combine more than one RES technology, optimizing their power flows in order to obtain efficiencies higher than that could be obtained from a single RES [2]. Thus, the RES HPS can address limitations in terms of RES power variability, variable load demand, efficiency, reliability, emissions and implementing costs [3,4]. Because the electricity consumption in the residential sector represents more than 30% of the overall energy consumption [2], the importance of optimization strategies of the energy usage in a smart house is obvious [5, 6]. Besides RESs, the RES HPS can contain one conventional energy sources, such a diesel generator, which will operate as backup energy source in standalone or grid connected mode [7]. In last decade, the diesel generator is replaced by the FC systems [8,9] due to its disadvantages such as greenhouse emissions, high maintenance costs, and low scalability to meet changing in load demands [10]. Thus, the RES/FC HPS with hydrogen storage ensures eco-friendly operation, being scalable for small- and medium scale power applications [11, 12, 13]. Consequently, the hydrogen production via the eco-friendly process of water electrolysis is proposed as alternative to expensive and unsafe solution of hydrogen storage in tanks [12] Anyway, some issues related to the FC use (such as fuel starvation phenomenon, safe operation, high cost of the membrane and catalyst, and so on) are s till in researchers’ attention [13]. However, the RES/FC HPS can provide multifold advantages based on EMU strategies proposed [14]. The LF control loop was proposed for a FC HPS (without support of the RESs) in [15], but this strategy could be also used in control of an active power filter for a solid oxide FC [16] or efficient operation of combined heat and power system [17]. The EMU strategy proposed here based on the LF and MEPT control loops set the fueling rates considering the average (AV) of the FC power requested on the DC bus and then generates the maximum net FC for these fueling rates. Thus, the efficiency of the whole HPS increase, the fuel consumption is minimized, and the ESS behavior and life-cycle are improved. The AV value of the FC net power and then the reference current are evaluated in the LF control block based on mean-value block, but other signal processing techniques could be used as well [3,8]. The reference current is used to set one of the fueling rates. The MEPT controller harvests the maximum available FC net power that can be generated by searching the MEP via the other fueling rate. The advantages of the four fueling topologies based on the LF and MEPT loops are shown in this paper. Note that EMU meta-rules to ensure the energy dispatch between the DC bus and the electrolyzer or plug-in electrical vehicle during an energy surplus (RES power higher than load demand) is only mentioned here, being intensively studied in the literature [11-14,18]. Also, the optimization of the load demand through scheduling the non-priority loads is outside the scope of this paper. The LF control is proposed here to make face efficiently to all load profile if the energy sources and backup source are properly designed [7-12,19]. A PEMFC is used as backup source because of the advantages of this technology related to other FC technologies [13-15,20]. In addition, the hydrogen is abundant in the nature and can be obtained by reforming the natural gas, ethanol, methanol, biogas, and so on, or by water electrolyzing. Besides the eco-friendly operation of the RES/FC HPS, the economic aspect makes the PEMFC a competitive technology in comparison to diesel generator, offering a reduced maintenance effort and cost [5-6,21]. So, integrating RES, FC and ESS into a RES/FC HPS via a multiport power converter (see Figure 1), the hybrid source will have the energy generation/storage support to make face to energy consumption and variability of the RES power that are not always synchronized with the load demand. Thus, EMU strategies are important in optimizing the energy management of the energy sources [1-4, 9, 11-15, 21]. The main goal of the EMU strategies must to ensure the load demand. Besides this, the other specific goals of the EMU strategies must be related to fuel consumption, energy efficiency of the energy sources that must to be also safe operated, life cycle of the HPS, and cost [11-15, 21-22]. Consequently, some variables of the HPS (named in Figure 1 as state signals; for example the state-of-charge (SOC) level of the ESS devices) and control variables (for example the DC bus voltage, FC current, load demand, and RES power level) must to be inputs of the EMU strategy. Besides these signals, the protection and user signals are used to safe operation of the HPS. The studies above mentioned analyze the energy efficiency of the EMU strategies using dynamic models of the HPS based on short-term simulations with time scale of seconds [23] or minutes [24]. The voltage regulation on the DC bus of the PV/FC HPS is approached by classical linear or non-linear control algorithm (for example, based on the differential flatness principle [25]). Different energy management strategies are proposed in the literature. For example, the fuzzy logic control was used in [26] to monitor the SOC of the ESS devices, improving the utilization costs and lifetime of the battery and hydrogen system as well. Also, the management of power flows between the FC and ESS in HPS grid connected is analyzed based on a fuzzy logic control [22, 27] or ANFIS control [28]. The energy efficiency of a standalone RES/FC HPS based on EMU strategy to make face to RES power variations was investigated in [24,29]. In long-term analysis, the main goal of the EMU strategies is focused on meeting the load demand considering in the energy dispatch other specific goals such as the cost and the energy efficiency, and state signals such as the SOC level of the ESS devices, etc. An adaptive predictive strategy for a RES/FC HPS to meet the above goals is analyzed with a time scale of hours in [30]. Three EMU strategies to meet the load demand and increase the energy efficiency based on long-term simulations throughout one year are proposed in [12,31]. Also, the expected lifetime of a standalone HPS was evaluated in [32]. The energy strategies mentioned above are mainly based on a monitoring of the power flows and states of the ESS devices to decide the energy dispatch between the FC and battery to the load [21,33]. In this paper, the battery operates in CS mode to minimize its size, the load demand being sustained based on the LF control proposed by the RESs and FC system. Also, it can be noted that the dynamics of the energy sources and especially of the power interfaces are frequently neglected [34]. The power converters are modeled in this paper using components from the SimPowerSystems library of the Matlab - Simulink® and not using an AV model that is recommended for a long-term analysis of the HPS behavior and performance. Thus, a short-term analysis will be performed here based on the sample frequency of the 10 kHz, which is more than enough to efficiently use the real profile of the RES power based on advanced Maximum Power Point Tracking (MPPT) control with high search speed and good tracking accuracy [23,35]. The MEPT technique based on the Extremum Seeking (ES) control scheme is used here to control one of the fueling rates [36], but note that any MEPT advanced technique could be used as well. The ES control is perturbed – based scheme, using dithers with different frequencies or orthogonal dithers to implement Single- Input Double-Outputs (SIDO) ES control scheme [37]. Note that only the performances of a PEMFC system are tested in [35-37], while here the performances of a whole RES/FC HPS are shown. Thus, besides the EMU that includes the LF and MEPT control schemes, the main components of the standalone RES/FC HPS includes a PV array, a set of wind generators, a FC system as backup source, a short – term ESS based on li-ion batteries and ultracapacitors, and a long-term ESS that consists of an electrolyzer and a hydrogen storage system with pressurized tanks (only this is shown in Figure 1), DC loads and AC load interfaced with inverters (named the equivalent load), and a multiport power converter or independent power interfaces for each energy sources and load. Summarizing, the research directions in field of the RES/FC HPS that are approached in the studies mentioned above are the following [38]: • The RES need advanced control techniques to harvest all power available, operating the RES close to MPP. This control is mandatory due to poor efficiency of solar PV, which is the main impediment in encouraging its use until the PV technology ...

Context 9

... from further deterioration if the RESs will be used more intensive in energy production. Furthermore, the solar and wind energy are free and clean, but available at variable levels related to the local environment parameters (insolation, wind speed and temperature). Besides, the long lifetime and low maintenance requirements recommend these types of RESs [ 1]. The combinations of PV and WT systems with ESS for the surplus of RES energy or lack of power under load peaks have a widespread use [2, 3, 4, 5, 6, 7, 8, 9 – 10]. The main feature of RES HPS is to combine more than one RES technology, optimizing their power flows in order to obtain efficiencies higher than that could be obtained from a single RES [2]. Thus, the RES HPS can address limitations in terms of RES power variability, variable load demand, efficiency, reliability, emissions and implementing costs [3,4]. Because the electricity consumption in the residential sector represents more than 30% of the overall energy consumption [2], the importance of optimization strategies of the energy usage in a smart house is obvious [5, 6]. Besides RESs, the RES HPS can contain one conventional energy sources, such a diesel generator, which will operate as backup energy source in standalone or grid connected mode [7]. In last decade, the diesel generator is replaced by the FC systems [8,9] due to its disadvantages such as greenhouse emissions, high maintenance costs, and low scalability to meet changing in load demands [10]. Thus, the RES/FC HPS with hydrogen storage ensures eco-friendly operation, being scalable for small- and medium scale power applications [11, 12, 13]. Consequently, the hydrogen production via the eco-friendly process of water electrolysis is proposed as alternative to expensive and unsafe solution of hydrogen storage in tanks [12] Anyway, some issues related to the FC use (such as fuel starvation phenomenon, safe operation, high cost of the membrane and catalyst, and so on) are s till in researchers’ attention [13]. However, the RES/FC HPS can provide multifold advantages based on EMU strategies proposed [14]. The LF control loop was proposed for a FC HPS (without support of the RESs) in [15], but this strategy could be also used in control of an active power filter for a solid oxide FC [16] or efficient operation of combined heat and power system [17]. The EMU strategy proposed here based on the LF and MEPT control loops set the fueling rates considering the average (AV) of the FC power requested on the DC bus and then generates the maximum net FC for these fueling rates. Thus, the efficiency of the whole HPS increase, the fuel consumption is minimized, and the ESS behavior and life-cycle are improved. The AV value of the FC net power and then the reference current are evaluated in the LF control block based on mean-value block, but other signal processing techniques could be used as well [3,8]. The reference current is used to set one of the fueling rates. The MEPT controller harvests the maximum available FC net power that can be generated by searching the MEP via the other fueling rate. The advantages of the four fueling topologies based on the LF and MEPT loops are shown in this paper. Note that EMU meta-rules to ensure the energy dispatch between the DC bus and the electrolyzer or plug-in electrical vehicle during an energy surplus (RES power higher than load demand) is only mentioned here, being intensively studied in the literature [11-14,18]. Also, the optimization of the load demand through scheduling the non-priority loads is outside the scope of this paper. The LF control is proposed here to make face efficiently to all load profile if the energy sources and backup source are properly designed [7-12,19]. A PEMFC is used as backup source because of the advantages of this technology related to other FC technologies [13-15,20]. In addition, the hydrogen is abundant in the nature and can be obtained by reforming the natural gas, ethanol, methanol, biogas, and so on, or by water electrolyzing. Besides the eco-friendly operation of the RES/FC HPS, the economic aspect makes the PEMFC a competitive technology in comparison to diesel generator, offering a reduced maintenance effort and cost [5-6,21]. So, integrating RES, FC and ESS into a RES/FC HPS via a multiport power converter (see Figure 1), the hybrid source will have the energy generation/storage support to make face to energy consumption and variability of the RES power that are not always synchronized with the load demand. Thus, EMU strategies are important in optimizing the energy management of the energy sources [1-4, 9, 11-15, 21]. The main goal of the EMU strategies must to ensure the load demand. Besides this, the other specific goals of the EMU strategies must be related to fuel consumption, energy efficiency of the energy sources that must to be also safe operated, life cycle of the HPS, and cost [11-15, 21-22]. Consequently, some variables of the HPS (named in Figure 1 as state signals; for example the state-of-charge (SOC) level of the ESS devices) and control variables (for example the DC bus voltage, FC current, load demand, and RES power level) must to be inputs of the EMU strategy. Besides these signals, the protection and user signals are used to safe operation of the HPS. The studies above mentioned analyze the energy efficiency of the EMU strategies using dynamic models of the HPS based on short-term simulations with time scale of seconds [23] or minutes [24]. The voltage regulation on the DC bus of the PV/FC HPS is approached by classical linear or non-linear control algorithm (for example, based on the differential flatness principle [25]). Different energy management strategies are proposed in the literature. For example, the fuzzy logic control was used in [26] to monitor the SOC of the ESS devices, improving the utilization costs and lifetime of the battery and hydrogen system as well. Also, the management of power flows between the FC and ESS in HPS grid connected is analyzed based on a fuzzy logic control [22, 27] or ANFIS control [28]. The energy efficiency of a standalone RES/FC HPS based on EMU strategy to make face to RES power variations was investigated in [24,29]. In long-term analysis, the main goal of the EMU strategies is focused on meeting the load demand considering in the energy dispatch other specific goals such as the cost and the energy efficiency, and state signals such as the SOC level of the ESS devices, etc. An adaptive predictive strategy for a RES/FC HPS to meet the above goals is analyzed with a time scale of hours in [30]. Three EMU strategies to meet the load demand and increase the energy efficiency based on long-term simulations throughout one year are proposed in [12,31]. Also, the expected lifetime of a standalone HPS was evaluated in [32]. The energy strategies mentioned above are mainly based on a monitoring of the power flows and states of the ESS devices to decide the energy dispatch between the FC and battery to the load [21,33]. In this paper, the battery operates in CS mode to minimize its size, the load demand being sustained based on the LF control proposed by the RESs and FC system. Also, it can be noted that the dynamics of the energy sources and especially of the power interfaces are frequently neglected [34]. The power converters are modeled in this paper using components from the SimPowerSystems library of the Matlab - Simulink® and not using an AV model that is recommended for a long-term analysis of the HPS behavior and performance. Thus, a short-term analysis will be performed here based on the sample frequency of the 10 kHz, which is more than enough to efficiently use the real profile of the RES power based on advanced Maximum Power Point Tracking (MPPT) control with high search speed and good tracking accuracy [23,35]. The MEPT technique based on the Extremum Seeking (ES) control scheme is used here to control one of the fueling rates [36], but note that any MEPT advanced technique could be used as well. The ES control is perturbed – based scheme, using dithers with different frequencies or orthogonal dithers to implement Single- Input Double-Outputs (SIDO) ES control scheme [37]. Note that only the performances of a PEMFC system are tested in [35-37], while here the performances of a whole RES/FC HPS are shown. Thus, besides the EMU that includes the LF and MEPT control schemes, the main components of the standalone RES/FC HPS includes a PV array, a set of wind generators, a FC system as backup source, a short – term ESS based on li-ion batteries and ultracapacitors, and a long-term ESS that consists of an electrolyzer and a hydrogen storage system with pressurized tanks (only this is shown in Figure 1), DC loads and AC load interfaced with inverters (named the equivalent load), and a multiport power converter or independent power interfaces for each energy sources and load. Summarizing, the research directions in field of the RES/FC HPS that are approached in the studies mentioned above are the following [38]: • The RES need advanced control techniques to harvest all power available, operating the RES close to MPP. This control is mandatory due to poor efficiency of solar PV, which is the main impediment in encouraging its use until the PV technology will be improved. • The power losses in power interfaces have been substantially reduced using advanced topologies, switches and appropriate control scheme, eventually integrated in multiport power converter structure. In general, the energy efficiency of power converters used in HPS is higher than 95%. • The hybrid batteries/ultracapacitors ESS is used in HPS to ensure the power flow balance on the DC bus, but their life-cycle need to be improved through innovative technologies, too. The CS mode proposed here for the ESS can improve the life-cycle of the batteries stack, besides other advantages such as reduced size and low costs of the ESS. • The ...