Efficiency of unitized reversible fuel cell systems

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 5 8 0 3 e5 8 1 5

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/he

Efficiency of unitized reversible fuel cell systems Hiroshi Ito a,*, Naoki Miyazaki b, Masayoshi Ishida b, Akihiro Nakano a a Research Institute for Energy Conservation, National Institute of Advanced Industrial Science and Technology (AIST), 1-2-1 Namiki, Tsukuba 305-8564, Japan b Department of Engineering Mechanics and Energy, University of Tsukuba, 1-1-1 Tennoudai, Tsukuba 305-8573, Japan

article info

abstract

Article history:

A pilot-scale unitized reversible fuel cell (URFC) system was installed in our laboratory

Received 7 December 2015

(AIST) and its efficiency was comprehensively evaluated. This URFC system was composed

Received in revised form

of a proton exchange membrane (PEM)-based cell/stack and balance of plant (BOP)

28 January 2016

adaptable for both electrolysis (EL) and fuel cell (FC) operation modes. First, the efficiency of

Accepted 31 January 2016

this URFC system was evaluated at both the stack level and system level. Next, the effi-

Available online 14 March 2016

ciency was also evaluated by considering the amount of recovered heat during EL and FC modes assuming combined heat and power (CHP) applications. Then, the calculated effi-

Keywords:

ciency of the URFC system at each operation mode was compared with that of specialized

Hydrogen energy system

devices of PEM electrolyzer and PEM fuel cell as reference standards. The comparison re-

Balance of plant

veals that the stack performance of the operation modes of the URFC were comparable to

System design

that of the respective specialized device for EL and FC, whereas the system performance

Efficiency

taking into account the BOP energy consumption requires improved efficiency for the FC

Heat recovery

mode of the URFC system. Finally, analysis of the system efficiency including heat recovery from the URFC system was used to evaluate quantitatively the potential of heat recovery, revealing that recovery and utilization of thermal energy are essential for improving the efficiency. It was identified that the round-trip efficiency of the system would be improved from 0.182 to 0.589 by utilizing thermal energy. Copyright © 2016, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.

Introduction Attention is currently focusing on renewable energy as a countermeasure to the depletion of fossil fuels and to global warming. Renewable energy sources (RES) such as solar and wind have significant potential, but have major drawbacks related to their fluctuating and intermittent nature. When the proportion of power output from RES such as photovoltaic (PV) or wind power is considerably lower than the capacity of the

electricity grid network, the fluctuation of RES can be balanced by conventional power generation. In the recent rapid, widespread introduction of unstable RES (PV and wind power) implemented in numerous countries, the capacity of unstable RES might reach the limitation of grid balancing. This problem will be critical in small-scale grids in rural and island areas. Under these circumstances, energy storage is considered a solution for stabilizing the supply of electricity [1]. Hydrogen is a unique storage medium for unstable RES because it offers many advantages for the storage of RES

* Corresponding author. Tel.: þ81 29 861 7262; fax: þ81 29 851 7523. E-mail address: [email protected] (H. Ito). http://dx.doi.org/10.1016/j.ijhydene.2016.01.150 0360-3199/Copyright © 2016, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.

5804

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 5 8 0 3 e5 8 1 5

Nomenclature Istack istack FBLW

stack current, A stack current density, A m2 flow rate of air supplied from air-blower, m3 s1 (L min1) FH2 flow rate of circulating hydrogen, m3 s1 (L min1) pressure of produced/supplied hydrogen from/to PH2 stack, Pa (abs) power consumption of air-blower, W PBLW Ploss_AC/DC power loss related to AC-DC converter (conversion efficiency of AC-DC converter: 0.85), W power consumption of pump 1 (circulation pump PP1 for cooling water), W power consumption of pump 2 (circulation pump PP2 for electrolysis water), W power consumption of pump 3 (circulation pump PP3 for hydrogen), W Qloss_EL heat loss to ambient during electrolysis operation, J (kWh) Qloss_FC heat loss to ambient during fuel cell operation, J (kWh) Qrcvr_EL recovered heat from the stack with cooling water during electrolysis operation, J (kWh) Qrcvr_FC recovered heat from the stack with cooling water during fuel cell operation, J (kWh) Tstack stack temperature, K ( C) operation time of electrolysis, s (h) tEL operation time of fuel cell, s (h) tFC cell voltage, V Vcell Vcell_avg average cell voltage calculated by dividing Vstack in the number of cells in series, V overall stack voltage, V Vstack WBOP_EL total energy consumption of balance of plant (BOP) during electrolysis operation (¼ ðPP1 þ PP2 þ Ploss_AC=DC Þ  tEL ), J (kWh)

output; 1) hydrogen can be produced with water electrolysis from intermittent RES (e.g., PV and wind power), 2) produced hydrogen can be stored in numerous forms (e.g., compressed, liquefied, metal hydride) without self-discharge over time, 3) carbon-free electricity production can be achieved using a fuel cell, and 4) system power (kW) and stored energy (kWh) can be independently optimized. According to literature [2e7], hydrogen utilization systems have been extensively studied for several decades. Up to now, the major component of such systems has been a small-scale distributed energy system typically consisting of an RES, water electrolyzer, hydrogen storage apparatus, and a fuel cell. PVs and wind turbines are most common RES for these systems. The lower operating temperature of proton exchange membrane fuel cells (PEMFCs) (~80  C) enables excellent start-up performance, a simplified BOP, and a compact total system. However, also due to low operating temperature, PEMFCs are sensitive to impurities in the hydrogen. Power

WBOP_FC total energy consumption of balance of plant (BOP) during fuel cell operation (¼ ðPP1 þ PP3 þ PBLW Þ  tFC ), J (kWh) Wstack_EL energy input into the stack during electrolysis operation (¼Vstack ,Istack  tEL ), J (kWh) Wstack_FC energy output from the stack during fuel cell operation (¼Vstack ,Istack  tFC ), J (kWh) Greek symbols DHH2 cons enthalpy of consumed hydrogen during fuel cell operation calculated using amount and standard enthalpy of formation of hydrogen (HHV), J (kWh) DHH2_prod enthalpy of produced hydrogen during electrolysis operation calculated using amount and standard enthalpy of formation of hydrogen, J (kWh) hstack_EL stack efficiency during electrolysis operation (cf. Table 2) hstack_FC stack efficiency during fuel cell operation (cf. Table 2) hstack_RT stack efficiency of round-tip (cf. Table 2) hsys_EL system efficiency during electrolysis operation (cf. Table 2) hsys_FC system efficiency during fuel cell operation (cf. Table 2) hsys_RT system efficiency of round-trip (cf. Table 2) h0 sys _EL system efficiency including heat recovery during electrolysis operation (cf. Table 2) h0 sys _FC system efficiency including heat recovery during fuel cell operation (cf. Table 2) h0 sys _RT system efficiency including heat recovery for round-trip (cf. Table 2)

production by a fuel cell introduced into a hydrogen utilization system can be supplied by stored pure hydrogen. Consequently, PEMFCs are promising candidates as power production devices in hydrogen storage systems. In general, a hydrogen utilization system is connected to either an AC-grid or a local DC-bus [2e5]. Although the operation strategy depends on the specific purpose of the system, there must be no overlap time between the electrolysis (EL) operation mode and fuel cell (FC) operation mode regardless of the type of application. Proton exchange membrane (PEM) electrolyzers and PEMFCs both use a common proton exchange membrane (PEM) as the electrolyte, and have a similar cell/stack design. From the technical viewpoint, it is possible to establish a unitized cell/stack of these two electrochemical devices. A unitized reversible fuel cell (URFC) based on PEM has been studied for several decades [8e11] as an energy device for space [12e14] or terrestrial applications [15,16]. As an energy-conversion device in stationary systems, URFCs have

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 5 8 0 3 e5 8 1 5

several advantages over discrete installation of both an electrolyzer and a fuel cell; 1) lower cost of total system, 2) higher operating ratio, 3) reduced maintenance, and 4) smaller footprint. Because the overall round-trip electrical efficiency of hydrogen utilization system is estimated at 30% at most (i.e., the ratio of power output from FC and power input to EL), the thermal energy utilization must be considered for improving the utility value of the system. PEMFC-based combined heat and power (CHP) systems have been extensively studied over the past several decades [17e24], and several commercial products are already available [25,26]. The maximum electrical efficiency of a PEMFC (including natural-gas reforming) is about 40%, and overall efficiency including thermal energy utilization exceeds 90% [25,26]. In a typical PEMFC-based CHP system, the stack is cooled by water (liquid) and has a heat recovery system (i.e., heat exchanger, cooling water loop, and hot water tank). Several previous works [27e31] discussed the relation between operating parameters (i.e., stack temperature and stoichiometry of supplied gases) and the stack performance of PEMFC system (i.e., electrical and thermal energy output). The change in the gas stoichiometry (i.e., gas flow rate) accompanies the change in the blower (or compressor) power, and brings the change in the parasitic energy consumption and the system efficiency as a result. However, to the best of our knowledge, there is lack of information about the change of system efficiency owing to the change of operating parameters. Most of previous studies on URFC have been concerning the optimization of cell components [32e40]. However, there is no previous study examining the system performance of URFC including parasitic energy consumption or comparing URFC system performance with conventional discrete system of EL and FC. The verification by the system level is indispensable to confirm the utility of URFC. A pilot-scale URFC system was installed in our laboratory and comprehensively evaluated. This system included not only a cell/stack but also a BOP that consists of pumps, blower, accumulation tank, humidifier, heat exchanger, and pipelines. With this system, we examined the performance not only of the cell/stack but also of the total system by evaluating the energy consumption of the BOP. In addition, we also evaluated the thermal energy recovered from the cell/stack during both EL and FC operation modes. For this purpose, three different efficiencies were defined and calculated for both the EL and FC modes: stack efficiency, system efficiency, and system efficiency including heat recovery. Also calculated was the round-trip efficiency defined by merging the efficiencies of both EL and FC modes. The performance of the URFC was compared with that of an individual PEM electrolyzer and a PEMFC. As for FC operation, the dependency of the efficiencies on the flow rate of gases (H2 and air) was examined on the basis of experimental results. Results can reveal information about the possibility of a URFC system as a component of a hydrogen utilization system.

Description of the URFC system The URFC stack and its BOP were designed and manufactured by Takasago Thermal Engineering Co. (Japan) [16] in the

5805

development of this URFC system as a main component of a stationary energy system in a large-scale building. In this pilot system, the nominal hydrogen production rate at EL operation mode was 1.0 Nm3/h and nominal power production capacity at FC mode was 0.5 kW. The system included a URFC stack, pipeline system, valves, pumps, air-blower, chiller, and two power devices (a DC-power and a DC-load), all of which were enclosed in a cabinet whose dimensions were 1280 (width)  740 (length)  1361 (height) mm (Fig. 1). Also enclosed within this cabinet was a system to control the stack current (Istack), stack temperature (Tstack), hydrogen pressure (PH2), and valve switching and to monitor all the operating parameters such as stack voltage (Vstack), flow rates of working fluids (H2, Air, and water), and so on.

URFC stack The URFC stack consisted of 10 rectangular cells, with an active electrode area of 250 cm2, and each cell consisted of a membrane electrode assembly (MEA) and bipolar plates. The MEA is a composite of a catalyst-coating membrane (CCM) and gas diffusion layers (GDLs) on both sides of the CCM. Table 1 lists the specifications of the cell and stack. The total weight of this stack including both end-plates (made of stainless steel) was about 55 kg. The stack temperature (Tstack) was measured with a sheathed thermocouple (Type T) inserted into the body of the end-plate with its tip at the center of the electrode area. The stack had six pipeline connections for fluids (Fig. 2), that is, the inlet and outlet lines of each fluid, namely, H2, O2 (or air), and cooling water. Each fluid line in the stack passed through the bipolar plates and membranes. Manifolds were designed to distribute both gases (H2 and either O2 or air) to one side of each cell. Each bipolar plate had a terminal to measure the cell voltage (Vcell).

Balance of plant (BOP) system Fabrication of an efficient URFC system must include the design of the BOP. To ensure an overall compact system, lines in the piping system must be shared by both operation modes as much as possible. Fig. 2 is a piping and flow diagram of the present system, showing six major line systems: inlet and outlet lines of H2 (left side of URFC stack in Fig. 2), inlet and outlet lines of either air or O2, electrolysis water circulation line (right side of URFC stack), and cooling water circulation line (upper side of URFC stack). The outlet line of H2 and that of either air or O2 were shared by both the EL and FC operation modes. In the EL mode, deionized (DI) water stored in the O2-water separator tank was supplied to the URFC stack using a circulation pump for electrolysis water (P2 in Fig. 2). The produced oxygen at the anode by EL was released from the stack with liquid water and fed back to the separator tank to remove the water. Oxygen gas was released from the tank and dried (cooled) with a heat exchanger and exhausted to the outside atmosphere. Hydrogen was simultaneously produced at the cathode. Because liquid water migrated from the anode (O2 side) to the cathode (H2 side) through the membrane by electro-osmosis drag during EL, hydrogen gas was also released with liquid water. Hydrogen was separated with

5806

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 5 8 0 3 e5 8 1 5

Fig. 1 e Photos of the URFC system installed in AIST; (a) overview, (b) view of components inside the cabinet.

Table 1 e Specifications of URFC stack. Cell/Stack configuration Electrode active area Number of cells in series Operation parameter Hydrogen production rate Power production (DC) Temperature of cell/stack Gas pressure of H2

250 cm2 10 Operation mode Electrolysis (EL) Fuel cell (FC) Electrolysis (EL) Fuel cell (FC) Electrolysis (EL) Fuel cell (FC)

1.0 Nm3/h (17 NL/min) 0.8 kW 40-50  C 60-70  C 0.1e1.0 MPa (abs) 0.15 MPa (abs)

liquid water at the H2-water separator tank and finally dried with a heat exchanger the same as for oxygen. Hydrogen pressure (PH2) was regulated between 0.1 and 1.0 MPa (abs) using a back pressure valve, while the valve in the H2 inlet line was closed during EL operation. Because the line pressure of the O2 side was fixed around 0.15 MPa (abs) throughout the EL operation, a differential pressure existed between the anode and cathode when H2 was compressed at the anode. During long-term operation, an increase in impurities in the circulation water for electrolysis cannot be avoided, resulting in an increase in electrical conductivity. Therefore, part of the circulation water was channeled into the bypass line for deionization using a water purifier. Because the operation

Fig. 2 e Piping and flow diagram of URFC system.

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 5 8 0 3 e5 8 1 5

temperature of the water purifier was limited to a maximum 40  C, the water in the bypass line was cooled with a heat exchanger located upstream of the purifier when the temperature of electrolysis water exceeded 40  C. As a result of purification, the electrical conductivity of the circulation water was kept under 1 mS cm1. Because the pressure loss of electrolysis water through the cell/stack increased with bubble production rate (i.e., current density istack), the flow rate decreased with istack even under constant power of pump P2 (PP2). In the design of this system, the minimum flow rate of the electrolysis water was set at 0.1 L/min per cell. In the FC operation mode, H2 gas and air were respectively supplied to the anode and cathode. Note that the anode and cathode during the FC operation were the reverse of those during EL. In the FC mode, the recirculation system was applied to supply H2 gas to the anode, that is, residual H2 released from the stack was recirculated with a recirculation pump (P3 in Fig. 2), and the amount of supplied H2 required was only for the reaction. In steady-state, the stoichiometric ratio of H2 was thus nearly 1 regardless of istack. At rated operation, the flow rate of recirculated H2 was 25 L/min, which corresponds to a stoichiometric ratio of 2.9 at istack ¼ 0.5 A cm2 operation (i.e., Istack ¼ 125 A). To maintain H2 at sufficiently high purity, the recirculation line was opened and H2 was fully recharged every 5 min. In the cathode of the FC mode, air was supplied by the air blower (BLW) and on the path to the stack was humidified with a membrane humidifier acting as a moisture exchanger between the air inlet and outlet lines. The dew point of inlet air was reached at about 60  C under rated flow rate of air (48 L/min). The flow rate of air was set from 40 to 70 L/min by changing the BLW power. In this flow rate range, the stoichiometric ratio of O2 in the air ranged from 1.9 to 3.4 at istack ¼ 0.5 A cm2. During the both EL and FC modes, Tstack was controlled by using circulating cooling water. In this cooling water loop, when Tstack reached a target temperature for operation, cold water was supplied to the heat exchanger from the chiller. The temperature of the cooling water was measured both at the inlet and outlet of the stack, and the thermal energy was calculated using this temperature difference and the flow rate of cooling water.

Efficiencies In this study, the performance of the URFC system was evaluated at the stack level and system level. In addition, considering CHP application with URFC, the heat recovery with cooling water was included in the evaluation. Three different efficiencies were defined and calculated for both the EL and FC modes: 1) stack efficiency (hstack), 2) system efficiency (hsys), and 3) system efficiency including heat recovery (h0 sys). Also calculated was the round-trip (RT) efficiency defined by merging the efficiencies of both EL and FC modes as defined in Table 2. The energy flow of the URFC system is schematically shown in Fig. 3. The efficiencies were evaluated on the basis of energy amount (J or KWh), and no loss related to hydrogen storage was considered here. The hstack represents the energy conversion between electricity and H2 only at the stack, when the BOP energy consumption is disregarded. In the case of EL operation,

5807

hstack_EL was defined as the ratio of the enthalpy of produced H2 (DHH2_prod ) during a certain period (tEL ) of EL operation to the energy input into the stack during EL operation (Wstack_EL), which corresponds toIstack  Vstack  tEL . The stack efficiency of the fuel cell (hstack_FC) was a kind of reciprocal of hstack_EL, that is, the ratio of energy output from the stack (Wstack_FC ¼ Istack  Vstack  tFC ) to the enthalpy of consumed H2 (DHH2_cons ) during a period (tFC ) of FC operation. The stack efficiency of round-trip (hstack_RT) was obtained by multiplying hstack_EL and hstack_FC. In the calculation of efficiencies, tEL and tFC were independently defined so as to equate the value of DHH2_prod and DHH2_cons . The enthalpy change of H2 was calculated based on the high heating value (HHV) of H2 (i.e., 285.83 kJ/mol). The hsys represents the performance of the system level considering BOP power consumption. In EL operation, the circulation pump for cooling water (P1) and for electrolysis water (P2) were operated. The combined power of these two pumps (PP1 þ PP2) was considered the BOP power consumption during EL operation (WBOP_EL ¼ ðPP1 þ PP2 Þ  tEL ). In addition, as discussed below (Section 3.4), in order to compare hsys of EL mode of the URFC (hsys_EL) with a specialized PEM electrolyzer, power conversion loss of the AC-DC converter (rectifier) (Ploss_AC/DC) was also included in WBOP_EL. In the definition of hsys_EL, WBOP_EL was added to the energy input for electrolysis (Wstack_EL). In FC operation, the circulation pump for H2 (P3) was switched on instead of P2, and the BLW was switched on to supply air to the cathode. However, in the case of hsys_FC, power loss by the conversion from DC to AC was not considered. Therefore, the total energy consumption of BOP during FC operation (WBOP_FC) consisted of the combined power of P1, P2, BLW, and operation time (i.e., ðPP1 þ PP3 þ PBLW Þ  tFC ) and was subtracted from the power output of FC in the calculation of the hsys of the fuel cell (i.e., hsys_FC). The same as hsys, the hsys of RT (i.e., hsys_RT) can be obtained by multiplying hsys_EL and hsys_FC, since DHH2_prod was equal to DHH2_cons . Assuming a CHP application of the URFC system, thermal energy recovered from the stack was considered. The recovered thermal energy (heat) was calculated based both on the temperature difference in cooling water between the outlet and inlet and on the flow rate of the water. In EL operation, the ratio of recovered heat was calculated by dividing the recovered thermal energy (Qrcvr_EL ) by the total energy input for electrolysis (Wstack_EL þ WBOP_EL), and then added to hsys_EL to obtain the system efficiency including heat recovery during electrolysis operation (h0 sys_EL). The system efficiency including heat recovery during FC operation (h0 sys_FC) was similarly calculated, that is, the recovered thermal energy during FC operation (Qrcvr_FC ) was divided by the input hydrogen enthalpy (DHH2_cons ) and then added to hsys_FC. For the evaluation of the RT system efficiency including heat recovery (h0 sys_RT), the ratio of recovered heat during each operation (EL/FC) was individually added to hsys_RT. In the present URFC system, a chiller was used not only to control the temperature of both the cooling water and electrolysis water (at the bypass line) but also for drying the exhaust gases (H2 and air/O2). The energy consumption of the chiller was higher than that of any of the pumps. Because the chosen cooling method depends on the system (e.g.,

5808

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 5 8 0 3 e5 8 1 5

Table 2 e Definitions of efficiency. Efficiency Stack efficiency

System efficiency

System efficiency including heat recovery

Operation mode

Definition DHH2 prod Wstack_EL

Electrolysis (EL)

hstack

EL

¼

Fuel cell (FC)

hstack

FC

¼ DHstack_FC H2 cons

Round-trip (RT)

hstack_RT ¼ hstack_EL  hstack_FC

Electrolysis (EL)

hsys

EL

prod ¼ Wstack_ELH2þW BOP_EL

Fuel cell (FC)

hsys

FC

¼

Round-trip (RT)

hsys_RT ¼ hsys_EL  hsys_FC

Electrolysis (EL)

h0sys

EL

¼ hsys

EL

Qrcvr EL þ Wstack_EL þWBOP_EL

Fuel cell (FC)

h0sys

FC

¼ hsys

FC

Qrcvr FC þ DH H2 cons

Round-trip (RT)

h0sys RT

W

DH

¼

Wstack_FC WBOP_FC DHH2 cons

Wstack_FC WBOP_EL þQrcvr EL þQrcvr Wstack_EL þWBOP_EL

FC

Fig. 3 e Schematic of energy flow from power input into electrolysis (EL) mode (left edge) to power output from fuel cell (FC) mode (right edge).

installation of a cooling tower for a larger scale system), the energy consumption of the chiller was not considered in the present calculations.

were measured at steady-state conditions. Thus, all the results except for continuous operation (Fig. 4) represent the performance at steady-state operation.

Continuous operation of the URFC system

Results and discussion In this study, as the first step in evaluating the URFC performance, the operating parameters (istack, Vstack, Tstack, and PH2)

In real applications, URFC operation requires quick switching of the operation mode, while minimizing the degradation of the performance caused by the switching. Evaluation of the

Fig. 4 e Change over time of hydrogen pressure (PH2), stack temperature (Tstack), stack voltage (Vstack), and stack current density (istack) during continuous operation of URFC system from electrolysis (EL), switching, to fuel cell (FC) mode.

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 5 8 0 3 e5 8 1 5

long-term durability was not been a subject of this study. The longest continuous operation was about 3 days. The total operation time was over 100 h and the number of switching times was 50, during which no severe trouble or degradation was observed. Fig. 4 shows representative results of continuous operation for about 2.5 h including switching between EL and FC modes. In this case, the URFC was operated first at the EL mode for about 60 min, and then switched to FC mode for about 60 min, and then switched again to EL mode. In the first EL mode, the URFC was operated at the rated condition (Tstack ¼ 50  C, PH2 ¼ 1.0 MPa, and istack ¼ 1.0 A cm2) and all the operating parameters (istack, Vstack, Tstack, and PH2) were stable, when Tstack was regulated by the cooling water and PH2 by the back pressure valve. In the switching procedure, dry nitrogen (N2) was introduced into both sides of the stack, and remaining water and gases were discharged. The switching procedure took about 5e10 min. At the beginning of FC mode, Tstack remained around 50  C, which was rather low for FC operation, and thus there was a possibility of flooding occurring at high istack. To avoid such flooding, istack was increased step by step as shown in Fig. 4. Tstack increased with increasing istack and reached 70  C. When Tstack > 70  C, stack cooling with circulating water was activated for regulating Tstack. Small spikes observed in PH2 during FC mode were caused by the draining and recharging of H2 in the recirculation line. At the rated operation of FC (Tstack ¼ 70  C, and istack ¼ 0.5 A cm2) (110e130 min in Fig. 4), again the operating parameters were stable, the same as in the EL mode. Flow rate of produced H2 was monitored with a digital flow meter. Under the PH2 range in the EL operation in this study (0.1e1.0 MPa), no relationship was observed between PH2 and Faraday (current) efficiency, which was near 1 at istack ¼ 1.0 A cm2. As noted before (Section 2.2), the stoichiometric ratio during FC operation was always near 1.

Current-voltage characteristics Fig. 5 shows the measured current density (istack) e average cell voltage (Vcell_avg) characteristics of EL and FC under

Fig. 5 e Current density (istack) e average cell voltage (Vcell_avg) characteristics of EL and FC mode at various stack temperature (Tstack). PH2 during EL was 1.0 MPa.

5809

different Tstack, where Vcell_avg was obtained by dividing Vstack in the number of cells (i.e., 10). The heat source for the stack was the self-heating (joule heat) resulting from the overpotential during both reactions of EL and FC. Because the overpotential during EL operation was smaller than that during FC, Tstack during EL mode was limited to 50  C, whereas Tstack during FC mode reached 70  C. In the case of EL operation, PH2 was set at 1.0 MPa. As expected, the EL performance was improved at higher Tstack, because the overpotential of both activation and ohmic decreased as Tstack was increased. As for FC operation, a significant difference between two different Tstack is evident at istack > 0.2 A cm2. The rapid decrease in Vcell_avg (i.e., Vstack) at Tstack ¼ 60  C indicates “flooding”, that is, condensed liquid water in the cell/stack accumulated at the electrode and hindered the mass transport of reactive gases to the electrode surface. The dew point of air passing through the membrane humidifier was estimated at around 60  C under the constant flow rate of the air regardless of Tstack. Therefore, RH of the air in the cathode gas channel would be reached around 100% when Tstack ¼ 60  C.

Efficiency analysis As described in Section 2.3, three different efficiencies have been defined for evaluating the URFC system at different levels, namely, hstack, hsys, and h0 sys. Based on the operation data under stable conditions, these three efficiencies were calculated for both EL and FC modes at different Tstack as summarized in Table 3. Fig. 6 shows the composition of energy produced by EL mode at Tstack ¼ 40 and 50  C. In this figure, the sum total energy corresponds to the input energy (Wstack_EL þ WBOP_EL) (see Fig. 2), and WBOP_EL includes power loss at AC-DC conversion. Here, the ratio of DHH2_prod and total energy represents hsys_EL, and the ratio of (DHH2_prod þQrcvr_EL ) and total energy represents h0 sys _EL. As Tstack was increased, hstack_EL increased due to improved current e voltage (IeV) performance of the cells (i.e., reduction of overpotential). hsys_EL and h0 sys _EL were also improved with the increase in Tstack (Table 3). Fig. 7 shows the composition of produced energy by FC mode at Tstack ¼ 60 and 70  C. The sum total input energy was equal to DHH2_cons (Fig. 3). The maximum istack for stable operation was 0.4 A cm2 when Tstack ¼ 60  C, whereas it was 0.5 A cm2 when Tstack ¼ 70  C as shown in Fig. 5. Thus, the total produced (consumed) energy differed between Tstack ¼ 60  C and 70  C. The ratio of net energy output from FC (i.e., Wstack_FCWBOP_FC) and total energy corresponds to hsys_FC, and the ratio of ½ðWstack_FC  WBOP_FC Þ þ Qrcvr_FC  and total energy corresponds to h0 sys_FC. In general, due to smaller overpotential, hstack_FC should be higher at lower istack. However, hstack_FC at 60  C was lower than that at 70  C (Table 3). The dew point of humidified air might be around 60  C regardless of Tstack, and thus the cells suffered from flooding when Tstack ¼ 60  C even when istack ¼ 0.4 A cm2 hsys_FC at 60  C was also lower than that at 70  C, despite h0 sys _FC being comparable at both values of Tstack. As described in Section 2.3, WBOP_FC consists of PP1, PP3, and PBLW, and was constant regardless either Tstack or istack. As shown in Fig. 7, the WBOP_FC component in Wstack_FC was significantly high, namely, 37% at

5810

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 5 8 0 3 e5 8 1 5

Table 3 e Calculated efficiencies of each operation mode of URFC at various stack temperature. Electrolysis (EL) mode Fuel cell (FC) mode Operating condition Stack temperature Current density Stack power (DC) H2 pressure (abs) Efficiencies Stack efficiency

Tstack istack Wstack_EL/Wstack_FC PH2 Electrolysis hstack_EL Fuel cell hstack_FC System efficiency Electrolysis hsys_EL Fuel cell hsys_FC System efficiency including heat recovery Electrolysis h'sys_EL Fuel cell h'sys_FC

Fig. 6 e Energy output contributions in EL operation (1 h) of URFC system at two different Tstack, when Istack ¼ 250 A (istack ¼ 1.0 A cm¡2), PH2 ¼ 1.0 MPa. Tstack ¼ 70  C, and 53% at 60  C. In summary, reduction of BOP power is imperative for achieving higher hsys_FC. In order to achieve stable operations and higher performance of the system, Tstack at the rated operation was fixed at 50  C for EL and at 70  C for FC mode.

Comparison of efficiencies with discrete devices Our research collaborator, Savannah River National Laboratory (SRNL) in the U.S., installed a separate commercial PEM electrolyzer and a PEMFC [41]. The SRNL PEM electrolyzer (HOGEN® RE S40; Proton Onsite) had a stack of 20 cells in series, and the active area of the electrode was 93 cm2. Its nominal hydrogen production rate was 1.05 Nm3/h at 12.8 bar (1.28 MPa), which was nearly identical with the EL mode of the URFC evaluated here. The SRNL PEMFC (GenCore™ 5B48; Plug Power) had a stack of 63 cells in series whose active area was 280 cm2. Its nominal power output was 5 kW, which was about 10 times larger than that with the FC mode of the URFC on the basis of net power output. Because these were standard models of a PEM electrolyzer and PEMFC, both were suitable as reference standards of the URFC performance assessment in our current study. Here, we collected the operating data for long-term operation of the SRNL PEM electrolyzer and the PEMFC and

40  C 1.0 A/cm2 4.6 kW 1.0 MPa 0.747

50  C 1.0 A/cm2 4.5 kW 1.0 MPa 0.773

0.619

0.640

0.724

0.738

60  C 0.4 A/cm2 0.6 kW 0.15 MPa

70  C 0.5 A/cm2 0.8 kW 0.15 MPa

0.395

0.451

0.189

0.284

0.787

0.768

Fig. 7 e Energy output contributions in FC operation (1 h) of URFC system at Tstack ¼ 60  C when Istack ¼ 100 A (istack ¼ 0.4 A cm¡2), and at Tstack ¼ 70  C when Istack ¼ 125 A (istack ¼ 0.5 A cm¡2).

calculated the efficiencies of these two individual systems. The calculated efficiencies are listed in Table 4 together with representative efficiencies of the URFC at the rated operation. First, hstack of the URFC and the separate SRNL systems were compared for each operation mode. The hstack_EL of the URFC was higher than that of the SRNL PEM electrolyzer. The hstack_EL mainly depends on the IeV performance of the stack, and this difference in hstack_EL might be partially attributed to the difference in Tstack and istack between these two systems. In general, IeV performance of electrolysis is improved with higher Tstack and lower istack. Tstack of the SRNL PEM electrolyzer was lower and thus istack was higher than the EL mode of the URFC. In terms of istack, because the SRNL PEM electrolyzer was a fully commercial product, the hydrogen production rate might have priority over efficiency, and thus istack at rated operation was higher than that of the URFC. However, hstack_EL of the URFC at Tstack ¼ 40  C was still superior to that of the SRNL PEM electrolyzer (Table 3). This difference in hstack_EL might also be attributed to specifications of the MEA such as species and load amount of catalyst, membrane thickness, and properties of the gas diffusion layer. Further discussion on the MEA is impossible at present, however, because details of MEA specifications are not clear for either the URFC or the

0.579 0.589

0.768 0.738

0.640

0.182

0.284

0.465

0.269

0.370 0.465

0.172

0.404 0.234 0.451 0.348

50 C 1.0 A/cm2 4.4 kW 1.0 MPa 0.773 Efficiencies

System efficiency including heat recovery

System efficiency

Electrolysis Fuel cell Round trip Electrolysis Fuel cell Round trip Electrolysis Fuel cell Round trip

Tstack istack Wstack_EL/Wstack_FC PH2 hstack_EL hstack_FC hstack_RT hsys_EL hsys_FC hsys_RT h'sys_EL h'sys_FC h'sys_RT Stack temperature Current density Stack power (DC) H2 pressure (abs) Stack efficiency Operating condition

PEMFC

~55  C 0.5 A/cm2 5.5 kW 0.55 MPa ~35 C 1.5 A/cm2 5.8 kW 1.5 MPa 0.580 70 C 0.5 A/cm2 0.8 kW 0.15 MPa

  

PEM electrolyzer Fuel cell (FC) mode

URFC

Electrolysis (EL) mode

Table 4 e Comparison of efficiencies between present URFC system and separate PEM electrolyzer and PEMFC as reference standards.

Discrete system

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 5 8 0 3 e5 8 1 5

5811

SRNL PEM electrolyzer. In regard to FC operation, hstack_FC of the URFC at Tstack ¼ 70  C was slightly superior to that of the SRNL PEMFC at Tstack near 55  C (Table 4). The same as in electrolysis, IeV performance of a fuel cell is improved with higher Tstack. This slight difference in hstack_FC between the URFC and SRNL PEMFC might be attributed to the difference in Tstack. In fact, hstack_FC of the URFC at Tstack ¼ 60  C was inferior to that of the SRNL PEMFC, even though istack of the URFC was lower than that of the SRNL PEMFC. hstack_RT of the URFC was significantly higher than both the SRNL PEM electrolyzer and PEMFC. In summary, the above comparison of stack efficiencies reveals that the stack performance of both EL and FC modes of the URFC were definitely comparable to those of the respective specialized devices for EL and FC. Secondly, hsys of the URFC was compared with that of the separate SRNL PEM electrolyzer and PEMFC systems. In the analysis of hsys, we need to examine the system configuration and data measurement point in detail. Because an air-cooling system was applied to both SRNL PEM electrolyzer and PEMFC, they could be operated without an outer cold-heat source and did not have a heat recovery system. In contrast, installation of an outer cold-heat source (i.e., cooling tower) has been assumed in the present URFC system. In particular, cooling of the bypass line (upstream from the water purifier) was essential for EL mode. At the same time, the water cooling system was applied for the stack, and heat recovery with water was easily achieved by the URFC. The measurement points for operating data differed between our URFC and the SRNL discrete systems. In our URFC system, we could measure DC input to EL operation, DC output from FC operation, and individual power consumption of P1, P2, P3, and BLW. Because the conversion efficiency of the AC-DC converter (rectifier) used for EL operation was identified (0.85), we could estimate the AC power input for EL mode of the URFC, and use it in the calculation of hsys_EL. As for the SRNL PEM electrolyzer, AC input to the system and DC input to the stack could be measured individually. The difference between AC input and DC input corresponds to the sum of the power loss of AC-DC converter and the energy consumption of BOP (pumps and fans), although the power consumption of each component of BOP was not monitored individually. In the case of the SRNL PEMFC system, the stack was cooled by circulating water (or refrigerant) that was subsequently cooled by an air-fan in the cabinet. For this system, DC output from the stack and DC output from the entire system could be measured separately, although power consumption of each component of BOP could not be measured individually. The difference between DC output from the stack and that from the system was power used for the BOP components in the system (pumps, blower, and fan). Heat removed from the stack by cooling water could be calculated using the flow rate and the temperature difference between the inlet and outlet of the cooling water, and was evaluated as recovered heat (Qrcvr_FC ) in the present comparison. For the sake of comparison between each operation mode of URFC and SRNL PEM electrolyzer and PEMFC as accurately as possible, as described in Section 2.3, hsys_EL was considered the power loss of the AC-DC converter for the both cases of EL mode of the URFC and SRNL PEM electrolyzer, whereas the power conversion from DC to AC was not considered in hsys_FC.

5812

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 5 8 0 3 e5 8 1 5

The same as hstack_EL, hsys_EL of URFC was higher than that of the SRNL PEM electrolyzer. To examine hsys_EL more comprehensively, the configuration and operation of the system must be considered. The stack of the SRNL PEM electrolyzer was mainly cooled by circulating electrolysis water, and the flow rate of the water was much higher than the amount needed for electrolysis. The pipeline of circulating electrolysis water was cooled by an air-fan installed in the cabinet. In addition, a hydrogen drier was built in the SRNL PEM electrolyzer system and might have a high energy consumption [42]. In the evaluation of hsys_EL, WBOP_EL of the SRNL PEM electrolyzer included the power consumption of all BOP components (power pump, air-fan, and drier). In the analysis of hsys_EL of the URFC system, as described in Section 2.3, the power consumption of the chiller was not included, although that of P1 and P2 was included. Thus, hsys_EL of the URFC might be somehow overestimated compared with that of the SRNL PEM electrolyzer. The hsys_FC of the URFC was inferior to that of the SRNL PEMFC, even though hstack_FC was superior. This lower hsys_FC must be due to the higher BOP energy consumption in the URFC system than in the SRNL PEMFC system. (Table 5 lists the measured data of power consumption of pumps P1, P2, and P3 at the rated operation, and Table 6 summarizes the relation between power consumption and boosted flow rate of the air-blower (BLW)). The power consumption of BLW (PBLW) was much higher than that of the pump, and the portion of PBLW  tFC in Wstack_FC reached about 25% at the rated flow rate (48 L/min). Therefore, reducing PBLW is critical to improve the hsys_FC of the URFC to be equivalent to that of the SRNL PEMFC. In this analysis, hsys_RT corresponds to the RT efficiency of energy from the total input into EL to the net output from FC via H2 (see Fig. 3). Due to contribution from the higher hsys_EL, hsys_RT of the URFC (0.182; Table 4) was slightly higher than that of the discrete system (0.172). Finally, h0 sys of the URFC was compared with that of SRNL PEM and SRNL PEMFC systems. By taking into account the recovered heat during EL operation (Qrcvr_EL ) into hsys_EL, h0 sys _EL of the URFC (0.738) was improved by about 10% (compared with hsys_EL ¼ 0.640). However, h0 sys _EL of the SRNL PEM electrolyzer (0.465) did not change from hsys _EL (0.465), because that electrolyzer system did not have a heat recovery system. In the case of the SRNL PEMFC, as described above, heat removed from the stack by cooling water was evaluated as recovered heat (Qrcvr_FC ), the same as the URFC. h0 sys _FC was thus improved from hsys _FC for the FC mode of the URFC (0.768 and 0.284, respectively) and the SRNL PEMFC (0.579 and 0.370, respectively). Because Qrcvr_FC of the SRNL PEMFC was lower

Table 5 e Power consumption of each circulation pump in URFC system. Notation

PP1 PP2 PP3

Component

Power consumption at rated operation [W]

Circulation pump for cooling water Circulation pump for electrolysis water Circulation pump for H2

90 49 8

Table 6 e Relationship between flow rate and power consumption of air-blower (BLW). Flow rate of air-blower (BLW) FBLW [L/min]

Power consumption PBLW [W]

40 48 60 70

154 196 266 301

than that of the URFC, h0 sys _FC of the SRNL PEMFC (0.579) was lower than that of the URFC (0.768). The reason for this low Qrcvr_FC of the SRNL PEMFC is not clear at present, though air flow caused by the fan in the cabinet might have increased the heat loss (Qloss_FC ). Consequently, a significant difference in h0 sys _RT exists between the URFC system (0.589) and the SRNL PEM electrolyzer and SRNL PEMFC (0.269). On the other hand, we need to consider the quality of thermal energy (i.e., exergy) depending on the heat source temperature. The temperature of Qrcvr_EL was 50  C at the rated operation, which is relatively low as a heat source temperature for CHP applications. When the utilization of Qrcvr_EL was omitted from the system, h0 sys _RT was slightly decreased to 0.492 (data not shown). To utilize the thermal energy recovered from the system, Qrcvr_FC would be rather more important than Qrcvr_EL . Nevertheless, the temperature range of the recovered heat (Qrcvr_EL or Qrcvr_FC ) was relatively low as 50e70  C. Considering the previous study [43] and commercialized residential CHP systems [25,26], promising use of this low-grade heat is a radiant floor heating or hot water supply for kitchen or bathroom. Further examination of hsys suggests that power reduction and optimization of BOP are essential for improving hsys _EL and hsys _FC, because further improvement in hstack_EL and hstack _FC is expected to be minimal. Nevertheless, hsys_RT might reach 0.25 to 0.30 at the most, even considering expected improvement in hsys _EL and hsys _FC. The potential of heat recovery was identified quantitatively by evaluating h0 sys. In summary, recovery and utilization of thermal energy are essential for improving not only efficiency but also utility value of the URFC system.

Effect of BOP power on FC performance As discussed in section Section 3.2, WBOP_FC of the URFC system was significantly high, and thus its reduction would be essential for improving hsys_FC. On the other hand, compared with the performance of EL operation, the performance of FC operation should be sensitive to operating conditions, such as Tstack, flow rate and relative humidity (RH) of introduced gases. Here, the effect of PBLW and PP3 on FC performance was investigated experimentally. At rated operation of the FC mode in the URFC system, FBLW was fixed at 48 L/min, and the PBLW at this flow rate was 196 W, which is about 25% of the total FC output (0.795 kW). Reduction of PBLW is therefore critical for improving hsys_FC. In FC operation of the present URFC system, FBLW could be increased from 40 to 70 L/min. Table 6 summarizes the relationship between FBLW and PBLW. Fig. 8 shows the hstack_FC and hsys_FC calculated by combining measured data of FC performance at

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 5 8 0 3 e5 8 1 5

different FBLW and corresponding PBLW. When Tstack ¼ 70  C, the FC performance was severely degraded at FBLW ¼ 70 L/min, and the operating data could not be measured. This suggests that the membrane resistance had increased significantly due to severe drying. In contrast, when Tstack ¼ 60  C, the FC performance degraded at FBLW ¼ 40 L/min, possibly due to severe flooding in the stack. As noted above, the dew point of introduced air through the membrane humidifier was assumed to be about 60  C at the rated flow rate (48 L/min) and might be changed depending on FBLW, that is, the dew point would decrease with the increase in FBLW. However, it is not clear the relation between the dew point and FBLW in detail, because we could not obtain any data of dew point or cell resistance. When Tstack ¼ 70  C, as shown in Fig. 8(a), hstack_FC was relatively independent of FBLW. However hsys_FC, which includes WBOP_FC, decreased with increasing FBLW (i.e., PBLW). When Tstack ¼ 60  C, as shown in Fig. 8(b), although hsys_FC remained

Fig. 8 e Dependency of stack efficiency (hstack_FC) and system efficiency (hsys_FC) on flow rate of air supplied from the air-blower (FBLW) at Tstack ¼ 70  C (a) and 60  C (b), when Istack ¼ 100 A (istack ¼ 0.4 A cm¡2).

5813

relatively unchanged, hstack_FC was improved with the increase in FBLW, which corresponded to the improvement in IeV performance. This suggests that the improvement in stack performance and the increase in PBLW actually offset each other. In Japan, residential CHP systems under the standard name of “Ene-Farm” were launched onto the market in May 2009, just after its demonstration project [24]. Although there are two options for fuel cell type (PEMFC and SOFC), most commercial products have been PEMFC based type. The EneFarm system consists of a PEMFC (0.7 kW power capacity), gas reformer (fueled by natural gas or propane gas), hot water tank (200 L), and BOP. In the period of the demonstration project (2005e2009), based on the support by the government, considerable effort went into the optimization and costreduction of BOP components, such as pumps, air-blower, sensors, valves, rectifier, and heat exchanger. Several manufacturers in Japan developed efficient air-blowers that had been specially designed for PEMFCs in Ene-Farm applications. Because the power capacity of a PEMFC in the Ene-Farm (0.7 kW) is similar to that of the FC mode of our current URFC system (0.5 kW), the nominal flow rate of the air-blower for the Ene-Farm is 50 L/min, which is comparable to that of our URFC system (48 L/min). However, the energy consumption of the air-blower in the Ene-Farm is less than 30 W at the rated operation [44], which is significantly lower than that of the air-blower in our URFC system (196 W, Table 6). The boost pressure of the air-blower in the Ene-Farm ranges from 3 to 15 kPa and is 12 kPa at the rated operation. Because the pressure drop in the present URFC stack was 15 kPa and the total pressure drop including pipelines reached about 30 kPa, the air-blower in the Ene-Farm could not be applied to this URFC. Nevertheless, assuming that the air-blower of the EneFarm could be applied to our URFC system, hsys_FC could be improved to 0.378 (from 0.284), which is comparable with the SRNL PEMFC system. In conclusion, optimization of the airblower is critical for improvement of hsys_FC. Next, we focused on the effect of flow rate of hydrogen recirculated by P3 pump (FH2) on the FC performance. FH2 could be changed in the range from 15 to 25 L/min, and fixed at 25 L/min at the rated operation. Fig. 9 shows the cell voltage variation during FC operation at FH2 ¼ 15 and 25 L/min, when Istack ¼ 125 A (istack ¼ 0.5 A cm2). At FH2 ¼ 15 L/min, the voltage of every cell was lower than that at FH2 ¼ 25 L/min, and the voltage of several cells (No.3 in particular) was significantly degraded. Also, hstack_FC ¼ 0.451 at FH2 ¼ 25 L/min, and degraded to 0.408 at 15 L/min. Note that regardless of FH2, the total flow rate of H2 supplied to the stack was nearly equal to the hydrogen amount needed for the reaction (stoichiometric ratio was near 1). Nevertheless, hydrogen distribution might be insufficient when FH2 ¼ 15 L/min. Considering that hstack_FC ¼ 0.447 at FH2 ¼ 20 L/min, which was close to the hstack_FC value of 25 L/min (0.451), FH2 at rated operation (25 L/ min) was sufficient for hydrogen distribution to each cell in the stack of our URFC. In contrast, because the PP3 (Table 5) was very low compared with the total FC output (Table 4) and the change in PP3 could not be detected with respect to the change in FH2, the effect of PP3 on hsys_FC was negligible. It is important to evaluate the efficiencies under off-design (partial load) conditions. However, efficiency analysis at off-

5814

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 5 8 0 3 e5 8 1 5

references

Fig. 9 e Cell voltage during FC operation of URFC system at a flow rate of recirculating H2 (FH2) of 25 and 15 L/min, when Istack ¼ 125 A (istack ¼ 0.5 A cm¡2).

design conditions could not be executed because enough data was not obtained at present.

Conclusions A pilot-scale URFC system was installed in our lab (AIST) and tested comprehensively. This URFC system was composed of a cell/stack and BOP that can adapted for both electrolysis (EL) and fuel cell (FC) operation modes. Based on measured data, the performance of the URFC system was evaluated at the stack level and system level, that is, stack efficiency and system efficiency were evaluated separately for each operation mode. Furthermore, the calculated efficiencies of this URFC system were compared with those of separate PEM electrolyzer and PEMFC as respective reference standards for EL and FC. Results showed that the stack performances of both the EL and FC modes of the URFC were comparable to those of the PEM electrolyzer and PEMFC, respectively. Analysis of the system efficiency considering BOP power consumption revealed that the performance of EL operation of the URFC system was comparable to that of the PEM electrolyzer, whereas the efficiency of the FC mode needs significant improvement to become comparable to that of the PEMFC. Analysis of the system efficiency including heat recovery from the URFC system quantified the potential of heat recovery. In conclusion, recovery and utilization of thermal energy are essential for improving not only the efficiency but also the utility value of the URFC system.

Acknowledgment The authors gratefully acknowledge financial support from the Ministry of Economy, Trade and Industry (METI) of Japan under the Japan-U.S. Cooperation Project for Research and Standardization of Clean Energy Technologies.

[1] Zakeri B, Syri S. Electrical energy storage system: a comparative life cycle cost analysis. Renew Sustain Energy Rev 2015;42:569e96. [2] Gahleitner G. Hydrogen from renewable electricity: an international review of power-to-gas pilot plants for stationary applications. Int J Hydrogen Energy 2013;38:2039e61. [3] Santarelli M, Calı` M, Macagno S. Design and analysis of stand-alone hydrogen energy systems with different renewable sources. Int J Hydrogen Energy 2004;29:1571e86. [4] Barbir F. PEM electrolysis for production of hydrogen from renewable energy sources. Sol Energy 2005;78:661e9. [5] Zoulias EI, Glockner R, Lymberopoulos N, Tsoutsos T, Vosseler I, Gavalda O, et al. Integration of hydrogen energy technologies in stand-alone power systems analysis of the current potential for applications. Renew Sustain Energy Rev 2006;10:432e62. [6] Deshmukh SS, Boehm RF. Review of modeling details related to renewably powered hydrogen systems. Renew Sustain Energy Rev 2008;12:2301e30. [7] Yilanci A, Dincer I, Ozturk HK. A review on solar-hydrogen/ fuel cell hybrid energy systems for stationary applications. Prog Energy Combust Sci 2009;35:231e44. [8] LaConti AB, Swette L. Special applications using PEMtechnology. In: Vielstich W, Lamm A, Gasteiger HA, editors. Handbook of fuel cells, 4. Chichester: John Wiley & Sons; 2003. Ch.55. [9] Pettersson J, Ramsey B, Harrison D. A review of the latest developments in electrodes for unitised regenerative polymer electrolyte fuel cells. J Power Sources 2006;157:28e34. [10] Dihrab SS, Sopian K, Alghoul MA, Sulaiman MY. Review of the membrane and bipolar plates materials for conventional and unitized regenerative fuel cells. Renew Sustain Energy Rev 2009;13:1663e8. [11] Gabbasa M, Sopian K, Fudholi A, Asim N. A review of unitized regenerative fuel cell stack: material, design and research achievements. Int J Hydrogen Energy 2014;39:17765e78. [12] Baldwin R, Pham M, Leonida A, McElroy J, Nalette T. Hydrogen-oxygen proton-exchange membrane fuel cells and electrolyzers. J Power Sources 1990;29:399e412. [13] Barbir F, Molter T, Dalton L. Efficiency and weight trade-off analysis of regenerative fuel cells as energy storage for aerospace applications. Int J Hydrogen Energy 2005;30:351e7. [14] Sone YA. 100-W class regenerative fuel cell system for lunar and planetary missions. J Power Sources 2011;196:9076e80. [15] Maclay JD, Brouwer J, Samuelsen GS. Dynamic analysis of regenerative fuel cell power for potential use in renewable residential applications. Int J Hydrogen Energy 2006;31:994e1009. [16] Kato A, Masuda M, Takahashi A, Ioroi T, Yamaki M, Ito H. Durability investigation of a MEM-type unitized reversible cell. ECS Trans 2009;25(1):1271e8. [17] Gigliucci G, Petruzzi L, Cerelli E, Carzisi A, La Menfola A. Demonstration of a residential CHP system based on PEM fuel cells. J Power Sources 2004;131:62e8. [18] Peacock AD, Newborough M. Impact of micro-CHP systems on domestic sector CO2 emissions. Appl Therm Eng 2005;25:2653e76. [19] Paepe MD, D'Herdt P, Mertens D. Micro-CHP systems for residential applications. Energy Convers Manag 2006;47:3435e46. [20] Ren H, Gao W, Ruan Y. Optimal sizing for residential CHP system. Appl Therm Eng 2008;28:514e23.

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 1 ( 2 0 1 6 ) 5 8 0 3 e5 8 1 5

[21] Thomas B. Benchmark testing of micro-CHP units. Appl Therm Eng 2008;28:2049e54. [22] Shaneb OA, Coates G, Taylor PC. Sizing of residential mCHP systems. Energy Build 2011;43:1991e2001. [23] Barelli L, Bidini G, Gallorini F, Ottaviano A. Dynamic analysis of PEMFC-based CHP systems for domestic application. Appl Energy 2012;91:13e28. [24] New Energy Foundation (NEF). Progress report on the largescale stationary fuel cell project in Japan. 2009. http://ieahia. org/pdfs/Case-Studies/Task18_Japan_Residential_FC_Report. aspx [page retrieved on Nov, 2015]. [25] http://www.toshiba.co.jp/product/fc/english/index.htm [Toshiba, page retrieved on Nov, 2015]. [26] http://panasonic.co.jp/ap/FC/en_index.html [Panasonic, page retrieved on Nov, 2015]. € nig P, Weber A, Lewald N, Aicher T, Jo € rissen L, Ivers[27] Ko e E, et al. Testing and model-aided analysis of a 2kWel Tiffe PEMFC CHP-system. J Power Sources 2005;145:327e35. [28] Torchio MF, Santarelli MG, Nicali A. Experimental analysis of the CHP performance of a PEMFC stack by a 24 factorial design. J Power Sources 2005;149:33e43. [29] Santarelli MG, Torchio MF, Calı` M, Giaretto V. Experimental analysis of cathode flow stoichiometry on the electrical performance of a PEMFC stack. Int J Hydrogen Energy 2007;32:710e6. [30] Shabani B, Andrews J. Experimental investigation of a PEM fuel cell to supply both heat and power in a solarhydrogen RAPS system. Int J Hydrogen Energy 2011;36:5442e52. [31] Migliardini F, Capasso C, Corbo P. Optimization of hydrogen feeding procedure in PEM fuel cell system for transportation. Int J Hydrogen Energy 2014;39:21746e52. [32] Ioroi T, Yasuda K, Siroma Z, Fujiwara N, Miyazaki Y. Thin film Electrocatalyst for unitized regenerative polymer electrolyte fuel cells. J Power Sources 2002;112:583e587805. [33] Yim SD, Park GG, Sohn YJ, Lee WY, Yoon YG, Yang TH, et al. Optimization of PtIr electrocatalyst for PEM URFC. Int J Hydrogen Energy 2005;30:1345e50.

5815

[34] Song S, Zhang H, Ma X, Shao ZG, Zhang Y, Yi B. Bifunctional oxygen electrode with corrosion-resistive gas diffusion layer for unitized regenerative fuel cell. Electrochem Commun 2006;8:399e405. [35] Jung HY, Huang SY, Ganesan P, Popov BN. Performance of gold-coated titanium bipolar plates in unitized regenerative fuel cell operation. J Power Sources 2009;194:972e5. [36] Doddathimmaiah A, Andrews J. Theory, modeling and performance measurement of unitized regenerative fuel cells. Int J Hydrogen Energy 2009;34:8157e70. [37] Grigoriev SA, Millet P, Dzhus KA, Middleton H, Saetre TO, Fateev VN. Design and characterization of bi-functional electrocatalytic layers for application on PEM unitized regenerative fuel cells. Int J Hydrogen Energy 2010;35:5070e6. [38] Hwang CY, Ishida M, Ito H, Maeda M, Nakano A, Hasegawa Y, et al. Influence of properties of gas diffusion layers on the performance of polymer electrolyte-based unitized reversible fuel cells. Int J Hydrogen Energy 2011;36:1740e53. [39] Hwang CY, Ishida M, Ito H, Maeda M, Nakano A, Kato A, et al. Effect of titanium powder loading in gas diffusion layer of a polymer electrolyte unitized reversible fuel cell cells. J Power Sources 2012;202:108e13. [40] Ito H, Abe K, Ishida M, Hwang CY, Nakano A. Effect of through-plane polytetrafluoroethylene distribution in a gas diffusion layer on a polymer electrolyte unitized reversible fuel cell. Int J Hydrogen Energy 2015;40:16556e65. [41] Motyka T. Savannah river national laboratory regenerative fuel cell project. 2008. DOE report SRNL-STI-2008e00388. [42] Harrison KW, Martin GD, Ramsden TG, Kramer WE, Novachek FJ. The wind-to-hydrogen project: operational experience, performance testing, and system integration, Technical report of national renewable energy laboratory. 2009. NREL/TP-550-44082. [43] Gandiglio M, Lanzini A, Santarelli M, Leone P. Design and optimization of a proton exchange membrane fuel cell CHP system for residential use. Energy Build 2014;69:381e93. [44] http://www.hiblow-usa.com/files/pumps/15.pdf [HiBlow USA, page retrieved on Nov, 2015].