Experimental integration of a foam-based floating photovoltaic (floatovoltaic) system with an anion exchange membrane electrolyzer for 5 kW-Scale green hydrogen production
1. Introduction
Hydrogen (H2) is gaining recognition as a key component in the ongoing energy transition, particularly in sectors where direct process electrification is challenging, such as energy-intensive industries, transportation, and long-term energy storage [1,2]. Unlike conventional batteries, hydrogen serves multiple functions beyond electricity storage, including use as fuel [3], a feedstock for fertilizer production [4], a potential alternative to electric vehicle (EV) batteries [5], and a critical input in various chemical industries [6]. Its versatility positions hydrogen as a cornerstone for a sustainable and decarbonized energy ecosystem [2]. The urgency to curb fossil fuel consumption has escalated due to rising carbon dioxide (CO2) emissions. The resultant climate destabilization as the last decade (2015–2024) has caused the highest temperatures ever recorded in post-industrial revolution era, with 2024 marking the hottest year and surpassing the 1.5 °C upper limit set by the Paris Agreement in 2015 [7,8].
Among renewable energy sources critical to mitigating anthropogenic climate change, solar photovoltaic (PV) technology has emerged as the most cost-effective and rapidly expanding option in recent years [9,10]. Despite PV’s effectiveness in emission reduction at low penetration levels, its inherent intermittency poses challenges as its share of penetration increases, requiring reliable storage solutions [11,12]. Moreover, the mismatch between PV output and real-time load demands underscores the need for efficient energy storage systems [13,14]. In this context, integrating PV with H2 production via electrolysis has demonstrated promising results in improving efficiency and reducing energy dependence. Recent works suggest that by strategically scheduling electrolyzers to coincide with surplus PV energy, up to 75 % of the electrolyzer’s energy demand can be met directly by solar power, thereby decreasing grid dependency and boosting system efficiency [15].
Electrolysis remains the cornerstone of green H2 production, with ongoing efforts focused on improving efficiency, reducing costs, and enhancing durability [16]. The three predominant electrolysis technologies comprise of alkaline electrolysis (AEL), proton exchange membrane (PEM) electrolysis, and solid oxide electrolysis (SOE); each offering unique advantages and challenges. AEL is an established technology known for low capital costs and long service life, but its slow dynamic response and reliance on highly concentrated electrolytes limit its applicability, especially with intermittent energy sources such as PV systems [16,17]. In contrast, PEM electrolyzers are more efficient and responsive, making them suitable for dynamic coupling with renewable sources [18]. PEM electrolysis, however, requires expensive precious metal catalysts (e.g., platinum and iridium) and operates in highly acidic environments, raising concerns about cost and durability [17,19]. SOE electrolyzers, which operate at high temperatures, offer superior efficiency by applying thermal energy to reduce electrical consumption, yet material degradation and long-term stability issues hinder their widespread deployment [17,19].
Anion exchange membrane (AEM) electrolyzers have gathered significant attention as a next-generation H2 production method due to their ability to operate with non-precious metal catalysts while maintaining high efficiency up to 80 % [19,20] and relatively high current densities of 0.2–2 A/cm2 [21]. Recent studies have shown the potential to surpass these benchmarks significantly. For example, a system using ethylene tetrafluoroethylene (ETFE)-grafted ionomers achieved a remarkable experimental current density of 11.2 A/cm2 at 2 V, highlighting tremendous potential [22]. Similarly, a tungsten disulfide (WS2) superstructure cathode enabled continuous operation at 1 A/cm2 for over 1000 h with minimal degradation, demonstrating promising durability [23]. Electrodes used in AEM systems can also be manufactured from low-cost, recycled or earth-abundant materials, offering both economic and sustainability benefits for large-scale H2 production [24]. A notable example includes a nickel-copper phosphide/nickel sulfide catalyst, which enabled stabled operation at 230 mA/cm2 with 95–97 % performance retention over 300 h, indicating the effectiveness of non-precious metal systems [25].
There remain, however, some significant technological hurdles that must be addressed in AEM electrolyzers. Current AEM materials degrade over time due to carbonation and hydroxide attack, leading to performance losses [26]. A comprehensive review identified degradation mechanisms occurring at the material and interfacial levels, emphasizing how temperature, clamping pressure, and electrolyte composition contribute to stability issues [27]. AEM systems also exhibit lower operational current densities and shorter lifetimes compared to PEM electrolyzers, limiting their use in high-demand scenarios [28]. Electrochemical impedance spectroscopy has confirmed that losses in catalyst activity and ionic transport contribute significantly to this performance gap [29]. Nevertheless, research into advanced membrane materials and ionomers is producing encouraging results. For instance, poly (biphenyl alkylene)-based AEMs have enabled stable operation for over 3500 h at 1 A/cm2, underscoring AEM technology’s emerging viability as a bridge between traditional alkaline and PEM electrolyzers [28,30].
Coupling PV with H2 production via electrolysis represents a promising pathway for large-scale renewable energy storage and application. When combined with an electrolyzer, PV-generated electricity can be used to produce H2, thereby efficiently storing energy that would otherwise be curtailed [31]. An emerging PV technology, floating PV (FPV), addresses land-conflict associated with conventional ground-based PV systems by installing PV on water bodies [32]. FPV offers several advantages, including increased energy production due to water-based PV modules cooling [33], reduced land requirements [32], and water conservation through evaporation prevention [34]. The same body of water can provide a water source for PV panel cooling and H2 production, optimizing resource use and enhancing the system synergy. These advantages make the FPV-AEM electrolyzer integration a compelling approach to green H2 production. Previous studies have explored FPV’s potential in H2 production, though most have focused on simulations [[35], [36], [37]], with limited experimental work on PV-AEM electrolysis systems at the kilowatt scale [15].
Despite advances in membrane chemistry and small-scale PV-electrolyzer prototypes, no prior work has experimentally validated a foam-based FPV system coupled with a kilowatt-scale AEM electrolyzer or quantified surplus solar energy utilization for off-grid hydrogen production. The present work fills this gap by evaluating the practical feasibility and performance of a 7 kW foam-based flexible FPV system integrated with an AEM electrolyzer, focusing on system efficiency and operational characteristics. The experimental methodology assesses stack-level and system-level energy conversion efficiency, analyzes hydrogen and oxygen purity, examines power-conversion losses and overall electrical efficiency, and investigates hydrogen production scheduling to optimize operation under surplus PV generation. The contributions of the present work include: (i) the first experimental demonstration at kilowatt scale of a foam-based FPV system coupled to an AEM electrolyzer, (ii) simulation-based quantification of hydrogen production driven by surplus photovoltaic energy in off-grid operation; and (iii) detailed analysis of system‐level energy losses with recommendations for direct DC coupling to minimize conversion stage losses.
The paper is structured as follows: Section 2 outlines the methodology, materials, system components, and experimental setup. Section 3 presents key performance results, including efficiency, hydrogen purity, and the role of AEM stacks as energy storage for PV systems. Section 4 discusses the broader implications of FPV-AEM integration, compares the findings to previous studies, and identifies current limitations and areas for future research to improve system performance in real-world deployment.
2. Material and methods
2.1. System description
The system under investigation is depicted in Fig. 1 and comprised of an AEM electrolyzer, an FPV system, a power conversion system integrated with energy storage, and data acquisition tools.
Fig. 1. FPV-AEM stack system components. (a) Complete system setup: 1-Inverters and power management box; 2-Floating solar panels system; 3-DC power supplies; 4-Electrolyzer setup; (b) AEM electrolyzer setup: 5-Water level sensor; 6-Oxygen tank; 7-Water flowmeter; 8-Heat exchanger; 9-Temperature probes; 10-AEM electrolyzer stack; 11-Mixing pumps; 12-Hydrogen tank; 13-Circulation pump. (c) Zoom on the AEM electrolyzer stack. (d) Schematic of the system setup with the different power stages.
2.1.1. AEM electrolyzer
The AEM electrolyzer was manufactured and supplied by Cipher Neutron (Toronto, Canada). The electrolyzer stack consists of 27 cells, with maximum operating voltage of 2.2 V per cell. The membrane has an active area of 100 cm2, supporting current densities ranging from 0.7 A/cm2 to 2 A/cm2, with an operational temperature range between 5 °C and 70 °C. A 1 M potassium hydroxide (KOH) solution serves as the electrolyte.
The AEM stack is incorporated into a hydrogen production system flow designed for stable and efficient hydrogen generation. The system is mounted on a dedicated platform and comprises two pressurized electrolyte tanks, four DC electrolyte pumps, two heat exchangers, two liquid level sensors, and two flow meters as shown in Fig. 1. The pressurized tanks have a dual function: storing the electrolyte solution and collecting the hydrogen and oxygen gases generated during the reaction. Each tank is equipped with a pressure release valve to ensure safe operation under varying pressure conditions. Two of the DC pumps are dedicated to mixing the electrolyte before the reaction begins, ensuring solution homogeneity, while the remaining two pumps circulate the electrolyte through the stack. Heat exchangers regulate the electrolyte temperature by transferring thermal energy via hot water circulation around the electrolyte. The heat exchangers can be activated or left idle. Level sensors monitor electrolyte levels in the tanks, and flow meters measure electrolyte flow through the system. To prevent corrosion, the platform, tanks, and connecting tubes are constructed from stainless steel 316 L (SS316L).
2.1.2. FPV system, power conversion unit, and data acquisition
The AEM electrolyzer was powered by a 7-kW FPV system installed on a pond in Ilderton, ON, covering only 2.8 % of the 1470 m2 total pond area (42 m × 35 m). The FPV system is equipped with an experimental foam-based design that leverages the lightweight nature of flexible monocrystalline PV modules to reduce the structural complexity typically associated with low-density polyethylene (LDPE) platforms in conventional rigid module-based FPV systems [38]. This approach enhances buoyancy while minimizing material requirements and assembly effort. In this experiment, the PV modules were sourced from Renogy [39], each with a rated power output of 175 W. A total of 40 modules were deployed to achieve the system’s 7 kW DC capacity. The assembly process was conducted on-site, following a previously established design framework in the literature, where the foam structure is adhered to the PV modules using a marine-grade sealant to ensure durability and water resistance [38,40]. The FPV system was configured into four arrays, each consisting of ten PV modules. Within each array, five modules were connected in series to form a string, and two such strings were connected in parallel.
The FPV system powering the AEM electrolyzer is supported by three single phase Victron Energy (Netherlands) inverters [41], each rated at 2.5 kW (3 kVA). These inverters are configured in a Y-connected three-phase arrangement, to provide a 3-phase output (230 V Line-Neutral/400 V Line-Line) with a total capacity of 7.5 kW (9kVA), ensuring compatibility with the electrolyzer’s power requirements. The system is integrated with four Victron Energy maximum power point tracking (MPPT) charge controllers [42], each assigned to one FPV array to optimize energy harvesting and maintain efficient power conversion. The inverters convert the generated DC power into AC, supplying a stable voltage for the electrolyzer’s operation. Additionally, a 10-kWh battery storage system, composed of two 5 kWh (48 V, 100Ah) batteries [43], is incorporated to stabilize power delivery, compensate for fluctuations in solar generation, and maintain continuous operation, as the system functions entirely off-grid.
The three-phase design of the PV inverter system was chosen based on the 10 kW AC-DC converter’s [44] operating input voltage and the supplier’s recommended optimal DC voltage (40–50 V, ∼5 kW) across the AEM electrolyzer. The DC power supply operates at a three-phase 380VAC ±15 % (line-to-line), 50 Hz input and provides a regulated 50 V DC output, with a maximum of 200 A AC current capacity. This converter ensures that the AC energy supplied by the inverters is efficiently transformed into the stable DC power required for electrolyzer operation.
Several sensors were integrated into the system to monitor and record data throughout its operation. The Cerbo GX [45], a Victron Energy monitoring device with advanced communication capabilities, was connected to the PV system to track DC power generation and AC power transmission from the inverters to the DC power supply. This DC power supply unit features a built-in display for real-time monitoring of operational voltage and current, while a digital multimeter was used to measure the electrolyzer stack’s operating voltage. Additionally, a digital temperature-controlled outlet [46] was used to monitor the electrolyzer’s temperature and regulate the heater’s operation. The temperature probes were attached, using insulated tape, to the exterior of the stainless-steel piping rather than immersed directly in the electrolyte. Environmental conditions, including solar irradiation (±1 %), ambient temperature (±0.2 °C), and wind speed (±3 %), were recorded using an on-site Lufft meteorological station (Lufft, Germany) [47], ensuring comprehensive data collection for system performance analysis.
Oxygen (O2) and hydrogen (H2) output flowrates were determined using two complementary measurement techniques. First, a gas displacement setup with a 3.43 L beaker was used where the time required to fill the beaker was recorded. Second, a Dwyer RMA-22-TMV rotameter (ITM Instruments, Canada) [48] was used with completely opened valves. Although the rotameter is designed for air, the measured data can be calibrated to accommodate other gases by using Charle’s law shown in equations (1), (2) [49].
(1)(2)
F is the gas flow read by the rotameter, c is the calibration factor determined by comparing measurements (3.25 for H2 and 1.25 for O2), and k is the temperature correction factor given by the ambient temperature and standard temperature and pressure (STP) value.
The oxygen and hydrogen were first passed sequentially through a condenser and a desiccant dryer to remove excess heat and moisture, thereby enabling accurate gas flow measurements. Gas composition was then determined using a Micro GC Varian CP-4900 (California, USA) [50]. For sampling and analysis, 1 L gas bags from Environmental Sampling Supply (California, USA) [51] were used. Three gas bags were collected for each gas, with samples taken from both the O2 and H2 streams. These samples were analyzed to obtain an average composition, ensuring more reliable and consistent results.
2.2. System operation and parameters
Two separate experiments were conducted outdoors on different days to assess the system’s performance. The first experiment was conducted on July 08, 2024, where the system operated without a heater, while the second experiment occurred on July 22, 2024, with the system connected to a controlled heater to accelerate the electrolyte heating process. During both experiments, the mixing pumps were activated for 5 min to mix the electrolytes between the two tanks. Following this, the flow pumps were turned on after the mixing pumps were switched off, and they remained active throughout the entire experiment. After an additional 5 min of running the flow pumps to establish consistent electrolyte flow, the power supply was activated to initiate the electrolysis reaction. Hydrogen and oxygen began to flow once sufficient pressure was built in the tanks, and the operational parameters were recorded for analysis. For both experiments, the DC power supply was configured in constant-current mode at 100 A (1 A/cm2) to maintain a fixed production rate and protect the AEM stack from overcurrent as its resistance decreased with temperature. In the non-heated run, absence of active cooling meant that unchecked temperature rises could exceed safe operating limits; consequently, the electrolyzer was powered off when the stack approached its critical temperature threshold.
The system parameters are classified into three categories: operational variables, environmental variables, and performance metrics. The operational variables include user-controlled parameters, such as the voltage and maximum current supplied to the electrolyzer, the type of power source used, the electrolyte temperature and flow rate, its type and concentration, and the number of cells in the stack. Detailed information on the input parameters and their corresponding values used is presented in Table 1. Environmental variables include solar irradiation, wind speed, and ambient temperature, all of which were recorded during the experiments. The performance metrics, derived from operational and environmental data, include H2 and O2 production rates, purity levels of the produced gases, electrical energy consumption and efficiency at distinct stages of the system, the energy content efficiency of the generated hydrogen, and the thermal performance of the system.
Table 1. Operational input parameters.
| Input Parameter Electrolyzer | Value | Unit |
|---|---|---|
| Total Stack Power | 5 | kW |
| PV Power | 7 | kW |
| Inverter Size | 7.5/9 | kW/kVA |
| Battery Size (51.2 V) | 2 × 200 | Ah |
| Number of Cells in the Stack | 27 | |
| Voltage per Cell Range | 0–1.86 | V/cell |
| Maximum Current Density | 2 | A/cm2 |
| Power Source Type | DC – AC – DC | |
| Electrolyte Type | 1 M KOH | |
| Electrolyte Temperature | 22–70 | °C |
| Electrolyte Flow Rate | 500 | L/min |
2.3. System performance model
2.3.1. Energy conversion performance model
Several energy parameters were investigated for the AEM-FPV coupled electrolyzer, including energy consumption at various power stages (PV system level, inverter level and DC power supply level), electrical energy conversion efficiency at these stages, energy content conversion efficiency, and the specific energy consumption (SEC) of H2 production. The energy consumption at each power stage (depicted in Fig. 1d) was calculated by multiplying the voltage by the operational current at that stage, where both voltage and current were measured by installed sensors. The electrical energy conversion efficiency, denoted as (%), represents the electrical losses occurring at each power stage. This efficiency is calculated by dividing the electricity output at a specific stage by the electricity input at the same stage, as shown in equation (3).
(3)
The energy content conversion efficiency, denoted as (%), in this context refers to the ratio of the raw energy that can be extracted from 1 kg of generated H2 to the energy consumed to produce 1 kg of H2, (kWh/kg), as shown in Equation (4). The raw energy content of 1 kg of H2 is considered as the higher heating value of H2, which is 142 MJ/kg or 39.44 kWh/kg [52]. This value is used to calculate the energy content conversion efficiency at each stage of the system.
(4)
The energy required to generate 1 kg of H2 or specific energy at each power stage is calculated by multiplying the power at that stage by the time required to generate 1 kg of H2, , as shown in equation (5). In contrast, the time to generate 1 kg of H2 is determined using the ideal gas law, as presented in equation (6). In this equation, (J·mol−1·K−1) represents the ideal gas constant, (K) is the temperature, (L/h) is the volumetric flow rate, (atm) is the pressure, and (kg/mol) is the molar mass of H2. This relationship calculates the time needed to generate a specific amount of H2, which is essential for determining energy consumption at each power stage.
(5)(6)
2.3.2. Thermal model of the electrolyzer
A simplified thermal model was developed to assess the thermal performance of the electrolyzer operating in an outdoor environment and to evaluate the effectiveness of a heater in a non-insulated electrolyzer. In this model, both the electrolyzer and the platform were treated as a rectangular plate exchanging thermal energy with the surrounding environment via convection and radiation. Additionally, the thermal energy absorbed by the system due to solar exposure was considered. The thermal properties of stainless steel 316 L (SS316L) [53] were applied in the model, and a quasi-steady-state analysis was conducted using the measured data collected during the experiments.
The thermal model was used to calculate the thermal efficiency (), as shown in equation (7). In this equation, (J) represents the energy consumed by the stack, while (J) refers to the energy lost by the system through heat exchange mechanisms. The energy loss, , is calculated through a net heat flow balance, as shown in equation (8).
(7)(8)
(J) represents the heat exchanged within the electrolyzer between two timesteps and is calculated using the specific heat and thermal properties of the electrolyte. (J) and (J) correspond to the heat exchanged between the electrolyzer and the environment via convection and radiation, respectively. Together, , , and account for the heat losses in the system. (J) represents the solar heat gained or absorbed by the system, as the electrolyzer was operated outdoors under direct sunlight. In the experiment where the electrolyzer was connected to a heating system, the energy from the heater was incorporated into , contributing to the overall heat energy input into the system.
2.4. Enhancing off-grid PV system performance in winter-dominant regions through hydrogen production
The main challenge of designing an off-grid solar PV system for an electrolyzer in winter-dominant regions such as Canada is the total size of the PV system due to seasonal variations in solar energy availability causing the peak load demand to coincide with low irradiation periods in winter [54]. Similarly, designing a fully off-grid residential system is challenging in this climate type due to the same reasons [55].
Winters are particularly harsh in Canada, with temperatures often dropping below 0 °C (which have a positive impact on PV output with temperature coefficient (Pmax) −0.34 %/0C [56]), however there are also significantly lower solar irradiance levels compared to summer, which has a much larger negative effect on PV performance than the temperature coefficient. During December and January, solar energy generation can be up to 2.5 times lower than in peak summer months at the same location, as shown in Fig. 2a. Meanwhile, a typical residential house with a heating, ventilation, and air conditioning (HVAC) system maintains a relatively stable energy demand throughout the year, often increasing in winter due to heating needs [57]. As shown in the load profile of an ideal residential building, the winter energy requirement is not significantly different from summer, making off-grid PV system optimization difficult. Due to the imbalance between lower energy production and steady demand, optimization software such as HOMER [58] and SAMA [59] often recommend oversized PV arrays and battery storage to ensure winter load coverage if the system is fully off-grid. A study in Nunavik Quebec highlights PV system challenges in cold regions due to seasonal energy mismatch. In winter, demand exceeds supply, while summer sees surplus energy. A 70 m2 PV array produced 13,679 kWh/year, but only 28 % was used directly [60]. This oversize PV-battery system results in higher installation costs and significant surplus energy generation during summer, which cannot be efficiently stored or used unless a grid connection is available or the energy can be stored using alternative methods.
Fig. 2. Methodology for sizing an electrolyzer for surplus PV energy utilization in winter-dominant regions: (a) Weather patterns and PV energy generation rate for 1 kW PV module in London, Ontario, (b) Load profile of a typical residential house and the surplus energy from an optimized off-grid PV system modeled using HOMER, and (c) Flowchart illustrating the integration and sizing of an electrolyzer for off-grid residential and commercial systems.
For an optimized off-grid PV designed to supply an ideal residential home in London, Ontario, with an annual energy demand of 9.12 MWh, a system configuration of 29.2 kW PV capacity and 85 kWh of battery storage is required to ensure reliability. Due to seasonal variations in solar availability, however, this system may produce up to 73.6 % surplus energy annually, as illustrated in Fig. 2b. This significant excess generation poses economic challenges for achieving full energy independence, as it leads to increased system costs and underutilized energy during peak production periods. A potential solution for addressing surplus energy in off-grid residential and commercial systems, is the integration of an electrolyzer, so that excess PV energy can be converted into H2 (and O2), which can be stored for future use. The electrolyzer operates only when surplus energy exceeds its minimum operational threshold, ensuring it runs efficiently within its optimal range.
Determining the appropriate electrolyzer size and operating conditions requires careful consideration of several factors, including the cost of the electrolyzer stack, available surplus energy, operational hours, and hydrogen production rate. Since this analysis does not focus on using H2 directly to power loads or fuel cells, conventional optimization tools like HOMER are not applicable. Instead, a sensitivity analysis of different electrolyzer stack sizes and varying minimum operating loads can help determine the most efficient configuration for surplus PV energy utilization. The process assessed electrolyzer sizes (0–30 kW) and minimum loading percentages (10 %–50 %), evaluating their impact on surplus energy use. The minimum loading percentage represents the fraction of electrolyzer power that must be available as surplus for the electrolyzer to turn on, preventing frequent switching under very low power conditions with low efficiency. This study examines how surplus energy changes with electrolyzer size and load thresholds, identifying the optimal size of an electrolyzer given a fully off-grid PV-battery system. Additionally, the electrolyzer’s capacity utilization factor (CUF); a parameter that quantifies how often is the electrolyzer powered on throughout the year; was analyzed to ensure efficient operation under real-world constraints.
The proposed system prioritizes economic feasibility by ensuring that the electrolyzer operates only when beneficial, without requiring expensive additional infrastructure. By diverting excess electricity toward hydrogen production, the system enhances overall off-grid PV system’s efficiency and increases economic viability. This strategy helps overcome a major limitation of off-grid solar systems, particularly in climates where winter conditions significantly reduce PV generation.
4. Discussion
4.1. Performance and feasibility of a PV-powered AEM hydrogen generation system
The experimental and simulation results demonstrate the feasibility of using solar PV to power an AEM electrolyzer for H2 production. The AEM stack achieved a stack-level energy conversion efficiency ranging from 73.3 % to 86.2 % based on the higher heating value (HHV) of H2. The corresponding specific energy consumption decreased from 53.81 kWh/kg to 45.77 kWh/kg. At the system level, the combined PV-AEM configuration exhibited an energy conversion efficiency of 48.17 %–61.90 %, with a specific energy consumption ranging from 82.87 kWh/kg to 63.72 kWh/kg.
The lower values of the H2 production efficiencies obtained in this study aligns with values reported in the literature. Previous studies have documented AEM stack efficiencies between 69.64 % and 79.26 % [16,28,63] (HHV basis), with specific energy consumption between 76.77 kWh/kg and 48.25 kWh/kg [16,[62], [63], [64], [65], [66], [67]]. System-level analyses have found HHV-based efficiencies between 58.26 % and 67.71 % [68], with specific energy consumption varying from 83.44 kWh/kg to 50.01 kWh/kg [62,65,67]. This study has arguably achieved one of the lowest reported specific energy values for an AEM electrolyzer, with a stack-level specific energy of 45.77 kWh/kg and a conversion efficiency of 86 %. These findings suggest that the studied AEM stack is among the most efficient currently available and could contribute toward achieving the International Renewable Energy Agency’s (IRENA) 2050 target of reducing AEM-specific energy consumption to below 42 kWh/kg [20]. It should be noted that the specific energy was obtained while conducting the experiment outdoors with wind speed between 0.7 and 2 m/s, ambient temperature between 24 °C and 26 °C, and relative humidity between 51 % and 57 %.
Despite some inefficiencies identified in the PV-AEM system, the overall system performance remains within the expected range based on literature values. Further improvements in efficiency could be achieved by minimizing thermal losses and directly supplying DC power from the PV array to the electrolyzer stack, eliminating losses associated with inverters and power conditioning stages. To meet IRENA’s 2050 system-level specific energy target of 45 kWh/kg, losses in the investigated system must be reduced by at least 18.72 kWh/kg. Removing the inverter stage alone could eliminate approximately half of these losses as depicted in Fig. 8, making the 2050 target more attainable. Clearly, moving to direct DC power is an appropriate strategy from both an energy perspective [69] and life cycle analysis [70]. The comparison of data from this study, past studies and IRENA goals are shown in Table 3.
Table 3. Comparison of key performance indicators (KPI) investigated in this study with literature data and IRENA 2050 KPI targets.
| AEM Key Performance Indicator (KPI) | Literature Range | Reference | This Study’s Performance (FPV-AEM) | IRENA 2050 Target [20] |
|---|---|---|---|---|
| Specific Energy Stack (kWh/kg) | 76.77–48.25 | [16,[62], [63], [64], [65], [66], [67]] | 45.77 | <42 |
| Specific Energy System (kWh/kg) | 83.44–50.01 | [62,65,67] | 63.72 | <45 |
| Stack Efficiency (%) | 69.64–79.26 | [16,28,63] | 86.2 | >88 |
| System Efficiency (%) | 49.25–57.24 | [68] | 61.9 | No goal |
| Hydrogen Flow (m3/h/kW) | 0.19–0.21 | [16,64] | 0.28 | No goal |
| Oxygen Flow | Data Not Available | 0.16 | No goal | |
| Hydrogen Purity | 99.5–99.99 | [64,65,71] | 99.22 | 99.9999 |
4.2. Hydrogen and oxygen flowrates, and large-scale PV-AEM integration
The hydrogen flowrate varied between 0.17 m3/h/kW and 0.28 m3/h/kW, surpassing the range of 0.19 m3/h/kW to 0.21 m3/h/kW reported in the literature. These results indicate that the analyzed AEM stack has the potential to generate more H2 at the system level than previously studied AEM stacks. Conversely, the system generated O2 at rates between 0.09 m3/h/kW and 0.16 m3/h/kW. The measured hydrogen purity was 99.2 %, close to the 99.5 % reported in the literature [16,64,65,71], but further purification is necessary to meet the industrial standard of 99.999 % [72].
One of the key contributions of this work is demonstrating the feasibility of powering AEM electrolyzers with PV systems at a kilowatt scale. While previous research has focused on membrane development [[73], [74], [75]] or Watt-scale AEM stacks [15,76], this study is the first to document a combined PV-AEM stack at this scale. The ability to integrate PV and AEM at a larger scale opens opportunities for off-grid solar PV system optimization, particularly in winter-dominant regions where solar irradiation is lower, and excess energy during summer remains underutilized. In areas with limited grid access, insufficient incentives for grid-tied solar PV, or unjustly priced grid buy-back prices [9,77], distributed H2 production using a PV-AEM system could serve as a decentralized energy solution. Individual buildings or small communities could generate and store H2 on-site using excess energy from oversized PV systems, reducing reliance on centralized energy sources and enhancing energy resilience [55,[78], [79], [80]].
In the case of FPV systems, an additional benefit is the use of local water sources as the primary feedstock for the electrolyzer. The synergistic approach of coupling FPV with AEM hydrogen production is particularly advantageous. The FPV reduces input costs and provides cooling benefits to PV modules [33,40], while also mitigating evaporation [34,81], contributing to sustainable water management. This synergistic coupling of FPV and AEM hydrogen production reduces input costs and maximizes resource use, making FPV-hydrogen systems particularly well suited for land-constrained or water-rich regions.
Beyond the technical demonstration, this work offers a pathway for policymakers and industry stakeholders to deploy decentralized green hydrogen solutions. Supporting pilot projects that co-locate water-surface PV arrays with onsite electrolyzers can alleviate land-use pressures and foster local fuel generation in remote or grid-limited areas. Moreover, targeted incentives and standards for direct DC-coupled electrolyzer systems could recover substantial conversion losses; potentially improving round-trip efficiency by more than 15 %; strengthening the economic case for green hydrogen. Recognizing hydrogen as a dispatchable resource in off-grid and microgrid regulations would further enable diesel-replacement strategies and bolster community resilience. Looking ahead, integrating these empirical performance data with comprehensive techno-economic models will be essential to guide large-scale implementation, refine system design, and inform regulatory frameworks that accelerate PV-driven hydrogen deployment.
4.3. Limitations and future work
Despite its promise, the PV-AEM system has limitations that need to be addressed to enhance performance. One key challenge is the reliance on solar irradiance, which fluctuates with seasonal and weather conditions. In colder climates, limited solar availability in winter poses challenges for consistent H2 production without excessive battery storage, hybridization or grid reliance. Future studies should explore deliberate oversized PV-electrolyzer configurations to optimize load planning in fully off-grid PV-AEM systems.
Another major limitation is power conversion inefficiency, particularly in the DC-AC and AC-DC transitions. Since AEM electrolyzers are inherently DC devices, they could be directly powered by PV arrays, as demonstrated in a recent small-scale study [15]. Future research should focus on designing and evaluating DC converters for large-scale AEM electrolyzers, potentially eliminating conversion losses. Investigations could assess whether AEMs operate better with a direct PV connection, as they have been shown to function at as low as 20 % of their rated capacity [62], or in a battery-supported PV system.
Intermittent operation of the electrolyzer, inherent in off-grid PV systems, results in a low-capacity utilization factor. Operating the electrolyzer at partial load, however, could extend its lifetime, which is a rich area of future research. For example, an AEM electrolyzer operating at 30 % capacity as shown in Fig. 9 would function for approximately 1739 h annually, equating to a 2.87-year lifetime, considering current membrane longevity (∼5000 h) [20,72]. More research is needed to improve membrane durability and develop strategies for maintenance and replacement.
The absence of an industrial-grade gas collection and purification system limited H2 purity. Additionally, heat losses due to a lack of insulation reduced system efficiency. Future research should focus on improving thermal efficiency by modeling and experimentally analyzing the system’s thermal behavior and developing optimally insulated enclosures for safe residential deployment.
The present work was constrained by its short-term experimental nature, which does not account for long-term system degradation or maintenance requirements. The impact of environmental conditions such as temperature and humidity on performance was not fully explored. A systematic sensitivity analysis should be conducted to assess the influence of wide ranges of wind speed, temperature, and humidity on specific energy consumption, thereby confirming the robustness of the reported minimum value under diverse environmental conditions. Future work should include long-term performance assessments, particularly in winter-dominant regions, to better understand system behavior over extended periods. Such investigations could involve controlled experiments during winter months alongside complementary thermal and electrical modeling under sub-zero, low-irradiance conditions. This line of inquiry will enable quantification of seasonal effects on H2 production and overall system efficiency. The resulting insights would guide targeted design modifications such as enhanced insulation, tailored electrolyte formulations, or the integration of auxiliary heat sources to guarantee dependable year-round operation. Additional studies should investigate the economic viability of PV-AEM systems in real-world residential applications and explore financial models that support their adoption. Furthermore, optimizing FPV integration with H2 production in diverse geographic regions, accounting for variations in solar irradiance, wind speed, and temperature, will help refine system design and implementation.
5. Conclusions
This study demonstrated the feasibility of integrating solar PV with an AEM electrolyzer for H2 production at a large scale. The results showed that the AEM stack achieved a high energy conversion efficiency, ranging from 73.3 % to 86.2 % (using H2 HHV), with a minimum specific energy consumption of 45.77 kWh/kg—one of the lowest reported values for AEM electrolyzers. While the overall PV-AEM system exhibited lower efficiencies due to conversion losses, direct DC coupling and thermal loss mitigation were identified as key opportunities for system performance improvement. Additionally, the system produced H2 with a purity of 99.22 %, highlighting the need for close-loop gas collection and purification systems to meet industrial-grade standards of 99.999 %. The findings underscore the potential of PV-powered AEM electrolysis as a viable solution for decentralized H2 production, particularly in off-grid applications. In regions with limited grid infrastructure or poor compensation of solar prosumers, excess solar energy that is often curtailed or underutilized can be stored in the form of H2, enhancing overall system operation, and resilience. Moreover, floating PV systems present an integrated approach where water availability can be leveraged both as an electrolyzer feedstock and as a cooling mechanism to improve PV efficiency. Looking forward, optimizing power electronics to enable direct DC coupling, improving system thermal insulation, and optimizing system operation for economic performance will be crucial in making PV-AEM systems more competitive. As hydrogen continues to emerge as a key pillar of the energy transition, its integration with renewable power sources offers a pathway to reducing dependence on fossil fuels, balancing seasonal energy supply fluctuations, and fostering a more sustainable and distributed energy landscape. The demonstrated efficiency and flexibility of PV-AEM electrolysis brings the technology one step closer to realizing the International Renewable Energy Agency’s 2050 targets for cost-effective and low-carbon H2 production.
May 17, 2025 at 04:33PM
https://www.sciencedirect.com/science/article/pii/S0360319925024383?dgcid=rss_sd_all