1. Introduction
Hydrogen has garnered significant attention as an energy carrier in response to escalating global air pollution. Recognized for its cleanliness and versatility, it serves as a promising energy source capable of providing heat and power across various sectors [
1,
2]. As the most abundant element in the universe, hydrogen’s applications span clean energy, renewable storage, transportation decarbonization, industrial use, and more [
3,
4]. Advancing green hydrogen is essential for enabling low-temperature fuel cells as viable alternatives to battery electric systems in combating climate change [
5].
Hydrogen does not naturally exist in its diatomic molecular form and must be synthesized using energy sources such as fossil fuels. Common methods include steam reforming of methane at high temperatures [
6], gasification [
7], and even emerging approaches like thermo-catalytic plastic waste conversion [
8]. Natural gas, rich in methane, remains a key feedstock, reacting with steam (700–1000 °C) to yield hydrogen [
9] through the following steps:
Hydrogen is often classified by its production method: grey (from fossil fuels, emits CO
2), blue (uses CCS), and green (via electrolysis using renewables) [
10,
11]. Green hydrogen is the most environmentally friendly. Electrolysis splits water into hydrogen and oxygen using electricity [
12]:
Among all current methods for hydrogen production via water electrolysis, proton exchange membrane electrolysis cells (PEMECs) have emerged as a leading technology due to their ability to deliver high-purity hydrogen, operate at elevated current densities, respond swiftly to fluctuating energy inputs, and maintain a compact, scalable design [
13,
14,
15].
This review provides a focused evaluation of numerical simulation approaches aimed at enhancing PEMEC performance, with an emphasis on advancements in modelling techniques and their practical applications. It covers a range of computational models—from zero-dimensional to fully three-dimensional frameworks and from single-phase to multiphase simulations—analyzing outputs such as polarization curves, efficiency predictions, and multiphysics coupling [
13,
14,
15,
16]. Through critical analysis of these methods, this review identifies key strengths, limitations, and research gaps that require further refinement to improve predictive accuracy and support experimental validation.
A key novelty of this work lies in its integrated approach, linking recent material developments—including membrane innovations, catalyst architectures, and porous transport layers—with their direct influence on numerical simulation results. Unlike prior reviews that treat materials and simulations independently, this study emphasizes the mutual dependency between material properties and modelling accuracy, showing how simulations can also guide future material design and system-level optimization.
Furthermore, this review delves into advanced multiphysics and two-phase flow models, which are essential to understanding internal PEMEC transport dynamics. Recent 3D, non-isothermal, two-phase studies have provided valuable insights into gas–liquid interactions, electrochemical kinetics, and spatial mass transport—crucial for optimizing PEMECs at high-current densities for commercial applications.
To support this modelling emphasis, this paper explores how material configurations and component properties—such as membrane conductivity, catalyst layer porosity, and flow-field geometry—affect simulation fidelity and real-world performance [
17,
18,
19,
20,
21]. In doing so, it highlights emerging challenges, such as the need for validated two-phase models, scalable catalyst structures, and high-fidelity numerical frameworks that balance accuracy and computational cost.
This review is structured to deliver a comprehensive and systematic overview of PEMEC simulation research.
Section 2 introduces the fundamentals of water electrolysis, comparing key hydrogen production technologies.
Section 3 outlines the core governing equations used in PEMEC modelling, while
Section 4 examines critical components—including membranes, catalysts, and flow fields—and their role in cell behaviour.
Building on this foundation,
Section 5 discusses modern numerical approaches from low- to high-dimensional models and from single-phase to two-phase simulation strategies.
Section 6 evaluates flow-field modelling and the trade-offs between simulation accuracy and computational demand.
Section 7 compares single-phase and two-phase PEMEC models, emphasizing the importance of capturing detailed transport phenomena such as gas evolution, velocity fields, and pressure distribution. This is supported by benchmarking results against experimental data, highlighting the value of two-phase modelling in replicating real-world behaviour.
This paper concludes with
Section 8, which summarizes the key findings and research directions for advancing PEMEC scalability and performance, followed by
Section 9, which discusses current limitations and the need for further model validation, particularly under dynamic operating conditions.
By connecting modelling advancements with material innovation and system optimization, this review aims to provide a comprehensive roadmap for the development of cost-effective, efficient, and scalable PEMEC technologies for green hydrogen production.
6. Mathematical Modelling and Simulation
Modelling and simulation play a crucial role in understanding the complex electrochemical and transport processes in proton exchange membrane electrolyzer cells (PEMECs). Numerical simulation models provide detailed insights into system behaviour, helping to optimize design and operating conditions. Over the years, various models—ranging from zero-dimensional analytical models to three-dimensional CFD simulations—have been developed to quantify the key physical and electrochemical phenomena occurring in PEMECs. This section presents the governing equations, boundary conditions, and initial conditions used in these models, along with a comparison of their assumptions and predictive capabilities. A particular focus is placed on the differences between single-phase and multi-phase models, which play a crucial role in capturing gas evolution and water transport mechanisms. In addition, emerging methodologies such as multi-scale modelling, computational fluid dynamics (CFD), and machine learning (ML)-based optimizations are discussed [
14,
16,
27,
78,
91,
92].
Table 5 summarizes various modelling approaches for PEMEC simulations, categorizing them based on their complexity and the physical aspects they incorporate.
- i.
Zero-dimensional model
Choi et al. [
17] developed a zero-dimensional model to analyze PEM electrolysis cells based on steady-state mass balances and electrochemical kinetics. The net molar production or consumption rates of water, hydrogen, and oxygen are governed by Faraday’s law, expressed in differential form as:
where
is the molar production or consumption rate of species
[mole/s],
is local current density [A/m
2],
is the active electrode area [m
2], and
is Faraday’s constant [C/mole]. These flow rates are directly related to the electrochemical reactions governed by the cell potential. Under the assumption of no mass transport limitations, the current density can be linked to the cell overpotential (
) using a simplified form of the Butler–Volmer equation, as stated below [
93]:
where
represents the anode exchange current density (under the well-mixed condition at the anode).
is the overpotential at the anode chamber that can be expressed as a function of temperature (
T), current density (
, and anode exchange current density (
, alternatively.
and
are the stoichiometric coefficient for electrons in the anode and transfer coefficient which have the values of 2 and 0.5, respectively [
94]. Assuming these values, the anode overpotential can be expressed explicitly as Equation (6).
The overpotential can be written in the form of cathode parameters (
,
, and
) where
and
take values of −2 and 0.5, respectively (Equation (7)).
To simulate the membrane potential (
, the divergence of current density is considered to be zero along with the steady-state conditions (no change over time). Consequently, the current density becomes directly proportional to the membrane potential, with electrolyte conductivity (
) serving as the factor equalizing this proportionality (Equation (8)).
Considering all available potential sources, e.g., Nernst potential, anode, cathode, and membrane overpotentials and losses in an electrolysis cell (see [
17]), an expression can be employed to determine the overall voltage (
) in relation to the current density (
). This relationship (Equation (9)) yields a polarization plot.
where
,
,
, and
are the Nernst potential, membrane thickness, membrane conductivity, and interfacial resistance, respectively. In the work of [
17], this curve was plotted and compared with the experimental findings in [
95]. However, their model did not consider heat transfer and lacked porous membrane simulations, limiting its accuracy [
17]. In PEM modelling (both PEMFC and PEMEC), the term “single-phase” refers to water existing solely in liquid or vapor form within the system, whereas multi-phase models incorporate gas bubble formation and liquid–gas interactions.
Table 5.
Evolution of numerical modelling for PEMEC—from simplified to advanced approaches.
Table 5.
Evolution of numerical modelling for PEMEC—from simplified to advanced approaches.
Model | Study Description | Key Output | Ref |
---|
0D models | Simplified electrochemical kinetics in PEMEC with no heat and mass transfer considerations. | Analyzed overpotentials across a range of current densities. | [20] |
Heat transfer and fluid flow simulation in PEMEC without electrochemical kinetics or liquid–gas interaction. | Investigated pressure drop in bipolar plates numerically and experimentally. | [80] |
Gas–fluid flow simulation in PEMEC with no electrochemical kinetics. | Presented velocity and pressure contours and compared results with single-phase flow data. | [95] |
Single-phase 3D models | Heat and mass transport modelling in PEMEC with simple electrochemical kinetics (no gas–liquid interaction). | Hydrogen generation in high-temperature PEMEC analyzed through polarization curves. | [23] |
Comprehensive PEMEC model incorporating detailed electrochemical kinetics (no gas–liquid interaction). | Examined polarization curves at different temperatures and analyzed the effect of exchange current density. | [96] |
PEMEC modelling with multiple fluid flow configurations (no gas–liquid interaction). | Investigated the effect of different flow configurations on polarization performance. | [24] |
Two-phase 3D models | PEMEC modelling under low-temperature conditions with detailed electrochemical kinetics. | Conducted an analytical study of PEMEC behaviour under low- and high-potential conditions. | [15] |
Non-isothermal, transient PEMEC model. | Studied temperature stabilization under various operational conditions. | [97] |
PEMEC simulation including gas–liquid interactions, integrated with electrochemical kinetics. | PEMEC simulation including gas–liquid interactions, integrated with electrochemical kinetics. | [14] |
- ii.
Single-phase model
In 2009, Nie et al. [
78] performed a three-dimensional numerical simulation to assess the fluid flow in the bipolar plates of a PEMEC using a single-phase model. The principles and governing equations in a PEMEC are very similar to the those in the work of Ubong et al. [
41] for a PEM fuel cell (PEMFC) operating at elevated temperatures with some minor distinctions. The objective of a PEMEC is to generate hydrogen from water with the aid of an external voltage, while in the PEMFC, hydrogen is used as fuel water to generate voltage. The membrane in the PEMFC is still used to separate the anode and cathode compartments and allows for the transport of protons (H
+) from the anode to the cathode but in reverse. Hydrogen is typically fed into the anode, and oxygen (or air) is fed into the cathode. At the anode, hydrogen is oxidized, producing protons and electrons. The protons migrate through the PEM to the cathode, where they combine with oxygen and electrons to form water. The flow of electrons through an external circuit generates electric power [
41,
72,
73]. The electrochemical reactions in a PEMFC are shown in Equation (10).
The mass conservation equation in a single-phase flow within homogeneous and isotropic porous media (PEM elements, e.g., membrane, PTL, CL) can be expressed in a general form as Equation (11) [
26].
where
,
, and
are the velocity vector, density, and porosity, respectively.
represent the source term accounting for the generation or consumption of mass within the system. The momentum equations for the single-phase flow in porous media can be written in the form of Equation (12) [
98].
where
and
represent pressure and the source term in momentum equations, respectively.
represents the tension tensor, which in Newtonian fluid can be expressed in terms of the velocity gradient (Equation (13)) [
26].
where
represents the dynamic viscosity of the gas.
Within porous media, Darcy’s law can be applied to relate the pressure drop to the gas velocity. The source term in Equation (12) can be expressed as Equation (14) [
77]:
where
and
are the permeability and porosity of porous media, respectively. Permeability can be determined using the Kozeny – Carman model [
26]:
where
represent the particle diameter of porous material. The value of these two parameters may vary in different materials and across different sections in the PEMEC (See
Table 6).
Beyond porosity and permeability, tortuosity ( significantly influences effective transport properties in porous electrodes by accounting for the complexity of internal pore paths. Similarly, electrical conductivity governs charge transport in PTLs and CLs and must be accurately represented in multiphysics models to predict electrochemical performance.
The molar concentrations of the active species in the process of electrolysis (hydrogen, oxygen, and water) are determined by solving the conservation of species (Equation (16)) [
77].
where
and
are molar concentration and effective diffusion coefficient of each species.
is the function of cell operating conditions like temperature and pressure and can be expressed as Equation (17) [
77].
where
and
denote the reference temperature and pressure.
is the reference diffusivity of hydrogen, oxygen, and water, which are 1.1
10
−4, 3.2
10
−5, and 7.35
10
−5 (
), respectively [
77,
91].
refers to the source term for each species and can be expressed as Equation (18) [
96]:
where
M is the molecular weight of the species, which has the value of 18, 2, and 32 [gr/mole] for water, hydrogen, and oxygen, respectively.
In steady-state conditions, there is a simple balance between proton current in the membrane (
and electron current in the electrodes (
, which is stated in Equation (19) [
93].
where
and
can be calculated by Ohm’s law [
93].
The next important governing equations used to determine the proton and electron potential and are expressed as Equations (20) and (21) [
77]
where
and
are the membrane (electrolyte) electrical potential and effective ion conductivity, respectively, while
and
are corresponding parameters in the solid phase (electrode).
and
are the source terms for the protonic and electrical potentials, respectively, which can be simply related to the current density (
), as shown in Equation (22) [
41,
77].
The effective electronic and membrane conductivities can be estimated using Bruggeman’s approach, which serves as a reliable approximation [
97]. As the effective membrane conductivity is also a function of temperature (
) and humidification degree in the membrane (
), the following expression can be used for it [
96,
99].
The last governing equation that determines the temperature distribution within the PEM cell is the energy equation in porous media. It can be expressed as Equation (24) [
100].
where
and
are specific heat at constant pressure and effective thermal conductivity.
There are various models to calculate the effective thermal conductivity, the most important of which are parallel, series, and geometric. One of the easiest models is using the parallel model, in which the fluid and the solid are parallel to the heat flow. It is a linear correlation, as shown below [
101]:
where
and
are the thermal conductivity of a solid (catalyst electrode) and liquid electrolyte (membrane), respectively. In this model, heat flows independently through the liquid and solid. In the series model, heat passes through each layer in sequence. Using a series model produces the lowest value for effective thermal conductivity [
101]. A brief comparison between the calculated amount of effective thermal conductivity using various methods is presented in
Table 7.
The source term in the energy equation (
may vary across different sections (e.g., anode CL, cathode CL, solid phase, and membrane). Neglecting the heat sources produced by irreversible reactions and entropic heat generation because of the reactions (in anode and cathode CL), the source term for different elements in a PEM can be simplified as shown in
Table 8 [
89].
Ruiz et al. [
102] developed a three-dimensional mathematical model to investigate key electrochemical and thermal phenomena in high-temperature PEMECs (above 100 °C). Their study highlighted several advantages of operating at elevated temperatures, including enhanced electrode kinetics, reduced overpotentials, lower thermodynamic energy requirements, and a decrease in reversible voltage. Additionally, the model incorporated a more detailed heat and mass transfer analysis in conjunction with electrochemical kinetics [
102].
That study further examined the impact of three different flow channel configurations—parallel, serpentine, and multi-serpentine—on hydrogen production and temperature distribution. Using a single-domain model in ANSYS Fluent 13 with non-isothermal flow properties, their results showed that the multi-serpentine design offered superior performance by promoting a more uniform temperature distribution. Notably, the model assumed a single-phase flow, neglecting the presence of liquid water due to the elevated operating temperature.
As discussed in the preceding section, the design of the flow field significantly impacts cell performance [
48]. Toghyani et al.’s [
77] study illustrated that altering the arrangement of the flow field led to changes in the distribution of reactants and products transferring toward the PEMEC outlet. The researchers employed a 3D model to simulate the electrolyte in three different arrangements: a parallel flow field (with and without metal foam as a flow distributor), a double serpentine flow field, and a simple channel filled with metal foam. The most recent arrangement exhibited superior performance in hydrogen production, temperature distribution, current density, and pressure drop [
77]. According to their findings, reducing the permeability of the porous media results in an increased pressure drop in the system and enhanced performance of the electrolyzers. Consequently, the optimal permeability for the foam should be carefully selected. In terms of porosity, they modelled this property as a macro-scale feature rather than a micro-structural one. They adopted a single-phase, non-isothermal, and steady-state assumption to streamline their 3D numerical simulation using the finite volume method. Additionally, they illustrated that the spiral flow field leads to greater uniformity in hydrogen distribution [
77,
91]. A schematic correlation between the relevant physics and parameters involved in analyzing a PEMEC (either in single-phase or multi-phase flow) is illustrated in
Figure 5.
- iii.
Two-phase model
Single-phase mass transport is not a realistic assumption in most PEMEC situations due to the limitations of the model [
26]. Numerous researchers have endeavoured to develop comprehensive models aiming to furnish precise insights into critical parameters within PEMEC, such as current density, volume fraction, temperature, and flow velocity. These models should possess the capability to analyze PEMEC not only under high operating temperatures and low-current densities, where single-phase models suffice, but also under low-temperature conditions with high-current densities, where single-phase models fail to deliver accurate data [
26,
27,
41,
77,
91]. Zinser et al. [
103] introduced a two-phase model to account for the transportation of oxygen and water within the anodic PTL in a PEMEC (1D half-cell model). The crucial drying-out process, which significantly influences water management in both PEMEC and PEMFC [
28], was investigated in the PTL using pressure distribution and saturation plots. To enhance transport behaviour, a suitable range of macro-structural features, including porosity, permeability, and geometrical features, was identified. This observation aligns with the findings of Kalinnikov et al. [
30], who, while considering the limitations of two-phase mass transport in the PTL for different regimes, noted similar improvements [
30]. One of the key aspects of studying two-phase mass transport is identifying the critical current density profile and distinguishing the stable regime (with a steady-state profile) from the unstable one (characterized by drying-out) [
30,
103]. Integrating an electrochemical model that includes simulations of coupled thermal-fluid dynamics, species transport, and electrochemistry into a two-phase flow provided a more comprehensive understanding of species transport and accurate predictions of overall cell performance [
15]. These insights directly aid in optimizing PEMEC stack design by improving PTL structures, catalyst layer configurations, and flow-field geometries to enhance reactant distribution and minimize performance losses.
In 2023, several researchers dedicated their efforts to 3D modelling of PEMEC, aiming to explore the influence of various operational factors such as current densities and temperature on overall PEMEC performance. They also investigated the impact of flow-field configurations on mass transport [
14,
16,
26,
27]. A common thread in these studies involved incorporating more sophisticated mass transport considerations into the two-phase flow (comprising water and gas) and conducting numerical simulations using commercial software such as ANSYS Fluent and COMSOL Multiphysics. These simulations varied boundary conditions and initial conditions across different geometries of the PTL and channel arrangements. This approach enabled the analysis of the bubble detachment process during different stages of hydrogen production, including the initial phase, instability, deformation, and separation. Factors such as gravity, buoyancy, and surface tension in the microchannel were considered during this analysis [
27]. Additionally, researchers explored gas bubble accumulation on the catalyst using integrated mass transport combined with a classic model. Gas bubble accumulation at the CL reduces effective reactant transport, leading to localized starvation, increased ohmic losses, and uneven current density distribution. Optimized PTL structures and improved two-phase flow management can mitigate these losses, enhancing overall PEMEC performance. An example of this is Jiang et al.’s work [
26], where the study revealed that gas discharge could be enhanced by using a combination of PTL and CL to reduce heat and mass loss.
In almost all of the two-phase model simulations, the electrochemical model that was used was the Butler–Volmer equation, which was elaboratively described in previous sections. To simulate two-phase systems, various methods have been employed in different studies. The two most significant models are the mixture model (homogeneous equilibrium) [
14,
15] and the Eulerian model [
26]. While the Eulerian model solves separate continuity and momentum equations for each phase, offering detailed phase tracking, it requires significantly higher computational resources. In contrast, the mixture model assumes local equilibrium and treats the two phases as a single homogeneous medium, reducing computational cost at the expense of phase separation accuracy. The choice between these models depends on the required resolution of gas–liquid interactions and available computational power.
Various methods are available for simulating flow properties within the governing equations. For instance, the Darcy model is utilized to simulate porous media [
30,
103]. Additionally, the volume of fluid (VOF) model, applicable to both single-phase and multi-phase flows, is employed to study phenomena such as bubble growth [
104]. The choice of model relies on parameters such as the nature of the flow (e.g., bubbly, annular) and computational sources. The governing equations for these models are described in the following section. Most two-phase PEMEC models assume a bubbly or slug flow regime in the PTL, with gas bubbles forming due to electrochemical reactions. In Eulerian models, phase interactions such as surface tension, drag force, and capillary effects are explicitly included, whereas mixture models assume local phase equilibrium. Typical boundary conditions include constant pressure at the outlet, velocity inlets for water feed, and no-slip conditions at solid interfaces. These assumptions influence the accuracy and computational cost of the simulations and should be selected based on the specific PEMEC operating conditions [
14,
15].
In this model, separate continuity equations are solved for each phase (liquid and gas), assuming no water evaporation. This implies that the rate of mass transfer between the two phases is considered negligible [
26]. A dimensionless quantity known as volume fraction (
) is defined to determine the relative number of different phases within the flow, which is shown in Equations (26) and (27) [
26]:
where
and
are liquid and gas volume fractions, respectively. It can be easily concluded that
. Thus, the continuity equations for liquid and gas in the flow are Equations (28) and (29), respectively [
105].
,
represent the source terms arising from electrochemical reactions for the active species. These values resemble those in the single-phase model (Equation (18)). Considering the mass change rate between liquid and dissolved water
within the anode catalysts, the source term associated with water consumption undergoes a slight modification, as expressed in Equation (30) [
26].
Similar to the single-phase model, the momentum conservation equation in the Eulerian two-phase model can be developed as Equations (31) and (32).
where
and
are tension tensors for liquid and gas, which in Newtonian fluid can be expressed in terms of velocity gradient (Equations (33) and (34)) [
106].
The terms
and
denote the dynamic viscosity of the liquid and gas (Pa·s).
and
represent the source terms for the liquid and gas phases. In the context of PEMECs, these terms originate from two distinct sources: (i) viscous resistance between porous walls and fluid and (ii) interaction drag force between liquid and gas [
106]. An effective model proposed by Schiller and Naumann is employed to simulate these interactions within various elements of a PEMEC, such as channels, PTL, and catalyst layers, as comprehensively described in [
26].
Similar to mass and momentum conservation equations, a set of energy equations can be used to determine the temperature distribution for both gas and liquid phases in PEMECs, as outlined in Equations (35) and (36) [
26,
107].
where
represent the energy that is released due to the electrolysis of water and varies throughout different sections within a PEMEC. The main parameters involved in determining the value of
are current density (
, membrane conductivity (
and solid conductivity
. A summary of heat sources in different layers of a PEMEC is presented in
Table 7. The heat generation within the channels can be considered zero [
26].
To determine the mass fraction of the species in the cell, one equation should be considered, which closely resembles Equation (16). If water evaporation is negligible, the species should be considered to be reduced to hydrogen and oxygen only. In this context, the mass fraction of hydrogen (
is determined by solving Equation (37) [
26,
107].
where
and
are the mass diffusion coefficient and hydrogen mass crossover. The mass fraction of oxygen can be calculated simply by subtracting from 1.
In this model, the two phases are treated as a single homogenous mixture with combined properties, including velocity (
, pressure (
, and temperature (
[
108]. Applying this technique significantly reduces the computational cost as the number of governing equations is halved [
15,
108]. The governing equations in the mixture model are derived by averaging on the Navier–Stokes equations. A separate continuity equation, along with a set of algebraic closure equations for relative velocities, is developed to calculate the volume fraction of the dispersed phase. [
109]. Liquid water acts as the continuous phase, while gas is considered the dispersed phase. On the cathode side, the dispersed gas is a combination of hydrogen and water vapor, whereas on the anode side, hydrogen substitutes for oxygen in the mixture [
15].
By defining two new parameters, namely the mass-averaged velocity (
) and mixture density (
), the continuity equation in the mixture model can be formulated, as shown in Equation (38) [
108].
where
is calculated by the densities of the species with their corresponding volume fractions, as outlined in Equation (39).
where
,
, and
are volume fraction, density, and the velocity vector of the active species in the flow. The mixture density (
) can be expressed by Equation (40) [
15,
108].
The momentum equation in the mixture model can be derived by combining the momentum equations (Equation (12)) for the continuous and dispersed phases in the flow [
110]. The general form of the momentum equation in the mixture model is outlined as Equation (41) [
15,
16].
where
is the tension tensor for the mixture flow and can be expressed in terms of the mass-averaged velocity gradient (Equation (42)) [
16,
77].
represents the source term in the momentum equation.
where
is the mixture’s dynamic viscosity and can be expressed as Equation (43) [
15,
16,
108].
The term
in Equation (41) denotes the drift velocity for the disperse phase
n, calculated as the difference between the disperse phase velocity and the mixture velocity. (
. Utilizing the same methodology, the energy and species transport equations can be derived for the mixture model, with comprehensive details provided in [
15,
108,
110,
111]. Recent advances in two-phase PEMEC modelling highlight the need for a balance between computational efficiency and physical accuracy. While Eulerian models provide detailed phase tracking, mixture models offer a computationally feasible alternative. Studies continue to refine these approaches by incorporating more realistic boundary conditions, improved mass transport considerations, and coupling with electrochemical simulations to enhance the predictive accuracy of PEMEC performance. After reviewing the governing equations for single-phase and two-phase models, it is useful to summarize the key trade-offs between these approaches.
Table 9 provides a comparative analysis of single-phase and two-phase (Eulerian and mixture) models in terms of computational cost, accuracy, and suitability for different operating conditions. This comparison highlights the strengths and limitations of each method, assisting in the selection of an appropriate model for PEMEC simulations.
To provide a clearer visual comparison,
Figure 6 presents a bar chart illustrating the computational cost, accuracy, bubble dynamics, and suitability of each modelling approach at different current densities. This comparison (
Table 9 and
Figure 6) establishes the fundamental differences between single-phase and two-phase modelling approaches. However, these general trade-offs must be examined in more detail by analyzing specific performance outputs such as polarization plots, pressure distribution, velocity fields, and species volume fractions. The next section expands this analysis by comparing PEMEC simulation outputs from various models with experimental data.
To complement the discussion above,
Table 10 provides a summary of selected PEMEC simulation studies, highlighting the modelling approaches, software platforms, and notable features used in the literature.
7. A Comparative Analysis Between the PEMEC Models
This section evaluates the performance of different PEMEC models by comparing key electrochemical and transport parameters. The analysis includes polarization plots, pressure distributions, velocity fields, and species volume fractions to assess the trade-offs in accuracy, computational cost, and physical realism. The results from different numerical studies are compared with experimental data, providing insights into how well each model captures real-world PEMEC behaviour.
A key benchmark for PEMEC models is their ability to reproduce polarization curves, which offer critical insights into electrochemical performance.
Figure 7 presents a comparative analysis of polarization plots derived from a 0D model [
17], a single-phase model [
77], a two-phase mixture model [
15], and experimental data [
114]. These results, obtained under similar operating conditions (80 °C and 1 atm), highlight the limitations of simpler models and the advantages of incorporating two-phase transport effects.
The agreement between model predictions and experimental data depends on how well each model captures key transport phenomena, including gas evolution, bubble formation, and flow distribution. These effects become particularly significant at high-current densities, where single-phase models fail to represent real operating conditions accurately. To address these limitations, two-phase models—Eulerian and mixture-based approaches—are employed.
Both Eulerian and mixture models simulate gas–liquid interactions, but they differ in terms of accuracy and computational cost. The Eulerian model treats liquid and gas as separate phases, allowing for detailed bubble tracking but requiring substantial computational resources. The mixture model, on the other hand, assumes local phase equilibrium, approximating gas–liquid interactions with reduced complexity. While Eulerian models offer high-fidelity analysis, mixture models provide a practical alternative for large-scale simulations where phase separation effects are less dominant [
14,
15,
16,
26].
While polarization plots provide valuable electrochemical insights, they do not fully capture the transport dynamics within a PEMEC. Understanding how liquid and gas phases interact within the flow field is crucial for optimizing performance, particularly at high-current densities, where gas evolution significantly alters transport properties. To investigate these effects, several studies have compared single-phase and two-phase flow models within PEMEC half-cell domains—most notably those by Nie et al. [
108], which provided key reference cases for evaluating gas–liquid interactions and their impact on cell performance. As shown in
Figure 8, In their two-phase simulation of a PEMEC anode domain, they investigated pressure distribution and gas–liquid interactions using a mixture model approach. The study revealed a high-pressure region at the inlet and a lower-pressure outlet, with oxygen bubbles forming and influencing the transport properties along the channel. The reported pressure gradient ranged from 96 to 107 kPa, illustrating the critical influence of bubble dynamics on hydrodynamics and performance predictions.
While pressure distribution provides insights into transport resistance and hydrodynamic performance, the volume fraction of water (H
2O) also plays a critical role in defining reactant availability, species transport, and electrochemical reaction efficiency. In PEMECs, oxygen bubble formation progressively reduces the local water concentration, affecting catalyst layer hydration, membrane conductivity, and overall transport properties. To explore this behaviour, Nie et al. [
108] simulated the distribution of water volume fraction in a two-phase flow field. Their results, presented in
Figure 9, revealed high water content at the inlet region (approximately 0.9–1.0), which gradually decreased along the channel length to around 0.5–0.6 at the outlet due to gas bubble formation. This spatial variation in water content highlights the transition from single-phase to two-phase flow and reinforces the importance of accurately modelling liquid–gas interactions in PEMECs.
Beyond water distribution, oxygen evolution plays a crucial role in PEMEC operation, influencing reactant availability, pressure distribution, and overall system efficiency. The accumulation of gas bubbles introduces additional transport resistance and alters local flow dynamics, making accurate modelling of oxygen volume fraction essential. Nie et al. [
108] investigated this phenomenon through two-phase simulations using the mixture model, as shown in
Figure 10. Their results revealed a clear trend: oxygen volume fraction begins near zero at the inlet—reflecting the absence of bubbles in the incoming liquid stream—and gradually increases along the flow field as electrochemical reactions proceed. By the outlet, the oxygen fraction reaches approximately 0.9–1.0 due to continuous gas generation and transport. This behaviour illustrates the dynamic impact of bubble formation on species transport and validates the importance of multiphase modelling in capturing PEMEC performance under realistic operating conditions.
As shown in
Figure 11, while the previous analyses focused on two-phase transport phenomena, it is also essential to evaluate single-phase flow conditions to establish a baseline for pressure distribution in PEMECs. Single-phase models, where only liquid water is present without gas evolution, provide insights into flow uniformity and pressure gradients in the absence of bubble-induced transport resistance. As reported by Nie et al. (2009) [
78], single-phase simulations reveal a gradual decline in pressure from the inlet to the outlet under steady-state flow conditions, with liquid water introduced at a constant flow rate (e.g., 60 mL/min). These trends help isolate the fundamental hydrodynamic behaviour of the cell before introducing multiphase interactions. The observed pressure distribution serves as a computationally efficient reference for understanding flow resistance and guiding flow-field design in PEMECs.
By establishing this single-phase baseline, the impact of bubble formation and gas-phase interactions in two-phase models can be more effectively isolated and analyzed, leading to a deeper understanding of multiphase transport in PEMECs. While single-phase models provide valuable insights into pressure distribution and liquid flow dynamics, they fail to account for critical phenomena such as gas evolution, bubble-induced transport resistance, and phase separation, which significantly influence PEMEC performance at high-current densities. Previous studies [
104,
108] have demonstrated that neglecting two-phase effects leads to overly optimistic performance predictions and an underestimation of transport limitations, particularly in the anode flow field. Consequently, two-phase models have become the preferred approach for accurately capturing real-world PEMEC behaviour, as they incorporate gas–liquid interactions, bubble dynamics, and their impact on mass transport [
99,
105]. Nevertheless, single-phase models remain useful as a computationally efficient first step for parameter studies and design optimization, particularly at lower current densities, where gas evolution is minimal. The findings presented in this study further reinforce the necessity of multiphase modelling for high-fidelity simulations while demonstrating that single-phase models still serve as a valuable starting point for fundamental hydrodynamic analyses in PEMEC research.
8. Conclusions, and Future Research Directions
This review critically examined the advancements and challenges associated with proton exchange membrane electrolyzer cells (PEMECs), focusing on materials, system configurations, and numerical modelling approaches. The core research question addressed in this study was how improvements in electrocatalysts, flow-field designs, and computational simulations contribute to enhancing PEMEC efficiency, durability, and scalability. Through extensive analysis, key areas for optimization were identified, particularly in electrocatalyst development, flow-field design, and computational modelling.
Significant progress has been made in the development of high-efficiency catalysts and advanced membrane materials, yet the dependence on costly noble metals, particularly iridium, remains a major limitation. Addressing this issue requires further exploration of non-precious metal alternatives and nanostructured catalyst architectures, which have shown promise in reducing costs while maintaining electrochemical activity [
114]. Additionally, innovative flow-field architectures, such as serpentine and interdigitated designs, have demonstrated the potential to enhance reactant distribution and gas management, reducing mass transport losses and improving overall cell performance [
78,
108]. However, the scalability of these configurations remains an ongoing research challenge, particularly in large-scale PEMEC applications.
Computational modelling has become a crucial tool for predicting and optimizing PEMEC performance by offering insights into gas evolution, bubble dynamics, and electrochemical reactions. The application of two-phase flow models, such as Eulerian and mixture-based approaches, has provided a deeper understanding of multiphase interactions within PEMECs, including pressure distribution, velocity fields, and phase separation effects. The comparative analysis in this study confirmed that multiphase modelling significantly improves accuracy over single-phase models, reinforcing its importance in PEMEC research. While single-phase models serve as a useful hydrodynamic baseline, their inability to capture oxygen evolution and gas-phase interactions highlights the critical role of two-phase modelling in accurately predicting real-world PEMEC behaviour.
However, discrepancies between numerical predictions and experimental results highlight the need for further model validation to improve predictive accuracy and practical applicability [
14,
15]. Refining two-phase flow simulations remains a critical research area, particularly for improving PEMEC efficiency at high-current densities. As demonstrated in this study, bubble formation and gas evolution significantly impact flow resistance, reactant transport, and local pressure variations. While existing mixture models provide reasonable accuracy, further improvements are needed to accurately capture phase separation, electrolyte transport, and membrane hydration under dynamic conditions, which directly affect PEMEC efficiency and operational stability. Future research should integrate high-fidelity CFD simulations with experimental datasets to bridge this gap and refine model parameters based on real-world operating conditions. Despite significant advancements, challenges remain in developing cost-effective and durable catalysts, scalable electrode configurations, and experimentally validated numerical models. To address these challenges, future research should explore alternative catalyst compositions, refine electrode architectures, and leverage machine-learning-assisted modelling to optimize reaction kinetics, enhance transport properties, and improve predictive accuracy in numerical simulations. Furthermore, future research should prioritize establishing standardized experimental validation frameworks to ensure consistency and comparability across numerical models. The integration of in situ diagnostics, such as operando imaging or spectroscopy, could also offer deeper insight into dynamic behaviours like bubble evolution, flooding, and degradation. Exploring hybrid systems that combine PEMECs with real-time energy management platforms or smart control algorithms will be crucial for optimizing performance under fluctuating renewable energy conditions. These efforts will help bridge the gap between theoretical modelling and practical application, accelerating the commercialization of PEMEC technologies.
Additionally, improving membrane electrode assembly durability and refining two-phase flow simulations will be critical to better capturing real-world operating conditions. The integration of PEMECs with renewable energy sources necessitates optimized system control strategies to accommodate variable power inputs and ensure stable hydrogen production [
1,
3]. Overcoming these challenges is essential for advancing PEMEC technology toward widespread industrial deployment, ensuring its role as a cornerstone in the hydrogen economy.
By overcoming these barriers, PEMECs can be further optimized for industrial-scale hydrogen production, accelerating the transition toward a sustainable and decarbonized energy future. Continued advancements in catalyst engineering, membrane durability, and multiphysics modelling will be crucial to achieving this goal. Furthermore, integrating PEMECs with intermittent renewable energy sources requires improved system control strategies to ensure stable and efficient operation under dynamic load conditions. This study provides a roadmap for future research, guiding the development of more efficient, durable, and economically viable PEM electrolyzers for green hydrogen production.