The study employs a sophisticated 3D, two-fluid, multiphase model. The core value of such a model is its ability to resolve the complex interactions between the gas and liquid water phases, particularly the transport of liquid water from the porous electrodes into the flow channels; a phenomenon critical for accurate performance prediction, especially at high current densities.
It is therefore highly concerning that the model relies on Assumption (3) in Section 2.2.1: “The amount of liquid water in channels is fixed as zero…”. This is implemented as a Dirichlet boundary condition (s_lq = 0) at the GDL-channel interface.
This boundary condition is unphysical for a multiphase model and creates a fundamental contradiction:
It severs the liquid transport pathway: The model solves for liquid water transport within the GDL/MPL/CL but artificially prevents this liquid from entering the channel. This is equivalent to imposing an infinitely efficient drying mechanism at the boundary, which does not exist in a real operating fuel cell.
It invalidates key conclusions: This assumption will systematically:
Under-predict liquid saturation in the regions of the porous media adjacent to the channels.
Over-predict reactant gas transport to the catalyst layers by reducing the simulated mass transport resistance.
Skew the water balance, affecting the predicted membrane water content and its distribution.
Render the model incapable of predicting channel flooding, a key failure mode.
Consequently, this calls into question several major findings:
The conclusion that the cell operates primarily in the “ohmic polarization region” (Section 3.2.3) may be an artifact of suppressed concentration overpotentials due to artificially facile water removal.
The quantitative values and uniformity indices for reactant concentration (Section 3.2.1) are likely overly optimistic.
The analysis of membrane dehydration (Section 3.2.4) is based on an unbalanced water budget, as the critical process of liquid water back-diffusion is misrepresented.
The quantified 3.75% performance loss due to coolant flow direction (Section 3.3) is suspect, as the cell’s sensitivity to temperature is intimately tied to water phase change, which is not physically modeled at the boundary.
Questions for the authors:
*What was the rationale for selecting a Dirichlet condition (s_lq = 0) instead of a more physical condition (e.g., a continuity condition or a convective flux) that would allow liquid water to enter the channel?
**Have you performed any sensitivity analysis to demonstrate that the key results (e.g., current density distribution, membrane water content) are robust and not qualitatively dependent on this specific, and highly restrictive, boundary condition?
***Given that this condition effectively prevents the simulation of a primary mass transport loss mechanism (channel and GDL flooding), how do you justify the model’s predictive capability for the high current density cases (1.2 – 1.8 A/cm²) where such phenomena are expected to be dominant?