Elsevier

Journal of Power Sources

Volume 331, 1 November 2016, Pages 462-474
Journal of Power Sources

Validation of pore network simulations of ex-situ water distributions in a gas diffusion layer of proton exchange membrane fuel cells with X-ray tomographic images

https://doi.org/10.1016/j.jpowsour.2016.09.076Get rights and content

Highlights

  • X-ray tomographic images of liquid water in a GDL are compared to simulations.

  • Pore network and full morphology simulations reproduced well the water distribution.

  • Pore network extraction parameters were benchmarked using full morphology.

Abstract

Understanding and modeling two-phase flows in the gas diffusion layer (GDL) of proton exchange membrane fuel cells are important in order to improve fuel cells performance. They are scientifically challenging because of the peculiarities of GDLs microstructures. In the present work, simulations on a pore network model are compared to X-ray tomographic images of water distributions during an ex-situ water invasion experiment. A method based on watershed segmentation was developed to extract a pore network from the 3D segmented image of the dry GDL. Pore network modeling and a full morphology model were then used to perform two-phase simulations and compared to the experimental data. The results show good agreement between experimental and simulated microscopic water distributions. Pore network extraction parameters were also benchmarked using the experimental data and results from full morphology simulations.

Introduction

Water management is a crucial aspect of Polymer Electrolyte Membrane Fuel Cells (PEMFC) operation. Water produced in excess by the oxygen reduction reaction needs to be removed from the cell to prevent flooding while the polymer membrane must stay well hydrated to conduct protons. Gas diffusion layers (GDL) are one of the electrodes porous layers [1] and a key component with respect to the water management. During fuel cell operation, liquid water can also appear in GDLs as a result of condensation [2]. Its effect is to impede the gas flux towards the catalyst layer where the electro-chemical reactions take place, decreasing the fuel cell performance. The biphasic, electrical, thermal and mechanical behaviors of GDLs must be therefore understood and well characterized in order to optimize fuel cell performance. In this respect, it is desirable to develop efficient and accurate numerical tools to simulate two-phase flows in GDLs.

Biphasic simulations in PEMFCs porous media have been widely investigated [3], [4]. The simulation methods can be divided into direct methods such as Lattice-Boltzmann and (mesoscale) geometry based methods like the full morphology [5] and pore networks models [6], [7], [8]. Mesoscale methods are preferred here in order to keep the computational cost low to allow for using images large enough to be representative of the GDL structure. It should be noted that the pore network models (PNM) have been frequently used to investigate two-phase flows in GDL, e.g. Refs. [2], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22]. Therefore, it is particularly important to confirm that PNM are well adapted to describe two-phase flows in fibrous materials, especially for the case of the capillarity driven regime that is expected to prevail in GDLs [9]. This is not obvious a priori because the microstructure of fibrous materials is quite different from the structure of granular materials or porous rocks mostly considered in previous applications of PNM, e.g. Refs. [23], [24].

It is usual to distinguish the structured pore networks from the unstructured ones. A structured pore network is constructed on a given lattice, typically a cubic lattice for 3D simulations. The pores are located on the nodes of the lattice and connected by narrower channels, also referred to as links, corresponding to the constrictions of the pore space. In a simple cubic network the channels are aligned with the three main directions of a Cartesian coordinate system. By, contrast, the unstructured networks offer the possibility to respect much more closely the local geometry of a given microstructure. Starting from a digital image of the “real” microstructure, the network is directly constructed from the image using appropriate numerical techniques. It is the approach taken in the present paper.

In this context, the main objective of the present paper is to investigate the capabilities of image based unstructured pore network and full morphology models to predict water distributions in fibrous materials such as the ones used in GDL. These simulation methods have already been successfully used to predict capillary pressure curves and saturation levels, e.g. Ref. [10] where a simple structured cubic network is used. The novelty in the present effort is to develop an image based unstructured pore network and to look at the three dimensional water distributions and not only to slice averaged saturation levels.

Extracting an unstructured pore network from a 3D digital image of a microstructure is described in a number of publications [25], [26], [27], [28], [29]. Most methods are based on medial axis analysis or the consideration of maximal balls. Several articles reviewed and discussed these methods [30], [31], [32], [33]. As far as we are concerned, we develop a method based on watershed segmentation. Watershed segmentation [34], [35] or similar methods [36], [37] are used in several articles as an image analysis tool to segment pores without doing pore network extraction [38], [39], [40]. Recently a few approaches use watershed segmentation or similar methods as part of a pore network extraction procedure [26], [28], [29], [41], [42]. The principle of watershed segmentation is to find constrictions, defined in a robust way as watershed lines. It has several advantages: it is a well-known tool in image analysis, it is robust and off-the-shell open source codes are available. It also offers an explicit degree of freedom regarding the pore merging, thanks to pore markers defined by the user.

In order to validate our pore network extraction procedure and to study the impact of the pore merging degree of freedom, we compare pore network simulations with experimental data and full morphology simulations. Full morphology simulations are in fact suggested as a useful benchmark to test pore network extraction procedures.

The experimental data are 3D microscopic liquid water distributions from water injection experiments obtained by X-ray tomographic microscopy [11]. X-ray tomography has been used in several works to visualize liquid water in GDLs [23], [24], [43], [44], [45], [46], [47], [48], [49]. Liquid water distributions in GDL obtained by X-ray tomography have also been used in some previous works in conjunction with numerical simulations, e.g. Ref. [9]. However, the present work is the first one, to the best of our knowledge, to propose a detailed comparison between PNM simulations and X-ray tomography images of the liquid water distribution in a GDL.

The aim of this article is thus to assess the predictability of pore network and full morphology models to simulate capillarity dominated biphasic flows in GDLs and to study a pore network extraction procedure based on watershed segmentation.

Regarding two-phase flows in GDL, one can distinguish in situ two-phase flows from ex-situ two phase flows. In situ refers to GDL in an operating fuel cell whereas ex situ refers to GDL typically used in characterization experiments involving the sole GDL (as opposed to the GDLs in the membrane electrode assembly (MEA) for the in-situ case). A typical objective of ex-situ experiments is to characterize the capillary pressure curve, an important parameter of the porous media classical two-phase flow model. In situ two-phase flows are a priori more complex than the ones observed in typical ex-situ experiments because of the coupling with heat transfer and the significance of phase change phenomena. Also, obtaining images of in situ liquid distribution from X-ray microscopy is more challenging. For these reasons, it is natural to first develop comparisons between simulations and experiments for the simpler ex-situ conditions. In this paper, an ex-situ configuration is considered.

Section snippets

Material

Gas diffusion layer material of the type SGL™ 24BA (SGL Carbon, D) is chosen for this study. SGL™ gas diffusion layer materials are composed from carbon fibers and a micro-porous carbonaceous binder. The pore size of the binder is typically less than 1 μm. SGL™ 24 has an uncompressed thickness of 190 μm and an uncompressed porosity of 84%. The suffix “BA” means that the material is hydrophobized with 5% (w/w) of PTFE and that there is no microporous layer (MPL).

X-ray tomographic microscopy

Experimental water distributions

Benchmarking network extraction parameters against full morphology results

As mentioned before pore network extraction from an image is a complex procedure which involves making a choice regarding the definition of pores. We discuss in this section the consequences of the choice of the pore merging parameter. We also suggest that comparing pore network two-phase simulations with full morphology simulations can help verify that the extracted pore network is valid.

The extracted pore network is not unique because the markers choice gives a degree of freedom on pore

Discussion

The agreement between the experiments and the simulations is considered as good but is not perfect. Possible model improvements are briefly discussed in this section.

Uncertainties regarding the inlet conditions are considered as a major cause of discrepancies between the simulations and the experiment. In this work we imposed inlet boundary conditions considering the hydrophilic membrane as a water reservoir. Accordingly, water enters the GDL where the fibers in contact with the membrane

Conclusion

Pore network and full morphology simulations of quasi-static two-phase flow in a gas diffusion layer were performed. A pore network extraction method based on watershed segmentation was developed. The geometry of the tomographic image was analyzed to identify pores and constrictions. A comparison between full morphology and pore network two-phase simulations was performed. This validated the extraction procedure. The extraction computations are fast (40min for a 700 million voxels image) and

Acknowledgements

The authors gratefully acknowledge the funding from the EU project IMPALA (“IMprove Pemfc with Advanced water management and gas diffusion Layers for Automotive application”, project number: 303446) within the Fuel Cells and Hydrogen Joint Undertaking (FCHJU), and SGL Carbon for the supply with gas diffusion layer materials.

References (59)

  • P.K. Sinha et al.

    Liquid water transport in a mixed-wet gas diffusion layer of a polymer electrolyte fuel cell

    Chem. Eng. Sci.

    (2008)
  • K.J. Lee et al.

    Liquid water distribution in hydrophobic gas-diffusion layers with interconnect rib geometry: an invasion-percolation pore network analysis

    Int. J. Hydrogen Energy

    (2014)
  • K.J. Lee et al.

    Steady saturation distribution in hydrophobic gas-diffusion layers of polymer electrolyte membrane fuel cells: a pore-network study

    J. Power Sources

    (2010)
  • K.-J. Lee et al.

    Pore-network analysis of two-phase water transport in gas diffusion layers of polymer electrolyte membrane fuel cells

    Electrochim. Acta

    (2009)
  • P.K. Sinha et al.

    Pore-network modeling of liquid water transport in gas diffusion layer of a polymer electrolyte fuel cell

    Electrochim. Acta

    (2007)
  • E.F. Médici et al.

    Evaporation, two phase flow, and thermal transport in porous media with application to low-temperature fuel cells

    Int. J. Heat. Mass Transf.

    (2013)
  • P.A. García-Salaberri et al.

    Effective diffusivity in partially-saturated carbon-fiber gas diffusion layers: effect of through-plane saturation distribution

    Int. J. Heat. Mass Transf.

    (2015)
  • I.V. Zenyuk et al.

    Electrochemistry Communications Probing water distribution in compressed fuel-cell gas-diffusion layers using X-ray computed tomography

    Electrochem. Commun.

    (2015)
  • A. Rabbani et al.

    An automated simple algorithm for realistic pore network extraction from micro-tomography images

    J. Pet. Sci. Eng.

    (2014)
  • D. Wildenschild et al.

    X-ray imaging and analysis techniques for quantifying pore-scale structure and processes in subsurface porous medium systems

    Adv. Water Resour.

    (2013)
  • K.E. Thompson et al.

    Quantitative computer reconstruction of particulate materials from microtomography images

    Powder Technol.

    (2006)
  • R.C. Atwood et al.

    Analysis of pore interconnectivity in bioactive glass foams using X-ray microtomography

    Scr. Mater

    (2004)
  • Bazylak

    Liquid water visualization in PEM fuel cells: a review

    Int. J. Hydrogen Energy

    (2009)
  • R. Flückiger et al.

    Investigation of liquid water in gas diffusion layers of polymer electrolyte fuel cells using X-ray tomographic microscopy

    Electrochim. Acta

    (2011)
  • H. Markötter et al.

    Investigation of 3D water transport paths in gas diffusion layers by combined in-situ synchrotron X-ray radiography and tomography

    Electrochem. Commun.

    (2011)
  • J. Pauchet et al.

    Performance loss of proton exchange membrane fuel cell due to hydrophobicity loss in gas diffusion layer: analysis by multiscale approach combining pore network and performance modelling

    Int. J. Hydrogen Energy

    (2012)
  • P.A. García-Salaberri et al.

    Effective diffusivity in partially-saturated carbon-fiber gas diffusion layers: effect of local saturation and application to macroscopic continuum models

    J. Power Sources

    (2015)
  • W.B. Lindquist

    The geometry of primary drainage

    J. Colloid Interface Sci.

    (2006)
  • B. Straubhaar et al.

    Water transport in gas diffusion layer of a polymer electrolyte fuel cell in the presence of a temperature gradient. Phase change effect

    Int. J. Hydrogen Energy

    (2015)
  • Cited by (49)

    • Experimental and simulation analysis of liquid capillary fingering process in the gas diffusion layer

      2023, Journal of Power Sources
      Citation Excerpt :

      The experimental method is useful for revealing the phenomenon of liquid water capillary fingering within the GDL [21], but the simulation method for simulating liquid water movement in the GDL is useful for solving the microscopic characteristics of two-phase transport, which is a useful tool for control and study of geometric and other parameters provide the opportunity to address fundamental causal relationships that are not explained by experimental studies [22]. The volume of fluid (VOF) [23–40], pore-network modeling (PNM) [41–61], and lattice Boltzmann method (LBM) [62–93] are the main simulation methods for simulating liquid phase movement in the GDL, and the specific study contents are listed in Table 1. The main factors impacting the internal liquid water movement for GDL are GDL thickness [53,54,72], fiber structure [30,57,79,92], pore distribution [26,28,45,46,69,83], compression deformation [26,27,32,34,49,59,66,67,70,88], PTFE content and distribution (wettability) [26,31,36,39,40,43,44,55,56,60,64,65,67,68,73,76,86,87,89–93].

    View all citing articles on Scopus
    View full text