1. On the influence of gdl compression on the performance of liquid-feed DMFCs under low cathode flooding conditions P. A. Garc í a-Salaberri, M. Vera Dept. of Thermal and Fluids Engineering Universidad Carlos III de Madrid http://fluidos.uc3m.es

2.  Motivation  Anode DMFC  Assembly Compression  Numerical Model  Implementation  Cases Study  Discussion of Results  Comparison with Experimental Data  Under-Rib Convection  GDL/Rib Electrical Contact Resistance  Conclusions & Future Work P.A. García-Salaberri/M. Vera – ModVal 111 Outline

3. Anode DMFC       [1] [2] P.A. García-Salaberri/M. Vera – ModVal 112  Rate limiting electrode under low cathode flooding conditions

4. [3] Assembly Compression  Most Flexible Element Concentrated in the GDL  Rib/Channel Pattern Uneven Distribution Prevent Reactants Leakage & Good Component Contact

5. Reduces GDL/BP Contact Resistance Two Opposite Effects Hinders Convective & Diffusive Transport through the GDL An Optimum Compression Exists Assembly Compression  Most Flexible Element Concentrated in the GDL  Rib/Channel Pattern Uneven Distribution Prevent Reactants Leakage & Good Component Contact

6. Step 1: FEM Model to simulate Assembly Process [4]  Non-Linear Orthotropic Mechanical Properties (TORAY ® Carbon Paper) Step 2: 2D/1D Finite-Volume based CFD Model [5,6]  Empirical Anisotropic Transport Properties (TORAY ® Carbon Paper) Diffusivity, Permeability, Capillary Pressure and Electrical Conductivity GDL/Rib Electrical Contact Resistance f(ε) f(p c ) P.A. García-Salaberri/M. Vera – ModVal 115 Numerical Implementation

7.  Two Case Studies: Serpentine vs. Parallel CFD Computational Domain [7] Anode GDL (2D Model)  Two-Phase  Non-Uniform Transp. Prop. & GDL/Rib R c CLs, MPLs, PEM & Cathode GDL (1D Model)  Locally Coupled to the 2D Anode GDL Model

8. Bipolar Plate Material  Two cases study: Graphite Stainless steel  Two Case Studies: Graphite vs. Stainless steel Metallic Much Higher R c

9. Influence of Compression Increases as the Cell Voltage is Lower  Cell Performance is compared with Experimental Data for a Graphite Serpentine Flow Field (Ge, 2005 [8]) Optimal Compression Ratio I vs. CR (Different Voltages) OPTIMAL COMPRESSION Trade-off between Ohmic Losses at Low CRs & Mass-Transport Limitations at High CRs Graphite: CR ≈ 5-10%

10. Serpentine Vs. Parallel Serpentine Parallel The Detrimental Effect of Compression is Much Higher for the Serpentine Flow Field Why This Difference?

11. Peclet Number  Under-Rib Convection can be an Important Transport Mechanism in the Anode of Liquid-Feed DMFCs Fixed Values

12. Parallel Rib/Channel Current Rib (-) Channel (--)Velocity Magnitude  Parallel: Compression affects only I rib due to the dominant role of Diffusion Channel unaffected Rib

13.  Serpentine: I rib & I ch are affected because the Decrease of the effect of URC with Compression in addition to diffusion; Higher Influence on Performance Serpentine Rib/Channel Current Rib (-) Channel (--)Velocity Magnitude

14.  Serpentine: I rib & I ch are affected because the Decrease of the effect of URC with Compression in addition to diffusion; Higher Influence on Performance  I rib > I ch : In Agreement with Experimental Data (Almheiri & Liu, 2014 [9]) Serpentine Rib/Channel Current Rib (-) Channel (--)Velocity Magnitude High Influence URC Rib & Channel

15. Graphite Vs. Metallic BP Graphite Metallic The CR requiered to Maximize the Cell Performance is Much Higher for Metallic Plates Graphite: CR ≈ 5-10% [8] Metallic: CR ≈ 30% [10]

16. Conclusions Future Work 1.Under-rib convection seems to be a significant mass transport mechanism in the anode of DMFCs and makes the detrimental effect of compression more pronounced. 2.The optimal compression ratio is strongly influenced by the bipolar plate material. 1.Two-phase phenomena at the cathode (LBM). 2.Extension to a fully-3D Model. 3.GDL/CL electrical contact resistance. P.A. García-Salaberri/M. Vera – ModVal 1115

17. [1]Ph. Krügera, H. Markötter, J. Haußmann, M. Klages, T. Arlt, J. Banhart, Ch. Hartnig, I. Manke, J. Scholta, Power Sources 2011; 196:5250–5255. [2]http://www.ualberta.ca (University of Alberta, Canada).http://www.ualberta.ca [3]Z.Y. Su, C.T. Liu, H.P. Chang, C.H. Li, K.J. Huang, P.C. Sui, Power Sources 2008; 183:182–192. [4]P.A. García-Salaberri, M. Vera, R. Zaera, Int. J. Hydrogen Energy 2011; 36:11856–11870. [5]M. Vera, Power Sources 2007; 171:763–777. [6]P.A. García-Salaberri, M. Vera, I. Iglesias, Power Sources 2014; 246:239–252. [7]T.S. Zhao, C. Xu, R. Chen, W.W. Yang, Prog. Energ. Combust. 2009; 35:275–292. [8]J. Ge, Doctoral dissertation, University of Miami, USA, 2005. [9]S. Almheiri, H. Liu, Power Sources 2014; 246:899–905. [10]Z.Y. Yuan, Y.F. Zhang, W.T. Fu, Z.P. Li, X.W. Liu, Fuel Cells 2013; 13:794–803. P.A. García-Salaberri/M. Vera – ModVal 1116 References

18. 2D Model

19. 2D Model (Cont.)  Phase-Change Source Term  Capillary Pressure

20. 1D Model

21. Porosity Field GDL/Rib Contact Pressure FEM Model (Step 1)

22. Methanol Crossover  Parasitic Current Decreases as the GDL is Compressed ↑CR → ↓C ml,acl In qualitative agreement with García-Díaz et al., 2012 [9]: I p = 185 mA/cm 2 @ 480 Kpa I p = 63 mA/cm 2 @ 1380 KPa

23. RibChannel RibChannel GDL/CL Contact Resistance Low Pressure High Resistance ChannelRib Under-Rib Convection  The Large Influence of Under-Rib Convection is consistent with Experimental Data (Almheiri & Liu, 2014 [11]) The GDL/CL Electrical Contact Resistance can Accentuate the Differences in the Current Distribution Numerical Results Experimental Data [11]

24. Peclet Therm. Eq.