This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/3.0/).
A comprehensive mathematical model of the performance of the cathodesupported solid oxide fuel cell (SOFC) with syngas fuel is presented. The model couples the intricate interdependency between the ionic conduction, electronic conduction, gas transport, the electrochemical reaction processes in the functional layers and on the electrode/electrolyte interfaces, methane steam reforming (MSR) and the water gas shift reaction (WGSR). The validity of the mathematical model is demonstrated by the excellent agreement between the numerical and experimental IV curves. The effect of anode rib width and cathode rib width on gas diffusion and cell performance is examined. The results show conclusively that the cell performance is strongly influenced by the rib width. Furthermore, the anode optimal rib width is smaller than that for cathode, which is contrary to anodesupported SOFC. Finally, the formulae for the anode and cathode optimal rib width are given, which provide an easy to use guidance for the broad SOFC engineering community.
Solid oxide fuel cell (SOFC) is considered to be one of the more promising new energy technologies [1–3]. In the last decade, there has been a great deal of research aiming to increase the performance of a single SOFC by looking for new electrodes and electrolyte materials, optimizing electrode microstructure or improving electrode and electrolyte preparation techniques [4–10] and so on. Currently, a single SOFC may produce a power density of 2.45 W·cm^{−2} at 800 °C and 0.7 V [11]. However, when multiple cells are connected by an interconnector (bipolar plates) to form a SOFC stack, the performance of SOFC at the stack level decreases by 50% or more compared to a single level SOFC [12,13]. There are many factors limiting the performance of SOFC as part of a stack. One of the primary factors is the additional losses caused by the interconnector geometry. Herein, additional losses are created by two main aspects: (i) the interconnector used in a stack can cause a significant contact resistance between the electrode and the interconnector; (ii) the interconnector limits gas diffusion in electrode.
Small grooves in interconnectors are defined as channels, which are commonly used to carry the fuel and air gas flow. The ribs, which separate and define the channels, make direct contact with the electrodes. In designing the layer architecture, there is a tradeoff that must be considered between the rib and channel sizes. On one hand, wider ribs and ribs covering a bigger fraction of the cell area may reduce the interface resistance to current flow by increasing the electrodeinterconnect contact area and reducing the current path through the possibly high resistance electrode material. Hence, such ribs will give a better conduction of the electrical current and reduce ohmic losses. On the other hand, the chemical species do not diffuse as well underneath wide ribs. Narrow ribs are needed to facilitate more uniform distribution of reactive gases across the area of the electrolyte surface and thus to promote electrochemical performance. The implications of the tradeoff to the cell performance can be very significant. It was reported that the power density was only 0.76 W·cm^{−2} at 0.7 V operating voltage for the combination of 2.6 mm channels and 2.4 mm ribs at the cathode side, while that for 4 mm channels and 1 mm ribs was 1.03 W·cm^{−2} [12]. Clearly, optimizing rib and channel sizes are of the utmost importance to realize the full potential of SOFCs.
Several authors have investigated the impact of the rib size on SOFC performance. Tanner and Virkar [14] proposed a simplified model to examine the effect of the interconnect design on the effective ohmic resistance of the membraneelectrode assembly in an SOFC stack. Lin et al. [15] provided a phenomenological model and analytical expressions to estimate the rib effects on the concentration and ohimc polarizations of anodesupported SOFC stacks. Ji et al. [16] showed that the terminal output of a SOFC stack depended strongly on the contact resistance. Jeon et al. [17] proposed a detailed microstructural model and examined systematically the influence of the rib width, pitch width and the contact area specific resistance (ASR_{contact}) on the stackcell performance. Noh et al. [18] showed that by modifying the current collection configuration, cell performance was improved by more than 30%. Based on systematic examination of the stack cell performance as a function of various influencing factors, Liu et al. [19] proposed a simple expression for the optimal rib width as a function of the pitch width and ASR_{contact} that was independent of the porosity, layer thickness and conductivity of the electrode and easy for SOFC engineers to use.
Unfortunately, the existing numerical studies on the rib design optimization have been carried out with two assumptions: (i) equal ASR_{contact} for the anodeinterconnect and cathodeinterconnect interfaces and (ii) equal width of the optimal anode and cathode rib widths. However, as demonstrated experimentally by Kornely et al. [12] and Dey et al. [20], ASR_{contact} for the anodeinterconnect and cathodeinterconnect interfaces can be very different. Therefore, the optimal anode rib width isn't equivalent to the optimal cathode rib width as the optimal rib width depends on ASR_{contact} [19,21]. Even if ASR_{contact} for the anodeinterconnect and cathodeinterconnect interfaces are equal, the optimal rib widths for anode and cathode should be different. Previous studies have already shown very different effects of the cathode and anode ribs on the mass transport in the electrodes. For anodesupported SOFC, the anode gas transport was only mildly affected by the anode rib, which is benefit from thick anode, while an oxygen depletion zone of 0.46 mm was found with a cathode rib width of only 0.8 mm due to thin cathode thickness limiting the oxygen diffusion to the area under rib [19]. On the contrary, for cathodesupported SOFCs, the anode rib has an important influence on the anode gas transport; the minimum hydrogen concentration under anode rib is only about one third of that under anode channel [22]. The existing experimental and theoretical findings indicate clearly that the optimal design of the stack cell should use different widths for the anode and the cathode ribs.
We have developed a comprehensive two dimensional model for an anodesupported SOFC with hydrogen as fuel and optimized the anode and cathode rib widths without the above two assumptions [21]. The numerical results show that the output current density depends strongly on the rib widths and the optimal rib widths for the anode and the cathode are quite different. Moreover, the optimal widths for both the anode and cathode ribs are found to be only sensitive to the contact resistance and the pitch width. Finally, the formulae for optimal rib width are given. However, these formulae are only valid for an anodesupported SOFC. The effects of the cathode and anode ribs on the mass transport for cathodesupported SOFC are different to that for anodesupported SOFC, which are demonstrated by references [19,22].
On one hand, from the aspect of anode concentration loss, the effect of anode rib width on the anode concentration loss is rather limited for anodesupported SOFCs due to the thick anode providing a wide alleyway allowing fuel to penetrate under the ribs. However, it is a quite different situation for a cathodesupported SOFC. The anode rib width has a serious influence on anode concentration loss since thin anode of cathodesupported SOFC limits fuel diffusion to the area under rib. Consequently, the optimal anode rib width for cathodesupported SOFC should be relatively small in comparison with that for anodesupported SOFC in order to reduce anode concentration loss. On the other hand, from aspect of cathode concentration loss, the thin cathode of anodesupported SOFC limits the oxygen diffusion to the area under rib and increases cathode concentration loss, while a thick cathode in a cathodesupported SOFC promotes oxygen diffusion to the area under rib. Therefore, the optimal cathode rib width of an anodesupported SOFC should be smaller than that of cathodesupported SOFC. As a result, the formulae for optimal rib widths of anodesupported SOFC given in our previous paper [21] can't be applied to a cathodesupported SOFC. There is an urgent need to investigate the optimal rib width for cathodesupported SOFCs. Due to such a research need, the objective of this work is to study the optimal rib width of cathodesupported SOFCs to provide guidelines for cathodesupported SOFC fabrication.
Compared with cathodesupported SOFCs, anodesupported SOFCs have attracted more attention. However, cathodesupported SOFCs show various advantages over anodesupported SOFCs. One is the low cost of cathode supporting materials such as strontiumdoped lanthanum manganese. Moreover, when operating on hydrocarbon fuels with low steamtocarbon ratio, a relatively thin anode would prevent it from depositing carbon. In addition, a thin anode layer would also provide benefits in terms of tolerance to volume contraction/expansion resulting from the accidental anode redox cycles.
In this paper, the previously developed model [21] is modified to investigate the performance of a cathodesupported SOFC. Different from the previous study with hydrogen as fuel, syngas is used, which consists of methane (CH_{4}), steam (H_{2}O), carbon monoxide (CO), hydrogen (H_{2}) and carbon dioxide (CO_{2}). The methane steam reforming and water gas shift reaction are taken into account in this model. The optimal anode and cathode rib widths for different ASR_{contact} are obtained and formulized to provide an easytouse guidance for designing the ribchannel layout.
ModelGeometric Model
Figure 1a shows a repeating cell unit of a cathodesupported cell stack. As demonstrated before [19,23], a 2D model is equivalent to a 3D model when discussing the effects of rib widths on the fuel cell performance as they provide essentially the same results. Therefore, a 2D model (Figure 1b) is used here. Due to symmetry, we select half of the repeating unit of stack as our computational domain, which is shown by the red line in Figure 1b. The computational domain consists of five layers: (1) an anode current collector layer (ACCL) with relatively large Ni and YSZparticles and high porosity to decrease gases transport resistance; (2) an anode functional layer (AFL) with fine Ni and YSZparticles and low porosity to reduce the active losses by affording abundant three phase boundaries (TPBs); (3) an electrolyte layer; (4) a cathode functional layer (CFL) with fine LSM and YSZparticles and low porosity and (5) a cathode support layer (CSL) with relatively large LSMparticles and high porosity. In Figure 1b, d_{rib} is one half of the interconnect rib width and d_{pitch} is the pitch width (the sum of one half of the rib width, d_{rib}, and one half of the channel width, d_{channel}).
Modeling Mass Transport
The gases transport in SOFC porous electrodes is complex and includes three distinct masstransfer mechanisms: Knudsen diffusion, molecular diffusion, as well as viscous flow. Thus, the dusty gas model (DGM) is required to estimate accurately the transport of gases in SOFC porous electrodes. However, the fluxes of different species are coupled with one another in the DGM and further coupling of the DGM with the mass conservation equations and the bulk chemical reactions is cumbersome to perform. In our previous study [24], the DGM in the form of an Fick's model (DGMFM) was derived from the DGM, which can give an explicit analytical expression for the flux of each species. Thus, DGMFM is adopted to describe the gases transport in SOFC porous electrodes. The molar mass conservation equation of species i is given by:
∇⋅Ni={0in CSL ACCLRiin AFL CFLwhere Ni is the molar flux of species i; R_{i} is the molar rate of production (+) or consumption (−) of species i due to chemical/electrochemical reactions. N_{i} can be calculated by the DGMFM:
Ni=−Di¯∇ci−ciki¯μ∇pDi¯=1−xi∑j≠ixjDijeffDiKeff(1−xi∑j≠ixjDijeff+xi1−xi∑j≠ixjDjKeff+(1−xi)DiKeff)−1ki¯=k+μ1−xi∑j≠lxjDjKeffDiKeff(RTctot(1−xi∑j≠lxjDijeff+xi1−xi∑j≠ixjDjKeff+(1−xi)DiKeff))−1where D̄i and K̄i are the equivalent diffusion coefficient of species i and the equivalent permeability coefficient, respectively; c_{i} is the molar concentration of species i;
ctot(=∑jcj) is the total molar concentration of the mixture; x_{i}(=c_{i} / c_{tot}) is the molar fraction of species i ; p is the total gas pressure; μ is the viscosity coefficient;
DiKeff is the effective Knudsen diffusion coefficient of species i and
Dijeff is the effective binary diffusion coefficient; R is the universal gas constant; T is the absolute temperature; k is the permeability coefficient.
Permeability coefficient k and the viscosity coefficient of fluid μ can be calculated by the following expressions [25]:
k=ɛ345(1−ɛ)2rel2μ=∑i=1mxiμi∑j=1mxjΦijΦij=18(1+MiMj)−0.5[1+(μiμj)0.5(MiMj)0.25]2where ε is the porosity; r_{el} is the mean electronic conducting particle's radii; m is the number of components in the fluid; Φ_{ij} is a dimensionless number; M_{i} (M_{j}) is the molar mass of species i (j); μ_{i} (μ_{j}) is the viscosity coefficient of species i (j) ; which can be derived from Sutherland's law:
μi=μrefTref+crefT+cref(TTref)1.5where T_{ref} is the reference temperature in Kelvin; μ_{ref} is the reference viscosity (Pa·s) at reference temperature T_{ref}; T is the temperature in Kelvin and c_{ref} is the Sutherland's constant.
Dijeff and
DiKeff can be evaluated by the following equations [26–28]:
Dijeff=ɛτ3.198×10−8T1.75p(νi1/3+νj1/3)2(1Mi+1Mj)0.5DiKeff=ɛτ23rg8RTπMirg=23ɛ1−ɛ1ϕel/rel+ϕio/riowhere τ is the tortuosity factor; r_{g} is the pore radius; ν_{i} is the diffusion volume of species i; r_{io} (r_{el}) is the mean electronic (ionic) conducting particle's radii; φ_{el} (φ_{io}) the volume fraction of the electronic (ionic) conducting particles in the composite electrode material.
In this study, the fuel consists of methane (CH_{4}), steam (H_{2}O), carbon monoxide (CO), hydrogen (H_{2}) and carbon dioxide (CO_{2}). MSR and WGSR are taken into account in porous anode, described respectively as:
CH4+H2O↔CO+3H2CO+H2O↔CO2+H2
The rate of MSR occurring on the surface of the catalyst particle can be estimated by following formula [29–31]:
Rr=SANi[0.0636T2exp(−27063T)cCH4cH2O−3.7×10−23T4exp(−232.78T)cCOcH23]where
SANi is the volumetric active surface area of Ni particles (m^{2}·m^{−3}).
The shift reaction occurs wherever the gas is present, which is considered to be at equilibrium in the porous anode due to the high working temperatures. On the basis of experimental data, a shift reaction rate expression can be formulated as [29–31]:
RS=ɛ(1.199T2exp(−12509T)cCOcH2O−67.7T2exp(−16909T)cCO2cH2)
Modeling Electrical Conduction
The electronic charge transfer in the electrodes (ACCL, AFL, CFL and CSL) is governed by:
∇⋅iel=∇⋅(−σeleff∇φel)={0in ACCL or CSL−Scurrentin AFLScurrentin CFL∇⋅iio=∇⋅(−σioeff∇φio)={Scurrentin AFL0in electrolyte−Scurrentin CFLwhere i_{el} (i_{io}) is the electronic (ionic) current density vector; φ_{el} (φ_{io}) is the electronic (ionic) potential; S_{current} the current source.
S_{current} may be calculated by the equation below:
Scurrent=λTPBitranswhere λ_{TPB} is the TPB length per unit volume (m^{−2}) or unit area (m^{−1}); i_{trans} is the transfer current based on per TPBs length or area.
i_{trans} is the local charge transfer current density and can be calculated as according to the empirical ButlerVolmer equation.
For Ni/YSZ AFL TPBs, i_{trans} is derived as [32,33]:
itransan=irefanexp(−EH2R(1T−1Tref))(pH2TPBpH2OTPBpH20pH2O0)[exp(2αfanFRTηactan)−exp(−2βranFRTηactan)]
For LSM/YSZ CFL TPBs, i_{trans} is derived as [32,33]:
itransca=irefcaexp(−EO2R(1T−1Tref))(pO2TPBpO20)0.25[exp(2αfcaFRTηactca)−exp(−2βrcaFRTηactca)]where α_{f} and β_{r} are the forward and reverse reaction symmetric factor, respectively; E_{H2} and E_{O2} are the activation energies for the anode and cathode electrochemical reactions, respectively;
irefan and
irefca are often deduced from experiments or assigned empirically at the reference temperature of T_{ref};
pH20 and
pH2O0 are the partial pressure of H_{2} and the partial pressure of H_{2}O at the fuel channel/anode interface, respectively;
pO20 is the partial pressure of O_{2} at the air channel/cathode interface;
pH2TPB and
pH2OTPB are the partial pressure of H_{2} and the partial pressure of H_{2}O at the anode TPBs, respectively and
pO2TPB is the partial pressure of O_{2} at the cathode TPBs. Here
ηactan and
ηactca are anode and cathode activation polarization respectively, defined as:
ηactan=φel−φio−ηconcan=φel−φio−RT2Fln(pH20pH2O0pH2OTPBpH2TPB)ηactca=φio−φel−ηconcca=φio−φel−RT4Fln(pO20pO2TPB)where F is the Faraday constant;
ηconcan(ηconcca) the anode (cathode) concentration polarization.
The volumespecific TPB length λ_{TPB} can be written as [21,34]:
λTPBV=2πmin(rel,rio)sin(θ2)nnioZio−elpelpion=1−ɛ43πrel3(nel+(1−nel)γ3)γ=riorelZio−el=Z2(1+rio2rel2)ϕel/relϕel/rel+ϕio/rioPk=(1−(3.764−Zk,k2)2.5)0.4Zk,k=Zϕk/rkϕel/rel+ϕio/rio
The areaspecific TPB length λ_{TPB} can be written as [21,34]:
λTPBA=4πrelsin(θ2)nrelnelpelwhere θ is the contact angle between particles (θ is assigned as 30° [35]); Z_{io−el} is the average coordination number between ionic conductor particles and electronic conductor particles (the number of contacts between a ionic conductor particle and electronic conductor particles ); n_{el} (n_{io}) is the number fraction of electronic (ionic) particles n_{io}=1−n_{el} ; γ is the size ratio of ionic particles to electronic particles and n is the number of particles per unit volume. Because n_{el} or n_{io} is rather difficult to measure, it is necessary to use volume fraction instead of number fraction. The relation between number fraction and volume fraction can be described as:
nel=ϕelγ31−ϕel+γ3ϕel
The effective electronic and ionic conductivities of composite electrode are strong relied on microstructures like particles size, porosity, volume fractions. The coordination number theory based percolation micromodel reveals the relationship between the effective electrode properties and the microstructure parameters [36–39]. Considering a binary system with random packing of spheres, corresponding to the electrodeparticles (denoted as el) and electrolyteparticles (denoted as io), The effective electric conductivity of kphase is estimated as [40]:
σkeff=σk0(ϕk−ϕkt1+ɛ/(1−ɛ)−ϕkt)2where
σk0 is the electric conductivity of kmaterial in dense solid; φ_{k} the volume fractions of kparticles in the solid structure;
ϕkt is the threshold volume fraction of kparticles, which is determined by [35,38]:
Zϕelt/relϕelt/rel+(1−ϕelt)/rio=1.764Zϕiot/rio(1−ϕiot)/rel+ϕiot/rio=1.764where Z is the average coordination number for each particle. Z is set to 6 for a random packing of spheres [35,41,42].
σk0 for Ni, LSM and YSZ may be estimated as [17,43]:
σNi0=3.27×106−1065.3TσLSM0=4.2×107Texp(−1150T)σYSZ0=6.25×104exp(−10300T)
Boundary Conditions (BCs)
Proper settings of the boundary conditions are required to solve these coupled partial differential equations correctly. The boundary settings for the mass transport equations are shown in Table 1. The boundary settings for the electronic and ionic charge transfer equations are shown in Table 2. The molar concentrations at the ACCL/channel interface or CSL/channel interface, C^{0}, are related to the molar fractions by the ideal gas equation of state. The total gas pressure at the electrode/channel interface is set at 1 atm. The contact resistance is set on the interface between the interconnect ribs and the electrodes. That is, local current densities cross the rib/anode (i_{rib→ACCL}) and the cathode/rib (i_{CSL→rib}) interfaces are determined by:
irib→ACCL=φe,rib/ACCL−φe,ACCL/ribASRcontactaniCSL→rib=φe,CSL/rib−φe,rib/CSLASRcontactcawhere
ASRcontactan(ASRcontactca) is the area specific contact resistance at the ribanode (cathode) interface; φ_{e,rib/ACCL} (φ_{e,ACCL/rib}) is the electric potential at the rib (anode) side of the anoderib boundary; φ_{e,CSL/rib} (φ_{e,rib/CSL}) is the electric potential at the cathode (rib) side of the cathoderib boundary. It should be pointed out that the contact resistance is an effective model parameter that may come from various origins such as the oxide scale formation of the interconnect material and the loose contact between the cell components.
Numerical Solution
The finite element commercial software COMSOL MULTIPHSICS^{®} (COMSOL, Inc., Burlington, MA, USA) was used in the present study to solve the required partial differential equations (PDEs) of the gas transport equation, ionic conduction equation and electronic conduction equation. The discretized PDEs were solved by using the direct solver with a relative convergence tolerance of 1 × 10^{−6}. Structured mesh elements were used and consisted of 720 rectangles with 10,474 degrees of freedom.
Numerical Validation
Nickel or silver meshes are the typical current collectors used for single cell testing. The meshes may be regarded as interconnect ribs with small pitch widths. A pitch width of 0.2 mm and a rib width of 0.05 mm corresponding to the experimental description were used here for the numerical model validation. Model parameters are listed in Tables 3–5.The simulated IV relations are compared with the experimental results of Wang et al. [44] in Figure 2. As shown in Figure 2, the theoretical results agree with the experimental measurements very well. Hence, the numerical model is used in the following to optimize interconnect rib. Moreover, except for those specified explicitly for each testing case, the model parameters listed in Tables 3–5 are used throughout the paper.
Results and DiscussionThe Effect of Rib Width on the Cell Performance
In order to investigate the influence of the rib width on the cell performance, four models were established with the same settings as described above except the size of interconnector geometry structure. The pitch width (d_{pitch}), the anode rib width
(driban) and the cathode rib width
(dribca) for four models are listed in Table 6.
Figure 3 shows the H_{2} concentration distribution in the anode of a cathodesupported SOFC. It is obvious that the H_{2} concentration gradients in the vertical direction are small, which is beneficial for a thin anode. However, the H_{2} concentration gradient in the horizontal direction is quite large. This is because the anode thickness is very thin which limits the H_{2} diffusion to the area under rib. Comparing Figure 3a with Figure 3b, it isn't difficult to find that the smallest H_{2} concentration is only 0.56 mol m^{3} for the CellB, which is only about one tenth of that for the CellA. Thus, the anode rib width has a significant impact on the fuel diffusion for cathodesupported SOFC, which is quite different to that for anodesupported SOFC [21]. For CellB, the output current densities for V_{op} of 0.7, 0.6 and 0.5 V are respectively 0.4166, 0.5099 and 0.5992 A·cm^{−2}, as shown in Table 7. However, the output current densities of CellA are 35.1%, 36.2% and 37.5% more than that of CellB for V_{op} of 0.7,0.6 and 0.5 V, respectively. Clearly, the anode rib width has a significant impact on the cell performance.
As shown in Figure 4a, although there is the change of O_{2} concentration in the vertical direction for CellC, the distributions of O_{2} concentration is reasonably uniform near the cathode/electrolyte interface. Figure 4b, however, shows a quite different situation for CellD. The change of O_{2} concentration near the cathode/electrolyte interface of CellD is quite large, decreasing from 1.5 mol·m^{−3} under the channel to 0.25 mol·m^{−3} under the rib, as can be seen in Figure 4b. Similarly to anode rib, the O_{2} diffusion is strongly influenced by the cathode rib. For CellD, the output current densities for Vop of 0.7, 0.6 and 0.5 V are respectively 0.4722, 0.5575 and 0.6290 A·cm^{−2}, as given in Table 7. However, the output current densities of CellC are 10.5%, 18.1% and 26.7% more than that of CellD for V_{op} of 0.7, 0.6 and 0.5 V, respectively. Obviously, it is very necessary to optimize the cathode rib width for increasing cell performance.
Expressions for the Optimal Rib Widths
As described above, the cell performance depends strongly on the rib width and a suitable choice of the rib width is very important for high cell performance. An optimal rib width is obtained for a given pitch width by changing rib width to achieve the maximum cell current density [47]. Our previous study demonstrated that the optimal rib width is only sensitive to the ribelectrode contact resistance and pitch width [19,21]. Therefore we only discuss the relationship between the optimal rib width and ribelectrode contact resistance and pitch width in the following section.
By comparing Figure 5 with Figure 6, it has been observed that the optimal anode rib width is smaller than the optimal cathode rib width for the same pitch width and contact resistance, which is contrary to the anodesupported situation. The reason is that cathode thickness is about 8 times the anode thickness for a cathodesupported SOFC. Thus, the influence of ribs on gas diffusion at the anode side is more serious than that at the cathode side for a fixed rib width. In order to reduce the concentration losses, the anode rib width should be relatively smaller than the cathode rib width.
For a given pitch width, the optimal rib width is found to be dependent approximately linearly on the contact resistance, such as shown in Figures 5 and 6. The relationship between the contact resistance and the optimal rib widths can be formulized as:
drib=A+B×ASRcontactwhere the intercept, A, and the slope, B, are only dependent on the pitch width. As easy to use engineering design guidance, Table 8 lists the parameters A and B for a range of pitch width from 2 mm to 5 mm.
Figures 7 and 8 show the optimal anode and cathode rib widths of anode and cathodesupported SOFCs, respectively, with the pitch width as 2 mm. Obviously, the optimal anode rib width of the anodesupported SOFC is about twice as large as that of the cathodesupported SOFC. However, the optimal cathode rib width of anodesupported SOFC is only about one half of that of cathodesupported SOFC. It can be safely concluded that the formulae of optimal rib widths for anodesupported SOFC can't be applied to cathodesupported SOFC. This is just the problem that this work solves. The above analysis also reveals the importance and necessity of this study.
Conclusions
We have described a comprehensive mathematical model for the performance of a cathodesupported SOFC with syngas fuel. The model takes into account the contact resistance between the electrode and rib and the dependence of the effective electrode properties on the microstructure parameters of the porous electrode. The impact of anode rib width and cathode rib width are studied. The numerical results show conclusively that the output current density depends strongly on the rib width and the rib design optimization is of high engineering importance. Systematic optimization of the anode rib width and cathode rib width for the maximum output current have been carried out. The formulae for the anode and cathode optimal rib width are given, which provides an easy to use guidance for the broad SOFC engineering community.
We gratefully acknowledge the financial support of Jiangsu Province Colleges and Universities Natural Science Projects (13KJB480003), the Jiangsu University of Science and Technology (35321101) and the National Science Foundation of China (21106058).
Conflicts of Interest
The authors declare no conflict of interest.
ReferencesRazbaniO.AssadiM.AnderssonM.Three dimensional cfd modeling and experimental validation of an electrolyte supported solid oxide fuel cell fed with methanefree biogasChenD.ZengQ.SuS.BiW.RenZ.Geometric optimization of a 10cell modular planar solid oxide fuel cell stack manifoldShiraziA.AminyavariM.NajafiB.RinaldiF.RazaghiM.Thermaleconomicenvironmental analysis and multiobjective optimization of an internalreforming solid oxide fuel cellgas turbine hybrid systemLiuQ.DongX.XiaoG.ZhaoF.ChenF.A novel electrode material for symmetrical SOFCsSuzukiT.HasanZ.FunahashiY.YamaguchiT.FujishiroY.AwanoM.Impact of anode microstructure on solid oxide fuel cellsYangL.LiuZ.WangS.ChoiY.ZuoC.LiuM.A mixed proton, oxygen ion, and electron conducting cathode for sofcs based on oxide proton conductorsHuangJ.XieF.WangC.MaoZ.Development of solid oxide fuel cell materials for intermediatetolow temperature operationChenG.YouH.X.KasaiY.SatoH.AbudulaA.Characterization of planer cathodesupported sofc prepared by a dual dry pressing methodLiuM.DongD.ZhaoF.GaoJ.DingD.LiuX.MengG.Highperformance cathodesupported sofcs prepared by a singlestep cofiring processWangB.LaiZ.Fingerlike voids induced by viscous fingering during phase inversion of alumina/pes/nmp suspensionsMukhopadhyayM.MukhopadhyayJ.Das SharmaA.BasuR.N.Engineered anode structure for enhanced electrochemical performance of anodesupported planar solid oxide fuel cellKornelyM.LeonideA.WeberA.IversTifféeE.Performance limiting factors in anodesupported cells originating from metallic interconnector designBlumL.MeulenbergW.A.NabielekH.SteinbergerWilckensR.Worldwide sofc technology overview and benchmarkTannerC.W.VirkarA.V.A simple model for interconnect design of planar solid oxide fuel cellsLinZ.StevensonJ.W.KhaleelM.A.The effect of interconnect rib size on the fuel cell concentration polarization in planar sofcsJiY.YuanK.ChungJ.ChenY.Effects of transport scale on heat/mass transfer and performance optimization for solid oxide fuel cellsJeonD.H.NamJ.H.KimC.J.Microstructural optimization of anodesupported solid oxide fuel cells by a comprehensive microscale modelNohH.S.HwangJ.YoonK.KimB.K.LeeH.W.LeeJ.H.SonJ.W.Optimization of current collection to reduce the lateral conduction loss of thinfilmprocessed cathodesLiuS.SongC.LinZ.The effects of the interconnect rib contact resistance on the performance of planar solid oxide fuel cell stack and the rib design optimizationDeyT.SingdeoD.BoseM.BasuR.N.GhoshP.C.Study of contact resistance at the electrodeinterconnect interfaces in planar type solid oxide fuel cellsKongW.LiJ.LiuS.LinZ.The influence of interconnect ribs on the performance of planar solid oxide fuel cell and formulae for optimal rib sizesKongW.SuS.C.GaoX.ZhangD.H.YuZ.D.Optimization of the anode current collector layer thickness for the cathodesupported solid oxide fuel cellLiuS.KongW.LinZ.Threedimensional modeling of planar solid oxide fuel cells and the rib design optimizationKongW.ZhuH.FeiZ.LinZ.A modified dusty gas model in the form of a fick's model for the prediction of multicomponent mass transport in a solid oxide fuel cell anodeLiJ.KongW.LinZ.Theoretical studies on the electrochemical and mechanical properties and microstructure optimization of microtubular solid oxide fuel cellsFullerE.N.SchettlerP.D.GiddingsJ.C.New method for prediction of binary gasphase diffusion coefficientsVeldsinkJ.W.VersteegG.F.van SwaaijW.P.M.van DammeR.M.J.The use of the dustygas model for the description of mass transport with chemical reaction in porous mediaToddB.YoungJ.B.Thermodynamic and transport properties of gases for use in solid oxide fuel cell modellingLehnertW.MeusingerJ.ThomF.Modelling of gas transport phenomena in sofc anodesKleinJ.BultelY.GeorgesS.PonsM.Modeling of a sofc fuelled by methane: From direct internal reforming to gradual internal reformingHouK.HughesR.The kinetics of methane steam reforming over a Ni/αAl_{2}O catalystCostamagnaP.HoneggerK.Modeling of solid oxide heat exchanger integrated stacks and simulation at high fuel utilizationZhuH.Y.KeeR.J.Modeling distributed chargetransfer processes in sofc membrane electrode assembliesLiuS.KongW.LinZ.A microscale modeling tool for the design and optimization of solid oxide fuel cellsCostamagnaP.CostaP.AntonucciV.Micromodelling of solid oxide fuel cell electrodesAliA.WenX.NandakumarK.LuoB.J.ChuangK.T.Geometrical modeling of microstructure of solid oxide fuel cell composite electrodesJanardhananV.M.HeuvelineV.DeutschmannO.Threephase boundary length in solidoxide fuel cells: A mathematical modelChenD.HeH.ZhangD.WangH.NiM.Percolation theory in solid oxide fuel cell composite electrodes with a mixed electronic and ionic conductorChenX.J.ChanS.H.KhorK.A.Simulation of a composite cathode in solid oxide fuel cellsWuJ.J.McLachlanD.S.Percolation exponents and thresholds obtained from the nearly ideal continuum percolation system graphiteboron nitrideSuzukiM.OshimaT.Coordination number of a multicomponent randomly packed bed of spheres with size distributionBouvardD.LangeF.F.Relation between percolation and particle coordination in binary powder mixturesShiY.X.CaiN.S.LiC.Numerical modeling of an anodesupported sofc button cell considering anodic surface diffusionChenT.LiT.MiaoH.WangW.G.WuY.Operation of NiYSZ anode supported solid oxide fuel cell on different simulated syngasChenD.BiW.KongW.LinZ.Combined microscale and macroscale modeling of the composite electrode of a solid oxide fuel cellZhaoF.VirkarA.V.Dependence of polarization in anodesupported solid oxide fuel cells on various cell parametersSelleriT.NajafiB.RinaldiF.ColomboG.Mathematical modeling and multiobjective optimization of a minichannel heat exchanger via genetic algorithmFigures and Tables
Schematic of (a) a cathodesupported SOFC stack cell unit and (b) crosssection of SOFC stack cell unit.
Comparison of the theoretical and experimental I–V relationships. The open symbols and solid lines denote respectively the measured and the calculated results. The thickness of anode support layer and cathode current collector layer are 500 and 45 μm respectively.
Distributions of H_{2} in (a) CellA and (b) CellB.
Distributions of O_{2} in (a) CellC and (b) CellD.
Dependence of the optimal anode rib width on the contact resistance and pitch width.
Dependence of the optimal cathode rib width on the contact resistance and pitch width.
The optimal anode rib width of anode and cathodesupported SOFC.
The optimal cathode rib width of anode and cathodesupported SOFC.
Boundary settings for mass transports in electrodes. “Insulation” means no flux through the boundary.
Equations
Boundary
ACCL/channel interface
AFL/electrolyte interface
All others
Fuel transfer
BC type
gas molar concentration
(H_{2}) inward molar flux
(H_{2}O) inward molar flux
insulation
BC
cH20,cCH40,cCO0,cCO20,cH2O0
−itransanλTPBA/2F
itransanλTPBA/2F

Equations
Boundary
CSL/channel interface
CFL/electrolyte interface
All others
Air transfer
BC type
(O_{2}) molar concentration
(N_{2}) molar concentration
(O_{2}) inward molar flux
(N_{2}) inward molar flux
insulation
BC
cO20
cN20
−itranscaλTPBA/4F
0

Boundary settings for the electronic and ionic charge transfer equations. “Electric Insulation” makes the normal component of the electric current zero.
Equations
Boundary
Rib/CSL interface
Rib/ACCL interface
CFL/Electrolyte interface
AFL/Electrolyte interface
All others
Electronic transfer
BC Type
(φ_{e,rib/CSL}) Reference potential
(φ_{e,rib/ACCL}) Reference potential
Inward current flow
Inward current flow
Electric insulation
BC
V_{op}
E_{0} = 1.08 V
itranscaλTPBA
−itransanλTPBA

Ionic transfer
BC Type


Interior current source
Interior current source
Electric insulation
BC


−itranscaλTPBA
itransanλTPBA

Model parameters [44,45].
Parameter
Value
Parameter
Value
Temperature, T (°C)
850
Reaction symmetric factor for cathode,
αfca,βrca
0.2, 0.8
Operation voltage, V_{op} (V)
0.7
Exchange transfer current density of cathode
i0,refca(Am−1)
7 × 10^{−5}
Tortuosity factor, τ
4
Activation energies for the anode, E_{H2}(J mol^{−1})
120 × 10^{3}
Fuel composition (molar fraction) at the anode/channel interface
CH_{4}(%)
10
Reaction symmetric factor for anode,
αfan,βran
1, 0.5
CO(%)
8
Exchange transfer current density of anode,
i0,refan(A m−1)
8.0 × 10^{−4}
CO_{2} (%)
17
The volumetric active surface area of Ni particles in AFL,
SANi (m^{2}·m^{−3})
1.05 × 10^{6}
H_{2}O (%)
5
The volumetric active surface area of Ni particles in ACCL,
SANi ( m^{2}·m^{−3})
4.13 × 10^{5}
H_{2} (%)
60
Area specific contact resistance at the ribanode interface,
ASRcontactan (Ω·cm^{2})
0.01
Activation energies for the cathode, E_{O2} (J mol^{−1})
130 × 10^{3}
Area specific contact resistance at the ribcathode interface,
ASRcontactca (Ω·cm^{2})
0.01
Micro structure parameters and thickness of each layer [34,46].
Layer
r_{el}(m)
ϕ_{el}
ε
Thickness (μm)
γ
ACCL
1 ×10^{−6}
0.5
0.45
45
1
AFL
0.5 × 10^{−6}
0.5
0.3
10
1
Electrolyte



10
1
CFL
0.5 × 10^{−6}
0.5
0.3
20
1
CCCL
1 × 10^{−6}
1
0.45
500
1
Diffusion volume and coefficients of viscosity coefficient.