Simulation of Solid Oxide Fuel Cell Anode in Aspen HYSYS—A Study on the Effect of Reforming Activity on Distributed Performance Profiles, Carbon Formation, and Anode Oxidation Risk

A distributed variable model for solid oxide fuel cell (SOFC), with internal fuel reforming on the anode, has been developed in Aspen HYSYS. The proposed model accounts for the complex and interactive mechanisms involved in the SOFC operation through a mathematically viable and numerically fast modeling framework. The internal fuel reforming reaction calculations have been carried out in a plug flow reactor (PFR) module integrated with a spreadsheet module to interactively calculate the electrochemical process details. By interlinking the two modules within Aspen HYSYS flowsheeting environment, the highly nonlinear SOFC distributed profiles have been readily captured using empirical correlations and without the necessity of using an external coding platform, such as MATLAB or FORTRAN. Distributed variables including temperature, current density, and concentration profiles along the cell length, have been discussed for various reforming activity rates. Moreover, parametric estimation of anode oxidation risk and carbon formation potential against fuel reformation intensity have been demonstrated that contributes to the SOFC lifetime evaluation. Incrementally progressive catalyst activity has been proposed as a technically viable approach for attaining smooth profiles within the SOFC anode. The proposed modeling platform paves the way for SOFC system flowsheeting and optimization, particularly where the study of systems with stack distributed variables is of interest.


Introduction
Fuel cells convert the chemical energy in fuel directly into electricity and heat, without combustion, leading to high efficiencies with low or even zero emissions. The SOFC is becoming a mature technology and can make the commercial breakthrough if cost targets can be met by achieving cost reductions through volume manufacturing, improved lifespan/performance, and lower cost materials [1][2][3]. Research and development in the last twenty years have led to significant advances in all areas of the technology including cell, seal, interconnect, and stack design, as well as peripheral components and the entire balance of plant (BoP) [4,5]. Manufacturing achievements have led to defect identification and minimization, quality control, and scale-up of stack components and the entire stack assembly manufacture.
The SOFC system is appropriate to operate on a pipeline fuel such as reticulated natural gas with its well-established supply infrastructure throughout the world. For such a fuel, minimal fuel processing is required, which includes desulphurization of the fuel to remove sulphur compounds steam reforming (MSR) reaction and accompanying mildly exothermic WGS reaction with heat available from the fuel cell reaction is appropriately captured with this approach. The corresponding composition profile under current load cannot be generated within the PFR module as this module only works with kinetic schemes and it is not possible to add the fuel cell reactions to the PFR as conversion reactions based on fuel utilization. An option to include the reaction kinetics of the electrochemical oxidation of hydrogen is also not available in the software. Nevertheless, the reaction extents and accompanying heat exchanges can be calculated in the spreadsheet module and linked to the PFR module. Firstly, this allows generation of open-circuit composition profiles of the internally reformed gas which sets the boundary for the Nernst voltage profile under load, after accounting for the extents of the fuel cell reactions. Secondly, the current density and composition profiles can be calculated within the spreadsheet block using appropriate correlations.
Previous work [9][10][11][12][13]16] largely focused on the issues that can predict and improve the fuel cell operation in terms of current generation and voltage losses. For instance, the effect of air flow rate, steam to carbon ratio (S/C), current density, fuel utilization (Uf), inlet temperatures, or operating pressure have been extensively investigated. By contrast, in this work we have mainly focused on an analysis of processes that significantly affect anode performance and lifetime and consequently impact on the SOFC system as a whole. We analyzed anode performance for various levels of reforming activity. Three cases are considered: (i) Full reforming activity, (ii) 1/3rd reforming activity, and (iii) 1/6th reforming activity. Reduced reforming activity may be the result of engineered design [20] or may result from progressive degradation of the anode from poisoning or sintering due to nickel coarsening over the useful life of the stack [4], which extends the reaction zone and requires more of the anode segment from the leading edge to fully convert methane. Reforming kinetics reported by Ahmed and Föger [18] and WGS reaction kinetics reported by Tingey [19] were employed as the reaction rate details for PFR, leading to 1D pseudo-homogeneous results. The three different levels of activity were assigned by reducing the Arrhenius factor in the rate expression by the reduction factors 0.33 (~1/3rd) and 0.67 (~1/6th). In physical terms this signifies loss of reforming activity by poisoning or sintering. For these levels of activity, we assess the anode oxidation risk and carbon formation potential on the anode, both of which have severe life-limiting consequences on the anode [7].
This paper contributes to the SOFC research considering its two novel contents including; (i) the novel simulation methodology: The simulation approach proposed for complicated internal reforming SOFC process offers simplicity and calculation speed without compromising the internal operations details. This is of particular interest for SOFC system modeling and design where several operational concepts including heat/mass transfer and electrochemical and fuel reformation reactions interactively occur at wide time and length scales. (ii) The understanding of distributed reformation potentials in controlling SOFC performance profiles. The incremental reformation process is demonstrated to be a promising strategy to moderate the undesirable gradients of SOFC internal profiles. This is promising to achieve higher homogeneity in temperature and concertation profiles inside the SOFC stack that subsequently offers enhanced current and voltage profiles. This is crucially important for SOFC efficiency and durability. In this paper, we demonstrate internal fuel reformation as an opportunity not only for heat integration and external reformer cost reduction but also for thermal management goals that eventually results in fuel cell longevity.

Simulation Methodology
The SOFC stack is simulated in the following way: Internal reforming of methane on the SOFC anode is modeled in a PFR module, using the reforming kinetics of Ahmed and Föger [18] and the WGS reaction kinetics of Tingey [19]. The electrochemical conversion of hydrogen and carbon monoxide are modeled as chemical conversions in a conversion reactor module. The associated electrical aspects including cell voltage, air utilization, and fuel utilization at a given operating current are calculated in a spreadsheet tool of Aspen HYSYS. The spreadsheet block in Aspen HYSYS is essentially an Excel-based spreadsheet with features of exporting data to and importing data from other modules of the flowsheet. This provides facility for post-processing of the calculated or entered values of the process variables. Computations of cell voltage entailing calculations of electrical losses, both ohmic and overpotential using empirical correlations [21] and calculation of Nernst voltage from the concentrations of the reacted gas are carried out in the spreadsheet, with compositions imported from the PFR module. Similarly, the sum of electrical losses and entropy change, after allowing for losses to the surroundings from the stack, is calculated and this value is exported to the energy stream linked to the PFR, as heat available for direct internal reforming.
The average stack operating temperature is obtained by a trial-and-error method. An average stack temperature is assumed for calculating Nernst voltage, operating voltage, and electrical losses. The assumed temperature is then compared to the value returned by the PFR, which simulates internal reforming and utilizes heat generated from the fuel cell reaction to compute the temperature profile within the fuel cell and the composition profile of the internally reformed fuel. The amount of heat generated, i.e., the electrical and entropy losses are calculated at the assumed average temperature. These steps are iterated until agreement is reached between the assumed average cell temperature and the calculated average cell temperature.
The composition of utilized gas is obtained by matching the degree of conversion in the conversion reactor, with the fuel utilization level calculated in the spreadsheet, based on fuel flow rate and operating current. Since the electrochemical conversion of H2 and CO are modeled as combustion reactions, there is a temperature rise in the reactor due to the exothermicity of the combustion reactions. Since the exothermicity of the fuel cell reactions have already been accounted for by calculating the electrical losses and entropy change of the fuel cell reaction and entering this value as heat input to the internal reforming PFR, temperature rise in the conversion reactor is suppressed by use of the HYSYS object SET, to set the temperature of the stream leaving the conversion reactor to be the same as the temperature of the stream leaving the PFR. Figure 1 and Table 1 show the integration of the modules representing the anode operation and the set of equations used in the calculation of all distributed variables within the stack-temperature, current density, and composition of all chemical species on the anode side. The reactions taking place in the reaction modules, shown in Figure 1, are as follows. Two parallel reactions take place in the internal reforming module (PFR) including MSR and WGS, presented by Equations (1) and (2), respectively: CH4 + H2O = CO + 3H2 (1) Two parallel electrochemical reactions take place in the fuel cell reactor module (conversion reactor). The main reaction is hydrogen oxidation (Equation (3)) that may occur with the CO oxidation (Equation (4)), simultaneously. H2 + 0.5O2 = H2O (3) CO + 0.5O2 = CO2 (4)

Equations/Parameters Comment
Average current drawn based on given fuel rate and desired fuel utilization.
Heat of electrochemical reaction for given fuel consumption, calculated in the spreadsheet block.
Temperature change due to heat release from fuel oxidation.
Open-circuit operation (OCO) composition and temperature profiles from the internal reforming module (PFR) based on MSR and WGS reaction kinetics.
profiles using fuel utilization.
Overpotential losses were calculated using an empirical equation [21]; a = 4.43, y = 0.77, z = -0.15, and Eb = 10,560 kJ/mol, determined by regression analysis of experimental measurement of cell overpotential at various current densities and fuel utilization at the three temperatures 750, 800, and 850 °C.
The cell ohmic resistance is independent of fuel utilization as expected [21], where te is the electrolyte thickness (µm) and A = 21,428 µm/Ω and B = 7776 K are constants, with their values determined empirically [21].

Results and Discussion
The base case is a system operating at 65% fuel utilization. Input parameters for the base case simulation and also estimated results returned by the flowsheet calculations are presented in Tables  2 and 3, respectively.

Distributed Profiles
A high degree of internal reforming of methane is desirable as it reduces the prereformer size/cost and contributes to the cooling of the cell and therefore, reduces the cooling air flow requirement. However, the internal fuel reformation process may undesirably cause higher gradients in the distributed variables profiles. Stack operation homogeneity is of crucial importance from both efficiency and lifetime viewpoints. Figure 2 shows when the kinetics of reforming is too rapid, the cell temperature drops sharply near the cell inlet preventing an even distribution of current density and temperature. The steep temperature gradient is caused by the cooling impact associated with the endothermic steam reforming of methane on the anode. Figure 3 shows the reforming in this case is so fast that it is completed within the first 40% of the cell. Clearly, the maximum temperature gradient can be reduced if the sharp drop in temperature can be avoided, improving all temperaturedependent profiles including species concentration, current density, and overpotentials distributions, all of which have interacting effects. An effective way of achieving this objective is by using catalysts with lower overall reforming activity in anode and/or designing an anode with progressively increasing local activity along the cell length. A reduction of the anode activity to 1/3rd of that of the fast reforming anode results in a relatively uniform temperature profile as shown in Figure 2. A key technical question is whether the fuel reformation should be fully accomplished in the first half of the cell length. If not, there are significant opportunities to design anode reforming activity targeting a more homogenous operation. Accordingly, the objective of achieving smoother temperature profiles in the operating cell must also be assessed from another viewpoint. The activities are compared to the basis case activity (fast reformation). For instance, 1/3rd activity means 66% less activity (lower rate) compared to the basis case. We have adjusted this via E (activation energy value) adjustment in the simulations. With activities lower than the basis case, the catalyst still may offer sufficient reformation as the methane consumption ( Figure 3) and hydrogen generation profiles show. The speed of reformation, however, is declined while it is compensated by residence time (axial length). This is the distributed conversion in contrast to the sharp conversion seen in basis case. To further reduce the temperature differential, an anode with a reforming activity 1/6th of that of the base case fast reforming anode is considered. This results in a more uniform temperature profile, Figure 2, but comes at a cost of increasing methane slip as shown in Figure 3. While a more homogenous electrochemical reactor is achievable under distributed/progressive reforming conditions, an immediate concern relevant to the reduced activity is the possibility of reduction in the generated current due to less local H2 availability in the first 30%-40% of the cell length. In order to assess this rigorously, the overall current produced over the cell surface must be calculated via integration of the local currents. Since no variation of current in cell width direction is assumed, that is a reasonable assumption for co-flow cell, two-dimensional integration can be replaced by one dimensional integration over the cell length. In such a case, therefore, the modeling approach proposed in this paper suffices. The surface under current density profiles in Figure 4 compares the total current production under different reforming approaches. Even though current production considerably drops in the cell inlet region when fast reforming occurs, the overall current production variation for fast and slow reformation activities can be reasonably ignored, indicating that cell efficiency will not be compromised for homogeneity. Methane reforming profile (Figure 3) gives an indication of how the anode exhaust stream might be post-processed. In a fast reforming method, methane is completely consumed inside the stack leaving anode tail gas mainly including, hydrogen, CO, and a significant amount of steam. In such a case anode gas recycling is an appropriate process strategy. For slow/distributed fuel reforming, anode tail gas might contain some methane and less hydrogen and CO compared to the fast reforming. This can be understood by interpretation of methane and current profiles simultaneously. As current generations are equal, the H2 consumption in all cases are almost the same. For a given methane rate at inlet, therefore, higher methane fraction in cell outlet indicates lower amount of hydrogen and CO. This becomes even more considerable for an activity as low as 1/6th of the base case. Therefore, an after-burner can be designed in the system to achieve high quality heat from the tail gas. It may be concluded that the progressive fuel reforming offers some advantage for a CHP system in which both high quality heat and homogenous stack performance are desirable. Stack exhaust temperature is reduced with progressive reforming activity compared to fast reforming, but the stack temperature profile for most of the anode length is higher as shown in Figure  2. While the latter is expected due to less steeper cooling along the cell, the reduction in stack exhaust temperature can be explained from a consideration of the associated current density profile. As shown in Figure 4, the current density profile is steepest for the fast reforming case resulting in higher joule heating effect. A lower joule heating effect results in lower increase in temperatures for the distributed reforming cases, to the point that the temperature at the end of the cell is lower for distributed reforming, even though the temperatures near the inlet are higher. A more uniform current density profile results in the case of lower reforming activity as expected which is beneficial from the point of stress reduction.

Anode Oxidation
The SOFC anode is susceptible to oxidation by steam in the reaction mixture according to Equation (5): In this paper, anode oxidation risk is determined on the basis of industrial experience [22] with nickel-based catalysts, where it reported finding that steam-to-hydrogen ratios greater than 6-8 increases the risk of nickel oxidation in nickel-based steam reforming catalysts. Analyses, therefore, are based on the local partial pressures as a characteristic indicator rather than estimation of rate for reaction 5. Distributed profiles show that the risk of anode oxidation is particularly high at high fuel utilization, where the partial pressure of steam in the reaction mixture is high, i.e., at high pH2O/pH2. Anode oxidation risk profile along the length of the cell is shown in Figure 5 for 65% and 75% fuel utilization with three levels of reforming activity. Anode oxidation risk is highest at the anode inlet for internal reforming anodes where enough hydrogen has not been generated ( Figure 6). The risk increases with slow reforming activity. At higher utilization, more hydrogen is utilized, further lowering the H2 content and increasing the H2O content, thereby increasing the risk of anode oxidation compared to lower utilization. In such a case, anode tail gas recycle may worsen the situation by introducing more steam upstream. This is an additional reason why tail gas in the slow internal reforming case is recommended to be burnt rather than being recycled. In general, anode gas recycling may change the pH2O and pH2 balance in favour of steam.

Carbon Formation
Carbon formation on the SOFC anode, as a challenging issue for SOFC degradation, might occur principally via two routes: Boudouard reaction and methane cracking. In this work, carbon formation activity is calculated based on the thermodynamical equilibrium estimated using the local temperature and compositions of the relevant species participating in the respective reactions. Carbon may form by cracking or dissociation of a methane molecule into carbon and hydrogen molecules according to the reaction in Equation (6): For this reaction, the carbon activity for dissociation of a methane ( ) can be calculated by using Equation (7): The Boudouard or carbon disproportionation reaction can be presented as Equation (8): Similar to the methane cracking, a carbon activity can be defined for Boudouard reaction ( ) by using the local concentration/partial pressure of the gases involved, as presented in Equation (9): Simulated distribution of the carbon formation risks from Boudouard reaction and methane cracking are depicted in Figure 7 and Figure 8, respectively. Figure 7 shows that under the operational conditions used in this study and fuel utilization ranging from 65% to 75%, the probability of carbon formation through Boudouard reaction is low and well below 1 regardless of the fuel reforming pattern, i.e., fast or distributed reformation patterns. This is primarily due to the fact that this reaction is not thermodynamically benefited by the elevated temperature specifically above 700 °C. The trend along the anode length can be explained with respect to the CO and CO2 concentrations profiles and the anode thermal behaviour. The risk is relatively high at lower fuel utilization levels due to relatively higher and lower levels of CO and CO2, respectively, in the anode gas mixture and also lower local temperature, all enhancing the Boudouard reaction chance to occur. Note that according to Equation (9), at any given temperature, the higher the ratio of (pCO) 2 /pCO2, the higher the carbon activity. Near the inlet, CO is low, as methane reforming has not progressed much. Further down, the ratio depends on how much CO is formed by reforming and how much CO2 is formed by WGS (Equation (2)). The WGS equilibrium is also affected by the current draw, as the WGS equilibrium shifts to the right as more H2 are consumed by the hydrogen electrochemical oxidation reaction (Equation (3)). It will also be affected by electrochemical oxidation of CO to CO2, but the extent of this reaction is generally small, as this reaction is not as fast as H2 oxidation. Moreover, due to equal stoichiometry, this effect is accounted for by the WGS equilibrium reaction (Equation (2)). As the WGS reaction generates one mole of hydrogen per mole of CO, the electrochemical and chemical balance is unaffected whether the CO conversion is modeled as a WGS or as electrochemical oxidation. In a recent work [23], the effect of different reaction kinetics and equilibrium of the methane steam reforming reaction and WGS reaction were shown to have significant effect on the concentration profiles along the cell length. Temperature change along the cell length also affects the carbon activity as it changes the value of the equilibrium constant.   Figure 8 shows a very high risk of carbon formation by methane cracking near the anode inlet where methane concentration is high. The risk increases for lower fuel utilization and for slower reforming activity. The carbon formation activity is calculated based on thermodynamic equilibrium; in practice, whether carbon will be formed will depend on the kinetics of the reactions involved.

Conclusion
A model of a SOFC cell incorporating a 1D model of the anode has been developed in Aspen HYSYS. A salient feature of this model is its ability to predict simultaneous direct internal reforming on the anode and electrochemical reaction with these two reactions thermally integrated. In spite of being based on empirical correlations that makes modeling platform computationally efficient it successfully captures the distributed variables associated with stack level performance, particularly with respect to internal reforming kinetics and electrical performance. This has been achieved through interlinking of the in-built PFR module with a spreadsheet block inside the Aspen HYSYS environment. Cases for various levels of reforming activity have been compared to demonstrate the effect and relative advantages and disadvantages in terms of temperature and current density profiles. Two technically challenging aspects of SOFC operation: Anode oxidation risk and carbon formation potential have been evaluated. The methodology proposed in this paper is flexible to deploy more detailed fundamental correlations such as explicit equations based on exchange current.