Stack model validation

In analogy to the treatment of the stacking model Jo (s) = 0 is valid, if the structural entities are embedded in matrix material. Compact material, again, may require a correction because of the merging of particles from abutting structural entities... 

Despite the availability of quite sophisticated fuel cell models with well-written code and convenient user interfaces, the fuel cell developer or engineer must be a critical user. As mentioned above, obtaining experimental data on the behavior of fuel cells (especially internally and at the micro-level) can be difficult, time-consuming, and expensive. Unfortunately this has lead to a dearth of accurate and detailed data of sufficient quality and quantity to allow thorough validation of the mathematical models. Much of the data on fuel cell performance reported in the literature is, while phenomenologically often interesting, insufficiently accurate and accompanied by far too little detail on the test conditions to be usable for model validation. In particular, with much of the cell and stack taken at modest utilization, it is almost impossible to infer kinetic data without spatially resolved data on current density, temperature and species concentrations. As a consequence, the validity of fuel cell models must be critically considered for each use. The user of the model must be thoroughly familiar with the assumptions and limitations embedded in the models. 

Further enhancement of the design tool is continuing. The next steps are to validate the model with experimental data and then extend the model to stack module and stack analysis. NETL now operates SOFC test facilities to generate detailed model validation data using well-characterized SOFC test specimens. These steps should make it possible to create a model that accurately predicts the performance of cells and stacks so that critical design information, such as the distribution of cell and stack stresses, can be provided to the fuel cell design engineer. 

If other statistical models of polydispersity should prove more appropriate than the paracrystalline stack, validations of the first-zero method may be carried out in analogy to the one presented here. 

PEFC) stacks, components and entire systems, in off-grid, and grid-connected configurations, with a capacity of up to 100 kW electrical power output. The facility consists of an automated and computerised fuel cell test station, gas analysers, a multi-axial vibration system which is housed in a walk-in environmental chamber (for controlling temperatures, humidity, shocks and vibrations) and ancillary equipment. The data obtained are complementary to and validate fuel cell simulations and models with reference to operation modes, components and system characteristics 1 ... 

A further criticism of the BET theory is the assumption that the heat of adsorption of the second and higher layers is equal to the heat of liquefaction. It seems reasonable to expect that polarization forces would induce a higher heat of adsorption in the second layer than in the third, and so forth. Only after several layers are adsorbed should the heat of adsorption equal the heat of liquefaction. It is, therefore, difficult to resolve a model of molecules adsorbed in stacks while postulating that all layers above the first are thermodynamically a true liquid structure. The apparent validity of these criticisms contributes to the failure of the BET equation at high relative pressures (P/Pq > 0.35). However, in the range of relative pressure leading to coverage near W/ = 1, the BET C values... 

In the light of the above, we recommend that while verification is applied, validation of the cell and stack calculations in comparison to carefully designed experiments must take priority in the fuel cell modeling community. Only by proper validation of the 3 -D calculations using, at least, the spatial distribution of temperature and current measurements, the computer simulations can take its proper role in design analysis and improvement with relevance to industrial application. Needless to say, the computation time must be reduced for practicality purposes. 

Another issue that needs to be addressed is the accurate calculations of the transients of stack operations under variable loading due to changes in power utilization demand and/or under start-up and shut-down conditions. Tracking fast transients, especially during the start-up process, requires at least second order accurate temporal resolution which will impose additional computational cost on stack simulations. It seems that in the near future the best alternative would be to use reduced order physics based models such as those presented in Section 5.2 with appropriate empirical input and experimental validation to get the most benefit out of computational studies. 

---