Modeling full-scale monolithic catalytic converters: Challenges and possible solutions

Modeling full-scale monolithic catalytic converters using state-of-the-art computational fluid dynamics algorithms and techniques encounters a classical multiscale problem: the channels within the monolith have length scales that are similar to 1-2 mm, while the converter itself has a length scale that is similar to 5-10 cm. This necessitates very fine grids to resolve all the length scales, resulting in few million computational cells. When complex heterogeneous chemistry is included, the computational problem becomes all but intractable unless massively parallel computation is employed. Two approaches to address this difficulty are reviewed, and their effectiveness demonstrated for the computation of fullscale catalytic converters with complex chemistry. The first approach is one where only the larger scales are resolved by a grid, while the physics at the smallest scale (channel scale) are modeled using subgrid scale models whose development entails detailed flux balances at the imaginary fluid-solid interfaces within each computational cell. The second approach. makes use of the in situ adaptive tabulation algorithm, after significant reformulation of the underlying mathematics, to accelerate computation of the surface reaction boundary conditions. Preliminary results shown here for a catalytic combustion application involving 19 species and 24 reactions indicate that both methods have the potential of improving computational efficiency by several orders of magnitude.