The role of numerical simulation to drive the catalytic converter development becomes more important as more efficient spark ignition engines after-treatment devices are required. The use of simplified approaches using rather simple correlations for heat and mass transfer in a channel has been widely used to obtain computational simplicity and sufficient accuracy. However these approaches always require specific experimental tuning so reducing their predictive capabilities. The feasibility of a computational fluid dynamics three-dimensional (3D) model coupled to a surface chemistry solver is evaluated in this paper as a tool to increase model predictivity then allowing the detailed study of the performance of a catalytic converter under widely varying operating conditions. The model is based on FLUENT to solve the steady-state 3D transport of mass, momentum and energy for a gas mixture channel flow, and it is coupled to a powerful surface chemistry tool (CANTERA). Checked with respect to literature available experimental data, this approach has proved its predictive capabilities not requiring an ad hoc tuning of the parameter, set. Heat and mass transfer characteristics of channels with different section shapes (sinusoidal, hexagonal, and squared) have then been analyzed. Results mainly indicate that a significant influence of operating temperature can be observed on Nusselt and Sherwood profiles and that traditional correlations, as well as the use of heat/mass transfer analogy, may give remarkable errors (up to 30% along one-third of the whole channel during light-off conditions) in the evaluation of the converter performance. The proposed approach represents an appropriate tool to generate local heat and mass transfer correlations for less accurate, but more comprehensive, ID models, either directly during the calculation or off-line, to build a proper data base.
