This content is not included in your SAE MOBILUS subscription, or you are not logged in.
Computational Fluid Dynamics and Its Application to Catalyst Exhaust Systems
Published June 01, 1992 by ISATA - Dusseldorf Trade Fair in United Kingdom
Event: ISATA 1992
In the last few years, computational fluid dynamics (CFD) has evolved from its roots in mathematics into a powerful tool for engineering. Improvements in computers, from small P.C.s to large main-frames, means that ever larger problems may be solved. Developments in the CFD codes themselves have produced faster, more robust algorithms and the codes have become much more user-friendly, enhancing their appeal to engineers.
Researchers in the automotive emissions field are turning their attention to CFD as a means of predicting the emission from a range of catalytic systems, without the need for expensive experimental testing. Currently, there are no accurate predictive techniques capable of handling the extremely transient nature of a catalytic exhaust system. Researchers throughout the world, however, have modeled simple catalytic systems (such as one- dimensional, adiabatic reactors), and are using them as a basis for more complex models.
Complete modelling of the whole system requires i) an understanding of the flow and heat transfer of the exhaust system, particularly in the converter; and ii) a thorough knowledge of the chemistry of catalytic converters. At this point in time, little is known about the fluid dynamics within converters, and even less about the kinetics of the catalytic reactions. As experimental programs reveal more about both these areas, faster computers and more robust codes will be needed in order to solve transient models of ever increasing complexity.
Research engineers, Oh and Cavendish, a General Motors Corporation Research Laboratories in the USA developed their own code based initially on single pellet studies. This one-dimensional model was then extended to a three-dimensional transient model through the work of Chen et al., where it was used on monolithic converters with flow maldistribution.
The General Motors Corporation approach to the transient three-dimensional problem of monolithic converters was to adopt a dual-solver technique, treating the monolith as a porous medium with gas flow. With this approach, the fluid flow and the mass transfer equations were solved, using a finite element solver, for the gases in the void fraction of the monolith. The resulting temperature and concentration fields were then passed to a separate solver which calculated the solid temperature distribution and used the reaction kinetics to calculate a new concentration field. The modified temperature and concentration fields were then passed back to the fluids solver and the process iterated to convergence. This was repeated for each time step in a transient calculation.
The model was the first of its kind to consider a three-dimensional non- adiabatic monolithic converter operating, transiently, under conditions of non-uniform flow. The model's predicted results gave good qualitative agreement, but owing to a lack of accurate kinetic data for the reactions considered, the quantitative agreement was only good in some areas.
Many other models have been formulated by various researchers but these mainly considered one-dimensional models which were adiabatic and/or had uniform flow. Other workers considering three-dimensional models and/or non- uniform flow have been Flytztani-Stephanopoulos et al., Becker and Zygourakis, Zygourakis and Leclerc and Schweich.
With the advent of more user-friendly, versatile, commercially available CFD codes, researchers are adapting these codes to the particular needs of catalytic converters rather than spending time and money writing bespoke codes of their own. Researchers Lai and Kim et al. have used a commercially available code to model the flow through various converter designs. Their results have indicated the expected extent of flow maldistribution within various converter designs. Their experimental program will aim to validate the model predictions beyond just pressure drop correlations. Developments are still needed in the area of reaction kinetics, whose highly nonlinear mathematical equations do not easily integrate into these codes. The largest single problem in this field is the lack of reliable and accurate experimental data with which to validate any predictions.