A theoretical and experimental study of automotive catalytic converters
In response to the increasingly widespread use of catalytic converters for meeting automotive exhaust emission regulations considerable attention is currently being directed towards improving their performance. Experimental analysis is costly and time consuming. A desirable alternative is computational modelling. This thesis describes the development of a fully integrated computational model for simulating monolith type automotive catalytic converters. Two commercial CFD codes, PHOENICS and STAR-CD, were utilised to implement established techniques for modelling the flow field in catalyst assemblies. To appraise the accuracy of the flow field predictions an isothermal steady flow rig was designed and developed. A selection of axisymmetric inlet diffusers and 180o expansions were tested, with the velocity profile across the monolith, the wall static pressure distribution along the inlet section and the total pressure drop across the assembly being measured. These datum sets were compared with predictions using a variety of turbulence models and solution algorithms. The closest agreement was achieved with a two-layer near wall approach, coupled to the fully turbulent version of the RNG k-ε model, and a nominally second order differencing scheme. Even with these approaches the predicted velocity profiles were too flat, the maximum velocity being as much as 17.5% too low. Agreement on pressure drops was better, the error being consistently less than 10%. These results illustrate that present modelling techniques are insufficiently reliable for accurate predictions. It is suggested that the major reason for the relatively poor performance of these techniques is the neglecting of channel entrance effects in the monolith pressure drop term. Despite these weaknesses it was possible to show that the model reproduces the correct trends, and magnitude of change, in pressure drop and velocity distributions as the catalyst geometry changes. The PHONETICS flow field model was extended to include the heat transfer, mass transfer and chemical reactions associated with catalysts. The methodology is based on an equivalent continuum approach. The result is a reacting model capable of simulating the three-dimensional distribution of solid and gas temperatures, species concentrations and flow field variables throughout the monolith mat and the effects that moisture has on the transient warm-up of the monolith. To assess the reacting model’s accuracy use was made of published light-off data from a catalyst connected to a test bed engine. Comparison with predicted results showed that the model was capable of reproducing the correct type, and time scales, of temperature and conversion efficiency behaviour during the warm-up cycle. From these predictions it was possible to show that the flow distribution across the monolith can significantly change during light-off. Following the identification, and subsequent modelling, of the condensation and evaporation of water during the warm-up process it was possible to show that, under the catalyst conditions tested, these moisture effects do not affect light-off times. Conditions under which moisture might affect light-off have been suggested. Although the general level of model accuracy may be acceptable for studying many catalyst phenomena, known deficiencies in the reaction kinetics used, errors in the flow field predictions, uncertainty over many of the physical constants and necessary model simplifications mean that accurate quantitative predictions are still lacking. Improving the level of accuracy will require a systematic experimental approach followed by model refinements.