AbstractAutomotive catalytic converters are increasingly used to reduce emissions from internal combustion engines to comply with emission regulations. Maldistributed flow across the catalyst affects its warm up, light off time, ageing, and conversion efficiency. This thesis concerns flow distribution in automotive catalytic converters and methods to improve CFD predictions.
Previous studies showed that modelling the monolith flow resistance using the Hagen- Poiseuille’s formulation under predicted flow maldistribution. The predictions were improved by incorporating an additional pressure loss term ½ρV2, where V is transverse velocity just upstream of a monolith channel, for oblique entry of the flow into the monolith known as the entrance effect. Further improvement was obtained by incorporating the critical angle of attack method. However, there was no experimental evidence to support these oblique entry loss formulations. There also remained the possibility that under prediction of flow maldistribution might be due to the failure to predict flow in the diffuser accurately.
A one-dimensional oblique angle flow rig was designed and built to measure the effect of oblique entry flow losses in monoliths. Experiments were performed at different angles of attack (α), using different lengths of substrate and a methodology was developed to obtain the oblique flow entrance losses. The results showed that the pressure loss attributed to the entrance effect increased with the angle of attack. The entrance effect was also found to be dependent on channel Reynolds number and substrate length. The theoretical assumption of ½ρV2 predicts accurately at low Reynolds number but looses its validity at high Reynolds number. From the experimental studies, an improved correlation for the entrance effect has been derived as a function of major controlling variables, i.e., angle of attack, length of the substrates and Reynolds number.
A two-dimensional rig was designed to measure the flow field using PIV in a 2-D diffuser placed upstream of two different length substrates. The results showed that the flow in a wide angle diffuser consisted of a central core, free shear layer and recirculation regions. The near-field region was found similar to that of a plane jet. The flow field was found to be independent of Reynolds number. Increasing the substrate length resulted in a flattening of the axial profiles close to the substrate face.
A CFD study was undertaken to predict maldistributed flow at the exit of the substrate for an axisymmetric catalyst model by incorporating the measured entrance effect correlation. A fixed critical angle of attack (αc,F) approach was used whereby the entrance effect is assumed constant for α>αc,F. Incorporating the entrance effect with αc,F= 810 improved the prediction of maldistribution in the flow profiles.
A 2-D CFD study was undertaken to predict the flow distribution in the diffuser and downstream of the substrate. A comparison of the CFD predictions in the diffuser using different turbulence models showed that all the turbulence models used in this study over predicted the width of the central core region and the V2F turbulence model gave velocity predictions that compared best with PIV. Incorporating the entrance effect improved the predictions close to the diffuser-substrate interface and downstream of the substrate.
|Date of Award||2008|
|Supervisor||Stephen Benjamin (Supervisor)|
- automotive catalytic converters