Automotive catalytic converters are used extensively in the automotive industry to reduce toxic pollutants from vehicle exhausts. The flow across automotive exhaust catalysts is distributed by a sudden expansion and has a significant effect on their conversion efficiency. The exhaust gas is pulsating and flow distribution is a function of engine operating condition, namely speed (frequency), load (flow rate) and pressure loss across the monolith. The aims of this study are to provide insight into the development of the pulsating flow field within the diffuser under isothermal conditions and to assess the steady-state computational fluid dynamics (CFD) predictions of flow maldistribution at high Reynolds numbers.
Flow measurements were made across an automotive catalyst monolith situated downstream of a planar wide-angled diffuser in the presence of pulsating flow. Cycle-resolved Particle Image Velocimetry (PIV) measurements were made in the diffuser and hot wire anemometry (HWA) downstream of the monoliths. The ratio of pulse period to residence time within the diffuser (J factor) characterises the flow distribution. During acceleration the flow remained attached to the diffuser walls for some distance before separating near the diffuser inlet later in the cycle. Two cases with J ~ 3.5 resulted in very similar flow fields with the flow able to reattach downstream of the separation bubbles. With J = 6.8 separation occurred earlier with the flow field resembling, at the time of deceleration, the steady flow field. Increasing J from 3.5 to 6.8 resulted in greater flow maldistribution within the monoliths; steady flow producing the highest maldistribution in all cases for the same Re.
The oblique entry pressure loss of monoliths were measured using a one-dimensional steady flow rig over a range of approach Reynolds number (200 < Rea < 4090) and angles of incidence (0o < α < 70o). Losses increased with α and Re at low mass flow rates but were independent of Re at high flow rates being 20% higher than the transverse dynamic pressure.
The flow distribution across axisymmetric ceramic 400 cpsi and perforated 600 cpsi monoliths were modelled using CFD and the porous medium approach. This requires knowledge of the axial and transverse monolith resistances; the latter being only applicable to the radially open structure. The axial resistances were measured by presenting uniform flow to the front face of the monolith. The transverse resistances were deduced by best matching CFD predictions to measurements of the radial flow profiles obtained downstream of the monolith when presented with non-uniform flow at its front face.
CFD predictions of the flow maldistibution were performed by adding the oblique entry pressure loss to the axial resistance to simulate the monolith losses. The critical angle approach was used to improve the predictions, i.e. the oblique entry loss was limited such that the losses were assumed constant above a fixed critical angle, αc. The result showed that the perforated 600 cpsi monolith requires the entrance effect to be restricted above αc = 81o, while the losses were assumed constant above αc = 85o for the ceramic 400 cpsi monolith. This might be due to the separation bubble at the monolith entrance being restricted by the smaller hydraulic diameter of the perforated monolith thus limiting the oblique entry loss at the lower incidence angle.
|Date of Award||2012|
|Supervisor||Carol Roberts (Supervisor)|
- catalytic converters
- pulsating flow
- exhaust catalysts