Previous studies have shown that an automotive catalyst is generally deactivated by two mechaD:isms; ~isoning and/or sintering. Poisoning is a chemical process, and sintering is induced by exposure to high temperatures. Many authors have shown the .differences between fresh and aged catalyst samples, but due to the destructive natl¥'e of-analysing samples the time dependency ofthe deactivation is unclear. This thesis describes an experimental procedure which reveals the time dependency of catalyst deactivation. It also presents a mathematical model, described uSing CFD (Computational Fluid Dynamics) that predicts the deactivated state of the catalyst. The model was validated by the experimental findings. .' The experimental procedure involved aging a selection of catalyst cores on an engine test bed with varying engine operating conditions to encourage the different deactivation mechanisms to act. Three deactivation regimes were created, LTD100 (mainly poisoning), HTD40 (sintering and poisoning) and HT40 (mainly sintering); they each separately deactivated 7 cores held within a core holder. The aged cores were subjected to light-off tests (warm-up tests to see conversion response, including steady state), and XRF (X-Ray Fluorescence) analysis to examine the contaminant levels within the cores. The early work of Oh et al (1983) along with papers from Matsunaga etal(1998) and Baba et al (2000) provided the starting point for two separate poisoning and sintering . models. The poisoning and sintering models were combined to create an overall deactivation model that used a parameter, e to account for the loss of area due to simultaneous poisoning and sintering. When the catalyst was fresh, ewas I and as it deactivated e reduced towards zero. e was ~en incorporated into a..... three:'way catalysis scheme to proportionally reduce the rates of reaction. This resulted in a delayed and degraded light-off when modelling the experimentally aged catalysts. The mathematical model also estimated the accumulated phosphorus by using a poisoning rate derived from Oh et al (1983). The experimental investigation succeeded in its attempts to degrade the catalyst cores. A decrease in activity was noted both for the light-off and the steady state conversion of all aged samples. Light-off temperatures were increased by as much as 80°C when aged. A linear increase in phosphorus was noted with time. No saturation limit was reached, indeed the most severely degraded core still accumulated phosphorus at a similar (linear) rate as when fresh. The high temperature aging regime captured more phosphorus than the low temperature aging regime. The mathematical model largely predicted the findings ofthe experimental proc~dure. The predicted light-off seemed to correlate well in most cases with the experimental data, although the steady state conversion was lower in the experimental cases. The poison accumulation predictions underestimated the phosphorus found on the catalyst cores. This was thought to be due to the predicted phosphorus occupying an active .:'.. site, and the measured amounts of phosphorus including the 'excess phosphorus', or phosphorus which had built up in layers upon itself. Differences betw~en the HID . and the HT regime's effect on light-offwere also predicted quite well. Both the predictions and the experimental cases showed a trend in T50 temperatures at early time intervals. Large increases in T50 were noted between the fresh and least aged samples (- 40 to 60°C), then subsequent aged light-off curves were more closely packed. The spread of the T50 for the a,ged samples typically covered 20°C. This indicated that catalysts lose the majority oftheir activity very early on, probably as a result ofsintering. The main experimental fmding of the work is the linear time history of deactivation and a major contribution is the development of a combined poisoning and sintering model. The model has great scope in helping to understand the processes of catalyst deactivation. Because the model is described using a commercfally available CFD code, the model can be extended to describe deactivation mechanisms in 2D and 3D.