Catalyst Deactivation - Spent Hydroprocessing Catalysts - page 87

ensures an ideal mixing, in which both radial and axial dispersion or back mixing are
neglected. Without such simplification, the development of models would be rather complex
and difficult. The complexity of modeling on reactor level is confirmed by the results published
by Tamm et al. [197] , which indicate significant differences in the catalyst deactivating
patterns between the inlet and outlet of the reactor. On the basis of microscopic evaluations,
these authors observed that the V and Ni deposition profiles exhibited either U or M shape, i.e,
the maximum of metal deposition was located either at the catalyst particle surface or inside
the catalyst particle. This was dependent on the location of catalyst particles in the fixed-bed.
Thus, the M shape profile was observed in the reactor inlet, and the U shape in the reactor
outlet. The maximum of metal deposition shifted inward with the decreasing temperature,
decreasing H 2 pressure, decrease in pellet diameter and increase in the average pore
diameter.
The model developed by Khang and Mosby [282] was based on pore-filling by metals and its
suitability was verified using the macroporous catalyst for applications in the trickle-bed
reactor. Deactivation process could be expressed by two adjustable Thiele modules, i.e., one
assuming the bulk diffusivity and kinetics, and the other effecting diffusivity and kinetics in
deactivated pores. The model was suitable for predicting deactivation curves for the HDS and
HDM reactions in a good agreement with the experimental data before less than 50% of pores
were filled with metals. The model data showed a reasonably good agreement with the results
obtained in pilot plant such as the HDM activity and metal loading between the inlet and outlet
of the fixed-bed reactor. A modified form of the model developed by Khang and Mosby [282]
was used by Togawa et al. [122] to analyze five sets of the deactivation data obtained from the
commercial operation employing ARDS process using a Kuwait residue as the feed. After
modification, the model was applicable; up to 80% of the pores were filled with metals
compared with 50% observed by Khang and Mosby [282] . An improved simulation using the
modified model was also reported by Kam et al. [283] on the catalysts.
The integrated mechanistic reactor model developed by Kam et al. [284] considered the initial
rapid catalyst deactivation by coke deposition. In this case, a distinction was made between the
“soft” coke formed during very early stages and the “hard” coke formed in the steady state of
catalyst deactivation. At the same time, during the start of run period, the deactivation by
metals was much less evident, whereas in the middle of run, and particularly before the end
was approached, the metal deposition was the dominant mode of deactivation. Figure 4.25
compared the simulation data with those obtained in pilot plant during more than 400 days on
stream. These results are for the second reactor of the four-reactor process. For HDS, a good
agreement between the predicted and measured results was obtained for less than 4,000 h on
stream, whereas the large discrepancies were observed for the HDAs results. Similarly, a good
prediction was made for HDV and hydrodenickelization (HDNi). However, the deviation of
the HDM data became more evident beyond 4,000 h on stream. In an effort to further advance
their model, Kam et al. [284,285] assumed that a part of the V and Ni were removed from