Catalyst coke level

The design frequency of regeneration is normally from three to six months for semi-regenerative units, and one reactor every 24 hours in cyclic units. For either case, an increase in regeneration frequency would result in a reduction in average catalyst coke level. Thus, gasoline yields would increase and catalyst requirements decrease. 

A large quantity of hydrogen-rich separator gas is normally recycled with the feed stream. Recycle rates may vary from 2,000 to 10,000 MSCF/B. The recycle gas serves to suppress catalyst coke make but normally has relatively little direct effect on gasoline yields or catalyst requirement. However, at lower recycle levels, where an increase in recycle rate may significantly increase reactor hydrogen partial pressure, the effect is similar to a small increase in total... 

In another study by the same group, K, Ca and CaK promoters were added to Rh supported on MgAl204 spinel. Both modified and unmodified catalysts produced similar H2 and CO reformate concentrations of 23 and 25 vol%, respectively. The different modifiers did affect carbon production on the catalysts. The unpromoted Rh catalyst showed carbon levels of 0.03 wt% carbon, where the RhK catalyst had 0.02 wt% carbon, the RhCa 0.015 wt%, and the RhCaK only 0.01 wt%. Bimetallic Pt Rh with Li, Ba and Li-Ba modifiers supported on MgAl204 spinel were also examined for H2 and CO yields and carbon formation resistance, with similar results to previous work for the yields. The unpromoted Pt-Rh catalyst showed carbon levels of 0.01 wt% carbon, where the promoted Pt-Rh catalysts all showed reduced coking levels of 0.005 wt%. 

The conversions are predicted quite accurately over a wide range of coke levels (from 0.7 to 2.7 wt %). There are some minor discrepancies in the coke predictions. However, these predictions are within the experimental accuracies of the tests. The FFB conversions are reproducible within 2 numbers, with 20% accuracy for coke yields. Similarly, the MAT data are reproducible within 3 numbers with 30% accuracy for coke yields. These comparisons suggest that the models could be used to fit the catalyst intrinsic parameters (kj, A,) from the transient tests and then predict the steady state riser performance. Indeed, we have done this exercise and we predict a reverse ranking for catalysts A and C in going from transient tests to the riser performance, as observed in the riser experimental data. 

In this study we address hydrocracking of VGO at moderate hydrogen pressures (30 bar) and elevated temperatures (450°C) using catalysts with little or no acidity [1]. The moderate pressures are attractive from a capital investment point of view. A potential drawback could be that the severe conditions lead to considerable coke deposition on the catalyst. In order to control the level of catalyst coking a careful balance of catalyst and process parameters is a prerequisite. 

The relative number of active sites as deduced from thiophene HDS (of spent catalysts) is also shown in Figure 4 for the Mo/Si02 catalysts. Fully in line with the XPS results, independent of the coke level the number drops to about 25% of that of the fresh catalyst. Comparison of the coverage of 50% and the remaining number of active sites (25%) suggests that the coke is not randomly distributed over the catalyst but is to some extent primarily deposited onto the active phase. The latter conclusion can also be deduced from the results of the XPS study [7J. 

Since DBT does not affect the coking rate, it is possible to measure HDS activity while coking the catalyst with pyrene. Results are shown in Fig. 7 for three repeat tests of HDS activity as a function of run length. The three tests were operated for different periods of time 40 hours, 65 hours and 110 hours. The resultant levels of carbon for the samples aged 40 and 110 run hours fit (9.5% wt and 13.5% wt, respectively) the data in Fig. 6 perfectly. However, the carbon level found for the 65 run hour aged sample was somewhat larger than expected, 15.7% wt vs. the expected 12.5% wt. The reason for deposition of the additional coke is unresolved. The data in Fig. 7 (solid boxes) show an unexpected activity drop between run hour 40 and 50. This activity drop is most likely caused by coke deposition on the catalyst (coke is the only source of deactivation during these runs ). We... 

Allowing for some spread in the data it seems as if little deactivation is caused by the first 4-5% wt carbon deposited, after which there is an exponential activity decline. This deactivation behaviour is of course indicative of the way in which coke is deposited on the catalyst surface, The initial deposition of coke mainly takes place on the bare A1203 surface, i,e. does not interfere with the active phase as demonstrated in a previous paper [6], At higher coke levels we observe an exponential activity decline indicative of a fouling type of deactivation rather than selective poisoning. 

The coke levels found for deactivated res id catalysts range from 10-20% wt on a spent catalyst basis (12-35% wt on a fresh catalyst basis). On the basis of Fig. 8 we would thus expect residual HDS activity of coke resid catalysts to be in the order of 10-40%. 

The apparatus has been shown to yield reproducible data on catalyst activity and selectivity as unambiguous functions of coke level. 

We continue to rely extensively on the two-step (initiation - propagation or autocatalytic) model 4) to evaluate data on coking rates. Two rate constants are involved fc for the deposition of coke on a "clean" surface, i.e., with no coke around and k2 when coke is deposited adjacent to another coke deposit. The former rate constant is for an initiation step (or "non-catalytic" coking), while the latter is for the propagation step (or coking catalyzed by the presence of the coke "product") hence, typically, k2 > ki. A third parameter used in the model is M, which represents the maximum amount of coke which can be deposited on the catalyst. In terms of these three parameters, the coke level expected in a pulse reactor after the passage of R amount of reactant is given by ... 

The values of the three parameters can be obtained by fitting the experimental data to Equation 2. The form of (R) is S-shaped, and the location of the inflexion point (the maximum "rate" of coke deposition) plays an important role in the coke / activity / selectivity behavior of a catalyst in general (4). The coke level, at the inflexion point of the curve can be obtained from the fitted values of the three parameters. 

Activity. We concentrate on the conversion of the "actual" feed reactant, cumene, as the measure of activity. We note how the conversion changes when different amounts of different coking additives (decane, naphthalene) are mixed with the feed, and pulsed over a catalyst of different coke levels. We also report data on the conversions of the additives decane or naphthalene under the same conditions. As mentioned earlier, cumene conversion is obtained by carrying out a benzene-ring balance on the contents of the sample collector after each pulse procedure, while conversions of naphthalene or decane are obtained by comparing peak areas with and without catalyst. 

Here a is the conversion corresponding to the fresh, uncoked, catalyst while b denotes a measure of the sensitivity of the catalyst activity to coke level. We use the parameters a and b to quan the role of the added species on the catalyst activity. 

Figure 9. Activity of catalyst for decane conversion (by comparison with non-catalytic run) in lS%cumene - 85%decane mixture as a function of coke on the catalyst. The straight line indicates the best fit for coke levels less than 7%.

The performance of the coked catalyst is quantified by the initial conversion, a, and the sensitivity to coke, h, viz., the intercept and slope of the conversion - coke relationship, at least for coke levels less than . The effect of reaction mixtures on a and b dq)ends upon the nature of the feed and the species added. When the additive is less-rapidly coking, then values of a and h for the feed itself change with additive concentration. When the species added is more-rapidly coking, then a and b for the feed itself are independent of the additive concentration, i.e. independent of A- But since only a more-rapidly coking species would be an additive for a typical accelerated-coking test, it can be concluded that the presence of the added species does not influence the activity, at least upto coke levels equal to... 

---