RESID DESULFURIZATION-CONCLUSION SIMPLE CHANGES REDUCE CATALYST DEACTIVATION, PRESSURE-DROP BUILDUP

Nov. 20, 1995
Hiroki Koyama, Eiichi Nagai, Hidenobu Torii, Hideaki Kumagai Japan Energy Corp. Kurashiki, Okayama, Japan In response to problems experienced in its resid hydrodesulfurization (HDS) unit, Japan Energy Corp. conducted commercial trials involving several operational changes. These changes solved the problems of pressure-drop increase, catalyst deactivation, and formation of hot spots in the catalyst bed.
Hiroki Koyama, Eiichi Nagai, Hidenobu Torii, Hideaki Kumagai
Japan Energy Corp.
Kurashiki, Okayama, Japan

In response to problems experienced in its resid hydrodesulfurization (HDS) unit, Japan Energy Corp. conducted commercial trials involving several operational changes.

These changes solved the problems of pressure-drop increase, catalyst deactivation, and formation of hot spots in the catalyst bed.

Part 1 of this series described the unit's operating problems and the cold-flow model tests that evaluated potential solutions (OGJ,Nov. 13, p. 82). This article concludes the series by describing the effects of the courses of action taken in the resid HDS unit.

COMMERCIAL TRIALS

To prevent the formation of hot spots, the methods confirmed to improve liquid distribution in the cold-flow model experiments were implemented in the Mizushima refinery commercial reactors. These changes included:

  • Improving catalyst-loading procedures
  • Using cylindrical catalyst particles
  • Installing new liquid distributors
  • Implementing product oil recycle.

Fig. 1 (29379 bytes) shows the effect of each operational change on the temperature deviation in the first bed.

The temperature deviation is measured as the average of the standard deviation in temperature and is an index of the degree of mal-distribution. A small number indicates good liquid distribution.

As shown in Fig. 1 (29379 bytes), before improving liquid distribution, a hot spot occurred within 2 months. During the first trial run, an increase in mass flow rate in the reactors, brought about by recycling product fuel oil, delayed hot spot occurrence a little.

In the second trial run, the sock-loading method was improved to avoid forming a slope on the bed surface. The catalyst was scattered over the surface of the bed rather than dropped in the center. This method greatly improved liquid distribution.

Finally, changing the catalyst shape to cylindrical, rather than trilobe, and replacing the existing liquid distributors maintained good liquid distribution throughout the third trial run.

Uniform radial distribution of coke and metal deposits in the first bed also verified the improvement in liquid distribution. The unit has experienced no hot spot problem since the changes. It has been demonstrated, therefore, that good liquid distribution prevents the formation of hot spots.

Shaped catalysts have higher volume activity than cylindrical catalysts when a reaction rate is controlled by diffusion) The shaped catalysts, therefore, have been commonly used for residue hydrodemetallization.

Because the effect of even liquid distribution is important to catalyst performance in large-scale commercial reactors, catalyst shape should be carefully selected to maximize catalyst utilization.

PRESSURE-DROP BUILDUP

Pressure-drop buildup had been a frequent problem in the first bed of the commercial reactor. The solid line in Fig. 2 (21604 bytes)(shows a typical curve for pressure-drop increase in the first bed.

Pressure drop started increasing about halfway into a run and grew exponentially with time. The unit was shut down before the pressure drop reached the stress tolerance of the catalyst bed support.

When the unit first experienced such a serious pres-sure- drop increase, inspection revealed that iron sulfide fines had plugged the surface of the catalyst bed. The iron sulfide particles suspended in the feedstock are so fine that most of them pass through a mechanical feed filter and carry over to the reactors.

To combat this problem, Japan Energy placed a large-size catalyst with high voidage on top of the small catalyst to increase the capacity for solid deposition at the top of the bed.

When the fraction of vacuum resid in the feed was increased, however, the amount of iron solids deposition increased and again caused a pressure-drop increase in the first bed.

Catalyst samples were taken at various bed depths to investigate the cause of the pressure-drop increase. The solids were separated from the catalyst particles and their components were analyzed.

A large amount of solids had deposited at the top of the layer of small catalyst particles. This led to the conclusion that the solids had deposited throughout the layer of large catalysts and plugged the small-catalyst layer.

The solids comprised not only iron sulfide, but also coke and other inorganic compounds. This indicated that dispersing the inorganic solids over the reactors would be effective in pre- venting pressure-drop increase.

Cold-flow model tests using a slurry of fine solids showed that the rate of solids deposition on the catalyst particles was proportional to the catalyst's specific surface area (the ratio of the particle surface area to the particle volume). This value varies with catalyst size and shape. (The catalyst voidage also varies with the shape.)

This result suggests that it is possible to control the dis- tribution of inorganic solid deposits along the bed, as well as the ultimate bed voidage, by properly selecting the size and shape of catalysts, depending on the bed positions.

Several catalysts with different sizes, shapes, and voidages were layered in the first bed. The catalysts in the upper layers have a higher capacity for solid deposition. The catalysts in the middle layers have a medium tolerance against solid deposition, but they decrease the deposition rate.

To prevent coke formation, some extent of catalyst activity should remain at the end of a run. The activity and pore structure of the catalyst in each layer, therefore, were carefully selected.

The dotted line in Fig. 2 (21604 bytes) shows that adopting an appropriate catalyst combination prevented a pressure-drop increase in the first bed.

The distribution of iron solids in the reactors are compared in Fig. 3 (34126 bytes). Before the catalyst adjustments, about 30% of the iron sulfide that had been carried over to the reactors had accumulated on the first bed. The new catalyst combination decreased carryover to 15%.

This demonstrates that dispersing the inorganic solids throughout the reactors also is effective in preventing a pressure-drop increase in the first bed.

Fig. 3 (34126 bytes) also shows that dispersing the solids did not increase the amount of iron solids deposited in the lower beds. This is consistent with the fact that no serious pres-sure-drop increase has been observed in the lower beds since the catalyst combination was changed.

The authors conclude that, before the catalyst adjustment, only smaller solid particles had passed to the lower beds, because most of the larger particles, which deposit quickly on the catalysts, had distributed in the first bed.

CATALYST DEACTIVATION

Activity tests of the catalysts used in the commercial reactors were conducted in a bench-scale reactor.2 The aged catalyst samples were taken from the second through fourth beds, where the HDS catalyst was packed.

The aged catalysts were Soxhlet-extracted with toluene, then dried. The activity tests were conducted on fresh and aged catalysts using Arabian Heavy atmospheric residue at 360 C. and 12 MPa.

Table 1 (11484 bytes) summarizes the chemical analyses and HDS activities of typical aged catalyst samples, relative to fresh catalyst. The catalyst in the fourth bed contained the most coke and showed the lowest HDS activity.

The large amount of coke on the catalyst in the fourth bed significantly decreased the active sites on the catalyst surface and restricted the diffusion of reactants in the catalyst pores. These effects caused the highest degree of deactivation.2

Comparing the catalyst activities in the fourth bed at different residue conversions, the authors conclude that, at high residue conversion, catalyst deactivation in the fourth bed is controlled by coke deposition.2 Because the catalyst activity in the fourth bed has the greatest influence on total reactor activity, it is important to minimize its deactivation resulting from coke deposition.

Coke is thought to be produced by the precipitation of large-molecule hydrocarbons, such as asphaltenes, when their solubility in oil is decreased.3 4 An increase in the conversion of vacuum residue increases the aromaticity of asphaltenes and decreases the aromaticity of maltenes.5 Consequently, the solubility of the asphaltenes in the maltenes decreases.

Absi-Halabi, et al., proposed that a decrease in asphaltene solubility is partly responsible for the steady build-up of coke on the catalysts, subsequent to the initial rapid coke deposition caused by the absorption of asphaltenes on the acidic sites of an alumina support.4 This may explain why an increase in residue conversion increases the amount of coke in the reactor exit.

BED-TEMPERATURE PROFILE

If the solubility of asphaltenes at the reactor exit can be controlled, coke deactivation in the fourth bed may be minimized. An increase in the aromaticity of the asphaltenes, resulting from increased residue conversion, can be controlled by hydrogenation of unstable intermediates, or cracked asphaltene radicals, which easily polymerize or condense.

The hydrogenation rate is usually slow compared to the rate of polymerization or condensation in fixed-bed reactors. For this reason, controlling the residue conversion rate, or the production rate of the cracked asphaltene radicals, at various reactor positions may be effective in hydrogenating those intermediates sufficiently. The residue conversion in each bed can be controlled, to some extent, by adjusting the temperature in each bed.

Two different reactor temperature profiles were tested in the commercial runs. Fig. 4 (25083 bytes) compares the reactor temperature profiles at the middle of the runs.

The temperature in each bed was controlled by adjusting the reactor inlet temperature or the flow rate of quench hydrogen injected between the beds. The weighted-aver-age bed temperature was controlled to maintain residue conversion at the reactor exit nearly constant throughout the runs. An increase in the reactor inlet temperature, therefore, increases residue conversion in the upper beds and decreases it in the lower beds.

Fig. 5 (25132 bytes) shows catalyst activities vs. metal deposition on the catalysts in the third and fourth beds at two different temperature profiles. The catalyst activity of each bed was calculated from the correlation between hydrogen consumption and bed temperature increase.

Fig. 5 (25132 bytes) suggests that the catalyst in the fourth bed was heavily deactivated by coke fouling. As the reactor inlet temperature was increased, the rates of catalyst deactivation in the third bed, and especially the fourth bed, declined.

This effect suggests that decreasing the residue conversion rate in the fourth bed decreased the rate of coke formation. Although the catalyst in the second bed showed slightly higher deactivation rates, the total catalyst activity was maintained at a higher level.

The authors conclude that controlling asphaltene solubility in the reactors by adopting an appropriate reactor temperature profile is effective in minimizing catalyst deactivation resulting from coke deposition in the fourth bed. In addition, this solution maximizes average catalyst activity for the run. It should be noted that solving the hot spot problem in the first bed has enabled Japan Energy to increase the reactor inlet tem- perature.

The result of this combination of simple operational changes allowed Japan Energy to increase run length in its resid HDS unit to 6 months.

REFERENCES

1. Cooper, B.H., Bonnis, B.B.L., and Moyse, B., OGJ, Dec. 8, 1986, p. 39.

2. Koyama, H., Nagai, E., and Kumagai, H., to be published in ACS Syrup. Series, 1996.

3. Wiehe, I.A., Ind. Eng. Chem. Res., Vol. 32, No. 11, 1993, p. 2,447.

4. Absi-Halabi, M., Stanislaus, A., and Trimm, D.L., Applied Catalysis, Vol. 72, Elsevier Sience, Amsterdam, 1991, p. 193.

5. Takatsuka, T., Wada, Y., Hirohama, S., and Fukui, Y., J. Chem. Eng. Japan, Vol. 22, No. 3, 1986, p. 298. Copyright 1995 Oil & Gas Journal. All Rights Reserved.