Frans L. Plantenga
Akzo Chemicals BV Amersfoort
The Netherlands
Yoshimasa Inoue, Naoyuki Torihara
Nippon Ketjen Co. Ltd.
Tokyo
Feedstock contaminants, as well as maldistribution of oil and incorrect catalyst loading in the guard reactor, can lead to serious malperformance of hydrotreaters and hydrocrackers.
However, proper liquid distribution, catalyst loading, and selection of catalyst-in particular, a novel iron and scale-removal catalyst-can improve the operation of hydroprocessing units.
In the past, many resid units in Japan have been operated with heavy feeds at very high severity. High levels of contaminants like Ni, V, Fe, Na, and Cl, combined with high levels of coke precursors and high operating temperatures, have made the operation of the guard reactors of these resid units especially critical.
Small deviations from the ideal liquid distribution and random plugging in portions of the catalyst bed lead to rapid coke formation (and hot spots). Once such areas are plugged, they accelerate the complete plugging of a guard reactor.
These problems initiated a research program at Nippon Ketjen Co. Ltd. The targets of this program are to find the causes of the plugging and to solve or improve the operation by better catalyst technology.1 2 The problems that were studied are:
- Liquid distribution
- Catalyst grading and loading procedures
- Catalyst selection.
LIQUID DISTRIBUTION
It is well known that the feed oil and H2 should be evenly distributed over the total surface area of the catalyst bed. Moreover, the linear mass velocity of the oil and gases should be high enough to ensure complete wetting of the catalyst particles and constant "renewal" of the oil around the particles.
To provide a good distribution of the oil and gas at the reactor inlet, a well-designed distributor plate-operated at its design conditions-is absolutely necessary. Many commercial units are not equipped with adequate distribution systems, causing malperformance of these units.
Although the design of distribution systems is not the subject of this article, it is essential to start with a homogeneous liquid distribution at the top of the reactor.
The catalyst system can have some corrective effect, but will never restore the oil distribution to ideal. The oil flow in a clean catalyst bed at start-of-run (SOR) can be affected by the catalyst in two ways:
- By the method of catalyst loading
- By the shape and size-grading of the particles.
The best method of catalyst loading is denseloading. This should preferably be done by experienced operators with high-quality equipment and technology.
Denseloading prevents the problems that can arise with sock loading. A well-known problem with sock loading occurs if the sock is positioned in the center of the reactor and moved up without leveling the catalyst.
The result is a high density in the middle and particle orientation towards the reactor wall. Both will promote a preferential flow along the reactor wall.
To study the effects of catalyst-particle size and grading, Nippon Ketjen constructed a large cold-flow model pilot plant. Different distributor systems and catalyst systems can be installed in this 4-m long, 60-cm diameter model reactor, shown in Fig. 1 .
Liquid and gas can be circulated through this plant with velocities typical of commercial reactors. The model liquids are chosen so that the viscosity and surface tension match those of typical refinery feeds.
A liquid collector mounted at the bottom of the reactor measures the flow through the various sections of the bottom. The collector, shown in Fig. 1, was divided into 21 sections of equal surface area.
Three different catalyst loadings were compared at different linear velocities (Fig. 2). The inhomogeneity of the distribution is expressed as s/x x 100%, where "s" is the standard deviation of the flow per section and "x" is the average flow.
Fig. 2 shows that grading with different shapes and different layers can improve the distribution, especially in the low mass velocity range (< 2,000 lb/sq ft/hr).
It is obvious that relatively high mass velocities (2,000 lb/sq ft/hr) produce better liquid distribution. The critical velocity depends on both the liquid viscosity and the gas-to-oil ratio.
Optimal liquid distribution is the result of:
- Proper design of the unit and the feed distributor
- Proper catalyst loading with denseloading and grading of the catalyst bed.
CATALYST FOULING
A catalyst bed is a filter. The presence of particulate matter in the feed will result in the gradual laydown of those particulates in the voids between catalyst particles, preferentially in the top layer of the bed.
This layer will eventually become dense and cause reactor plugging.
Even with feed filters, particulates can still be present in the feed, e.g., coke fines from furnace tubes, scale from after-filter feed lines, and scale from filter breakthrough.
One way to improve this situation is to spread the deposits over a larger volume of the catalyst bed. This can be done by size-grading layers of catalysts with different void fractions. The catalyst with the larger diameter has the greater interparticle void fraction.
To demonstrate this dispersion effect, the pressure-drop buildup over three different catalyst grading systems was studied.
In a small cold-flow model unit (Fig. 3), oil containing particulate matter can be circulated over a catalyst bed, and the pressure drop can be measured.
Fig. 4 shows the three catalyst systems studied. System A is without grading; System B is the classical grading with a different diameter hydrotreating catalyst; and System C is loaded with a new guard catalyst, Ketjenfine (KF) KG-1.
The difference between Curve A and Curve B (Fig. 5) clearly illustrates the benefits of grading on pressure-drop buildup.
The further improvement seen with KF KG-1 (Curve C) is not because of a larger void volume, but because this catalyst can absorb macroscopic particles present in the feed (diameter < 200 m).
Hence, particulates can be deposited not only in the void volume between the particles, but also in the pore volume of the catalyst.
Fig. 6 shows the particulate distribution as a function of the catalyst-bed depth for the three systems studied.
The conclusions are very clear:
- No grading results in plugging of the top layer only.
- Grading results in a larger void volume in which deposits are layed-down.
- Application of a superporous catalyst produces the largest storage capacity for scale particulates.
IRON SCALE
The problem of Fe compounds and Fe particulates has become particularly apparent in atmospheric and vacuum residue processing. It is believed that most iron originates from corrosion.
Residues contain considerable amounts of naphthenic acids, which cause corrosion. The final result of this corrosion is Fe naphthenates and Fe particulates in the feed. The exact structure of these "compounds" has not been further elucidated.
Fe compounds have been found in residues in considerable amounts (typically 5-50 ppm), even after the feed filter.
If an Fe-containing feed is passed over a conventional catalyst bed, both the Fe particulates and the Fe naphthenates are converted to FeS.
Part of this FeS will be deposited in the intraparticle volume and part of it on the outer surfaces of the catalyst particles. None of the FeS will penetrate into the catalyst particles, not even into large-pore resid demetallization catalysts.
Most of the FeS will be deposited between the catalyst particles, and if this occurs at high temperatures as applied in some resid units, the FeS will act as a cokemaking (dehydrogenation) catalyst.
Table 1 shows some typical chemical analyses of fines deposited between catalyst particles, proving a relative abundance of Fe, S, and coke (C). The composition of the fouling pattern is further elucidated by performing an electron probe microanalysis (EPMA) on a "fused" lump of catalyst.
Fig. 7 shows an element density distribution map for six elements. From these pictures it is clear that:
- Nickel and vanadium are present mostly inside the demet catalyst (as expected).
- FeS is present only between the particles and as a very thin film around the catalyst particles.
- Coke is present mainly between the catalyst particles, with the highest concentration between the FeS particles.
HEAVY METALS
Heavy metals such as Ni and V are present mainly in residues and, in smaller quantities, in heavy vacuum gas oils (HVGOs). All of this Ni and V is contained in large organic molecules with porphyrin structures.
Under hydroprocessing conditions, these molecules are demetallized and the metals deposited in the catalyst pores as Ni and V sulfides.
Special demetallization catalysts are used to remove the metals from residues and trap them.
OTHER IMPURITIES
Other impurities sometimes present in feeds are Pb, As, and Si. All of these impurities are more-or-less severe poisons for the catalyst, and they are all absorbed by conventional catalysts.
Akzo has developed a special type of catalyst for trapping silicon compounds, which come from antifoaming agents in cokers.
CATALYST DEVELOPMENT
With the aforementioned fouling mechanisms in mind, Nippon Ketjen has developed a catalyst with the following functions:
- Capacity to remove large scale particles (0-200 m)
- Activity for removal of organic iron compounds
- No activity for removal of Ni + V.
KF KG-1 5B and KF KG-1 3B are the results of this development program. KF KG-1 is a spherical catalyst that is available in two diameters: 5 mm and 3 mm.
The ability to trap scale particles was already demonstrated with the cold-flow model test. The conditions in a pilot test to demonstrate the improvement in pressure-drop buildup are described in Table 2.
Under resid mild hydrocracking operating conditions, the pressure-drop buildup of KF KG-1/KFR-30 was compared with KFR-30 alone. KFR-30 is a dual-function resid hydrodesulfurization (HDS)/hydrodemetallization (HDM) catalyst.
Fig. 8 shows the pressure at the reactor inlet required to keep the reactor outlet at constant pressure. The KF KG-1 system provides stable operation, whereas the system without KF KG-1 plugs relatively quickly.
The results of the analyses of two basket samples, which have been in use side-by-side in a commercial unit for 8 months, showed on a weight basis that KF KG-1 5B picked up 15 times more Fe than the conventional demet catalyst. On a volume basis, this number is 30.
Another notable difference is in the Ni + V pickup. KF KG-1 removes almost no Ni + V, whereas a considerable amount is loaded on the demet catalyst.
An analysis of the Fe profiles of the spent catalysts shows that FeS is deposited throughout the KF KG-1 particles, whereas it hardly penetrates normal demetallization catalyst particles.
KF KG-1 effectively removes Fe scale and organic Fe, and also has the ability to store the deposited FeS in the catalyst pores. Moreover, other particulate matter can be trapped, thereby decreasing the chance of blocking the oil/gas flow.
Ni + V REMOVAL
Ni and V can be effectively removed by different types of catalysts, depending on the feed quality.
KF VGO-Demet was developed to provide a good Ni + V removal and storage capacity while retaining the good HDS activity and stability of small-pore HDS catalyst.
The function of KF VGO-Demet is to remove more metals (Ni + V) and to protect the downstream catalysts, thereby increasing cycle length.
Commercial operation has shown that 30-50% more Ni + V can be removed using KF VGO-Demet. Fig. 9 shows the effect of KF VGO-Demet on the distribution of metals over the catalyst bed.
COMMERCIAL EXPERIENCE
KF KG-1 was introduced in 1989 (at first, only in resid operation). Since that time, more than 19 batches have been delivered to 13 units.
In several units, it has been applied in consecutive cycles. In one particular resid train, the guard reactor had completely coked up during each past run.
The application of KF KG-1, however, led to an almost free-flowing catalyst during unloading and a considerable savings of time and effort.
KF KG-1 is also applied in some vacuum gas oil and coker gas oil (CGO) units. A unit in Japan, treating CGO, experienced pressure-drop buildup. The pressure drop was attributed to iron scale and coke fines accumulating in the first reactor.
Analyses of the fines showed Fe, S, and C as major constituents. The average particle size was 66 m, despite the fact that 50 m feed filters were used. Moreover, many particles over 100 m were detected.
In this unit, the pressure drop has been monitored carefully over the last three cycles (1 year each). In the last cycle, KF KG-1 was loaded in the first reactor, replacing the normal top-layer catalysts (also 5 mm and 3 mm diameter). The 15 cu m of KG-1 was about 2% of the total catalyst volume.
Fig. 10 shows the normalized pressure drop as a function of time-onstream for the three cycles. The result is striking. The 1988 and 1989 cycles without KF KG-1 show the same rate of pressure-drop buildup. During the 1990 run, the rate of pressure-drop buildup is reduced by more than 50%.
REFERENCES
- Fujita, K., Torihara, N., Miyauchi, Y., and Inoue, Y., Proceedings Tocat 1, July 1-5, 1990, Tokyo.
- Fujita, K., and Torihara, N., Nippon Ketjen Seminar, Tokyo, 1990.
Copyright 1991 Oil & Gas Journal. All Rights Reserved.
