Catalyst separation technology improves FCC gasoline yields

Tiffin E. Johnson, Terry L. Goolsby, Michael A. Silverman M.W. Kellogg
Technology Co.
Houston

Dennis C. Kowalczyk
Refining Process Services
Cheswick, Pa.

Howard F. Moore
Marathon Ashland Petroleum LLC
Catlettsburg, Ky.

MagnaCat uses simple, high field-strength permanent magnets to separate, remove, and discard the oldest, most metals-laden and least active catalyst from the fluid-catalytic cracking (FCC) inventory.

The MagnaCat unit at Marathon Ashland Petroleum LLC's Canton, Ohio, refinery began operations in 1996.

MagnaCat uses dry, magnetic-separation techniques to separate equilibrium catalysts by age and metal content. This withdrawal of older catalysts results in higher gasoline yield, reduced hydrogen production, improved coke selectivity, and lower catalyst and additives consumption. Increased FCC operating margins at Canton of $0.30-$0.60/bbl are attributed to the magnetic-separation prodedure.¹

Valero Refining Co. plans to begin MagnaCat operations on its 73,000 b/d heavy-oil cracking unit in Corpus Christi, Tex., in the fourth quarter of 1998.

MagnaCat principles

FCC operators add fresh catalyst on a near continuous basis to replace losses through the cyclones and to maintain the activity of the unit's circulating inventory at an acceptable level. As a result, the inventory of catalyst particles in a commercial FCC unit represents a broad age and activity distribution. Newly added particles are relatively fresh and active, and particles that have been in the unit for many months (even up to a year) are catalytically dead.

The particles that have been in the unit the longest are the most deactivated and least selective for cracking to desired liquid products. This lower activity is the result of extended exposure to the hydrothermal deactivating environment of the regenerator, which reduces zeolite-surface area and crystallinity.

Many FCC units include atmospheric or vacuum residue in the feedstock to upgrade more heavy oils to transportation fuels. This inclusion results in a catalyst inventory that is contaminated with metals such as nickel, vanadium, and iron.

The oldest particles have the greatest concentration of accumulated metals, particularly nickel and iron, because the oldest catalyst particles have had the most contact with the metals-contaminated feed. The distribution of the accumulated nickel and iron on catalyst will be similar to the age distribution of the inventory.

Fig. 1 [34,723 bytes] shows the particle age distribution of a perfectly back-mixed catalyst inventory for two daily makeup rates. The greater magnetic susceptibility imparted by the higher concentration of deposited metals on the older catalyst allows MagnaCat to separate it from the newer, less-magnetic catalyst particles.

The process at Canton begins with a slip stream of FCC equilibrium catalyst flowing from the regenerator through a boiler feedwater catalyst cooler. A slide valve regulates the catalyst in a dilute-phase transport to the MagnaCat catalyst hopper.

The slightly fluidized hopper catalyst flows through a cooling water catalyst cooler to further reduce the catalyst temperature from 260 to 100° F. The catalyst then moves to the magnetic-separator feed hopper in a dense-phase mode. Fig. 2 [43,081 bytes] shows the roller magnet assembly, which is the heart of the MagnaCat process.

Equilibrium catalyst is distributed onto a moving belt which has a high field strength permanent magnet in the form of a roller at its far end. As the equilibrium catalyst passes into the magnetic field, the most magnetic catalyst is retained on the belt.

The nonmagnetic catalyst, on the other hand, is discharged from the end of the belt into a chute and returned to the FCC unit. After the most magnetic catalyst leaves the roller's magnetic field, it is discharged into a second chute and discarded.

Fig. 3 [38,666 bytes] shows how MagnaCat changes the age distribution of the regenerator catalyst by withdrawing spent catalyst from the FCC unit.

Removal of oldest catalyst

MagnaCat cleans up the catalyst inventory. The oldest catalyst particles (oldest 10%) have metals-concentration levels that exceed the average metals levels on equilibrium catalyst by 2-5 times. Particles with high levels of metals produce significant amounts of coke and hydrogen via nonselective thermal cracking and dehydrogenation reactions.

Coke and hydrogen production occur even though metals become partially deactivated toward these reactions with age. The oldest particles still experience fresh metals deposition in the same way as fresh catalyst.

Two ratios from Canton's equilibrium-catalyst separation data demonstrate MagnaCat's ability to remove the tail end of the catalyst age distribution curve:

• The ratio of magnetic susceptibility for magnetic reject relative to recycle, which is a measure of the magnetic gradient of the catalyst inventory.
• The ratio of iron content for magnetic reject relative to recycle.

This work was performed on samples of Canton equilibrium catalyst taken prior to the MagnaCat start-up and near the end of the MagnaCat implementation period. For this analysis, lab separation data were used to ensure that the sample separation conditions (including ratio of magnetic reject to recycle) were held constant.

Results indicated that after the implementation of MagnaCat, the magnetic susceptibility ratio dropped from 2.02 to 1.28, and the iron content ratio dropped from 1.34 to 1.09. The drop in these ratios resulted from a rejection of the oldest catalyst particles. Even when the tail of the FCC catalyst distribution is removed, the magnetic susceptibility gradient is still sufficient to allow ongoing magnetic separation of the remaining catalyst inventory.

Large catalyst inventory units will benefit the most from shaving the tail off of the catalyst-particle residence-time distribution curve. Consider two FCC units with the same catalyst makeup rate, one with a large inventory and low-percentage daily replacement, and the other with a small inventory and high-percentage daily replacement. Fig. 1 shows that the large inventory unit has a higher percentage of older, highly magnetic particles.

Therefore, there will be a greater difference in activity between the magnetic reject and recycled catalyst for a large inventory unit than for a small inventory unit. Qualitatively, MagnaCat allows a large inventory FCC to enjoy the benefits of a small inventory unit by significantly lowering the average age of the catalyst in the inventory.

Shaving off the tail of the catalyst age distribution curve permits the unique character of the catalyst, whether it be gasoline, olefins, or coke selectivity, to be realized without being diluted by large amounts of dead catalyst in the inventory.

Pilot plant studies

To gain greater insight into the performance benefits of magnetically separated catalyst, microactivity (MAT) and pilot unit studies were carried out at the Kellogg Technology Development Center. A commercial, metals-contaminated equilibrium catalyst was magnetically fractionated using the standard 10/10/40/40 magnetic separation protocol.

The 10/10/40/40 protocol comprises weight percent fractions: the 10% most magnetic fraction (oldest catalyst particles, highest metals content), the 10% second most magnetic fraction (second oldest catalyst particles, second highest metals content), the 40% third magnetic fraction (good activity, contains some deactivated catalyst particles, lower metals content), the 40% least magnetic fraction (highest activity, lowest metals contents).

Analysis of the catalyst fractions from this separation showed a large magnetic susceptibility gradient from the most to the least magnetic samples indicating good potential for MagnaCat applications (Fig. 4 [32,687 bytes]). Table 1 [35,343 bytes] contains metals analyses for the equilibrium catalyst, the 20% most magnetic fraction, and the 80% least magnetic fraction.

The significant difference in metals content, particularly nickel and iron, confirms the observed magnetic susceptibility gradient. The vanadium concentration gradient from least to most magnetic indicates that mobile vanadium species are not formed in the partial combustion regenerator of this particular FCC unit.

Results of the MAT tests using the separated fractions demonstrated the expected lower catalytic activity of the higher metals catalyst fraction (Fig. 5 [35,228 bytes]). These data confirm previous magnetic separation results published by Ashland.²

An FCC pilot plant study using the magnetically separated catalyst was conducted with a 1/3 b/d circulating riser pilot unit. The sour heavy gas-oil feedstock had an API gravity of 19.5° API, a concarbon number of 1.1 wt %, and a sulfur content of 2.11 wt %.

Comparison of pilot unit results for the least magnetic 80% and the FCC equilibrium catalyst are shown in Fig. 6 [24,374 bytes]Fig.7 [24,596 bytes] and Fig. 8 [31,404 bytes]. The least magnetic fraction has a slightly higher gasoline selectivity but significantly lower hydrogen and coke selectivities than the equilibrium catalyst.

Removal of 20% of the most magnetic catalyst from the equilibrium catalyst results in significantly higher conversion (Fig. 9 [31,205 bytes]). These conclusions are similar to those reported in other studies.³

Again, the reduction in hydrogen yield and higher conversion is accomplished by removing the older catalyst fraction, which had a high nickel content.

Previous Canton results

Results from the Canton initial testing period (April and August 1996) are shown in Table 2 [117,435 bytes] and Table3 [57,618 bytes].¹ During this period, metals were controlled in the 2,000-3,000 ppm nickel and vanadium range. Fresh catalyst addition rates averaged between 0.20 and 0.25 lb/bbl of fresh feed, a 25% reduction compared to the preMagnaCat rate of 0.32 lb/bbl.

Table 2 shows four sets of mass-balanced Canton FCC unit yields corresponding to the test runs. At similar feed rates, about 23,000 b/d, the conversion increased significantly as MagnaCat came on-line. The conversion increase was accompanied by an increase in gasoline, C₃, and C₄ yields.

Table 3 shows catalyst MAT results for Canton's equilibrium catalyst during these four test runs. These data confirm the increased gasoline, decreased coke, and decreased hydrogen selectivities observed from the commercial test-run data.

Canton update

The commercial MagnaCat unit at Marathon Ashland's Canton refinery continues to add value to Canton's operations.

After reviewing this performance, refinery management decided to take greater advantage of the MagnaCat unit and reduce catalyst addition rates to a minimum. For the period of September 1996 through December 1997, addition rates were targeted at 0.18 lb/bbl. Fig. 10 [44,084 bytes] shows that this resulted in an average metals-on-catalyst levels of about 4,000 ppm.

Impressively, this 25%+ increase in metals on the catalyst resulted in little or no loss of gasoline selectivity (Fig. 10). MagnaCat allows the refinery to minimize catalyst addition rates with no loss in catalytic performance, while minimizing catalyst disposal and undesirable metals-catalyzed reactions.

MagnaCat began operations in April 1996, and it is estimated that MagnaCat treated the full catalyst inventory by July 1996. During this transition period, total C₄ yield declined as gasoline yield/selectivity improved.

The ratio of C₄ olefins to C₄ saturates increased, which implies more selective catalytic cracking and more profit for the refiner. As MagnaCat continued to operate through 1997, the low level of C₄ production and high level of C₄ olefins has been maintained (Fig. 11 [31,140 bytes]).

Catalyst additives

Many operators are concerned about the retention of their catalyst additives in conjunction with MagnaCat. Experience at the Marathon Ashland Canton refinery demonstrates that the magnetic-separation technique can improve the retention of SO _x additives, which potentially reduces catalyst additive costs at Canton.

The Canton FCC unit currently uses SO_x abatement additives to control the flue-gas SO_x content. This additive was also used prior to the installation of MagnaCat.

If an additive is recycled with the least magnetic catalyst, it must exhibit a low magnetic susceptibility or it will be rejected with the high-metals catalyst. Measurements show that the magnetic susceptibility of the Desox additive is -0.52 x 10^-6 electron mass units/g (emu/g). This indicates that the SO_x abatement material is diamagnetic (i.e., repelled by a magnet) or exhibits no magnetic properties. This information alone suggests that the Desox additive will remain in the least magnetic fraction and be recycled to the FCC.

Fig. 12 [36,782 bytes]shows the concentration of Desox material for the equilibrium catalyst and the least magnetic recycle at Canton from April to August 1996.

Before and after the MagnaCat unit was installed, Desox material was added at a constant rate to ensure compliance with environmental standards. Fig. 12 shows that prior to MagnaCat, the content of the additive fluctuated between 1% and 1.5% in the FCC equilibrium catalyst. After the MagnaCat start-up, the concentration of the Desox additive in the FCC equilibrium catalyst inventory increased to 1.5-2.5 wt %.

Analysis of the least-magnetic fraction indicated a generally higher Desox concentration than in the equilibrium catalyst. This indicated that the SO_x additive was being recycled to the unit with the least magnetic recycle catalyst. Not all the SO_x additive was recycled to the FCC, however. A small amount of this material goes to the most magnetic reject stream. The additive in this cut was found to be deactivated by metals contamination from nickel, vanadium, and iron. These metals increased the magnetic susceptibility of the additive material and allowed it to be removed with the most magnetic fraction. The rising Desox concentration at constant Desox addition rate indicates that most of the additive is being retained in the least magnetic recycle fraction when the equilibrium catalyst is separated in the MagnaCat unit. Less SO_x additive makeup is required for a given level of flue-gas desulfurization when using MagnaCat.

References

1. Goolsby, Terry L., Moore, Howard F., Kowalczyk, Dennis C., Zampieri, Mario L., and Bussey, B. Karl, "FCC Unit Optimization Using the MagnaCat Process," NPRA Paper AM-97-32, San Antonio, Mar. 16-18, 1997.
2. Kowalczyk, Dennis C., Campagna, Robert J., Hettinger, William P., and Takase, Shinji, "Magnetic Separation Enhances FCC Unit Profitability," NPRA Paper AM-91-51, San Antonio, Mar. 17-19, 1991.
3. Anderson, S., and Myrstad, T., Applied Catalysis, 159, 1997, pp. 291-304.

The Authors