Post-riser quench improves FCC yields, profitability

George P. Quinn
Amoco Petroleum Products
Naperville, Ill.

Michael A. Silverman
Stone & Webster Engineering Corp.
Houston

Yorktown FCC Unit (42474 bytes)

Performance tests of a post-riser quench system have shown increases in gasoline yield; reductions in fuel gas, butadienes, and pentadienes; and improved oxidation stability of FCC gasoline.

Following commercial trials in six of Amoco Petroleum Products' refineries for some 10 years, the system is now being installed by other refiners.

With today's modern FCC process, the highly selective catalytic reactions occur in about 1.5 to 3 sec, with riser outlet temperatures of 980 F. to 1,000 F., in straight vertical risers.1 Once these reactions are completed, the catalyst is rapidly separated from the products using a riser termination device, such as rough-cut cyclones.

At this point, the products normally are discharged into the dilute phase of the FCC disengager. Here, the vapor may be at elevated temperatures for 5-40 sec. Such conditions allow thermal cracking reactions to proceed. These reactions degrade the products formed by the selective catalytic reactions that occurred in the FCC riser.

Elimination of this post-riser thermal cracking is essential to achieving the most desirable product slate. Most of Amoco's FCC units (FCCUs) have rough-cut cyclone systems in which there is significant vapor residence time in the disengaging vessel.

One option that will reduce the impact of these thermal reactions is to "hard-connect" the riser cyclones to the reactor cyclones. Amoco concluded, however, that this was not the best option because of potential process problems.

These potential problems included catalyst carry-over to the main fractionator (because there is no surge capacity in the system), reactor heat-up, differential metal growth, and coking in the reactor.

Another option-the one Amoco chose to pursue-is to reduce the temperature in the dilute phase. This technology involves quenching the dilute phase, thereby reducing the thermal cracking rates. The approach is very effective and virtually eliminates post-riser cracking.

During the period that Amoco developed the quench technology, Stone & Webster Engineering Corp. developed a state-of-the-art riser termination device. Stone & Webster's Ramshorn device and Amoco quench technology complement each other very well (details of this technology will be discussed later).

The Ramshorn device is operating in five FCC units. Another eight are in various stages of design and construction.

Background

Since its introduction in 1942, fluid catalytic cracking has become the primary gasoline-production process in the modern refinery. As the technology developed from low-activity amorphous silica alumina catalysts to today's high-activity zeolitic catalysts, the process has changed significantly.

Specifically, the very active and highly selective modern catalysts have dictated shorter contact time between the hydrocarbons and the catalyst. Moreover, the temperature of the cracking reaction has increased gradually.

As a result of these changes, catalytic cracking reactions now occur in a vertical riser with short residence times and high riser outlet temperatures.1 In fact, some FCCUs run with riser outlet temperatures as high as 1,040 F.

Fig. 1 (39795 bytes) shows examples of three state-of-the-art designs for a modern FCC reaction system.

While many configurations use existing catalyst technology effectively, the essential features of a modern FCC reaction system are:

• High-temperature full CO combustion

• Low carbon on regenerated catalyst

• High-efficiency feed injection technology

• Straight, vertical riser with 1.5 to 3-sec reaction time.

Systems that achieve these objectives, or those that rapidly separate hydrocarbon and catalyst (like that shown in Fig. 1 (39795 bytes)), will accomplish virtually pure catalytic cracking in the optimum riser time. Yields from such systems are highly selective.

Having achieved the objective of cracking the feed over an active zeolitic catalyst in the optimum manner, most units then discharge the product into the dilute phase of the FCC reactor.

Numerous studies have shown that this post-riser vapor residence time leads to thermal cracking and some continued catalytic cracking in the disengager vessel.2 The relative amounts of thermal and catalytic cracking in the dilute phase depend on the separation device employed.

Conventional ballistic separators, such as T-shaped devices, have been shown to deflect large amounts of product vapor directly to the stripper bed. Because the product and catalyst contact one another in the disengager vessel, extensive thermal and catalytic cracking take place.

Rough-cut cyclones, on the other hand, efficiently separate catalyst and discharge catalyst-free vapor into the disengaging vessel. In these systems, the principal post-riser reaction is thermal cracking.

The primary product of catalytic cracking reactions is high-octane gasoline. By discharging this gasoline product into the dilute phase at 950 F. to 1,000 F. for 5-40 sec, the gasoline product is thermally cracked to fuel gas (primarily C2s).

Other products of thermal reactions, such as butadienes, are produced as well. Butadienes that reach the alkylation unit will dramatically increase acid consumption.

Pentadienes also can be produced by thermal cracking in the dilute phase. These products are extremely reactive and have been linked to deterioration of the oxidation stability of FCC gasoline.

To put it simply, after achieving nearly perfect catalytic cracking in the riser, the post-riser reactions then thermally decompose the products in the dilute phase.

Quench technology

Fig. 2 (42172 bytes) shows examples of two Amoco FCCUs with rough-cut cyclone systems employing post-riser quench, and a Ramshorn axial-cyclone system.

Rough-cut cyclone systems have been well documented. Radiotracer studies of these systems show 1.0-2.0 sec of vapor residence time from riser outlet to rough-cut cyclone outlet.

The Ramshorn system separates the vapor from the catalyst at the top of the riser by inducing a 180 turn around a gas outlet tube. This tube is slotted on the leading (inner) sides. As the vapor/catalyst mixture rotates around the gas outlet tube, centrifugal and viscous forces separate the catalyst from the vapor.

The catalyst exits through the slots in the top of the stripper bed via diplegs. After passing through the gas outlet tube slots, the essentially solids-free product gases are ducted to a confined vapor space at the upper cyclone inlets, thereby minimizing backmixing in the disengager vessel.

Radiotracer testing of the Ramshorn axial cyclone system has shown 0.7-sec vapor residence time from the riser outlet to the Ramshorn outlet.3 Another advantage of the Ramshorn system is that it operates at a significantly lower delta P than a rough-cut cyclone system.

The Ramshorn separator provides 98+% separation efficiency. In addition, its diplegs do not have to be buried in the catalyst bed. The catalyst can then be stripped using prestripping steam rings in the dilute phase.

Both the rough-cut cyclone and Ramshorn systems are ideally suited to use with quench technology because the vapor exiting the separator is catalyst-free, and the quench point immediately follows catalyst/vapor separation.

Quench objectives

The negative impact of dilute-phase, post-riser thermal reactions on FCC yields is well known. Vapor-line sampling experiments were conducted at several units and illustrated this impact.

In one experiment, one sample was taken from a probe inserted into the reactor over the vapor outlet from the rough-cut cyclone. A second sample was taken from the reactor vapor line just a few feet from the plenum.

Delta yields (across the dilute phase) were:

• C2- +1.3 wt %

• Gasoline, -1.1 vol %.

Note that, across the dilute phase, fuel gas (C2 2 ) yields increased significantly and gasoline yields decreased substantially.

Traveling thermocouples were placed in about the same locations and showed a temperature decrease of about 20 F. Clearly, cracking reactions were occurring. The primary product of these reactions was fuel gas.

Because the catalyst had been essentially eliminated by the rough-cut cyclones, these reactions obviously were thermally induced. The riser outlet temperatures (950-1,020 F.) were high enough to support thermal reactions.

Radioactive tracer tests conducted on several units indicated that the vapor residence time varied from 5 to 40 sec in the dilute phase, depending on the unit design. This range of times will result in varying degrees of post-riser thermal cracking.

Upon reviewing this information, a program was developed to find a way to diminish the thermal reactions by either reducing the time or the temperature of the cracked product in the dilute phase.

Consider a traditional expression for thermal reactions:

Thermal products ~ (time) (exp (-E/RT))

In this case, the impact of thermal reactions on the products should be related directly to time.

As previously discussed, Amoco looked at an experiment in which the rough-cut cyclones were "hard-connected" to the upper cyclones to reduce the residence time. With this design, however, there was concern about mechanical reliability and catalyst carry-over to the fractionator.

Amoco also recognized that, while this approach could significantly shorten the residence time in the dilute phase, it would not reduce the time in the upper cyclones, the plenum, or the vapor line to the fractionator. (The radiotracer testing had shown that the time from the upper reactor cyclones through to the fractionator could be 3-10 sec.)

Eventually, the cost and complexity of the modifications to reduce the dilute-phase residence time prompted Amoco to find another approach. The other option is to reduce the temperature in the dilute phase and, as it turns out, also in the upper cyclones, the plenum, and the reactor vapor line.

The most often quoted activation energy for thermal cracking reactions is on the order of 100,000 BTU/lb-mole. Applying that value to the equation mentioned previously leads to the conclusion that the products of thermal reactions in the dilute phase will be halved for every 25 F. drop in dilute phase temperature.

Decreasing the temperature by 50 F., therefore, will reduce the products of thermal reactions by 75%. A 75 F. reduction, likewise, will decrease the yield of thermal-reaction products by 88%.

The appealing aspect of this approach is that it is clearly less expensive than hard-connecting the cyclones, in addition to being straightforward and simple. There were, however, some concerns that had to be addressed.

The first concern was that condensation would occur, leading to excessive coking in the reactor vessel. Laboratory tests and simulations were conducted to set the appropriate operating criteria to ensure that this did not occur.

The second concern was that adding a cooling medium to the process would increase the mass flow through the cyclones and into the fractionator. Careful simulation showed that it was possible to select a quench material that would result in either no change or even a decrease in flow volume, despite increased mass flow.

Amoco determined that light cycle oil (LCO) satisfied this criteria. As a result of using this as the quench material, there would be little or no increase in pressure drop through the system.

Finally, there was concern about the impact on the fractionator. Again, simulation showed that, while an adjustment to the heat-removal circuit duty would be necessary, its impact would be small.

More importantly, the reduction in fuel-gas production would actually reduce volumetric flow through most of the trays in the column, and jet flooding would not be a problem.

Quench application

For the first trial of this post-riser quench technology, Amoco selected its Yorktown, Va., FCCU. A schematic of this unit is shown in Fig. 3 (42474 bytes).

The unit was originally an Orthoflow B type, with the reactor vessel below the regenerator. Stone & Webster revamped the unit in 1987, however, to an external vertical-riser configuration.

A unique feature of this design is the external rough-cut cyclone. Cracked products from the rough-cut cyclone are transferred via a vapor line to the reactor vessel. This design enabled Amoco to inject the quench material through a dispersion nozzle directly into the cracked vapor in the vapor transfer line, as shown in Fig. 3 (42474 bytes).

Before installing the quench system, the riser outlet temperature in this unit was 950 F. While this temperature is considered moderate, the residence time of about 20 sec indicated that there was adequate time for thermal cracking to occur. Vapor-line sampling also confirmed this potential.

While the unit was on-line, quench was installed using a hot tap into the vapor line. LCO from the fractionator was selected as the quench material.

Appropriate valving, thermocouples, and control loops also were installed. Performance tests were run several times, with the quench feed both on and off, to quantify the effect of cooling the dilute phase. The results of these tests are shown in Table 1 (13476 bytes).

As expected, the major impact was on C2 2 and gasoline. Specifically, fuel gas yield decreased about 0.4 wt %, while gasoline yield increased 0.6 vol %.

Also of note, the temperature drop across the dilute phase went from 10 F. to almost zero. These results were achieved with only 45 F. cooling of the dilute phase.

Based on this successful trial at Yorktown, Amoco installed dilute-phase quench at five of its other FCCUs. The results of performance tests at two other FCCUs are shown in Table 2 (11044 bytes).

In both these cases, there was a significant reduction in fuel gas production and an accompanying increase in gasoline yield.

Further improvements

While the initial performance testing clearly indicated the expected reduction in fuel gas and increase in gasoline, it failed to show other significant yield shifts. One might question whether the quench material was cracking, or whether there would be any decrease in conversion.

To answer these questions, Amoco initiated additional extensive performance testing, including further vapor-line sampling. These tests supported the previous conclusion that, in the absence of dilute-phase quench, the reactions are almost exclusively thermal.

The vast majority of the molecules in the dilute phase were gasoline and lighter; hence, the thermal reactions involved primarily converted products from the catalytic reactions in the riser.

The highly aromatic LCO and decant oil are relatively unreactive in the dilute phase; therefore, the conversion of 430+ F. boiling range material to gasoline and lighter streams is rather low.

This testing also established that other thermal reactions occur in the dilute phase of FCC reactors that do not employ quench. Diolefins-both butadienes and pentadienes-are formed in the dilute phase when thermal reactions are allowed to proceed.

When post-riser quench was introduced to stop these thermal reactions, a 50% drop in diolefins production was measured. Some representative data are:

• C4==, -40 to -50%

• C5==, -50 to -70%.

The effect of butadienes on acid consumption, in both sulfuric and hydrofluoric acid alkylation units, is well known.4 The impact of pentadienes on acid consumption in alkylation would be similar if a refiner were planning to practice amylene alkylation to reduce olefins and Rvp for reformulated gasoline production.

A reduction in diolefins yield was expected because testing showed that the dilute-phase reactions eliminated by quench were primarily thermal. A surprising result was that quench dramatically improved the oxidation stability of the finished blended gasolines.

Selected data on the oxidation stability of FCC gasoline before and after dilute-phase quenching are included in Table 3 (11333 bytes).

The chemistry of FCC gasoline is complex, and the properties that affect oxidation stability are not clear. Certainly, the 50% reduction in highly reactive pentadienes plays a role. A reduction in other thermal reaction products probably also plays a role in the improved gasoline oxidation stability.

Performance improvement

Fig. 4 (27223 bytes) shows the basis for a predictive tool to simulate the effect of quench in the FCC process.5 In this figure, the dry gas reduction resulting from the addition of quench is expressed by the rate expression: scf dry gas/bbl fresh feed/sec of post-riser residence time (scf/bbl-sec). This rate constant measures the thermal reactions that have been eliminated.

In Fig. 4 (27223 bytes), the rate constant is plotted on a logarithmic scale vs. the riser outlet temperature. Note that, if this is replotted vs. the inverse of the absolute temperature in degrees Rankine, as in Fig. 5 (27569 bytes), the slope implies an activation energy of 93,000 BTU/lb-mole.

This is in excellent agreement with literature values for the activation energy of thermal reactions (about 100,000 BTU/lb-mole), and is further confirmation of the thermal nature of the reactions in the dilute phase.

Fig. 4 (27223 bytes) can be used directly as a design tool for simulating the effect of post-riser quench. Assume, for example, that a unit has a dilute-phase residence time of 15 sec, a riser outlet temperature of 1,000 F., and utilizes a 50 F. post-riser quench. Referring to Fig. 4 (27223 bytes), a rate constant of about 5 scf/bbl-sec would be used. In this case, the quench will reduce the fuel gas by 5 scf/bbl fresh feed/sec of post-riser residence time.

Because the dilute-phase residence time is 15 sec, a fuel gas reduction of 75 scf/bbl would be expected, as explained previously. As suming a fuel gas molecular weight of 18, this leads to the conclusion that the fuel gas reduction will be 1.1 wt % of feed.

As illustrated earlier, the vast majority of the reduction in fuel gas will be converted to gasoline. This will result in a gasoline yield increase of about 1.3 vol %.

Table 4 (17648 bytes) shows some examples of the expected shift in fuel gas and gasoline yield for a unit with a 15-sec residence time operating between 975 F. and 1,025 F. and using 25-75 F. of post-riser quench.

Clearly, at higher riser-outlet temperatures, the benefits become rather large. At 1,000 F. using 50 F. of quench, the reduction in fuel gas and the increase in gasoline are worth about 15/bbl, or about $2.5 million/year for a 50,000 b/d FCCU.

There also is the additional impact of reduced diolefins production on alkylation acid consumption. If this 50,000 b/d FCCU had an associated 7,000 b/d sulfuric acid alkylation unit charging all of the C4 olefins, the reduced diolefins yield would result in an annual fresh acid savings of $250,000 (at an acid price of $50/ton).

The value of reduced pentadienes and improved gasoline oxidation stability will depend on each refiner's operation.

Table 5 (17544 bytes) shows three cases indicating that many units experience even further advantages of post-riser quench. Case A shows the same example FCCU with 15-sec dilute-phase residence time and 1,000 F. riser outlet temperature.

In Case B, 50 F. post-riser quench has been employed. It is worth noting that wet gas yield decreased 7% compared to Case A. For the refiner who has an FCCU limited by wet gas compressor capacity, this 7% reduction is significant.

The most obvious way to utilize the additional wet gas compressor capacity is to increase feed rate. For a 50,000 b/d FCCU limited by gas compressor capacity, quench could be used easily to increase feed rate by 3,500 b/d.

Another way that additional wet gas compressor capacity could be used is to increase the riser outlet temperature. Case C shows how increasing riser outlet temperature by 15 F. to 1,015 F. essentially utilizes the 7% gas compressor capacity that post-riser quench made available.

Looking more closely at Case C, it shows that, as riser temperature increases, conversion increases, as expected. At these severe cracking conditions, however, operation in overcracking mode would be anticipated; thus, gasoline yield should have decreased. Instead, this case shows that, by stopping the thermal cracking reactions in the dilute phase, gasoline yield remains almost constant.

The incremental conversion from the increased riser temperature, therefore, comprises primarily propylene and butylenes, which can be used for chemical feedstocks and alkylation. Moreover, FCC gasoline RON is increased.

Post-riser quench technology can enhance the yield benefits of an FCCU, in addition to improving gasoline oxidation stability and allowing increases in either feed rate or riser temperature. For a 50,000 b/d FCCU, these benefits can total as much as $6 million/year at current market prices.

References

1. Evans, Richard E., and Quinn, George P., Fluid Catalytic Cracking: Science and Technology, Studies in Surface Science and Catalysis, Vol. 76, Ch. 15, p. 563.

2. Silverman, Michael A., "Riser Termination," Stone & Webster Engineering Corp., 1st FCC Forum, May 11-13, 1994, The Woodlands, Tex.

3. Ross, J.L., Hibble, P.W., and Dharia, D.J., "Catalytic Cracking Technologies to Maximize Gasoline, Diesel, and/or LPG," Grace Davison FCC Technology Conference, Athens, Sept. 27-30, 1994.

4. Cosyns, Jean, Nocca, Jean Luc, Keefer, Pamela S., and Masters, Kenneth R., "Ultimate C4/C5 Olefin Processing Scheme for Maximizing Reformulated Gasoline Production," NPRA annual meeting, 1991, San Antonio.

5. U.S. Patent No. 5,089,235.

The Authors