TEXAS PLANT FIRST TO ISOMERIZE N-BUYTLENES TO ISOBUTYLENE
Thomas P. John
Texas Olefins Co.
Houston
Stephen P. Thomas
Phillips Petroleum Co.
Bartlesville, Okla.
Environmental policy and changing markets around the world necessitate more flexibility in the upgrading of light olefins.
Since mid-1991, Texas Olefins Co. has operated a commercial-scale process for the skeletal isomerization of nbutylenes to produce isobutylene for methyl tertiary butyl ether (C) feedstock. This article describes the process, operating experience, economics, and applications in the refining and petrochemical industries.
BACKGROUND
In August 1991, Texas Olefins began operating the world first commercial-scale development unit for the isomerization of n-butylenes to isobutylene. The 2,400 b/d unit quickly evolved into a reliable operation.
This operation demonstrates the technical and commercial feasibility of using olefin isomerization to upgrade the value of mixed-C4 streams from steam crackers and refineries. This is significant because a global surplus of normal butylenes is developing.
By-product C4S from a worldwide increase in ethylene demand and steam-cracker capacity contribute to the n-butylene surplus. Likewise, the quantity of refinery C4 olefins is increasing because of increased worldwide fluid catalytic cracking (FCC) capacity for upgrading heavy oils to meet gasoline demand.
Some refiners are even altering FCC operations to make more isobutylene for MTBE, thus also increasing their n-butylene yields.
As the supply of n-butylenes increases, the demand for them is declining. Environmental considerations discourage their use in gasoline directly. On the other hand, a strong demand for isobutylene has been created by the environmental benefits of oxygenates such as MTBE.
Thus, as strategies for producing low-emissions gasoline develop, isomerization of n-butylenes to isobutylene for MTBE production is a significant economic upgrade.
SKELETAL ISOMERIZATION
Skeletal isomerization technology for olefins has been in development for quite some time. Phillips Petroleum Co., Texas Olefins, and many other companies have extensive research histories in the area (OGJ, Nov. 23, 1992, p. 36; Mar. 23, 1992, p. 42).
A variety of catalysts has been developed to effect the transformation, but it has taken current market forces to bring these to commercial application.
Skeletal isomerization of olefins proceeds by a carbenium ion mechanism, so acidic catalysts are required. A variety of reactions can occur on these catalysts, however, so careful balancing of catalyst properties and process conditions is necessary to achieve maximum production of desired branched-chain olefins.
In the absence of side reactions, reactants in an olefin-isomerization process approach thermodynamic equilibrium over the catalyst. Thus, the conversion of normal olefins to iso-olefins is limited to the equilibrium value, which is a function of the reaction temperature.
Ideally the same product composition is achieved regardless of the isomer distribution in the feed. There are minor differences in the heats of reaction for the various isomers, however, so minor heat differences are observed, depending on feed. composition.
Because olefin isomerization is an equilibrium process, the only way to achieve high conversion is to remove iso-olefins and recycle normal olefins.
A significant, competing side reaction with virtually all olefin-isomerization catalysts is disproportionation. Two moles of butylene, for example, react to produce propylene and pentenes. Fortunately in most instances these disproportionation products are also of high value.
Some polymerization of olefins also tends to occur on these catalysts, but this reaction yields lower-valley heavy hydrocarbons and coke on the catalyst.
The current interest in olefin isomerization has centered around C4 and C5 olefins. Potential applications include isomerizing FCC C4 and C5 olefins for increased production of ethers (MTBE, ethyl tertiary butyl ether [ETBE], and tertiary amyl methyl ether [TAME]), and for the production of additional isobutylene from steamcracker C4 streams.
SKIP PROCESS
A simplified flow diagram illustrates the Texas Olefins/Phillips skeletal isomerization process (SKIP) for converting normal butylenes to isobutylene (Fig. 1).
Hydrocarbon feed containing linear butylenes is first vaporized and mixed with steam. The resultant mixture is heated to reaction temperature (480-550 C., or 900-1,025 F.), using a feed/effluent exchanger and a feed heater.
The mixture then enters the fixedbed reactor. In the reactor, a portion of the linear butylenes is converted, at essentially atmospheric pressure, to isobutyl and by-product light ends and C5 + compounds. Minor amounts of coke and a heavy oil are also formed. The catalyst is inexpensive and extremely hardy.
The reactor effluent is partially cooled by heat exchange with the feed and further cooled using a waterquench column. Vapor from the quench column is compressed, cooled, and sent to a separator, where the vapor and water phases are removed from the liquid-hydrocarbon phase.
Vapor from the separator is compressed and fed to the depropanizer to recover isobutylene from the light ends. The water phase from the separator is combined with the water/heavy-oil mixture from the quench column and sent to a coalescer, where the water and heavy oil are separated before exiting the process.
The hydrocarbon phase from the separator is fed to a depropanizer to remove light ends. Depropanizer bottoms are fed to a depentanizer, where the mixed butylenes are the overhead product. Depentanizer bottom product is C5+ material.
All water from the process is sent to water treatment or recycled to the reactor feed.
The process can be run as once through if there is a high-value use for the raffinate stream once the iso-olefins are removed. Alternatively, the lean stream can be recycled, and additional branched olefins can be produced, as shown in Figs. 2 and 3.
REGENERATION
The process is cyclic and requires two reactors. One catalyst bed is on-line while the other is regenerated. Cycle times are typically a few hours. The catalyst is regenerated by controlled oxidation. Regeneration occurs fast enough to be complete well before the other bed is off-line.
As shown in Fig. 4, during regeneration a stream of hot steam and air is passed through the catalyst bed. The steam/air mixture is heated to reaction temperature using a feed/effluent exchanger and a regeneration heater.
As coke is removed from the catalyst, regeneration gases from the reactor are mixed with additional air and sent to a catalytic oxidizer where combustible materials are converted to carbon dioxide and water.
Effluent from the catalytic oxidizer is heat exchanged with incoming regeneration gas and passed through a condenser to remove a major portion of the water. Uncondensed material is burned with fuel in the regeneration heater.
Condensed water is sent to water treatment or reused as boiler feedwater.
ENVIRONMENT
The spent catalyst is a nonhazardous waste.
It has been demonstrated that the waste water can be successfully treated in a conventional, biological waste water treatment plant. Further, Texas Olefins believes the waste water can be recycled within the process.
Carbon monoxide emissions from catalyst regeneration are controlled with catalytic or combustion oxidation. Fired-heater emissions can be minimized by using clean fuel gas and low-NOx burners.
The volatile organic compound (VOC) emissions are primarily fugitive emissions from valves and flanges. Because the unit is fairly simple, these emissions are not great.
FEEDSTOCK FACTORS
High-purity linear olefins are the preferred feedstock, but saturated hydrocarbons have no detrimental effect on the reaction other than to act as a diluent. As the SKIP process essentially establishes an equilibrium among the various olefinic structural isomers, any iso-olefins in the feed limit conversion and should be minimized.
Extensive testing of potential feedstock contaminants has not identified any permanent catalyst poisons. Compounds such as dienes, acetylenes, oxygenates, and extractive-distillation solvents are coke-formers at feed concentrations in the range of thousands of parts per million.
These coke-formers will cause the catalyst to lose activity temporarily. Catalyst activity can be restored, however, by regeneration.
No elaborate feedstock preparation is necessary to run the process, but if an MTBE unit is to be placed upstream of a SKIP unit, a two-stage MTBE process is recommended.
PROCESS PERFORMANCE
For n-butylenes, once-through conversion is typically in the range of 35% and selectivity to isobutylene, 80-85%.
The small amount of coproducts formed are predominantly usable products, namely isopentenes and propylene.
Selectivity to propylene and lighter products is typically 5-10% and selectivity to C5 and heavier products is 8-12%.
OPERATIONAL EXPERIENCE
In early 1991, Texas Olefins began operating a SKIP pilot plant. In August 1991, piping changes to an existing unit at Texas Olefins' Houston plant were completed to allow commercial development of the process. The nominal capacity of the unit is 2,400 b/d of isobutylene product.
Despite use of old equipment not optimized for the process, the unit has performed well and quickly demonstrated commercial reliability.
The unit was down during the summer of 1992 for planned modifications, including automation of the reactor switch valves and other improvements. Otherwise, it has operated consistently, with minimum mechanical downtime and no significant process problems.
The catalyst has been proven extremely durable. No special feed or catalyst treatment is used. Although there have been instances of bed coking because of high concentrations of coke-formers in the feed, catalyst activity was successfully restored to previous levels each time with an extended regeneration.
Waste water from the process is successfully treated in Texas Olefins' biological waste water treatment plant.
Reactor switch-valve operation has been very reliable. Readily available valves and operators are used. This is a key factor in being able to use a simple, cost-effective, reactor-system design.
The commercial unit has operated successfully at a variety of feed rates, cycle times, and operating parameters. It has processed C4 streams from all over the world in a highly integrated petrochemical complex.
Although the commercial unit matches the pilot-plant operation well, Texas Olefins continues to operate the pilot plant for purposes of process improvement. Phillips is conducting additional studies in its laboratories.
ECONOMICS
A process-economics study for a new plant found capital investment for a SKIP unit sized to feed a 3,000 b/d (118,000 metric tons/year [mtyl) MTBE unit to be about $21 million, inside battery limits, for U.S. Gulf Coast installation.
This value includes the installed cost for all equipment required for the process, plus normal project overheads and contingencies.
This 3,000 b/d plant would consume about 3,000 b/d (95,200 mty) of normal butylenes and produce more than 2,400 b/d (76,200 mty) of isobutylene. Feed is assumed to be 90% linear butylenes and 10% saturates.
This capital investment estimate and the operating-cost figures to follow are based on once-through operation. Because the process is equilibrium-limited, recycle will increase production and, as will be demonstrated, reduce the unit cost of producing iso-butylene.
UTILITIES' COSTS
Table 1 shows estimated utilities consumption and unit costs for a SKIP unit. Typical U.S. Gulf Coast values are assumed.
Estimates of utilities costs per unit of product are given in Table 2. Steam figures include the steam added with the feed during the process plus that added during regeneration.
OPERATION COSTS
Estimated operating costs for a 3,000 b/d plant are shown in Table 3. The cost basis for the utilities estimates is shown in Table 1.
Feedstock and by-product values are based on projected market conditions and are higher than current prices. These are detailed in Table 4. The basis for other economic factors included in the evaluation are also shown in Table 4.
APPLICATIONS
An important use for skeletal isomerization will be to increase the amount of MTBE that can be produced from a C4-olefin stream.
Table 5 shows the operating costs from Table 3, expressed as cost per unit of MTBE produced. Methanol feed and incremental utilities and catalyst costs for the MTBE unit are included.
CAT-CRACKER C4s
A study was made using SKIP to increase the production of MTBE from an FCC C4 stream. Table 6 shows the composition of the feed assumed for purposes of the study. The recycle scheme in Fig. 2 was also assumed.
As shown in Fig. 2, a purge is required to remove the saturated hydrocarbons, which pass through the reactor unaffected and build up with recycle.
An extractive distillation unit can be used to remove most of the saturates from the feed to the SKIP unit. This option would decrease the size of both the SKIP and MTBE units.
For comparison, the cost of producing MTBE from the contained n-butylenes was calculated for two different levels of purge, and for a single case using extractive distillation.
In the high-purge case, the purge rate was set so that the feed rate to the SKIP unit was approximately equal to that of a once-through scheme (no recycle). Feed to the SKIP unit in the low-purge case was 2.4 times that rate.
The extractive-distillation case was designed assuming performance of that unit so that 89% of feed n-butylenes were recovered as SKIP feed.
The analysis of each case included both capital and operating costs for the appropriate SKIP, MTBE, and extractive-distillation units. U.S. Gulf Coast, inside battery limits costs for a complete SKIP unit, a two-stage MTBE plant, and a furfural-water extractive distillation unit were estimated.
Incremental capital costs were calculated for the MTBE plant, as it was assumed that, without SKIP, the refiner would build a smaller MTBE unit to extract the contained isobutylene. No attempt was made to optimize the extractive distillation unit, so it is likely that the economics for that case could be significantly improved.
The basis of the economic calculations in this analysis is the same as that described earlier for the calculation of SKIP operating costs.
Results of the analysis are shown in Fig. 5. Note that as recycle increases (from high to low purge), the costs/gallon MTBE produced, for both the MTBE and SKIP units, decrease. Incremental MTBE unit costs do not decrease much, but SKIP costs decrease significantly as the amount of recycle increases (low purge rate).
SKIP costs are dramatically lower for the case that includes the extractive distillation unit. These savings more than offset the costs for the extractive distillation unit, resulting in the lowest total cost/gal MTBE. Extractive distillation might not be cost-effective for every application, but it appears to be cost-effective for dilute feeds such ass FCC C4s.
STEAM CRACKER C4s
Another study evaluated using the SKIP to increase MTBE production from a steam-cracker C4 stream. Again, the effect on the incremental MTBE cost of varying recycle levels was determined.
These cases assumed a SKIP-feed composition as shown in Table 6, in the column labeled "After MTBE unit." The cases also represent processing a raffinate from an MTBE unit fed with a steam-cracker C4 stream selectively hydrogenated to remove butadienes.
In this study the recycle loop involved a second MTBE unit, as shown in Fig. 3. Recycle was varied so that the feed rate to the SKIP unit was 1.5, 2, and 3 times the flow to the unit in a once-through, no-recycle case (1 x).
The cost of the incremental MTBE produced in each of these cases was determined as in the previous study. Results are presented in Fig. 6.
Note that as recycle increases, the cost/gal MTBE produced, for both the MTBE and SKIP units, decreases. (This cost varies little between the 2 x and 3 x cases.) MTBE yields increase significantly throughout the entire range. It is clear that recycle can be used to optimize MTBE production and economics.
For comparison, the case of a steamcracker that used a single, large MTBE plant (as in Fig. 2) was studied. For a new installation, this would reduce the capital investment in MTBE capacity.
This case (2 x ') was sized so that the flow through the SKIP unit was the same as in the 2 x case. Feed for the case is shown in Table 6 in the column labeled "Before MTBE unit."
Fig. 7 presents a comparison of the 2x and 2x' cases. The MTBE unit costs show the expected economy of scale for the 2 x ' case. SKIP-unit costs are higher for the 2 x ' case, however, so that total MTBE costs differ little.
MTBE production is about 10% less in Case 2 x ' because the concentration of n-butylenes in the purge stream is higher with only one MTBE unit.
The SKIP process is available for license from either Phillips Petroleum Co. or Texas Olefins Co. Work continues to bring the C5 skeletal isomerization process to the same stage of development.
ACKNOWLEDGMENT
The authors acknowledge the valuable assistance of their colleagues in generating the data used in this article and its preparation. Thanks to Bob Schoppe, Larry Goodwin, Terry Funkhouser, and Dan Tschopp at Texas Olefins, and Harold Hunt and Iqbal Ahmed at Phillips.
Copyright 1993 Oil & Gas Journal. All Rights Reserved.