ONLINE MONITORING � IMPROVE FCC GASOLINE � THE WRONG BEARING

AAuugguusstt 22000011Reprinted from:

Process Technology—Refining

A. Katona, Mol-RT Refinery, Szazhalombatta,Hungary; T. Darde, Pall Europe, Saint Germain-en-Lage, Cedex, France; and T. H. Wines, PallCorp., East Hill, New York

C austic carryover from a treater can causedownstream contamination of fluid catalyticcracked (FCC) gasoline; Result: off-

specification product. Caustic forms a very stableemulsion with gasoline that results in high levels ofsodium, water and phenols in the final product. Newdevelopments in coalescing technology have made itpossible to separate very stable emulsions anddispersions.

In the following case history, the Mol-RT’s Sza-zhalombatta Refinery experienced emulsion problemsfrom caustic carryover. Eager to resolve this condi-tion, refinery engineers investigated several tech-nologies to break the emulsion and mitigate the prob-lem. One option considered was a new high-efficiencyfluoropolymeric liquid / liquid coalescer to optimizethe caustic treating system. This case history detailsthe benefits when using present and new separationequipment to control gasoline emulsion problems.

Caustic problems. TheMol-RT Refinery in Sza-zhalombatta, Hungary,operates a caustic-treat-ing unit to remove mer-captans from FCC gaso-line. This unit processes130 m3/h (19,624 bpd) ofFCC gasoline. In thisprocess, FCC gasoline isdelivered to two f ixed-bed reactors, each con-taining a catalystimpregnated onto anactivated carbon sub-strate. Cobalt and vanadium pthalocyanines areused as catalysts due to their high activity and sta-bility in the oxidation reaction for mercaptans andlow solubility in petroleum fuels.1 Caustic (3°Baumé) and oxidizing air are injected into the FCCgasoline upstream of the reactor. In the reactor,

mercaptans are extracted into the caustic and thenconverted to disulfide oils by oxidation and cat-alytic action. Also, phenols are extracted into thecaustic phase. The reactor effluent flows to a three-phase separator where air, disulfides and regen-erated caustic are separated. Regenerated causticis recycled back to the reactor. Fig. 1 shows a sim-plified schematic of the caustic-treating unit. Theprocess parameters for the Mol-RT Refinery caus-tic treater are presented in Table 1.

The reactor effluent contained a quantity of car-ried-over caustic that resulted in hazy gasoline prod-uct, high costs for caustic makeup and corrosion ofdownstream piping. However, the hazy gasoline wasproblematic. It had to be blended off or reprocessed;otherwise, it would cause the final product to beoff-specification for sodium. Likewise, the carry-over caustic could also react with methyl tertiarybutyl ether (MTBE) that was blended into the gaso-line downstream of the caustic-treating system.

SEPARATION OPTIONSTo reduce the caustic excursions downstream

of the treating unit, Mol-RT considered several options,which include settlingtanks, mesh-pad beds,electrostatic separators,sand beds and waterwashing:

Settling tanks weredeemed not viable due tothe high stability of thecaustic-fuel emulsion. Astable emulsion containsvery small droplets thatdo not settle efficiently,and excessive settlingtime and /or tank volumewould be required.

Mesh pads operateby the principle of inertial impaction whereby thedroplet momentum is great enough so the dropletleaves the streamlines of the fluid flow and impactsmetal f ibers or plates in the mesh pads. Thesedroplets then coalesce into large drops that separateby gravity. Mesh pads do not work well when the

Reprinted from HYDROCARBON PROCESSING magazine, August 2001 issue, pgs. 103-108. Used with permission.

Improve haze removal for FCC gasoline

This refinery used advanced coalescer techniques to separate very stableemulsions and dispersions by mechanical means

Table 1. Caustic treater process parameters.

Application Removal of caustic from gasoline

Process flowrate 130 m3/h (19,624 bpd)Temperature 40°C (104°F)Pressure 6.5 barg (94.3 psig)Gasoline viscosity 0.55 centistokes @ 38°CGasoline density 0.68 g/mlCaustic density 1.02 g/mlInterfacial tension 12.4 dyne/cmCaustic water pH 12.4Total suspended solids < 1 ppmwInjected caustic concentration 3° Baumé (2.3 % wt)

HYDROCARBON PROCESSING / AUGUST 2001

interfacial tension is low (< 20 dyne/cm) and thedroplets are very small leading to poor droplet captureefficiency. Also at flowrates reduced from the originaldesign, the inertial separation force is greatly low-ered leading to poor separation.

Electrostatic precipitators separate caustic waterfrom hydrocarbons by mean of an electric charge cre-ated by a high voltage source. While electrostatic sep-arators can be effective, they require a high operatingcost for electricity. Furthermore, they require significantinitial capital expenditures and can have high main-tenance costs when electrodes short circuit due to caus-tic slugs (increased conductivity) or scale buildup onthe electrodes.

Sand beds are able to separate caustic from hydro-

carbon fuels by acting as coalescerswhereby the small caustic dropletsadsorb onto the sand and form largerdrops that are also separated bygravity. In practice, however, sandbeds are prone to severe problems offluid channeling through the bedcaused by bed compaction and crack-ing. This often leads to poor sand bedperformance and high manpowercosts to load and unload the bed.

Water washing usually requireslarge extraction towers that have ahigh initial capital expense and addi-tional separation equipment toremove any carryover water from theextraction process. This process con-sumes water that will require dis-posal or further treatment for re-useafter the extraction process.

Liquid/liquid coalescers havethe same low operating costs regard-less of the amount of water or causticcharged to the unit. Based on thesuperior efficiency and low cost ofseparation, the Mol-RT Refinerydecided to run field trials with high-efficiency polymeric liquid/liquid(L /L) coalescers.

L/L coalescer system. The tradi-tional L /L coalescers have usedglass-fiber media, which work wellfor emulsions that have interfacialtensions greater than 20 dyne/cmand for systems that have neutralwater as the dispersed phase. Newcoalescer media, constructed withnovel formulated polymers and fluo-ropolymers, are effective for emul-sions having interfacial tensions aslow as 0.5 dyne/cm and for harshchemical environments such as thecaustic/ fuel system.2,3 High-effi-ciency polymeric coalescer have pro-duced clean petroleum fuels withsodium levels below 0.5 mg/l and freewater concentrations of <15 ppmv.

A high-efficiency L /L coalescerin the horizontal configuration is shown in Fig. 2.The system consists of a pre-filter section followedby a horizontal coalescer cartridge stage with a set-tling zone that relies on the difference in densityfor separation of the coalesced droplets. The fluidenters at the side of the housing and flows from theinside of the coalescer cartridges radially outwardcausing the enlargement or coalescing of the inletdispersion into large droplets in the outlet stream.These coalesced droplets then flow axially in the hor-izontal direction through a settling zone. The dis-persed caustic-phase coalesced droplets settle down-ward by gravity and are collected in a sump locatedat the bottom of the housing. Purified gasoline leavesat the top of the housing.

Fig. 1. Caustic treating Unit at the Mol-RT Refinery.

Table 2. Caustic- fuel interfacial tension data gathered from field tests.

Plant location Hydrocarbon Interfacial tension, dyne/cm

U.S. Pacific Heavy catalytic cracked (HCC) gasoline 0.7U.S. Pacific Light catalytic cracked (LCC) gasoline 0.6U.S. Midwest Heavy catalytic cracked (HCC) gasoline 0.5U.S. Gulf Coast Fluid catalytic cracked (FCC) gasoline 5.0U.S. Gulf Coast Light gas oil (LGO) 3.9U.S. Gulf Coast Heavy catalytic naphtha (HCN) 4.3Canada Refined oil (RO) 3.6–4.7Canada Heavy catalytic naphtha (HCN) 0.8Canada Light catalytic naphtha (LCN) 10.2Canada Kerosine 2.4Singapore Kerosine 8.4England Fluid catalytic cracked (FCC) gasoline 12.0Hungary Fluid catalytic cracked (FCC) gasoline 12.4

HYDROCARBON PROCESSING / AUGUST 2001

The L /L coalescing system oper-ates in three stages: separation ofsolids, coalescence and separation ofcoalesced drops:

Separation of solids. Solidscan increase the stability of anemulsion; consequently, removingsolids can make coalescing easier.Generally, this step can be achievedby a separate cartridge filter sys-tem or by a re-generable backwashf ilter system for high levels ofsolids. In addition, the filtrationstage protects the coalescer andincreases service life.

Coalescence. The next step isprimary coalescence. In this stage,the pore dimensions begin with avery fine structure and then becomemore open to allow for void space forthe coalescing droplets. In the pri-mary coalescence zone, the inletdroplet dispersion containing finedroplets in the size range of 0.2 to 50microns (�m) is transformed into asuspension of enlarged droplets inthe size range of 500 to 5,000 �m.

The coalescence mechanisminvolves the adsorption of dropletsto the coalescer fibers, followed bytranslation along the fibers and col-lisions at the junctures betweenf ibers. In these collisions, thedroplets merge together or coalesce.The viscous drag of the bulk fluid stream then causesthe enlarged drops to disengage from the fibers. Thisprocess is repeated a number of times through thecoalescer depth until the large coalesced drops exitthe coalescer media. The necessary condition thatdroplet-fiber adsorption occur for coalescing has beensupported by many sources.4,5

Separation of coalesced droplets. Once thedroplets have been coalesced, they are now assumedto be as large as possible for given flow conditions. Sep-aration is achieved by using a settling zone, which relieson the difference in densities between the coalesceddroplets and bulk fluid. Caustic is separated in a col-lection sump that can be manually drained on a periodicbasis or equipped with an automatic level control anddrain system. Estimation of the coalesced drop sizeand required settling zone is best determined throughpilot-scale tests at field conditions.

Surfactants. They are naturally present in crude oil,and thus, are found in refined petroleum products.During the oxidation process and the caustic recircu-lation in the sulfur-removal process, surfactants can beconcentrated to high levels. Surfactants present incaustic treaters include: sulfides, mercaptides, naph-thenic acids, cresylic acids and phenol homologs.6

Petroleum naphtha sulfonates have also been identi-fied as natural occurring petroleum surfactants thatare especially detrimental to glass-fiber conventionalcoalescers.7 The surfactants can adsorb at the solid /liq-

uid interface (coalescer fibers) or at the liquid / liquidinterface (water/oil).

When surfactants concentrate on the coalescerfibers, this is known as “disarming.” The coalescerfibers are shielded from the passing aqueous droplets;this results in poor separation efficiency. Generally,the disarming phenomenon does not occur unless theinterfacial tension between water and fuel is less than20 dyne/cm. When specially formulated polymeric coa-lescer medium was used in place of glass fiber, dis-arming was not observed.2 ,3 The coalescing perfor-mance of a polymeric medium can be greatly enhancedby modifying surface properties that can not be donewith glass-fiber medium.

Surfactants can also concentrate at the water/fuelinterface. This condition can lead to very smalldroplets and stable emulsions. To separate these typesof emulsions, special consideration must be directed tothe pore size and distribution in the coalescer mediato intercept and coalesce these fine droplets. Fieldtests conducted at caustic-treating units have uncov-ered similar results showing very low interfacial ten-sion of caustic- hydrocarbon emulsions as shown inTable 2.

Pilot-scale L/L coalescer test. To evaluate the siz-ing and performance of the new fluoropolymeric coa-lescer to separate the carried-over caustic from theFCC gasoline at the sulfur-removal system outlet, apilot-scale test coalescer unit was used at the Mol-RT

Fig. 2. High-efficiency L/L coalescer system.

Table 3. Pilot-scale L/L coalescer results.

Flowrate, Test Volume of Visual Concentration of Caustic water lpm duration, caustic water appearance, caustic solution recovered, liters

min coalesced, liters inlet/outlet coalesced, ppmv caustic water/ m3 gasoline

5 50 0.74 Hazy/Clear 3,000 3.0011 115 4.00 Hazy/Clear 3,162 3.1620 120 7.15 Hazy/Clear 2,979 2.9823 1,405 64.63 Hazy/Clear 2,000 2.00

HYDROCARBON PROCESSING / AUGUST 2001

Refinery. As shown in Fig. 3, it contained one 6-in.(152.4-mm) length high-efficiency liquid / liquid testcoalescer and one 10-in. (254-mm) length pre-filter.

A slipstream was sent through the test unit, and thequality of the influent and the effluent was analyzed.If the effluent quality was acceptable, then the flowthrough the unit was increased and the fluid qualityagain analyzed. The flow through the test unit wasincreased until the effluent quality was unacceptable.In this case, the test was based on visual observations;as long as the effluent was “bright and clear,” then theflowrate was increased. The results of the testing aresummarized in Table 3.

In analyzing the data, the maximum flow that wasrun through the test unit was 23 lpm (6.1 gpm). AKarl Fischer total water analysis was conducted byMol-RT Refinery; results showed that the gasolineat the outlet of the test coalescer had less than 15ppmv of free water. The water concentrations mea-sured were close to the water saturation value (251ppmw @40°C) at the test flowrate of 23 lpm. It wasalso found that when the gasoline from the coalesceroutlet was cooled to 5°C, it remained clear. Theamount of caustic water collected by the test coa-lescer was 2 liters of caustic solution per m3 of gaso-line (2,000 ppmv).

Data showed that the caustic/FCC-gasoline emul-sion was a highly stable as indicated by the interfacialtension measurement of 12.4 dyne/cm. Any emulsionwith an interfacial tension of less than 20 dyne /cmis considered very stable and difficult to separateusing, settling tanks, mesh pads, sand beds or con-ventional glass-fiber coalescers. Based on the resultsfrom the field tests at Mol-RT Refinery, the installa-tion of high-efficiency fluoropolymeric liquid / liquid

(L / L) coalescers was recommended. The coalescerwould be intended to:

• Produce consistent “bright and clear” gasoline• Recover caustic solution• Reduce downstream corrosion• Eliminate potential reaction of caustic with MTBE

blending.

Full-scale operation. Mol-RT installed a L /L coa-lescer system that contained 17 high-efficiency coa-lescer cartridges that were 40 in. in length and four in.in diameter. A pre-filter was also installed to removesolid particulates upstream of the coalescer. The siz-ing of the full-scale system was based on the pilot-scaleL/L coalescer tests and was at an equivalent flux to14 lpm through the test coalescer. This was well belowthe maximum test flowrate of 23 lpm and facilitatedpossible variations in the caustic water properties thatmay affect coalescing. The fluoropolymeric L /L coa-lescer cartridges were specially designed to removecaustic carryover from hydrocarbon streams.

The coalescer unit was started up in November 1995.After several months of operation, the following resultswere observed:

• Approximately 2,000 ppmv of caustic solutionrecovered by test coalescer (2 liter caustic/m3 gasoline)

• No haze in effluent. Gasoline was “bright andclear.” Samples were cooled to 5°C with no hazeobserved.

The service life of the pre-filters was 6 months beforerequiring change-out and the L/L coalescers operatedfor 18 months. In this service, the service life for thepre-filter and coalescers exceeded the initial designrequirements, and therefore, represented a low overalloperation expense.

Fig. 3. Schematic of the pilot-scale L/L coalescer test unit.

2200 Northern BoulevardEast Hills, New York 11548-1289

888.873.7255 toll free516.484.5400 phone516.484.0364 fax

Pall Corporation has offices and plants throughout the world in locations including:Argentina, Australia, Austria, Belgium, Brazil, Canada, China, France, Germany, Hong Kong,India, Indonesia, Ireland, Italy, Japan, Korea, Malaysia, Mexico, the Netherlands, NewZealand, Norway, Poland, Puerto Rico, Russia, Singapore, South Africa, Spain, Sweden,Switzerland, Taiwan, Thailand, United Kingdom, United States, and Venezuela. Distributorsare located in all major industrial areas of the world.

Visit us on the Web at www.pall.com

Select-A-FAX* 800.664.7255for the latest information delivered to any fax machine, anywhere inthe world. This automatic system operates 24 hours a day, sevendays a week. Call 1.516.942.0523 from outside the USA.

HP-MOLRT

© Copyright 2002, Pall Corporation. Pall, are trademarks of Pall Corporation. ® Indicates a Pall trademarkregistered in the USA. is a service mark of Pall Corporation. *Select-A-FAX is aregistered trademark of CyberData, Inc.

Phenol removal. Another advantage detected inthe test was the removal of phenol. During tests, phe-nol concentration in the FCC gasoline was approxi-mately 330 ppm in the coalescer influent. Because thephenol is concentrated in the aqueous phase, the coa-lescer was able to reduce the gasoline phenol content.The effluent phenol concentration was 270 ppm, rep-resenting an 18% reduction in phenol content in thegasoline. The phenol content in the coalesced waterwas 14,500 ppm.

From an economic standpoint, the benefits receivedby the caustic recovery justify the installation of thecoalescer, with payback in less than one year. Otherlarge benefits are anticipated due to the haze reduc-tion in the gasoline and lowered downstream corro-sion. Over the four years of operation, the averageannual savings for caustic has been estimated at 126,000 lb/yr on a dry basis. New developments in coa-lescing technology have made it possible to separatevery stable emulsions and dispersions by mechani-cal means. �

LITERATURE CITED1 Leitao, A. and A. Rodrigues, “Studies on the Merox process: Kinetics of n-butyl mercap-

tan oxidation,” Chemical Engineering Science, 1989, p. 1245.2 Brown, R. L., and T. H. Wines, “Improve suspended water removal from fuels,” Hydro-

carbon Processing, December 1993, p. 95. 3 Wines, T. H., and R. L. Brown, “Difficult liquid-liquid separations,” Chemical Engineer-

ing, December 1997, p. 104.4 Basu, S., “A study on the effect of wetting on the mechanism of coalescence,” Journal of

Colloid and Interface Science, 1993, Vol. 159, p.68.5 Jeater, P., E. Rushton, E., and G. A. Davies, “Coalescence in fibre beds,” Filtration & Sep-

aration, March/April 1980, p. 129.6 Suarez, F. J., “Pluses and minuses of caustic treating,” Hydrocarbon Processing, October

1996, p. 117.7 Hughes, V. B., “Aviation fuel handling: New mechanism insight into the effect of surfac-

tants on water coalescer performance,” Second International Filtration Conference, SanAntonio, April 1997.

LK/2M/01-02 Article copyright © 2002 by Gulf Publishing Company. All rights reserved. Printed in USA.

Antal Katona is the head of Motor Fuels Pro-duction Department Head, at the MOL DanubeRefinery. He has been working on productionfacilities at Danube Refinery of MOL-RT. for 24years, and has experience with crude, aro-matics, visbreaker, FCC, HFA, MTBE, and HDSplants operation. Mr. Katona has participatedin major development projects of FCC, HFA,MTBE and fuel blender units, as well as, theiradvanced process control technology. He

holds engineering degrees from Petrik Lajos Chemical Engi-neering College and Baku University of Petrochemical Technol-ogy. Mr. Katona is an invited speaker and panel member ofnumerous international-refining forums. Currently working, he isacquiring an MBA degree in economics.

Thomas H. Wines is a senior marketing man-ager for the Fuels and Chemicals Group at PallCorp., East Hills, New York. His work historyat Pall Corp. includes positions as senior staffengineer, staff engineer, and senior test engi-neer in the Scientific and Laboratory ServicesDepartment. His experience includes over 12years of global filtration troubleshooting inthe refinery, gas-processing, and chemicalindustries. Mr. Wines is a specialist in thefields of liquid-gas and liquid-liquid coalescing. He holds a BSdegree in chemistry from Fordham University and an MS degreein chemical engineering from Columbia University. He is completingstudies for a PhD in chemical engineering at Columbia University,and is a member of the American Institute of Chemical Engineersand the American Chemical Society.

Thierry Darde is the Global Marketing Man-ager for the hydrocarbon business at PallCorp. Dr. Darde holds Ph.D. and DsC degreesin chemical engineering from ENSIC NancyUniversity, France. Dr. Darde began his careeras a research engineer at the Shell ResearchCentre (KSLA) and was a process engineer atShell’s Reichstett Refinery. Later, he was theplant manager of the Reichstett Refinery andwas the head of Economics at Shell’s Berre

Refinery. In 1995, he joined Pall Corp. and is the European mar-keting manager for the hydrocarbon business.