+ All Categories
Home > Documents > hydroprocessing-depropanizer

hydroprocessing-depropanizer

Date post: 14-Oct-2014
Category:
Upload: sathish-kumar
View: 479 times
Download: 0 times
Share this document with a friend
12

Click here to load reader

Transcript
Page 1: hydroprocessing-depropanizer

1 March/April 1999

HYDROPROCESSINGOPTIONS

Reliable minimum-capital-investment revamps requirepushing the plant to thelimit of the existing equip-

ment. While various small equipmentdesign errors are tolerable at moderateequipment capacity, they are not whenthe unit is pushed.1,2 Process designfor reliable low-cost revamps mustfocus on the following areas:• Test-run: quantifying existing equip-

ment performance• Alternative flow schemes: maximiz-

ing existing equipment utilization• Reliable revamps: detail equipment

design.All phases of a revamp process

design must be addressed to meetboth the low cost and reliability objec-tives.3,4 Shortcutting this processresults in either higher investmentcosts due to unnecessary equipmentmodifications or unreliable revampdesigns being installed because exist-ing equipment problems were notidentified. This article will address theimportance of test-runs in the designof refinery revamps. FCC units areused to highlight the importance ofthe test-run in quantifying equipmentperformance. (Photo 1).

A well-planned comprehensivetest-run is the most important step in

any revamp, yet it israrely performed. Oftenit is considered anunnecessary cost. To-day, there is an overrid-ing belief that ease-of-use computer modelscan be used by inexpe-rienced personnel; andthat all data can be rec-onciled by the appro-priate software algo-rithms. Unfortunately,refinery equipmentmay not operate perthe model assumptionsor the presumed origi-nal equipment design.Installed equipmentperformance must bemeasured. For in-stance, knowing thatthe FCC wet gas com-pressor is limiting thethroughput or unitconversion is not thesame as knowing whatcauses the limitation.Accurate operatingdata is required toquantify existing equipment perfor-mance and pinpoint specific causes ofequipment operating problems.

While management informationsystems contain volumes of data, oftenthis data is not complete or worse, it

Pushing Plant Limits. Test Runs, Plant Expectationsand Performance ConfidenceBy SCOTT W. GOLDEN, Process Consulting Services Inc., Houston, TX

S P E C I A L F O C U S O N

Photo 1. Revamps Push Equipment Limits

Page 2: hydroprocessing-depropanizer

may not be accurate. Pushing refineryunit operations to the major equip-ment limits requires a differentapproach to a revamp design. Reliablelow-cost revamps must address themyriad of potential detail equipmentdesign considerations that cause unitsnot to perform. Revamp processdesign must quantify existing equip-ment performance and eliminate bot-tlenecks that prevent maximumequipment utilization. Ultimately,revamps must be both cost effectiveand reliable.

Conventional Project Process DesignA conventional project process designis executed by a series of activities thatare often performed by differentgroups of individuals.5 First a heat-and-material balance based on prede-termined revamp objectives is com-pleted. Existing equipment evaluationthen follows. Generally, all this work iscompleted in the design office. If thecalculated existing equipment capacityis insufficient, the equipment isreplaced or paralleled. In thisapproach, the office-based calculationof the plant performance sets thefuture performance expectation. Engi-neering calculations are performedbased on presumed equipment perfor-mance, vendor supplied estimates ofperformance, industry averages or thespecific E&C’s design standards.Revamped process equipment perfor-mance is often a small differencebetween two large numbers (futurerequirements– current capabilities).The conventional project’s processdesign determines the future require-ments based on the criteria previouslymentioned. In meeting the revampobjective, equipment design assump-tions often force unnecessary equip-ment modifications or underestimateexisting equipment capacity. To pre-vent revamp failures, engineering andconstruction firms add in safety mar-gins to the changes required. The safe-

ty margin required to adequately coverthe errors in the conventional project’sapproach is often greater than the realchanges required.

Reliable Revamps-Minimum InvestmentBy contrast, revamp process designfocuses on measuring the currentequipment performance with a com-prehensive well-planned and executedtest run (Figure 1). Gathering specifictest-run data and the subsequentanalysis is the basis for an accuratebench-mark of current equipmentperformance. The term test-run hascome to mean different things. How-

ever, in this era of computers andadvanced process control equipment,a test-run generally means gatheringdata from the computer and sending itto the engineering contractor’s processdesigners. Our definition of test-run isan accurate set of composition, pres-sure and temperature data on allprocess streams that impact the unitheat and material balance includingdata needed to quantify the majorprocess equipment performance.Generally, revamp requirements are anincremental change to current mea-sured (not calculated) performanceand not an average or presumed per-formance. The existing equipment

2 WORLD REFINING March/April 1999

F O C U S O N : HYDROPROCESSING OPTIONS

PRELIMINARY REVAMPPROCESS OBJECTIVE

FIELD REVIEW

COMPREHENSIVETEST RUN

TEST RUNDATA EVALUATION

EXISTING EQUIPMENTREVIEW

DEVELOPMENT OF ALTERNATIVEREVAMP FLOW SCHEME

FINAL OBJECTIVES

DESIGN PACKAGE

COST ESTIMATE& SCHEDULE COST ESTIMATORS

REVAMP ENGINEERS

REFINERY STAFF +REVAMP ENGINEERS

REVAMP ENGINEER

REVAMP ENGINEER

REVAMP ENGINEERS

REFINERY STAFF +REVAMP ENGINEERS

REFINERY STAFF +REVAMP ENGINEERS

REFINERY DEFINEDOBJECTIVES

Figure 1. Revamp Approach

Page 3: hydroprocessing-depropanizer

March/April 1999 WORLD REFINING 3

F O C U S O N : HYDROPROCESSING OPTIONS

performance must be quantified. Thisminimizes revamp errors and cost, aswell as identifies equipment perfor-mance problems. Higher confidence inyour current equipment performanceallows the revamp engineer to pushthe equipment limits further.6 Thisminimizes investment for any givenrevamp objective.

Test Runs-Why Are They Needed?An accurate bench-mark requires aplant test-run. Revamping requires aknowledge of the absolute magnitudeof the equipment capabilities stillunused. The standard informationavailable to plant owners from processcontrol systems and reports is trenddata, which often is not adequate toidentify deficiencies. Hidden offsets intrend data do not affect period-to-peri-od operating performance compar-isons. However, hidden offsets canhave a dramatic impact on revampanalysis. The surest way to have a suc-cessful, low investment revamp is tocarefully define your current operating

basis. This approach is shown for sev-eral FCC product recovery revamps.Case study examples highlight theimportance of test-run data whenrevamping for reliability and mini-mum investment cost.

Performing a comprehensivetest-run is a necessity. The time spentdetermining real plant limitationsdirectly relates to the revamp capitalcost. Test-run data simplify engineer-ing calculations and help avoid calcu-lation errors. Ease-of-use engineeringsoftware calculations do not necessari-ly represent reality. For instance, sys-tem hydraulic calculations involveequivalent lengths of pipe, friction fac-tors, exchanger pressure drop algo-rithms, and oil physical and transportproperties. In theory, a heat exchangerperformance can be rigorously calcu-lated by a computer program. Howev-er, the reality for many petroleumrefinery exchanger services is that theactual fouling factor value controls theservice heat transfer coefficient. It isthe largest resistance to heat transfer.User input fouling factors are, at best,

estimates based on a similar service inanother unit. Often the estimates usedfor design are so low, that they makethe installed exchanger surface areaunworkable. Additionally, calculatedexchanger tube and shell-side pressuredrop assumes the exchanger is clean.Fouling can dramatically increaseexchanger pressure drop.

Many FCC catalytic sections arebeing revamped to increase conversionand/or product selectivity.7 Theserevamps make more C3-C12 boilingrange products from the reactor. Theadditional lower boiling products mustbe fractionated in the gas plantcolumns. Figure 2 shows the depro-panizer in the FCC gas plant. Theoverhead product from this column isprocessed in an ethylene plant C3 split-ter to produce chemical grade propy-lene and propane for sales. The FCCdepropanizer overhead product C4

content is an important control para-meter. The depropanizer column isreboiled with heat from the FCC mainfractionator slurry pumparound.Increased reactor C3-C12 productionalways increases gas plant reboiler heatrequirements. The proposed depro-panizer heat and material balance, forthe revamp, shows higher reboiler dutyto meet the C3 recovery and overheadproduct C4 composition target.

The conventional project’sapproach rigorously models the TEMAAKT kettle with one of the several com-mercially available shell-and-tubeexchanger models. The exchangerequipment specialist uses the specificE&C’s design criteria for this servicewhich is a total shell and tube-side foul-ing factor of 0.005 Ft2-hr-°F/Btu. Therigorous exchanger model indicates theexisting depropanizer reboiler will meetfuture duty requirements. This conclu-sion is based on assumed fouling fac-tors and office-based calculations.However, the measured total foulingfactor calculated with accurate fieldmeasured data averages 0.009. Basedon the measured equipment perfor-

Figure 2. Depropanizer Kettle Reboiler

Page 4: hydroprocessing-depropanizer

4 WORLD REFINING March/April 1999

F O C U S O N : HYDROPROCESSING OPTIONS

mance, the exchanger is either under-surfaced or has high fouling. Anexchanger’s performance one week afterstartup is not an indicator of averageoperation and should not be the basisfor a revamp service exchanger rating.

Data, Numbers and Test RunsThe proper place to start a revamp,whatever the objective, is with a wellplanned and executed test-run of theoperating unit. Test-run planningidentifies specific data to be gatheredso that equipment performance can bequantified. The specific data is identi-fied on a “flag sheet.” “Why perform atest run, when we have data from ourcontrol system?” is the inevitable ques-tion that arises from management.This question is natural. After all, hav-ing spent what is most likely millionsof dollars for an advanced manage-ment information system, manage-ment expects that the numbers fromthe system will be usable for anydesired purpose. This question, whilenatural, arises from misconceptionsabout management information sys-

tems and the numbers they report.Management information systems(MIS) gather numbers from plantinstrumentation systems, store thenumbers and manipulate the numbersinto a variety of reports. At best, statis-tical methods may be used to fill in themissing numbers or to attempt to cor-rect obvious errors.

Using only the MIS (or plantinformation, PI) data to perform arevamp requires several assumptionsfor this approach to be successful.Three assumptions are of particularimportance. First, that nearly all of theinformation coming in is correct. Sec-ond, that statistical methods will cor-rectly identify the faulty data and fix itbased on a relationship with the cor-rect data. And last, that all therequired data to quantify equipmentperformance has been gathered. Theseassumptions often fail. First, plant pre-ventive maintenance to get instru-ments to read to a correct absolutereading is rarely done in a refinery.8 Itis generally not necessary for day-to-day operations. Having a trend line on

a plot is the objective of the PI and thisis how it is used by the experiencedoperators. These trend lines are greatfor showing if today’s operation is bet-ter or worse than yesterday’s, lastweek’s or the same month last year.Biases and offsets, built into the trendline for an entire time period, do notmatter in making comparisonsbetween time periods. In looking atabsolute limits of equipment perfor-mance being pushed in a revamp,absolute values are critical.

Second, advanced plant infor-mation systems all have means to rec-oncile plant data. Whatever themethod, the assumption is that someof the data is incorrect and must bereconciled. Reconciled data simplymeans the incorrect numbers are cal-culated from some relationship. It maybe a simple material balance (flowin=flow out) or the heat and materialbalance (HM&B). However, eachincreasing level of sophistication(material balance is less complex thanthe HM&B) must use more plant mea-sured data. Again, the under-lyingassumption of data reconciliation isthat computer models can correct anypiece of data given enough informa-tion. The need to reconcile the data iscaused by numerous problems frominstrument failures to incorrect instru-ment location in the field vs. the flowdiagram. The test-run is a structuredeffort to gather comprehensive compo-sition, temperature and pressure dataaround each major piece of equip-ment.9 If data reconciliation isrequired, it is applied locally where ithas a reasonable chance of working.Plant-wide data reconciliation is stillonly a dream.

Third, often the data required toquantify equipment performance is notroutinely gathered by the MIS systems.Returning to our FCC depropanizershown in Figure 2, a properly plannedand executed test-run must gather therequired data to quantify thedepropanizer reboiler performance.

Figure 3. Kettle Reboiler Flooding

Page 5: hydroprocessing-depropanizer

March/April 1999 WORLD REFINING 5

F O C U S O N : HYDROPROCESSING OPTIONS

Rigorous heat exchanger modelsuse the exchanger mechanical configu-ration and the process fluid physicaland transport properties to rate theheat exchanger. Heat exchanger pro-grams “rate” the exchanger based onuser input fouling factors and processconditions. If the data is not accuratethe exchanger rating will not be cor-rect. While errors may not matterwhen a new plant is built and theexchanger is significantly over-sur-faced, low cost revamps must fully uti-lize the existing equipment perfor-mance. When determining if a piece ofequipment can be pushed, these dataerrors can lead to wrong conclusions.

Computer ModelsEngineering computer models are use-ful tools, however, they are not apanacea. As an example, rigorous heatexchanger models require measuredpressure and temperature data to beproperly calibrated. The two majorproblems with rigorous heat exchang-er models are the user-input (guessed)fouling factors and calculated pressure.The exchanger fouling factor must becalculated from plant flowrates andtemperatures. Often the fouling factoris defined by the user based on an“industry” average, which is not accu-rate for the specific service. Addition-ally, the calculated tube-side pressuredrop from the exchanger models arenot correct unless it has been calibrat-ed with field measured pressure drop.This is especially true for fouling ser-vices. FCC slurry pumparound is ahigh boiling condensed aromaticsstream containing coke and catalystfines. The coke and catalyst contentvary depending on numerous factors.Catalyst and coke foul heat exchang-ers. The fouling tendency is unit spe-cific. Numbers should never be takenas accurate data and the data must becomplete.

In our earlier example, a mini-mum-cost revamp objective would beto increase depropanizer reboiler ser-

vice duty without having to replacethe exchanger. Figure 3 shows thedepropanizer kettle reboiler systemwith some relative equipment dimen-sions and the test-run measured pres-sure and temperature data. Kettlereboiler performance is affected by theshell and tube-side process flows. Dataanalysis and properly interpreting themeasured data is important. The mea-sured pressure drop from the columnto the reboiler is 2 psi. The kettlereboiler baffle elevation difference isonly 7'-6" to the reboiler vapor return.Distillation column kettle reboilerdesign requires the height of liquid inthe bottom of the column equal to thepressure drop from the column to thereboiler. Assuming the feed to thereboiler has a density of 0.68, the mea-sured column to reboiler pressuredrop requires 7' of liquid static head toovercome the 2 psi pressure drop.The measured pressure drop impliesthe liquid level in the bottom of thecolumn is near the vapor return noz-zle. This affects both the reboiler andcolumn performance. This problem

will limit the exchanger service dutyand must be corrected or the exchang-er service duty cannot be increasedbeyond current performance.

Returning to the example shownin Figure 2, increasing FCC reactorconversion requires increased depro-panizer reboiler duty. A conventionalproject’s process design approach hasconcluded that the reboiler surfacearea is adequate for the revamp condi-tions. Figure 4 shows the relative loca-tion of the depropanizer reboiler in theslurry pumparound circuit and somehydraulic data. Increasing the depro-panizer reboiler service duty by 20%requires increased slurry circulationrate to the exchanger. Increased slurrycirculation rate will impact thepumparound system hydraulics. Theexisting hydraulic limitations will dic-tate whether higher slurry flowratesare possible. Test-run measured pres-sure drop and valve position data helpquickly identify opportunities. Re-vamp process design uses this data inconjunction with the calculatedexchanger fouling factor to determine

Figure 4. Slurry Pumparound Hydraulics

REACTOREFFLUENT

Page 6: hydroprocessing-depropanizer

6 WORLD REFINING March/April 1999

F O C U S O N : HYDROPROCESSING OPTIONS

future performance. The followingpotential modifications will allow theslurry pumparound flowrate to beincreased:•Decrease tube side passes from 4 to 2• Increase slurry pumparound control

valve size•Pump impeller size or turbine speed

Prior to determining the mostcost-effective and reliable modifica-tion, the specific exchanger servicemust be considered. Previous operat-ing history on the exchanger indicatesa fouling rate that requires theexchanger be taken out of service after12 months. FCC slurry pumparoundservice has relatively high foulingrates. Fouling service exchangers areaffected by the process fluid, tube sidevelocity, fouling rate and fouling fac-tor. These are related. Higher exchang-er tube velocity lowers the fouling rateand the fouling factor. Conversely,lower tube velocity will increase thefouling rate. Industry “standards” statethe velocity in this service should be5-8 ft/sec, however, some refiners haveoperated at 13 ft/sec with minimal

tube erosion. FCC exchanger designs(hydraulics permitting) using slurrypumparound should be operated at 8ft/sec or higher to minimize the foul-ing rate and the fouling factor.

Process flow sheet models andrigorous equipment evaluation pro-grams are necessary tools for revampprocess design. They should never beused without comprehensive planttest-run data. Rigorous exchangermodels can be used to quantify tubevelocity, fouling factors and exchangersurface to meet future conditions. Inthe depropanizer reboiler case, theconventional project’s approach rec-ommended decreasing the number oftube passes to eliminate the slurrypumparound hydraulic limitations. Infact, lowering tube velocity willincrease the rate of fouling and reduceexchanger run length. The depro-panizer reboiler measured fouling fac-tor of 0.009 is above what can be tol-erated if future service duty requirementsare to be achieved. The only reliablemodifications to the depropanizerreboiler system is pump and/or control

valve changes to increase the tubevelocity and reduce fouling. Decreas-ing the number of tube passes is not areliable solution. Whether pump mod-ifications are feasible is situation spe-cific. However, decreasing the numberof tube passes aggravates an existingproblem. Reliable minimum-costrevamp designs are not determined byrote application of rules. There is onlyone given in a revamp and that is“there are no rules, only the specificsof the situation.”

Case Studies—Use Of Test Run DataIn recent years, FCC unit convertersection revamps have increased con-version through higher riser feed tem-peratures, higher catalyst/oil ratio, andimproved converter section hardwareand catalyst technology. These con-verter section revamps have affectedmajor FCC product recovery equip-ment. A product recovery revamp torecover the lighter reactor productswill need to evaluate the integratedmain fractionator, wet gas compressorand gas plant performance. Threeparts of the product recovery sectionthat are significantly impacted are theslurry pumparound system, gas plantreboilers and main fractionator heatbalance. These case studies will high-light examples where test-run data wasused to minimize capital investment,identify under-performing equipmentand implement reliable revamps inFCC product recovery section.

Slurry Pumparound Steam Generators-ExampleIncreasing conversion typically in-creases the slurry pumparound heatremoval requirements. Figure 5shows the four main fractionatorpumparound heat removal systems.Figure 6 shows the slurry pump-around system. The slurry pump-around system removes riser super-heat with two identical parallel steamgenerators (third is used as a spare)

Figure 5. Main Fractionator Pumparounds

Page 7: hydroprocessing-depropanizer

March/April 1999 WORLD REFINING 7

F O C U S O N : HYDROPROCESSING OPTIONS

and the debutanizer reboiler. The totalslurry pumparound system flowrategoes to the two steam generators witha portion of the steam generator outletused to reboil the debutanizer. Theremainder is by-passed around thedebutanizer reboiler. A converter sec-tion revamp will have the followingimpact on the slurry pumparound sys-tem: • Increase total slurry pumparound

duty• Increase the debutanizer reboiler

duty-more gasoline and alky feedSlurry pumparound steam gen-

erators can produce 150-600 psigsteam. The steam pressure is a func-tion of refinery steam system balanceand the investment cost. In this casesteam is produced at 250 psig fromtwo parallel kettle (TEMA AKT) steamgenerators. Typical industry total foul-ing factors for this service vary from0.003-0.006. These fouling factorsresult in a service U-value between 75-130 Btu/hr-ft2-°F. Applying this designcriteria to our system results in a cal-culated steam generator surface areaequal to twice that installed. The exist-ing steam generator rating shows thesteam generators under-surfaced,when they appear to be operating atmuch higher service U-values.

Figure 7 is a schematic of thekettle steam generator. The boiler feedwater rate, steam rate, and the specifictest-run data collected is indicated.The measured temperatures andflowrates are used to model the steamgenerator. The calculated service U-value is 240 Btu/hr-ft2-°F. Therefore,the exchanger was exceeding thedesign heat duty based on industrystandard service U-values by almost100%. The steam generator slurrypumparound inlet flow resulted in atube velocity of 13 ft/sec. Theexchangers had been operating forseveral years under these conditions.Every 12-18 months an exchanger istaken out of service for cleaning andthe spare exchanger put in service.

The implications of the test-run mea-sured data is important when deter-mining what equipment should bemodified.

Prior to assuming that the calcu-

lated U-value is correct, let’s use thetest-run data to check for consistency.First, the steam and the boiler feedmeter showed 88,000 lb/hr and92,000 lb/hr, respectively. Assuming a

Figure 7. Slurry Steam Generator

Figure 6. Slurry Pumparound Exchangers

Page 8: hydroprocessing-depropanizer

8 WORLD REFINING March/April 1999

F O C U S O N : HYDROPROCESSING OPTIONS

reasonable blow-down of 2% thesteam/water data are consistent. Basedon the measured service U-value andtheoretical calculations of shell andtube-side heat transfer coefficients atotal fouling factor of 0.0005 is calcu-lated. This is essentially a cleanexchanger. Flowmeters can be wrong,although both the slurry andwater/steam flows would all need tobe in error.

Comprehensive test-run dataincludes information to reconcile dataerrors. The test-run data included con-trol valve pressure drop and % open.Control valve data can be used toidentify major flowrate errors. Usingboth the boiler feed water and steamcontrol valve pressure drop, % open,

and valve Cv curve, the steam pro-duction was calculated at approxi-mately 90,000 lb/hr. Additionally theexchanger tube-side calculated pres-sure drop based on the metered slurrypumparound flowrate was 24 psi. Themeasured pressure drop was 22 psi.Therefore, the metered slurrypumparound flowrate also appear tobe correct.

The test-run data and subse-quent data shows high tube velocityresults in very high service U-values.The high slurry tube velocity keepsthe exchanger tubes clean. Revampingthis system by adding an additionalsteam generator in parallel will requirehigher slurry pumparound flow rate,otherwise the existing exchanger tubevelocity will decrease. Decreasing tubevelocity to 8 ft/sec lowers the calculat-ed service U-value clean to 200Btu/hr-ft2-°F. A qualitative revampconsideration must be made to deter-mine the impact of lower tube velocityon the existing exchanger fouling rateand the fouling factor. While operatingat a tube velocity above 13 ft/sec does

not follow design averages, it nonethe-less is reality. Minimizing capitalinvestment in this FCC would involvemaintaining the high slurry steam gen-erator tube velocity. Without accuratetest-run data this conclusion wouldnot be possible. If one wants rotedesign practices they should stick tograssroot projects and avoid thevagaries of revamps.

Debutanizer ReboilerDebutanizer reboiler duty increaseswhen the FCC unit conversionincreases. The duty increase is a func-tion of reactor yield shift and thedebutanizer fractionation objectives ofalky feed recovery, alky feed C5 olefincontent, and gasoline Rvp. Generally,the debutanizer reboiler duty increasesbetween 30–70% when conversion isincreased. Accurately measuring thedebutanizer service U-value is impor-tant. Figure 8 shows the measuredtemperature and pressure data for thereboiler. The low tube-side pressuredrop indicates low velocity (start-of-run). Low velocity will increase thefouling rate and increase the averagefouling factor over the exchanger runlength. This exchanger has a history ofsevere fouling. Debutanizer reboiler(using slurry) fouling can be high onboth the shell and tube-sides,although high shell-side fouling usual-ly results from using 700°F slurry.

Table 1 shows the calculateddebutanizer reboiler fouling factors forthree different FCC units using slurryas reboiler heat. The numbers varysignificantly. Variation is caused byslurry oil physical properties (APIGravity=-3.0 to 4.0), catalyst fines,coke fines, tube-side velocity andexchanger design.

Revamping the converter sectionincreases debutanizer reboiler duty sig-nificantly. Therefore, the slurrypumparound flowrate must beincreased to meet higher reboiler duty.This debutanizer reboiler has a calcu-lated fouling factor of 0.016. Rigorous

Figure 8. Debutanizer Reboilers

Table 1. Debutanizer Reboiler (Fouling Factor)

Unit # Fouling Factor(Ft2-hr-°F/Btu)

1 0.006

2 0.016

3 0.009

Page 9: hydroprocessing-depropanizer

March/April 1999 WORLD REFINING 9

F O C U S O N : HYDROPROCESSING OPTIONS

exchanger calculations indicate thetube velocity is approximately4 ft/sec.10,11 Ultimately, the revampengineer will have to make a decisionabout the relationship between foulingfactor and tube velocity. Reducing theoverall fouling factor from 0.016 (2months operation) to 0.009 significant-ly affects the service U-value. It is pos-sible to meet future reboiler duty forthis service by increasing tube velocityand decreasing the average fouling fac-tor to 0.009. Increasing slurrypumparound flowrate would increasethe exchanger tube velocity, which willreduce the fouling factor. Minimum-cost revamps require engineering deci-sions that are not part of any ease-of-use computer model.

Stripper (De-ethanizer Reboiler)Gas plant stripper columns remove C2

and H2S prior to feeding thedepropanizer or the debutanizer. Thereboiler duty largely tracks the % gaso-line, and to a lesser extent, the alkyfeed production. Increasing gasolineproduction by 20% will increase thestripper reboiler duty by 20-35%.Large changes in alky feed productionalso increase the stripper duty. Figure9 shows a stripper reboiler systemusing FCC main fractionator LCOpumparound (Figure 5) as reboilerheat. This is a relatively common gasplant/main fractionator heat integra-tion scheme. Often, the stripperreboiler performance is limited byLCO pumparound heat availability orpoor stripper reboiler draw/piping sys-tem design.

Figure 10 shows the reboilersystem and test-run measured streamtemperatures. The unit is designed as aonce-through reboiler system. Thestripper column bottom tray liquidfeeds two identical parallel reboilers.The following assumptions are madewhen this system was designed:• 100% of bottom tray liquid feeds the

two reboilers• 50% of the bottom tray liquid flow

to each reboiler• 50% of the LCO pumparound total

circulation rate flow to each reboilerThe plant piping and column

internal design will dictate whether

the above design assumptions are cor-rect. Test-run measured temperaturedata should be used to identify once-through and parallel reboiler operatingperformance. Evaluating the measured

Figure 9. Deethanizer (Stripper) Reboilers

Figure 10. Deethanizer Reboiler Temperatures

–TEMPERATURE, °F

Page 10: hydroprocessing-depropanizer

test-run temperatures shows that allthe above assumptions are incorrect.

A once-through thermosyphonreboiler is essentially one theoreticalstage of fractionation. These reboilersare designed to take advantage of largetemperature differences between the

bottom tray liquid and bottom prod-uct. Using a once-through reboilermaximizes LMTD. FCC, saturate gasand delayed coker strippers (de-etha-nizers) all have 40-60°F temperaturedifferences between the bottom trayliquid and the column bottom prod-

ucts. If bottom tray liquid by-passesthe reboiler, then the reboiler returntemperature must be higher to meetthe column bottom product C2 controlobjective. In this case (more often thannot), a valve tray is used to draw liq-uid to the reboiler system. Active traysshould never be used to draw liquid toa once-through reboiler because theywill leak. Figure 11 represents theactual operation of this reboiler sys-tem. The bottom product temperatureis 12°F colder than the reboiler vaporreturn (measured at the column). Thisindicates that some of the bottom trayliquid leaks through the tray deck.

The leakage rate can be estimat-ed with a computer model. Replacingthe active tray with a seal-welded col-lector tray will ensure all the liquidgoes to the once-through reboiler.

The reboiler design assumesequal process stream flows to bothreboilers. Evaluating the measuredtemperatures shows the tube andshell-side process flows are not equal,although, the exact imbalance cannotbe determined. The two reboiler outletvapor stream temperatures vary by 5°Fand the LCO pumparound outlet tem-peratures are not equal. This reboilersystem was designed improperly,resulting in under-utilized exchangersurface area. A revamp should consid-er opportunities to fully utilize existingequipment.

FCC Main Fractionator Heat BalanceIncreasing reactor conversion shiftsproduct yields from LCO and decantoil to gasoline and lighter product.Reducing the LCO and decant oil prod-uct yields, shifts the main fractionatorheat balance. Figure 5 shows an FCCmain fractionator pumparound system.When gasoline yield is increased, thereflux ratio in the gasoline/LCO frac-tionation section must be maintained.The top pumparound duty must beincreased to maintain gasoline/LCOseparation. Concurrently, the LCO

10 WORLD REFINING March/April 1999

F O C U S O N : HYDROPROCESSING OPTIONS

Figure 11. Reboiler Draw System

Figure 12. Main Fractionator Gasoline/LCO Fractionation

–TEMPERATURE, °F

Page 11: hydroprocessing-depropanizer

March/April 1999 WORLD REFINING 11

F O C U S O N : HYDROPROCESSING OPTIONS

pumparound duty decreases becausemore column internal vapor must flowto the top pumparound. At the sametime, the LCO pumparound dutydecreases, the gas plant stripper reboilerduty increases. In this case, the LCOpumparound supplies the stripperreboiler heat, then the reactor conver-sion increase creates a fundamentalmain fractionator/gas plant heat integra-tion system design problem. Minimum-cost revamps must address the integrat-ed system affects of shifting yields.

Figures 12 and 13 show a poten-tial revamp that addresses both theincreased main fractionator toppumparound heat removal and theincreased stripper reboiler dutyrequirements.12 Assuming the stripperreboiler design uses two parallel reboil-ers, then one can be converted to toppumparound service. The feasibility ofthis will be unit specific, however, thisspecific modification has been imple-mented successfully. The toppumparound draw temperature is typi-cally 325°F vs. an LCO pumparounddraw temperature of 440°F. Therefore,the top pumparound exchanger LMTDis less than the LCO pumparoundexchanger. The use of series reboilershas advantages that are not apparentunless the reboiler system is thorough-ly evaluated.

Using top pumparound heat forone of the two parallel reboilersrequires a thorough evaluation of thereboiler system. Thermosyphon serviceheat transfer coefficient is affected byboth a heat flux limitation and theshell-side percent vaporization. Theheat transfer coefficient drops whenvaporization increases beyond25–30%. Stripper reboilers using oneheat service (Figure 9) have a vaporiza-tion rate between 40-50%. Therefore,the inherent stripper reboiler serviceU-value is low due to high percentvaporization. Also, the heat flux inthese services is limited to about13,000-15,000 Btu/hr-ft2. Dependingon the LCO pumparound circulation

rate, the flux limitation often preventstaking advantage of the higherexchanger LMTD. Once the exchangerflux limit is reached, the service U-value decreases as the LMTD isincreased. Higher LCO pumparound

circulation rate increases LMTD butdecreases the service U-value. Heat fluxlimitations result in no additional strip-per reboiler duty.

Figures 12 and 13 modifica-tions address both the main fractiona-

Figure 13. Deethanizer Reboiler Revamp

Photo 2. Maximize Equipment Utilization

Page 12: hydroprocessing-depropanizer

Copyright © Hart Publications • 4545 Post Oak Place, Suite 210 •Houston, TX 77027 • 713/993-9320

F O C U S O N : HYDROPROCESSING OPTIONS

tor heat balance problem and aninherently poor reboiler systemdesign. All of the bottom tray liquid iswithdrawn with a seal-welded collec-tor tray to ensure maximum toppumparound exchanger LMTD. Thetop pumparound reboiler supplies40–60% of the total stripper columnreboiler duty. The top pumparoundreboiler return feeds one side of thecolumn bottom, which is partitionedwith a baffle. The liquid from the toppumparound reboiler return-side ofthe baffle feeds the LCO pumparoundreboiler. This series reboiler arrange-ment reduces vaporization to lessthan 30% in both reboilers, thereforethe service U-values go up. The resultis maximum service U-value and min-imum exchanger surface area. Anadditional benefit is improved energyrecovery. The main fractionator heatbalance problem is corrected byincreasing top pumparound duty anddecreasing LCO pumparound dutyfor the stripper reboiler.

Equipment CostsTest-run data and analysis are a vitalpart of the revamp process. Revampprocess design quantifies equipmentperformance, identifies existing unitequipment bottlenecks and economi-cally eliminates the bottleneck.13

Understanding how the integrated sys-tem works and modifying the systemmaximizes use of existing equipment.Process design controls the majority ofcapital investment, not project execu-tion and procurement. Project executionand procurement are important. Never-theless, these functions change a smallproject (<20–30 MM$) installed cost byonly 10–15%. However, process designdetermines what equipment must bemodified. It is minimizing the pump,piping, heat exchanger and tower mod-ifications that have the biggest impacton a revamp installed cost. (Photo 2).Revamp process design must focus onmeasuring existing equipment perfor-mance and conceptual aspects of cir-

cumventing the limitations. Test-runsare the starting point of a revamp, nottheoretical calculations. ■

AcknowledgmentPrepared for Presentation at theProcess Optimization Conference GulfPublishing Company, Houston, Texas,7-10 April 1997

References1. Lieberman, N. P., and Lieberman,

E.T., “Design and Installation Pit-falls Appear in Vacuum ColumnRetrofit,” Oil and Gas Journal, Aug.21, 1991, pp. 75-79.

2. Lieberman, N.P., and Lieberman, E.T., “Inadequate Inspection Cause ofVacuum Tower Revamp Failure,”Oil and Gas Journal, Dec. 14, 1992,pp. 33-35.

3. Bloch, H. P., “How the Best Petro-chemical Petrochemical Companieswill Achieve Reliability,” HydrocarbonProcessing, July 1996, pp. 83-86.

4. Golden, S. W., “Revamping FCC’s-Process and Reliability”, PetroleumTechnology Quarterly, Summer1996, pp.85-93.

5. Golden, S. W., “Minimize CapitalInvestment for Refinery Revamps,”Hydrocarbon Processing, Jan. 1997,pp. 103-112.

6. Kowalczyk, D., and Golden, S. W., “FCC Optimization–A Minimum

Capital Approach,” Fuel Technologyand Management, March/April1996, pp.37-45.

7. Golden, S. W., “Approaching the Revamp Project,” Hydrocarbon Tech-nology Quarterly, Autumn 1995,pp.47-55.

8. Sloley, A. W., “Avoid Problems Dur-ing Distillation Column Startup,”Chemical Engineering Progress,July 1996, pp.30-39.

9. Golden, S. W., “Temperature, Pres-sure Measurements Solve ColumnOperating Problems,” Oil and GasJournal, Dec. 25, 1995.

10.Sloley, A. W. and Martin, G. R.,“Effectively Design and SimulateThermosyphon Reboiler Systems:Part 1,”Hydrocarbon Processing, June1995, pp.67-78.

11.Sloley, A. W. and Martin, G. R.,“Effectively Design and SimulateThermosyphon Reboiler Systems:Part 2,”Hydrocarbon Processing, July1995, pp.101-110.

12.Golden, S.W., Sloley, A. W., andFleming, P. B., “Revamping FCCUnit MainFractionator Energy Sys-tems,”Hydrocarbon Processing, Nov.1993, pp.43-50.

13.Golden, S. W., Schmidt, K. D., and Martin, G. R., “Field Data, NewDesign Correct Faulty FCC TowerRevamp,” Oil and Gas Journal, May31,1993, pp.54-60.

The authorScott W. Golden is a chemical engineer withProcess Consulting Services Inc., Houston, TX.His work includes field troubleshooting andapplying fundamental chemical engineeringskills to improve refinery profitability. The com-pany provides revamp process design, optimiza-tion and troubleshooting to the refinery indus-try. Mr. Golden was previously a refineryprocess engineer and distillation system trou-bleshooter. He has a BS in chemical engineeringfrom the University of Maine. He has authoredand co-authored more than 75 technical papersconcerning refinery unit troubleshooting, designand simulation.


Recommended