Subscriber access provided by UNIV OF UTAHEnergy & Fuels is published by the American Chemical Society. 1155 Sixteenth StreetN.W., Washington, DC 20036Published by American Chemical Society. Copyright American Chemical Society.However, no copyright claim is made to original U.S. Government works, or worksproduced by employees of any Commonwealth realm Crown government in the courseof their duties.ArticlePredictive Modeling and Optimization for anIndustrial Penex Isomerization Unit - A Case StudyMohanad M. Said, Tamer Samir Ahmed, and Tarek M. MoustafaEnergy Fuels, Just Accepted Manuscript Publication Date (Web): 23 Nov 2014Downloaded from http://pubs.acs.org on December 1, 2014Just AcceptedJustAcceptedmanuscriptshavebeenpeer-reviewedandacceptedforpublication.Theyarepostedonline prior to technical editing, formatting for publication and author proofing. The American ChemicalSocietyprovidesJustAcceptedasafreeservicetotheresearchcommunitytoexpeditethedissemination of scientific material as soon as possible after acceptance. Just Accepted manuscriptsappear in full in PDF format accompanied by an HTML abstract. Just Accepted manuscripts have beenfully peer reviewed, but should not be considered the official version of record. They are accessible to allreaders and citable by the Digital Object Identifier (DOI). Just Accepted is an optional service offeredtoauthors.Therefore,theJustAcceptedWebsitemaynotincludeallarticlesthatwillbepublishedin the journal. After a manuscript is technically edited and formatted, it will be removed from the JustAccepted Web site and published as an ASAP article. Note that technical editing may introduce minorchangestothemanuscripttextand/orgraphicswhichcouldaffectcontent,andalllegaldisclaimersandethicalguidelinesthatapplytothejournalpertain.ACScannotbeheldresponsibleforerrorsorconsequencesarisingfromtheuseofinformationcontainedintheseJustAcceptedmanuscripts. 1 ` Predictive Modeling and Optimization for an Industrial Penex Isomerization Unit - A Case Study Mohanad M. Saida, Tamer S. Ahmeda,*, Tarek M. Moustafaa

a Chemical Engineering Department, Faculty of Engineering, Cairo University Giza 12613, Egypt * Corresponding author: Tel.: +20 114 292 4407; E-mail address: Tamer.S.Ahmed@cu.edu.egAbstract.ThisworkpresentsamodelforUOPHydrogenOnceThrough(HOT)Penex ProcessusingAspenHYSYSPetroleumRefiningmodule.Themodelreliesonroutinely takenindustrialdataofprocessstreamsduringnormaloperatingconditions.Acquireddata sets have been tested and screened to ensure data validity for building the model and avoiding erroneous results. A reaction network with 20 reactions and 19 components has been used for the reactors model. The reactors model has been validated using 4 months of industrial plant data.In addition,rigoroustray-to-traysimulation ofisomeratestabilizerhasbeenutilizedto match the performance of plant stabilizer. The model validated has been used for studying the effects of each process variable on plant performance. In addition, the model has been used in optimizingtheoperating conditionsof the process.This optimizationshowedapotential for notable fuel savings in the process. Page 1 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 2 ` 1.Introduction Sincethe70s,eliminationofleadcompoundsfromgasolinepoolhasincreased interestintotheisomerizationoflightstraightrunnaphthasoastopreservetheoctane number ofproduced gasoline1. Recent and future regulations put strict limits on benzene and aromaticlevelsinmotorgasoline.Forexample,theU.S.EnvironmentalProtectionAgency MSATII(MobileSourcesAirToxicsPhase2)regulationsrequiresdecreasingtheaverage benzene content of U.S. gasoline pool to 0.62 vol.%2,3. This urged refiners to search for other sources of gasoline with a lower aromatic content than catalytic reforming.Isomerization is thoughttobeoneoftheeffectivesolutionstoproducemotorgasolinecompatiblewith environmental regulations. Isomerate (isomerization product) is highly desirable with respect to environmental regulations due to its zero benzene content and high octane number. Isomerizationreactionsareexothermicequilibrium-limitedreactions.Asconversion isfarfromequilibriumconversion,anincreaseinreactortemperatureleadstoincreasein reactionvelocityandsubsequentincreaseinconversion.However,onceequilibriumis approached,increasingtemperaturedecreasesconversionduetothedecreaseofreaction equilibriumconstant4.Figure1showsthechangeofconversion(iso-paraffinsyield)with temperature. Asindicatedin Figure1,conversion increaseswithtemperatureuntilacertain temperature (optimum temperature) is reached, then iso-paraffins yield decreases. INSERT FIGURE 1 Currently,threecatalysttypesareusedcommerciallyfornaphthaisomerization (Figure2).Allofthemareplatinumcontainingcatalysts6:1)Zeolitecatalysts:zeolite catalystshavethelowestactivityamongisomerizationcatalysts,hencetheyareusedat highertemperaturesthatareunfavorablewithrespecttoisomersyield.However,these Page 2 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 3 ` catalysts have high resistance to feed impurities and can be regenerated. Zeolite catalyst units use firedheaters forfeedand hydrogenheatingup tothereaction temperature.Theprocess also employs a high ratio of hydrogen to hydrocarbon for hydrotreating and dearomatization offeedstock.Accordingly,arecyclegascompressorandaproductseparatorareused;2) Chlorinated alumina catalysts: These are the most active isomerization catalysts providing the highest octane number and isomerate yield. These catalysts require continuous injection of a chlorine compound (CCl4) to maintain catalyst activity. In addition, they are very sensible to impurities(oxygen,sulfurandnitrogencompounds).Therefore,feedhydrotreatingand dryingisamandatory.Lowhydrogentohydrocarbonratioisrequired.Hence,neithera recycle gas compressor nor a recycle gas is needed; 3) Sulfated zirconia: These catalysts have theadvantagesofbothprevioustypes.Theyaremoreactivethanzeolitecatalysts,hence favoringhigherisomersconversions.Inaddition,theyareresistanttoimpuritiesand regenerable.However,unitsusingsulfatedzirconianeedarecyclegascompressoranda product separator. INSERT FIGURE 2 As seen in Figure 2, chlorinated alumina catalysts provide the highest octane number and isomerate yield. By good operation of upstream hydrotreating unit and feed dryers, a long servicelife(morethan10years)couldbeachieved2.UOPandAxensarethelicensorsof processesusingchlorinatedalumina-basedcatalysts.UOPlicensestheprocessunderthe nameof"PenexProcess".ThefirstPenexprocesswasbroughtonstreamatBorgerTexas Refinery using I-3 catalyst7.Mostoftheliteratureconcentratesondevelopingnewtypesofcatalysts8.Afew research work paid attention to kinetic modeling of isomerization reactions and a fewer paid Page 3 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 4 ` attentiontomodelingisomerizationreactionsoverchlorinatedaluminacatalysts. Consequently,optimizationforindustrialisomerizationunitsisveryscarceintheopen literaturesincetherigorouskineticmodelingforthereactoristheheartoftheoptimization process. Besl et al.9 discussed briefly the optimization of Penex process in a German refinery. Ontheotherhand,DudleyandMalloy10developedasimplekineticmodelbasedonlyon isomerizationandcrackingreactionsthatwasusedinoptimizingaprocessthatusesAlCl3 liquidcatalyst.Ahari and coworkers11investigatedtheeffectsof methyl cyclopentane inthe feedofisomerizationfeedusingaprocessmodeldevelopedbyHYSYS.Inaddition,they studiedexperimentallythehydrogenpartialpressureeffectonPtmordenitezeolitecatalyst activityandconversionofn-paraffinsandproposedkineticequationsforn-C5andn-C6 conversion12. Brito et al.13 studied experimentally the performance of Pt-Ni/mordenite zeolite catalystswithdifferentmetalproportionsandkineticmodelforcatalystdeactivationwas proposed.Koncsagetal.14proposedakineticmodelforC5/C6isomerizationoverPt/H-zeolite atindustrial conditions. Surlaet al.15used a single eventmethodologyto establish a kineticmodelforC5/C6isomerizationoverchlorinatedaluminacatalyst.Finally,adetailed kinetic model that is suitable for the three major catalyst types was proposed by Chekantsev etal.8.Thereactionnetworktheyproposedcontained36reactions.Theauthorsillustrated thatthedifferencesinreactionratesoverdifferentcatalystsaremodestexceptforfew isomerizationreactions.Themodeltheyproposedagreedwellwithexperimentalresultsof the three catalyst types. Verylittleattentionwaspaidintheliteraturetotheapplicationofkineticmodeling andoptimizationtoanexistingindustrialunit.Inthiscontext,wepresentherekinetic modeling for an existing industrial Penex isomerization unit using Aspen HYSYS Petroleum Refining isomerization reactor model. In addition, the model has been used for studying the effect of different process variables on process performance and for process optimization. Page 4 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 5 ` 2.Process Description Figure 3 shows a process flow diagram for Penex process. Feed naphtha and make-up hydrogenfirstpassthroughdrierstoeliminateanytracesofwaterbecausewaterisa permanent poison of Penex catalyst. Then, naphtha and hydrogen are mixed priorto heating themixtureuptoreactiontemperature.Maintainingaproperhydrogenpartialpressureis required inside the reactor to prevent coke deposition on catalyst. The reactor charge mixture isheatedbyexchangingheatwiththesecondandfirstreactoreffluent,respectively.A chlorinecompound"CCl4"isinjectedintothereactorchargetoprovideacidsiteson catalyst'ssurfacethatisrequiredforisomerizationreaction.Thefeedisbroughtuptothe reactiontemperaturethroughafiredheater.Theeffluentofthefirstreactoristhencooled throughexchangerspriortoenteringthesecondreactortoremoveheatgeneratedby exothermic reactions in the first reactor bed so that to favor equilibrium limited isomerization inthesecondreactorbed.Thereactors'effluentisthenfedtoastabilizertoseparatelight gases(C4- andhydrogen)fromtheproductstream.Theoverheadgasesissenttoapacked bedscrubberthatemploysacausticwashtoneutralizehydrogenchlorideformedfromthe decompositionofthechlorinecompound.Finally,theproducedgasesaresenttovapor recovery for LPG production. Thestabilized isomerate may be sent directly to gasoline pool or may undergo fractionation to maximize the octane number of the isomerate. INSERT FIGURE 3 3.Process Chemistry Praffin Isomerization. Paraffinisomerization is the main reaction in the process. As mentioned before, paraffin isomerization is an exothermic equilibrium limited reaction so that Page 5 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 6 ` a higher conversion is favored at low temperatures. Table 1 lists the research octane numbers (RON) of n-C5 , n-C6 and their isomers. It is clear that multi branched isomerssuch as 2,2--dimethyl butane (2,2 DMB) and 2,3-dimethyl butane (2,3 DMB) have higher octane numbers than single branched isomers such as 2-Methyl Pentane (2MP) and 3-Methyl Pentane (3MP). Thus, formation of multi-branched isomers is highly desirable. INSERT TABLE 1 Naphthenes isomerization. Naphthene isomerization reaction is also an equilibrium-limitedreaction.Cyclohexane(CH)andmethylcyclopentane(MCP)existinequilibriumat reaction conditions but formation of MCP increases as temperature is increased. Benzene Saturation. Benzene saturation to cyclohexane is a highly exothermic rapid reaction.Thehighheatofreactionofbenzenesaturationaffectstheconversionofthe exothermic isomerization reaction that is favored at low temperature. This puts a limit on the amount of benzene that can be tolerated in reactor feed. All benzene in the feed is saturated completelyintheleadreactor.Actually,theleadreactortemperatureisalwaysadjustedto maximizeisomerratios.Ifthebenzenecontentoffeedincreasesthenalowerreactorinlet temperature is used to maximize isomers' conversion attained at lead reactor outlet and vice versa. Page 6 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 7 ` Naphthene ring opening. Naphthenes may undergo ring opening and form paraffins at reactor conditions. Ring opening increases with increasing reactor bed temperature. Hydrocracking.Asreactortemperatureincreases,hydrocrackingrateincreases. Largemolecules(C7)areeasiertocrackthansmallermolecules.AsC5/C6isomerization approachesequilibrium,theextentofhydrocrackingincreasesleadingtolowerliquidyield and an increase in stabilizer overhead gas (C4-). 4.Reactor Model AspenHYSYSv.7.3PetroleumRefiningisomerizationreactormodulewasusedin developing the process model. The module contains a detailed kinetic model of reactions that occur inisomerizationprocess.Thereactormodel containsrate equationsforisomerization, benzenesaturation,ringopening,hydrocracking,andheavy(C7+)reactions(Figure4). The rateexpressionforeachreactionclassiscodedtomatchliteraturedata.Allreactionsare irreversible except for isomerization and benzene saturation. Typically, the reaction network consistsof20reactionsand8ofthemarereversible.Eachreactionclassisfirstorderwith respecttotheprimaryreactant andreaction-classrate equationhas adenominator following Langmiur-Hinshelwood-Hougen-Watsonmechanism.Thereactionschemecontains hydrocarbons up to C7. Higher carbon components are mapped into six ring C8 naphthenes. In reality,isomerizationfeedusuallycontainstraceamountsofC7+ components.Reactionrate equations are expressed in the model as: Reaction Rate = Global activity Reaction-class Activity Heterogeneous Reaction Rate (1) Reactormodelcanbetunedtomatchplantreactorperformanceviatwoschemes: basictuningandadvancedtuningschemes.Basictuningincludesthetuningoftheactivity parametersofthereactorsuchasglobalactivity(overallreactoractivity)andspecific Page 7 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 8 ` reaction-class activity parameters (activity of each type of reactions, e,g,: hydrocracking class reactions).Global activity parameter affects the rates of all the reactions. On the other hand, theactivityofeachspecificreaction-classcanbeadjustedviathespecificreaction-class activity parameter (e.g.: hydrocracking activity parameter for hydrocracking-class reactions). Ifbasic tuning isnot sufficient,kineticparametersof each individualreactionmaybetuned to match plant performance using the advanced tuning. ThePenexunitunderinvestigationisaHOT(Hydrogen-Once-Throughunit).The processflowdiagramoftheunitisidenticaltothatshowninFigure3.Thecurrentworkis only concerned with the isomerization reactors and the downstream stabilizer. Dryers and gas scrubber modeling are beyond the scope of this study. No changes or special procedures were appliedforbuildingthemodelandthemodelwascalibratedaccordingtosamplesthatare routinely taken during normal operation. INSERT FIGURE 4 IndustrialData.Data from the industrial unit under investigation were gathered and wereorganizedintodatasets.Eachdatasetrepresentsanoperatingday.Eachdataset includesacomponentanalysisofallinputandoutputstreams(make-uphydrogen,feed naphtha, isomerate and stabilizer off-gas) and component analysis of the lead reactor product. In addition, each data set includes the operating conditions of the reactors and the stabilizing column. C7+ de-lumping. Isomerization feed usually contains a little amount of C7+ hydrocarbons and this amount could be controlled via the upstream naphtha splitter. Figure 5 shows the varaiation of the amount of C7+ in feed during study. The C7+ is mostly around 1 Page 8 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 9 ` wt%. The C7+lump in isomerization feed is expected to be mostly normal heptane, heptane isomers and C7naphthenes, with almost no toluene since the boiling point of tolueneis 110C, which is far from that existing in the isomerization feed.ForaccuraterepresentationofC7+lump,thede-lumpingproceduredevelopedby Riazi18wasusedfortheC7+ lumpinthecurrentstudy.Theprocedureusescorrelationsfor estimating the paraffins, naphthenes and romatics (PNA) composition of petroleum fractions usingonlybulkproperitiessuchasspecificgravity(SG)andmolecularweight(MW).The results of PNA composition calculations for C7+ in feed are presented in Table 2. The results reinforcethepostulationthattolueneisnegligableintheisomerizationfeed.Therefore,the C7+ fraction is de-lumped into C7 paraffins and C7 naphthenes with known percentage of each hydrocarbon class. INSERT FIGURE 5 INSERT TABLE 2 Properitiesofapetroleumfractionofknowncompositioncouldbecalculatedfrom the properities of constituent pure componentsby applying the proper mixing role: = (2) Usingthisprinciple,anoptimizationalgorithmcanbeusedtoestimatethe compositionofapetroleumfractionbyminimizingtheerrorbetweentheknownproperities ofthatfractionandthecalculatedproperities.Inthiswork,thefollowingobjectivefunction was minimized: Page 9 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 10 ` = 100(3) This isthe same principle that was usedin the literature to estimate the composition of catalytic reformer feed19,20. However, in the current work, a limited number of properties of C7+ are available (SG, MW, Reid Vapor Pressure (RVP) and RON). This limits the number of compositions that can be estimated. Usingthepreliminaryde-lumpingobtainedthroughPNAcompositioncalculation, two additional equations were added for compositions estimate. In this work, the C7+fraction wasde-lumpedintofivepesudocomponents:multi-branchedheptaneparaffins(MBP7), singlebranchedheptaneparaffins(SBP7),normalheptane(NP7),fiveringC7 naphthenes (5N7),andsixringC7naphthenes(6N7).Propertiesofthesepesudocomponentswere obtained from Aspen properities data bank. Mole fractions of pesudo components wereused for the calculation of RVP and MWand volume fractions were used for SG and RON. Table 3 shows a comparison between available properities and calculated properities for C7+ in feed and product streams. Estimated compositions of C7+are shown in Table 4. INSERT TABLE 3 INSERT TABLE 4 Data Screening. Model quality depends mainly on the data used. As indictaed earlier, datasetsusedfordevelopingthisprocessmodelwereobtainedduringnormaloperation .Obtainingconsistentdatafromindustrialunitsmaysometimesbeanextremedifficult missionduetofrequentchangesinfeedandprocessconditions.Datasetsindaysthat Page 10 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 11 ` witnessedanyup-normalchangesinprocessoperationwereexcluded.Inaddition,amass balancetestwas applied to alldatasets andany datasetshavinga mass balanceerrormore than 2% wereexcluded. In order to ensure data verification, a hydrogen balance was applied alsotodatasetsbycalculatingthehydrogenweightineachstreamthroughsummingthe hydrogencontributionofeachcomponentandanysetshavinganerrormorethan3%were omitted.TaskarandRiggs20usedthefollowingformulatocalculatetheweightfractionof hydrogen in component CiHj (H factor): = + (4) The weight of hydrogen in each stream could be calculated by: = (5) Apartfromdatasetsindaysthatwitnessedcapacitychanges,mostdatasetswere quite consistent having an average mass balance and hydrogen balance error of -0.969% and-1.481%, respectively. ModelCalibrationandParametersEstimation.Parameterestimationisthemost criticalstepinmodelbuilding.Awell-calibratedmodelproducessignificantandrepeatable predictions over a wide range of operating conditions. Improper calibration of reactor model mayleadtoanovercalibratedmodelwithapoorpredictivepower.AspenHYSYS isomerizationreactormodelenablesusertomatchplantperformancethroughbasictuning andadvancedtuningbyadjustingkineticparametersforeachreaction.Modeldevelopers claim that basic tuning of reactor model is sufficient to match plant data. In this work, a basic tuningschemewasappliedinordertoavoidovercalibrationofreactormodel.Advanced Page 11 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 12 ` tuningmayrequireextremelyaccurateplantdatathatwerenotfeasibleinourcase.For accurate representation of plant reactor, the upcoming steps were followed: A- Generating streams:Eachdatasetwasusedtogenerateastreamrepresentingtheleadreactorchargeattheday that each data set represents. B- Running the model: Thereactormodelrequiressomemechanicaldatatorun.Thesedataincludereactor dimensions and catalyst properties (catalyst bed porosity and bulk density).Then the model is allowed to run generating an output stream. C- Model calibration: Reactor model calibration is mainly adjusting the activity parametersto match plantdata. In thiswork,thereactormodelwascalibratedwiththeaidofAspenHYSYSOptimizerby minimizing the following objective function: f = +

(6) Table5liststheadjustmentfactorsusedforreactormodelcalibrationandapplied bondsforeachparameter.Itwasfoundthatacceptablemodelperformanceisreachedwhen using those bonds during calibration. It is important not to include yields of all components in reactorcalibration.PashikantiandLiu21showedthatincludingallmeasurementsinreactor calibration may result in a poor calibrated model that responds wildly even to small changes toinputvariables.Table6showsthereactormeasurementsthatwereincludedin optimization function and weighting factors used with them. Page 12 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 13 ` INSERT TABLE 5 INSERT TABLE 6 Largerweightingfactorswereappliedfortermswhereacloserfitisrequired.For example,thereactormodelisrequiredtofittheplantisomerratios.Therefore,larger weightingfactorswereappliedtonormalandiso-paraffins.Onthecontrary,thelowest weightingfactorwasgiventoreactortemperaturerisesincetheleadreactortemperatureis frequentlyadjustedtomaximizeisomerratiosinthereactoreffluentandhencethereactor temperature rise usually fluctuatesduring operation. In the availabledata sets, an average of reactor inlet and outlet temperature is only given. Therefore, temperature rise may be the least reliable data point in each data set and applying a high weighting factor for temperature rise may result in poor calibration. The three previous steps were repeated for each data set and activity parameters were calculatedforeachdatasetindividually.Almostallactivityparameterswerefoundtovary withinanarrow range for different data sets. Consequently, the average values of calculated activity parameters are expected to be quite satisfactory for model calibration. Thesameprocedurewasusedforlagreactorcalibration.Thecompositionofthe effluent of the calibrated lead reactor was used to simulate the composition of lag reactor feed atplantconditions.Thelagreactorwascalibratedbyminimizingerrorbetweenmodeland plant data using the same objective function. Routine sampling in the industrial unit does not includeasampleforlagreactoreffluent.Thus,thecompositionoflagreactoreffluentwas calculated through back-mixing of isomerate and stabilizer off-gas streams. The same activity parameterswereusedformodelcalibrationandsamebondswereusedexceptforglobal activityparameter.Inisomerizationunits,theleadreactorcatalystlosesactivitybeforelag Page 13 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 14 ` reactor. When lead reactor performance becomes unsatisfactory, catalyst in the lead reactor is replacedwithfreshcatalystandtheorderofreactorsisreversedthatthenewloadedlead reactorlagstheexistinglagreactor(whichisnowtheleadreactor).Therefore,lagreactor catalyst is always more active than lead reactor catalyst. Due to this fact, a higher upper bond wasgiventoglobalactivityparameterduringlagreactorcalibration.Table7showsthe average activity parameters estimated through calibration process for both reactors. INSERT TABLE 7 Modelvalidationandtesting.Figures6and7showthemodelperformanceversus plantyieldsforthe23datasetsusedinreactormodelcalibrationprocessforleadandlag reactors,respectively.Themodelpredicationsaresatisfactoryforbothreactor.Itshouldbe noted that a closer fit may be achieved with advanced tuning of kinetic parameters (especially therateconstantsofringopeningreactions).However,advancedtuning requiresstrictplant measurements, which were not available. INSERT FIGURE 6 INSERT FIGURE 7 In order to ensure model capability of predicting plant performance, the model yields were compared with plant yields for the next 4 months after calibration. Chlorinated alumina catalysts lose activity very slowly during normal operation (the catalyst used in the industrial unitunderinvestigationwasloadedfromabout13yearsandisstillbeingusedwithgood Page 14 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 15 ` activity).Thus,changesincatalystactivityduringthe4monthsarenegligibleandre-calibrationofreactors'modelsisnotrequired.Figures8and9showthatthemodelwas useful in tracking plant performance for the 4 months after the calibration. INSERT FIGURE 8 INSERT FIGURE 9 StabilizerModel.Although this study is concerned with the optimization of process variablesofreactionsectiononly,aprecisestabilizermodelisalsorequiredforaccurate prediction of isomerate yield. The standard inside-out method was used for stabilizer model22. Theinside-outmethodconvergesquicklywithawidevarietyofspecifications.Sincethe isomerate stabilizer is used to adjust the RVP of isomerate product by limiting the amount of C4-intheisomerateproduct,itfunctionsverysimilartothefunctionofade-butanizer column.DataprovidedbyKaes23showthattheoverallefficiencyofade-butanizeris85-90%.Since actualplant stabilizerhas30 trays,the modelstabilizer should contain about26 theoretical trays. AccordingtotheguidelinesprovidedbyKaes23,stabilizerspecificationswere adopted. Thefunctionofisomeratestabilizeristostabilizeisomerateproductbyseparating light ends fromit.Therefore, the RVPofisomerate is anindicationofthe recovery oflight ends and degree of separation achieved. Since the stabilizer operates with full reflux, another specificationwasneededforbuildingstabilizermodel.Therefore,theoverheadcondenser temperature wasspecified asaperformance specificationsincethecondenseroperateswith significantvaporproductflows23.Figure10showsacompleteprocessflowdiagramofthe process model. Page 15 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 16 ` INSERT FIGURE 10 5.Process Variables Having established a process model for Penex unit, the next step is studying the effects of each process variable to predict the process performance for a given change in a process variable. Studying the effects of each process variable is essential for process optimization. The following studywasconductedusingtheaveragecompositionofindustrialunitfeedduringthestudy period (6 months). LeadReactorInletTemperature.Isomerizationreactionsareequilibriumlimited reactions; hence there is a maximum conversion (equilibrium conversion) that could be attained at each temperature (Figure 1). Because isomerization reactions are exothermic, the equilibrium conversiondecreasesastemperatureincreases.Wheneverthereactionisfarfromequilibrium conversion,anincreaseintemperatureleadstoincreaseinreactionvelocityandanincreasein conversion(isomersyieldandhenceRON).However,onceequilibriumisreached,increasing temperatureincreasesbackwardreactionvelocityandreducesconversion.Thisisreflectedin Figure 11A, which shows the effect of lead reactor temperature on RON. Actually, i-pentane and 2,2-DMB are the components with the greatest effect on RON of isomerate. Therefore, the effect oftemperatureon(I-C5/C5)%and(2,2-DMB/C6)areidenticalwiththatofRON(Figure11B and 11C). Although, the isomerization of 2,3-DMB is still far from equilibrium (Figure 11D), but ithaslittleeffectonRON.Inaddition,increasingtemperatureincreasestherateofother reactions.Increasedhydrocrackingandringopeningreactionsleadstoincreasedhydrogen consumption(Figure11E).Finally,theincreaseinhydrocrakingleadstothedecreaseofof Page 16 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 17 ` isomerate yield (Figure 11F). Isomerate wt.% was defined in this study as the wt% of isomerate to the total reactor effluent. INSERT FIGURE 11 Lag Reactor Inlet Temperature. For the lag reactor, the effect of the inlet temperature is similar in trend to that of the lead reactor (Figures 12 A-F). INSERT FIGURE 12 Hydrogentohydrocarbonmoleratio.Hydrogenisrequiredforcompletingthe reactionsanddecreasingcokedepositiononcatalystsurface.Generally,increasinginlet hydrogenleadstohigherhydrogenpartialpressureinsidethereactorwhichincreases hydrocracking.Consequently,isomerateyield(Figure13A)andtheRONofisomerate(Figure 13B) decrease due to increased hydrocracking of paraffins isomers (lower paraffin Isomerization Number(PIN=i-C5/C5+(2,2-DMB+2,3-DMB)/C6))(Figure13C)).Therefore,thereactors should be operated at the lowest possible hydrogen to Hydrocarbon (H2:HC) ratio. However, at anyinstance,theH2:HCratioshouldnotbelowerthan0.05moleH2/moleHydrocarbonas claimedbyprocesslicensor.TheH2:HCmoleratioisusuallymeasuredatreactoroutletfor Hydrogen Once Through Penex units. Page 17 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 18 ` INSERT FIGURE 13 Feedrate(LHSV).Atthesamereactorinlettemperature,thelowertheliquidhourly space velocity (LHSV), the higher the PIN (Figure 14A). On the contrary, the higher the LHSV, the higher the yield of isomerate due to lower hydrocracking (Figure 14B). This is expected since increasing feed rate decreases contact time with catalyst, which results in lower reactions rates. INSERT FIGURE 14 Feedcomposition.A)Methylcyclopentaneandcyclohexane:Generally,cyclic compounds adsorb on catalyst surface reducing active sites available for other reactants. Hence, an increase in methylcyclopentane (MCP) and cyclohexane (CH) % causes slightly lower isomer ratiosinisomerate(Figures15Aand15B).Ontheotherhand,theRONofisomeratewillbe higher due to the increase of MCP or CH in isomerate, which has a relative high RON (Figures 15C and 15D). This contradicts what was reported by Ahari et al.11. They investigated the effects of MCP in isomerization feed and claimed that increasing the amount of MCP in feed results in reduction of isomerateRON. However, theauthors carried outthat study usinga mixture ofn-hexane and MCP as a feed on a different catalyst operating at higher temperature. Therefore, at theseconditions,thenaphtheneisomerizationreactionisdrivenstronglyinthedirectionof formingCH,whichhasarelativelylowerRONcomparedtoMCPandmulti-branchedhexane isomers.Inaddition,thehigheroperatingtemperatureleadstoincreasedringopeningofMCP Page 18 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 19 ` andCH,whichleadtolowerRON.Accordingly,thedifferencesinfeedcompositionand operating conditions may be the reason for the contradicting results. IsomerateyieldincreaseswiththeincreaseofMCPorCHcontentinfeed.Bothcyclo componentsdonotundergohydrocrackingasparaffins,howeverringopeningreactionsmay occur.Generally,ringopeningreactionisfavoredslightlycomparedtohydrocrackingwith increasingtemperature,becauseofitshigheractivationenergy(e.g.MCPringopeninghasa higher activation energy compared to C6 paraffins hydrocracking). Moreover, reduction of active sitesavailableforparaffinsreducesparaffinshydrocracking,andhenceincreasesliquidyield (Figures 15E and 15F).Methylcyclopentaneandcyclohexanearealsomajorhydrogenconsumersthroughring opening.Therefore,hydrogenconsumptionincreasesastheamountofMCPorCHinthefeed increases. It should be notedthat the overall hydrogen consumptionmay notincrease(Figures 15G and 15H) due to that the increase of MCP or CH content in feed is on the account of other componentsincludingbenzenethatisthemajorhydrogenconsumer.However,thehydrogen consumptioninthelagreactorshowsanincreasewiththeincreaseinMCPorCHcontent (Figures 15I and 15J). INSERT FIGURE 15 B)Benzene:Figure16Ashowstheactualvariationofbenzenecontentinthefeed.As mentionedearlier,benzeneishydrogenatedcompletelytoCHintheleadreactor.Benzene saturationishighlyexothermicreactionthatisunfavorablebytheequilibriumlimited Page 19 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 20 ` isomerization.Figure16Bshowstheeffectoffeedbenzenecontentontheleadreactor temperaturerise.Smallchangesinfeedbenzenecontentgreatlyaffecttheleadreactor temperature rise. However, this does not affect isomerate RON much (Figure 16C) since benzene isconvertedtoCHthathasamoderateRONormayundergoisomerizationtoformMCPthat has higher RON.Ontheotherhand,theexothermicheatofbenzenesaturationaffectsPINsignificantly (Figure 16D). In actual operation, the lead reactor inlet temperature is always varied in order to maintainaconstanttemperaturerisethroughthereactorbed.Theallowabletemperatureriseis increasedwithfeedrateinordertoobtainareasonableconversionofnormalparaffins.This operating scheme may be effective but it is very tedious, especially when wide variations in feed benzene content occur.Moreover,theincreaseinfeedbenzenecontentleadstoanincreaseinisomerateyield (Figure 16E) since the produced CH or MCP undergo slow ring opening and adsorb on catalyst surfacereducingactivesitesavailable forpraffinshydrocracking.Finally,benzenesaturationis themajorhydrogenconsumingreactioninPenexprocess.Figure16Fshowstheincreasein hydrogen consumption with increase of benzene in feed. INSERT FIGURE 16 6.Model application to process optimization From previous analysis, it can be concludedthat feed composition and feed rate arethe dominant process variables. Benzene composition may be the most important variable. As seen Page 20 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 21 ` inFigure16D,atconstantinlettemperaturetotheleadreactor,thePINisgreatlyaffectedby variationoffeedbenzenecontent.Thus,continuousmonitoringoffeedbenzenecontentis alwaysrequired.Although,feedbenzenecontentdoesnotchangewidely(Figure16A),evena small change in benzene has a great impact on isomerization degree. Refiners offset this impact by continuous variation of lead reactor inlet temperature so as to obtain a fixed temperature rise through lead reactor. The amount of allowable temperature rise is increased as feed rate increases toincreasereactionsrates.Thisoperatingtechniquemaybeeffective,howeverifthefeed witnesses frequent variations in benzene level, this operating scheme will be nearly impossible. For more effective operation and more profits, some refiners developed rigorous models for real time optimization of penex process9. This shows the necessity of developing rigorous models in modern refineries. AnotherimportantvariableistheH2:HCmoleratioinsidethereactor.Asindicated before,hydrogenisrequiredforcompletingisomerizationandreducingcokelay-downon catalyst.However,anincreaseinhydrogenpartialpressureinsidethereactorsleadstoan increaseinhydrocrackingandreductionofisomerateyield.Therefore,thereactorsshouldbe operatedwiththelowestpossiblehydrogenpartialpressure.However,asaruleofthumbto avoidcokedeposition,thereactorsshouldnotbeoperatedatH2:H.Cratiolessthan0.05 measured at lag reactor outlet at any time. Refinersfacetwooperationalscenariosinoptimizingrefineryprocesses.Thefirst scenariois"what-if"scenariowheretherefinerswanttopredicttheprocessperformanceifa change occurred to one or more of the process variables. This scenario has been covered in the previousanalysisofoperatingvariables.Theotherscenarioisthe"how-to"scenario.Refiners Page 21 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 22 ` usuallyaimatmaximizingprofits.Hence,thefrequentquestioninrefiners'mindsis"Howto increase profit?" Inordertooptimizetheprocess,thenotesobtainedfromthefirstscenarioshouldbe implemented.Table8presentsthebaseoperatingconditionsoftheinvestigatedindustrialunit andthelimitboundsinducedbyprocesslicensor.Itisclearthatthereactorsareoperatedwith H2:HCratiomuchhigherthanrequired.Then,thefirststepwasreducinghydrogenpartial pressuretothelowestpossiblevaluewiththesamereactors'inlettemperature.Actually,the lowestpossibleH2:HCratioatthereactors'outletwas0.0865whichisstillfarfromthe allowable minimum. INSERT TABLE 8 LowerH2:HCratioscouldbeobtainedbuteitheratlowerleadreactortemperatureor higher lag reactor inlet temperature because the amount of make-up hydrogen used affects heat transfer coefficients in both the hot and cold combined exchangers and consequently affects the outlettemperatures.Havingobtainedthelowestpossiblehydrogenpartialpressureinreactors, thenextstepwasvaryingtheleadandlagreactorinlettemperaturesimultaneouslysoasto obtain the optimum operating point. Figures 17, 18 and 19 show the obtained results. From Figure 17, the lower the reactors inlet temperatures, the higher the isomerate yield. Therefore,itisalwaystherefiner'sdecisiontoraisethereactorstemperaturetoincreasethe isomerizationdegree(Figure18)toobtainahigherRON(Figure19)withthesacrificeof Page 22 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 23 ` isomerateyield.Table9showsacomparisonbetweenplantyieldsatbasepointandoptimum operating point. INSERT FIGURE 17 INSERT FIGURE 18 INSERT FIGURE 19 INSERT TABLE 9 Results shown in Table 9 indicate that about 210 kg/hr can be added to the total regular isomerateyieldbyfollowingtheoptimizationscheme.Savingsinisomerateyieldaremainly attributedtodecreasedhydrocrackingbyreducinghydrogenpartialpressureinsidereactors.In addition,itisshownthatthereactorchargeheaterdutyhasdeclinedduetoreducedinlet temperatureofleadreactorandreducedloadonheaterbyeliminatingapartofmake-up hydrogenthatwasusedoriginally.However,thedecreaseinreactorcharge-heaterdutyis overcome by the increase in stabilizer bottom reboiler, resulting in apparent energy deficiency of about96,635kJ/hr.However,followingtheoptimizedoperatingschemewillnotonlyimprove isomerate yield and RON, but it will also reduce the consumption of make-up gas. Make-up gas fromnaphthareformermaybedirectedtofuelgassystemintherefinery.Thenewoperating schemewilladdabout161kg/hrofmake-upgastothefuelgassystem.Thiswillprovide approximately 11.3 x 106 kJ/hr. This may not only cover the increase in stabilizer bottom reboiler duty, but also it may cover the total energy consumption of the fired heaters in the Penex unit. It should be noted here that savings in energy and product yield are proportional to plant capacity. Page 23 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 24 ` Finally,themodelmaybeusedtopredicttheplantperformanceatvariousoperating points.Themodelisextremelyusefulforpredictingsuitableoperatingconditions,especially whenplantexperiencesvariationsinfeedbenzenecontent.Thiscanbeachievedby incorporatingthemodelintoarealtimeoptimization(RTO)scheme.Inthiscase,itis recommended that the model be finely tuned to match plant performance closely. 7.Conclusions AprocessmodelforanindustrialPenexprocesswasdevelopedusingAspenHYSYS Petroleum Refining isomerization reactor module. The model could track the plant performance satisfactorily. In addition, the model was used for studying the effect of each process variable on process performance. Among all process variables, benzene feed content and H2:HC ratio were themostprominentfactorstoaffecttheprocessperformance.Finally,themodelwasusedforoptimizingprocessatsteadystateconditions.Resultsobtainedfromthemodelshowedthat considerablesavingsinproductyieldcanbeachievedbyloweringhydrogenpartialpressure inside reactors to the lowest possible practical value. The surplus in make-up gas may be directed tofuelgasresultinginsignificantfuelsavings.Themodelmayalsobeincorporatedinareal time optimization (RTO) scheme. In this case, the model should be finely tuned to match plant performance closely. NomenclatureA6 Benzene C1 Methane Page 24 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 25 ` C2 Ethane C3 Propane i-C4Iso butane n-C4 Normal butane i-C5Iso pentane n-C5Normal pentane n-C6Normal hexane n-C7 Normal heptane CHCyclo hexane 2,2-DMB 2,2 Dimethyl butane 2,3-DMB 2,3 Dimethyl butane H factorifactor of component i MCH Methyl cyclo hexane MCP Methyl cyclopentaneMBP7 Multi branched heptanes 2MP2-Methyl pentane 3MP3-Methyl pentane 5N5Cyclo pentane 5N7Five ring, seven carbon naphthene Page 25 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 26 ` 6N7Six ring, seven carbon naphthene NP7Normal heptane SBP7Single branched heptanes Temperature rise in reactor model Temperature rise in plant reactor Property of a petroleum fraction Property of a pure component i Mole , volume or mass fraction of component i Known property of C7+ fraction Weighting factor for component i Mass fraction of component i in model outlet stream Mass fraction of component i in plant reactor outlet stream PINParaffin Isomerization Number = i-C5/C5 + (2,2-DMB+2,3-DMB)/C6 References (1) Moulijn, J. A.; Makkee, M.; van Diepen, A. E. Chemical Process Technology; 2nd ed.; Wiley, 2013. (2) Deak, V. G.; Rosin, R. R.; Sullivan, D. K. In AICHE 2008 Spring National Meeting; New Orleans, LA, USA, 2008. (3) Laredo, G. C.; Castillo, J.; Cano, J. L. Fuel 2014, 135, 459467. Page 26 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 27 ` (4) Smith, J. M.; van Ness, H. C.; Abbott, M. M. Introduction to Chemical Engineering Thermodynamics; 6th ed.; McGraw-Hill, 2001. (5) Yasakova, E. A.; Sitdikova, A. V; Achmetov, A. F. Oil Gas Bus. 2010. (6) Meyers, R. A. Handbook of Petroleum Refining Processes; 3rd ed.; McGraw-Hill, 2004. (7) Dean, L. E.; Harris, H. R.; Belden, D. H.; Haensel, V. Platin. Met. Rev. 1959, 3, 911. (8) Chekantsev, N. V.; Gyngazova, M. S.; Ivanchina, E. D. Chem. Eng. J. 2014, 238, 120128. (9) Besl, H.; Kossman, W.; Crowe, T. J.; Caracotsios, M. Oil Gas J. 1998, 96, 6164. (10) Dudley, R. E.; Malloy, J. B. Ind. Eng. Chem. Process Des. Dev. 1963, 2, 239244. (11) Ahari, J. S.; Ahmadpanah, S. J.; Khaleghinasab, A.; Kakavand, M. Pet. Coal 2005, 47, 2631. (12) Ahari, J. S.; Khorsand, K.; Hosseini, A. A.; Farshi, A. Pet. Coal 2006, 48, 4250. (13) Brito, K. D.; Sousa, B. V.; Rodrigues, M. G. F.; Alves, J. J. N. Brazilian J. Pet. Gas 2008, 2, 18. (14) Koncsag, C. I.; Tutun, I. A.; SAFTA, C. Ovidius Univ. Ann. Chem. 2011, 22, 102106. (15) Surla, K.; Guillaume, D.; Verstraete, J. J.; Galtier, P. Oil Gas Sci. Technol. 2011, 66, 343365. (16) Ghosh, P.; Hickey, K. J.; Jaffe, S. B. Ind. Eng. Chem. Res. 2006, 45, 337345. (17) Perdih, A.; Perdih, F. Acta Chim. Slov. 2006, 53, 306315. (18) Riazi, M. R. Characterization and Properties of Petroleum Fractions; 1st ed.; ASTM International, 2005. (19) Mahdavian, M.; Fatemi, S.; Fazeli, A. Int. J. Chem. React. Eng. 2010, 8, A8. (20) Taskar, U.; Riggs, J. B. AIChE J. 1997, 43, 740753. (21) Pashikanti, K.; Liu, Y. A. Energy & Fuels 2011, 25, 53205344. (22) Russell, R. A. Chem. Eng. - New York 1983, 90, 5359. (23) Kaes, G. L. Refinery Process Modeling: A Practical Guide to Steady State Modeling of Petroleum Processes; 1st ed.; Athens Printing Company: Athens, GA, USA, 2000.Page 27 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 28 ` Tables Table 1: RON of normal and iso paraffins16,17 Component RON n-pentane 62 i-pentane 92 2-Methyl Pentane(2MP) 73.4 3-Methyl Pentane (3MP) 74.5 2,2 dimethyl butane (2,2-DMB) 91.8 2,3 dimethyl butane (2,3-DMB) 105.8 n-hexane 24.8 Page 28 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 29 ` Table 2: PNA composition of C7+ fraction Hydrocarbon classVol.% Paraffins 74.24 Naphthenes 23.53 Aromatics 2.22 Page 29 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 30 ` Table 3: Available and calculated Properties of C7+ fraction in feed and product streams PropertyC7+ (Feed)C7+ (Product) GivenCalc.GivenCalc. SG0.69150.70420.6830.709 MW100.19899.666100.19899.70 RVP (psi)1.971.972.12.187 RON55558282 Page 30 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 31 ` Table 4: Estimated composition of C7+ fraction in feed and product streams CompnentFeed (wt.%)Product (wt.%) MBP7048.2 SBP752.727.3 N-C721.2 0 5N710.112.2 6N71612.3 Page 31 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 32 ` Table 5: Adjustment factors used for reactor model calibrationParameter Range of deviation from the base Global activity0.1-1 Isomerization activity0.1-1.1 Hydrocracking activity0.1-1.1 Hydrogenation activity0.1-1.1 Ring opening activity0.1-1.1 Heavy activity0.1-1.1 Page 32 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 33 ` Table 6: Terms included in objective function for reactor model calibration and applied weighing factors TermApplied weighing factor Temperature Rise (C)1 I-Pentane (wt%)5 N-Pentane (wt%)5 Cyclo Pentane (wt%)2 2 Methyl Pentane (wt%)5 3 Methyl Pentane (wt%)5 2,2 Di-Methyl Butane (wt%)5 2,2 Di-Methyl Butane (wt%)5 N-Hexane (wt%)5 Methyl Cyclo Pentane (wt%)2 Benzene (wt%)2 Cyclo Hexane (wt%)2 C7+ (wt%)2 Page 33 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 34 ` Table 7: Estimated average activity parameters for lead and lag reactors ParameterLead ReactorLag reactor Global activity0.9122.402 Isomerization activity0.8271.092 Hydrocracking activity1.0230.968 Hydrogenation activity0.9390.995 Ring opening activity1.0541.042 Heavy activity1.071.059 Page 34 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 35 ` Table 8: Operating conditions of the unit at the base operating point and limit bounds induced by process licensor Process VariableBase Operating PointBounds H2:HC ratio0.1565min. 0.05 Lead Reactor Inlet T124 C105 - 204C Lag Reactor Inlet T120C105 - 204C Page 35 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 36 ` Table 9: Process performance at base and optimum operating point Lead Reactor Inlet T (C) Lag Reactor Inlet T (C) Make-up Hydrogen Flow (kg/hr) Reactor Charge Heater Duty (kJ/hr) Stabilizer Reboiler Duty (kJ/hr) Isomerate Yield (kg/hr) PIN Base Operating Point 124120614.21,018,2759,452,82931,6931.18 Optimum Operating Point 116117453721,6459,846,09431,9031.184 Page 36 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 37 ` Figures Figure 1: Effect of temperature on isomers yield5 Page 37 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 38 ` Figure 2: Comparison of the activity of different isomerization catalysts5 Page 38 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 39 ` Figure 3: UOP Penex process6 Page 39 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 40 ` Figure 4: Isomerization model reaction network Page 40 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 41 ` Figure 5: Variation of C7+ wt% in the feed 012345625-Feb 16-Apr 5-Jun 25-Jul 13-SepC7+ wt.%Page 41 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 42 ` Figure 6: Plant versus model yields with model calibration data sets forlead reactor 05101520253035400 5 10 15 20 25 30 35 40Model (wt%)Plant (wt.%)i-C-5%n-C55N52,2DMB2,3DMB2MP3MPN-C65N6Page 42 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 43 ` Figure 7: Plant versus model yields with model calibration data sets for lag reactor 0510152025303540450 5 10 15 20 25 30 35 40 45Model (wt%)Plant (wt.%)i-C-5%n-C55N52,2DMB2,3DMB2MP3MPN-C65N6Page 43 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 44 ` Figure 8: Plant versus model yields for the 4 months after calibration for lead reactor 05101520253035400 5 10 15 20 25 30 35 40Model (wt%)Plant (wt%)i-C-5%n-C55N52,2DMB2,3DMB2MP3MPN-C65N6Page 44 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 45 ` Figure 9: Plant versus model yields for the 4 months after calibration for lag reactor 05101520253035400 5 10 15 20 25 30 35 40Model (wt%)Plant (wt%)i-C-5%n-C55N52,2DMB2,3DMB2MP3MPN-C65N6Page 45 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 46 ` Figure 10: Penex isomerization unit model Page 46 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 47 `

Figure 11: A- Effect of Lead reactor inlet temperature on RON; B- Variation of (I-C5/C5)% withleadreactorinlettemperature;C-Variationof(2,2DMB/C6)%withleadreactorinlet temperature; D- Variation of (2,3DMB/C6)% with lead reactor inlet temperature; E- Effect of leadinletreactortemperatureonhydrogenconsumptioninleadreactor;F-Effectoflead reactor inlet temperature on isomerate yield. Lag reactor inlet temperature = 120 C, H2:HC = 0.1565, A6 = 2.99 wt.%, LHSV = 1.15 hr-1 84.0684.0784.0884.0984.184.1184.1284.1384.1484.15105 110 115 120 125RONLead Reactor Inlet Temp. C(A)72.672.87373.273.473.673.874105 110 115 120 125(I-C5/C5)%Lead Reactor Inlet Temp. C(B)25.82626.226.426.626.82727.227.427.627.8105 110 115 120 125(2,2DMB/C6)%Lead Reactor Inlet Temp. C(C)8.38.358.48.458.58.55105 110 115 120 125(2,3DMB/C6)%Lead Reactor Inlet Temp. C(D)892894896898900902904906908105 110 115 120 125Hydrogen consumption (STD_m3/hr)Lead Reactor Inlet Temp. C(E)95.1895.1995.295.2195.2295.2395.2495.2595.2695.2795.28105 110 115 120 125Isomerate wt.%Lead Reactor Inlet Temp. C(F)Page 47 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 48 ` Figure12:A-Effect of Lag reactor inlet temperature onRON; B-Variation of (I-C5/C5)% withlagreactorinlettemperature;C-Variationof(2,2DMB/C6)%withlagreactorinlet temperature; D-Variation of (2,3DMB/C6)% with lag reactor inlet temperature; E-Effect of leadinletreactortemperatureonhydrogenconsumptioninlagreactor;F-Effectoflag reactor inlet temperature on isomerate yield. Lead reactor inlet temperature = 124 C, H2:HC = 0.1565, A6= 2.99 wt%, LHSV = 1.15 hr-1 84.0284.0484.0684.0884.184.1284.14106 111 116 121 126RONLag Reactor Inlet Temp. (C)(A)76.2576.376.3576.476.4576.576.55106 111 116 121 126(I-C5/C5)%Lag Reactor Inlet Temp. (C)(B)32.332.432.532.632.732.832.9106 111 116 121 126(2,2DMB/C6)%Lag Reactor Inlet Temp. (C)(C)8.488.58.528.548.568.588.68.628.64106 111 116 121 126(2,3DMB/C6)%Lag Reactor Inlet Temp. (C)(D)00.511.522.533.54106 111 116 121 126Hydrogen consumption (STD_m3/hr)Lag Reactor Inlet Temp. (C)(E)95.17595.1895.18595.1995.19595.295.20595.2195.21595.22110 115 120 125Isomerate wt.%Lag Reactor Inlet Temp. (C)(F)Page 48 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 49 `

Figure 13: A- Effect of H2:HC ratio on isomerate yield; B- Effect of H2:HC ratio on RON of isomerate;C-EffectofH2:HCratioonPIN.Leadreactorinlettemperature=124C,lag reactor inlet temperature = 120 C, A6 = 2.99 wt%, LHSV = 1.15 hr-1 9595.295.495.695.89696.296.40.07 0.09 0.11 0.13 0.15 0.17Isomerate wt.%H2:HC (A)84.1284.1384.1484.1584.1684.1784.1884.1984.284.2184.220.07 0.09 0.11 0.13 0.15 0.17RONH2:HC(B)1.17951.181.18051.1811.18151.1821.18251.1830.08 0.1 0.12 0.14 0.16PINH2:HC (C)Page 49 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 50 ` Figure14:A-Effectoffeedrate(LHSV)onPIN;B-Effectoffeedrate(LHSV)on isomerate yield. Lag reactor inlet temperature = 120 C, H2:HC = 0.1565, A6 = 2.99 wt% 1.051.061.071.081.091.11.11105 110 115 120 125PINLead Reactor inlet T1.03 hr-1 1.15 hr-1 1.3 hr-1(A)95.1495.1695.1895.295.2295.2495.2695.2895.3105 110 115 120 125Isomerate wt.%Lead Reactor T1.03hr-1 1.15hr-1 1.3 hr-1(B)Page 50 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 51 ` 1.1791.17951.181.18051.1810 2 4 6 8PINFeed MCP content, wt% (base 4.08%)(A)1.179751.17981.179851.17991.179951.181.180051.18011.180150 0.5 1 1.5 2PINFeed CH content, wt% (base 1.18%)(B)83.958484.0584.184.1584.284.2584.30 2 4 6 8RONFeed MCP content, wt% (base 4.08%)(C)84.0984.184.1184.1284.1384.1484.1584.1684.1784.180 0.5 1 1.5 2RONFeed CH content, wt% (base 1.18%)(D)94.99595.195.295.395.495.50 2 4 6 8Isomerate wt.%Feed MCP content, wt% (base 4.08%)(E)95.195.1295.1495.1695.1895.295.2295.2495.260 0.5 1 1.5 2Isomerate wt.%Feed CH content, wt% (base 1.18%)(F)Page 51 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 52 `

Figure 15: A- Effect of feed MCP content on PIN; B- Effect of feed CH content on PIN; C- EffectoffeedMCPcontentonRON;D-EffectoffeedCHcontentonRON;E-Effectof feedMCPcontentonisomerateyield;F-EffectoffeedCHcontentonisomerateyield;G- Effect of feed MCP content on total hydrogen consumption; H- Effect of feed CH content on total hydrogen consumption; I:- Effect of feed MCP content on hydrogen consumption in the lagreactor;J-EffectoffeedCHcontentonhydrogenconsumptioninthelagreactor.Lead reactor inlet temperature = 124 C, lag reactor Inlet temperature = 120C, LHSV = 1.15hr-1 8908959009059109159209259300 2 4 6 8 Total Hydrogen Consumption(STD_m3/hr)Feed MCP content, wt% (base 4.08%) (G)9029049069089109129140 0.5 1 1.5 2Total Hydrogen Consumption (STD_m3/hr)Feed CH content, wt% (base 1.18%)(H)22.22.42.62.833.20 2 4 6 8Hydrogen Consumption in lag reactor (STD_m3/hr)Feed MCP content, wt%(base 4.08%)(I)2.62.652.72.752.82.850 0.5 1 1.5 2Hydrogen Consumption in lag reactor (STD_m3/hr)Feed CH content, wt% (base 1.18%)(J)Page 52 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 53 `

Figure16:A-Variation of feed benzene content; B-Effect of feed benzene content on lead reactortemperature rise; C-Effect of feedbenzene contenton isomerate RON;D-Effect of FeedbenzenecontentonPINintheleadreactor;E-Effectoffeedbenzenecontenton isomerate yield; F- Effect of feed benzene content on hydrogen consumption in lead reactor. Lead reactor inlet temperature = 124 C, lag reactor inlet temperature = 120C, LHSV = 1.15 hr-1 1.01.52.02.53.03.54.04.55.026-Jan 17-Mar 6-May 25-Jun 14-Aug 3-OctA6 wt%(A)0510152025303540450 2 4 6Temperature Rise in lead reactorFeed benzene content, wt% (base 2.99%)(B)83.8583.983.958484.0584.184.1584.284.250 2 4 6RONFeedbenzene, wt% (base 2,99%)(C)1.061.0651.071.0751.081.0851.091.0951.10 2 4 6PINFeed benzene content, wt% (base 2.99%)(D)94.494.694.89595.295.495.695.8960 2 4 6Isomerate wt%Feed benzene content, wt% (base 2.99%)(E)020040060080010001200140016000 2 4 6Hydrogen Consumption in lead reactor (STD_m3/hr)Feed benzene content, wt% (base 2.99%)(F)Page 53 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 54 ` Figure 17: Variation of isomerate yield with reactors inlet temperatures 11211411611812012212496.3796.3996.4196.4396.4596.4796.49117115113111109107Lag reactor T, oCIsomerate wt %Lead Reactor T, oC96.49-96.596.47-96.4996.45-96.4796.43-96.4596.41-96.4396.39-96.4196.37-96.39Page 54 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 55 ` Figure 18: Variation of Paraffin Isomerization Number with reactors inlet temperatures 1121151181211241.171.1721.1741.1761.1781.181.1821.1841.186117115113111109107Lag Reactor Inlet T, oCPINLead Reactor Inlet T, oC1.184-1.1861.182-1.1841.18-1.1821.178-1.181.176-1.1781.174-1.1761.172-1.1741.17-1.172Page 55 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960 56 ` Figure 19: Variation of isomerate RON with reactors inlet temperatures 11211411611812012212484.0884.184.1284.1484.1684.1884.284.2284.24117114111108Lag reactor inlet T, oCRONLead reactor inlet T, oC84.24-84.2584.22-84.2484.2-84.2284.18-84.284.16-84.1884.14-84.1684.12-84.1484.1-84.1284.08-84.1Page 56 of 56ACS Paragon Plus EnvironmentEnergy & Fuels123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960