Maleic Anhydride from Mixtures of n-Butenes and n-Butane:Effective Reaction Kinetics

W. M. Brandsta1dter and B. Kraushaar-Czarnetzki*

CVT - Institute of Chemical Process Engineering, University of Karlsruhe (TH), Kaiserstrasse 12,76128 Karlsruhe, Germany

The conversion of mixtures of n-butane and n-butenes to maleic anhydride over an industrialvanadium-phosphorus oxide (VPO) catalyst was experimentally investigated. Specifically, theconversion of a synthetic raffinate II mixture was studied. The reaction kinetics were modeledon the basis of a reaction network consisting of eight individual reactions and by using hyperbolicrate equations. Calculations and experimental results are shown for a temperature range from400 to 450 C and for molar fractions of the synthetic raffinate II in the feed that were variedfrom 1% to 2%. The kinetic model shows good agreement with the experimental results. Thevalue of the maximum yield of maleic anhydride is insensitive to changes in temperature andhydrocarbon feed fractions and amounts to 48%. Increases in temperature, however, reducethe amount of catalyst required to reach the maximum yield. The conversions of pure 1-buteneand pure n-butane lead to maximum yields of 53% and 56%, respectively. The efficiency of amaleic anhydride process will, therefore, strongly depend on the choice of feedstock and, inparticular, on the availability of n-butane at the designated production site.

1. Introduction

Maleic anhydride is one of the intermediate productswith the highest forecasted increase in demand in thenear future.1 Nowadays, it is predominantly producedthrough the partial selective oxidation of n-butane.

Processes based on n-butane/n-butene mixtures haveno importance at present; howewer, they may becomean economic alternative as raw materials in the future.Mixtures of C4 hydrocarbons are inevitably coproducedupon the manufacturing of ethylene and propylene insteamcracker plants. Hence, C4 production capacitiesare directly coupled to the production capacities ofethylene and propylene, which are steadily increasing.The C4 components of the steamcracker effluent aren-butene isomers, n-butane, isobutane, isobutene, andbutadiene. Isobutene can be removed selectively bymeans of etherification to methyl tert-butyl ether (MTBE)or ethyl tert-butyl ether (ETBE). The separation of theother components is costly because their boiling pointsare very close. Various strategies have been developedto increase revenues by recovering certain componentsof the stream by physical or chemical means or byrecycling the hydrogenated stream to the feed of thecracking units to reduce the amount of fresh feedrequired.2 Other possibilities, such as hydroformylation,telomerization, or metathesis of parts of the C4 streams,are presented in ref 3.

Here, we consider a fixed-bed maleic anhydrideprocess in which all linear C4 hydrocarbons except fordienes and acetylenes are converted in one stage. TheC4 cut of the steamcracker effluent after etherificationof isobutene and extraction of butadiene, denoted asraffinate II, typically contains 16% mol/mol of n-butane, 75% mol/mol of n-butenes, and 6% mol/mol of

isobutane, as well as small amounts of residual isobuteneand isopentenes.4,5 The utilization of raffinate II couldlead to a better valorization of C4 steamcracker streamsand would cope with the growing demand for maleicanhydride. By using butadienesalso recovered from theC4 streamsmaleic anhydride could be further processedto plasticizers by means of the Diels-Alder reaction3at the same site.

The investigation of a process based on mixed C4hydrocarbons has interesting scientific aspects as well.The conversion behaviors of n-butenes and n-butane aredifferent with respect to reaction rates, intermediates,and selectivities, resulting in different overall reactorperformances. In addition, the composition of industrialC4 mixtures may vary depending on the origin of thefeedstock and the preceding processes. Therefore, it isnecessary to consider the reactions of n-butane andn-butenes individually to obtain a kinetic model thatcan cope with different kinds of mixtures.

Here, we report on the development of a kinetic modelfor the conversion of n-butane and n-butenes overindustrially developed vanadium-phosphorus oxide(VPO) catalysts.

For the oxidation of pure n-butane, the well-knowntriangular network, considering the species n-butane,maleic anhydride, and carbon oxides, is widely used. Theproposed rate equations, however, differ much fromauthor to author: pseudo first-order reaction rates wereused in refs 6 and 7, power rate laws were used in ref8, combinations of hyperbolic and power law rateequations are presented in refs 9-11, Langmuir-Hinshelwood and Eley-Rideal approaches were used inrefs 8, 11, and 12, and redox mechanisms (also calledMars-van Krevelen mechanisms) were used in refs 13and 14. An overview of the kinetic models that useseparate n-butane oxidation and catalyst reoxidationequations are presented in ref 15. In a comparison ofthe various models, Becker12 and Uihlein11 found thatthe Eley-Rideal approach is the one that is best suited.

* To whom correspondence should be addressed. Tel.:49-721-608-3947. Fax: 49-721-608-6118. E-mail:kraushaar@cvt.uka.de.

5550 Ind. Eng. Chem. Res. 2005, 44, 5550-5559

10.1021/ie050099o CCC: $30.25 2005 American Chemical SocietyPublished on Web 06/14/2005

The resulting rate equations are of the type

The reaction rate for the conversion of hydrocarbon i tohydrocarbon j shows a dependency on the partialpressure of hydrocarbon i (pi), the partial pressure ofoxygen (pO2), and the partial pressures of n-butane (pbta)and water (pH2O) as inhibiting species.

Concerning the conversion of pure n-butenes, a reac-tion network as depicted in Figure 1 has been proposedin refs 7 and 16. Pseudo-first-order rate equations wereused in these works. In ref 17, a Langmuir-Hinshel-wood type model was successfully used in the oxidationof 1-butene and 1,3-butadiene for low conversion levels.None of these attempts led to a quantitative descriptionof the catalyst behavior at different temperatures andfeed compositions in the interesting range of industrialapplication (high conversion levels).

Except for ref 7, only one article was found about theconversion of raffinate II: In ref 5, the commercialBayer process for the production of maleic anhydride ispresented, but no quantitative reaction kinetics isshown. Quast7 proposed a reaction network by super-position of the networks for the pure hydrocarbons asdisplayed in Figure 2. However, only starting experi-ments were conducted, and the network has not yetbeen investigated enough.

In the present work, it was necessary to validate theproposed reaction network for the conversion of themixtures or to find a suitable replacement and toquantify the reaction kinetics. The quantification of thereaction kinetics also requires the proposal of the typeof individual rate equations. For the latter, it was alsonecessary to study the conversion of the pure n-butenes.Finally, the aim was to determine the quantities of thekinetic parameters.

2. Experimental Methods and Data Evaluation

The catalyst was supplied by industry and is calledVPO-50 throughout the article. It consisted of activeVPO shells on steatite carriers with a weight fractionof 50%. The catalyst bodies had the shape of hollowcylinders with a height of 4 mm, an outer diameter of 5

mm, and an inner diameter of slightly >1 mm. Thecatalyst was pretreated by a well defined start-upprocedure lasting several days. The kinetic measure-ments were carried out afterward. The stability of thecatalyst was checked by repeated measurement underdefined conditions in regular intervals. The reactor wasfilled with fresh catalyst as soon as the repeatedvalidation measurements showed significant deviationsfrom the reference.

The experimental setup for the kinetic analysis isdisplayed in Figure 3. The fixed-bed reaction zone hada height of 1.5 m and a diameter of 15 mm consistingof catalyst bodies mixed with nonreactive steatitespheres with a diameter of 2-3 mm.

The temperature in the catalyst bed was controlledby six independent heating zones along the reactor andwas monitored by 13 axially distributed thermocouples.By using different set points for the six heating zones,it was possible to operate the reactor in a well-definednon-isothermal mode. Experiments that were carriedout non-isothermally and their results are reported inref 18. The results presented here stem from isothermaloperation only. The deviations of the measured temper-atures from the desired temperature were

based on an experimental observation of the conversionsof the pure components and the conversion of themixture: The conversion of n-butane was stronglyinhibited by the presence of n-butenes and 1,3-butadi-ene. It was, therefore, assumed that n-butane had notbeen converted until no more 1,3-butadiene was presentin the gas phase. The concentration of water wascalculated from the oxygen balance and was cross-checked by solving the hydrogen balance.

In the evaluation of the experiments, the amount ofthe individual carbon-containing species in the gasphase was expressed in terms of normalized dimension-less concentrations of species i according to

where the molar flow of species i is related to the molarflow of hydrocarbons at the reactor inlet. By consideringthe amount of carbon atoms in each species i (C,i), thesum of all normalized dimensionless concentrations ata certain position equals unity because of the elementconservation of carbon. The values of the individualdimensionless concentrations range between zero andone. In the case of the reaction products, Yi alsorepresents the yield of this species. A second dimension-less concentration definition was used for the oxygencontaining species,

with the same properties as Yi, only related to oxygen.The conversion level and the reactor selectivities weredefined as

respectively. The concentration profiles were mostlyplotted as functions of the modified residence time tmodwhich equals the ratio of the amount of active mass tothe actual volume flow through the reactor.

Most of the experiments were carried out with a feedapproximating the typical composition of raffinate II.4,5We used 16% mol/mol of n-butane. The linear buteneswere represented by 75% mol/mol of 1-butene. As a

substitute for the branched hydrocarbons in raffinateII, the laboratory feed contained 9% of nitrogen. Air,1-butene 2.0, and n-butane 2.5 (both from MesserGriesheim) were fed individually to the reactor withwell-defined flow rates. The total flow rate through thereactor was varied in the range of 10 to 30 mL/s relatedto normal temperature and pressure (0 C and 1.013bar). The pressure in the reactor had a constant valueof 1.3 bar.

Experiments were carried out in a temperature rangeof 360-450 C and hydrocarbon feed fractions of 1.0-2.0%. The conversion of the pure components as wellas mixtures thereof was used for parameter estimations.To show the influence of temperature, hydrocarbon feedfractions, and composition of the hydrocarbon feed onthe reaction performance, the results for the reactionconditions depicted in Table 1 are presented in thearticle.

Since the normalized concentrations are related to thecarbon content, the 9% of nitrogen is not considered inthe yields. The concentrations and yields are, therefore,related to the amount of the linear C4 hydrocarbons,only. The yields for the use of a real raffinate II mixtureare about

Figure 3. Experimental setup comprising a fixed-bed reactor and an analytical section.

Yi )C,i

4ni

nHC,0(2)

Zi )O,i

2ni

nO2,0(3)

XHC ) 1 - YHC (4)

RSi )Yi

XHC(5)

Table 1. Reaction Conditions for Which the Results ArePresented in the Articlea

type xbtaHC xbte

HC xHC,0 T, C

Variation of Reaction TemperaturerII 18% 82% 0.9% 400

410420430440450

Variation of HC ContentrII 18% 82% 0.9% 420

1.4%1.8%

Variation of n-Butane/n-Butenes Contentsbta 100% 0% 1.0% 4200.25 25% 75% 1.0%rII 18% 82% 0.9%bte 0% 100% 1.0%

a The symbols xbtaHC, xbte

HC, and xHC,0 mean molar fraction ofn-butane in the hydrocarbon feedstock, molar fraction of n-butenesin the hydrocarbon feedstock, and molar hydrocarbon fraction inthe feed, respectively.

Yireal 0.91Yi (6)

5552 Ind. Eng. Chem. Res., Vol. 44, No. 15, 2005

since it was found in preliminary experiments thatisobutane shows the same conversion rate as n-butanebut is directly oxidized, mainly to carbon oxides.

The conversion of carbon monoxide to carbon dioxidecan be neglected at the typical reaction conditions inthe presence of VPO catalysts.19 The oxidation reactionof an arbitrary C4 hydrocarbon i with y hydrogen atomsand z oxygen atoms to carbon oxides results in

The stoichiometric constant si represents the amountof formed carbon monoxide from the oxidation of speciesi to carbon oxides. A value of si ) 1 means that onlycarbon monoxide is formed, whereas si ) 0 representsthe classical definition of total oxidation by formingcarbon dioxide as the only carbon-containing species.

We used an isothermal, plug-flow reactor model forthe evaluation of the reaction kinetics since deviationsfrom an isothermal operation were small. Plug flow wasassumed for two reasons: (1) A bidisperse packing bedwas used (catalyst particles and inert material), whichleads to higher packing densities, higher turbulence,and, therefore, lower deviations from plug flow. (2) Theexperimental data was obtained by using several fillingsof the reactor with different dilution ratios and differentflow rates. There were no significant differences in theconcentration profiles obtained from measurements ofthe reference operation conditions, despite the use ofdifferent dilution ratios and flow rates. We thereforeassumed that the plug-flow model is valid. Since nosignificant changes in the profiles were observed whenthe flow rate was altered, it was assumed that externalmass transport resistances were negligible.

The determination of the kinetic and stoichiometricparameters was carried out by means of nonlinearregression, where the parameters are found by a least-squares minimization of the error in fitting the data.More precisely, chi-square minimization was carried outwith individual standard deviations i for each mea-sured molar fraction xi. The standard deviations wereapproximated by an error estimation

The fixed amount abs represents the detection limit andwas chosen to be 0.01%. The relative amount rel is therelative error of the measured molar fractions and wasassumed to be 5%. The parameter estimation wascarried out with the optimization toolbox of Matlab. Thetest for the identification of the global minimum wascarried out by varying the initial guesses of the modelparameters in ranges with physically meaningful valuesand estimating the parameters. Most of the estimationsled to the same model parameters; however, some ofthem were significantly different but resulted in muchhigher residual values of the error, and these wererejected. It was, therefore, assumed that the globalminimum was found.

After the minimization, the linear approximations ofthe confidence limits of the individual model parameterand of the crosscorrelation coefficients were calculatedby methods presented in ref 20. A probability of 95%was chosen for the confidence limits. This means thatthe probabilities of the true model parameters to be inthe confidence limits are 95%. The heat of reactionswere estimated by correlations found in refs 21 and 22.

3. Results and Discussion

The results of the partial oxidation of 1% mol/mol ofsynthetic raffinate II in air at a temperature of 410 C

Figure 4. Partial oxidation of 1% mol/mol of synthetic raffinate II in air at 410 C. The concentration profiles (Yi) are plotted against themodified residence time (tmod), and the reactor selectivities (RSi) are plotted against the conversion level of the hydrocarbons in the feed(XHC). The symbols represent the experimental values, and the lines represent the calculations based on the reaction kinetics.

C4HyOz + (4 - 2si + y4 - z2)O2 f4siCO + 4(1 - si)CO2 +

y2H2O (7)

i ) abs + relxi (8)

Ind. Eng. Chem. Res., Vol. 44, No. 15, 2005 5553

are presented in the plots in Figure 4. The concentrationprofiles of n-butenes and n-butane as well as thecalculated sum concentration of both are shown inFigure 4a. At tmod ) 0, the sum concentration ) 1. Sincethe concentrations are related to carbon, the fractionsof 75% of n-butenes and 16% of n-butane of the syntheticraffinate II mixture lead to dimensionless concentra-tions of 82% and 17.5%, respectively. The n-butenes arealmost completely converted at a residence time of 0.1(g s)/mL. At this residence time, the conversion level ofn-butane is below a value of 15%. Because of thedifficulties in determining the concentrations of n-butane and 1,3-butadiene simultaneously, the experi-mental values of the concentrations of these specieshave large uncertainties for residence times

stock. Therefore, the approximations in the experimen-tal determination of the concentrations of n-butane and1,3-butadiene are sufficiently good.

The maximum yield of maleic anhydride amounts to48% for synthetic raffinate II. The yield related to areal raffinate II mixture, therefor, amounts to 43%.

The reactor selectivities as functions of the conversionlevel of raffinate II are shown in Figure 4b. Then-butenes are converted at low residence times. Theconversion of n-butane is insignificant as long as n-butenes are present. The selectivity behavior of raffinateII at low conversion levels is, therefore, solely character-ized by the behavior of the n-butenes. The oxygen profileis presented in Figure 4d. It shows that the conversionlevel of oxygen is 80% after most of the hydrocarbonshave been converted.

All experiments carried out at temperatures between360 and 450 C with pure n-butenes as well as withmixtures of n-butenes and n-butane were used for theparameter estimation. Measurements with pure n-butane in the feed were also used for the parameterestimation, but only those measurements that werecarried out after the catalyst had been used for theconversion of n-butenes were considered. The reactionnetwork displayed in Figure 2 was found to representall significant reactions of the oxidation of mixtures ofn-butenes and n-butane.

The Eley-Rideal approach, as it was proposed for theconversion of pure n-butane (eq 1), was extended byinhibition expressions for n-butenes and IP, resultingin rate equations of the type

The reference pressure p+ ) 26 kPa has been introducedso that the rate constants ki,j are independent of thepower of the partial pressure of oxygen and the rootamounts to values 1. The rate constants were assumedto obey Arrhenius law

The inhibition constants bi and the stoichiometricconstants si were found to be independent of tempera-ture.

The kinetic model consists of 24 model parametersthat were determined by the use of 1096 experimentallydetermined concentrations. The highest value of thecrosscorrelation parameters is 0.85 for the correlationof the rate constant kbte,COx and the inhibition constant

Figure 6. Effect of the reaction temperature (in C) on concentration profiles and reactor selectivities in the oxidation of 1% mol/mol ofsynthetic raffinate II in air.

ri,j )ki,jpixpO2/p+

1 + bbtapbta + bbtepbte + bIPpIP + bH2OpH2O(9)

ki,j ) ki,j exp(- EA,i,jRT ) (10)

Ind. Eng. Chem. Res., Vol. 44, No. 15, 2005 5555

bbte. The crosscorrelation parameters are, therefore, all

Parts e and f of Figure 6 show the reactor selectivitiesof IP and maleic anhydride, respectively. The reactorselectivities are almost independent of temperature. Thecalculations and the experimental values show goodagreement. At conversion levels of 90%, the maximumreactor selectivities are achieved.

3.2. Variation of the Hydrocarbon Fraction inthe Feed. The results of the experiments with variationof the hydrocarbon fraction in the feed and the corre-sponding calculations are presented in Figure 7. Theindividual curves are labeled as explained in Table 1.The concentration profiles of n-butenes and n-butaneare plotted against the modified residence time in partsa and b of Figure 7, respectively. The decrease in thedimensionless concentrations with increasing residencetime is slower for higher hydrocarbon fractions in thefeed. This can be attributed to the inhibition of thereactions by adsorbed hydrocarbon species and is con-sidered in the Eley-Rideal approach by the inhibitionexpressions containing the bi constants. The reactorselectivities of IP and maleic anhydride are plotted inparts c and d of Figure 7, respectively. The experimentalprofiles show no clear dependency on the hydrocarbonfraction in the feed: the profiles are similar. Solely, themaximum reactor selectivity of maleic anhydride for araffinate II fraction of 2.0% is slightly lower than the

corresponding values for smaller fractions. The calcula-tions show good agreement with the experimentalvalues.

3.3. Variation of the n-Butane/n-Butene Ratio.The concentration profiles and reactor selectivities thatwere determined in the experiments with varyingcompositions of the hydrocarbon mixture and the cor-responding calculated values are presented in Figure8. The concentration profiles of n-butenes as functionsof the modified residence time are shown in Figure 8a.The different amounts of n-butenes in the feed lead todifferent values for the dimensionless concentrations attmod ) 0. The profiles show a good agreement betweenexperiments and calculations. Also showing good agree-ment are the profiles that are presented in Figure 8b,where the concentration profiles of n-butane are plotted,except for the case of the conversion of pure n-butane.In this case, the calculated values are higher than theexperimental ones. The experimental values were ob-tained in an experiment with a freshly activated catalystthat had not yet been used in the oxidation of n-butenes.It was found that the catalyst behavior changes ir-reversibly after the catalyst was used for the conversionof n-butenes. The calculations, however, are based onthe reaction kinetics for the conversion of the mixturesafter the irreversible change.

Figure 8. Effect of different compositions of the hydrocarbon feedstock on the concentration profiles and reactor selectivities at 420 Cand 1% mol/mol of hydrocarbons in the feed. The keys show the abbreviations that are explained in Table 1.

Ind. Eng. Chem. Res., Vol. 44, No. 15, 2005 5557

The concentration profiles of the intermediate prod-ucts are plotted in Figure 8c. Calculated and experi-mentally determined values are in good agreement. Themaximum yields as well as the corresponding residencetimes decrease with increasing relative amounts ofn-butenes in the feed. Additionally, there is an increasein the conversion levels at which the reactor selectivitiesapproach zero. The highest deviations between thecalculated and experimental values are observed for25% of n-butenes in the hydrocarbon feed mixture andmay be explained with the uncertainty in the experi-mental determination of the 1,3-butadiene concentra-tion.

The reactor selectivities of maleic anhydride aredisplayed in Figure 8f. The initial selectivities of maleicanhydride (for X f 0) decrease with increasing relativeamounts of n-butenes. The calculated values for thepure n-butane conversion for conversion levels of 90%are lower than the experimental values. This deviationcan be attributed to the different behavior of the catalystafter the conversion of n-butenes. The yields of maleicanhydride are presented in Figure 8d. The highest yieldof 56% was achieved with pure n-butane, the lowestyield was achieved with a value of 48% with thesynthetic raffinate II mixture. In the case of puren-butenes, the maximum yield lies at a value of 53%.The reduction of the yield maximum upon the conver-sion of raffinate II can be attributed to the fact thathigher residence times are required to convert the less-reactive n-butane in raffinate II. Maleic anhydride hasalready been formed during the consumption of then-butenes at lower residence times and undergoesconsecutive reactions, while the conversion of n-butaneis still proceeding. In the experiment with a hydrocarbonmixture containing 25% mol/mol of 1-butene, the maxi-mum yield could not be achieved. Higher relativeamounts of n-butenes in the feed allow for shorterresidence times to reach the maximum yield.

4. Conclusions

The reaction network displayed in Figure 2 as wellas the reaction kinetics with rate equations of the typepresented in eq 11 is suitable to describe the oxidationof mixtures of n-butenes and n-butane to maleic anhy-dride. The kinetic model is in good agreement with theexperimental results. The value of the maximum yieldof maleic anhydride is insensitive to changes in tem-perature and hydrocarbon feed fraction and amountsto 48%. By increasing the temperature, however, theamount of catalyst required to reach the maximum yieldcan be reduced. The conversion of pure 1-butene andpure n-butane led to maximum yields of 53% and 56%,respectively. The efficiency of a maleic anhydride pro-cess will, therefore, strongly depend on the choice of thefeedstock and, in particular, on the availability ofn-butane at the designated production site.

Acknowledgment

The authors thank Max-Buchner-Forschungsstiftungfor financial support.

Nomenclature

bi ) inhibition constant related to species i, in MPa-1EA,i,j ) activation energy of the reaction of species i to

species j, in kJ/mol

ki,j ) rate constant for the conversion of i to j related tothe mass of active material, in mol/(s kg MPa)

ki,j ) preexponential factor in Arrhenius law for thereaction of species i to species j

related to the mass of active material, in mol/(s kg MPa)ni ) molar flow rate of species i, in mol/snHC,0 ) molar flow rate of hydrocarbons in feed, in mol/snO2,0 ) molar flow rate of oxygen in feed, in mol/sp+ ) 26 kPa, reference pressurepi ) partial pressure of species i, in kPapO2 ) partial pressure of oxygen, in kPari,j ) reaction rate for the conversion of i to j per mass of

active material, in mol/(s kg)R ) 8.314 J/(mol K), ideal gas constantsi ) stoichiometric constant for the oxidation of species i

to carbon oxidesRSi ) reactor selectivity of species iT ) temperature, in Ktmod ) modified residence time, in (g s)/mLXHC ) conversion levelxi ) molar fraction of species ixbta

HC ) molar fraction of n-butane in the hydrocarbonfeedstock

xbteHC ) molar fraction of n-butenes in the hydrocarbonfeedstock

xHC,0 ) molar hydrocarbon fraction in the feedYi ) normalized concentration of species i related to carbonYHC ) sum of the normalized concentrations of n-butenes

and n-butaneZi ) normalized concentration of species i related to oxygen

Greek Symbols

j,i ) number of atoms of type j in species ii ) standard deviation of the molar fraction of species iabs ) detection limit in the determination of molar

fractionsrel ) relative error in the determination of molar fractions

Literature Cited

(1) Arpentinier, P.; Cavani, F.; Trifiro, F. The Technology ofCatalytic Oxidations. 1. Chemical, catalytic & engineering aspects;Editions Technip: Paris, France, 2001.

(2) Kaufmann, D. Valorization of Raffinate II. In C4 ChemistrysManufacture and Use of C4 Hydrocarbons, Proceedings of theDGMK Conference, Aachen, Germany, Oct 6-8, 1997; Keim, W.,Ed.; DGMK Tagungsbericht 9705.

(3) Wiese, K.-D.; Nierlich, F. Getting out of the box: Furtherupgrading of C4 streams. In C4/C5-Hydrocarbons: Routes to highervalue-added products, Proceedings of the DGMK Conference,Munich, Germany, Oct 13-15, 2004; Ernst, S., Ed.; DGMK-Tagungsbericht 2004-3.

(4) Weissermel, K.; Arpe, H.-J. Industrielle Organische Chemie,5th ed.; Wiley-VCH: Weinheim, Germany, 1998.

(5) Heller, H.; Lenz, G.; Thiel, R. The Bayer Process for theManufacture of Maleic Anhydride from Butenes and as a By-Product of the Phthalic Anhydride Production. In Advances inPetrochemical Technology; Chemical Engineering Symposium;London, U.K., May 11-12, 1977; Vol. 50.

(6) Brandstadter, W. M.; Kraushaar-Czarnetzki, B. Pilot plantprocessing of n-butane to maleic anhydride above the explosionlimit. In Creating Value from Light OlefinssProduction andConversion, Proceedings of the DGMK Conference, Hamburg,Germany, Oct 10-11, 2001; Emig, G., Ed.; DGMK-Tagungsbericht2001-4.

(7) Quast, A. Partielle Oxidation von C4-Kohlenwasserstoffenzur Darstellung von Maleinsaureanhydrid und Furan. Ph.D.Dissertation, Universitat Karlsruhe, Karlsruhe, Germany, 1999.

(8) Lerou, J. J.; Mills, P. L. Du Pont butane oxidation process.Precis. Process Technol. 1993, 175-195.

(9) Sharma, R. K.; Cresswell, D. L. Kinetics and fixed-bedreactor modeling of butane oxidation to maleic anhydride. AIChEJ. 1991, 37, 39-47.

5558 Ind. Eng. Chem. Res., Vol. 44, No. 15, 2005

(10) Wellauer, T. P. Optimal Policies in Maleic AnhydrideProduction through Detailed Reactor Modelling. Ph.D. Disserta-tion, Eidgenossische Technische Hochschule Zurich, Zurich, Swit-zerland, 1985.

(11) Uihlein, K. Butanoxidation an VPO-Wirbelschichtkataly-satoren. Ph.D. Dissertation, Universitat Karlsruhe, Karlsruhe,Germany, 1993.

(12) Becker, C. Katalytische Wandreaktorkonzepte fur MSA-Synthese und Methanol-Dampfreformierung. Ph.D. Dissertation,Universitat Stuttgart, Stuttgart, Germany, 2002.

(13) Bej, S. K.; Rao, M. S. Selective oxidation of n-butane tomaleic anhydride. 2. Identification of rate expression for thereaction. Ind. Eng. Chem. Res. 1991, 30, 1824-1828.

(14) Varma, R. L.; Saraf, D. N. Selective oxidation of C4hydrocarbons to maleic anhydride. Ind. Eng. Chem. Prod. Res. Dev.1979, 18, 7-13.

(15) May, A. Das Reaktor-Regenerator-System als neuesKonzept fur die getrennte Reaktionsfuhrung bei der selektivenOxidation von n-Butan zu Maleinsaureanhydrid. Ph.D. Disserta-tion, Universitat Erlangen-Nurnberg, Erlangen, Germany, 2002.

(16) Varma, R. L.; Saraf, D. N. Oxidation of butene to maleicanhydride. J. Catal. 1978, 55, 361-372.

(17) Cavani, F.; Centi, G.; Manenti, I.; Riva, A.; Trifiro, F.Oxidation of 1-butene and butadiene to maleic anhydride. 1. Roleof oxygen partial pressure. Ind. Eng. Chem. Prod. Res. Dev. 1983,22.

(18) Brandstadter, W. M.; Casaretto, P.; Kraushaar-Czarnetzki,B. Maleic anhydride from mixtures of n-butane and n-butenes. InChances for Innovative Processes at the Interface between Refiningand Petrochemistry, Proceedings of the DGMK Conference, 2002;DGMK-Tagungsbericht 2002-4.

(19) Dente, M.; Pierucci, S.; Tronconi, E.; Cecchini, M.; Ghelfi,F. Selective oxidation of n-butane to maleic anhydride in fluid bedreactors: Detailed kinetic investigation and reactor modelling.Chem. Eng. Sci. 2003, 58, 643-648.

(20) Press, W. H.; Teukolsky, S. A.; Vetterling, W. T.; Fannery,B. P. Numerical Recipes in C: The Art of Scientific Computing,2nd ed.; Cambridge University Press: New York, 1992.

(21) Liley, P. E.; Thomson, G. H.; Friend, D. G.; Daubert, T.E.; Buck, E. Perrys Chemical Engineers' Handbook, 7th ed.;McGraw-Hill International Editions: New York, 1998; Chapter2: Physical and Chemical Data.

(22) Lucas, K.; Luckas, M. VDI-Warmeatlas, 9th ed.; VDIVerlag: Duesseldorf, Germany, 2002; Chapter Da: Berechnungs-methoden fur Stoffeigenschaften.

Received for review January 24, 2005Revised manuscript received May 10, 2005

Accepted May 12, 2005

IE050099O

Ind. Eng. Chem. Res., Vol. 44, No. 15, 2005 5559