Catalytic reforming

xúc tác reforming

Catalytic Reforming

The catalytic reforming process consists of a number of reactions which take place

on bifunctional catalysts for converting the hydrocarbons contained in naphthafractions to monocyclic aromatics

Naphthenes with six carbon atom rings are subjected to dehydrogenation.Naphthenes with five carbon atom rings are subjected to isomerization followed

by dehydrogenation, usually called dehydroisomerization The alkanes go throughcyclization followed by dehydrogenation, usually called dehydrocyclization.Simultaneously, the hydrocarbons and especially the alkanes undergo parallel, com-peting reactions of isomerization and hydrocracking with conversions sometimescomparable to the reactions producing aromatics

There are two ways in which catalytic reforming may be used One option is toprocess the heavy fractions of straight run naphthas in order to increase their octanerating by 40–50 units The other way is to process a narrow fraction of gasoline such

as C6-C8 or C7-C8 From the obtained reformate (called in this case BTX) are thenseparated the aromatic hydrocarbons (mainly benzene, toluene and xylenes), for thepetrochemical industry This second process is also called aromatization

Both processing options are performed in the same units, working under lar operating conditions The presentation that follows will be referring to bothoptions at the same time

simi-The final part of the chapter (Section 13.9) will present the catalytic processesused for converting hydrocarbons obtained from the aromatization, in order toincrease the production of those hydrocarbons that present a higher interest forthe petrochemical industry: hydrodisproportionation or dealkylation of tolueneand isomerization of xylenes

In contrast with this classical image of catalytic reforming, a new process hasbeen developed, the feed of which is the propane-butane fraction An extra step,



First time communicated at the ‘‘American Institute of Chemical Engineers Summer National Meeting’’, Denver, Colorado 21–24 August 1988 [1].

dehydropolymerization, is added before cyclization and aromatization This newprocess will be presented at the end of the chapter, but it is too early to estimateits impact on global processing.

As mentioned above, catalytic reforming consists of reactions of dehydrogenation,dehydroisomerization, and dehydrocyclization leading to the formation of aromatichydrocarbons The thermodynamics of other concurrent reactions, mainly involvingalkanes, of isomerization and hydrocracking were examined in previous chapters.Specific to catalytic reforming processes is the introduction of hydrogentogether with hydrocarbon feed into the reactor system Molecular ratios of 2 to 5hydrogen/hydrocarbons are used in order to decrease the rate of coking of thecatalyst Since hydrogen is a product of the aromatization reactions, its presence

in the feed to the reactor displaces the thermodynamic equilibrium of the reactions.This effect is accounted for in the following calculations and discussions

The equilibrium calculations were performed for pressures between 2 and 40bar in order to compare both older processes working at pressures of around 20–30bar and newer processes working at much lower pressures

The variations in heat of formation and entropy for the most important reactions ofcatalytic reforming were calculated based on the thermodynamic constants published

by Stull et al [2](see Table 13.1)

Using the calculation method developed by us and presented in Section.1.2, theequilibrium of dehydrogenation of the cyclohexanes is shown inFigure 13.1 Thehydrocarbons having similar values for the equilibrium conversion are groupedtogether

The data of Figure 13.1 refer to stoichiometric conditions However, if anexcess of hydrogen is present in the system, further calculations will be necessary.The equilibrium constant for the dehydrogenation of cycloalkanes is given by:

For an excess of n moles of hydrogen per mole of cycloalkane, the equilibriumconstant becomes:



So far this process is known by its trade names, such as ‘‘Cyclar’’ of BP-UOP or ‘‘Aroforming’’ of IFP Names such as ‘‘dehydropolymerization of lower alkanes’’ and ‘‘catalytic poly-reforming’’ are our suggestions.

Since at any given temperature the two equilibrium constants are equal, Eq.(13.1) and (13.2) can be equalized and, denoting

Figure 13.2 gives the value of a, for the equilibrium conversion x in metric conditions, (as read from the plots ofFigure 13.1),at a given temperature and

stoichio-Table 13.1 Thermodynamic Data for Cyclohexanes Dehydrogenation

Figure 13.1 The thermodynamic equilibrium of dehydrogenation of cyclohexanes instoichiometric conditions.

4 C8 alkylcyclohexanes $ xylenes þ 3H2, C9-C11 hexanes$ alkylbenzenes þ 3H2

alkylcyclo-Figure 13.2 The influence of hydrogen excess upon the equilibrium of dehydrogenation anddehydroisomerization of cycloalkanes

pressure and an excess of n moles of hydrogen The equilibrium conversion in thepresence of an excess of n moles of hydrogen is then calculated using Eq (13.3).Example Calculate the cyclohexane-benzene equilibrium at 468C, 33 bar, and

an excess of 5 mol hydrogen per mol feed

ANSWER: In stoichiometric condition, Figure 13.1gives x ¼ 0:90

For x ¼ 0:90 and n ¼ 5 ,Figure 13.2gives a ¼ 0:93

According to Eq (13.3), the equilibrium conversion with excess of hydrogen,

dehydroiso-Using the same methodology as in the case of the cyclohexanes, the equilibriumfor stoichiometric conditions is shown in Figure 13.3 Due to the fact that thestoichiometry of the reaction is identical to that for cyclohexanes, the expression

in Eq (13.4) and consequently the graph in Figure 13.2 for alkylcyclopentanes are

Table 13.2 Thermodynamic Data for Dehydro-isomerization

of Alkylcyclopentanes

Reaction

ð
H0

800Þr,kcal/mol

(
S0

800Þr,cal/mol deg49.24 85.414

cis 46.16 90.684trans 48.18 90.524cis 47.98 90.524trans 47.44 90.524

Figure 13.3 Thermodynamic equilibrium of dehydroisomerization of alkylcyclopentanes instoichiometric conditions:

also the same Thus, the equilibrium in the presence of hydrogen excess can becalculated in a similar way

13.1.3 Dehydrocyclization of Alkanes

Variations in the heat of formation and in the entropy for the reaction of cyclization were calculated for several typical hydrocarbons using the same source ofthermodynamic data [2] The results are given inTable 13.3

dehydro-In the calculations for the conversion of octanes, only the methylheptanes weretaken into account, since dimethylhexanes are found in much smaller quantities inthe straight-run gasoline (see Section 5.1.4)

The equilibrium conversion for dehydrocyclization under stoichiometric ditions is plotted inFigure 13.4

con-Following the same logic used in obtaining Eq (13.4), the equivalent sion for dehydrocyclization is:

13.1.4 Isomerization and Hydrocracking

Consequently, the degree of influence depends on their relative rates

The rate of aromatization of alkanes is the slowest Therefore their overallconversion to aromatics is influenced to a large extent by the reactions of isomeriza-tion and hydrocracking The rate of aromatization of the alkylcyclopentanes is muchhigher Therefore they are less affected by such reactions Finally, for alkylcyclohex-anes, the rate of aromatization is very fast, so that the reactions of isomerization andhydrocracking may be completely ignored The isomerization of alkanes has beenexamined inChapter 11.The composition of various heptane isomers may be taken

as representative of the higher alkanes According toFigure 11.4,at the temperatures

Table 13.3 Thermodynamic Data for the Dehydrocyclization of

Alkanes

Reaction

ð
H0

800Þr,kcal/mol

ð
S0

800Þr,cal=mol  deg63.77 105.02260.85 107.872

62.56 109.762

61.93 108.772

60.96 108.19258.31 106.34259.24 109.522

61.96 109.12259.31 107.27258.66 106.142

58.53 108.412

60.68 108.592

Source: Ref 2.

practiced in catalytic reforming, the equilibrium corresponds to 14% n-heptane,40% 2- and 3-methylheptanes, the rest being dimethylpentanes and trimethylbu-tanes In conclusion, isomerization diverts a high proportion of alkanes towardsmolecular structures that can no longer undergo dehydrocyclization.

Furthermore, the reactions of hydrocracking, undergone mainly by the alkanes, lower the conversion to aromatics even more

iso-The thermodynamic equilibrium of the overall reactions is influenced also bythe fact that the hydrocracking reactions are highly exothermic while the reactions ofaromatization are all highly endothermic As a result, the hydrocracking reactionslead to an increase of the outlet temperature and therefore an increase in the con-version to aromatic hydrocarbons, especially in the last reactor of the unit

The isomerization equilibrium of the alkylaromatic hydrocarbons produced inthe process becomes important, especially for the xylenes, the proportions of whichalmost always correspond to the thermodynamic equilibrium of their isomerization.Since the equilibrium composition corresponds to approx 20% ortho-, 20% para-,and 60% meta-xylene, the isomerization of meta-xylene becomes an important com-Figure 13.4 The thermodynamic equilibrium for dehydrocyclization of alkanes in stoichio-metric conditions

mercial process It will be presented separately, at the end of this chapter (Section13.9.5).

The calculations and results obtained in the previous sections lead to the followingconclusions concerning the thermodynamic limitations to the process of catalyticreforming

The temperatures do not change significantly with the process type and rangebetween 410C and 420C for the exit from the first reactor and about 480–500C forthe exit from the last reactor

In contrast, the pressure in the system varies more significantly Older unitswere operated at a pressure between 18 and 30 bar while the units built after 1980operate at pressures between 3.5 and 7 bar

The molar ratio hydrogen/hydrocarbon varies between 8 and 10 for the olderunits and was lowered to 2 to 5 for the more modern ones

Considering an effluent temperature of 480C from the last reactor of the unitand a molar ratio hydrogen/hydrocarbon of 4, the equilibrium conversions werecalculated comparatively for two levels of pressure, 20 bar and 2.5 bar The follow-ing conclusions may be drawn:

1 The equilibrium conversions of cyclohexane and alkylcyclohexanes will behigher than 99%, both at lower pressures of 3.5–7.8 bar and at higherpressures of 20 bar The conversion will drop to  90% for cyclohexaneand to  95% for methylcyclohexane at pressures of 30–35 bar

2 The equilibrium conversions of alkylcyclopentanes are somewhat lessfavorable The equilibrium conversion of methylcyclopentane at a pressure

of 20 bar reaches only 86% but increases to over 99% at pressures of 3.5–7bar The equilibrium conversion of higher alkylcyclopentanes will bealmost quantitative at both low pressures and pressures of around 20 bar

3 The equilibrium conversions of alkanes to aromatics are the least favored.For n-hexane, the conversion will reach only 30% at a pressure of 20 bar,decreasing to 10% at a pressure of 30 bar The equilibrium conversion ofn-hexane would reach 85–90% only at lower pressures of 3.5–7 bar Theconversion of heptanes and octanes to ethylbenzene will correspond toabout 85% at a pressure of 20 bar, becoming almost complete at pressures

of 3.5–7 bar The conversions to xylenes and higher hydrocarbons willreach 85% at a pressure of 40 bar, increasing to 92–95% at 20 bar andbecoming almost complete in low pressure processes

It should be noted here that the graphs presented inFigures 13.1–13.5allow anestimation of the thermodynamic limits to conversion, at the exits of any of the three(or four) reactors of the commercial catalytic reforming units

The first catalyst used for industrial catalytic reforming consisted of 9% denum oxide on alumina gel The plant started in 1940 under the name of

molyb-‘‘Hydroforming.’’

The use of platinum instead of molybdenum oxide was patented by V Haensel

of UOP in 1949 and the first plant using such a catalyst started working in the sameyear under the name of ‘‘Platforming.’’

The overwhelming advantages of platinum when compared to molybdenumoxide led to a complete replacement of the latter

The catalyst of platinum on -alumina support underwent various ments, both with respect to the support and by promoting the platinum with othermetals such as iridium, palladium, tin, and rhenium

improve-Catalytic reforming catalysts patented by different manufacturers werereviewed by Aalund [3] and presented in the monograph of Little [4]

The dual function character of the catalysts for catalytic reforming is provided

by the acid centers of the support, which catalyze the reactions of isomerization andhydrocracking, as well as by the metallic centers—platinum associated with othermetals dispersed on the support—which catalyze the dehydrogenation reactions Amore detailed analysis of the reaction mechanisms (see Section 13.3) has to considerthe formation of complex metal–acid centers In order to achieve maximum effi-ciency for the process, a balance must be found between the acidic and dehydrogen-ating functions of the catalyst

The most frequently used support is-alumina (-Al2O3), and its appropriateacidic level is achieved through a treatment with HCl or sometimes with HF.Hydrochloric acid is preferred, because the final acidity is easier to control.Sometimes CCl4 or organic chlorides are used instead of HCl

In the catalytic reforming process, traces of water tend to eliminate the chloric acid fixed on the support; therefore a certain amount of hydrochloric acid isFigure 13.5 The effect of the hydrogen excess on the dehydrocyclization of alkanes

hydro-continuously injected into the reactors in order to maintain the required acidity level

of the catalyst

In the past [5], amorphous aluminosilica was used as support but alkalinesubstances needed to be added to compensate for its strong acidity Since therequired acidity was quite difficult to achieve, the use of amorphous aluminosilicawas abandoned

The progress achieved in the synthesis of zeolites led to attempts to incorporatethem in the preparation of the alumina support, erionite [6] in particular In labora-tory studies, this led to an increase in octane number by 3 to 7 units [7]

The activity of erionite was explained in some studies by a bifunctionalmechanism of cyclization [8], or in other studies by cyclization on the external sur-face of the erionite crystals [9,10]

Such catalysts were prepared by mixing bohemite with 15–35 wt % H-erionite,followed by an impregnation with a solution of H2PtCl6 [6]

The publications concerning laboratory results do not provide sufficient mation to decide to include zeolites in the preparation of the supports for catalyticreforming catalysts However, the multiple possibilities offered by zeolytes and theadvances in their technology may increase their use in preparing more efficientsupports for bifunctional catalysts

infor-Currently, -alumina is used almost exclusively as support for commercialreforming catalysts Their preparation involves strict, proprietary rules that ensurethe reproducibility of their activity and porosity

The two main preparation methods are only mentioned here, since a detaileddescription of the preparation of catalytic reforming catalysts is the subject ofspecialized publications [188]

One method prepares the alumina of the desired structure and porosity Afterbeing washed, dried, and formatted, it is impregnated with the solutioncontaining the metallic compound The treatment with HCl that follows,sets the desired acidity One option is to use chloroplatinic acid, whichaccounts also for the final acid treatment

The other method is coprecipitation by mixing the solutions containing thesoluble alumina precursor with that of the platinum compound

The first method is preferred for reforming catalysts because it allows a moreprecise adjustment of the characteristics of the final catalyst Furthermore, this seems

to be the only suitable method when preparing bi- and polymetallic catalysts, ing a more precise dosage of the promoters

allow-At the beginning the only metal used was platinum To improve its dispersion

on the surface, a treatment with 0.1–0.5 wt % S was performed, usually within thereactors, just after loading the unit with a new batch of catalyst [4]

The content of platinum in catalyst varies between 0.2 and 0.6 wt %, usuallybeing between 0.3–0.35 wt % The content rarely exceeds this range

The content of chlorine usually varies between 0.8–1.3 wt %

An overview of the main characteristics of some bi- and polymetallic catalysts

is given inTable 13.4where data of six typical U.S patents are presented

It should be emphasized that the exact role of various metals added to num is not clear Existing publications tend to justify their presence mainly by their

plati-effect on the activity, selectivity, and stability of the catalysts Nevertheless, thepublished data allow some interesting assumptions and conclusions to be drawn.The addition of metals of the platinum group such as palladium or iridium hasthe effect of augmenting the reactions promoted by platinum The addition of acertain amount of iridium proves to be more efficient than increasing the content

of platinum by the same amount, as shown inFigure 13.6

Research on palladium was carried out mainly in Russia, with the purpose ofreplacing platinum, while research on iridium at the Institut Francais du Petrole led

to the preparation of highly efficient bimetallic catalysts such as RG422 and RG423,

as well as polymetallic catalysts with platinum, iridium, and rhenium

There are several ways of expressing the behavior in time, i.e., the performancestability of a catalytic reforming catalyst Usually, the stability is expressed by thedecrease in time of the octane rating of the reformate Sometimes it is expressed bythe increase of temperature required for maintaining a constant activity or also, bythe decrease of the selectivity, as expressed by the decrease in the volume of the liquidphase and of the hydrogen obtained

and of a catalyst promoted with rhenium

The effect of iridium is explained by its higher hydrogenating activitycompared to that of platinum This prevents the formation of coke deposits onthe active surface of the catalyst This theory is supported by experimental results

on platinum catalysts promoted with other metals, as shown inFigures 13.8and13.9

Typicalexample Range

Typicalexample

Typicalexample Range

Typicalexample

num The activity of iridium increases continuously with the Ir/Pt ratio, that ofrhenium reaches a maximum at about Re=Pt ¼ 1:4, while the activity of germaniumdecreases continuously.

Similar results were obtained by Stoica and Raseev [13] when studying thepromotion by rhenium of an industrial catalyst containing 0.35 wt % platinum.The maximum activity was obtained for 0.6 wt % Re, which corresponds to aratio Re=Pt ¼ 1:7

catalysts developed by the French Institute of Petroleum The results show thatwhen compared with a classic catalyst containing 0.6 wt % Pt, coke formation isabout 40% lower on a catalyst containing 0.6 wt % Pt and 0.6 wt % Re, it is byabout 67% lower on a catalyst containing 0.6 wt % Pt and 0.08 wt % Ir, while it ishigher on a catalyst containing 0.6 wt % Pt and 0.22 wt % Ge Both graphs inFigure13.8 and 13.9 show the advantages of rhenium, which is the most frequently usedpromoter in the preparation of catalysts for catalytic reforming Rhenium is alsoadded to catalysts based on platinum and iridium

The effect of rhenium on reducing the rate of catalyst aging is shown inFigure

Research and Engineering Co is also given for comparison

A similar effect of reducing the formation of coke is observed for tin [14–16].Furthermore, it decreases the amount of gas formed by decreasing the acidity of theacid centers

The catalysts containing Sn were prepared in Russia at Riazan under the nameSPR-2, as spheres of 1.5–1.8 mm diameter or as PR-42, as extrudates Both catalystswere prepared in two forms:

Figure 13.6 The effect of platinum and iridium concentrations on catalyst aging:* - 0.035

% Pt;* - 0.60 % Pt; ~ - 0.35 % Pt þ Ir Test conditions: Feed ASTM 95–200C; PONA50/-/42/8 % vol; pressure 10 bar; temperature 500C; H2=HC¼ 4 mol/mol (From Ref 12.)

A: containing 0.38–0.40 wt % Pt, 0.25 wt % Sn, and 1.3 wt % Cl2

B: containing 0.6 wt % Pt, 0.40 wt % Sn, and 1.3 wt % Cl2

In the preliminary experiments the catalyst PR-42B gave the best results It wassuccessfully tested in an industrial unit [14] and it gave a better performance than thecatalyst KR-110K (platinum-rhenium on alumina) The spherical catalyst SPR-2Balso gave good results, comparable to those obtained using multimetallic catalysts[17]

Figure 13.7 Effect of promoters on catalyst stability (From Ref 11.)

Figure 13.8 Hydrogenation activity vs metal/platinum ratio for iridium, rhenium, andgermanium Benzene þ 3H2! cyclohexane; P ¼ 1bar; t ¼ 500C; space velocity 20 g/g/h; H2

=HC¼ 20 mol/mol (From Ref 12.)

Figure 13.9 The dynamics of coke formation for various bimetallic catalysts n-C7H16 !toluene þ gases þ coke; p ¼ 5 bar; t ¼ 500C; space velocity ¼ 29 g/g/h; H2=Hc¼ 2 mol/mol.(From Ref 12.)

Studies performed in Russia, investigated the promotion by cadmium as well asthe general influence of the method of preparation on catalyst performance [19] Thestudies examined promotion by 0.2–0.3 wt % Cd on the platinum-rhenium catalysts[18] Although the higher relative volatility of cadmium leads to its elimination fromthe system, its effect seems to persist [20,21] This could be explained by its influence

on the dispersion of Re-Pt, which is maintained even after cadmium had left thecatalyst by volatilization

The dispersion of platinum and promoting metals in crystallites form plays avery significant role in the activity of reforming catalysts [4] In the case of platinum-rhenium catalysts, the platinum crystallites may have dimensions between 8 and 100A˚ [22] The size of crystallites in a fresh catalyst should be smaller than 35 A˚,preferably below 24 A˚ Current trends are to achieve even smaller dimensions(nanometer sizes) for the dispersed noble metals particles on the support surface.Catalysts containing the noble metals as nanoparticles, will have higher activity(larger number of metallic sites) and lower costs (lower loading of noble metals)than the traditional ones

One of the main roles of the promoters is to prevent the agglomeration ofcrystallites, a phenomenon known as ‘‘platinum sintering.’’

The dispersion effect can be observed inFigure 13.11, which shows the ratiobetween the amount of accessible platinum and the total amount of platinum forthree catalysts: a nonpromoted catalyst, a Pt Ge=Al2O3 catalyst, and a Pt Ir=Al2O3catalyst The success of germanium as a promoter in maintaining the accessibility ofplatinum in time is very clear, justifying its use in catalyst preparation

Additional details concerning the structure, characterization and testing of thereforming catalysts are given in the monographs edited by Antos et al [189–191]

Figure 13.10 The time variation of the performance of several catalysts (Feed: alkanenaphtha, p ¼ 10:5 bar, average temperature 499

C, F1 octane number of the reformate 102.5.)

13.3 REACTION MECHANISMS

Catalytic reforming was shown to consist of a number of reactions catalyzed by thetwo functions of the catalyst

whileFigure 13.12B gives a similar sketch for heptanes This reaction mechanismwas analyzed by Raseev and Ionescu [23–25]

The sketches of Figures 13.12A and 13.12B require some explanations.The precise mechanism of dehydrogenation of the six atom rings on Pt cata-lysts is still not completely understood [26] Thus it was not possible to either confirm

or deny the existence of the hexa-diene intermediate, already identified in the hydrogenation on molybdenum oxide catalysts For this reason such a step was notincluded in Figures 13.12 A and B

de-The identification of the presence of cyclohexene is insufficient to confirm thedehydrogenation in stages or to reject the opposite theory, which considers themechanism by which the molecule is adsorbed ‘‘parallel’’ to or ‘‘flat’’ on the catalystsurface, simultaneously on several centers—the ‘‘multiplet’’ theory of Balandin [27].Cyclohexene may result from a parallel reaction occurring on pairs of sites, in placeswhere the number of adjacent sites does not allow the adsorption on the multiplet.This aspect was mentioned by us earlier [23] Consequently, the two mechanismsmay take place in parallel, with the adsorption on multiplets having a dominant role.This latter mechanism is the only one accepted by various authors [28]

Figure 13.11 The effect of promoters on platinum efficiency (From Ref 12.)

The direct dehydrogenation of alkylcyclopentanes to aromatic hydrocarbons

by enlargement of the cycle catalyzed by platinum was considered possible [29] evenwithout the intervention of the acidic sites In order to check this assumption, weprepared a catalyst of platinum on activated carbon [30] Although the catalyst didnot have any acidity, cyclohexene and benzene still formed from methyl-cyclo-pentane at 400C in a batch reactor The conversions however were two orders ofmagnitude lower than for a typical Sinclair Baker RD150 catalyst with 0.35 wt % Pt

in the same conditions The conclusion is that the presence of the acid sites isessential in the aromatization of methylcyclopentane

The isomerization of cycloalkanes with five and six atoms in the ring followsthe reactions suggested in Figures 13.12A and 13.12B The presence of cyclohexene

Figure 13.12A Reactions of hexanes on bifunctional catalysts (hydrocracking reactions arenot included)

Figure 13.12B Reactions of heptanes on bifunctional catalysts (hydrocracking reactionsare not included)

and of methylcyclopentene has been identified in the study on methylcyclopentanearomatization [30].

It should be mentioned that the opening of the ring in methylcyclopentaneleads to the formation of 2-methylpentene in agreement with the predominantformation of tertiary ions on acidic sites The reaction is then followed by anisomerization to 3-methylpentene and n-pentene [30]

Our study confirmed the sequential character of the reactions The ratio of2-methylpentene to 3-methylpentene in the reaction product was almost two timeslarger than given by the thermodynamic equilibrium

The ring-opening reaction is followed by reactions of hydrocracking of the alkenes and also of the iso-alkanes

iso-The cyclization mechanism was examined in more detail by Raseev and Stoicawithin a study on the conversion of n-heptane on bifunctional catalysts in the unit[13] shown inFigure 13.13.The catalyst was diluted with inert material and the unitcould be operated up to space velocities of 1000 g/g.hour

At these high space velocities, cis- and 2, cis- and

trans-heptene-3, and isoheptene were identified No heptene-1 was found.Figure 13.14shows thevariation of total heptenes, methylcyclohexane, methylcyclopentane, and toluenewith the weight hourly space velocity, while Figure 13.15 shows the variation ofindividual heptenes with the WHSV The latter figure shows that the alkenesreach a maximum before methylcyclohexane, while toluene increases continuously.This fact supports the successive transformation

n-heptane $ heptenes $ methylcyclohexane $ toluene

as shown also inFigure 13.12B

The formation of intermediate alkenes was signaled by other authors [26,31] aswell

In most of the published studies it was observed that the 5-carbon atoms ringsare formed in parallel with the 6-carbon rings Their ratio depends on the character-istics of the catalyst

With the exception of the methylcyclopentane-cyclohexane system, the ability of formation of five- or six-carbon atom rings is almost the same This isreflected by the following equilibrium compositions calculated from thermodynamicdata at a temperature of 480C:

prob-In the system methylcyclopentane-cyclohexane, the equilibrium is displacedtowards the formation of methylcyclopentane; the equilibrium conversion at 480C

corresponds to only 0.084 cyclohexane Furthermore, the equilibrium of genation of cyclohexane to benzene is less favorable than for the higher homologs (see

benzene when subjecting to catalytic reforming naphthas rich in alkanes

The type of ring resulting from the cyclization step depends on the manner inwhich the hydrocarbon is adsorbed on the active sites of the catalyst Several assump-tions were developed on this subject, ranging between that involving the positioning ofthe molecule of alkane between the platinum atoms (Kazanski [32]) and the adsorption

of the atoms at the two ends of the carbon chain on two adjacent sites [13] It is difficult

to select from among these hypotheses From the energy point of view, for bons with seven or more carbon atoms, both types of rings have equal chance Thisagrees with the experimental results and with most current opinions

hydrocar-Figure 13.13 Continuous bench unit for the study of catalytic reforming

The other reactions taking place on bifunctional catalysts, i.e., isomerizationand hydrocracking, were already presented in detail inChapters 11and12.

The formation of benzene from C6hydrocarbons, as a result of demethylationand hydrocracking, is of special interest The sequence of reactions may be deduced

con-verts to cyclohexane, which then dehydrogenates to benzene

The smaller value and the position of the maximum yield of pentane at higher values of the space velocities than the maximum of cyclohexanesuggests that methylcyclopentane appears from the isomerization of cyclohexane,rather than by cyclization of n-hexane

methylcyclo-At the beginning of this chapter we suggested the names ‘‘catalytic ing’’ and ‘‘catalytic dehydropolyaromatization’’ for the process with commercial

polyform-Figure 13.14 The formation of heptenes, methylcyclohexane, methylcyclopentane andtoluene on a UOP-R11 catalyst in a continuous unit p ¼ 10 bar, t ¼ 500C, H2=C7H16¼

6:5: (From Ref 13.)

names ‘‘Cyclar’’ and ‘‘Aroforming.’’ This process involves the reactions of genation of C3-C5alkanes on metal sites, followed by the dimerization of the formedalkenes on the acid sites, and subsequent cyclization and dehydrogenation toaromatic hydrocarbons The reaction schematic developed by us is shown inFigure 13.16.

dehydro-The process uses catalysts based on zeolite support, which do not favor theformation of polycyclic hydrocarbons and implicitly the formation of coke

Figure 13.16 The reaction scheme of the aromatization process of C3–C5 alkanes onbifunctional catalysts

Figure 13.15 The conversion of n-heptane to C6-hydrocarbons on a Pt/Re Catalyst at p ¼

10 bar, t ¼ 500C, H2=hydrocarbon ¼ 1:5 mol/mol/ (From Ref 13.)

However, since the operating conditions are more severe than in classic ing, the catalyst requires continuous regeneration, either by continuously passing thecatalyst through a regenerator (Cyclar) [33,34], or via the cyclic operation of thereactors (Aroforming) [35].

reform-Details on the commercial implementation of these processes are given inSection 13.10

The influence of external diffusion was studied by Raseev and Stoica [13] in a tinuous bench unit by varying the linear velocity of the reactants through the catalystbed at constant space velocity, shape, and size of the catalyst particles and constantparameters of the process A plug flow through the catalyst was ensured It has beenfound that external diffusion influences the process only at Reynolds numbers lowerthan 4 These were calculated for the diameter of the catalyst granules, at constantconditions of 500C, 25 bar, molar ratio H2=HCof 6.5, and space velocity of 16.5 gn-heptane/g catalyst  hour

con-Since in commercial reactors the Reynolds number is always above this value,the external diffusion has no influence on the yields obtained in commercialconditions

Diffusion through the pores of catalysts depends very much on the pore ture of the particular catalysts, which makes it difficult to draw general conclusions.Previously published data [36–38] have shown a significant influence of the diameter

struc-of the catalyst granules on the overall rate struc-of the process The rate decreases with theincrease in the average diameter of the catalyst, which demonstrates the influence ofinternal diffusion

Experiments carried out using a Sinclair-Baker RD150 and UOP R11 catalysts

of various dimensions led to the conclusion that internal diffusion begins to influencethe overall reaction rate at granule diameters larger than 0.4–0.63 mm [13] Theprocess conditions were: temperature of 500C, pressure of 25 bar, H2=HC molarratio of 6.5, and WHSV of 50 g/g catalyst  hour

Since the size of the catalyst particles in commercial reactors is larger thanthese values, one may conclude that internal diffusion will reduce the overall reactionrate of industrial processes Naturally, the effect of internal diffusion may bedecreased or even completely eliminated by using a more porous structure for alu-mina support This explains the evolution of industrial catalysts from having a totalpore volume of 0.48 cm3/g catalyst in 1975–1978 to a value of 0.6–0.7 cm3/g catalyst

at the present time

Usually, reforming catalysts are prepared as extruded cylinders with diameters

of about 1.5 mm (or sometimes 3 mm) or as spheres with diameters of 2–3 mm.The use of zeolite support with bimodal pore distribution and macrocirculationpores can reduce or completely eliminate the slowing effect of the internal diffusion

on the reaction rate

The decrease of the diameter of the catalyst particles lowers the effect ofinternal diffusion and increases the pressure drop in the reactor(see Figure 13.17)

This increases the cost of compressing the recycle hydrogen-rich gases The optimum

shape and size of the catalyst granules are determined as a function of the specificprocess operating conditions.

13.4.2 The Reaction Kinetics

The first kinetic models for catalytic reforming were proposed by Smith [39] and byKrane et al [40] and were reviewed by Raseev et al [23–25]

Smith [39] expressed the rate of reaction by the equations:

con- For hydrocracking of alkanes and cycloalkanes:

By solving this system of differential equations, the time dependence of thecomposition of the reaction mixture may be calculated

Although the method published by Krane was one of the first tion methods, it is still used today for model development The models areused to improve the performance of commercial catalytic reforming units[41,42]

calcula-The approaches of Smith and Krane do not take into account that the catalyticreforming process takes place in adiabatic conditions Therefore, the equations have

to be solved simultaneously with those describing the evolution of temperature(deduced from heat balances) along the reactor length

Burnett [43,44] improved the approach He started from the following reactionscheme:

This scheme is debatable because of the well-known fact that alkanes do notconvert directly to aromatic hydrocarbons However, the scheme was used by theauthor for computer modeling, leading to plausible results.

Jorov and Pancenkov [45–47] used a simplified model, similar to that used bySmith:

hydrocracking products P Ð N Ð A

Table 13.5 Reaction Rate Constants for the

Calculation of Catalytic Reforming Kinetics by the

Method of Krane et al

A – aromatics, N – cycloalkanes, P – alkanes; the subscripts

represent the number of carbon atoms in molecule.

Source: Ref 40.

Again, they did not differentiate between the behavior of alkylcycloalkaneswith five and with six carbon atoms in the ring, or between the behavior of normal-and iso-alkanes Instead, they introduced several approximations as well as some

‘‘coefficients of adjustment’’ in order to fit the calculations to industrial data Thisgives an empirical character to the whole approach The results were used by theauthors for developing recommendations on the distribution of catalyst among thereactors and for the optimum temperature profile

Henningen and Bundgard-Nielson [48] developed a more complex model, thattakes into account differences in the behavior of cycloalkanes with five and with sixcarbon atoms in the ring It also considers differences in the behavior of normal- andiso-alkanes The flowchart of the model is:

d ¼

1

Cpðn þ 1Þðk1PNP
HNP!Cþ k1PIP
HIP!Cþ k2PACH
HACH!NP

þ k3PNP
HNP!ACHþ k4PACP
HACP!NPþ k5PNP
HNP!ACP

þ k6PIP
HIP!ACHþ k7PACH
HACH!IPþ k8PIP
HIP!NP

þ k9PNP
HNP!IPþ k10PIP
HIP!ACPþ k11PACP
HACP!IP

þ k12PACH
HACH!ACPþ k13PACP
HACP!ACH

In the above equations:  is the reaction time, pi the partial pressure of thecomponent i, Cpthe heat capacity, n the hydrogen/hydrocarbon ratio, kithe reactionrate constant expressed using the Arrhenius equation,
Hi!jthe heat of reaction forthe conversion i to j

The values for the pre-exponential factors, the activation energies, and the heat

of reaction recommended by the authors for the above model are given in Table 13.6

A major improvement brought by the model developed by Henningen was thedistinction made between cycloalkanes with five and with six carbon atoms in thecycle, and above all, the addition of the heat balance equation Considering theprocess as nonisothermal during calculation brought the simulation results closer

Table 13.6 Values for the Parameters of Eqs 13.11–13.17: Rate Constants,

Pre-exponential Factors, Activations Energies, Heat of Reaction

Reaction

Relative rateconstant, t¼ 500 C,

pH2¼ 30 bar

Pre-exponentialfactor A (naturallogarithm, ln A)

Activationenergy, E,kcal/mol

Heat ofreaction,
H,kcal/mol

Consequently, Jenkins and Stephens proposed the introduction of exponentialterms for pressure in order to correct the rate constants suggested by Krane:

ki¼ k0

i

p300

 i

ð13:18Þwhere kiis the corrected rate constant, k0i is the original rate constant given by Krane

at the pressure of 300 psig, p is the system pressure in psig, and is the correctioncoefficient If the pressure is expressed in bars, the ratio within the brackets becomes(p/21)

The values for the exponent suggested by Jenkins for different reaction typesare:

Ancheyta-Jua`rez and Anguilar-Rodriguez [41] of the Mexican PetroleumInstitute generalized Krane’s model further, adding a correction for temperature

to Jenkins’ pressure corrections Thus, Eq 13.18 became:

ki¼ k0

i

p21

 exp EiR

1

T0
1T

ð13:19Þwhere T0is the reference temperature and T is the temperature corresponding to theconstant ki The values for the activation energies Eiof various reactions were takenfrom Henningen(see Table 13.6)

The reference temperature corresponds to the temperature at which Kranecalculated the rate constants given inTable 13.5 However, Krane [40] mentionedthat for C5-C7 hydrocarbons the rate constants were obtained under isothermalconditions at 500C, while for the C8-C10 hydrocarbons the conditions wereadiabatic in the temperature interval of 515–470C In that case, the reference tem-perature should be taken as the temperature equivalent to the adiabatic mean rate,Eqs (2.163) or (2.164) For the present case, it is difficult and of no practical value tocalculate the reference temperature with a high degree of precision for each reaction.Indeed, since the average rate depends on the activation energy of each reactionunder consideration, the error using an average temperature is lower than the experi-mental error associated with the evaluation of the rate constants

In general, the temperature equivalent to the adiabatic mean rate, tEVMA, issomewhat higher than the arithmetic average for a specified interval Hence, thesame value of 500C could be accepted for C8-C10 hydrocarbons as for the C6-C7hydrocarbons

The model of Krane [40], improved by Jenkins [49], Henningen [48], andAncheyta [41] was used by AguilarRodriguez and Ancheyta-Jua`rez [42] for computermodeling of the process of catalytic reforming The results were in remarkableagreement with the experimental effluent compositions and the dynamics of thereaction temperatures

In all the above reported studies, the reactions were considered to be homogeneous, which was reflected in the expressions used for the rate equations Thedeveloped reaction models were shown to be satisfactory for reactor design and foroptimization purposes.

pseudo-The kinetics of catalytic reforming was treated as a heterogeneous catalyticprocess in 1967 by Raseev and Ionescu [51] within a study of the catalytic reforming

of C6-hydrocarbons in a bench unit with a plug flow, isothermal reactor connected to

a gas chromatograph

The reactions considered were:

The reaction rates for the five participants were expressed as a function of thefraction of catalyst surface occupied by reactants without distinguishing between thetwo types: acid and metallic sites:

i¼Pbipi

bipi

ð13:21Þ

where bi are the adsorption coefficients

Despite such simplifications, the limited number of experimental points thatwere generated proved insufficient for evaluating the parameters of the kineticmodel

was performed by the research team of Martin and Froment for C6carbons [52] and by Van Trimpant, Martin, and Froment for C7-hydrocarbons[53]

-hydro-The authors used the following scheme for C7-hydrocarbons without ing the hydrocracking reactions:

consider-n-P7¼ normal paraffin with 7 C atoms (heptane), SBP7 ¼ single-branched C7paraffins (alkanes), MBP7¼ multi-branched C7 alkanes, 5N7¼ C7 Naphthenes(cyclo-alkanes) with 5 C atoms in the ring, 6N7 ¼ C7 naphthene, with 6 C atoms

in the ring

The experimental data were obtained in an isothermal tubular reactor using apresulfided catalyst containing 0.593 wt % Pt and 0.67 wt % Cl2 on -Al2O3 Thefeed was 99% n-heptane The 996 experimental points obtained allowed the authors

to verify their kinetic model and to determine the values of the reaction rate stants and the adsorption constants The final equations were:

con- For isomerization of alkanes

r0i – reaction rates in kmol/kg catalyst hour in the absence of any coke deposits,

pA and pB– the partial pressures of the initial and final components of thereactions,

The pre-exponential factors and the activation energies are given inTable 13.7

The adsorption factors

The adsorption on acid centers—:

ð13:28Þwhere

bMCH ¼ 0:27 bar
1; A ¼ 8:34  109bar;
H0¼ 96:93 kJ=mol

The authors considered the dehydrogenation of methylcyclohexane to toluene

to be the only reaction taking place by adsorption on the metallic centers All otherreactions involved adsorption on the acid centers The authors justified this separa-tion by the fact that in isomerization, cyclization, and hydrocracking, the rate-controlling steps are those involving the acid centers This is in agreement with otherpublications [54–56]

The decay of catalyst activity was studied by many researchers [28,53,56–62] Mostauthors suggested various mathematical expressions that take into account thedecrease of the rate constants due to the formation of coke

Rabinovich et al [28] proposed equations of the following type:

ki¼ k0 1

1 þiCi

ð13:29Þwhere k0is the rate constant for the fresh catalyst, Ci is the coke content andiis adeactivation constant

Equations similar to Eq (13.29) have been written separately for the acid andfor the metallic sites of the catalyst However, the authors did not give any values forthe constants in order to apply this method

Tanatarov et al [57] suggested a deactivation equation of the following form:

Ostrovsky and Demanov [58] suggested the following form:

where c, cmax are the quantities of coke and M,K are the rate constants for theformation of coke on metallic and acid sites The experimental values obtained bythe authors gave a ratio ofM=Kffi 15.

The kinetics of coke formation was also addressed in the paper on the catalyticreforming of C6-C7 hydrocarbons quoted above [53] The following equation wasdeveloped for C7-hydrocarbons:

R0c¼ k1PT P5N7

p2 H

þ k2PT PMCH

p2 H

þk5pIH pB

p2 H

ð13:33Þ

In these equations, R0cis the initial reaction rate for the fresh catalyst, expressed in kgcoke /kg catalyst hour while p are the partial pressures The following subscripts areused for hydrocarbons: T – toluene, 5N7– cyclopentanes with seven carbon atoms,

H – hydrogen, MCH – methylcyclohexane, NP7 – normal-heptane, MCP – cyclopentane, IH – isohexanes, and B – benzene

methyl-The values of the pre-exponential factors of the five rate constants and theircorresponding activation energies are given below:

Rate constant

Pre-exponential factor, A,

kg coke/kg catalyst hour

Activation energy,kJ/mol

where these reactions correspond to the two terms in Eq (13.33)

In commercial plants, the space velocity is mostly constant For a constantspace velocity the ratio kg catalyst/kg feed per hour is constant, and it is given below:

The value obtained for Ccmay also be replaced in Eqs (13.39) and (13.35) inorder to calculate the decreased reaction rate due to the formed coke.

The amount of coke formed up to a certain distance zj from the inlet of theisothermal reaction system or, the corresponding time tj, may be obtained by writing

Eq (13.40) in terms of the durationj:

cRc ¼ ec Cc
1

cRcj ¼ e c Ccðz j ; j Þ
1

Dividing and regrouping results in:

iso-The mathematical model was based on the following reactions:

ðaÞðbÞðcÞðdÞðeÞðfÞAssuming first order homogeneous reaction, the rates were expressed as:

This approach was similar to the one discussed above [53]

The processing of experimental data, summarized inFigure 13.18, led to thefollowing expression for the rate of formation of coke:

Cis the coke contained in the catalyst in wt %

pis the partial pressures in MPa

subscripts:

Pis for alkanes

Ais for aromatics

H2 is for hydrogen

Figure 13.18 Coke formation vs relative reaction time The lines represent data calculated

by using Eq 13.45 (From Ref 59.)

used for the mathematical modeling The lines ofFigure 13.19indicate that values of

i and consequently the values of the deactivation functioni are reaction specific.This conclusion is different from that formulated by other authors [53], whoaccepted the same value for all reactions considered in the model

Analyses of the effect of the process parameters should account for the presence ofthe two types of active sites that catalyze different types of reactions, which aresometime competing reactions

Values for the apparent activation energies for the main reactions taking place incatalytic reforming on platinum catalysts were selected and calculated [23–25] frompublished data [63–66] They are given below:

These data reflect the fact that temperature has a higher effect on the reactions

of cyclization and hydrocracking than on the reactions of dehydrogenation ofcycloalkanes

Consequently, increasing temperature will enhance the conversion of alkanes by favoring the reactions of hydrocracking, while for alkanes it will enhancethe conversion for the reactions of cyclization and dehydrogenation

cyclo-The following values of apparent activation energies for the conversion ofn-heptane were derived from our research on some pure hydrocarbons [13] (seeTable 13.8)

The values of the activation energies depend on pressure and on the actualcatalyst as specified in the table above This makes the comparison with otherpublished data quite difficult

This is the reason for which the influence of temperature on conversion is oftenrepresented by graphs where all other operating parameters are kept constant

Table 13.8 Activation Energies of n-Heptane Conversion on

Different Catalysts

Catalyst Pressure, bar

Apparent activation energy, kJ/molOverall conversion Conversion to tolueneSinclair-Baker

[13,50,51] Figures 13.20 and13.21show such plots for the catalytic reforming of heptane and methylcyclopentane on a Pt-Re catalyst based on the experimentalresults of Stoica and Raseev [13].

n-These graphs confirm the well-known fact that the conversion to aromaticsincreases with temperature, especially for hydrocarbons with seven or more carbons

in the molecule For hydrocarbons with six atoms, thermodynamic limitations ence the final conversion as shown above As a result, higher temperatures are used

influ-in catalytic reforminflu-ing, although they are limited by coke formation Hydrocrackinflu-ing

is also intensified by temperature increase, but to a lesser extent than aromatizationand therefore it does not become a limiting factor

Both cases (n-heptane and methylcyclopentane) confirm the formation ofcycloalkanes as intermediates, their concentrations passing through a maximum asfor any successive process

Figure 13.20 The variation of effluent composition for the catalytic reforming of n-heptane,versus temperature Catalyst Pt-Re, p ¼ 10 bar, H2/n-C7¼ 6:5 mol/mol, space velocity ¼ 2 g/

g catalyst  hour (From Ref 13.)

For n-heptane feeds, the isomerization reaction leads to a ratio of i-C7H16=n-C7H16 larger than 1 (see Figure 13.20) However, the concentration of heptanesdecreases as the concentrations of hydrocracking products and of toluene increase.For methylcyclohexane, the decrease in the concentration of i-C6H14 is lessobvious and it is almost exclusively related to the formation of hydrocrackingproducts.

It is useful to plot the influence of temperature on the octane rating of thereformate, sometimes showing also the volume percent of the liquid fractionobtained Such graphs are shown inFigures 13.22and13.23

It was not possible to develop a general correlation expressing the influence oftemperature [4] owing to the wide differences between the operating conditions and

Figure 13.21 The variation of effluent composition for the catalytic reforming of cyclopentane, vs temperature Catalyst Pt-Re, p ¼ 10 bar, H2=HC¼ 6:5 mol/mol, spacevelocity ¼ 2 g/g catalyst  h (From Ref 13.)