Section III-Paper 4

REACTIONS I N CATALYTIC REFORMING OF NAPHTHAS

BY

H. G. KRANE," A. B. GROH," B. L. SCHULMAN"" AND J. H. SINFELT""

ABSTRACT. Catalytic reforming is today's most important refinery process for octane improvement;
consequently, large incentives exist to design and operate it efficiently. The most efficient use of
reforming requires a thorough knowledge of changes in molecular composition over a range of operating
conditions. Such knowledge has been gained for hydrocarbons containing up to ten carbon atoms by
reforming individual hydrocarbons and whole naphtha.
n-Heptane was reformed at pressures from 150 to 450 psig., and n-hexane, methylcyclopentane,
benzene, methylcyclohexane, and toluene were reformed at 300 psig. The rate of cyclization of
n-paraffins increases fourfold when the number of carbon atoms increase from six to seven. Cyclization
rate is independent of pressure, but hydrocracking and isomerization rates increase as pressure is raised.
Whole naphtha was reformed at 0.7 to 5 space velocity, 100 to 450 psig, 880 to 960"F., and 2 : l to
8 :1 hydrogen :hydrocarbon mol ratio. Conversion of paraffins through cyclization and hydrocracking
increases with molecular weight. Conversion of naphthenes is rapid; few remain even at high space
velocity. Loss of aromatics through dealkylation and hydrogenation becomes noticeable at low space
velocity. Over-all conversion of paraffins containing six to ten carbon atoms increases as pressure is
reduced and temperature is raised. Yield of aromatics correspondingly increases. Hydrogen :hydro-
carbon ratio has little effect on conversion at low space velocity and low pressure.
Kinetic models have been developed for n-heptane and for the principal reactions of whole naphtha.
Twenty rate equations for naphtha give conversion and formation of each molecule containing one to
ten carbon atoms ; fifty-three reaction steps are covered. Calculated product compositions and yields
compare well with experimental data.

RESUME. Parmi les procédés de raffinage ayant pour but l'amélioration de l'octane, la réformation
catalytique est actuellement le plus important; la poursuite de méthodes perfectionnées et de meilleurs
moyens d'utiliser celles-ci se trouve donc encouragée. La réformation vraiment efficace présuppose une
connaissance approfondie des changements encourus par la composition moléculaire dans toute l'étendue
de conditions d'opération. On peut affirmer qu'une telTe connaissance est actuellement acquise dans le
domaine d'hydrocarbures possédant jusqu'à dix atomes de carbone; ce but a été atteint dans la réforma-
tion expérimentale d'hydrocarbures individuels et de naphtes entiers.
La réformation de n-heptane a été exécutée sous pressions de 150 & 450 psig (pounds per square
inch gage, livres par pouce carré de gauge), celle de n-hexane, methylcyclopentane, benzène, methyl-
cyclohexane et toluène sous la pression de 300 psig. On a trouvé que la vitesse de formation de structures
cycliques en partant de n-parafines croît au quadruple lorsque le nombre d'atomes de carbone
augmente de six à sept. La vitesse de la formation de cyc!es est indépendante de la pression; par contre,
celle d'hydrocracking et d'isomérisation croît en même temps que la pression.
Le naphte entier a été réformé à des viiesses spatiales de 0.7 à 5 sous pressions 100-450 psig,
880-960°F et avec quotients moléculaires hydr0gène:hydrocarbure de 2 : l à 8:l. La conversion de paraf-
fines par formation de cycles et par hydrocracking croît avec le poids moléculaire. La conversion de
naphtènes se poursuit rapidement; il en reste peu, même à des vitesses spatiales élevées. La perte de
composés aromatiques, par désalkylation et hydrogénation, devient perceptible à des vitesses spatiales
basses. La conversion totale des paraffines comprenant de six à dix atomes de carbone croît en raison
inverse de la pression et en raison directe de la température. Le rendement en composés aromatiques

*Standard Oil Company, Indiana, Whiting, Indiana, U.S.A.:

**&so Research and Engineering Company, Linden, New Jersey, U.S.A.

39
40 FIFTH WORLD PETROLEUM CONGRESS

augmente parallèlement. La valeur du quotient hydrogène:hydrocarbure influe faiblement sur la con-

version lorsque les valeurs de vitesse spatiale et de pression sont basses.
Des modèles cinétiques ont été déduits pour le cas du n-heptane, ainsi que pour les réactions prin-
cipales du naphte entier. Pour ce dernier, vingt équations de vitesse fournissent les valeurs de conversion
et de formation pour toute molécule contenant de un à dix atomes de carbone; on a considéré cinquante-
trois degrés divers de réactions. Les compositions calculées du produit et les rendements théoriques sont
en accord avec les données de l’expérience.

Introduction I n order to develop a more thorough knowledge and

Catalytic reforming of naphtha has become the most a more detailed kinetic interpretation of reforming
important refinery technique for raising octane num- reactions, changes in molecular composition both of
ber. Among catalytic conversion processes, reforming individual hydrocarbons and of whole naphtha were
capacity is second only to that of catalytic cracking. studied over a range of conditions used in regenerative
I n view of this capacity, there are large incentives to platinum reforming, as exemplified in the Ultraform-
design and operate reformers efficiently. The most ef- ing and Powerforming processes [ 7,8,9, 10, 11] jointly
ficient use of reforming requires a thorough knowledge licensed by our two companies. Such studies were
of how feed composition and operating conditions af- aimed at defining the major reactions of individual
fect product compositions and yields. Selection of feeds, hydrocarbons containing up to ten carbon atoms, and
design and operating conditions of reformers, and at developing a kinetic model that predicts molecular
further processing of the reformate all depend upon the composition and yields for use in reforming studies.
changes in molecular composition that occur in re-
forming reactions. Experimental
I n most catalytic reforming processes, the reactions Individual hydrocarbons and whole naphtha were
are promoted by fixed beds of platinum-on-alumina reformed by experimental techniques that differed in
catalysts. The main reactions are dehydrogenation, detail but followed the general process scheme shown
isomerization, cyclization, and hydrocracking [ 1, 21. in Figure 1. Liquid feed was mixed with hydrogen or
Naphthenes dehydrogenate to aromatics. Paraffins and
naphthenes isomerize to form compounds of the same RECYCLE G A S

number of carbon atoms in different arrangement.

r---- ----- 1
I
Paraffins cyclize by ring-closure to naphthenes. Paraf-
fins also hydrocrack to other paraffins with fewer car-
bon atoms. Other reactions take place to a lesser ex-
tent; aromatics hydrogenate to naphthenes, naphthenes GAS
REACTOR
hydrodecyclize to paraffins, and cyclic compounds de-
alkylate. These reactions have been explored with in-
dividual hydrocarbon reactants [31, and mechanisms SEPARATOR

have been proposed [4, 51. Simplified kinetic descrip-

tions of the reactions of paraffins, naphthenes, and LIQUID

aromatics have also been described [6]. FIG. 1-Process Flow.

EXPERIMENTAL REFORMING CONDITIONS

Individual
Hydrocarbons Whole Naphtha
Apparatus Isothermal Adiabatic and Isothermal
Catalyst, Wt. % Platinum 0.3 0.6
Pressure, psig 150,300,450 100 to 460
Temperature, “F 930 880 to 960
Hydrogen : Hydrocarbon mol ratio 3 :1, 6 :1 2:1to8:1
Space Velocity, weight liquid per hour
per weight catalyst 6 to 200 0.7 to 6
 REACTIONS I N CATALYTIC REFORMING OF NAPHTHAS 41

hydrogen-rich recycle gas and vaporized. The gases Space velocity was varied to cover a range of conver-
were passed over a fixed bed of catalyst in a reactor. sions of reforming.
Reactor effluent was analyzed before or after cooling Experimental conditions in the isothermal pilot plant
and separating into gaseous and liquid products. Feeds were varied over a wide range to obtain the effects of
and products were analyzed by mass, ultraviolet, and operating variables. Space velocity was varied by
infrared spcctroscopy and by gas chromatography. changing the amount of catalyst while holding feed
The experimental conditions for reforming individual rate constant. Conditions for individual experiments
hydrocarbons and whole naphtha are given in Table I. were selected on the basis of an interlocking factorial
Individual hydrocarbons were reformed in a labor- design [ 131. Products from selected tests were analyzed
atory-scale isothermal unit. Compounds studied were in detail.
n-hexane, methylcyclopentane, and benzene in the six-
carbon group, and n-heptane, methylcyclohexane, and Reaction Patterns
toluene in the seven-carbon group. Catalysts containing Experimental results showed how hydrocarbon type,
0.3 weight per cent platinum on alumina were diluted number of carbon atoms, and process variables affect
with inert beads to help provide isothermal conditions. reforming reactions. By the use of low conversions to
Normal heptane was reformed at 150, 300, and 450 reduce secondary reactions, the rates of the primary
psig; the other compounds were reformed at 300 psig. steps were obtained for individual compounds contain-
Hydrogen :hydrocarbon mol ratio was 5 : 1 for paraf- ing six and seven carbon atoms. Reactions of hydro-
finic and naphthenic feeds, 3: 1 for aromatic feeds. carbons containing up to ten carbon atoms were given
Once-through hydrogen was used. Space velocity was by the results with whole naphtha. Major process vari-
varied between 5 and 200 weights of liquid feed per ables were studied to determine the effects of reaction
hour per weight of catalyst in order to limit conversions conditions on the conversion of these compounds.
to 10 to 40%.
Individual Hydrocarbons
A Mid-Continent naphtha boiling between 200 and Product compositions were obtained when reforming
363°F was reformed in an adiabatic pilot plant proc- six- and seven-carbon hydrocarbons. From these data,
essing one barrel per day and in an isothermal pilot rates of reaction were determined by plotting fractional
plant processing one-quarter barrel per day. I n the conversion to specific components versus the reciprocal
single reactor of the adiabatic unit, catalyst was placed of moles of feed per hour per weight of catalyst, and
in four equal beds separated by reheat sections con- estimating the slope of the curve at zero conversion.
taining inert material. I n the isothermal unit, catalysts Rate data are summarized in Table II.
in a single bed were diluted with inert material to help The main reactions of paraffins proceed at different
provide isothermal conditions. rates for each compound. The rate of isomerization of
In the adiabatic pilot plant, pressure, kinetic average n-hexane is more than twice that of hydrocracking.
temperature, [ 121, and hydrogen: hydrocarbon ratio Cyclization proceeds only one-fourth as fast as hydro-
were held constant. Recycle gas provided the hydrogen. cracking. Thus, conversion to aromatics is low. n-Hep-

TABLE II
REACTION RATES
(Gram Moles per Hour per Gram of Catalyst)
Methylcyclo- Methylcyclo-
n-Hexane n-Heptane pentane hexsne
Isomerization
n- to iso-Paraffins 0.12
-
0.16 - -
five- to six-carbon Rings - 0.13 -
Cyclization 0.012 0.06 - -
Hydrodecyclization - - 0.09 -
Hydrocracking 0.06 0.06 - -
Dehydrogenation - - - 1.4
42 FIFTH WORLD PETROLEUM CONGRESS

tane isomerizes and hydrocracks at about the same

HEAVY P A R A F F I N S
rate as n-hexane. The rate of cyclization, four times
eo -
that of n-hexane, leads to much greater conversion
to aromatics.
Reactions of naphthenes show marked differences
between five-carbon and six-carbon rings. Methylcyclo- 50, l I I I l 1
hexane readily dehydrogenates to toluene; no other sig-
nificant reactions were observed. On the other hand,
methylcyclopentane isomerizes to six-carbon rings and
undergoes ring opening with hydrogenation to paraf-
fins. The isomerization reaction is followed by rapid o 20

dehydrogenation to aromatics. The rate of the com-

Y
Y
/-
,SEVEN ,SIX -
JFIVE

peting ring-opening reaction is 0.7 that of isomeriza- w !

tion; consequently, the formation of aromatics is less
that that of six-carbon ring naphthenes. 0 NAPHTHENES

Benzene and toluene conversions are much slower

than paraffin and naphthene conversions. The ob-
served products, mainly a spectrum of paraffins, are
the result of ring-opening and hydrocracking. The for-
. O t \
mation of small amounts of toluene from benzene, and
of benzene and ethylbenzene from toluene, is evidence
of addition and removal of side chains.

Whole Naphtha
Changes in composition that occur when whole
naphtha is reformed provide a basis for deducing the
reforming reactions of hydrocarbons containing up to
O 10 20 30 40 50 60 70 80 90 100
ten carbon atoms. Mid-Continent naphtha and five
A C T I V I T Y I SPACE V E L O C I T Y
reformed products were analyzed to obtain the hydro-
carbon type and the amount of each molecule contain- FIG. 2-Conversion of Who!e Naphtha.
ing one to nine carbon atoms. Molecules containing ten
containing ten or more carbon atoms, whereas 60%
or more carbon atoms were combined. The reformed remain of those containing eight carbon atoms.
products covered a wide range of conversions. Data
The cracked and hydrogenated products of the
were obtained at different catalyst activities that re-
heavy paraffins appear as light paraffins containing one
flected changes in coke content of the catalyst. Activity,
to seven carbon atoms. As the heavier molecules dis-
defined as per cent of activity of coke-free catalyst,
appear, less six- and seven-carbon paraffins are formed,
was monitored by periodic tests. and they themselves begin to be converted. Heptanes
Composition of the naphtha and reformed products disappear first; hexanes would disappear on further
are plotted in Figure 2 against reforming severity, in reforming. Paraffins of one through five carbon atoms
terms of catalyst activity divided by space velocity. The increase continuously. These molecules result not only
disappearance and formation of molecules of each hy- from heavier paraffins, but also from naphthenes and
drocarbon type can be interpreted on the basis of the aromatics by removal of side chains and by ring open-
studies of individual hydrocarbons. ing followed by hydrocracking.
Heavy paraffins with eight to ten carbon atoms Conversion of naphthenes to aromatics is essentially
quickly convert by hydrocracking and cyclization. The complete at low severity. Detailed analyses of other re-
heavier the paraffin, the faster the conversion. For ex- formed products indicate that the remaining naph-
ample at a severity of 25, only 20% remain of paraffins thenes are about 90% alkylcyclopentanes.
 REACTIONS IN CATALYTIC REFORMING OF N A P H T H A S 43

Xrornatics are formed quickly by naphthene dehy- 3. The distributions of isomers in each carbon-atom
::rogenation. Ninety per cent of the aromatics present group are plotted against reforming severity. Distribu-
a t a seberity of 90 are formed previously at 25. Aro- tions do not change significantly beyond low-severity
matics of seven, eight, and nine carbon atoms increase reforming. Thermodynamic equilibrium apparently is
slowly after swift initial formation. Those containing established among isomers of the same number of
ten or more carbon atoms decrease a t high severity by carbon atoms [ 14,151.
removal of side chains, and by ring opening. Although Distributions of paraffin isomers, on the other hand,
arcnatics of less than ten carbon atoms undergo the
vary over a wider range of reforming severity as shown
s a r e reactions, there is a net production of these com-
in Figure 4. The proportion of branched-chain com-
pounds. They too would decline on further reforming.
Spectral analyses show that higher aromatics gen- pounds increases with severity and with molecular
crdly consist of a benzene ring with one or more nor- weight. The branched-chain molecules approach ther-
mal-alkyl groups. Some of these compounds containing modynamic equilibrium with straight-chain molecules
e:ght, nine, and ten carbon atoms are shown in Figure at high severity.

100 I I I

DIMETHYLBENZENES
80 - - O
V 80 IS0 H EPTAN ES

--u- - - -O--
z TRIMETHYLBEN ZENES
!- \ '
-
O u
"L
\'
I- a
6
a
c
LL
40 L
c %
METHYLETHYLBENZENES
1 5
O
" n-HEXANE

7- -- -0,
n-HEPTANE
l- I
a z
I
O
Z m
O 0 a 0
m
u 100 I
I I I I I I
a
O IL
O ISON ON AN ES
LL
o - I-z 801
I-
- DIMETHYLETHYLBENZENES
W
Z o ISOOCTANES
W TETRAMETHYLBENZENES
O
n- BUTYLBENZENE
a DIETHYLBENZENES
J

-
A

A d-
I I
O L I O
O 50 IO 0 O 50 100
ACTIVITY /SPACE VELOCITY A C T I V I T Y / SPACE V E L O C I T Y
FIG. 3-Aromatic Isomers. FIG. "Paraffin Isomers.
44 FIFTH WORLD PETROLEUM CONGRESS

Effects of Process Conditions preting the effects of reforming whole naphtha where
Pressure, temperature, and hydrogen :hydrocarbon the primary reactions are not obvious.
ratio can affect the relative rates of reforming reactions Detailed compositions of products from whole naph-
and the overall conversion of hydrocarbons. Reforming tha reformed over a broad range of conditions were ob-
n-heptane at low conversions shows how primary re- tained. ßecause more than one variable was changed
actions are affected by pressure. Reforming whole at a time, to coníorm to the experimental design, the
naphtha at higher conversions includes reverse reac- data were correlated by multiple regression on a digital
tions, and shows effects of process variables on overall computer. I n this technique, the method of least
conversion of hydrocarbons. squares is used to obtain empirical equations relating
The effects of pressure on conversion on n-heptane component yields to the variables. Plots derived from
to cyclics, to lower-carbon paraffins, and to isoheptane these equations reveal the effects of specific variables
are shown in Figure 5. Rate of cyclization is essentially on conversion.
independent of pressure. Rates of hydrocracking and Effects of pressure and temperature on reforming
of isomerization increase 50% for the three-fold in- paraffins containing six and more carbon atoms and on
crease. These results are particularly useful in inter- all aromatics are shown in Figure 6. Conversion of

-
I I 1 I
SIX-CARBON AND
L I G H T E R PARAFFINS
/ IS O H E P T A N E S
/

I I./ v

O I 2 3 0 I 2 3 O I 2 3
100 / S P A C E V E L O C I T Y
Data were obtained with a catalyst containing coke and. hence, rates obtained from these data
are not consistent with rates shown in Table II.
FIG. 5-Effect of Pressure on Conversion of n-Heptane.
 REACTIONS I N CATALYTIC REFORMING OF NAPHTHAS 45

100 PSIG 450 P S I G

60
I I I l I I
I
SIX-CARBON A N D HEAVIER S I X - C A R B O N AND HEAVIER
PARAFFINS PARAFFINS
40 - 1

w
W
II.
20

IL
O

w o
6 80 r I I I I I I
5
AROMATICS 9 3 5 0 ~ AROMATICS
O
0
o
a 60
W -/ /7----
/-------

885OF
-
v>
W
J
,040 - -
/e---
/
935.F
-
885°F -

20 - -

O I I l I I I

A C T I V I T Y / SPACE V E L O C I T Y
FIG. /;Effects of Pressure and Temperature on Conversion of Whole Naphtha.
70 I I
paraffins is more complete at low pressure and at high H2: HYDROCARBON R A T I O

temperature. The same conditions produce the most 2:1 O00

aromatics. At 450 p i g , a large amount of paraffins
n
apparently remains unconverted at high severity. Pro- W
AROMATICS
duction of aromatics is correspondingly low.
Although the data from reforming n-heptane indi-
cated no effect of pressure on rate of cyclization, for-
mation of aromatics from whole naphtha varies with 40
pressure. The higher formation of aromatics at low
pressure probably is caused by both ( a ) reducing the
rate of hydrocracking while not affecting the rate of - I I'. FIVE-THROUGH-SEVEN-CARB~
cyclization of paraffins, as shown by the data on n-hep-
tane, and ( b ) reducing hydrogenation of aromatics at O
high severity. More complete conversion of paraffins
obtained at low pressure also results from less hydro-
genation of aromatics and ring opening. PARAFFINS
Effects of hydrogen: hydrocarbon ratio on conversion
of paraffins and formation of aromatics at 100 p i g are O t
O 10 20 30 40 50 60
shown in Figure 7. Although severities were not low A C T I V I T Y / SPACE V E L O C I T Y
enough to establish initial rates, a fourfold change of FIG. 7-Effeet of Hydrogen: Hydrocarbon Ratio
the ratio at higher severities has little effect on over-all on Conversions.
46 FIFTH WORLD PETROLEUM CONGRESS

conversions. The ratio does have a pronounced effect also cyclizes to the naphthene and hydrocracks. The
on the kinds of aromatics produced; low ratio favors naphthene dehydrogenates t o an aromatic and hydro-
formation of the lighter aromatics. Data at 450 psig decyclizes to both normal- and isoheptane. The aro-
and 885'F suggest that increasing hydrogen: hydro- matic hydrogenates to the naphthene.
carbon ratio reduces the rate of conversion of paraffins For reforming at 925'F. and a hydrogen:hydrocar-
to aromatics. Thus hydrogen :hydrocarbon ratio has a bon mol ratio of 5 : 1, these patterns were used to de-
different effect on conversions at different pressures. velop the five rate equations shown on the following
page.
Reaction Kinetics In these equations, the symbols are :
nC, = Moles n-heptane per mole hydrocarbon
The data give product yields and compositions for
feed.
specific feeds. For general application, the data should
iC, = Moles isoheptane per mole hydrocarbon
be correlated so that product yields and composition
feed.
can be determined for any feed. The best correlations
N = Moles naphthene per mole hydrocarbon
are those based on a kinetic model of the reforming re-
feed.
actions. Such a model comprises a set of differential
A = Moles aromatic per mole hydrocarbon
equations that describes the rate of each reaction in
feed.
terms of composition and reaction conditions. When
H = Mol fraction hydrogen.
integrated, these equations give the composition
C = Moles of iso- and n-heptane converted to
changes across the whole reaction zone.
cracked products.
The ideal kinetic model describes every reaction step
by including ali intermediate steps that may involve
MSV = Molal space velocity, gram moles per hour
per gram of catalyst
only trace components. Most models must compromise
T = Total pressure, atmospheres.
this ideal, either for lack of data or to avoid over-com-
The rate equations are pseudo-first order with respect
plication. The large number of components makes a
to hydrocarbon concentration, expressed as moles per
kinetic model for whole naphtha exceedingly compli-
cated unless some of the reactions are neglected or mole of hydrocarbon feed. Relative rates for primary
combined into over-all steps. A model for individual reactions of n-heptane and for the effect of total pres-
hydrocarbons can consider more details for a given de- sure on cracking and isomerization were derived from
gree of complexity. Even with compromises for simpli- the present studies at low conversion. From published
city, kinetic models are powerful tools for describing data on reforming n-heptane, [3] coefficients for sec-
complex composition changes in reforming. ondary reactions were established. The ratios of co-
efficients for forward and reverse reactions in most part
Kinetic Model for n-Heptane correspond closely with values predicted from thermo-
A relatively simple kinetic model for reforming n- dynamic equilibria.
heptane has been developed to describe the reaction Calculated conversions agree well with the published
steps between n-heptane, isoheptane, naphthene inter- data. This comparison, shown in Figure 9, illustrates
mediate, cracked products, and aromatic product. The the value of the kinetic approach in describing reaction
reaction patterns are indicated in Figure 8. n-Heptane patterns. Although many of the rate equation coeffi-
isomerizes to isoheptane, cyclizes to a naphthene and cients were obtained from data at low conversions, the
hydrocracks. Isohcptane isomerizes to n-heptane, and model predirts the results of reforming at high conver-
sions for 100 to 500 psip. The reduced conversion to
aromatics observed at high pressure is accounted for
in the model by increased heptane cracking and by in-
creased hydrogenation of aromatics after they are
formed.
Other models are being tested to get an even better
FIG. &-Reaction Patterns for Kinetic Model of fit to the data. An unpublished non-linear regression
Reforming n-Heptane. technique suggested by Professor G. E. P. Box of Prince-
 REACTIONS IN CATALYTIC REFORMING OF NAPHTHAS 47

-
d(nC,) - - [0.060 t 0.061n0*37 t 0.026~'."] nc, i ]
[ 0 . 0 2 3 ~ ~ * ~ 'iC, t 0.0337r.H.N
( 1 )
d-
MSV

--
d(iC,) -
1
- [,.O60 t t 0.015~0.33] ic, + [ ~ ~ ] + 0.015.rr.H.N
0 . 0 6 1 ~ ' ~ nC,
d(-)
MSV

díN)

dí-)
1
= - [0.048n.H + 1.7
1 N t 0.060.nC7 t 0.O60.iC7 + 4.5 x 10-6773.H3.A

MSV

= [0.026n0."] nc, t [0.0i5n0*"] ic,

1

ton University is being used to determine the best carbon number. Aromatics hydrogenate and decyclize
model. to paraffins of the same carbon number, and dealkylate
to aromatics of lower carbon number.
Kinetic Model for Whole Naphtha All reactions are presumed to be pseudo-first order
A kinetic model has been developed for reforming with regard to the hydrocarbon. The hydrocracking
naphtha that describes the principal reactions under reactions are assumed to occur by cleavage of one car-
mild reforming conditions. The rate equations give the bon-carbon bond to yield two lighter paraffins.
ronversion and formation of each paraffin from one ?'he kinetics are expressed by fifty-three first-order
to ten carbon atoms, each naphthene from six to ten reaction steps, which are given in Table III. I n each
carbon atoms and each aromatic from six to ten car- reaction step, P, N, or A denotes paraffin, naphhene,
bon atoms. Isomcr distribution within each carbon- or aromatic and the subscript denotes carbon number.
number group is not accounted for. The constant for the step is the coefficient in the rate
Three sequences of reactions have been considered. equation expressed in the form:
Paraffins hydrocrack to paraffins of lower carbon num-
ber and cyclizc to naphthenes of the same carbon
number; cyclization of hexanes is neglected. Naph-
thenes hydrodecyclize to paraffins of the same carbon
number, dehydrogenate to aromatics of thc same rar-
bon number, and dealkylate to naphthenes of lower
48 FIFTH WORLD PETROLEUM CONGRESS

IO 0 I I 1 I I TABLE III
80 n- HEPTANE - REACTION STEPS AND THEIR RATE CONSTANTS
Rate Constant, Rate Constant,
Reaction Step k x 10* Reaction Step k x IO2
O Pio 3 Ni0 2.54 Px + Na 1.33
W Pio 3 PO + P i 0.49 Pa + P? f Pi 0.19
Ps -+ Pa + Pz
w
U Pi0 + Pa + P2 0.63 0.26
Pi0 + P7 + P3 1.09 PS + + P3
P6 0.43
+ Pa + P a + 2P4
W
z Pi0 PI 0.89 0.35
;
I Pi0 + 2P3 1.24
a N n -+ Pa 0.47
w
I Nio 4 Piu 0.64 Na j N? 0.09
l I l l l I 1 Niu 3 N s 1.34 Na + A8 21.60
c
Ni" 3 Nh 1.34
AH+ Pa
L
O Ni0 N7 0.80 0.16
v> Ni0 + Ai0 24.50 As + A? 0.01
w Aio + Pio 0.16 P7 + N7 0.68
P? + P a t Pi
-I
O Aut + As 0.06 0.14
I Aiu + As 0.06 P7 -+ + P2
P5 0.18
O I O0 + A7 -+ f P3
2 1 I
CRACKED HEPTANES
l I I
~ O O P S I (I 1 Aio
Pa + Ns
0.00
1.81
P7
N? -+
P4
P7
0.32
0.20
LL BO Ps + Ps + P i 0.30 N7 -+ AT 9.03
W
a Ps 3 P? + P2 0.39
u> 60 Ps + Pa 4- P3 0.68 A i + P? 0.16
W
Ps + Ps + P4 0.56
Pa -+ Ne
-J
O 40 0.00
2 Ns + Ps 0.64 Pa + + Pl
P6 0.14
20 N8 + Ns 1.27 Pa + +Pz
P4 0.18
Nn + N7 1.27 Pa + 2P3 0.27
Nn + As 24.50
O
Ne + Pe 1.48
As + Ps 0.16 Ne -+ Aa 4.02
As + An 0.06
As + A7 0.06 A0 + NB 0.46
Ps -+ +
P 4 Pi 0.12
20 P; + +P3 P2 0.16
/g-v-v -c

O This rate equation describes how heptane is formed

O 10 20 30 40 50 60
from hydrocracking of paraffins of higher carbon num-
IOO/MOLAL SPACE VELOCITY
ber, from hydrodecyclization of naphthenes and aro-
925 'F. 5 : l HYDROGEN : HYDROCARBON )
matics of the same carbon number, and how it disap-
FIG. 9-Agreement Between Experimental Data and pears by hydrocracking to lighter paraffins and by
Calculated Conversions of n-Heptane.
cyclizing to naphthenes.
The rate-equation coefficients were determined pri-
where R denotes moles of reacting component per mole marily from the data on reforming whole naphtha
of liquid feed, A/SV denotes activity/space velocity, under adiabatic conditions. Figure 10 compares com-
and k is the rate constant. In this form of the equation, positions calculated from the rate equations to those
the coefficient k includes factors for the effects of tem- observed when reforming Mid-Continent naphtha.
perature, pressure, hydrogen rate, and molecular 'The calculated compositions were obtained by numer-
weight of the feed. ical integration of the equations on an IRM 705 digital
The equations for all the reaction steps are combined computer. T h c calculated compositions agree well with
into twenty simultaneous differential equations, which the observed; the maximum deviations occur in pre-
comprise the mathematical model. A typical equation dicting heptanes at intermediate severities and pen-
is that for heptanes: tanes at high severity.

dP7
--
dA -
0.0109 Pl, + 0.0039 P, + 0.0019 P, + 0.0020 N, + 0.0016 A, - 0.0122 P7
-
sv
 REACTIONS I N CATALYTIC REFORMING OF NAPHTHAS 49

c5

LIGHT PARAFFINS ONE-CARBON

IO

I I I l I I I I I
o 25 I
W I I I I I I I I

c
W
LL

-I
HEAVY PARAFFINS
LL
O
u)
20
W
J
O
z
O
9
SEVEN-CARBON
a
W
a
u)
-- --
W
-I
O 5
5

O
25 I
I l I 1 l
y
I
o
AROMATICS

'lr,
O
O IO

FIG. 10-Agreement
'NINE

20
~
AND HEAVIER

30
I

40
I o

50 60
I

ACTIVITY / SPACE VELOCITY

SIX-,C~BO~L

70 00

Between Experimental Data and Calculated Conversions

of Whole Naphtha.
I ~

90
I100
50 FIFTH WORLD PETROLEUM CONGRESS

An application of the kinetic model that is of par- Selection of the best feeds for reforming is an im-
ticular interest to the refiner is the calculation of portant economic factor in refinery operations. Selec-
reformate yields at various octane numbers. Through tion has been greatly facilitated by the quantitative
blending corrclations similar to those that have been knowledge of reaction patterns of components in
developed [6] octane number can be related to aro- naphtha.
matics concentration. Calculated and experimental In Ultraforming and Powerforming, continuous
yield-octane relationships are shown in Figure 11 for 'catalyst regeneration without interruption of naphtha
the Mid-Continent naphtha and for a Kuwait naphtha processing provides a wide latitude in choice of pres-
with an ASTM boiling range of 125 to 360°F. The sure, terperature, and hydrogen :hydrocarbon ratio.
Kuwait naphtha contains about 20 per cent Pentanes Knowledge of the effects of such variables on yields,
and hexanes and about 70 per cent total paraffins. catalyst requirements, and equipment needs permits
Agrcement of calculated and experimental results is choice of the best operating conditions for each re-
good. I n the calculation for Kuwait naphtha, the finery situation.
kinetic model is applied to a feed of greatly different Planning studies can be made much easier by includ-
composition than the Mid-Continent naphtha used to ing the kinetics correlation in a mathematical model of
establish the rate constants. catalytic reforming combined with other processes.
Many process schemes, including recycle or reproces-
951 I I I I I I sing of portions of reformate, can be evaluated rapidly
œ by a digital computer; lengthy and expensive experi-
E mental work is thus avoided.
3
w
L
90- For any conversion process, evaluation and planning
O m can be made with greater confidence and accuracy
z
d when the calculation procedures recognize funda-
(JY 85-
W
z mental molecular changes. Designs and operation of
;3 other refinery processes that now depend upon gross
2
w empirical correlations would benefit by this approach.
80-
I
L
O
I- Acknowledgement
z KUWAIT
0 75- ï h e authors gratefully acknowledge the efforts and
œ contributions of many individuals in the Standard Oil
W
a
W

2
70-
\ Company (Indiana), the Esso Research and Engi-
neering Company, and the Humble Oil and Refining
-J
O Company.
>
65 L I 1 I I 1 I
References
1. Haensel, V., and Donaldson, G . R., Industrial and
Engineering Chemistry, 43, No. 9, 2102 (Septem-
ber 1951).
2. Heineman, H., Mills, G. A., Hattman, J. B., and
Conclusion Kirsch, F. W., Industrial and Engineering Chem-
The fundamental information established by these istry, 45, No. 1, 130 (January 1953).
studies has been used to advantage in present opera- 3. Hettinger, W. P. Jr., Keith, C. D., Gring, J. L., and
tions and future planning involving over 500,000 bar- Teter, J. W., Industrial and Engineering Chem-
rels per day of Ultraforming and Powerforming istry, 47, No. 4, 719 (April 1955).
capacity. Much of the new knowledge should also 4. Mills, G. A., Heineman, H., Milliken, T. H., and
broaden understanding of catalytic reforming in Oblad, A. G., Industrial and Engineering Chem-
general. istry, 45, No. 1, 134 (January 1953).
 REACTIONS I N CATALYTIC REFORMING OF NAPHTHAS 51

5 . Donaldson, G. R., Pasik, L. F., and Haensel, V., portement des produits purs s’est étendue ensuite à
Industrial and Engineering Chemistry, ‘47, No. 4, celle de naphta entier et a servi à en poser les bases.
731 (April 1955). L’interprétation des résultats obtenus sur heptane
6. Smith, R. B., presented before National Meeting et sur naphta entier a été effectuée suivant un modèle
of the American Institute of Chemical Engineers, cinétique complexe impliquant 53 réactions dans le
Salt Lake City, September 21-24 (1958). second cas et cela malgré des simplifications impor-
7. Forrester, J. H., Corin, A. L., and Malloy, J. B., tantes comme le fait de négliger la distribution en iso-
Oil and Gas Journal, 52, No. 49, 139, April 12 mères.
(1954). O r il semble, d’après les travaux antérieurs et les
8. Roberts, J. K., Thiele, E. W., and Shankland, résultats mêmes des essais des auteurs, que, dans cer-
R. V., Proceedings of Fourth World Petroleum tains cas, cette simplification devrait conduire à des
Congress, June 6-15 ( 1955). divergences importantes et ne pourrait être justifiée.
9. Petroleum Refiner, 37, No. 9, 223 (September C’est le cas des dérivés naphténiques à cycle 5 et de
1958). ceux à cycle 6 qui ont un comportement notablement
10. Holt, P. H. II, and Haig, R.R., Petroleum Refiner, différent; ainsi, par exemple, le méthyl-cyclopentane
36, No. 9,221 (September 1957). peut, soit subir I’hydrodécyclisation, soit s’isomériser
11. Petroleum Refiner, 37, No. 9, 221 (September en cyclohexane avec déshydrogénation subséquente
1958). rapide en benzène, alors que le dérivé à cycle 6 ne
12. Seelig, H. S., and Malloy, J. B., American Institute subit pratiquement que la déshydrogénation en ben-
of Chemical Engineers Journal, I, No. 4,528 (De- zène. Une preuve supplémentaire de ce comportement
cember 1955). différent est le fait que l’analyse des naphtènes restant
13. Reitzer, B. J., and Brooks, J. A., presented before dans le produit réformé à faible sévérité indique une
National Meeting of the American Chemical So- composition à 90% d’alkylcyclopentane.
ciety, San Francisco, April 13-18 (1958). Or, malgré le fait que las auteurs n’ont pas tenu
14. Farkos, A., (Editor) “Physical Chemistry of Hy- compte de ces éléments, la concordance entre les
drocarbons,” Academic Press (New York), Vol. I résultats expérimentaux et ceux calculés est très bonne.
p. 363, 1950. Est-ce que cette concordance existe parce que les
15. Pitts, P. M. Jr., Connor, J. E. Jr., and Leum, L. N., valeurs des constantes des réactions au départ des
Industrial and Engineering Chemistry, 47, No. 4, naphtènes tiennent compte d’une certaine répartition
770 (April 1955). des isomères ou bien est-ce dU à une influence
négligeable de cette différence sur le résultat global?
This paper was presented on June 4, 1959, by O n peut, dans ce dernier cas, se demander si une
H. G. KRANE. simplification du modèle cinétique ne serait pas
Discussion possible et si elle n’aboutirait pas à une méthode de
C. THONON (Institut Français du Pétrole, Rueil- prédiction de l’évolution du processus de reforming
Malmaison, France). Atteindre au perfectionnement plus maniable que celle présentée dont la complexité
des procédés par une analyse fondamentale des réac- limite certainement l’emploi.
tions qu’ils impliquent est une voie bien ardue lorsqu’il D’autre part, il serait intéressant de savoir si la
s’agit d’un système aussi complexe que celui mis en méthode appliquée à d’autres naphtas que celui du
oeuvre dans le “reforming’’ catalytique et l’on doit Mid-Continent, qui a servi à son élaboration, con-
féliciter les auteurs non seulement de l’avoir tentéNe, duit à des résultats aussi concordants quant à la
mais encore d‘être arrivés à des résultats remarquables. distribution entre les divers constituants, l’exemple du
I1 faut souligner que cette étude n’est pas seulement brut de Kuwait ne permettant pas de le juger en
une analyse globale du processus de reforming mais première approximation.
est basée sur l’analyse des réactions composantes pour H. G. KRANEreplies. There are three questioni
en estimer les constantes cinétiques fondamentales lit asked. The first one: Is the agreement between the
leur synthèse ultérieure. data and the calculation duc to the fact that the
L’étude entreprise d’abord pour l’examen du coni- differences in C5 and in C, ring naphthenes are
52 FIFTH WORLD PETROLEUM CONGRESS

accounted for in the constants, or is it because these possibly for the heavy aromatic distribution between
differences are negligible in the actual reforming carbon number, where we could account for these
result? differences by different operating conditions under
I think it’s possibly a little bit of both. Actually, which this naphtha was formed.
the constants do represent the average behavior for A. L. SALUSINSZKY (Standard V a c u u m Refining
the type of naphthene distribution we see in the Co., Willamstown, Australia). The authors gave in
naphtha we study. We know there are variations; the paper the definition of severity as a fraction of
in many cases these are not large but nevertheless space velocity and catalyst activity. I wish to ask for
the agreement has always been good, and we suspect the exact definition of this fraction.
that as you go to the higher carbon numbers some H. G. KRANEreplies. With the catalyst used in
of these differences probably aren’t as great. that study, if it had zero coke on it, activity would
The second question is: Is there a simplification be 100 and space velocity would be the weight-
of the model possible so that it is easier to use? Yes, hourly space velocity, that is pounds of oil per hour
simplified models have been presented. This model per pound of catalyst. As coke is deposited on the
was designed specifically to answer certain questions catalyst, of course, the activity term gets less and
that required that we go to this degree of complexity. less.
We wanted to evaluate schemes where we would have R. J. HAWES(Tidewater Oil Co., Delaware City,
to recycle a reprocess reformate to see if there was Del., U.S.A.). I’d like to congratulate the authors
possibly any advantage for combination processing,
on an excellent work, excellently presented. I n the
and in order to evaluate these accurately we found practice of platinum reforming, I think we’ve all
that we had to be able to predict the composition noticed an increase in boiling range. This, I presume,
of the reformate in terms of the carbon number and is a very small part of the reactions, but I wonder
the carbon type, that is the PONA within each whether the authors would have any comments on
carbon number of the reformate products, and we the type of reactions that caused this apparent in-
could see no other way of doing it than to incor- crease in molecular weight.
porate each of the reaction steps involved in the
kinetic model. I would also be very much interested in the gen-
While it’s complicated, the ability to use high- eral type of analytical methods they use-not the
speed electronic digital computing makes its actual details, but the general type.
use quite rapid, and we found it convenient to use. H. G. KRANEreplies. With regard to forming small
The question was asked, how the results on other amounts of polymeric material, yes, these reactions
naphthas compare. The Kuwait naphtha is just occur; we could not account for them in the model
one example, this is true. We have compared the and we handled their effects in our reforming prob-
results on a yield-versus-octane basis on many differ- lems empirically, on top of what we have here.
ent types of naphthas ranging all the way from pure With regard to the analytical procedure, there are
paraffins to highly naphthenic stocks, and the cor- several different types; I’ll talk about the one type
respondence between calculated yields at various that was used and I think some of my co-authors
octane numbers has always been good. We do not might mention some of the other types. Where large
have a great many detailed analytical breakdowns quantities of naphtha are reformed, such as in the
of the products of different types of naphthas that whole naphtha study and the composition curves that
allow us to make the detailed product composition you saw, products in fairly large quantities were
comparison that we showed in the slide for the separated by silica gel to separate the aromatics frac-
Mid-Continent naphtha. We have, however, several tion from the saturate fraction. These in turn were
of these. We have them for light cuts for Mid- fractionated by very sharp fractionation into approxi-
Continent naphtha, in which the agreement is excel- mate carbon number cuts. These were analyzed by
lent-this is five different boiling range cuts-and mass spectrography and isomer distribution was ob-
we also have a comparison for a light Arabian type tained by ultraviolet and infrared spectrography;
stock, where the agreement was very good except also on some of the lighter ends, gas chromatography
 REACTIONS IN CATALYTIC REFORMING OF NAPHTHAS 53

was used. I think some of my co-authors might could be very major. But with cyclic regeneration
mention some other techniques. as we practice it, with a more or less average coke
F. KUNREUTHER (Shell Oil Co., New York, N . Y . level on the catalyst, this does not affect the accuracy
U.S.A.). I also want to say that the paper, I thought, of our calculations.
was an excellent presentation of a kinetic model of With regard to the activation energy, I think the
catalytic reforming. I would like to ask about two primary differences would be between the naphthene
points which appear to be simplifications, which dehydrogenation reactions and the hydrodecyclization
might not be justified in certain extreme cases, and and hydrocracking reactions as a whole, because these
I’d like to have the authors’ comments. two do not have far different activation energies.
We recognize that the model is not a universal one
The first one is that the change in activity of the
catalyst is incorporated with the space velocity. I and not applicable over an extreme temperature
range, but over ranges of interest we can make
think there is evidence that as activity decreases the
rates of various reactions decrease-not to the same certain fudge factor type of adjustments where we
can get by despite these differences, but they are two
degree-and, therefore, it seems to me that activity
very legitimate questions.
decline should be incorporated in the rate constant
equation. A. I. SNOW(Sinclair Research Laboratory, Harvey,
Ill., U.S.A.). The same treatment and the same
The second point is that the activation energy of kinetic constants were used for all stocks ranging
the various reactions is probably not equal, and
from pure paraffins apparently to full range feeds.
therefore using a single, average kinetic temperature You might expect the differences are due to competi-
in an adiabatic reaction system might not be justified tion of adsorption of various species on catalytic
when you are comparing two reaction systems in sites. I wonder whether the authors would care -to
which the temperature sequence over three or four comment with reference to this point. Are such ad-
different reactors is varied to a large degree.
sorptions considered unimportant in comparing, say,
H. G . KRANEreplies. These are very legitimate pure hydrocarbons individually and in the full range
comments and I think it emphasizes the point that feed?
there are always simplifications you make to meet H. G . KRANEreplies. These competitive adsorp-
the problem at hand. Rion effects do seem to be important, but note the
With regard to the activity term being different way the rate equations were written regarding pres-
for different reactions as your catalyst deactivates, sure effects-you will notice the rates were not par-
yes, this is true. However, in our processes we ticularly proportional to pressure, and we think the
operate with an average coke level on a catalyst situation that arises here is that you have a catalyst
where this is not particularly significant as far as the rather heavily covered with these hydrocarbons, so
selectivity differences of the reactions go. When that the adsorption effects depend upon the relative
you’re designing a reforming cycle where you’re partial pressure of these components in the gas phase,
going out for long times and you eventually stop and and we have essentially accounted for this in the way
regenerate because of a yield loss, these differences the rate equations are formulated.