NAPHTHALENE VIA HYDRODEALKYLATION

Comparison of Pilot and Commercial Plant Data with an Analog Computer Model

C , R . A N D E RSS0 N , Sun Oil Go., .Warms Hook, Pa.

D . E, LA M B , University of Delaware, Newark, Del.

Production of naphthalene via hydrodealkylation in an adiabatic tubular reactor was simulated using an
analog computer. A mathematical model was developed based on available kinetic data and on pilot plant
results in which commercial plant charge stocks were used. Fifteen separate reactions were included in the
final model to describe the hydrodealkylation of commercial charge stock. Coefficients in some of the
kinetic equations were established by parameter estimation techniques using an analog computer. The
model was used to predict temperature distributions and reactor effluent concentrations for a commercial
plant in which unreacted product was recycled to the feed. The effects of reactor inlet temperature, hydro-
gen purity, and gas-oil ratio on plant operation were studied. Predictions of plant performance obtained
through use of the model were in accord with commercial plant operating experience.

N RECENT YEARS, increasing efforts have been directed toward
I analysis of process behavior for purposes of simulation and
optimization. Despite notable successes in this area, formid-
able difficulties often arise in the application of theoretical
techniques to practical process situations. These difficulties
may result from incomplete knowledge of fluid mechanics and
reaction kinetics, or, in some cases, from uncertainty of rate-
controlling steps. Establishment of reaction kinetics is par-
ticularly difficult in petrochemical processes because of the
presence of large numbers of compounds which may undergo
a variety of simultaneoiis and consecutive reactions.
T o overcome these difficulties, it is important that process
analysis together with process simulation be initiated in the
early stages of process development. Analysis and simulation
begun a t the bench scale level can lead to better pilot plant
design and can indicate the most significant experiments to be
performed with bench and pilot scale equipment. T h e
interplay of analysis, simulation, and experiment throughout
the development of a process can often result in significant
economies in the design and operation of a commercial plant.
A case history of the application of analysis and simulation
techniques to the development of the hydrodealkylation reactor
used in Sun OiP Co.'s naphthalene process is described in this
paper. Although the analysis is limited to steady-state con-
siderations, results have proved useful not only in design and
operation of the reactor but also in specification of control
devices and safety features.
Process Description
The salient features of Sun Oil Co.'s process for the produc-
tion of naphthalene from catalytic gas oil have been described
( 7 ) . T h e principal components of the process are shown in
Figure 1, and can be grouped into three categories: feed
preparation section, conversion section, and naphthalene
purification section.
In the feed preparation section catalytic gas oil is distilled
to produce a 440' to 515' F. true-boiling-point cut. An
aromatic concentrate containing essentially all of the mono-
methyl- and dimethyl-substituted naphthalenes present in the
gas oil is extracted from this cut. The extract is catalytically
desulfurized before being used as feed to the conversion section
of the process.
T h e conversion section contains the tubular hydrodealkyla-
tion reactor shown in Figure 2. Feed to the reactor consists
of extract from the feed preparation section of the process
mixed with the hydrogen-rich gas and a reactor recycle stream
containing unconverted methylnaphthalenes. A series of
reactions takes place within the reactor, which result in de-
alkylation of alkyl-substituted dicyclic aromatics to produce
naphthalene. Monocyclic aromatics in the reactor are simul-

HIGH CETANE
DIESEL FUEL GAS
COMPONENT MOLE 'A
H2
CH4
-- 35-80
20-60
ACID GAS
-TO SULFUR
RECOVERY FRESH FEED
COMPONENT
ALKYLNAPHTHALENES 40-80 LIQUID PRODUCT
ADIABATIC
NON-NAPHTHALENES:
ALKYL- NAPHTHALENE 30-70
BENZENES
HYDROAROMATICS
INDENES
SATURATES

NAPHTHALENE 0-25
ALKYL
H E A V ~FUEL NAPHTHALENE 75-100

Figure 1. Principal components of naphthalene process Figure 2. Hydrodealkylation reactor

VOL. 3 NO. 2 APRIL 1964 177

taneously dealkylated to yield primarily benzene, toluene, and
xylene. Adiabatic operation results in significant tempera-
ture rise through the reactor.
The reactor effluent is sent to the purification section of the
process. Here in a series of distillations the reactor effluent
is separated into high purity naphthalene, a recycle stream
containing unconverted methylnaphthalenes, high octane
gasoline, and a small amount of heavy fuel. The naphthalene
cut is clay-treated and then rerun to produce a product con-
taining 99+% naphthalene.
Model Requirements and Background Information
The process analysis described here is restricted to the
hydrodealkylation reactor. An objective of the analysis was
prediction of the variation in concentration through the
tubular reactor for each important chemical species together
with the temperature variation through the reactor. I t was
necessary to do this over a range of the following variables:
1. Naphthalene formers in charge 40-80 wt. yo
2. Hydrogen-oil feed ratio 7-1 5 moles/mole
3. Hydrogen feed stream purity 35-80 mole yo H2
4. Gas-oil feed ratio 13-25 moles/mole
5. Reactor inlet temperature 1100-1250' F.
6. Reactor residence time 5-30 sec.
T h e gas stream consisted of a mixture of hydrogen and
methane, whereas the oil stream consisted of substituted aromat-
ics which had been extracted in the feed preparation section
of the process. Optimization of both reactor design and
reactor operating conditions is greatly aided by prediction of
the effects of these variables on concentration and temperature
distributions through the reactor.
A second related objective consisted of calculating the
sensitivity of reactor effluent temperature and concentration
to changes in reactor inlet conditions. Knowledge of this
sensitivity is used in control system design and in specifying
conditions which must be met to ensure safe plant operation.
Development of a mathematical model to satisfy these ob-
jectives was complicated by two factors. First, a large number
of compounds are present in the reactor feed stream and a
variety of simultaneous and consecutive reactions takes place
in the reactor. Simplifications involving grouping together of
similar compounds are essential to obtain a useful model.
Second, the kinetic rate constants and heats of reaction were not
known for many of the reactions which take place in the reactor.
I t was therefore necessary to develop a model in which these
parameters were adjustable. Experimental pilot plant in-
formation was used to determine how complex the mathe-
matical model need be-Le., how many chemical reactions
must be included-and to estimate the reaction rate constants
and heats of reaction.
The pilot plant reactor consisted of a tube 2 inches in diam-
eter and 36 inches long. Adiabatic operation was obtained
by holding radial temperature gradients through the tube
wall close to zero by means of a series of electrical windings.
Composition of both reactor feed and product streams was
determined by distillation to obtain narrow boiling fractions
followed by vapor phase chromatography and low ionizing
voltage mass spectroscopy. Temperature distributions through
the reactor were obtained using a thermocouple which could
be moved along the longitudinal axis of the reactor.
Design throughput of the pilot scale reactor was 500 ml. per
hour of liquid charge a t 20 to 1 gas-oil mole ratio. Tempera-
ture and composition of feed to the reactor were varied, as
were gas-oil ratio, hydrogen purity, and reactor residence time.
The requirements which any mathematical model of a
process must satisfy are usually the result of compromise be-
tween rigor and model complexity or availability of basic
178 l&EC PROCESS DESIGN A N D DEVELOPMENT
process information. Requirements of the naphthalene reactor
model are summarized as :
Prediction of pilot plant temperature profiles to within =tl
standard deviation of the measured profile. In most cases
this is approximately 1 5 ' F.
Prediction of concentration of naphthalene and methyl-
naphthalenes in the pilot plant reactor effluent to within
f15% of their measured values.
Information upon which the model was based came both
from literature sources and pilot plant operation. Much work
has been reported on the thermal hydrodemethylation of
methylnaphthalenes and methylbenzenes (2, 3, 5, 8 ) . How-
ever, kinetic information for other hydrodealkylation and
cracking reactions which take place in the reactor is not avail-
able in the literature.
In early attempts to develop a satisfactory model it was
assumed that the kinetics of all reactions which do not lead to
production of naphthalene could be ignored, and that the
thermal effects of these reactions could be approximated by
assuming their rate of heat production to be proportional to that
of the naphthalene-producing reactions. Results of analog
computer simulation based on this simple model showed that
the model was incapable of adequately describing pilot plant
operation. Inclusion of the kinetics of nonnaphthalene-
producing reactions was required in the development of a more
sophisticated model to satisfy the requirements.
Model Development
The mathematical model of the hydrodealkylation reactor
was based on the following assumptions:
The reactor was assumed to be adiabatic.
Plug flow with no axial dispersion of heat or mass was as-
sumed throughout the reactor.
Physical properties of the reacting mixture, such as density
and heat capacity, were assumed constant throughout the
reactor.
I t was assumed that first-order kinetic expressions could be
used to describe the rates of dealkylation reactions which take
place in the reactor. This was possible because of the large
excess of hydrogen present in the reactor. The rate of hydro-
demethylation of monomethylnaphthalenes has been described
( 3 ) by the equation:
Although hydrogen is consumed during the course of reaction,
the change in hydrogen concentration is sufficiently small to
permit simplification of the rate expressions to the following
form :
ri = k,Zi (2)
Under no conditions was the change in hydrogen concentration
sufficiently great to cause errors of more than 1OYo in any of the
first-order rate constants. Silsby and Sawyer (8)have calcu-
lated the reaction rate constant in Equation 2 from toluene
demethylation data obtained a t various concentrations. The
rate constant was found to have essentially the same value for
all hydrogen-toluene ratios greater than 4.
I t was assumed that the set of reactions which take place
within the reactor can be represented by the reaction sequence
shown in Figure 3. These reactions are divided into three
groups. The first group consists of compounds which when
dealkylated yield naphthalene. In Figure 3 some of the
reactants in this group are lumped together and represented
by a single letter. For example, Crepresents the mole fraction
of the sum of 1,4-, 1,5-, and 1,8-dimethylnaphthalene plus some
other naphthalene-producing compounds. It was assumed
that these compounds dealkylate a t the same rate to produce
1-methylnaphthalene and similar compounds represented by
A , which upon further dealkylation yield naphthalene, shown
by N . Similarly, 2,3-, 2,6-, and 2,7-dimethylnaphthalene plus

Figure 3. Kinetic model
some similar compounds are represented by D. These com-
pounds were assumed to dealkylate a t the same rate to produce
2-methylnaphthalene and similar compounds, indicated by B.
1,2-, 1,3-, 1,6-, and 1,:-dimethylnaphthalene and similar com-
pounds are represenied by E, and ethylnaphthalenes with
similar compounds are represented by F.
The second group of reactions involves compounds which
dealkylate to yield compounds other than naphthalene. These
include monocyclic aromatics such as alkyl indans, indenes,
and benzenes. As in the case of the first group of reactions
it was assumed that all compounds represented by the same
letter react a t the same' rate.
T h e cracking reactions, Z + L, J -c L , and X + X',con-
stitute the third group of reactions shown in Figure 3. T h e
first reaction represents cracking of alkyl hydroaromatics to
form alkyl benzenes. The second reaction shows cracking of
alkyl indenes to produce alkyl benzenes. In the last reaction
small amounts of saiiirated hydrocarbons crack to produce
methane and ethane. I t was assumed that the rates of all
cracking reactions can be represented by first-order rate ex-
pressions. I t is known that the second and third groups of
reactions produce >70yo of the heat evolved in the reactor,
although the reacting compounds in these groups comprise
<50 mole yoof the oil in the reactor feed stream.
By making use of the above assumptions the following differ-
ential equations were derived to describe the variations in
concentration and temperature which occur along the length
of the tubular reactor.
(3)
(4)
dC
- = -,k1C (5)
de
(7)
dZ'
- = kgZ
de
where
ki = cy; exp (- 2)
and
P =G[?+,+l
011 1
Molar concentrations of hydrogen, methane, and oil a t the
reactor inlet are used to calculate the ratios in Equation 20.
Beta is assumed constant throughout the reactor. The in-
dependent variable 0, which appears in the differential equa-
tions, is the residence time of a differential fluid element in
the reactor. I t is defined as the distance which the element
has traveled from the reactor inlet divided by the fluid linear
velocity.
Specification of charge stock composition and reactor inlet
temperature determines the initial conditions required by
Equations 3 to 18. Specification of hydrogen-oil ratio and
hydrogen purity permits calculation of p. Solution of the
set of equations gives the concentration of each chemical species
shown in Figure 3 and temperature as functions of position
through the reactor. However, before solutions can be ob-
tained it is necessary to determine the values of all constants
which appear in the equations. These include the Arrhenius
pre-exponential factors, ai, the activation energies, Eai, and
the heats of reaction, (-A&)<. These constants must be
known for each of the 15 reactions shown in Figure 3.
Parameter Estimation
Estimation of the constants in Equations 3 to 19 constituted
the most difficult phase in the development of the model.
Values of some of the constants are available in the literature,
but others were obtained by parameter estimation studies
using an analog computer.
Activation energies for reactions involving hydrodea!kyla-
tion of aromatics have been reported by several investigators
( 2 , 3, 5, 8 ) . Silsby and Sawyer ( 8 ) report a value of 50 f 5
kcal. per gram mole for the hydrodealkylation of toluene.
Betts and Popper ( 3 )report 50.4 and 53.5 kcal. per gram mole
for the hydrodealkylation of 1-methyl- and 2-methylnaphtha-
lene, respectively. Although the activation energy depends
upon the type of aromatic which is dealkylated, and the
number, location, and length of the substituted alkyl groups,
the differences in activation energy are small. A single value
VOL. 3 NO. 2 APRIL 1964 179
 of activation energy equal to 52 kcal. per gram mole was used
for E l to Elz.
-4 higher activation energy is associated with the cracking
reactions. Sachanen (7) has reported an activation energy
of 65 kcal. per gram mole for the cracking of Cg to ClO paraffins.
Nelson (6) reports activation energies in the range of 60 to
70 kcal. per gram mole for the thermal cracking of gas oils,
naphthas, and recycle stocks. Accordingly a value of activa-
tion energy to 65 kcal. per gram mole was used for E13 to E I 5 . TEMPERATURE

T h e values reported for the rate constants ( 2 , 4 )for the hydro- FUNCTION GENERATOR
1
dealkylation of methylbenzene and the methylnaphthalenes e
were not directly applicable for this study, since none of the
components in Figure 3 represents pure components. Figure 4. Block diagram of analog circuitry
T h e values for the heats of reaction for ( - A H R ) ~through
( - A H R ) ~were calculated by group contribution methods as
given by Hougen and Watson ( 4 ) . Estimation of the remain-
ing heats of reaction was complicated by the fact that some of
the reactants shown in the reaction sequence of Figure 3 are
not pure compounds but rather are groupings of similar
compounds. Estimates of ( - A H R ) ~through ( - A H R ) ~ E were
obtained from the analog computer parameter estimation
studies described below. n
W
After a review of the literature 22 parameters remained to be N

specified: a1 to a15,and ( - A H R ) ~to ( -AHR)lg. If a n analytical

3 30
f
solution of Equations 3 to 20 were available, these parameters a
0
could in principle be established by substituting 22 sets of 5 20
t
reactor effluent concentration and temperature data into the 0
solution and solving for the unknown parameters. However, !j10
because of the temperature dependence of the reaction rate I
constants, Equations 3 to 18 are nonlinear and analytical 80
solutions are not available. The following procedure involv- 0 5 10 15 20
ing use of an analog computer was therefore adopted. RESIDENCE TIME, e (SECONDS)
Figure 5. Typical concentration profiles
Values of initial conditions and required by the model
were calculated for several pilot plant runs. After choosing
values of the undetermined parameters, Equations 3 to 20
were solved using a n analog computer. Values of concentra-
tion and temperature obtained from the computer solution
were plotted as functions of 0. The calculated temperature
c
profile was then compared with that measured in the pilot L' 1300
plant reactor. The concentrations computed a t the reactor e
*
outlet were compared with analyses of the pilot plant reactor w
K
effluent stream. By adjusting the undetermined parameters, 2
a n attempt was made to bring the computed results into 5
a 1250
agreement with experimental measurements. This was a W
n
tedious task and proved unsuccessful because of the large I
number of undetermined parameters and the complex coupling P
of the equations. The method might have been proved 1200
satisfactory had a n analog computer equipped with high
speed repetitive operation been used.
MODEL GENERATED PR
A new approach to the problem was then conceived. The 1150 1
energy and material balance equations were uncoupled in the 0 5 IO 15 20
computer circuit so that the temperature obtained from the RESIDENCE TIME, e (SECONDS)
energy balance was no longer used to calculate the reaction
Figure 6. Temperature profiles for fresh feed runs
rate constants. Instead, the reaction rate constants were
calculated using the experimentally measured pilot plant
temperature profile which was reproduced in the analog
computer by using a function generator. A block diagram
of the analog computer circuit showing the uncoupling is
presented in Figure 4. Table 1. Pilot Plant Operating Conditions
Uncoupling of energy and material balance equations per- Run No.
200B 200C 204 205
mitted the undetermined pre-exponential factors to be estab-
lished independently of the heats of reaction. Pre-exponential n, p.s.i.g. 400 400 400 400
0, sec. 20 20 20 20
factors in the analog model were adjusted to obtain agreement Moles Hl/mole oil 12.2 12.3 11.8 11.9
between the reactor effluent concentration distribution as Moles CH4/mole oil 11.9 11.9 10.8 10.8
calculated by the analog computer and as measured in a series Inlet temperature, O F. 1177 1200 1184 1205
Charge stuck Fresh aromatic 50% fresh extract
of pilot plant runs. The 15 pre-exponential factors were extract +
507' recycle
evaluated from 60 pieces of data obtained from seven pilot

180 I&EC PROCESS DESIGN A N D DEVELOPMENT

plant runs a t different :reactor inlet conditions. I t was possible conditions for these runs are given in Table I. Measured and
to determine some of the pre-exponential factors independently predicted product distributions in the reactor effluent for run
of others because some experimental reaction data involving 205 are compared in Table IV. Experimental and computed
pure components were available, and also because after de- temperature profiles are given in Figure 7 . .4greement be-
coupling of the energy and mass balances some of the reactions tween the model and pilot plant results for both the temperature
proceeded independently of others because of the consecutive profile and product distributions is very good and well within
nature of parts of the reaction sequence. Typical of the runs the model criteria specified earlier.
used are 200B and .%00C. Operating conditions for these T h e critical test of the model came when it was used to
runs are shown in Table I. predict the product distribution in the effluent stream of a
The undetermined heats of reaction were next established, commercial scale reactor. The excellent agreement between
by adjusting heats of reaction in the analog computer model to reactor effluent product distribution predicted from the model
bring the temperature profile computed from the energy and commercial operating experience is shown in Table V.
balance into agreement with the temperature profiles measured
in the plant reactor.
T h e product distributions of the naphthalene-forming com-
pounds in the reactor (effluent which were calculated from the Table II. Pilot Plant Data Used in Model Development
model are shown in Table 11. Agreement between pilot plant Compared with Model Results
product distributions and model results is well within the Fresh Aromatic Extract, Normalized
requirements specified for the rnodel. Similar agreement Mole 7 0
between model and experiment was obtained for the remaining 200B Product 200C Product
compounds indicated in Figure 3. Typical concentration Pilot Pilot
profiles obtained from the analog computer simulation are Model Component Charge plant Model plant Afodel
shown in Figure 5. A comparison of the temperature profiles hT 4.0 41.8 37.52 63.7 64.03
A 12.1 1 1 . 1 11.84 6.0 5.45
presented in Figure 6 again shows excellent agreement between B 25.5 34.7 36.56 26.0 26.65
model and experiment The slight differences in shapes of the C 1.6 0.45 0.11
measured and calculated temperature, profiles are probably D 23.3 9.37 3.34
d u e to a combination of factors. These may include small E 27.7 3.24 0.27
F 5.8 1.02 0.15
heat losses, axial diffusion, slightly nonuniform velocity
profile, presence of small amounts of reactive contaminants,
Sum ( C +
D f E +
F) 12.4 14.08 4 . 3 3.87
and approximations introduced into the model by grouping
Table 111. Kinetic Constants Used in Analog Model
of compounds haviq; similar reactive properties. These
factors were not included in the model in the interest of sim- Reference ai X 70-", Eq, ki at 1700' F., (-AHR)~,
Subsfript Set.-' Kcal. /G. Set.-' B.t.u./Lb. Mole
plicity and also because the model provides a prediction of reac- z (400P.S.Z.G.)Mole (400 P.S.Z.G.) Reacted
tor performance satisfactory for design purposes.
T h e activation energies together with the final values of the
pre-exponential factoris and heats of reaction for each of the
1
2
3
0.640
0.465
0.465
52i0 ;:y;;
0.00492 15,000
15,000
15.000
4 0,640 0,00492 15;OOO
naphthalene-forming reactions are given in Table 111. Also 5 0.424 0.00326 19,700
shown in Table I11 are the values of the reaction rate con- 6 0.249 0.00191 19,700
stants a t l l O O o F. calculated from Equation 19.

Table IV. Comparison of Product Distributions from Pilot

Use of Model for Prediction Plant and Model
T h e first test of the model was to predict results of pilot plant 50% Fresh Extract +
50% Recycle,
Normalized Mole 7%
runs which were not used in the parameter estimation studies.
205 Product
Pilot plant runs 204 and 205 are typical of those chosen for this
Pilot
purpose and were o n e in which the fresh feed to the reactor Model Component Charge plant Model
was replaced by a feed stream with a composition approximat- N 11 .o 48.8 45.52
ing that expected during operation with recycle. Operating A 17.6 10.7 11.25
B 37.9 33.6 35.21
C 0.9 0.26
D 13.3 5.33
E 16.0 1.85
F 3.3 0.58
-
Sum ( C +D f E f F) 6.9 8.02

Table V. Comparison of Product Distributions from Com-

mercial Plant and Model
Normalized Mole %
Component Charge Commercial Model
N 20.1 81 . O 81.54
A 9.2 2.3 1.90
B 28.1 15.1 15.49
C 1.2 0.02
I I D 17.2 1.02
11500L-----L- 5 IO 15 20 E 20.5 0.02
RESIDENCE TIME, e ( SECONDS) F
Sum ( C f D + E + F)
3.7
1.6
-
0.01
1.07
Figure 7. Temperature profiles for recycle runs

VOL. 3 NO. 2 APRIL 1964 181

 for any reactor feed stock by making use of the analog computer
model.

/
CHARGE: FRESH FEED 8.280 The model can also be used effectively to calculate the
magnitude of variations in reactor effluent concentration and
temperature resulting from small changes in inlet conditions.
When this information is expressed in terms of sensitivities or
steady-state gains, it is of great value in planning plant oper-
ating and in control system synthesis.

Nomenclature
= average heat capacity of reaction mixture, B.t.u./lb.
RESIDENCE TIME, e RESIDENCE TIME, e mole ’ F.
E, = activation energy, cal./g. mole
Figure 8. Reactor temperature and naphthalene cornposi- [Hz] = molar hydrogen concentration in Equation 1, Moles
tion vs. residence time for different values of B H ~ / c u ft.
.
k’ = 1.5-ordpr reaction velocity constant defined by Equa-
tion 1, (moles/cu. ft.) -1 12 sec. -l
The model was also used to determine plant operating k = first-order reaction velocity constant defined by
Equation 2, sec.-l
conditions which could lead to serious control problems. R = gas constant, 1.987 cal./g. mole K.
T h e highly exothermic nature of some of the hydrodealkyla- Y‘ = rate of reaction as defined by Equation 1, moles/cu.
tion and cracking reactions results in significant temperature ft./sec.
rise in the adiabatic reactor, and presents the possibility of Y = rate of reaction as defined by Equation 2, moles/mole
oil/sec.
runaway reaction. I t is essential that the upper temperature absolute temperature, K. ’
T =
limit of construction materials used in the reactor not be ex- t = temperature, O F.
ceeded. Factors which are important in determining the u = fluid linear velocity, feet/sec.
temperature rise through the reactor are: x = distance from reactor inlet, feet
[Z] = molar concentration of 1-methyl- or 2-methylnaph-
Numerical value of p thalene in Equation 1, moles/cu. ft.
Reactor inlet temperature Z = mole fraction of any of species pictured in Figure 3,
Oil feed composition moles i mole oil
Reactor residence time
HZ
oil
= hydrogen-oil mole ratio
Variations in /3 constitute one of the most serious threats to
satisfactory commercial plant operation. Temperature rise _H_4 -
C - methane-oil mole ratio
oil
in the reactor increases as (3 decreases. A decrease in p results
from a decrease in gas-oil ratio or a n increase in hydrogen
Hz
-
oil
. ~~
+ CoilH I = gas-oil mole ratio
~

.~~

purity. In a commercial plant sudden decreases in gas-oil a = pre-exponential factor in Arrhenius expression Equa-
ratio are possible in the event of compressor failure. The tion 19, set.-'
analog computer model of the reactor was used to determine p = defined by Equation 20,
the minimum values of p which could be tolerated for various (lb. mole total
charge compositions and reactor inlet temperatures. Tem-
perature and naphthalene concentration profiles through the
reactor for various values of p are shown in Figure 8. For the
case of fresh oil feed a t an inlet temperature corresponding to SUBSCRIPT
that in Figure 8, p > 2 p 0 was required to ensure that the reactor i = arbitrary index
outlet not exceed materials of construction specifications. A
somewhat lower value of is permitted when an oil feed- literature Cited
containing recycle is used mainly because of the lower con- (1) Barbor, R. P., Oil Gas J . 60,123 (May 21,1962).
centration of nonnaphthalene-forming hydrocarbons in the (2) Betts, W. D., Popper, F., Silsby, R. I., J . Appl. Chem. 7, 497
H (157)
feed. For this case the upper limit was not exceeded when \-’-’/.
(3) Zbid., 8, 509 (1958).
using p > 1.2p0. The effect of reactor residence time on (4) Hougen, 0. A., Watson, K. M., “Chemical Process Principles.
maximum temperature rise is less important because many of Part Two,” pp. 756-804, Wiley, New York, 1947.
the highly exothermic reactions go nearly to completion during ( 5 ) Morii, H., Hashimoto, A., Tominaga, H., Uneo, K., Bull.
Japan Petrol. Znst. 4, 28 (1962).
a single pass through the reactor. (6) Nelson, W. L., “Petroleum Refinery Engineering,” 4th ed.,
There is an economic incentive for making /3 low and inlet pp. 626ff, McGraw-Hill, New York, 1958.
temperature high in order to maximize conversion to naphtha- (7) Sachanen, A. N., “Conversion of Petroleum,” 2nd ed., pp.
73-111, Reinhold, New York, 1948.
lene and minimize reactor size and recycle flow. I t is therefore (8) Silsby, R. I., Sawyer, E. W., J . Appl. Chem. 6, 347 (1956).
necessary to find the optimum combination of inlet tempera- RECEIVED
for review June 6, 1963
ture, p, and residence time which maximizes conversion to ACCEPTEDOctober 24, 1963
naphthalene without risking the possibility of a runaway Chemical Reactor Simulation Symposium, 50th Meeting, Ameri-
reaction. Such an optimization study can easily be carried out can Institute of Chemical Engineers, Buffalo, N. Y . ,May 1963.

182 l & E C PROCESS D E S I G N A N D D E V E L O P M E N T