Slmulation of Catalytic Reactions in Shale Oil Refining: Crudes. These

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Ind. Eng. Chem. Process Des. Dev.

7004, 23, 827-833 027

Medina, A. G.; Mccermott, C.; Ashton. N. Chem. Eng. Sci. 1979, 3 4 , 861. Toor, H. L. AIChE J . 1984~1,10, 448, 460.
Nagata, I. J . Chem. Eng. Data 1982~1, 7 , 360. Toor, H. L. AIChE J . 1984b, IO, 545.
Nagata. I. J . Chem. Eng. Data 1982b, 7 , 367. Weiiand, R. H.; Taylor, R. Chem. Eng. Educ. 1982, 16(4) 158.
Oldershaw. C. F. Ind. €ng. Chem. 1941, 13, 265. Wilke, C.R. Chem. €ng. Prog. 1950, 46(2) 95.
Sakata, M.; Yanagi. T. I . Chem. E. Symp. Ser. No. 56, 1979. 3.2121. Wilson, G. M. J . Am. Chem. Soc.1984, 86, 127.
Slattery, J. C. “Momentum, Energy and Mass Transfer in Continua”; McOraw- Young, G. C.; Weber, J. H. Ind. Eng. Chem. Rocess Des. Dev. 1972, 1 1 ,
Hili: New York, 1972. 440.
Smith, E. D. “Design of Equilibrium Stage Processes”: McGraw-Hill: New Zuiderweg, F. J.; Harmens, A. Chem. Eng. Sci. 1958, 9 , 89.
York, 1963; Chapter 15 by Fair, J. R.
Smith. L. W.; Taylor, R. Ind. Eng. Chem. Fundam. 1983, 22, 97.
Taylor, R.; Smith, L. W. Chem. Eng. Commun. 1982, 1 4 , 361. Received for review June 2, 1983
Toor, H. L. AIChE J . 1957, 3 , 198. Accepted November 28, 1983

Slmulation of Catalytic Reactions in Shale Oil Refining

Tek Sutlkno’
Midwest Research Institute, Kansas City, Missouri 6 4 110

Stanely M. Walas
Department of Chemical and Petroleum Engineering, University of Kansas, Lawrence, Kansas 66044

Process yield data from advanced catalytic reaction of Paraho shale oil have been simulated. The properties and
the required refining severity of shale 011 are different from those of petroleum crudes. The simulation was made
to evaluate if mathematical models developed for hydrotreating, hydrocracking, and fluid catalytic cracking of
petroleum-based oils can be applied for simulation of these reaction processes used for shale oils. Preliminary
results indicate that refining of shale oil by these processes can be adequately simulated by the models derived
for petroleum-based oils. However, further process data at several reaction conditions of a variety of shale oils
are needed to confirm the applicabilities of these models.

Introduction though several research investigations of shale oil refining


The use of shale oil as an alternate feedstocks for the have been conducted, these investigations are feasibility
production of transportation fuels has been the subject of studies in nature, and experimental yield data from cat-
several investigations (Frost and Cottingham, 1974; Lovell alytic reaction processes of shale oil refining are very lim-
et al., 1981; Sullivan et al., 1978). It is generally concluded ited. A literature survey indicated that the pilot plant
that surface-retorted shale oil such as Paraho shale oil can refining data reported by Sullivan et al. (1978) are the most
be refined to various specification-grade transportation comprehensive at present; three pilot refining schemes
fuels-diesel fuel, gasoline, and jet fuel. The properties shown in Figures 1to 3 were used in this pilot plant study
of shale oil are, however, significantly different from pe- of shale oil refining. In light of their comprehensiveness,
troleum crudes, and the refining of shale oil is more com- those pilot plant data are used as the bases for present
plex and severe than that of petroleum. Advanced cata- simulation study.
lytic reaction processes are required to convert shale oil As shown in Figures 1to 3, the major catalytic reaction
to transportation fuels. processes in shale oil refining are hydrotreating, hydro-
Most of the published models for simulations of ad- cracking, fluid catalytic cracking, and reforming. Only
vanced catalytic reaction processes in oil refining are de- hydrotreating, hydrocracking, and fluid catalytic cracking
veloped from and verified with experimental data on pe- data are simulated in this study; the reforming process is
troleum-based feedstocks. These models can be used for not included because shale oil-derived naphtha is report-
the development of designs, operations, and refining edly identical with petroleum-based naphtha (Lovell et al.,
strategy development. Because of the specific properties 1981; Sullivan et al., 1978). Hydrotreating, hydrocracking,
of shale oil and the refining severity required, this paper and fluid catalytic cracking processes in shale oil refining
evaluates the applicabilities of the readily developed perform different functions, and simulation of each of these
models for simulation of catalytic reaction processes in processes requires specific mathematical framework. The
shale oil refining. This applicability evaluation is necessary mathematical models for simulations of these processes
if any of the petroleum refining reaction models is to be have been selected from the literature; these models and
used for designs, yield predictions, or optimization of the the results of simulations are described next in the order
respective reaction process in shale oil refining. of hydrotreating, hydrocracking, and fluid catalytic
Process models for catalytic reactions in oil refining are cracking.
generally complex and involve several parameters. The A. Hydrotreating
values of these parameters must be known if the model In refining shale oil to transportation fuels, the main
is used for simulation or yield prediction purposes. Al- purpose of hydrotreating is removal of nitrogen; therefore,
0196-4305/84/1123-0827$01.50/0 0 1984 American Chemical Society
828 Ind. Eng. Chem. Process Des. Dev., Vol. 23, No. 4, 1984

.C" Table I. Hydrotreating of Whole Shale Oil at Several


Product Nitrogen Levels
inlet nitrogen concentration 2.18% (wt)
Vrhoie catalyst: ICR 106, on stream time 1300-1800 h
reactor temperature, average 767 OF
total pressure 2200 psig
hydrogen partial pressure 1850 psia
total gas in 7000 SCF/bbl
recycle gas 5000 SCF/bbl
Outlet Nitrogen Concentrations at Various LHSV
outlet nitrogen concentration, ppm 640 30 1.3
LHSV, l / h 0.6 0.3 0.2
1 Battamr y Jet Fuei time on stream, h 24 48 48
Recycle

Gilbert, 19731, effective catalyst wetting model (Mears,


1 , C i i s butanes and iighter; Used for H2 plant feed, gasoline
blending, refinery f u e l . ond liquified petroleum gas (LPG).
19741, and others. Recently, a phenomenological model
2 . Some naphtha from hydrotreating units Qiso used for H2 plant feed. developed for trickle-bed reactors has been reported by
3. Foul gor and water treoted t> recover NH3 ond ruifur.
Crine et al. (1980). Because of the complexity of the
Figure 1. Shale oil refining-hydrotreating and hydrocracking trickle-bed reactor, some of the models are sophisticated
(Sullivan et al., 1978). and involve several multiphase transport phenomena and
reaction mechanisms. Verifications of these models are
difficult and require clear identifications of all parameters
affecting the performance of the reactor. However, con-
fusion among authors still exists with respect to the the-
Whole
Dewatered oretical interpretation of the experimental data [e.g.,
Marangozis (1980) and Shah (1980)l. In the present sim-
ulation of a commercial hydroprocessing trickle-bed re-
actor, simple correlating models were used rather than
sophisticated ones because experimental data obtained
from real reactors often are not accurate enough to dis-
criminate between the prediction of simple and more so-
phisticated models.
Among the models which have been evaluated with re-
spect to the predictive accuracy, the effective catalyst
wetting model developed by Mears (1974) has been re-
1 . C i i s butanes and lighter; Used for H2 piant f e e d , gosoline ported to correlate the experimental data well [Montagna
b l e n d i n g , refinery fuel, and i i q u i f i e d petroleum gas ( L P G ) . et al. (1977) and Gianetto (1978)l. It has been adopted
2 . Foul gos and water treoted to recover NH3 and sulfur.
here for the simulation of shale oil hydrotreating. This
Figure 2. Shale oil refining-hydrotreating and fluid catalytic model is a semiempirical one, but no reliable theoretical
cracking (Sullivan et al., 1978). model is available. The model can be briefly described as
follows
Whole
Dewotered ;'4 dC - --k i W W
Hydro- C' _ (1)
treating
qGq dx LHSV
Goroi ine
where C = concentration, mol/cm3, x = dimensionless
distance, z / L , ki = first-order reaction rate constant, cm3
of liquid/(cm3 of reactor volume), h, 7 = catalyst effec-
tiveness, Aeff = A,/A,, dimensionless, where A , = wetted
area of catalyst and A, = total area of catalyst, and LHSV
= liquid hourly space velocity, cm3/h cm3 of catalyst. For
4 scale-up purposes
Refinery Fuel
Aeff = 1 - exp(-1.36GaL0~05WeL0~2(~c/6L)0~75)
(2)
1 . Ca i s butanes ond iighter; Used for hydrogen plant feed, gasoline where GaL = d:gpL/pL2, dimensionless Galileo number,
2.
b l e n d i n g , refinery f u e l , and l i q u i f i e d petroleum go' (LPG).
Foul go' and woter treated to recovery NH3 ond sulfur.
d, = diameter of catalyst particle, cm, g = acceleration by
gravity, cm/s2, p L = liquid density, g/cm3, WL = liquid
Figure 3. Shale oil refining-delayed coking and hydrotreating viscosity, g/(cm)(s), WeL = GL2dp/BLpL, dimensionless,
(Sullivan et al., 1978).
Weber number, G L = liquid mass velocity, g/(cm2)(h),BL
a process model capable of predicting the reduction level = liquid surface tension, g/s2, and 6c = critical value of
of nitrogen content is essential. In the pilot study of surface tension for a given cracking, g/s2.
Sullivan et al. (1978))surface-retorted Paraho shale oil was Combining eq 1 and 2 and integrating, one obtains
hydrotreated in trickle bed reactors in which the liquid oil
and the gaseous hydrogen flow downward through a fiied 1 n Ci
-z- kill
[l - ~ X ~ ( - K L ~ ~ ~ ( L H S (3)
V)~~~J]
Co LHSV
catalyst bed. For trickle-bed reactors used in the hydro-
processing of petroleum crudes, several process models where K = f(pL, BL, 6c, pL, d,) and L = reactor length, cm.
developed for performance prediction have been published. As shown in eq 3, three constants, (ki, 9,and K) must
Comprehensive reviews of these models are presented by be evaluated before the level of conversion can be com-
Satterfield (1975), Gianetto et al. (1978))and Shah (1979). puted. Two sets of conversion data at different liquid
Included in this review are the hold-up model (Henry and hourly space velocities are enough to evaluate these con-
Ind. Eng. Chem. Process Des. Dev., Vol. 23, No. 4, 1984 829

matical model capable of predicting the boiling range of


the product is essential.
Several publications on the modeling of hydrocracking
have appeared. A comprehensive review has been made
by Choudhary and Saraf (1975). Qader and Hill (1969)
showed experimentally that hydrocracking of gas oil fol-
lows fit-order kinetics. Qader et d. (1970) have reported
first-order rate constants for light and heavy petroleum
oils in the hydrocracking process. Although it is reasonable
to expect that the hydrocracking of the 650 to 850 O F
fraction of the hydrotreated shale oil will follow first-order
kinetics, verification is desirable and could be checked with
adequate hydrocracking test data of shale oil, but such
0 Hydrotreating of Shale Oil experimental data are not available.
at 767' F
0 Hydrotreating of Petroleum-
Zhorov et al. (1971) developed several chemical process
Bared Charge Stock10/ schemes to describe the hydrocracking conversion of pe-
troleum feedstock into gas, naphtha, diesel fuels, and
10 100 i ,000 10,000 residue. These process schemes contain stoichiometric
Calculated Unconverted N i t r o g e n , ppm coefficients describing the distribution of hydrocracking
Figure 4. Parity plot of model predictions vs. experimentalresults. products. At a given temperature, the stoichiometric
coefficients are relatively constant over a range of residence
stants since ki and 7 are lumped together. times. However, for more severe hydrocracking, these
In the hydrotreating of crude shale oil, Sullivan et al. coefficients vary significantly and have not been properly
(1978) reported the nitrogen content reduction data at a correlated.
fixed reaction temperature for three values of LHSV. Stangeland (1974) has developed a kinetic model capable
These data are shown in Table I. of predicting hydrocracker yields satisfactorily and is ac-
The two lumped constantrs, KL and kiq, were evaluated tually being used to simulate and optimize commercial
analytically from the fmt two conversion data (0.6 and 0.3 hydrocracker yields. This model is used here for simula-
LHSVs). The values of kill and KL0.4are solved to be 2.14 tion of shale oil hydrocracking. Stangeland (1974) assumed
and 4.2, respectively, and eq 3 becomes that the feed and product streams contain a continuum
of compounds that can be characterized solely by boiling
2.14 point. These compounds are then segregated into fixed
{I- e~p[-4.2(LHSV)~.~l}(4) boiling ranges of 50 O F each. Each is characterized solely
by the true boiling point (TBP) of the end of the range.
Note that the parameter K is lumped together with The change in the amount of the ith compound is the
since the length of the pilot reactor was proprietary and result of its cracking plus the contribution from the
K could not be evaluated separately. Equation 4 was used cracking of all heavier compounds.
to predict the nitrogen removal level at 0.2 LHSV; the
result is shown in Figure 4. It appears that the agreement - - -k,f,(t) + i-1
-dfi(t) CPijkjfj(t) (5)
between the experimental and predicted values is fairly dt j=l
good. Comparing the experimental and the predicted where f i = ith component with f i as the heaviest compo-
values of nitrogen removal in hydrotreating of petroleum- nent, ki = first-order rate constant, and Pij = fraction of
based charge stocks (Figure 4), it is clear that an effective light component f i formed by cracking a heavier component
catalyst wetting model can simulate hydrotreating of shale
oil as well as it simulates hydrotreating of petroleum. fr
Rewrite eq 5 in a single matrix differential equation
However, more experimental data with known values of
relevant parameters must be generated to evaluate the $0) = -(I - P)KF(t) (6)
individual values of the constants.
B. Hydrocracking where F ( t ) = vector containing the weight fractions of all
As shown in Figure 1, the heavy fraction (650 to 850 O F ) components I = identity matrix, P = lower triangular
of the hydrotreated shale oil was refined further in an product distribution matrix, and K = diagonal matrix
extinction recycle hydrocracking unit. Hydrocracking having the ki)s on the diagonal. If all cracking rate con-
convertes heavy hydrocarbons into high grade jet fuels or stants are different, the solution to eq 6 is
gasoline, reformer charge stock, and light gases. Pilot shale F(t) = DE(t) (7)
oil refining studies of Sullivan et al. (1978) indicated that or in the matrix form
hydrocracking is the only process scheme capable of ef-
ficiently maximizing jet fuel production; in the process of
Figure 1 the main purpose of the extinction recycle hy-
drocracker is to produce jet fuels from the heavy fraction
of the hydrotreated shale oil.
Hydrocracking of shale oil is conducted in a trickle-bed
reactor; thus, hydrocracking and hydrotreating processes
are hydrodynamically similar. However, the primary
function of hydrotreating is to remove impurities in the The procedure to compute Dij, Pij, and ki is rather
feedstocks while that of hydrocracking is mainly to upgrade lengthy and complicated, but it is described in detail by
the feedstocks by lowering the boiling range. The per- Stangeland (1974). Primarily, the model has three ad-
formance of a hydrocracker is generally based on the true justable coefficients: A, B, and C. Coefficient A relates
boiling point (TBP) of the product; therefore, a mathe- to reaction rate constants kj of each individual component;
830 Ind. Eng. Chem. Process Des. Dev., Vol. 23, No. 4, 1984

Table 11. Pilot Plant Hydrocracking of 650-850 OF Fraction 0.0, to 2.0 for C (Stangeland, 1974). The values of the
of Hydrotreated Shale Oil parameters used here for hydrocracking of hydrotreated
operating conditions shale oil are well within the ranges of those used suc-
catalyst ICR 106 cessfully to represent the hydrocracking yields of petro-
reactor temperature, average 785 O F leum-based feedstocks.
total pressure 2350 psig C. Fluid Catalytic Cracking
hydrogen mean pressure 2129 psia
total gas in 8700 SCF/bbl Catalytic cracking is a widely used process for converting
recycle gas, SCF/bbl 7300 SCF/bbl heavy oils into more valuable gasoline and lighter products.
liquid hourly space velocity 1.0/h Its role in the oil refining industry is described in a recent
per pass conversion 0.5351 review by Venuto and Habib (1979). In the pilot plant
C,+ products" accum wt, % study conducted by Sullivan et al. (1978), and 650 O F +
C4 group 2.26 fraction of the hydrotreated shale oil was catalytically
C5 - 175 O F 7.70 cracked to produce gasoline.
175 to 315 O F 19.90 Several process models for prediction of gasoline yield
315 to 560 O F 53.50 have appeared, for example by Reif et al. (1961), Schneider
a All temperatures shown are true boiling point temperature. et al. (1969), and others. A widely used model which is
capable of predicting the conversion level and the gasoline
0 Doto from Shale Gor Oil
A Data f r o m Petroleum Gar O i l %
yield was initially developed by Weekman (1968, 1969).
-Model Predictions This model was based on a three-lump cracking kinetic
scheme which includes a catalyst decay function. The
900
model has been shown to predict adequately the perform-
ances of various types of cracking reactors including fixed
bed, moving bed, and fluidized bed types. It has been used
as a mathematical framework to analyze many aspects of
the catalytic cracking process, for example, the effect of
molecular composition of feedstocks on cracking kinetics
(Nace et al., 1971), the effect of nitrogen poisoning and
recycle (Voltz et al., 1972),and the selectivity behavior of
catalytic cracking in various types of reactors (Weekman
and Nace, 1970). The model was also utilized by Paraskos
et al. (1976) and Shah et al. (1977) to describe the per-
formance of a transfer line catalytic cracking reactor with
0 IC 20 30 40 50 60 70 80 90 100
reasonable success. Elnashaie and Hennawi (1979) used
Y e d 3f Cq- V i r ic the same concepts to develop an integrated mathematical
Figure 5. Comparison of measured and predicted yields for ex-
model for parametric analysis of a steady state fluid cat-
tinction recycle hydrocracking of shale and petroleum gas oils. alytic cracking unit. Weekman (1979) has reported the use
of this model in the optimization ofcommercial catalytic
coefficient B relates to the product distribution function crackers.
( P J ;and coefficient C affects the product distribution and The three-lump model of Weekman has a key weakness,
determines the yield of butane. namely, that each time the charge stock is changed, a new
A computer program has been developed here to use set of rate constants and the catalyst decay function must
Stangeland's model for data simulation of shale oil hy- be determined from laboratory experimental data. This
drocracking. However, only one pilot run was made of is due to the fact that the model does not take detailed
hydrocracking of 650 to 850 O F fraction of hydrotreated account of the molecular compositions of the charge stocks.
shale oil, for which the operating conditions and product Since variation of compositions of charge stock is common
yields are given in Table 11. The data in this table were in refineries, the experimental efforts required to deter-
generated from a pilot hydrocracker operation in an ex- mine the constants in the model for each charge stock are
tinction recycle mode. In this mode, the fresh feed and quite expensive. In an effort to correct this drawback,
the recycle constitute the total feed to the reactor. Jacob et al. (1976) developed a ten-lump model which
The yield data in Table I1 were applied with the de- includes the cracking characteristic of the charge stocks.
veloped computer program to obtain the results shown in In the present simulation of shale oil catalytic cracking,
Figure 5. The TBP values were calculated from the the three-lump model is adopted since the avilable pilot
computer program with a specific set of values for A , B , plant data of shale oil catalytic cracking are not enough
and C. As shown in Figure 5, this specific set gives rea- to allow evaluation of the constants of the ten-lump model.
sonable agreement between the calculated and observed Details of the three-lump model are as follows. Week-
TBP values and was the best among several other sets man (1968) represented the cracking process by the simple
which had been tried. While the values of A , B , and C reactions
could have been determined by a nonlinear regression or
other optimization methods to best fit the experimental
data, these methods were not used because experimental
A, -
ko
a1A2+ azA3 (94
k2
data are too limited in number for statistically meaningful A2 ----* AS (9b)
evaluation of A , B, and C. As shown in Figure 5, the
agreements between the experimental and the predicted Al represents the gas oil charged, A2 represents the C5-410
values are fairly good for both shale and petroleum gas oils. O F gasoline product, and A, the butanes, dry gases, and
The values of the three parameters of the model are A = coke. For an isothermal, vapor phase, plug flow reactor
0.9, B = 0.75, and C = 0.01. For hydrocracking of feeds- with negligible intraparticle diffusion, the cracking reaction
tocks of petroleum source, the ranges of these parameters can be described in terms of two coupled continuity
are known to be -1.0 to 1.0 for A , -2.0 to 1.0 for B, and equations given below.
lnd. Eng. Chem. Process Des. Dev., Vol. 23, No. 4, 1984 831

Table 111. Fluid Catalytic Cracking of 650 OF+ Fraction of


Hvdrotreated Shale Oil
operating conditionsa
run no. 1 2 3
reactor temperature, "F 930 930 975
weight hourly space velocity, h 7.85 7.86 7.71
y1 and y2 represent the instantaneous weight fractions of catalyst oil weight ratio 6.71 4.39 7.10
gas oil and gasoline, respectively. VI is the vapor velocity severity 0.855 0.558 0.921
in ft/s. The rate terms, R1 and R2, can be expressed as conversion, w t % 69.25 61.25 76.97
Product Yields, Weight Fraction
Ribi,t) = & ~ ( t ) ~ i ~ (12) coke, butane, and lighter 0.2038 0.1699 0.2717
gasoline (C5 - 430) 0.4887 0.4426 0.4981
R2b2,t) = k24WY2 (13) gas oil remaining 0.3075 0.3875 0.2303
4(t) is the catalyst decay function. Equations 12 and 13 a Catalyst Davison CBZ-1.
can be normalized into dimensionless equations.
At the exit of a fluid bed reactor (x = l), with y1 Ix=o =
1, eq 21 can be solved as

By dividing eq 17 by eq 18, one gets


where

dY1 KO

p v = the vapor density, lb/ft3 at reactor conditions, VI =


With the boundary condition of no initial gasoline (yl
= 1, y2 = 0), the solution of eq 23 is
the reactor volume, ft3, Fo = the oil charge rate, lb/s, t,
= the vapor phase residence time, s, t, = the catalyst
residence time (s) under oil exposure condition, 0 = t 'It,
where t 'is the clock time, s, x = z / L , where z is the axial
distance (ft) in the reactor and L is the total reador length,
ft, p o = initial charge density at reactor conditions, lb/ft3
(vapor density if no reaction), ho= density of liquid charge where
at room temperature, lb/ft3, s = Fo/pLoVr,and kl = alko.
p,,'s in eq 14 and 15 strictly should have been p:s; how-
ever, the model assumes the vapor density variation due
to cracking reaction is absorbed in the second-order
cracking reaction approximation. For steady-state oper-
ation the time derivatives of eq 14 and 15 are zero.
Moreover, 4 can be adequately expressed as
4(t) = (16)
A combination of eq 22,24, and 25 can be used to predict
CYis the decay velocity constant. Equations 14 and 15 can yield fractions from a fluid bed catalytic cracker. However,
be written in eq 17 and 18 by transforming du = @/s dx the constants in eq 22 and 24 must be evaluated first if
dyl/du = -K0I2 (17) they are used for yield prediction. Two conversion data
points are necessary for evaluation of all constants in the
dy2/du = KlY12 - K2Y2 (18) models.
A total of five runs was reported by Sullivan et al. (1978).
where KO= P & O / ~ L ~K1
, = P O Q . I ~ O / P L ' ,K2 = P O ~ ~ P L ~ . Three of these runs were conducted with a charge stock
The stretched time u is a function of the nature of the of 1300 ppm nitrogen content. The other two runs were
catalyst decay function and the catalyst residence time. made with two charge stocks with nitrogen contents of 385
With the assumptions of perfectly mixed catalyst flow and ppm and 870 ppm. These last two runs were made at the
plug gas flow in the fluidized bed reactor, the average same conversion level and therefore could not be used for
velocity constant (k)can be defined as the evaluation of constants in the model. Moreover, con-
version level in catalytic cracking process is dependent
k = k o ~0 m I ( 0 ) ~ d
(BO ) (19) upon the inlet nitrogen content, and a different level of
nitrogen content may require a different set of values for
The internal age distribution I(0) for a perfectly mixed the kinetic constants.
vessel is e"; 4(0) in eq 19 is e-x0with X = at,. k can be The present data simulation of shale oil catalytic
solved as cracking are based on the three runs with 1300 ppm ni-
trogen charge. Table I11 displays the operating data and
the product yield fractions.
The data from run no. 1and 2 were used to evaluate the
With this expression for the average reaction velocity constants of the three-lump model. Because some of the
constant, eq 14 can be solved as constants in the model cannot be solved explicitly and the
integration must be made numerically, a computer pro-
gram using a successive substitution method was developed
to evaluate the constants. The values of the various con-
832 Ind. Eng. Chem. Process Des. Dev., Vol. 23, No. 4, 1984

B = a parameter in Stangeland's hydrocracking model, di-


-
I , 2 . 3 Run Numberr in Catalytic
mensionless
Crocking of Shale Bared Go5 O i l
0 Gar Oil C = a parameter in Stangeland's hydrocracking model, di-
0.81 A Gasoline mensionless, or concentration, mol/cm3
Gasoline Y i e l d Doto from
E 0.71
0
Petroleum Eared Gar O i i w
D = production distribution matrix defined in eq 7
u. D,, = coefficients of matrix D
5 0.61
i
$$
2.00
/ d = diameter of catalyst particle, cm
8 = oil charge rate, lb/h
F ( t ) = vector containing the weight fractions of all compo-
nents, dimensionless
f , = weight fraction of ith component in hydrocracking feed
f 0.3
= 205.4/hi or hydrocrackage, dimensionless
= 77, l/hr
i 0.2 = 35.9/hr
Ga = dimensionless Galileo number
= 0,23/hr GL = liquid mass velocity, g/cm2 h
g = acceleration by gravity, cm/s2
00 0.1 0.2 0.3 0.4 0 5 0.6 0.7 0 8 0 9 1 00 Z = identity matrix
Predicted Product Weight Fraction
Z(8) = internal age distribution function of catalyst, dimen-
sionless
Figure 6. Parity plot of model predictions vs. experimental resulta. K = diagonal matrix having the ki's on the diagonal, or a
constant defined in eq 3
stants evaluated are shown in Figure 6. With the con- KO = (poko)/pLo,gas oil rate constant, l / h
stants evaluated from run no. 1and 2, the model was used K l = (Poalko)/PL
to predict the yield fractions from run no. 3. The exper- k, = mean average reaction velocity constant of ith reaction
imental and the calculated values are shown in Figure 6. ki = reaction rate constant of ith reaction in both hydro-
It can be seen that the calculated conversion is lower than treating and hydrocracking models; k, is the reaction ve-
the experimental value. This is expected because the re- locity constant of ith reaction at 8 = 0 in the model for fluid
catalytic cracking
actor temperature in run no. 3 is 45 O F higher than the L = reactor length, cm
reactor temperatures of the first two runs from which the LHSV = liquid hourly space velocity, l / h
constants were evaluated. The cracking reaction is en- P,,= a distribution parameter of product from hydrocracking
dothermic and high temperatures should increase the rate RIP2= kinetic rate functions as defined in eq 14 and 15
constants. Compared to the calculated and experimental r1 = Kl/KO
value agreements obtained from the three-lump model for r2 = &/KO
petroleum-based charge stocks (Figure 6 ) )the agreement s = F o / p L o V,, liquid hourly space velocity based on liquid
for shale catalytic cracking seems reasonable. However, charge density and reactor volume, l / h
more experimental data with better accuracy and con- t ' = clock time, s
sistency are definitely required to confirm the applicability t, = catalyst residence time, s
t, = vapor phase residence time, s
of the three-lump model or even the ten-lump model for U, = vapor velocity, ft/s
yield predition of shale oil catalytic cracking. V, = reactor volume, ft3
D. Conclusion W e L = dimensionless Weber number
x = dimensionless distance
While the selected models seem to simulate well the y 1 = instantaneous gas oil weight fraction
hydrotreating, hydrocracking, and fluid catalytic cracking y 2 = instantaneous gasoline weight fraction
process data of Paraho shale oil, the present applicability y 3 = instantaneous weight fraction of light gases and coke
assessment of these models for shale oil refining must be z = axial distance in the reactor, ft
at best regarded as preliminary in nature. Constants or Greek Letters
coefficients in the selected models have been evaluated
from the limited refining data of Paraho shale oil. New
values of these constants or coefficients may be necessary
a = decay velocity constant, 1 h
6 = liquid surface tension, g/s I
6, = critical value of surface tension for a given packing, g/s2
to fit the refining data of other shale oils with chemical e = instantaneous weight fraction converted
composition significantly different from Paraho shale oil. 8 = t'/tc
Moreover, the present simulation study has been limited X = extent of catalyst decay group, at,
to a single temperature of each individual reaction process. 9 = catalyst effectiveness, dimensionless
Thus, for simulation of similar reaction processes in re- p L = liquid density, g/cm3
fining of shale o& from other sources or at other operating pLo =. liquid density at room temperature, lb/ft3
conditions, further simulation studies with more well-de- p o = initial charge density at reactor conditions, lb/ft3
signed experimental data are needed to confirm the ap- p , = vapor density, lb/ft3
pL = liquid viscosity, g/cm s
plicabilities of the selected models. Nevertheless, the $ ( t ) = catalyst decay function, dimensionless
present work indicates that pilot data on hydrotreating,
hydrocracking, and fluid catalytic cracking of Paraho shale Literature Cited
oil can be closely simulated by available models. Choudhary, N.; Saraf, D. N. Ind. Eng. Chem. Prod. D e s . D e v . 1975, 1 4 ,
74.
Crine, M.; Marchot, P.; L'Homme. G. A. Chem. Eng. Sei. 1980, 35. 51.
Nomenclature Elnashaie, S. S. E.; El Hennawi, I . M. Chem. Eng. Sei. 1979, 3 4 , 1113.
Frost, C. M.; Cottingham, P. L. U . S . Bur. Mines Rep. Invesf. 1974, B W I -
A = a parameter in Stangeland hydrocracking model; di- 7844.
mension1ess Ganetto, A.; &Mi, G.; Specchi, 0 . ; Sicardi, S. A I C M J . 1978, 24(6),1087.
A, = gas oil weight fraction Henry, H. C.; Gilbert, J. B. Ind. Eng. Chem. Process D e s . D e v . 1973, 12,
A2 = gasoline weight fraction 328.
Jacob, S. M.; Gross, B.; Voltz, S. E.; Weekman, V. W. Jr., AIChE J . 1978,
A3 = butane, coke and gases weight fraction 22.
a1,.a2= coefficients of catalytic cracking reaction aa defined Lovell, P. F.; Fryback, M. G.; Reif, H. E.; Schwedock, J. P. Hydroc8rbon Pro-
In eq 9, lb of product/lb of feed cess. May 1981.
A& = ratio of wetted area to the actual surface area of catalyst, Marangozis, J. Ind. Eng. Chem. Process D e s . D e v . 1980, 19, 326.
Mears, D.E. A&. Chem. Ser. 1974, No. 133, 218.
dimensionless Montagna, A. A.; Shah, Y. T.; Paraskos. J. A. Ind. Eng. Chem. Process
dimensionless
E' = ( p , Vr)/(Fotc), Des. Dev. 1977, 16, 152.
Ind. Eng. Chem. Process Des. Dev. 1904, 23, 833-845 833

Nace, D. M.; V o k , S. E.; Weekman, V. W., Jr. Ind. Eng. Chem. Process “Refining and Upgrading of Synfuels from Coal and Oil Shales by Ad-
Des. Dev. 1971. 70. 530. vanced Cataiytic Processes,” First Interim Report, U.S. Dept. of Energy,
Paraskos, J. A.; S&h, Y.T.; McKinney, J. D.; Carr, N. L. Ind. Eng. Chem. April 1978, UC-gOd, FE-2315-25.
Process Des. Dev. 1976, 75. 165. Venuto, P. 6.; Habiie, E. T., Jr. “Fluid Catalytic Cracking with Zeolite
Qader, S. A.; Hill, G. R. Ind. Eng. Chem. Process Des. Dev. 1969, 8, 98. Catalysts”; Marcel Dekker: New York, 1979.
Qader, S. A.; Singh, S.; Wiser, W. H.; Hill, G. R. J. Inst. Pet. 1970, 56(550), Voltz, S. E.; Nace, D. M.; Jacob, S. M.; Weekman, V. W.. Jr. Ind. Eng.
187. Chem . Process Des. Dev. 1972. 7 1 , 26 1.
Reif, H. E.; Kress, R. F.; Smith, J. S. Pet. Refiner 1961. 40(5), 237. Weekman, V. W., Jr. Ind. Eng. Chem. Process Des. Dev. 1968, 7, 90.
Salterfield. C. H. AIChE J. 1975, 27(2). Weekman, V. W., Jr.; Nace, D. M. AIChE J. 1970, 76(3), 397.
.
Schnaider, G. J.; Mukhim, I. I.; Chueva, M. A.; Kogan, Y. S. Khim Tekhno/. Weekman, V. W., Jr. Ind. Eng. Chem. Process Des. Dev. 1969, 8, 305.
Topi. hfasei. 1969, NO. 7 , OI. Weekman, V. W., Jr. AIChEMonog. Ser. 1979, 75(11).
Shah. Y. T. “Gas-Liquid-Soli Reactor Design”; McGraw-Hill: New York, Zhorov, Yu. M.; Panchenkov, G. M.;Tatarintseva, G. M.; Kuzimin, S. T.; Zen-
1979.
kovski, S. M. Int. Chem. Eng. 1971, 77(2), 258.
Shah, Y. T. Ind. Eng. Chem. ProcessDes. Dev. 1980, 79, 328.
Shah, Y. T.; Huling, 0. P.; Paraskos, J. A.; McKinney, J. D. Ind. Eng. Chem.
Process Des. Dev. 1977, 16, 89.
Stangeland, 6. E. Ind. Eng. Chem. ProcessDes. Dev. 1974, 73, 74. Received for review June 13, 1983
Sullivan. R. F.; Stangeland, 6. E.; Rudy, C. E.; Green, D. C.; Frumkin, H. A. Accepted January 16, 1984

Thermophysical Properties of Complex Systems: Applications of


Multiproperty Analysis

Michael R. Brul6”

Kerr-McGee Corporation, Oklahoma City, Oklahoma 73 125

Kenneth E. Starllng

University of Oklahoma, Norman, Oklahoma 730 79

The “therm-trans” method is presented for predicting characterization parameters (e.g., critical properties) required
in multiparameter corresponding-states correlations. Multiproperty analysis of thermodynamic and transport data
is used to extract the characterization parameters through simultaneous calculation of the property data by an
equation of state and a viscosity correlation. The property data needed are similar to the inspection data obtained
for pseudocomponent fractions resolved from standard TBPdistillation analyses. This new approach for determining
correlation parameters altogether circumvents the need for critical properties per se. Application to complex
petroleum and coal-liquid fractions is demonstrated.

Prediction of Thermophysical Properties the presence of heteroatomic and attached functional


A modified Benedict-Webb-Rubin (BWR) equation of groups such as -N-, -NH,and -OH. Such intermolecular
state has been applied to predict the thermodynamic behavior cannot be described accurately by conventional
properties of the simpler hydrocarbons found in petroleum three-parameter corresponding-states correlations.
and coal-derived fluids (Brul6 et al., 1979, 1982). This Characterization of Complex Fluids
three-parameter corresponding-states correlation has been The problem of characterization further compounds the
used in conjunction with a conformal-solution model to problem of correlation. A standard characterization pro-
describe the vapor/liquid-equilibrium (VLE) bahavior of cedure must be available for resolving a complex fossil
model multicomponent mixtures representative of the fluid, with hundreds of diverse distillable and nondistill-
actual systems found in fossil-fuels processing (Watanasiri able organic compounds, into a pseudocomponent mixture
et al., 1982; Brul6 et al., 1983; Brul6 and Corbett, 1984). with 20 or so fractions representing the overall properties
Equation-of-state characterization parameters and ideal- of the parent full-range fluid (Brul6 et al., 1981). Once a
gas thermodynamic properties have been estimated, by full-range fossil fluid is resolved into effective pseudo-
using empirical correlations, as functions of the average component fractions, characterization parameters that
measurable properties of fractions such as normal boiling enable properties correlations to predict accurate ther-
point and specific gravity (Brul6 et al., 1982). mophysical properties must be determined for each of the
Many problems remain in predicting complex-system fractions.
thermophysical properties. Correlations must be capable As complex fluids increase in molecular weight, the
of predicting properties of high-molecular-weight, multi- determination of characterization parameters at the normal
functional organic compounds at mild-to-severe operating boiling point becomes less feasible. This presents an im-
temperatures and pressures. Multipolar and associating passe, especially for many of the pure model compounds
effects exhibited by some coal fluids are attributable to used for correlation development, since only low-temper-
0196-430518411123-0833$0 1.5010 0 1984 American Chemical Society

You might also like