EXAMPLE 11.9.1.

INFLUENCE OF INTERNAL DIFFUSION LIMITATIONS IN CATALYTIC REFORMING

Catalytic reforming is one of the main operations of petroleum refining. It aims at boosting the
research octane number of a straight run naphtha from 50-60 to 90-100 and processes it with a large
molar ratio hydrogen/hydrocarbons in a 3- or 4-bed adiabatic reactor on bifunctional Pt/Re-alumina
or Pt/Sn-alumina catalysts. The process is globally endothermic and the flow leaving a bed is reheated
before entering into the next. The naphthenes are dehydrogenated into aromatics while the n-
paraffins isomerize into i-paraffins and also undergo dehydrocyclization into aromatics. Sotelo-Boyas
and Froment [2009] simulated a three bed adiabatic reforming reactor with radial flow, linking the
possibility of internal diffusion limitations to the various reactions. Interphase gradients were
negligible, given the high flow velocities. The naphtha feed contained some 200 components in the
range C5-C12. The reactions catalyzed by the acid alumina were written in terms of elementary steps
of carbenium ion chemistry, as explained in Chapter 2. The dehydrogenations of the naphthenes and
paraffins, catalyzed by the metal function, were expressed on the molecular level. The reaction
network was generated by computer, using the Boolean matrix approach illustrated in Chapter 2. The
number of independent rate parameters of the acid catalyzed steps was reduced through the single
event concept to 21, that for the reactions on the metal function to 24 (16 rate parameters and 8
adsorption equilibrium constants) by accounting for the nature of the C-atoms surrounding the double
bond. These parameters were derived from a detailed data base obtained by 24 experiments carried
out in the pressure range 5 to 9 bar and temperature range 400-500C in an isothermal fixed bed
reactor with plug flow and filled with catalyst crushed to 0.5-0.7 10-3 m, to avoid internal diffusion
limitations. At 500C the number of component responses per experiment was 56.

622 CHAPTER 11: FIXED BED CATALYTIC REACTORS

The reactor model is very similar to that presented in Example 11.9.1.B but the pressure drop is
calculated from the Ergun equation and there is no solid diluent. Two versions were considered: a
homogeneous and a heterogeneous, accounting for diffusion limitations in the catalyst particles with
diameter 3.2 10-3 m. The diffusion fluxes were expressed by the Stefan-Maxwell equations for
multicomponent mixtures or by the Wilke approximation. The flux equations were integrated by
means of orthogonal collocation, mentioned in Chapter 3. The set of fluid field equations was stiff and
was integrated using Gears routine. The number of continuity equations per phase (fluid or solid)
amounted to 69, representing the variation along the beds of 69 pure components and Groups of
Isomers (GOI) with equilibrium among their members. Fig. 11.9.1.C-1 shows the effect of the diffusion
limitations for just one of the reactions, the dehydrogenation of the feed component n-octane, in
terms of its effectiveness factor profile through the beds. This factor is much smaller than one and
increases towards the exit of each bed because the lower temperatures there affect more the reaction
rates than the diffusion rates. The increase is strongest in the first bed in which the fast endothermic
dehydrogenations are prevailing. Fig. 11.9.1.C-2 illustrates the difference between the simulations
with and without diffusion limitations for the total yield of the aromatics, naphthenes, C5+ n-paraffins
and C5+ i-paraffins. These quantities are calculated from the yields of the individual components. The
yield of naphthenes (kg naphthenes in the flow per 100 kg naphtha fed) is lower in the homogeneous
case than in the heterogeneous case because they are converted faster in the absence of diffusion
limitations. The opposite is true for the formation of aromatics which has an

BED 1 BED 2 BED 3 1

0.8

0.6

0.4

0.2

0 10 20 30 40 50

Space time, Kgcat hr/Kmol HC fed

Figure 11.9.1.C-1

Profiles of the effectiveness factor for the conversion of n-octane in a 3-bed adiabatic catalytic
reformer with radial flow. Sotelo-Boyas and Froment [2009].

11.10 TWO-DIMENSIONAL HETEROGENEOUS MODELS 623

BED 1 BED 2 70
ps.homog. 60 heterog.

50

BED 3

Aromatics

40

30

20

10 Naphthenes

C5+ iso-Paraffins

C5+ n-Paraffins

0 10 20 30 40 50

Space time, Kgcat hr/Kmol HC fed

Figure 11.9.1.C-2
Yield profiles in catalytic reformer with radial flow for various fractions. Dashed curve: pseudo-
homogeneous model; Full curve: heterogeneous model. Sotelo-Boyas and Froment [2009].

effectiveness factor exceeding one. These two figures suffice to illustrate the necessity of accounting
for these limitations to achieve accurate simulations and predictions for real processes. An approach
based on effective rates, implicitly combining reaction and diffusion, is not recommended. With the
levels of modeling and the computational means presently available there is no reason any more to
resort to this approach.

11.10 TWO-DIMENSIONAL HETEROGENEOUS MODELS

In Section 11.7.5 a pseudo-homogeneous model was presented and applied that accounts for radial
variations in the structure of a packed bed. The model can easily be extended to account for inter- and
intraparticle gradients of concentration and, if required, also for temperature gradients. Such a two-
dimensional heterogeneous model needs to distinguish between the fluid phase and the solid phase
as a whole not just a particle to correctly account for the heat transfer in radial direction. That
requires a distinction between the

effective radial conductivity fer for the fluid phase and that for the gas phase, ser [de Wasch and
Froment, 1971]. This concept of fer and ser was introduced as far back as 1950 by Singer and
Wilhelm [1950] but strangely enough all subsequent

work made use of the global er concept discussed in Section 11.7. The preceding

624 CHAPTER 11: FIXED BED CATALYTIC REACTORS

considerations led de Wasch and Froment [1971] and Papageorgiou and Froment [1995] to the
following mathematical model:

For the fluid phase:

Cfi Cfi 1 Cfi s

z z r r r r ers r gv si fi
 f cp uz Tf ur Tf r r r er f hf av ( ss Tf )

For the solid phase:

kgav (Csi Cfi ) s (1 i i

hf av ( ss Tf ) r r r er s s (1 ) ( Hi i i)

with the boundary conditions:

at z = 0:

at r = 0:

at r = Rt:

Cfi = C0i, Tf = T0

rfi ri 0, r r 0

rfi Csi 0, Tf = Ts = Tw.

These equations have to be solved simultaneously with the Reynolds averaged Navier-Stokes
equations and the Ergun-equation given in Section 11.7.5 to account for the radial void profile and
generate both axial and radial flow velocity components, ur and uz. In these equations it is assumed
that the heat transfer through the fluid and solid phase occur in parallel [de Wasch and Froment, 1971;
Dixon and Cresswell, 1979; Dixon, 1985]. fer and ser can be calculated from the correlations of de
Wasch and Froment [1972] and Zehner and Schlnder [1972]. The internal diffusion limitations appear
in those equations by means of the effectiveness factor , which is obtained by numerical integration
of the diffusion and reaction equations inside the particles, as discussed in Chapter 3.
Papageorgiou and Froment applied this model to the phthalic anhydride synthesis, discussed under
Section 11.7.5. The Figs. 11.10-1(a) and (b) show the radial conversion and temperature profiles at a
given axial distance in the

11.10 TWO-DIMENSIONAL HETEROGENEOUS MODELS 625

(a) (b) (c) Figures 11.10-1

Radial conversion (a) and temperature (b) profiles in two-dimensional heterogeneous model with
porosity and velocity profiles. (c) Comparison of radially averaged axial temperature profiles for 2D
pseudo-homogeneous and heterogeneous models. dt/dp = 5.8; Rep = 175 [Papageorgiou and
Froment, 1995].

reactor. The difference between the gas and solid temperature is only 2-3C, even for this very
exothermic process. This is generally so in industrial reactors because of the high flow velocity. Fig.
11.10-1(c) compares the radially averaged axial temperature profile of the 2D heterogeneous model
with that of the 2D pseudo-homogeneous model of Section 11.7.5. The difference between the two
simulations is significant.

The results shown here reflect accurately controlled operating conditions. An increase in the feed
temperature or in the temperature of the salt bath cooling system may lead to much more
pronounced profiles and differences between the two models discussed here. The model predictions
are very sensitive to the heat transfer parameters and the validity of some among these may be limited
to relatively narrow operating conditions. The detailed CFD models could contribute in the future to
checking and improving existing heat and mass transfer correlations derived in the sixties and
seventies on the basis of simple plug flow models. Given the irregular boundaries between fluid and
solid and the location of the heat source or sink inside the solid the task is extremely complicated.

626 CHAPTER 11: FIXED BED CATALYTIC REACTORS

PROBLEMS

11.1 (a) The kinetics of the catalytic reaction A R + S are given by

dx kKA[ A R (ps / K)] W (1 KA pA KR Ks ps )

FA0

The reaction is carried out isothermally in a packed bed reactor with plug flow at 275C. The feed
contains 0.155 mole water per mole reactant. Water is not adsorbed on the catalyst and acts purely
as an inert diluent. Given the following data,

Ft0 = 4.2 kmol/h (total flow rate) b = 1500 kg/m

dt = 0.05 m pt = 3 bar

K = 0.589 [bar]

k= 4.3593 [kmol/kg cat. h] KA = 0.43039 [bar -1]

KR + Ks = 2.8951 [bar-1]
calculate the length of the reactor required to reach an exit conversion of (i) 40 percent, (ii) 70 percent.

(b) Suppose the reaction is carried out under the same conditions in a multitubular reactor. The tube
length is 3 m. The total feed per tube is 4 kmol/h. An annual production of 20,000 metric tons of
product is required. The molecular weight of the product is 44. One year on stream is equivalent to
8000 h. Determine the number of tubes required to meet the production.

11.2 Discuss whether the following is correct or wrong:

(a) The relationship conversion versus reactor length is linear for a zero-order reaction without heat
effect carried out in an adiabatic reactor.

(b) In an adiabatic reactor the relationship of temperature versus conversion is a straight line for first-
order reactions only.

(c) Optimization of a multibed adiabatic reactor turns out to be roughly equivalent with distributing
the catalyst in equal amounts over the different beds.

11.3 (a) Consider an isothermal fixed bed reactor with axial mixing superposed on plug flow
conditions in which an irreversible first-order reaction takes place. Show that, for a given set of
operating variables, the effect of axial diffusion decreases with increasing reactor length.

PROBLEMS 627

(b) Given the values

us = 0.01 m/s dp = = 0.4 b = Pea = 2 k=

0.004 m 1200 kg/m

1 10-5 m/kg cat. s

Compute the CA profiles in tubular reactors with axial mixing as a function of total length and compare
with the plug flow profile.
(c) Verify for this case if a bed depth of 50dp is sufficient for eliminating axial mixing effects.

(d) On what basis has the 50dp rule been established?

(e) Explain why the concentration profiles under axial mixing conditions

never converge to the plug profile, not even in the limiting case L .

11.4 Calculate the heat transfer parameters of the two-dimensional pseudohomogeneous models
for the design of the reactor for hydrocarbon oxidation of Section 11.7.3, using the correlations given
in Section 11.7.1. Compare the value er calculated from the expressions given by (a) Kunii and Smith,
(b) Zehner and Schundler. Determine their sensitivity with respect to the solid conductivity. Additional
data:

s = g = p = = dp = dt = Tm = =

1.163 10 -3 kJ/m s K 4.99 10-5 kJ/m s K 0.8

0.95 0.003 m 0.0254 m 382C 0.38

11.5 Check whether multiple steady states can occur in the case of the

hydrocarbon oxidation of Section 11.7.3, but considering only the reaction A B. Required data:

-H = s = Tf =E =R = Cf = De =

1,285,350 kJ/kmol 1.163 10-3 kJ/m s K 655 K

1113,044 kJ/kmol 8.3144 kmol K

0.4145 10-3 kmol/m3 5 10-4 m2/h

628 CHAPTER 11: FIXED BED CATALYTIC REACTORS

11.62 A tubular, fixed bed catalytic reactor is to be used for a highly exothermic reaction, and the
preliminary design must consider the possibility of a hot spot. The following parameters have been
selected in the initial design estimates:

Tad 600 1.5 dimensionless adiabatic temperature rise ref

E 3200 dimensionless activation energy RTRe f 2 400

kBCA0V 3 A0

UA 22.5

reaction rate group or the number of reactor units

number of heat transfer units

(a) Will there be an excessive hot spot for the set of parameters given in the preliminary design?

(b) An obvious (if expensive) way to overcome a hot spot problem is to add diluents to the reactor
feed. What must the dilution be to achieve a design 10 percent safer than the hot spot minimum
dilution?

(c) Indicate how the two parameters of the hot spot analysis are altered by the following design
choices:

Decrease inlet (reference) temperature by 10 K. Decrease the tube diameter by 20

percent.

Increase the reactor length by 20 percent.

Change the catalyst to lower the activation energy by 40 percent. (d) The reactor designed in
part (b) is expected to undergo two changes

during extended operation: the catalyst activity will decrease by 30 percent and the heat transfer
coefficient will decrease by 20 percent. The loss in catalyst activity will be compensated by raising the
reactor temperature enough to keep kB constant. Investigate the probability of developing a hot spot
in the reactor under the revised conditions.

11.7 Fixed Bed Reactor for Styrene Production (a) Introduction

Styrene is produced by catalytic dehydrogenation of ethylbenzene on an iron catalyst. The reaction is
endothermic and reversible and takes place

2 This problem was contributed by Prof. J. H. Olson, University of Delaware.

PROBLEMS 629

with an increase in the number of mole. Consequently, styrene conversion is favored by high
temperatures, low pressures, and dilution of the feed by means of an inert component, such as
benzene or, more generally, steam. The steam also serves as a heat carrier, reducing the temperature
drop in adiabatic operation.

(b) Reactor

Operating conditions Molar feed rates:

Ethylbenzene: 707 kmol/h Styrene: 7.104 kmol/h Benzene: 0.293

kmol/h Toluene: 4.968 kmol/h Steam: 7777 kmol/h

Inlet temperature of the mixture: T0 = 620C Inlet pressure: 2.37 bar

Reactor dimensions Inner diameter: 7 m

(W / FEB )total = 62 kg cat. h/kmol Additional data

Catalyst equivalent diameter: dp = 0.0055 m Catalyst internal void fraction: s = 0.4 Tortuosity of the
catalyst: = 3

Catalyst density: s = 2500 kg/m3 cat. Void fraction of the bed: B = 0.4312

Bulk density of the catalyst bed: B = 1422 kg/m

(c) Kinetics

The main reaction is:

C6H5 CH2 CH3 C6H5 CH = CH2 + H2 (1)

3 parallel reactions occur:

C6H5 CH2 CH3 C6H6 + C2H4 (2)

C6H5 CH2 CH3 + H2 C6H5 CH3 + CH4 (3)

C6H5 CH = CH2 + 2H2 C6H5 CH3 + CH4 (4)

630 CHAPTER 11: FIXED BED CATALYTIC REACTORS

Reactions (1), (2), and (3) were assumed to be both catalytic and thermal. Reaction (4) was assumed
to be catalytic. Only the main reaction (1) was assumed to be reversible.

The following rate expressions were proposed for the above catalytic (rcj) and thermal (rtj) reactions
[Lee, W.J., Froment, G.F., Ind. Eng. Chem. Res., 47, 9183 (2008)]:

pST pH 2 1 EB EB

eq

c1 1 KEB pEB KH 2 pH 2 KST pST 2

k2KEB pEB

c2 1 KEB pEB KH 2 pH 2 KST pST 2

k3KEB pEB KH 2 pH 2

c3 1 KEB pEB KH 2 pH 2 KST pST 2

k4KST pST KH 2 pH 2

c4 1 KEB pEB KH 2 pH 2 KST pST 2

pST pH 2 t1 t1 EB

eq

t2 kt2 pEB
t3 kt3 pEB

where

ki Aexp RT

kti tiexp Rti

Kj Ajexp RTj

for i = 1, 2, 3, 4 and j = EB, ST, H2.

PROBLEMS 631

The rate parameters are given in Table 1, the thermodynamic parameters in Table 2.

TABLE 1

RATE PARAMETERS

TABLE 2
THERMODYNAMIC PARAMETERS

The following relation is used for the calculation of Keq:

oK exp RT

and

Keq K (Po )

where Po is the standard state pressure (1 bar) and i , with i the i

stoichiometric coefficients.

(d) Physicochemical data

Go Ho TSo

with

Ho H298.15 298.15CodT

So S298.15 298.15 Co dT
632 CHAPTER 11: FIXED BED CATALYTIC REACTORS

The heat capacities are calculated in the form of cubic functions of T.

Cpj aj bjT cjT2 djT3

The coefficients are given in Table 3 [Reid, R.C., Prausnitz, J.M., and Sherwood, T.K., The Properties of
Gases and Liquids, 3rd ed., McGraw Hill, New York (1977)].

TABLE 3

COEFFICIENTS IN THE CUBIC HEAT CAPACITY EQUATION AND STANDARD HEATS AND GIBBS FREE
ENERGIES OF FORMATION

The mixture viscosity can be calculated using Wilke's approximation. The viscosities of ethylbenzene,
styrene, benzene, and toluene can be calculated using the corresponding-states method of Thodos
[Perry, R.H., Green, D.W., Perry's Chemical Engineers' Handbook, 7th ed., McGraw-Hill (1997)]. The
viscosities of hydrogen and steam can be calculated using the Chapman-Enskog equation.

The inert steam being largely present, Wilke's equation can be used for the calculation of the diffusion
coefficients of the different species in the gas mixture. The binary diffusion coefficients can be
calculated from the atomic diffusion volumes, according to:

0.00143T1.75

AB 12 13 1 3 AB A B
(e) Reactor model

The reactor is assumed to be adiabatic with plug flow. Axial dispersion can be ignored. Mass transfer
limitations inside the catalyst pellet are to be accounted for by calculating the effectiveness factors.
Heat transfer limitations inside the catalyst pellet can be neglected. The catalyst activity is assumed
to be constant. Use the conversion of ethylbenzene and the

REFERENCES 633

conversions of ethylbenzene to the different products in the set of continuity equations. Use the Ergun
equation to describe the pressure drop.

(f) Simulation

Simulate the profiles of conversion into styrene, benzene, and toluene, and the temperature and
pressure profiles in the reactor.

11.8 Determine the units of the rate coefficients k1, k2, k3, and of the adsorption coefficients in
the rate equations of Example 11.9.1.A, Table 11.9.1.A-2.

11.9 (a) Repeat the calculation of the primary methane steam reformer presented in Example
11.9.1.A. Assume constant effectiveness factors based on the calculations shown in Figure 11.9.1.A-7.

(b) Repeat the calculation as in (a), but using the 2-dimensional reactor model of Section 11.10. Verify
the radial temperature uniformity in the reactor.