I-A METHANOL TO GASOLINE PROCESS

by
Sebastian Joseph
and
Yatlsh T. Shah
Chemical and Petroleum Engineering Department
University of Pittsburgh
Pittsburgh, PA 15261

CONTENTS
.__

Page
I-A

METHANOL TO GASOLINE PROCESS . . . . . . . . . . . . . . . . . . . . . .

I-!

SUMMARY

1-5

esmeeebueooaeeeowemoeeeeooeeQoeeleo~oeee~eeeoeo~o

Introduction

I-6

.ete,aeo~eomt~e,mo,oelte.oDe~,~eeleeo.oo~oo

Methanol to Gasoline Process

1-8

Reaction Mechanism ................................

I-|0

Fluldized Bed Reactor

1-13

Case Study

m e e l o o J m ~ o m e a l a e t

Fixed Bed Reactor

Case

o.,,,,eo,eeooleeeeooQooaeooeDeee~oo

, - ,

, . ,

,,

,,

, . .

1-25
1-29

u..,.I...ee~lee,eloee~og|ooeeelem6oooge

S t u d y ..........................................

1-41

Comparison of Fixed and Fluidlzed Bed Reactors ..........

1-45

Conclusions

1-54

Nomenclature

e.e,,,lJla,,.omoe,e.,.~e.mee,le,oeoooeo,eoee

References

e o o o e m u o a

, , e .

1-56
t

.o

. - e

Appendix I-A ............................................

I-2

, . o o e e

1-58
I-A-I

LIST OF FIGURES

Page
I-A-1

Gasoline from Coal or Natural Gas via M T G ...............

I-A-2

Schematic Representation of Fluidized Bed Reactor

and Regenerator System ..................................

1-15

Representation of Fluid Bed Reactor According to the

Model of Fryer and Potter ...............................

1-17

Typical Concentration Profiles of Reactants and

Products in Fluidized Bed Reactor .......................

1-27

I-A-5

Schematic Representation of Fixed Bed Reactor ...........

1-30

i-A-6

Concentration Profiles of Reactant and Products in

Fixed Bed Reactor .................................
. .....

1-43

I-A-7

Product Distribution of Hydrocarbons in Fixed Bed.......

1-44

I-A-8

Temperature Profile in Fixed Bed Reactor ...... . ...... ...

1-47

I-A-9

Conversion as a Function of WHSV for Fixed and

Fiuidized Bed Reactors ........................ . .........

1-49

STY as a Function of WHSF for Fixed and Fiuidized

Bed Reactors ............................................

1-50

Selectivity to Olefins in Fixed and Fluidized Bed

Reactors as a Function of W~SV ..........................

1-51

Selectivity to Olefins as a Function of Methanol

Conversion in Fixed and Fluidized Beds ..................

1-53

I-A-3

I-A-4

I-A-!0

I-A-If

I-A-i2

I-3

1-6

LIST OF TABLES

Page
I-A-I

Heats of Reaction for Major Reattion Steps in

Methanol Conversion to Hydrocarbons .....................

1-9

Properties of Cracking Catalysts Used in Cold Flow

Studies .................................................

1-22

Comparison of Experimental and Predicted Rise

Velocities ..............................................

1-23

Comparison of Experimental and Predicted

Conversions .............................................

1-26

Experimental and Predlcted Olefins Distribution,

mol % ...................................................

1-26

I-A-6

Effect of Bubble Diameter on Conversion .................

1-28

I-A-7

Effect of Temperature

1-28

I-A-8

Reactions in the Kinetic Scheme of Mihail, et.

a l . ( 1 9 8 3 b ) .............................................

1-32

Comparison of Experimental and Predidted Re8ults

for Fixed Bed Reactors ..................................

1-41

Experimental and Predicted Results for PDU ..............

1-42

I-A-2

I-A-3

I-A-4

I-A-5

I-A-9

I-A-10

on Conversion .....................

I-4

SUMMARY

lwo reactor types have been proposed for Mobil's Methanol to Gasoline
process - Fluldized Bed and Fixed Bed Reactor.

In this report, these two

reactor systems were modeled and computer programs developed for the reactor
models.

The fluidized bed was modeled using the Fryer-Potter counter current

backmixing model, while the fixed bed was modeled using a pseudo-homogeneous
reactor model.
plant data.

Results from the computer simulations were compared with pilot

Although the conversions were correctly predicted, the product

distributions were different due to the unknown Si02/AI203 ratio of the MTG
catalyst used in the pilot plant studies,

A comparison of the two reactor

types was carried out as regards conversion and selectivity to oleflns.

The

comparison showed that at the same WHSV (Weight Hourly Space Velocity) the
fixed bed reactor gave a higher conversion and Better selectivity towards
olefins than the fluldized bed.

I-5

Introduction
Mobil has developed a novel process for the conversion of methanol to
high octane gasoline in excellent yields using a shape selective zeolite
catalyst called ZSM-5 (Meisel et al. (1977)).

The coupling of this methanol-

to-gasoline (MTG) process with the well-established commercial technology for

the conversion of coal (or any carbon source) to methanol provides a new route
for the conversion of coal to gasoline (Figure 1-A-I).
MTG process is the ZSM-5 zeolite catalyst.

The key element in the

The pore structure of ZSM-5

consists of intersecting channels with diameters of about 6 A which are just

large enough to produce hydrocarbons boiling in the gasoline range.

Mobil has

proposed two reactor systems for this process-fixed bed and fluidized bed.
The first commercial plant for the MTG process is a 14,000 BPD gasoline plant
i n New Z e a l a n d ,

b a s e d on t h e f i x e d

bed r e a c t o r

concept

(Penick et al.

(1983),

and K a g g i n ( 1 9 8 5 ) ) .
The p u r p o s e

of the present

work i s

t o model t h e two r e a c t o r

systems that

are used for the MTG process, namely, the fluldized bed reactor and the fixed
bed reactor and to develop computer programs for these models.

The simulation

of these reactors can lead to an evaluation of reactor performance as well as

validating kinetic mechanisms developed for the MTG reaction.

The programs

are written in a modular form in FORTRAN and the codes have been implemented
on the University of Pittsburgh DEC System-f099 computer.

The programs can be

easily adapted to the ASPEN simulator by converting them into user models.
Included in this report are the model development for the fluidized and fixed
bed reactors, description of the numerical methods used to predict the
performance of the units, source code of the programs, instructions for their
use and a sample problem showing both the input data and the resulting output.

I-6

COAL

O}'Oea--

Cool

~{IIB

Synthedn
@on
{CO ~ H2)

MTO --~GASOUHE
Pre~nn

/~etheneH

Prec~

WeQor

I--t
!

~t~om

dATURALGA$~ 2ofnrming
~t~em
Protons

~ynthosln
, (~OS

(CO ~

~othonoO

~OW

MTG --~-GA$OUNE
Proco~

H2)
Wa~r

Figure .I-A-I:

Gasoline from co~l or natural gas via MTG'(from Penick e t a l . , 1983).

Methanol to Gasoline Process

In the MTG process, methanol is converted to hydrocarbons and water.

The

process is highly selective and can be represented by the following overall

reaction:

nCH30H

(1)

hydrocarbons + nH20

Quantitatively, I00 tons of methanol is converted to nearly 44 tons of

hydrocarbons and 56 tons of water.

The hydrocarbons produced contain 95% of

the energy in the methanol feedstock; the eothermic heat of reaction contains
the re,minlng energy.
The conversion of methanol to hydrocarbons over ZSM-5 can be represented
by the following sequence of steps (Chang and Silvestri (1977)):
-H20

2CH30H

CH3OCH 3

+H20

-H20
CH3OCH 3

C2-C5
olefins

C2-C 5
olefins

Paraffins
Aromtics
Cyclopara f fins

The intlal dehydration reaction is rapid and essentially at equilibrium over a

wide range of conditions.

Due to the selective nature of the ZSM-5 catalyst,

no hydrocarbons above CI0 are formed under M~G conditions.

The unique catalyst and reaction mechanisms impose several design
constraints

in the design

of the reactor

I-8

systems

for

t h e MTG p r o c e s s .

These

include the highly exothermic nature of the reaction, the need for essentially
complete methanol conversion, steam deactivation of the catalyst, the "bandaging" phenomenon, and durene formation.
a.

Thermochemistry
The MTG reaction is highly exothermic and the heat of reaction is in

the range 1510-1740 (kJ/kg of methanol) depending on the particular product

distribution.

This heat of reaction, if uncontrolled, would give an adiabatic

temperature rise of about 600C.

The following table gives the heats of

reaction for the major reaction steps in the conversion of methanol to

hydrocarbons (Chang and Silvestri (1977)).

Table 1-A-i:

Heats of Reaction for Major Reaction Steps in Methanol

Conversion to Hydrocarbons @

Reaction

CH3OH

(-AH r)
kcal

~[ CH3OCH 3 + ~1 H20

I CH3OCH 3

+ (CH2)olefin s ~i H20

(CH2)olefin s
CH30H

(CH2)hydrocarbons

(CH2)hydrocarbons + H20

% of total heat
o f reaction

2.410

22.5

4.466

41.8

3.814

35.7

10.69

i00.0

@based on I gmole of methanol and t = 371C

b.

Complete Methanol Conversion

The major products of the MTG reaction are hydrocarbons and water.

Hence, any unconverted methanol will dissolve into the water phase and
recovery of this methanol would entail adding a distillation step to process
the very dilute water phase.

Thus, essentially complete conversion of

methanol is sought.

I-9

c.

Catalyst Deactivation
The ZSM-5 catalyst undergoes two types of aging which contribute to a

gradual loss of catalyst activity.

A reversible loss results from coke formed

on the catalyst as a reaction product.

The second type of deactivation

results from one of the reaction products, steam.

Low reactor temperatures

and low partial pressures of water minimize the aging and favor a longer
catalyst life.
d.

Ba nd-Ag i ng
This phenomenon occurs only in fixed beds with fresh catalysts, where

the reaction occurs over a relatively small zone In a fixed bed.

This

reaction front moves down the catalyst bed as the coke deposits first
deactivate the front part of the bed.
permits a fixed-bed
frequent

regeneration

e.

The use of a sufficient catalyst volume

d e s i g n i n which o n - s t r e a m p e r i o d s a r e

l o n g enough t o a v o i d

of catalyst.

Durene F o r m a t i o n
Among the aromatic compounds formed in the MTG reaction is durene

(l,2,4,5-tetramethyl-benzene).
79 C.

Durene has a relatively high freezing point of

GasOline obtained directly from the reactors could contain 4 to 7 wt%

durene.

These high concentrations of durene could lead to problems with

solids build-up on carburetors during cold starts.

Durene is mostly formed by

alkylatlon of lower molecular weight aromatics with methanol.

partial

pressures

and h i g h r e a c t i o n

temperatures

Low methanol

tend to reduce the durene

l e v e 1.
R e a c t i o n Mechanism
Chang (1983)
postulated

has d i s c u s s e d

in the literature

the various

reaction

f o r t h e ~frG p r o c e s s .

b a s e d on t h e f o l l o w i n g a s s u m p t i o n s :

I-lO

mechanisms t h a t

have b e e n

Most o f t h e models a r e

(a) Methanol and Dimethyl ether

(DME) are always at equilibrium and can

be treated as a single kinetic species,

(b) Olefins

can be treated as a single kinetic species,

(c) The sum of paraffins and aromatics

can be treated as a single kinetic

species.
Assumption

(a) has been shown to be valid by the observation

wide range of conversions,

rapid than the subsequent

the initial step of ether form~tlon is much more

olefin-forming

step and is essentially at

equilibrium

Chen and Reagan (1979) discovered

a utocata!ytlc

over ZSM-5.

A+B

that over a

that the reaction was

The following scheme was proposed:

where
A = oxygenates

(methanol + DME);

B = olefins;

CPmng (1980) modified this scheme to account

where

A+B

B+C

A = oxygenates;

+ aro~tics

for the homologation

B = (:CH2) ; C = o!efins;

1-11

C = paraffins

D = paraffins

of olefins.

+ aromatics

Anthony (1981) pointed out an inconsistency in the derivation of the

expression for the reaction rate for the above scheme.

Futhermore, data at

only one temperature was analyzed and hence extrapolation to other

temperatures is not possible.
Recently, Mihall et al. (1983a, 1983b) have developed a detailed kinetic
model for the conversion of methanol to olefins and hydrocarbons.

The kinetic

model for methanol to olefins is obtained for only one temperature.

However,

for the conversion of methanol-to hydrocarbons, kinetic parameters (Arrhenlus

frequency factor and activation energy) are obtained from non-lsothermal
operation.

This detailed kinetic scheme involves 53 reaction steps and 37

kinetic species (including radical intermediates).

A kinetic scheme based on kinetic lumps has been proposed recently by
Chang et al. (1984) based on the following mechanism:

k
I

where

k
2

A = oxygenates; B = olefins; C = paraffins + aromatics

These lumps are defined on a water-free basis.

The disappearance of

oxygenates and oleflns are assumed to be first-order.

The kinetic parameters

k I and k 2 were determined over a range of temperatures and Si02/AI203 ratios

in the catalyst.

For a SIO2/AI203 ratio of ~ 450, they obtained:

k I

10366
-I
1.09x107 exp (.- ~ .) s

k2

0.98

-I
s

1-12

The data for oiefin disappearance

effect.

(k 2) showed little or no temperature

The data of Chen and Reagan (1979) for k I is well correlated by the

above expression for k I.

a.

Effect of Pressure
Chang et al. (1978) found that the main effect of increasing pressure

in the MTG s}~thesls was to promote the for~etlon of higher aromatics,

especially durene and to increase gasoline yield.

From their data, Chang et

al. (1978) concluded that the for~mtion of higher a r o ~ t l c s

was not p r i ~ r i l y

a result of increased contact time but represented an intrinsic change in

reaction selectivity with increased pressure.

Liederm~n et al. (1978) took

into account the inhibition effects caused by the adsorption of reactants

and/or products with an increase in pressure by a slmple expression of the
type (I/I+KAP A) where KA is an adsorption constant and PA is the sum of the
partial pressure of methanol, dimethyl ether and water.

They calculated K A to

be 1.5.
b.

Effect of Zeolite SIO2/AI203

Chang et al. (1984) studied the effect of temperature, pressure,

contact time and catalyst Bronsted acidity on olefln selectivity and found
that olefin for~mtion could be decoupled from aromatization via a combination
of high temperature and low catalyst acidity (high Si02/AI203 ratios).

similar observation regarding this effect of Si02/AI203 ratio has been

observed by Kikuchi et. al. (1984).
Fluldized Bed Reactor
Though the New Zealand MTG plant was based on the fixed bed reactor
concept, Mobil plans to use the fluidized bed reactor concept for future
plants (Penick et a!. (1983)).
commissioned in West G e r ~ n y

To this end, a i00 BPD pilot plant has been

(Flatow et alo (1984)).

1-13

Mobil initially carried out laboratory scale studies in a 4.13 cm

diameter

fluidized bed reactor

diameter

fluldlzed bed was used for carrying out studies under cold flow and

reactive conditions
diameter
a

(Voltz et al.

(Kam et al. (1978)).

fluid bed has been achieved

Process

(1976)).

Later a 10.12 cm

The erection of a I00 BPD, 60 cm

(Flatow et al.

(1984)).

DescriPtion

The process development

unit (PDU) fluid bed reactor first built by

Mobil consisted of a 4.12 cm diameter column with 5 zones which were heated by
an electrical

resistance

of the unit was 73.6 cm.

furnance

(Voltz and Wise (1976)).

During operation,

The total height

the charge stock and nitrogen

carrier gas were pumped through a preheater coll where the charge ~ s
vaporized.

The reaction occured in the dense fluid bed which contained four

umbrella-shaped

baffles to minimize gas by-passlng.

fluid bed development

The next stage in the

consisted of the following design features

(Figure I-A-

2).

10.2 cm diameter by 762 cm high adiabatic dense fluid bed reactor.

External llne for recirculating entrained catalyst to the reactor.

Intermittent

regeneration of catalyst to maintain day-to-day steady

state operation.
b.

Fluidized Bed Reactor Model

Several models have been proposed in the literature

reactors.

A review of these models is given by Yates (1983).

for fluidized bed

Most models are

based on the two-phase theory of fluidlzatlon which states that "all gas in
excess of that necessary to Just fluidlze the bed passes through in the form
of bubbles."

The various models differ in their assumptions

1-14

regarding the

"0 Product Recovery

'R

Disengager
16" X 3'

=.~ Flue Gas

.~,~
'~

Re~clor

~,

Regenera,or

10" x 2 i/2'
Air + N2

4" x 25'

i0" x 2 112'

Liquid Feed

Vapor Feed
t

Figure I-A-2:

Schematic representation of fluidized bed reactor and

regenerator system.
1-15

exact nature of the phases (bubble, cloud-wake and emulsion) and degree of gas
mixing in the phases.

The model types can be divided into:

(i) simple or arbitrary models based on empirical correlations obtained

with sm~ll scale equipment and
(li) models based on bubble dynamics which describe reactor behavior in
terms of the known physics and hydrodynamics of fluid beds.
For the present purposes of modeling, the countercurrent backmixlng model
of Fryer and Potter (1972), which is essentially an elaborate version of the
Kunii-Levenspiel bubbling bed model (Kunil and Levenspiel (1968a, 1968b)) will
be used.
c.

This model is described in the next section.

Countercurrent Backmixln~ Model of Fryer-Potter
The assumptions ~mde in this model are as follows:

(i) Bubbles are of one size and are evenly distributed in the bed.
(il) Each bubble drags along with it a wake of solids, creating a
circulation of solids in the bed, with upflow behind bubbles and downflow in
the rest of the emulsion.
(iii) The emulsion stays at minimum fluidizlng conditions; thus the
relative velocity of gas and solid remain unchanged.
(iv) Interphase gas exchange occurs in two stages:

from bubble to cloud

and from cloud to the particulate phase.

The model of Fryer and Potter (1972) is shown in Figure I-A-3.

The

superficial fluldizlng gas velocity is:

(2)

U = UGB + UGp + UGC

Since bubble and cloud-wake rise together at the same absolute velocity

1-16

UGC = fwSmfUGB .....................................

(3)

The backmixing model assumes that the relative velocity of gas to solids in
the particulate phase is the same as at incipient iluidization, i.e.

UGp

UGBfw (l-emf)
+

[l-~B(l+fw)]emf

Umf

[l-gB(l+fw)](l-~mf)

= --.

emf

(4)

Exit Gas Concentration

z=H
Phase

Bubble

Cloud-Wake

Superficial
Gas Velocity
Fraction
Volume

~B

Porosity

1.0

Gas Exchange
Coefficient

Figure I-A-3:

Particulate

UGp~

fwgB

l-gB(l+f w)

~mf

emf

+ KBC +

+ KCp

z=O

Inlet Gas
Concentration

Ci

Fluidizing Gas
Velocity

Representation of Fluid Bed Reactor According to the Model

of Fryer and Potter (1972).

1-17

Thus the superficial gas velocities in the bubble and particulate phases can
be obtained from equations (2), (3) and (4) as:

UGB = U - Umf[l-eB(l+f w)]

(s)

UGp -- Umf[l-eB(l+fw)]

(6)

[l+mff w] - Ufwemf

Backmixing of gas (i.e., negative UGp) is predicted if U exceeds a critical

value, Ucr, where:

U
cr

~ = [ i + ~ ][i
Umf
emff w

(7)

- B (i+fw)]

The coefficients, for gas exchange between bubble and cloud-wake phase and
cloud-wake and particulate phase are given by Kunii and Levenepiel (1968b) as
follows:
gl/4DGI/2

K c- 4.5 ( - - )

+ 5.85 (

(8)

s/4 )

DB

DB

I/2
cmfDGUA
KCp = 6.78 ( ~
)

(9)

DB

The bubble rise velocity, uA can be obtained from the following equation:

u A - U-Urn.f [l-eB(l+fw)]

+ 0.71(gD B)

1/2

(10)

and the bubble volume fraction is given by

UGB
E

(ll)

me

uA

1-18

The value of fw is taken as approximately unity except that when eB > 1/3, fw
is reduced:

1 - eB
f

2s B

(12)

Thi~ modification is introduced to ensure that the cloud-wake phase should

never exceed the dense phase, i.e.:

SBf w = l-e B (1+f w)

(~3)

The model equations are developed for steady-state, irreversible,

constant-volume gas phase reactions of the first order.

The contact of

reactant with the very smell amount of solids which =my be dispersed in the
bubble p~mse is neglected.

The ~ t e r i a l balances on reactant gas in the

bubble, cloud-wake and particulate pPmses, respectively, are:

-dC B

d---T-:
-de C

KBc(CB-Cc)eB

UGB

(14)

Kcp(Cc-Cp) CB+KBc(Cc-C B) ~B+kCcfw~B

d---f-:
-dCp

d--i-- :

UGC

(15)

Kcp(Cp-Cc)SB+kCp[I-SB(I+fw) ]

UG~

(16)

The boundary conditions are written for 5ackmixing conditions (i.e., when
U is sufficiently lares to cause downflow of particulate gas.

From equation

(7) this is seen to occur for U > 3-11Umf , which in most industrial situations
lls valid).
f

1-19

(a) at the distributor

level (z=O) all bubble gas is considered to derive

from the incoming gas

c B = ci

({7)

The remainder of the incoming gas combines with downflowing particulate phase
gas to constitute the cloud-wake gas

-UGp Cp + (U-UGB) C i = UGC C C

({8)

(b) At the top of the bed (z=H) gas leaving the bed is considered to be
made up of all the bubble gas and some cloud-wake gas wlth the remainder of
the cloud-w~ke gas providing the downflowing gas in the particulate phase

Cp

cc

(19)

The required expression for the exit gas concentration is

UCH = UGB CB + (U - UGB) CC

(20)

An analytical solution to the above boundary value problem has been provided
by Fryer and Potter (1972).

As stated earlier, this model is basically an

extension of the Kunii-Levesplel

bubbling bed model and the model equations

for the bubbling bed model can be achieved by simply ignoring the throughflow
of gas In the cloud-wake and particulate phases (i.e. UGC - O, UGp - 0).

1-20

Note

that some authors have mistakenly assumed the resulting expressions to imply
completely mixed cloud-wake and particulate phase gas, which is obviously not
true.
d.

Slugging in Fluldized Beds

Slugging occurs in deep fluldized beds of high aspect ratio where the

continuation of bubble coalescence leads eventually to the formRtlon of

bubbles whose diameter is equal to that of the bed itself.

Such bubbles are

called slugs and their hydrodynamic properties are different from those of
freely moving bubbles.

Stewart and Devidson (1967) have proposed a criterion

to determine the onset of slugging.

The following equation gives the minimum

excess gas velocity at which slugs will form in a tube of diameter D T.

(U-Umf) : 0.07 (gDT)I/2

(21)

From cold flow studies in a i0.12 cm diameter fluldized bed, Yam et al.
(1978) determined Umf to be 0.15 cm/sec for a particle size range of 70-75
~m.

From equation (21) the minimum excess gas velocity to cause slugging is

calculated to be 7.1 cm/sec.

In view of the fact that gas velocities in the

range 30-60 cm/sec were used under reacting conditions, it seems that slugging
would definitely occur in the fluldlzed bed reactor.

However, gem et al.

(1978) found from their cold flow studies that, with two different catalyst
types= DI0 and CBZ-I, slugging in the bed was less pronounced wlth DI0
compared to CBZ-I.

The properties of these two catalysts are given in Table

I-A-2.

1-21

Table I-A-2:

Properties of Cracking Catalysts Used in Cold Flow

Studies (Kam e t a l . (1978)).
CBZ-I

DIO

Surface Area, m2/gm

290

335

Pore Volume, cc/gm

0.577

0.975

Packed Density, gm/cc

0.64

0.63

Loosely Packed Density, gm/cc

0.53

0.47

Particle Density, gm/cc

1.035

0.73

Real Density, gm/cc

2.569

2.53

Physical Analysis

Furthermore, contrary to what was observed with CBZ-I, slugging with DIO
catalyst decreased as the superficial gas velocity increased.

They explained

this unusual phenomenon on the basis of the maximum stable bubble size that
could occur in the column.

For DIO, this size ~ms determined to be 7.6 cm,

while that for CBZ-I wms 12.1 cm.

The maximum stable bubble size for DIO is

less than the reactor diameter (10.12 cm) while that for CBZ-I is larger.
Hence, less slugging was observed with DIO compared to CBZ-I.

Another

indirect proof to show that slugging was absent comes from the bubble velocity
measurements carried out by Kam etal.

(1978).

The rise velocity of

continuously generated slugs is given by

USA = U - Umf +

~ble

0.35 (gDT)I/2

I-A-3 shows the experimental values of Ram e t a l .

velocities predicted by equation (22).

1-22

(22)

(1978) and rise

Table i-A-3:
U, cm/sec

Comparison of Experimental and Predicted Rise Velocities

Experimental, cm/sec

Predicted, cm/sec

15.2

56.1

49.9

21.3

84.1

56.0

42.7

167.9

77.4

From Table i-A-3, it is seen that equation (22) grossly underpredicts the
rise velocities.
equal volume.

It is known that slugs rise more slowly than bubbles of

Hence, in view of the fact that the Mobil catalyst used for the

MTG process closely resembles DIO rather than CBZ-I, it is assumed that
slugging is absent in the fluidized bed reactor and the backmixing model of
Fryer and Potter is used for modeling the reactor.
e.

Model Assumptions
The following assumptions are made in the development of the reactor

model:

i)

The temperature is constant along the length of the reactor.

Even

though the reactor was operated in an adiabatic manner, a very uniform

temperature profile was obtained during the pilot plant operation (Kam et al.
(1978)).
ll)

Hence, this is a reasonable assumption.

Though there is a slight expansion of the gas phase (about 20%),

the superficial gas velocity is assumed to remain constant along the length of
the reactor.
iii)

The mass transfer coefficients for all species can be taken to be

equal (Levenspiel et el. (1978)).

iv)

The bubble diameter, DB, is assumed to be equal to the maximum

stable bubble size that can be achieved in the fluidlzed bed.

1-23

f.

Input Parameters
To solve the model equations for the countercurrent

backmixing model,

the following parameters need to be known:

Solid particles

Umf, Era.
f

Gas

DG

Operating Parameters

U, H, d t

Kinetic Data

Hydrodynamic parameter

DB

The velocity at minimum fluidizatlon conditions,

Umf, has been reported

to be 0.15 cm/sec for the MTG catalyst (Kam and Lee (1978)).
minimum fluidization,

The voidage at

Cmf, was calculated from the following equation

PfL

Emf = i

PfL

where

p
P

= density of fluidized bed at minimum fluidizlng conditions,

0.224 g m / c c
pp
Hence,

= density of catalyst particle, 0.481 gm/cc

Cmf is calculated to be 0.533.

The gas phase diffuslvity,

DG, w~s calculated by means of the Chapm~n-

Enskog formula for a binary gas mixture.

T3 / 2

(IlMA + l I M B ) I l l

DAB = 0.0018583
Pt O2AB nAB

where

DAB

,,, bulk dlffuslvity, cm2/s

= temperature,

MA , M B

Pt

molecular weights of gases A & B

- total pressure of the gas mixture, atm

1-24

~AB' CAB = constants in the Lennard-Jones potential-energy function

for the molecular pair AB
~AB

: collision integral

Thus, D G was calculated to be 0.18 cm2/sec.

Fryer and Potter (1972) made an

extensive survey of the effect of the variables DG, Smf , fw and Hmf in the
ranges of values: 0. i to 1.0 cm2/sec, 0.4 to 0.6, 0.5 to 1.0 and 25 to 200 cm
respectively, and found their effect s,-~ll in comparison with k, DB, Umf and
U.

Hence, the estlm~tion of D G is not very critical.

The operating parameters, U, H and d t were taken from the Mobil report

(Kam and Lee (1978)).

The kinetic model of Chang et al. (1984) ~ s

the kinetic data were obtained from that paper.

used and

The bubble diameter D B is an

extremely critical parameter in these calculations a n d it was taken to be the

Irmximum stable bubble size (7.5 cm).
g.

Method of Solution
Even though the kinetic expressions are flrst-order,

the model

equations are solved numerically so that even non-linear rate expressions can
be included later, if so desired.

The model equations (14), (15) and (16)

together with the boundary conditions,

boundary value problem.
soft,re

(17)~ (18) and (19) constitute a

This boundary value problim is solved using the

package COLSYS (Ascher et a l. (1981)).

Case Study
Simulations of the fluidized bed were performed for various experimental
runs found in the report of Kam and lee (1978).

Tee following table compares

the experimental methanol conversions and those predicted from the Fryermodel.

1-25

Table I-A-4:

Comparison of Experimental and Predicted Conversions

Run No.@

U
cm/s

Experimental
Conversion, %

Predicted
Conversion, %

CT-231-I-2

39.23

CT-231-I-II

47.28

99.42

99.28

CT-231-I-17

25.88

99.98

99.92

i00.0

99.7

@From the appendix of the report by Kam and Lee (1978)

As can be seen, the conversions predicted by Fryer and Potter's model are
in good agreement with the experimental conversions.

The reaction model that

is used here is a "lumped" kinetic model and hence the entire product
distribution cannot be obtained.

Only selectlvlties of "oleflns" and

"paraffins and aromatics" can be presently obtained.

The concentration

profiles of the various species are shown in Figure I-A-4.

The following

table shows the experimental and predicted olefins distribution.

Table I-A-5:

Experimental and Predicted Oleflns

Distribution, mol %

Run No.
CT-231-2-62

Experimental
29.44

Predicted
1.5

The reason for the large discrepancy between experimental and predicted
olefins selectivity is probably due to the nature of the catalysts used.
stated earlier, the SiO2/Al203 ratio in the MTG catalyst can have a
significant influence on the product distribution.

1-26

As

A mooh+dm~
0

C poraffin~+aromatlc~

~.~-

I
P~

ol~flnm

~).P~

0.4

~).6

O.B

DimenBionleBB length
Figure I-A-4:

Typical concentration profiles of reactants and' products in f l u i d i z e d bed reactor

(WHSV = I . I , T = 672.6K, P = 2.7 atm).

The bubble diameter

countercurrent

is an extremely

critical

parameter

in the

backmixing model and the following table shows the effect of

bubble diameter on conversion.

Table I-A-6:

Effect of Bubble Diameter on Conversion.

Run No. CT-231-I-2
Conversion,

DB, cm
7.5

99.7

8.5

99.4

10.2

98.5

A bubble size of 10.12 cm indicates

theoretically,

the countercurrent

be used to model the fluidized

are different

that the bubble is a slug and

backmlxing model of Fryer and Potter cannot

bed since the hydrodynamic

from those of bubbles.

Slugging bed models

Raghuraman and Potter (1978) or Yates and Gregoire

appropriate
indicated

Table I-A-7.

properties of slugs
llke those of

(1980) would be the

models to use and they would predict conversions

in Table I-A-6.

Table I-A-7:

lower than those

The effect of t~mperature on conversion

The conversion

is shown in

increases with an increase in temperature.

Effect of Temperature

on Conversion

(U = 39.23 cm/s)

Temperature,

OF

Conversion,

650

95.84

7OO

98.94

762

99.?0

1-28

Fixed Bed Reactor

a.

Process Description
In the fixed bed process,

(Figure I-A-5).

two adiabatic reactors are used in series

In the first reactor, methanol is partially dehydrated to an

equilibrium mixture of methanol,

dimethyl ether and water over a dehydration

catalyst.

-).

2CH3OH

(CH3)20 + H20

This reaction is thermodynamically

limited and does not go to completion.

The

effluent from the first reactor containing a near equilibrium mixture of

methanol,

dimethyl ether and water is fed to the second reactor where methanol

and dimethyl ether are converted to hydrocarbons

catalyst.

and wa~er over a zeolite

Light gases are recycled to the second reactor to reduce the

adiabatic temperature

rise.

The heat of reaction is removed from the product

and recycle gases downstream to the reactor.

b.

Fixed

Bed Reactor Modeling

I n modeling the fluidized bed, the lumped kinetic model of Chang et

al. (1984) was used for the MTG reaction.
temperature

This model was used since the

in the fluidized bed was uniform despite the adiabatic mode of

operation and the heat balance was not required in modeling the fluidized
bed.

However,

there is a significant

temperature

rise in the fixed bed

reactor and hence the heat balance equation has to be included in the reactor
model equations

to model the temperature rise in the bed.

The heat of

reaction is o~e of the key terms that appears in the heat balance equation.
Since the MTG reaction consists of many parallel and consecutive steps, it is
not possible to assign a fixed value for the overall heat of reaction.

1-29

Global

RecycleGas__ ~ _ _ _ .
Methanol -

Prehealer

,. . . . . . . . .

',

;_!

- ~ Gas

,.~-

- Preheater
1,3" ID Reactor

C:)

/ I
60 cc of Catalyst

p//

210cc of Catalyst

L
Dehydration
Reactor
Figure I-A-5:

Conversion
Reactor
Schematic representation of fixed bed reactor.

High PressuFe
Separator
Liquid

of the heat of reaction will not lead to a satisfactory

temperature
detailed

profile.

fit of the

Hence, a rigorous design of the fixed bed requires

reaction scheme.

Such a detailed scheme would lead to a set of

continuity equations

for the various species which would directly predict the

product distribution

and the effect of temperature.

Furthermore,

the overall

heat of reaction would be correctly calculated from the heats of reaction for
the individual

reactions,

at all stages of conversion.

section on reaction mechanism,

Mihail et al. (1983b) have developed a detailed

kinetic model for the MTG reaction.

ether at equilibrium with methanol,
oxygenates

As stated in the

Their model is as follows:

generates carbene;

the carbene attacks the

giving light olefins, which then form higher olefins.

through carbenium ions form aromatics and paraffins.

the dimethyl

The olefins,

The various steps that

are involved in the mechanism are outlined in Table I-A-8.

This kinetic model

was not used for the fluidized bed reactor model since the use of this model
would lead to a set of Iii (37 species x 3 phases) non-linear differential
equations which are essentially stiff in nature and constitute a boundary
value problem.

Computational

time as well as memory requirements

for this

problem would be large and hence the lumped kinetic model was adopted for the
fluldized bed model.
c.

Dehydration Reactor
As shown in Figure I-A-5, the PDU consisted of a dehydration reactor,

where methanol is converted to an equilibrium mixture of methanol,

ether and water.

dimethyl

This reactor is also operated in an adiabatic manner.

Equilibrium methanol conversions and adiabatic reactor exit temperatures

as

functions of reactor inlet temperature have been calculated by Chang et al.

1978),

Their results can be correlated by the following expressions:

w

1-31

Table I-A-8:

I).

Reactions

in the kinetic scheme of Mihail et al. (1983b)

Methanol reactions

I).

CH3OH

CO + 2H 2

II).

zii).

IV)

2).

2CH3OH

CH3OCH 3 + H20

3).

CH3OCH 3

2:CH 2 + H20

Light olefins f o r m t l o n

4).

:CH 2 + CH3OH

C2H 4 + H20

5).

:CH 2 + CH3OCH 3

C2H 4 + CH3OH

6).

:CH 2 + CH3OCH 3

C3H 6 + H20

Hi~her olefins forms tion from carbene and light olefins:

7).

:CH 2 + C2H 4

C3H 6

8).

:CH 2 + C3H 6

C4H 8

9).

:CH 2 + C4H 8

C5HI0

I0).

:CH 2 + C5HI0

C6H12

II).

:CH 2 + C6H12

C7H14

Methane

12),

V).

from carbene and oxygenates:

formation from carbene and h)~dro~en:

:CH 2 + H 2

CH 4

Carbenium ions formation from olefins:

+A-H +

13)

C2H 4

.
+

~H5

-A+A-H +

14)

C3H6

C3H7

-A-

1-32

Table I-A-8 (continued)

+A-H +
+

14).

C3H 6

C3H7

-A+A-H +

15).

16).

C4H8

C~H9

+A-H +

CsHI0

+
C5HII

-AVI).

Carbenium ions attack on light olefins giving higher olefins

(oligomerization)
17)

+ 7 + C3H 6
C3H

18).

+
C4H 9 + C2H 4

+A+
_A-'H+
+A+
_A-H +

C6H12

C6H12

+A19).

C~H
9~

+ C3H 6

20)

c5R11 + c2H4

2-

+
~_A-H +

C7H14

+A

c7 14
_A-H+

vI ).

Carbenium ions attack on higher olefins giving paraffins and dienes:

21).

+
C2H 5 + C6H12

+A~_
_A-H +

C2H 6 + C6HI0

+A22).

,23)

C2H 5 + C7H14

C3 H 7 + C6H12

~
_A-H +
+A
~+
_A-H +

1-33

C2H 6 + C7H12

C3H8 + C6HI0

Table I-A-8 (continued)

24).

C3H
+ 7 + C7H14

25).

C4H
+ 9 + C6H12

C3H 8 + C7H12

+A+
_A-H +
+A~
_A-H +

C4HI0 + C6HI0

26).

C~H 9 + C7Hi4

27).

+
C5HII + C6H12

+A
+
~-- +
-A H

C5H12 + C6HI0

+
C5HII
+ C7H14

+A
+
~_
_A-H +

C5H12 + C7H12

28).
VIII).

+A
-'-+

-A H

C4HI0 + C7H12

Carbenium ions attack on dlenes ~ivin~ paraffins and cyclodienes:

+A-

29).

C~H7 + C6HI0

30).

+
C3H 7 + C7H12

~
_A-H+
+A
+
._
_A-H +

C6H8 + C3H8

C7HI0 + C3H 8

+A-

31).

C4H
+ 9 + c ~ 10

C~8+C~I

~+

_A-H +
+A-

32)

C4 H 9 + C7H12

4
_A-H +

C7HI0 + C4HI0

+A
33).

C5H!I
+
+ C6HI0

w-

C6H 8 + C5H12

-A H

34).

+
C5HII
+ C7H12

+A
_A-H +
1-34

C7HI0 + C5H12

Table I-A-8 (continued)

IX)

Carbenium ions attack on cyclodienes giving paraffins and aromatics:

+A35).

+ 7 + C6H 8.
C3H

C6H 6 + C3H 8
_A-H +

36).

+A+

+
C3H 7 + CTHI0

C7H 8 + C3H8

_A-H +

x).

+A+
_A-H +

37).

+
C4H 9 + C6H8

38).

C~H 9 + C7HI0

+A+
-A- H+

39).

C~HII + C6H8

+A
_A-H +

40).

+
C5HII + C7HI0

+A+
~_A-H+

C6H 6 + C4HI0

C7H8 + C4Hl0

C6H6 + C5H12

C7H 8 + C5H12

Aromatics condensation:
41).

2C6H 6

4-

42).

2C6H 6

CloH 8 + C2H 4
C9H8 + C2H 4 + C

1-35

Table I-A-8 (continued)

El).

Aromatics a Ikylation:
43).
44).
45),
46).
47).
48).
49).
50).

XII).

CH30H + C6H 6

CH3OH + C7H 8

CH30H + C8HIo

C9H12 + H20

CH3OH + C9H12

CIOH14 + H20

CH3OH + CIOH14

CIIHI6 + H20

CH3OH + CIIHI6

CH30H + CIoH8

CIIHI0 + H20

CH30H + CIIHI0

C12H12 + H20

C7H 8 + H20
C8HI0 + H20

C12H18 + H20

Paraffins demethanlza t lon:

51).

C5H12

C4H 8 + CH 4

52).

C4HIO

C3H 6 + CH 4

53).

C3H 8

C2H 4 + CH 4

1-36

Feed

Pure Methanol

Equilibrium
Conversion (%)
Exit Temperature
under Adiabatic
Conditions (K)

Crude Methanol (16 wt % water)

-0.0315T* + 99.788

-0.042T + 101.80

0.8165T + 225.14

0.833T + 193.62

*T is in degrees Kelvin, and is the inlet temperature to the dehydration

reactor.
d.

Model Assumptions
The followin~ assumptions are made in the development of the fixed

bed reactor model:

i) the gas phase is assumed to be in plug flowo

Since the L/D rat lo for

the Eixed bed is greater than 7, this is a reasonable assumption.

li) Since the kinetic parameters for the reaction scheme of Mihail e t a ! .
(1983b) were obtained adopting a pseudohomogeneous model for the reactor, the
same reactor model is used in the present case i.e. no distinction is made
between the fluid and solid temperature

This model assumes dlffusiomal

resistances and intrapartlc!e resistances to be negligible.

This assumption

is quite restrictive and it has been made due to the absence of any
experimental data regarding catalyst particle size and conversion.

For the

dehydration reaction of methanol, S~mbb and Gates (1972) found that the
effectiveness factor decreased from 0.93 to 0.62 as the mean catalyst (Hmordenite crysta!lites) pore length increased from 5.9 to 16.6 ~o

In view of

the fact that the catalyst particle sizes are not reported in the MTG
experimental studies, this assumption of negligible intraparticle diffusion
had to be made.
iii) An average pressure is used in the calculations since in most cases
the pressure drop in a fixed bed reactor is relatively small (Froment and
Bischoff (1979)).

1-37

e.

Model Equations
Simulating the fixed bed reactor requires the following set of

continuity equations for the components to be integrated, together with the

balance equation.

heat

Rj

ddz-~"

where

dt
4

alJ

NR

= number of reactions

NS

= number of species

d~T
dz

dt

I
NS

{ . ~ Z

Fj

J=l

r dt
rl)
4

2
J-l, NS

(23)

NR
(24)

(-AH i) r i + q(z)~d t}
i=l

Cpj
I
aj

NS
Here,

NR
= (E
i=l

ri = kI ~

(25)

cj

J=l
where

a~

is the order of the Ith reaction with respect to the Jth species.

Fj
Cj = NS ,,
Z
Fj
J-1
The

Pt
RT

--

(26)

heat of reaction is the algebraic sum of the heats o[ formation of

r e a c t a n t s and products:
NS

-Atl i - -

Z alj Hfj
J=l

l,

NR

(27)

where

Hfj

"

Hfj

T
Cpj dT
f
Tref
1-38

J=I,NS

(28)

The boundary conditions for equations (23) and (24) are

at z=0

F.j = 5 0

j--I$ Ns

T = Tin

e.

Method of Solution
The kinetic scheme of Mihall et al. (19835) leads to a set of

continuity equations for the reacting components that are ~ t h e ~ t i c a l l y stiff

in nature because of the orders of magnitude of difference between the
concentrations of molecular and ionic species.

Ordinary numerical integration

routines llke the Runge-Kutta method fail to work for systems of stiff
differential equations.

A more detailed discussion on stiff differential

equations can be found in the book by l~v!s (1984).

Recently, numerical

integration routines for sets of stiff differential equations have been worked
out (Gear (1971)).

For the purposes of integrating the above set of model

equations, a code aw_ilable at the University of Pittsburgh Computing Center

(Hind~mrsh)for solving systems of stiff differential equations (MTH:DLSODE)
w~s used.
f.

Parameter Es ti~m tlon

On examining the model equations (23) and (24), it is apparent that

the following parameters need to be known to perform the simulation:

Kinetic parameters - k,E
Physico-chemlcal parameters - Cpj, Hfj o
Operating parameters- dr, Tin , 5 o , L
The kinetic parameters are obtained from Mihail et el. (1983b) and they are
listed in Appendix I-A.

The physico-chemical parameters Cpj and

obtained from the property data hank in Reid et el. (1977).

1-39

o
Hfj

are

In those cases,

,.

where the compounds are not listed, the parameters Cpj and

Hfj

are

calculated by group contribution methods.

The heat capacity Cp was calculated by the method of Rihani and
Doralswamy (1965).

This method is based on the equation

E ni a

+ ~ nlb i T + I niCiT2 + E hid i T 3

i
i
i

where ni represents the number of groups of type i.

The parameters ai, bi, c i

and d i are group contribution parameters.

Example: 1,5 hexadiene contains 2 - ~

Group i

ni

ai

!
ffi

CH 2

and 2 - CH 2 groups.

bixl02

cixl04

dix106

--C :, CH2

-CH 2

0,2773

3.4580

-0.1918

0.004130

0.3945

2.1363

-0.I197

0.002596

Thus, Cp = 1.3436 + 11.1886 x 10 -2 T - 0.623 x 10 -4 T 2 + 0.013452 x 10-6 T 3

for 1,5-hexadiene.
The heat of formation at standard conditions Hf,298 Is calculated by the
method of Verma and Doraiswamy

Example:

(1965).

1,5 hex~diene

1-40

H f,298 = "~ (group contribution

due to

C = CH 2 ) + 2 (group contribution due

to CH 2

= 2 x 15.02 + 2 (-4.94)
= 20.16 kcal/gmole

The parameters

Cpj and

Bfjo
o

obtained from the property data bank in

Reid et ai. (1977) and calculated by the above group contribution methods are
listed in Appendix

i-A-9.

Case Study
The fixed bed reactor simulation was first performed for the case of no
recycle gas and for a fixed bed mlcroreactor which was essentially an
isothermal

reactor (Chang et al. (1978)).

The following table shows the

comparison between the experimental and predicted results.

Table I-A-9 Comparison of Experimental and Predicted Results

for Fixed Bed Reactor T = 644.26, P = 1 arm
WHSV = 1.0

Conversion,

Hydrocarbon Distribution,
wt%
Methane & Ethane
Propane
n-Butane + Isobutane
C2-C 4 0 l e f i n s
C5+ Nomaromatics
Aromatics
A6-AI0
All +

P/though the conversions

Experimental

Predicted

99

i00

1.0
8.7
21.9
1.7
30.5

19.50
8.0
4.80
24.7
0.8

35.7
0.5

23.4
10.70

are correctly predicted,

for the two cases are completely different.

1-41

the product distributions

The total aromatic distribution

is somewhat in agreement, but the non-aromatic distributions are totally

different.

As was seen in the case of the fluidized bed reactor as well, the

discrepancy is mainly due to the differing Si02/AI203 ratios in the two

catalysts.

A similar observation regarding product selectivity has been made

by Zhang and Ou (1984).

Typical concentration profiles along the length of

the reactor are shown in Figures I-A-6 and I-A-7.

For the case of the actual PDU, a dehydration reactor was used in
conjunction with the MTG reactor and light gases were recycled to reduce the
rise in temperature.

As noted by Yurchak et ai.(1979), the composition of the

recycle gas changed during the cycle.

A typical recycle gas composition for

purposes of simulation is as follows:

Methane

40 mol%

Hydrogen

I0 mol%

Carbon Monoxide

I0 mol%

Ethylene

i0 mol%

Ethane

i0 mol%

Propylene

i0 mol%

Propane

I0 mol%

A comparison of experimental and predicted results is shown in the following

table.

Table I-A-10

Experimental and Predicted Results for PDU

Experimental

Predicted

Outlet temperature
from dehydration
reactor (K)

683.71

696.04

Outlet temperature
from MTG reactor (K)

723.26

741.99

Conversion (%)

I00

99.8

1-42

~ o u ~ . - -

. . . . .

,.

.
I

III

III

0
CB
!
4~

0
tD
f~

HYI)ROCARO01~S- , ~

o~

O.~

0.4
DimenBion]eBB

Figure I-A-6:

0.6
Axial

O.B
DiBtanee

Concentration profiles of reactant and products .in fixed bed reactor (single pass,
I~HSV -- 0.8, P = I arm, T = 644. 261{).

,~ loo

II

III IIIIII II

i SOIll.
INS
/-rARAmNS

i 411.

A~C

E Ill.
o

Figure I-A-7:

o.s

0.4
Dimensionless

0.6
Axial Distance

o.e

Product distribution of hydrocarbons in fixed bed (same conditions as in Figure I-A-6).

by Mihail et al. (1983b) had a SiO2/AI203

ratio greater than 20 (no specific

number is mentioned).

of the selectivities

Hence,

comparisons

on the basis of

the lumped and detailed kinetic models cannot be made with any certainty due
to the differing

SiO2/AI203 ratios.

Thus, a comparison of the two reactor models

lumped kinetic model of Chang et al. (1984).
considered

is made on the basis of the

The following factors are

in the comparison:

i) The same WHSV (Weight Hourly Space Velocity)

reactors.

in both

WHSV is defined as

WHSV =

Inlet flow of reactant

weight of catalyst

Identical WHSV's can be obtained

(kg/hr)
(kg)

in both reactors by simultaneously

the inlet flow and the weight of catalyst.

suggests

is maintained

changing

A sound basis for comparison

that the weight of catalyst in both reactors be kept constant

not necessarily

then

(but

the same) and changes in WHSV be effected by changing the

inlet flow of reactant.

The weights of catalyst used in the two reactors are

those used in the pilot plant studies of Voltz and Wise (1976) and Kam and Lee
(1978).

The implication

of this assumption

is that the geometrical

dimensions

of the two reactors are different.

Weight of catalyst in fluid bed = 16027 gm

Weight of catalyst

ii) Isothermal

in fixed bed = 132 gm

operation is assumed in both reactors.

reasonable assumption

for the fluldized bed reactor.

reactor is operated in an adiabatic manner,

1-46

This is a

Though the fixed bed

the use of the lumped kinetic

750-

II

?O0!
...,j

~D

650-

~100"
0

o.~

i i

I i i

o.,~

lllill

o.~

|I

i lq i

i__

o.e

DimenBion]eBB Axia]I Dietenee

Figure I-A-8:

Temperature profile in fixed bed reactor (recycle gas ratio 3.3, WHSV = 3.3, P = 19 atm).

model precludes

the use of a heat balance equation in the model equations, and

hence an isothermal mode of operation is adopted for the fixed bed reactor as
well.
ill) The inlet temperature and pressure are the same for both reactors
and there is no recycle stream for either reactor.
With the above constraints,

the conversion of methanol,

selectivity and

space time yield (STY) are determined as a function of WHSV for the two
reactors.

STY is defined as follows:

STY

moles methanol converted/time

kg catalyst

The conversion and selectivity as a function of WHSV are shown in Figures

9 and I-A-IO.

I-A-

The decrease in conversion is much more rapid In the case of

the fluidized bed reactor.

The fixed bed reactor Is a plug flow reactor and

the decrease in conversion with WHSV is due to the decrease in space time.
For the fluid bed, as WHSV increases,

the portion of the gas passing through

the bed as bubbles increases and a decreased efficiency of contact of reactant

with the catalyst particles results since the exchange coefficient between
bubble and cloud-wake remains unchanged in this case.

Hence, the drop in

conversion is more marked in the case of a fluid bed.

Figure

selectivity

to B(oleflns) as a function of methanol conversion.

selectivity to B decreases with conversion for both reactors.

any given conversion,
a fixed bed.

I-A-If shows the

the selectivity

to B in a fluid

bed

The
However, for

is less than that in

It should be noted that the fixed bed reactor model assumed no

mass or heat transfer resistances or pore diffusion effects.

The p e r f o r m n c e

of the fluldlzed bed as regards the selectivity to the intermediate product,

B(oleflns),

is typical for reactions occurring in 8erles (Levensplel et a l.

1-48

100

.,,q

.~

....

..........

-.,

....

%
%

Q)

%
%
%

c)

%
l

r~

l
%
%
%
%
k

Figure I-A-9:

I.~

2.5

Conversion as a function of WHSV for fixed and f l u i d i z e d bed

reactors.
(P = 2 atmo T = 673K).

1-49

II

'1
80

--F/~d

Bed R e a c t o r

- - - l q u l d t s a d Bed meactor

- ~o/

~
~

5o

if! ,o
~

3o

r~

20

Figure l-A-lO:

STY as a function of WHSV for fixed and fluidized bed reactors

(P = 2 atm, T = 673K).

1-50

-g

@#

.~o (}oi

.//

.,'"

l,m m m ~ m m ~ ~
0,5

Figure ! - A - I I :

I.~

2.5

Selectivity to olefins in fixed and fluidized bed reactors as

a function of WHSV (P = 2 at.m, T = 673K).
1-51

(1978)).

The STY as a function of WHSV is shown in Figure I-A-12 and it can

be seen that with increasing space velocity the STY for the fluldized bed is
less than that for the fixed bed.

This is due to the increased by-passing of

gas in the form of bubbles resulting in a decrease in conversion.

One way of

avoiding this situation is to decrease the bubble size by introducing

internals into the fluid bed which would help in breaking up the large
bubbles. Factors such as conversion and selectivity are not the only
considerations

in the selection of a reactor.

Since the MTG reaction is such

a highly exothermic one, reaction heat management is a principal

consideration.

Some of the advantages

the fluidized bed reactor has over the

fixed bed reactor in this regard are (Yates (1983)):

(a) Solid particles are in continuous motion and are normslly very well
mixed, hence "hot spots" are rapidly dissipated and the bed operates in an
essentially l s o t h e r ~ l

~nner.

(b) Due to the very high bed-to-surface

heat transfer that can be

achieved, again as a result of particle motion, temperature control is seldom

a problem.
(c) The fluld-llke properties

of the gas-solld mixture enable the s o l i d s

to be transferred without difficulty from one vessel to another.

Hence, the

catalyst can be regenerated without taking the reactor off-stream, as would be

the c a s e

for a f i x e d

bed reactor.

Offsetting the above advantages are some disadvantages

are so severe that fluidized bed reactors cannot be used.
(a) Erosion of bed internals,

that in some cases

These are:

heat transfer coils, valves, etc. caused by

the "sand blasting" action of the solids.

1-52

%
%
%

%
%

%
%
%

im

%
%

%
%

o
G

'N,
%
%
%

O
%

of N tha e!

Fi gure I-A=12:

(g)

SelectJvJt~ to olefins as a function of methanol conversion

in fixed and fluidized beds (P = 2 atm~ T = 673K).

1-53

(b) Loss of very fine particles through cyclone plugging and attrition.
However, Mobil ha3 claimed that it has developed a material which is not very
susceptible to attrition (Haggin (1985)).
(c) By-passing of the solid by gas bubbles which can severely llmlt the
conversion on a once-through basis.
(d) Reliable scale-up is difficult to achieve due to the limited
understanding of the physics of fluidizatlon.

In fact, due to this reason

Mobil decided to offer only their fixed bed reactor concept to New Zealand and
carry out pilot plant studies in a I00 BPD plant before licensing the
fluidized bed reactor concept (Penick et al. (1983)).
Conclusions
Modeling of the fluidlzed and fixed bed reactiors for the MTG process was
performed in this study.

The countercurrent backmixing model of Fryer and

Potter (1972) was used to model the fluidized bed, while the one-dlmenslonal
pseudo-homogeneous model was used for the fixed bed.

Model predictions for

methanol conversion compared well with experimental conversions.

However,

product selectivities from the model simulations were not in good agreement
with experimental results.

This was mainly due to the fact that the

SI02/AI203 ratio in the 2SM-5 catalyst used in the pilot plant studies has not
not been reported.

The Si02/AI203 ratio in the 2SM-5 catalyst has a major

influence in determining product selectivity.

Thus, a knowledge of the

SI02/AI203 ratio is essential for correctly predicting product

selectivlties.

A comparison of the fixed and fluidized bed reactors for the

MTG process was done on both a quantitative and qualitative basis.

This

comparison revealed that though the fixed bed reactor gave high conversion and
a higher selectivity to intermediate product (olefins) at a given WMSV as

1-54

compared

to the fluidized bed, the fluidized bed reactor was better in terms

of reaction heat management.

-DD

Nomenclature:

CB

reactant concentration

CC

reactant concentration in cloud-wake gas, moles/cc

CH

reactant concentration

Cpj

specific heat of component j, kcal/kmol.K

Cp

reactant concentration

Ci

reactant concentration in inlet gas, moles/cc

dt

reactor diameter,

DB

diameter of sphere having the bubble volume, cm

DG

gas phase diffusion coefficient,

fw

ratio of wake volume to bubble volume

Fj

molar feed rate of species J, kmol/s

gravitational acceleration,

Fjo

inlet molar feed rate of species j, kmol/s

height of bubbling bed, cm

Hfj

heat of formation of species j, kcal/kmol

in bubble gas, moles/cc

in exit gas, moles/cc

in particulate phase gas, moles/cc

cm2/sec

cm/sec 2

Hfj

heat of formation of species j, kcal/kmol

-AH i

heat of reaction for reaction i, kcal/kmol

Hmf

height of fluidized bed at minimum fluidizlng conditions, cm

k,kl,k 2

first order reaction rate constants,

KBC

volumetric rate of gas exchange between bubble and cloud-wake per

unit bubble volume, sec -I

Kcp

volumetric rate of gas exchange between cloud-wake and particulate

phase per unit bubble volume, sec -l

Pt

operating pressure, atm

q(z)

heat removal per unit length of reactor, kcal/m.s

rI

rate of reaction per unit volume, kmol/m3s

gas constant, m 3 atm/kmol K

Rj

total rate of change of the amount of component J, kmol/m3s

1-56

see

-I

temperature, K

Tin

inlet temperature to reactor, K

ua

bubble rise velocity, cm/sec

superficial velocity of fluidizing gas, cm/sec

Ucr

superficial gas velocity above which hackmixing occurs, cm/sec

UGB

superficial gas velocity in bubble phase, cm/sec

UGC

superficial gas velocity in cloud-wake region, cm/sec

UGF

superficial gas velocity in cloud-w~ke region cm/sec

Umf

superficial gas velocity at incipient fluidization, cm/sec

distance along reactor, m

Greek Letters
sij

stolchiometrlc coefficient of component j with respect to the ith

reaction

eB

fraction of bed volume occupied by bubbles

Emf

voidage fraction at incipient fluidlzatlon

1-57

References
Anthony, R.G., "A Kinetic Model for Methanol Conversion to Hydrocarbons",
Chem. Eng. Sci., 36, 789 (1981).
Ascher, U., J. Christiansen and R.D. Russel, "Collocation Software for
Boundary-Value ODE's", ACM Trans. on Math. Soft~are, 7__ (2), 209 (1981).
Chang, C.D., Hydrocarbons from Methanol, Marcel Dekker Inc., New York (1983).
Chang, C.D., "A Kinetic Model for Methanol Conversion to Hydrocarbons", Chem.
Eng. Scl. 35, 619 (1980).
Chang, C.D. and A.J. Silvestri, "The Conversion of Methanol and Other Ocompounds to Hydrocarbons over Zeolite Catalysts", J. Catal., 47, 249 (1977).
Chang, C.D., J.C.W. Kuo, W.H. [ang, S.M. Jacob, J.J. Wise and A.J. Silvestrl,
"Process Studies on the Conversion of Methanol to Gasoline," Ind. Eng. Chem.
Process Des. Dev. I__/.7,255 (1978).
Chang, C.D., C.T.W. Chu and R.F. Socha, '~ethanol Conversion to Oleflns over
ZSM-5 I. Effect of Temperature and Zeolite SIO2/A1203", J. Catal., 86, 289
(1984).
Davis, M.E., Numerical Methods and Modeling for Chemical Engineers, John Wiley
& Sons, New York, (1984).
Chen, N.Y. and W.J. Reagan, "Evidence of Autocatalysls in Methanol to
Hydrocarbon Reactions Over Zeolite Catalysts", J. Catal., 59, 123 (1979).
Flatow, D., N. Davlduk and H. Heiart, "Conversion of Methanol to Gasoline,
Engineering and Construction of the Demonstration Plant", DOE Report No.
DOE/NBM-500323, March (1984).
Froment, G.F. and K.B. Bischoff, Chemical Reactor Analysis and Design; John
Wiley & Sons, New York (1979).
Fryer, C. and O.E. Potter, "Countercurrent Backmixlng Model for Fluldlzed Bed
Catalytic Reactors. Applicability of Simplified Solutions", Ind. Eng. Chem.

Fundam., II (3), 338 (1972).

Gear, C.W., Numerical-Initial Value Problems in Ordinary Differential
Equations, Prentice-Hall, Englewood Cliffs, N.J. (1971).
Haggln, J., "First Methanol-to-Gasoline Plant Nears Startup in New Zealand",
Chem. and Eng. News, March 25, 1985, p. 39.
Hindmrsh, A., Mathematical and Statistical Division, lawrence Livermore
La bora t orles.
Kam, A.Y. and W. Lee, "Fluid Bed Process Studies on Selective Conversion of
Methanol to High Octane Gasoline", DOE Report No. FE-2490-15, April (1978).
Kikuchi, E., S. Ha tanaka, R. Hamana, and Y. Morita, "The Acidic Properties of
ZSM-5 Zeolite and its Catalytic Activity in the Conversion of Methanol", Int.
Chem. Eng., 24 (I), 146, (1984).

1-58

Kunii, D. and O. Levenspiel, "Bubbling Bed Model. Model for the Flow of Gas
Through a Fiuidized Bed", Ind. Eng. Chem. Fundam., 7, 446 (1968a).
Kunii, D. and O. Levenspiel, "Bubbling Bed Model for Kinetic Processes in
Fluidized Bed. Gas-Solid M~ss and Heat Transfer and Catalytic Reaction, " Ind
Eng. Chem. Process Des. Dev., 7_, 481 (1968b).
Levenspie!, O., N. Baden and B.D. Kulkarni, "Complex First-Order Reactions in
Fluidized Reactors: Application of the KL Model", Ind. Eng. Chem. Process
Des. Dev. 17, 478 (1978).
Lieder~n, D., S.M. Jacob, S.E. Voltz and J.J. Wise, "Process Variable Effects
in the Conversion of Methanol to Gasoline in a Fluid Bed Reactor," Ind. Eng.
Chem. Process Des. Dev. 17, 340 (1978).
Meisel, S.L., J.P. McCullough, C.H. Lechthaler and P.B. Weisz, "Gasoline from
Methanol in One Step", Chem. Tech. February, 86 (1976).
Mihail, R., S. StraJa, Gh. Morla, G. Musca and Gr. Pop, "Kinetic Model for
Metb~nol Conversion to Olefins" Ind. Eng. Chem. Process Des. Dev., 22 (3), 532
(1983~).
Mihali, R., S. Straja, Gh. Morla, G. Musca and Gr. Pop., "A Kinetic Model for
Methanol Conversion to Hydrocarbons" Chem. Eng. Sci., 38, 1581 (19835).
Penick, J.E., Wo L e e and J. Maziuk, "Development of the Methanol-to-Gasoline
Process" in Chemical Reaction Engineering-Plenary Lectures (J. Wei and C.
Georgakis eds.) ACS S~posium Series No. 226, 19 (1983).
Raghuram~n, J. and O.E. Potter, "Countercurrent Backmixing Model for Slugging
Fluidized-Bed Reactors, " AIChE J. 24, 698 (1978).
Raid, R.C., J.M. Prausnitz and T.K. Sherwood, The Properties of Gases and
Liquids, McGraw-Hill, New York (1977).
Rihani, D.N. and L.K. Doraiswamy, "Estimation of Heat Capacity of O r p h i c
Compounds from Group Contributions", Ind. Engo Chem. Fundam., 4__,17 (1965).
Stewart, P.S.B. and J.F. Devidson, "Slug Flow in Fluidized Beds", Powder

Tech., !, 6! (1967).
S~bb, E.A. and B.C. Gates, "Diffusion, Reaction and Fouling in H-Mordenite
Crystallites. The Catalytic Dehydration of Methanol", Ind. Eng. Chem.
Fundam., I!, 541 (1972).
Verma, K.K. and L.K. Doraiswamy, "Estimation of Heats of Formation of Organic
Compounds", Ind. Eng. Chem. Fundam., 4__,389 (1965).
Voltz, S.E. and J.S. Wise, "Development Studies on Conversion of Methanol and
Related Oxygenates to Gasolines", Final Report, prepared for ERDA under
Contract No. E(49-18)-1773, November (1976).
Yates, J.G., Fundamentals of Fluldlzed-Bed Chemical Processes, Butterworths,
London (1983).

1-59

Yates, J.G. and J.Y. Gregoire, "Catalytic Oxldatlon of O-Xylene in a Slugging

Fluldlzed Bed," Chem. Eng. Sci. 35, 380 (1980).
Yurchak, S., S.E. Voltz and J.P. Warner, "Process Aging Studies in the
Conversion of Methanol to ~ s o l i n e in a Fixed Bed Reactor", Ind. Eng. Chem.
Process Des. Dev., 18, 527 (1979).
Zhang, Z. and G. Ou, "Investigation and Evaluation of a Methanol-toHydrocarbon Catalyst in a Laboratory Scale Reactor", Int. Chem. Eng., 24, 368
(1984).

1-60