Previous Page

Chapter 7 CHEMICAL TECHNOLOGY 421

REACTOR

Fluidization by a liquid
When V,, increases, fluidization remains homogeneous and E~ decreases
regularly with V,, according to an empirical law established by Richardson
and Zaki:
v,, = (1 -Ep)n (7.22)
Ut

with ut corresponding to the terminal falling velocity of a particle of average

diameter dp. The exponent n is always positive and depends on the Reynolds
number, the particle shape and on the ratio dpfD,with D being the reactor
diameter.
Fluidization by a gas
When V,, increases to over (I/,&,, two cases can occur:
(1) Fluidization remains homogeneous, in which case the Richardson and
Zaki relation is applicable. Fluidization is said to be particulate. This is rather
infrequent and corresponds to very particular conditions which are seldom
found in industrial practice.
(2) Fluidization becomes heterogeneous with gas bubbles appearing prac-
tically as soon as (VS&, is exceeded. This is the normal gas-solid fluidization
regime, sometimes called aggregative fluidization. When aggregative fluidiza-
tion occurs it should be regularized, i.e. both slugging and channeling should
be eliminated. Slugging is when gas bubbles pressed against the walls of the
reactor cause slugs to occur with solids thrown out of the bed. The problem is
generally overcome either by increasing the ratio between bed diameter and
height, or by placing internals (heat exchange tubes for example) in the bed.
Channeling is when preferential paths are created in the bed and is remedied
by designing an appropriate distributor.
In the aggregative fluidization regime an important fact should be noted.
The volume fraction occupied by the gas in the phase consisting of the homo-
geneous gas-particle emulsion (also called the dense phase) is identical to
(1 - E ~ ~ i.e.
) , the value of the fluidization threshold. Once the fluidization
threshold is reached, it is thus as if the excess gas in relation to (I/,&, passed
totally through the bed in the form of bubbles. This fact is utilized in modelling
the process.

7.1.4.3 Multi-Phase Catalytic Reactors

a. Fixed Beds with Two-PhaseFlow
Three configurations can be imagined according to the respective flow direc-
tions of the gas and liquid phases:
cocurrent downflow systems (Fig. 7.14A),
cocurrent upflow systems (Fig. 7.14B),
counter-current systems (Fig. 7.14C).
422 Chapter 7. CHEMICAL REACTOR
TECHNOLOGY

A B
Gas

i-.i
Liquid
Fquid

Y A ir’
Gas + Liquid
Gas Liquid
-Liquid

-Fixed catalyst
bed

t
Liquid
J
Figure
7.14 Various types o f fixed bed two-phase flow catalytic reactors.

Cocurrent downflow is the most commonly used technology as it is the

easiest to implement. In somewhat the same way a s for packed columns, the
efficiency of fixed beds with cocurrent two-phase downflow depends essen-
tially on the distribution of the liquid phase at the top of the catalyst bed.
Distribution plates are used similar to those described in Chapter 2.
Cocurrent upflow allows liquid phase distribution problems to be avoided,
assuring good wetting of the catalyst particles throughout the bed. This fluid
circulation mode is recommended in particular for low capacity operations
(pilot unit for example).
 Chapter 7 CHEMICAL REACTOR
TECHNOLOGY 423

Counter-current flow has recently found attractive applications with the

designation “reactive distillation”, used especially in the synthesis of methyl
tert-butyl ether (MTBE). The catalyst (here, ion exchange resins in the form of
beads 1 mm in diameter) must be implemented in a particular way to allow
counter-current flow. The advantage of the set up is that the reaction (iso-
butene + methanol) and the separation of the reaction product (MTBE) are
carried out at the same time. This type of technology is applicable to other
reactions in which the separation of the reaction product allows the thermo-
dynamic equilibrium to be shifted.
In actual practice, the cocurrent downflow configuration is undoubtedly
the most frequently used.
Cocurrent downflow
Fixed bed two-phase flow catalytic reactors (often termed trickle beds)
have been the subject of a great deal of research due to their importance in
industry. There are a large number of publications on flow regimes, pressure
drop and liquid hold-up calculations, and gas-liquid and liquid-solid mass
transfers. These different aspects will be examined, with special emphasis on
the important points for practical reactor design @hase distribution for exam-
ple) and modelling.
1. Flow regimes
There are essentially three types of flow depending on the relative flow
rates of the liquid and gas phases:
a flow with the gas phase continuous and the liquid dispersed for low liq-
uid flow rates,
a flow with the liquid phase continuous and the gas dispersed in the form
of bubbles for high liquid flow rates,
a pulsating or high interaction flow for average liquid and gas flow rates.
Various diagrams have been proposed to represent these regimes. The L/G
versus G diagram by Charpentier et al. (1978) (Fig. 7.15) is presented as an
example. L and G are the superficial mass flow rates (L = pLVs, and G = pcVs,).
The diagram has the drawback of using the same variable C in the abscissa
and in the ordinate, which somewhat masks the influence of the liquid flow
rate L . On the other hand, the authors of the diagram have used the corrective
terms h and w to take fluid physical properties into account:

(7.23)

y = - Pwater [- -
CLL (P;;er)’]”’
(7.24)
PL Pwater

In addition, Charpentier et al. have shown the influence that liquid foama-
bility can have.
424 Chapter 7. CHEMICAL TECHNOLOGY
REACTOR

L
- hv
G
10:

10'

Figure
7.15 Diagram of two-phase
cocurrent downflow
10
regimes through a
catalytic bed.

rJ
Mist
1
10-2 lo-' 1
G / h(kg/m'.s)

Note that industrial reactors work either in the pulsating or in the trickling
regime, while pilot plant reactors always work in the trickling regime. The pul-
sating regime is also often called the high interaction regime and in actual
practice an attempt is made t o approach these conditions which lead to bet-
ter contact between the fluids and the catalyst.
It is possible to estimate the minimum value of L required to achieve com-
plete wetting of the catalytic bed by using the relation proposed by Ng (1986):

(7.25)

Relations put forward by the same author can also be used t o estimate the
transition between the trickling and pulsating regimes.
2. Pressure drop and liquid hold-up
Since the research by Larkins in 1961, numerous pressure drop and liquid
hold-up results have been published, for example by Charpentier et al. (1978)
and more recently Wild et al. (1991). The Larachi correlation (1990) for calcu-
lating pressure drops in a two-phase downflow fixed bed was taken from the
review article published by Wild et al. This correlation employs the following
dimensionless groups:

K = XG(ReLWe,)0.25 (7.26)
VSL
 Chapter 7 CHEMICAL REACTORTECHNOLOGY 425

ReL= PL
- VSLdp
WeL=
PL G L d p
-
PL 0

The correlation is then written as follows:

(7.27)

with the use of the hydraulic diameter

+
F L G = ~ - ' . ~ ( 3 1 . 317.5 X K-0'5) (7.28)
pL, the liquid saturation ratio, can be calculated from the same parameters by
means of the equations below:
pL = 1 - r = 1.22 xXeo.I5x ReL0.*x Wet15 (7.29)
Moreover, there are the following expressions for the reactor volume frac-
tions occupied by the liquid and the gas respectively:
eL= pL(l - cp) and E~ = (1 - pJ(1 - cp) (7.30)
Lastly, the pressure drop calculation per unit of bed length is obtained from
the following equation which contains one term due to friction and another
related to gravity:
(7.31)

Note that these correlations, like all those found in the literature, were
obtained in small diameter columns (less than 30 cm). Their validity for indus-
trial dimensions (diameter from 1 to 3 m) is therefore uncertain, and conse-
quently the values obtained with the formulas should be used with caution.
These reservations are, however, not overly troublesome in actual practice as
a comfortable safety margin is commonly kept t o take into account the
expected fouling of the catalytic bed and therefore the increase in pressure
drop with time.
3. Modelling
Given the uncertainty about the set of parameters describing two-phase
flow inside a catalytic bed, it is advisable to use the simplest model possible
t o attempt to describe the operation of this type of reactor.
As an initial approximation, the gas and liquid phases might be considered
to flow at linear velocities identical all over a cross-section (plug flow). The
possible influence of turbulent diffusion superimposed on plug flow must then
be considered.
Expressions are found in the literature that allow axial Peclet numbers to
be evaluated for both the gas and the liquid phase. The focus is generally on
the conversion of the reactant in the liquid phase, so the liquid phase Peclet
number will be more important. (Pe), usually ranges between 0.3 and 1.
426 Chaoter 7 CHEMICAL REACTOR
TECHNOLOGY

If the criterion defined by Mears is applied, for considering that axial diffu-
sivity is negligible, i.e.:

(7.32)

with H the total height of the catalytic bed and C,, the key reactant concen-
tration in the liquid phase, for 99% conversion, i.e. for C,,(O)/C,,(H) = 100, the
result should be:
H 20
- > - 2 . 3 2~= 300 (7.33)
d,, 0.3

This condition is readily achieved in actual practice, since catalyst parti-

cles generally have a diameter of between 1 and 4 mm. Thus if the plug flow
assumption is adopted for the two phases, equations expressing molar bal-
ances can then be written for each of the components taking part in the reac-
tion. As mentioned earlier, one of the important points to consider during the
design of such reactors is phase distribution at the top of each catalytic bed.
This problem will be dealt with during the presentation of concrete examples.

Cocurrent upflow
This type of fixed bed reactor where the liquid and gas phases pass
through the catalytic bed from bottom to top is only seldom employed on an
industrial scale. However, it is often used in small pilot plant units. The reason
is that it may be difficult to keep the catalytic bed stationary when the two-
phase mixture passes through it and any catalyst particle movement due to
flow leads to unacceptable attrition. In contrast, one of the advantages of the
technology can be to provide a simple and effective solution to the liquid dis-
tribution problem encountered in downflow reactors.
The same as for downflow, there are various flow regimes whose limits can
not yet be precisely predicted. The bubble regime with a continuous liquid
phase would seem to be the most common in actual practice. According to
a number of published data, this regime is established for low gas flow rates,
i.e.:
C > 1 kg/m2.s
The pulsating regime is believed to exist above this value.
A number of correlations have been suggested to calculate the pressure
drop. However, given the relative uncertainty of the formulas, it is undoubt-
edly preferable to calculate a pressure drop by excess in an initial approxima-
tion simply by taking the static height of liquid in the bed.
The liquid phase distribution does not pose any of the difficult problems
encountered in downflow. In contrast however, suitable distribution of the gas
phase will have to be achieved and the catalytic bed will have to be kept sta-
tionary in actual practice.
 Chapter 7. CHEMICAL TECHNOLOGY 427
REACTOR

Counter-currentflows
There are only very few industrial facilities implementing fixed bed cat-
alytic reactors with two fluid phases (gas and liquid) circulating counter-
currently. This technology can, however, afford certain advantages.
It should be noted that one of the constraints of this type of reactor is the
occurrence of the flooding phenomenon which limits the flow rate of each of
the phases that is to pass through the catalytic bed. Reference can be made to
Chapter 2 on this point, where these concepts are discussed regarding packed
columns. In order to increase the range of allowable flow rates, larger catalyst
particles than those generally selected for fixed beds will have to be used.
Ideally, catalyst particles would be in the shape of conventional packing ele-
ments such as Berl saddles or Raschig rings. The consequence of larger sized
catalyst particles could be lower efficiency.
In contrast, working with counter-current flow can be advantageous for cer-
tain reactions, reversible reactions for example. In the gas phase, a reaction
product can in fact be eliminated and the equilibrium can be shifted. This is
taken advantage of in reactive distillation as used in producing MTBE (see
VOl. 3).

b. Three-phaseFluidized Beds
Three-phase or ebullating bed reactors can be schematically represented as
shown in Figure 7.16. The liquid and gas phases pass through the reactor from
bottom to top and keep the catalytic solid in suspension. The catalyst parti-

1 1 ) Catalyst level
* i ** ) * ) 1
**) )** * ) * )
;* )
:
) * 2 + L,Q I UD
I
Fi * * * )* * ) )*
)
with
Sketch of a three-phase 7 * * * *1 ) catalyst (*)
1 )
) * in suspension
fluidized bed or :>** * )*)*)*
ebullating bed. * I * ) * ,* ) >

~ ) * )* * * ) *
*f,’* I * ) *
428 Chapter 7 CHEMICAL REACTOR
TECHNOLOGY

cles generally range from 1 to 5 mm in size. The two fluid phases exit at the
upper part of the reactor without entraining the solid which remains inside the
reactor in the form of a fluidized layer.
It should be noted first of all that the understanding of three-phase flu-
idized beds is still incomplete. The following section gives some of the infor-
mation in the literature that can be used for an initial description of these sys-
tems. The correlations reported were generally obtained with air-water
systems and in small sized apparatuses, so they should be applied to indus-
trial cases only with extreme care.

Flow regimes
In an initial approximation, the three-phase systems could be described as
a bed fluidized by the liquid in which the gas phase flows the same as in a bub-
ble column. This is just about true when the liquid velocity is relatively high
and the gas velocity remains low. However, when the gas velocity is high and
the liquid velocity is low the situation is completely different, with occurrence
of a pulsating type regime. For intermediate gas and liquid velocities, a grad-
ual transition from the fixed bed to the completely fluidized bed can be
observed.
Minimum fluidization velocity
There are in fact two minimum fluidization velocities, each one corre-
sponding to the flow of a single fluid phase (gas or liquid). In Section 7.1.4.2.c,
the part headed “Minimum superficial fluidization velocity” explains how to
calculate these values. Here, the focus is on determining the interaction
between the two fluid flows and the onset of the fluidization phenomenon.
This is generally expressed in the form of a minimum velocity of the liquid,
VsLm, for a certain gas flow rate, such as in the Begovich and Watson equation
for example:

dp VSLm P L = 5.12 x (Ar)~662(Fr)~o~1’s

(7.34)
(Re),, =
CLL

(Ar>,= pLd;(pp CLt

- p3g (Archimedes number)

(Fr), = -
V2SG (Froude number)
gdp
The above relation is not applicable for low gas flow rates.

Bed expansion and hold-up

With the flow of the two fluid phases, the granular solid reaches the flu-
idized state subsequent to an expansion of the initial fixed bed. Here the situ-
ation is a little more complicated than when there is a single fluid phase. It has
in fact been observed that the increase in gas flow rate could in some cases be
 Chapter 7 CHEMICAL TECHNOLOGY 429
REACTOR

accompanied by a contraction of the fluidized bed. This is true when particles

smaller than 2.5 mm are fluidized first of all by the liquid alone with the gas
phase then gradually added. For low increasing gas flow rates, a slight con-
traction of the fluidized bed is generally observed, then an expansion when
the gas flow rate takes on larger values.
The Kim et al. relation can be utilized to calculate bed expansion:

+
( E ~ E ~ =) 1.40 (Fr)p'7(We)0.078= 1 - E P (7.35)
and is valid for beds that undergo no initial contraction.
In the expression above the Froude and Weber numbers are defined as fol-
lows:
V S G PL
V2SL We= -
(Fr),= -
(J
gdP
The liquid phase bed hold-up can be calculated by means of a formula
established by the same authors:
E~ = 1.504 (Fr)~0.086(Fr)~234(Re)~0.082(We)0.092

Pressure drop
The pressure drop through the bed alone can be calculated simply based
on the static pressure, i.e.:
AP = + PLEL + pcE3Hg

with H being the height of the fluidized layer. The AP due to the distributors,
which have to be designed with care, must be added to the AP value above.

7.2 Reactors Used in Refineries

This section will cover examples of the main types of reactors that have been
presented previously, with reference made to industrially important processes
(see Vol. 3).

7.2.1 Steam Cracking

Steam cracking produces ethylene first of all, but also propylene as well.
Collaterally,depending on the feed, it also produces a C, cut rich in olefins and
butadiene and a C,, cut with a high aromatics content. It is therefore an essen-
tial process for petrochemistry. As its name indicates, it achieves pyrolysis in
the presence of steam to take advantage of the high temperature instability of
paraffins and naphthenes compared with respect to olefins and aromatics.
The two major reactions are the cutting of the carbon-carbon bond and
dehydrogenation. They are endothermic and reversible, and therefore pro-
430 Chapter 7 CHEMICAL REACTOR
TECHNOLOGY
t
t I
 Chapter 7 CHEMICAL TECHNOLOGY 431
REACTOR

moted at high temperature. Side reactions of olefin condensation in heavier

products leading to coke formation are superimposed on the two main reac-
tions. The main reactions are promoted by lowering the pressure, the higher
the hydrocarbon partial pressure, the more significant the side reactions
become (Albright et al., 1988).
The recommendations that can be drawn from thermodynamics and kinet-
ics can be summarized as follows:
high heat input at high temperature (higher than SOO'C),
hydrocarbon partial pressure as low as possible,
very short residence time at high temperature in order to prevent side
reactions,
vigorous quenching at the reactor outlet to stop all reactions.
The technique selected to fulfill these different requirements consists in
sending a mixture of hydrocarbons and steam through tubes heated in a fur-
nace. The hydrocarbons are raised to a high temperature and pyrolyzed, and
the resulting products are separated after sudden quenching in a special
exchanger called a TLX (transfer line exchanger) (Fig. 7.17) (Chauvel et al.,
1989). The tubular reactor is in fact best suited t o meeting the requirements
listed above. Given the required temperature, it is difficult to achieve heat
transfer by any means other than radiation. Regarding the low hydrocarbon
partial pressure, it is necessary to introduce dilution steam while keeping the
total pressure slightly higher than atmospheric at the tube outlet.
A wide variety of feeds can be used in steam cracking, ranging from ethane
to vacuum gas oil. Product yields evidently depend on the feed cracked and
this is one of the prime economic elements to be taken into account. The opti-
mum operating conditions will not be the same for all feeds, but the same fur-
nace can be used to process different feedstocks, however. Flexibility with
respect to feeds and therefore operating conditions is consequently crucial.
It has been mentioned that side reactions generate heavy products which
ultimately lead to coke formation. Coke is laid down on the inside walls of the
tube, making regular decoking necessary by combustion of the deposits. The
coke deposition rate increases with the molecular weight of the feed cracked.
For a vacuum gas oil, the operating cycle can become shorter than 10 days,
which is hardly acceptable in industrial practice.
Due to the above-mentioned constraints, steam cracking furnace design
has improved constantly, particularly since 1960 when ethylene demand
began a period of high growth. Pyrolysis furnace manufacturers have made
great efforts to improve furnace performance. From the first furnaces built
during the Second World War up to around 1960, furnaces were designed to
operated with relatively long residence times and low severities. Tubes with
an inside diameter ranging from 90 to 140 mm were laid out horizontally and
reached a total length of some 100 m (Fig. 7.18A). The second generation of fur-
naces made an appearance around 1960, with the use of vertical tubes in the
radiation section which allowed higher temperatures to be attained. The
development in cracking tube design (Fig. 7.18) met the following aims:
432 Chapter 7. CHEMICAL REACTOR
TECHNOLOGY

reduce residence time,

increase outlet temperature,
reduce pressure drop.
The first logical approach is therefore to reduce tube length, with the atten-
dant increase in heat flux. Since heat flux can not be increased indefinitely
(maximum 100 kW/m2), tube diameter must therefore soon be reduced and
tubes must be placed in parallel. This has been a general design trend as
shown in Table 7.2. Tubes with multiple inlets and variable diameters can be
noted in particular. They allow wide variations in cross-sectional area as the
reaction proceeds. This type of configuration has a number of advantages:
(1) In the first part of the reactor, several small diameter tubes in parallel
allow rapid heat transfer in order to reach the reaction temperature. In this
section of the reactor, side reactions can proceed only slowly and the pressure
drop is of little significance.
(2) In the last part of the reactor there is a single large diameter tube, thereby
reducing pressure drop. This is important since that is where the side reactions
are the most rapid. Furthermore, the coke deposit due to side reactions will be
laid down without reducing the tube cross-section too much (Mol, 1984).

c :

II
._...

+ t

-
Figure
7.18 Different types of tubes used in steam cracking.
A. Horizontal tube. B. Vertical tube with uniform diameter. C. Multiple inlet
 Inside Length Residence Ethylene
Designation diameter time capacity Type Constructor
(see Fig. 7.18)
(mm) (m) (s)

90-140 100 >1 Horizontal, uniform diameter (A)

SRT-1 0.6-0.7 30000 Vertical, uniform diameter (B)
SRT-2 HS 0.3-0.35 25 000 Multiple inlet tubes (C) Lummus
SRT-2 HC 0.5-0.6 45 000 Multiple inlet tubes (C)
SRT-3 0.27-0.45 50 000 Multiple inlet tubes (C)
SRT-4 HS 0.33 Multiple inlet tubes (C)
SRT-4 HC 0.25 Multiple inlet tubes (C)
SRT-5 0.23 Multiple inlet tubes (C)
LSUC 50-125 60-70 0.3-0.4 50 000 Constant D
LSSC 50-150 70-80 0.3-0.4 70 000 Variable D, multiple inlets (C) Linde and Selas
LSCC 60-150 45-60 0.2-0.4 70 000 Variable D
USC-u 50-75 0.35 Variable D type U (D) Stone & Webster .U

USC-w 50-95 0.35 Variable D t m e W E )

MK 0.5-0.6 35 000 Vertical, uniform diameter (B)
GK-I 0.3-0.4 50 000 Multiple inlet tubes (C) KTI
GK-I1 0.3-0.4 90 000 Multiple inlet tubes (C)
Millisecond 50-75 10-12 0.08-0.10 Single vertical tube (0 Pullman-Kellom
HMC 50-75 0.3-0.35 MultiDle inlet tubes Selas-Hitachi
M-TCF 0.3-0.6 50 000 Vertical, elliptical cross-section Mitsubishi
Table
1 7.2 I Characteristicsof the major steam cracking furnace tubes. P
0
0
434 Chaoter 7. CHEMICAL REACTOR TECHNOLOGY

Some constructors such as Kellogg (Ennis et al., 1975), have taken this
development to an extreme, producing very short (10-12 m) small diameter
tubes to obtain very short residence times of approximately one one-hun-
dredth of a second (Fig. 7.18F).
Furnaces are designed to have the most uniform heat flux distribution pos-
sible in the radiation section. The tubes are placed vertically in the center of
the radiation section and are heated by burners along the walls and sometimes
a few burners on the furnace hearth or midway up the walls (Fig. 7.19).
Combustion gases ordinarily circulate from bottom to top with the convection
section located at the top of the furnace. The reverse is also possible
(Fig. 7.19D). However, the quenching exchanger (TLX) is usually located above
the radiation section. The radiation sections are sometimes laid out in parallel
in the same furnace and the combustion gases are grouped together in the
same convection section (Fig. 7.20).
Tubes are either placed on the same plane (Fig. 21A) or arranged in a stag-
gered pattern in two parallel planes close to the center of the radiation cham-
ber (Fig. 7.21B). For tubes of the multi-inlet variable diameter type, the two lay-
outs will coexist as shown in Figure 7.21C. The advantage of placing the tubes
in the same plane is that it leads to much more homogeneous heat flux distri-
bution at the circumference of the tubes. Thus, for the same average flux of
70 kW/m2,the local flux varies from 75 to 122% of the average flux at the sur-
face of a tube placed on the same median plane. Meanwhile, if the tube is
placed in a staggered pattern, the local flux varies from 64 to 147% of the aver-
age flux (see Chapter 5). The advantage of the staggered layout is that it
requires a smaller radiation section, with a gain in size of up to 20%.
Complementary to the cracking tube where the reaction occurs at high
temperature, the quenching exchanger at the furnace outlet is also very impor-
tant. All the chemical reactions must in fact be stopped as rapidly as possible,
in particular the side reactions leading to heavy products. As soon as the efflu-
ent exits the furnace, it goes through an exchanger specially designed to lower
the temperature to around 600°C in a fraction of a second. Then it is cooled
further by mixing with a heavy fraction (quenching oil) recycled from the pri-
mary fractionating column. The function of the TLX is also to generate HP
steam at high temperature. Figure 7.22 gives an example of the structure of
such an exchanger which is often seen in actual practice. This type of equip-
ment is used to generate steam between 100 and 140 bar (Schmidt, 1994).
Research has also been done on further shortening the residence time
while at the same time increasing the temperature, hitherto limited by tube
resistance. This is possible only by implementing new ceramic materials.
Other attempts have been made to increase the surface area available for heat
transfer by providing the tubes with internal fins or by giving the tubes non-
circular cross-sections of a varying degree of complexity. Nevertheless, this
aspect is not the only one to be considered given all the constraints of the sys-
tem. In fact although reactor optimization is an important factor, the prof-
itability of the process overall also depends on factors related to operation,
such as decoking frequency.
 TECHNOLOGY 435
Chapter 7. CHEMICAL REACTOR

Combustion gases
A B
TLX
Convectior
zone

Burners

C D

Burners

Furnace feed

Figure
7.19 Differenttypes o f Furnaces.
A. Burners solely on the walls o f the radiation section (Selas, Lummus, Stone
436 Chapter 7 CHEMICAL REACTOR
TECHNOLOGY

Perspective view of a
“Millisecond’’ furnace
(Ke llogg).

Feed

B &&oooooooooooy

Figure
7.2 1 Tube layout inside the radiation section.
A. Two staggered tubes. B. A single layer of tubes. C. Variable diameter tubes
 Chapter 7. CHEMICAL TECHNOLOGY 437
REACTOR

__i
t
B
Vapor

t
Furnace effluent - Water + steam
J

1
Water

1
Figure
7.22 “Schmidt” type quenching exchanger (Twr).
A. Close up ofthe head ofthe dual tubes making up the exchanger between the
438 Chaoter 7 CHEMICAL REACTOR
TECHNOLOGY

The problem of furnace and quenching exchanger coking has been more
particularly a subject of attention in the past ten years. New materials have
been sought for furnace tubes in order to reduce the coke deposition rate.
Moreover, new designs for quenching systems at the furnace outlet have
helped overcome the problems of using heavy feeds such as vacuum gas oils.
Here it should be noted that the trend is to allow for the possibility of using
varied feeds in the future, given the uncertainty as to feedstock availability
and price.
In order to keep up with these developments and reinforce them, important
research has been done on the kinetics of the numerous reactions involved in
steam cracking. The aim is to simulate the reactor as a whole and in particu-
lar predict the influence of feed characteristics (Froment, 1992).
As mentioned earlier, the uncertainty about the future mainly concerns the
feedstocks that will be available and their price. New technologies are already
being studied so as to be able to use the feeds available at the lowest cost,
such as petroleum residues if necessary (Duncan et al., 1992).

7.2.2 Dimerization
In Volume 3, Chapter 9, oligomerization processes were presented and among
them the Dimersol process of olefin dimerization by homogeneous catalysis. It
is a homogeneous liquid phase process whose technology is a good example
of a continuous stirred tank reactor. In the case of propylene dimerization to
produce a mixture of hexenes to be added to the gasoline pool, the reaction
can written in simplified form as follows:

The reaction is exothermic (AH= -20 kcal/mol) and second order versus
the propylene concentration.
The catalyst consists of an organo-metallic complex with a nickel base that
is soluble in the liquid hydrocarbon phase. It is injected at a very low concen-
tration at the same time as the feed containing the propylene. The reaction
temperature is moderate, between 40 and 50°C. The operating pressure is
adjusted to have a liquid phase in the reactor (between 10 and 30 bar). Under
these conditions the reaction is relatively slow and requires residence times of
several hours to obtain a conversion of approximately 90%.
A tubular reactor would be more efficient than a stirred tank reactor, since
the reaction is second order. However, given the large reaction volume
required and the heat exchange that has to be accomplished, the stirred tank
reactor proves to be the most economical, even if several stages need to be
placed in series in order to reduce the total reaction volume.
The reactor is made up of a vertical cylindrical drum with external circula-
tion via a heat exchanger (Fig. 7.23). Several reactors of this type are generally
placed in series. The volume of each drum is determined from the desired
overall conversion, which will be distributed in each stage.
 Chapter 7. CHEMICAL TECHNOLOGY 439
REACTOR

+ Hexenes

I
I Water

Reactor
Volume = V

/-I Recycle

Catalyst

Propylene feed

Figure
7.23 Sketch o f the reactor for hexene production by propylene dimerization.

The recirculation flow rate of each stage must meet two objectives:
(1) Eliminate the heat produced by the reaction by passing through an
external exchanger. The temperature difference that the recycle undergoes
through the exchanger must be only a few degrees (less than l0.C).
(2) Assure mixing inside the reactor so as to approach perfect mixing. It is
accepted that t o do so, the recirculation time must be approximately ten times
shorter than the residence time 8.
Thus the following equations link the different variables of the system, with
the subscript i corresponding to the number of the reaction stage.

(7.36)

(7.37)

with the following definitions:

Q F volume flow rate of the reaction phase, assumed constant
QRi recycle flow rate of stage i
C, propylene concentration in the fresh feed
Xi conversion at the exit of stage i
440 Chapter 7 CHEMICAL REACTOR
TECHNOLOGY

pL density of the reaction phase, assumed constant

C,, specific heat of the reaction phase
AHR enthalpy of the dimerization reaction
The successive reaction stages can be chosen with identical volumes but
different partial conversions, or with the same partial conversion and different
volumes. The second choice has the advantage of leading to identical exchang-
ers for each stage.
The overall conversion achieved is approximately 90%,but it can be advan-
tageous to pursue the reaction further. In this case, a tubular reactor can be
contemplated, given the slight thermal effect at this stage of the conversion
which is very slow. A simple tube, without any heat insulation can then be
used (Fig. 7.24).

7.2.3 Gas Scrubbing by Chemical Absorption

7.2.3.1 Sour Gas Scrubbing
As already pointed out, a good example of a gas-liquid reactor is the chemical
absorption of sour gases CO, and/or H,S. Gas scrubbing is found in the hydro-
gen manufacturing process (see Vol. 3) or in facilities for producing natural gas
which often contains both H,S and CO,. The principle of these processes is
absorption under pressure in a liquid phase, followed by desorption at a lower
pressure. Ordinarily however, the liquid phase is a basic solution that reacts
rapidly with CO, and H,S. Advantage is thus taken of the fact that gas-liquid
mass transfer is accelerated by the chemical reaction in the liquid phase (see
Section 7.1.3.1). The absorbent solutions used are aqueous solutions of
monoethanolamine (MEA), diethanolamine PEA), methyl-diethanolamine
(MDEA) or basic carbonates (K,C03 for example).
Some processes are purely physical, using solvents such as sulfolane,
propylene carbonate, N-methylpyrrolidone or acetone. The advantage of phys-
ical processes appears mainly at high pressure, since purely physical solubil-
ity increases proportionally with pressure. On the other hand, the capacity of
a solution reacting with CO, depends only slightly on the pressure.
Figure 7.25 shows a comparison of CO, equilibrium curves with different
liquid phases. It can clearly be seen that at low CO, partial pressures the sys-
tems implementing a chemical reaction are superior. The solubility curves for
these systems include two parts:
the first with a low slope resulting from the chemical reaction that is lim-
ited by the chemical equilibrium to a greater or lesser extent;
the second part with a slope determined by the solubility of CO, in the
solvent without reactant.
As for the absorption rate, according to Section 7.1.3.1 the aim is the fastest
reaction possible in order to obtain a high acceleration factor. If the different
amines are compared from this standpoint, the most reactive is MEA, followed
by DEA and MDEA which is the least reactive of the three.
 Chapter 7 CHEMICAL REACTOR
TECHNOLOGY 441
-
-LA-
cu
a,
U
Ba,
L
442 TECHNOLOGY
Chapter 7 CHEMICAL REACTOR

-
-7Comparison o f various
CO, absorption solvents.

B. N-methylpyrrolidone
(40°C).
C Aqueous solution o f
S2 SI 50 C O content
~ in 15% wt MEA (40°C).
theliquid phase D. Sulfinol process
(free + combined) solvent (sulfolane).
E. Aqueous caustic soda.

The reaction that takes place during CO, absorption can be written as fol-
lows:
CO, + 2 RNH, + RNHCOORNH,
As long as the number of moles of CO, absorbed is smaller than half the
number of moles of amine present in the solution before the onset of absorp-
tion, it can be considered that only the above reaction takes place. The reac-
tion rate can be expressed by:
r = k[CO,J.[RNH,J
Due to the reaction stoichiometry (1 mole of sour compound for 2 moles of
amine), the recommended amine circulation flow rates are such that the molar
ratio of sour gas to amine is between 0.3 and 0.4. This constraint is often
expressed in percent of approach to equilibrium in the amine solution exiting
the absorber, between 65 and 75%.
The kinetic constant k varies considerably depending on the amine. The
following relative values can be noted:
MEA 30
DEA 15
MDEA 1
Operation is therefore conducted in such a way that the reaction can be
considered instantaneous. Under these conditions the flux of A, (CO, or H2S)
can be expressed as follows (Fig. 7.3):
"1Ix=0 = ~ I L E C ~ L (7.38)
or expressing the acceleration factor E for an instantaneous reaction:
VIA, + vZA, *A,

(7.39)
 TECHNOLOGY
Chapter 7 CHEMlCAL REACTOR 443

For the same equipment it will thus be advantageous to increase C;, and
c,,b. Although c,,b can be fixed freely (the amine concentration for example),
it should not be forgotten that when the viscosity increases, the transfer coef-
ficient k , , decreases. As a result, an optimum value of C,, will have to be cho-
sen, which is approximately 15 to 30%weight for MEA or DEA in an aqueous
solution. Moreover, it should be noted that to obtain the same value of C2Lbl
less weight of MEA is required than of DEA or MDEA. However, because of the
higher reactivity of MEA (reaction with COS and corrosion), DEA or MDEA are
often preferred.
It may be advantageous to use a solvent other than water in order to
increase the value of CTL The Sulfinol process utilizes sulfolane as a solvent,
in which CO, is over three times more soluble.
The technologies employed for both absorption and desorption are of the
counter-current column type (with plates or packing), similar to those
described in Chapters 1 and 2. Plate columns are generally chosen for large
sizes. The diameter is conventionally calculated based on the superficial
velocity uG.It is recommended to apply a safety coefficient of 80% to the value
of uG to take any foaming into account. The number of plates is generally cal-
culated on the basis of the equilibrium curves, taking the thermal effect
related to the reaction into account. For packed columns the diameter is cal-
culated from the flooding correlations. Meanwhile, the packing height is cal-
culated on the basis of gas-liquid mass transfer coefficients, taking into
account the acceleration due to the reaction.
A number of precautions must be taken concerning materials because of
the aggressivity of the gas or liquid phase, especially in the hottest sections.
Generally speaking the column shells are made of carbon steel, whereas the
plates, reboilers and some pumps are made of stainless steel alloy. Packings
may be made of ceramic or preferably of polypropylene which is lighter.

7.2.3.2 Clauspol Process

Another example of a gas-liquid reactor can be found in the Clauspol process
(see Vol. 3). In this case the Claus reaction is carried out:
2SH, + SO, + 3s + 2 H 2 0
at low temperature in a liquid phase containing a soluble catalyst. Here H,S
and SO, are absorbed simultaneously in the liquid phase and the reaction
takes place within the liquid phase. This reaction is very rapid and the con-
centration profile near the gas-liquid interface can be represented as indicated
in Figure 7.26 given the relative solubility of H,S and SO, (Deschamps et al.,
1972). The SO, concentration is practically constant within the liquid phase,
while the H,S concentration rapidly approaches zero inside the liquid film. The
reaction can then be treated as a very rapid first order reaction and therefore
limited by the gas-liquid mass transfer. The operation is carried out in a
packed column whose volume is calculated from the interfacial area and the
gas-liquid mass transfer coefficients. The process employs packing, often of
the 3/4 inch ceramic lntalox saddle type.
444 Chapter 7, CHEMICAL TECHNOLOGY
REACTOR

Concentrations

Figure
SO, and H2S concen-
tration profiles during
simultaneous absorp
tion in a liquid phase
containing a dissolved
catalyst (Clauspol pro-
cess).

INTERFACE

7.2.4 Aliphatic Alkylation

The aliphatic alkylation processes have already been presented and so the
only aspect considered here is the reactor (see Vol. 3 for details on the process
flow schemes).
From the standpoint of the chemical reaction, the principal one is between
isobutane and light olefins (butenes mainly):
H H H CH3H H H
I I I I I I I
H3C-C-CH3 + CH,=C-C-CH, -+ H3C-C-C-C-C-CH3
I I I l l 1
CH3 H H CH3H H
which is catalyzed by a strong acid (sulfuric or hydrofluoric acid). Conse-
quently, there are two categories of processes depending on the acid used:
the Stratco and Exxon processes with sulfuric acid,
the Phillips and UOP processes with hydrofluoric acid.
 Chapter 7. CHEMICAL TECHNOLOGY 445
REACTOR

Whatever the acid used, the main characteristics are:

a reaction in the presence of two liquid phases (hydrocarbons and acid),
a rapid reaction in the acid phase, controlled by reactant transfer (isobu-
tane and olefins) from the organic phase to the acid phase,
the reaction must proceed at low temperature (lower than 10°C for sul-
furic acid) in order to prevent side reactions between the olefins.
Referring to the discussion in Section 7.1.3.2 about reactors with two fluid
phases, the first thing to do is calculate the Hatta number to determine the
regime. This has been done for the case of sulfuric acid which has been more
thoroughly studied experimentally (Borges, 1995). The Hatta number is at
least approximately 40. As a result, it is such that the reaction occurs wholly
in the acid film near the interface and the rate of reaction is controlled by the
interfacial area between the two phases.
Furthermore, the solubility of isobutane and of butene-1 is:
2 x lo-* mol/m3 for isobutane and
5x mol/m3 for butene-1.
The concentration profiles near the interface are presented schematically
in Figure 7.27. Since isobutane is greatly in excess, its concentration in the acid
film varies only slightly and the reaction becomes pseudo-first order. The
butene concentration profile can be calculated by means of a simple molar bal-
ance in an acid film element. The following is obtained:

(7.40)

with:
k pseudo-first order kinetic constant
gBn diffusivity of butenes in the acid
CBH concentration of butenes in the hydrocarbon phase
CBA concentration of butenes in the acid phase
x distance from the interface
According to recent estimates (Christensen et al., 1993), the ratio k / 9 B Acan
be taken as equal to 7 x lOI3 m-*. CBH is approximately 3.6 x lo3 mol/m3 at the
reactor inlet. As such, for x = 1 ym, the butene concentration is reduced by a
factor of approximately 4000. The butenes are therefore totally consumed in
the film near the interface and consequently do not take part in side reactions
in bulk of the acid phase.
Hence the total butene absorption flux, NB,which controls the main reac-
tion rate can be expressed by the relation below:

(7.41)

with A, being the interfacial area per unit of volume (m2/m3).

446 Chapter 7 CHEMICAL REACTOR
TECHNOLOGY

Concentrations

i-C4 - Acid film

b

--
1 - 1 - 1 1

I I1 1 II1

Droplet
ACID PHASE
HYDROCARBONS

Olefins
CBH

Figure
7.27 Isobutane and butene concentration profile near the interface.

In this case the determinative parameter is the overall interfacial area

developed in the reactor and should be as large as possible. The choice of
reactor type must therefore be dictated by this requirement.

7.2.4.1 Processes Using Sulfuric Acid

a. Stratco Process
The Stratco reactor consists of a horizontal drum equipped with a turbine of
the marine propeller type at one end and a tube bundle in the rest of the vol-
ume (Fig. 7.28). The mixture of isobutane plus olefins and the sulfuric acid
returning from the settler are introduced in the front close to the propeller that
stirs the whole reactor. The two phases are sucked in by the propeller and form
an emulsion that circulates along the drum walls and then goes through the
whole length of the tube bundle which eliminates the heat of reaction. Since
the acid is in excess, it is thus the continuous phase. Recent measurements of
the interfacial area versus the phase ratio have shown that the maximum inter-
facial area is obtained with an acid ratio of approximately 75% volume.
A horizontal drum located above the reactor serves as a settler and the set-
tled acid returns into the reactor by gravity. Hence the phase ratio in the reac-
tor is established by itself at a value ranging from 50 to 60% volume of acid,
subsequent to the gravity circulation between the reactor and the settler. The
hydrocarbon phase contains alkylate, a large amount of excess isobutane (70-
80%) and a very low olefin concentration, thereby limiting polymerization side
reactions.
Chapter 7. CHEMICAL TECHNOLOGY 447
REACTOR
448 Chaoler 7 CHEMICAL REACTOR
TECHMOLOGY

Several reactors are often coupled with the same settler whose volume is
calculated to give a settling time between 30 min and 1 h. The higher the acid
ratio, the slower the settling related to the emulsion viscosity. Recycling the
organic phase in the settler allows the settling time to be shortened and sec-
ondary products (soluble oils) to be reduced at the same time.
Inside the reactor, whose volume is approximately 35 m3 (for production of
approximately 15 m3/h of alkylate), the high circulation flow rate induced by
the propeller would normally make concentrations homogeneous throughout
the volume. However, due to the rate of reaction, olefins disappear apprecia-
bly between the inlet in the turbine and the arrival in the tube bundle where
partial coalescence of the emulsion is promoted. The emulsion residence time
in the peripheral reactor zone between the turbine and the exchanger is
approximately 30 s. The average residence time in the reactor is approxi-
mately 1000 s.
Due to the mechanical power input by the turbine, approximately
7.5 kW/m3 of reaction mixture, the mixture constitutes an emulsion of fine
hydrocarbon droplets within the acid phase. Based on the recent results
obtained by Am Ende et al. (1995), the size of the hydrocarbon droplets can be
estimated at between 50 and 100 pm. From the calculation of olefin depletion
in these droplets presented by Christensen et al. (1993), the following is
obtained by combining equations (7.40) and (7.41):

(7.42)

dg:average diameter of hydrocarbon phase droplets.

A maximum contact time of 15 s is thus required for a conversion of 99.9%.
The contactor can not be considered as a stirred tank reactor, since the
reaction time is much shorter than the average residence time. As such, the
reactor volume may be broken down into two zones:
the peripheral zone where the emulsion circulates with a plug type flow
and over 99% of the olefin conversion takes place;
the recirculation zone through the tube bundle where the heat exchange
is accomplished.
The reaction products are withdrawn to the settler between these two
zones. The variation in olefin concentration in the hydrocarbon phase can be
schematically represented as shown in Figure 7.29.

b. Exxon/Kellogg Pvocess
The Exxon process uses cascade type reactors originally designed by Kellogg.
In series mechanical mixers are placed in neighboring compartments incorpo-
rated in a horizontal drum (Fig. 7.30). One of the characteristics of this process
is the fact that the heat of reaction is eliminated by evaporation of part of the
excess isobutane, feeding the cascade of mixers. Like the isobutane, the sulfu-
ric acid is introduced at the cascade inlet and circulates from one compart-
 Chapter 7. CHEMICAL TECHNOLOGY 449
REACTOR

To settler Acid from settler

t Peripherical zone
I
Tube bundle zone

-It
t
Hydrocarbons
Olefin concentration
Turbine

Periphery

L Contactor length 0

Figure

I I
7.29
Schematic variation in olefin concentration in the hydrocarbon phase inside
the Stratco contactor.

ment to the next by overflowing. The olefin plus isobutane mixture is divided
up into aliquot parts which are introduced in each compartment in parallel.
Each compartment is vigorously stirred to generate the interfacial area
required for mass transfer and disperse the olefins within the hydrocarbon
phase. The phases to be mixed are introduced just below the turbine which
also draws the acid-rich emulsion back in. A gas phase consisting essentially
of evaporated isobutane exits at the top of each compartment. At the gas-liq-
uid interface, partial settling of the emulsion has already taken place and the
overflow to the next compartment is therefore richer in hydrocarbons than
the emulsion produced in the turbine. This partial settling allows a higher
ratio in each compartment between the acid and hydrocarbon phases than the
ratio that would be dictated by the respective flow rate of the phases.
The same process as described earlier operates at the hydrocarbon-acid
interface, causing the feed olefins to disappear rapidly. Here circulation times
are shorter than in the Stratco contactor, but olefin concentrations are also
lower in the hydrocarbon phase.
450 Chapter 7. CHEMICAL REACTOR
TECHNOLOGY

l
I
A Cooling fluid - Vapor compressor Debutanizer

isobutane
recycle Acid recycle
I
x

Figure
7.30 I Diagram of the Exxon sulfuric acid alkylation reactor (Albright, 1990).
A. Overall view of the reactor. 8.Close up of the contact cells.

7.2.4.2 Processes Using Hydrofluoric Acid

Alkylate production using hydrofluoric acid a s a catalyst is implemented in
two processes: Phillips and UOP. They are different from sulfuric acid pro-
cesses in that they d o not employ mechanical stirring systems. The low vis-
cosity of HF, reduced even more by high the isobutane solubility, allows the
phases t o be emulsified much more easily.

a. Phillips Process
This process flow scheme was given in Volume 3, so the reactor alone will be
considered here. It is a tubular reactor with two liquid phases (Fig. 7.31). The
reaction mixture is introduced at the base of the vertical tube at high velocity
by several injectors. The kinetic energy provided by the injected hydrocar-
bons allows the phase to be finely dispersed in the acid which is recycled by
 Chapter 7 CHEMICAL TECHNOLOGY 451
REACTOR

1 Figure
7.31 Sketch of the Phillips HF alkylation reactor.

gravity from the settler located in the upper part of the reactor. The hydro-
carbon phase is kept dispersed throughout the reactor by means of perforated
plates. The circulation of the phases is thus the result of injecting the organic
phase and of the difference in specific gravity between the emulsion and the
recycled acid.
The acid to hydrocarbon volume ratio can vary from 1 to 4, depending on
operating conditions. Due to olefin dilution by a significant isobutane recycle,
the reactor is practically isothermal. The contact time in the reactor is approx-
imately 20 to 40 s.
The calculation performed earlier for sulfuric acid can be applied by adapt-
ing the values of the physical constants. Few details are given on this subject
in the literature. However, it can be assumed that reaction rates are approxi-
mately the same a s with sulfuric acid if not higher, since the temperature and
the isobutane concentration in the acid phase are higher. In addition, butene
diffusivity in HF is much greater because of the viscosity, which is 130 times
lower than that of sulfuric acid.
452 Chapter 7. CHEMICAL REACTOR
TECHNOLOGY

Alkylate

.
Water
Settler

Reactor
Sketch of the UOP HF
alkylation reactor with
one reaction stage.

Acid

lsobutane recycle

Olefins
+ lsobutane
~

b. UOP Process
In this process the reactor consists of the shell part of a heat exchanger in a
vertical position (Fig. 7.32). The hydrocarbon mixture is introduced at differ-
ent levels in the exchanger by nozzles, which assure fine dispersion. The acid
returning from the settler, by gravity similar to the previous case, enters the
lower part of the reactor.
The principle therefore remains the same as for the Phillips process, except
that the temperature is controlled by heat exchange inside the reactor itself.
Depending on the capacity of the unit, one or two reactors may be used.
When there are two reactors, they work in series with settlers associated with
each one (Fig. 7.33). The acid and hydrocarbons exiting from the first settler
are sent to the second reactor. In contrast, the fresh feed is divided up between
the first and the second reactor, thereby increasing the isobutane to olefin
ratio in each reactor.
Alkylation by HF has also been proposed in Stratco contactors.

7.2.5 Hydrogen Production

The various hydrogen production processes were presented in Chapter 14,
Volume 3. They are in fact a sequence of successive steps, most of which con-
 Chapter 7 CHEMICAL TECHNOLOGY 453
REACTOR

Alkylate

8 Settler

Reactor
stage 2

Figure
Sketch of the UOP HF
alkylation reactor with L
two reaction stages.

Reactor
stage 1

m
A -

Olefins -

Fresh lsobutane recycle

lsobutane - _
L

sist of catalytic reactors. Some of these steps will be taken as examples of cat-
alytic reactors: steam reforming, CO conversion and methanation, and their
essential technological elements will be presented.

7.2.5.1 Steam Reforming

This is generally the first step in a hydrogen production process and uses
methane or light hydrocarbons (LPG or naphtha) as feed. Methane reforming
will be discussed here, which is currently the most widespread.
454 Chapter 7 CHEMICAL TECHNOLOGY
REACTOR

The catalytic reaction that takes place is written as follows:

CH, + HZO % CO + 3H2
Since it is a highly endothermic reaction (AH=49 kcal/mol), it must be car-
ried out at the highest temperature possible (over 800°C at the reactor outlet).
The main reaction is accompanied by another important reaction:
CO + HZO % CO, + Hz
which is exothermic (AH = -10 kcal/mol).
The reactor effluent is thus composed of a mixture of hydrogen, carbon
monoxide, carbon dioxide and unconverted methane and steam. Moreover, it
is impossible to prevent some carbon from being formed and laid down on the
catalyst.
The catalyst, with a nickel on alumina or calcium aluminate base, is most
frequently in the shape of rings. The outside dimensions vary from 15 to
25 mm, whereas the inner orifice has a diameter of 6 to 9 mm. Some catalysts
are in the shape of beads or pellets perforated with 4, 5 or 6 holes (Fig. 7.34).

Figure
7.34 Common shapes o f steam reforming catalysts: rings or pellets perforafed with
holes.

The catalyst is loaded in vertical tubes placed inside the radiation chamber
of a furnace (see Chapter 5). The types of furnace found in actual practice are
similar to those described for steam cracking with a vertical cracking tube,
since with the catalyst this is the only acceptable tube position.
The special steel tubes have an inside diameter of between 80 and 130 mm
and a length of between 10 and 15 m. They can be arranged in one or two ver-
tical planes inside the radiation chamber. The furnaces differ mainly in the lay-
out of the burners along the furnace walls or at the bottom or top of them. The
resulting temperature profiles depend evidently on burner distribution in
space (Meyers, 1997). Burners regularly arranged along the vertical walls of
the combustion chamber would seem to give the best temperature profile
along the tube (Dybkaer, 1995). The tube skin temperature, which may exceed
900°C, controls the heat flux supplied to the catalyst for the reaction. The flux
must be especially high in the part of the tube located between 20 and 40% of
its length as this is where the rate of reaction is the fastest.
 Next Page

Chapter 7. CHEMICAL TECHNOLOGY 455

REACTOR

Tubes filled
with catalyst

Sketch o f a steam
reforming furnace with
side burners (called a
terrace furnace)
(Meyers, 1997). Burners

*
Outlet collector

Because of the high temperatures at which the tubes work, care must be
taken to eliminate excessive mechanical stress. The expansion in tube length
is in fact approximately 25 cm at the working temperature. So that this elon-
gation can occur, the tubes are supported at the bottom where they are con-
nected to a collector (Fig. 7.35). Most of the tube weight is supported by a
spring connected to the upper flange in order to reduce buckling risks. The
tubes are fed from the top by flexible lines to allow them to expand freely.
Sometimes the bottom of the tubes is connected to the collector by flexible
lines as well (Dybkaer, 1995).
The catalyst is loaded inside the tubes with the greatest care. Because of
the large number of tubes in parallel, it is crucial for the pressure drop in each
tube to be exactly the same so as to obtain uniform flow rate distribution. This
means loading the same amount of catalyst in each tube and producing no cat-
alyst fines during the loading procedure. This is made possible by weighing
exactly the same amount of catalyst placed in each tube and filling it by means
of a cloth sleeve (sock) that allows the catalyst to be introduced gently into
each tube. It should be noted that the ratio between the inside tube diameter
and the characteristic catalyst dimension does not always meet the criterion
given in Section 7.1.4.2.a in the part on “Sizing the catalyst bed”, i.e. d,/d,, > 20.
As a consequence, the formulas presented for calculating the pressure drop or
mass transfer coefficients must be used with care. The Ergun formula allows