CatalysisToday, 1 (1987) 587-607 587

Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands

MALEIC ANHYDRIDE FROM C-4 FEEDSTOCKS USING FLUIDIZED BED REACTORS

R. M. CONTRACTOR and A. W. SLEIGHT

Central Research and Development Department, E. I. du Pont de
Nemours & Company, Wilmington, DE 19898, USA

INTRODUCTION
The predominant commercial route to maleic anhydride from C-4
feedstocks is the vapor phase oxidation of n-butane over
vanadium-phosphorous-oxide(VP01 catalyst using fixed bed
reactors. Fluidized bed processes have been developed (refs. l-4)
because of some advantages offered by fluidized bed reactors over
fixed bed reactors.
In this chapter, we attempt to provide a comprehensive
critical review of fluidized bed processes for producing maleic
anhydride from C-4 feedstocks. The catalyst for the production of
maleic anhydride, the reaction kinetics, the reactor design, heat
removal, product recovery, and product purification features of
the commercial fixed bed processes have been discussed in the
previous chapters of this issue. We discuss the catalyst and
engineering aspects of fluidized bed processes with emphasis on
the differences with the corresponding fixed bed systems and their
impact on the overall economics of maleic anhydride production.
The single vessel, dense phase fluidized bed reactors as well as
circulating systems involving lean phase fluidization are
reviewed.

FLUIDIZED BED PROCESSES

The common features of maleic anhydride production by the
fixed bed or the fluidized bed route include the reactor for
partial oxidation of C-4 feedstocks, the maleic anhydride recovery
and the purification systems. There are, however, some important
differences, primarily in the reactor design, which affect the
investment and production costs. The large exotherm and the
thermal sensitivity of the reaction makes the reactor design of
critical importance.

Advantases of the Fluidixed Bed System

In fluidized bed reactors, air is fed to the reactor bottom,

0920-5661/67/$07.35
0 1987Elaevier~ien~ePublishersB.V.
fluidizing the catalyst. n-Butane is vaporized and injected
directly into the fluidized catalyst bed. Rapid mixing of the
catalyst particles in the fluidized bed prevents formation of
localized hot spots resulting in uniform temperature throughout
the bed with precise temperature control, even though hydrocarbon
and air are fed separately. Use of fine catalyst particles
results in better utilization of the particle surface and in
reduced temperature and concentration gradients within the
particles. Yield losses associated with hot spots are avoided,
and the catalyst life is increased. The heat of reaction is
easily removed by direct steam generation in the cooling coils
immersed in the catalyst bed. Cooling surface area is relatively
small because of high heat transfer coefficient on both the water
side and the process side of the coils. In contrast, the heat
removal and temperature control in the fixed bed system requires
the use of complex multitubular reactor, low feed concentration
and expensive heat transfer equipment.
Hydrocarbon concentrations well within the flammable region
are used by introducing the feed directly in the fluidized bed.
The heat capacity of the catalyst and the rapid heat transfer by
the movement of the catalyst particles prevents formation of a
flame front. Close to stoichiometric concentration (about 4 mol%
n-butane) of hydrocarbon in air is used. In contrast, in the
fixed bed reactors, n-butane concentrations of less than 2 mol%
are used to prevent formation of an explosive mixture. The
reduced air rate in fluidized bed processes reduces the compressor
investment and utilities costs, reduces the reactor diameter, and
forms a more concentrated product stream. This, in turn, reduces
the size of the effluent cooler, permits recovering a substantial
fraction of maleic anhydride by condensation, and reduces the size
of the absorbers for recovering the remaining product in the gas
stream.
High per pass conversion to maleic anhydride is used on a once
through basis. However, all n-butane oxidation processes operate
at less than 100% conversion in order to maximize yield. Because
of the lower air requirement of the fluidized bed process, the
vent gas from the scrubler has a high heating value and is
incinerated thermally without the need for a catalytic incinerator
or addition of supplementary fuel.
Loading and unloading of the catalysts is much easier in the
fluidized bed systems. The fluidized bed reactors are scalable to
very large diameter permitting economies of scale. The maximum
capacity of a single, multitubular fixed bed reactor is limited to
about 20,000 metric tons/year. Capacity of about 40,000 metric
tons/year is possible with the fluidized bed reactors. Fluidized
bed processes result in lower investment particularly for a large
scale plant.

Commercial Fluidized Bed Processes

A number of companies have developed fluidized bed processes
for maleic anhydride production. In 1970, Mitsubishi started a
commercial fluidized bed plant based on the oxidation of butenes
in crude C-4 fraction. Badger, Alusuisse Italia/LummusCrest and
Sohio/UCB have demonstrated fluidized bed processes for n-butane
oxidation on large demonstration plants and offer them for
licensing. These processes use single-vessel,dense phase
fluidized bed reactors. The general outline of the fluidized bed
process discussed above apply to all of the developed processes.
Different companies have their own proprietary product recovery
and purification systems developed for their fixed bed processes.
Essentially the same systems are used for their fluidized bed
processes with appropriate modifications and improvements
attributable to higher concentration of maleic anhydride in the
effluent of the fluidized bed reactor. The companies use their
own proprietary catalyst and reactor design, the details of which
are not available. Some specific features and the claimed
advantages of the processes are summarized below.

(i) Mitsubishi orocess. A schematic flow diagram for this

process is shown in Figure 1: The process (ref. 1) uses crude C-4
fraction from a naphtha cracker as a feed. The plant has a
nameplate capacity of 18,000 metric tons/year. It is the only
commercial plant using a fluidized bed rector for the production
of maleic anhydride.
In the reactor section, C-4 fraction is fed with air into the
reactor for conversion to maleic anhydride in contact with the
fluidized catalyst. Concentration of C-4 fraction in feed air is
described as high. The heat of reaction is removed by cooling
coils in the reactor generating high pressure (430-710 psi) steam.
The reacted gasl containing maleic anhydride is absorbed by
water in a quench tower so that an aqueous solution of maleic acid
is formed.
590

Off-gas from the quench tower, containing a small amount of

byproducts is incinerated in the wastegas combustion chamber
generating more high pressure steam.
In the dehydration/purification section, maleic acid is
concentrated and dehydrated by their process which operates
effectively in a short time. Crude maleic anhydride is purified
by a simple distillation to meet required specifications.
Their catalyst is said to possess high selectivity, long
durability under high C-4 concentration, and low attrition.
The process is considered feasible and economical, especially
when cheap and rich C-4 hydrocarbons are available from a naphtha
cracker and when a large amount of surplus high pressure steam can
be used at the same site.

Waste Gas
Product
(Fluid)

Waste Gas Combustor

Quench Tower

1 @j!Ef!&

Dehydrator Detillation Column Pelletlzer

Fluidized-Bed Reactor

F ig. 1. Mitsubishi Chem ical process from reference 1.

(ii) ALMA process. The ALMA Process was developed jointly by

Alusuisse Italia and Lummus Crest. A flow diagram of the ALMA
Process (ref. 2) is shown in Figure 2. It uses n-butane as a feed
stock. n-Butane and air are fed separately in the fluidized bed
reactor to give about 4 mol% n-butane concentration.
 591

Fig. 2. ALMA Process from reference 2.

The process uses their proprietary anhydrous maleic anhydride

recovery system using an organic solvent. The organic solvent has
a higher boiling point than maleic anhydride. 'The exclusion of
water eliminates byproduct formation experienced with an aqueous
recovery system. It allows the entire maleic anhydride production
to be separated from the reactor effluent in a single absorption
step.
The process has been demonstrated on a 3,000 metric tons/year
semi-commercial plant using proprietary attrition resistant
catalyst manufactured in commercial production facilities. The
process is claimed to have high yield, low air rate and catalyst
inventory, and negligible activity decline or catalyst losses by
attrition. Alusuisse Italia has reported that it is proceeding
with plans for its own ALMA plant. The process is also offered
for licensing.
Their economic analyses show that for a 40,000 metric
tons/year capacity plant, the maleic anhydride transfer price is
roughly 12% lower for the ALMA process relative to their fixed bed
process. They claim that the cost improvements of their process,
combined with the raw material savings for n-butane versus benzene
allow the ALMA process to compare favorably with fully depreciated
benzene-based existing plants.

(iii) Badser process. Badger developed their process (ref. 3)

with Denka Chemical Corporation. A flow diagram of the Badger
Process is shown in Figure 3. They operated a demonstration plant
592

in 1981 at Denka's MA plant in Houston, Texas. Information on the

demonstration plant reactor size is not available. The process is
offered for licensing.

To Incinerator Dehydrator Refining

Fluid Bed
Reactor
n

Butane

Maleic Anhydride

Residue
Air

0_7

Air
Compressor

Fig. 3. Badger Process from reference 3

In the Badger Process, a substantial fraction of maleic

anhydride is condensed and recovered as a liquid by cooling the
reactor effluent and the rest of the product is absorbed in water
as maleic acid. The acid is dehydrated to anhydride preferably
using a water entraining agent such as xylene. They claim that
the fluidized bed catalytic oxidation of n-butane using the Badger
Process produces maleic anhydride at good yields with reduced
utilities requirements and with a substantially lowered capital
investment. This is attributed to discovery and development of
attrition resistant catalyst whose performance does not change
with time, together with an optimum fluidixed bed reactor design.

(iv) Sohio/UCB process. A flow diagram of the process is

shown in Figure 4. The oxidation catalyst for the process was
developed by Sohio (ref. 4). They claim that the catalyst gives
over 50% yield of maleic anhydride on a once-through basis.
As with the Badger Process, up to about 50% of the maleic
anhydride is continuously condensed out by cooling the effluent
gases below the dew point of the maleic anhydride, but above the
dew point of water. Their specially designed condenser never
needs to be put out of service for washing. The remaining product
is absorbed in water as maleic acid.
 593

Absorpiion Drying Dehydration Distillation

Tail Gas Pure Maleic Anhydride

Residue
Crude Maleic Anhydride

Fig. 4. Sohio/UCB Process from reference 4.

They prefer not to use any organic solvent as an entrainer for

dehydration step. Instead the crude maleic acid solution is first
evaporated under vacuum, and afterwards thermally transformed to
anhydride in a dehydration reactor specially developed to reduce
the isomerization of maleic acid to fumaric acid.
Sohio has tested their fluidixed bed oxidation reactor system
in a large-scale pilot plant facility. UCB has commercially
operated the recovery and purification system with maleic
anhydride produced in other fixed bed processes. Sohio has
recently announced (ref. 5) plans for a commercial plant based on
the SohioAJCB Process.

Limitations of the Fluidixed Bed Systems

In spite of these advantages of the fluidized bed processes,
no new commercial plant based on fluidixed bed technology has been
built for maleic anhydride production since Mitsubishi Chemicals
started its plant in 1970. Many maleic anhydride producers are
not convinced of the overall merits of the technology. One reason
for the lack of acceptance of the technology may be insufficient
growth in demand for maleic anhydride to justify construction of a
plant large enough to achieve economies of scale. Furthermore,
there are some technical problems with the fluidixed bed processes
relative to fixed bed systems, and not everyone is persuaded that
all the problems have been solved. The problems include catalyst
attrition, low yields, and scale-up uncertainties. Discussion of
these problems and the approaches taken to solve them are covered
in the next few sections.
594

CATALYST FOR FLUIDIZED BED

Catalyst
As with the fixed bed processes, VP0 catalyst is used for the
fluidized bed processes to produce maleic anhydride from C-4
feedstocks. To achieve better gas-solid contact in fluidized bed,
it is a common practice to use fine particles in the size range of
about 20 to 150 microns. The catalysts are preferably prepared by
spray drying to form readily fluidixable spheroidal particles.
Although there are numerous patents and publications on VP0
catalyst for oxidation of C-4 feedstocks to maleic anhydride, only
a few patents (ref. 6-7) exist on VP0 catalyst specifically for
fluidized bed reactors. The chemical properties of the VP0
catalyst precursor used for spray drying is essentially the same
as that for preparing fixed bed catalysts. The spray dried VP0
catalyst particles do not have sufficient attrition resistance to
withstand forces in large fluidized bed reactors. Discussion on
catalyst attrition in fluidized bed reactors and approaches taken
to overcome the attrition problem follows.

Attrition
(i) Source of Attrition. Particles are subject to attrition
wherever they are in motion, and particularly when the motion is
abruptly changed. Besides such kinetic stresses, attrition may
also be caused by thermal and chemical stresses in the reactor.
Although the particles are subject to kinetic stresses in all
areas of a fluidized bed reactor, the principle areas of attrition
are the high velocity jets of the gas distribution grid zone and
the cyclone entrance or within its barrel where particles impact
the walls and rub against the barrel.

(ii) Attrition Tests. Many methods of assessing attritability

of particles have been employed in the past (ref. 8). These
methods are often used arbitrarily, an empirical correlation
between the testing method and the commercial reactor being
established. In some cases, the test is intended to simulate the
stresses in a particular type of operation. For example, the
submerged jet attrition test (ref. 8) intended to simulate
stresses in the grid zone, is often used to assess attrition
resistance of the catalyst in a fluidised bed reactor. The
attritability of a particle is likely to depend on parameters such
as the forces experienced by the particles, the rate of
application thereof and the physical properties of the particles.
However, much remains to be understood about the controlling
mechanism of attrition, laboratory assessment of attrition, and
the relationship of the laboratory test data with particle
attrition in the commercial reactors.

(iii) Effects of Attrition. For the gas velocities and the

particle size typically used in fluidized bed catalytic reactors,
the catalyst entrainment from the fluidized bed to the cyclones
can be as high as 4 to 8 Kg/meter3 of gas. The multistage cyclone
systems used in these reactors recover essentially 100% of the
course particles entrained in the gas stream and return them to
the bed. However, a small fraction of the fine particles (~45,~)
and a large fraction of the dust (<~O,U particles) produced by
attrition escape the cyclone system and are carried out of the
reactor in the effluent stream. Periodic make-up is required to
maintain desired catalyst inventory in the bed. In addition to
the cost of make-up catalyst, the catalyst losses entail
additional costs and problems in the downstream processing and
catalyst disposal.

(iv) Reactor Modifications to Minimize Attrition. Reactor

design modifications are often employed to minimize catalyst
attrition in fluidized beds. Sometimes a shroud is installed
around each orifice in the grid plate to decrease the effective
jet velocity to which the particles are exposed. Catalyst
attrition in the cyclone area is minimized by reducing the cyclone
inlet velocity and increasing cyclone diameter on the first stage
of the multistage cyclone system, Attrition in the cyclones can
be eliminated by replacing the cyclone system with porous metal
blow-back filter system (ref. 9). Besides higher costs and
operational complexity associated with the use of such filters,
there is a potential problem of uncontrolled reaction occurring on
the filter surface. Moreover, any dust produced by attrition in
other areas of the fluidized bed would remain in the bed,
affecting the bed hydrodynamics and gas-solid contacting
adversely.

Preparation of Attrition Resistant Catalyst

Spray drying of the VP0 catalyst leads to particles with very
poor attrition resistance, entirely unsuitable for a fluidized bed
596

reactor even if the reactor modification described above are

employed. A commonly used approach to increase attrition
resistance of the catalyst is to add 20 to 50% by weight of
colloidal silica to the catalyst precursor before spray drying
(ref. 6). When such a spray dried catalyst is calcined, the
silica sinters and forms an attrition resistant particle.
However, the presence of silica usually causes high selectivity
losses for oxidation of n-butane to maleic anhydride (ref. 7). A
novel approach to produce attrition resistant catalyst that do not
cause selectivity losses was recently reported (ref. 10). In this
approach, the catalyst precursor is slurried with freshly prepared
polysilicic acid in ratios to produce a final solid composition
having only 5 to 10% silica by weight. The slurry is spray dried
into a chamber where the temperature is 250C. This small amount
of silica imparts attrition resistance, but does not lead to
detectable degradation of catalytic properties. Catalyst
particles prepared by this technique have a different
microstructure than that found for catalysts prepared with larger
amount of colloidal silica (Figure 5). The technique results in a
thin layer of silica at the periphery of the spray dried
particles. The layer is very durable, yet it is porous to
reactants and products.

50% Colloidal Silica

10% PSA
CatalystParticles

PorousMicrospheres
45-150 pm

Fig. 5. Attrition resistant catalyst: Two concepts, from reference

10.
All of the companies offering fluidized bed processes for
maleic anhydride production claim low catalyst losses in their
process. The approaches taken to reduce catalyst losses are not
disclosed.
 597

YIELDS IN FLUIDIZED 3ED

Reaction Pathway Considerations
The reader is referred to the chapter on fixed bed reactors
for a discussion on reaction kinetics for the catalytic oxidation
of C-4 hydrocarbons. In this section, impact of the reaction
network on the fluidixed bed reactor performance is discussed.
It is generally accepted (refs. 11-13) that the VP0 catalyst
oxidizes n-butane to maleic anhydride via redox mechanism of its
surface layers in a manner consistent with the Mars-van Krevelen
model (ref. 14). According to the model, the reaction is
comprised of two steps:

- Reaction between catalyst in oxidized form and n-butane in which

the oxidized catalyst becomes reduced;
- The reduced catalyst becomes oxidized again by oxygen from the
gas phase.

Under steady state conditions, the rate of the two steps must
be the same, but the rate is frequently controlled by just one of
the two steps. In fluidixed bed reactors, the goal is to operate
using stoichiometric ratio of oxygen to butane. However, the
catalyst activity and selectivity decreases as the catalyst begins
to reduce from its optimum oxidation state (ref. 13). A somewhat
higher than stoichiometric ratio of oxygen to n-butane is
generally used to insure against catalyst over reduction. On the
other hand, as shown by Trifiro and coworkers (ref. 121, an
increase in residual oxygen to n-butane ratio increases catalytic
degradation of maleic anhydride to CO,. Also, there may be a
vapor phase degradation of maleic anhydride in the presence of
excess oxygen in the free board of the fluidized bed.
The reaction network below, involving parallel and series
reaction steps, is the simplest of those proposed for oxidation of
n-butane on VP0 catalyst.
Kl
n-butane -> maleic anhydride

At low conversions, the parallel reaction is considered to

be the primary path to selectivity loss: but at high conversion and
high concentration of maleic anhydride, the series reaction
598

becomes more significant. At near total conversion, the yield of

maleic anhydride becomes very low. Fixed bed, plug flow reactors
are thus favored over fluidixed bed reactors owing to significant
backmixing effects in fluidized beds. Increasing the reaction
temperature increases the rate of the parallel reaction of CO,
formation more than the rate of maleic anhydride formation. Thus,
local hot-spots in the rector have a negative effect on the
selectivity to maleic anhydride. The ease of controlling hot-spots
is one of the advantages of fluidized bed reactors over fixed bed.
The above discussion on the reaction network suggests the following
steps to achieve high maleic anhydride yields in fluid bed
reactors.
Achieve good mixing of the reactants and the catalyst in the
grid zone of the fluidized bed reactors to avoid pockets lean in
butane, oxygen, or the catalyst.
Maintain good mixing in the main bed, as measured locally.
Minimize gas back-mixing and yet maintain the reactor
temperature under control.
Control catalyst inventory to avoid near total conversion and
associated maleic anhydride degradation.
Insure minimum excess oxygen in the product gas to minimize
maleic anhydride degradation.
Maintain narrow gas residence time distribution.

Hydrodynamics Considerations
(i) Grid Zone. Good mixing of the catalyst and the reactants
can be achieved in the grid zone by premixing reactant gases and
using the largest number of holes per unit area of the gas
distribution grid. This would entail a complex grid design to
insure constant ratio of the gaseous reactant in each hole and
uniform gas distribution over entire grid zone. Premixing of the
reactant gases would form an explosive mixture. Thus a
grid/sparger configuration is commonly used where air is
introduced through the grid zone and n-butane is introduced
directly in the bed through a sparger. As the gas solid mixing is
generally good in the grid zone and deteriorates further
downstream in the bed because of the growth of bubbles, design of
this area is very critical. Zones of low oxygen concentration or
low n-butane concentration should be minimized. Mitsubishi has a
patent (ref. 15) on approach to premix the reactant in the
 599

explosive range and avoid flame propagation by use of a fluidised

bed of inert particles ahead of the catalyst bed.

(ii) Main Bed. Most of the excess gas, above that required
for minimum fluidixation, passes through the fluidised bed in
fast-moving bubbles. Bubbles formed at the distributor plate
begin to coalesce rapidly as they rise through the bed. The
resulting larger diameter bubbles have a correspondingly higher
bubble rise velocity and give wider gas residence time
distribution. Bubbles not only coalesce and grow, they also
breakup presumably because large bubbles become hydrodynamically
unstable. Thus, there appears to be a maximum stable bubble size.
Gas distributors with more holes per unit area will produce
smaller bubbles at the distributor and will rise to greater height
in the bed before reaching the maximum stable size than for
distributor with fewer holes per unit area.
A rising bubble carries solids with it in its wake. During
its rise through the bed, solids are exchanged with the emulsion
phase solids. To compensate for the net upflow of solids, a
gradual downflow of the emulsion phase occurs resulting in
circulation and intense mixing of the solids in the bed. The
downward moving solids bring with them gases in the interstices
and pores of the solids and adsorbed on the solid surface, causing
a considerable gas backmixing. In summary, in a bubbling bed, an
increase in gas velocity increases the amount of gas going-up in
bubbles and thus increases solids and gas backmixing as well as
gives wider gas residence time distribution. A wide gas residence
time distribution implies that a considerable amount of gas leaves
the reactor in a very short time in rapidly rising large bubbles,
causing low conversion. The gases in the emulsion phase suffer
high conversion and selectivity losses. Bypassing of gases is
limited by the extent to which gas is exchanged between bubble and
emulsion phase. Several approaches have been suggested in the
literature to estimate the gas exchange (ref. 16-17). Treatment
of these is outside the scope of this article.

(iii) Aooroaches to Minimize Backmixinq. Several approaches

are used in fluidixed bed reactors to minimize the gas backmixing
and to obtain narrow gas residence time distribution.
Addition of fine particles in the fluidized bed decreases the
bubble diameter at any given gas velocity and the size of the
600

maximum stable bubble, as well as increase voidage of the emulsion

phase. This decreases the bubble rise velocity and thus decreases
backmixing and gives narrower gas residence time distribution. It
is a common practice in many commercial catalytic reactors to add
up to about 25 to 30% fines (20 to 45~) in the bed containing
coarser particles (45 to 150~). Adding more than 35 to 40% fines
begins to change the bed hydrodynamics adversely because the
particle mixture shifts from aerable "group A" powder to cohesive
"group C" powder (ref. 18).
Another approach often taken in commercial catalytic fluidixed
bed reactors besides adding fines is to operate the bed in
turbulent mode. As the gas flow rate through a fluidixed bed is
continuously increased, the violent bubbling action at first
increases. As discussed above, bubbles become larger while
coalescence and splitting events are more and more frequent.
Pressure fluctuations aiso grow in amplitude. However, beyond a
certain superficial gas velocity, the pressure fluctuations begin
to decrease, large bubbles begin to disappear, and the bed changes
from bubbling to turbulent regime. Viewed from one perspective,
voids or bubbles in the turbulent regime are splitting as rapidly
as they are coalescing so that the mean void size is small and the
discrete bubbles in the usual sense cannot be distinguished,
Viewed from another perspective the turbulent regime is
intermediate between the bubbling regime where voids or bubbles
are dispersed in a solid-rich emulsion, and a fast fluidization
regime where solid-rich particle clusters are dispersed in a
continuous phase consisting almost entirely of gas and where the
amplitude of pressure fluctuations levels off at a value much less
than that obtained in a violently bubbling bed. Its key
hydrodynamic features include decreasing pressure fluctuations,
narrowing of gas residence time distribution, decreasing void size
and associated decrease in solid and gas backmixing with
increasing gas flow rate. An increase in the fines level in the
bed decrease the velocity at which the transition from bubbling to
turbulent regime occurs.
Yet another approach sometimes recommended is to use staging.
Installation of horizontal perforated plates or baffles in the bed
can be used to limit the bubbles size and is considered
particularly useful for slow reactions and when an adequate
concentration of fines cannot be maintained. Staging, however,
adds considerable complication, and quantitative design criteria
 801

are not available. It is therefore not used frequently

specifically to control bubble size.

FLUIDIZED BED SCALE-UP

The size and mobility of the bubbles change according to the
changes in reactor size making it difficult to predict the reactor
performance on scale-up. Solids mixing induced by bubbles is
often overshadowed in large beds by mixing induced by "gulf
streaming" of the bed emulsion phase. Typically an upflow of the
bed is induced in the central region of the bed with downflow
along the walls. In very large beds, this effect is thought to
form mixing cells, much like thermal convection cells in liquids
with central upflowing regions. Gulf streaming can be induced by
a non-uniform distribution of the gas in the grid zone, Once an
unequal gas or bubble flow has been established creating lateral
density gradients, bulk flow of the emulsion phase of the type
characterized as "gulf streaming" is set up. It is conceivable
that the unequal bubble flow is established by other reasons as
wall. "Gulf streaming" creates wide gas residence time
distribution and associated loss in conversion as well as
significant solid and gas backmixing and corresponding loss in
selectivity.
Most literature references on fluidized bed scale-up mention
poorer gas solid contacting, wider gas residence time distribution
and increased backmixing with increased diameter of the reactor,
Figure 6 shows a plot by J. J. Van Deemter (ref. 191 of reactor
diameter versus gas and solid mixing coefficient data for a large
number of commercial and pilot-scale catalytic cracker regenerator
and other reactors using Geldart's (ref. 18) "Group A" particles,
It clearly shows an increase in gas and solid backmixing with
increased reactor diameter. Scale-up of "group A* fluid beds far
catalysis is notoriously difficult when high conversion of
reactants are desired (20). Figure 7 taken from reference 20
illustrates the usual progression from bench scale to commercial
design. The lower curve in the figure suggests how this
progression has sometimes led to a poor reactor efficiency (ratio
of apparent kinetic constant to intrinsic constant),
A strategy sometimes recommended to minimize these effects of
scale-up is to use the same hydraulic diameter by use of vertical
internals or baffles in the larger diameter reactor. This criteria
should be used with caution. A reliable, general quantitative
602

treatment of baffle effects is not available. In commercial

reactors, baffles have apparently been rarely used to improve
gas/solid contacting. When internals are installed for other
reasons such as heat transfer tubes, however, they undoubtedly do
improve contacting.

0
0
0

8
8
8
0
J ?
8 0

80
0

8
P
0
0
Radial
Mixing

0 i

I Broad
Particle
Size Range
Narrow
I

MLO n
MBO +
MS 0
I 1 I 1 I I I,
I
0.1 1 10
Vessel Diameter, D (m)

Fig. 6. Mixing coefficient in fluidized beds from reference 19.

ML-Longitudinal gas mixing coefficient. MB-Gas backmixing
coefficient. MS-Solids mixing coefficient.

On scale-up, the grid zone design has to contend with added

structural constraints which may make it difficult to insure that
adequate mixing of the reactants is achieved. Changes in the grid
 603

zone have oftenbeen needed after a commercial plant is built to

achieve adequate reactor performance.

0.8L No Scale-Up Loss

Fig. 7. Effect of Scale-

up on reactor efficiency
from reference 20.

Bench Pilot

Reactor diameter (m)

There are several fluidixed bed models proposed for scaling-up

fluidized bed reactors to a commercial scale. Werther has applied
(ref. 21) his two-phase model to synthesis of maleic anhydride from
C-4 feedstocks and shown that using the rate constants derived from
the data on a 45 cm diameter reactor, the calculated yield and
conversion for a 15 cm bed are close to the expected value but not
for a 4 cm reactor. Dutta et al (ref. 221, examined scale-up of a
fluid bed reactor for catalytic air oxidation of n-butane to maleic
anhydride based on the projection of three different models. Some
of their conclusions are summarized below.
For each of the examined models, at least one parameter needs
to be adjusted or tuned by the observed yield data, even in a
bench-scale unit.
Design and scale-up of a commercial catalytic reactor from
laboratory-scaleexperiments are strongly influenced by the
uncertainty regarding the limiting bubble size in beds which
contain relatively small and light catalyst particles.
Many of the commercial catalytic reactors are high-velocity,
bubbling fluidixed beds (UC > 100 Umf) with baffles/internals.
Direct applicability of available reactor models to the design
604

and scale-up of such systems is questionable. Nevertheless,

the models do serve as sound logical tools in projecting the
expected interactions between a large number of design and
operating variables.
- The identified shortcomings of the available models reaffirm
the necessity to incorporate in-house expertise and experience
obtained at full or semi-commercial scale, in order to achieve
a successful scale-up.
The last conclusion is the most significant. All the
companies offering fluidized bed processes indicate they have
demonstrated their process on large fluid bed reactors, but not
everyone has disclosed the size of their process demonstration
unit. Some scale-up uncertainties remain even after experience is
gained on 50 to 75 cm diameter pilot plants.

CIRCULATING SOLIDS REACTORS

Recently Du Pont disclosed (ref. 10) a new reactor concept for
selective oxidation of n-butane to maleic anhydride which reduces
or eliminates most of the negatives of fluidized bed reactor while
maintaining its advantages. Additional performance advantages are
derived by separating two steps of the redox process.
Figure 8 shows a schematic of the circulating solids riser
reactor used for n-butane oxidation. VP0 catalyst is continuously
circulated around the loop. n-Butane is oxidized by the catalyst
in the lean-phase, riser section which has plug flow
characteristics. The catalyst is reoxidized in the dense phase
fluidized bed regenerator. Very high n-butane concentration in
the feed gas is possible, and high selectivities are achieved. It
is believed that an important selectivity loss pathway involves
highly active surface species such as Os- and O-. Presumably some
of these nonselective pathways are eliminated in this reactor by
carrying out the oxidation reaction in the absence of gas phase
oxygen. Advantages claimed for the process are summarized below.
Together with the attrition resistant catalyst described earlier,
the reactor is claimed to give higher than 70% maleic anhydride
yields, lower investment, and superior economics relative to
alternatives. Recently, Monsanto and Du Pont announced (ref. 23)
plans to jointly develop this process.
 605

Advantaqes of Circulatino Solids Riser Reactor

l Separate catalyst oxidation and reduction zones
- Independent control of two zones with catalyst in transient
cyclic state
- High selectivity. Low oxygen concentration in reactor and
control of nonselective oxygen species
- Concentrated product streams
- High hydrocarbon concentration in feed
- Product gas separate from regenerator off gas
- High throughput
- Low catalyst inventory
- No explosibility problem

l Riser reactor zone

- High selectivity. Plug flow. No hot spots.
- No free board burning
- High turndown ratio
- Ease of scale-up

l Fluidized bed regenerator zone

- High heat transfer coefficient
- Good temperature control

CONCLUSIONS
Fluidixed bed processes for production of maleic anhydride
offer many advantages over fixed bed systems. The main advantages
are: (a) high heat transfer coefficient; (bf ease of temperature
control and elimination of hot spots: and (c) high C-4
concentrations in the feed. These advantages result in a lower
investment process, particularly for a large scale plant. Several
companies have developed fluidized bed processes. All claim to be
able to produce a lower cost maleic anhydride. Catalyst losses
due to attrition have been reduced but generally at some penalty
of lower selectivity. Yield losses associated with backmixing are
minimized but not eliminated. The negative impact of lower yields
would depend on n-butane price and the value of byproduct steam at
a given site. Some scale-up uncertainties remain except when
costly and time consuming experience with very large demonstration
plants is gained. Du Pont's attrition resistant catalyst and the
circulating solids riser reactor technologies show a potential for
606

a major improvement in overall economics relative to the

incremental improvementsof the fluidixed bed processes.

Off Gas Maleic Anhydride

Air

Fig. 8. Du Pont's
circulating solids
riser reactor from
reference 10.
Inert
Gas

Stand

t_ Butane

S. Ushio, Chemical Engineering, September 20 (1971) 107.

; S. C. Arnold, G. D. Suciu, L. Verde and A. Neri, Hydrocarbon
Processing September (1985) 123.
3 G. S. Schaffel, S. S. Chen, and J. J. Graham, Proceedings
Second World Congress of Chemical Engineering, Vol. 2 (1981) 4.
4 Sohio/UCB S.A., Maleic Anhydride, Hydrocarbon Processing,
November (1983) 109.
5 K. Brooks, Sohio Explores Butane Uses, Chemical Week,
December 17 (1986) 9.
6 H. Kuroe, Japan Patent 60,64,622, (1985); Chem. Abstr. 103
(1985) 71817.
7 P. R. Blum and M. L. Nicholas, US Patent 4,317,778 (1982).
8 C. R. Bemrose and J. Bridgwater, Powder Technology, 49 (1987)
97.
9 M. N. Schwendeman, R. McBride, D. Reuter, and M. Isaacs,
Chemical Processing, July (1983) 100.
10 R. M. Contractor, H. E. Bergna, H. S. Horowitz,
C. M. Blackstone, 8. Malone, C. C. Torardi, B. Griffiths, U.
Chowdhry, and A. W. Sleight, Catalysis Today 1 (1987) 49.
11 M. Pepera, J. L. Callahan, M. J. Desmond, E. C. Millberger, P.
R. Blum, and N. J. Bremer, J. Am. Chem. Sot., 107 (1985) 4883.
12 F. Cavani, G. Centi, A. Riva, and F. Trifiro, Catalysis Today 1
(1987) 17.
13 B. K. Hodnett and B. Delmon, Ind. Eng. Chem. Fundam. 23 (1984)
465.
 607

14 P. Mars and D. W. Van Krevelen, Chem. Eng. Sci. Suppl. 3 (1954)

41.
15 K. Matsuura and H. Imai, Japan Patent 74 11,374 (1974); Chem.
Abstr; 82 (1975) 17453.
16 J. F. Davidson and D. Harrison, "Fluidized Particles,"
Cambridge University Press (1963).
17 D. Kunii and 0. Levenspiel, "FluidizationEngineering," Wiley,
New York (1969).
18 D. Geldart, Powder Technology, 7 (1973) 285.
19 J. J. Van Deemter, "Fluidization"edited by J. R. Grace and J.
M. Matsen, Plenum Press (1980) 69.
20 A. M. Squires, M. Kwauk, and A. A. Avidan, Science 230 (1985)
1329.
21 W. Bauer and J. Werther, Proceedings of 7th International
Symposium on Chemical Reaction Engineering, October (1982).
22 S. Dutta, S. C. Arnold, G. D. Suciu, and L. Verde, I. Chem. E.
Symposium Series No. 87, 517-526.
23. Chemical Marketing Reporter, January 26 (1987).