This article was downloaded by: [171.67.34.

205]
On: 14 March 2013, At: 22:03
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Catalysis Reviews: Science and

Engineering
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/lctr20

Reverse-Flow Operation in Fixed Bed

Catalytic Reactors
a a
YURII SH. MATROS & GRIGORI A. BUNIMOVICH
a
Matros Technologies 14963 Green Circle Drive Chesterfield,
Missouri, 63017
Version of record first published: 16 Aug 2006.

To cite this article: YURII SH. MATROS & GRIGORI A. BUNIMOVICH (1996): Reverse-Flow Operation in
Fixed Bed Catalytic Reactors, Catalysis Reviews: Science and Engineering, 38:1, 1-68

To link to this article: http://dx.doi.org/10.1080/01614949608006453

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions

This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation
that the contents will be complete or accurate or up to date. The accuracy of any
instructions, formulae, and drug doses should be independently verified with primary
sources. The publisher shall not be liable for any loss, actions, claims, proceedings,
demand, or costs or damages whatsoever or howsoever caused arising directly or
indirectly in connection with or arising out of the use of this material.
 CATAL. REV.-SCI. ENG., 38(1), 1-68 (1996)

Reverse-Flow Operation in
Fixed Bed Catalytic
Reactors
Downloaded by [171.67.34.205] at 22:03 14 March 2013

YURII SH. MATROS" and GRIGORI A. BUNIMOVICH

Matros Technologies
14963 Green Circle Drive
Chesterfield, Missouri 63017

I. INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

11. BACKGROUND OF THE TECHNOLOGY ..................... 4

111. ANALYSIS OF MATHEMATICAL MODELS . . . . . . . . . . . . . . . . . . . 10
A. One-Dimensional Models ................................ 10
B. Simplified Approaches at Low and High
Reversal Frequencies .................................... 14
C. Exothermic Reaction Performance in One Catalyst Bed . . . . . . . 19
D. Use of Inert Ceramic Packing and Heat Removal . . . . . . . . . . . . 26
E. Complex Exothermic and Endothermic Processes . . . . . . . . . . . . 28
F. Temperature Nonuniformity .............................. 32
G. Influence of Catalyst Dynamics ........................... 33
IV. EXPERIMENTS ............................................ 37
V. INDUSTRIAL APPLICATIONS, RESULTS, AND
FUTURE PROSPECTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
A. Catalytic Incineration of VOC ............................ 48
B. Sulfur Dioxide Oxidation ................................ 50
C. Selective Reduction of NO, .............................. 52
D. Sulfur Production by Catalytic Claus Method . . . . . . . . . . . . . . . 52

*To whom correspondence should be sent at the above address. Fax: (314) 391-3658.
1

Copyright 0 1996 by Marcel Dekker, Inc.

 2 MATROS AND BUNIMOVICH

E. Heat Recovery of Lean Gases and Fuels . . . . . . . . . . . . . . . . . . . 53

F. Natural Gas Reforming or Partial Oxidation . . . . . . . . . . . . . . . . . 54
G. Synthesis of Methanol and Ammonia ...................... 55
H. Catalytic Dehydrogenation ............................... 56
VI. CONCLUDING REMARKS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
NOTATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
ACKNOWLEDGMENT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
REFERENCES ............................................. 60
Key Words:Catalyst dynamics; Fixed bed of catalyst; Mathematical
modeling; Reverse-flow operation; Unsteady-state catalytic process.
Downloaded by [171.67.34.205] at 22:03 14 March 2013

I. INTRODUCTION

Transition from the usual steady-state mode of continuous processes

toward forced unsteady-state conditions (FUSC) has been discussed in chem-
ical engineering literature since the beginning of the 1960s [1-121. FUSC
can be created by periodic variations of temperature, composition, or other
parameters to the inlet of a chemical unit.
Mathematically, a set of possible unsteady or cyclic states of a dynamic
system includes within it a narrower set of steady states. Therefore, the max-
imum value of an object function (Ju)obtained at optimal unsteady-state
conditions is not lower than that for optimal steady state (J?),or:
Jll 2 Js
This inequality is trivial. The only difficulty a chemical engineer faces is the
selection of unsteady-state operating conditions that can provide a substantial
advantage when compared to optimal steady-state operation.
When performing forced unsteady processes in a heterogeneous catalytic
reactor, one can outline two major factors responsible for positive effects [13]:
1. Dynamic properties of the catalyst: Unsteady conditions in the gas
phase can give rise to such changes in state, composition, and struc-
ture of the catalyst which induce an selectivity and/or activity in-
crease compared to steady-state operation.
2. Dynamic characteristics of a whole reactor system: Forced variation
of inlet parameters is aimed at formation of optimum temperature
and composition distributions in the reactor which cannot be ob-
tained in any steady-state regime.
Initial investigation of dehydrogenation of ethanol to ethylene and di-
ethyl ether conducted by Denis and Kabel [14] and Wandrey and Renken
[15] opened a series of experimental works which demonstrated improved
 REVERSE-FLOW OPERATION 3

catalyst performance for nearly all large-scale heterogeneous catalytic pro-

cesses (see Refs. 13, 16-20). The list of such processes includes sulfur di-
oxide oxidation over vanadium catalyst, ammonia synthesis over promoted
iron catalyst, hydrogen sulfide oxidation by SO, on bauxite catalysts, Fisher-
Tropch synthesis over ruthenium and cobalt catalysts, CO oxidation on dif-
ferent metallic and oxide catalysts, etc. Typically, the experimental system is
an isothermal fixed bed reactor exposed to periodic oscillations of the inlet
gas mixture composition.
Accumulated experience to this point allows an affirmative answer to
the question: ‘‘Periodical operation in chemical reactor: are global improve-
ments attainable?” [19]. However, more efforts are needed to elucidate what
type of unsteady-state conditions will provide maximum effect and how large
its value will be.
Downloaded by [171.67.34.205] at 22:03 14 March 2013

FUSC can also be created by circulation of catalyst between two or

more catalytic reactors or in one reactor with a fluidized bed of catalyst. In
the mid-SOs, Du Pont developed the so-called Riser process for maleic an-
hydride production from n-butane [21-241. This partial oxidation reaction is
carried out in a reactor with rising catalyst flow and in a second fluidized bed
reactor where the catalyst is oxidized by air. Making use of the dynamic
properties of vanadium phosphorous oxide catalyst makes it possible to
achieve maleic anhydride selectivity as high as 80%.
Reverse-flow operation (RFO) in a fixed bed catalytic reactor, which
was described by one of the authors of this review as early as 1977 [25], is
an example of FUSC which creates an effect due to appropriate use of whole-
reactor dynamics. The process improvement under unsteady-state conditions
can still be achieved even in the case when the catalyst quickly adjusts to
the changes of the temperature and reaction mixture composition, and a
steady-state approach to the reaction kinetics is valid. The basic mode of
operation is very simple and does not include forced oscillations of inlet gas
temperature or composition but simply comprises a periodic reversal of cold
gas flow to a preheated fixed catalyst bed (Fig. 1).
During the last 10 years a number of papers have been published which
are devoted to studies and development of reverse flow reactors. More than
20 academic research teams from different countries have contributed. Some
of the papers only contain theoretical analysis on the basis of simple first-
order reaction kinetics. Others report attempts to simulate the real catalytic
systems and to study them experimentally. The idea of RFO appeared to be
attractive for practical use and some chemical companies tried to practice it
on a pilot or commercial scale. The scope of possible applications includes
both exothermic processes such as sulfur dioxide oxidation for sulfuric acid
production or methanol synthesis and endothermic processes like hydrocar-
bons dehydrogenation. A number of publications relate to purification of
waste industrial gases from volatile organic compounds (VOC), as well as
nitrogen oxides and SOz. Some papers discuss the results of industrial oper-
ation. This indicates current interest from both academic researchers and ap-
 4 MATROS AND BUNIMOVICH

plication engineers. On the other hand, the lack of a generalized view on the
issue seems to be perceptible. The surveys [13, 20, 261 published earlier
related mainly to Russian works and now are out of date. In this paper we
have set for ourselves the objective of making up for this deficiency and
presenting critical review of known works beginning from the development
of basic ideas of RFO performance for various applications. This review
examines the results of theory, mathematical modeling, experimental studies,
and practical experience of industrial applications.

11. BACKGROUND OF THE TECHNOLOGY

Most of the studies reviewed are devoted to reverse-flow operation for

Downloaded by [171.67.34.205] at 22:03 14 March 2013

one-route exothermic reaction. The experimental temperature profiles (Fig. 2)

for SO, oxidation over vanadium oxide catalyst [27, 13, 201 clarify this typ-
ical RFO application. The experimental rector had an inner diameter of 0.17
m and catalyst bed height of 0.44 m. The flow diagram differed from that
shown in Fig. 1 only by the inclusion of an inlet gas preheater. Five ther-
mocouples were installed along the reactor length and simple devices for
chemical analysis of the inlet and outlet SO, concentration were included.
The air/SO, mixture was fed into the reactor with a total flow rate of 3800
L/h and an inlet SO, concentration equal to 7.5% vol.
For start-up, the inlet gas at a temperature of 400°C was fed into the
reactor for several hours and a steady-state regime [Fig. 2(a), line 11 was
obtained. Then, the gas flow direction was changed and the temperature of
the inlet gas was reduced to 200°C. This temperature corresponds to a neg-
ligible rate of SO, oxidation reaction over vanadium catalyst. Temperature
profiles 2 and 3 [Fig. 2(a)] at 60 and 120 min after flow reversal are shown

i
V V
FIG. 1. Principal diagram of reverse-flow reactor: V and V ’ , switching
valves; C, catalyst bed.
 REVERSE-FLOW OPERATION 5

100
0.00 0.10 0.20
-
-
0.30
..
..
.111.
1
0.40
. - - I-00.00
0j
0.10 0.20 0.30 0.40
Catalyst b e d Zength, m Cafalyst b e d l e n g t h . m

FIG. 2. Evolution of temperature profiles in experimental reactor for SO2

Downloaded by [171.67.34.205] at 22:03 14 March 2013

oxidation (adopted from Boreskov et al. [27]): (a) first semicycle; (b) second semi-
cycle; 1, steady state at downward flow direction; 2 and 3, 60 and 120 min after
feeding of cool gas; 4 and 5, 180 and 240 min after first flow reversal.

by lines 2 and 3. Slow cooling of the fraction of the catalyst bed bordering
the gas inlet was observed while the temperature in the center of the bed
became higher than in steady-state conditions. In about 2 h after cool gas
began feeding into the reactor, a second reversal of gas flow was done. The
process during this next 2-h period [Fig. 2(b)] is characterized by a slow
cooling of the left (inlet) part of the bed and heating of the right (outlet) part
that had been previously cooled down prior to the reversal. Temperature pro-
files which were obtained after several cycles were close to those shown in
Fig. 2(b).
Two principal features of the observed phenomena are worth noting:
1. The difference between maximum and inlet gas temperature
(T,,, - Ti,) during the cycling is substantially higher than the
adiabatic temperature rise, AT,,, corresponding to the maximum
possible heat release in a steady-state process. (For these experi-
ments: T,,, - Ti, - AT,, SOOC). -
2. After the temperature plateau, where the process conditions are
close to equilibrium, the temperature drops in the outlet part of the
bed.
The first observation means that catalyst serves not only to accelerate the
chemical reaction but also as a heat exchange and heat accumulation media
(heat sink), which is able to collect and transfer the stored energy of the
reaction to the cooler inlet gas. This makes it possible to provide continuous
autothermal operations without external gas preheating in advance of the cat-
alyst bed.
The second observation is the result of a large catalyst bed heat capacity
 6 MATROS AND BUNIMOVICH

compared with that of the gas mixture. Steady-state exothermic processes in

an adiabatic reactor can provide only a temperature rise along with catalyst
bed length. [Some temperature decrease in steady-state experiments-Fig.
2(a), line 1- is associated with insufficient compensation for heat losses in
the laboratory reactor.] Rising temperature is not optimal for a reversible
reaction like SO, oxidation or ammonia synthesis because conversion is lim-
ited by the equilibrium condition at maximum temperature. To obtain a higher
conversion, several catalytic stages with intermediate gas cooling are rou-
tinely employed. Reverse-flow operation permits higher conversion than for
steady-state operation in a single catalyst bed. For these experiments, cycle-
averaged conversion in unsteady-state conditions was 5% higher than for
steady state [13, 271.
These features determine some of the advantages of the reverse process
Downloaded by [171.67.34.205] at 22:03 14 March 2013

in comparison with traditional steady-state operation. RFO is attractive for

treatment of diluted gases which have an adiabatic temperature rise of about
15"-2OoC. RFO has been already commercialized for processes involving the
catalytic incineration of noxious organic compounds in industrial waste gases
(see, e.g., Refs. 20, 28-30), SO, oxidation after nonferrous metal smelters
[20, 31-33], and selective NO, reduction [34, 351. For reversible reactions
such as SO, oxidation, and methanol and ammonia syntheses, the RFO allows
a simplified reactor design, eliminates all shell and tube heat exchangers, and
decreases pressure drop through the system [ 13, 201.
As early as 1942 the Russian physicist D. A. Frank-Kamenetski exper-
imentally discovered the possibility of continuous "reaction zone migration"
in a packed bed of Cu/Al,O, catalyst with periodic flow reversal of an iso-
propanol/air mixture. Only a short description of this experiment was pub-
lished in his well-known book [36]. However, a negative conclusion about
possible practical use of this phenomenon of moving reaction zone with re-
verse-flow operation was made at that time [36].
The development of RFO applied to exothermic reactions began from
the works of a research group from the Institute of Catalysis in Novosibirsk,
Russia, in the mid-70s [25, 371. These authors were the first to pay attention
to possible wide application of RFO in industrial catalysis. They discussed
theoretical and applied problems related to RFO including analysis of various
mathematical models and experimental examples (see Table 1 in the Exper-
iments section). In Russian certificates [38-451 and Western patents [46],
Boreskov and Matros, with coworkers, described various reactor configura-
tions and methods for process improvement.
Reverse-flow operation for endothermic catalytic reaction has been dis-
cussed by Heggs with coworkers from Leeds University, GB [47, 481. Their
reactor system (Fig. 3) included two catalyst beds and two four-way switching
valves. The process of ethylbenzene dehydrogenation over iron oxide catalyst
was chosen. Feeding of the reactants, ethylbenzene and steam, was periodi-
cally changed to high temperature steam which served to reheat the bed that
had cooled during the endothermic reaction. The catalyst is used as a heat
 REVERSE-FLOW OPERATION 7

Products
0

f
Downloaded by [171.67.34.205] at 22:03 14 March 2013

Et hylbenzene
and steam

FIG. 3. Flow diagram of a reactor system for endothermic process of ethyl-

benzene dehydrogenation (adopted from Heggs [47]): 1, catalyst bed; 2, four-way
switching valve.

regenerator as in the process discussed above, but, in contrast, the heat of the
product gas is not recovered for subsequent inlet gas preheating. This type
of operation we may designate as an “open-loop” RFO, contrary to the
‘ ‘closed-loop” operation to be discussed in relation with exothermic
processes.
The reactor scheme shown in Fig. 3 was discussed previously by Gav-
alas [49, 501, but for highly exothermic processes of partial oxidation. The
author tried to estimate on the basis of a simplified cell model the potential
of periodic flow reversal for catalyst bed cooling.
Recently two chemical companies, ABB Lummus Crest and UCI [Sl],
showed the efficiency of open-loop reverse-flow operation for catalytic par-
affin dehydrogenation. In accordance with the traditional Houdry technology
this endothermic process is performed under periodic heating of catalyst bed
by air where cocurrent flow of air and hydrocarbon is used. Change to coun-
tercurrent flows of air and hydrocarbons allows for an increasing temperature
profile along the bed length, which results in more favorable equilibrium
conditions, higher conversion, and better use of heat [511.
RFO can be used for combined performance of endothermic and exo-
thermic reactions. This process has been tested by Blanks et al. [52] in pilot
plants for hydrogen production by partial oxidation of natural gas. Highly
endothermic steam reforming of natural gas was combined with exothermic
methane combustion. Reverse-flow operation made it possible to achieve high
temperature in the catalyst bed at low average difference between outlet and
inlet gas temperature; thus methane consumption during exothermic combus-
 8 MATROS AND BUNIMOVICH

tion was very low. A similar process for air or oxygen-water-steam natural
gas autothermal reforming has been previously described in a Russian certif-
icate [41].
Figure 4 shows some possible flow diagrams based on the closed-loop
reverse-flow concept. Often, especially for the purpose of VOC oxidation,
part of the catalyst at the boundaries of the bed is replaced by the inert
packing material [Fig. 4(a)]. In this case the temperature in the reactor can
be easily elevated, which offers a substantial reduction in the quantity of
catalyst needed for an irreversible reaction. Such a reactor configuration can
also be applied to partial oxidation of natural gas [52] to exclude an unfa-
vorable catalytic reaction of synthesis gas methanation occurring at low
temperatures.
In a case of very low inlet concentration, preheating of gases between
Downloaded by [171.67.34.205] at 22:03 14 March 2013

two packed beds is used. The reactor can be configured by installing an

electric heater, fuel burner, or hot gas injection into the reactor center [Figs.
4(d) and 4(f)].
Figures 4(b), 4(c), and 4(e) show possible process types with interme-
diate heat removal. Energy recovery can be accomplished by passing hot
reacted gas through one or more intermediate coolers [Figs. 4(b) and 4(e)]

FIG. 4. Various arrangements of reverse-flow reactor: 1, catalyst bed; 2, re-

fractory packing bed; 3, switching valve; 4, heat exchanger for reaction energy util-
ization; 5 , heater; 6, reactant or energy intermediate feeding; 7, catalyst bed having
unidirectional flow; 8, heat exchanger for inlet gas preheating; 9, rotary plate-type
catalyst bed.
 REVERSE-FLOW OPERATION 9

[40, 201 or by withdrawal of a part of the gas from reactor center [Fig. 4(c)].
The last type of operation has been examined [29, 53-54, 1731.
Flow diagrams shown in Figs. 4(f) and 4(g) were recommended for the
processes where two or more gas components should be mixed. This has
been discussed [34, 551 as an example of selective catalytic reduction of NO,
by ammonia.
In Fig, 4(h) the scheme of a combined reverse-flow reactor and shell-
and-tube heat exchanger is presented. This type of operation is applicable for
SO, oxidation in sulfuric acid production [20, 32, 331. Sulfur dioxide is ox-
idized initially in a reverse-flow reactor [Fig. 4(h), 13 and further in an ad-
ditional reactor (7). The direction of flow in this second reactor is not
changed. The gas is slightly preheated in the heat exchanger (8) before oxi-
dation. Such a process arrangement allows for creating favorable thermody-
Downloaded by [171.67.34.205] at 22:03 14 March 2013

namic conditions in the reactor (7). This provides a high degree of conversion
at high SO, concentration in the inlet gas and, simultaneously, enables op-
eration without fuel consumption at low SO, concentration.
Rotation of the catalyst bed can potentially be applied instead of periodic
flow reversal. Figure 4(i) shows this scheme with gas moving along the axis
of a rotating cylinder containing plate-type catalyst. This process arrangement
and the scheme with radial gas flow which is perpendicular to the cylinder
axis have been cited in Ref. 53 as possible methods for VOC control. Use
of rotating catalytic reactor-regenerators in open-loop processes for CO and
H, oxidation and NO, reduction has been discussed in the literature [56, 571.
“Unusual” types of RFO have been considered by Agar and Ruppel
[58] and Matros [20]. Both references proposed utilizing the adsorption ca-
pacity of a catalyst bed instead of its heat capacity.
In Ref. 20 a hypothetical example of a selective process with two par-
allel reaction routes is described. If the formation of a useful product proceeds
through the reaction of one of the components with an intermediate which
has high adsorption capacity onto the catalyst, the selectivity can be drasti-
cally increased for the RFO. The profile of intermediate distribution has a
bell-shaped form like the temperature profile in Fig. 2. This is more favorable
for the selective reaction than the steady-state monotonous profile.
The authors [58, 591 proposed carrying out selective reduction of NO,
by ammonia on the basis of the flow diagram shown in Fig. 1. Cycling also
includes periodic changing of inlet gas composition as well as periodic feed
gas flow reversal. In the first phase, one of the reactants, NH,,which is highly
adsorbed by the catalyst, is injected in excess of reaction stoichiometry. This
NH, accumulates on the inlet boundary parts of the bed and the wavefront
of NH, adsorption gradually moves toward the catalyst bed outlet. In the
second phase, ammonia is stopped; however, the second component (nitrogen
oxide) continues to react with the excess of ammonia previously adsorbed.
Then, the valves are switched and the first and second phases are repeated
in a different gas flow direction. As a result, an adsorption zone with high
surface ammonia concentration is formed and moves along the catalyst bed
 10 MATROS AND BUNIMOVICH

back and forth like the high temperature zone shown in Fig. 2. Due to am-
monia stored on the catalyst, the process is relatively insensitive to fluctua-
tions of gas flow and NO, concentration [58, 591. There are other potential
benefits specific for NO, reduction, such as low ammonia slip emission and
low requirement for uniform ammonia distribution over the reactor cross
section.
To finish this short discussion, it is necessary to mention RFO applica-
tions in noncatalytic chemical processes. One is the early technology of VOC
incineration in regenerative thermal oxidizers [60]. This process uses two
packed beds of refractory ceramic material as a heat sink and a hollow com-
bustion (or retention) chamber between them. Low-temperature polluted air
is heated in one packed bed, exposed to VOC destruction or alteration in the
hollow combustion chamber where natural gas is burned, then cooled in the
Downloaded by [171.67.34.205] at 22:03 14 March 2013

second ceramic bed before exiting the incinerator. The flow direction through
this system is periodically reversed. A similar process [61] was tested as early
as 1948 for oxidation of atmospheric nitrogen to nitric oxides. There are
American and Japanese patents [62-651 where several methods have been
proposed for improving the operation in regenerative thermal oxidizers. In
particular, use of a catalyst bed instead of a combustion chamber has been
claimed [62-651.

111. ANALYSIS OF MATHEMATICAL MODELS

A. One-Dimensional Models

Deterministic continuum models are commonly used for investigations

of steady-state and dynamics behavior of fixed bed reactors, as well as for
reactor design. These models are constructed from experimental studies of
flow patterns, heat and mass transfer in packed beds, and the previous ex-
perience in reactor modeling. Systems of models of different degrees of com-
plexity have been discussed in a number of reviews, most recently in Refs.
20, and 66-68.
The reverse-flow reactor is a new subject of interest. A number of works
are related to mathematical modeling and simulation of RFO aimed at de-
scribing the principal features of the phenomena occurring in a reverse-flow
reactor rather than designing an actual catalytic process. Apparently, several
simple models are capable for qualitative description of the moving temper-
ature and concentration profiles observed during the reverse-flow operation
[69, 701.
These models should include terms of heat and mass transfer by flow
convection and reaction kinetics. These factors have been considered in the
simplest plug-flow model of steady-state process in an adiabatic fixed bed
[81]. Flow reversal leads to slow periodic movement of temperature fields.
Therefore, a transient term related to the temperature of the fixed bed should
 REVERSE-FLOW OPERATION 11

be also included. Volumetric heat capacity of solid catalyst is commonly three

orders of magnitude higher than that for gas. Taking this into account, heat
accumulation in the gas phase can be excluded from the simple model. This
factor influences only the reactor behavior at the very beginning of each
reversal period.
The last factor which should be taken into account, and which results
in a diversity of possible simple models, is heat dissipation [69, 701. Formally
this dissipation can be defined by several means such as axial heat conduc-
tivity, interphase heat transfer, or heat transfer inside a catalyst pellet. The
role of those factors consists of averting unlimited catalyst temperature rise,
which results from the large difference between heat and mass capacities [69,
701.
If the heat dissipation is described by the term of axial heat conductivity,
Downloaded by [171.67.34.205] at 22:03 14 March 2013

a pseudohomogeneous model can be formulated. For a single reaction without

changing the number of moles of reactants, it can be written as:
l3T a2T aT
(E,P,C, + E,PsCJ -=
at
kf-
at -
p;c,u -
dl
+ (-AH)W(x,T)
dX ax
E,P, --$= -up," - + W(x,T)
ai
In the reverse-flow reactor the linear velocity U changes in time. For a
constant time between switches, tCi2,a square-wave mode of the linear ve-
locity variation is used:

where u > 0 is the absolute value of linear velocity, n = 1, 2 . . . is the number

of switches, and tc is the cycle duration. Boundary conditions can be given
by :

at 1=0 (4)

1 -k K(t)
x = X
2
atl=L
 12 MATROS AND BUNIMOVICH

be cyclically repeated after a number of switches (at n-

Supposing that the moving temperature and concentrations profiles will
a),the following
formula for cycle averaged heat balance (at c, = const) can be derived:
Tinp-I- ATadk = To,, (6)
where:
Downloaded by [171.67.34.205] at 22:03 14 March 2013

X and Toutare the cycle-averaged conversion and temperature,

respectively

AT,, is the adiabatic temperature rise

Equation (6) is an analog of the heat balance for steady-state adiabatic
reactors.
The system of Eqs. (1)-(2) can be reduced assuming quasi-steady-state
mass balance. As a result, the left side of Eq. (2) is equal to zero:
ax
0 = Up: 7 - W(X,T) (7)
dl
A pseudohomogeneous model, Eqs. (1)-(5), was used for studies of
RFO for exothermic [20, 29, 711 and endothermic [72] processes.
Another simple model is a two-phase plug-flow model that takes into
account only a limitation of interphase heat transfer as a heat dispersion
mechanism:

This model has also been used in studies of RFOs for exothermic [73-771
and endothermic [47] processes.
In the case of a real reactor, the factors considered above may work in
concert, and a special discussion is needed about their relative influences.
Also, mass transfer limitations due to inter- and intraparticle diffusion be-
 REVERSE-FLOW OPERATION 13

come important when it is necessary to derive quantitative information about

conversion or other process parameters.
If temperature gradients inside catalyst pellets are not high, a more gen-
eral two-phase model accounting for longitudinal heat dispersion in a “solid”
phase and mass transfer limitations inside and outside the catalyst particle is
applicable [20, 66, 79, 801 for reverse-flow reactor modeling. The model
description is given as follows:
Downloaded by [171.67.34.205] at 22:03 14 March 2013

E
axs = -up; -
- 8% + p,)a,p,(x, - x,)
at ai
The boundary conditions at

can be given as:

a T,
x = 0, T = Tinp,and A,,f - = 0 at 1 = 0
ai
a TS
-= 0 at I = L
ai

Effectiveness factor can be determined from experimental data or us-

ing a routine procedure of integrating the ordinary differential equations de-
scribing diffusion and reaction inside the isothermal catalyst particle (see,
e.g., Ref. 81).
The nature of axial effective heat conductivity of the catalyst bed (A,,,f)
has been discussed elsewhere [66, 861. In other papers [117, 137, 1491, axial
heat conductivity in the gas phase is considered instead of heat conductivity
in the solid phase. The experimental correlations for heat and mass transfer
parameters (ao7 Po, A,,,,) can be found elsewhere [82-861.
Assuming quasi-steady state for gas concentrations and temperature [20,
30, 31, 87, 88, 1321 is an appropriate way to simplify the model (11)-(16).
In special cases considered below, steady state can be applied to all variables,
thus allowing the highest level of simplification of the equations which leads
to analytical estimates of reactor parameters. On the other hand, more so-
phisticated models taking into account heat and mass transfer along the re-
 14 MATROS AND BUNIMOVICH

actor radius and unsteady state of catalyst are considered elsewhere [138,
139, 143, 1721.
The profiles under flow-reversal operation are usually calculated by sev-
eral solutions of the model equations along the time intervals 0 5 t 5 tCi2.
One direction of flow (U = u) is considered. The first solution is determined
at initial conditions: u(')(O,l,) = do',where w = {Ts, T,, x,, x,} is any variable,
do)are given initial profiles. The initial data for each subsequent solution,
wu+')(O,l), depend on the preceding:
v~+l)(o,z) = uytc,2,L - I),
where j = 1, 2 . . . is the number of the solution. The calculations are finished
when a cycling convergence criterion or balance condition (6) is satisfied.
Certain difficulties related to the large expenditure of computational time
Downloaded by [171.67.34.205] at 22:03 14 March 2013

have been discussed [76, 771. Those problems are caused by steep tempera-
ture or concentration gradients and the large number of cycles required for
solution convergence. The numerical procedure is substantially simplified if
an assumption about quasi-steady states of fast variables T,, x,, and x, is made
[74-771. After that, linear Eqs. (12) and (14) along the space interval can be
solved analytically [87, 881, and the system of Eqs. (11)-(16) is reduced to
one parabolic partial differential equation coupled with algebraic equations
for the gas temperature and concentration profiles.
Bhatia [76] and Gupta and Bhatia [77] compared various numerical tech-
niques for solution of the heterogeneous plug-flow model Eqs. (8)-(10) as-
suming quasi-steady-state concentration and gas temperature profiles. They
transformed steady-state differential equations describing rapid variables (x
and T,) profiles to integral equations. The resulting integrodifferential equa-
tion with respect to T,(t) was solved by the explicit Runge-Kutta method,
with the aid of perturbation procedure [76] and a finite element collocation
technique [77]. The perturbation procedure used by Bhatia [76] consists of
an approximation of the solid temperature profiles by a series in the power
of a small parameter Z = tc,z.The authors truncated the series after the first-
order term. This method allowed a reduction in computational time by a factor
of almost 10. Unfortunately, it did not yield sufficient accuracy at high cycling
periods because of the algebraic complexity of a higher-order expansion. The
finite element method reduces the computational time by a factor of at least
4-7 and can be used for any value of cycle time [77]. Various numerical
methods used for RFO analysis with more general one-dimensional models
[for examples, Eqs. (11)-(16)] have been described elsewhere [80, 87, 91,
117, 129, 132, 135, 1671.

B. Simplified Approaches at Low and High Reversal Frequencies

Figures 5(a)-5(c) show the temperature and SO, conversion profiles

calculated for the conditions close to the experimental ones for Fig. 2. Equa-
 REVERSE-FLOW OPERATION 15

tions (11)-(16) were used and, additionally, quasi-steady states for gas phase
concentrations and temperature were assumed. The function W(T,, x,) is de-
scribed by a kinetic model [89]. The profiles in Figs. 5(a)-(c) reflect three
typical stabilized regimes obtained at medium, short, and long cycle duration.
As one can see from Fig. 5, the reaction proceeds predominantly in two
narrow moving zones. The first zone belongs to a portion of the temperature
profile rising from about 380°C, and the second lies in the similar portion of
the falling temperature profile.
At high reversal frequencies [Fig. 5(b)] temperature profiles oscillate
inside very narrow limits. This means that the cycle time is substantially
shorter than the time needed for the reaction zone to travel through the
catalyst bed
Downloaded by [171.67.34.205] at 22:03 14 March 2013

On the other hand, the cycle period is sufficiently larger than the gas flow
residence time, so quasi-steady state of the gas temperature and concentration
is still valid. The assumptions described above have been used by Boreskov
et al. [71] and Matros [20] to obtain a so-called relaxed steady-state model.
The term “relaxed steady state” was originally discussed by Bailey [6] in
his analysis of the possible types of forced unsteady-state processes. Boreskov

KJ

100

Length of bed (dimensionless) Length of bed (dimensionless) Length of bed (dimensionless)

.- 0

0 0

Length of bed (dimensionless) Length of bed (dimensionless) Length of bed (dimensionless)

FIG. 5 . Temperature and conversion profiles during a semicycle for reverse

flow operation: (a) L = 0.44 m, t, = 120 min; (b) L = 0.44 m, t, = 12 min; (c) L =
0.88 m, t, = 240 min; 1, 2, and 3, beginning, half, and end of the reversal period,
l’, conversion profiles immediately before switching.
 16 MATROS AND BUNIMOVICH

all the terms of Eq. (1) over a cycle, taking the limit at t, -
et al. [71] used a pseudohomogeneous model, Eqs. (1)-(5). After averaging
0, accounting
for two possible flow directions using Eq. (3), the following equation de-
scribing temperature profiles was obtained:

where w = W/Cinp;arrows indicate two different flow directions, and x is

conversion of the reactant. The expressions for the conversion profiles are
obtained from Eq. (7):

dx,
Downloaded by [171.67.34.205] at 22:03 14 March 2013

u __ = -w(x,,T)
dl
The boundary conditions are similar to eqs. (4) and (5).
For the converged cycling, the steady-state form of Eq. (17) can be
applied to describe the process at high reversal frequencies. Chumakova and
Zolotarskii [90] proved that the solution of the resulting system of ordinary
differential equations is symmetric, namely, T(Z) = T(L - I ) and x,(l) =
x-(L - I>. Temperature profile has a symmetrical bell-shaped form, so the
maximum temperature T,,, is achieved at the center of catalyst bed. The order
of Eq. (17) can be reduced after integration along the interval [OC] and
elimination of w(x,T) using mass balance Eqs. (18). Accounting for symmetry
of the solution, a reduced form is:

and boundary conditions can be represented as:

at 1 = 0: x, = 0; x, = x,,,; and T = To,, = Tinp+ ATasouI (20)
at 1 = L J 2 : x, = x, (21)
Principally the same approach has been used by Eigenberger and Nieken
[53] in their analysis of RFO for catalytic incineration of volatile organic
compounds. They observed that high-frequency switching makes reverse-flow
operation similar to a steady-state process in an U-tube reactor with indirect
countercurrent heat exchange.
Haynes with coworkers [72] used a relaxed steady-state model for en-
dothermic process modeling. In this case a convection term was not excluded
and a numerical solution of reduced equations is not as easy as for the sym-
metrical case. Nevertheless, the solution can be found much faster than as-
suming a transient model, and a large number of parametric dependencies
can be studied in a reasonable time [72]. Ivanov et al. [91] used this approach
for analysis of two parallel exothermic reactions. Bhatia [76] derived a re-
 REVERSE-FLOW OPERATION 17

laxed steady-state model as a zero-order problem and used it in combination

with the regular perturbation technique. Equations of relaxed steady states for
the heterogeneous model Eqs. (11)-(16) have been formulated recently by
us [171].
Boreskov et al. [71] and Matros [20] considered a first-order irrevers-
ible reaction with w(T&) = ko exp(-E/RT)(l - x). In this case a relationship
between maximum temperature and outlet conversion can be determined from
the equation:

An integral equation similar to Eq. (22) has been given recently by Nieken
et al. [173]. These authors accounted also for external mass transfer limitation.
Downloaded by [171.67.34.205] at 22:03 14 March 2013

The integral in the left side of Eq. (22) can be found approximately at
high energy of activation using what is known as Frank-Kamenetski substi-
tution [36]:

where:
0 = (T - T,)/(RT; E ) is dimensionless temperature
b = RTJE
Another limiting case is low switching frequency in a long packed bed
[see Fig. 5(c)].
In this case, a rising temperature profile has a nearly constant shape and
moves with a fairly constant velocity during a large part of the cycle. This
situation can be approached by the model of steady-state reaction front move-
ment. Such a model describes the so-called creeping front phenomenon,
which has been extensively studied in a number of publications. Because
RFO performance for an exothermic reaction has common features with this
phenomenon of creping front, we include a short review related to its analysis.
As early as in 1959 Wicke and Vortmeyer [92] studied experimentally
the structures of a combustion front in a coal particulate bed with air blowing
through the bed. This system is quite close to a catalytic fixed bed reactor.
Further, Padberg and Wicke [93] and Fieguth and Wicke [94] experimentally
observed creeping profiles for CO oxidation over a platinum catalyst. Padberg
and Wicke [95] wrote the steady-state model along a coordinate system mov-
ing together with temperature and conversion profiles. This approach was
initially applied in the classical combustion theory developed by Zeldovich
and Frank-Kamenetski [96, 36, 971.
This theory assumes a constant velocity of heat front propagation (uF)
and considers a steady traveling wave solution depending on the single var-
iable r = 1 + u,t. Assuming invariability of the moving temperature and
 18 MATROS AND BUNIMOVICH

conversion profiles leads to an idealization of the catalyst bed by an infinite

length system. A reduced system for a quasihomogeneous model (1)-(2) can
be represented by:
Xef d2T dT ATadw(x,T)
ucg& dr2
(1 - yo) -
dr
+ U
=o

with boundary conditions:

and x - 0 (26)

- dT
Downloaded by [171.67.34.205] at 22:03 14 March 2013

at r +a: --0 (27)

dr
where o = uF/uis dimensionless velocity.
The order of Eq. (24) can be reduced, and a simple relationship between
velocity and maximum temperature of the heat front is derived as follows:

Equation (28) is an energy balance for steady-state reaction front movement.

cases eg/y -
It was originally obtained by Wicke and Vortmeyer [92]. In many practical

be eliminated.
1 - lop7and the last fraction in the right side of Eq. (28) can

The heat front can move in the direction opposite to the reaction mixture.
In this case the reaction heat will be accumulated in the catalyst bed, and
outlet (or maximum) temperature will be less than its adiabatic value: AT,,
xc(TF)> TF - Tinpat uF c 0. A standing wave with uF = 0 is also possible.
All these types of creeping phenomena were studied by Wicke and coworkers
[92-951. Modeling studies were performed in [95, 98-1111. Rhee, Lewis,
and Amundson [ 1011 performed detailed numerical analysis of the phenom-
enon on the basis of a two-phase model, Eqs. (11)-(14), accounting in ad-
dition for conductivity and diffusion in the gas phase. Numerical calculations
of heat front propagation and analytical solutions for irreversible reaction
kinetics were presented in Refs. 102-104. Vortmeyer [lo51 presented a sem-
iempirical correlation for the velocity of the reaction front movement, which
was further generalized by Vortmeyer and Jahnel [98]. These authors [98]
studied the influence of radiation heat transfer. Simon and Vortmeyer [lo61
conducted an experimental study of ethane oxidation over a supported pal-
ladium catalyst and tried to elucidate the influence of heat losses on heat front
parameters. Particular reaction systems have been studied experimentally and
numerically [107, 1081. Noskov et al. [lo91 numerically studied the influence
 REVERSE-FLOW OPERATION 19

of intraparticle heat conductivity and diffusion. Two parallel reactions as well

as reaction with irreversible adsorption and catalyst deactivation were con-
sidered recently by Il'in and Luss [110, 1111.
Attempts at obtaining analytical expressions for the movement velocity
have been undertaken in some of the first studies of the phenomena [98, 99,
102, 1051. A series of papers [69, 112-1161 published in Russia contain a
comprehensive qualitative analysis of the creeping phenomena have been
summarized recently in the book by Kiselev [78].
Several methods for approximate determination of reaction front param-
eters are known [78]. Frequently applied is a so-called narrow reaction zone
approach [96], which includes an approximate linearized T(x) dependency in
the vicinity of maximum temperature. For a pseudohomogeneous model, Eqs.
(25)-(28), and reversible first-order reaction this linearized form may be rep-
Downloaded by [171.67.34.205] at 22:03 14 March 2013

resented as:

Assuming a high value of the second term in brackets and combining

(29) with (25) and (28) yields the integral equation:

that can be solved approximately using Eq. (23). Such a method has been
applied by Kiselev with coworkers [78, 112-1161.

C. Exothermic Reaction Performance in One Catalyst Bed

Temperature and conversion profiles (Figs. 2 and 5 ) are formed after

preliminary preheating of the catalyst bed to a temperatures where the reac-
tion rate is reasonably high. This start-up operation can be characterized as
a reaction ignition. With longer cycles, such that t, > 2(L/u,), extinction of
the reaction will occur since the reaction front leaves the catalyst bed. After
that, reverse-flow operation is "useless" because of low temperatures and
very low conversions at any point in the catalyst bed. Shorter cycle duration
leads to wider margins of stability; however, the process can still be extin-
guished even at short cycle duration.
Figure 6 shows stabilized axial temperature and conversion profiles and
Fig. 7 represents dependencies of maximum temperature and outlet conver-
sion on residence time in the catalyst bed. We did these calculations on the
basis of a relaxed steady-state model, Eqs. (18)-(21) and first-order reversible
reaction kinetics. Similar results have been obtained [20, 711 for an irrevers-
ible reaction. These examples demonstrate the possibility of having three
stabilized cyclic solutions at low inlet temperatures (see lines 1-3, Fig. 6;
and lines 1-4, Fig. 7).
 20 MATROS AND BUNIMOVICH

The numerical procedure [71] used to calculate the profiles in Fig. 6

does not allow for evaluating the process stability. Indirect conclusions about
the stability can be drawn from numerical simulation using the dynamic
model [20, 291. Two extreme regimes, upper and lower (Fig. 6, lines 3 and
lo), are stable. A n intermediate regime (line 2, Fig. 6) is unstable. It is char-
acterized by a lower conversion than in the upper regime. A qualitative picture
of steady-state regimes (Fig. 6) observed at high reversal frequencies seems
to be realistic also for long cycles, although an intermediate unstable solution
for the latter case has not been found in known studies of reverse-flow
reactors.
The multiplicity leads to an S-shaped form of parametric dependencies
[Figs. 7(a), and 7(b)]. This type of curve has been numerically studied by
Boreskov et al. [71], Matros [20], Eigenberger and Nieken [29], Snyder and
Downloaded by [171.67.34.205] at 22:03 14 March 2013

Subramaniam [117], and Young et al. [88]. Recently Chumakova with co-
workers [90, 1181 carried out a detailed parametric investigation of a relaxed
steady-state model for one irreversible reaction.
As one can see from Fig. 7, in a long catalyst bed (7, = L/u > 3 s), the

20

FIG. 6. (a) Temperature and (b) conversion profiles for three possible states
of a reactor. Calculation on the basis of relaxed steady-state model for reversible
first-order exothermic reaction. AT,, = 100°C; h,f/(u2p~c,)= 0.166 s; k,, = 3.13.10'
s I ; E = 84 kJ/mole; E - = 160 kJ/mole; ko- = 9.8.10"' s-
 REVERSE-FLOW OPERATION 21

2
Downloaded by [171.67.34.205] at 22:03 14 March 2013

1 .o

0.6

r* 0.5

0.3

0.0

r,, s
(b)
FIG. 7. (a) Maximum temperature and (b) conversion versus space time of
gas in the catalyst bed at various inlet gas temperatures. 1, Tin,,= 0°C; 2, Tinp=
100°C; 3, Tinp= 200°C; 4, Ti,lp= 300°C; 5, Tinp= 300°C. Parameters correspond to
Fig. 6.

maximum temperature ( T K )and outlet conversion (xou,)tend to approach some

asymptotic values Tm : and x:. Snyder and Subramaniam [ 1171 designated
these conditions as a "thermodynamic limitation" in contrast to those that
are close to process extinction, which is referred to by these authors as "ki-
netic limitation" condition.
An estimation of maximum temperature limit (Tkm) for the first-order
irreversible reaction and low inlet temperature has been derived by Boreskov
et al. [71] from Eqs. (22) and (23). A more general estimation for first-order
reversible reaction is given by:
 22 MATROS AND BUNIMOVICH

E
R(T:~)~
~ )equilibrium conversion at Ti"'
x ~ ( T : is
m = 2 is a constant numerical coefficient [20]
Parameter Ze, Zeldovich number, characterizes dimensional temperature rise
in a catalyst bed. Expression (31) is implicit with respect to maximum tem-
perature but it can be easily solved using simple iterations.
Zolotarski [119] obtained a simple estimation for maximum outlet con-
version in the relaxed steady state:

X,(T;"')) -
-"I
Downloaded by [171.67.34.205] at 22:03 14 March 2013

E -E -

where E- is a reverse reaction activation energy. It is seen that maximum

conversion depends in a unique fashion on maximum temperature and can
only decrease as Tkmincreases. The difference between outlet and equilibrium
conversions, Ax = x!u": - xe(Tg"'), will be higher for reactions with higher
ratio of the reverse and direct activation energies.
As early as 1979 Broeskov et al. [112] obtained analytical estimations
for maximum temperature in a steadily moving reaction front. The simplest
pseudohomogeneous model, Eqs. (24)-(27), gives a dependency similar to
Eq. (31). Kiselev and Matros [69] obtained estimations of maximum tem-
perature ( T F )in the creeping front for a more general case accounting for
diffusion inside catalyst pellet, heat conductivity of pellet, and heat transfer
between solid and gas phases. This estimation can be represented by the
equation:

where qi(TF) is a parameter dependent on the temperature and diffusion co-

efficient inside the catalyst pellet which can be treated as a weight-average
effectiveness factor along the heat front coordinate; A,f is an effective heat
conductivity corresponding to the approximate process description by the
pseudo homogeneous model. Effectiveness factor qr depends on the maxi-
mum temperature as [78]:

where:
'J'(T) = R , ,/y, Thiele-type modulus for reversible reaction
m, = parameter and index representing the geometry of catalyst pellet:
 REVERSE-FLOW OPERATION 23

rn = 0 for flat plate,

m = 1 for cylinder (no axial diffusion), and
rn = 2 for sphere
j, = chT; j , = lo(q) (where lois a zero-order modified Bessel func-
tion), and
j , = (shT)/'P.
Kiselev with coworkers [69, 70, 781 obtained an estimation of kcf that
includes three main factors responsible for heat dissipation in the catalyst
bed. They are: nonzero axial heat conductivity (As,ef > 0), limitation of intra-
particle heat transfer (aoc m), and limitation of heat transfer by conductivity
inside catalyst pellet, [A,/(R~c,pi)] c a).This estimation can be written in the
form:
Downloaded by [171.67.34.205] at 22:03 14 March 2013

Equation (35) describes growth of heat dispersion while decreasing the

effects of interphase heat transfer and intraparticle heat conductivity. The
meaning of the multiplier:

is that, in a moving temperature front, influence of intra- and interparticle

heat transfer to overall heat dissipation decreases as the creeping velocity
decreases. Addition of this multiplier means also that no opposite direction

conductivity of the solid phase when As,ec -

of front movement (oy c 0) is possible in a case of negligible axial heat
0.
Equations similar to Eqs. (33)-(35) can be derived from the relaxed
steady-state model formulated for general situation including the influence of
interphase and interparticle heat transfer. The only differences are: my = 1,
and the denominator in the square root is two times less than in Eq. (33).

(35), at o y = 1 and A, -
Note that the formal representation of effective heat conductivity, Eq.
~0 has been disclosed by Vortmeyer and Shaefer

[ 1201. This representation has been discussed in an application to reverse-

flow reactor by Kunitskii et al. [122], Bunimovich et al. [123], Matros [20],
and, recently, by Nieken et al. [173]. Note also that, if the effective heat
transfer coefficient in a heterogeneous plug-flow model, Eqs. (8)-( lo), is
estimated as
(UC,PbJ2
a,[= T
L f

Eq. (35) yields the result known from the studies of heat regenerators [121]
as a zero-order approximation for the heat transfer coefficient.
At linear velocities as high as 0.1-1.0 m/sec, commonly used in indus-
trial catalytic reactors, the axial heat conductivity has no pronounced effect.
 24 MATROS AND BUNIMOVICH

Influence of intraparticle heat conductivity is exhibited only at elevated mass

flow rate (up:), typical for high-pressure processes. In this case a substantial
increase in the heat transfer coefficient cxo is achieved that is the same as
increasing the Biot number:

BiT = aoRpc,p,
AP

At large values of Bi,, heat transfer is limited by heat conductivity in the

catalyst particle [20, 661.
Boreskov et al. [71] and Matros [20] obtained the estimates for critical
conditions corresponding to process extinction in a case of high switching
frequency. For first-order reversible reaction a minimum space time (7:) re-
quired to provide the high-temperature regime is given by the equation:
Downloaded by [171.67.34.205] at 22:03 14 March 2013

Parameter can be taken from Eq. (35) at o y = 1.

The influence of switching frequency on the process stability is esti-
mated [53, 122-1241 by a simple linear equation:

In Eq. (37) the velocity of heat front propagation uF can be determined from
Eqs. (30) and (33); and L", from Eq. (36).
Approximate expressions, Eqs. (31)-(37), coincide well with a number
of known RFO simulation results obtained for simple exothermic reactions
[20, 29, 30, 73, 76, 88, 1171. This is due to the fact that in a long reactor
operating at low inlet temperature, the cycle duration does not greatly influ-
ence such parameters as maximum temperature T,,, and average conversion
-
xout[20, 88, 1171. This empirical rule is valid for intermediate cycle duration:

L L
2 - E 5 t, 5 2 - y
U U

At long cycles as well as at very short cycles [SO], reactor extinction is

observed.
The main results of parametric analysis of RFO made on the basis of
estimations, using Eqs. (33)-(39) and known simulation studies are:
a. The lower the rate constant of the reaction, the higher the maximum
temperature [20, 29, 30, 761. This can be seen from Eqs. (31) and
(33). On the other hand, at high maximum temperature, critical
space time, Eq. (36), increases and the system approaches the ex-
tinction limit.
 REVERSE-FLOW OPERATION 25

b. The higher the activation energy, the higher the maximum temper-
ature and the critical reactor length. In their simulation studies Ei-
genberger and Nieken [29], Bhatia [76], and Matros et al. [30] var-
ied the activation energy while holding the fixed reaction rate
constant at a certain temperature. In this case the numerical depen-
dency T,,,,,(E) has a maximum at some intermediate activation en-
ergy. The same observation results also from Eqs. (31) and (33).
c. Higher axial conductivity (A& or a lower external heat transfer
coefficient (ao)leads to a decrease in maximum temperature. At high
values of the overall coefficient of heat conductivity, kef, the extinc-
tion of the reaction occurs in accordance with Eqs. (31) and (35)-
(37), and numerical simulation results [20, 29, 761. Note that the
effective heat conductivity has no noticeable effect on a steady state
Downloaded by [171.67.34.205] at 22:03 14 March 2013

in adiabatic packed bed reactor.

d. The value of minimum adiabatic temperature rise AT$ correspond-
ing to the process extinction can be determined for any set of reactor
parameters [20, 1261 and can be controlled by proper selection of
reactor design parameters. The policy for such a selection is deter-
mined by Eqs. (31), (36), and (37).
e. Maximum temperature is not affected significantly by the inlet tem-

temperature, Ze -
perature in a ___domain of low Tlnpaccording to Eq. (31). At high inlet

- 0 and Eq. (31) degenerates to the equation

Tkm To,, = T,, - ATadx::, so the maximum temperature ap-
proaches the temperature at the reactor outlet. These limiting prop-
erties were demonstrated by numerical studies [20, 30, 80, 1171. In
Ref. 117 great attention is paid to the investigation of T,,, versus
T,npdependencies for various situations. Also this work studied the
ability of RFO for energy accumulation or enthalpy trapping. Max-
imum energy accumulation is provided at intermediate inlet tem-
peratures. A minimum for the T,,, versus Tlnpcurves was found [30,
1171. This does not agree with the Eq. (31), which predicts monot-
onous dependency Tkm(TInp). The explanation lies in the fact that the
temperature profile movement during each half-cycle acquires larger
magnitude at low inlet temperature and this leads to higher temper-
ature gradients.
f. For reversible reactions large catalyst pellet size often leads to a
higher conversion. This is one of the most pronounced results of
the theory described above. Specifically, pellet size influences the
effective heat conductivity kcf and average effectiveness factor 3,.
Analysis of these parameters gives a maximum temperature de-
crease and corresponding increase in conversion despite a decrease
in the effectiveness factor. However, small catalyst pellet size favors
stability of operation and reduces packed bed length. The effect of
pellet size on the SO, oxidation reaction was discussed by Buni-
movich et al. [123] and Sapundzhiev et al. [127].
 26 MATROS AND BUNIMOVICH

g. Higher linear velocity leads to an increase in maximum temperature

if the residence time is large and unchanged. Once the length of a
packed bed is fixed, extinction will occur at both high and low linear
velocities. This follows from Eqs. (35)-(36) and simulation studies
[90, 1181 which show that the multiplicity ___domain forms an isolated
___domain in a parametric plot (T,,,, u). Chumakova and Matros [118]
have revealed also that there exists a second multiplicity ___domain at
very low linear velocity when a reverse-flow reactor turns to a system
close to a continuous stirred tank reactor. These exotic conditions are
not covered by estimation from Eqs. (31)-(36).
Snyder and Subramaniam [117] noted a remarkably long duration of the
extinction process. At short periods of switching (120 sec), it required more
than 100 flow reversals to see a noticeable temperature decrease. This ob-
Downloaded by [171.67.34.205] at 22:03 14 March 2013

servation is not surprising when taking into account the characteristic time
for temperature stabilization
L 2E,C,p,
7, F=Z ____
Ad
which can be evaluated from Eq. (17). This time is as long as several hours
for usual process conditions.
The theory described above is restricted by the assumption of quasi-
steady-state temperature and concentration in the gas phase [see Eq. (7)]. Van
den Bussche et al. [80] provided a numerical study of methanol synthesis on
the basis of the model given by Eqs. (11)-(16). These authors showed that
mass and heat capacity of the bulk gas phase and mass capacity of gas inside
catalyst pellets lead to a decrease in maximum temperature in comparison
with the quasi-steady-state approach. The reason for this relates to mass ac-
cumulation along the heat front length. Trapping of reactants in the gas phase
causes a decrease in the ingress of fresh reactant into the high-temperature
zone. This spreads out the reaction and heat generation over a larger section
of the catalyst bed and gives some improvement in conversion during the
main part of each semicycle. Immediately after a flow reversal the uncon-
verted gas is displaced from the void volumes in the boundary section of the
bed and this causes a decrease in overall conversion.

D. Use of Inert Ceramic Packing and Heat Removal

In a reactor with inert front and back sections surrounding the catalyst
[Fig. 4(a)], the temperature after inert ceramic packing can be estimated by
an almost linear expression:
 REVERSE-FLOW OPERATION 27

where index inr relates to the inert packing; Lin, is the length of one of the
inert beds, and i$ is effective heat conductivity of the inert packing. Pre-
heating of inlet gas in the inert packed bed is improved by an increase in the
length of such bed or a decrease in its effective heat conductivity. It enables
a substantial reduction in catalyst bed length for irreversible processes such
as VOC oxidation [20, 29, 30, 126, 1731. For a reversible reaction, over-
loading of the inert packing may induce a drop in conversion because of a
rise in maximum catalyst bed temperature [20, 88, 1251.
Young et al. [88] demonstrated some conversion improvement for a
reversible reaction occurring in a reverse-flow reactor with one or two beds
compared with optimal steady-state conditions in an adiabatic packed bed
reactor. Similar comparison made earlier by us [128] showed that, with the
same catalyst loading, a one-bed reverse-flow reactor can produce a higher
Downloaded by [171.67.34.205] at 22:03 14 March 2013

conversion than a two-bed traditional reactor with interstage cooling operated

at optimum conditions which include inlet temperatures and the ratio between
amounts of catalyst in the two beds.
A detailed modeling study of a method for heat utilization [Fig. 4(b)]
was conducted by Boreskov and Matros with coworkers [20, 87, 1291 for a
first-order irreversible reaction. The reaction parameters were chosen close to
those for methane oxidation over a metal oxide catalyst. It was assumed [129]
that part of the gas passes through a heat exchanger to maintain a constant
temperature Ti:.np,2at the inlet of second catalyst bed. The influence of reactor
parameters on the degree of heat recovery:

was examined, where AT, is a cycle averaged difference between inlet and
outlet gas temperatures in heat exchanger 4 [see Fig. 4(b)]. 6, is the fraction
of gas passing through the heat exchanger. Typical temperature profiles for
this type of operation are shown in Fig. 8(a). Maximum heat recovery, which
corresponds to some optimal set t, and T,np,Z, increases when linear velocity
and packing bed length increase.
Recently, Sapundzhiev et al. [ 1301 published the results of mathematical
modeling of methane combustion over perovskite catalyst in a reactor with
periodic flow reversal. The reactor configuration assumes that part of the inlet
gas is preheated in an intermediate heat exchanger, the energy for preheating
being removed from the center of the reactor. Results of simulation given in
Ref. 130 agree with RFO theory discussed above as well as with earlier
results of modeling a similar system [129].
Figure 8(b) shows an example of temperature profiles in the packed bed
with partial gas withdrawal from the bed center [Fig. 4(c)] proposed by Ei-
genberger with coworkers [29, 53, 54, 1731. With a long ceramic bed, the
outlet reactor temperature can be nearly equal to the inlet temperature, so all
the reaction heat is removed via the gas discharged from the reactor center.
 28 MATROS AND BUNIMOVICH

Maximum fraction of this gas flow aw can be determined as (Tinp+ AT,,

[53], and factor
xout)/Tmdx q h is defined primarily by the efficiency of the heat
recovery unit installed downstream of the gas withdrawal point.
Two types of the heat recovery processes considered above have not
been compared in the literature. Presumably, the process with gas withdrawal
[Fig. 8(b)] is simpler in operation than that with an intermediate heat ex-
changer [Fig. 8(a)]; however, for the latter case a higher energy potential can
be generated due to a major portion of hot gas passing through the heat
recovery unit.

E. Complex Exothermic and Endothermic Processes

Modeling studies have been conducted for several examples of complex

Downloaded by [171.67.34.205] at 22:03 14 March 2013

reaction kinetics.
Authors [34, 351 have discussed selective nitrogen oxide reduction by
ammonia in an oxygen-containing gaseous mixture. It was noted that for any
set of reactor parameters there exists an optimum cycle duration correspond-
ing to minimum NO, and NH, concentrations at the reactor outlet.

A + v,B -+ S + 0, A + v2B -
Zagoruiko et al. [132] discussed the following reaction scheme:
E + 0, A + v,E 2 S +D

0 - 00
Depth 01 ihe bed, dimensionless Depth of the bed, dimensionless

FIG. 8. Temperature profiles in reactors with intermediate heat removal: (a)

gas after heat recovery is returned to the reactor [flow diagram in Fig. 4(b)]; (b)
portion of the gas is withdrawn without returning [Fig. 4(c)]; 1, 2, and 3, beginning,
middle, and end of a semicycle, respectively.
 REVERSE-FLOW OPERATION 29

where all steps are exothermic and the two first steps are irreversible. The
authors [132] also assumed rather high values of reaction rate constants at
ambient temperature. This is typical for the process of selective hydrogen
sulfide oxidation over bauxite catalyst. As in the case of NO, reduction [35],
the pronounced optimum in dependency of the desired yield of product S on
the reversal frequency was found [132]. At high inlet temperature this yield
becomes thermodynamically controlled and a steady-state mode of operation
may be preferable over periodic flow reversal [132].

thermic reactions A - B and C -

Multiplicity phenomena in the case of two independent first-order exo-
.Dwere studied by Ivanov et al. [91]. This
type of kinetics can describe catalytic oxidation of volatile organic com-
pounds with an excess of oxygen. The calculation shows a complex multi-
plicity of converged reactor states in a case when reaction rate constants for
Downloaded by [171.67.34.205] at 22:03 14 March 2013

two parallel processes differ considerably. The difference in reaction rates is

so large that slowly reacted component A is practically nonreactive at a certain
temperature, assuring complete conversion of the second component C. Three
states are stable, and between them two unstable regimes occur. This possible
structure of cyclic steady-states was mentioned later [54] and experimentally
observed by Nieken, Kolios, and Eigenberger [165, 1731.
Zagoruiko with coauthors [133-1361 carried out theoretical analyses of
the catalytic Claus process (hydrogen sulfide oxidation by SO,) involving
sulfur condensation and evaporation. The overall rate of liquid sulfur accu-
mulation or evaporation was assumed to be determined only by the rate of
sulfur transport to the internal particle surface. Also a linear dependency of
bulk catalyst activity on the liquid sulfur fraction within a particle and neg-
ligible sulfur condensation between the catalyst particles were assumed.
The rate of H,S oxidation reaction is very high even at ambient gas inlet
temperature, but sulfur condensing at low temperatures blocks active catalyst
surface and the reaction stops because of catalyst deactivation. The transient
processes after the reaction mixture contacts the “fresh” catalyst include
three consecutive dynamic stages [135]: (1) fast (within seconds) change in
reactant composition to form quasi-steady-state composition profiles corre-
sponding to the initial temperature; (2) relatively fast (10s of minutes) tem-
perature change until a quasi-steady-state temperature distribution is attained
in the catalyst bed; and (3) very slow (several hours) movement of the con-
densed sulfur and temperature fronts through the catalyst bed. In a reverse-
flow reactor, periodic evaporation of condensed sulfur proceeds from the out-
let parts of the catalyst bed. Though it is difficult to remove all sulfur
condensed within catalyst pellets at the bed edges, after a certain period a
balance between the amount of sulfur condensed and evaporated is attained
[135, 1361. Thus, a self-sustained regeneration of the catalyst can be created.
Figure 9 shows profiles of temperature and relative amounts of liquid sulfur
in catalyst pellets along the bed length for a stabilized reverse-flow operation
[ 1351.
A detailed modeling study of endothermic reaction performance with an
 30 MATROS AND BUNLMOVICH

open-loop reactor configuration (Fig. 3) has been done by Haynes et al. [72].
The problem of multiplicity and stability does not exist in this case because
only one stable state can be attained at any initial temperature. The authors
[72] characterized the intensity of thermal regeneration by Peclet number for
steam flow:

where the additional index st means the parameters corresponding to the

steam flow. Haynes et al. [72] have constructed a dependency of cycle av-
eraged conversion on Pest and compared it with the same dependency for a
steady-state process attained for the case of blending the reactant with steam
and the resultant mixture passing through catalyst beds two times longer than
Downloaded by [171.67.34.205] at 22:03 14 March 2013

one bed in a reverse-flow operation. The authors [72] distinguished three

regions in the Peatdomain, characterized by qualitatively different temperature
profiles. At low steam flow rate, reverse-flow operation leads to a decrease

8,%
B
225

200

175

150

125
I .
0 a2 0.4 0.6 0.8 I

XLS

0.8

0.6

0.4

0.2

0
I
0 02 0.4 0.6 0.8 I

%L

FIG. 9. Profiles of temperature (0) and relative concentration of liquid sulfur

(x,.~)along the catalyst bed length (adopted from Zagoruiko et al. [135]): A , B, and
C, start, middle, and end of semicycle respectively; arrow, direction of gas flow.
 REVERSE-FLOW OPERATION 31

in conversion compared to steady-state operation because energy is predom-

inantly removed from catalyst bed by the reverse stream of the reactants. At
a moderate steam flow rate corresponding to approximate equality of reactant
and steam Pe numbers, the regeneration energy becomes sufficiently high to
heat the reactant stream and cycle-averaged conversion approaches that of
steady state. At high steam flow, the temperature increases along the bed
length, then flattens, and nearly complete conversion can be attained. This
complete conversion is very difficult to achieve in steady-state operation be-
cause the temperature decreases at low steam flow rates and residence time
decreases at high steam flow rates. One of the conclusions from the analysis
by Haynes et al. [72] was that the efficiency of RFO increases at low cycle
duration.
This conclusion had been drawn previously by Heggs, who presented
Downloaded by [171.67.34.205] at 22:03 14 March 2013

the results of mathematical modeling of ethylbenzene dehydrogenation on

iron oxide catalyst [47]. Schematic representation of the reaction set studied
[47] is:
+ H2
--
EB t S (39)
EB P, (40)
EB + H, P, (41)
Reaction (39), describing specifically dehydrogenation of ethylbenzene (EB)
to styrene (S), is highly exothermic and reversible. Irreversible reactions of
ethylbenzene destruction (40), and ethylbenzene hydrogenation (41) give un-
desirable products (PI, Pz), reaction (41) being exothermic. The higher the
temperature, the lower the selectivity. Heggs [47] did not give a detailed
modeling study of this complex system, demonstrating only the possibility of
an increase in both conversion and selectivity in a reverse-flow reactor system
(Fig. 3). The decrease in cycle duration was noted to be favorable for a
conversion increase but resulted in a decrease in process selectivity.
The same reactions set, (39)-(41), was used by Snyder and Subraman-
iam [137] for studying the process of ethylbenzene dehydrogenation in a
reverse-flow reactor with boundary ceramic packing and internal steam in-
jection [possible flow diagrams are shown in Figure 4(f) and 4(g)]. Steam
injection into one point located in the center of the catalyst bed results in
high local temperature and low selectivity toward styrene formation. To over-
come this limitation the authors [ 1371 suggested multiple steam injection in
several points symmetrical about the center of the catalyst bed. During a
semicycle, steam is fed only into the points located in the half of bed nearest
to the ethylbenzene feed inlet. To prevent coke formation, a fraction of the
steam is added to the reactor inlet. Figure 10 shows the temperature and
conversion profiles calculated [137] for such a process. The authors [137]
concluded that this method of steam introduction achieves the same or higher
selectivity and yield as steady-state operation, and at substantially lower
steam consumption.
 32 MATROS AND BUNIMOVICH

E;: Temperature Nonuniformity

Two articles [138, 1391 have examined spatial nonuniformity in a re-

verse-flow reactor. The authors were interested in possible nonuniform tem-
perature fields during exothermic processes with low adiabatic temperature
rise in a large heat-insulated reactor with a diameter on the order of meters.
A two-dimensional heterogeneous model with first-order reaction kinetics was
used with the assumption of low heat capacity of the reactor walls. To esti-
mate the influence of the void fraction distribution E(r), a flat radial profile
of reactor pressure drop AZ’(r) was assumed, and the radial distribution of
superficial velocity u(r) was calculated from the Ergun equation at a given
total volumetric mass flow rate:
Downloaded by [171.67.34.205] at 22:03 14 March 2013

where f , and fi are functions independent of fluid velocity.

A monotonous approximation of the radial porosity profile was applied,
describing increase of porosity in a peripheral zone with a width of 5-10
catalyst pellets.
Three different types of reactor behavior were found [138]. If only heat
losses are included in the consideration and porosity distribution is uniform,
a smooth temperature decrease in the periphery of the reactor is observed.
At low adiabatic temperature rise, even small heat losses can lead to
extinction of the reaction near the wall and then a very slow movement of
the cold zone from the wall toward the reactor center (see Fig. 11).The time
for complete reactor extinction may take several days. During this transition

1000 0.4

h
Y 0.3
v

-z
E3 800 -E
x
v1
0

a 0.2 ;
E ._
0

c r
U 600 >
C
rn 0.1 $

400 0.0
0 1 2 3 4
Axial Position ( r n )

Steam:EB c. 8: Tmix - 61 1K

FIG. 10. Temperature and conversion profiles in reverse-flow reactor for

ethylbenzene dehydrogenation to styrene (adopted from Snyder and Subramaniam
[137]). Steam feed is split between five axial symmetric locations.
 REVERSE-FLOW OPERATION 33

the conversion along the reactor axis can remain unaffected by the wall
effects.
Substantial temperature gradients have been observed for the case of
nonuniform porosity. For that case, two different stable regimes typical for a
one-dimensional model (see Figs. 6 and 7) can “coexist” simultaneously in
the reactor. In the central part of the bed, high temperature and nearly com-
plete conversion are attained, while near the wall, where local porosity is
higher, the temperature is close to inlet temperature and conversion is about
zero. The overall cycle-averaged conversion can have a stable value of about
0.5 that is impossible for stabilized periodic operation described by a one-
dimensional model. It was found [ 1381 that increasing the packed bed length
allows a compensating negative impact of radial nonuniformity on the overall
conversion.
Downloaded by [171.67.34.205] at 22:03 14 March 2013

G. Influence of Catalyst Dynamics

Reverse-flow operation induces substantial oscillating concentrations

and temperature fields in the catalyst bed. If an exothermic reaction proceeds,
each catalyst pellet at the boundary sections of the bed experiences heating
when the converted gas mixture passes through and cooling when the reactant
gas mixture passes through (Fig. 5). The extent of temperature and compo-

0 YC’,’ , . . . , , . , . . , . . , . , , . . . . , . , , , , , . . . . . . . , , ~ . . . . , \L%.l
0 2 4 6 a 10

Reactor diameter. rn

FIG. 11. Dynamics of reactor extinction. Temperature profiles along the di-
ameter of the reactor for SO, oxidation. Inlet concentration of SO, = 1.8%; u = 0.3
m/s;L=2m;l,t=0.2h,2,t= llh,3,t=24h,4,t=37h,and5,t=5Oh.
(Example calculated by N.V. Vernikovskaia, Institute of Catalysis, Novosibirsk,
Russia.)
 34 MATROS AND BUNIMOVICH

sition oscillations depends on the pellet position in the bed, the cycling period,
and other process parameters.
The cycle duration time scale, defined as the time required for a heat
wave to pass through a bed, varies typically from several minutes for high-
pressure processes to several dozens minutes for normal-pressure processes
such as sulfur dioxide oxidation. Obviously, if the reaction steps or side cat-
alytic processes which define the catalyst state dynamics occur on a time
scale comparable with the cycle duration, then the catalyst state will be sub-
stantially nonsteady.
The importance of the catalyst state dynamics was shown experimentally
for sulfur dioxide oxidation over vanadium catalysts [20, 1401 and methanol
synthesis over Cu/ZnO catalyst [ 1411. Some indication of the strong influence
of catalyst dynamics was also demonstrated for RFO in NO, reduction over
Downloaded by [171.67.34.205] at 22:03 14 March 2013

vanadiumhitanium catalyst [35].

Kiselev [78, 1141 presented a theoretical study of surface dynamic in-
fluence on the phenomena of stationary heat wave movement for a case of
the simplest two-stage reaction mechanism:
A + **A* (42)
(43)
First adsorption stage (42) is exothermic whereas the second desorption stage
(43) is endothermic, total reaction heat being equal to the sum of specific
heat generated by adsorption and desorption processes (QT = qa + qd). The
inertia of surface concentration is determined by the parameter
(1 - E)S,Z
K =
Cinpy
where S, is specific surface of the catalyst pellet, 2 is maximum density of
adsorbed species, and CinPis inlet concentration of reaction component A. At
high values of parameter K, the adsorption capacity is high enough to create
a continual difference between the consumption of inlet component A and the
production of outlet species B. If heat wave movement coincides with the
direction of gas flow, the outlet concentration of the product B is less than
the conversion of the product A. Another important consequence is a lag
of endothermic desorption stage (43) behind the exothermic adsorption stage
(42). This results in an increase in maximum temperature TFin comparison
with the creeping front movement at the condition of fast reaction.
Oruzheinikov et al. [142] conducted a numerical study of the reverse-
flow operation on the basis of reaction mechanisms (42)-(43). They con-
cluded that catalyst state dynamics leads to a decrease in maximum temper-
ature which formally contradicts the Kiselev analysis [114]. The possible
reason for such a catalyst dynamics influence can be explained by the periodic
recovery of active sites (*) in a reverse-flow reactor.
Attempts have been made to simulate NO, selective reduction while
 REVERSE-FLOW OPERATION 35

taking into account the high ammonia adsorption capacity of vanadium cat-
alysts. The amount of ammonia adsorbed can be about 0.29% by weight for
a typical vanadium pentoxide catalyst. This feature is the basis of the idea of
isothermal reverse-flow operation proposed by Agar and Ruppel [58]. These
authors described a simple dynamic model of the reaction including two
stages, as follows:
NH, + * -z= NH,* (44)
NH,* + NO + 1/40, * N, + 3/20, + * (45)
Kinetic parameters were roughly determined from adsorption break-
through experiments. The isothermal reactor model took into account un-
steady-state behavior of the adsorbed ammonia, surface and gaseous diffusion
inside spherical catalyst pellets, intraparticle mass transfer, and convection.
Downloaded by [171.67.34.205] at 22:03 14 March 2013

Calculated two-dimensional concentration profiles illustrated the ability of the

proposed process to trap adsorbed ammonia in the center of the catalyst bed.
Noskov et al. [35] used the reaction system (44)-(45) for simulating the
performance of a nonisothermal reverse flow reactor for NO, reduction. It
was shown that for central ammonia injection [Fig. 4(f)], a certain amount
of adsorbed ammonia remains continually in the central part of the bed. This
allows the reaction to occur in that half of the bed when it is not directly fed
by ammonia during the semicycle. The authors [35] pointed out that the
ammonia adsorption leads to a maximum temperature decrease and conver-
sion rate increase compared to steady-state catalyst performance. This ex-
plains the observed discrepancies between parameters for industrial plant op-
eration versus results calculated on the basis of steady-state kinetics [35].
Detailed analysis of catalyst state dynamics has been conducted recently
[143, 1721 on an example of SO, oxidation over vanadium catalyst in a
reverse-flow reactor. The dynamic kinetic model of SO, oxidation developed
by Balzhinimaev with coworkers [144, 1451 includes both the main catalytic
reaction steps and side processes involving changes in the catalyst state. This
model was also used [147] to explain the results of periodic concentration
forcing experiments by Briggs et al. [146].
At reaction conditions (temperature over about 400°C) the active com-
ponent of the vanadium catalyst is in a melt phase distributed inside the pores
of a silica carrier. It was found [148] that the value of Henry’s constant for
SO, is about 10’ times larger than those of SO, and O,, and that the melt has
an extremely high absorption capacity for sulfur trioxide. This defines a rather
long characteristic time, about 10-200 min, for the process of melt saturation
by sulfur trioxide. Estimation of the characteristic time for binuclear vana-
dium complex transformations in the bed [143, 1721 gave a value of several
minutes. Both these times scales are comparable to the cycle and substantially
higher (by a factor of 10z-103) than the times of SO, and 0, accumulation.
Thus, for reactor modeling, only the inertia of the liquid phase concen-
tration of SO, and of the reaction intermediates needs to be accounted for
 36 MATROS AND BUNIMOVICH
Downloaded by [171.67.34.205] at 22:03 14 March 2013

*
S O O ~ , . , 8 ,. 8~. I
4'0
Cycle period. rnin
I I I I I I I I
$0
I ~
110~
~ ~
~ ' - -
180
'
~ ~
~ ' ~
* ~
~

FIG. 12. Effect of cycle duration and inlet SO, concentration on average
conversion (a) and maximum temperature (b): 1-3, prediction with unsteady-state
kinetic model; 1'-3', prediction with stationary kinetic models. Inlet SO, concentra-
tion = 9% (1, l'), 6% (2, 2'), and 3% (3, 3'). (Adopted from Bunimovich et al. [143,
1721.)

while the remaining concentrations may be considered to be at quasi-steady

state.
In order to analyze the effects of various factors exerting influence on
dynamic properties of the catalyst, simulation results obtained under various
assumptions on the reaction kinetics model have been compared. Special
attention was paid to signs and values of deviations Ax = xy -
xi and AT,,,,, = Tkax- TLaxcorresponding to possible inaccuracy in the pro-
cess simulation with the steady-state kinetic model giving the average con-
version x: and maximum temperature T",,,.
As an example, the effects of cycle duration and input sulfur dioxide
concentration [143, 1721 are demonstrated in Fig. 12. Both dynamic and
steady-state kinetic models predict similar behavior. As a result of catalyst
state dynamics, the maximum temperature in the bed decreases [compare
solid and dashed lines on Fig. 12(b)]. The reaction mechanism
 REVERSE-FLOW OPERATION 37

steps and the accumulation of SO, in the melt decrease the degree of catalyst
reduction in the inlet part of bed and increase the reaction rate in the front
heated zone. This, coupled with the heat released due to SO, absorption in
the opposite part of the catalyst bed, creates favorable conditions for the
reversible exothermic reaction. Both effects reduce temperature gradients as
well as the maximum temperature. On the other hand, dynamic processes in
the catalyst melt increase the degree of catalyst reduction in the outlet part
of the bed where the temperature decreases. Reoxidation of the catalyst causes
a sulfur dioxide release and lower average conversion. This negative effect
of catalyst dynamics was shown to be more pronounced at SO, concentrations
lower than 6%. At higher SO, concentrations and longer cycles, catalyst dy-
namics lead to higher conversion.
One possible way to increase conversion when processing feeds with
Downloaded by [171.67.34.205] at 22:03 14 March 2013

low SO, concentration is to replace the boundary areas of the catalyst bed
with inert packing. This produces higher catalyst temperatures and eliminates
low-temperature sections of the bed where sulfur dioxide release would be
possible.

IV. EXPERIMENTS

In Table 1 a summary of known experimental investigations is presented.

Six different catalytic processes have been studied using small bench-scale
laboratory units (examples 4, 15, and 17 in Table 1) and larger laboratory
and pilot reactors (examples 2, 3, and 16).
The purpose of most experiments was to test the principal RFO features
such as recuperation of the reaction heat in the catalyst bed, autothermal
processing of low-temperature gases and fuels, and improvement in conver-
sion for reversible reactions. These features were initially demonstrated in
the experiments made in Boreskov Institute of Catalysis, Russia, for SO,
oxidation over vanadium catalyst, VOC oxidation over metal oxide catalyst,
and NO, reduction on vanadia-titania catalyst (examples 1, 5, 10 and 12 in
Table 1). The experiments were carried out in a simple laboratory stand with
a diameter of 0.175 m. Minimum adiabatic temperature rise obtained with
this type of installation was about 40°C. Experimentally observed influences
on reactor behavior of operating parameters such as flow rate, feed compo-
sition and temperature, and duration of cycle, were in a good qualitative
agreement with theoretical predictions. Commercial catalysts were proven to
be stable enough under conditions of oscillating temperature and gas com-
position. These principal results have been confirmed later [29, 149, 169,
1731 in experimental studies of reverse-flow operation for VOC oxidation.
In recent VOC oxidation experiments by Nieken et al. [173] and van de
Beld et al. [169], the oxidation of the different volatile organic compounds
and their mixtures has been tested. Examples of complex reactor behavior
for simultaneous propane and propylene oxidation were discussed 11731. Van
 38 MATROS AND BUNIMOVICH

TABLE 1

Characteristics of
Example Reactor Catalyst
Process no. Reference Catalyst diameter, m volume Composition Pressure
~~~ ~

SO? 1 Boreskov et Commercial 0.175 10-30 L 1.7-9% SO2 Atmospheric

oxidation to al. (1982) vanadium in air mixture
so, ~ 7 1 catalyst,
rings 8 mm

2 Boreskov et Vanadium 1.6; 2 5.3-14.7 0.7-9% SO,; Atmospheric

o,,
Downloaded by [171.67.34.205] at 22:03 14 March 2013

al. (1982) catalyst, 8 and m’ X-IS%

~401 mm grains; 2.8 the rest N2
Matros 8 and 25
(1989) (201 mm rings
Bunimovich
et al., (1900
11231

3 Isozaki et al. Commercial 1.9 6; 7.1 0.7-1.1% Atmospheric

(1990) [151] vanadium m’ SOJ5-13% 0,
catalyst and 5-9%
SO,l8-14% 0 2

4 Levina et al. 4 types of 5 mm 1.74 g 656.8% Atmospheric

(1990), [152] vanadium so,;
catalyst 12% 0,

VOC S Boreskov et Industrial 0.175 12-50 L 0.0S-0.14% Atmospheric

oxidation al (1984) copper butane and
1281 chromate propane in air
Matros catalyst,
(1989) [20], grains 6 mm
Matros et al. and rings 13
(1993) 1301 mm
 REVERSE-FLOW OPERATION 39

Experimental Studies of Reverse-Flow Operation in Catalytic Reactors

Characteristics
inlet gas Space of process
Linear time in Ratio: Cycle Maximum efficiency,
Temperature, Flow rate, velocity, catalyst inert/ duration, temperature, results of
"C m3/h AT,,, "C misec bed, sec catalyst min "C experiments

20 -200 20-50 49-260 0.04-0.6 3.2-10 - 1.5-240 470-600 90-95% SO,

conversion;
demonstration
of process
feasibility. In-
fluence of op-
erational pa-
rameters,
AT,,, t,, U.
40-50 1500- 20- 260 0.03-0.4 6.5-20 - 20-120 450-640 SO, conver-
Downloaded by [171.67.34.205] at 22:03 14 March 2013

3000 sion was 95-

97% at inlet
so,
concentra-
tion 0.7-296.
Develop-
ment of the
commercial
process.
40-50 2500- 20- 180 0.3-0.35 6-7 - 30 580-620 SO, conver-
3600 and 450- sion was 95-
470 97% at inlet
SO,
concentra-
tion 0.7-2%.
Develop-
ment of the
commercial
process.
- 1.5 mW - - 1.7 - 20-80 - Study of va-
see nadium cata-
lysts behavior
under reverse-
flow
operation.
Ambient 25-65 38-112 0.42-0.75 0.7-6.4 0-4.7 15-90 340-600 I Complete
VOC oxida-
tion, demon-
stration of
process effi-
ciency at low
and oscillated
VOC con-
centrations.
Parametric
study influ-
ence of AT,,,
t,, u, inert
sections.

(continurd)
 40 MATROS AND BUNIMOVICH

TABLE 1

Characteristics of
Example Reactor Catalyst
Process no. Reference Catalyst diameter, m volume Composition Pressure
voc 6 Eigenberger Pd 0.05 0.7 L Propane and Atmospheric
oxidation and Nieken honeycomb methane in
(1988) [29], catalyst with air
(1994) (531 width of
channel 1
mm

voc 7 Nieken, PI on 0.05 1.6 L Propane and Atmospheric

oxidation Kolios, and alumina propylene
Eigenberger honeycomb mixtures in air
Downloaded by [171.67.34.205] at 22:03 14 March 2013

(1994) [173] catalyst with

square
channels of 1
mm and 2
mm

voc 8 van de Beld Granular 0.145 10.7-16.5 L Ethene, 1-5

oxidation et al. (1994) catalyst, propane, and
(1691; O.OS% P d k - methanol in air
40,

van de Beld
and
Westerterp
(1993) [149]

voc 9 Sapund- Copper 0.175 60 L O.OS-O.lS%, Amhient

oxidation zhiev el al., chromate butane in air
(1988) [139] catalyst, 5-
mm grains

Methane 10 Matros et &Copper 0.17s 0.13 m’ 0.45-1.2% Ambient

oxidation (1988) [87], chromate methane in
(1989) 1201 catalyst, air
rings 15 and
25 mm

11 Chaouki et 0.2% Pdl 0.045 0.7 L 1.38 and Ambient

al. (1994) alumina 1.8% methane
~701 catalyst in air
 REVERSE-FLOW OPERATION 41

Continued
Characteristics
inlet gas Space of process
Linear time in Ratio: Cycle Maximum efficiency,
Temperature, Flow rate, velocity, catalyst inert/ duration, temperature, results of
“C m’/h AT,,, “C m/sec bed, sec catalyst min “C experiments
Ambient 2.4 1-2 290 - 800 Test of hon-
eycomb inert
packing and
catalyst, par-
tial hot gas
withdrawal
for heat
removal.
Ambient 6 30-160 0.85-1.7 0.94-0.47 1-1.7 1 170-60 Study of the
influence of
operational
Downloaded by [171.67.34.205] at 22:03 14 March 2013

parameters,
experi-
mental show-
ing of
multiple
steady state
for complex
voc
mixtures
Ambient 6-36 m’i 0-110 0.1-0.6 1.1-6.5 0.54 0.8-35 370-630 Extended
h study of influ-
ence of oper-
ating
parameters.
Comparison
with one-
dimensional
model includ-
ing overall ra-
dial heat
transfer
coefficient.
Ambient 15-50 45-160 0.17-0.58 4.3-14 0 40-200 400-500 Measuring the
radial temper-
ature profiles,
comparison
with two-di-
mensional
mathematical
model.
Ambient 60 122-300 0.4-0.6 9-14 0 60 80 - 1000 Lean meth-
andair mix-
ture combus-
tion; about
50% of reac-
tion heat is
utilized.
Ambient 1.4-2.8 370 and 0.28-0.5 0.8-1.6 0.49 130- 600-660 Comparison
480 200 min with a two-di-
mensional
model for
methane oxi-
dation
reaction.

(continued)
 42 MATROS AND BUNIMOVICH

TABLE 1

Characteristics of
Example Reactor Catalyst
Process no. Reference Catalyst diameter, m volume Composition Pressure
NO, 12 Bobrova et Industrial 0.175 8.5-15 L 3.7-10.5 Ambient
reduction a1 , (1988) vanadium g/m’ NO,
~341 catalyqt and 2-6 pim’
NI-I,
in air

13 Noskov et Honeycomb Square 11.2 L 2.5-8 g/m’ Ambient

al., (1993) catalyst with 0.15 NO, and
WI ch a n n eI stoich. NH, in
Downloaded by [171.67.34.205] at 22:03 14 March 2013

width 4 mm air

Methanol 14 Matros, Industrial 0.068 2.7 L 3.9% CO+CO, 5; 8 MPa

synthesis (1989) [20] copper-zinc in hydrogen
catalyst mixture

15 Neophy- Copper-zinc - 30.6 g 3 5 7 % CO 5 MPa

tides and calalyst, and CO, in
Froment, grain size 1 h ydrogcn
(1992) mm mixture
~411

Partial 16 Blanks et al., Methane 0.57 127 L, 26% CH, in 0.2 MPa
oxidation of (1990) [52] steam air mixture
methane reforming
catalyst: Nii
A120,

CO and C,H, 17 Punvano et 0.1% Pt on 0.01 35 cm’ 4-6% C,H, in 0.1 MPa
oxidation al., (1994) y-A1,03 air
[I661

de Beld et al. [169] indicated strong influence of methanol adsorption on

overall reactor efficiency during the oxidation of this component.
Experimental data were obtained [20, 87, 1501 for heat recovery after
catalytic combustion of poorly oxidizing lean methane/air mixtures. A labo-
ratory unit with intermediate cooler [see Fig. 4(b)] provides for recovery of
about 50% of reaction heat. Another scheme for using heat by interstage gas
 REVERSE-FLOW OPERATION 43

Continued
Characteristics
inlet gas Space of process
Linear time in Ratio: Cycle Maximum efficiency,
Temperature, Flow rate, velocity, catalyst inert/ duration, temperature, results of
"C m3/h AT,,, "C m/sec bed, sec catalyst min "C experiments
30-120 27-60 30-95 0.3-0.63 0.5-1.7 3.6- 15-50 340-430 97-99% NO,
4.5 removed,
feasibility of
reverse-flow
operation for
NO,
reduction.
30-40 35-50 30 0.5-0.7 0.7-1 2 5-14 240-350 Use of honey-
comb catalyst
and inert
Downloaded by [171.67.34.205] at 22:03 14 March 2013

packing for
reverse-flow
NO,
reduction.
80-120 17.5- - 1.3-2.7 0.25-0.53 0 6-8 270-290 Yield of
35.3 methanol
3.3%; catalyst
productivity
6-15 kg/
(1 day)
120-140 0.2-0.4 - - 0.35-0.6 0 1.5-16 240-280 Detailed mea-
surement of
outlet gas
composition,
catalyst dy-
namics effects
on reactor
behavior.
-70-120 -225 - 0.25 1.2 4 120 -900 Feasibility of
RFO for hy-
drogen gener-
ation, com-
bining endo-,
and exother-
mic reactions.
Ambient - 2200-3400 - - -2 15-30 -700 Run-away
phenomena in
a water-
cooled tubular
reactor

withdrawal [Fig. 4(c)] was tested [29, 53, 1731. Recently, Nieken with coau-
thors [165, 1731 experimentally verified different ways to control maximum
temperature in a reverseiflow reactor. They showed that withdrawing hot gas
[Fig. 4(c)], combined with an increase in axial heat conductivity in the central
zone of the catalyst bed, prevented catalyst overheating at high VOC con-
centration and maintained high temperature at low VOC concentrations. An-
 44 MATROS AND BUNIMOVICH

other recent work by Purwono et al. [166] (example 17, Table 1) gave the
results of experiments with an externally cooled catalyst tube. These exper-
iments were performed at a very high adiabatic temperature rise, and runaway
phenomena in a reverse flow reactor were studied.
Laboratory tests of a reverse-flow methanol synthesis (example 14, Table
1) briefly described in Matros’ book [20] demonstrated high methanol yield
at low inlet gas temperature.
Eigenberger with coworkers [29, 53, 165, 1731 used a catalyst and inert
material made of commercial ceramic honeycomb monoliths structures. Later,
Noskov et al. [35] carried out experimental tests of honeycomb catalyst and
ceramic packing for NO, reduction. This type of packing was found to be
very promising as it provides for laminar flow at high linear velocity, which
results in a more favorable ratio of heat transfer to pressure drop than for
Downloaded by [171.67.34.205] at 22:03 14 March 2013

random packing 1531.

Laboratory reactors used in the most of the experiments (Table 1) with
diameters 0.05-0.18 m are nonadiabatic, and the experimental results may
not be scaled up to industrial plants having diameter larger than 1 m. To
describe the temperature profiles in those nonadiabatic experiments, van de
Beld et al. [149] used a one-dimensional model which includes in the energy
balance an overall wall heat transfer coefficient U,. This coefficient can be
specified from special experiments made in the absence of a reaction [149,
169, 1701 or simply by adjusting of experimental and calculated temperature
profiles. Apparently, the values of U, depend on the reactor radius because
of the radial temperature gradients present. The experimental radial temper-
ature profiles in the laboratory reactor with a diameter of 0.175 m were
discussed [139]. These profiles were fitted on the basis of a two-dimensional
model. The same model has been used recently by Chaouki et al. [170] for
describing their experiments with methane oxidation in a reactor with a di-
ameter of 4.5 cm.
Blanks et al. [52] carried out the only known experimental investigation
(example 14, Table 1) of a RFO for combined endo- and exothermic reac-
tions. Figure 13 presents the temperature profiles obtained in this work. The
temperature peak relates to the fast exothermic reaction of methane combus-
tion, followed by a slow temperature decrease due to the endothermic reac-
tions of water or/and CO, methane reforming. The optimal process strategy
consists in spatial localization of endo- and exothermic reaction zones in the
same part of reactor volume, namely, at the inlet edge of the catalyst bed.
Otherwise, as has been noted [52],oxygen can be consumed in the inert
packing before the catalyst bed, which causes cooling of the catalytic zone
and allows carbon deposition inside the catalyst pellets. This results in me-
chanical destruction of the catalyst. To prevent this problem, about 10% of
the natural gas feed was burned before the reactor. This moderates heat gen-
eration inside the reactor and allows better temperature control. The paper by
Blanks et al. does not contain an extensive analysis of a process performance.
The calculated temperature profile (line 2 in Fig. 13) was obtained on the
 REVERSE-FLOW OPERATION 45
Downloaded by [171.67.34.205] at 22:03 14 March 2013

FIG. 13. Temperature profiles for partial oxidation of natural gas (adopted
from Blanks et al. [52]).

basis of a one-dimensional heterogeneous model, Eqs. (11)-(14), and on

reaction kinetics which include three main steps: methane combustion, re-
forming, and water-gas shift reaction. Twelve model parameters including
rate constants were adjusted to fit the pilot plant experiments.
SO, oxidation experiments on large-scale pilot plants [140, 20, 123, 1511
(examples 2 and 3, Table 1) were carried out over a wide range of gas
compositions and linear velocities, and with different catalysts. In Table 2,
the main performance parameters for the pilot reactors are given along with
the calculated parameters determined using the system of Eqs. (11)-(14) and
a kinetic equation [89]. Experimental data for examples 1-3 (Table 2) were
taken from Ref. 151, examples 8 and 9 from Ref. 122, and other examples
from Refs. 20 and 140. Calculated parameters in Table 2 were given else-
where [123, 143, 1721.
In agreement with theoretical predictions, maximum temperature de-
creases with increasing catalyst pellet size (examples 10 and l l , Table 2) and
decreasing linear velocity (examples 8 and 9, Table 2). A n increase in sulfur
dioxide inlet concentration causes a steep rise in maximum temperature and
a decrease in conversion. In general, the one-dimensional heterogeneous
model with quasi-steady-state reaction kinetics is capable of predicting cor-
rectly the main parameters for both pilot and industrial reactors. This conclu-
sion was confirmed recently by Gosiewski [168], who compared his own
calculations with the experimental data [20, 1401.
However, as can also be seen from Table 2, in most cases the calculated
maximum temperature and conversion are higher than those obtained exper-
imentally. Especially large discrepancies are found at low sulfur dioxide con-
centration. Examples 1-2 and 4-5 (Table 2) show that at SO, concentrations
 Downloaded by [171.67.34.205] at 22:03 14 March 2013

TABLE 2
Main Parameters of SO, Oxidation Pilot Plant Reactors and Simulation Results
Average conversion,
(xa>and
maximum temperature
Average inlet (TIll2X)

concentration, Experiment Simulation

% Gas flow Catalystlinert Catalyst
Example Reactor linear packing bed size, Cycle T,,,, T,,X,

no. SO, 0, diameter, m velocity, mls height, m mm duration, min xa, % "C x,, % "C
1 0.71 11 1.9 0.35 2.3f0.65 11 20 94.5 463 97.9 470
2 0.9 7 34 95.5 475 96.7 488
3 1.1 5 36 95.0 500 94.8 510
4 1.o 10 1.6 0.36 2.710.45 9 20 95.5 450 97.7 494
5 1.3 0.20 20 95.5 460 97.0 493
6 1.7 10 2.0 0.23 2.010.3 9 30 95.0 505 95.6 527
7 2.3 11 30 92.5 545 92.3 560
8 2.3 15 7.0 0.30 3.010.3 25 26 95.5 540 95.1 541
9 2.5 15 0.15 38 96.2 530 96.0 525
10 3.2 12 2.0 0.25 2.0f0.3 9 25 91.5 590 90.0 596
11 3.2 12 0.20 4.6f0.3 25 50 94.0 520 92.5 560
12 4.2 12 3.6 0.13 1.710.2 35 91.0 580 90.0 588
13 8.4 13 0.06 120 85.0 610 86.2 612
Note. Adapted from Refs. 143 and 172.
 REVERSE-FLOW OPERATION 47

of 0.7-1.3%, the observed conversion is 1.2-3% lower than the calculated

conversion. Deviation from this trend in example 3 (Table 2) may be attrib-
uted to significant oscillations in gas composition typical of feed streams
originating from metallurgical plants.
The Bunimovich et al. papers [143, 1721 concluded that the influence
of catalyst state dynamics on reversed-flow reactor performance explains the
observed differences between measurements and simulation, which are es-
pecially large at low sulfur dioxide concentrations. This is demonstrated by
the results shown in Fig. 12.
The direct indication of substantial unsteady-state catalyst behavior was
obtained in pilot plant tests [140] (example 2, Table 1). Figure 14 shows the
changes of concentrations of sulfur oxides at the reactor outlet,
C g , and C g 3 , during a semicycle for the given inlet sulfur dioxide concen-
Downloaded by [171.67.34.205] at 22:03 14 March 2013

tration, C&. The time interval shown in Fig. 14 is equal to the period be-
tween flow reversals, tciz.Consequently, the pattern shown in Fig. 14 is re-
peated each tciz.It is seen from the Fig. 14 that initially sulfur trioxide is
almost absent at the reactor outlet, but that by the middle of the half-cycle
its concentration becomes higher than the inlet SO, concentration. The ma-
terial balance on sulfur is satisfied only on a cycle-averaged basis. Obviously,
sulfur trioxide content in the catalyst bed changes rather slowly, i.e., the
catalyst state cannot rapidly follow the quickly changing state of the reaction
gas.
Similar observations were made [ 1521 where a significant (up to 2 min)
delay in SO, outlet was observed for a space time 1.7 s. The experimental
setup used included three successive isothermal plug-flow reactors at 480°,
420°, and 380°C, and a preconverter yielding about 90%SO, conversion. Gas

Time, min.

FIG. 14. Time dependencies of outlet SO, and SO, concentration during a
half-cycle (pilot plant data adopted from Boreskov et al. [140]).
 48 MATROS AND BUNIMOVICH

flow through the reactors was reversed cyclically. For a first semicycle the
flow after the preconverter fed initially into the high temperature reactor. For
another semicycle, nonconverted SO, gas passed through the low-, then mod-
erate-, and then high-temperature reactor. Thus, the experimental model of
reverse-process performance in the ‘ ‘falling temperature’’ zone was created.
Average conversion obtained during a preconverted mixture feeding was
lower compared with that for steady-state operation.
The investigation (example 13 in Table 1) by Neophytides and Froment
[ 1411 concerned primarily experimental assessment of the dynamics of Cu/
ZnO catalyst in a reverse-flow reactor for methanol synthesis. Using an on-
line quadrupole mass spectrometer, these authors obtained a number of ac-
curate measurements of effluent composition at various cycle duration and
inlet reactor temperatures. Figure 15 [ 1411 shows the methanol production
Downloaded by [171.67.34.205] at 22:03 14 March 2013

rate at the reactor exit and the rate of carbon oxide consumption during a
semicycle. Only at the end of a semicycle is carbon consumption from COX
equal to its transformation into methanol. For about 90 sec after flow reversal
less carbon is produced than consumed, but later the methanol production
rate exceeds that of COXconsumption. Even more complicated time depen-
dency was observed for CO, consumption and water production rates, which
are equal to each other under steady-state conditions but differ in reverse-
flow operation. The authors explained these dependencies by assuming the
formation of a relatively stable surface formate from CO, and H,, and then
its decomposition at high temperature. Also, a rather high catalyst capacity
with respect to water should exist at low temperature.

V. INDUSTRIAL APPLICATIONS, RESULTS, AND FUTURE

PROSPECTS

Three commercially used processes are known where periodic flow re-
versal in a catalytic packed bed reactor is employed. These are oxidation of
volatile organic compounds (VOC) for purification of industrial exhaust
gases, SO2 oxidation for sulfuric acid production, and NO, reduction by am-
monia in an industrial exhaust gas. A short review of the results of these
applications is given below. Also, the prospects of future use of RFO for
other catalytic processes are discussed.

A . Catalytic Incineration of VOC

Today, catalytic incineration of VOC is the broadest area of application

of reverse-flow reactors. One of the most commonly used techniques for VOC
abatement is homogeneous oxidation in regenerative thermal oxidizers (RTO).
This process provides for relatively high heat recovery efficiency. However,
the high-temperature combustion, up to 1000°C, is accompanied by nitrogen
 REVERSE-FLOW OPERATION 49

0 7m 200 300 400 500

Tine (s)

FIG. 15. Experimental measurements of consumption rate of CO, + CO

Downloaded by [171.67.34.205] at 22:03 14 March 2013

(. . .) and production rate of methanol (-) for a reverse-flow methanol synthesis

reactor (adopted from Neophytides and Froment [1411).

oxides formation and in many cases does not appear to be optimum with
respect to energy consumption.
Catalytic oxidation occurs at a much lower temperature than homoge-
neous combustion, down to about 200"C, and provides for almost complete
VOC removal without NO, formation. However, the conventional catalytic
incineration plant that includes a shell-and-tube heat exchanger operates au-
tothermally, i.e., without external energy consumption, only if the total adi-
abatic temperature rise from VOC oxidation exceeds 50"-70"C. Decreasing
the AT,, requires an increase in heat exchange surface that in turn requires
bigger heat exchangers with additional external heat losses. These losses can
become large enough to cause loss of autothermal operation. The conven-
tional catalytic incineration process is difficult to use for treatment of gases
with varying flow rates and concentrations, even if automatic control is
provided.
Using reverse-flow operation in a reactor loaded with catalyst and re-
fractory ceramic beds [45, 63-65, 153, 1541 is a reasonable way of avoiding
drawbacks of traditional VOC abatement processes. In such a reactor, VOC
oxidation proceeds at relatively low temperatures and no fuel is required for
self-sustaining operation with AT,, as low as 20"-30°C.
Mathematical simulation allows selection of optimal process parameters
such as linear velocity of gas flow, inert packing and catalyst bed lengths,
dimensions of particles, and switching frequency [20, 29, 30, 131, 169, 171,
1731. Honeycomb catalyst and inert structure [53, 1731 can be useful.
Nowadays, dozens of VOC catalytic oxidation units employing the RFO
concept are in operation in Western Europe, mainly designed and built by the
Danish company Haldor Topsoe; and in Russia, engineered by Boreskov In-
stitute of Catalysis. Expansion of this technology to the USA is being un-
 50 MATROS AND BUNIMOMCH

dertaken by Monsanto Enviro-Chem. System, Inc. [ 159, 1741. Operation of

Russian plants has been discussed [20, 131, 157, 1581. VOC removed in these
plants includes C,-C, hydrocarbons, esters, ketones, aromatics, alcohols, var-
ious solvents, etc. Flow rate range is from 500-600 m3/h to 16,000-20,000
m3/h. These installations are used for purification of both process gases such
as gases from plasticizer production characterized by a relatively high con-
centration of pollutants, and vent gases from paintinddrying sites having very
low VOC concentrations [157]. Most of the plants treat gas flows with fluc-
tuating flow rates and VOC loadings. For example, the unit installed at a
chemical plant in Uzbekistan [158] treats a gas stream containing HCN, ac-
rylonytrile, acetaldehyde, and many other components with the gas flow rate
changing from 565 to 4520 m3/hr-i.e., by a factor of 8-and VOC concen-
tration from 1.3 to 13.6 g/m3-i.e., by a factor of 10. Under these conditions,
Downloaded by [171.67.34.205] at 22:03 14 March 2013

operation of a traditional catalytic incinerator becomes unstable while a RFO

unit shows stable self-sustaining operation.
In Table 3, comparative data on different methods for VOC incineration
[159] are presented. The evaluation was done for typical plants purifying
contaminated air with 0.1-0.5 g/m' of VOC at flow rates ranging from 500
to 50,000 m3/h. Catalytic reverse-flow reactors reduce operating costs by a
factor of 5 in comparison with a conventional recuperative catalytic unit and
provide for 1.6 times reduced operating costs as compared to the relatively
efficient regenerative thermal oxidizer.

B. Sulfur Dioxide Oxidation

Sulfur dioxide oxidation is one of the oldest catalytic technologies. The

optimum process scheme and mode of operation generally depend on the
source of the SO,-containing gases. Lean gas (0.5-3.5% SO,) is typically
encountered in nonferrous metallurgical smelting operations. After stages of
dry and wet cleaning, the exhaust gas temperature is about 4O0-6O0C while

TABLE 3
Comparison Between Various VOC Control Systems
Method Capital costs Operating costs
1. Catalytic, recuperative 1 1
2. Thermal, recuperative 1.8 2.0
3. Thermal, regenerative 1.6 0.5
4. Catalytic regenerative (catalytic
reverse process) 1.o 0.2
Note. Estimates from Ref. 159.
 REVERSE-FLOW OPERATION 51

the minimum reaction temperature is 370"-4OO0C. Utilization of these gases

by traditional methods of sulfuric acid production relies on a multibed cata-
lytic reactor with preheating and interstage cooling. This method is considered
uneconomical because of significant fuel consumption required for processing
the gas with an adiabatic temperature rise of 15"-100°C. Quite often in de-
veloping countries (and even in Canada) such gases are released into the
atmosphere untreated. Normal gas (3.5-11% SO,) is obtained by sulfur burn-
ing, oxidation of hydrogen sulfide, roasting of pyrite ores, or from the recov-
ery of spent sulfuric acid. This gas is traditionally used as the feedstock for
the classical sulfuric acid process consisting of multiple catalytic beds with
interstage cooling. High conversion, above 99.7%, which is both desired and
most likely required in order to eliminate SO, pollution, is achieved using
the double contact-double absorption (DC-DA) systems, which involve SO,
Downloaded by [171.67.34.205] at 22:03 14 March 2013

absorption between two stages of catalytic oxidation. Since the absorption

process occurs at relatively low temperature, these systems also require heat
exchangers for intermediate gas cooling and heating. The large number of
heat exchangers and interconnecting pipes make the units complex and cum-
bersome; they also have high pressure drop.
Mathematical simulation studies and pilot plant tests [20, 31-33, 1231
showed that the reverse-flow operation permits a substantially simplified re-
actor scheme, allows decreased metal weight, and generates lower pressure
drop with a catalyst charge and conversion efficiency close to the conven-
tional steady-state technology.
The unsteady-state process of SO, oxidation patented by Boreskov In-
stitute of Catalysis [38, 461 is widely used today throughout the former Soviet
Union, mostly for processing of low-concentration gases in nonferrous met-
allurgy. Now several industrial plants are processing gases with flow rates
from 30,000-40,000 m3/h up to 100,000 m'h, and inlet SO, concentration
varying from 1.25% to 3-4.5% [20, 32, 33, 1231. Industrial reactors are in
operation also in Bulgaria and China. Different reactor configurations are
used. The simplest flow diagram [Fig. 1, 4(a)] is used for treating gases with
SO, concentration below 2-3%. At higher concentration of SO2, a reactor
with one or two intermediate heat exchangers to remove excessive heat is
recommended. For SO, concentrations above 5-7%, the method presented in
Fig. 4(h) is used. The additional catalyst bed with unidirectional flow serves
to smooth oscillations in temperature and gas composition after the primary
reverse-flow reactor and to increase conversion from 92-93% to 97-98%.
Economic evaluation [32] of the use of a RFO in sulfuric acid production
showed a decrease in metal needed by a factor of 1.5-5, 30-50% reduction
in pressure drop, 20-80% lower capital costs, and 5-20% lower operating
costs.
In Refs. 151 and 160 the results of pilot plant tests for double contact-
double absorption systems with a RFO in either of the conversion stages are
presented. The outlet concentration of SO, after the DC-DA system was
about 400 ppm. This is in line with values from typical commercial plants
 52 MATROS AND BUNIMOVICH

used in nonferrous metallury. The benefits expected include 30% lower op-
erating costs and 15% lower equipment costs [151].

C. Selective Reduction of NO,

During the design of an industrial process special measures are taken to

prevent formation of ammonia salts at the low temperatures in the catalyst
bed and inert packing [35]. The decomposition of ammonium nitratehitrite
deposits can proceed explosively, causing strong local overheating inside the
reactor. As a method for preventing this phenomenon, injection of ammonia
into the high-temperature zone in the reactor center is used [flow diagrams
in Figs. 4 ( 9 and 4(g)].
One industrial unit is operating in Russia. It purifies about 11,200 m3/h
Downloaded by [171.67.34.205] at 22:03 14 March 2013

of off-gases containing 2-14 g/m3 NO, remaining after absorption of acids

[34, 351. A two-bed reactor with interstage injection of ammonia water [Fig.
4(f)] is used. Loaded into the reactor are a common commercial V,O,/Al,O,
granular catalyst and porcelain Rashig rings as an inert packing. Inlet gas
temperature is 30”-40°C, reversal frequency is 4-6 h-’, and maximum tem-
perature in the reactor is 300”-400°C. Inlet NO, concentration fluctuates
slowly and requires adjustment of the quantity of injected ammonia water no
more often than every 2 h. Even though the inlet NO, concentration can vary
widely, the outlet concentration does not exceed 30-70 mg/m’. This high
purification efficiency is obtained by keeping optimum temperatures in the
reactor due to automatic control of cycle frequency. Another important reason
appears to be a “smoothing” effect of ammonia adsorption and reaction in
the catalyst [35].
To remove NO, from flue gases with NO, concentration ranging from
0.5 to 2 g/m3, honeycomb inert material and catalyst can be advantageous. It
allows autothermal operation at a minimum NO, concentration that provides
for an adiabatic temperature rise of about 10°C [35]. The optimum channel
size of a honeycomb structure for minimizing the pressure drop across the
reactor was estimated to be about 2-3 mm [35].

D. Sulfir Production by Catalytic Claus Method

The Claus process is the main method of sulfur production from hydro-
gen sulfide containing gases. Traditionally, the Claus process unit consists of
a furnace for homogeneous oxidation of hydrogen sulfide into sulfur and SO2,
and a catalytic stage where hydrogen sulfide is oxidized by SO, producing
elemental sulfur and water. This stage usually includes two or three beds of
alumina catalyst with intermediate sulfur condensation and gas preheating
between the beds. Sulfur recovery in such an unit is not higher than 96% for
two beds and 98% for a three-bed catalytic reactor. In recent years, several
processes for tail gas cleanup following Claus process plants were developed.
 REVERSE-FLOW OPERATION 53

Among them is the SULFREEN process [161] based on the Claus reaction
at low temperature (120"-160"C). In this process the sulfur formed is con-
densed directly in the catalyst bed and is periodically removed by evaporation
with hot air.
The idea and technical evaluation of the catalytic reverse-flow Claus
process were developed by Matros and Zagoruiko with coworkers [39, 133-
1361. A catalytic reactor with one bed can replace two catalytic converters
with intermediate sulfur removal followed by gas preheating. Use of a re-
verse-flow reactor instead of the two-bed stationary Claus process provides
an equal or better degree of sulfur recovery (95-97%) and lower capital cost
for the unit (1.3-1.5 times less), and eliminates the need for energy con-
sumption to preheat the gas fed to the system. Total production cost reduction
was evaluated to be about 20-25%.
Downloaded by [171.67.34.205] at 22:03 14 March 2013

Figure 16 shows the process schematic of the so-called double Claus

reverse-process [ 134, 1621. This configuration includes three catalyst beds
and two intermediate sulfur condensers as well as switching valves for gas
flow reversal in internal and external reactor loops. Initially gas enters the
first reactor (3a in Fig. 16) after a waste heat boiler (1) at 300"-400°C and
passes into the second (3a) and third (3B) converters at 120"-160°C. In the
first converter, hydrolysis of COS and CS, proceeds predominantly. After
intermediate sulfur removal the cold gas enters the second converter operating
with periodic flow reversal. After cooling and condensation of sulfur, the gas
is fed into the third converter where sulfur is condensed directly in the packed
catalyst bed as in the SULFREEN process. Periodic regeneration of convert-
ers 3B and 3a is carried out by hot gases from the waste heat boiler. The
optimum reversal period in the central reactor is about 20-200 min, while
reversal in the external loop should be performed approximately once a day.
At comparable environmental efficiency ( ~ 9 9 % )with a system of Claus +
SULFREEN units, the double reverse process provides capital costs saving
as high as 30-40%, lower energy consumption, and a total sulfur production
cost that is lower by 25-30% [134].

E. Heat Recovery of Lean Gases and Fuels

Different industries emit gases containing volatile organic compounds,

methane, or carbon monoxide. Large amounts of methane are liberated from
coal mines, biomass landfills, and various refuses. Potentially, these gases can
be used for energy generation; however, they often have low or fluctuating
concentration of fuel value components that hinders their use in conventional
fuel burners. Papers [40, 129, 130, 1701 discuss the use of a catalytic reactor
with periodic flow reversal as a combustion device for lean gaseous fuel.
Matros [20] cites the results of an evaluation of a methane oxidation reactor.
The conclusion was that it is possible to create an economically attractive
process for using heat from the combustion of methane vented from coal
 54 MATROS AND BUNIMOVICH

mines. A reactor with linear gas flow velocity 0.5-1 m/s could be used, space
time in the catalyst bed would be no more than 1.5 sec, maximum temperature
about 750"-8OO"C, pressure drop about 300 mm of water column, and the
efficiency of heat recovery 60-70%. Inexpensive base metal catalysts could
be used in such a plant.

E Natural Gas Reforming or Partial Oxidation

A conventional tubular reformer for hydrogen or synthesis gas produc-

tion by steam reforming of natural gas represents a complicated and large
undertaking with only a small portion of the unit volume being occupied by
catalyst. Only about a half of the furnace duty is transferred through the tubes
whereas the other half leaves the reformer furnace as hot flue gas. As a result,
Downloaded by [171.67.34.205] at 22:03 14 March 2013

significant quantities of ecologically harmful nitrogen oxides and carbon di-

oxide are emitted into the atmosphere. An expensive heat recovery system is
required because of the high temperature of the processed gas (about 790"-
850°C). It is difficult to produce a synthesis gas with a composition which is
optimum for methanol synthesis, Another process modification, so-called au-
tothermal reforming, includes preliminary combustion of a fraction of the
natural gas followed by methane steam conversion in an adiabatic packed
bed reactor. This allows reduced natural gas consumption in comparison with
a tubular steam reformer; however, it requires rather large amounts of costly
cryogenic oxygen. The outlet gas temperature from the reformer unit is high
(850"-99O"C), and an expensive heat recovery system is also required. Re-
cently, a so-called heat exchange reformer concept was developed (see, e.g.,
Rostrup-Nielsen et al. [163]). Heat for this endothermic reaction is provided
by heat exchange with the product gas coming from an additional adiabatic
packed bed reactor. To enhance heat generation, oxygen is mixed with natural
gas in the adiabatic reactor. The outlet gas temperature is low so a rather high
degree of heat recovery is obtained.
RFO units for methane conversion [41] or partial oxidation [52] will
probably compete in the future with the above-mentioned commercial pro-

FIG. 16. "Double" Claw reverse-process flow diagram (adopted from Ma-
tros and Zagoruiko [134]): 1, Claw furnace; 2, sulfur condenser; 3, catalytic con-
verter; 4, switching valve; 5, acid gas; 6, air; 7, sulfur; 8, tail gas to incineration.
 REVERSE-FLOW OPERATION 55

cesses of hydrogen and synthesis gas production. RFO allows use of heat
from the exothermic methane oxidation and water-gas shift reactions for the
endothermic methane conversion reactions. It provides a very high level of
heat utilization. The difference between the temperatures of the initial feed
mixture and final products is no more than 30"-50°C while the temperatures
inside the reformer reach about 1000"- 1200°C. This makes possible natural
gas consumption per ton of ammonia or methanol production close to theo-
retical minimum values. Estimates for the reverse-flow reformer show that
the mole ratio between hydrogen and carbon oxides at the reformer outlet
can be close to ideal for synthesis gas for methanol production. Flexibility
of plants for ammonia or methanol production can be achieved through avoid-
ing a heat recovery unit. In comparison with the heat exchange reformer,
simplification of reactor construction and decreased oxygen and natural gas
Downloaded by [171.67.34.205] at 22:03 14 March 2013

consumption can be obtained.

G. Synthesis of Methanol and Ammonia

Conventional units for performing these reversible exothermic processes

include a recuperative heat exchanger for initial gas preheating, an adiabatic
reactor with several catalyst beds with addition of cold gas between the beds,
and a synthesis gas cooler or separator. Recirculation of the process gas is
used, and a fraction of the recycle gas is continuously withdrawn for purging
inert gas and hydrogen. Reverse-flow operation can be efficiently applied both
in the primary production loop and for utilization of purge gases from the
recycle loop.
Mathematical simulation of a methanol synthesis reactor [20] showed
that the optimum process configuration for the primary process loop is a
reactor with intermediate heat removal [Fig. 4(b)]. This configuration enables
the recovery of about 50% of the reaction heat. The methanol yield appears
to be even higher than that achieved in a steady-state reactor with four adi-
abatic packed beds and intermediate feed injections.
Purge gas has relatively low CO and CO, concentrations which are total
0.4% at the beginning of a reactor run using fresh catalyst and about 4% at
the end of the run. Steady-state operation without initial gas preheating is
possible only at combined CO and CO, concentration higher than about 2%,
which means uneconomical production of methanol from purge gases by the
traditional process. Reverse-flow operation allows the use of purge gases con-
taining only about 0.4% of CO and CO, without additional energy consump-
tion and can increase methanol production by 5-7%.
Use of RFO in an ammonia synthesis reactor is expected to be efficient
in the primary synthesis loop as well as for purge gases utilization [20, 1641.
Ammonia synthesis at linear velocities around 1m/sec in annular bed reactors
has been recommended [164]. With this design the production of ammonia
in 1 catalyst bed is equivalent to the production of 2-3 beds operating under
 56 MATROS AND BUNIMOVICH

steady-state conditions. Substantial simplification of both the reactor and the

unit process flow sheets can be achieved [20].

H. Catalytic Dehydrogenation

The highly endothermic process of ethylbenzene dehydrogenation to sty-

rene is performed commercially in one- or two-stage adiabatic packed bed
reactors and is characterized by a very high steam-to-hydrocarbon ratio at the
reactor inlet. This ratio is commonly equal to 7-10 and is dictated by the
necessity to maintain reaction temperature at 65O"-70O0C, a level that pro-
vides thermodynamically favorable conditions for high conversion of ethyl-
benzene. Substantial dilution of the inlet gas mixture by superheated steam
leads to a large pressure drop across the reactor unit, requires substantial
Downloaded by [171.67.34.205] at 22:03 14 March 2013

product condenser capacity, results in large-sized plants, and what is even

more important, causes high energy consumption. Calculation by Heggs [47]
showed that a regenerative catalytic reactor system with countercurrent pe-
riodic preheating of the packed bed by superheated steam provided a 6.3%
increase in conversion of ethylbenzene to styrene at higher selectivity than
two identically sized steady-state adiabatic reactors with interstage heating.
Snyder and Subramaniam [137] showed that periodic flow reversal of the
hydrocarbon feed along with steam introduction between axially symmetric
locations allowed a two fold reduction in the steam-to-ethylbenzene feed ra-
tio. This decrease means a corresponding decrease in energy consumption
and a reduction in the cost of the technology as a whole.
Use of countercurrent period feeding of hydrocarbons and air for cata-
lytic dehydrogenation of propene or isobutane [51] achieves a 20% saving
for hydrocarbon feed in a large plant producing isobutylene.

VI. CONCLUDING REMARKS

Reverse-flow operation in a fixed bed catalytic reactor can be considered

as an engineering technique which successfully realizes the general idea of
forced unsteady-state conditions in heterogeneous catalytic reactors. On the
other side, this technique itself generated an entirely new class of industrial
catalytic processes, both exothermic and endothermic. This is governed by
unusual features of reverse-flow operation such as a combination of the re-
action and heat accumulation/exchange in the fixed catalyst bed, and almost
unlimited reaction temperature rise at a low inlet temperature, efficient re-
action performance in a moving zone, an approach toward optimum temper-
ature distribution for reversible reactions, and the possibility of exploiting
dynamic properties of the catalyst.
Many of these essential features are well represented by simple one-
dimensional models assuming quasi-steady-state reaction kinetics. Theory
 REVERSE-FLOW OPERATION 57

based on those models has been well developed and presented in a number
of studies. The simplifications of the process models at low and high reversal
frequencies allow analytical estimates of the main reactor parameters. Com-
bined with numerical simulation, this provides for a basic understanding of
optimum process conditions, multiplicity, and stability. Several recently pub-
lished works contain modeling of more complex systems including several
reactions, catalyst state dynamics, and spatial nonuniformity of the catalyst
bed. It appears that these studies will be continued in the future, especially
in light of the latest trends in literature toward the development of dynamic
kinetic models. Reverse-flow reactors with an additional catalyst function
such as adsorbing and accumulating the reactants should be studied as a
possible technique for NO, abatement and other processes. New models for
systems which combine heterogeneous and homogeneous reactions would be
Downloaded by [171.67.34.205] at 22:03 14 March 2013

of interest for possible RFO applications for high-temperature catalytic com-

bustion and methane conversion. A procedure enabling the analysis of an
“unstable” intermediate regime (Fig. 6) is among the problems requiring
solution within the scope of relatively simple approaches.
Experimental studies primarily using simple laboratory tubular reactors
confirmed the principal results of the theory based on quasi-steady-state ki-
netics and have proved the applicability and potential efficiency of reverse-
flow operation. Some of the experimental data related to the processes of SO,
oxidation, synthesis of methanol, and selective reduction of nitrous oxides,
indicate slow changes of the catalyst state during operation. Future experi-
mental work certainly will be focused on the investigation of new reaction
systems including endothermic reactions. More accurate experiments coupled
with mathematical modeling are needed to cope with the complexity of the
models which are necessary for quantitative description of reactor
performance.
Reverse-flow operation has been applied industrially for incineration of
VOC, SO, oxidation in sulfuric acid production, and NO, reduction by am-
monia. Technical evaluation based on the results of mathematical modeling
have shown potential efficiency of this technique for at least six other indus-
trial processes. Important advantages of the RFO reactor technologies in com-
parison with traditional steady-state processes are:

The possibility of using the fixed catalyst bed as a heat and mass
accumulator, and as a regenerative countercurrent heat exchanger
which allows autothermal operation of exothermic processes at low
and fluctuating inlet concentrations of a reactant.
Creation of conditions thermodynamically favorable for reversible
reactions such as a declining temperature profile for exothermic and
increasing profile for endothermic process.
Intensive energy trapping that allows efficient recovery of exothermic
reaction heat or substantially decreases energy consumption for en-
dothermic processes.
 MATROS AND BUNIMOVICH

The possibility of combining endo- and exothermic reactions in one

catalyst bed, which provides energy economy.
A smaller reactor size, simplification of the unit flow diagram, and
reduction in the average reactor temperature, which provides for a
decrease in pressure drop and heat losses.

NOTATION

specific external surface of particles per packed bed volume, m-l

inlet concentration, mol .m-3, NTP
gas-phase heat capacity, kJ .kg-' - K-'
solid-phase heat capacity, kJ - kg-' - K-'
activation energy, kJ * mol-'
Downloaded by [171.67.34.205] at 22:03 14 March 2013

activation energy of reverse reaction

reaction rate constant, s-l
bed length, m
minimal length of catalyst bed providing high-temperature perfor-
mance in reverse-flow operation, m
axial coordinate, rn
number of half-cycle
heat of adsorption, kJ - mo1-I
heat of desorption, kJ * mol-'
I + u,t; coordinate of moving front, m
gas constant, kJ * mol-' - K-'
radius of catalyst particle, m
temperature, K
maximum temperature, K
maximum temperature in creeping front, K
inlet temperature, K
average outlet temperature, K
maximum temperature in relaxed steady state, K
maximum temperature in relaxed steady state at infinite catalyst
bed length, K
time, s
cycle duration, min
half-cycle duration, min
linear velocity (NTP), m * s-'
absolute value of linear velocity (NTP), m - s - '
creeping front velocity, m . s-l
rate of chemical reaction, mole * m-3 * s-l
rate of chemical reaction referred to inlet concentration of reac-
tant, s-I, for first-order reversible reaction
 REVERSE-FLOW OPERATION 59

X fractional conversion
XS fractional conversion on the surface of catalyst pellet
XI3 conversion in the gas phase
-
X average outlet conversion
x,(T) equilibrium conversion at temperature T, for first-order reaction:
Kp(T) where Kp(T) is equilibrium constant
Xe(T) =
1 + KP(T)
Greek Letters

heat transfer coefficient, kJ * m-2 - K-l- s-l

mass transfer coefficient, me s-l
adiabatic temperature rise, "C
Downloaded by [171.67.34.205] at 22:03 14 March 2013

pressure drop
heat of reaction, kJlmole
porosity of packed bed
fraction of gas phase in packed bed; E , = E + (1 - E)E,
void fraction of catalyst pellet
fraction of solid phase in packed bed; E, = (1 - ~ ) ( - 1 eP)
effectiveness factor
effective heat conductivity of packed bed, kW -
m-'.K-'
heat conductivity of catalyst pellet, kW - m-' * K-'
effective heat conductivity of packed bed solid phase, kW * mP1-
K- '
gas-phase density, kg * rnp3
gas-phase density at normal temperature and pressure, kg m-3
solid-phase density, kg m-3
Llu; space time, s
L,*lu;minimum space time in catalyst bed providing high-
temperature performance, s

Groups of Parameters

0 = UF/U

w = W/C,",
UstLCg,stPg,st
Pest =
&,st
 60 MATROS AND BUNIMOVICH

TEm - Tin- AT,,x!:

Ze =
R(T;~)*

Subscripts and Superscripts

left-side flow
c right-side flow
0 att=O
e equilibrium state
F heat front
g gas phase
1 i-th reaction
inp inlet
Downloaded by [171.67.34.205] at 22:03 14 March 2013

inr inert packing

j j-th gas-phase component
lim limit
m m-th intermediate
P particle
R relaxed steady state
S solid phase
st steam

ACKNOWLEDGMENT

The authors wish to thank Mr. David McCombs, Monsanto Enviro-

Chem Systems, Inc., and Dr. Vadim Strots, Matros Technologies, for useful
discussion and their kind help and assistance in preparation of the manuscript.

REFERENCES

1. Symposium on Cyclic Processing Operations, 151st Meeting of the ACS, Ind.

Eng. Chem., Proc. Des. V e v . , 6, 2-48, 1967.
2. F. J. M. Horn and R. C. Lin, Ind. Eng. Chem., Proc. V e s . Dev., 6, 21 (1967).
3. J. M. Douglas and D. W. T. Rippin, Chem. Eng. Sci., 21, 305 (1966).
4. J. E. Bailey and F. J. M. Horn, Ber. Bunsenges Phys. Chem., 73, 274 (1969).
5. J. E. Bailey and F. J. M. Horn, Ber. Bunsenges Phys. Chem., 74, 611 (1970).
6. J. E. Bailey, Chem. Eng. Commun., 1 , 111 (1973).
7. S. Bittani, G. Fronza, and G. Guardabassi, IEEE Trans. Aut. Control.,AC-18,
33 (1973).
8. G. Guardabassi, A. Locatelli, and S. Rinaldi, Optimization Theory Appl., 14,
1 (1974).
9. N. Watanabe, K. Onogi, and M. Matsubara, Chem. Eng. Sci., 36, 809 (1980).
10. W. H. Ray, Ind. Eng. Chem. Proc. Des. Dev., 7 , 422 (1968).
 REVERSE-FLOW OPERATION 61

11. R. L. Laurence and G. Vasudevan, Ind. Eng. Chem. Proc. Des. Dev., 7, 427
(1968).
12. G. R. Meira, J. Macromol. Sci.-Rev, Macromol. Chem., 20, 207 (1981).
13. G. K. Boreskov and Yu. Sh. Matros, Catal. Rev.--Sci. Eng., 25, 551 (1984).
14. G. H. Denis and R. L. Kabel, Chem. Eng. Sci., 25, 1057 (1970).
15. V. C. Wandrey and A. Renken, Chem. Ztg., 95, 343 (1971).
16. A. Renken, Int. Chem. Eng., 24, 202 (1984).
17. A. Renken, Int. Chem. Eng., 33, 61 (1993).
18. P. L. Silveston, Periodical operation of chemical reactors-a review of the
experimental literature, in Reactions and Reaction Engineering (R. A. Mash-
elkar and R. Kumar, eds.), Indian Academy of Sciences, Bangalore, India,
1987, p. 217.
19. R. Yadav and R. G. Rinker, Chem. Eng. Sci., 44, 2191 (1989).
20. Yu. Sh. Matros, Catalytic Processes under Unsteady State Conditions, Elsev-
Downloaded by [171.67.34.205] at 22:03 14 March 2013

ier, Amsterdam, 1989.

21. R. M. Contractor, H. E. Bergna, H. S . Horovitz, C. M. Blackstone, U. Chow-
dhary, and A. E. Sleight, Butane oxidation to maleic anhydride in a recircu-
lating solids reactor, in Catalysis-1 987 (J. W. Ward, ed.), Elsevier, Amster-
dam, 1988, p. 645.
22. R. M. Contractor and A. E. Sleight, Catal. Today, 1, 587 (1987).
23. R. M. Contractor and J. Chauoki, Circulating fluidized bed as a catalytic
reactor, in Circulating Fluidized Bed Technology-IZI (P. Basu, M. Horio, and
M. Hasatani, eds.), Pergamon Press, Toronto, Ont., 1990, Vol. 1, Chap. 1.
24. R. M. Contractor, D. I. Garnett, H. S. Horowitz, H. E. Bergna, G. S. Patience,
J. T. Schwartz, and G. M. Sister, A new commercial scale process for n-butane
oxidation to maleic anhydride using a circulating fluidized bed reactor, in New
Development in Selective Oxidation II (V. Cortex Corberan and S. Vic Bellon,
eds.), Elsevier, 1990, p. 233.
25. Yu. Sh. Matros, Prospects of use of unsteady-state processes in catalytic re-
actors, JVKhO im. D.I. Mendeleeva [Journal of All-Union Chemical Society],
22, 576 (1977) (in Russian).
26. Yu. Sh. Matros, Chem. Eng. Sci., 45, 2097 (1990).
27. G. K. Boreskov, G. A. Bunimovich, Yu. Sh. Matros, and A. A. Ivanov, Kinet.
Catal., 23, 402 (1982).
28. G. K. Boreskov, Yu. Sh. Matros, V. I. Lugovskoy, G. A. Bunimovich, and V.
I. Puzhilova, Teor. Osn. Khim. Technol. [Theoretical Fundamentals of Chem-
ical Engineering], 18, 328 (1984).
29. G. Eigenberger and U. Nieken, Chem. Eng. Sci., 43, 2109 (1988).
30. Yu. Sh. Matros, A. S. Noskov, and V. A. Chumachenko, Chem. Eng. Pro-
cessing, 32, 89 (1993).
31. Yu. Sh. Matros, G. A. Bunimovich, and G. K. Boreskov, Unsteady-state per-
formance of sulphur dioxide oxidation in production of sulfuric acid, in Fron-
tiers in Chemical Reaction Engineering, Vol. 2 (L. K. Doraiswarni, and R. A.
Mashelkar, eds.), Wiley Eastern, New Delhi, 1984.
32. Yu. Sh. Matros, Unsteady-state oxidation of sulfur dioxide in sulfuric acid
production, Sulphur, 183, 23 (1986).
33. Yu. Sh. Matros and G. A. Bunimovich, Reverse process of sulfur dioxide
oxidation in sulfuric acid production, in Sulphur 1990, Preprints of Interna-
 62 MATROS AND BUNIMOVICH

tional Conference, Cancun, Mexico, April 1-4, 1990 (British Sulphur Cor-
poration Ltd, eds.), Gosport, Hants, 1990, p. 249.
34. L. N. Bobrova, E. M. Slavinskaya, A. S. Noskov, and Yu. Sh. Matros, React.
Kinet. Catal. Lett., 37, 267 (1988).
35. A. S. Noskov, L. M. Bobrova, and Yu. Sh. Matros, Catal. Today, 17, 293
(1993).
36. D. A. Frank-Kamenetski, Difision and Heat Exchange in Chemical Kinetics,
Princeton University Press, Princeton, NJ, 1955.
37. G. K. Boreskov, Yu. Sh. Matros, 0. V. Kiselev, and G. A. Bunimovich, Per-
formance of heterogeneous catalytic process in unsteady-state conditions,
Doklady Akademii Nauk SSSR, [Doklady Chemical Technology], 237, 160
(1977).
38. G. K. Boreskov, Yu. Sh. Matros, V. Yu. Volkov, and A. A. Ivanov, The process
of sulfur dioxide oxidation in sulfur trioxide, Auth. Certificate USSR 994400,
Downloaded by [171.67.34.205] at 22:03 14 March 2013

Off, Bull., No. 5, 1983.

39. G. K. Boreskov, Yu. Sh. Matros, A. I. Oruzheinikov, A. V. Filatov, V. I.
Volkov, A. I. Toporovskii, V. A. Orlov, and N. N. Kundo, The process of
elementary sulfur production, Auth. Certificate USSR 977852, Off, Bull., No.
46, 1984.
40. G. K. Boreskov, Yu. Sh. Matros, and V. S. Lakhmostov, The process of heat
utilization, Auth. certificate USSR 1160201, Off. Bull., No. 21, 1985.
41. G. K. Boreskov, A. S. Noskov, Yu. Sh. Matros, I. A. Zolotarskii, and V. N.
Menshov, The process of hydrogen production, Auth. Certificate USSR
1188095, OfiBull., No. 40, 1985.
42. G. K. Boreskov, N. M. Zhavoronkov, 0. S. Chekhov, Yu. Sh. Matros, V. N.
Orlik, A. M. Alekseev, and Yu. A. Sokolinskii, The process of ammonia syn-
thesis, Auth. Certificate USSR 865796, Off. Bull., No. 35, 1981.
43. G. K. Boreskov, Yu. Sh. Matros, I. A. Zolotarsky, V. S. Lakhmostov, I. A.
Ryzhak, and Yu. V. Lender, The process of methanol synthesis, Auth. Certif-
icate USSR 1249010, OffBull., No. 29, 1986.
44. G. K. Boreskov, Yu. Sh. Matros and V. S. Lakhmostov, The process of sulfur
trioxide production, Auth. Certificate USSR 865796, 1982.
45. G. K. Boreskov, Yu. Sh. Matros, and V. I. Lugovskoy, Method of VOC pu-
rification, Auth. Certificate USSR 849594, Off, Bull., No. 14, 1982.
46. Yu. Sh. Matros, G. K. Boreskov, V. S. Lahmostov, Yu. V. Volkov, and A. A.
Ivanov, Method of producing sulfur trioxide, U. S. Patent 4,478,808, 1984;
Canadian Patent 1,154,933, October 10, 1983; Japan Patent 60-18604, June
11, 1985; German Patent 4050468A1, July 20, 1982.
47. P. J. Heggs, Trans. Inst. MC, 8, 115 (1986).
48. P. J. Heggs and N. Abdullah, Chem. Eng. Res. Des., 6 , 258 (1986).
49. G. R. Gavalas, Znd. Eng. Chem. Fundam., 10, 71 (1971).
50. G. R. Gavalas, AIChE J., 17, 787 (1971).
51. R. J. Gartside, R. J. Feldman, C. Ercan, and F. M. Dautzenberg, New
CATOFINa process provides higher conversion and selectivity, presented at
the World Conference on Refinery Processing and Reformulated Gasoline, San
Antonio, TX, March 22-24, 1994, ABB Lumuss Crest, 14 pp.
52. R. F. Blanks, T. S. Wittrig, and D. A. Petterson, Chem. Eng. Sci., 45, 2407
(1990).
 REVERSE-FLOW OPERATION 63

53. G. Eigenberger, and U. Nieken, Chem. Ing. Tech., 63, 781 (1991); English
version in Int. Chern. Eng., 34, 4 (1994).
54. U. Nieken, G. Kolios, and G. Eigenberger, Chem. Ing. Tech., 64, 826 (1992).
55. German Patent 3,505,354 A l , 1985.
56. M. Sadakata, M. Navada, M. Imagava, T. Furusava, and D. Kunii, Int. Chem.
Eng., 21, 303 (1981).
57. M. Kotter, H.-G. Lintz, and T. Turek, Chem. Eng. Sci., 47, 2763 (1992).
58. D. Agar and W. Ruppel, Chem. Eng. Sci., 43, 2073 (1988).
59. D. Agar and W. Ruppel, Chem. Ing. Tech., 60, 731 (1988).
60. F. G. Cottrel, Purifying gases and apparatus therefor, U. S. Patent 2,171,733,
June 21, 1938.
61. N. Gilbert and F. Daniels, Ind. Eng. Chem., 40, 1719 (1948).
62. R. Houston, Regenerative incinerator systems for waste gases, U. S. Patent
3,870,474, March 11, 1975.
Downloaded by [171.67.34.205] at 22:03 14 March 2013

63. A. Fujii, Aircleaner, Japan Patent 52-49424, 1977.

64. A. Fujii, Aircleaner, Japan Patent 52-49425, 1977.
65. A. Fujii, Aircleaner, Japan Patent 52-49426, 1977.
66. Yu. Sh. Matros, Unsteady Processes in Catalytic Reactors, Elsevier, Amster-
dam, 1985.
67. G . F. Froment and H. Hofman, Design of fixed bed gas-solid catalytic re-
actors, in Chemical Reactions and Reactor Engineering (J. J. Carberry and
A. Varma, eds.), Dekker, New York, 1986.
68. S. Fey0 de Azavedo, M. A. Romero-Agawa, and A. P. Wardle, Trans IChemE,
68, Part A , 483-502 (1990).
69. 0. V. Kiselev, and Yu. Sh. Matros, Intrinsic energy removal in the heat front
moving in catalyst bed, Fizika Goreniia i Vzryva [Combust. Explos. Shock
Waves], 23, 167 (1987).
70. 0. V. Kiselev, Yu. Sh. Matros, and N. A. Chumakova, Phenomena of heat
front propagation in fixed catalyst bed, in Propagation of Heat Waves in Het-
erogeneous Media (Yu. Sh. Matros, ed.), “Nauka”, Siberian Branch, Novo-
sibirsk, 1988 (in Russian).
71. G. K. Boreskov, G. A. Bunimovich, Yu. Sh. Matros, 0. V. Kiselev, and I. A.
Zolotarski, Doklady Akademii Nauk SSSR [Doklady Chemical Technology],
268, 646 (1983).
72. T. N. Haynes, Ch. Georgakis, and H. S. Caram, Chem. Eng. Sci., 47, 2927
(1992).
73. A. Gawdzik and L. Rakovski, Chem. Eng. Sci., 43, 3023 (1988).
74. A. Gawdzik and L. Rakovski, Computers Chem. Eng., 13, 1165 (1989).
75. A. Gawdzik and L. Rakovski, Intern. Chem. Eng., 32, 192 (1992).
76. S . K. Bhatia, Chem. Eng. Sci., 46, 361 (1991).
77. V. K. Gupta and S . K. Bhatia, Computer Chem. Eng., 15, 1165 (1991).
78. 0. V. Kiselev, Theoretical Study of the Phenomena of Heat Waves Movement
in Catalyst Bed, Russian Academy of Sciences, Institute of Catalysis, Novo-
sibirsk, 1993 (in Russian).
79. G. F. Froment, Reversed flow operation of fixed bed catalytic reactor, in Un-
steady State Processes in Catalysis (Yu. Sh. Matros, ed.), VPS BV, Utrecht,
1990, pp. 57-89.
 64 MATROS AND BUNIMOVICH

80. K. M. Van den Bussche, S. N. Neophytides, I. A. Zolotarskii, and G. F. Fro-

ment, Chem. Eng. Sci., 48 (12), 3335-3345 (1993).
81. G. F. Froment and K. B. Bishoff, Chemical Reactor Analysis and Design, 2nd
ed., Wiley, New York, 1990.
82. M. E. Aerov, 0. M. Todos, and D. A. Narinskii, Columns with Packed Bed,
“Chimia,” Leningrad, 1979 (in Russian).
83. A. R. Balakrishnan, and D. C. T. Pei, Ind. Eng. Chem. Process. Des. Dev, 18,
30 (1979).
84. A. R. Balakrishnan and D. C. T. Pei, Chem. Eng. Sci., 30, 293 (1975).
85. N. Wakao and S. Kaguei, Heat and Mass Transfer in Packed Beds, Gordon
and Breach, New York, 1982.
86. Yu. Sh. Matros, V. I. Lugovskoy, B. N. Ogarkov, and V. B. Nakrohin, Teo-
reticheskie Osnovy Khimicheskoi Tekhnologii, [Theoretical Fundamentals
Chemical Engineering], 12, 291 (1978).
87. Yu. Sh. Matros, A. G. Ivanov, and L. L. Gogin, Teoreticheskie Osnovy Khim-
Downloaded by [171.67.34.205] at 22:03 14 March 2013

icheskoi Tekhnologii [Theoretical Fundamentals Chemical Engineering], 22,

481 (1988).
88. B. Young, D. Hildebrandt, and D. Glasser, Chem. Eng. Sci., 47, 1825 (1992).
89. G. K. Boreskov, R. A. Buyanov, and A. A. Ivanov, Kinet. Catal., 8, 153
(1967).
90. N. A. Chumakova, and I. A. Zolotarskii, Reverse-flow operation of a fixed-
bed catalytic reactor. Analysis of relaxed steady-state regimes, in Unsteady
State Processes in Catalysis (Yu. Sh. Matros, ed. ), VPS BV, Utrecht, 1990.
91. Yu. V. Ivanov, A. S. Noskov, Yu. Sh. Matros, and G. A. Bunimovich, Teo-
reticheskie Osnovy Khimicheskoi Tekhnologii [Theoretical Fundamentals
Chemical Engineering], 26, 228 (1992).
92. E. Wicke and D. Vortmeyer, Z. Electrochem Ber. Bunsenges. Phys. Chem., 63,
145 (1959).
93. E. Wicke and G. Padberg, Chem. Ing. Tech., 40, 1033 (1968).
94. P. Feiguth and E. Wicke, Chem. Ing. Tech., 43, 604 (1971).
95. G . Padberg and E. Wicke, Chem. Eng. Sci., 22, 1035 (1967).
96. Ya. B. Zeldovich and D. A. Frank-Kamenetski, Doklady Akademii Nauk SSSR
[Doklady Physical Chemistry], 12, 100 (1938); To the theory of uniform flame
propagation, Ibid., p. 693.
97. Ya. B. Zeldovich, G. I. Barenblatt, V. B. Librovich, and G. M. Makhviladse,
The Mathematical Theory of Combustion and Explosions, Nauka, MOSCOW,
1980 (in Russian); English version published by Consultants Bureau, New
York, 1985.
98. D. Vortmeyer and W. Jahnel, Chem. Eng. Sci., 27, 1485 (1972).
99. D. Vortmeyer and W. Jahnel, Chem. Ing. Tech., 43, 461 (1971).
100. G. Eigenberger, Chem. Eng. Sci., 27, 1909 (1972); ibid., 1917 (1972).
101. H-K. Rhee, R. P. Lewis, and N. R. Amundson, Ind. Eng. Chem. Fundam., 13,
317 (1974).
102. E. D. Gilles, Chem. Eng. Sci., 29, 1211 (1974).
103. E. D. Gilles and W. Ruppel, Model reduction in chemically reacting system,
in Dynamic and Modeling of Reactive System, Univ. of Wisconsin, Madison,
1979.
104. V. Pinjala, Y. C. Chen, and D. Luss, AIChE J . , 34, 1663 (1988).
 REVERSE-FLOW OPERATION 65

105. D. Vortmeyer, Z. Elektrochem. Ber. Bunsenges. Phys. Chem., 65, 282 (1961).
106. B. Simon and D. Vortmeyer, Chem. Eng. Sci., 33, 109 (1978).
107. C. Sharma and R. Hughes, Chem. Eng. Sci., 34, 613-634 (1979).
108. H. V. Doesburg and W. D. Jong, Chem. Eng. Sci., 31, 45-58 (1976).
109. A. S. Noskov, V. I. Drobishevich, 0. V. Kiselev, L. V. Yausheva, and Yu. Sh.
Matros, Doklady Akademii Nauk SSSR [Doklady Chemical Technology], 269,
1139 (1983).
110. A. V. Il’in and D. Luss, AZChE J., 38, 1609 (1992).
111. A. V. Il’in and D. Luss, Ind. Eng. Chem. Res., 32, 247 (1993).
112. G. K. Boreskov, Yu. Sh. Matros, and 0. V. Kiselev, Doklady Akademii Nauk
SSSR [Doklady Chemical Technology], 248, 160 (1979).
113. G. K. Boreskov, Yu. Sh. Matros, and 0. V. Kiselev, Kinet. Catal., 20, 773
(1979).
114. 0. V. Kiselev, Physika Goreniia z Vzyiva [Combust. Explos. Shock Waves],
Downloaded by [171.67.34.205] at 22:03 14 March 2013

23, 38 (1987).
115. 0. V. Kiselev and Yu. Sh. Matros, Flame front propagation in fixed catalyst
bed, Physika Goreniia i Vzryva [Combust. Explos. Shock Waves], 16, 25
(1980).
116. 0 . V. Kiselev and Yu. Sh. Matros, Physika Goreniia i Vzryva [Combust. Ex-
plos. Shock Waves], 21, 66 (1985).
117. J. D. Snyder and B. Subramaniam, Chem. Eng. Sci., 48, 4051 (1993).
118. N. A. Chumakova and Yu. Sh. Matros, About relaxed steady-state operation
in catalytic process model, in Boundaly Problems for Partial Differential
Equations, Editions of the Institute of Mathematics, Novosibirsk, 1990 (in
Russian).
119. I. A. Zolotarskii, unpublished results, private communication, 1986.
120. D. Vortmeyer and R. J. Shaefer, Chem. Eng. Sci., 29, 485 (1974).
121. H. Hausen, in Heat transfer, in Counterflow, Parallel Flow and Cross Flow
(A. J. Willmott, ed.), McGraw-Hill, New York 1983, Chap. 11-19.
122. V. Ya. Kunitskii, G. K. Boreskov, Yu. Sh. Matros, G. A. Bunimovich, A. A.
Balashov, M. A. Polischuk, V. S. Lakhmostov, and V. Z. Charnii, Start-up and
exploiting the industrial reactor for sulfur dioxide oxidation in unsteady-state
conditions, in Unsteady-State Processes in Catalysis, Part 2, (Yu. Sh. Matros,
ed. ), Institute of Catalysis, Novosibirsk, 1983 (in Russian).
123. G. A. Bunimovich, V. 0. Strots, and 0. V. Goldman, Theory and industrial
application of SO, oxidation reverse-process for sulfuric acid production, in
Unsteady State Processes in Catalysis (Yu. Sh. Matros, ed. ), VPS BV,
Utrecht, 1990.
124. J. Thullie and A. Burghardt, Application of the flow reversal reactor to the
methanol synthesis, in Unsteady State Processes in Catalysis (Yu. Sh. Matros,
ed. ), VPS BV, Utrecht, 1990.
125. Ch. Sapundzhiev, G. Grozev, and D. Elenkov, Chem. Eng. Technol., 13, 131
(1990).
126. Ch. Sapundzhiev, G. Grozev, and D. Elenkov, Chem. Eng. Technol., 14, 209
(1991).
127. Chr. Sapundzhiev, G. A. Bunimovich, and Yu. Sh. Matros, Commun. Depart.
Chem. Bulg. Acad. Sci., 17, 298 (1984).
128. G. A. Bunimovich and Yu. Sh. Matros, Unsteady-state performance of exo-
 66 MATROS AND BUNIMOVICH

thermic reaction in fixed catalyst bed reactor, in Proceedings of the 1st All-
Union Conference “Unsteady-State Processes in Catalysis, ’’ Institute of Ca-
talysis, Siberian Branch of the USSR Academy of Sciences, Novosibirsk, 1979
(in Russian).
129. G. K. Boreskov, Yu. Sh. Matros, and A. G. Ivanov, Dokl. Akad. Nauk SSSR,
288, 429 (1986).
130. C. Sapundzhiev, J. Chaouki, C. Guy, and D. Klvana, Chem. Eng. Commun.,
125, 171 (1993).
131. Yu. Sh. Matros, A. S. Noskov, and V. A. Chumachenko, in Catalytic Purifi-
cation of Waste Industrial Gases (V. N. Parmon, ed.), “Nauka,” Siberian
Branch, Novosibirsk, 1991 (in Russian).
132. A. N. Zagoruiko, Yu. Sh. Matros, V. Ravi Kumar, and B. D. Kulkami, Chem.
Eng. Sci., 47, 4315 (1992).
133. Yu. Sh. Matros and A. N. Zagoruiko, Doklady Akademii Nauk SSSR [Doklady
Downloaded by [171.67.34.205] at 22:03 14 March 2013

Chemical Technology], 294, 1424 (1987).

134. Yu. Sh. Matros and A. N. Zagoruiko, Sulphur recovery by the Claus reverse-
process, in Sulphur 1990, Preprints of International Conference, Cancun, Mex-
ico, April 1-4, 1990 (British Sulphur Corporation Ltd., ed.), Gosport, Hants,
1990.
135. A. N. Zagoruiko, A. S. Noskov, V. I. Drobyshevich, L. V. Yausheva, I. V.
Malakhova, and Yu. Sh. Matros, Teoreticheskie Osnovy Khimicheskoi Tekhn-
ologii [Theoretical Fundamentals Chemical Engineering], 23, 209 (1989).
136. A. N. Zagoruiko, Mathematical modeling and technological aspects of Claus
reverse process, in Unsteady State Processes in Catalysis (Yu. Sh. Matros,
ed.), VPS BV; Utrecht, 1990.
137. J. D. Snyder and i;. Subramaniam, Chem. Eng. Sci., 49, 5585 (1994).
138. G. A. Bunimovich, Ch. Sapundzhiev, Yu. Sh. Matros, V. 1. Drobishevich, L.
V. Yausheva, D. Elenkov, and G. Grozev, Mathematical simulation of tem-
perature unhomogeneities in a flow reversal reactor, in Unsteady State Pro-
cesses in Catalysis (Yu. Sh. Matros, ed. ), VPS BV, Utrecht, 1990.
139. Ch. Sapundzhiev, G. A. Bunimovich, V. I. Drobishevich, G. Grozev, L. V.
Yausheva, Yu. Sh. Matros, and D. Elenkov, Teoreticheskie Osnovy Khimi-
cheskoi Tekhnologii [Theoretical Fundamentals Chemical Engineering], 22
(3), 349-355 (1988).
140. G. K. Boreskov, Yu. Sh. Matros, G. A. Bunimovich, A. A. Balashov, Yu. V.
Filatov, V. P. Kozlov, N. F. Khripunov, V. S . Epifanov, A. E. Popov, Yu. M.
Soloviev, A. V. Safonov, and 0. N. Smirnova, Pilot study of unsteady-state
method of sulphur dioxide oxidation, in Unsteady-State Processes in Chemical
Reactors, Institute of Catalysis, Siberian Branch of the USSR Academy of
Sciences, Novosibirsk, 1982 (in Russian).
141. S. G. Neophytides and G. F. Froment, Ind. Eng. Chem. Res., 31, 1583 (1992).
142. A. I. Oruzheinikov, Yu. V. Ivanov, Yu. Sh. Matros, and A. P. Gerasev, Kinet.
Catal., 27, 954 (1986).
143. G . A. Bunimovich, N. V. Vernikovskaya, V. 0. Strots, Yu. Sh. Matros, and
B. S. Balzhinimaev, Influence of vanadium catalyst dynamic properties on the
parameters of reverse-process of sulfur dioxide oxidation, preprint of the In-
stitute of Catalysis, Russian Academy of Sciences, Novosibirsk, 1992 (in
Russian).
 REVERSE-FLOW OPERATION 67

144. B. S. Balzhinimaev, A. A. Ivanov, L. B. Lapina, V. M. Mastichin, and K. I.

Zamaraev, Faraday Discuss. Chem. SOC.,6, 133 (1989).
145. A. A. Ivanov and B. S . Balzhinimaev, React. Kinet. Catal. Lett., 35, 413
(1987).
146. J. P. Briggs, R. R. Hudgins, and P. L. Silveston, Chem. Eng. Sci., 32, 1087
(1977).
147. P. L. Silveston, R. R. Hudgins, S . Bogdashev, N. Vernijakovskaja, and Yu.
Sh. Matros, Chem. Eng. Sci., 49, 335 (1993).
148. B. S. Balzhinimaev, V. E. Ponomarev, G. K. Boreskov, A. A. Ivanov, V. S.
Sheplev, and E. M. Sadovskaya, React. Kinet. Catal. Lett., 28, 81 (1985).
149. L. van den Beld and K. R. Westerterp, Purification of waste air in a reverse
flow reactor, in Preprint Volume of First Topical Conference on Industrial
Chemical Engineering Technology, (J. B. Cropley, Coord.), St. Louis Annual
Meeting of AIChE, 1993.
Downloaded by [171.67.34.205] at 22:03 14 March 2013

150. L. L. Gogin, A. G. Ivanov, and V. S. Lachmostov, Utilization of high-grade

heat produced in detoxication of industrial waste gases using unsteady-state
catalytic method, in Unsteady-State Processes in Catalysis (Yu. Sh. Matros,
ed. ), VSP BV, Utrecht, 1990.
151. C . Isozaki, T. Katagiri, Y. Nakamura, S. Hirabayashi, S . Yabe, and T. Yamaki,
Demonstration of D C D A sulphuric acid plant based on unsteady-state pro-
cess, in Unsteady State Processes in Catalysis (Yu. Sh. Matros, ed.), VPS BV,
Utrecht, 1990.
152. L. M. Levina, V. 0. Strots, S. A. Popov, and Yu. Sh. Matros, React. Kinet.
Catal. Lett., 42, 73 (1990).
153. J. Wojciechowski, Method of catalytic cleaning of gas, Eur. Patent 0,037,119,
1980.
154. J. Wojciechowski and J. Haber, Appl. Catal., 4 , 275 (1982).
155. Yu. Sh. Matros, A. S. Noskov, A. N. Zagoruiko, and 0. V. Goldman, Com-
parison of two types of catalytic reverse-flow reactors, Teoreticheskie Osnovy
Khimicheskoi Tekhnologii [Theoretical Fundamentals Chemical Engineering]
(1994, in press).
156. K. Miiller, Brenstofi- Warme-Krafi, 37, 85 (1985).
157. Yu. Sh. Matros, G. A. Bunimovich, and A. S . Noskov, Catal. Today, 17, 261
(1993).
158. G. V. Vanin, A. S . Noskov, G . Ya. Popova, T. V. Andrushkevich, and Yu. Sh.
Matros, Cataly. Today, 17, 251 (1993).
159. Yu. Sh. Matros, D. E. McCombs, V. 0. Strots, G. A. Bunimovich, and C.
Roach, Catalytic reverse-process for VOC control: Experimental data and re-
actor simulation, in Environmental Issues and Solutions in Petroleum Explo-
ration, Production and Refining, Proceedings of the International Petroleum
Environmental Conference, March 2-4, 1994, Houston, TX (K. L. Sublette,
T. M. Harris, and F. S. Manning, eds. ), PennWell Books, 1994, pp. 44-53.
160. A. A. Balashov, Yu. V. Filatov, V. P. Kozlov, V. S. Epifanov, Yu. M. Soloviev,
and 0. N. Smirnova, Khimicheskaya Promyshlennost [Chemical Industry], 12,
730 (1987).
161. Hydrocarbon Processing, 58, 140 (1979).
162. Yu. Sh. Matros and A. N. Zagoruiko, Process of sulfur recovery, Soviet Patent
Application 4,131,457, 1986.
 68 MATROS AND BUNIMOVICH

163. J. R. Rostrup-Nielsen, I. Dybkjaer, and L. J. Christiansen, Steam-reforming.

opportunities and limits of the technology, in NATO ASI Chemical Reactor
Technology for Environmentally Safe Reactors and Products (H. de Lasa, ed.),
Kluwer, Dordrecht, 1992, p. 249.
164. A. P. Gerasev and Yu. Sh. Matros, Theoreticheskie Osnovy Khimicheskoi Tech-
nologii [Theoretical Fundamentals Chemical Engineering], 25, 821-827
(1991).
165. U. Nieken, J. Kolios, and G. Eigenberger, Chem. Eng. Sci., 49, 5507 (1994).
166. S . Punvono, H. Budman, R. R. Silveston, and Yu. Sh. Matros, Chem. Eng.
Sci., 49, 5437 (1994).
167. L. van de Beld and K. R. Westerterp, Chem. Eng. Technol., 17, 217 (1994).
168. K. Gosiewski, Chem. Eng. Proc., 32, 233 (1993).
169. L. van de Beld, R. A. Borman, 0. R. Derkx, B. A. A. van Woezik, and K.
R. Westerterp, Ind. Eng. Chem. Res., 33, 2946 (1994).
Downloaded by [171.67.34.205] at 22:03 14 March 2013

170. J . Chaouki, G. Gui, C. Sapundzhiev, D. Kusohorsky, and D. Klvana, Ind. Eng.

Chem. Res., 33, 2957 (1994).
171. Yu. Sh. Matros, G. A. Bunimovich, Znd. Eng. Chem. Res., 34, 1630 (1995).
172. G. A. Bunimovich, N. V. Vernikovskaya, V. 0. Strots, Yu. Sh. Matros, and
B. S. Balzhinimaev, Chem. Eng. Sci., 50, 565 (1995).
173. U. Nieken, G. Kolios, and G. Eigenberger, Catul. Today, 20, 335 (1994).
174. Yu. Sh. Matros, G. A. Bunimovich, S. E. Patterson, and S. F. Meyer, Catal.
Today, 21, (1995, accepted for publication).