4140 Ind. Eng. Chem. Res.

1999, 38, 4140-4143

COMMENTARIES

Chemical Reaction Engineering

In 1956 a group of European chemical engineers came Here is an example of a catalytic rate equation
together in Amsterdam for a symposium1 to exchange derived from the LHHW theory. For a first-order
ideas on a wide range of topics, all dealing with chemical reversible reaction A a R, dual-site mechanism, surface
reactors. Professor van Krevelen suggested that they reaction controlling, no product R in the feed, the rate
name the theme of their conference chemical reaction equation, from H&W2
engineering, CRE. This name has taken hold, and today
CRE takes its place as one of the key branches of
chemical engineering. Let me briefly tell what the
thinking was before this development, what was dis-
-rA )
ksLKA
(1 + CAKA + CRKR + CIKI + ‚‚‚)
(
2
CA -
CR
K )
tinctive about the CRE approach, and finally what has (1)
happened since.
The petroleum processing industry blossomed early The corresponding conversion expression for a plug flow
reactor, from H&W3 is

[
in this century, and engineers were needed to serve its
goals. To meet this demand, the profession of chemical
1 2
engineering was founded and chemical engineering
CW (π
+K n +Kn ) A A0 I I0

π( F )
departments were created to teach and prepare these ) +
engineers. Although this may seem to be an overly
simplistic view, still, it is true that the petroleum (1 + K1 )
industry was the most important force driving the 1
creation of our profession. 2( + K n + K n )(K A A0 I I0 R - KA)nA0
π
In the first half of our century studies in chemical +
(1 + K1 )
2

](
engineering focused on physical operations. New design
methods were developed for flow systems, heat exchang-

)
ers, absorbers, extractors, distillation columns and the (KR - KA)nA02 nA0 (KR - KA)2x2
like, and many textbooks were devoted to these subjects. ln - -
13 1 1
This approach was called unit operations.
At that time no general methods had been developed (
1+
K )
nA0 - 1 + x
K
21+
K ( ) ( )
1
to guide the design of reactors. Instead, a subject called
unit processes was created as a means for the orderly (
π
-
R A)
2 + KAnA0 + KInI0 (KR - KA)x (K - K )2n x
A0
classification of processes involving chemical reactions. 1 12
However, these chemical processes seemed so different,
group from group (hydrogenations, thermal cracking,
1+
K ( )
1+
K ( )
(2)
sulfonations, chlorinations, and so on) and even from
reaction to reaction within a group, that each process If we bypass the LHHW model, and instead use a
had to be treated uniquely. Intuition and past practices simple first-order kinetic model, we get the following
were the prime guides for the practitioner in his quest equation in place of eq 1:
for a good and efficient design for his particular reaction.
1 k′1 CR,eq
The Petrotech Approach Aa R, ‚‚‚ -r′A ) k′1CA - k′2CR, ‚‚‚ K ) )
2 k′2 CA,eq
The year 1947 was a momentous year, for it was then (3)
that the first book on the engineering of chemical
In addition, in place of the cumbersome expression of
reactors appeared. It was Hougen and Watson’s Chemi-
eq 2, this model gives
cal Process Principles: Part III, Kinetics and Catalysis.

( )
I refer to this book as H&W.
k′CA0W xA,eq
This book guided the study of the subject in the k′t′ ) ) xA,eq ln (4)
petroleum processing and chemical industries and in FA0 xA,eq - xA
universities in the United States. In part, its methods
and philosophy are still evident today. I call this the Compare the simplicity of eq 3 with eq 1 and eq 4
petrotech approach. Its unique feature is that it is based with eq 2. Also note that the above H&W equations are
exclusively on Eyring’s theory of absolute reaction rates. written without spelling out the meaning of the symbols,
This was applied by Langmuir-Hinshelwood and Hou- but only to show their comparative complexity.
gen-Watson to reactions of all types: homogeneous, As mentioned earlier, the theory of absolute reaction
solid-catalyzed gas phase, gas/solid, and fluid/fluid. I rates was used, not just for solid-catalyzed systems but
designate this theory as LHHW. for all reaction types. As another example, consider a
10.1021/ie990488g CCC: $18.00 © 1999 American Chemical Society
Published on Web 09/11/1999

particle which shrinks as it reacts with gas:
A(g) + B(s) a R(g) + S(g)
With slow surface reaction controlling, the LHHW
model gives the following rate equation, from H&W:4
Area(L′1)2k′1
-rA ) × the overall problem with a kinetic model appropriate
(1 + CAKA + CRKR + CIKI + ‚‚‚)2
( )
CRCSKRKS
CAKA - (5)
Ksurf
which represents the rate of shrinkage of a particle.
Alternatively, the shrinking core model gives the fol-
lowing conversion vs time performance expression:
t
) 1 - (1 - xB)1/3
ttotal
I suggest that the adventurous reader try to develop the
equivalent conversion-time expression from the LHHW
rate equation, eq 5.
A second and much more serious criticism on the use
of the LHHW model by engineers is illustrated with the
following detailed example laid out by H&W5 on the
hydrogenation of iso-octenes (called codimers):
C8H16 + H2 a C8H18
codimer
To find a rate equation, 40 experimental runs were
made to test 18 possible LHHW mechanisms. These
were as follows:
(a) adsorption of codimer controlling
(b) adsorption of hydrogen controlling
(c) desorption of octane controlling
(d) surface reaction controlling
(e)-(r) ... 14 additional mechanisms
By least-squares analysis H&W concluded that only one
mechanism, mechanism (h), fit the data. However,
Chou6 by a statistical analysis showed that other
mechanisms fitted the data equally well.
Now, to tell that one of these specific mechanisms
represents reality, one must show that 17 of the 18
mechanisms tested do not fit the data. This was not
done in this example and has never been done for any
example since.
This fact shows that choosing one of these 18 mech-
anisms simply represents a curve-fitting procedure.
With this in mind the engineer should pick the simplest
model to fit the data. This is the CRE approach, and
this often leads to the use of an nth order rate equation.
Reactor Design from CRE
In approaching the design of a reactor system the
engineer has to answer a number of important prelimi-
nary questions before embarking on his detailed calcu-
lations. These questions are as follows:
(1) Do I have the right reactor type in mind: should
it be plug flow, mixed flow, recycle, multistage, or what?
(2) What temperature progression should I aim for:
constant, rising, falling, etc., and would this require heat
exchange, maybe multistaged?
(3) For a catalytic reaction, what size of particle
should be used? This tells what type of reactor should
be used: packed bed, fluidized, etc.
Ind. Eng. Chem. Res., Vol. 38, No. 11, 1999 4141
(4) Does the catalyst deactivate, and if so, does it
deactivate rapidly or slowly?
All these questions require dealing with flow and
contacting and heat-transfer and mass-transfer factors.
Unfortunately, the LHHW rate expressions, such as eqs
1, 2, or 5, cannot in practice incorporate these physical
rate phenomena. On the contrary, CRE does deal with
for each class of system (fluid/fluid, fluid/solid, biochemi-
cal, solid catalytic, etc.) In many cases the nth order
rate model is chosen because it can account in a simple
way for all the physical factors. For example, an nth
order rate form can account for the following:
(1) Nonideal flow of fluid in the reactor by the axial
dispersion model (Danckwerts, Wilhelm), by the tanks-
in-series model (MacMullin, Denbigh), or by combined
models.
(6) (2) Fluid/fluid systems in which the two-film mass-
transfer model is combined with nth order kinetics
(Hatta, van Krevelen, Sharma).
(3) Pore diffusion in catalysts (Damkohler, Thiele,
Zeldovich, Wagner, Weisz, Prater, Wheeler).
(4) Deactivation of catalysts, and also tell which
temperature-time progression is best to use (Szepe,
Aris).
Let me illustrate this approach by considering the
combined effect of pore diffusion and deactivation on
reactor behavior.
For an nth order reaction, mth order deactivation, and
strong pore diffusional resistance (effectiveness factor
) ), we write
{
A f R with
-r′A ) k′CAna [mol reacted/kg of cat.‚s]
da
)kdam [1/s]
dt
where t is the reactor run time and a is the catalyst
activity, with the fresh catalyst having an activity of a
) 1.
You may think that this is too simple a model to be
of any practical use. Not so. Actually, it is the only model
which has been able to account for reaction with both
pore diffusion and deactivation effects. To illustrate, for
first-order kinetics this model gives the following per-
formance equations for plug flow of gas through a bed
of catalyst.
}
For no pore resistance,  ) 1 CA0
ln ) k′τ′,
and no deactivation, kd ) 0 CA
CA0W
where τ′ )
FA0
}
For no pore resistance, )1
but first-order deactivation, kd * 0, m ) 1
CA0
ln ) k′τ′ exp(kdt)
CA
For strong pore resistance,  ) 1/MT
and no deactivation, kd ) 0 }
CA0 k′τ′
ln )
CA MT
4142 Ind. Eng. Chem. Res., Vol. 38, No. 11, 1999

For strong pore resistance,  ) 1/MT

and first-order deactivation, kd * 0, m ) 1 }
ln
CA0 k′τ′
CA
)
MT
exp -
kdt
2 ( )
where

x
k′Fs
MT ) Thiele modulus ) L
Deff
Figure 1.
These simple performance expressions have been
shown by Levenspiel7 to fit the reported data. There is
no way to get any useful analytical expressions such as
the above by using the LHHW model. And, without an
analytical expression, you have no insight to guide you
to different and better operating conditions.
Reactor behavior is governed by four interrelated
quantities. These are shown in Figure 1.
For laboratory research, you choose the type of your
experimental reactor, hence the flow pattern; you mea-
sure the input and the output and from this you can
evaluate the kinetics. In process design, you specify the
input, you know the kinetics and flow pattern, and from
this you can predict the output.
Today, many researchers use physical chemistry and
quantum thermodynamics to come up with very detailed
kinetics: for example, some have used as many as
60 000 individual elementary reaction steps, while at
the same time completely ignoring the question of
looking for the best contacting or flow model. They just Figure 2.
assume that the plug flow model should be used. They
it was not practical to incorporate into its equations any
should realize that their reactor predictions are only as
of the transport resistances.
good as the cruder of the two above-mentioned factors,
The second step came in 1958 with the CRE approach,
as illustrated in Figure 2.
which was designed to account for all the physical and
Often, the lack of knowledge of the expected flow
chemical rate factors. To do this, it set aside the theory
pattern in the reactor is the main cause of uncertainty
of absolute reaction rates, and in its place adopted
in the design of reactors, not the kinetics.
simple kinetic rate expressions appropriate for this or
that system. This approach gave tractable conversion
CRE Today expressions.
Today, at the end of the century, the computer enables
Today, nearly every chemical engineering department the researcher to try the most complicated models that
in the world includes in its curriculum an undergradu- he desires, those involving multiple elementary reac-
ate course devoted to the study of reactors. However, tions and/or computational fluid dynamics.
an AIChE survey has shown that, just before the But we must be wary when we wield the impressive
introduction of the CRE approach, only 18% of all power of the computer. The findings from such analyses
departments in the United States taught the subject at will usually be of little generality and will only apply
all. The simplicity of the CRE approach made this to the system actually studied. Also, the researcher is
subject teachable. able to incorporate into his model many more factors
Because of its prestige, some chemical engineering (for example, 60 000 rate expressions) than he has
departments have changed their names to departments reliable experimental values for. What does that say for
of CRE. Also, the name has been embraced by all sorts his predictions?
of related and not so related subjects. However, at its This type of study is fine for simulation exercises, but
core CRE aims to bring together the study of reaction when it comes to reactor design the predictions of such
kinetics and the transport phenomena so that the analyses should be viewed with caution.
practioner can tell what kind of process to focus upon,
so as to end up with a well-behaved reactor setup which Acknowledgment
operates economically and stably while producing most
of what is wanted and least of what is not wanted. I thank Professors Ken Bischoff, Mike Dudukovic,
Steve Szepe, and Richard Turton who have read this
paper and have made useful comments.
Recapitulation

The first step in reactor design came in 1947 with the Literature Cited
petrotech approach which used the theory of absolute (1) Chemical Reaction Engineering, the First Symposium of the
reaction rates as its kinetic model for all reaction types. European Federation of Chemical Engineers; in Chem. Eng. Sci.
This was a mechanistic model and was so complex that 1958, 8, 1-200.

(2) Hougen, O. A.; Watson, K. M. Chemical Process Principles,
Part III; Wiley: New York, 1947; p 919, eq 37.
(3) Hougen, O. A.; Watson, K. M. Chemical Process Principles,
Part III; Wiley: New York, 1947; p 929, eq b.
(4) Hougen, O. A.; Watson, K. M. Chemical Process Principles,
Part III; Wiley: New York, 1947; p 1065, eqs 17 and 22.
Ind. Eng. Chem. Res., Vol. 38, No. 11, 1999 4143
(5) Hougen, O. A.; Watson, K. M. Chemical Process Principles,
Part III; Wiley: New York, 1947; pp 943-958.
(6) Chou, C. H. Least Squares. Ind. Eng. Chem. 1958, 50, 789.
(7) Levenspiel, O. The Chemical Reactor Omnibook; OSU
Bookstores: Corvallis, OR, 1996; Chapter 31.
Octave Levenspiel
Chemical Engineering Department
Oregon State University
Corvallis, Oregon 97331
IE990488G