8.

3 The Continuous Flow Stirred Tank Reactor (CSTR) 269

Substitution of numerical values gives of a reactor network must be proportioned such

1 that equal increments in conversion occur across
t = [ - £ n ( l - 0.9)] - 18.6 sec each leg. For this case, too, the network acts as
0.124 if it were a single plug flow reactor with a volume
This value is considerably less than the reactor equal to the sum of the constituent reactor
space time and differs from it by 38%. volumes. Thus for any series-parallel combina-
tion of plug flow reactors, one can treat the whole
system as a single plug flow reactor with a volume
8.2.3 Series-Parallel Combinations of equal to the total volume of the individual re-
Tubular Reactors actors, provided that the fluid streams are dis-
In order to achieve increases in production tributed in a manner such that streams that
capacity or to obtain higher conversion levels, combine have the same composition. Hence, for
it may be necessary to provide additional reactor any units in parallel, the space times or VR/FA0
volume through the use of various series-parallel must be identical. Other flow distributions would
be less efficient.
combinations of reactors.
Consider j plug flow reactors connected in
series and l e t / i , / 2 , / 3 , . . . / ) , . . . J) represent the 8.3 THE CONTINUOUS FLOW STIRRED
fraction conversion of the limiting reagent, TANK REACTOR (CSTR)
leaving reactors 1, 2, 3 , . . . / , . . . , j . For each of Continuous flow stirred tank reactors are widely
the reactors considered above, the appropriate used in the chemical process industry. Although
design equation is 8.2.7. For reactor j , individual reactors may be used, it is usually
dfA preferable to employ a battery of such reactors
VRi = (8.2.18) connected in series. The effectiveness of such
\-r. batteries depends on the number of reactors
The total reactor volume is obtained by used, the sizes of the component reactors, and
summing the individual reactor volumes. the efficiency of mixing within each stage.

dfA
, total - F P2 r J
P ]
~\ (8.2.19)
- A Jj-A~- A)J

From the principles of calculus, the quantity in Continuous flow stirred tank reactors are
brackets can be rewritten as a single integral. normally just what the name implies—tanks into
Hence which reactants flow and from which a product
stream is removed on a continuous basis.
CFSTR, CSTR, C-star, and back-mix reactor
Thus j plug flow reactors in series with a total are only a few of the names applied to the
volume VRtioial give the same conversion as a idealized stirred tank flow reactor. We will use
single reactor of volume T^>total. the letters CSTR as a shorthand notation in this
When plug flow reactors are connected in textbook. The virtues of a stirred tank reactor
parallel, the most efficient utilization of the total lie in its simplicity of construction and the rela-
reactor volume occurs when mixing of streams tive ease with which it may be controlled. These
of differing compositions does not occur. Con- reactors are used primarily for carrying out
sequently the feed rates to different parallel legs liquid phase reactions in the organic chemicals
270 Basic Concepts in Reactor Design and Ideal Reactor Models

industry, particularly for systems that are cha- molecules have short residence times, there is a
racterized by relatively slow reaction rates. If it rapid response at the reactor outlet to changes
is imperative that a gas phase reaction be carried in the reactor feed stream. This characteristic
out under efficient mixing conditions similar to facilitates automatic control of the reactor.
those found in a stirred tank reactor, one may Because the mixing process makes the prop-
employ a tubular reactor containing a recycle erties of the entire reactor contents equal to
loop. At sufficiently high recycle rates, such those of the effluent stream, there will be a step
systems approximate CSTR behavior. This sec- change in fluid composition and temperature at
tion is concerned with the development of design the point where the feedstream enters the reactor.
equations that are appropriate for use with the In the present chapter we will restrict the dis-
idealized stirred tank model. cussion to cases where there is no temperature
change on entrance to the reactor unless other-
8.3.1 Individual Stirred Tank Reactors wise noted. For these isothermal CSTR's the
8.3.1.1 Basic Assumptions and Design Relation- drop in reactant concentration as the fluid
ships. The most important feature of the enters the reactor implies that in the vast
CSTR is its mixing characteristics. The ideal- majority of cases the volume average reaction
ized model of reactor performance presumes rate will be low by comparison to that in an
that the reactor contents are perfectly mixed so isothermal plug flow reactor. Consequently,
that the system properties are uniform through- when both operate at the same temperature, a
out. The effluent composition and temperature CSTR will have to be significantly larger than
are thus identical with those of the reactor con- a PFR in order to effect the same composition
tents. This feature greatly simplifies the analysis change in a given amount of fluid. (Exceptions
of stirred tank reactors vis a vis tubular reactors occur in the case of autocatalytic reactions.)
for both isothermal and nonisothermal opera- However, by using a battery of stirred tanks in
tion. It is not difficult to obtain a good approxi- series, differences in total volume requirements
mation to CSTR behavior, provided that the can be significantly reduced. Moreover, be-
fluid phase is not too viscous. The approxima- cause of the simplicity of their construction,
tion to CSTR behavior is valid if the time stirred tank reactors normally cost much less
necessary to disperse an entering element of per unit of volume than tubular reactors. Thus,
fluid (e.g., a shot of dye or radioactive tracer) in many cases, it is more economic to employ a
uniformly throughout the tank is very much Targe stirred tank reactor or a battery of such
shorter than the average residence time in the reactors than it is to use a tubular reactor. Size
tank. comparisons of these two types can be treated
Unlike the situation in a plug flow reactor, the quantitatively, and we will return to this subject
various fluid elements mix with one another in in Sections 8.3.1.3 and 8.3.2.3. First, however, we
a CSTR. In the limit of perfect mixing, a tracer must develop the basic design equations.
molecule that enters at the reactor inlet has Consider the schematic representation of a
equal probability of being anywhere in the continuous flow stirred tank reactor shown in
vessel after an infinitesimally small time incre- Figure 8.5. The starting point for the develop-
ment. Thus all fluid elements in the reactor have ment of the fundamental design equation is again
equal probability of leaving in the next time a generalized material balance on a reactant
increment. Consequently there will be a broad species. For the steady-state case the accumula-
distribution of residence times for various tracer tion term in equation 8.0.1 is zero. Furthermore,
molecules. The character of the distribution is since conditions are uniform throughout the
discussed in Section 11.1. Because some of the reactor volume, the material balance may be
8.3 The Continuous Flow Stirred Tank Reactor (CSTR) 271

Outlet molal flow rate

Fraction conversion of
limiting reagent at outlet
Outlet concentration of
limiting reagent
F F
A out ~ A0 ( t)

F
A in = l n l e t molal flow rate
[A in = Fraction conversion
of limiting reagent
at inlet
CA in = Inlet concentration
of limiting reagent

Figure 8.5
Schematic representation of CSTR indicating process variables.

written over the entire reactor. Hence,

z /Rate of flow of \ /Rate of disappearance \

/ Rate ofr j aflow of
r N
\ / , . A / f ^ ^ •• l
I = reactant+ out of + of reactant uby reaction (8.3.1)
I reactant into reactor/ » ' ' 'J '
\ reactor \ in the reactor /

In terms of the symbols indicated in Figure 8.5, The last equation may be rewritten in terms
of the fraction conversion as
^ m = FAout + (-rAF)VR (8.3.2)
where we again emphasize that the appropriate ^ o ( l - L m ) = ^ o ( l - L o u t ) + (~rAF)VR
volume is that physically occupied by the re- (8.3.3)
acting fluid. The quantity (— rAF) is the rate of
disappearance of reactant A evaluated at reactor where FA0 is again the molal flow rate corre-
outlet conditions. sponding to zero conversion. Rearrangement
272 Basic Concepts in Reactor Design and Ideal Reactor Models

gives reference volumetric flow rate % and a corre-

JA out JA in sponding concentration {CA0\
(8.3.4)
F AO (~rAF) }R ut JA in)
(8.3.5)
Equation 8.3.4 is an extremely useful expression %~ i-rAP)
relating in a simple manner the reactor volume,
In terms of the reactor space time
the molal flow rate at zero conversion, the
change in fraction conversion accomplished
in the reactor, and the reaction rate. A knowl- T =
edge of any three of these quantities permits the (-rAF) ' AF)
fourth to be calculated directly. For reactor (8.3.6)
design purposes the two problems of primary
interest can be readily solved using this equation. This equation differs from that for the plug
flow reactor (8.2.9) in that for a CSTR the rate is
1. The size reactor needed to perform a specified evaluated at effluent conditions and thus appears
task under specified operating conditions outside the integral.
may be determined. It is particularly convenient to choose the
2. For a reactor of a given size one may deter- reference conditions at which the volumetric
mine either the conversion achieved for a flow rate is measured as the temperature and
specified flow rate and temperature or the pressure prevailing at the reactor inlet, because
quantity of material that can be processed to this choice leads to a convenient physical inter-
a given degree of conversion at a specified pretation of the parameters f^ and CA0 and,
temperature. in many cases, one finds that the latter quantity
cancels a similar term appearing in the reaction
Equation 8.3.4 may also be used in the analy- rate expression. Unless otherwise specified, this
sis of kinetic data taken in laboratory scale choice of reference conditions is used throughout
stirred tank reactors. One may directly deter- the remainder of this text. For constant density
mine the reaction rate from a knowledge of the systems and this choice of reference conditions,
reactor volume, flow rate through the reactor, the space time x then becomes numerically equal
and stream compositions. The fact that one may to the average residence time of the fluid in the
determine the rate directly and without integra- reactor.
tion makes stirred tank reactors particularly
Since one is almost always concerned with
attractive for use in studies of reactions with
liquid phase reactions when dealing with stirred
complex rate expressions (e.g., enzymatic or
tank reactors, the assumption of constant fluid
heterogeneous catalytic reactions) or of systems
density is usually appropriate. In this case equa-
in which multiple reactions take place.
tion 8.3.6 can be written as
Equation 8.3.4 is completely general and in-
dependent of whether the reaction occurs at
constant density (5A = 0) and of whether the
P dCA
^ A in ^A out
Jc
feed stream and the reactor contents have iden-
tical temperatures. The effective reactor volume (for constant density systems only)
is independent of the particular geometry giving
(8.3.7)
rise to this volume. All that is required is that
the contents be well mixed. We now wish to consider some examples that
If the molal flow rate at zero fraction conver- indicate how to employ the above equations in
sion is written in terms of the product of a reactor design analyses.
8.3 The Continuous Flow Stirred Tank Reactor (CSTR) 273

ILLUSTRATION 8.6 DETERMINATION OF This volume is appreciably larger than the

REQUIRED CSTR VOLUME UNDER volume of plug flow reactor calculated in Illus-
ISOTHERMAL OPERATING tration 8.3 for the same reaction conditions and
CONDITIONS—LIQUID PHASE REACTION fraction conversion. However, the cost of such a
reactor would be considerably less than the cost
Consider the Diels-Alder reaction between 1,3-
of a tubular reactor of the size determined in
butadiene (B) and methyl acrylate (M) discussed
Illustration 8.3.
in Illustration 8.3.
B + M -> C
8,3.1.2 Mean Residence Time in Stirred Tank
If the operating conditions used in that illustra-
Reactors. Stirred tank reactors differ from plug
tion are again employed, determine the volume flow reactors in that not all fluid elements remain
of a single continuous stirred tank reactor which in the CSTR for the same length of time. The
will give 40% conversion of the butadiene when characteristics of the residence time distribution
the liquid flow rate is 0.500 m 3 /ksec. function are treated in Chapter 11. In this sub-
section we consider only the problem of deter-
Solution
mining the average residence time of a fluid
For the conditions cited the reaction rate element in an ideal CSTR. This problem is
expression is of the form considerably simplified by the fact that the fluid
properties are uniform throughout the reactor
-rB = kCB0(\ - / B ) C A l c l 3 i 0 (A) and equal to those prevailing at the exit. Thus
with the mean residence time in an ideal CSTR is
k = 1.15 x 10~ 3 m 3 /mole-ksec given by
CB0 = 96.5 moles/m3 t = (8.3.8)
^AICI3,O = 6-^3 moles/m 3
where the volumetric flow rate is evaluated at
Equation 8.3.6 may be used as the basic design
effluent conditions.
relationship.
The effluent volumetric flow rate is also given
dfB by
i = (B) % = %(l + SAfAF) (8.3.9)
(-rBF) {-rBF)
where Yo is the volumetric flow rate evaluated
Combining equations A and B gives
at a composition corresponding to zero fraction
CBOIB JB conversion. Combining equations 8.3.8 and 8.3.9
= •
T
gives
kCB0(l — fB)C A\CI3,O ^(1 — / B
VR T
Substitution of numerical values gives t = S
A/AF) SA/AAF
O40 (8.3.10)
(1.15 x 10" 3 )(l - 0.4)(6.63)
Unlike the situation in the PFR, there is always
= 87.4 ksec = 24.3 hr a simple relationship between the mean resi-
dence time and the reactor space time for a
From the definition of the space time and the
CSTR. Since one normally associates a liquid
inlet volumetric flow rate,
feed stream with these reactors, volumetric ex-
VR - T^o - (87.4)(0.500) = 43.7 m 3 pansion effects are usually negligible (SA = 0).
274 Basic Concepts in Reactor Design and Ideal Reactor Models

In this situation the mean residence time and the For most of the commonly encountered types
reactor space time become identical. of reaction rate expressions the rate decreases
monotonically with increasing fraction con-
8.3.1.3 Relative Size Requirements for an In- version. The fact that a step change in composi-
dividual Continuous Stirred Tank Reactor and a tion occurs as a fluid enters an ideal CSTR
Plug Flow Reactor. In the development of the
final reactor design for a proposed production
requirement, the chemical engineer must con-
sider a variety of reactor types and modes of
operation. Several factors must be considered
in the development of the final design, some of Path followed in
which may be only peripherally related to the plug flow reactor

kinetics of the reaction in question. Many of

Rate
these factors are implicit in the questions posed
in Section 8.0.1. Since a variety of operating
conditions, modes of operation, and reactor
types can be used to accomplish a specified task, Rate prevailing in
individual CSTR
it is not possible to generate a simple logical
procedure that can be followed to arrive at a
truly optimum design. A knowledge of the per-
formance characteristics of individual ideal re-
Reactant concentration
actors (and combinations thereof) and sound
engineering judgment based on previous design Figure 8.6a
experience are both useful in arriving at a work- Reaction rate versus reactant concentration plot
able design. The nonanalytical reasoning that for typical reactions—single ideal reactors.
must perforce enter into design calculations is
beyond the scope of this textbook. This capa-
bility comes only with the accumulation of
experience and interaction with other individ-
uals who have long been regularly engaged in the Inlet conditions-
practice of reactor design. We can, however,
indicate the quantitative considerations that
have an important bearing on the economics of
Rate
the proposed design. The choice of the reactor
Effluent from reactor 1
network employed to carry out the desired con-
version will play an important role in that it will Effluent from
specify the size of the units needed and the distri- reactor 2

bution of products emerging from the reactor. Effluent from

reactor 3
In this section we turn our attention to the
problem of determining relative size require-
ments for a single ideal CSTR and a PFR where
Reactant concentration
both reactors operate isothermally at the same
temperature. The analysis applies to systems in Figure 8.6b
which only one reaction occurs to a significant Reaction rate versus reactant concentration plot
extent. for typical reactions—three reactors in series.
8.3 The Continuous Flow Stirred Tank Reactor (CSTR) 275

implies that for these cases the reaction rate will the area under a plot of the reciprocal reaction
be much smaller in this type of reactor than it rate ( — l/rA) versus fraction conversion was
would be in a tubular reactor being used to equal to the ratio VR/FA0 for a plug flow reactor.
accomplish the same composition change. Con- In the case of a CSTR equation 8.3.4 indicates
sequently, the CSTR must be significantly larger that on a similar plot, the quantity VR/FA0 is
than the PFR. These considerations are evident equal to the rectangular area shown in Figure
from a brief consideration of Figure 8.6a. In a 8.7. Thus, the ratio of the rectangular area to the
plug flow reactor one moves from right to left area under the curve is equal to the ratio of
along the rate curve as he or she proceeds from reactor volumes if identical molal flow rates and
inlet to outlet. In a CSTR, on the other hand, conversion increments are employed. For any
the reaction rate is constant throughout the rate expression that decreases monotonically
reactor and equal to that prevailing at the outlet. with increasing fraction conversion, the CSTR
Except in the case of autocatalytic reactions (see will always require a larger reactor volume than
Section 9.4), this rate corresponds to the lowest the corresponding PFR used to accomplish the
point on the PFR rate curve and implies that a same task.
larger volume will be required to accomplish
the same composition change between inlet and
outlet streams. Reciprocal,
rate curve
It is possible to reduce the disparity in reactor
volumes by using several tanks in series so that
one obtains stepwise changes in composition as
one proceeds from tank to tank. This situation is
depicted in Figure 8.6b for the case of three
CSTR's in series. In this case it is readily ap-
parent that each of the CSTR's would have a
larger volume than that of a PFR necessary to
effect the same composition change, so that the
total volume of the three CSTR's would exceed
that of a single PFR used to effect the same
overall change in composition. (We have shown
earlier that the three PFR's in series would be
equivalent to a single PFR with a volume equal
to the sum of those of the individual PFR's.) As
one increases the number of CSTR's in series, J A out

the disparities in total volume become less and, Fraction conversion, fA

in the limit, as the number of CSTR's approaches
infinity the battery will approach PFR behavior. Figure 8.7
For now, however, we wish to limit our con- Reciprocal rate versus fraction conversion plot for
siderations to a comparison of the relative sizes determination of VR/FA0 for CSTR.
of a single CSTR and a PFR under isothermal
operating conditions. If one has a knowledge of the analytical form
In order to be able to compare relative reactor of the reaction rate expression, equations 8.2.7
sizes one needs a knowledge of the form of the and 8.3.4 may be used to determine the relative
reaction rate expression in either graphical or reactor volumes required. In order to indicate
analytical terms. In Section 8.2.1 we showed that the utility of these equations, let us consider the
276 Basic Concepts in Reactor Design and Ideal Reactor Models

general class of reactions that follow simple nth- 3. The higher the order of the reaction, the
order kinetics with n normally lying between greater the ratio of sizes at a fixed conversion
zero and three. Following Levenspiel's analysis level. For zero-order reactions reactor size is
(11), we will presume that the inlet fraction con- independent of reactor type.
version is zero and that the volumetric expansion 4. Variations in fluid density on reaction can
parameter (3A) has some arbitrary value. Thus have significant effects on the size ratio, but
the effects are secondary when compared to
- fA)" the variations in reaction order. For positive
— r4 = (8.3.11)
SAfA)n values of the expansion parameter dA, the
volume ratio is increased, for negative values
For these conditions the general design equation
of SA, the volume ratio decreases. However,
for a plug flow reactor (8.2.7) becomes
the fact that in practice CSTR's are used only
vn 'Mi +sAfA)ndfA for liquid phase reactions makes this point
(8.3.12) academic.
,40/PFR

while that for a single CSTR (equation 8.3.4) The larger volume requirement of the CSTR
becomes does not necessarily imply extra capital costs,
especially for reactions that occur at ambient
+ sA/Ay pressure. However, the fact that the required
(8.3.13)
/csTR kC"A0 \ l - f A CSTR volume increases rapidly at high con-
version levels leads to some interesting optimiza-
Dividing equation 8.3.13 by equation 8.3.12 gives
tion problems in reactor design. The chemical
engineer must find the point at which he obtains
I
AO /CSTR 1-/4
an economic trade-off between the high fraction
conversion obtained in a large reactor versus the
dAfAy dfA
VRC%\

/PFR
p - fAr
low conversions in a small reactor. In the first
situation the equipment costs will be high and
Jo
(8.3.14) the product separation costs and raw material
costs low; in the second case equipment costs
Levenspiel (11) has evaluated the right side of will be low and the other costs high. The optim-
equation 8.3.14 for various values of n and SA. ization problem may be further complicated by
His results are presented in graphical form in allowing the number of CSTR's employed to
Figure 8.8. For identical feed concentrations vary, as we will see in Section 8.3.2.3.
(CA0) and molal flow rates (FA0) the ordinate of
In Illustrations 8.3 and 8.6 we considered the
the figure indicates the volume ratio required
reactor size requirements for the Diels-Alder
for a specified conversion level.
reaction between 1,4-butadiene and methyl
There are several useful generalizations which acrylate. For the conditions cited the reaction
may be gleaned from a thorough study of may be considered as a pseudo first-order reac-
Figure 8.8. They include the following. tion with 5A = 0. At a fraction conversion of
0.40 the required PFR volume was 33.5 m 3 , while
1. If the reaction order is positive, the CSTR is the required CSTR volume was 43.7 m 3 . The
larger than the PFR at all conversion levels. ratio of these volumes is 1.30. From Figure 8.8
2. The higher the fraction conversion involved, the ratio is seen to be identical with this value.
the greater the disparity between the sizes of Thus this figure or equation 8.3.14 can be used
the CSTR and the PFR. in solving a number of problems involving the
8.3 The Continuous Flow Stirred Tank Reactor (CSTR) 277

100

~2-
Io

0.01

Figure 8.8
Comparison of performance of single CSTR and plug flow reactor for the
nth-order reactions

A -> products, — rA = KCnA

The ordinate becomes the volume ratio KCSTR/VPFR or the space time ratio
T CSTR /T PFR if the same quantities of identical feed are used. (Adapted from
Chemical Reaction Engineering, Second Edition, by O. Levenspiel. Copyright
© 1972. Reprinted by permission of John Wiley and Sons, Inc.)

comparison of the performance of an individual 8.3.1.4 CSTR Performance Under Nonsteady-

CSTR with that of a plug flow reactor for systems State Conditions. During start-up and shut-
that obey nth-order kinetics. Mixed second- and down periods and during shifts from one steady-
third-order reactions can also be handled in state operating condition to a second, the design
terms of the nth-order model if stoichiometric equations developed in Section 8.3.1.1 are no
ratios of reactants are used. The techniques longer appropriate. In these cases the starting
employed in solving problems using the figure point for the analysis must be the generalized
will be discussed in Section 8.3.2.3 after a corre- material and energy balance equations con-
sponding figure has been developed for use with taining accumulation terms (equations 8.0.1 and
multiple identical CSTR's connected in series. 8.0.3). In order to indicate the general approach
278 Basic Concepts in Reactor Design and Ideal Reactor Models

to the nonsteady-state analysis of reactors, we For liquid phase reactions where SA = 0, the
wish to consider briefly the relations that govern following expression is appropriate.
the transient behavior of an individual CSTR
operating under isothermal conditions. In some Kn
ln ~ ^ t (8-3.19)
cases it is possible to obtain easily analytical dt
solutions describing the approach to the steady
Combining equations 8.3.18 and 8.3.19 gives
state because of the uniform composition of the
reactor contents. This situation is in distinct dCAF ,„
contrast to that prevailing in a plug flow reactor ^ (8.3.20)
dt
where one must invariably resort to numerical
solutions of the transient material balance rela- There are several interesting forms of equation
tions in order to describe the approach to steady- 8.3.20 that correspond to various limiting con-
state conditions. ditions. For example, if both VR and %n are
For nonsteady-state operating conditions the time invariant, we have the situation corre-
generalized material balance on reactant A is as sponding to a shift from one steady-state
follows. operating condition to a second, and the quantity

T r^ . . Disappearance
rsr . .
InputA = Output + u . K
-f Accumulation
by reaction

F = F A- (— r \V (8.3.15)
A in -* A out ' V AF/ R dt

where the reaction rate is evaluated at the con- VRl%n is just the reactor space time T. Hence,
ditions prevailing at the reactor outlet and NA for this case,
is the total number of moles of species A in the C — C
dC AF
reactor at time t. NA may change as a result of (8.3.21)
changes in the volume occupied by the reacting dt
fluid or as a result of changes in the composition It is readily apparent that equation 8.3.21 re-
of the fluid. Since duces to the basic design equation (equation
8.3.7) when steady-state conditions prevail.
= CAFVR (8.3.16)
Under the presumptions that CA in undergoes a
then step change at time zero and that the system is
isothermal, equation 8.3.21 has been solved for
various reaction rate expressions. In the case of
first-order reactions, solutions are available for
If the molal flow rates are written as the pro- both multiple identical CSTR's in series and in-
duct of a concentration and a volumetric flow dividual CSTR's (12). In the case of a first-order
rate, and if equation 8.3.17 is combined with irreversible reaction in a single CSTR, equation
equation 8.3.15, 8.3.21 becomes
dC AF cA in -kCAF (8.3.22)
dt T T
dC4
+ (8.3.18) This equation is an ordinary linear differential
dt dt
8.3 The Continuous Flow Stirred Tank Reactor (CSTR) 279

equation with constant coefficients if CA in and counter stepwise variations in temperature as

the rate constant are time independent. In this one moves from one reactor to another in the
case the solution may be obtained by separation cascade.
of variables and integration. Each of the individual CSTR's that make up
the cascade can be analyzed using the techniques
dC AF
rcAF
Jc* =J and concepts developed in Section 8.3.1. The
present section indicates how one may mani-
pulate the key relations developed earlier to
where C* is the reactant concentration in -the obtain equations that simplify the analysis of a
tank at time zero. Integration gives cascade of ideal CSTR's.
We begin by indicating a few generalizations
d* _ U + 1-)c* that are relevant to the treatment of batteries
of stirrfed tank reactors. Consider the cascade
of ideal CSTR's shown in Figure 8.10. For any
individual reactor denoted by the subscript i
= t (8.3.24) the basic design equatidn developed earlier as
k + - equation 8.3.4 is appropriate:
T
or JA out JA in Ji Ji-1
r
(8.3.26),
— r AF — rAi
r^% ^ A in
kx
where the subscripts used are as indicated in the
(8.3.25) figure and (— rAi) is the rate of disappearance
where the first term on the right is just the steady- of species A in reactor i. We wish to emphasize
that FA0 is the molal flow rate of reactant A
state effluent concentration.
~under conditions that would correspond to zero
fraction conversion. We should also note that
8.3.2 Stirred Tank Reactors in Series the conditions within any individual reactor
are not influenced by what happens in reactors
In order to reduce the disparities in volume or
downstream. The conditions of the inlet stream
space time requirements between an individual
and tho£e prevailing within the reactor itself
CSTR and a plug flow reactor, batteries or
are the only variables that influence reactor
cascades of stirred tank reactors ard employed.
performance under either steady-state or tran-
These reactor networks consist of a number of
sient conditions.
stirred tank reactors connected in series with
If desired, the last equation can also be
the effluent from one reactor serving as the
written in terms of the reactor space time for
input to the next. Although the concentration
the fth reactor as
is uniform within any one reactor, there is a
progressive decrease in reactant concentration
as ohe moves from the initial tank to the final vRi 3.3.27)
tank in the cascade. In effect one has stepwise — rAi
variations in composition as he moves from one where CA0 is the reactant'concentration corre-
CSTR to another. Figure 8.9 illustrates the sponding to zero conversion at the inlet tem-
stepwise variations typical of reactor cascades' perature and pressure.
for different numbers of CSTR's in series. In the Equations 8.3.26 and 8.3.27 are generally
general nonisothermal case one will also en- applicable to all types of CSTR cascades. If one
 /AO

T-l ^

fAO fA, fA2 fAZ

Fraction conversion

Figure 8.9
Schematic representation of reciprocal rate curve for cascade of three
arbitrary size CSTR's.

K , T,

Figure 8.10
Notation for a cascade of N CSTR's in series.

280
8.3 The Continuous Flow Stirred Tank Reactor (CSTR) 281

recognizes that the use of such cascades is systems. In this section we presume a knowledge
almost invariably restricted to liquid systems of the reaction rate expression as a function of
and that in such systems density changes the concentration of reactant A, at least in
caused by reaction or thermal effects are usually graphical terms. That is, we presume
quite small, then additional relations or simpli-
fications can be developed from these starting (8.3.30)
equations. In particular this situation implies
that at steady state the volumetric flow rate where g{CA) denotes some arbitrary function.
between stages is substantially constant. It also For present purposes we assume that the CSTR's
implies that for each reactor, Jt = xb and that all operate at the same temperature and that
the following relation between concentration g{CA) is known at this temperature.
and fraction conversion is appropriate: Equation 8.3.29 may be written in the form

(8.3.28) c —
CAi = CAO(1 - ft) W(i- 1) (8.3.31)
where CAi is the concentration of reactant A
leaving the ith reactor. We will make the as- \n graphical terms this equation indicates that
sumption of constant density throughout the a plot of( — rAi) versus CAi is a straight line with
remainder of Section 8.3.2. We should also recall a slope — 1/ij that cuts the abscissa at C^-^.
that the development of equation 8.3.26 pre- An Analysis of the reactor design problem
sumed steady-state operation and thus this involves the simultaneous solution of equation
assumption is also made throughout Sections 8.3.30 and several equations of the form of
8.3.2.1 to 8.3.2.3. equation 8.3.31 (one for each reactor). These
In view of equation 8.3.28 and these assump- equations are the basis for the solution of the
tions, equation 8.3.27 can be rewritten as two types of problems with which the reactor
designer is most often faced in the analysis of
batteries of ideal CSTR's.

1. What is the final effluent composition from

This equation is_the one that is most appropriate a network of such reactors? (We might also
for our use in the next two subsections, in that require the composition of the stream leaving
we will find it convenient to work in terms of each reactor.)
reactant concentrations instead of conversion > 2. What combination of ideal CSTR's is best
levels. suited to achieving a specified conversion?
Each of these problems will be considered in
8.3.2.1 Graphical Approach to the Analysis of turn. Consider the three ideal CSTR's shown in
Batteries of Stirred Tank Reactors Operating at Figure 8.11. The characteristic space times of
Steady State. Even in reaction systems where these reactors may differ widely. Note that the
it is not possible to determine the algebraic form direction of flow is from right to left. The first
of the reaction rate expression, it is often possible step in the analysis requires the preparation of
to obtain kinetic data that permit one to express a plot o»f reaction rate versus reactant concen-
graphically the rate as a function of the concen- tration based on experimental data (i.e., the
tration of one reactant. Laboratory scale CSTR's generation of a graphical representation of
are particularly appropriate for generating this equation 8.3.30). It is presented as curve I in
type of kinetic data for complex reaction Figure 8.11.
282 Basic Concepts in Reactor Design and Ideal Reactor Models

(I) Curve representing kinetic data

c
Ao
Concentration

Figure 8.11
Plots used in the graphical analysis of cascades of ideal CSTR's.

Now, for the first reactor, equation 8.3.31 through the point where the ordinate is zero
becomes and the abscissa is CA0. However, only the point
CAO CAI of intersection of the straight line with the curve
-r, = ~ (8.3.32) representing equation 8.3.30 has physical mean-
ing. It represents the conditions f hat must
For a specified inlet concentration this equation prevail in the first reactor. Hence the solution
indicates that a plot of —i\ versus CA1 is a to the first part of our problem consists of
straight line with slope (—1/iJ that passes drawing straight line EF through point E with
8.3 The Continuous Flow Stirred Tank Reactor (CSTR) 283

slope ( —1/ij). CAl is the point of intersection bination of CSTR's that is best suited to
of this straight line with curve I, as shown in achieving a specified conversion level. We will
Figure 8.11. begin by considering the case of two arbitrary
Now that CA1 is known, it is evident that a size ideal CSTR's in series operating under
similar process can be used to find CA2 because, isothermal conditions and then briefly treat the
in this case, equation 8.3.31 becomes problem of using multiple identical CSTR's in
series. Consider the two cascade configurations
- r 2 = °A1 ~ CA2
(8.3.33) shown in Figure 8.12, taken from Levenspiel (13).
For the first reactor, equation 8.3.26 becomes
Thus we may construct straight line GH in
Figure 8.11, by drawing a line of slope ( — 1/T2) Ji Jo
(8.3.34)
through the point with an ordinate of zero and AO
an abscissa of CAl. The intersection_of this line while, for the second, it becomes
with curve I gives us the effluent concentration
from the second reactor. This same procedure VR2 f2 ~ h
can be repeated for any other reactors that may FAO (-rA2)
be part of the cascade. The straight line JK was The relations indicated by these two equations
constructed in this fashion for the present case. are shown graphically in Figure 8.12 for two
Let us turn now to the second of the problems alternative configurations. In both cases the
mentioned earlier—determination of the corn- cascade operates between the same initial and

/=o

Rectangular area
measures size of
first unit

•Of second unit

Figure 8.12
Graphical representation of variables for two CSTR's. (Adapted from Chemical Reaction
Engineering, Second Edition, by O. Levenspiel. Copyright © 1972. Reprinted by permission
of John Wiley and Sons, Inc.)
284 Basic Concepts in Reactor Design and Ideal Reactor Models

final conversion levels. The same arbitrary reac-

tion rate expression is appropriate in both
instances. As the composition of the effluent
from the first reactor is changed, the relative
size requirements for the two individual reactors
also change, as does the total volume required.
The size ratio is determined by the ratio of the
two shaded areas and the total volume by the
sum of these areas. The total reactor volume is
minimized when rectangle KLMN is made as
large as possible so the problem of selecting the
optimum sizes of the two reactor components
in this sense reduces to that of selecting point
/i so that the area of rectangle KLMN is
maximized. Levenspiel (13) has considered this Fraction conversion
general problem. We need consider only the
Figure 8.13
results of his analysis. Maximization of rectangles applied to find the
For reaction rate expressions oTthe nth-order optimum intermediate conversion and optimum
form it can be shown that there is always one sizes of two CSTR's in series. (Adapted from
and only one point that minimizes the total Chemical Reaction Engineering, Second Edition,
volume when n > 0. This situation is obtained by O. Levenspiel. Copyright © 1972. Reprinted
when the intermediate fraction conversion f1 is by permission of John Wiley and Sons, Inc.)
selected so that the slope of the curve represen-
ting the reaction rate expression at this con- second-order reaction where 99% conversion
version level is equal to the slope of the diagonal is desired, the minimum total volume required
of rectangle KLMN, as shown in Figure 8.13, is only about 3% less than if equal volume tanks
adopted from Levenspiel. Once this conversion had been used. For 99.9% conversion the
level is known, equations 8.3.34 and 8.3.35 may difference is only about 4%. Generally, the
be used to determine the required reactor sizes. savings associated with this small a reduction
As Levenspiel points out, the optimum size in total volume requirements would scarcely
ratio is generally dependent on the form of the be adequate to cover the extra costs of engineer-
reaction rate expression and on the conversion ing, installing, and maintaining two tanks of
task specified. For first-order kinetics (either different sizes. The argument for uniformity in
irreversible or reversible with first-order kinetics tank sizes becomes even stronger when cascades
in both directions) equal-sized reactors should composed of more than two reactors are consid-
be used. For orders above unity the smaller ered. Consequently, except in those rare cases
reactor should precede the larger; for orders where there are compelling reasons to the
between zero and unity the larger reactor should contrary, the reactor designer tends to employ
precede the smaller. Szepe and Levenspiel (14) multiple identically sized CSTR's in working
have presented charts showing the optimum up design specifications. However, it may be
size ratio for a cascade of two reactors as a advantageous to run the various CSTR's at
function of the conversion level for various different temperatures.
reaction orders. Their results indicate that the For the case of multiple equal-sized reactors
minimum in the total volume requirement is an in series, the problem of determining the reactor
extremely shallow one. For example, for a simple sizes necessary to achieve a specified degree of
8.3 The Continuous Flow Stirred Tank Reactor (CSTR) 285

conversion can be solved by a trial and error or, at 25 °C,

procedure. In this case the lines in figures analo- r = 9.92CI + 0.1984CB (A)
gous to Figure 8.11 will all be parallel to one
another. Consequently, one draws a number of where the rate is expressed in kilomoles per cubic
parallel lines equal to the number of CSTR's he meter per kilosecond when concentrations are
or she intends to use, with the first line passing expressed in kilomoles per cubic meter.
through the inlet composition. When the slope Equation A is presented in graphical form as
used provides the necessary match with the curve M in Figure 8.14.
specified final effluent composition, this slope For 87.5% conversion the concentration of
may be used to determine the necessary reactor benzoquinone in the effluent from the last
volume. Illustration 8.7 indicates the use of this reactor in the cascade will be equal to
technique. (1 - 0.875)(0.08) or 0.010 kmole/m 3 .
For the case where the cascade consists of only
ILLUSTRATION 8.7 DETERMINATION OF a single reactor, only a single straight line of
CSTR SIZE REQUIREMENTS FOR the form of equation 8.3.31 is involved in the
CASCADES OF VARIOUS S I Z E S - graphical solution. One merely links the point
GRAPHICAL SOLUTION on curve M corresponding to the effluent con-
Consider the Diels-Alder reaction between centration of benzoquinone with the point on
benzoquinone (B) and cyclopentadiene (C), the abscissa corresponding to the feed concen-
which was discussed in Illustration 8.1. tration. The slope of this line is equal to ( — 1/T)
or ( — i^/VR). In the present instance the slope
B + C -+ adduct is equal to
If the reaction occurs in the liquid phase at
25 C, determine the reactor volume require- (2.976 - 0) x 10- 3 /(0.01 - 0.08)
ments for cascades of one and three identical or
-0.0425 ksec" 1
CSTR's. The rate at which liquid feed is supplied
Thus
is 0.278 m3/ksec. Use the graphical approach
outlined previously. The following constraints 0.278
yVR 6.54 n r
are applicable. 0.0425 0.0425
r = kCBCc with k = 9.92 m3/kmole-ksec For the case where the cascade consists of
C c o = 0.1 kmole/m 3
CB0 = 0.08 kmole/m 3 three identical reactors in series, a trial and error
approach is necessary to determine the required
Conversion desired—87.5% reactor size. One starts from the inlet concen-
Solution tration and draws a line linking this point on
the abscissa with some point J on curve M. One
The graphical approach requires a plot of reac- then draws a straight line parallel to the first,
tion rate versus the concentration of the limiting but passing through the point on the abscissa
reagent (benzoquinone). In order to prepare this corresponding to the benzoquinone concentra-
plot it is necessary to relate the two reactant tion at pointJ. This straight line intersects curve
concentrations to one another. From the initial M at some point K. One then repeats the pro-
concentrations and the stoichiometric co- cedure by drawing yet another parallel line
efficients, through the point on the abscissa corresponding
Cc = CB + 0.02 to the benzoquinone concentration at K. If the
Thus intersection of this straight line with curve M
r = kCB{CB + 0.02) occurs at a reactant concentration of 0.010
286 Basic Concepts in Reactor Design and Ideal Reactor Models

120
1 I I I I
- Solution for single CSTR
-Solution for cascade of three CSTR's
~ 100 -

*E 80
3

60

£ 40

20

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09

Benzoquinone concentration (kmoles/m3)

Figure 8.14
Graphical solution for cascade of CSTR's.

kmole/m 3 , our initial choice of slope was correct. quite generally applicable tp the steady-state
If not, one must choose a new point J and repeat analysis of CSTR cascades, it is not highly
the procedure until such agreement is obtained. accurate in numerical terms, particularly when
Figure 8.14 indicates the construction for this the generation of the curve representing the
case. The slopes of the straight lines in this reaction rate expression involves graphical or
figure are equal to (18.2 - 0) x 10" 3/(0.034 _ 0.08) numerical differentiation of rate data. Algebraic
or —0.396 ksec" 1 . Thus the volume of each rhethods are capable of greater accuracy than
individual CSTR is equal to 0.278/0.396 or graphical ones if the functional form and con-
0.70 m 3 . The combined volume of all three stants involved in the reaction rate expression
CSTR's is thus 2.1 m 3 , which is a volume re- are knpwn. In this subsection we will again
duction of more than a factor of three when consider a cascade of CSTR's where the reactor
compared to the single CSTR case. In Section volumes are not necessarily equal Jto one another.
8.3.2.3 we will see that such large volume re- We will again use the nomenclature shown in
ductions are typical of the use of cascades of Figure 8.10 and start from equation 8.3.29,
CSTR's. rearranged as follows.

CAi - (8.3.36)
8.3.2.2 Algebraic Approach tq the Analysis of
Batteries of Stirred Tank Reactors Operating at Starting with the first reactor and using algebraic
Steady State. Although the graphical ap- iteration, we proceed to analyze the cascade for
proach presented in the previous subsection is particular forms of the reaction rate expression.
8.3 The Continuous Flow Stirred Tank Reactor (CSTR) 287

For example, in the case of first-order kinetics 8.3.36 as

ii + klCA(ti = CMl-1) (8.3,.37)

-AN =
CAN C
AO
— ^AO ~~ Z^ (kiTi) 3.3.43)
i= 1
Thus CAV-i)
CM
(8.3..38) which is again appropriate for nonisothermal,
~ 1 + kn
or nonequal volume cases. If the cascade is iso-
thermal and all reactor volumes are equal,
(8.3 .39)
1 + k.r, CAN = CAn - Nkz (8.3.44)
and
where T is the space time or mean residence time
CAO for an individual CSTR.
CA, =
/C22T 2 k2T2) Consider now the general second-order case
(8.3.40) where equation 8.3.36 becomes
Finally, (8.3.45)
CA(N-l) or

CM =
CAO
(1 + + /C2T2) • • • (1 + kNTN) when we have discarded the negative root be-
(8.3.41) cause reactant concentrations cannot be
negative. Thus
These relations are valid regardless of whether
the reactors all operate at the same temperature
or at different temperatures. In the case where (8.3.47)
the cascade is isothermal and all reactors have
the same size, - 1 + yfl
C2 =
CAO
(8.3.42)
(1 + kxf -1+ /l+4/c 2 T 2
where T is the space time for an individual 2k?2<<2
T
CSTR. (8.3.48)
One may use the same general approach when
the reaction kinetics are other than first-order. and so on. Although a general expression for
However, except in the case of zero-order CAN would be rather complex algebraically,
kinetics, it is not possible to obtain simple closed involving N nested square roots, the solution for
form expressions for CAN, particularly if unequal the isothermal case is readily obtained graph-
reactor volumes are used. However, the numer- ically by the procedure outlined in the previous
ical calculations for other reaction orders are subsection. A number of other graphical solu-
not difficult to make for the relatively small tions have been discussed in the reactor design
number of stages likely to be encountered in literature by various individuals (15-18).
industrial practice. The results for zero-order Illustrations 8.8 and 8.9 indicate how the
kinetics may be determined from equation techniques developed in this and the previous
288 Basic Concepts in Reactor Design and Ideal Reactor Models

section may be used in the design analysis of This value is considerably larger than that
cascades of stirred tank reactors. calculated in Illustration 8.7 for a nonstoichio-
metric feed ratio, thus indicating the potential
ILLUSTRATION 8.8 DETERMINATION OF desirability of using an excess of one reagent
REACTOR SIZE REQUIREMENTS FOR A when it appears to be a positive power in the
CASCADE OF CSTR's—ALGEBRAIC rate expression. In any economic analysis of a
APPROACH process, however, the costs of separation and
recovery or disposal of the excess reagent must
Consider the Diels—Alder reaction between be taken into account.
benzoquinone (B) and cyclopentadiene (C), dis-
cussed in Illustrations 8.1 and 8.7. Case II—Two Identical CSTR's in Series
B + C -> adduct In this case it will be necessary to determine the
If one employs a feed containing equimolal con- concentration in the effluent from the first
centrations of reactants, the reaction rate ex- reactor in order to determine the required reactor
pression can be written as size. One way of proceeding is to write the design
equation for each CSTR.
r = kCcCB = kCB
CBO(/BI -0)
Determine the reactor size requirements for (A)
/Bl)2
cascades composed of one, two, and three iden-
tical CSTR's. Use an algebraic approach and CB0(/B2 "
assume isothermal operation at 25 °C where the (B)
/cCg O (l —
reaction rate constant is equal to 9.92 m 3 /
kmole-ksec. Reactant concentrations in the feed For identical reactors equations A and B may
are equal to 0.08 kmole/m 3 . The liquid feed rate be combined to give
is equal to 0.278 m3/ksec. The desired degree of
JB1 JB2 JB
conversion is equal to 87.5%.
(1 - (1 " hl
Solution For fB2 = 0.875,
For the specified degree of conversion the
effluent concentration of benzoquinone must be
(1 - 0.875)2(/*i) = (0.875 - - fB1)2
equal to (1 - O.875)(O.O8) = 0.01 kmole/m 3 . This equation may be solved by trial and error
Case I—Single CSTR or by graphical means to determine the com-
position of the effluent from the first reactor.
In this case equation 8.3.45 can be solved for the Thus fB1 = 0.7251 and CBl = 0.08(1 - 0.7251) =
reactor space time directly. 0.02199 kmole/m 3 . Either equation A or equa-
Q40 ~ CAI tion B may now be used to determine the re-
T =
kC2 quired reactor size. Hence

or (0.278)(0.08)(0.7251)
0.08 - 0.01 VR1 = = 3.36 m 3
9.92(0.08)2(l - 0.7251):
y
R\
= 70.56 ksec
9.92(0.01)2
The total volume of the two reactors is 6.72 m 3 ,
Thus
which is considerably less than half that required
VR1 = 70.56(0.278) = 19.6 m 3 if only a single CSTR is employed.
8.3 The Continuous Flow Stirred Tank Reactor (CSTR) 289

Case III—Three Identical CSTR's in Series This equation may be solved numerically by trial
and error, recognizing thalfBl must lie between
In this case there are two intermediate unspec-
0 and 0.875. The appropriate value of fB1 is
ified reactant concentrations instead of just the
0.6285. Equation F may now be used to deter-
single intermediate concentration encountered
mine that fB2 = 0.8038. With a knowledge of
in Case II. At least one of these concentrations
these conversions, equation C, D, or E may be
must be determined if one is to be able to ap-
used to determine the required reactor volume.
propriately size the reactors. In principle one
Thus
may follow the procedure used in Case II where
the design equations for each CSTR are written 0.278(0.08)(0.6285)
VR1 = ~ = 1.60 m 3
and the reactor space times then equated. This 9.92(0.08)2(l - 0.6285)
procedure gives three equations and three un-
knowns (VR1, fBU and fB2). Thus, for the first The total volume of the cascade is then 3 (1.60)
reactor, or 4.8 m 3 , which is again a significant reduction
in the total volume requirement but not nearly
/-,. as great as that brought about in going from one
kC2B0(l - fBl)2 to two CSTR's in series.
For the second, Obviously this approach is not easily extended
to cascades containing more than three reactors
2- fBl) and, in those cases, an alternative trial and error
(D)
- fB2)2 procedure is preferable. One chooses a reactor
volume and then determines the overall fraction
and, for the third, conversion that would be obtained in a cascade
of N reactors. When one's choice of individual
VR3 ^ C BO (0.875 - fB2)
(E) reactor size meets the specified overall degree of
i/- kC2 (\ — 0 RIS}2
conversion, the choice may be regarded as the
Combining equations C and E gives desired solution. This latter approach is readily
amenable to iterative programming techniques
0.875 - fB2 fB1 using a digital computer.
(0.125)2 (1 - far)2
or
(0.125)2/B1 ILLUSTRATION 8.9 DETERMINATION OF
/ „ = 0.875 - (F) OPTIMUM REACTOR SIZES FOR A
A
v ~~ /BI)
CASCADE OF TWO CSTR's
Combining equations C and D gives
Consider the Diels-Alder reaction between
JBI _ JB2 ~ JBI benzoquinone (B) and cyclopentadiene (C) dis-
(G) cussed earlier in Illustrations 8.1, 8.7, and 8.8.
(1 - fBl)2 " (1 - fBl)2
Now fB2 may be eliminated from the last B + C -> adduct
equation by using equation F. We wish to determine the effect of using a
2 cascade of two CSTR's that differ in size on the
(0.125) / B1
0.875 - -ft Bl volume requirements for the reactor network.
fB (1 - fBl)2 In Illustration 8.8 we saw that for reactors of
2
(1 - fm)2 (0-125) /fll equal size the total volume requirement was
1 - 0.875 +
(1 - fBi)2 6.72 m3. If the same feed composition and flow
290 Basic Concepts in Reactor Design and Ideal Reactor Models

rate as in the previous illustration are employed Thus,

and if the reactors are operated isothermally
0.02224(0.702)
at 25 °C, determine the minimum total volume Ri = 2.77 m 3
required and the manner in which the volume ~ (9.92)(0.08)2(l - 0.702)2
should be distributed between the two reactors. 0.02224(0.875 - 0.702)
An overall conversion of 0.875 is to be achieved. m
3
(9.92)(0.08)2(l - 0.875)2
Solution and
For the conditions cited the reaction rate ex- + VR2 = 2.77 + 3.88 = 6.65 m 3
pression is of the form
This total differs from that for equal-sized
r = kC2E reactors by only 0.07 m 3 or approximately 1%.
where k = 9.92 m 3 /kmole-ksec. The benefits that would ensue from this small
From the basic design relationship for a change would probably be far outweighed by
CSTR (8.3.4), the disadvantages associated with having re-
actors of different sizes, such as the costs of
y
Rl ftB\ engineering design, construction, fabrication,
(A)
kCB0(l — and inventory of spare parts. In general, the
and economics associated with using reactors of
different sizes are offset by the concomitant dis-
Y
R2 fB2 - ftBl 0.875 - fB1 advantages. For further treatment of the prob-
FB0 2
kC B0(\ - fB2)2
Ul - 0.875)2 lem of optimization of a two-tank CSTR cas-
(B) cade, consult the papers by Crooks (19) and
Denbigh (20).
Thus the total reactor volume in the cascade is
given by

V -4- VV
F Rn
gu rI fm
jB1 '0.875 -
Rl + R2 ~— kC~2 7j 7~x2 8.3.2.3 Size Comparisons Between Cascades of
(0.125)2
Ideal Continuous Stirred Tank Reactors and Plug
(Q Flow Reactors. In this section the size require-
It is this sum that we desire to minimize. The ments for CSTR cascades containing different
easiest approach to finding this minimum is to numbers of identical reactors are compared with
plot the quantity in brackets versus fB1. The that for a plug flow reactor used to effect the
minimum in this quantity then gives the mini- same change in composition.
mum total volume, and the value of fB1 asso- One may define a space time for an entire
ciated with the minimum may be used in cascade (ic) in terms of the ratio of the sum of
equations A and B to determine the optimum the component reactor volumes to the inlet
distribution of the total volume between the volumetric flow rate. Hence
two reactors. The minimum occurs when fB1 =
0.702.
Now vRl
T, = = I T,- (8.3.49)
FBo = CBOro = (0.08)(0.278)
- 0.02224 kmole/ksec If all component reactors have the same volume
8.3 The Continuous Flow Stirred Tank Reactor (CSTR) 291

and thus the same space time (T),

(8.3.57)
Tc = NT (8.3.50) A out

These relations support our earlier assertion

For a first-order reaction, we showed that for that for the same overall conversion the total
a cascade composed of equal-sized reactors, volume of a cascade of CSTR's should approach
equation 8.3.42 governed the effluent composi- the plug flow volume as the number of reactors
tion from the nth reactor. in the cascade is increased.
C4 1 For a finite (low) number of CSTR's in series,
(8.3.51) equation 8.3.52 can be rewritten as
CA0 (1 + kzf
where T is the space time for an individual re-
actor. Combining equations 8.3.50 and 8.3.51
and rearranging gives The ratio of equations 8.3.58 and 8.3.57 gives the
relative total space time requirement for a cas-
(8.3.52) cade of stirred tank reactors vis a vis a plug flow
CAN N reactor.
If the right side of this equation is expanded in L
c for N reactors L
c for N reactors
a binomial series, c
cfor JV=oo ''plug flow

N(N - 1) r o
N - 1
AN 2! CAAN J cascade
N(N - 1)(JV - 2) //cr CA0
£n
C I
^ A out/ plug flow
(8.3.53) (8.3.59)

and if ekXc is expanded in series form, If the effluents from the two streams are to be
identical and if equimolal feed rates and com-
positions are employed, the ratio of space times
2! 3! becomes equal to the ratio of total volume re-
(8.3.54) quirements. Thus, for constant density systems
where CAout = CAN = CAO(1 - fA),
a term by term comparison in the limit as N
approaches infinity indicates that
^ Total, cascade
r..
CA
kT
e < as N -> oo (8.3.55) -fA)
(8.3.60)
or
Figure 8.15, reproduced from Levenspiel (21),
as N -> oo (8.3.56) is in essence a plot of this ratio versus the frac-
tion conversion for various values of N, the
This relation is identical with that which number of identical CSTR's employed. The
would be obtained from equation 8.2.10 for a larger the value of N, the smaller the discrepancy
plug flow reactor with first-order kinetics. in reactor volume requirements between the
292 Basic Concepts in Reactor Design and Ideal Reactor Models

30
First—order
reaction

* ' * 0
>
\ ,
10
1

^ •

. > <

AT ^^*****"
^ ^
<
——
— ^

— 1

—— - --I -

~N = ^ "o s
1.0
0.01 0.1
= C /C
A A0

Figure 8.15
Comparison of performance of a series of N equal-size CSTR reactors with
a plug flow reactor for the first-order reaction

R, = 0

For the same processing rate of identical feed the ordinate measures the
volume ratio VWCSTR -S/^PFR or the space time ratio T WCSTRS /T PFR directly. (Adapted
from Chemical Reaction Engineering, Second Edition, by O. Levenspiel.
Copyright © 1972. Reprinted by permission of John Wiley and Sons, Inc.)

CSTR cascade and a PFR reactor. Note how with the greatest change occurring on addition
rapidly PFR behavior is approached as N in- of a second tank.
creases. Levenspiel has also included lines of Although the major thrust of the material
constant kz on this figure, and these lines may be presented in this subsection has concerned the
useful in solving certain types of design prob- relative size requirements for CSTR cascades
lems, as we will see in Illustration 8.10. and plug flow reactors, the practicing chemical
Levenspiel (22) has prepared a similar plot for engineer will be more concerned with the relative
second-order reactions (Figure 8.16). It is based economics of the two alternative reactor net-
on a generalization of equation 8.3.46 for N work configurations. In this regard it is worth
identical reactors in series and the integral form repeating that the additional capital costs asso-
of the plug flow design equation for second- ciated with the larger volumes of CSTR's are
order kinetics. We again see that increasing the relatively small, particularly when the units are
number of reactors in the cascade causes the designed for operation at atmospheric pressure.
total volume discrepancy between the cascade Consequently, a plot of the total costs associated
and a plug flow reactor to diminish rapidly, with the use of N reactors versus the number N
8.3 The Continuous Flow Stirred Tank Reactor (CSTR) 293

100
X Second—order
xX C\ •
reaction

\
V
\

\
/
/ i 4>
10
/ / >

\s
y X
y y

7? . f / X <*>
r /
yt /
^^
^.

"^
— i«
—-
•»»

T-
i
/ x^
\
• • — I

y>\fN = °° (plug flow) s

1.0
0.01 0.1 1.0
CA/CA0

Figure 8.16
Comparison of performance of a series of N equal-size CSTR reactors with a
plug flow reactor for elementary second-order reactions

2A -> products
A + B -> products, CA0 = Cfl0

with negligible expansion. For the same processing rate of identical feed the
ordinate gives the volume ratio 1^NCSTRS/^PFR or the space time ratio TNCSTRS^PFR
directly. (Adapted from Chemical Reaction Engineering, Second Edition, by O.
Levenspiel. Copyright 1972. Reprinted by permission of John Wiley and Sons,
Inc.

will usually look something like that shown in the cascade. Consequently, the reactor designer
Figure 8.17. One obtains an economic trade-off must consider cascades containing different
between the costs associated with the high vol- numbers of reactors in the search for an eco-
ume requirements when very few reactors are nomic optimum.
employed and the additional engineering, fabri- The following illustration indicates how Fig-
cation, installation, and maintenance costs ures 8.15 and 8.16 are used in handling simple
incurred by using a larger number of reactors in reactor design calculations.
294 Basic Concepts in Reactor Design and Ideal Reactor Models

1. What degree of conversion can be obtained

in the tubular reactor?
2. What reactor size would be required to
achieve the same conversion if a single ideal
CSTR were employed?
3. What degree of conversion would be obtained
in a single CSTR equal in size to the tubular
reactor?
Cost
4. If two identical ideal CSTR's in series are
employed (each with a volume equal to that
determined in part 2), by how large a factor
can the flow rate of the feedstream be in-
creased while maintaining the conversion
level constant at the value used in parts 1 and
2?
5. If one employs these same two ideal CSTR's
in series and maintains a constant feed rate,
1 2 3 4 5 6 7
what conversion is achieved?
Number of tanks
Solution
Figure 8.17
Interaction between the cost decline associated For an equimolal feed the reaction rate expres-
with reduced total volume and cost increases sion can be written as
arising from design, construction, installation, and
maintenance of a number of reactors.

A. Conversion for Plug Flow

The reactor space time is given by
ILLUSTRATION 8.10 USE OF THE DESIGN
CHARTS FOR COMPARISON OF 2.2
T = = 19.80 ksec
ALTERNATIVE REACTOR NETWORKS 0.1111
Consider the Diels-Alder reaction between ben- Consequently, the characteristic dimensionless
zoquinone (B) and cyclopentadiene (C) discussed rate group for the second-order reaction and
in the last three illustrations. initial conditions is given by
B + C -+ adduct kCB0T = (9.92)(0.1)(19.80) = 19.6
At 25 °C the reaction is first-order in each In Figure 8.16 the line corresponding to plug
reactant with a rate constant of 9.92 m 3 / flow is that where VNCSTR.JVPFK = 1. The inter-
kmole-ksec. A feed stream containing equimolal section of this line and the line for a value of
quantities of B and C (0.1 kmole/m 3 ) is to be kCB0T = 19.6 gives the desired degree of con-
processed at a rate of 0.1111 m3/ksec. A tubular version—95%.
reactor (assume plug flow) with an effective
volume of 2.20 m 3 is to be employed in the B. Size Requirement for CSTR for Identical
processing operation. Processing Task
Use the design charts in Figures 8.8, 8.15, and For the same feed rate and initial concentrations
8.16 to determine the following. the ordinates of Figures 8.8 and 8.16 reduce to
8.3 The Continuous Flow Stirred Tank Reactor (CSTR) 295

the volume ratios for the two reactors. Hence, at E. Increase in Conversion Arising from the Use
the same fB, we see that VCSTR/VPFR = 20. Thus of a Cascade of Two CSTR's at a Specified Feed
K:STR = 20(2.20) = 44 m 3 . Rate
If the feed rate is maintained constant while the
C. Conversion in a CSTR of the Same Size as number of reactors is doubled, the overall space
the Tubular Reactor time for the cascade will double. In the present
For a CSTR equal in volume to the tubular case kCB0TN = 2 = 2(350) = 700. From Figure
8.16, at this value of the dimensionless group
reactor, one moves along a line of constant
and N = 2, it is seen that fB = 0.99.
kCBOr in Figure 8.16 in order to determine the
conversions accomplished in cascades composed
of different numbers of reactors but with the
same overall space time. The intersection of the 8.3.2.4 Analysis of CSTR Cascades under Non-
linefcCfloi = 19.6 and the curve for N = 1 gives steady-State Conditions. In Section 8.3.1.4 the
fs = 0-80. equations relevant to the analysis of the transient
behavior of an individual CSTR were developed
D. Increase in Processing Rate Arising from the and discussed. It is relatively simple to extend the
Use of a Cascade of Two CSTR's at a Specified most general of these relations to the case of mul-
Degree of Conversion tiple CSTR's in series. For example, equations
The values of the group kCBOz that correspond 8.3.15 to 8.3.21 may all be applied to any indi-
to 95% conversion and one or two CSTR's in vidual reactor in the cascade of stirred tank re-
series may be determined from Figure 8.16. They actors, and these relations may be used to
are approximately 350 and 70, respectively. analyze the cascade in stepwise fashion. The
Thus, difference in the analysis for the cascade, how-
ever, arises from the fact that more of the terms
*r = 2 =2 70 in the basic relations are likely to be time variant
- 0.20 when applied to reactors beyond the first. For
(kCB0T)N=i 350
example, even though the feed to the first re-
Since actor may be time invariant during a period of
nonsteady-state behavior in the cascade, the
feed to the second reactor will vary with time as
and
the first reactor strives to reach its steady-state
condition. Similar considerations apply further
= 2-
TN=1 downstream. However, since there is no effect of
variations downstream on the performance of
then upstream CSTR's, one may start at the reactor
where the disturbance is introduced and work
= 10
o.2O downstream from that point. In our generalized
notation, equation 8.3.20 becomes
Consequently, the processing rate for the cas-
cade will be an order of magnitude greater than -Ai
that for a single CSTR. Note that operation of (8.3.61)
dt
the two reactors in parallel would have merely
doubled the processing capacity. Hence there is where the reaction rate is evaluated at conditions
a very strong case for operating with the units in prevailing in reactor i and where we have pre-
a series configuration. sumed both a constant density system and a