CHf CGf

CHn CGn
CAf CAf CAf = CAn
Feature Report
Engineering Practice CAn CAn

CA(n–1) – CAn
Xn =
CA(n–1)

CSTR Design
CAn–1 CAn–1

for Reversible
CA1 – CA2
X2 =
CA1

CA1 CA1

Reactions
CA0 – CA1
X1 =
CA0

CA0 CA0

CAo CAo
Here, a design approach for continuous Xf =
CA0 – CAf CA0 = M > 1.0 Product:
CA0 CGf and CHf
stirred-tank reactors is outlined for three Plug-flow Multiple backmixed
reactor reactors
cases of second-order reactions
Figure 1. Conversion in plug-flow reactors and
CSTRs for second order reactions is shown here,
Ralph Levine with conversion per stage shown for the CSTR case

M
ultiple CSTRs (continuous
stirred-tank reactors) are
advantageous in situations
where the reaction is slow;
two immiscible liquids are present and
require higher agitation rates; or vis-
cous liquids are present that require
high agitation rates. Unlike in plug-
flow reactors, agitation is easily avail-
able in CSTRs. In this article, batch
C Af C v CHf
and plugflow reactors are analyzed
and compared to multiple CSTRs.
The number of reactors required in
a CSTR system is based on the con-
version for each stage. When the final
stage obtains the fraction of uncon-
verted reactant that is equal to the
desired final value from the plug-flow
case, the CSTR system is complete.
The volumetric efficiency of mul-
tiple CSTRs is expressed as a function
of conversion per stage and gives the
total conversion required. In this ar-
ticle, we will apply this to reversible
second-order reactions.
k 
2nd-order, reversible reactions
r  kF C A2 0 1
X f F
The first case presented here is a
kinetic process requiring a double
C
stage, the number of stages of equal and CH0 are equal to zero, and the fol-
volume, as well as the volumetric ef- lowing expressions are true:
ficiency of the CSTR stages and the
plugflow reactor. CGf  C A 0 X f
The reactor design is developed by (2b)
selecting a conversion in the first stage.
Then, the second-stage conversion is CHf = C A 0 X f
 (2c)
equal to that of the first stage, since
it requires an equal volume. This pro-
C  C
2 2
cedure is continued until the fraction Xf
A0 A0 Gf
of reactant exiting each reactor stage
 (2d)
reaches the desired value in the last
stage, or slightly less than the plugflow
case, as illustrated in Figure 1. The ratio may be different, as, for ex-
The kinetic rate conversion of a re- ample, the concentration of G in the
versible bi-molecular reaction at con- product may be five times that of H.
stant temperature and flowrate is rep- However, we assume that the products
resented by Equation (1). The reaction are equal (CGf = CHf).
is illustrated below (nomenclature is 
defined on p. 49. Expressing reaction rate. The rate
equation can be modified to include
2AjG  H conversion and equilibrium constant
terms. Substituting Equations (1), and
−r = kF C A 2 − kR CG CH
(2d) into Equation (1) give an expres-
(1) sion for rate.
kR  kF K

K C A0 X f
2 2
 (2)
(3a)
C X  C
2 2

component (2A) to be fed to a reactor,  A0 f A0 Af (2a)

¥ ¨ X 2  K 1 ·´
and producing two products (G and  C C C C
r  kF C 2 ¦1 2 X f  © ¸µ
v f
G0 Gf H0 Hf
A0 ¦
H). The design may be calculated for
§ ©
ª K ¸¹µ¶
both CSTR and plug-flow reactors, de- Assume that G and H compounds are
termining the conversion in the first not present in the feed. Therefore, CG0  (3b)
46 Chemical Engineering www.che.com september 2009
 ¨¥ K 1 ´ 2 · V tion (11a), based on each stage having
r  kF C A2 0 ©¦ µ X f 2 X f  1¸ v F A0 e

k C X  ln ¨ª1 X f X e ·¹
(6b) the same volume and conditions.
ª§ K ¶ ¹ 2
¥ X ´ C C Ae C A1 C Ae
(3c) Volume of each CSTR stage. An ex- ¦¦1 1 µµ  A 2 –
At equilibrium, the net reaction rate pression for the first stage of a CSTR § X e ¶ C A1 C Ae C A 0 C Ae
equals zero. is given in Equation (7). The first stage  C C (11a)
conversion, X1, occurs in each of the  A2 Ae

¥ K 1´ 2 C A 0 C Ae
0¦ µ Xe 2 Xe  1 successive stages (X2, X3, and so on), 
§ K ¶ (4a) and each has the same volume and re- Continue this process for the nth stage
action temperature. to obtain the following equations.
Using the quadratic equation, Equa-
V1 X ¥
n
tion (4a) is simplified to Equation (4c).  1 X ´ C C Ae
vC A 0 r (7) ¦¦1 1 µµ  An
 2 4  K 1 X e ¶ C A 0 C Ae
2
2p K §  (12)
Xe
2  K 1 K Substituting Equation (5c) into (7) and
(4b) rearranging gives Equation (7a). ¥ X ´ ¥C C ´
n log ¦¦1 1 µµ  log ¦¦ An Ae
µµ
1 p 1 1 K V1 CA 0 kF X1 X e § Xe ¶ § C A 0 C Ae ¶
Xe  
 K 1 K v 1  X1 X e (7a)  (12a)
(4c)
n
Equation (7a) has only one indepen- ¥ X1 ´ X f C An C Ae
The quadratic equation can also be dent variable (V1). If each stirred reac- ¦¦1 X µµ  1 X  C C (12b)
used to simplify Equation (3c), result- tor stage is to be of equal volume and § e ¶ e A0 Ae

ing in Equation (5a). volumetric flowrate, then the result is

a constant conversion per stage. That Total volume of all stages. Substi-
2 p 4 4 ¨ª1 1 K ·¹ is, each stage, when at a fixed set of tute Equation (7a) into (13a).
Xf 
2  K 1 K conditions, has the same conversion
VT  nV1
(5) from each stage, expressed as: (13a)
X1 = X2 = X3... = Xn
1 p 1 1 K VT V ¨ X X ·
Xf  kF C A 0  n 1 kF C A 0  n© 1 e ¸
 K 1

K
(5a)
Number of stages. Conversion for
stage 1 is expressed by equation (8).
v v ©ª1  X1 X e ¸¹

The reaction rate expression can then (13b)

be expressed as Equations (5b) and X1  C A 0 C A1 C A 0 (8) VT
k C
(5c). v F A0
The equilibrium conversion is based  ¥C C ´ (13c)
(
−r = kF CA2 0 X e − X f ) on time to reach a net reaction rate log ¦¦ An Ae
µµ
§ C A 0 C Ae ¶ X1 X e
(5b) of zero, which may be calculated by  –
Equation (4b) or (9). log ¨ª1  X1 X e ·¹ 1  X1 X e
(
−r = kF CA20 Xe ⎡⎣1 − X f X e ⎤⎦ )
(5c) X e  C A 0 C Ae C A 0 (9) By definition, CAn = CA0(1 – Xf), where
Xf is the total conversion in the nth
Stirred reactor in a batch or plug- Subtract Equation (9) from (8) and di- stage or the final desired conversion of
flow reactor. The batch reactor case vide by (9) to obtain Equations (10a) the plug-flow reactor. By the method
and the ideal continuous plugflow and (10b). used to obtain Equation (10), the fol-
case are given in Equation (6). lowing equation is similarly derived.
The reaction time is t for the batch Substitution of Equation (12b) into
case, and V/v for the plugflow case. e 1  A1 Ae C A0
X X  C C (10)
(13c) gives Equation (13d).
X e − X1 C A1 − C Ae V
= ⌠ dX
V Xf
= T kF C A 0
 ⌡ (6) X (10a)
vC A 0 0
−r e
C A 0 − C Ae  v (13d)

C A1 C Ae 

log ¨ª1 X f X e ·¹
–
X1 X e
Substituting Equation (5c) into (6) and 1  X1 X e  log ¨ª1  X1 X e ·¹ 1  X1 X e
rearranging, gives Equation (6a). C A 0 C Ae (10b)

V Xf dX The exit concentration, CA1, can be Volumetric efficiency

(
k C2 X =
vC A 0 F A 0 e
) ∫ 0
(
1 − X f Xe ) calculated from Equation (10b). Also, Since VT/v in Equation (13c) is resi-
the exit concentration from the second dence time, as is V/v in Equation (6b),
 (6a) stage, CA2, can be calculated from Equa- for CSTRs, these terms are equiva-
Chemical Engineering www.che.com september 2009 47
 Table 1.Volumetric Efficiency for Equation (14a)

Engineering Practice (14a)

X1/Xe Xe 1–(X1/Xe) log[1–(X1/Xe)] (V/VT)Xe V/VT

lent. The volumetric flowrate is the
0.1 0.7 0.9 –0.046 0.948 1.355
same in all cases (a batch operation
for one complete reaction cycle). Thus, 0.2 0.7 0.8 –0.097 0.893 1.275
the ratio of comparison should be V for 0.3 0.7 0.7 –0.155 0.832 1.189
plugflow or batch operation (reaction
0.4 0.7 0.6 –0.222 0.766 1.095
volume and time only) compared to
VT for multiple CSTRs. This ratio (V/ 0.5 0.7 0.5 –0.301 0.693 0.990
VT), volumetric efficiency is expressed 0.6 0.7 0.4 –0.398 0.611 0.873
as Equation (14), and is derived from
0.7 0.7 0.3 –0.523 0.516 0.737
Equations (6b) and (13d).
V 0.8 0.7 0.2 –0.699 0.402 0.575
v kF C A 0 X e 0.9 0.7 0.1 –1.000 0.256 0.366
VT
 k C (14) TABLE 1. For any ratio of conversion per stage to equilibrium conversion, this table
v F A0 provides the corresponding volumetric efficiency, based on Equation (14a)



2.303 log ¨ª1 X f X e ·¹
¥ M 1´ The volumetric efficiency is calculated
X1 X e
–

log ¨ª1 X f X e ·¹ 0  X e 2 ¦¦ µµ X e  M as Equation (23).
1  X1 X e log ¨1  X X · § KCB0 ¶
(18a)
ª 1 e ¹ V
Using the quadratic equation, Equa- 
VT
tion (18a) becomes (18b). (23)
V 2.303 log ª1  X1 X e ¹
¨ · 2.303 ¨1  X 1 X e ·
 ¥ M 1´
2 © ¸ – log ¨ª1  X1 X e ·¹

VT X e  X1 X e
(14a) M 1
p ¦¦ µµ 4 M X e ª©  X 1 X e ¸¹
KCB0 § KCB0 ¶
– log ¨ª1  X1 X e ·¹ Xe 
(18b)
Reversible production of a
2
dimer from twin reactants
Volumetric efficiency is independent Using Equations (17) and (18a), an ex-
In another alternate but similar case,
of the initial or final concentration pression for Xf is found.
Equation (15) is modified for double
and velocity constant at constant tem- ¥ M 1´ components that are reversibly re-
perature, as well as overall conver- r  kF CB2 0 X f2 ¦¦ µµ X f  M acted to form a dimer, as shown in the
sion. It is dependent on only the ratio § KCB0 ¶ (19a)
reaction below. As in this last case,
of the first stage conversion compared Equation (19a) can be simplified to
there is only one product.
to the equilibrium conversion. Calcu- Equation (19b) using the quadratic
lations for Equation (14a) are shown equation. 2Aj P
in Table 1. 2 −r = kF C A 0 2 − kR CP
¥ M  1´ ¥ M 1´ (24)
¦¦ µµ p ¦¦ µµ 4 M
Reversible production of a
§ KCB0 ¶ § KCB0 ¶
(19b) CPf = C A 0 X f
dimer from two reactants Xf  (25)
Another case exists, where two com- 2
 k 
K C A0 X f
2

ponents are reversibly reacted to form r  kF C A 0 2 1 X f F

a single product, a dimer, rather than

two products (as shown in the reaction

r  kF CB2 0 X e ¨ª1 X f X e ·¹ (19c)  (26a)
r  kF C A 0 ª X f  2 KC A 0 X f  1·¹
¨2 2
below). This case is similar to the pre- V1  kF CB0 X e
vious case, but but with only one prod-  ln ¨ª1  X1 X e ·¹(20a)  (26b)
uct, as shown below in Equation (15). v
At equilibrium, the rate is zero.
The volume of each backmixed stage
C  Dj P is equal. 0  X e 2  2 KC A 0 X e  1
(27a)
Using the quadratic equation, Equa-
r  kF CB CD kR C p
(15) V1  kF CB0 X e X1 X e tion (27a) becomes (27b).

CB0 X f  CB0 CBf  C pf
v 1  X1 X e (20b)
 (16a)
1 KC
2
X e  1 KC A 0 p A0
1
C CBe (27b)
C pf  CB0 X f
(16b) 1  X1
X e  B1 (21)
CB0 CBe
1 KC
2
X f  1 KC A 0 p A0
1
¨ Xf · (28a)
CBn CBe
r  kF CB20
©ª

© 1 Xf
M Xf  ¸
KCB0 ¸¹
CB0 CBe

 1 X f Xe r  kF C A 0 2

¨1 X X ·
ª e ¹

f
(28b)
 ¨ª1  X1 X e ·¹
n
 (17) (22) For a plugflow reactor, the following
At equilibrium, the rate is zero. expression is true.
48 Chemical Engineering www.che.com september 2009
 Summary of Equations Subscripts
[6] Nomenclature 0 Initial conditions
B Dj P  S
1,2,3 First, second and third stages
 M  1 p  M  1 4 M 1 K
2
K C Concentration, A For component A
Xe  moles/unit volume
C For component C
2  K 1 K k Reaction rate
constant D For component D

 M  1 p  M  1 4 M 1 K e Equilibrium conditions
2
K K Equilibrium constant
Xf  M Initial mole ratio of D/B f Overall or final conditions
2  K 1 K n Number of stages F Conditions for forward reaction
2AjG  H G For component G
r Reaction rate
1 p 1 1 K t Reaction time H For component H
Xe  V Reactor volume j Any stage in the series of reactor stages
 K 1 K v Volumetric flowrate n The nth stage
1 p 1 1 K X Conversion P For component P
Xf  R Conditions for reverse reaction
 K 1 K T The total of all n stages
C  Dj P
2 References Author
M 1 ¥ M 1´
p ¦¦ µµ 4 M Ralph Levine is a retired
1. Levenspiel, O., “Chemical Reaction Engi- chemical engineer currently
KCB0 § KCB0 ¶ neering,” John Wiley & Sons, Inc., 1962. working as a consultant
Xe  2. Levine, R., Hydro. Proc., July 1967, pp. for plants, design or opera-
2 158–160. tions and R&D (578 Arbor
2 Meadow Dr., Ballwin, Mo.
¥ M 1´ ¥ M 1´ 3. Levine, R. A New Design Approach for 63021; Email: ralphle2000@
¦¦ µµ p ¦¦ µµ 4 M Backmixed Reactors — Part I, Chem. yahoo.com). Levine earned a
Eng. July 1, 1968, pp. 62–67.
§ KCB0 ¶ § KCB0 ¶ B.S.Ch.E. from the City Uni-
Xf  4. Levine, R. A New Design Approach for versity of New York, and did
2 Backmixed Reactors — Part II, Chem. graduate work at Louisiana
Eng. July 29, 1968, pp. 145–150. State University and the Uni-
2Aj P versity of Delaware. Levine later served as an
5. Levine, R. A New Design Approach for engineer for the U.S. Army Chemical Corps. He
1 KC
2
X e  1 KC A 0 p A0
1 Backmixed Reactors — Part III, Chem. has worked for DuPont, Cities Service Co., and
Eng. Aug 12, 1968, pp. 167–171. most recently, Columbian Chemical Co. Levine
has filed several U.S. patents during his career,
1 KC 6. Levine, R. CSTRs: Bound for Maximum
2
X f  1 KC A 0 p A0
1 Conversion, Chem. Eng. Jan. 2009, pp. and is a published author, with his work fea-
30–34. tured in Chemical Engineering and Hydrocar-
CPD-5 8/20/07 10:34 AM Page 1 bon Processing.

V C A 0 kF X e

v

 ln ¨ª1 X f X e ·¹ (29)
The expression for multiple CSTRs is
given as Equation (30). The best way to heat and cool
 V1 C A 0 kF

X1 X e (30) the most corrosive materials.
v 
1 X 1 Xe
¨1  X X ·  1 X X 
n
ª
1 e ¹ f e

 C C (31)
 An Ae

C A 0 C Ae
Nothing controls
The volumetric efficiency is found to temperatures of
be Equation (32). corrosives and high
purity materials better
V ¥ 2.303 ´ ¨1  X1 X e · than AMETEK fluoropoly-
  ¦¦ µµ © ¸ (32)
VT § X e ¶ ©ª  X1 X e ¸¹ mer heat exchangers. Not
glass. Not silicon carbide
– log ¨ª1  X1 X e ·¹
or graphite units. Available
The last two cases presented here are with steel or non-metallic
reversible and have only one product. shells, diameters from 3"
The differences between these cases to 14" and metric designs,
are the values calculated based on AMETEK heat exchangers
the quadratic equation for both Xe include TEMA/ANSI nozzle and
and Xf. All second order reactions that end connections. To learn more
are reversible and produce one or two call (302) 456-4431 or visit:
products require the quadratic equa- www.ametekfpp.com
tion for the calculation of Xe and Xf for
each case. A summary of these equa-
tions is presented in the box above. ■
Edited by Kate Torzewski
Circle 27 on p. 62 or go to adlinks.che.com/23018-27
Chemical Engineering www.che.com september 2009 49