Continuous stirred tank reactor models

Dr. M.J . Willis

Dept. of Chemical and Process Engineering, University of Newcastle.
e-mail: mark.willis@ncl.ac.uk
Tel. 0191 222 7242
Written: November, 1998
Updated: April, 1999; March, 2000
Aims and objectives
Chemical reactors are the most influential and therefore important units that a
chemical engineer will encounter. To ensure the successful operation of a
continuous stirred tank reactor (CSTR) it is necessary to understand their
dynamic characteristics. A good understanding will ultimately enable effective
control systems design. The aim of these notes is to introduce some basic
concepts of chemical reaction systems modelling and develop simulation
models for CSTR's . Non-linear and linear systems descriptions are derived.
Introduction
To describe the dynamic behaviour of a CSTR mass, component and energy
balance equations must be developed. This requires an understanding of the
functional expressions that describe chemical reaction. A reaction will create
new components while simultaneously reducing reactant concentrations. The
reaction may give off heat or my require energy to proceed.
The mass balance (typical units, kg/s)
Without reaction, the basic mass balance expression for a system (e.g tank)
is written:
Rate of mass flow in Rate of mass flow out = Rate of change of mass
within system
Writing the mass balance expression for a stirred tank
Consider a well-mixed tank of liquid (figure 1). The inlet stream flow is F
in
(m
3
/s) with density
in
(kg/m
3
). The volume of the liquid in the tank is V (m
3
)
with constant density (kg/m
3
). The flow leaving the tank is F (m
3
/s) with
liquid density (kg/m
3
).
Table 1 summarises each term that appears in the mass balance.
Figure 1. Mixed Tank of Liquid
Rate of mass flow in Rate of mass flow out Rate of change of
mass within system
inlet flowrate density exit flowrate density
dt
density) d(volume
F
in

in
F
dt
) d(V
Table 1. The terms in the mass balance for the stirred
tank system.
Referring to table 1 the mass balance is,
dt
) d(V
F F
in

(1)
For liquid systems equation (1) normally can be simplified by making the
assumption that liquid density is constant. Additionally as V =Ah then,
dt
Ad(h)
F F
in
(2)
The component balance (typical units, kg/s)
To develop a realistic CSTR model the change of individual species (or
components) with respect to time must be considered. This is because
individual components can appear / disappear because of reaction
(remember that the overall mass of reactants and products will always stay
the same). If there are N components N 1 component balances and an
overall mass balance expression are required. Alternatively a component
balance may be written for each species. A component balance for the j
th
chemical species is,
Rate of flow of j
th
component in rate of flow of j
th
component out
+ rate of formation of j
th
component from chemical reactions
= rate of change of j
th
component
Adding a chemical reaction to the stirred tank model
Assume that the reaction may be described as, A B, i.e. component A
reacts irreversibly to form component B. Further, assume that the reaction
rate is 1
st
order. Therefore the rate of reaction with respect to C
A
is modelled
as,
dt
) d(C
A

A
kC - (3)
The negative sign implies that C
A
disappears because of reaction (the
specification of kinetic expressions describing chemical reactions is
explained in more detail in Appendix 1).
Writing the component balance for the stirred tank model
If the concentration of A in the inlet stream is C
Ain
(moles/m
3
) and in the
reactor C
A
(moles/m
3
). Table 2 summarises the terms that appear in the
component balance for reactant A.
Rate of flow of
A in
Rate of flow of
A out
Rate of change
of A caused by
chemical
reaction
Rate of change
of A inside the
tank
Molecular weight
inlet flowrate
conc. of A
Molecular weight
outlet flowrate
conc. of A
{Conc. of A}
Molecular weight
Molecular weight

dt
A) of Conc. d(volume
M
wA
F
in
C
Ain
M
wA
FC
A
- kVC
A
M
wA
M
wA
dt
) d(VC
A
Table 2. The terms in the component balance for the
stirred tank system with reaction A B (first order
reaction rate).
Note:
(1) The units balance. Each term in the table has units of kg / s. Verifying
this is true may seem trivial but it is always worth checking.
(2) Rate of reaction and reaction order. Reaction rate is proportional to the
concentration of reactants raised to some powers. The actual expression is
obtained by a best-fit to experimental data. There is no necessary
connection between stoichiometry of the reaction and reaction order. The
proportionality constant (k) is temperature dependent. The units of k depend
upon the order of the reaction. This is because reaction rate should always
have the units moles per unit time per unit volume.
(2) Units of the component balance expression. As the molecular weight
term appears in each term of the component balance it may be cancelled.
The component balance will then have the units (mole /s).
Referring to table 2 the component balance differential equation is (in terms
of moles /s),
dt
dC
V
dt
) d(VC
FC F
A A
A in

A
kVC -
Ain
C (4)
The energy balance (typical units, J/s or watts)
Consider the CSTR shown in figure 2. The energy balance expression is
given by:
Rate of energy flow in rate of energy flow out
+ rate at which heat added due to reaction = rate of change of energy
within system
Flow in
(Fin)
Flow out
(F)
Cooling Coil
(Fc, Tc)
Liquid Reaction
A ->B
Figure 2. CSTR with cooling coil removing energy (Q)
Writing the energy balance for the stirred tank model
The reaction (A B) is assumed to be exothermic. A cooling coil is used to
remove any heat generated by reaction. The rate of heat removal by the
cooling coil is Q (J /s or Watts). Fluid specific heat and density are assumed
constant. Table 3 summarises each term that appears in the energy balance.
Rate of flow of
energy into
CSTR
Rate of flow of
energy out of
CSTR
Rate at which
energy is
generated due
to chemical
reaction
Rate of change
of liquid energy
mass flowrate
Specific heat
Temp. diff.
mass flowrate
Specific heat
Temp. diff.
Rate of
consumption of
reactant the
heat of reaction
dt
e) temperatur d(volume
heat specific density

F
in

in
C
p
[T
in
-
T
datum
] FC
p
[T-T
datum
] - kVC
A
H
dt
d(T)
V C
p

Table 3. The terms in the energy balance for the stirred

tank system with reaction A B (first order reaction
rate). The table neglects the expression for the rate of
heat removal using a cooling coil (the mathematical
expressions that describe this are developed later in the
notes).
Note:
(1) Rate of flow of energy in /out. This is given by mass flow (F
in

in
or

F) *
specific heat (C
p
) * Temperature difference (T
in
T
datum
) or (T T
datum
). If the
mass flowrate and the specific heat are constant the datum temperature will
disappear in the energy balance expression (see equation 4).
(2) Heat of reaction. This is the difference in energy required to break the
bonds in the reactants when compared to the energy required to break the
bonds in the products. The heat of reaction (H) is negative for an exothermic
reaction and positive for an endothermic reaction. In other words, if a reaction
is exothermic then heat is given out and if it is endothermic then heat is taken
from the system.
(3) Rate of change of energy (E). When deriving a model that is to be used
to study process dynamics as well as the implementation and testing of
process control strategies, the energy balance is generally posed in terms of
"rate of change of temperature with respect to time". The expression for the
rate of change of energy or heat content of a liquid is obtained as follows
1
,

1
For a rigourous derivation the reader is referred to standard thermodynamic textbooks.
dt
dT
C V
dt
dT
MC ] T [T
dt
d
MC )] T (T [MC
dt
d
dt
dE
p p datum p datum p

Referring to table 3 the energy balance differential equation is (assuming F
in
=
F),
dt
dT
V C
dt
d(VT)
T) F(T
p in

p p
C Q H C
A
kVC - (5)
(4) Arrhenius temperature dependence. The effect of temperature on the
reaction rate k is usually found to be exponential,
k =k
o
e
-E/RT
(6)
where k
o
a pre-exponential (or Arrhenius) factor, E the activation energy, T is
the reaction temperature and R the gas law constant.
Modelling the rate of heat transfer through a cooling coil / jacket
Equation (5) models the rate of heat removal through a cooling coil or jacket
as Q (J / s). Adjusting 'Q' (via manipulation of the coolant flow) will regulate
temperature in a CSTR. Therefore to develop a more realistic model of the
system Q must be related to the flowrate through the coil or jacket (F
c
). To
develop the model a number of assumptions are made:
Density and specific heat of the coolant are constant.
Coolant dynamics are be ignored (they are assumed fast when compared
to the temperature dynamics of the liquid in the CSTR).
The area of the coil multiplied by the overall heat transfer coefficient is
approximated as A
c
U =

c
F

where and are appropriately defined
constants.
The logarithmic mean temperature difference is approximated using an
arithmetic mean.
Given these assumptions, the two expressions that describe the rate of heat
transfer are,

,
_

2
] T [T ] T [T
F Q
) T (T C F Q
c(out) c(in)
c(in) c(out) p c

c
(7a,7b)
The objective is to obtain an expression for Q that relates F
c
, T and T
c(in)
. This
can be obtained by using equation (7a) to eliminate T
c(out)
from equation 7b.
This yields,
p
c(in)
C
F
) T (T F
Q

2
1
c
c
c
F +

+
(8)
Resulting in the following energy balance expression,
dt
d(T)
V
C
F
) T (T F
T) F(T
p
c(in)
in p
c
c
c
p
C
F
H C

+
2
kVC -
1
A
(9)
Modelling the dynamics of the reactor wall
In certain CSTRs (generally small vessels) the wall dynamics can have a
significant effect on the thermal control and stability of a CSTR. If this is the
case then an energy balance expression should be developed describing the
rate of change of wall temperature with respect to time. If it is assumed that
the wall temperature is the same at any point. Table 4 summarises each term
that appears in the energy balance.
Rate of energy
transfer (reactor
to wall)
Rate of energy
transfer (wall to
air / jacket or
insulation)
Rate of change
of metal
temperature
h
i
A[T
-
T
w
] h
o
A
w
[T
w
-T
o
]
dt
) d(T
V C
W
W pW W

Table 4. The terms in the energy balance for the reactor

wall.
Note:
(1) The value of the film coefficients can be obtained from appropriate heat
transfer correlation's.
Referring to table 4 the energy balance differential equation for the wall is,
dt
) d(T
V C ) T (T A h ) T A(T h
W
W pW o w w o w i W
(10)
If the thermal capacity of the wall is significantly lower than that of the vessel,
i.e. (
w
C
pw
V
w
) << (C
p
V) then it may be possible to ignore the dynamic effects
of the wall, i.e make a pseudo-steady-state assumption.
Linearisation
Control system design and analysis is based upon linear process
representations. To obtain an approximate (linearised) model of the CSTR
Taylor series expansion may be used. For a function of one variable the
Taylor series expansion is,
...... ) (
! 2
1
) ( ) ( ) (
2
2
2
+
1
]
1

+
1
]
1

+
ss
x
ss
x
ss
x x
dx
d
x x
dx
d
x x
ss
ss
(11)
where x
ss
is the point at which the linearsiation is made (this is normally the
steady-state operating point). This result generalises for m variables.
Linearisation of CSTR equations
Assume that the reaction may be described as, A B. The reaction rate is
1
st
order. There is perfect level control (F
in
=F). The reaction is exothermic
and heat is removed from the system using a cooling coil. The dynamics of
the coil and the reactor wall are negligible when compared to the liquid in the
vessel. The model describing the rate of change of composition and
temperature in the system is then given by,
dt
dC
V C F
A
A

A o
)VC
RT
E -
exp( k - ) (
Ain
C (12)
dt
d(T)
V
C
F
) T (T F
T) F(T
p
c(in)
in p
c
c
c
p
C
F
H C

+
2
)VC
RT
E -
exp( k -
1
A o
(13)
A linear model will be developed around the steady-state operating point. The
linearisation will be respect to C
A
and T (the output variables), C
A(in)
, T
(in)
and
F (possible system disturbances) and F
c
(the variable that can be used to
regulate the system).
Linearisation of the component balance
The component balance is a function of F T, , C , C
Ain A
i.e.,
dt
dC
V C F F) T, , C , (C
A
A Ain A

A o
)VC
RT
E -
exp( k - ) (
Ain
C (14)
The Taylor series expansion is given by,
' '
Ain
Ain
'
A
A
'
A
F
F
T
T
C
C
C
C dt
dC
ss ss
ss ss
1
]
1

+
1
]
1

+
1
]
1

+
1
]
1

'
(15)
where x =x - x
ss
is used to represent a deviation from the steady-state
operating point. Table 5 summarises the partial differential equations.
A
11
C
11
A
12
C
13
ss
1
]
1

A
C
ss
1
]
1

Ain
C
ss
1
]
1

T
ss
1
]
1

F
ss
1
]
1

+ )
RT
E
exp( k
V
F
o
ss
1
]
1

V
F
ss
1
]
1

A
2
o
)C
RT
E
exp(
RT
E k
ss
1
]
1

V
C C
A A(in)
Table 5. Partial differential equations for the
linearised CSTR component balance expression. (The naming
convention used for the various terms such as e.g. A
11
will be explained later in the notes.
Linearisation of the energy balance
The energy balance is a function of
c A in,
F C T T, F, and ,
dt
d(T)
V
C
F
) T (T F
F) F(T F , C , T T, F,
p
c(in)
in c A in p
c
c
c
p
C
F
H C

+
2
)VC
RT
E -
exp( k - ) (
1
A o
The Taylor series expansion is given by,
'
c
c
'
A
A
'
in
in
'
F
F
C
C
T
T
T
T
F
F dt
dT
ss ss ss
ss ss
1
]
1

+
1
]
1

+
1
]
1

+
1
]
1

+
1
]
1

' '
(16)
where x = x - x
ss
is used to represent a deviation from the steady-state
operating point. Table 6 summarises the partial differential equations.
While the expressions in Table 5 and 6 appear complex, upon substitution of
relevant constants and the steady-state operating conditions each differential
term reduces to a single numeric value. The two linear equations can then be
written in the following general form (which is known as a state-space
description),
Cd Bu Ax x + +
!
(17)
Where x
!
is a vector of differentials, i.e.
T
' '
A
dt
dT
dt
dC
[ ] x
!
, u is a vector of
manipulated variables, i.e. u =[F
c

], d is a vector of disturbance terms, i.e. d =

[C
A(in)
, T
(in)
, F] and A, B and C are matrices whose elements are the partial
differential terms evaluated at the steady-state operting point.
For the reactor model considered, the state-space model is given by,
1
1
1
]
1

1
]
1

+
1
]
1

+
1
]
1

1
]
1

'
'
'
23 22
13 11 '
c
21
'
'
A
22 21
12 11
F
T
C
C C 0
C 0 C
F
B T
C
A A
A A
(in)
A(in)
0
x
!
(18)
C
23
C
22
A
21
B
21
ss
1
]
1

F
ss
1
]
1

in
T
ss
1
]
1

A
C
ss
1
]
1

c
F
ss
1
]
1

V
T T
(in)
ss
1
]
1

V
F
ss
1
1
]
1

)
RT
E
exp( k
C
H) (
o

ss
css
css
in c
css
css
css
T T
1
1
1
1
1
]
1

1
]
1

+
+

2
) (
) )(
2
(
2
V F
F C V
C
F
F
C V
F
p
p p

A
22
ss
1
]
1

T
ss
c
c
c
p
1
1
1
1
]
1

+
+

+
2
V F
F C V
F
)C
RT
E
)exp(
RT
E
k
C
H) (
-
V
F
p
A 2 o
ss

1
(
Table 6. Partial differential equations for the
linearised CSTR energy balance expression.
Studying the stability of a CSTR
There are two methods that may be used to study stability characteristics of a
CSTR. The first is to use graphical analysis to obtain an insight into system
dynamics (Van Heerden, 1953) while the second is to use the linearised
differential equation model and analyse the system characteristics using root
locus analysis. The advantage of the second method is that it provides a
means by which open as well as closed loop system characteristics may
easily be studied.
Stability and state-space models
Taking Laplace Transforms of the state equation (17) yields,
) Cd( ) Bu( ) Ax( x(s) s s s s + + or,
) Cd( ) Bu( x(s) A I s s ] - [s + or,
) Cd( A I ) Bu( A I x(s) s ] - [s s ] - [s
-1 -1
+ (19)
The term ] - [s A I is the characteristic equation. In the complex plane, it is the
position of the roots of this characteristic equation which define the stability
properties of the process. Elementary mathematics will establish that the
roots of ] - [s A I are the eigenvalues of the A matrix. Calculation of
eigenvalues is straightforward in Matlab the command being eig(A).
CSTR case study example #1
2
The inlet concentration of the reactant and product are C
A(in)
and C
B(in)
(mole/l)
respectively while the outlet concentration is C
A
and C
B
(mole/l). The reaction
is reversible with the forward and reverse reaction being first order where the
reaction rates k
1
(s
-1
) and k
2
(s
-1
) depend on the reactor temperature via the
Arrhenius relationship,

,
_

RT
1
E
exp
1
C
1
k ,

,
_

RT
2
E
exp
2
C
2
k
For this system there is no cooling coil or jacket. The manipulated variable is
the inlet temperature (T
(in)
) which may be used to control either temperature
(T) or product composition, C
B
.
(1) Verify that the following mechanistic model description is valid and then
simulate the system.

2
Taken from, Economou, C.G., Morari, M. and Palsson, B.O. (1986) IMC 5,
Extension to non-linear systems, Ind. Eng. Chem. Process Des. Dev., 25, pp 403-411.
B
C
2
k
A
C
1
k
A
C
A(in)
C
dt
A
dC
+

) (
B
C
2
k
A
C
1
k
B
C
B(in)
C
dt
B
dC
+

) (

) (
) (
) (
T
(in)
T
B
C
2
k
A
C
1
k
Cp
H
dt
dT

+

, is specified as the time constant of the vessel and the model parameters
are given by (note that some of the units per litre , l):
Value of system parameters Steady-state
C
A(in)
=1.0 C
B(in)
=0.0 C
1
=5 x 10
3
T
i
=427
=60 H
R
=20920 C
2
=1 x 10
6
T
o
=430
=1 mole / l C
p
=4184 E
1
=41840 C
A
=0.492 mole /
l
R =8.314 E
2
=62760 C
B
=0.508 mole / l
Table 4 Parameter values for CSTR simulation
(3) Determine the stability characteristic of the process around the normal
operating point.
References
(Van Heerden - Ind. Eng. Chem., Vol 45, 1953)
Appendix 1
Terminology used in describing kinetics (rate of reaction)
Moles and Atoms
A mole of a chemical compound means Avogadros number of molecules
(6.0225 x 10
23
molecules per mole). The molecular weight of a compound is
the mass of 1 mole and is equal to the sum of the masses of its atoms,
M =m
k
A
k
(9)
Where m
k
is the number of atoms of the k
th
element in a compound, A
k
is the
atomic weight of the k
th
element, and the summation is over the number of
elements in a compound.
Example
Compute the molecular weight of sulphuric acid (H
2
SO
4
)
K Element Atomic weight,
A
k
1 H 1
2 S 32
3 O 16
M =m
1
A
1
+m
2
A
2
+m
3
A
3
=2(1)+1(32)+4(16)=98 g/mole
Chemical Reactions
Consider the following reaction,
CH
4
+H
2
0 =CO +3H
2
This states that 1 mole of CH
4
reacts with 1 mole of H
2
0 to give 1 mole of CO
and 3 moles of H
2
. The mass of each compound is equal to the number of
moles of the compound multiplied by the molecular weight.
CH
4
H
2
0 CO H
2
No. of Moles 1 1 1 3
Molecular weight (g/mole) 16 18 28 2
Mass (g) 16 18 28 6
The mass of the reactants is 34 g and the products 34 g.
Overall reaction rate,
The overall reaction rate is the rate of change of moles of any component
due to chemical reaction divided by the stoichiometric coefficient,
=(1/v
j
)(dn
j
/dt)
Example
For the reaction,
CH
4
+H
2
0 =CO +3H
2
the rate of production of H
2
in a reactor operated at steady-state is 9 moles
/hr. What is the rate of consumption of H
2
O ?
For the components we have,
Component No. j
CH
4
1
H
2
0 2
CO 3
H
2
4
For H
2
we have,
=(1/3)(dn
4
/dt) =(1/3)* 9 moles /hr =3 moles /hr
For H
2
O we have,
v
2
=dn
2
/dt =3 moles /hr * (-1) =-3 moles /hr
i.e. the rate of consumption of water is 2 moles / hour
Nomenclature and units
Symbol Description Units
F Flow m
3
/s
h height m
Density kg/m
3
A Area of vessel m
2
V Volume m
3
M Mass of liquid in
reactor
kg
t Time s
C Concentration mole/m
3
M
w
Molecular weight kg/mole
k Reaction rate
term
units vary
C
p
Specific heat J /kg/k
T Temperature K
h Film heat transfer
coefficient
W/m
2
K
U Overall heat
transfer
coefficient
W/m
2
K
E Activation energy J
R Gas law constant J /mole/K
H Heat of reaction J /mole
Subscripts
Symbol Description
c Coolant (coil or jacket)
w Wall of reactor