See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/221231879

Modeling, Simulation and Control of Chemical Industrial Reactor

Conference Paper · January 2008

DOI: 10.1007/978-90-481-3656-8_39 · Source: DBLP

CITATIONS READS

0 718

3 authors:

Zdenka Prokopová Petr Silhavy

Tomas Bata University in Zlín Tomas Bata University in Zlín
74 PUBLICATIONS   197 CITATIONS    42 PUBLICATIONS   120 CITATIONS   

SEE PROFILE SEE PROFILE

Radek Silhavy
Tomas Bata University in Zlín
53 PUBLICATIONS   138 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Use Case Points Dataset (Software Development Effort Estimation, Software Size) View project

All content following this page was uploaded by Zdenka Prokopová on 08 February 2017.

The user has requested enhancement of the downloaded file.

 Modeling, Simulation and Control of Chemical
Industrial Reactor
Z. Prokopova, P. Silhavy, R. Silhavy
Tomas Bata University in Zlin,
Nad Stranemi 4511,
760 05 Zlin, Czech Republic
[email protected], [email protected], [email protected]

Abstract – The presented paper is focused on analysis, Design of controllers by algebraic approach became
mathematical modeling, simulation and control of reactor which is a useful and popular discipline in control theory during the
used in the chemical and tanning technology. The contribution last decade. On the beginning the ring of polynomials was
brings complex analysis of continuous-flow circulation reactor for used for control design. All controllers are obtained via linear
the cyclohexane production. A mathematical dynamic model was
Diophantine equations in an appropriate ring. The necessity
derived and the optimal parameters were computed. Then a
comparison of several different methods of controller design was of robust control was naturally developed by the situation
performed. One type of controller was obtained by classical method when the nominal plant (used in control design) differs from
the second one by algebraic approach via solutions of Diophantine the real (perturbed) one. A suitable tool for parameter
equations in the ring of polynomials. The third type was obtained uncertainty is the infinity norm H. Hence, a polynomial
via solutions of Diophantine equations in the ring of stable and description of transfer functions had to be replaced by another
proper rational functions - robust controllers. All simulations were one. A convenient description adopted from [7], [8], [9] is
performed in the standard Matlab-Simulink environment. a factorization approach where transfer functions are
expressed as a ratio of two Hurwitz stable and proper rational
I. INTRODUCTION
functions (Rs). Then conditions of robust stability can be
Material and energy balances are the key issues of easily formulated in algebraic parlance.
mathematical models of chemical reactors and processes. The
II. ANALYSIS AND MATHEMATICAL MODEL
combination with chemical kinetics and transport effects an
intellectual basis for chemical reactor design can be obtained. Presented reactor is a continuous-flow circulation reactor
The behavior of real chemical process is obviously too for the cyclohexane production. Cyclohexane, as an important
complex for a perfect, complete and whole mathematical base material for polyamide manufacturing, makes by
analysis. Various kinds of approximations and simplifications catalytic hydrogenation of benzene. A large number of
are commonly used to obtain results reasonably simple but reaction heat transfers during hydrogenation process.
sufficiently accurate. Cardinal simplifications of the Therefore, benzene and cyclohexane mixture hydrogenation
mathematical description can be utilized by the following proceed in continuous time circulating reactor under
presumptions: hydrogen excess. The presented mathematical model should
- Process does not change its properties during the respect the following conditions:
observation period, - Reaction mixture is perfectly mixed by circulation.
- Relation between input and output variables of the - The volume of reaction mixture is constant.
process is linear, - All technological parameters in reactor are constant.
- Process variables can be measured continuously and - Heat transfer on the both sides of wall is ideal.
have a continuous-time trend. The hydrogenation mechanism of benzene is supposed to
A useful and interesting application of reactor design for be a first-order reaction:
chemical engineers can be seen in [1], [2]. There are
emphases on numerical solutions which are needed for most
practical problems in chemical reactor design. Sophisticated k
numerical techniques and methods are rarely necessary.  3H 2  (1)
The aim of the presented approach is to make the Benzene Cyclohexane
rB  k  cB
techniques understandable and easily accessible to design
engineers and the methodology enables the deeper and
comprehensive focus on the chemistry and physics of the where k is a constant of proportionality known as the rate
process. constant and rB is the rate of the reaction. There is not the
There are many reputable authors and superior texts on rate at which a particular component reacts. Components
chemical reactions and theirs kinetics, engineering and reactor benzene and hydrogen are consumed by the reaction and thus
analysis and design, e.g. [3], [4], [5], [6]. are "formed" at a negative rate. The material balance for the
component benzene satisfies the following differential III. CONTROL DESIGN
equation (for constant volume V of reactor):
Control systems depicted in Fig. 1. and Fig. 2. are
dc considered for design of one degree of freedom – feedback
qv .cBv  q.cB  V .k .cB  V B (2) (FB) and two degrees of freedom – feedback feed-forward
dt
(FBFW) controllers.
v
where qv and q are input and output flow and cBv and cB are
y
input and output concentration of component benzene. w u
The temperature balance in the reactor satisfies the Q B
differential equation: P A
n
qv . v .c pv .Tv  ( H R )V .k .c B 
dT (3)
q. .c p .T  F . (T  Tx )  V . .c p
dt
Fig. 1. Feedback control system
where the relation (H R )V .k.cB is the heat generated by
w
the reaction and F. (T  Tx ) represents the heat transfer into R
surroundings. P v
The rate constant k for elementary reactions is expressed
by Arrhenius equation: Q u B y

P A
E
k  k0T n exp   (4) n
 RT 

where parameter n=0 (or n=0.5, 1 for the special cases of

reactions) depends on used theoretical model. E is the Fig. 2. Feedback - feed-forward control system
activation energy and the fraction E/R is called the activation
temperature. Gw G
Transfer functions w  and v  v represent the
After some mathematical manipulations the equation (2) Fw Fv
of the benzene concentration variation takes the form: reference and disturbance signal, respectively.

E A. Classical approaches
dcB qv q 
  c Bv   c B  k0e RT  c B (5)
dt V V
1) Naslin method
The method is determined for the low degrees transfer
Similarly, variation of the reaction mixture temperature in
B
the reactor is: functions in the form . The controller has defined structure
A
dT qv k .c q in advance. We choose the controller handling that maximal
 T  ( H R ). B  T overshooting not exceed limitations. Procedure can be
dt V v  .c p V
(6) described in the next steps:
F .
 (T  Tx ) -We have the transfer function
V . .c p
Bs 
G s 
K

a n s ...a1 s 1 As 
(7)
The time response of the reaction mixture concentration n
depends on the input flow of the benzene, the concentration
of input stream cBv , the volume of reactor filling, the specific
- The controller has defined structure
heat capacity of reaction mixture cp and the rate constant k.
The solution of ordinary differential equation (5) requires Q s 
C s 
P s 
an initial condition: cBv = c B0 at t=0. (8)
From all mentioned parameters the input flow stream into
the reactor qv and its initial concentration cBv can be - We have to calculate the characteristic polynomial
changed.
AP + BQ = c0+c1s+ ...+cmsm (9)
 - For: ci2  α ci-1 . ci+1, where the relation between α and 2) Robust control
overshooting is defined by table. For maximal In this case linear transfer functions are no more
overshooting y  5% is α = 2.0. represented as a ratio of two polynomial but two elements of
an another ring. For the purposes of robust control in the
The method provide only parameters borders but their
sense of this contribution any transfer function G(s) of a
availability is necessary to test by simulation.
linear system can be expressed as a ratio of two elements:

bs  Bs  a s  bs 

2) Method of optimal module
The method was worked out by Oldenburg and Sartorius. G s   As  ; Bs 
The principle is that the amplitude frequency characteristics a s  As  ms  ms  (13)
have to be flat in the area of lowest frequency. The method is ms  smn ; n  max dega;degb
very simple for the controlled systems with the transfer
function in the form G s 
K Elements A, B,... constitute a sub ring Rm(s) of the ring of
.
a n s n ...a1 s  a 0 all Hurwitz stable and proper rational functions Rs(s). A class
of PID like controllers is generated by first and second order
For n=3 and PI controller there is exists explicit
systems.
formulation of controller parameters:
For m = 0 subring Rm(s) expands to the traditiona1 ring
Rs(s) [7]. An infinity norm (convenient for uncertainty) in
1 a12  a 2 1 a13  2a1a 2  a3 both rings is defined by
r1   , r0   . (10)
2 K a1a 2  a3 2K a1a 2  a3
G  sup G s   sup G  j 
B. Modern approaches Re s 0 E

 
1 (14)
G1 G 2  1 sup G1 s   G 2 s 
G 2 2 2
1) Pole-placement control G 2 Re s 0
Basic signification in the theory of linear control has the
characteristic equation of closed control loop. There are
This norm is the radius of the smallest circle containing
known polynomials of numerator and denominator of
the Nyquist plot of the transfer function. Almost all
controlled system in the characteristic equation but doesn’t
mathematical models differ from physical systems.
known polynomials of controller transfer function. The design
B s 
of controller was reduced for solving Diophantine equations Let G s  
As 
be a nominal plant and consider a family
in the modern interpretation, where:
B s 
of perturbed systems G s  
- a(s) p(s) + b(s) q(s) = c(s) has solution
As 
where
 Highest common divisor HCD (a,b) divide c
- If exists pp(s), qp(s) equation a(s) p(s) + b(s) q(s) = c(s)
A A    1 B  B    2
p s  p p ( s )bs t s  or A A  B  B   
(15)
q s  q p ( s )a s t s 
(11)

The control design in Rm(s) is proposed in [11], [12], [13].

where t(s) is arbitrary polynomial; HCD (a,b) = 1 For the given nominal plant G(s) in the form of (13) it
- Particular solution we can obtain by e.g. indefinite consists of the following steps:
coefficient method.
Solution: Step 1: All stabilizing (feedback) controllers are given by all
- Estimation of degrees solutions of the equation:

deg p  deg b1 AP  BQ  1 (16)

degc dega deg b
degq  dega 1 (12)
deg p  degcdeg b in the parameter form expressed by the fraction:
degq  dega 1 degc  dega deg b

Q Q0  AT
 (17)
- Calculation of particular solution: By comparison of left P P0  BT
hand side and right hand side of Diophantine equation we
can obtain set of linear algebraic equations. where P0,Q0 X Rm(s) are particular solutions and T is arbitrary
ps  p p ( s )bs t s  in Rm(s).
qs  q p ( s )a s t s  
- Calculation of all solutions: t(s) is
 Step 2: Feedback controllers (if necessary) are given in a
arbitrary polynomial. similar way by all solutions of
 Fw S  BR  1 (18) The approximation was obtained by a four-parameter
method adopted from [13] applied to step responses.
Step 3: Compute P X Rm(s) such that
1.2
Fw divides P for asymptotic tracking problems (19) 1
Fv divides P for disturbance rejection problems
1

Step 4: For perturbed plants choose such P, Q in which fulfils 0.8

the conditions

y(t) [kmol/m3]
2
0.6
 1 P0  BT  2 Q0  AT 1
P0  BT (20) 0.4
or  1
Q0  AT 3
0.2

4
Steps 1 - 3 represent the first and simplest case of robust 0
5
design. All solutions of Diophantine equations in Rm(s) are
expressed as functions of the parameter m> 0. The va1ue of -0.2
0 5 10 15 20
this parameter strongly influences dynamics as well as t [min]
robustness of the proposed control system. Fig. 3. The output y time responses to input u step changes,
u=-0.08(1), -0.07(2), -0.05(3), 0.05(4), 0.08(5)
IV. SIMULATION OF MATHEMATICAL MODEL

The continuous-flow circulation reactor for catalytic V. RESULTS OF CONTROLLED SYSTEMS SIMULATION
hydrogenation of benzene with V=0,5 m3 volume was used
for the simulation of the concentration and temperature One of the main problems of the catalytic hydrogenation
characteristics. Initial input flow into the reactor was of benzene in continuous-flow circulation reactor is
qv=0.0067 m3s-1 with the initial concentration of benzene progressive degeneration of reaction surfactant. This is cause
that the reaction is slowing down, the rate constant k is
cBv =1.9 kg/m3. Initial rate constant was considered k=
decreasing and the concentration of benzene in output mixture
1.616.e14 s-1, density of mass =985 kg m-3, specific heat is increasing. The aim is to keep the concentration of benzene
capacity cp = 4.050 J/kg.K, transfer heat coefficient E= under the critical value. One of the possibilities how to solve
435.00 W/m2.K, surface F=5.5 m2, activation energy E=4.8e4 the problem is reduction of input flow q of the benzene and
J mol-1, initial mass temperature Tv=320 K and temperature of cyclohexane mixture into the reactor.
surrounding Tx=290 K.
A. Classical control algorithms
Steady-State Analysis Results
The reactor steady-state characteristics were obtained by Simulation 1: Naslin method, maximal overshooting 5% (α =
solution of equations (4)-(6). The initial conditions were 2.0), set point w=0.002.
computed by a standard optimization method - iterative Naslin method
procedure. 0.05

0.045
Dynamic Analysis Results
0.04
For dynamic analysis purposes one input and one output
choices were considered. Then were defined input u and 0.035

output y as deviations from their steady-state values – u(t) =

output

0.03
q(t) – qs(t) and y(t) = cB(t) – cBs(t). 0.025
Graphical interpretation of simulation experiments – the
output y time responses to input u step changes are depicted 0.02

on Fig. 3. 0.015

0.01
Model approximation
For control requirement, described continuous time 0.005

circulating reactor was approximated by the second order 0

0 5 10 15 20 25 30 35 40
system in the form: time

Bs  Fig. 4. Results of Naslin method

H s 
1.9

71s 10 s0.5 As 
2
(21)
Simulation 2: Method of optimal module, set point w=0.002. Simulation 5: Robust approach, FB, m=10, set point w=0.002.
x 10
-3 Robust approach - FB m=10
Method of optimal modul 4
0.6

3.5
0.5

3
0.4
2.5

output
0.3
output

0.2
1.5

0.1 1

0 0.5

0
-0.1 0 5 10 15 20 25 30 35 40
0 5 10 15 20 25 30 35 40 time
time
Fig. 8. Results of robust approach – FB, m=10
Fig. 5. Results of Method of optimal module

B. Modern control algorithms Simulation 6: Robust approach, FBFW, m=2, set point
w=0.002.
Simulation 3: Polynomial approach, set point w=0.002. x 10
-3 Robust approach - FBFW m=2
4
x 10
-3 Polynomial approach
7
3.5

6
3

5
2.5
output

4 2
output

3 1.5

1
2

0.5
1

0
0 5 10 15 20 25 30 35 40
0 time
0 5 10 15 20 25 30 35 40
time
Fig. 9. Results of robust approach – FBFW, m=2
Fig. 6. Results of polynomial approach
Simulation 7: Robust approach, FBFW, m=10, set point
Simulation 4: Robust approach, FB, m=2, set point w=0.002. w=0.002.
-3 Robust approach - FBFW m=10
x 10
-3 Robust approach - FB m=2 x 10
4
4

3.5
3.5

3 3

2.5
output

2.5
output

2 2

1.5 1.5

1 1

0.5 0.5

0 0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
time time

Fig. 7. Results of robust approach – FB, m=2 Fig. 10. Results of robust approach – FBFW, m=10
 CONCLUSION

There are several different methodologies presented in the

paper. The main aim was to compare classical and modern
approaches to control design of one a bit special technological
process. Controllers obtained by classical methods have an
advantage in simple deduction of controller’s parameters but
control behaviors from the system have relatively big
overshooting. The polynomial approach is much better, but
the calculations of controller’s parameters are more
complicated and controlled outputs are very dependent on the
approximation model, which we used for controller design.
The robust methodology enables to tune and influence the
robustness and control behavior by a single scalar positive
parameter m. It means that control output is not so dependent
on the approximation model of the system. We can see (Fig.7
– Fig.10) that the control behavior is improving with the
increasing of parameter m value. For m=2 is overshooting
bigger than we are able to accept but for m=10 is
overshooting minimal. The control behavior is a little bit
better when we use two degree of freedom control loop
(feedback and feedforward control system).
The design methods were developed for SISO systems
generally. All computations and simulation were performed in
the MATLAB+SIMULINK environment.

ACKNOWLEDGMENT

This research has been supported by the Ministry of

Education of the Czech Republic in the range of the project
No. MSM 7088352102. This support is very gratefully
acknowledged.

REFERENCES
[1] F. Froment, and K. B. Bischotf, Chemical Reactor Analysis and
Design, 2nd ed., Wiley, New York. 1990.
[2] M. B. King, and M. B. Winterbottom, Reactor Design For Chemical
Engineers, Chapman & Hall, London. 1998.
[3] R. M. Felder, R. W. Rousseau, Elementary Principles of Chemical
Processes, 3rd ed., Wiley, New York. 2000.
[4] O. Levenspiel, Chemical Reaction Engineering. 3rd ed., Wiley, New
York. 1998.
[5] L. D. Schmidt, The Engineering of Chemical Reactions, Oxford
University Press, New York. 1998.
[6] J. M. Smith, Chemical Engineering Kinetics. McGRAW-HILL, New
York. p. 131. 1956.
[7] M. Vidyasagar, Control System Synthesis: A Factorization Approach.
MlT Press, Cambridge , 1985.
[8] J.C. Doyle, B.A. Francis, A.R. Tannenbaum. Feedback Control
Theory. Maxwell McMillan, 1992.
[9] V. Kučera, “Diophantine Equations in Control – A Survey”,
Automatica, 29, No.6, pp. 1361-75. 1993.
[10] M. Morari, E. Zafiriou, Robust Process Control. Prentice Hall, New
Jersey. 1989.
[11] R. Prokop, J.P. Corriou. “Design and Analysis of Simple Robust
Controllers”. Int. JournalofControl, Vol. 66, No. 6, pp. 905-921.
1997.
[12] R. Prokop, P. Dostál. “Design of Simple Robust Controllers”. In 10th
International Conference on PC´95. Tatranské Matliare, pp. 103-108.
1995.
[13] K.J., Astrom, T. Hagg1und, PID Controllers: Theory, Design and
Tuning, 343 p. Instrument Society of America. 1995.

View publication stats