Dynamic Simulation and

Optimization using EMSO

– Lecture 5 –
Simulation of tubular reactors.

Argimiro R. Secchi

PEQ/COPPE-UFRJ
Solutions for Process Control and Optimization
January, 2013 1
Dynamic Modeling of Tubular Reactors

Assumptions:
• Constant physical properties (, , cp);
• Newtonian fluid;
• Angular symmetry;
• vr = v = 0
2
Model classification based on physical-chemistry principles

Microscopic Model

Multiple Gradients Model

Maximum Gradient Model

Macroscopic Model

3
 Microscopic Model

(.v )  0 vz  vz (r , t )

Ci Ci
 vz (r, t )  ( Ci v)  ( D(l ) Ci )  Ri
t z
T T
 cp   c p vz (r, t )   c p ( v T )  ( k (l ) T )  ((l ) v(l )  (t )(vt ) )  Sr
t z

v
  v v   [  v v]  P   (l )  2 v   g
t

Turbulence Model (simple example):

v ci  J i( t )  D( t ) Ci

Computational Fluid
 c p v T   q ( t )   k ( t ) T Dynamic (CFD)

 v v   ( t )    ( t ) v
4
 Recommended Numerical Method

Finite Volume:
Consist in carry out balances of properties in elementary volumes (finite
volumes), or in an equivalent form in the integration over the elementary volume
of the differential equation in the conservative form (or divergent form, where the
fluxes appearing in the derivatives).

 Use of CFD software (Computational Fluid Dynamics)

5
Simple example using finite-volume method:
Reaction-diffusion equations of a spherical catalytic particle

 C 1   2 C 
  2 r    C (r , t )
2 First-order selections
 t r r  r 
 C
  0 ; C (1, t )  1
 r r 0

 C ( r , 0)  C0


dV  4r 2 dr

e e e
C 2  2 C 
 t r dr    r r     2 r 2 C dr

w w w

6
 Simple example using finite-volume method:
e

 dr
2
C r e
3
Cp  w
 2 2  C r 2 dr mean value
e
 re  rw  w
 dr
2
r
w

 C  CE  C p C p  CW
e
dC p 3
 3 3 r 2   C C
 
2
Cp
dt  re  rw   r  w

r r  re re r r  rw rw

dC p
 AW CW  Ap C p  AE CE   2C p p = 2, ..., N  1
dt

3rw2 3  re2 rw2  3re2

AW  Ap   AE 
 re3  rw3  rw  re  rw   e rw 
3 3 
r  re3  rw3  re
7
 Simple example using finite-volume method:
Boundary conditions

C C Ce  C p 1 Cp
0  
r rw r re r f r f

dC
Resulting systems:   AC  b where A is a tridiagonal matrix
dt  
dC
Or a non-linear system for reaction of order  1:   F ( C)
dt  
8
 Multiple Gradients Model

D  D( t )  D( l ) k  k (t )  k (l )   (t )  (l )

Ci C
 vz (r, t ) i  ( D Ci )  Ri
t z
T T
 cp   c p vz (r, t )  ( k T )   v  Sr
t z
vz
  P 2vz  g
t
Boundary Conditions:

vz, Ci0 , T0
reaction zone

z=0 z=L
z = 0¯ z = 0+ z = L¯ z = L+

9
 Component Mass Balance
(removing time average notation)

Ci  2Ci 1   Ci  Ci

 D z (r , t ) 2   r D R (r , t )   vz ( r , t )  Ri
t z r r  r  z

Ci (0, r , t )
vz (r , t ) Ci 0 (t )  vz ( r , t ) Ci (0, r , t )  D z (r , t )
z
Ci
( L, r , t )  0 no reaction at exit
z
Ci
( z , 0, t )  0 Symmetry
r
Ci Impermeable wall
( z, R, t )  0
r
Ci ( z , r , 0)  Ci _ in ( z , r ) Initial condition

10
 Energy Balance
(removing time average notation)

T  2T 1   T  T
 cp  k z (r , t ) 2   r k R (r , t )    c p vz ( r , t )  H r RA
t z r r  r  z

k z (r , t ) T (0, r , t )
vz (r , t ) T0 (t )  vz (r , t ) T (0, r , t ) 
 cp z
T
( L, r , t )  0 No reaction at exit
z
T
( z , 0, t )  0 Symmetry
r
T
k R ( R, t ) ( z , R, t )  U Tw  T ( z , R, t )  Heat exchange with the wall
r

T ( z , r , 0)  Tin ( z , r ) Initial condition

11
 Momentum Balance
(removing time average notation)

vz P 1   vz 
   r 
t z r r  r 

vz Symmetry
(0, t )  0
r
vz ( R , t )  0 Fixed wall

vz ( r , 0)  vz _ in ( r ) Initial Condition

12
 Additional assumptions:
- constant effective diffusive coefficients
- constant velocity

Ci  2Ci D R   Ci  Ci

 DL 2   r 
 z v  Ri
t z r r  r  z

T  2T k R   T  T
 cp  kL 2   r    c v  H r RA
t z r r  r  z
p z

Ci (0, r , t ) k L T (0, r , t )

vz Ci 0 (t )  vz Ci (0, r , t )  D L vz T0 (t )  vz T (0, r , t ) 
z  cp z
Ci T
( L, r , t )  0 ( L, r , t )  0
z z
Ci T
( z , 0, t )  0 ( z , 0, t )  0
r r
Ci T
( z, R, t )  0 kR ( z , R, t )  U Tw  T ( z , R, t ) 
r r
Ci ( z , r , 0)  Ci _ in ( z , r ) T ( z , r , 0)  Tin ( z , r )
13
 Method of Lines
- spatial discretization (finite differences, finite volumes, finite elements)
- time integration
Example: finite differences in axial direction and orthogonal collocation in radial

dCi , j ,k C  2Ci , j ,k  Ci , j 1,k   n 1   Ci , j 1,k  Ci , j 1,k 

 D L  i , j 1,k   4 D R    k k ,m
u B  Ak ,m  i , j ,m 
C  v z    Ri , j ,k
dt  z 2   m  0   2 z 

dT j ,k  T j 1,k  2T j ,k  T j 1,k   n 1   T j 1,k  T j 1,k 

 cp  kL    4 kR    uk Bk ,m  Ak ,m  T j ,m    c p vz    H r RA, j ,k
dt   z 2
  m0   2z 

where: n 1 u  up
ur 2
lm (u )   Ak .m
dl (u )
 m k Bk ,m 
d 2lm (uk )
p  0 um  u p du du 2
pm n 1

j = 1,2,...,N k = 1,2,...,n y (u )  P (  , )
n 1 (u )   lm (u ) ym
m 0

14
 Method of Lines
Boundary conditions

 C  Ci 0 (t )  k L  T2,k  T0 (t ) 
vz Ci 0 (t )  vz Ci ,1,k  D L  i ,2,k  vz T0 (t )  vz T1,k 
 2 z   
 c p  2z 

Ci , N 1,k  Ci , N ,k  0 TN 1,k  TN ,k  0

n 1 n 1

 A0,mCi, j ,m  0
m0
A
m0
0, m T j ,m  0

n 1
k R  An 1,mT j ,m  U Tw  T j ,n 1 
n 1

A
m0
n 1, m Ci , j ,m  0
m 0

Ci , j , k  Ci _ in , j , k T j ,k  Tin , j ,k
j = 1,2,...,N k = 1,2,...,n
Results in a DAE system!

15
 Another Multiple Gradients Model
(ignoring radial gradients) – PFR with axial dispersion
R

Ci  2Ci C
 D L 2  vz i  Ri
 C (r , z, t ) r dr
i

t z z Ci ( z , t )  0
R

C (0, t )  r dr
vz Ci 0 (t )  vz Ci (0, t )  D L i 0
z
Ci
( L, t )  0
z

Ci ( z , 0)  Ci _ in ( z )
R

T  2T T 2U  T (r , z, t ) r dr
 cp  k L 2   c p vz  Tw  T   H r RA T ( z, t )  0
t z z R R

k L T (0, t )
 r dr
vz T0 (t )  vz T (0, t )  0

 c p z
T
( L, t )  0
z

T ( z , 0)  Tin ( z )
16
 Method of Lines
- spatial discretization (F.D., F.V., orthogonal collocation on finite elements)
- time integration
Example: finite differences in axial direction

dCi , j  C  2Ci , j  Ci , j 1   Ci , j 1  Ci , j 1 
 D L  i , j 1 
 zv   Ri , j
dt  z 2
  2z  j = 1,2,...,N
dT j  T j 1  2T j  T j 1   T j 1  T j 1  2U
 cp  kL 
 2    c p vz 

 Tw  Tj   H r RA, j
dt  z   2 z  R

 C  Ci 0 (t )  k L  T2  T0 (t ) 
vz Ci 0 (t )  vz Ci ,1  D L  i ,2 vz T0 (t )  vz T1   
  c p  2z 
 2 z 
Ci , N 1  Ci , N  0 TN 1  TN  0

Ci , j  Ci _ in , j T j  Tin , j

17
 Maximum Gradient Model
(ignoring axial dispersion) – PFR without axial dispersion

Ci C
 vz i  Ri
t z

Ci (0, t )  Ci 0 (t )

Ci ( z , 0)  Ci _ in ( z )

T T 2U
 cp   c p vz  Tw  T   H r RA
t z R

T (0, t )  T0 (t )

T ( z , 0)  Tin ( z )

Results in an EDO system!

18
 Macroscopic Model

V
dCi
 Ci 0 (t ) vz S  Ci vz S  Ri V
 C ( z, t ) S dz
i

dt Ci (t )  0
L

 S dz
Ci (0)  Ci _ in 0

dT  T ( z, t ) S dz
 cp V   c p vz S T0 (t )   c p vz S T  U At (Tw  T )  H r RA V T (t )  0
dt L

 S dz
T (0)  Tin
0

Results in a EDO system!

19
Method of Orthogonal Collocation with
EMSO

• Boundaries
• Internal Points
• Alfa and Beta

EMSO Plugin Roots

Simulator A and B Matrices

• Jacobi roots
• A and B matrices

DD as Plugin (Type=“OCFEM”, Boundary="BOTH”, InternalPoints=5

alfa=1, beta=1)
Plugin: ocfem_emso.dll
20
 Fixed-bed Reactor with Axial Dispersion
(reaction of order m)

y y 1  2 y
   Da y m

 x Pe x 2

Boundary conditions:

1 y
  1  y (, 0) or y (, 0)  1
Pe x x 0

y
0
x x 1

Initial conditions:

y (0, x)  0
21
 PDE
Method of Lines: D.F. and Orthogonal Collocation

Example: add Plugin ocfem_emso.dll and

execute flowsheets of files FDM_ss.mso,
OCM_ss.mso and OCFEM_ss.mso, and
compare results of discretizations. Repeat
for the dynamic simulation in files
FDM_din.mso e OCM_din.mso.

22
 Comparing Results
1.2000 1.2000

1.0000 1.0000

0.8000 0.8000

0.6000 0.6000

0.4000 0.4000

0.2000 0.2000

0.0000 0.0000
0.0000 0.2000 0.4000 0.6000 0.8000 1.0000 0.0000 0.2000 0.4000 0.6000 0.8000 1.0000

OCM by EMSO  1  1 Method of Finite Differences

Number of internal points: 5 Number of internal points: 6000
y(x=1) = 0.151475 (error of 0.038%) y(x=1) = 0.15155 (error of 0.087%)

y(x=1) = 0.151418 (exact)

23
 Case Study

• Production of acetic anhydride in adiabatic PFR

– Acetic anhydride is often produced by reacting acetic acid with ketene,
obtained by heating acetone at 700-770oC.
– A important step is the vapor phase cracking of acetone to ketene and
methane:
C H 3C O C H 3  C H 2 C O  C H 4
– The second step is the reaction of ketene with acetic acid.

CH 2CO  CH 3COOH   CH 3CO  O

Ref: G. V. Jeffreys, A Problem in Chemical Engineering Design: The Manufacture of

Acetic Anhydride, 2nd ed. (London: Institution of Chemical Engineers, 1964)

24
 Case Study
Problem Definition
– The first production step is carried out in a vapor phase reaction
of acetone in an adiabatic PFR.

where A = acetone; B = ketene and C = methane

A BC
– The reaction is of 1a order in relation to acetone in the cracking
reaction, with Arrhenius constant given by:
• k – seconds-1  34222 
k  exp  34.34  
• T – Kelvin  T 
25
 Case Study
Process Description

– Reactor geometry
• adiabatic continuous tubular reactor;
• bank of 1000 tubes of 1 in sch. 40 with cross section of 0.557 m2;
• total length of 2.28 m;
– Operating conditions
• feed temperature 762oC (1035 K);
• operating pressure: 1.6 atm
• feed flow rate of 8000 kg/h (137.9 kmol/h);
– Composition
• acetone, ketene and methane
• feed of pure acetone
– Kinetics
• first order reaction,
• pre-exponential factor (k0): 8.2 x 1014 s-1
• activation energy (E/R): 34222 K
• heat of reaction: -80.77 kJ/mol
26
 Case study
– Production of acetic anhydride –

Example: run FlowSheet in file

PFR_Adiabatico.mso and plot steady-state
temperature and composition profiles.
Show also the evolution of the
temperature profile. Discuss the type and
quality of discretization.

27
 Exercise
Solve the reaction-diffusion problem in a spherical catalytic
particle, given by:
y 1   2 y 
 2  r    2 1/ 2
y
t r r  r 
y y (t , r ) r 1  1 y (t , r ) t 0  0
0
r r 0

  2 (Thiele modulus)

28
 References
• Himmelblau, D. M. & Bischoff, K. B., "Process Analysis and Simulation - Deterministic Systems", John Wiley &
Sons, 1968.
• Finlayson, B. A., "The Method of Weighted Residuals and Variational Principles with Application in Fluid Mechanics,
Heat and Mass Transfer", Academic Press, 1972.
• Villadsen, J. & Michelsen, M. L., "Solution of Differential Equation Models by Polynomial Approximation", Prentice-
Hall, 1978.
• Davis, M. E., "Numerical Methods and Modeling for Chemical Engineers", John Wiley & Sons, 1984.
• Denn, M., "Process Modeling", Longman, New York, 1986.
• Luyben, W. L., "Process Modeling, Simulation, and Control for Chemical Engineers", McGraw-Hill, 1990.
• Silebi, C.A. & Schiesser, W.E., “Dynamic Modeling of Transport Process Systems”, Academic Press, Inc., 1992.
• Biscaia Jr., E.C. “Método de Resíduos Ponderados com Aplicação em Simulação de Processos”, XV CNMAC,
1992
• Ogunnaike, B.A. & Ray, W.H., “Process Dynamics, Modeling, and Control”, Oxford Univ. Press, New York, 1994.
• Rice, R.G. & Do, D.D., “Applied Mathematics and Modeling for Chemical Engineers”, John Wiley & Sons, 1995.
• Maliska, C.R. “Transferência de Calor e Mecânica dos Fluidos Computacional”, 1995.
• Bequette, B.W., “Process Dynamics: Modeling, Analysis, and Simulation”, Prentice Hall, 1998.
• Fogler, H.S., “Elementos de Engenharia de Reações Químicas”, Prentice Hall, 1999.

29
 Special thanks to
Prof. Rafael de Pelegrini Soares, D.Sc.
Eng. Gerson Balbueno Bicca, M.Sc.
Eng. Euclides Almeida Neto, D.Sc.
Eng. Eduardo Moreira de Lemos, D.Sc.
Eng. Marco Antônio Müller

For helping in the preparation of this material

For supporting the ALSOC Project.

30
 EP 2013

... thank you for your attention!

http://www.enq.ufrgs.br/alsoc

Solutions for Process Control and Optimization

Process Modeling, Simulation and Control Lab

• Prof. Argimiro Resende Secchi, D.Sc.
• Phone: +55-21-2562-8307
• E-mail: arge@peq.coppe.ufrj.br
• http://www.peq.coppe.ufrj.br/Areas/Modelagem_e_simulacao.html

31