Computers chem. Engng Vol.20, Suppl.,pp.

S1619-SI624,1996
Pergamon
Copyright 1996ElsevierScience Ltd
S0098-1354(96)00275-X Printedin GreatBritain.All fightsreserved
0098-1354/96$15.00+0.00

DYNAMIC SIMULATION OF REACTIVE DISTILLATION:

AN MTBE CASE STUDY
STANY SCHRANS,1 SJOERD DE WOLF
Shell Research and Technology Centre Amsterdam
P.O. Box 38000, 1030 BN Amsterdam
AND RICHARD BAUR2
Institut ffir Systemdynamik und Regelungstechnik
UniversitEt Stuttgart
PfMfenwaldring 9, 70550 Suttgart

Dynamic simulations of a 15 stage MTBE reactive distillation column have been per-
formed with the SPEEDUPflowsheeter. The simulations confirm that the column config-
uration with fixed bottom outflow rate can display steady-state multiplicity. Changes in
the column operation (feed flow rate or composition) can cause a jump from one steady
state to another or trigger oscillations. The observed steady-state multiplicity remains
when the reboiler temperature is fixed rather than the bottom flow rate.

1. I n t r o d u c t i o n

The technology of integrating reactors with separators has made significant progress in recent
years. In reactive distillation products are made by reaction and distilled in s i t u . These pro-
cesses often have better conversion and selectivity, and lower capital and operating costs than
the corresponding reactor-followed-by-separation technology. Several processes containing re-
active distillation columns have now commercial applications, the Eastman Kodak process
for the production of methyl acetate, and the catalytic distillation of MTBE being the most
well-known examples.
Reactive distillation processes may display a variety of dynamic phenomena. ASPENPLUS
simulations have resulted in the discovery of steely-state multiplicity in an M T B E reactive
distillation column (Jacobs and Krishna, 1993, Nijhuis et al., 1993, Hauan et al., 1995).
Experimental evidence for steady-state multiplicity in a pilot TAME reactive distillation
column of Neste Oy have been reported by Bravo et al. (1993), while sustained oscillations
of process variables have been observed in a lab-scale reactive distillation column for MTBE
production (Sundmacher and Hoffmann, 1995).
Despite all these indications of dynamic phenomena, most of the modelling and sim-
ulations of reactive distillation processes reported in the open literature have been of the
steady-state type. Rigorous dynamic simulations are needed to investigate the relevance
of the steady-state multiplicity and the process sensitivity to disturbances in operational
parameters. Dynamic modelling of reactive distillation processes is thus required for the de-
velopment of a control scheme (see, for example, Roat et al., 1986), startup and shutdown
procedures, and will moreover permit a more robust operation of the column.

2. T h e R e a c t i v e Distillation Process for M T B E P r o d u c t i o n

The process studied in this paper is the reactive distillation of Methyl Tert-Butyl Ether
(MTBE). MTBE is an octane booster and clean-air component in gasoline, and is reputed
to be the world's fastest growing chemical. The reactive distillation technology for MTBE
production has been developed in the early 80's (Smith and Huddleston, 1982).
1Author to whom correspondence should be addressed, e-mail: schransl@ksla.nl.
2Work performed during a traineeship at the Koninklijke/Shell-Laboratorium, Amsterdam.

c~cE zo:~3(e~-c S1619

S1620 European Symposiumon ComputerAidedProcessEngineering--6.Part B

MTBE is formed in a highly selective reversible exothermic reaction from isobutene and
methanol over an acid ion-exchange resin catalyst according to the reaction

isobutene + methanol ~ MTBE. (1)

A rate equation for the reaction over the ion-exchange resin has been proposed by Rehfinger
and Hoffmann (1990):

( alB 1 aMTBE~
rate = mcatqacidk! \ aMeOH Keq a~eOH / ' (2)

where meat is the catalyst loading, qacid the catalyst ion-exchange capacity, and ai the activity
coefficient of the corresponding component. The forward reaction rate constant k! and the
equilibrium constant Keq have been fitted experimentally by Rehfinger and Hoffmann (1990).
Typically, isobutene is one of the components of a C4-raflinate feed. The vapor-liquid equilib-
rium is highly non ideal (including two binary azeotropes and one ternary reactive azeotrope).
No liquid phase separations are known to occur.

3. S t e a d y - s t a t e Results

Using ASPENPLUS Jacobs and Krishna (1993) performed steady-state simulations of an

MTBE reactive distillation column with 15 trays, 8 of which were reactive. They discovered
that when the bottom flow rate is fixed certain column configurations have, besides a high
isobutene conversion steady state, also a low isobutene conversion steady state. Hauan et
al. (1995) confirmed these results and presented a mechanistic explanation for the existence
of the two conversion branches.
Even though this steady-state multiplicity in simulations is by now well established, the
stability and relevance of these solutions has not yet been investigated with a dynamic sim-
ulator. Abufares and Douglas (1995) performed dynamic simulations of an MTBE reactive
distillation column (with only 1 reactive tray) with the SPEEDUP flowsheeter but did not
report the existence of multiple steady states.
The MTBE reactive distillation column used in our simulation study is shown in Fig. 1;
it is similar to the one of Jacobs and Krishna (1993) (the inerts being n-butane rather than
n-butene). The column contains 2 rectification trays, 8 reactive trays, 5 stripping trays, a
total condenser (with fixed reflux ratio and pressure) and a partial reboiler. In the reboiler
it is possible to fix either the bottom flow rate or the temperature. The C4 feed is located
on the lowest reactive tray, C8, whereas the location of the methanol feed can be varied (and
distributed) over the column. The chemical reactions are kinetically limited, and the vapour
and liquid are assumed to be in equilibrium. Our steady-state and dynamic simulations were
performed with the SPEEDUP flowsheeter.
Fig. 2 shows the isobutene conversions in the steady state for different methanol feed
locations for fixed bottom flow rate and fixed reboiler temperature.
The left graph of Fig. 2 shows the results when the bottom outflow rate is fixed to
674 kmol/h. When the methanol is fed on the trays C7 or C8 multiple steady states are
possible; amongst them one with a high and one with a low isobutene conversion. The steady-
states with the lower isohutene conversions have large negative reaction rates on the lowest
reactive trays, indicating MTBE decomposition on these trays. The right graph of Fig. 2
shows the isobutene conversions for the different steady states when the reboiler temperature
is fixed to 410 K. In this case, every methanol feed location results in multiple steady states.
 European Symposium on Computer Aided Process Engineering--6. Part B S1621

MTBE

R~ell~~ttum

Figure 1: Reactive distillation column configuration in dynamic simulations. The C4 [eed is

always on the lOth tray, C8. The reboiler outflow rate or temperature can be fixed with a
PI-controller.

Con[IB] (%) Con[IB] (%)

100 100 .....-

~l d ~ po4 m ~ ~

60 J S ab ) 60 : S ab s
-~ -- U ~st 3.bl ~ k 40 ~ " P-" : I.J ist ibl
40 ~ ,.,, ,.. ,..~,..~

2C 20 L : L

R1R2 C1C2C3C4C5C6C7C8 $1S2 $3 $4 $5 R1R2 CIC2C3C4C5C6C7C8 $1S2 $3 $4 $5

Figure 2: Isobutene conversions as a function of the methanol feed location for fixed bottom
outflow rate and fixed reboiler temperature.

The unstable steady states were, in all cases, only found by using the full dynamic capabilities
of SPEEDUP.
The behaviour of this reactive distillation column is to a large extent determined by
the existence of steady states for different operating conditions. This is a complex problem
since many operating variables can be changed. Fig. 3 shows some examples of phase space
diagrams when only one of these variables is changed for the column configuration with fixed
bottom flow. The dotted vertical line corresponds to the base case column, i.e. the column
with methanol fed on C7 and bottom flow given by 674 kmol/h. These diagrams seem to be
of a forbidden complexity, and suggest that a variety of dynamic phenomena are liable to
occur, making the column ditficult to operate.
S1622 European Symposiumon ComputerAided ProcessEngineering--6. Part B

Reflux Ratio Catalyst Mass

100
i lOO i
60
/
60 6 i
, ~
1
40 4O i

20
i/\ .......... 60 /J, 20
1

:
i '
1
: i [t )nne]
! 0 '
5 10 15 20 25 30 35 40 0.5 1 1.5 2 2.5

Figure 3: Steady states isobutene conversions for different reflux ratios and catalyst loadings.

Fig. 3 also reveals some striking differences between reactive and non reactive distillation.
The isobutene conversion (on the high conversion branch) is not a monotonically increasing
function of the reflux ratio. Thus, unlike for non reactive distillation, column performance
is not optimized at infinite reflux. Similarly, increasing the catalyst loading (or equivalently
the catalyst activity) does not necessarily lead to a better conversion.

4. D y n a m i c S i m u l a t i o n s

Figs. 4 and 5 show some results of dynamic simulations of this MTBE reactive distillation
column with methanol feed located at C7 when the bottom flow rate is fixed to 674 kmol/h.
The left graph of Fig. 4 shows the response to a +10 % methanol feed flow pulse between
t - 3 h and t -- 4 h. After the disturbance the column returns to the high isobutene conversion
steady state. The right graph of Fig. 4 shows the response to a similar feed pulse of 12.5%.
As a result of this disturbance, the column jumps from the high conversion operation point
to the low conversion operation point. Even though the operating condition at t > 4 h is
identical to the initial operating conditions, the column operates in a different steady state.

100

90
431 100

90
430

i/,
420
80 80
0

70 ~o 410'

60 tO

0

J-~MeOHfeed

,5 10
Time (h).
15
'ol
20
60 ~ " ~

0
MeOH feed
5
, Time(h),
10 15
40C
20

Figure 4: Response to a disturbance in the methanol feed flow rate. Left, a I0 % increase
during I h; right, a 12.5,% increase during I h. ( : conversion; .... : temperature).

The left graph of Fig. 5 shows the response of the column to a change in the composition
 European Symposium on Computer Aided Process Engineering--6. Part B S1623

961
390 365

, ..... ~',;
95'
I
;

ao[~
[~" I 370 g4[ gl 360
70 o l ' ' i ' i I t ,1!l ,t /. ,' ! W N I! m

6o[--J V V ,,I V ,,I V v 360 93]

,Sco.mp . ._1 IB comp Time,(h) ~55
o 10 20 30 40 0 lo 2'0 30 40

Figure 5: Response to a disturbance in the C4-feed composition. Left, a 4 %increase in the

isobutene composition, followed by a 5 % increase. Right, a 2.5 % increase in the isobutene
composition ( : conversion; . . . . : temperature).

of the (74 feed. First, at t = 3 h the mole fraction of isobutene in the C4 feed is increased
with 4%, while the total C4 flow rate is kept constant (just as the methanol feed flow rate):
the process variables start to oscillate. Next at t = 25 h the isobutene mole fraction in the C4
feed is increased a little more: the isobutene conversion now drops to a steady-state value
of ca. 53 %. The right graph of Fig. 5 shows the response of the column to a 2.5 % increase
in the isobutene composition of the (74 feed. This change in operating condition results in
damped oscillations of the process variables.

5. Conclusions

Dynamic simulations of an MTBE reactive distillation column have been performed using
the SPEEDUP flowsheeter. Fixing the bottom outflow rate or fixing the reboiler temperature
leads to multiple steady states. Dynamic simulations have shown that this reactive distillation
column is 'unstable' for certain disturbances of the feed. This work illustrates that, in this
case, dynamic simulations should be taken into account during the first steps of the design
process in order to avoid dynamic surprises. This work also illustrates the power of dynamic
simulations compared to steady-state simulations.
6. List of Symbols

aIB Activity coefficient of isobutene

aMeOH Activity coefficient of methanol
aMTBE Activity coefficient of MTBE
Keq Equilibrium constant
kl Forward reaction rate constant 1/s
mcat Catalyst loading of a tray kg
qacid Ion-exchange capacity of the catalyst eq(H+)/kg

7. References

A.A. Abufares and P.L. Douglas, 1995, Mathematical modelling and simulation of an M T B E
catalytic distillation process using SPEEDUP and ASPENPLUS, Trans IChem., 72, 3.
S1624 EuropeanSymposiumon ComputerAidedProcessEngineering--6.PartB

J.L. Bravo, A. Pyh~ilahti, and H. J/irvelin, 1993, Investigations in a catalytic distillation pilot
plant: vapor/liquid equilibrium, kinetics, and mass-transfer issues, Ind. Engng. Chem. Res. 32,
2220.
S. Hauan, T. Hertzberg, and K.M. Lien, 1995, Multiplicity in reactive distillation of MTBE,
presented at ESCAPE-5, Comp. Chem. Engng., 19, S-327.
R. Jacobs and R. Krishna, 1993, Multiple solutions in reactive distillation for methyl tert-
butyl ether synthesis, Ind. Engng. Chem. Res., 32, 1706.
S.A. Nijhuis, F.P.J.M. Kerkhof, and A.N.S. Mak, 1993, Multiple steady states during reactive
distillation of methyl tert-butyl ether, Ind. Engng. Chem. Res., 32, 2767.
A. Rehfinger and U. Hoffmann, 1990, Kinetics of methyl tertiary butyl ether liquid phase
synthesis catalyzed by ion exchange resin-I Intrinsic rate expression in liquid phase activi-
ties, Chem. Engng. Sci., 45, 1605.
S.D. Roar, J.J. Downs, E.F. Vogel, and J.E. Doss, 1986, The integration of rigorous dynamic
modeling and control system synthesis for distillation columns: an industrial approach, in
Chemical Process Control-CPC III, M. Morari and T.J. McAvoy (Eds), Elsevier, NY.
L.A. Smith and M.N. Huddleston, 1982, New MTBE design now commercial, Hydrocarbon
Processing, March 1982, 121.
K. Sundmacher and U. Hoffmann, 1995, Oscillatory vapor-liquM transport phenomena in a
packed reactive distillation column for fuel ether production Chem. Engng. J., 57, 219.