Sie sind auf Seite 1von 9

8704 Ind. Eng. Chem. Res.

2008, 47, 8704–8712

Dynamic Control of a Column/Side-Reactor Process


Devrim B. Kaymak† and William L. Luyben*,‡
Department of Chemical Engineering, Istanbul Technical UniVersity, 34469, Maslak, Istanbul, Turkey,
Department of Chemical Engineering, Lehigh UniVersity, Bethlehem, PennsylVania 18015

If the temperature range suitable for reasonable chemical reaction kinetics does not match the temperature
range suitable for vapor-liquid equilibrium, reactive columns are not economically attractive. One way to
overcome this temperature mismatch is use a flowsheet that features a distillation column with multiple side
reactors. The column operates at a pressure that gives temperatures favorable for separation, while the reactors
operate at temperatures (and pressures) favorable for reaction kinetics. This paper discusses the controllability
of a column/side-reactor process based on the steady-state design studied by Kaymak and Luyben.1 The
performances of two different control structures are explored. The first uses two temperature controllers, one
with direct action and the other with reverse action. The second control structure uses an internal composition
analyzer. One of the main conclusions is that the opposite actions of controllers in the two-temperature control
structure result in a composition breakthrough that moves the column to a different operating condition. The
composition and temperature controllers in the second structure have the same action and provide reasonably
effective control.

1. Introduction change with temperature. At a temperature of 320 K, which is


typical of the reflux drum temperature in a condenser using
Reactive distillation can yield significant economic benefits cooling water, the relative volatilities between adjacent com-
in some systems. However, for reactive distillation to be ponents are R320 ) 2. As temperature increases, the relative
attractive, the temperatures that are good for reaction must match volatilities decrease. A value of R390 ) 1 means that all
the temperatures that are good for vapor-liquid separation. components have the same relative volatility at 390 K. The
Therefore traditional reactive distillation is not effective in many temperature of 390 K is selected because this temperature gives
chemical systems because of a mismatch in temperatures. favorable kinetics in terms of both the forward specific reaction
Kaymak et al.2 demonstrated quantitatively that reactive distil- rate and the chemical equilibrium constant in the numerical case
lation becomes less attractive as the temperature mismatch used to illustrate the issues.
increases. Although we studied a wide range of R390 values in the design
To overcome this temperature mismatch, a flowsheet com- paper,1 in this control paper we only deal with the R390 ) 0.95
bining a distillation column with external side reactors can be case in which there is considerable reaction/separation mismatch.
considered where the column operates at temperatures favorable The vapor pressure PjS of component j is a function of
for separation, while the reactors operate at temperatures temperature as given in the Antoine equation.
favorable for chemical kinetics. There are several papers in the
literature dealing with the design of combined column/reactor BVP,j
systems.3-7 However, we are not aware of any paper in the ln PSj ) AVP,j - (2)
T
literature that discusses the controllability of these column/side-
reactor configurations. where AVP,j and BVP,j are constants over a limited temperature
range. The Antoine constants for the case considered here are
In this paper we explore the dynamic control of a column/
given in Table 1. The detailed calculations to obtain these values
side-reactor process based on a recent design paper.1
are explained in a previous paper.2
The process configuration is given in Figure 1. There is a
2. Process Studied
distillation column operating at its optimum pressure and
The process studied has an ideal exothermic liquid-phase temperature for separation (320 K in the reflux drum). The
reaction where products C and D are formed from the reactants reactants A and B are fed to the column in two pure reactant
A and B in several side reactors linked to a distillation column. fresh feed streams F0A and F0B, both with a flowrate of 12.60
mol/s. There are 17 trays and a partial reboiler in the stripping
A+BSC+D (1) section below fresh feed streams F0A, and a rectifying section
The reactors operate at pressures and temperatures conducive with 7 trays and a total condenser above fresh feed streams
for reaction kinetics. The side reactors are connected to a F0B. There is a middle zone with 7 trays between these two
distillation column that operates at a pressure conducive for sections, where three total liquid trap-out trays are installed
vapor-liquid separation. No reaction occurs within the column. at different locations. The lower and upper trap-out trays are
The parameter used by Kaymak et al.2 to quantify reaction/ immediately below the fresh feed streams of reactants A and
separation temperature mismatch is R390, which is a measure B, respectively. It is assumed that the number of trays
of how the relative volatilities among the four components
Table 1. Vapor Pressure Constants
* To whom correspondence should be addressed: E-mail: WLL0@ R390 constant A B C D
Lehigh.edu. Tel.: 610-758-4256. Fax: 610-758-5057.

Istanbul Technical University. 0.95 AVP 12.34 15.80 8.89 19.26

Lehigh University. BVP 3862.00 5189.23 2534.77 6516.46

10.1021/ie701705m CCC: $40.75  2008 American Chemical Society


Published on Web 10/04/2008
Ind. Eng. Chem. Res., Vol. 47, No. 22, 2008 8705
catalyst is 0.5, so the vessel is half-filled with liquid and
half-filled with solid.
The reaction occurs at the catalyst site, so that is where
the heat of reaction affects the catalyst temperature. There
is also heat transfer from the catalyst to the liquid. This means
that there will be a temperature difference between the liquid
and the solid even under steady-state conditions. The energy
equation for the solid catalyst in the nth lump is
dTcat,n -λFliqVliq ⁄ Mliq
) (kFxA,nxB,n - kRxC,nxD,n) -
dt FcatVcatCPcat
UA
(T - Tliq,n) (3)
FcatVcatCPcat cat,n
The energy equation for the liquid in each lump includes the
convective terms associated with flow of liquid in and out of
the lump at their respective temperatures.
dTliq,n MliqFliq,n UA
) (T - Tliq,n) + (T - Tliq,n)
dt FliqVliq liq,n-1 FliqVliqCPliq cat,n
(4)
The last term in both of these energy balances comes from the
Figure 1. Column/side-reactor process.
heat transfer between the catalyst and the liquid.
Two assumptions have been made to calculate the values of
between each liquid trap-out tray is the same. Thus, the heat transfer coefficient U and area A needed for these equations:
second trap-out tray is at the midpoint of this middle zone. (i) the temperature difference between the liquid and solid is
This column is just used for separation; no reaction occurs ∼1.5 K at steady state, and (ii) the catalyst particles are spherical
in any part of the column. The light product C leaves in the with a 0.002 m diameter.
distillate, while the heavy product D is removed in the The reaction directly affects the compositions in the liquid
bottoms. The dynamic component balances for the column phase. The forward and reverse specific reaction rates follow
are given in the design paper.1 the Arrhenius Law.
All the reaction takes place in three side reactors. These
reactors are adiabatic plug-flow reactors, and the holdups of kF ) aFe-EF ⁄RTcat (5)
each reactor are the same (VR ) 80 kmol). The vapor from
the tray below each trap-out tray flows up through the kR ) aRe-ER ⁄RTcat (6)
chimney to the tray above with no vapor-liquid contacting. The overall reaction rate is based on concentrations in mole
Each liquid trap-out tray collects all the liquid coming down fractions and liquid holdups in moles. To avoid confusion, the
from the tray above the trap-out tray. Liquid from each tray- specific reaction rates used in this paper (with the reactor half-
out tray is pumped to a high enough pressure so that the full of catalyst) are twice those used in the design paper in which
material stays liquid at the higher temperatures in the external catalyst was not considered.
side reactors. Reactor effluents are fed back to the column
on the tray below the trap-out tray from which they were withdrawn. dxi,n Fliq,n
Since we are interested in the dynamic behavior of the ) (x - xi,n) ( (kFxA,nxB,n - kRxC,nxD,n)
dt FliqVliq ⁄ Mliq i,n-1
system for this control paper, the tubular reactor model used
(7)
in the flowsheet must be dynamic, not just steady-state as in
the design paper. The spatial variation in temperature and Since the reaction is exothermic, some vapor is produced
composition variables is approximated by using a lumped as the liquid from the high-pressure reactor flashes into the
model. With a reasonable number of lumps, the lumped model low-pressure column. This results in an increase of the vapor
gives a steady-state profile similar to the real steady-state flow rate at each external reactor location and a corresponding
profile we get using the rigorous steady-state equations. A decrease of the liquid rate below the external reactor location.
lumped model of a distributed system is equivalent to using The quantity of this vapor is calculated from the heat
a series of CSTRs to approximate a tubular reactor. Each generated by the reaction occurring in the external reactor,
lump has its dynamic component and energy balances with the flow rate of material into the reactor and the thermal
variables needed to be numerically integrated in time: dT/ properties. The steady-state vapor and liquid rates are constant
dt, dx/dt, etc. through the stripping and rectifying sections since equimolal
In these reactors, both the process liquid and the catalyst overflow is assumed. However, there are different liquid and
have thermal capacitance, so the temperatures of these two vapor rates in the middle sections of the column. The physical
phases can be dynamically different in each lump with heat and chemical parameter values are given in Table 2. Figure
transfer between the solid and liquid phases. Of course the 2A gives the composition and temperature profiles for
liquid compositions change dynamically from lump to lump column. The left graphs are steady-state results using rigorous
because of the convective flows in and out and because of ordinary differential equation models for the three reactors.
the reaction. The kinetics used is based on the volume of The right graphs are steady-state results using the lumped
the liquid. That brings up the issue of specifying the amount reactor model. The column profiles are essentially identical.
of liquid and the amount of solid catalyst in a given total Figure 2B gives composition and temperature profiles for
vessel volume. It is assumed that the void volume of the each of the three reactors using either the rigorous reactor
8706 Ind. Eng. Chem. Res., Vol. 47, No. 22, 2008

Table 2. Physical and Chemical Parameters 3. Control Structures


activation energy of reaction cal/mol In practical applications it is desirable to use inferential
forward 30000 temperature measurements whenever possible instead of direct
reverse 40000
specific reaction rate at 366 K kmol/(s-kmol) composition measurements. Composition analyzers have higher
forward 0.016 cost, require more maintenance, and can introduce deadtime into
reverse 0.016/(KEQ)366 the control loop. Therefore, we first explore a control structure
heat of reaction, λ kJ/mol - 41.8 that does not have any composition analyzer. The performance
heat of vaporization, ∆Hv kJ/mol 29 of this two-temperature control structure will then be compared
molecular weight of the mixture, Mliq g/mol 50
catalyst density, Fcat kg/m3 2000 with a structure that uses a composition analyzer.
liquid density, Fliq kg/m3 800 The control structures are single-input-single-output (SISO)
ideal gas constant cal/(mol-K) 1.987 structures with PI controllers for temperature and composition
catalyst heat capacity, CPcat kJ/(kg-K) 0.50 and P-only controllers for levels with gains of 2. The
liquid heat capacity, CPliq J/(kg-K) 2.93
Tyreus-Luyben tuning method is used to tune the temperature
heat-transfer coefficient, U kJ/(s-K-m2) 0.02
heat-transfer area, A m2 749 and composition controllers. The ultimate gain and ultimate
period necessary for this method are obtained using the relay-
model (solid lines) or the lumped reactor model (dashed feedback test. Two first-order measurement-lags of 60 s each
lines). These profiles are quite similar. are used in all temperature loops, while a 3 min deadtime is

Figure 2. (A) Composition and temperature profiles for column, (B) composition and temperature profiles for reactors (length ) 8.6 m).
Ind. Eng. Chem. Res., Vol. 47, No. 22, 2008 8707
used for the composition analyzer. All the control valves in the by the reflux flowrate. Reflux ratio is controlled by measuring
process are designed to be half-open at steady state. the reflux flowrate, multiplying this by the reciprocal of the
The effectiveness of the control structures studied in this paper desired reflux ratio and sending this signal to a remote-set
is demonstrated by using dynamic simulations and subjecting cascade flow controller on the distillate stream. The col-
the processes to production rate changes (∆F0j or ∆VS) and feed umn pressure is assumed to be constant and controlled by
composition disturbances (∆z0A and ∆z0B). condenser heat removal. Note that the liquid levels on the
As with any control system, factors such as criticality of the three trap-out trays are held constant by manipulating the
online analyzers and failure of measurements would need to be three valves in the liquid streams from the reactors back to
considered in any actual plant installation. These factors have the column (controllers not shown in Figure 3). The pressures
not been considered in this study. in the liquid-filled reactors are set by the pump head.
3.1. Control Structure CS7. Control structure CS7 is based The selection of the trays to temperature control is the main
on a paper by Roat and co-workers8 and is shown in Fig- issue in this structure. From the rigorous steady-state simulation,
ure 3. the gain matrix between the inputs (two fresh-feed flowrates)
Two temperature controllers manipulate the two fresh-feed and the outputs (the temperatures on all trays) is calculated
flowrates to maintain the temperatures on two trays. Reboiler numerically. Using these steady-state gains, the locations of the
heat-input is flow controlled and serves as the production- most sensitive trays whose temperatures are to be controlled
rate handle. The base level is controlled by manipulating the are selected by applying singular value decomposition (SVD)
bottoms flowrate, while the reflux drum level is controlled analysis. Some engineering judgment must be also exercised
in this selection. The procedure is discussed in detail in a
previous paper.9
The steady-state gains (KF0A and KF0B) between tray
temperatures and feed flow rates and their related SVD results
for CS7 structure are given in Figure 4. The steady-state gains
between tray temperatures and the F0A feed are negative (an
increase in feed flowrate decreases tray temperature) through-
out the column except for a small region in the lower
stripping section that has very small positive values. On the
other hand, the steady-state gains between tray temperatures
and the F0B feed are mostly positive, especially at the most
temperature sensitive regions. Note that the most sensitive
temperature measurements to the change in the fresh feed
streams are in the rectifying section with a negative maximum
at tray 27 for F0A and with a positive maximum at tray 26
for F0B. However, since these two trays are so close to each
other, it is not possible to use both of them.
To overcome this problem, we look for secondary sensitive
regions for both gains (second peaks that are smaller than
the biggest ones). By looking at the curves in Figure 4, these
secondary peaks are found in the upper stripping section (tray
16 for both gains). Although the problem defined above may
be solved by selecting one of these trays to pair with one of
Figure 3. Control structure CS7. the fresh feed stream, a bigger problem still exists. While

Figure 4. Gains and SVD for CS7.


8708 Ind. Eng. Chem. Res., Vol. 47, No. 22, 2008

both primary and secondary sensitive trays have negatiVe Table 3. Tuning Parameters for Control Structure CS7
gains for changes in F0A, the gains are positiVe for changes pairing span KC τI
in F0B. Therefore there will be a difference in the actions of
F0A/T16 10 1.37 37.95
control loops, and this may result in difficulties for this F0B/T26 10 0.06 8624
control structure. Kaymak and Luyben9 pointed out that the
actions of the two temperature controllers used in a reactive
produces an increase in the temperature on both control trays
distillation column should be the same (e.g., both positive).
(tray 16 and tray 26). The tray 16 temperature controller has
Controlling an upper tray in rectifying section by manipu-
direct action, so it increases the F0A feed flowrate. However,
lating the fresh-feed flowrate F0B (the upper-feed flowrate)
the tray 26 temperature controller has reverse action, so it
and controlling a tray in the stripping zone by manipulating
decreases the F0B feed flowrate. Not enough reactant B is
the fresh-feed flowrate F0A (the lower-feed flowrate) sounds
added in a timely enough manner to prevent light reactant A
more reasonable than the reverse. Therefore the F0A/T16 and
from going to the upper section of the column. Once this
F0B/T26 pairings are first selected as the control loops instead
breakthrough occurs, the column moves to a different
of the alternative pairings of F0A/T27 and F0B/T16.
operating condition where the impurity of A in the distillate
Figure 5 gives relay-feedback test results for both loops used
stream increases and product purities are lower. Thus, we
in this structure. Note the difference in the time scales for the
can conclude that control structure CS7 is not a good choice
two loops. Although the amplitude and period look reasonable
for controlling this column/side-reactor process.
for the F0A/T16 pairing, the shark-tooth shape of the tray 26
3.2. Control Structure CS5. As mentioned in the previous
temperature response is characteristic of a process with inverse
section, it is desirable to use inferential temperature measure-
response. This is confirmed by the controller gain and reset time
ments instead of direct composition measurements whenever
values calculated from test data. As given in Table 3, the values
possible. However, the results show that the control structure
for the F0A/T16 loop have reasonable values. However, the
CS7 is not an effective control structure at least for the
controller for tray 26 has an unrealistically large reset time of
process studied here, which is the optimum economic steady-
8624 min.
state design. It appears that some direct composition informa-
Therefore, the alternative structure with the F0A/T27 and F0B/
tion about reactant inventory inside the system is required
T16 pairings is checked by applying the relay-feedback test. This
for a more effective control system since the column is
structure also results in inverse response as revealed in the huge
designed for neat operation. Thus, another control structure
reset time of the F0A/T27 loop. Thus, switching the pairing does
CS5 that includes one composition analyzer is explored in
not help solve the problem.
Figure 6 shows the response of CS7 to (5% step changes in this section.
production rate (vapor boilup) using the F0A/T16 and F0B/T26 This control structure differs from control structure CS7 only
pairings. The response of the system is stable, but it takes a in how the fresh feed streams and the reboiler heat inputs are
long time for T26 temperature to go back to its setpoint for both manipulated. Figure 7 shows that the temperature of a tray near
positive and negative disturbances. Such a long time to settle the bottom of the column is used to infer bottoms product purity,
down is a problem related to the inverse response and the huge and it is controlled by manipulating the vapor boilup. Tray 16,
reset time. A bigger problem is that even fairly small 5% which has the biggest steady-state gain (KVS) in the stripping
disturbances result in large changes in the purities of the product section, is selected. Distillate purity is not controlled. The fresh
streams xD,C and xB,D. While one of the purities settles down to feed F0B is flow controlled and serves as the production rate
a new steady state above its setpoint, the other one goes to a handle.
new steady state that is far below its setpoint. A composition analyzer is used to detect the inventory of
This problem occurs because the two temperature control- component A in the system. Feedback can be used to prevent
lers in the CS7 structure give different initial responses the gradual buildup or depletion of reactant A. Thus, the flowrate
because of their opposite actions. An increase in vapor boilup of the other fresh feed F0A is manipulated by a composition

Figure 5. Relay-feedback test results for CS7 pairings.


Ind. Eng. Chem. Res., Vol. 47, No. 22, 2008 8709

Figure 6. A (5% step change in production rate handle ∆VS.


controller than maintains the concentration of component A on reasonable for both loops. These results are also confirmed
a selected tray. by the controller gain and reset time values given in Table
To see which tray is the best for detecting component A 4. There is no inverse response (huge reset times) problem.
concentration, a disturbance sensitivity approach was explored. There is also no action-related problem because both loops
The compositions of both products are driven to their correct have the same action. The unusual spikes in the composition
values by varying reflux ratio and heat input at different feed on tray 20 are caused by the step changes in the F0A feed,
flowrates. The composition and temperature profiles have to which is pure component A and is fed directly on tray 20.
change to give “perfect” composition control of the two To check the effectiveness of this control structure, quite
products. The changes of composition of component A through large (20% step changes in the production rate (fresh-feed
the column at ( 20% of original feed flowrates are shown in stream F0B) are applied to the system. Figure 10 shows the
Figure 8. From this figure, the tray with the least changing response of the CS5 structure to these disturbances. Both
composition (tray 20) is selected as the preferred one to control controlled variables go back to their setpoints for both
by manipulating the flowrate of the other fresh feed F0A. This positive and negative disturbances in a short time. Also, even
tray is also the effluent of the second (middle) reactor returned with these very large (20% step changes, the purities of the
back to the column. product streams xD,C and xB,D settle down smoothly to steady-
Figure 9 gives relay-feedback test results for both loops: state values that are within (1% of the 95% specification
F0A/xA,20 and VS/T16. The amplitudes and periods look value. Therefore the CS5 structure is a much better choice
for controlling this column/side-reactor process compared to
CS7 structure.
The responses of CS5 with tray 20 composition controlled
for changes in feed compositions are given in Figures 11 and

Figure 7. Control structure CS5. Figure 8. Disturbance sensitivity results.


8710 Ind. Eng. Chem. Res., Vol. 47, No. 22, 2008

Figure 9. Relay-feedback test results for CS5 pairings.

Table 4. Tuning Parameters for Control Structure CS5 the 3% feed stream impurity. However, the purity of the distillate
pairing span KC τI stream xD,C goes out of the specification limits for a 5% feed
stream impurity.
F0A/xa,20 0.5 4.79 49.87
VS/T16 10 1.30 7.04
4. Conclusion
12. In Figure 11 the composition of the fresh feed F0A is changed
from pure A to either 95 mol % A with 5 mol % B (solid lines) Two different control structures have been studied for the
or 90 mol % A with 10 mol % B (dashed lines). Control control of a column/side-reactor process with three external
structure CS5 handles this type of disturbance quite well. The reactors. The chemical system studied has two reactants and
purities of both products are maintained within 1% of the desired two products and operates without any excess reactant. The
95% specification. However, as the impurity amount is increased effectiveness of each control structure is demonstrated using
to 10 mol % B, the purity of the bottoms stream xB,D moves to disturbances in production rate and fresh-feed compositions.
the limit of the specifications which is 1% below from the For the first control structure CS7, two tray temperatures are
setpoint. Thus, it can be said that this purity of the bottoms controlled by manipulating the two fresh-feed streams. The
stream will deviate by more than 1% for larger feed composition vapor boilup (or reboiler duty) is the production rate handle.
disturbances. Simulation results indicate that this control structure is not a
Figure 12 gives the responses of CS5 to 3% and 5% good choice for controlling this process. The main problem is
impurities of A in fresh-feed stream F0B. The control structure the disharmony between the actions of the loops. While one of
keeps product purities within 1% of their specified values for the loops has a direct action, the other loop has a reverse action.

Figure 10. Step change ((20%) in production rate handle ∆F0B.


Ind. Eng. Chem. Res., Vol. 47, No. 22, 2008 8711

Figure 11. Impurity of B in fresh feed composition ∆z0A.

Figure 12. Impurity of A in fresh feed composition ∆z0B.

A secondary problem is that one of the loops shows the control structures CS5 and CS7. For both flowsheets the use
characteristic of a process with inverse response. of control structure CS7 results in similar conflicts between
For the second control structure CS5, an internal composi- the actions of the loops and inverse response problems for
tion analyzer is used, and the fresh feed stream F0B is the one of the loops. These problems affect the robustness of
production-rate handle. This control structure shows better control structure CS7 significantly. On the other hand, control
responses to disturbances because of several reasons. First, structure CS5 could easily handle disturbances that are typical
a composition analyzer that supplies information about in most chemical processes for both the column/side-reactor
reactant inventory inside the system that can be used to process and the reactive distillation column.
improve the controllability in a “neat” process. In addition,
the similarity in the actions of the two loops and the absence We have not explored in this paper the possibility of changing
of the inverse response behavior result in a more effective the design to provide a more controllable process. Instead of
control structure compared to CS7. using the economic optimum design, controllability may be
In a previous paper,10 the same control structures have been improved by changing design parameters such as the number
used for a reactive distillation column in the same two and size of reactors and the number of trays in the various
reactants-two products system. Although the column/side- sections of the column. The suboptimal design may result in a
reactor process and the reactive distillation column are process that can be controlled by the two-temperature control
different in terms of reaction-separation interactions, simi- structure, which would eliminate the need for analyzers. We
larities have been observed in their dynamic responses to plan to explore this possibility in a future paper.
8712 Ind. Eng. Chem. Res., Vol. 47, No. 22, 2008

Nomenclature xi,n ) composition of component i in lump n (mole fraction i)


A ) heat transfer area between catalyst and liquid (m ) 2 z0i,j ) composition of fresh feed stream i (mole fraction j)
aF ) pre-exponential factor for the forward reaction (mol-sec-1- Greek Symbols
mol-1)
aR ) pre-exponential factor for the reverse reaction (mol-sec-1- R390 ) relative volatility at 390 K
mol-1) λ ) heat of reaction (kJ-mol-1)
AVP ) vapor pressure constant Fcat ) catalyst density (kg-m-3)
B ) bottoms flow rate in the column (mol-sec-1) Fliq ) liquid density (kg-m-3)
BVP ) vapor pressure constant τI ) reset time (min)
CPcat ) heat capacity of catalyst (kJ-kg-1-K-1) Literature Cited
CPliq ) heat capacity of liquid (kJ-kg-1-K-1)
D ) distillate flow rate in the column (mol-sec-1) (1) Kaymak, D. B.; Luyben, W. L. Optimum design of a column/side
reactor process. Ind. Eng. Chem. Res. 2007, 46, 5175–5185.
EF ) activation energy of forward reaction (cal-mol-1)
(2) Kaymak, D. B.; Luyben, W. L.; Smith, O. J. Effect of relative
ER ) activation energy of reverse reaction (cal-mol-1) volatility on the quantitative comparison of reactive distillation and
F0j ) fresh feed flow rate of reactant j (mol-sec-1) conventional multi-unit systems. Ind. Eng. Chem. Res. 2004, 43, 3151–
Fliq,n ) flow rate of liquid in lump n (mol-sec-1) 3162.
(3) Schoenmakers, H. G.; Buehler, W. K. Distillation column with
KC ) controller gain
external reactors - an alternative to the reaction column. Ger. Chem. Eng.
kF ) specific reaction rate of forward reaction (mol-sec-1-mol-1) 1982, 5, 292–296.
kR ) specific reaction rate of reverse reaction (mol-sec-1-mol-1) (4) Jakobsson, K.; Pyhälahti, A.; Pakkanen, S.; Keskinen, K.; Aittamaa,
KEQ ) equilibrium constant J. Modelling of a side reactor configuration combining reaction and
distillation. Chem. Eng. Sci. 2002, 57, 1521–1524.
KFoj ) steady-state gain related to the fresh feed j (K-mol-1-sec)
(5) Gadewar, S. B.; Tao, L.; Malone, M. F.; Doherty, M. F. Process
Mliq ) molecular weight of the liquid mixture (kg-kmole-1) alternatives for coupling reaction and distillation. Chem. Eng. Res. & Des.
PSj ) vapor pressure of component j (bar) 2004, 82, 140–147.
R ) perfect gas law constant (cal-mol-1-K-1) (6) Ouni, T.; Jakobsson, K.; Pyhälahti, A.; Aittamaa, J. Enhancing
productivity of side reactor configuration through optimizing the reaction
R ) reflux flowrate (mol-sec-1)
conditions. Chem. Eng. Res. & Des. 2004, 82, 167–174.
t ) time (sec) (7) Citro, F.; Lee, J. W. Widening the applicability of reactive distillation
Tcat,n ) temperature of catalyst in lump n (K) technology by using concurrent design. Ind. Eng. Chem. Res. 2004, 43,
Tliq,n ) temperature of liquid in lump n (K) 375–383.
(8) Roat, S. ; Downs. J.; Vogel, E.; Doss, J. Integration of rigorous
Ti ) column temperature on tray i (K) dynamic modeling and control system synthesis for distillation columns.
U ) heat transfer coefficient between catalyst and liquid (kJ-s-1- In Chemical Process Control-CPC III; Elsevier: Amsterdam, The Neth-
m-2-K-1) erlands, 1986.
Vcat ) volume of catalyst in each lump (m3) (9) Kaymak, D. B.; Luyben, W. L. Evaluation of a two-temperature
control structure for a two-reactant/two-product type of reactive distillation
Vliq ) volume of liquid in each lump (m3) column. Chem. Eng. Sci. 2006, 61, 4432–4450.
VR ) molar holdup of the reactor (moles) (10) Kaymak, D. B.; Luyben, W. L. Quantitative Comparison of dynamic
VS ) vapor boilup (mol-sec-1) controllability between a reactive distillation column and a conventional
xB,j ) bottoms composition of component j in liquid (mole fraction multi-unit process. Comput. Clum. Eng. 2008, 32, 1456–1470.
j)
xD,j ) distillate composition of component j in liquid (mole fraction ReceiVed for reView December 13, 2007
j) ReVised manuscript receiVed August 14, 2008
xi,j ) composition of component j in liquid on tray i (mole fraction Accepted September 2, 2008
j) IE701705M

Das könnte Ihnen auch gefallen