Professional Documents
Culture Documents
a r t i c l e i n f o a b s t r a c t
Article history: This work develops a dynamic, first principles-based model of a reactive distillation column used for ben-
Received 14 October 2013 zene hydrogenation of a reformate stream and investigates different control structures for this process.
Received in revised form 3 January 2014 The model is used initially to develop and evaluate a feedback control strategy which provides good reg-
Accepted 7 January 2014
ulatory performance for small disturbances, however, it tends to be sluggish for significant disturbances
Available online 28 February 2014
in the feed composition. In order to address this point, adding a feedforward controller to the feedback
structure has also been investigated. However, the feedforward controller can only be implemented if
Keywords:
composition measurements of the feed are taken. As online composition measurements are expensive
Reactive distillation
Packed column
in practice, several different scenarios have been investigated where samples of the feed are taken and
Dynamic modeling subsequently analyzed in a lab, as represented by measurement time delays. Simulation results show
Feedback control that adding feedforward control to the feedback scheme can be very beneficial for this process, however,
Feedforward control this is only the case if the composition disturbance measurements do not involve a significant time delay.
1. Introduction toluene has a high octane rating (RON) and should be retained in
the final product.
Automotive emissions are a significant contributor to poor air
Benzene (100 RON) + 2H2 → cyclohexane (83 RON) (1)
quality [1]. As such, specifications for automobile fuels obtained
from petroleum have received increasing levels of attention from Toluene (120 RON) + 3H2 → methylcyclohexane (75 RON) (2)
the Environmental Protection Agency (EPA). Benzene is one of the
compounds that is regulated as it is a carcinogen and the EPA In order to avoid problems related to the selectivity of the cata-
requires all refiners to limit the amount of benzene in gasoline to lyst, the reformate stream is split into light and heavy components
0.62 vol% [2]. While benzene in the gasoline pool results from a vari- in the conventional process (Fig. 1a). As benzene is a reasonably
ety of sources, the main contributor is the reformer unit resulting light component of this mixture, it is mostly concentrated in the
in significant amounts of benzene present in reformate streams. distillate, and accordingly, is hydrogenated before being sent to
As the reformate stream is used to boost octane rating, there are the gasoline pool. The downside of this process is that a high cap-
economic objectives that have to be taken into account while com- ital investment is needed. Reactive distillation (Fig. 1b) offers an
plying with environmental regulations. alternative route for solving this problem. By combining reaction
One option to remove benzene is to hydrogenate in the presence with separation it is possible to selectively react one component in
of a catalyst. However, a problem arises as the catalyst used for a specified region of the column while suppressing unwanted reac-
the reaction is not exclusively selective for benzene, and toluene, tions of other components. Furthermore, additional savings can be
which is present in the reformate stream in considerable quantities, achieved as the heat of reaction can directly be used for separation
will also be hydrogenated. Toluene hydrogenation is undesirable as of the mixture.
While reactive distillation (RD) can have significant advantages
over traditional designs, there are also challenges that need
to be considered. The simultaneous presence of reaction and
separation phenomena can result in complex dynamic behavior.
∗ Corresponding author at: Center for Biotechnology and Interdisciplinary Studies,
Combining reaction and separation into a single vessel results in
Rensselaer Polytechnic Institute, Troy, NY 12180, United States.
fewer manipulated variables, thus increasing interactions between
Tel.: +1 518 276 2138; fax: +1 518 276 3035. control loops [3]. RD columns have been observed to be very
E-mail address: hahnj@rpi.edu (J. Hahn). sensitive to changes in feed concentration. This is a crucial aspect
0959-1524/$ – see front matter © 2014 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jprocont.2014.01.005
114 V. Mahindrakar, J. Hahn / Journal of Process Control 24 (2014) 113–124
Fig. 1. Schematics of (a) conventional reaction/separation process and (b) reactive distillation.
models, the vapor and liquid phase are assumed to be in equi- multi-loop control system [27]. A major concern with multi-loop
librium. NEQ stage models use rate-based equations to describe control is the presence of process interactions, i.e., each manipu-
the mass transfer occurring in conventional distillation columns. lated variable may affect multiple controlled variables, however,
NEQ stage models are generally based on the use of rigorous these interactions are not taken into account for the control
Maxwell–Stefan equations for estimating heat and mass transfer structure design. Multi-loop control systems may not provide
rates across the interface. Many papers have presented EQ models satisfactory control in some scenarios and multivariable control
[6,8,12,13,19–21] and NEQ models [16,17,20–22] for reactive dis- strategies, such as model predictive control and decoupling control
tillation. Most of the NEQ models developed for reactive distillation can provide better control. However, multi-loop control is the
are generally steady state models [20–22]. However, Peng et al. [23] most widely used control strategy for distillation columns because
have compared the results of dynamic NEQ and EQ model, and con- of its simplicity, both in terms of maintenance and controller
cluded that the results are similar for their case. Contrary to this, tuning. As no work has been done on modeling and control of
Baur [17] pointed out that the responses from the models may differ benzene hydrogenation via reactive distillation, this work focuses
quantitatively and the dynamics are influenced by column speci- on traditional control strategies and advanced control will be
fications. NEQ models are generally more challenging to simulate investigated in the future.
[17,23] and require thermodynamic properties for the calculation The commonly used control structures for reactive distillation
of mass transfer coefficients and interfacial areas. are similar to those of conventional distillation columns. Skogestad
The dynamic behavior of a distillation column is strongly influ- and co-workers [28–30] have discussed the selection of controlled
enced by fluid hydraulics in the column. This is even more so in and manipulated variables. Most distillation columns generally use
reactive distillation columns, as liquid hold-ups, and liquid res- either a 4 × 4 control or a 5 × 5 control structure. These configura-
idence times are important for determining the conversion and tions refer to the number of manipulated variables and the number
selectivity of the reactive distillation column. Very few dynamic of controlled variables used. Eq. (3) presents the set of manipulated
models consider both liquid and vapor holdup in the columns. variables u and controlled variables y that are used in 5 × 5 systems
The vapor holdups are generally neglected because of the low in no particular order of correlation; see Fig. 2 for an illustration
density of vapor in comparison to liquid. Also, considering vapor of the variables. Columns having a 4 × 4 control system frequently
holdups leads to additional computational difficulties in the model.
As such, most dynamic models consider only dynamic liquid hold
up or in some cases a constant liquid holdup [23]. However, Choe
and Luyben [24] suggest that vapor holdup should be considered
for dynamic models of columns operating at pressures greater
than 5–10 atm. Equations governing the vapor and liquid flows,
and hold-ups in a packed column have been discussed by many
papers: Bemer and Kalis [25] have given equations for liquid hold-
up and pressure drops in irrigated columns while Mackowiak [26]
has extensively reviewed methods for determining vapor flow in
packed columns.
do not use the pressure at the top of column, P1 , as a controlled The relative gains are arranged to form the matrix
variable [30]. ⎡ ⎤
11 12 ··· 1n
⎛ ⎞ ⎛x ⎞ ⎢ ···
⎥
L1 ⎢ 21 22 2n ⎥
=⎢ ⎥
1
⎢. ⎥ (5)
⎜ ⎟ ⎜x ⎟ .. .. ..
⎜ VN , QN ⎟ ⎜ N ⎟ ⎣ .. ⎦
⎜ ⎟ ⎜ ⎟ . . .
u=⎜
⎜ D ⎟
⎟ y=⎜ ⎟
⎜ M1 ⎟ (3) n1 n2 ··· nn
⎜ L ⎟ ⎜ ⎟
⎝ N ⎠ ⎝ MN ⎠ Controlled and manipulated variables are paired such that the
V2 , QD corresponding relative gains are positive and as close to one as
P1
possible. While the above definition of RGA may seem difficult to
estimate directly for real systems, the RGA can be determined from
The 5 × 5 system shown in Fig. 2 is generally used for columns an open-loop gain matrix. The procedure for computing the RGA
with a total condenser. For these columns, the pressure is typi- from an open-loop gain matrix, K, is given by
cally controlled by manipulating the condenser heat removal as
T
the condenser temperature and the column pressure are directly H = (K −1 ) (6)
linked for a total condenser. Partial condensers are used when there
=K ⊗H (7)
are very light components in the column feed that would require
a high column pressure or a low condenser temperature. In a col- where ⊗ denotes the Schur product (element by element multipli-
umn with a partial condenser, vapor is removed from the condenser cation) [27].
as a vapor stream. The pressure of the column is strongly influ- The controllers used for feedback control loops are generally
enced by the outlet vapor stream flow rate. Luyben [31] and Hori controllers of PID-type, where PI controllers are the most com-
and Skogestad [32] have discussed control structures for columns monly used ones. Luyben [31] has pointed out that flow controllers
with partial condensers. Sloley [33] has extensively reviewed con- that regulate the inventory of a column, e.g., the distillate and bot-
trol strategies that can be used for columns based on the type of tom flow, should be proportional-only controller as the inventory
condenser. in the column is sufficiently large to overcome the effect of offsets
Recycle loops in chemical plants are known to significantly alter that may occur due to the use of proportional-only controllers.
the control and dynamics of process networks [34–36]. Recycle
streams increase the overall time constants, thus “slowing down” 2.2.3. Feedforward control
its overall response [34]. Designing a control structure for process Feedback control does not take corrective action until after
networks with recycle can be challenging because recycle streams deviations in the controlled variables occur. As the effects of feed
can induce a time scale separation, where the dynamics of the pro- disturbances will only be detected after a while, this lack of predic-
cess evolves at a fast time scale, and the dynamics of the overall tive control can limit the overall column performance, especially if
process with recycle at a slower time scale [37]. Dynamical analy- the column includes large time constants or time delays. One option
sis and control of such process networks with recycle have received is to also include feedforward control in addition to feedback con-
considerable attention [37]. One of the first design procedures was trol in a control structure. Feedforward control systems measure
proposed by Buckley [38] and has been widely used in industry the disturbance variables and take corrective action before upsets
for many years [39]. The first step is to design a control struc- of the controlled variables can be recorded. The main disadvan-
ture that handles the inventory of the entire process (liquid levels tage of feedforward control is that the disturbance variable must
and gas pressures). This “hydraulic” structure provides smooth be measured online which is not always feasible, physically or for
flow rate changes. Fast-acting proportional-only level controllers economic reasons.
provide the most simple and most effective way to achieve this The basic idea for feedforward controller is to measure the
flow smoothing [39]. The second step is to close the “product qual- disturbance affecting the system and compute a change of the
ity” loops. These loops typically use slower proportional–integral manipulated variable such that the effect of the disturbance on
controllers to hold product streams as close to the specification the controlled variable is canceled by the change of the manipu-
as possible. Larsson and Skogestad [40] and Luyben et al. [41] have lated variable [44]. Feedforward controller designs are thus based
discussed plant-wide controller design procedures for a large num- on process models. A feedforward controller transfer function Gf
ber of measurements and control loops. When applied to a smaller for a system can be given by [27]
single unit scale, they mirror the steps that have been mentioned
Gd
above. Gf = − (8)
Gp Gmf
(11)
Table 2
Open loop transfer functions.
R QN Q1
Table 3
RGA for RD column at different feed benzene concentrations of (a) 6 vol%, (b) 3 vol%, and (c) 11 vol%.
(a) Feed benzene concentration of 6 vol% (b) Feed benzene concentration of 3 vol% (c) Feed benzene concentration of 11 vol%
R QN Q1 R QN Q1 R QN Q1
3.2.2. Feedback controller design benzene concentrations for a significant period of time. As such it
Both the reflux drum and reboiler drum need controllers to reg- was hypothesized that adding feedforward action to this structure
ulate the flows and maintain specified liquid levels in the vessels. will improve the performance. The off-spec concentration of ben-
The reflux drum also holds vapor which needs to be regulated such zene in the product stream can be reduced if the flow of hydrogen
that the pressure of the column is maintained. Since the stream to the column is regulated according to the feed composition.
flow rates regulate the inventory of the column, P-only controllers Based on the disturbance variables, manipulated variables, and
have been used. These three proportional controllers were tuned controlled variables, Gp and Gd from Eq. (8) are defined as follows:
via Ziegler Nichols tuning relations. A PI controller was used for
maintaining the reflux drum outlet vapor flow rate V1 in order V1
Gp = (15)
to avoid an offset in the column pressure at the top (P1 ). PI con- FH2
trollers were also used for the temperature point control loops. All
the PI controllers were tuned using internal model control (IMC) V1
Gd = (16)
tuning relations [41]. The transfer functions shown in Table 2 were zC6 H6
used to compute the controller parameters for temperature control
loops. A transfer function (13) was obtained for the fresh hydrogen These transfer functions were determined using the MATLAB
feed (FH2 ) control loop by passing step change inputs to the model system identification toolbox on simulated data for open-loop step
without the recycle stream. responses:
V1 75.265
GPwithout recycle = = (13) 99.39
FH2 1 + 678.4 s Gp = (17)
1 + 3269.9 s
An IMC controller was designed based on transfer function (13)
which was determined for the distillation column without the −7.457 × 104
Gd = (18)
recycle loop in accordance with design procedure described for 1 + 3580.2 s
plant-wide control [38–40]. Table 4 shows the controller tuning
In order to keep the process realistic, feedforward control with
parameters that were derived and used for the control loops oper-
different levels of measurement delay
mf :
ating on the model. The values of c were chosen by adjusting the
desired speed of the closed-loop response. The faster the desired
Gmf = e−
mf s (19)
response, the lower the value of c . As faster response can lead
to larger overshoots, c needs to be chosen to achieve a trade-
off between speed of response and potential for overshoot. Chien were used for the measurement transfer function. Also, since the
and Fruehauf [48] have given the following general guideline to feedforward controller is represented by a lead-lag element, the
determine acceptable values of c for FOPTD systems, controller transfer function was augmented with a filter with a
time constant of 120 s in order to avoid large sudden changes in
> c >
(14) the manipulated variable. The resulting dynamic feedforward con-
troller is given by
where
is the time delay. The values of c chosen for the PI con-
trolled loops are listed in Table 4. 1 + 3269.9 s 1
Gf = 750.3 (20)
1 + 3580.2 s 1 + 120 s
3.3. Feedforward controller
If a static feedforward control law would be considered then this
Most conventional columns use feedback-only control [30]. would result in
As reactive distillation columns can be viewed as an exten-
sion of conventional columns, the first approach was to design a Gf = 750.3 (21)
feedback-only control structure, appropriately tune the controllers,
and observe the performance. However, as will be shown in Section Fig. 3 shows the control structure used for the column along
4, feed composition disturbances affect the column performance with the feedforward controller for controlling the fresh hydrogen
adversely, i.e., the product does not meet the specifications for feed.
120 V. Mahindrakar, J. Hahn / Journal of Process Control 24 (2014) 113–124
Table 4
Feedback controller settings.
Manipulated variable Controlled variable Kc i (s) c (s) Type of controller and tuning method used
1.766 × 107
Q1 T1 1887.1 110 PI (IMC)
c
2.305 × 108
QN T19 5817.4 582 PI (IMC)
c
342.15
R TN − 3593.3 610 PI (IMC)
c
9.013
FH2 V1 678.4 128.7 PI (IMC)
c
D h1 3340 – – P (Ziegler Nichols)
LN hN 3830 – – P (Ziegler Nichols)
V1 P1 −12.5 – – P (Ziegler Nichols)
Table 5
Steady state results for inlet and outlet streams.
4. Investigation and comparison of different control react as much benzene as possible while minimizing the amount
schemes for the column of toluene entering the stage containing catalysts packing. It can be
seen that the benzene mole fraction increases along the height of
4.1. Steady state results the column, until the reactive zone, where it decreases due to the
reaction. Very little toluene is present in the catalyst zone of the
The RD column model is assumed to initially operate at the same RD column and as a result 99.8% of the toluene from the reformate
steady state for any of the comparisons of different control schemes stream is retained in the bottom stream.
made in this section. This steady state corresponds to nominal feed
(composition and temperature) being fed to the column, where the
4.2. Feedback controller results
benzene concentration is 6 vol%. Table 5 includes the steady state
values of all feed and product streams.
The rigorous model built in gPROMS was augmented with the
Fig. 4 depicts the profile of benzene and toluene concentration
control structure shown in Fig. 3. The controllers specified in Table 4
at steady state in the column. The objective of the RD column is to
were used for investigation. The model combined with the con-
trollers was subjected to changes in the inputs, representing step
disturbances that occur after 1hr, to evaluate the performance of
Benzene molfraction the control schemes. The set-points of all the controlled variables
0.6 Toluene molfraction
remain unchanged throughout this investigation resulting in a reg-
0.5 ulatory control problem.
Fig. 5 depicts the benzene concentration in the product for the
Mole fraction
0.4
Catalyst column under feedback-only control subjected to step changes in
Zone
0.3 the temperature of ±5 K (Fig. 5(a)) and the feed flow rate of ±5%
(Fig. 5(b)). The responses indicate that the effect of the disturbances
0.2
on the product benzene concentration is not significant, i.e., only
0.1 small changes can be seen in the benzene concentration and the
0 concentration stays far below the allowable limit.
10 20 30 40 50 60 70
Stage number
One of most common disturbances for the benzene hydrogena-
tion process is a change in the feed composition. The benzene
Fig. 4. Profiles of benzene and toluene in RD column. concentration in the reformate stream can increase up to a value
V. Mahindrakar, J. Hahn / Journal of Process Control 24 (2014) 113–124 121
a) Feed temperature -5 K
b) 0.27
0.26
Feed flowrate +5%
Feed temperature +5K Feed flowrate -5%
Benzene vol%
Benzene vol%
0.26
0.255
0.25
0.25
0.245 0.24
0 5 10 15 0 5 10 15
Time (hours) Time (hours)
Fig. 5. Responses to step changes in (a) feed temperature and (b) feed flow rate.
of 11 vol% [4]. In order to evaluate such a scenario, a step change the benzene concentration. Use of feedforward control can reduce
was given to the feed benzene composition from 6 vol% to 11 vol% the overshoot and settling time. All subfigures in Fig. 6 include a
and the concentrations of all other components were reduced pro- comparison of the responses of the column for feedback-only con-
portionally. The graphs in Fig. 6 labeled “feedback-only” show the trol and feedforward–feedback control to a step disturbance in the
responses of the outlet benzene concentration and of the other feed composition from 6 vol% to 11 vol% benzene. It can be clearly
manipulated variables for a step change in the feed concentration. seen that there is significant improvement in the response time
The steady state benzene concentration meets the specifications. and the overshoot when feedforward control is added to the exist-
However, it can be seen that the responses have significant over- ing feedback control scheme. Additionally, the trajectories of all
shoot and also a large settling time under a feedback-only control manipulated variables also remain within reasonable bounds for
scheme. This situation presents a clear opportunity for feedforward feedforward–feedback regulatory control.
control in order to minimize the effect of the disturbance on the While Fig. 6 shows the positive impact that the addition of feed-
controlled variable. forward control can have on the process, the simulation assumed
that concentration measurements are available instantaneously.
4.3. Comparison of feedback and feedforward/feedback control This is not a very realistic assumption in practice unless an online
schemes analyzer is used. One commonly used alternative is that samples
from streams are taken and then analyzed in a lab, i.e., measure-
The simulations shown in Fig. 6 for feedback-only control ments will only be available at discrete points in time and with
involve significant overshoot and as a result a considerable amount a certain time delay. As such, it is important to know how much
of product does not meet the EPA specifications of 0.62 vol% [2] for of an effect measurement time delay has on the performance of
Fig. 6. Responses of controlled and manipulated variables for step change in feed composition: (a) benzene concentration of the product, (b) fresh H2 feed – FH2 , (c) condenser
duty – Q1 , (d) reboiler duty – QN , and (e) reflux ratio – R.
122 V. Mahindrakar, J. Hahn / Journal of Process Control 24 (2014) 113–124
Feedback-only that is occurring at some point during the operation among all cases
1 FF-FB with continuous measurements is not occurring for the feedback-only control scheme but instead
FF-FB with sampling time of 15min
FF-FB with sampling time of 30min for feedforward–feedback control with significant time delays. It
Benzene vol%
0.8 FF-FB with sampling time of 45min can be clearly seen from Table 6 that the larger the time delay for
FF-FB with sampling time of 60min
the composition measurement, the less of a benefit in the over-
0.6 all reduction of the offspec product can be achieved. At the same
time, the largest deviations from the target are occurring for long
0.4 measurement time delays. It is beyond the scope of this study to
evaluate these responses for different design specifications. How-
0.2
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
ever, in order to put the discussion of the performance for different
Time (hours) measurement delays into a more general perspective, it should be
pointed out that the dominant time constant of the systems is equal
Fig. 7. Response to step change in feed composition for feedback-only control and to 1.6 h. It can be concluded that the feedforward–feedback scheme
feedforward–feedback control for five different sampling times (continuous, 15 min, is superior to feedback-only control for the case of continuous mea-
30 min, 45 min, 60 min).
surements or measurements with a time delay of up to 15 min,
which corresponds to 15% of the dominant time constant. There is
the feedforward–feedback (FF–FB) control system. This case has only a marginal benefit to the feedforward–feedback scheme if the
been analyzed next. Fig. 7 shows a comparison of the responses for time delay is 30 min, corresponding to 31% of the dominant time
feedback-only control and combined feedforward–feedback con- constant, and it is questionable if there are any benefits of includ-
trol where the disturbance measurement transfer function Gmf uses ing feedforward control if the measurement time delay is 45 min
several different time delays corresponding to the sampling times. or more, corresponding to 46% of the dominant time constant.
As before, the step change disturbance occurs after an hour and the In addition to investigating a dynamic feedforward controller, a
system is initially at steady state. static feedforward controller has also been investigated. Table 6b
It can be clearly seen that longer measurement time delays sig- shows a comparison of the deviation values computed using the
nificantly degrade the advantages that the addition of feedforward static feedforward controller given by Eq. (21) instead of the
control has on the performance. In order to quantify the perfor- dynamic feedforward controller from Eq. (20). The results are close
mance of the responses shown in Fig. 7, a measure is defined to to those obtained by using a dynamic feedforward controller. In
quantify the area between the curve and the EPA specification of fact, the graphs of the responses overlap with those depicted in
0.62 vol% [2]: Fig. 7 for dynamic feedforward control and as such no separate fig-
∞ ure for the graphs is included. Thus, from an application point of
Dev = H(Bz vol% − 0.62) × (Bz vol% − 0.62) dt (22) view, a simple static feedforward controller could be used instead
0 of a dynamic feedforward controller without significant loss of per-
where H is the Heaviside function defined as formance.
One last point to consider is that this investigation focused on
0, x < 0 step disturbances as these are the most common disturbances for
H(x) = (23)
1, x≥0 the scenario investigated in this work. As such, it was appropriate
to model the measurements occurring from lab samples as contin-
Table 6 shows a comparison of the deviation value computed
uous samples with a time delay instead of using discrete samples
from Eq. (22) for different sampling times. Not surprisingly, it can
with time delays as the two will return identical results for step
be concluded that feedforward control reduces the upset condition,
disturbances. However, it should be pointed out that if the benzene
as measured by Eq. (22), for all investigated cases. Similarly, it is also
concentration disturbances would have had a different nature than
not surprising that the best performance is achieved for the com-
a step, that it would have been required to use discrete sampling
bined feedforward–feedback control structure that uses an online
and time delays. This was not necessary for the cases investigated
composition analyzer. However, the largest benzene concentration
in this work, though.
It was one of the goals of this investigation to determine the
Table 6 benefit of using a control scheme that combines feedforward and
Deviations, as measured by Eq. (22), for (a) dynamic feedforward controllers and (b)
feedback control over a feedback-only control scheme. The simu-
static feedforward controllers.
lation results indicate that a significant benefit only exists if upsets
(a) in the feed composition can be quickly detected.
Control structure Dev % reduct. Max Bz
vol% 5. Conclusions
Feedback-only 1074 0.85
FF–FB with continuous measurements 423 61% 0.66 Benzene hydrogenation via reactive distillation is a process
FF–FB with sampling time of 15 min 559 48% 0.76 that has found significant use in the process industries. However,
FF–FB with sampling time of 30 min 756 30% 0.97
no models of this process can be found in the open litera-
FF–FB with sampling time of 45 min 874 19% 1.02
FF–FB with sampling time of 60 min 917 15% 0.94
ture. This paper addresses this point by developing a dynamic
equilibrium-based model for a reactive distillation column used
(b) for the hydrogenation of benzene. Simulations were carried out to
determine transfer functions between manipulated and controlled
Control structure Dev % reduct. Max Bz
vol% variables. Control loop pairing was performed using RGA analy-
sis and the feedback controllers were tuned via IMC tuning (PI)
Feedback-only 1074 0.85
FF–FB with continuous measurements 444 59% 0.67
and Ziegler Nichols tuning (P). a model-based feedforward con-
FF–FB with sampling time of 15 min 543 49% 0.74 troller was also designed to reduce upset conditions caused by
FF–FB with sampling time of 30 min 756 30% 1.04 disturbances.
FF–FB with sampling time of 45 min 888 17% 1.13 Simulations indicate that the column performance for feed tem-
FF–FB with sampling time of 60 min 939 13% 1.04
perature and feed flow rate disturbances remains acceptable for
V. Mahindrakar, J. Hahn / Journal of Process Control 24 (2014) 113–124 123
∗
a feedback-only control scheme. However, the column has a sig- yj,i = yj+1,i (1 − ) + yj,i (A11)
nificant settling time for disturbances in the feed composition.
Feedforward control can reduce these upset conditions resulting Energy balance
from feed disturbances. However, it was shown that the use of a d(Mlj Hlj + Mvj Hvj )
feedforward–feedback control scheme is only beneficial if the time = Vj+1 Hj+1 + Lj−1 Hlj−1 − Lj Hlj − Vj Hvj − Hrxn,j
dt
delay associated with the feed composition measurement is small.
(A12)
In summary, this paper (1) presented the first detailed model of
a reactive distillation column for the hydrogenation of benzene, (2) Flow rate and holdups
designed and evaluated a feedback control scheme for the column,
Liquid
and (3) investigated the benefit of using a feedforward/feedback
control structure for different sampling times of the feed composi- Mj vollj = hj (HETP)A (A13)
tion measurement.
2/3
hj = 0.34a1/3 uj (A14)
Acknowledgment
Vapor
The authors gratefully acknowledge partial financial support Pj A(HETP)(ε − hj ) = Mvj Rgas Tj (A15)
from the American Chemical Society - Petroleum Research Fund
(Grant PRF# 50978-ND9). P0,j (1 − ε) u2v v
= (A16)
HETP ε3 dp K
Appendix A. 150
= + 1.75 (A17)
Rev
A.1. Model
−5
2hj
Condenser and reflux drum Pj = P0,j 1 − (A18)
dp a
Mass balance
Pj+1 = Pj + Pj (A19)
d(Ml1 + Mv1 )
= V2 − L1 − D − V1 (A1) Reaction rate
dt
kj,1 KA1 KH1 CAj,1 CHj
Component balances rxnj,Benzene = − (A20)
3
d(Ml1 x1,i + Mv1 y1,i ) (3KA1 CAj,1 + (KH1 CHj )1/2 + 1)
= V2 y2,i − L1 x1,i − V1 y1,i ∀i : 1 to n − 1 (A2)
dt kj,2 KA2 KH2 CAj,2 CHj
n
n rxnj,Toluene = − 3
(A21)
x1,i = 1; y1,i = 1 (A3) (3KA2 CAj,2 + (KH2 CHj )1/2 + 1)
i=1 i=1
Ea 1 1
ki,j = ki,0 exp − − (A22)
Vapor–liquid equilibrium Rgas Tj T0
x1,i l1,i = y1,i v1,i ∀i : 1 to n (A4) Reboiler
Energy balance Mass balance
[1] Mobile Source Air Toxics: Control of Hazardous Air Pollutants from Mobile
Sources, United States Environmental Protection Agency, EPA420-R-05-901,
n
n 2005, November.
xj,i = 1; yj,i = 1 (A9) [2] Control of Hazardous Air Pollutants from Mobile Sources: Final Rule to Reduce
Mobile Source Air Toxics, United States Environmental Protection Agency,
i=1 i=1 EPA420-F-07-017, 2007, February.
[3] G.J. Harmsen, Reactive distillation: the front-runner of industrial process inten-
Vapor–liquid equilibrium and Murphree efficiency sification: a full review of commercial applications, research, scale-up, design
∗ and operation, Chemical Engineering and Processing: Process Intensification
xj,i lj,i = yj,i lj,i ∀i : 1 to n (A10) 46 (9) (2007) 774–780.
124 V. Mahindrakar, J. Hahn / Journal of Process Control 24 (2014) 113–124
[4] Control of Hazardous Air Pollutants from Mobile Sources: 40 CFR Parts 59, 50, [25] G.G. Bemer, G.A.J. Kalis, A new method to predict hold-up and pressure drop
85 and 86, Environmental Protection Agency, EPA-HQ-OAR-2005-0036, 2006, in packed columns, Transactions of the Institution of Chemical Engineers 56
February. (1978) 200.
[5] R. Taylor, R. Krishna, Modelling reactive distillation, Chemical Engineering Sci- [26] J. Mackowiak, Fluid Dynamics of Packed Columns: Principles of the Fluid
ence 55 (22) (2000) 5183–5229. Dynamic Design of Columns for Gas/Liquid and Liquid/Liquid Systems,
[6] K. Alejski, F. Duprat, Dynamic simulation of the multicomponent reactive dis- Springer, New York, 2010.
tillation, Chemical Engineering Science 51 (18) (1996) 4237–4252. [27] D.E. Seborg, D.A. Mellichamp, T.F. Edgar, F.J. Doyle III, Process Dynamics and
[7] D.B. Kaymak, W.L. Luyben, Evaluation of a two-temperature control structure Control, Wiley, New York, 2010.
for a two-reactant/two-product type of reactive distillation column, Chemical [28] S. Skogestad, M. Morari, Control configuration selection for distillation
Engineering Science 61 (13) (2006) 4432–4450. columns, AIChE Journal 33 (10) (1987) 1620–1635.
[8] M.C. Georgiadis, M. Schenk, E.N. Pistikopoulos, R. Gani, The interactions of [29] S. Skogestad, P. Lundström, E.W. Jacobsen, Selecting the best distillation control
design control and operability in reactive distillation systems, Computers & configuration, AIChE Journal 36 (5) (1990) 753–764.
Chemical Engineering 26 (4) (2002) 735–746. [30] S. Skogestad, Dynamics and control of distillation columns – a critical survey,
[9] P. Kumar, N. Kaistha, Decentralized control of a kinetically controlled ideal reac- Modeling, Identification and Control 18 (3) (1997) 177–217.
tive distillation column, Chemical Engineering Science 63 (1) (2008) 228–243. [31] W.L. Luyben, Alternative control structures for distillation columns with par-
[10] G.R. Gildert, K. Rock, T. McGuirk, Advances in process technology through cat- tial condensers, Industrial & Engineering Chemistry Research 43 (20) (2004)
alytic distillation, in: Proceeding of the International Symposium on Large 6416–6429.
Chemical Plants, Antwerpen, 1998, pp. 103–113. [32] E.S. Hori, S. Skogestad, Control structure selection of a deethanizer column with
[11] M.G. Sneesby, M.O. Tade, T.N. Smith, Two-point control of a reactive distillation partial condenser, in: Proceedings of European Congress of Chemical Engineer-
column for composition and conversion, Journal of Process Control 9 (1) (1999) ing (ECCE-6), Copenhagen, 2007.
19–31. [33] A.W. Sloley, Effectively control column pressure, Chemical Engineering
[12] M.G. Sneesby, M.O. Tade, T.N. Smith, Steady-state transitions in the reactive Progress 97 (1) (2001) 38–48.
distillation of MTBE, Computers & Chemical Engineering 22 (7) (1998) 879–892. [34] M.M. Denn, R. Lavie, Dynamics of plants with recycle, Chemical Engineering
[13] M.G. Sneesby, M.O. Tade, R. Datta, T.N. Smith, ETBE synthesis via reactive dis- Journal 24 (1) (1982) 55–59.
tillation. 2. Dynamic simulation and control aspects, Industrial & Engineering [35] N. Kapoor, T.J. McAvoy, T.E. Marlin, Effect of recycle structure on distillation
Chemistry Research 36 (5) (1997) 1870–1881. tower time constants, AIChE Journal 32 (3) (1986) 411–418.
[14] D.A. Bartlett, O.M. Wahnschafft, Dynamic simulation and control strategy [36] J. Morud, S. Skogestad, Effects of recycle on dynamics and control of
evaluation for MTBE reactive distillation, in: Foundations of Computer-Aided chemical processing plants, Computers & Chemical Engineering 18 (1994)
Process Operation, AIChE Symposium Series, vol. 320, Snowbird, 1998, pp. S529–S534.
315–321. [37] A. Kumar, P. Daoutidis, Nonlinear dynamics and control of process systems with
[15] N. Sharma, K. Singh, Model predictive control and neural network predic- recycle, Journal of Process Control 12 (4) (2002) 475–484.
tive control of TAME reactive distillation column, Chemical Engineering and [38] P.S. Buckley, Techniques of Process Control, Wiley, New York, 1964.
Processing: Process Intensification 59 (2012) 9–21. [39] W.L. Luyben, Dynamics and control of recycle systems. 1. Simple open-loop
[16] A.M. Katariya, R.S. Kamath, K.M. Moudgalya, S.M. Mahajani, Non-equilibrium and closed-loop systems, Industrial & Engineering Chemistry Research 32 (3)
stage modeling and non-linear dynamic effects in the synthesis of TAME (1993) 466–475.
by reactive distillation, Computers & Chemical Engineering 32 (10) (2008) [40] T. Larsson, S. Skogestad, Plantwide control – a review and a new design proce-
2243–2255. dure, Modeling, Identification and Control 21 (4) (2000) 209–240.
[17] R. Baur, R. Taylor, R. Krishna, Dynamic behaviour of reactive distillation columns [41] M.L. Luyben, B.D. Tyreus, W.L. Luyben, Plantwide control design procedure,
described by a nonequilibrium stage model, Chemical Engineering Science 56 AIChE Journal 43 (12) (1997) 3161–3174.
(6) (2001) 2085–2102. [42] E.H. Bristol, On a new measure of interactions for multivariable process control,
˛
[18] A. Kołodziej, M. Jaroszyński, W. Sałacki, W. Orlikowski, K. Fraczek, M. Klöker, IEEE Transactions on Automatic Control AC-II (1966) 133.
A. Górak, Catalytic distillation for TAME synthesis with structured catalytic [43] F.G. Shinskey, Process Control Systems: Application, Design and Tuning,
packings, Chemical Engineering Research and Design 82 (2) (2004) 175–184. McGraw-Hill, Inc., New York, 1990 (Chapter 7).
[19] R. Jacobs, R. Krishna, Multiple solutions in reactive distillation for methyl tert- [44] B.W. Bequette, Process Control: Modeling, Design, and Simulation, Prentice Hall
butyl ether synthesis, Industrial & Engineering Chemistry Research 32 (8) Professional, New Jersey, 2003.
(1993) 1706–1709. [45] S. Toppinen, T.K. Rantakylä, T. Salmi, J. Aittamaa, Kinetics of the liquid-phase
[20] R. Baur, A.P. Higler, R. Taylor, R. Krishna, Comparison of equilibrium stage and hydrogenation of benzene and some monosubstituted alkylbenzenes over a
nonequilibrium stage models for reactive distillation, Chemical Engineering nickel catalyst, Industrial & Engineering Chemistry Research 35 (6) (1996)
Journal 76 (1) (2000) 33–47. 1824–1833.
[21] J. Peng, S. Lextrait, T.F. Edgar, R.B. Eldridge, A comparison of steady-state [46] E.S. Hori, S. Skogestad, Selection of controlled variables: maximum gain rule and
equilibrium and rate-based models for packed reactive distillation columns, combination of measurements, Industrial & Engineering Chemistry Research
Industrial & Engineering Chemistry Research 41 (11) (2002) 2735–2744. 47 (23) (2008) 9465–9471.
[22] A.P. Higler, R. Taylor, R. Krishna, Nonequilibrium modelling of reactive distilla- [47] W.L. Luyben, Evaluation of criteria for selecting temperature control
tion: multiple steady states in MTBE synthesis, Chemical Engineering Science trays in distillation columns, Journal of Process Control 16 (2) (2006)
54 (10) (1999) 1389–1395. 115–134.
[23] J. Peng, T.F. Edgar, R.B. Eldridge, Dynamic rate-based and equilibrium models [48] I.-L. Chien, P.S. Fruehauf, Consider IMC tuning to improve controller perfor-
for a packed reactive distillation column, Chemical Engineering Science 58 (12) mance, Chemical Engineering Progress 86 (1990) 33–41.
(2003) 2671–2680.
[24] Y.S. Choe, W.L. Luyben, Rigorous dynamic models of distillation columns, Indus-
trial & Engineering Chemistry Research 26 (10) (1987) 2158–2161.