Professional Documents
Culture Documents
The efficient control of an exothermic packed bed reactor is demonstrated in concert with the distributed
parameter state estimation of temperature and concentration profiles and on-line estimation of the catalyst
activity profile. The position-dependent catalyst activity is estimated using recursive least squares with a
variable forgetting factor, a distributed mass balance model and the current estimated temperature profile.
The temperature level in the reactor is regulated through control of the maximum estimated reactor
temperature using the reactor coolant temperature as the manipulated variable. These methods are demon-
strated both through closed-loop simulation and experiments with a laboratory-scale reactor for the partial
oxidation of methanol. The experiments included both setpoint changes and disturbance rejection. The
estimation and control system effectively overcomes errors in initial conditions, heat transfer parameters
and kinetic parameters.
The packed bed chemical reactor is one of the most numerical results. There are two possible approaches:
difficult challenges for a process control engineer. First, early lumping and late lumping of the system model. The
the reactor is a distributed system: the state variables of most straightforward method is to discretize the distri-
temperature and composition depend upon position as buted system at the earliest opportunity into an approxi-
well as time. Thus, the output variables are limited, the mate model usually consisting of a set of ordinary differ-
states to be controlled must be chosen from an infinite ential equations. Another approach to discretization is to
number of possibilities and the control actions must be pick a limited number of inputs and outputs and identify
based on a limited number of measurements. Often some an empirical system model relating the outputs to the
form of state estimation must be used in order to control inputs. In either case, the control problem can then be
unmeasured state variables and to overcome measure- solved by techniques for lumped multivariable systems.
ment errors. If the control designer desires to apply the full distri-
For many such reactors, the maximization of the yield buted parameter control theory, then ‘late lumping’ dis-
of the desired product is the key control objective. How- cretizes the equations for the implemented controller
ever, this can be difficult to measure directly within the after the control design has been completed’. In almost
time scale of typical reactor dynamics. In addition, the all previous work in the literature, early lumping has
manipulated variables often have complicated effects on been used for control of packed bed reactors.
reactor yield. These complications are due to the Arrhen- Several experimental investigations have been con-
ius temperature dependence of the reaction rate, non- ducted for control of different types of packed bed reac-
linear kinetic expressions, complex reaction networks tors: adiabatic reactor+, autothermal reactors4 and
and widely varying disturbance propagation rates in the non-adiabatic wall-cooled reactors@. Mandler et al.‘-*
solid and gas phases of the reactor. have used a collocation model based on a laboratory
In order to implement model-based control for a dis- non-adiabatic reactor for control studies via simulation.
tributed parameter system (DPS) such as a packed bed The previous ‘early lumping’ controller design tech-
tubular reactor, the system must be ‘lumped’ to obtain niques can be divided into two classes:
0959-1524/92/010023-20
0 1992 Butterworth-Heinemann Ltd J. Proc. Cont. 1992, Vol2, No 1 23
Control scheme for packed bed reactors. I: L. C. Windes and W. H. Ray
Methanol
feed
Feed pump
Pressure indicator
Electrical heater
large and unwieldly number of state variables. than those resulting from changes in catalyst activity.
Order-reduction methods can be employed. Gilles and coworkers (Gilles and Epple13, Gilles and
l Selection of a limited number of state variables Ruppel14, Epple15) have developed an approximate distri-
which are measured directly or determined on-line buted model for control problems with constant profile
via a state estimator. Either the model equations structure and applied it to packed bed reactors.
or identification methods are used to arrive at an In the present paper, we make use of late lumping and
input-output model for this low-order system. carry out the control system design making use of the full
distributed model. We are able to take advantage of the
distributed features of the reactor behaviour in the
One common approach, first used by Vakil et aLg, is to design. In order to demonstrate the applicability and
discretize the reactor equations by orthogonal colloca- performance of the design procedure, an experimental
tion and apply linear-quadratic state estimation and reactor having product yield limitations was controlled
control. Among those using a small number of input- using the proposed design procedure.
output variables, adaptive control has been used for both The specific reactor studied here is a pilot scale, jacket
SISO systemslo and MIMO application&‘. Internal cooled, single-tube reactor with an exothermic reaction
model control and structured singular value analysis taking place within porous catalyst pellets (Figure I). The
have been used to control the reactor outlet tempera- reactor is designed to carry out the partial oxidation of
ture’, and multivariable control design has been applied methanol to formaldehyde and suffers from a secondary
in cases where the molar flowrate or reactor conversion reaction of formaldehyde to carbon monoxide and
are also consideredg. One example of controlling a water. The experimental apparatus, the mathematical
tubular reactor in the presence of catalyst deactivationI model describing the system and the steady state and
attempted to maintain the reactor outlet conversion con- dynamic experiments identifying the model parameters
stant through use of the pseudo-steady state reactor have been fully described in earlier papers16J7; thus we
equations and did not consider reactor dynamics other only present a brief summary here.
The yield of formaldehyde (the intermediate product selecting disturbances encountered in industrial
in the partial oxidation) is strongly affected by higher situations;
temperatures, which lead to total combustion products; l reproducibility and successful modelling of the
thus, control of the bed temperature profile is essential input response.
for yield maximization, The material and energy In many control problems the goal is to accomplish a
balances for the proce@ are summarized in the Appen- successful setpoint change with a minimum of offset and
dix. Because the fluid species and fluid temperature dyna- interactions. However, for this reactor the setpoint is
mics are more than 1000 times faster than the bed tem- chosen to maximize an objective such as reactor yield.
perature, they are considered to be in quasi-steady state. Thus, after reactor start-up, the setpoint does not change
In order to provide a high performance control scheme except as reactor parameters change. Here, the more
for the packed bed reactor, a comprehensive, generally important task for the control system is to reject distur-
applicable distributed parameter control strategy has bances and compensate for time-varying parameters.
been developed which is robust to modelling errors and In order to select the best set of control inputs, a
catalyst deactivation and controls the reactor over a wide preliminary set of data is necessary. Because the accuracy
range of operating conditions. The overall control with which the reactor model predicts responses to
system package consists of four sections: changes in input and disturbance variables affects the
l temperature and composition state estimation performance of the estimator and controller, open-loop
0 parameter estimation experimentation was conducted for a wide range of
0 control values of all possible manipulated variables. Generally,
0 optimization. any variable that causes an axial shift in the hotspot of
These run on a process control computer in real time. the reactor is more difficult to model because the reactor
This paper emphasizes the parameter estimation and response is substantially affected by axial variations in
control studies which utilize the results of the state esti- catalyst activity caused by deactivation. Also, feed flow-
mation. The state estimation algorithm has been pre- rate causes problems as a control variable because of
sented in an earlier paper’*, and on-line optimization is parameter variations and additional non-linearities in
described in a companion paperlg. the input-output model. Although several types of dis-
turbance cause model parameters to change somewhat, a
constant parameter model has been chosen for model-
Design and implementation of the control based control system design.
scheme Let us review each of the input variables to evaluate
their suitability as a manipulated variable. The reactor
In addition to choosing which strategies and algorithm temperature, and thus reactor yield, is highly dependent
to implement, control system structural decisions must on the jacket wall temperature16. This fact favours wall
be made for input-output variables, measurements and temperature as a control variable. The wall temperature
control loops. Also, the dynamics of the experimental is able to control the temperature in general throughout
control variables and disturbances should be assessed. the reactor, but the entire temperature profile increases
or decreases in response to the corresponding changes at
Selection of manipulated input variables the wall. The location of the hot spot is affected little by
Several possibilities for combinations of control inputs, changes in wall temperature. Thus disturbances which
outputs and disturbances can be observed in the previous change the hot spot location (e.g. flow rate, inlet temper-
literature on packed bed reactor control. For the reactor ature) cannot be completely rejected by manipulating
in this study, the choices for manipulated (i.e., control) wall temperature alone. However, the wall temperature
variables and disturbances are: generally provides adequate control of the two most
l jacket wall temperature crucial outputs, the maximum reactor temperature and
l inlet feed temperature the yield at the reactor exit.
l inlet methanol mole fraction The inlet feed temperature is unfavourable as a control
l feed flow rate. variable in terms of the first criterion because it causes
The decision concerning which of these variables to only small changes in yield. Thus, a very large change in
choose as controls, which to allow to be disturbances, the inlet temperature would be needed to reject a small
and which to hold constant is based on the following upset in the external wall cooling system or to compen-
‘considerations: sate for deactivating catalyst. Manipulation of the inlet
the magnitude of the effect which the control vari- temperature gives control over location of the hot spot.
able has upon the desired output; However, changes in the inlet temperature can give rise
creation of ‘interesting’ dynamics (i.e., wrong-way to ‘wrong-way’ behaviour. The dynamic and steady state
behaviour) which results in a challenging control response of the reactor to this variable is highly depen-
problem; dent on the activity profile in the catalyst bed.
experimental constraints, especially the dynamic The inlet methanol concentration is much like wall
response of the proposed control variables; temperature in terms of controllability of the reactor
formulating a realistic situation, using controls temperature. However, changes in the inlet concent-
commonly available in industrial reactors and ration have only a small effect on the yield and have a
large effect on the production rate. Thus inlet concent- benefit was gained by the use of more than three temper-
ration should remain as large as safety constraints allow. ature measurements.
Feed flow rate has a large effect both on the yield and In choosing the most appropriate output variables,
the maximum reactor temperature, and it shifts the hot one must consider the control system objectives. The
spot markedly. The large change in the hot spot location objective function can take a rather general form includ-
might have detrimental consequences for model-based ing constraints and can be a function of a combination of
control if only one or two temperature measurements are several facets of reactor operation depending on the par-
used. Control action by feed flow manipulation would ticular process:utility consumption, exit temperatures,
not recover the original steady state around which a coolant temperature, pressure drop, material properties
controller may have been linearized. In addition, the etc. A special case is an objective which is a function of
flow rate is not necessarily available to change as a mani- weighted values of the products of the reactor. In the
pulated variable, because the amount of flow through the present case, reactor yield is chosen as the objective,
reactor is often specified by upstream process units or formaldehyde is assigned a unit value, and methanol,
design constraints for an entire plant. carbon monoxide, water and air have zero value. Thus
The actual dynamic response of the manipulated vari- the reactor yield of formaldehyde is maximized.
ables is important since it defines the constraints on Although the objective gives the overall goal of the
experimental control studies. Very often step changes are control system, we usually choose other output variables
not possible in manipulated variables due to equipment to be regulated in order to best achieve the control
lags, and realistic changes must be represented by a ramp system objective.
Some possible output variables are:
or other function rather than a step change. For our
l a composite objective based on amount of pro-
experimental reactor the rates of response of the possible
ducts (flow and exit composition), with reactor
manipulated variables are as follows:
yield as an example;
l 5°C mini for wall temperature heating or cooling
l the conversion at an interior axial location (has
at 25&26O”C; slower heating and faster cooling at
greater sensitivity to the manipulated variable
higher temperatures and vice versa;
than at the exit);
l 5°C mini for feed temperature with a time delay
l the axial composition profile;
of about 20 s; shows some overshoot indicating
l the maximum reactor temperature (i.e. hot spot
system of high order;
temperature)
l almost immediate response for changes in the
l the axial position of the hot spot;
methanol feed rate;
l the axial centreline temperature profile;
l a first order response with a time constant of l&
l the entire two-dimensional temperature profile;
30 s for changes in the air flow rate. l the temperature at the thermocouple nearest to the
Among the manipulated variables discussed here, the original steady state hot spot;
wall temperature and the inlet temperature are the most l one temperature from any fixed location (i.e. 5-10
likely to be chosen for control of an industrial reactor. cm upstream from steady state hot spot);
They are easily measurable and normally unaffected by 0 two temperature measurements (one upstream,
upstream or downstream process units. Variations in the one downstream) and the difference between them.
inlet temperature could be caused by use of a constant If the process outputs are directly measured tempera-
energy input preheater with a fluctuating feed flow or by tures (as in the last three examples), then feedback
heating the feed by exchange with a time-varying hot control loops could be implemented in a straightforward
stream. If the primary reactant is metered in as a pure manner. However, most of the outputs listed above can-
stream, then it can serve as a control input. The feed flow not be measured directly for real-time control, and their
rate can normally be measured, but may be subject to values must be computed based on a process model and
frequent transients if the feed is coming directly from an measurements which are readily available. The state esti-
upstream unit with no intermediate storage. In this case mator developed earlieri is used to generate these out-
the feed flow is constrained by the plant production and puts.
is a disturbance rather than a manipulated variable. Thus State estimation enables efficient use of only a few
reactor wall temperature is the primary control input measurements and a model of the process to generate an
chosen. output variable containing more valuable information
than measurements or a process model alone. For the
present reactor, the entire temperature and composition
Selection of output variables profiles of the reactor are estimated. However, it is con-
Previous reactor control experiments reported in the venient to use single numbers in feedback control loops.
literature have commonly used several evenly spaced The maximum reactor temperature, or hotspot tempera-
thermocouple measurements or an estimator with axially ture, is known to correlate well with reactor yield and
weighted concentrations as the output variables. Some- also provide a measure of the safety margin in reactor
times the exit temperature was used as a single measure- operation because it is a good indicator of the severity of
ment for adiabatic reactors; however, this usually is not the conditions under which the reactor is operating.
useful for wall-cooled reactors. Generally, only a small Therefore, the maximum reactor temperature (T,,,,,),
Parameter estimation
Parameter estimation
I
at t*+tD: 1 at t*:
Process recursive parametric
I
least 1 sensitivity
reactor
squares
Input experiments]
lYEST (z*.t*)
I
or
I 4
reactor
model
ut*+tD)
I T(r.z.t*)
(simulation]
J-i
T
meas
State estimation
X = f(r,z,t)
available through state estimation, has been chosen as the setpoint. Because the temperatures are rapidly and
the output variable for the control loop. The axial accurately measured, and the temperature profile estima-
location of T,,,,, is a promising second output variable for tion algorithm operates better for errors in the kinetic
a multivariable control strategy. parameters, attaining the estimated maximum tempera-
An alternative to using the maximum temperature esti- ture setpoint ensures that the reactor is operating near
mate for the output variable of the control loop would be the desired conditions even if the composition estimates
to use the exit composition estimates directly. However, are in substantial error.
there are deficiencies in this strategy. The first difficulty is
in specification of an appropriate setpoint for a multi-
component mixture. Because the reaction products are Parameter estimation
not independent of each other, only certain response A principal difficulty in achieving good predictions of the
surfaces are achievable. On the other hand, the reactor composition profiles in the reactor is the fact that the
selectivity is not the same in all situations either, so catalyst deactivates non-uniformly over time. Therefore,
specification of one component does not ensure the a necessary task is estimating the catalyst activity profile
desired amounts of others. If the overall yield or other on-line. These catalyst activity parameter estimates are
objective to be maximized is the control output variable, based on: a) infrequent composition measurements with
then it has a steady state maximum from which there are time delays; and b) the reactor temperature profile taken
only negative deviations from the setpoint. The resulting directly from the state estimation algorithm. A block
control problem is extremely non-linear, and the value of diagram in Figure 2 indicates the structure and data flow
the output can be attained by two different values of the of the parameter estimation algorithm which utilizes
input except at the maximum. The problem of maximiz- recursive least squares with a variable forgetting factor.
ing objectives such as yield can be dealt with more effecti- One or more parameters are estimated from each indepe-
vely through on-line optimization coupled with cascade nently measured component i’nthe reaction mixture. The
control. The output of an outer loop yield optimization current estimate of the catalyst activity profile is updated
algorithm can be the setpoint for the inner control loop by comparing the model predictions with the compo-
on T,,,. An on-line optimizing loop built around the sition measurements. These parameter estimates are used
controller described here has been developed and is to modify the reaction rate parameters used by the state
demonstrated in companion experiments to the ones estimation and optimization algorithms.
reported here’9. The composition samples are taken one at a time at a
Another important consideration is control system selected axial location in the reactor, and the compo-
performance in the presence of errors in the reaction rate sition analysis is available after a time delay (TV). The
parameters. The temperature estimates are more reliable mole fractions of all reacting species are assumed to be at
than the composition estimates. If there were substantial a quasi-steady state with respect to the instantaneous
errors in the assumed catalyst activity, direct feedback.of temperature profile at the time the sample was taken
the composition estimates would cause severe errors in from the reactor. The sequence of events in parameter
the control action in attempting to force the estimates,to estimation is given below. At sample time, t*:
1. obtain a composition sample (at location I*) and A recursive least squares algorithm with variable for-
send it to measuring device; getting factor20 implemented with U-D factorization is
2. at the same time, use the estimated temperature used to estimate the parameters. These and similar esti-
profile and the quasi-steady state non-linear distri- mation algorithms are also discussed by Ydstie and Sar-
buted mass balance equations for each component gent21. After calculating the prediction and the error as
to solve for the pseudo-steady composition pro- shown above, the algorithm is:
file;
3. use interpolation to extract and save the estimated Gain: pk- I$k
When the analysis of the composition sample is com- Forgetting factor: hk = 1 - (1 - &?Kk) Ek/x (7)
plete:
If h < h,i” then set h = h,,
5. compute the measurement value;
Covariance: Pk = (1 - Kk+‘)Pk-,/hk (8)
6. compare the measurement with the estimate and
calculate the resulting error; When the algorithm is implemented, Pk is replaced by
7. use recursive estimation to update the parameters; (VDU), where U is upper triangular with l’s on the
8. pass new parameter values to the state estimation diagonal, and D is diagonal. The user specifies the
algorithm where they are used in estimating suc- following values:
ceeding temperature and composition profiles; LA, - the minimum forgetting factor;
9. .pass new parameter values to the optimization 1 = oZ,N,, - a measure of the information content,
algorithm. where 02, is the expected measurement noise vari-
ance, and N,, is the nominal asymptotic memory
The parameter estimation algorithm is implemented in length (small values give a sensitive estimator);
the following manner. The predicted value of the kth Di - the diagonal elements of the initial values for
measurement is obtained by solving the non-linear alge- Pk, i = 1, M where A4is the number of parameters
braic equations for the composition at the collocation estimated.
points and interpolating to generate values at the mea-
surement location. The solution is based upon all para- Reactor hot spot temperature control
meters at the measurement time (Q, including the esti-
The block diagram for regulation of the maximum reac-
mated parameters which were updated at time tk_ 1 + q,
tor temperature is shown in Figure 3. The jacket wall
and which were being used in the model at time tk. The
temperature setpoint was chosen as the manipulated
predicted output of composition at time tk is given by:
input variable in the outer cascade loop to regulate the
maximum estimated reactor temperature. The jacket
(1)
wall temperature is maintained by a heater controlled
where ak are fixed parameters and ok_, are the para- with a local PID time-proportional controller in the
meters estimated on-line. inner loop. The state estimation and parameter estima-
Linearizing the equations for the composition measur- tion sections work together to provide good estimates of
ement output with respect to each estimated parameter: the maximum temperature.
The wall temperature was chosen as the manipulated
Yk = flakptkpek-d +
ay
a (ok - ok-l)
variable for the following reasons:
(2)
l The monotonic response of the reactor tempera-
the parametric sensitivity vector ay/aO is found by finite ture to wall temperature - an increase in wall
differences through one at a time perturbation of the temperature increases the temperature everywhere
parameters about ok_, and solving Equation (1). Taking in the reactor, and a decrease does the reverse;
the output to be the measured value, and substituting j$, l The yield in the reactor is strongly affected by the
the linearized equation for the parameters yields: wall temperature, and a wide range of control over
the methanol conversion and carbon monoxide
yk - ik =
ay
s (ok - ok-,)
production is possible;
l the wall temperature is often available as the vari-
able used to control the reactor in industrially
Defining E as the prediction error, 6 as the change in relevant situations.
the parameter vector for this time step, and + as the
parametric sensitivity, the standard form for the para- The ‘well-behaved’ input-output relationship between
meter estimates becomes: T, and T,,, is helpful for control design. There is no
inverse response, and the gain is always positive regard-
Ek = d?@k (4) less of the operating conditions.
Process
reactor
(experiments)
L
=U Y
k or meas
fQ= --_L * Parameter
74 - reactor
:s
2 Twall estimation
model
5
(simulation)
1 I
/
T(r.z.t*)
meas
e(t*+tD)
Maximum
-
State
temperature _ Tmax r;
L temperature
X = f (r,z,t)
setpoint
One feature of the wall temperature input is that it is perature equal to the measured coolant temperature. If
not constrained by an absolute value, but by the rate of the feed temperature was kept near the wall temperature,
change of the input. Since a cascade control system was it had little effect on reactor performance due to the
used, the rate of change of the wall temperature setpoint deactivation of the catalyst at the entrance of the bed.
was constrained within the maximum possible rate of The first portion of the bed functioned partly as a heat
change in the wall temperature itself. A proportional- exchanger bringing the gas coming through the bed close
integral digital controller algorithm with constraint on to the wall temperature.
the rate of change of the control action was found to
work satisfactorily for manipulation of the wall tempera-
ture setpoint in controlling maximum reactor tempera- Experimental methods
ture. The controller equations are:
On-line experiments have been performed to test the
Ek = Yset,k - Yk (9) control scheme shown in Figure 3. The state estimation
algorithm was used throughout in order to provide reli-
Au = Kc (Ek_, - Ek + l/T, L!t(Ek_, + Ek)/2) (10) able estimates of the reactor temperature and compo-
sition profiles. The rationale for the design of the experi-
If IAuJ > Aumax,then ments was to provide a basic test of the control system
and then to test the performance under difficult and
&k =
w (A4 AumaxlKc
+ Ek-l(l - WCW)
complicating circumstances.
1 + At/(24 (11) An enhanced version of the data acquisition and local
The control action, where u is the wall temperature control program which had been used to operate the
setpoint constrained within a safe level for maximum reactor during open-loop steady state and dynamic
coolant temperature, is: experiments” was used in closed-loop experimentation.
Important features are the on-line gas chromatograph
uk = min(u max,uk-I + 4 (12)
data analysis and extensive communication between the
real-time data acquisition package and the estimation
The input-output relationship betwen T, and T,,,,,was and control algorithms. The cycle time for the gas chro-
observed to be of high order, and therefore the gain must matograph samples was 12.5 min. The amounts of
be kept reasonably low to avoid oscillations. However, formaldehyde, carbon monoxide and unreacted metha-
another factor limiting the aggressiveness of the con- nol were measured for each sample and determined as a
troller gains is the inverse response of T,,,,, caused by fraction of the methanol fed to the reactor. The time
some flow rate disturbances. interval for temperature measurement sets was 6 s. The
For the control experiments in this work, the feed experiments were run entirely automatically by the
temperature controller attempted to keep the feed tem- computer using a protocol specified in a data file.
CI where
$ 600-
2’ : inflection point
f = atan (a(1 - Z’)) - atan (ct(z - Z”))
Al atan (a(1 - Z’)) - atan ( - crZ”>
;m$g
0 35 70 0 35 70
0.081 I
A2 at exit
0.04
0.02 c 1
0 20 40 60 80 100 120 140 160
Time (min)
.8
.6
.4
.2
0 35 70
Final estimated profile
I I I 1 1 I I I
0 10 20 30 40 50 60 70
estimated profile. Estimation of the preexponential Figure 7 Parameter estimation experiment: estimated composition
profiles compared with data points at the time of parameter updates
factor for carbon monoxide production based on the
fractional carbon monoxide conversion was quite reli-
able. the composition estimates are based upon a reasonable
Figure 7 shows the estimated composition profiles in temperature profile. Note that the control scheme
comparison to the data at the times indicated. The provides a separation between the parameter and state
numbers in the legends refer to the operating conditions estimation with resulting improved stability of the esti-
given in Table 1. Composition profiles l-4 show metha- mates.
nol conversions and by-product carbon monoxide pro- In examining the results in Figure 8 at the time of 25
duction considerably greater than the data during the min, we see that the first parameter update at 20 min
period of high estimated activity. However, beginning allows significant improvement in the estimated tempera-
with the fifth profile, good agreement is attained between ture profile versus the profile at a time of 12 min. How-
data and estimates. Thus, the parameter estimates ever, more severe conditions at lower flow rates (33 min)
quickly converge to values giving accurate estimates of emphasize that the kinetics are still in error. After 50 min
the entire composition profile in the reactor. It is import- of operation, the temperature profiles are quite well
ant to notice that not only is the exit composition of the represented under a variety of steady state conditions.
reactor predicted properly, but also the compositions at Note that even the more severe conditions of a wall
other sampling locations. temperature of 270°C and a low flow rate are accurately
Figure 8 shows the corresponding estimated tempera- estimated at time = 137 min.
ture profiles in comparison to the data at the times indi- It is concluded that reliable estimates of the reactor
cated on the graphs. The profiles are at steady state at behaviour and a reasonable approximation of the cata-
each of the different operating conditions of the experi- lyst activity profile in the reactor can be obtained on-line
ment. The kinetic parameter estimates are erroneous after about five composition measurements. This perfor-
enough to cause ‘runaway’ in the model alone if the mance is good even in the case of a poor guess for the
temperature measurements had not been used to esti- preexponential factors of the kinetic rate constants.
mate the temperature profile. Thus, the state estimation Operating conditions can be changed as needed, and
algorithm generates acceptable temperature estimates there is no requirement that the system be at steady state.
even in the face of significant parameter bias. This The estimated maximum reactor temperature is needed
robustness is important because it gives confidence that for control and optimization, and the maximum temper-
Number Time (min) T, = T;(‘C) u, CH,OH Flow (g SK’) Sample time (min) Sample location (cm) Comments
Initial conditions for reactor and estimation: T, = T, = 260, Y, = 0.05, flow = 1.O; temperature uniform at 26o’C; methanol flow begins at time = 0
Experiment Time (mm) r, CH,OH Flow (g s- ‘) TmnisetW) Sample time Sample location Comments
a
,/--
/
.___
..__. e--;__w--_
265
:,:
320 260
260
240
I/
I- ----
-------
5
Tmaximum
Estimate
Setpoint
10
255
250
0
-
-------
5
Twall
Setpoint
10 60
----
-------
65
Estimate
Setpoint
70
255-
60
I
65
------- Setpoint
I
70 75
Time (min) Time (min) Time (min) Time [min)
345 340, 262
b If
340 - ..___.... ~~..--~~~~
- Twall
------- Setpoint
- Twall
.------- Setpoint
I I
320 L
15 20 25 30 ‘“:- 80 85 90 75 80 85
316
--- Estimat e
It
314
325 --__-_
312 - Twall
------- Setpoint
320 1 I
30 35 40 45
335 A
330
325
320 I I I
45 50 55 60
scheme is that it accurately represent the yield perfor- monoxide are well represented as shown by the compari-
mance of the reactor. Both methanol conversion to son between the composition estimates and experimental
formaldehyde and the formaldehyde oxidation to carbon data in Figure 12.
320
Time = 27 min
Tset = 340
Yi = 0.05
F = 0.7
260
0 35
Time = 57 min
Tset = 325
320 320
280
260 260
0 35 0 35 70
Figure 11 Control experiment Cl: comparison between measured and estimated temperature profiles
Carbon monoxide value within 12 min. The true wall temperature was
o”“---y-Ei changed rapidly until it coincided with the setpoint out-
put of the controller, and then it quickly was adjusted to
the original steady state value. For the sampling time
interval of the experiment, no complicating factors were
!+I-+
observed due to using the modelled rather than the mea-
sured wall temperature response. The reactor tempera-
ture estimates quickly recovered from the original erro-
neous boundary conditions, and the T,,,,, estimates were
within 2°C of the data after rapid initial convergence.
0
Experiment C3 tests the estimation and control system
in the presence of errors in the parameters for heat
transfer. The radial Peclet number was chosen 25%
i 100 larger than indicated by off-line analysis of previous
data, and the interphase heat transfer coefficient was
chosen 40% lower than the base case model. These inac-
Figure 12 Control experiment Cl: comparison between composition curate parameters cause higher estimated reactor tem-
data and estimates at exit
peratures as well as a more sensitive model, and the state
estimator must compensate accordingly. It is important
The second control experiment, C2, demonstrates per- for the state estimator to be able to compensate adequa-
formance when the boundary conditions (wall tempera- tely for a somewhat mismatched model in order that the
ture) are not accurately known by the estimation and controller set the desired reactor temperatures, especially
control program. There is no direct measurement of the the T,,,. The procedure of experiment C3 was the same
wall temperature transmitted to the process control as C I a: the start-up of the reactor of a maximum temper-
computer. In this case, the wall temperature setpoint was ature of 325°C.
sent to the local controller, and the response of the local The results of this experiment were similar to Cla
loop is represented within the estimation and control except with more overshoot in the estimated T,,,,,and the
program as a simple first-order system. In addition, the wall temperature (cf Figure 14). This result could be
experiment verifies the accuracy of the closed-loop expected, since less heat transfer gives a more sensitive
control simulation which uses a first-order dynamic model having a higher steady state gain but a similar
model to represent the experimental response of the local time constant. Due to the model errors causing incorrect
wall temperature control loop. The results of experiment temperature estimates, the reactor was operated with 4-
C2 are shown in Figure 13, where an initial error of 10°C 5°C lower wall temperature than the previous experi-
in the wall temperature estimate caused a 4°C upset in the ments without deliberate model mismatch. This perfor-
maximum temperature, which returned to its original mance is a great improvement over the 1520°C errors
332 7
a
- Twall
270
- -- Estimate
----- Setpoint
268
266
264
260
- Tmax
258
--- Estimate
------- Setpoint
3221 I I I I I 2561 I I I I
0 2.5 5.0 7.5 10.0 12.5 0 2.5 5.0 7.5 10.0 1;
- Tmax - Twall
--- Estimate ----- Setpoint
_---_-- Setpoint
5 10 15
which occur in the model in the absence of state estima- hold the maximum temperature at 325°C during step
tion. changes in the flow rate (C4b,c,d). Then, in part C4e, an
The fourth section of control experiments (C4 a-f) operating change is made in the feed mole fraction meth-
demonstrates control of the maximum reactor tempera- anol from 0.05 to 0.04, and the setpoint for maximum
ture in the presence of significant errors in the kinetic temperature is reduced from 325 to 310°C. This type of
parameters. These experiments also demonstrate the esti- change is typical of what occurs if there is known change
mation of the kinetic parameters in a closed-loop control in the feed, and experience or off-line modelling indicates
setting, and the improvements in the composition state a corresponding change to be made in the setpoint. The
estimates resulting from adjustments in the kinetic para- final part of the experiment (C4f), is a change back to the
meters. The maximum temperature setpoints and inputs setpoint of Cla to demonstrate the difference in perfor-
for this experiment are shown in Figure 1.5. Experiment mance as the estimates of the kinetic parameters
C4a demonstrates start-up of the reactor to a maximum improve.
temperature of 325°C. The controller then attempts to First, this sequence was tested through closed-loop
340
rime = 12 min -‘-I Time = 37 min
..... .. . . . .. .
320 ‘. .
I’ ’ \\
/ \\
300 / \\
z I’
e 280
r
..
?
3
?\
m
!g 260 70 0 35
0 35
E
‘, 340 340
Time = 50 min Time = 62 min Time = 75 min
.z
z r~ . . .
I
I
A
: 320 ‘\ 320
t
s
300
/
280 ., /
/
+ 70
Figure 16 Control simulation C4: comparison between measured and estimated temperature profiles
\ zfyyy-g
100 ___--_-----_-------- ------- - _--_
“Eb-1
0 10 20 30 40 50 60 70 80
Time (min]
Methanol Carbon monoxide’
o*750w 80 o~02%i%hi-
0 0
Figure 17 Control simulation C4: estimates of catalyst activity and comparison between composition data and estimates at exit
The methodology described in this paper is useful for buted systems, for preventing reactor runaway, and for
controlling packed bed reactors and other similar distri- Qptimizing reactor performance measures such as pro-
- Tmaximum
---- Estimate
------- Setpoint
G
0
; 315-
- Twall - Twall
245[ , ---I_--- :etpo:nt ------- Setpoint
_---- Setpoint
300 I I , I
0 2.5 5.0 7.5 10.0 12. .5 0 2.5 5.0 7.5 10.01 50 52.5 55 57.5 60 62 i 50 52.5 55 57.5 60 62.5
Time (min) Time (min) Time (min) Time (min)
330. 266
330
- Twall
G .------- Setpoint
o 325 - Twall
f ________
Setpoir E :
1
zb 320 z :’
I 320.. :
x :
2 I
a 315
I- E
- Tmaximum k- - Tmaximum
315
3 1IJ _--_ Estimate
258 -
_____-_ Setpoint
---- Estimate
------- Setpoint
. - Twall
------- Setpoint
320 I I 2541
25 27.5 30 32.5 35 37.5 25 27.5 30 32.5 35 3 7.5
330
G
0 325
klf
,
z 320
%
E
aI 315
c-
_ Tmaximum
310 254 - - Twall
------- Setpoint
305 252 I I I , .
37.5 40 42.5 45 47.5 50 37.5 40 42.5 45 47.5 50
duct yield. The estimation and control system effectively oxidation of methanol for formaldehyde production, is
overcame incorrect initial conditions, erroneous heat itself an important industrial packed bed reactor process
transfer parameters, erroneous kinetic parameters and operated around the world and is representative of wall-
unmeasured disturbances. cooled packed bed processes. Thus the approach demon-
The example system chosen for the study, the partial strated in this work should be applicable to other packed
340
Time = 25 min 1
E
3
L”
$ 0 35 70
q
E
m
; 340 340
.E Time = 50 min Time = 75 min
t; 320 320 ,
u” f .
300 300
280 260 ’
.
.’
/*
7 2600%
2600 35 70 35
Figure 19 Control experiment C4: comparison between measured and estimated temperature profiles
1 I4=0
r1 integral time for controller
4 parametric sensitivity 6Bi,p %
cl0 = Pe,,
1Pe,,Pe,,(Bi,, + 3)
Subscripts
co carbon monoxide Dimensionless parameters:
M methanol
set setpoint
Tr = -tuo cj = 5 y, z A T, = g
L ‘MO y MO 0
Superscripts
estimated quantity I
+ at reactor exit Tr = 7 z=Z ,=r a=L
0 at reactor entrance 0 L rt rt
Appendix
B, = ( - AH~)rloKd?‘o, TOIL
GC,Jo
couo Go