Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Abstract
This paper describes the application of self-optimizing control to a large-scale process, the HDA plant. The idea is to select controlled
variables which when kept constant lead to minimum economic loss. First, the optimal active constraints need to be controlled. Next,
controlled variables need to be found for the remaining unconstrained degrees of freedom. In order to avoid the combinatorial problem
related to the selection of outputs/measurements for such large plants, a local (linear) analysis based on singular value decomposition
(SVD) is used for pre-screening. This is followed by a more detailed analysis using the nonlinear model. Note that a steady-state model,
in this case one built in Aspen PlusTM , is sufcient for selecting controlled variables. A dynamic model is required to design and test the
complete control system which include regulatory control. This is considered in the part II of the series.
r 2006 Elsevier Ltd. All rights reserved.
Keywords: HDA process; Self-optimizing control; Selection of controlled variable; Aspen PlusTM
1. Introduction
This paper deals with the selection of controlled
variables for the HDA process. One objective is to avoid
the combinatorial control structure issue for such largescale processes by using local methods based on the
singular value decomposition (SVD) of the linearized
model of the process.
The selection of controlled variables is based on steadystate economics and use the ideas of self-optimizing control
to nd the best set(s). Self-optimizing control is when an
acceptable (economic) loss can be achieved using constant
set points for the controlled variables, without the need to
reoptimize when disturbances occur (Skogestad, 2000). The
constant set point policy is simple but will not be optimal
(and thus have a positive loss) as a result of the following
two factors: (1) disturbances, i.e., changes in (independent)
variables and parameters that cause the optimal set
points to change, and (2) implementation errors, i.e.,
differences between the setpoints and the actual values of
Corresponding author. Fax: +47 7359 4080.
ARTICLE IN PRESS
A.C.B. de Araujo et al. / Control Engineering Practice 15 (2007) 12221237
1223
ARTICLE IN PRESS
1224
(1)
u0
g2 x; u0 ; dp0.
(2)
ARTICLE IN PRESS
A.C.B. de Araujo et al. / Control Engineering Practice 15 (2007) 12221237
1225
Table 1
Steady-state controlled variables selected by various authors
Y20b
Y71
Y48
Y49
Y62
Y68
Y72
Y28
Y5
Y19
Y64
Y70
Y73
Y69
Y27
Y16
Y26
Y46
Y74
Y53
Y54
Y55
Y56
Y75
Y76
Y77
Y78
Y58
Y57
(1)
(2)
(3)
(4)
(5)
(6)
(7)
8
x
x
x
8
x
9
x
13
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
Note: (1) Stephanopoulos (1984); (2) Brognaux (1992), Wolff (1994), Cao et al., Herrmann et al. (2003); (3) Ng and Stephanopoulos (1996); (4) Ponton and
Laing (1993); (5) Luyben et al. (1998), Luyben (2002); (6) Konda et al. (2005); (7) This work.
a
The total number of steady-state degrees of freedom is 13, so there are additional controlled variables, or xed inputs, which are not clearly specied by
some authors.
b
Y-variables refer to candidates in Table 4.
c
Active constraints found in this work.
(3)
1
J 1=2
uu G c
(4)
uopt
copt , where G is the
with z
steady-state gain matrix from the unconstrained degrees of
freedom u to the controlled variables c (yet to be selected)
and J uu the Hessian of the cost function with respect to the
u. Truly optimal operation corresponds to L 0, but in
general L40. A small value of the loss function L is desired
as it implies that the plant is operating close to its
optimum. The main issue here is not to nd the optimal
set points, but rather to nd the right variables to keep
constant.
ARTICLE IN PRESS
A.C.B. de Araujo et al. / Control Engineering Practice 15 (2007) 12221237
1226
1
1
,
2
2 sS1 GJ 1=2
uu
(5)
(6)
a
1
,
2 sS1 G2
(7)
where a sJ uu .
Thus, to minimize the loss, L, sS 1 GJ 1=2
uu should be
maximized or alternatively maximize sS 1 G; the latter is the
original minimum singular value rule of Skogestad (2000).
Originally, a MatLabTM model was used to obtain the
optimal variation Dcopt d, the steady-state gain matrix G
and the Hessian J uu , but in the present version Aspen
PlusTM is used instead (see the Appendix for details). The
use of a commercial owsheet simulator like Aspen PlusTM
demonstrates the practical usefulness of the approach.
4. HDA process description
In the HDA process, fresh toluene (pure) and hydrogen
(97% hydrogen and 3% methane) are mixed with recycled
toluene and hydrogen (Fig. 1). This reactant mixture is rst
preheated in a FEHE using the reactor efuent stream and
then to the reaction temperature in a furnace before being
fed to an adiabatic plug-ow reactor.
A main reaction and a side reaction take place in the
reactor as follows:
Toluene H2 ! Benzene Methane,
2 Benzene Diphenyl H2 .
8
9
Table 2
Typical number of steady-state degrees of freedom for some process units
Process unit
DOF
1 (feedrate)
n 1 split fractions (n is the number of exit streams)
0
1 (work)
0a
1 (holdup)
0a
1 (duty or net area)
0a + number of side streams
Add 1 degree of freedom for each extra pressure that is set (need an extra valve, compressor, or pump), e.g., in ash tank, gas phase reactor, or column.
ARTICLE IN PRESS
A.C.B. de Araujo et al. / Control Engineering Practice 15 (2007) 12221237
1227
5. Results
This section describes the self-optimizing control procedure applied to the HDA process model in Aspen PlusTM
starting with the degree of freedom analysis.
GREC
V5
31
V1
T2
FFH2
RX
FUR
32
HDA Process
RIN
QUENCHER
ROUT
TOTTOL
19
HX
29
V6
V3
22
FFTOL
GAS
T1
28
SEP
COMP
12
18
13
LIQ
V10
T3
P1
COND
23
15
T5
27
TREC
7
V11
C1
V13
P4
V4
D1
20
C3
14
D3
P2
C2
D2
F1
V15
26
21
P5
V14
17
B3
P3
V12
B2
30
B1
10
1228
Table 3
Stream table for the nominally optimal operating point for the HDA process
Stream
12
13
14
15
17
18
19
20
21
22
23
26
27
28
29
30
433.37
13.403
0
0
0
1.0841
14.438
0.0016
0
0
1809.1
2910.1
45.023
316.8
0.0101
0
0
276.36
15.946
9.43
1519.4
3219
498.76
26.948
15.328
1519.4
3219
498.76
26.948
15.328
0
0
276.33
0.0831
0
1.7618
23.464
449.13
25.914
15.325
0
0
0.0329
15.862
9.43
1517.7
3195.5
49.631
1.0342
0.0031
141.9
298.78
4.6405
0.0967
0.0003
0
0
276.33
0.0831
0
0
0
0.0329
15.862
9.43
0
0
0
300
0
0
0
0.0329
15.859
0.0073
0
0
0
0.0038
9.4227
1.0841
14.438
276.36
15.946
9.43
0.6777
9.0258
172.77
9.9684
5.8951
0.6777
9.0258
172.77
9.9684
5.8951
0
0
0
0.0038
9.4227
0.97
0.03
0
0
0
446.77
100.11
555
1
0.3551
0.0698
0.9301
1E 04
0
0
15.524
93.394
50
1
0.4846
0.3561
0.5727
0.0089
0.0623
2E 06
5081.1
120.25
530
0.932
88.217
0
0
0.9159
0.0528
0.0313
301.74
235.26
31.714
0.4557
9.8177
0.2878
0.6097
0.0945
0.0051
0.0029
5279.5
357.67
477.4
1
68.617
0.2878
0.6097
0.0945
0.0051
0.0029
5279.5
95
477
0.9023
90.323
0
0
0.9997
0.0003
0
276.41
223.92
80
0
7.2579
0.0034
0.0455
0.8711
0.0503
0.0297
515.6
95.359
530
0
10.044
0
0
0.0013
0.6263
0.3724
25.325
333
84
0
0.8809
0.3186
0.6708
0.0104
0.0002
6E 07
4763.9
124.89
555
1
99.2
0.3186
0.6708
0.0104
0.0002
6E 07
445.42
124.89
555
1
9.2752
0
0
0.9997
0.0003
0
276.41
224.02
50
0
7.2579
0
0
0.0013
0.6263
0.3724
25.325
325.6
30.75
0.0337
0.8809
0
0
0
1
0
300
100.27
555
0
1.8353
0
0
0.0021
0.9975
0.0005
15.899
289.14
555
0
0.2276
0
0
0
0.0004
0.9996
9.4264
566.56
82
0
0.7991
0.0034
0.0455
0.8711
0.0503
0.0297
317.26
95.359
530
0
6.1802
0.0034
0.0455
0.8711
0.0503
0.0297
198.33
95.359
530
0
3.8635
0.0034
0.0455
0.8711
0.0503
0.0297
198.33
95.578
487.4
0
3.8635
0
0
0
0.0004
0.9996
9.4264
565.56
32
0.005
0.7991
Stream
Mole flow [lb mol/h]
Hydrogen
Methane
Benzene
Toluene
Diphenyl
Mole fraction
Hydrogen
Methane
Benzene
Toluene
Diphenyl
Total ow [lb mol/h]
Temperature [1F]
Pressure [psi]
Vapor fraction
Enthalpy [M Btu/h]
31
32
B1
B2
B3
D1
D2
D3
F1
FFH2
FFTOL GAS
GREC
LIQ
PURGE RIN
ROUT
TOTTOL TREC
1519.4
3219
498.76
26.948
15.328
1809.1
2910.1
45.023
316.8
0.0101
0
0.0003
276.36
15.946
9.43
0
0
0.0329
15.862
9.43
0
0
0
0.0038
9.4227
1.0841
14.438
0.0016
0
0
0
0
276.33
0.0831
0
0
0
0.0329
15.859
0.0073
1.0841
14.438
276.36
15.946
9.43
433.37
13.403
0
0
0
0
0
0
300
0
1517.7
3195.5
49.631
1.0342
0.0031
1375.8
2896.7
44.99
0.9375
0.0028
1.7618
23.464
449.13
25.914
15.325
141.9
298.78
4.6405
0.0967
0.0003
1809.1
2910.1
45.023
316.8
0.0101
1518.8
3210
325.99
16.98
9.4331
0
0
0.0329
315.86
0.0073
0
0
0.0329
15.859
0.0073
0.2878
0.6097
0.0945
0.0051
0.0029
5279.5
1150
487.4
1
7.2524
0.3561
0.5727
0.0089
0.0623
2E 06
5081.1
1004.8
510
1
26.854
0
1E 06
0.9159
0.0528
0.0313
301.74
371.4
154
0
9.8177
0
0
0.0013
0.6263
0.3724
25.325
332.65
34
0
0.8801
0
0
0
0.0004
0.9996
9.4264
565.54
32
0
0.7981
0.0698
0.9301
1E 04
0
0
15.524
83.814
150
1
0.4846
0
0
0.9997
0.0003
0
276.41
223.52
30
0
7.2521
0
0
0.0021
0.9975
0.0005
15.899
283.61
30
0
0.2222
0.0034
0.0455
0.8711
0.0503
0.0297
317.26
97.982
160
0.0293
6.1802
0.97
0.03
0
0
0
446.77
100
605
1
0.3551
0
0
0
1
0
300
100
605
0
1.8353
0.3186
0.6708
0.0104
0.0002
6E 07
4763.9
94.979
476
1
100.36
0.3186
0.6708
0.0104
0.0002
6E 07
4318.5
124.89
555
1
89.925
0.0034
0.0455
0.8711
0.0503
0.0297
515.6
94.979
476
0
10.033
0.3186
0.6708
0.0104
0.0002
6E 07
445.42
123.85
505
1
9.2752
0.3561
0.5727
0.0089
0.0623
2E 06
5081.1
1201.2
500
1
11.118
0.2989
0.6317
0.0642
0.0033
0.0019
5081.1
1277.2
496
1
11.118
0
0
0.0001
0.9999
2E 05
315.9
110.72
555
0
2.0629
0
0
0.0021
0.9975
0.0005
15.899
288.76
675
0
0.2276
ARTICLE IN PRESS
ARTICLE IN PRESS
A.C.B. de Araujo et al. / Control Engineering Practice 15 (2007) 12221237
Table 4
Selected candidate controlled variables for the HDA process (excluding
levels)
Y1
Y2
Y3
Y4
Y5
Y6
Y7
Y8
Y9
Y10
Y11
Y12
Y13
Y14
Y15
Y16
Y17
Y18
Y19
Y20
Y21
Y22
Y23
Y24
Y25
Y26
Y27
Y28
Y29
Y30
Y31
Y32
Y33
Y34
Y35
Y36
Y37
Y38
Y39
Y40
Y41
Y42
Y43
Y44
Y45
Y46
Y47
Y48
Y49
Y50
Y51
Y52
Y53
Y54
Y55
Y56
Y57
Y58
Y59
Y60
1229
Table 4 (continued )
Y61
Y62
Y63
Y64
Y65
Y66
Y67
Y68
Y69
Y70
Y71
Y72
Y73
Y74
Y75
Y76
Y77
Y78
Table 5
Steady-state degrees of freedom analysis based on Table 2
Process unit
DOF
212
212
111
200
100
212
326
13
(11)
X5.
(12)
(13)
(14)
ARTICLE IN PRESS
A.C.B. de Araujo et al. / Control Engineering Practice 15 (2007) 12221237
1230
Table 6
List of manipulable variables
U1
U2
U3
U4
U5
U6
U7
U8
U9
U10
U11
U12
U13
U14
U15
U16
U17
U18
U19
U20
Table 7
Economic data for the HDA process based on Douglas (1988)
Manipulated variable
Steady state
Steady state
Steady state
Steady state
Steady state
Steady state
Steady state
Dynamic only
Steady state
Dynamic only
Steady state
Dynamic only
Steady state
Dynamic only
Steady state
Dynamic only
Steady state
Dynamic only
Steady state
Dynamic only
pben
ptol
pgas
pfuel
pcw
ppow
pstm
9:04$=lb mol
6:04$=lb mol
1:32$=lb mol
4:00 106 $=Btu
2:34 108 $=Btu
0:042$=Bhp
2:50 106 $=Btu
(level control)
(level control)
(level control)
(level control)
(level control)
(level control)
3.
4.
5.
6.
(level control)
(15)
5.4. Step 4. Optimization
(16)
(17)
(18)
(19)
ARTICLE IN PRESS
A.C.B. de Araujo et al. / Control Engineering Practice 15 (2007) 12221237
1231
Nominal
Disturbance
300
300
0.03
5.0
500
1150
0.9997
1 104
1 106
1:3 103
0:5 103
0:4 103
285
315
0.08
5.5
520
1170
0.9960
3 104
5 106
2 103
1 103
1 103
Table 8
Disturbances to the process
D1
D2
D3
D4
D5
D6
D7
D8
D9
D10
D11
D12
ARTICLE IN PRESS
A.C.B. de Araujo et al. / Control Engineering Practice 15 (2007) 12221237
1232
x 106
3.7623
3.7622
3.7622
xB,methane = 110-6
Qreboiler [Btu/h]
3.7622
3.7621
3.762
3.762
3.762
3.7619
3.7618
3.7618
10-7
10-6
10-5
xD,benzene
10-4
10-3
10-3
10-2
x 107
10
9
8
Qreboiler [Btu/h]
xB,benzene = 1.310-3
6
5
4
3
2
1
0
10-6
10-5
10-4
xB,benzene
x 106
10
9
8
Qreboiler [Btu/h]
xD,diphenyl = 510-4
6
5
4
3
2
1
0
10-6
10-5
10-4
10-3
xB,toluene
Fig. 4. Typical relations between reboiler duty and product purity: (a) stabilizer distillate; (b) benzene column bottoms; (c) toluene column bottoms.
ARTICLE IN PRESS
A.C.B. de Araujo et al. / Control Engineering Practice 15 (2007) 12221237
Table 9
Specications for distillation columns
Column/specication
Value
Comment
Stabilizer
Y53 xD;benzene
Y54 xB;methane
1 104
1 106
(A)
(B)
Benzene column
Y55 xD;benzene
Y56 xB;benzene
0:9997
1:3 103
Active constraint
(A)
Toluene column
Y57 xD;diphenyl
Y58 xB;toluene
0:5 103
0:4 103
(C)
(A)
(A) Determined by trade-off between energy usage and recovery (Fig. 4).
(B) xB;methane should be small to avoid methane impurity in distillate of
benzene column.
(C) Diphenyl should not be recycled because it may reduce the available
production rate if there is bottleneck in the plant.
1233
4.95
4.90
Profit [M$/year]
4.85
4.80
4.75
4.70
4.65
4.60
4.55
4.50
Fig. 5. Effect of disturbances (see Table 8) on optimal operation. Percentages in parentheses are changes with respect to the nominal optimum.
ARTICLE IN PRESS
A.C.B. de Araujo et al. / Control Engineering Practice 15 (2007) 12221237
1234
Table 10
Effect of disturbances on optimal values for selected variables
Variable
Unit
Nominal
D1
D2
D3
D4
D5
D6
D7
Prot
Y4
Y15
Y16a
Y19a
Y20a
Y21
Y26
Y28
Y29
Y45
Y46
Y49
Y53b
Y54b
Y55b
Y56b
Y57b
Y58b
Y62a
Y68
Y70a
k$/year
F
F
F
F
lb mol/h
lb mol/h
lb mol/h
4693.4
1201.15
1277.21
1150
95
300
446.59
445.27
0.3558
0.5729
0.8721
0.0491
0.6710
1 104
1 106
0.9997
1:3 103
5 104
4 104
500
454.39
5.0
4552.7
1198.20
1273.64
1150
95
285
431.29
429.78
0.3548
0.5742
0.8671
0.0544
0.6717
1 104
1 106
0.9997
1:3 103
5 104
4 104
500
443.20
5.0
4903.2
1202.89
1279.25
1150
95
315
470.33
468.91
0.3577
0.5707
0.8703
0.0511
0.6691
1 104
1 106
0.9997
1:3 103
5 104
4 104
500
474.93
5.0
4649.0
1204.66
1277.71
1150
95
300
476.29
474.95
0.3454
0.5854
0.8667
0.5419
0.6803
1 104
1 106
0.9997
1:3 103
5 104
4 104
500
473.22
5.0
4585.6
1206.66
1279.65
1150
95
300
460.03
458.44
0.3703
0.5622
0.8792
0.4534
0.6534
1 104
1 106
0.9997
1:3 103
5 104
4 104
500
485.53
5.5
4664.7
1196.44
1272.25
1150
95
300
446.75
445.27
0.3558
0.5730
0.8683
0.5322
0.6708
1 104
1 106
0.9997
1:3 103
5 104
4 104
520
564.09
5.0
4722.5
1201.88
1276.89
1170
95
300
444.73
443.23
0.3526
0.5767
0.8692
0.5205
0.6737
1 104
1 106
0.9997
1:3 103
5 104
4 104
500
460.82
5.0
4705.5
1199.33
1274.99
1150
95
300
445.46
443.90
0.3560
0.5727
0.8662
0.5549
0.6705
1 104
1 106
0.996
1:3 103
5 104
4 104
500
455.41
5.0
psi
hp
Active constraints.
Distillation specication.
Table 11
Candidate controlled variables with small losses in local analysis
Variable
Name
Nominal optimal
Optimal variation
Implementation error
Total span
Y4
Y26
Y29
Y30
Y35
Y36
Y37
Y40
Y45
Y46
Y47
Y49
Y50
Y68
Y69
1201.15
445.27
0.5729
0.0091
0.0996
0.0031
0.0033
0.0107
0.8721
0.0491
0.0318
0.6710
0.0107
454.39
0.9124
5.52
29.73
0.0125
0.000068
0.0059
0.0007
0.0003
0.000081
0.0071
0.0071
0.0023
0.0175
0.000081
109.69
0.0076
60.06
22.26
0.0001
0.0001
0.0001
0.0001
0.0001
0.0001
0.0001
0.0001
0.0001
0.0001
0.0001
4.54
0.01
65.57
52
0.0126
0.000168
0.006
0.0008
0.0004
0.000181
0.0072
0.0072
0.0024
0.0176
0.000181
114.23
0.0176
ARTICLE IN PRESS
A.C.B. de Araujo et al. / Control Engineering Practice 15 (2007) 12221237
1235
Table 12
Local analysis: minimum singular values for candidate sets of unconstrained controlled variables
Set
Variable 1
Variable 2
1=2
Fulla
I
II
III
IV
V
VI
VII
VIII
IX
X
XI
XII
XIII
XIV
XV
XVI
Y29
Y29
Y29
Y29
Y29
Y29
Y29
Y29
Y29
Y29
Y4
Y4
Y26
Y4
Y26
Y49
Y36
Y69
Y45
Y46
Y40
Y50
Y35
Y47
Y30
Y37
Y26
Y49
Y49
Y68
Y68
Y68
6.2523
2.2942
2.2523
2.2133
2.2102
2.2072
2.1981
1.8452
1.8344
1.7855
1.7149
1.2439
0.2008
1.3352
0.1198
1.2196
0.0198
6.3436
2.3331
2.2761
2.2545
2.2398
2.2201
2.2199
1.8247
1.8044
1.7851
1.6825
1.2815
0.1957
1.2902
0.1201
1.2785
0.0201
Table 13
Loss in k$/year caused by disturbances and implementation errors for the alternative sets of controlled variables from Table 12
Set
D1
D2
D3
D4
D5
D6
D7
ny1 a
ny2
Average
I
II
III
IV
V
VI
VII
VIII
IX
X
XI
XIIb
XIII
XIV
XV
XVI
70.40
86.16
100.01
118.45
136.60
143.54
149.94
140.83
150.37
151.61
163.29
188.09
162.78
193.80
179.48
233.26
5.37
10.91
13.22
16.04
16.92
19.70
22.01
23.40
25.25
31.07
43.10
55.86
37.49
61.99
43.24
188.87
14.41
25.78
35.40
38.22
48.46
48.47
58.42
59.81
67.70
70.11
97.70
125.35
88.99
131.70
89.21
259.70
4.57
18.98
26.66
39.52
53.16
58.17
67.39
85.09
96.31
99.91
133.87
169.45
144.73
157.08
183.32
364.56
12.85
27.11
55.52
60.30
69.07
79.12
79.27
81.44
83.30
88.29
104.15
128.55
128.55
137.96
155.35
186.68
12.57
13.31
13.60
37.98
41.48
51.23
64.68
76.60
85.55
106.15
127.00
151.18
124.42
154.38
122.78
171.82
9.66
17.77
21.82
43.17
78.59
106.07
112.07
118.25
136.07
141.18
150.84
178.46
148.47
188.23
159.47
224.66
5.33
5.33
5.33
5.33
5.33
5.33
5.33
5.33
5.33
5.33
243.97
243.97
176.86
243.97
176.86
25.46
3.37
33.58
10.92
4.57
16.17
12.02
12.05
12.03
3.40
4.19
176.86
25.46
25.46
302.04
302.04
302.04
15.39
26.55
31.39
40.40
51.75
58.18
63.46
66.97
72.59
77.54
137.87
140.71
115.31
174.57
156.86
217.45
ny1 and ny2 are the implementation errors associated with each variable in the set.
This is similar to the structure of Luyben (2002), but with control of active constraints.
7. Conclusions
This paper has discussed the selection of controlled
variables for the HDA process using the self-optimizing
control procedure. The large number of variable combinations makes it a challenging problem, and a local (linear)
analysis based on the SVD of the linearized model of the
plant was used to select good candidate sets for the
unconstrained controlled variables. Specically, 16 candi-
ARTICLE IN PRESS
A.C.B. de Araujo et al. / Control Engineering Practice 15 (2007) 12221237
1236
(20)
h!0
Ju E ii h E jj h Ju E ii h Ju E jj h Ju
,
hhT ij
21
where E ij is the zero matrix except for the ij-element which
is 1. The several functions of J in the denominator of (21)
are evaluated in a Calculator block in Aspen PlusTM and
taken to MatLabTM for the numerical calculation of
H ij q2 Ju=qu2j ji .
A.2. Optimal variation for the candidate variables
The optimal variation for the candidate variables
(spanci ) is used to scale the linear matrix G obtained by
linearizing the nonlinear model of the process. In this work,
it was used direct calculations from the nonlinear model of
the HDA process in Aspen PlusTM .
For each candidate controlled variable ci , it is obtained its
maximum optimal variation Dci;opt d due to variation in
disturbances. From the nonlinear model, the optimal parameters (inputs and outputs) for various conditions (disturbances and operating points) are computed. This yields a
lookup table of optimal parameter values as a function of
the operating conditions. From this, one can identify
Dci;opt d maxjcji;opt cnom
i;opt j,
j2D
(22)
(23)
References
Brognaux, C. (1992). A case study in operability analysis: The HDA plant.
Master thesis, University of London, London, England.
Buckley, P. S. (1964). Techniques of process control. New York, USA:
Wiley.
Cao, Y., & Biss, D. (1996). New screening techniques for choosing
manipulated variables. in Preprints IFAC 96, 13th world congress of
IFAC, volume M, San Francisco, CA.
Cao, Y., & Rossiter, D. (1997). An input pre-screening technique for
control structure selection. Computers and Chemical Engineering, 21(6),
563569.
Cao, Y., & Rossiter, D. (1998). Input selection and localized disturbance
rejection. Journal of Process Control, 8(3), 175183.
Cao, Y., Rossiter, D., Edwards, D. W., Knechtel, J., & Owens,
D. (1998). Modelling issues for control structure selection in a
chemical process. Computers and Chemical Engineering, 22(Suppl.),
S411S418.
Cao, Y., Rossiter, D., & Owens, D. (1997a). Input selection for
disturbance rejection under manipulated variable constraints. Computers and Chemical Engineering, 21(Suppl.), S403S408.
Cao, Y., Rossiter, D., & Owens, D. (1997b). Screening criteria for input
and output selection. In Proceedings of European control conference,
ECC 97, Brussels, Belgium.
Cao, Y., Rossiter, D., & Owens, D. (1998). Globally optimal control
structure selection using branch and bound method. In Proceedings of
DYCOPS-5, fifth IFAC symposium on dynamics and control of process
systems, Corfu, Greece.
Douglas, J. M. (1988). Conceptual design of chemical processes. New York,
USA: McGraw-Hill.
Halvorsen, I. J., Skogestad, S., Morud, J. C., & Alstad, V. (2003). Optimal
selection of controlled variables. Industrial and Engineering Chemical
Research, 42, 32733284.
Herrmann, G., Spurgeon, S. K., & Edwards, C. (2003). A model-based
sliding mode control methodology applied to the hda-plant. Journal of
Process Control, 13, 129138.
Konda, N.V.S.N.M., Rangaiah, G.P., & Krishnaswamy, P.R. (2005).
Simulation based heuristics methodology for plant-wide control of
industrial processes. In Proceedings of 16th IFAC world congress,
Praha, Czech Republic.
Luyben, W. L. (2002). Plantwide dynamic simulators in chemical processing
and control. New York, USA: Marcel Dekker Inc.
Luyben, W. L., Tyreus, B. D., & Luyben, M. L. (1998). Plantwide process
control. New York, USA: McGraw-Hill.
Maarleveld, A., & Rijnsdorp, J. E. (1970). Constraint control on
distillation columns. Automatica, 6, 5158.
ARTICLE IN PRESS
A.C.B. de Araujo et al. / Control Engineering Practice 15 (2007) 12221237
McKetta, J. J. (1977). Benzene design problem. Encyclopedia of chemical
processing and design. New York, USA: Dekker.
Ng, C., & Stephanopoulos, G. (1996). Synthesis of control systems
for chemical plants. Computers and Chemical Engineering, 20,
S999S1004.
Ponton, J. W., & Laing, D. M. (1993). A hierarchical approach to the
design of process control systems. Transactions of the Institution of the
Chemical Engineers, 71(Part A), 181188.
Skogestad, S. (2000). Plantwide control: The search for the self-optimizing
control structure. Journal of Process Control, 10, 487507.
1237