Decoupling Control Schemes

of Multivariable Systems
( 多变量系统的解耦控制 )

Dai Lian-Kui Shen Guo-jiang

Institute of Industrial Control,

Zhejiang University
 Last Lecture
 Definition of Relative Gain
 Calculation of Relative Gain
 Meaning of Relative Gain Matrix
 Calculation of Relative Gain Matrix
 CVs and MVs Pairing
 Problems of Multivariable Process
 Which is the best pairing of controlled and
manipulated variables?
 How does the interaction affect the stability of
the loops?
 How should the feedback controllers be tuned
in a multivariable scheme?
 Can something be done with the control
scheme to break, or minimize, the interaction
between loops?
 Interaction and Stability
Characteristic equations for loop 1(No.2
controller are in “M” model ):
1 + GC1Gv1G11 = 0
Characteristic equations for loop 2(No.1
controller are in “M” model ):

1 + GC 2Gv 2G22 = 0
Characteristic equations for complete system( both controllers are in “A” mode):

(1 + GC1Gv1G11 )(1 + GC 2Gv 2G22 ) - GC1Gv1GC 2Gv 2G12G21 = 0

 Interaction and Stability (cont.)
 The roots of the characteristic equation for each
individual loop are not the roots of the
characteristic equation for the complete system.
 For interaction to affect the stability, it must work
both ways.
 The interaction effect on one loop can be
eliminated by interrupting the other loop; this is
easily done by switching the controller to manual.
 Tuning PID Controllers
for 2*2 Interacting Systems
 If one loop is much faster than the other one,
the fast loop is tuned first with the other loop
in Manual. Then the slow loop is tuned with
the faster loop in Automatic.
 If both loops are about the same speed of
response, and one variable is more important
to control than the other one, detune the less
important loop by setting a small gain & a long
reset time.
 Other cases……
 Content
 When Necessary to Design Decoupler
 Linear Decoupler Design from Block
Diagrams
 Nonlinear Decoupler Design from Basic
Principles
 Application Examples
 Pairing Example
of a Blending Process
�y1 � �F � �u1 � �F1 �
� �= � � , � �= � �
�y2 � �C� � u2 � �F2 �

Steady-state model:
y1 = u1 + u2
C1u1 + C2u 2
y2 =
u1 + u2

Problem: how can you analyze

the coupling between two loop ?
 Pairing of a Blending Process
(cont.)
1. Obtaining steady-state process gain:
y1 = u1 + u2
�Dy1 � �1 1 � �Du1 �
y2 =
C1u1 + C2u2 � �= � �
� �
�Dy2 � �K 21 K 22 �
�Du2 �
u1 + u2

K 21 =
�
y2
=
( C1 - C2 ) u2 , K 22 =
�y2
=
( C2 - C1 ) u1
( 1 2) ( 1 2)
2 2
�
u1 u2 u + u �
u2 u1 u + u
 Pairing of a Blending Process
(cont.)
2. Obtaining relative gain matrix:
�Dy1 � �1 1 � �Du1 �
� �= � �
� �
�Dy2 � �K 21 K 22 �
�Du2 �

K11 1 u1 u2
m11 = =
K12 K 21 K 21 u1 u2
K11 - 1- y1
K 22 K 22 u1 + u2 u1 + u2
1 u1
= = ; u2 u1
u2 u1 + u2 y2
1+ u1 + u2 u1 + u2
u1
 Pairing of a Blending Process
(cont.)
3. Pairing CVs and MVs using RGM
F1 F2  If F1>F2, the correct
F1 F2 pairing is F － F1, C － F2;
F
F1 + F2 F1 + F2
 If F2>F1, the correct
F2 F1
C pairing is F － F2, C － F1.
F1 + F2 F1 + F2
If F2=F1, which is the
correct pairing ?
 When necessary to apply
decoupling for a coupling process
(1) Strong interaction exists （存在强耦合） :
even for the best pairing option, one or
more of the relative gains is far from unity,
(2) Both loops are about the same speed of
response, and both CVs are of the same
important （两回路动态特性接近，且两被
控变量均同等重要） .
 Problem Discussion
For the controlled system,
design your decoupling
control systems and
simulate your solution with
SimuLink. If
C1 F1 + C2 F2
F = F1 + F2 , C = ;
F1 + F2
Fm 0.5 C 1 Am e -5 s
= , = , = ;
F s + 1 C 4s + 1 C 2s + 1
Initial states:
F10 = 75, F20 = 25, C1 = 60%, C2 = 40%.
 Decoupler Design
 Decoupler （解耦器） Design Principle
 Linear Cascade Decoupler Based on
Transfer Function Matrix G(s)
 Linear Feedforward Decoupler from G(s)
 Nonlinear Steady-state Cascade
Decoupler Based on Basic Principles ( 基
于过程机理的非线性稳态解耦器 )
 Decoupler Design Principle

Design decoupler to cancel the Dy2 m Dy1m

= 0, =0
effects of the cross blocks, i. e., Dv1 Dv2
Block Diagram for a General 2*2
System with Cascade Decoupler

Decoupling Conditions ?
 Cascade Decoupler
for a General 2*2 System
 y1m ( s)   G11 G12  D11 D12  v1 ( s) 
  =    
 y2 m ( s)   G21 G22  D21 D22  v2 ( s ) 
Dy2 m Dy1m
= 0, =0
Dv1 Dv2
G11 D12 + G12 D22 = 0, G21 D11 + G22 D21 = 0

If D11 = 1, G12 G21

D12 = - , D21 = -
D22 = 1 G11 G22
 About Cascade Decoupler

Problem: (1) initial MVs’ value or “Man/Auto” mode switch;

(2) limit of MVs.
Block Diagram for a 2*2 System
with Feedforward Decoupler
 Feedforward Decoupler Design
for a 2*2 System
Dynamic decoupler:
Dy2 m G12 ( s) G21 ( s)
= 0, D12 ( s) = - , D21 ( s) = -
Dv1 G11 ( s ) G22 ( s)
Dy1m Steady-state decoupler:
=0
Dv2 K12 K 21
D12 = - , D21 = -
K11 K 22

Feedforward Control & Feedforward Decoupler

Existing Problems ?
 Nonlinear Steady-state
Complete Decoupler ( 全解耦器 )

Design principle: introduce new variables v1, v2 to

satisfy Dy2 Dy1 u1 = f1 (v1 , v2 )
= 0, = 0 and
Dv1 Dv2 = 0
Dv2 Dv1 = 0
u = f (v , v )
2 2 1 2
 Nonlinear Steady-state
Triangular Decoupler ( 三角解耦 )

Design principle: introduce new variables v1, v2 to

satisfy Dy2 m Dy1m u1 = f1 (v1 , v2 )
= 0 or = 0 and
Dv 1 Dv 2 u = f (v , v )
2 2 1 2
 Triangular Decoupler Design
for a Blending Process
�y1 � �F � �u1 � �F1 �
� �= � � , � �= � �
�y2 � �C� � u2 � �F2 �

Steady-state model:
y1 = u1 + u2
C1u1 + C2u2
y2 =
u1 + u2
u2
Let v2 = y1 = u1 + v2u1
u1
C1u1 + C2 v2u1 C1 + C2 v2
y2 = =
u1 + v2u1 1 + v2
Triangular Decoupler for the
Blending Process (cont.)
 Complete Decoupler Design
for the Blending Process
�y1 � �F � �u1 � �F1 �
� �= � � , � �= � �
�y2 � �C� � u2 � �F2 �

Steady-state model:
y1 = u1 + u2
C1u1 + C2u2
y2 =
u1 + u2

Let v1 = u1 + u2 y1 = v1
u2 y2 = C1 (1 - v2 ) + C2 v2
v2 =
u1 + u2
 Nonlinear Decoupler Design
Example (cont.)
�y1 � �F � �u1 � �F1 �
� �= � � , � �= � �
�y2 � �C� � u2 � �F2 �

Steady-state model:
y1 = v1
y2 = C1 (1 - v2 ) + C2 v2

Nonlinear decoupler:
v1 = u1 + u2
u2 u1 = v1 - v1v2
v2 =
u1 + u2 u2 = v1v2
 Implementation of Nonlinear
Decoupler

Decoupler:
u1 = v1 - v1v2
u2 = v1v2
 Simulation of Nonlinear
Decoupler for a Blending Process

( Please see …/Decoupling/NonlinearDecoupling.mdl )

 Simulation Results
of Nonlinear Decoupler
F, Fsp
115

110
T/hr

105

100

95
0 20 40 60 80 100 120 140 160 180 200

C, Csp
56

54

52
%

50

48
0 20 40 60 80 100 120 140 160 180 200
min
 Problem Discussion
There are two branches in the following pump. Please
design a control system to stabilize the flow in each branch,
and reduce the coupling between the branches.
 Summary
 Concept of Relative Gains
 Obtaining Process Gains and Relative
Gain Matrix
 Rule of CVs and MVs Pairing
 Decoupling Design from Block Diagrams
 Decoupling Design from Basic Principles
 Next Topic:
Boiler Control Schemes
 Process Diagram and Control Problems of
Boiler ( 锅炉设备的生产流程与控制问题 )
 Characteristic Analysis & Three-element
Control for Drum Level ( 汽包水位特性分析与
三冲量控制 )
 Steam Temperature Control ( 蒸汽温度控制 )
 详见《过程控制工程》 12.4 节 (P.214)
 Problem Discussion ( 《过程控制工程》思考
题与习题 12-3, P.223)