Control of Multiple-Input,

Multiple-Output Processes
Chapter 18

Multiloop controllers
• Modeling the interactions
• Relative Gain Array (RGA)
• Singular Value Analysis (SVA)
• Decoupling strategies
 Control of multivariable processes

 In practical control problems there typically are a

number of process variables which must be controlled
and a number which can be manipulated
Chapter 18

Example: product quality and through put

must usually be controlled.

• Several simple physical examples are shown in Fig.

18.1.
Note "process interactions" between controlled
and manipulated variables.
Chapter 18

SEE FIGURE 18.1

in text.
Chapter 18
 • Controlled Variables: xD , xB , P, hD , and hB

• Manipulated Variables: D, B, R , Q D , and Q B

Chapter 18
  In this chapter we will be concerned with characterizing process
interactions and selecting an appropriate multiloop control
configuration.

 If process interactions are significant, even the best multiloop

control system may not provide satisfactory control.
Chapter 18

 In these situations there are incentives for considering

multivariable control strategies

Definitions:

•Multiloop control: Each manipulated variable depends on

only a single controlled variable, i.e., a set of conventional
feedback controllers.

•Multivariable Control: Each manipulated variable can depend

on two or more of the controlled variables.

Examples: decoupling control, model predictive control

 Multiloop Control Strategy
•Typical industrial approach
•Consists of using n standard FB controllers (e.g. PID), one for
each controlled variable.

•Control system design

1. Select controlled and manipulated variables.
Chapter 18

2. Select pairing of controlled and manipulated variables.

3. Specify types of FB controllers.

Example: 2 x 2 system

Two possible controller pairings:

U1 with Y1, U2 with Y2 …or
U1 with Y2, U2 with Y1
Note: For n x n system, n! possible pairing configurations.
 Transfer Function Model (2 x 2 system)

Two controlled variables and two manipulated variables

(4 transfer functions required)
Chapter 18

Y1 ( s ) Y1 ( s )
 GP11 ( s ),  GP12 ( s )
U1 ( s) U 2 ( s)
Y2 ( s ) Y2 ( s )
 GP 21 ( s ),  GP 22 ( s )
U1 ( s ) U 2 (s)

Thus, the input-output relations for the process can be

written as:
Y1 ( s )  GP11 ( s )U1 ( s )  GP12 ( s )U 2 ( s )
Y2 ( s )  GP 21 ( s )U1 ( s )  GP 22 ( s )U 2 ( s )
 Or in vector-matrix notation as,
Y ( s )  GP ( s )U ( s)
Chapter 18

where Y(s) and U(s) are vectors,

Y1 ( s )  U1 ( s ) 
Y ( s)    , U ( s)   
Y (
 2  s ) U (
 2  s )
And Gp(s) is the transfer function matrix for the process

 G P11 (s) G P12 (s) 

G P (s)   
G
 P 21 (s ) G P 22 (s ) 
Chapter 18
 Control-loop interactions

• Process interactions may induce undesirable

interactions between two or more control loops.
Chapter 18

Example: 2 x 2 system
Control loop interactions are due to the presence
of a third feedback loop.
• Problems arising from control loop interactions
i) Closed -loop system may become destabilized.
ii) Controller tuning becomes more difficult
 Block Diagram Analysis

For the multiloop control configuration the transfer

function between a controlled and a manipulated
variable depends on whether the other feedback
Chapter 18

control loops are open or closed.

Example: 2 x 2 system, 1-1/2 -2 pairing
From block diagram algebra we can show
Y1 ( s )
 GP11 ( s ), (second loop open)
U1 ( s )
Y1 ( s ) GP12GP 21GC 2 (second loop closed)
 GP11 
U1 ( s) 1  GC 2GP 22
Note that the last expression contains GC2 .
Chapter 18
Chapter 18
Chapter 18

Figure 18.6 Stability region for Example 18.2 with 1-1/2-2 controller pairing
Chapter 18

Figure 18.7 Stability region for Example 18.2 with 1-2/2-1 controller pairing
 Relative gain array

• Provides two useful types of information:

Chapter 18

1) Measure of process interactions

2) Recommendation about best pairing of
controlled and manipulated variables.
• Requires knowledge of s.s. gains but not
process dynamics.
 Example of RGA Analysis: 2 x 2 system
• Steady-state process model,
Y1  K11U1  K12U 2
Y2  K 21U1  K 22U 2
Chapter 18

The RGA is defined as:

 12 
RGA   11 

 21  22 

where the relative gain, ij, relates the ith controlled

variable and the jth manipulated variable
open - loop gain
 ij 
closed - loop gain
 Scaling Properties:

i) ij is dimensionless
ii)   
ij ij  1.0
Chapter 18

i j

For 2 x 2 system,
1
11  , 12  1  11   21
K K
1  12 21
K11 K 22

Recommended Controller Pairing

Corresponds to the ij which has the largest positive

value.
 In general:
1. Pairings which correspond to negative pairings should
not be selected.
2. Otherwise, choose the pairing which has ij closest
to one.
Chapter 18

Examples:
Process Gain Relative Gain
Matrix, K : Array,  :

K11 0  1 0 
 0
 K 22 
 0 1 
 
 0 K12 
0 1 
K
 21 0   1 0 
 
K11
 0
K12 
K 22 
 1 0 
0 1 

 
 K11

0 
K 1 0 
 21 K 22  0 1 
 
 Recall, for 2X2 systems...
1
Y1  K11U1  K12U 2 11  , 12  1  11   21
K K
1  12 21
Y2  K 21U1  K 22U 2 K11 K 22

EXAMPLE:
Chapter 18

K K12   2 1.5
K   11   
K 21 K 22  1.5 2 
 2.29  1.29
  
 Recommended pairing is Y 1

 1.29 2.29  and U1, Y2 and U2.

EXAMPLE:
 2 1.5 0.64 0.36
K   
1. 5 2   0.36 0.64 
Recommended pairing is Y 1 with U1, Y2 with U2.
Chapter 18 EXAMPLE: Thermal Mixing System

The RGA can be expressed in terms of the manipulated variables:

Wh Wc
W  Wc Wh 
W  W Wc  Wh 
  c h

 Wh Wc 
T W  W Wc  Wh 
 c h

Note that each relative gain is between 0 and 1. Recommended

controller pairing depends on nominal values of W,T, Th, and Tc.

See Exercise 18.16

 EXAMPLE: Ill-conditioned Gain Matrix

y = Ku

2 x 2 process y1 = 5 u1 + 8 u2
Chapter 18

y2 = 10 u1 + 15.8 u2

specify operating point y, solve for u

Adj K
u=K y =-1
y
det K

K11K 22 K11K 22
RGA : 11  
K11K 22  K12K 21 det K

effect of det K → 0 ?
 RGA for Higher-Order Systems:
For and n x n system,
U1 U2  Un
Y1  11 12  1n 
Y2 21 22  2 n 

Chapter 18

    
 
Yn  n1 n1  nn 

Each ij can be calculated from the relation

 ij  K ij H ij
Where Kij is the (i,j) -element in the steady-state gain K:
matrix,
Y  KU

And Hij is the (i,j) -element of the H  K  

1 T
.
Note that,   KH
 EXAMPLE: Hydrocracker

The RGA for a hydrocracker has been reported as,

Chapter 18

U1 U2 U3 U4
Y1  0.931 0.150 0.080  0.164
Y2   0.011  0.429 0.286 1.154 

Y3  0.135 3.314  0.270  1.910 
 
Y4  0.215  2.030 0.900 1.919 

Recommended controller pairing?

 Singular Value Analysis
K = W  VT

 is diagonal matrix of singular values

(1, 2, …, r)
Chapter 18

The singular values are the positive square roots of the

eigenvalues of
KTK (r = rank of KTK)
W,V are input and output singular vectors Columns of W and
V are orthonormal. Also
WWT = I
VVT = I
Calculate , W, V using MATLAB (svd = singular value
decomposition)
Condition number (CN) is the ratio of the largest to the
smallest singular value and indicates if K is ill-conditioned.
 CN is a measure of sensitivity of the matrix properties to changes in a
specific element.
Consider  1 0
K  
10 1 
 (RGA) = 1.0
If K12 changes from 0 to 0.1, then K becomes a singular matrix, which
Chapter 18

corresponds to a process that is hard to control.

RGA and SVA used together can indicate whether a process is easy (or
hard) to control.

10.1 0 
 (K ) =   CN = 101
 0 0.1

K is poorly conditioned when CN is a large number (e.g., > 10). Hence

small changes in the model for this process can make it very difficult to
control.
 Selection of Inputs and Outputs
• Arrange the singular values in order of largest to
smallest and look for any σi/σi-1 > 10; then one or
Chapter 18

more inputs (or outputs) can be deleted.

• Delete one row and one column of K at a time and
evaluate the properties of the reduced gain matrix.
• Example:
 0.48 0.90 0.006 
K   0.52 0.95 0.008 
 0.90 0.95 0.020 
  0.5714 0.3766 0.7292 
W   0.6035 0.4093 0.6843 
 0.5561 0.8311 0.0066 
1.618 0 0 
   0 1.143 0 
 0 0 0.0097 
18

 0.0541 0.9984 0.0151

Chapter18

V   0.9985 0.0540 0.0068 

 0.0060 0.0154 0.9999 
Chapter

CN = 166.5 (σ1/σ3)

The RGA is as follows:

 2.4376 3.0241 0.4135 
 1.2211 0.7617 0.5407 
 
 2.2165 1.2623 0.0458 

Preliminary pairing: y1-u2, y2-u3,y3-u1.

CN suggests only two output variables can be controlled. Eliminate one input and one output
(3x3→2x2).
Chapter 18
Chapter 18
 Alternative Strategies for Dealing with Undesirable
Control Loop Interactions

1. "Detune" one or more FB controllers.

2. Select different manipulated or controlled variables.
e.g., nonlinear functions of original variables
Chapter 18

3. Use a decoupling control scheme.

4. Use some other type of multivariable control scheme.

Decoupling Control Systems

Basic Idea: Use additional controllers to compensate for process

interactions and thus reduce control loop interactions

Ideally, decoupling control allows setpoint changes to affect only

the desired controlled variables.

Typically, decoupling controllers are designed using a simple

process model (e.g. steady state model or transfer function model)
Chapter 18
 Design Equations:
We want cross-controller, GC12, to cancel out the effect of U2 on Y1.
Thus, we would like,
T12GP11U 2  GP12U 2  0
Chapter 18

Since U2  0 (in general), then

GP12
T12  
GP11
Similarly, we want G21 to cancel the effect of M1 on C2. Thus, we
require that...
T21GP 22U1  GP 21U1  0
GP 21
 T21  
GP 22
cf. with design equations for FF control based on block diagram
analysis
 Alternatives to Complete Decoupling

• Static Decoupling (use SS gains)

• Partial Decoupling (either GC12 or GC21 is set equal to
Chapter 18

zero)
Process Interaction

Corrective Action (via “cross-controller” or “decoupler”).

Ideal Decouplers:
GP12 ( s )
T12 ( s )  
GP11 ( s)
GP 21 ( s)
T21 ( s)  
GP 22 ( s )
 Variations on a Theme:

•Partial Decoupling:
Use only one “cross-controller.”

•Static Decoupling:
Chapter 18

Design to eliminate SS interactions

Ideal decouplers are merely gains:
K P12
T12  
K P11
K P 21
T21  
K P 22
•Nonlinear Decoupling
Appropriate for nonlinear processes.
Chapter 18
Chapter 18

Previous chapter Next chapter