Chapter 6
Multivariable control
Introduction to Multivariable Control
• Up to this point, the fundamentals of process dynamics and control have
been illustrated by single-input single-output (SISO) systems. The
processes encountered in the real world are usually multiple-input
multiple-output (MIMO) systems.
• To explore these concepts, consider the interacting, two-tank liquid-level
system in Fig. 1 where there is one input, the flow to tank 1 ( m1), and one
output, the level in tank 2 ( h2 ). In this figure, h2 is related to m1 by a
second-order transfer function.
• From the point of view of a SISO system, the relation between h2 and m1
may be represented by the block diagram in Fig.1 b. One may place a
feedback control system around the open-loop system of Fig. 1 b to
maintain control of H2.
SISO system: two interacting tank system
Fig. 1
MIMO system: two interacting tank system
Fig. 2a
MIMO system: two interacting tank system
• Now consider the process of Fig. 2a in which there are two inputs
(𝒎𝟏 and 𝒎𝟐 ) and two outputs ( 𝒉𝟏 and 𝒉𝟐 ).
• A change in 𝒎𝟏 alone will affect both outputs (𝒉𝟏 and 𝒉𝟐 ). A change in
𝒎𝟐 alone will also change both outputs.
• The interaction between inputs and outputs can be seen more clearly
by the block diagram of Fig. 2b. In this diagram, the transfer
functions show how the change in one of the inputs affects both of the
outputs.
MIMO system: two interacting tank system
Fig. 2b
MIMO system : CONTROL OF INTERACTING SYSTEMS
• The problem of controlling the outputs of a MIMO system will be
discussed by means of a 2x2 system shown in Fig. 3 .
Fig. 3
MIMO system : CONTROL OF INTERACTING SYSTEMS
• The problem can be extended to the case of more than two pairs of
inputs and outputs by the same procedure described here.
• The control objective is to control C1 and C2 independently, in spite of
changes in M1 and M2 or other load variables.
• Two control loops are added to the diagram of Fig. 3 as shown in Fig.
4 . Each loop has a block for the controller, the valve, and the
measuring element.
CONTROL OF INTERACTING SYSTEMS
Fig. 4
CONTROL OF INTERACTING SYSTEMS
• In principle, the multi-loop control system of Fig. 4 will maintain
control of C1 and C2. However, because of the interaction present in
the system, a change in R1 will also cause C2 to vary because a
disturbance enters the lower loop through the transfer function G21.
• Because of interaction, both outputs ( C1 and C2) will change if a
change is made in either input alone.
• If G21 and G12 provide weak interaction, the two-controller scheme
of Fig. 4 will give satisfactory control.
• In the extreme, if G12=G21=0, we have no interaction and the two
control loops are isolated from each other.
Decoupling
• To completely eliminate the interaction between outputs and set points,
two more controllers (cross-controllers) are added to the diagram of
Fig. 4 to give the diagram shown in Fig. 5 . In principle, these cross-
controllers can be designed to eliminate interaction.
Decoupling
Fig. 5
Decoupling
• From Fig. 5 , we may write by direct observation the following
relationships in the form of the matrix expression.
𝑪 = 𝑮𝑷 𝑴
𝑪𝟏 𝑮𝟏𝟏 𝑮𝟏𝟐 𝑴𝟏
=
𝑪𝟐 𝑮𝟐𝟏 𝑮𝟐𝟐 𝑴𝟐
Where:
𝑪𝟏 𝑮𝟏𝟏 𝑮𝟏𝟐 𝑴𝟏
𝑪= , 𝑮𝑷 = and 𝑴 =
𝑪𝟐 𝑮𝟐𝟏 𝑮𝟐𝟐 𝑴𝟐
Decoupling
𝑴𝟏 = 𝑮𝒗𝟏 𝑮𝒄𝟏𝟏 𝑬𝟏 + 𝑮𝒗𝟏 𝑮𝒄𝟏𝟐 𝑬𝟐
𝑴𝟐 = 𝑮𝒗𝟐 𝑮𝒄𝟐𝟏 𝑬𝟏 + 𝑮𝒗𝟐 𝑮𝒄𝟐𝟐 𝑬𝟐
𝑴 = 𝑮𝒗 𝑮𝒄 𝑬
𝑴𝟏 𝑮𝒗𝟏 𝟎 𝑮𝒄𝟏𝟏 𝑮𝒄𝟏𝟐 𝑬𝟏
=
𝑴𝟐 𝟎 𝑮𝒗𝟐 𝑮𝒄𝟐𝟏 𝑮𝒄𝟐𝟐 𝑬𝟐
𝑮𝒗𝟏 𝟎 𝑮𝒄𝟏𝟏 𝑮𝒄𝟏𝟐 𝑬𝟏
𝑮𝒗 = , 𝑮𝒄 = and 𝑬 =
𝟎 𝑮𝒗𝟐 𝑮𝒄𝟐𝟏 𝑮𝒄𝟐𝟐 𝑬𝟐
Where:
• 𝑮𝒗 -valve matrix and 𝑮𝒄 -controller matrix
Decoupling
• From Fig. 5 , we can write directly
𝑬𝟏 = 𝑹𝟏 − 𝑮𝒎𝟏 𝑪𝟏
𝑬𝟐 = 𝑹𝟐 − 𝑮𝒎𝟐 𝑪𝟐
𝑬 = 𝑹 − 𝑮𝒎 𝑪
𝑬𝟏 𝑹𝟏 𝑮𝒎𝟏 𝟎 𝑪𝟏
= −
𝑬𝟐 𝑹𝟐 𝟎 𝑮𝒎𝟐 𝑪𝟐
𝑮𝒎𝟏 𝟎 𝑹𝟏
𝑮𝒎 = and 𝑹 =
𝟎 𝑮𝒎𝟐 𝑹𝟐
Where:
• 𝑮𝒎 - measurement matrix and 𝑹- set point
Decoupling
𝑪 = 𝑮𝑷 𝑴 and 𝑴 = 𝑮𝒗 𝑮𝒄 𝑬
Therefore, 𝑪 = 𝑮𝑷 𝑮𝒗 𝑮𝒄 𝑬
Let 𝑮𝒐 = 𝑮𝑷 𝑮𝒗 𝑮𝒄
Then, 𝑪 = 𝑮𝒐 𝑬 substituting , 𝑬 = 𝑹 − 𝑮𝒎 𝑪
𝑪 = 𝑮𝒐 𝑹 − 𝑮𝒐 𝑮𝒎 𝑪
Solving for C from the above equation
𝑪 = 𝑰 + 𝑮𝒐 𝑮𝒎 −𝟏 𝑮𝒐 𝑹
−𝟏
𝑪𝟏 𝑰 𝟎 𝑮𝟏𝟏 𝑮𝟏𝟐 𝑮𝒗𝟏 𝟎 𝑮𝒄𝟏𝟏 𝑮𝒄𝟏𝟐 𝑮𝒎𝟏 𝟎 𝑹𝟏
= + 𝑮𝒐
𝑪𝟐 𝟎 𝑰 𝑮𝟐𝟏 𝑮𝟐𝟐 𝟎 𝑮𝒗𝟐 𝑮𝒄𝟐𝟏 𝑮𝒄𝟐𝟐 𝟎 𝑮𝒎𝟐 𝑹𝟐
MIMO block diagram
• Notice that the closed-loop behavior expressed by this matrix equation
is analogous to the closed-loop response of a SISO system, which may
be written
𝑮𝒐 𝒔
𝑪 𝒔 = 𝑹(𝒔)
𝟏 + 𝑮𝒐 𝒔 𝑮𝒎 𝒔
• Block diagram for MIMO control system in terms of matrix blocks.
Fig. 6
MIMO block diagram
• In this diagram, the blocks are filled with matrices. The double line
indicates that more than one variable is being transmitted. Each
block contains a matrix of transfer functions that relates an output
vector to an input vector.
Fig. 7
Decoupling
• In order for no interaction to occur between C and R in Fig. 5 (i.e., R1
affects only C1 and R2 affects only C2), the off-diagonal elements of
𝑰 + 𝑮𝒐 𝑮𝒎 −𝟏 𝑮𝒐 must be zero. Since 𝑰 and 𝑮𝒎 are diagonal,
𝑰 + 𝑮𝒐 𝑮𝒎 −𝟏 𝑮𝒐 will be diagonal if 𝑮𝒐 is diagonal. Multiplication of
the matrices in the expression for 𝑮𝒐 is shown below:
𝑮𝒐 = 𝑮𝑷 𝑮𝒗 𝑮𝒄
𝑮𝟏𝟏 𝑮𝟏𝟐 𝑮𝒗𝟏 𝟎 𝑮𝒄𝟏𝟏 𝑮𝒄𝟏𝟐
𝑮𝒐 =
𝑮𝟐𝟏 𝑮𝟐𝟐 𝟎 𝑮𝒗𝟐 𝑮𝒄𝟐𝟏 𝑮𝒄𝟐𝟐
𝑮𝟏𝟏 𝑮𝒗𝟏 𝑮𝒄𝟏𝟏 + 𝑮𝟏𝟐 𝑮𝒗𝟐 𝑮𝒄𝟐𝟏 𝑮𝟏𝟏 𝑮𝒗𝟏 𝑮𝒄𝟏𝟐 + 𝑮𝟏𝟐 𝑮𝒗𝟐 𝑮𝒄𝟐𝟐
𝑮𝒐 =
𝑮𝟐𝟏 𝑮𝒗𝟏 𝑮𝒄𝟏𝟏 + 𝑮𝟐𝟐 𝑮𝒗𝟐 𝑮𝒄𝟐𝟏 𝑮𝟐𝟏 𝑮𝒗𝟏 𝑮𝒄𝟏𝟐 + 𝑮𝟐𝟐 𝑮𝒗𝟐 𝑮𝒄𝟐𝟐
Decoupling
• Setting the off-diagonal elements to zero and solving for 𝑮𝒄𝟏𝟐 and 𝑮𝒄𝟐𝟏
give:
𝑮𝟏𝟐 𝑮𝒗𝟐 𝑮𝒄𝟐𝟐
𝑮𝒄𝟏𝟐 = −
𝑮𝟏𝟏 𝑮𝒗𝟏
𝑮𝟐𝟏 𝑮𝒗𝟏 𝑮𝒄𝟏𝟏
𝑮𝒄𝟐𝟏 =−
𝑮𝟐𝟐 𝑮𝒗𝟐
Example: two interacting tank system
Fig. 8
• Consider the above system where: 𝑨𝟏 = 𝟏, 𝑨𝟐 = 𝟏Τ𝟐, 𝑹𝟏 = 𝟏Τ𝟐, 𝑹𝟐 = 𝟐,
𝑹𝟑 = 𝟏
• For the two-tank, interacting liquid-level system shown in Fig. 8, develop
the block diagram for a MIMO system and Decuple the control system.
Example: two interacting tank system
• Material balances around tank 1 and tank 2 give the following
differential equations:
𝒅𝑪𝟏 𝑪𝟏 − 𝑪𝟐 𝑪𝟏
𝑨𝟏 = 𝒎𝟏 − −
𝒅𝒕 𝑹𝟏 𝑹𝟑
𝒅𝑪𝟐 𝑪𝟐 − 𝑪𝟏 𝑪𝟐
𝑨𝟐 = 𝒎𝟐 − −
𝒅𝒕 𝑹𝟏 𝑹𝟐
𝑪ሶ 𝟏 = 𝒎𝟏 − 𝟑𝑪𝟏 + 𝟐𝑪𝟐
𝑪ሶ 𝟐 = 𝟐𝒎𝟐 + 𝟒𝑪𝟏 − 𝟓𝑪𝟐
Example: two interacting tank system
• These equations may be written in matrix form as
𝑪ሶ = 𝑨𝑪 + 𝑩𝑴
−𝟑 𝟐 𝟏 𝟎
𝑨= and 𝑩 =
𝟒 −𝟓 𝟎 𝟐
𝑪ሶ = 𝑨𝑪 + 𝑩𝑴 solution of this differential equation is:
𝑪 𝒔 = 𝑺𝑰 − 𝑨 −𝟏 𝑩𝑴 𝒔 = 𝑮𝑷 𝑴
𝑮𝑷 = 𝑺𝑰 − 𝑨 −𝟏 𝑩
𝒔+𝟓 𝟒
𝟒 𝟐(𝒔 + 𝟑)
𝑮𝑷 =
(𝒔 + 𝟏)(𝒔 + 𝟕)
Example: two interacting tank system
𝒔+𝟓 𝟒 𝟒 𝟐(𝒔+𝟑)
𝑮𝟏𝟏 = , 𝑮𝟏𝟐 = , 𝑮𝟐𝟏 = and 𝑮𝟐𝟐 =
(𝒔+𝟏)(𝒔+𝟕) (𝒔+𝟏)(𝒔+𝟕) (𝒔+𝟏)(𝒔+𝟕) (𝒔+𝟏)(𝒔+𝟕)
• The block diagram can now be drawn as shown in Fig. 9
Fig. 9
Example2: decupling
• For the two-tank liquid-level
system of Example 1, determine
the controller transfer function
matrix 𝑮𝑪 needed to eliminate
interaction. The primary
controllers are to be
proportional; i.e., 𝑮𝑪𝟏𝟏 =
𝑲𝟏 and 𝑮𝑪𝟐𝟐 = 𝑲𝟐 . The
diagram of the control system
is shown in Fig. 10
Fig. 10
Example2: decupling
• The block labeled Controller contains the four transfer functions that
are the elements of 𝑮𝑪 .
• Assume 𝑮𝒗 is a unit diagonal matrix, i.e., 𝑮𝒗𝟏 = 𝑮𝒗𝟐 = 𝟏.
𝑮𝟏𝟐 𝑮𝒗𝟐 𝑮𝒄𝟐𝟐 𝑮𝟏𝟐 𝑮𝒄𝟐𝟐 𝟒 𝒔+𝟏 𝒔+𝟕
•𝑮𝒄𝟏𝟐 = − = − =− 𝑲𝟐 =
𝑮𝟏𝟏 𝑮𝒗𝟏 𝑮𝟏𝟏 𝒔+𝟏 𝒔+𝟕 𝒔+𝟓
𝟒𝑲𝟐
−
𝒔+𝟓
𝑮𝟐𝟏 𝑮𝒗𝟏 𝑮𝒄𝟏𝟏 𝑮𝟐𝟏 𝑮𝒄𝟏𝟏 𝟒 𝒔+𝟏 𝒔+𝟕
•𝑮𝒄𝟐𝟏 = − = − =− 𝑲𝟏 =
𝑮𝟐𝟐 𝑮𝒗𝟐 𝑮𝟐𝟐 𝒔+𝟏 𝒔+𝟕 𝟐 𝒔+𝟑
𝟐𝑲𝟏
−
𝒔+𝟑
Example2: decupling
• Having found the transfer functions for the cross-controllers, we can
now determine the nature of the uncoupled response of c1 to a change
in r1 and of c2 to a change in r2.
𝑮𝟏𝟏 𝑮𝒄𝟏𝟏 + 𝑮𝟏𝟐 𝑮𝒄𝟐𝟏 𝟎
𝑮𝒐 =
𝟎 𝑮𝟐𝟏 𝑮𝒄𝟏𝟐 + 𝑮𝟐𝟐 𝑮𝒄𝟐𝟐
𝑲𝟏
𝟎
𝑮𝒐 = 𝒔 + 𝟑
𝟐𝑲𝟐
𝟎
𝒔+𝟓
Example2: block diagram for decoupled MIMO system
Fig. 11
Example2: step response
• From Fig. 11, we may write directly
𝑪 = 𝑮𝟎 𝑬
𝑬=𝑹−𝑪
𝑪 = 𝑮𝟎 𝑹 − 𝑮𝟎 𝑪
𝑪𝟏 𝑮𝟎𝟏𝟏 𝟎 𝑹𝟏 𝑮𝟎𝟏𝟏 𝟎 𝑪𝟏
= −
𝑪𝟐 𝟎 𝑮𝟎𝟐𝟐 𝑹𝟐 𝟎 𝑮𝟎𝟐𝟐 𝑪𝟐
𝑪𝟏 = 𝑮𝟎𝟏𝟏 𝑹𝟏 − 𝑮𝟎𝟏𝟏 𝑪𝟏
𝑪𝟐 = 𝑮𝟎𝟐𝟐 𝑹𝟐 − 𝑮𝟎𝟐𝟐 𝑪𝟐
𝑮𝟎𝟏𝟏
𝑪𝟏 = 𝑹𝟏 (𝒔)
𝟏 + 𝑮𝟎𝟏𝟏
𝑮𝟎𝟐𝟐
𝑪𝟐 = 𝑹𝟐 (𝒔)
𝟏 + 𝑮𝟎𝟐𝟐
Example2: decupling
𝑲𝟏
𝑪𝟏 = 𝒔 + 𝟑 𝑹𝟏 (𝒔)
𝑲𝟏
𝟏+
𝒔+𝟑
𝟐𝑲𝟐
𝑪𝟐 = 𝒔 + 𝟓 𝑹𝟐 (𝒔)
𝟐𝑲𝟐
𝟏+
𝒔+𝟓
The result shows that the cross-controllers give two separate non-
interacting control loops, as shown in Fig. 12
Example2: decupling
Fig. 12
Example2: decupling Simulink implementation
STABILITY OF MULTIVARIABLE
SYSTEMS
𝑪 = 𝑰 + 𝑮𝒐 𝑮𝒎 −𝟏 𝑮 𝑹
𝒐
𝑨𝒅𝒋 𝑰 + 𝑮𝒐 𝑮𝒎
𝑪= 𝑮𝒐 𝑹
𝑰 + 𝑮𝒐 𝑮𝒎
𝑰 + 𝑮𝒐 𝑮𝒎 = 𝟎 characteristic equation
