Introduction to feedback control

October 6,2016 1
 Outline
• Feedback control concept
• Types of feedback controllers
• Controller design principles
• Tuning of the PID-controller
• Cascade, Feedforward and Ratio Control
• Adaptive control

March 23, 2009 2

 Feedback control concept

March 23, 2009 3

 Feedback control concept, a block diagram
d

Y (sp) e c Control Y(meas.)

+ Controller Process
element
-

Measuring
element

March 23, 2009 4

 Types of feedback controllers
• P-controller (Proportional)
• PI-controller (Proportional+Integral)
• PID-controller
(Proportional+Integral+Derivative)

March 23, 2009 5

 P-controller
• C(t)=K*e(t)
• A proportional controller is described by the
value of its proportional gain K.
• C(t)=controller output
• e(t)=error signal
e(t)=y(setpoint)-y(measurement)

March 23, 2009 6

 PI-controller
• Proportinal-integral controller
t
K
C (t )  Ke(t ) 
I  e(t )dt
0
• I is the integral time constant and it is an
adjustable parameter
• The integral term of a PI-controller causes its
output change as long as there is a non-zero error
signal

March 23, 2009 7

 PID-controller
• Proportional-integral-derivative controller
t
K de
C (t )  Ke(t )   e(t )dt  K D
I 0 dt
• With the presence of the derivative term, the PID
controller anticipates what the error will be in the
immediate future and applies a control action which
is proportional to the current rate of change in the
error

March 23, 2009 8

 Controller properties: P-controller 1/2
• Remaining offset between setpoint and
measured variable

March 23, 2009 9

 Controller properties: P-controller 2/2
• Accelerate the control system response
• Leaves nonzero steady-state offset

• Use when offsets are unimportant – or when

there is a natural integrator
– Liquid level controls, for example

March 23, 2009 10

 Controller properties: PI-controller
• Eliminates offset
• Response becomes oscillatory
• As Kc increases, the integral action
increases the propensity towards instability

• Use when offsets cannot be tolerated

March 23, 2009 11

 Controller properties: PID-controller
• No offset
• Derivative action stabilizes oscillation
• Derivative action amplifies noise
components in noisy signals

• Use when important to compensate natural

sluggishness (slow response, long time lags)
– and when the system is noise free
March 23, 2009 12
 Controller design principles
• Controller type
– P, PI, PID
• Parameters for the controller
– Performance criteria:
• Stability
• Speed
• Accuracy

March 23, 2009 13

 Performance criteria
• Stability, Speed, Accuracy
– Partly conflicting with each other
– E.g. increasing the speed of the control always makes
the control more unstable.
– Compromises must always be made
– High speed and good accuracy should be achieved, but
a certain distance to the limit of stability must be kept.
– If the controller was tuned very near the limit of
stability, changes in the process might make the system
unstable.

March 23, 2009 14

 Performance criteria
• Minimum rise time
tr
• Min. settling time ts
• Specified max.
Overshoot a1 < a1,spec
• Specified max. Decay
ratio a2/a1 < dr

March 23, 2009 15

 Tuning of the PID-controller
1. Limit of stability method
2. Step response method

March 23, 2009 16

 Limit of stability tuning method 1/2
• Use proportional control only and the
feedback control loop closed.
• Increase the proportional gain until the
system oscillates continuously
– Gain: K=Kkr
– The period of the oscillation Tkr

March 23, 2009 17

 Limit of stability tuning method 2/2
• Get the PID parameters from the table:
Controller Gain K Integrating time TI Derivating time TD
P 0.5 KKR - -
PI 0.45 KKR 0.85 TKR -
PID 0.6 KKR 0.5 TKR 0.12 TKR

March 23, 2009 18

 Step response method 1/3
• An open-loop step input test is carried out
• The process input is a step-like function,
with step height K1.
– Step input K1 causes a change with magnitude
K2 in output. A tangent is drawn in the turning
point of the response curve.

March 23, 2009 19

 Step response method 2/3

March 23, 2009 20

 Step response method 3/3

• Get the PID parameters from the table:

Controller Gain K Integrating time TI Derivating time TD
K1 T2
P K 2 T1 - -

K1 T2
PI 0.8  3 T1 -
K 2 T1

K1 T2
PID 1.2  2 T1 0.42 T1
K 2 T1

March 23, 2009 21

 Controller Tuning Example

March 23, 2009 22

 Processes with significant disturbances
• Pressure change in steam flow
• Change in the feed temperature

March 23, 2009 23

 Cascade control (1/2)
• Two PID control loops
• Flow controller to control the flow
• Temperature controller
adjusts the setpoint of
flow controller
• Output of TC (master)is
the set-point for the
FC (slave)
March 23, 2009 24
 Cascade control (2/2)
• The disturbance to be regulated must be
within the inner loop
• The inner control must respond much more
quickly than the outer loop

March 23, 2009 25

 Feedforward control
• With feedforward contol it is possible to
compensate to measured disturbances
before the process is affected

March 23, 2009 26

 Feedforward control

March 23, 2009 27

 Feedforward control - considerations
• Not useful if disturbance cannot be
measured
• Requires a process model
– Sometimes modelling is not possible
– The better a model represents the behaviour of
a process, the better feedforward controller will
be

March 23, 2009 28

 Ratio control 1/2
• Mostly used to control the ratio of flow
rates of two streams
• Both flow rates are measured but only one
can be controlled
– The uncontrolled stream is usually referred to
as wild stream

March 23, 2009 29

 Ratio control 2/2
• Used a lot in the process industry
• Process flow applications – example mixer:

March 23, 2009 30

 Adaptive control systems
• Adaptive is called a control system, which
adjusts its parameters automatically in such
a way as to compensate for variations in the
characteristics of the process it controls.
• The various types of adaptive control
systems differ only in the way the
parameters of controller are adjusted.

March 23, 2009 31

 Why adaptive controllers are needed? 1/2
• Most processes are non-linear
– The linearized models that are used to design
linear controllers depend on the particular
steady-state around which the process is
linearized.
– As the desired steady-state operation of a
process changes, the best values of the
controller’s parameters change.

March 23, 2009 32

 Why adaptive controllers are needed? 2/2
• Most processes are non-stationary i.e. their
characteristics change with time
– Reactions in the reactors

March 23, 2009 33

 Objective of the adaptation procedure
• The objective function that will guide the
adaptation mechanism to the best
adjustment of the controller parameters.
• A criterion to guide the adaptive tuning of
the controller

March 23, 2009 34

 Different types of adaptive control schemes
• Scheduled adaptive control
• Model-reference adaptive control
• Self-tuning controllers

March 23, 2009 35

 Scheduled adaptive control
• Controller parameters are adjusted
according to preprogrammed schedule

March 23, 2009 36

 Model-reference adaptive control
• the reference model that consists of a
reasonable model of how the process should
respond to a set-point change.

March 23, 2009 37

 Self-tuning controller 1/2
• Uses process input and output to estimate
recursively, on-line, the parameters of an
approximate process model.
• Thus as the actual nonlinear process changes
operating region or changes with time, an
approximate linear model is continuously updated
with new parameters.
• The updated model is then used in a prespecified
control system design procedure to generate
updated controller parameters.
March 23, 2009 38
 Self-tuning controller 2/2

March 23, 2009 39