Design of Feedback Controllers

Dr. Meera Balachandran

Department of Chemical Engineering and Materials Science
 What are we going to learn?
• How do we select the type of feedback
controller?
• How do we adjust the controller parameters
to achieve optimum response?
 3 Design Questions
1. What type of feedback controller should be
used?
2. How do we select the best values of the
adjustable parameters?
Controller tuning
3. What performance criteria should we use for
selection and tuning of the controller?
 PID Controller
• PID control law
 1 t de(t ) 
p (t )  p  K c e(t )  0 e(t )dt   D dt 
* *

 I
• Transfer function
P' ( s)  1 
 Gc ( s )  K c 1    D s 
E (s)  Is 
• Controller tuning
– Need to select PID parameters (Kc, tI, tD) that yield “good”
closed-loop response to disturbances and setpoint changes
– Only know: Kc > 0, tI > 0, tD >0
– Trial-and-error tuning difficult and time consuming
– Need methods to determine good initial parameter values
– Refine parameter values by trial-and error fine tuning
 Performance Criteria
• Steady state performance criteria
eg. Zero error at steady state
• Dynamic performance criteria
- Criteria that use only a few points of the
response
Simpler but approximate
- Criteria that use the entire closed loop response
from time t = 0 to t = very large
More precise but cumbersome
Closed-Loop Performance Criteria
1. Stability of closed-loop system
2. Minimization of disturbance effects
3. Rapid, smooth tracking of setpoint
changes
4. Elimination of steady-state offset
5. Avoidance of excessive control action
6. Robustness to changes in process
conditions
7. Robustness to errors in process model
 Simple Performance Criteria
• Overshoot
• Rise time
• Settling time
• Decay ratio
• Frequency of oscillation of transient response
Usually more than one objectives should be satisfied
May lead to conflicting response characteristics
Balanced by the designer
Characteristics of an Underdamped
Response

• Rise time
B C ±5% • Overshoot (B)
• Decay ratio
(C/B)
y(t)

D
• Settling or
response time
• Period (T)
Time
trise trt
Impact of Controller Tuning

e 4 s
G p ( s )  Gd ( s ) 
20 s  1
 Decay Ratio
• One – quarter decay criterion
• Decay ratio of 0.25
• Reasonable trade-off between a fast rise time
and a reasonable settling time.
 Time – Integral Performance Criteria

Based on the entire response of the process

from t = 0 until steady state has been reached
• Integral of the square error (ISE)
• Integral of the absolute value of error (IAE)
• Integral of the time – weighted absolute error
(ITAE)
 Time – Integral Performance Criteria

Integral of the square error (ISE)

• To suppress large errors

• Errors are squared and thus contribute more
to the value of the integral
 Time – Integral Performance Criteria

Integral of the absolute value of error (IAE)

• For suppressing small errors, IAE is better that

ISE
 Time – Integral Performance Criteria
• Integral of the time – weighted absolute error
(ITAE)

• To suppress errors that persist for a long time.

• Large t amplifies the effect of even small
errors in the value of the integral
 Time – Integral Performance Criteria
• Different criteria lead to different designs
• For the same time – integral criterion,
different input changes give different designs
 Dynamic Performance Criteria

Tuning for a Disturbance Input Tuning for a Set Point Input

 General Tuning Rules
• The controller gain (Kc) is inversely proportional to
the process gain (K).
• Kc decreases as the ratio of the time delay (q) and
the dominant process time (t) constant increases.
• Both the integral time (tI) and the derivative time (tD)
increase as q/t increases.
• A reasonable initial guess for the derivative mode is
tD = 0.25tI.
• Kc decreases as integral control to a proportional
controller. Kc increases as derivative control is added
to a PI controller.
 3 Design Questions
1. What type of feedback controller should be
used?
2. How do we select the best values of the
adjustable parameters?
3. What performance criteria should we use for
selection and tuning of the controller?
 Answer to the3 Design Questions
1. Define appropriate performance criterion
2. Compute the value of the performance
criterion using P, PI or PID controller with the
best settings for the adjusted parameters Kc,
τI and τD
3. Select the controller which gives the best
value for the performance criterion
 Drawbacks of Time – Integral Performance Criteria

• Mathematically rigorous, tedious

• Relies on models / transfer functions for the
process, sensor and final control element
which may not be known exactly
• Incorporates certain ambiguities as to which is
the most appropriate criterion and what input
changes to consider
 Controller Tuning Approaches
1. Use simple criteria eg. One quarter decay ratio,
minimum settling time etc.
Simple approach. Easy to implement on an actual
process. Usually provides multiple solutions and
appropriate solution is chosen using additional
specifications on the performance
2. Use time – integral criteria
Cumbersome, relies on mathematical models, time
consuming
3. Use proven empirical rules
 On-Line PID Controller Tuning
• Often a process model is not available for controller
tuning
• Need to determine tuning parameters directly from
plant data
• Continuous cycling method
– Utilizing a proportional controller, find the Kc value that
produces sustained oscillations. This is the ultimate gain Kcu.
– Determine the period of the sustained oscillations. This is
the ultimate period Pu.
– Kc and Pu can also be found from an available transfer
function model by direct substitution.
 Ziegler-Nichols Rules (Z-N)
• Proposed by Ziegler and Nichols (1942),for
Taylor Instrument Co.USA.
• A closed-loop tuning method
• The controller remains in the loop as an active
controller in automatic mode.
• The ultimate gain (KU) and ultimate period (PU)
be obtained from a closed-loop test of the
actual process / frequency response.
 Ziegler-Nichols Rules (Z-N)
1. After the process reaches steady state at the normal
level of operation, remove the integral and derivative
modes of the controller, leaving only proportional
control.
2. Select a value of proportional gain (KC), disturb the
system, and observe the transient response. If the
response decays, select a higher value of KC and again
observe the response of the system. Continue increasing
the gain in small steps until the response first exhibits a
sustained oscillation. The value of gain and the period of
oscillation that correspond to the sustained oscillation
are the ultimate gain (KU) and the ultimate period (Pu).
 Ziegler-Nichols Rules (Z-N)
3. From the values of ultimate gain (KU) and the
ultimate period (Pu), found in the previous
step, use the Ziegler- Nichols rules to
determine controller settings (Kc, τI and τD).
 Shortcomings of Continuous Cycling Method
• Time consuming for processes with large time
constants
• Process pushed to stability limit.
• Ultimate gain does not exist for first- and second-
order systems without time delays
• Generally not applicable to integrating and open-
loop unstable systems because stability is
achieved only for intermediate Kc values.

First two shortcomings can be overcome using relay
autotuning method to determine Kc and Pu (see text).
 Cohen and Coon Rules (C-C)
Process Reaction Curve Method
• Open-loop method
• The control action is removed from the
controller by placing it in manual mode and an
open-loop transient is induced by a step
change in the signal to the valve
• Proposed by Cohen and Coon (1953)
• Used as an alternative to the Z-N method.
 Cohen and Coon Rules (C-C)

1. After the process reaches steady state at the

normal level of operation, switch the
controller to manual.
 Cohen and Coon Rules (C-C)
2. With the controller
in manual,
introduce a small
step change in the
controller output
that goes to the
valve and record
the transient, which
is the process
reaction curve
 Cohen and Coon Rules (C-C)
3. Draw a straight line tangent to
the curve at the point of
inflection. The intersection of
the tangent line with the time
axis is the apparent transport lag
(Td); the apparent first-order
time constant (T) is obtained
from
T = Bu / S
where Bu is the ultimate value of
B at large t and S is the slope of
the tangent line. The steady-
state gain that relates B to M in
is given by
KP = Bu /M
 Cohen and Coon Rules (C-C)
4. Using the values of Kc, τI and τD from step 3, the
controller settings are found from the relations
given in Table
Cohen and Coon Rules (C-C)
 Shortcomings of Step Test Method
• Experimental test
performed under
open-loop conditions.
• Method is not
applicable to open-
loop unstable
processes
• Step response can be
sensitive to the
direction and
magnitude of the step
change if the process is
nonlinear.
 Summary
• Design questions
• PID Controllers
• Performance Criteria
• Tuning Methods