Controller Modes - Part 3
Controller
Composite Tuning
Controller
Prof. Sneh. K. Soni
Assistant Professor (Inst. & Control)
Room No. D306 (Third Floor)
[Link]@[Link]
Controller Tuning
What is controller tuning?
Why it is required?
What is “Good Control”?
Controller Tuning
Controller Tuning
Quarter Decay Performance?
Controller Performance
Controller Performance
Controller Tuning Techniques
Controller tuning techniques:
Closed-loop tuning method.
Open-loop tuning method.
Closed-Loop Tuning (Ziegler-Nichols)
Method
The continuous cycling method was proposed by
Ziegler and Nichols in 1942 and is the best known
methods for tuning PID controllers.
This is a closed-loop tuning methodology and is
also known as loop tuning or ultimate gain method
since the first step involves obtaining the closed-loop
ultimate [Link]).
Ultimate Gain (kcu) is the largest value of the controller
gain kc that maintains closed-loop stability when a
proportional-only controller is used.
Closed-Loop Tuning (Ziegler-Nichols)
Method
Step 1: Eliminate integral and derivative actions, i.e.
keep a proportional-only (P) controller.
Step 2: Set kc at a low value and place the controller in
automatic.
Step 3: Increase the controller gain kc by small
increments until the closed-loop system response
exhibit continuous cycling (or sustained oscillations
with constant amplitude). The period of the sustained
oscillations is referred to as the ultimate period, pu.
Step 4: The PID controller settings are calculated using
kcu and Pu based on Ziegler-Nichols (ZN) tuning
relations.
Ziegler-Nichols Settings
Ziegler-Nichols tuning relations:
Controller kc I D
P 0.5kcu --- ---
PI 0.45kcu Pu / 1.2 ---
PD 0.6kcu ---- Pu / 8
PID 0.6kcu Pu / 2 Pu / 8
Closed-Loop Tuning (Ziegler-Nichols)
Method
Advantages Disadvantages
Closed-loop experiments are Continuous cycling may be
used. objectionable since the process
The controller settings are is pushed to the stability limit.
easily calculated. Caution should be used for
Does not assume a model open-loop unstable processes.
structure for the process Some slow processes required
output. long time in the tuning.
Controller Tuning Strategy
We should emphasize that the settings
provided by either of the techniques have to
regarded as “good” first estimates and
subsequent fine-tuning should be considered.
Example
• Consider the following process:
2e s
g p (s ) ; g f (s ) 1; g m (s ) 1
10s 15s 1
• Design a PI & PID controller using ZN methods.
Example - ZN
Using P-controller
only, the closed-loop
system is brought to
the edge of the
stability limit. Pu= 12
The proportional gain
that yields sustained
oscillations and the
period of oscillations kcu= 8
are noted.
Controller kc I D
PI 3.63 10 ---
PID 4.70 6 1.5
Example - ZN
PI Controller
Set-point Response Disturbance Response
Example - ZN
PID Controller
Set-point Response Disturbance Response
Open-Loop Tuning (Cohen-Coon)
Method
• Consider the “open” (manual) control system:
d (s)
A
c (s ) g d (s )
s
ysp (s) e(s) + y (s)
g c (s ) g f (s )
m(s)
g (s ) +
+
-
ym (s)
g m (s )
• Introduce a step change of magnitude A in the manipulated
variable, c(s).
• Record the value of the output, ym(t), with respect to time.
Process Reaction Curve
Process Reaction Curve
B k
Find the parameters of a
ym(s)
FOPDT transfer function
from the process reaction
curve.
ym ( s ) k e t D s
g PRC ( s )
c( s) p s 1
K=B/A =B/S
tD time
Cohen-Coon Settings
Use the FOPDT parameters (k, , tD) to compute the
parameter settings for P, PI or PID controllers using
Cohen-Coon (CC) relations.
For P controller:
1 tD
kc 1
k tD 3
PI and PID Controller
For PI Controller
tD
1 tD 30 3
kc 0 .9 I tD
k tD 12 9 20
tD
For PID Controller
tD
4
32 6
1 tD D tD
kc 1.33 I tD 11 2
tD
k tD 4 13 8
tD
Cohen-Coon Settings
Advantages Disadvantages
The experimental test must be
Only a single experiment is performed in open-loop.
necessary. It may be difficult to determine the
It does not require a trial- slope at the inflection point accurately,
especially for noisy systems.
and-error procedure.
The method is based on the
The controller settings are assumption that the FOPTD model
easily calculated. approximates the actual process
behavior well.
It does not consider the dynamics of
all the component
Example
• Consider the following process:
2e s
g p (s ) ; g f (s ) 1; g m (s ) 1
10s 15s 1
• Design a PI & PID controller using CC Methods.
Example - CC
2
y(t)
The process 1.5
reaction curve is
obtained and its 1 2e 4s
parameters are g PRC (s )
determined. (13s 1)
0.5
They are used to
obtain CC settings.
0
0 20 40 time 60
Controller kc I D
PI 1.50 8.16 ---
PID 2.29 8.75 1.37
Example - CC
PI Controller
Set-point Response Disturbance Response
Example - CC
PID Controller
Set-point Response Disturbance Response
PI Controller Tuning Example
Unit-step disturbance responses for the closed loop system
(FOPTD Model: K = 1, d=4 and =20
