Control Chapter 7

CHAPTERS 7
TUNING OF FEEDBACK CONTROLLERS
IVÁN D. PORTNOY DE LA OSSA, M.E.,

M.Sc.
MECHANICAL ENGINEERING DEPARTMENT

INTRODUCTION
The tuning of a feedback controller refers to the adjustment of its parameters in
order to obtain a specified closed-loop response. Thus, before tuning the controller,
a process characterization must be performed. We will look at two methods for
carrying out process characterization:
• On-line (closed loop)
• Off-line (step-testing)
The difficulty for tuning increases with the number of parameters to be adjusted.
We will also look at some different specifications of control loop performance, e.g.;
Quarter Decay Ration (QDR) response, minimum integral of absolute error (min IAE)
response, etc. However, keep in mind that one single procedure cannot give the
best results for all process control situations.
PROCESS CHARACTERIZATION:
ON-LINE (CLOSED LOOP) METHOD:
Proposed by Ziegler and Nichols in 1942. It consists of two steps:
Step 1: Determination of the dynamic characteristics of the control loop. These

characteristics are represented by the ultimate gain, , and the ultimate period of
oscillation, , of the loop
Step 2: Estimation of the controller tuning parameters that produce a desired

response.
QUARTER DECAY RATIO (QDR) BY ULTIMATE GAIN
If transfer functions of all components of the loop are known, the dynamic
parameters and can be analytically determined. This is not usually the case, and we
must often determine them from the actual process by the following procedure:
1. Switch off the integral and derivative modes so as to have a P controller.
2. With the controller in automatic (closed-loop), increase the proportional gain
slowly until the loop goes oscillatory with constant amplitude. Record the value
of the gain that produces sustained oscillations as .
3. The period of oscillation is measured and recorded as .

For the desired response, Ziegler and Nichols specified a decay ratio of one-fourth,
i.e., the ratio of the amplitudes of two consecutive oscillations.
Once and have been determined, they are used to calculate the controller parameters
using formulas in the following table:
The QDR response is very desirable for disturbance inputs, because it prevents a
large initial deviation from set point without being too oscillatory. However, it is not
so desirable for step changes in set point because it causes a 50% overshoot.
This difficulty can be easily corrected by reducing the controller gain from the value
predicted by the formulas.
Another difficulty of using QDR tuning formulas is that (except for the case of the P
controller) the set of tuning parameters necessary to obtain such behavior is not
unique
OPEN-LOOP PROCESS CHARACTERIZATION
The Ziegler-Nichols on-line tuning method introduced before is the only one that
characterizes process by and . Most methods approximate processes to a first order
plus dead time (FOPDT) or a second order plus dead time (SOPDT) transfer function.
Where and .
As the controller concerns, is the process, and this is precisely what must be
approximated, if possible, as a FOPDT or SOPDT transfer function.
A FOPDT looks like:
While a SOPDT looks like:
For underdamped processes:

OPEN-LOOP PROCESS CHARACTERIZATION:
PROCESS STEP TESTING
The step test procedure is carried out as follows:

1. With the controller on manual apply a step change in the signal . The
magnitude of this change should be large enough for the consequent change in
the transmitter signal to be measurable, but not so large that the response will
be distorted by the process nonlinearities.
2. Record the response of from the time of introduction of the step test until the
system reaches a new steady state. Typically, a step test lasts between a few
minutes and several hours, depending on the speed of response of the process.
PROCESS STEP TESTING
PROCESS STEP TESTING:
APPROXIMATION TO FOPDT (FIT 3)
Once open-loop step response has been performed, the FOPDT parameters can be
determined by a fit 3 procedure:
∆𝑐 3
𝑘=
Δ𝑚
, 𝜏=
2
( 𝑡 2 − 𝑡 1 ) , 𝑡 0 =𝑡 2 − 𝜏
TUNING FOR QDR RESPONSE
In addition to on-line tuning formulas, Ziegler and Nichols developed off-line
formulas based on a FOPDT model fit to the process reaction curve.
TUNING FOR QDR RESPONSE
In using the formulas of Table 7-2.1, we must keep in mind that they are empirical
and apply only to a limited range of ratios of dead time to time constant. They are
most applicable for a range of of around 0.10 to 0.5.
As pointed out in the discussion of on-line tuning, the QDR formulas can be adjusted
to less oscillatory responses by reducing the proportional gain from the value given
by the tuning formula.
TUNING FOR MINIMUM ERROR INTEGRAL CRITERIA
Since QDR tuning parameters are not unique, Paul W. Murrill and Cecil L. Smith
developed new tuning formulas.
They used the FOPDT model parameters. Their specification for closed loop
response was to minimize the error or deviation of the controlled variable from set
point.
The integral of the error cannot be minimized directly, because a very large negative
error would be the minimum. In order to prevent negative values of the
performance function, the following formulations of the integral can be proposed:
• Integral of the absolute value of the error:
• Integral of the square of the error:

Unfortunately, the optimum set of parameter values is a function not only of integral
definition but also of the type of input (that is, disturbance or set point) and of its
shape (for example, step change, ramp, and so on).
The step change input is usually selected because it is the most disruptive. In terms of
the input type, we must select either set point or disturbance input for tuning,
according to which one is expected to affect the loop more often.
Most process controllers are considered regulators, except for the slave controllers in
cascade control schemes, which are servo regulators (in this case set point changes
are the most important).
MINIMUM IAE FORMULAS FOR DISTURBANCE INPUTS
Lopez et al. (1967) developed tuning formulas for minimum error integral criteria.
They are most applicable for a range of of around 0.10 to 1

MINIMUM IAE FORMULAS FOR SET POINT CHANGES
They are most applicable for a range of of around 0.10 to 1

MINIMUM IAE FORMULAS FOR DISTURBANCE:
PERFORMANCE
MINIMUM IAE FORMULAS FOR DISTURBANCE:
PERFORMANCE
OPEN-LOOP PROCESS CHARACTERIZATION: DAHLIN
SYNTHESIS TUNING FORMULAS
OPEN-LOOP PROCESS CHARACTERIZATION: LAMBDA
TUNING
5% Overshoot 0% Overshoot
P
PI
PID
TUNING: EXCERCISES:
7-15
TUNING: EXCERCISES:
7-15
TUNING: EXCERCISES:
7-15
TUNING: EXCERCISES:
7-15
CASE STUDY 2 (APPENDIX E)

Control Chapter 7

Uploaded by

Document Informationclick to expand document information

Copyright:

Available Formats

Control Chapter 7

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Control Chapter 7

Uploaded by

Copyright:

Available Formats

CHAPTERS 7

TUNING OF FEEDBACK CONTROLLERS

IVÁN D. PORTNOY DE LA OSSA, M.E.,

MECHANICAL ENGINEERING DEPARTMENT

Proposed by Ziegler and Nichols in 1942. It consists of two steps:

Step 1: Determination of the dynamic characteristics of the control loop. These

Step 2: Estimation of the controller tuning parameters that produce a desired

3. The period of oscillation is measured and recorded as .

While a SOPDT looks like:

For underdamped processes:

The step test procedure is carried out as follows:

• Integral of the square of the error:

They are most applicable for a range of of around 0.10 to 1

They are most applicable for a range of of around 0.10 to 1

You might also like