120
BioFlo/CelliGen 115
y
M1369-0050
User’s Guide
19.5
What is P-I-D Tuning?
Tuning consists of establishing controller settings (the proportional, integral, and
derivative constants) such that the controller provides proper control. If the
P-I-D
constants are incorrect, the control signal may be too weak for the parameter to ever
reach setpoint or, at the other extreme, the controller may respond excessively to small
errors, causing the actuator to oscillate between high and low values. Usable
P-I-D
constants must be determined for each
P-I-D
loop. The process is largely one of
calculated trial and error.
All loops that are configured with the
P-I-D
control mode must be tuned. When
delivered as part of an NBS system,
P-I-D
loops will have been tuned at the factory to
work correctly with the NBS-controlled instruments. For other applications, the user is
responsible for
P-I-D
tuning.
Tuning can be a complex task for those unfamiliar with the process, which is why a
trained engineer or technician normally performs this task. A number of textbooks
1
that explain the theory and describe the process could be useful for the
mathematically-inclined novice. The Ziegler-Nichols method, described in the
footnoted reference, is used at our production facilities.
The following suggestions are intended for novices. Be sure to refer to a textbook, and
consider utilizing the services of a technician.
•
Allow sufficient time for the task. Tuning is an iterative process. It consists of
configuring a loop with trial
P
,
I
and
D
values, evaluating loop response, then
readjusting the constants. The process is repeated until the loop responds fully and
without oscillation.
•
One usually begins with a trial
P
, setting
I
and
D
to zero. After
P
is established, a
similar iterative process establishes
I
.
•
Most fermentor probes respond too slowly or are too noisy to utilize the
D
term to
advantage. In most cases,
D
should remain at zero. Agitation is sometimes an
exception.
•
The magnitude of the control signal depends on the
P
,
I
and
D
constants. It also
depends inversely on a
Normalizing Constant.
1
For example, Chinks, F.G., Process Control Systems: Application, Design, and Tuning, McGraw-Hill
(1988), New York, Auckland, Bogota, London, Toronto, Sydney, Tokyo, Montreal.
