Feature
Story
Computational
Intelligence Comes
of Age:
University of Iowa discovers
new process control methods
Ferman Milster, PE, Associate Director, Utilities and Energy Management, University of Iowa
Awise person once told me that the
energy plant of the future will be
staffed with a human and a dog.
The human’s purpose will be to feed the
dog. The dog’s purpose will be to ensure
the human doesn’t touch the controls.
Indeed, I have observed a marvelous
transformation in process controls from
my first experience as a boiler sight glass
watchman using a manual feed water valve
to my current experience with advanced
digital process controls. However, I’m
not quite ready to turn it all over to the
computer…yet.
As the University of Iowa (UI) has
learned, district energy plants can benefit
from the use of automated ‘intelligent’
computer algorithms and programs to
become more efficient and reliable. Computational methods and software systems
are available to help system operators
and engineers discover, display and utilize
process ‘knowledge’ that until now was
too complex to understand or was hidden
from them.
etc. – that are not readily apparent to
plant operators and engineers. These relationships are automatically monitored,
and when the system detects that a process
relationship is deviating from ‘normal,’
it provides an early warning. Typically,
this will occur before process alarm set-
points are reached. Automated discovery
of these conditions provides more time
for operators and engineers to investigate
the abnormality. It may be that the condition is simply one that has not been
observed by the software, or it may be
actual degradation in a machine that, if not
Automated Knowledge
Discovery
These advanced systems can help us
learn the relationships between process
variables – flows, pressures, temperatures,
Courtesy University of Iowa.
The control room at the University of Iowa’s central plant.
