Revelations on Dynamic Process Analysis,
Advanced Control,
and Online Optimization
This
“open-to-all” class was taught by Dr. Robert Bartman of ProControl, Inc., to
Advanced Process Control Engineers from refineries of BP and
ConocoPhillips. Direct student quotes
follow. Note: comments on ProControl’s
Discover software below come from a highly experienced BP APC engineer.
DISCOVER is a great tool. It’s got lots of useful commenting and help features. I can’t wait to use it back at the plant.
Discover is intelligent software, with features that have
clearly been included in response to real-life modeling problems – for
example, the capability to back out effects of specific variables whose impacts
are already known. I liked the
flexibilities on the ‘Calculated Variables’ tab. I liked the representation (both graphical
trends and percentages) of the influence of all variables in multivariable
modeling. I also liked Discover’s inline
advice and comments, its tooltips, and its Help screens.
This software looks like it was built by
an engineer, for engineers. It
should be a lesson to people not to get ‘hung up’ on just the slickness of an
interface, but instead to offer great functionality
and (very important) embedded, practical, documentation.
The quality of this software, and its documentation, says something about
the ability of smaller, more focused, companies to deliver better quality products than
‘sprawling’ software houses.
My
boss sent me to the course to learn how to spend time seeking out new
opportunities, and where we could improve the economic performance of the plant
by adding appropriate Advanced Control applications. Almost all of our control engineers have
completed this course, and have made real improvements to the refinery’s
performance as a result.
After
courses from Shinskey, and a 10-day workshop on DMC,
this course has finally made me confident in dealing with process
dynamics, PID feedback, feedforward control, and the more complex regulatory
controls – the things I’ve been working with for over 5 years!
I
was sent to this course to learn how to be a control engineer. I’m going home a much more effective
control engineer. I now know what I can
do to make money. The highest payout for
this class is the huge step change in my knowledge that you have given me.
This
was an extremely worthwhile and very useful course. I didn’t even know what was possible before I
came.
I
might have been bumbling along for another 10 years before I acquired this same
knowledge and understanding of process control.
I certainly did not learn a lot in my first year!
Course
topics
that interested me most were: a)
your (smart) seat-of-the-pants PID tuning technique, and b) how to do plant tests
– only because these two will be easiest to immediately apply without more
software tools. In c)
feedforward control, seeing the pitfalls of not ‘synching’ properly was an
unexpected surprise.
d) Your constraint
control strategy (e.g., for maximizing process unit feedrate) was also
good. The feedback one usually gets when
approaching this kind of control scheme is “It’s a poor man’s DMC.” But such comment forgets that most
operators see DMC as a Black Box, and that constraint control the way you
presented it would be much more obvious to them. Folks also forget that, in only implementing
DMC or RMPCT, they’re wasting the capabilities of the DCS’s we paid for – and
are wasting big $$ on unnecessarily complex APC applications.
Given
my company’s moratorium on using Derivative, e) your coverage on ‘D’ in
a PID was excellent and enlightening.
Long live Derivative!
Process dynamics coverage interested me the most. This opens the door to a rigorous, systematic approach to solving control problems. Also liked the design of advanced controls on towers and furnaces, utilizing dynamic analysis, feedforward, cascades, constraint control, and DR. We have plenty of room to make improvements here.
There’s
not anything in this course I thought extraneous.
This
was a great class. You obviously know
your stuff.
I
had ***** *****’s class, purporting to cover the same subjects. It was worthless. He likely knew his stuff, but I didn’t leave
with any of it.
The
highest payout for this class is probably in HC Fractionator constraint
control, which is do-able if some instrumentation issues can be resolved – but
which wasn’t looking so good prior to this course (we were “blindly” building a
DMC application).
Editorial
comment: Many more technical – and enthusiastic –
comments are available upon request.
Here’s one on how to wring the most from our education:
I see the most effective scenario for our utilizing your
training as: 1) give our new
control engineers an easy, unit-specific problem to solve (which, being new,
they’ll struggle with). 2) Then
send them to DCS school, about 3 to 4 months after
arriving. 3) Give them a more
challenging control problem. Let them
read the manual and struggle, so that when 4) they go to your
school, they have a learning objective in mind.
There should be about 9 months between DCS school
and your ‘Revelations in Control’ course.
Again, the point is to have questions, an objective, a
problem to solve – to arrive “with an end in mind”. [Editorial
comment: based on much experience,
ProControl fully concurs with this ‘homegrown control engineer’ training
curriculum.]
The instructor was great at explaining topics in an easy-to-understand way. His delivery was interesting and animated – never dull. He was a great teacher, making it all interesting and easy to remember.
The
instructor didn’t just present material from many different angles, but
encouraged thought processes to ensure that the students internalized the
material. I liked his “No one gets left
behind” philosophy. The homework and
exams ensured that people took this course seriously.
Bob
has a very strong ability to demystify complex control concepts, or (more
correctly) to recast complex things into simple ideas we can all relate
to. He also adapts well to the class’s
current level of understanding.
Thanks
for the positive comments – you’ve really boosted my confidence in doing more
creative things back in the plant.
Level
was excellent for instruction, as it showed how difficult this problem really
is. This is another tool I’d really like
to get back in the plant.
Really a good tool.
Things like finding the right nonlinear level algo
look hard by ‘winging it’. It should be
easy with this software.
Level’s
a very clever, and insightful, tool!