Competition

Setanta wrote:

Interesting but some of us do this every day in their own field.   What
kind of systems do you think are notoriously difficult?


An applied controls book would be interesting.   People are alway
looking for application examples in their own field.  It takes a lot of
work to publish a book.  Tim would be the expert on this.   I know that
even magazine articles take a lot of work.

Herding cats comes to mind.

Peter Nachtwey

This is the point though

A lot of books are NOT about the practicalities - Some people write books on
control just to show their proficiency in Mathematics.

I can think of many BUT

Astrom and Hagglung have provided plently of material here - which I an in
possession of


The idea I have is to target energy usage as the primary optimisation -
using a variety of controllers to acchieve this - without introducing
operator - setup difficulties.

My language thoughts on this are to transform the more complex controller
types to an interface that emulates a PID controller for the operator - but
the underlying controller will actually be more complex (not just for the
sake of it)



I am fashioning a crosier as we speak :-)

That immediately introduces what many practitioners will perceive as a
fairly artificial criterion. Fact is that for process plants, energy
conservation tends to be a second-tier incentive for good control.

Bruce Varley wrote:

In fact, one often wants control goals that _maximize_ the energy
consumption -- at least the energy consumed by the actuators.  This is
because you're often putting a controller in there to increase
bandwidth, and that means pushing your actuator hard.

Where I've seen energy conservation used as a parameter it had more to
do with fooling an otherwise robustness-agnostic algorithm into toning
down it's control rule in hopes that the result wouldn't send a real
system flying into wild oscillations.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google?  See http://cfaj.freeshell.org/google/

"Applied Control Theory for Embedded Systems" came out in April.
See details at http://www.wescottdesign.com/actfes/actfes.html

proclaimed to the world:


I will give an example that I think demonstrates what you are saying
in mixed terms, at least for me. Tell me if this fits what you are
asserting.

We were redesigning a control system for a pumping station. The
station has VS pumps that take from a 30 ft wet well/storage tank. The
old system used PID controls of the three pumps in cascade. When a
single pump reached 80% capacity, then the controls would add another
pump and bring them to 40% each. During peak flows, the main system
would change the set point for tank level from the normal 20 ft mark,
down to 10 ft gradually over a period of 8 hrs or so. This scheme was
in use to spread peak loads out over the low times. The system also
had a algorithm that lowered the power consumption during peak periods
by overriding the set points if the total plant was coming close to
the peak power value set by the utility. They were under contract
which penalized you if you exceeded the peak.

Anyway, during the redesign, it was argued for doing away with PID at
the pumps because they had problems with stability. It place of PID
the other controls engineer on staff to the engineering firm insisted
that straight proportional control was all they needed and trying to
keep a set level was needless. I argued that even though the system
would be more stable, you would give away tank storage capacity equal
to the range of the proportional system. He argued this would be
slight because the proportional would be set at three feet. The system
was pushing it's capacity for storage anyway, but now it appeared to
be alright to give away 10% of that because they had trouble getting
PID stable.

Please apply your idea to this system. I assume you would say in my
system that I wanted to maximize energy usage of the pump system
because it responded faster and to a greater degree.


Here I am assuming you are arguing with me in my case, but you can see
that energy conservation needs to be better defined.

On Jan 15, 9:21 am, Paul M <PaulMatWiredogdotcom> wrote:

I find it hard to believe that a tank level control can't be made
stable.


Again,  this happens when there isn't a good understanding of what the
system is doing.

I don't think anyone is arguing the point that one can use the tank as
an accumulator so that peak demands are reduced.   This is done all the
time in hydraulic control systems.  It makes good sense.  Why size the
motors or pumps for peak loads when they can be run at an average load?
 This just takes a little more design work up front to optimize
performance.

I agree with Tim about the bandwidth.  I have only been asked once
about LQR type of control in 25 years and I don't think that guy really
know what he was asking for.  Most of the time I get asked about how
fast can I move from here to there and there are no concerns about the
power required.  A lot of my customers are happy to run the system in
saturation until the last millisecond before ramping down to a nice
stable stop with no overshoot.

Peter Nachtwey



Peter Nachtwey

Peter Nachtwey wrote:

   ...


The amount of inflow is uncontrolled, but the maximum rate of level
change is known: the interceptor is only so big, and flows by gravity.
The rate at which pumped flow can be allowed to change is very limited.
Do not underestimate the water-hammer effect in a few miles of 1-meter pipe.


I think that energy spent in the actuators is a small part of the energy
of the plant. Moving water is much more costly than moving valves.

   ...

Jerry
--
Engineering is the art of making what you want from things you can get.
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

On 16 Jan 2007 02:40:05 -0800, "Peter Nachtwey"


And it can, but the problems came from this being a 1970 built system
with three pumps that cut in and out. You need a control that takes
this into account. You need a system that takes kilotons of inertial
energy into account. You just can't try to stop 6 miles of water in a
three foot pipe to come to a stop very quickly. Early on, they cracked
a check valve trying. I have been in the pump pit when a pump is shut
down and water hammer slams the valve shut. It is scary. The building
shakes to the point where dust rains down from the roof.


This is not what I was asking or saying. Of course accumulators work.
That is what the tanks were there for. I was saying that control
redesign would make 10% of the tank capacity unusable for
accumulation. They used tank storage to keep from going over a peak,
set by the power company. Going over the peak during certain hours of
the day caused them to get fined by the contract. They used storage
tanks hold wastewater during the day and pump it at night when energy
costs were lower too.


But I was talking about an industry where energy conservation has a
lot to do with the process. Bandwidth has little to do with anything.
The OP brought up different concerns his design software looked at. I
don't know what kind of controls it was intended to be used with.

On Jan 18, 12:52 am, Paul M <PaulMatWiredogdotcom> wrote:

Try generating 10 Hz sine waves with an actuator with a 10 Hz band
width.

Peter Nachtwey

On 18 Jan 2007 17:56:20 -0800, "Peter Nachtwey"


Stop quoting out of context and go listen to some 60 cycle humm. :-)

Peter Nachtwey wrote:


Maybe these energy concerns are some kind of relics from the space era in
control engineering in the 1960s.

Z.H.

Nowadays we just use A Relativity Drive

Well, despite what many people might expect, in my experience for the
process industries  the profile of energy as a control improvement driver
has stayed pretty much where it always was in the scheme of things. What has
become hugely more important for CEs, and instrument guys as well, is
'licence to operate' issues. In the old days if your product was off spec
you just snuck it out into the marketplace, if your plant was smelly then
that was tough. Now either of these situations can put you out of business
if you're caught out by a regulatory authority. If you're looking for a
current-day hot area, then I'd look no further than LTO. However, the
transfer functions and smart algos still tend to be pretty much 'second
tier' behind the people and business issues, as they've always been.

Energy saving is coming online

optimisation -

It's still a second-tier incentive for good control.  If energy saving was a
plant's primary incentive, the solution is easy - turn off the power and go
home.  You can't get much better than that...

Cameron:-)

Not sure where your comin from Cameron - Sure - Energy saving is a secondary
consideration - BUT - tell your top management you can save them x amount of
dinari per blip and see what happens -

Setanta

Setanta wrote:


How can you save energy on a motion application?   The amount of work
required to move an object from point a to point b will not change no
matter what control algorithm you use.

Energy will not be saved by the control algorithm.  Motion control
requires that energy added to the system is equal to the work required
on a millisecond to millisecond basis.  I don't care what control
algorithm you use that balance must exist if one is going to follow a
motion profile.

P=F*v
P=M*a*v

Since the acceleration and velocity are carefully controlled at each
point along the path the power required at each point will be the same
and the energy required is just the sum of the power required at all
the points.

The power or energy optimization is done at design time by reducing
mass or friction.  

Peter Nachtwey

Not all applics involve motion control

Setanta wrote:


Ok, so you don't want the motion control guys to win.  That doesn't
sound fair.  ;) Fine.
So what do you have in mind?

On a motion control application, like a flying shear, one can save
energy by minimizing the length of travel required to synchronize and
make the cut and get back.   Minimizing the required distance minimizes
the required energy.
The question is how much does each part of the motion controller
contribute to the minimizing the required motion.  Is the the PID?
Feedback resolution?  Or is the motion profile itself.  Maybe it is all
of the above and more.

Perhaps someone knows of an application where the control algorithm is
the significant contributor to efficiency.

Peter Nachtwey