Re: Controller TYPE2b

["Calener" <calener@xxxxxxxx>]

I have tryed your advices in the system but with only one controller to
limit the error.
Working with one controller it get "stuck", although it get rid of all
oscillations using a small timestep (0.02h).
How much sorter has to be a timestep than the time constant of my system to
get rid of all the oscillations?
How could I know and change the time constant of my system?

I didn´t understand "the tank has enough inercia to de-couple the
controllers" (sorry, I am an inexpert)

On the other hand, I was checking the outputs and TRACE and a lot of times
the controller don´t work well, and most of them are in the ON criteria. In
adiction, sometimes the collector outlet temperature changes to a too much
high value, for example from 30 to 70ºC instead the control output is OFF
and the diference with tank temperature is bigger then the Upper input
temperature (Monitoring =100ºC).
It could be because the pump is OFF and the fluid temperature in the
collector increases but I think it is too much.

Bye and best regards

Alberto
U. de Oviedo
España


----- Original Message ----- 
From: Michaël Kummert <kummert@xxxxxxxxxxxxx>
To: <trnsys@xxxxxxxxxxxxxxxxxxx>
Sent: Wednesday, May 26, 2004 6:27 PM
Subject: Re: Controller TYPE2b



I sent a reply to a similar problem recently but the message is not in
the archive yet so here is a slightly edited copy/paste:

Oscillations are normal for an ON-OFF controller considering that you
are using a fixed time step and iterating during the simulation. Having
the controller "Stick" to one state is necessary in a simulation to
prevent endless oscillations.

The fact that the controller(s) are stuck during x% of the time is not
worrying per se because in most cases you can hope that errors will
cancel each others in average. Of course, some times you may end-up with
a controller stuck in the bad state for long periods of time, especially
if you have 2 controllers acting on the same variable. In your case I
would assume the tank has enough inertia to de-couple the controllers so
you should not get into that kind of problem.

You may want to check what happens during the iterations of one time
step taking into account the component order: Using a TRACE command on
the two controllers and the tank can help you understand why controllers
get stuck.

Sometimes, reducing the time step can help you. If you go down to a
timestep much shorter than the time constant of your system you should
get rid of all oscillations and the controller should never stick
(again, this may not be true if 2 controllers are acting in opposite
ways on the same variable). Another option is to use a proportional
(actually a PI or PID) controller.

I hope this helps,

Michaël Kummert


-- 
_________________________________________________________

Michaël Kummert

Solar Energy Laboratory - University of Wisconsin-Madison
1303 Engr Res Bldg, 1500 Engineering Drive
Madison, WI 53706

Tel: +1 (608) 263-1589
Fax: +1 (608) 262-8464
E-mail: kummert@xxxxxxxxxxxxx

SEL Web Site: http://sel.me.wisc.edu
TRNSYS Web Site: http://sel.me.wisc.edu/trnsys