PID BLOCK sampling time on AB PLC 1

[josedavidch]

Hi all,

I have been dealing with an industrial oven, where the regular process temperature controller were replaced by a PLC in order to automate some other task. The problem is that my PLC does not have auto-tune like some other models and then, I am dealing to get the PID block tunned. So far, I have used the same PI paramaeters as was used by the old temperature controller, but seems not work specialy because of the sampling time I am using to trigger the PID block. Short sample looks to be quite ustable, while long sample time produces high overshoot and since the oven is quite a good built unit, it wont decrease temperature fast enough (bad). So, without get into much detail, what is the thumb rule to determine the sampling time, which allow me as 2nd step, to play with PI parameters?

Thanks.

I have run a datacollection and pasted on

http://dexautomation.com/loopanalyzer.php.

Raw data is attached along with report.

 

[FrancisL]

How are you triggering the PID?
It should be run at a fixed sample rate that is not too fast but fast enough.
Typically you should sample 5 to 10 times the process time constant.
See this for example


Francis

http://www.controldraw.co.uk

http://www.s88control.blogspot.com

 

[josedavidch]

Thanks Francis,

I found my perfect sampling already. I also experimented that too fast normally does not affect the performance of the process but could be a waster of resources (as the note you referenced states). However, too slow will dramatically affect the process trend.

I also found this link incredible useful and I would recomend it any time I can:

http://www.controlguru.com

So far, I have been working with P=10000, I = 0.5 and validating D = 10. No sure yet if P=10000 is too high or if D is ok on 10, I would run more test to make sure P is sensitive enough and that D is high enough to avoid overshoot without insert oscilations.

The problem I have right now is that oven, when door is open, reduces the air temperature (this is force convection oven) in about 50F/min, but once you close door, it will recover quite fast even if power is 100%, however, as control senses temperature drop, it turns out to 100% and the combined effect of the existing heat on the oven plus the output on 100% when below 50F, will actually build up to much heat and produces a small overshoot. In the other scenario when oven is cold and set point moved to process temp, the PID works perfect, no overshot, no undershot nor oscilations.

 

[xnuke]

I'm glad you've worked out the sampling time issue.

If you're getting different process behavior under two different operating conditions, it's likely that your process is nonlinear. You can use different tuning parameters for the PID at the two different operating points (i.e., gain scheduling) or call two different PIDs tuned differently depending on the operating condition and switch between them (trying for a bumpless transfer in either case). Alternatively, you can decide which is more important, rejecting the disturbance of the door opening or heating up from a cold condition, and tune your PID for that condition to meet the requirements of the process.

xnuke
"Live and act within the limit of your knowledge and keep expanding it to the limit of your life." Ayn Rand, Atlas Shrugged.
Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[controlnovice]

With Temperature control,

first get rid of the I term.... You can do this by making it PD control, or making the I very high (in Min/Repeat...or make it very low for Repeat/Min). Once you get the P and D terms close, then you can slowly bring in the I term.

Unless I'm not thinking correctly, a P term of 10000 pretty much makes it On/Off control, so very little controlability, if any at all.

For the D term, depends on how big the oven is. Probably range from 15-60 seconds.......

______________________________________________________________________________
This is normally the space where people post something insightful.