Intergral Value in P&I Loop

I was just recently asked to figure out why the level control in one of our Evaporators was so erratic. I Trended the P&I Loop via PARC View for 24 hours to see what was going on. The set point is for 60 percent level and the Process variable seems to bounce up to 70% then to 60% then down to 50% then back to 60% and follows the same pattern. Over a period of time this oscillation seems to be quite equal in time or "constant oscillations". Each oscillation from peak (70%) to peak is about 2 minutes and 45 seconds on average. Right now the Proportional gain is set to .3 and the integral value or "reset time" TI is set to 10s. Would I not want to increase the integral time say to 20 seconds to slow the controller down to eliminate the oscillations? I'm new to PID tuning but I've also read a lot about it, but not coming from a mathematical background I get easily confused with the formula's involved. Please give some insight to my issue!!

Thanks, TEX

The value of your integral term can have different meanings depending on how you have your PID equation set independent or dependent.

Dependent = Minutes per repeat

Independent = repeats per sec.

Post your program if you are not sure. Tell us which PID is in question.

Some useful links for you..

http://www.ronbeaufort.com/WhatIsP.pdf

http://www.ronbeaufort.com/WhatIsI.pdf

http://www.ronbeaufort.com/WhatIsD.pdf

http://www.ronbeaufort.com/WhyNotPOnly.pdf

what happen on 3 or 6 months ago? history.

you need to trace the history,maintenance record?

are this happen only when running certain product?

if you change PID today ,next 3 month you change again,next 6 month change again?

during earlier start up the system them setting with best PID.

where the instrument PID record?

Right now the Proportional gain is set to .3 and the integral value or "reset time" TI is set to 10s.

Click to expand...

0.3 doesn't mean anything. What are the units? 0.3% per inch? 0.3volts per centimeter.
The integrator time constant seems to be too short. Try setting the integrator time constant to 1 minute or disabling the integrator. The P only control should not oscillate in normal level control. The level will vary slowly as the load changes. This will work unless there is some sort of dead time we don't know about.

One of the things I don't like about Allen Bradley PIDs is that the gain is always error in counts and output in counts so the PID is unitless. That is OK except know one tells use how the counts are scaled so the PID number are meaningless.

For instance a SCP block may scale so there are 100 counts per inch and the output may be 300 counts per volt.

Mickey said:

The value of your integral term can have different meanings depending on how you have your PID equation set independent or dependent.

Dependent = Minutes per repeat

Independent = repeats per sec.

Post your program if you are not sure. Tell us which PID is in question.

Click to expand...

What gets 'repeated'? Where are there repeat units cancelled out. For instance the ISA integrator has time constants for the integrator term. Motion controllers use a gain which is output/accumulated_error. Repeat is an old term that really doesn't apply to PIDs implement in PLCs today and it obscures understand what is really happening.

Ok I attached the program and the trend I used. The PD block is PD80:8

Apparently an engineer had trended this for 6-8 months and said its always been doing this. I haven't seen that trend, only the 48 hour one I've had going and its still oscillating between 50% and 70%. Maybe the PID was never tuned properly. You'll notice there's 8 body level PID controls. Each body is a certain stage in the evaps. What the evaps do is evaporate the water out of our liquor that's created from cooking our pulp in a digester with acid. After the pulp comes out of the digester it get washed with water to get rid of any red liquor on the pulp. We burn the red liquor in our recovery boiler, but because the red liquor has so much water in it we cant burn it until our evaporators get rid of all the water. #8 body is the first stage in the evaps and gets the highest amount of water to solids (liquor) ratio. As it goes through the stages (8-1) the water is taken out via Steam injection until what's left is a pure red liquor we can burn in our oil guns into our boiler. The PID is set up for dependant, which I'm not sure what it means. I need to take a damn class on PID control and loop tuning.

p.s. I know your going to ask why there's a huge swing in that trend, I'm trying to figure out what that is, they could be flushing the evaps

Repeat is an old term that really doesn't apply to PIDs implement in PLCs today

Click to expand...

That's because I'm old and thats what PLC5's instruction set calls it (The PLC5 is also old)

Your PID instruction is set for "Dependent PID Equation"

Did you read thru the above links?