Allen-Bradley PID not reaching setpoint

I have a PIDE set up for controlling a tank level. I've run the autotune on it and it is working fairly well except for one thing. No matter what I do, the level consistently remains 2% below the setpoint! It doesn't matter if I start off above or below the setpoint, the behavior is the same. So when I tell it to keep the tank 60% full, it will stay very level at 58%. While I could just add 2% to whatever the user inputs, this seems like a bad work around.

Any ideas why my PIDE is so consistently 2% shy of target?

Could be proportional droop. Are you using the Integral term?

Tom Jenkins said:

Could be proportional droop. Are you using the Integral term?

Click to expand...

Yes I am

Scaling error?

rupej said:

Scaling error?

Click to expand...

Almost! I was finally able to back track through the code enough to find that the PV wasn't directly the tank level at all. It was extrapolating the tank level based on the discharge flow with an assumed density based on what it believes is in the tank and uses the difference between the setpoint and the actual level to try and "Trim" up the difference.

It's pretty clever, but when the density of the liquid is slightly off the assumed density, then the minor "trim" it adds to the PV isn't enough to get it up to setpoint. The end result was that the tank would control very effectively at 2% below the desired level and the operators would be continually frustrated that the setpoint wasn't being reached.