Initial PID Kc setting from bump test using Logix500

Hello,

I've been tinkering with PIDs for a while now and have a pretty good understanding of all of the terms. I typically set my initial (first try) proportional gain using the "Raise it until the output oscillates and then lower it until it doesn't anymore" method. That's been pretty effective but can take an incredibly long time, especially when dealing with heating and cooling applications.

I'd like to learn how to get an initial (ballpark) Kc value from doing a bump test. I've read the article on controlguru.com about how to calculate this from dividing the change in the process variable (%) by the change in output (%) but it's still a bit unclear. Does anybody have any rules of thumbs for determining an initial Kc value from the bump test?

Also, controlguru.com suggests leaving the reset rate equal to the process time. Have you folks found that to be a good initial setting for reset?

Thanks!

Actually the controller gain is percent output per degree error and the system gain is degrees per % control output. What you must remember is you must subtract off the ambient temperature.

I wrote one of the articles for the controlguru. Here is a link to the original material.
http://deltamotion.com/peter/Mathcad/FOPDT/Mathcad - FOPDT.pdf
Note the dead time and the ambient temperature variable PVss.
The first 2 pages are the proof that the control guru formulas are correct. I show how the formulas for Kc and ti are calculated.

I also wrote some autotuning software. One of the examples is actual data from the controlguru site about the heat exchanger.
http://deltamotion.com/peter/Scilab/AutoTune/heatexchanger.gif

Thanks Peter. That's a great article- I've read it multiple times to try to absorb it but I'm coming up a bit short on the formula to calculate Kc. Specifically, I'm not sure about some of the variables.

τp is the time constant, right? Calculated by the time it takes the PV to make 63% of the change to steady state after the bump test?

θp is the dead time? The amount of time it takes to notice a change after starting the bump test?

τc is the greater of either 0.1*τp and 0.8*θp?

That's the derivation for the "Lambda Tuning" method, for anyone interested in further searching.

You are right.
The derivation for Kc is all there.
If you can model your temperature using two time constants ( SOPDT ) see this
http://deltamotion.com/peter/Mathcad/SOPDT/Mathcad - SOPDT1.pdf
It is hard to accurately determine the time constants for a SOPDT system but know this....
There are no FOPDT systems in reality.

Thank you for all you do, Peter.