Help with PI tuning, Please!

I have been fighting with the tuning of this PID loop for a week. Never had this much trouble before, but I am not a PID wiz, so anyones suggestions will be appreciated. PLC is a GE Fanuc Versamax Micro, programmed with Versapro. Application is control of irrigation system pressure via a submersible pump in a tank and a vfd controlled by the analog output of the plc. I am faulting on low pressure (drive not ramping up to meet demand), but I am having trouble finding a happy place where it doesn't just run away and overshoot.

Tuning Parameters block set up as follows:

P-3.50
D-0
I-0
sample period-0
bias-0
Db upper and lower-0
Up clamp-32000
Low Clamp-16000
Error term-SP-PV
Min slew-0
Deriv. action-Error
Output action-Positive

Help!

Scott

Try www.controlguru.com they have an excellent explanation for PID tuning.

schererent said:

I have been fighting with the tuning of this PID loop for a week. Never had this much trouble before, but I am not a PID wiz, so anyones suggestions will be appreciated. PLC is a GE Fanuc Versamax Micro, programmed with Versapro. Application is control of irrigation system pressure via a submersible pump in a tank and a vfd controlled by the analog output of the plc. I am faulting on low pressure (drive not ramping up to meet demand), but I am having trouble finding a happy place where it doesn't just run away and overshoot.

Tuning Parameters block set up as follows:

P-3.50
D-0
I-0
sample period-0
bias-0
Db upper and lower-0
Up clamp-32000
Low Clamp-16000
Error term-SP-PV
Min slew-0
Deriv. action-Error
Output action-Positive

Help!

Scott

Click to expand...

So what does 3.5 mean? Percent output per PSI? Dac Counts Out per error in counts? The numbers mean something. You should be able to calculate what the maximum error is going to be and adjust the proportional gain accordingly. Perhaps your problem is not a tuning problem at all. What if the VFD will not keep up with demand and you get a low pressure alarm even when the pump is on at 100%. What is 100% ? A flow or pressure rate ... what? What is the demand? I have exceeded my question quota.

Hello,

as stated, your problem is slightly vague, however initially it sounds as if your P term is too small, and the system can't react fast enough. Does it only throw a low pressure fault on startup? If so the pump us not able to pressurise the irrigation system before the fault is tripped. You'll need to allow some deadtime or a startup time on the fault trigger.

Anyway I have no idea exactly what your system is like, but I can still point you in the right direction. One of the quick and simple methods I use from time to time on the fly is called the Ziegler-Nichols Method, there are a few places to read up on it, but this one seems quite concise:

http://www.chem.mtu.edu/~tbco/cm416/zn.html

I also did a quick search and this site seems to be comprehensive without going into copious amounts of math or control theory:

http://www.jashaw.com/pid/tutorial/

Alternatively you can use software like Simulink to model your system and tune your PID controller.

Am no PID expert myself but one thing jumps out from your parameters, Sample Period cannot be zero. Sample Period is used throught the PID calculation and zero will seriously screw things up.

Bryan

Jeev said:

I also did a quick search and this site seems to be comprehensive without going into copious amounts of math or control theory:

http://www.jashaw.com/pid/tutorial/

Click to expand...

Quack, Quack, Quack. There is no real math or method that backs up what he says. Anybody can put up a website and claim to be an expert.

www.controlguru.com is much better. The tuning methods that are used there can be proven mathematically. After all what they are trying to do is sell auto tuning software.

ZN is from the dark ages too. Some systems can't or shouldn't be made to oscillate.

One really must understand the nature of the system they are tuning. A irrigation system is much different than a temperature control system in that the system is non-linear. If the VFD controls flow and indirectly controls pressure then the pressure will not increase in a linear way. The open loop gains is the change in pressure over the change in control signal. This information should be known or acquired impirically. Then the controller gain can be adjusted on-the-fly.

A simple irrigation system should be controllable with just a P gain. It would be better if the integrator was used. With out the integrator there will always be some error unless an integrator is used but it possible to calculate what the error will be. Until we know what 3.5 means we can only speculate.

This should be simple.

Thanks

I certainly appreciate all your responses. A few minutes at the controlguru, and I have a whole new understanding of the process. Thanks again. I will post when it is smoothed out and working!

BryanG said:

...Sample Period cannot be zero. Sample Period is used throught the PID calculation and zero will seriously screw things up.
Bryan

Click to expand...

This is 100% correct. Your PID loop will not work correctly with Sample Period value of 0.