PID tuning help

[ChrisP]

We have a system that we are programming for a controlled paint booth environment. We are controlling temperature and humidity in the booth at various setpoints depending on the current function in the booth. We are using RSLogix 5000. There is a cooling section, reheat section, and humidifier section in the process.
The problem: We are controlling the chilled water valve to get the temperature down to a predefined point to control humidity. There is a considerable amount of overshoot, and the system won't stabilize like we'd like it to. I think one of the problems is that there is a considerable dead time or lag between the time when we make a valve change and when the temperature actually starts to change and then when the temp starts to change, it's inertia continues for a while after the valve is modified. I'd like to see the valve to start to change when temp is approaching the setpoint, not waiting until it has past the setpoint.
We don't work with this type of system very often and haven't had much experience with tuning PID's. Also, it isn't feasible for us to purchase RSTune or other software, since we are only doing this one time for the setup and then will be turning the system over to the client for maintenance. I'd appreciate any help that you guys could give.

 

[Peter Nachtwey]

See the heat exchanger problem on the control guru website

www.controlguru.com

 

[Ron Beaufort]

welcome to the forum ...

the best way to get DETAILED help with this is to ZIP your project file (ACD) and post it (attach it) here - in this thread ...

 

[TridentAutomation]

have you been able to get the loop to start permanent osillation? because that is the first step. in most temperature loops the trend can take a very long time but once you find the time constant that you are working with then i can help you tune it further.

 

[kamenges]

I would tend to agree with Peter. There are more accurate and time efficient methods of system identification than driving the system to oscillation. However, that doesn't change the fact that the first step is system identification. Once you know the system gain(s), time constant(s) and deadtime, determining reasonable control loop settings becomes a much easier proposition.

Keith

 

[ChrisP]

Thanks for the link to the controlguru info. I am going to read through this info and try to implement what it says. If I continue to have difficulties, then I will post a zip of the acd file. From what I've read thus far, I can see that I have some measuring to do on the system after making a change from a stable state. I've still got more of it to read, but wanted to let you guys know that I've at least started on the resolution. Thanks for all of your help so far.

 

[Ron Beaufort]

but what if the setup actually IS incorrect? ...

Greetings Keith ...

the first step is system identification.

I highly respect your opinion - but I've got to disagree with you on this one ...

personally I've lost track of how many times the forum has tried unsuccessfully to help "tune" a PID - only to find out eventually that the original "timing" and/or the "triggering" and/or the "scaling" were out of whack ...

all of the "tuning calculations" being considered are fully dependent on a proper setup of the PID ... I respectfully suggest that the setup should be checked first ...

really we're only talking about two or three minutes of analyzing the program to check this out and confirm the setup ... but if the nuts-and-bolts setup is actually incorrect, it could easily take hours (if not days) of calculation and experimentation with the tuning values to reveal the underlying problem ...

consider this quote from the original poster:

We don't work with this type of system very often

that (in and of itself) is enough to make me mildly suspicious of whether the PID has been programmed correctly or not ...

party on ...