RS5000 PID Question

I'm trying to control a slow process where we only get an updated PV every 5-10 minutes. Right now we do a pseudo-automatic control that works ok, but could be a lot better. I was wondering if using a PID loop with a very slow update time would work.

Has anyone ever setup a PID loop in AB Compactlogix with that slow of an update time? Would triggering the PID to update right after a new sample is taken be a better approach?

Old No. 7 said:

I'm trying to control a slow process where we only get an updated PV every 5-10 minutes. Right now we do a pseudo-automatic control that works ok, but could be a lot better. I was wondering if using a PID loop with a very slow update time would work.

Has anyone ever setup a PID loop in AB Compactlogix with that slow of an update time? Would triggering the PID to update right after a new sample is taken be a better approach?

Click to expand...

Well I think that will start quite a debate !!!

In the meantime, a description of the process would help in determining if a PID or a "roll-your-own" control algorithm would work best....

How slow is this process? I am going to make a couple of assumptions here with the hope that you can either confirm or deny them.

I am going to assume that the process is predominantly a first order process. I will further assume that the process first order time constant is on the order of 30-60 minutes, meaning it will take the process 3-5 hours to settle out after a step change in output. Does any of this ring true?

Keith

What's keeping you from getting feedback data at a quicker rate?

A little more info...

We're controlling a process gas mixture. When a change is made to the mixture, it probably takes a couple minutes until you can see the effect of the change.

The bigger limitation is the gas sensors. They can only take a reading every so often (5-10 minutes) so it seems like a typical PID block wouldn't work very well if the PV is flat for 10 minutes, then gets a instant step change after the sample is taken, then flat for 10 more minutes, etc.

This seems like a good application for floating control.

Old No. 7 said:

..... it probably takes a couple minutes until you can see the effect of the change.....

....the gas sensors can only take a reading every so often (5-10 minutes)....

Click to expand...

what's the limitation - the delay before the gas mixture can be measured, or the measurement rate of the sensors ?

is this a potential "roll-up" in time, meaning that the control algorithm sees a random time delay of the asynchronous effect and measurement ?

can the gas sensors measure faster, or is it a sample time they need to analyse the gas ?

can the analysers be "reset" to synchronise with a CV output change ? this could minimise or stabilise the CV -> PV delay, which might make it easier to control.

In trying to appreciate a complex loop control, I often think about what a person would do faced with the same decisions....

Things I consider -

1. If the PV won't change for 5 minutes after I adjust a valve, why would I look at it every 100mS....

2. Can I use historical data to make an adjustment....

3. What stops me from "over-cooking" my response, sort of leads back to 2. A person would have gained "experience".

There are always challenges on some closed-loop controls, and sometimes PID control just can't be made to work efficiently.

I'm interested to see Tom expand on his "floating control"

the PID block should execute only when the PV updates and then calculate a new CV. I prefer to write my own PID blocks so I have a few that are tailored for each application