Suggestions for PID control

[mhump711]

I am setting up control for Hypochlorite dosing. The easy part is the dosing calculation for flow pacing but I would also like to setup trimming with a PID controller. We have an analyzer to measure the PPM of chlorine and I want to use this to lower/raise the dose as needed. The hard part is the analyzers we use update the PV every 5 minutes.

My question is what is the best way to setup this trimming control for a PV that doesn't update but every 5 minutes?

Options I know about:
1) I recently learned about IMC in RSLogix5000 but I know nothing about it.
2) Really conservative tune for PID
3) Add/Subtract every time the PV changes by a predetermined amount.

Thanks in advance. Sorry If it takes me a minute to respond...going on vacation till Tuesday.

 

[Lare]

I would forget PID for this.

Add / subtract if not on setpoint window is easier.



If sampling interval is at least 5 mins, then PID should be tuned relately slow and PID output is more and less only add / subtract some part.
Slow PID and own sub/add calculation with timer is allmost same. (PID is tuned more like P-term only controller).




Base dose calculation based on flow works usually well.

 

[OkiePC]

It has been my experience that you want to qualify the analyzer result before you apply it to your flow pacing slope. Don't use a PID, just roll your own logic.

Add a real number variable to use as a multiplier and start it off at 1.0 and multiply it by the result of your flow pace math.

When you receive fresh analyzer data, make sure it is legit (limit test it) and then calculate the error and multiply that error by another gain variable that can be summed with your multiplier. This second gain value is just to allow you to adjust how much of the signal error gets applied to your slope multiplier.

Then clamp the multiplier between reasonable limits. It is a good idea to display that multiplier somewhere and indicate when it is at one of the limits.

I have done this on two jobs and if the analyzer goes wonky because of a lack of maintenance or a sensor problem, you can end up way out of whack if you don't anticipate issues with the analyzer feedback.

 

[MaxK]

I am setting up control for Hypochlorite dosing.

Do you really think that every forum member is working on the Hypochlorite dilution system? And the system is identical to yours?
Without data (curves or something) flow rate, supplied Hypochlorite, measured ppm of chlorine, the discussion is pointless.

 

[parky]

I agree with Maxk, for example is this in-line dosing, batch tank dosing etc.
how long is a piece of string comes to mind.

 

[mhump711]

It has been my experience that you want to qualify the analyzer result before you apply it to your flow pacing slope. Don't use a PID, just roll your own logic.

Add a real number variable to use as a multiplier and start it off at 1.0 and multiply it by the result of your flow pace math.

When you receive fresh analyzer data, make sure it is legit (limit test it) and then calculate the error and multiply that error by another gain variable that can be summed with your multiplier. This second gain value is just to allow you to adjust how much of the signal error gets applied to your slope multiplier.

Then clamp the multiplier between reasonable limits. It is a good idea to display that multiplier somewhere and indicate when it is at one of the limits.

I have done this on two jobs and if the analyzer goes wonky because of a lack of maintenance or a sensor problem, you can end up way out of whack if you don't anticipate issues with the analyzer feedback.

This is an interesting idea. Limit testing and clamping I can see is a must.

Thanks everyone for the help

 

[MaxK]

It has been my experience that you want to qualify the analyzer result before you apply it to your flow pacing slope. ...

I think I get your idea. Let me state clearly what you mentioned:
1. The system is quite sensitive to faults and inaccuracies of the flow meter and analyzer.
2. The system itself is unstable and requires thoughtful tuning and limiter settings.

Do you agree?

Wouldn't it be safer to analyze a sample of flow meter and analyzer data and derive a formula or table of "compensating effects" (correction factors)?

 

[OkiePC]

It's not that it is unstable, but can, on occasions, give you bad info. You don't want to run away with bad info and let the dosing become ridiculous. Your calculated results should be clamped within reasonable limits, so that even if you have undetected but inaccurate analyzer data, you only adjust the slope of your scaling to a degree that limits the true dosage to within a range that is still safe.

A lot of these systems run for hours or even a full day unattended before a plant operator comes along to make a check with a handheld analyzer at which point anything that has gone wrong can be corrected.

 

[mhump711]

It's not that it is unstable, but can, on occasions, give you bad info. You don't want to run away with bad info and let the dosing become ridiculous. Your calculated results should be clamped within reasonable limits, so that even if you have undetected but inaccurate analyzer data, you only adjust the slope of your scaling to a degree that limits the true dosage to within a range that is still safe.

A lot of these systems run for hours or even a full day unattended before a plant operator comes along to make a check with a handheld analyzer at which point anything that has gone wrong can be corrected.

This is ideal. We want the the main calculation for the dose to do the work but within reason we also want to add/subtract as necessary. Sometimes we dose in more than one area so the need for trim control is necessary to the final injection would be helpful.

I have something working like this in another area but needed ideas because the analyzer gives me a number every 5 minutes. The analyzer is about 50-100ft (not good at gauging distance) after the injection point.

Sorry for the vagueness it has been requested to provide as little info as possible.

Again thank you everyone for the help