PID help advice

Hi,

I'm looking for some advice or opinions on a control loop that I have been asked to look at and possibly modify.

The present system has been there for 10+ years and as far as I am aware it's never successfully worked.

My employer has the current set up which is shown in the attached drawing.

Using raw water, an acid is added which is then circulated round a contact loop, after the loop, an Alkali is added.
The expected end result of the loop is to achieve a specific alkali Ph.

The loop has 2 Ph probes which are selectable via SACDA and 2 dosing pumps.
The acid dosing pump is controlled by a PID loop where the SP is entered in the SCADA and the PV is measured from the selected Ph probe.
The Alkali dosing pump is purely proportional to flow.
In the PID block, the derivative is switched off.

One probe is just after the acid dose and the 2nd is just after the Alkali dose.
My employer wants to be able to control the acid dose by monitoring the Ph probe which is just after the Alkali dose and is also over 10 minutes process time after the acid dose.
I've been told that the system was designed to work this way, even though they have had no successes in running it this way.

My wee head can't process how I can control an acid dose while monitoring the after an Alkali has been added, the Alkali will raise the value by 5.
There is also only one set of PID parameters.

Looking at it, firstly I'd assume that each probe would require its own PID parameters.
I have had a play with it, but can't seem to get the loop to react to change and then recover in an acceptable time scale.

Could someone give me an opinion on whether the system could operate from the Alkali probe with some PID tweaking?

Thanks in advance,

Colin

It's difficult for me to give you solid advice on this specific application with limited knowledge, but what I can tell you is that on loops with very slow response time, I have had better luck programming with plain 'ol logic than with the PLC softwares' PID instructions.

I can give you an example. I have an agricultural customer with very large barns containing up to a few hundred thousand birds. Maintaining a specific temperature inside the barns is very important. There are a lot of fans, evaporative coolers, and heaters. It can take up to an hour to notice a significant change in the feedback loop. It was easier for me to run several tests while changing my variables than to tune a traditional PID loop. It took a while though. In the end, I could predict how many fans and evaporative coolers I would need to turn on based on the outside temperature, humidity, and size of the barn. I could track the slope of the change-in-temperature line, and if it varied from what I expect, I can change the variables as needed.

Unless you're able to make significant changes to the process itself, that 13 minutes of dead time will be hard to overcome with feedback-only control.

A couple process-related questions to ask your employer/customer: (1) is it acceptable to control outgoing pH by adjusting the alkali dose to maintain pH (or is a fixed, proportional-to-flow dose required), or (2) if fixed alkali is required, can that be introduced first, and then final pH is controlled by subsequent acid dose immediately before outgoing pH is measured?

The general idea is to get your control objective measurement (pH probe) closer to the control element (dosing pump) in order to reduce transport delay. Otherwise you will need to use some type of model-based control as suggested by Zlinger to improve performance.

pH control is very non-linear. It is tough to control with PID, as Zlinger suggests. I would consider using floating control, with different steps in the dosage at different pH levels.

I don't see how you can control upstream pH for acid dosing using a probe after adding alkali dose. It seems to me that if you want to lower pH before the loop, which I assume changes the pH, you would need to have that loop independent and the second loop raise the pH as required.

I also don't see how this can be done the way your manager wants. How variable are the flow rates?

I would probably try to come up with some sort of PLC logic based algorithm to calculate the base dosing level (flow rate, etc.), with the feedback being used as a slow correction factor. But I would definitely use the probe closest to the control device.

You may need to use some sort of cascade approach. Use the second probe to correct the setpoint of the first algorithm.

The hardest thing about tuning this kind of system is that you have to wait a looonnnggg time to see how your changes affect the system.