pH Control

I need a some help on a process that I'm currently working on which involves the control of pH. This is my first time working with pH and I'm pretty concerned about how non-linear it is. I'm using a Rosemount 5081 that is connected to a Rosemount 3900 pH sensor. The 5081 is outputting a 4-20mA signal back to my control logix chassis. The 4-20mA signal is going to a 1756-IF8 and the controller is a 1756-L71 with 2Mbytes memory.

The pH sensor is located on the backside of a tank that is being mixed. There are two metering pumps that will pump either acid or base into the process before it goes into the tank for mixing. The customer has a pretty wide range for what is acceptable on pH, 5.5-9.5.

I going down the path of using a PIDE block but I'm not sure how to implement everything. Right now I'm trying to set something up so when the PV rises above 7 the acid pump kicks on and runs, which would be direct acting (E = PV-SP) and visa versa for the base pump, reverse acting (E = SP-PV).

Am I even on the right path? Do I need to take the titration curves into account? Can I do this with one PIDE block? If I'm just dead wrong let me know, any suggestion would be greatly appreciated. Thanks!

Actually, I wouldn't use PID for this, although many people do.

I would suggest using a comparison, and if the pH is high turn on the acid pump for an adjustable time, have an adjustable off time for the mixing to be effective, and then repeat as necessary. This is more likely to be stable and the tuning would be simpler for the operators. For low pH use a similar system for the caustic pump.

Here's an interesting article on linearizing pH curves.

http://www.controlglobal.com/articles/2005/509.html

I wrote a program for a similar process and ended up going with the same scheme that Tom suggested, pH is not fun. With that wide of a range you don't need to get very fancy.

Depending on alot of factors Ph is a difficult control to do in Logix, depending on the process/ agitators, tank size a small dose can have a large or small effect and you don't want to be chasing your tail all day long. It takes alot of looking into and understanding the process, sometimes can be trial and error which depending on the process might not be possible. In a few difficult cases I had in the past I did averaging for the control input before sending it into the PID. Lucky for you, you have a Rosemount Transmitter, so with a little setup you can do this averaging in there using a communicator so you eliminate the need to do addiditonal logic in the PLC. Look at the change rate of the product versus input as well as where you are sensing, proper placement of your transmitter is critical in this type of control.

In Ph in the past I have used feed forward feed reverse to limit my PID gains to use bigger swings when way out and tighter controls to maintain when close without swings. In one process I did PH +/- .3 was significant to the process which made controlling Ph an essential thing which made me go to alternate configurations for PID gains to manage it.

Good Luck, that one does sometimes present a challenge depending on conditions and process!

Use a commercial chemical controller - much easier. And then there is calibration of the sensor etcetera - built in to a commercial controller.

I've done quite a few water treatment plants over the years and last year had ten filter presses for a cement factory which had been running on manual control for 5(!) years since the plant swapped from S5 to S7 and nobody could get the control system working (similar problem to pH in that you're using a pump to fill the filter press with a slurry. When the press is full, continuing to pump starts to force the water contained in the slurry out of the filter. When you get to the point where nearly all the water has been pressed out the pressure in the filter goes up like a rocket).

What I do is to set two interim set-points one or two per cent before the actual target SP. I then ramp the delivery pump up to full whack and keep it there until I hit the first (more distant) SP. I then reset the ramp to zero and start it again until it reaches the second SP - if you've got the ramp rate and the distance bewtween the first and second SPs correct you won't reach maximum on the second run. When the second SP is reached I once again stop the ramp then switch over to conventional PI control.

How far away the the two interim SPs need to be depends on the size of the tank, how big your pump or screw is (delivery in kg/min etc.) and how much the incoming medium causes the pH value to stray from the desired SP.

Using this system I've had been able to hold +-0.1 pH even on some pretty horrible mixtures (cooking oil plants come to mind!).

Chemical Dose Control, pH Control, ORP Control etc.

For Future Reference.
Attached is a long proven set of functions for pH, ORP, Chlorine, and any other chemical dosing control.
The guts of chem dosing is:
Rule-1: You MUST control the DOSE RATE (mg/L) of chemical NOT the dose flow rate (L/h). See attached function DOSE_CALC giving the dose flow rate calc from dose rate. NOTE: ALL controller outputs are DOSE RATE (mg/L), as are the BASE_DOSE values in (mg/L).
Rule-2: As the systems are generally very NON-Linear you MUST use a base dose rate as a feed forward.
See attached PID_TRIM. What is written is to attach to a ControLogix 5000 PIDE. You will need to work out the pins if using another PLC.
The base dose rate is established empirically in the plant. If as in this example there is to be acid & caustic dosing, then you need to have 2 separate controllers, each with their own set of parameters. If both acid and caustic, then each needs to have an individual set point and there needs to be start dosing and stop dosing set points established for each of the acid and caustic, such that ONLY one can dose at a time. Generally the target for acid dosing will be say 7.5~8.5, while for caustic 5.5~6.5. So caustic starts dosing at say 5.8 and doses to set point 6.7. If pH rises > 7.2 then stop caustic.
Acid starts dosing at say 8.2 and doses to set point 8.0. If pH falls < 7.0 then stop acid.
Note that the dissolved solids content of the media being dosed is critical as to the base dose required. Where this varies widely, you will require a variable BASE_DOSE value. Perhaps in the normal stream a BASE_DOSE for acid is 2mg/L and 1mg/L for caustic. But if the dissolved solids content rises enormously, then the BASE_DOSE may need to rise significantly.
You will determine this at commissioning.
WIth the PID_TRIM controller, where the controller output is limited to a value of BASE_DOSE +/-Trim%, the controller output can be limited and thus be less susceptible to instability.
PID_TRIM incorporates a start-up delay as the pH (ORP, Cl) analyser requires time to register the effect of the dose (Transport lag + sample lag).
I have used this successfully for many years in the Water & Waste Water industry and it works well. You can control pH, you just have to understand the process! Note I also use this to control CO2 dosing for pH control. THe CO2 dose calculation is obviously a little different as we control kg/h mass flow rate, although the same principle applies with the controller OUTPUT in mg/L, converted by multiplying by the process flow rate to arrive at kg/h.
Note that this Trim controller can be used for any process and through feed forward control provides extremely stable process control.

you have 2 setpoints, so the acid starts when the PH is very close to 9.5 and stops when it is 9.4
the base pump same but at 5.5
See it as 2 totally different installations.

but the PB as small as possible until the output gets unstable, then control the P and I for stable operation.

Maybe someone might find this interesting:
http://www.omega.com/green/pdf/PHCONTROL_REF.pdf

Kalle