Flow control using dosing pump and Coriolis flow meter

[MikeyN]

We use dosing pumps that are similar to the Grundfoss DDA. The output is pretty accurate IMO. Instead of a flow meter, we use a flow switch with a .07 GPM trigger on the outlet of the pump to monitor how much chemical we have pumped into a tank. We use the pulses from the pump to our advantage in this setup. Since each pulse is the same amount of chemical each time, we just count the number of pulses to give us total output for one cycle. We have a line tech verify each pumps output once a week with a graduated cylinder, and for about 2 years now, we have never been more than 1 mL off of our target.

 

[Tom Jenkins]

We use dosing pumps that are similar to the Grundfoss DDA. The output is pretty accurate IMO. Instead of a flow meter, we use a flow switch with a .07 GPM trigger on the outlet of the pump to monitor how much chemical we have pumped into a tank. We use the pulses from the pump to our advantage in this setup. Since each pulse is the same amount of chemical each time, we just count the number of pulses to give us total output for one cycle. We have a line tech verify each pumps output once a week with a graduated cylinder, and for about 2 years now, we have never been more than 1 mL off of our target.

I agree. Use a positive displacement dosing pump and skip the flow meter.

 

[mylespetro]

We use dosing pumps that are similar to the Grundfoss DDA. The output is pretty accurate IMO. Instead of a flow meter, we use a flow switch with a .07 GPM trigger on the outlet of the pump to monitor how much chemical we have pumped into a tank. We use the pulses from the pump to our advantage in this setup. Since each pulse is the same amount of chemical each time, we just count the number of pulses to give us total output for one cycle. We have a line tech verify each pumps output once a week with a graduated cylinder, and for about 2 years now, we have never been more than 1 mL off of our target.

That's great to hear. I've typically worked only with Prominent pumps, but in my experience, dosing pumps seem to be right on target. We did do a rough check on this one the other day using a stopwatch while watching the level in a graduated vessel go down and it seemed to be a bit slow, but there was also about 4psi of back pressure on the vessel when it's in operation, which we also had applied during that test. That is probably going to be another hurdle to overcome, however during my tests today, there was no pressure on the vessel.

 

[drbitboy]

...the integral action was 6min/repeat. ...we lowered the integral action to 1min/repeat ...

Not that it matters at this point, but aren't those units repeat/minute?

 

[mylespetro]

Not that it matters at this point, but aren't those units repeat/minute?

Nope, they're definitely minutes/repeat. I didn't grab a picture of the tuning page, but lowering the number definitely increases the integral action, and the math instruction that calculates the integral term based on the value input from the HMI and put into the control word for the PID instruction also says "min/repeat."

I'll attach some of the pictures I have of the trend here.

The name of each picture is pretty descriptive, but I'll expand on them a little bit. Also note that not all of the trends are displayed at the same time scale, note the "Time range" at the top left of the window for each.

"PV noise in manual 50 percent": This is the fluctuation we experienced on the PV with the controller set to manual at 50% controller output (or Control Variable/CV). I'm aware that in this particular trend, it looks like a consistent oscillation, but that's just a coincidence.

"Change to auto": This is the trend after we changed the loop to auto with a setpoint of 200. The extreme noise on the PV at the beginning is what I was referring to in a previous post about the PV being very oscillatory/noisy until the CV hit ~12.5%. After that point it gets relatively smooth. Note, at the start of this trend, the integral term was 3min/repeat, but where the CV (blue/teal line) slope changes at around 1/4 into the trend, I had changed this to 1min/repeat, hence the change in slope of the CV.

"Reaching SP": This is where the loop finally approaches and reaches setpoint (200), it looks a bit smoother here, but again the time scale here is shorter (0.07 hours or 4.2 minutes). This is actually the loop responding to a disturbance, the water (for test purposes) that we were pumping out of a small vessel had run out so the PV dropped until we refilled. The reason all of the trends are so short is that I had to keep exiting the trend screen to silence/acknowledge alarms or change other things and the trend isn't persistent when it's not open.

"Decreased SP": Setpoint changed to 31, which is the first programmed setpoint for the batch process.

"Decreased SP 2": This is where it approximately reaches the new setpoint. Here you can see the oscillations above setpoint I was referring to. At the moment the picture was taken, the PV was 25.2335, but it was clearly bouncing between 30 and approximately 45. Again, this doesn't look too bad, but again note the time scale of the trend - 0.57 hours or 34 minutes, so we're at just under a half hour of actual trending in this photo. I could deal with oscillations like this if it didn't also take so long to get to this point.

EDIT: Not sure why my pictures uploaded rotated, sorry about that

 

[drbitboy]

Huh. Then I would say you can definitely go with more aggressive (lower) values for both PB and Ti.

The first trend is the most interesting: with the pump running at a constant speed, that trend may be defining the process capability, although it could also be process disturbance of course; there is a lot of noise in the measurement, and I don't think it has anything to do with pulsation from the pump (unless the PD pump is doing something like 15 strokes per hour at 50% CV).

Also, the process gain seems to be variable, but that could be from the noise in the PV; is pump speed linear with CV, and does 0% CV mean zero speed on the pump?

What is the velocity in the Coriolis meter? Is there a minimum limit to velocity below which that type of meter gives hinky readings?

Several posts mentioned air in the meter as a problem.

 

[Tom Jenkins]

The impact of backpressure and wear depends on the type of pump. Peristaltic types would be most affected, diaphragm types very little.

An example: https://www.lmipumps.com/en-us/chem...s4azfuLTY2O4JKQrSpaCvvr0yqNeo_uMaArKEEALw_wcB

 

[mylespetro]

The impact of backpressure and wear depends on the type of pump. Peristaltic types would be most affected, diaphragm types very little.

An example: https://www.lmipumps.com/en-us/chem...s4azfuLTY2O4JKQrSpaCvvr0yqNeo_uMaArKEEALw_wcB

Thanks for the info Tom, I'll take a look at your link after work.