Water System PID Control with a SLC 5/05

Good morning, all. I suppose a little bio is in order, to start: I recently began this new job as the plant's PLC Controls Technician (I know... fancy title...). Honestly, this is the first time that I have ever been paid for this type of work, although I have been utilizing PLC's for most of my career. I am a Master Electrician by trade and schooling, and have mainly self-taught the PLC world to myself. Yeah, I know that can be pretty scary, but I don't try to do anything I'm not comfortable with, and we have an outside contractor that we bring in for any high-level stuff. Or I search here for great answers and explanations. The plan is to get me up to speed over the next couple of years while weaning off utilizing the outside guys. But I digress... Here's my current problem:

Our process water system is supplied by two 90 gpm pumps driven by 7-1/2 hp motors controlled by AB PowerFlex 70's. Generally, these pumps take turns supplying the water to the system, unless our pressure drops to below 40 psi, in which case they both kick in to bring us back up. Under normal conditions, however, the analog signal is sent to the drive from a PID. Everything seems to be set up Ok and working, although I think that the system could use a good tuning. Here are the PID's parameters (sorry I don't have screen shots at the moment):

Control Block: N20:53; Process Variable: N20:14; and Control Variable: N20:20.

Kc= 2.0
Ti = 0.1
Td= 0.00
Loop Update = 0.10
Control Mode = E+SP-PV
PID Control = Auto
Time Mode = Timed
Limit Output CV = 0
Deadband = 0

Setpoint = 4915
Smax = 16383
Smin = 0
PV = 4915

CV% = 64
Output Max = 100
Output Min = 50
SE = 0

The only flags that are enabled are: Timed Mode (TM), Output Limiting (OL), and PID Enable (EN).

The 4-20 mA pressure transducer that is currently reading the output from the pumps is about 15 feet away. It will generally hold our Set Point of 60 psi, although you can watch and see a definite drop in the pressure at times under load. The system will eventually be a looped system, but for right now, it is ended approximately 350 feet from the feed.

We have a machine that, it is theorized, is running low on water. It is located in the opposite back corner of the plant from the feed, approximately 250+ feet of pipe away. I was instructed to add a new transducer to the line near where the mid-point will be, ~300+/- feet away. The engineers are theorizing that controlling the pressure from that point might be a more accurate spot in relation to what the system is doing on the back end.

Things that are known:
- If the system pressure gets to high (+95 psi), the pressure relief will blow off.
- As one would imagine, simply plugging the new transducer in for the PID's CV doesn't work: it causes the system to oscillate from ~45 psi to ~75 psi.
- Because of how the system is reading the new transducer, (over Ethernet through a RIO), it seems to have a better response time in the processor than the old, which is read directly off an analog input card on the SLC. (There actually appears to be points where that card "locks up" and freezes the PV, but I am not sure as to why. I believe this is most of our problem with the system, as it allows a drop in pressure. When it does start reading correctly again, it is usually bordering on kicking in the second pump.)
- The PLC enclosure looks like a family of large gerbils sleeps there. Too many hands have been in there over the years, and really the entire thing needs to be re-done. It is in the plan, but not for a long while, and we still need to run.
- I will not be able to run any step-tests on the system for a few weeks, as we will be running production.
- All of the scaling and set-up in the program seems to be ok, from what I have read on here and in the AB manual.

Things I do not know:
- What the hell I'm doing. Actually, where to start on trying to get things tuned in. I know that blindly stabbing at settings is not the correct way to do things, but it is all I have to go on for the moment.
- Whether this is the correct road to be traveling down or not. Where is the best spot to be controlling a system such as this?
- Whether all of the settings are where they should be. Are things set up correctly? Is there a better way in which to control it all? I'm just not sure at this point.
- Shouldn't there be some Differential setting? I realize that it is not always needed, but with a system such as this and the rapid changes that it sees, I would think that it would be a necessity... But what do I know?

I was pretty sure that I was going to need help in getting this thing tuned in from the start (our guy was not available this week), so I have been reading and gathering information prior to posting here. Hopefully, I have enough so that some guidance can be provided, although I will certainly do my best to gather anything that might be asked of me. My plan at the moment is to try some different parameters in the PID and see what happens. Anyone with any thoughts, ideas, suggestions, whatever - please don't hesitate to post.

Thank you all for your time in advance (and sorry about the length).

- Jeff

Having the feedback pressure 300+ feet away from the pump is not going to make control easier. It will allow you to balance the setpoint at that spot, but the loop pressure drop will still be the same overall - its just that your control will be harder to tune.

You are essentially asking how to tune the loop. For the SLC, I recommend a 3-part series published here by Ron Beaufort. In part 1, Ron doesn't get involved until at least on the 2nd scroll page of my screens so don't give up. If you read these missives and seem lost, well welcome the club. Experience helps a lot.

And here is Ron's web site on the subject. ( click on sample lessons, scroll down)

http://www.ronbeaufort.com/main_frame_page.htm

You have to define the problem better. What is the goal?

1 Control pressure at the pump? If so then what about the pressure losses between the pump and the machine 300+ feet away.

2 Control pressure at the machine? This will be difficult from 300 ft away.

Narlin said:

Having the feedback pressure 300+ feet away from the pump is not going to make control easier.

Click to expand...

Yes, the speed of sound/pressure waves in water is about 5 ft/millisecond so there will be a 60 millisecond dead time between the time the pump does something and the time it is detected.

3 Control flow to the machine 300+ ft away?

Does this pump only pump to the one machine 300+ ft away? If so can the -pump be moved closer to the machine?

----------------------------------------------------------------

If the pressure sensor is 300+ ft away from the pump the pressure will be difficult to control due to the dead time and how fast the pressure will increase and decrease. To stop the rapid swings from 45 to 75 psi you need to add an accumulator by the pressure sensor. The accumulator will slow down the pressure swings. The larger the the accumulator the slower the pressure will change so the pump can respond even with the 60 millisecond dead time.

An accumulator is like a capacitor that filters out voltage ( pressure ) ripples.

+1 on the accumulator. When I was working in water/wastewater, we would always run into projects where the civil engineer moved the pumps at the last minute to the opposite corner of the building, and the machine that was most sensitive to pressure fluctuations was now a couple hundred feet away. Add on a pressurized water tank to smooth out the fluctuations, and things start to run as expected.

I was at one water plant that had the pressure switch for the plant process water on the same line that ran to the bathroom. Flush the toilet, and the plant shut down.

Have you used any accumulators in FDA water systems? They would be possible but maybe pricey. We nearly always have a loop with the pump near to the main water tank followed by many drops. The goal is to assure that return water is at least XX psi in order to keep the sanitary spray ball working. There is a transducer at the pump supply and at the tank return. When users use the water, we obviously need to ramp up the pump to maintain the sprayball pressure. For a control scheme, I usually put the PID feedback a few feet from the pump, making it easier to tune. Then either automated, or manual, we open worst case ports and see what happens to the return pressure. The loop setpoint at the supply is then raised to accommodate this worst case.
.
I have an accumulator on a home water system (non-sanitary) and it works really well to smooth out pressure waves that used to cause pipe knocking.

Thanks, guys - as usual, that's some really great feedback.

As for Ron's stuff, I'm not sure that anyone could read through all that he has posted in a mere week, but I tried - I've read through most, including part one of the 3-parter. I just needed to start actually having some results for engineering...

The pumps supply water to the entire plant operations. Granted, not every machine utilizes this water (we also have an RO system supply, as well as a re-circulation/filter system for grinding, but it does supply a large amount. In fact, one of the next steps the head engineer and I plan on taking is to monitor the system machine by machine to try to see where the large dump is happening (I don't believe it's the bathroom, but it acts similar...). This will most likely be an on-going problem that is continuously revisited, due to system design (and it's subsequent *******ization over the years), but I'm confident that we can at least make the system better.

I had not thought of utilizing an accumulator at the other transducer, but that makes a lot of sense.

I did manage to do a brief test (twice) earlier this morning, with results about what you would expect. With the Kc = 1.0, the Ti = 0.5, and the Td = 0.07, I had better control out of it (a delta of about 9 psi, from 54 to 63 psi), but I had a screwy thing happen both times: a few minutes into the test, pressure fell off for no discernible reason, all the way down to 14 psi. I could not ascertain whether it was a control problem or a symptom of the plant running, but it happened both times.

I will continue to monitor and test and see if I can't get it better, one way or the other, and will let you guys know. Any other points that anyone might think of to have me ponder, I welcome it.

Thanks again.

- Jeff

Huh. Would not have thought that it would edit, "ba$tardization" ...