Speed control using level switches

[janner_10]

It's hard to program around a ***** design, get an analogue sensor or put up with hunting of sorts.

**** = S h ! T btw

 

[Phrog30]

it's hard to program around a ***** design, get an analogue sensor or put up with hunting of sorts.

**** = s h ! T btw

+1

 

[Phrog30]

Close, but the right thing to do is to take the average of the two speeds at the transitions

Nope, that isn't the right thing to do. The right thing is install the correct field devices. Either continuous level (analog) or two more digital devices, so LoLo, Lo, Hi, HiHi.

 

[Aardwizz]

The right thing is install the correct field devices.

"Right" is often relative.
The best TECHNICAL solution is, as you say, to get the proper sensors to give the proper control.

The best FINANCIAL solution may be to devise a control strategy to compensate for the current design.

There may be a host of factors that might favor using only two level switches over a level sensor, or flow meters on the intake and every outlet from the tank, or whatever else we can come up with. We don't know. The OP didn't say. So to claim that there is only one "right" solution is a bit of hubris on your part.

Not that this site, or the internet in general, usually worries about an excess of hubris.

 

[drbitboy]

Not that this site, or the internet in general, usually worries about an excess of hubris.

>snort<

 

[Phrog30]

The best FINANCIAL solution may be to devise a control strategy to compensate for the current design.

And how many times do people cheap out and then spend more time or waste more resources in the end, when they could do it "right" the first time? From my chair, it happens way more than it should.

 

[drbitboy]

And how many times do people cheap out and then spend more time or waste more resources in the end, when they could do it "right" the first time? From my chair, it happens way more than it should.

Judging by most of the questions on this forum, it is obviously not an insignificant number.


But once the hardware is in place, there can be money to be made compensating for losses in poor designs, and telling the OPs that their system is designed wrong is being like Microsoft:


There was once a pilot that was flying to Seattle, but it was foggy. After a while he flew by a tall building, and could see someone through an office window, so he flew by holding up a sign that said "Where am I." When he flew by again, the person in the office was holding up a sign that said "You're in an airplane." The pilot said "Ah!" then flew on a bit more, turned to the left and landed at the airport. His co-pilot said "how did that help?" The pilot said "The answer was 100% accurate and completely useless, so I realized that was Microsoft Tower, and I knew how to get to the airport from there."

 

[Peter Nachtwey]

I have been watching eating pop corn.
I give Phrog30 an A for knowing that there should be a real level sensor.
I give drbitboy a F for not knowing better and wasting too much time and not seeing the real problem. It should be obvious that the slower the rate of change in the pump speed the longer it will take to "ping-pong" between the two level switches. However, if the rate of change in pump speed isn't faster than the possible rate of change in outflow rate there could be serious over or underflow problems. NO ONE ASKED ABOUT THE HIGHEST RATE OF CHANGE IN THE OUTFLOW! That determines how fast the pump speed must change if the rate of change is high the level will "ping-pong" up and down at much shorter intervals with the outflow rate is not changing.



I like ping pong.

 

[drbitboy]

I give drbitboy a F

Ha ha, you waste time how you want; I'll waste time how I want.


These posts are puzzles to me; I am not in it to show how I would have designed it better. If I actually needed to do that, I would ask my brother before I asked here.

...NO ONE ASKED ABOUT THE HIGHEST RATE OF CHANGE IN THE OUTFLOW! That determines how fast the pump speed must change if the rate of change is high the level will "ping-pong" up and down at much shorter intervals with the outflow rate is not changing.

So @PeterN's all-caps CLAIM is that no one picked up that that was issue, huh? Really? Then what is this:

... It will take some tuning to set the [delta flow per second] to control the excursions, but that is expected.

?

Tag, you're it.

 

[drbitboy]

Close, but the right thing to do is [blah blah blah]

Whoops, sorry, Aardwiz, I misread your approach, and actually you and I are doing the same thing, so yours would work; it might take longer to get a number and settle, but it's the exact same math.

 

[CENTER]

I don't understand the big deal. I think keeping it simple is smart. I have been using this set up for years to pump out water trays. Using 2 cheap floats and a drive. Bottom float turns the pump on set your frequency so the level remains the same and when the next float goes high bump it up to 60 HZ. The flow remains constant but there are other small lets say like laundry tray pumps that will up the flow momentarily. No big deal cheap and efficient.

 

[drbitboy]

It should be obvious that the slower the rate of change in the pump speed the longer it will take to "ping-pong" between the two level switches.

Bzzt, thanks for playing.


The flowrate at which it enters a level excursion will be the same flowrate at which it will exit the excursion*, regardless of the ramp rate; the only thing the ramp rate affects is the duration and extent of the excursion.



* standard caveats/assumptions: constant (mean) process flow; insignificant level hysteresis.

 

[Peter Nachtwey]

@drbitboy, tuning?
What do you tune? The pump speed must be able to change as fast as the flow does.

 

[daba]

I really don't see the problem with hunting between the limit switches.

It's not like it is relay controlled, which will impose a "lifetime", it is a VFD, which will be quite happy to change speed regularly, or even start/stop ....

All you have to do is ensure that the tank will "fill" at the higher of the speed set-points with maximum outflow. The minimum speed may just as well be zero and the pump is therefore on/off. The level switches could even be moved closer together nearer the vertical centre of the tank, which will give a more consistent "level" if that is important to the outflows....

The biggest issue I see is that everything relies on a single level switch at each end, i.e. no Low-Low and no High-High level switches. One single failure of a level switch and you have a situation that you have no warning of, and cannot take appropriate actions if that situation arises.

 

[drbitboy]

@drbitboy, tuning?
What do you tune? The pump speed must be able to change as fast as the flow does.

Actually faster, but yes, that is the essence of the tuning i.e. picking the ramp rate.


OP never came back, I wonder if this is a homework exercise.


If it is indeed mis-designed and there are only discrete level sensors, then it does have a financial benefit: the integral of flowrate throughput (process out or controlled in) over long times is constant because it is equal to the integral of uncontrolled process flow, but power (and cost of power) is flowrate cubed, so bang-bang control will lose money compared to successful control that matches pump flow to process flow.


However, while the system will work for simple and slow process flow perturbations e.g. infrequent step changes, it could be unstable for certain (analogous to harmonics in analog feedback control) process behavior; the same could be true for feedback (e.g. PID) control of an analog level PV, but I suspect this control using discrete feedback is more finicky.