High Precision Water Level Switch

Depends on the flow characteristics desired and the environment...

https://www.flomatic.com/valves/automatic-control-valves/float-valve/index.asp?valveid=27

Yes they are control valves and are operated to control fluid levels.

I suspect that there is only one 600mm gate valve which will be positioned downstream of the flow meter so it does not create turbulence in the flow meter inlet pipe. It will be opened as quickly as possible to establish steady state flow in the minimum of time and will be closed slowly or after the tank has emptied to prevent pipe hammer. At this flow rate the gate disc will try to snatch fully closed when a large proportion of the surface area of the upstream edge is exposed to the stream at about ¾ closed. If the valve uses a hydraulic actuator (which can be opened much faster than with an electric motor) an 'over centre' valve is needed in the hydraulic control circuit to guarantee controlled closing without pipe hammer.

And I want a winning lottery ticket for my birthday, but there are too many unknown variables.

What is the diameter of the tank what will happen when the level Is sensed?

The parameters which I thought necessary, I narrated. But if other parameters are needed...

tank diameter is 2500 mm

When the trigger (sensing) point is passed, PLC will do some other operation. I will fit two such sensors at different height on the tank. The water will pass through some flow meter which will give pulses per unit volume passes through it.

When water passes over first sensor, PLC will register the pulse count of first sensor. when water passes over the second sensor, it will again register the pulse count of the flow meter. Subtracting, I will find precise volume between these levels.

Hope I described to sufficient length

Won't the volume of water between these two switches be measurable and constant? Do you plan to move the level sensors

either way..

4.90625 liters per millimeter of tank height.

here's one that might work

Thanks

Yes, I am looking for similar LS

This has repeatability of +/- 1 mm. I am looking for better one.

W.E Anderson division of Dwyer gives +/- 0.5 mm

I am looking for if any one offers even better

If you can only find one with repeatability on +/- 0.5 mm why not make an alteration to allow that spec to work, i.e. insert a filler in the tank over the required range that redices the effective cross sectional area by 4/5, so that what previously would have resulted in a change of 0.1mm would be amplified to a change of 0.5mm.

I expect evaporation, leaks and flow rate measurement uncertainties will come into play here.

If you want to find the precise volume between levels - you have not written how far apart the two level sensors are, 600mm?? - then geometry gives the answer; 0.6 metres height in 2.5 metres diameter is 4.908 m³ x 0.6 = 2.95 m³ , say 3 tonnes weight.

With 200mm diameter discharge pipe, that means 100m/s velocity in pipe.

That is 4.908 litres/mm height. How much water is left in tank, makes a difference to range & accuracy of any weighing system?

If you are trying to calibrate or check a flow meter, I am sure any manufacturer would run the outflow into a tank on scales and weigh that tank empty, then plus outflow - weighing is much the most accurate technique. Why not measure water level in tank with a ruler before & after the flow or have a sight glass [https://en.wikipedia.org/wiki/Sight_glass]?

If you are really trying to deliver a certain volume of water out of tank, how do you really know you have filled it back to top level for next shot? Why not fill a small tank of the required volume till it overflows (adjust volume by dropping stones in), then run it to empty?

Or have valves 600mm apart on tank side as overflow & delivery pipes.

A 1 metre column of water @ 20'C held to constant cross section will expand 2.6mm in height between 20'C & 30'C. Take off 10 % for expansion of steel tank. That is significant compared to 2 x 0.5mm error allowed for switches.

I didn't say that discharge pipe is 200 mm. Actually discharge pipe and valve will be 600 mm diameter, so velocity is discharge will be 10 m/she ripples will be in main 2.5 m diameter tank, but, I do not expect those in side pipe for level sensors

OK, You have a 20m high tank, which gives ~33s of flow & water is turned-on by a valve over 15s. I guess that means a dead time of seconds, plus a highly non-linear increase of flow plus 5s when it does not change much.

I guess 10s & 20% tank capacity gone before stabilisation, if ever really stable.

You have 2 bar of head - falling as tank empties. I think mean height of your 600mm? spread of switches in tank may significantly affect flow rate measured.

You have not disputed my 600 mm between level detectors [over 1 second], so with < +/- 0.5mm detector error, required accuracy is 1/600 - say 0.1%.

However, 10ms response time you specify is 1% of 1 second, which will translate to 1% difference of flow count That is a lot, unless both level detectors have exact same response time. Most of response time may be resistance-capacitance electronic with big tolerance 50%.

Also PLC inputs have a lag time (to avoid spurious operation), usually around 10 ms and the whole PLC has a cycle time of milliseconds. Often, the order in which PLC scans inputs is uncertain.

My suggestion would be to use a pressure transducer, at least to evaluate what is happening. Direct strain gauge will not have any hidden lags (4-20 mA outputs are often heavily lagged electronically)- also, as a single device, the time lag will be same at each pressure. N.B. Theoretically, if you put a steady ramp into a simple lag response system, the output settles to the same rate as input within a few multiples of the lag time.

The transducer could be calibrated against the sight glass as the tank is filled before test. In sun, sight glass may be hotter than tank - suggest test at night & drain-refill sight glass from tank before test.

I would use a fast rate DC voltage data logger recorder [e.g. USB instrument oscilloscope/data logger on laptop] for the pressure signal, not PLC, & arrange for marker pulses from the level switches on the trace. I would suppose that the PLC could still receive the switch signals & count the flow pulses with count stored at regular sample times. This would reveal the difference between dynamic switch operation & static operation - guess switch pick-up-drop-out differential can be tested by recording pressure & switch operation while filling [stop fill at a switch-off point, then drain slowly a little till switch-on occurs, refill a little till switch goes off -any hysteresis ]

We both came up with essentially the same solution but you posted while I was still writing GA

jhhassociates,

I gave you a GA for the magnetostriction solution. Pressure transducers (strain gauge) have hysteresis, linearity & drift limits & checking over 0.6m of 20m height span is a challenge even with calibration on the day.

OP writes he has a flowmeter, probably just for system proving [expensive if permanent] - if one always, why bother with the switches? Maybe he doubts flowmeter response rate & transient performance and thought level/time/count was an easy cross-check.

67model

You will likely find that a flow meter is the only way you will get the measure of accuracy you desire. It will be more predictable as you only have to factor in the repeatability of one instrument, rather than two. Also, if the liquid is not clean, or if it is clear, or tends to leave a residue, has waves on the surface, has unusual surface tension characteristics, you can rule out most easy to apply point level solutions. Weigh cells are also a tough application.

Of course, a flow meter will easily cost you maybe 10 times the point level install cost, so then you need to figure out just what is the real practical value of your accuracy specification, and decide if your instrumentation cost will justify that. Flow meters are also not easy to apply and use, the highest accuracy units have moving parts, which you would best avoid if possible.

The wall thickness might matter when you mount an internal sensor but other than that it really doesn't matter.

It would be useful to know a lot more about what you consider to be "water"; grey water, black water, sea water, potable water, pharmaceutical distilled water, cooling tower water, etc.

My WAG from the sparse information is to use load sensors under the tank and measure the weight. This way localized fluid slosh won't produce false triggers.

Water will be simple tap water, which may become little rusty after running through normal steel (not stainless steel) pipes

If the sensor is some externally fitted sensor like some ultrasonic, wall thick should surely matter. But if the sensor is protruding inside the tank, wall thickness should matter, as I think.

I agree with redfred here, the load sensors can get the accuracy required, the rest is just high speed electronic circuit...

https://www.mt.com/us/en/home/products/Industrial_Weighing_Solutions/AutomPrecision/High_Precision_Weigh_Modules.html

Load sensor will tell us nothing about the levels

Well you have to set it up first, but the weight of the water is translatable to the volume, and the volume creates the level... If you know the dimensions of the tank you can determine the weight of the water at any level....

I think you are pushing the limits of level sensing with specs that tight. The surface of the liquid won't be smooth when falling, so your sensing height repeatability of +/- 0.5mm will be meaningless.

I suggest you consider putting a flow meter on the output. This is a much more robust sensor than level sensors.

As I explained as reply to comment by J E Chicago above, I want to measure the volume between two level switches, so... one reference is level switches and other is the flow meter readings. Thus I am already using flow meter at output.

I know at high speed level drop, there will be small ripples in water surface. so I am trying get best accuracy by using best precision level switches available

You might need custom order to spec...

https://fpisensors.com/water-level-sensors/

A pressure sensor at the bottom of the tank may give you adequate level resolution and accuracy, depending on the depth. This would also let you program the levels in the PLC and do your fine tuning in software.

If the water level falls a 600mm per second I'd imagine the turbulence on the surface will exceed your precision requirement!
I think you to reevaluate the actual requirement.
Del
(Damn PW Slack beat me to it . S'pose I'll have to give him a GA now .... grrr ftzzz )

I have observed that even at this high flow rate, the water surface is not really turbulent, unless there is some vortex formed at very lower end. (My tank height will be 20 meters)

There will be some ripples on the water surface, which may be of 2 to mm high maximum.

But, if to avoid even this effect, if I provide one more external pipe of say 200 mm diameter joined at the bottom of the tank and keep top end of external pipe open to atmosphere (remember my tank diameter is 2500 mm) and put the level sensors in this external pipe, this ripple effect also will be nullified.

But now question is shall the water level in external pipe will exactly follow the level in the main tank? Or will it lag?

Is there any relation between main tank diameter and external pipe diameter, so that there will not be any lag?

What is the size of the discharge pipe? ...and what type of valve do you have? ....How fast can it go from fully closed, to fully open?

Discharge pipe size : 600 mm

valve opening or closing time 0-90 deg in 15 seconds

So how can you expect precise level control with a 15 sec response time for the valve to shut the water off without estimating the amount of water that will be added once the level control activates, and if that's the case then why would you need such a fast response time...A float valve would throttle the flow down as it approached the proper level...then the volume could be checked by weight and adjusted via sight glass...

I am measuring the volume between two level switches in dynamic conditions, so opening and closing of valve doesn't matter

So you want to measure the velocity?....and derive the volume from that?

The statement <...measuring the volume between two level switches in dynamic conditions...> sounds as though the measuring technique is incorrect.

Has a formal HazOp Study been carried out yet? What was the outcome?

So, what actually starts and stops the flow, given that the travel time of the stated valve is equivalent to half the height of the vessel during the level-drop activity, please?

This is a funny-old process. Is there a Process Engineer allocated to this job? Why is the individual asking to achieve the unachievable? Is it lack of experience?

A1) It will lag, as there would be flow between the <...external pipe...> and the <...main tank...>.

A2) Any <...relation...> could only be established by experiment. At the sort of tolerances being specified, temperature will be a criterion for repeatability, as the properties of <...water...> vary with temperature.

A weir remains a better device than a <...high precision level switch...> _{[whatever that really means]}.

<...the repeatability of the sensing height should be within less than 0.5 mm. Even if it is 0.1 mm, that is great....>

So, if temperature remains a variable, then, because no temperature tolerance has been specified, then:

• The density of water, and therefore the level, varies with temperature, which will affect the repeatability.
• There is no criterion on surface tension of the <...water...>, variations in which, introduced by the change in substances within it, will affect the trip point of the <...High Precision Level Switch...>.
• There is no criterion on the viscosity either, which is strongly influenced by temperature, and will have an effect on the rate of change of level as the water drains from the <...vessel...>.

It's starting to look as though a weir would be a better solution to maintaining a level than a <...High Precision Level Switch...>.

A floatless level switch is the first method that comes to mind one that I have installed many times is made by Omron.

If you have a PLC as I believe you did say then why not use a pressure sensor with a 4-20ma output and write the logic in the PLC for the two levels that way you can change the trip points at any time in the logic or by configuring thumbwheel switches. Wika make an excellent unit which I have installed in coal washing jigs.

The third type is the use of a magneto restrictive float transducer. As the float moves on the sensor probe the output changes from the sensor. I have installed 1v to 11v output probes on a coal washing jig to open the dump gates. They can also be obtained in 4-20ma versions.

All the above units have high accuracy and low hysteresis so any would suit your needs without going to Differential Pressure cells and their extra coast.

Pressure is a function of density and that varies with temperature and height, which means that the volume measured will vary with temperature, failing the repeatability criterion specified in the original post.

Gsuhas,

Your spec's, as PW Slack said, are incompatible at the extreme level change of 600 mm/sec. With the stated diameter of the tank you are dealing with a flow rate up to nearly 3,000 liters/sec., which makes me wonder if there is an error in your calculations. The specifications you give are for an accuracy of 1:40,000 and a response time of 0.8 millisecond.

Floats are out because of speed, any wetted sensor that depends on the water disappearing from its surface will be too slow also. You might find an ultrasonic sensor that meets these specifications, but I doubt it.

Discussion has raised questions about the possible turbulence of the water and ripples in its surface. In my experience the suggestion of a separate tube to isolate the sensor from these effects is good--I suggest one with a diameter of at least 150 mm.

--JMM

Yes, the units and numbers do not make sense at all. (AI language translation?)

All the process point level switches I've dealt with have some electronic circuit that has some inherent delay, but none provide a 'response rate' spec.
I suspect that the fastest response would probably be from magnetic float level switch, like something shown here:
https://koboldusa.com/products/level/level-monitoring/

Given that the internal switch/magnet is a physical initiation which doesn't change over time, its repeatability depends on the movement of the float as drops/rises with the dropping/rising water level.
A top mount style could be fastened to a depth adjusting rod with a screw adjustment to allow you to make small incremental adjustments to the exact empirical trip point needed. I would expect some hysteresis between falling and rising make/break switch action. Might take two float switches at the same level, one for rising level, one for falling level, to get such exacting trip points as you expect.
They don't cost a lot, so it's not a huge cost to experiment.
Just add a PLC DI input with a fast scan rate.

Are you trying to build a system to test/calibrate the flow meter?

"ripples will be in main 2.5 m diameter tank, but, I do not expect those in side pipe for level sensors"
But the level in the side pipe will oscillate same as any U tube ...
Del

You have a valid point, Del. I guess some people have never watched the tube of a coffee urn while another filled a cup of coffee.

A simple float operated magnetic reed switch with stainless or polymer ball is the cheapest and simplest solution.Use a stilling well to reduce turbulence at the sensor.

Inquire at this link for details on a very sensitive version of their switches.I can speak from experience that these are virtually bullet proof when properly applied.

This is only one brand,but there are many others to choose from.Use Google.

Of course,the logical way is through calculation using flow, as mentioned previously.

https://www.gemssensors.com/level/single-point-level-switches/float

Here is a link to another method.These can indicate as well as switch.:

http://www.dwyer-inst.com/Product/Pressure/DifferentialPressure/Gage-Switches-Dial/SeriesA3000#specs

A simple bubble tube level sensor immersed in a stilling well set deeply in the tank to prevent turbulence.

With an analog or digital transmitter it will provide constant feed back of the level,and the output can be programmed to open/close contacts at any point of the calibrated span.

Old school,but it still works.

Suggestion for eliminating all the triggering and electronic processing response times.

Ditch individual level switches and the possible errors introduced by how accurately you can position them. Install a Magnetostrictive level sensor in a stilling tube within the tank where it should be possible to get a resolution of 0.1mm with an accuracy of +0.5mm (Google WIKA data sheet LM20.01 for initial details but there are many brands on the market and others may be more suited to your application) Because you will always be taking readings with the float moving in the same direction at a constant speed you should be able to improve on the accuracy tolerances quoted. Talk to the manufacturers for more advice. The level sensor gives a 4-20mA output and I would suggest that you input to a PC rather than a PLC to minimize scan time errors. Digitize the 4-20mA computer input to the same resolution as the level sensor resolution output. If you fill the tank in 0.5 litre increments and plot the depth on the level sensor to a computer look up register you can eliminate errors caused by any irregularities created by variations in the cross sectional area of the tank at different heights. Using the look up table program two set points with a known volume between. Open the outlet valve and allow steady state flow conditions above the minimum turn down ratio of the flow meter to establish before triggering the upper set point using a compare function with the programmed preset. As the level reduces the second set point will be triggered. The signal and processing time from both set points will have the same time delay so they cancel each other out. The same cancellation technique applies to any processing delays inherent in triggering the start and stop of the readings from the flow meter pulse input.

The only complication I can envisage is if the Magnostritive sensor cannot cope with the speed of output generated with it's float moving at 600mm/sec. Again only the manufacturers will be able to advise you on this.

Check out this link for a LVDT level sensor,with nearly infinite resolution:

https://www.te.com/usa-en/products/sensors/position-sensors/linear-position-sensors-lvdt-lvit.html?tab=pgp-story

The simplest method to determine the volume-per/mm is to do a simple volume calculation of the tank and use the flow meter that you have.

You should be able to program the input/output values in the controller to keep the level exactly where you want it.Unless the tank volume changes due to scale,etc,the calculation will be very close.Of course,if you really want to split hairs,you could allow for volume change of tank and displacement of the water to correspond with each 1/1000 degree of temperature riser in degree Kelvin,of course.You would also have to factor in the tide times at your location,even if you are not near the coast.If the tank is open,you must allow for bird droppings,etc.

I am presuming that the tank is vertical,not horizontal.If horizontal, and level,the height change per volume is not linear,and you will have to calculate it.The calculation gets a little more complicated if the tank is horizontal and not level.

Plenty of programs on the net that will do the math for you.

The KISS System is always best.

How about a totally different approach? Attach a u-tube mercury manometer of sufficient height and diameter to the tank and do the measurements of level on the mercury. This would decrease the vertical height from 20 meters to less than 1 meter and reduce the speed of level change by a similar ratio. The manometer tube material can be chosen to be one which is not "wetted" by the mercury. Possibly a laser distance measuring gauge could be chosen for measuring the height, if accuracy and speed are adequate to the task....

JMM

You might investigate a gray scale glass scale and moving sensor (float activated).

The glass scales used on Co-ordinate Measuring machines have sub-micron accuracy and the sensor response times fall inside of 10ms. (but a glass scale of the length that you are requesting seems impractical)

You might be able to get a lower precision engraved gray scale on stable material for the length that you are seeking.

Alternative might be a full length glass/perspex rod (waveguide) such that fluid/air transition causes reflection and measure depth through the reflected pulse delay.

OK! My last suggestion:

Contact these people,they are the masters of level control.

https://www.dwyer-inst.com/Contact-Us/

If they don't have it or can't make it,you really don't need it anyway.I have provided a link and you must do some of the homework yourself,we can't so it all for you.

"Nuff said"