Horizontal Tank Level Measurement & Logging

[Joe_WaZoo]

Hello,

I need a little help or guidance from you level control gurus.

My customer has a horizontally mounted tank that is 144 inches long and 64 inches in diameter. The ends are flat. The customer would like to know the daily usage of the liquid in the tank via a measurement in the morning and another at the end of the day. They use a gravity feed system out of the tank and on some days they believe that they use only 10 gallons of liquid and other days they may use upwards of 100+ gallons. The use of measurement today is a long yard stick. They use that plus a look-up table to give the estimated tank volume.

We have used an ultrasonic sensor to measure the contents from the top of the tank looking down at the liquid, and we get pretty close to the volume, +/- 20 gallons depending on the changes in the analog output of the sensor. The customer has a linearization sheet of 500+ points of volume reference, and we can only linerize upto 100 points from our hardware. For example, we are talking about in the analog sensor output of 9.768mA to 9.695mA, this corresponds to 1294 gallons to 1308 gallons respectfully.

The customer wants accuracy of up to 1 gallon changes in this tank measurement. I say that is impossible given it is a horizontally mounted tank and that our sensor can not detect those extreme small changes.

What solutions would you recommend for this sensing application?

Thanks,
Joe_WaZoo

P.S. After all my internet research, I cringed at the sight of horizontally mounted tanks.

[nonuke]

You can install 4 load cells.

[OkiePC]

Even if you replaced the look up table with math to more accurately calculate the volume of the tank, I doubt you would get close enough with that type of sensing.

My first choice would be to use a flow meter to directly measure the actual usage, and the second choice would be to use load cells to weigh the tank (emery winslow)

[leitmotif]

Joe

If they use 100 gallon and they want a 1 gallon tolerance that right there is 1% tolerance. 1% tolerance gets expensive.

Seems to me the cheapest would be a city utility water meter - assuming compatiblity with meter components and liquid has constistancy of about that of water. With a 1% tolerance they may have to do temperature correction.

Is it possible the bean counters have you using $500.00 of metering to chase $0.05 of material??

Dan Bentler

[OkiePC]

We use these and they've been very accurate for us (measuring water at 40psi in a nice straight pipe section). Only issue is changing the batteries about once a year.

[leitmotif]

Originally Posted by OkiePC

We use these and they've been very accurate for us (measuring water at 40psi in a nice straight pipe section). Only issue is changing the batteries about once a year.

Have seen those and seem to work well - looked at specs most are 1% and some are 0.75%. Barely within customer desired spec. If customer can loosen up the spec these should work well.

Dan Bentler

[danw]

I'm too lazy to do the math.

For a tank of those dimensions, what is total capacity in gallons?

Dan

[JeffKiper]

Is the density the same all the time? If so can you use a pressure transducer to measure the force of the liquid?

[Paully's5.0]

Here is a good link I found a few months ago:

http://www.arachnoid.com/TankCalc/

It builds you a lookup table based on the parameters of the tank you specify, as well as the height increments you desire.

I agree that it will take a lot more time and money to get it within 1 gallon.

Is the tank mounted at true 0 degrees? If not, well that is another variable you have to contend with.

The above app does calculate a "tilted" horizontal tank. It's pretty slick.

[jimbo3123]

Originally Posted by OkiePC

Even if you replaced the look up table with math to more accurately calculate the volume of the tank, I doubt you would get close enough with that type of sensing.

My first choice would be to use a flow meter to directly measure the actual usage, and the second choice would be to use load cells to weigh the tank (emery winslow)

There's a simple formula for the volume that can easily be translated to PLC logic (assuming your PLC has built-in trig functions). I derived the formula a few years back (I can try to do it again) by calculating the integral from the bottom to the top of a semicircular shape.

As for the tolerance of your sensor, I'm sure you could look up the specified tolerance of your level sensor and multiply it by the length and width of the tank (at the middle for worst case). That would tell you what kind of accuracy to expect.

[Bitmore]

I would a suggest a differential pressure transmitter. check the link http://www.spiraxsarco.com/resources...am-boilers.asp

I know it is a tutorial for a boiler but at the end it shows the differential pressure application. Rosemount, Yokowgawa, and many others claim they can detect levels at the precision you require.

bitmore

[danw]

In a tank of those dimensions, a one gallon change in level at midlevel, where the worst error occurs due to tank geometry, is 0.05" level change.

Industrial ultrasonics can't do that. Industrial ultrasonics are generally good to ±0.25" not ±0.05".

Head pressure measurement with a true industrial grade DP transmitter will easily resolve 0.05"w.c., if the density (specific gravity) of the material in the tank remains constant. That can be iffy.

A pressure transmitter can't be in a flow or drain line because the flow will affect the head pressure value.

Load cells will make this measurement, too.

Dan

[Roy Matson]

A magneto-restrictive probe is probably the most accurate. I am working with one for a 26,000 gallon tank that has very fine resolution. Send me an e-mail I will send some information.
roy underscore matson at yahoo dot ca

[danw]

I agree, magnetostrictive is very accurate, accurate enough that MTS offers a model that talks Modbus, so that resolution isn't lost in the analog conversions.

Medium has to be self-cleaning, because the float will stick in goopy stuff that builds up film or residue on the probe shaft.

Dan

[bikerdude]

Joe,
Are you using a lookup table now?
Are you looking for math to determine the level in between of the lookup points? Say if 9.250 ma = 1000gal & 9.275ma = 1005gal. There is math available to figure out what the level is @ 9.260ma. I don't have it on this PC but I have it somewhere. If I'm way off here just ignore me! I recently had a tank level display installed by a vendor that agreed to let me "Help". The horizontal tanks are a pain for sure. What kind of customer insists on 0.1% accuracy? I know, I know... the kind that don't want to pay extra for it! The life of a "True" programmer would be boring without these challenges to keep the old grey matter pumping.

BD