Fuel Oil Level tank fluctuations 1

[mabbympb]

Hello,

We have a fuel oil tank that is approximately half full (300,000) litres.

This tank is only there for use as a standby, should our gas supply become interupted.

The tank is measureing the level using 3 rosemount level level transmitters. Looking at PI trends over a period of ten days (any ten days) the level measurements 'change' by a few thousand litres.

The trends show peak high levels at approximately 4pm each afternoon, and peak low levels at approximately 4am each morning.

can anyone help explain this please?

http://img266.imageshack.us/img266/8184/fueloillevel.jpg

 

[jmw]

A couple of thousand litres in 300,000 litres? i.e. about 1%?
And how does this track with temperature changes and corresponding density changes?
Do you have warm days and cold nights? How does this change seasonally?
You could get something close to 1% for a 10-15C swing.
What is the accuracy of the level transmitters and the effect of temperature on their accuracy?
(see

http://www.viscoanalyser.com

downloads for a temperature density calculation, density 12mf.xls)

JMW

http://www.ViscoAnalyser.com

 

[ivymike]

differential (fuel vs tank) thermal expansion/contraction seems the most likely cause.

 

[dcasto]

a seal tank? No moisture condensing/evaporating?

 

[BigInch]

Classic diurnal thermal expansion of the fuel.

0.0008 1/C * 300,000 Liters * 10 C diurnal temperature change
is 2667 liter change in oil volume.


"The top of the organisation doesn't listen sufficiently to what the bottom is saying." Tony Hayward X-CEO BP
"Being GREEN isn't easy." Kermit

http://www.youtube.com/watch?v=hpiIWMWWVco

http://virtualpipeline.spaces.live.com

 

[jmw]

"Diurnal", that's the word I should use!

JMW

http://www.ViscoAnalyser.com

 

[BigInch]

Yes, a its a good one. Of two or more urns.

"The top of the organisation doesn't listen sufficiently to what the bottom is saying." Tony Hayward X-CEO BP
"Being GREEN isn't easy." Kermit

http://www.youtube.com/watch?v=hpiIWMWWVco

http://virtualpipeline.spaces.live.com

 

[katmar]

If by "PI" you are inferring that the level gauge works by measuring the pressure then the apparent level fluctuations are not due to the fuel expanding and contracting in the tank. The pressure at the bottom of the tank is the classic
P = [ρ].g.h

Even if the tank level (i.e. h in the above equation) does vary with temperature, the pressure gauge will not "see" it because the product of [ρ] and h will remain constant.

The cause is more likely to be the effect of temperature on the filled capillary between the sensor diaphragm and the actual pressure transmitter. If the filler oil and the capillary material do not have exactly the same thermal expansion characteristics the pressure in the capillary will change with temperature and you effectively have a wandering zero on your level gauge.

Katmar Software
Engineering & Risk Analysis Software

http://katmarsoftware.com

 

[mabbympb]

thank you for your responses,
Yes, it is clear that this is the temperature effects on the capilliaries causing the change. The scale of the graph threw me slightly, however when you work it out it is only a 0.9% change from high to low.

I have a question for BigInch

In your calculation, what is the 0.0008? is it a density coeffecient per degree for fuel oil?

 

[25362]

In the way BigInch presented it, it appears to me that 0,0008/^oC is a cubic thermal expansion coefficient.

 

[BigInch]

Yes, a volumetric thermal expansion coefficient for an oil.

"The top of the organisation doesn't listen sufficiently to what the bottom is saying." Tony Hayward X-CEO BP
"Being GREEN isn't easy." Kermit

http://www.youtube.com/watch?v=hpiIWMWWVco

http://virtualpipeline.spaces.live.com

 

[jmw]

If you know your oil density at 15C you can use the spreadsheet I linked to above to get and idea of how the density changes with temperature. It also calculates the volume correction factors, I assume your tank contents ought to be referenced to standard volume for accounting?

JMW

http://www.ViscoAnalyser.com

 

[mabbympb]

jmw, thanks for the link i just couldnt figure it out



At 2pm today (Monday) I fixed a hosepipe to the capiliaries of the transmitters and doused them with water, and continued to do so for an hour. You can see a clear difference in the drop of level on the PI graph (

http://img186.imageshack.us/f/dieseltanks2.jpg/)

(Please ignore the yellow line, i have no idea why is shoots up?!)


To work out the difference in levels during the day the following formula is used:

Difference in volume = Coefficient of thermal expansion * difference in temperature * quantity in tank

If you take at a rough guess the highs and lows of the level being 293k litres and 287k litres respectively, then you can say that the daily average is 290k litres.

(I had to work this out in Fahrenheit, as I could only find the coefficient of thermal expansion of diesel in F and not C)
10 degrees C = 50 degrees F
25 degrees C = 77 degrees F (These are rough estimates at the ambient during day and night)
Coefficient of thermal expansion for diesel = 0.00046 per degree Fahrenheit

Put all this in to the equation

Difference in volume = Coefficient of thermal expansion * difference in temperature * quantity in tank
= 0.00046 * 27 degrees F * 290,000
=3601.8 litres

From this, you can say the measure levels due to the changes in ambient conditions will swing +/- 3601.8 litres. If you look at PI, this proves this theory.

http://img186.imageshack.us/f/dieseltanks2.jpg/

*
*
*
*
*
*
*
But one final thing, is this the whole volume of liquid expanding in the tank, or just the small amount in the capilliaries
ie. there is less liquid to heat and cool in the capillaries and so this is why you get this kind of fluctuation in level readings?? Ideally if this was a critical system you would have temperature compensation and density compensation, however the process is not that important to plant availability

 

[BigInch]

Its the temperature corrected volume of whatever you think the original volume was. We have no idea from where you got your 300,000 liter number.

"The top of the organisation doesn't listen sufficiently to what the bottom is saying." Tony Hayward X-CEO BP
"Being GREEN isn't easy." Kermit

http://www.youtube.com/watch?v=hpiIWMWWVco

http://virtualpipeline.spaces.live.com

 

[jmw]

Plus we have no idea when your level system was calibrated or how.

Your system can tell you about changes in volume but how accurate is its measurement of total volume?

Tank level is the first step in volume calculation.
You need your tank shape tables and then you need to know how the tank volume changes with temperature, as has already been indicated, and with actual fill - how much does the tank distort due to the mass of fuel in it and how does this affect volume.

The easy bit would seem to be noting the fact that volume changes with temperature because your volume measurement changes with temperature.

I'd suggest looking up the history of the sensor. Manufacturer's data, any relevant calibration certs etc. Chances are it hasn't been calibrated in a while.

How long has this fuel been sitting there. Has it stratified in the tank? has it begun to break down? got too much water in it?

What could happen to it?
It can be pilfered is one thing. It can degrade for another.
There ought to be a routine of taking top middle and bottom samples to look at quality. These samples should give you the analysis of the oil which should give you the density at 15C or 60F some where.
(by the way, if you have a 60F value, you can enter the data into the spreadsheet in degF, calculate the base density for 15deg C from that and then use the spreadsheet for your calculations. It isn't necessarily kosher, but it will do what you want.)

JMW

http://www.ViscoAnalyser.com

 

[BigInch]

I think what you should do is to constantly correct the indicated volume to 60 F, or the temperature of your choice, keep a record of the corrected volumes, additions of fuel to the tank, use of fuel from the tank (again all corrected to 60F) and have your system trigger an alarm when you have variations that cannot be explained by temperature changes, which could indicate a leak. You've got all that nice data, might as well do something useful with it.

"The top of the organisation doesn't listen sufficiently to what the bottom is saying." Tony Hayward X-CEO BP
"Being GREEN isn't easy." Kermit

http://www.youtube.com/watch?v=hpiIWMWWVco

http://virtualpipeline.spaces.live.com

 

[katmar]

But one final thing, is this the whole volume of liquid expanding in the tank, or just the small amount in the capilliaries ie. there is less liquid to heat and cool in the capillaries and so this is why you get this kind of fluctuation in level readings??

The liquid in the tank probably is expanding and contracting with temperature, although with the thermal inertia of 300,000 litres it will not heat up and cool down nearly as much as the ambient temperature. But this expansion is irrelevant. As I explained before, what the pressure transmitter "sees" is the tank's [ρ].g.h, which does not change with temperature. What you are seeing is the expansion and contraction of the filler oil trapped in the capillary tubes causing the zero to drift. Because the surface area to volume ratio of these tubes is enormously higher than the tank itself, this oil will track the ambient temperature more closely. If you can protect these capillary tubes from the sunlight during the day, and the cold wind during the night, or rain and snow, you will see much smaller fluctuations in the apparent tank level.

Katmar Software
Engineering & Risk Analysis Software

http://katmarsoftware.com