Fuel tank temperature level compensation

Q&A and troubleshooting
, , , ,
1

Hello.

I have built an off-grid solution for a costumer.
There we have a 2000L Diesel fuel tank, that is connected to a **GX Tank 140. **
The tank is installed in an unheated garage.

In the picture we can clearly se that the fuel level walks with the temperature in the garage.
= Normal density change due to temperatur = The level sensor is a pressure sensor that is in the bottom of the tank.
Is there any easy compensation for this?

image
image572×917 21.8 KB

Best regards!
//jimmy

2

This is an interesting question, and one I am professionally interested in being a chemical engineer.

Firstly, it is not actually to do with the density of the fuel.

Pressure = gravitational constant x density x fluid depth.

If you have a fixed mass of fuel in a fixed size of tank, then if the fluid warms up and the density falls by 10%, that means the volume increases by 10%, so in a fixed tank the depth increases by 10%. Density falls by 10%, depth increases by 10%, you have the same pressure reading even though the volume is higher. so the temperature density effect is not causing your measured level to vary.

The effect is most likely that the ambient temperature is causing the tank itself to expand and contract and this effect is larger. When the tank heats up, its floor area increases so a fixed volume of fluid will have a lower depth. Your data shows that at higher temperature you measure a lower pressure and hence lower volume.

To compensate this you need to

  1. Load the large firmware image and enable Node-RED.
  2. Create a graph of tank volume against temperature for a period with no fuel input or output and fit a curve to this so you know how the volume varies with temperature. Do this as a fraction of volume so say 0degC is 1500l and 20degC is 1350l then volume changes by -10% in 20degC or -0.5%/degC. This is % of volume, not tank level.
  3. You then need to choose a reference temperature.
  4. In Node-RED create a virtual tank called something like “Corrected Fuel”.
  5. Create a flow that reads the measured tank level, has a function that in the case above increases the volume by 0.5%/degC for every deC above zero or reduces the volume by 0.5%/degC below zero.
  6. Output the new volume calculation to the virtual tank.
3

Interesting problem. The graphs appear a bit out of phase too.

4

insulate the fuel tank?
Looks like you are getting some cold weather there, so that may also help stop the diesel from waxing.

5

Good point made by @MiPaul about being out of phase as heating and cooling of the tank will lag behind the air temperature. This means that any correlation between air temperature and fuel volume will be less precise. Also, it is unlikely that the whole tank contents heat and cool as rapidly as shown, the tank walls will be an average of the air and fuel temperatures and will react much more quickly. As @MikeD says, insulate the tank, that will cut down the cycling and make changes less and you could put a temperature sensor under the insulation which would then measure the tank wall temperature

6

The presented graph do imply fuel movement in/out ?

A physical variation of 20%, 1300 liters - 1700 liters in a 2000 liters tank, from temperature only seems absolutely impossible.

7

That would help explain why the level variations switch phases vs. temperature, and why the peaks in level are sharper and have gain over temperature, but I question the addition and usage would be at the same rate unless it’s an extremely big generator and a small bucket filling. I wonder if there is any barometric influences, maybe from a one-way vent?

8

Yeah, if it’s an interesting issue.

The temperature sensor is also in the other end of the garage from the generator. (that causes the temperature rise when running)

Meanwhile the diesel tank is 2m from the generator. So if temp rises 5 degrees it is more likely 15 degrees around the generator.

The tank is an dubble isolated plastic tank, with an open “hole”. So no one way vale ore anything like that.

But it is quite extreme sometimes. From 1650 to 1250. 25%. That can’t be explained with any temperature effekt on plastic?

Se attached, I can’t understand that extreme difference. And the sensor measures preshure difference, so the general weather pressure shall not have any difference.

Can the sensor be broken?

Screenshot_20251113_203532_Chrome
Screenshot_20251113_203532_Chrome1080×2340 120 KB

Screenshot_20251113_203602_Chrome
Screenshot_20251113_203602_Chrome1080×2340 116 KB

9

Assume the pressure sensor has its own connection and is not on the generator supply line because that would give pressure fluctuations.

10

An up to this task pressure sensor should have, at least, internal temperature compensation and refference voltage. For example, a plain automotive oil pressure sensor will need a lot of external signal processing before being usefull.

As @pwfarnell mentioned, at least make sure you power up the sensor from a quality, stable power supply. The GX Tank supply should be OK.

Do you have some more details about the sensor ? Code, type, datasheet ?

11

Se attached datasheet.
SGE250025.001

The sensor is connected directly to the GX Tank.
It have external stable 24VDC supply that feeds the control system of the site.
Every thing else works fine.

Is a big advanced system.
3 x multiplus
1 x wind generator
1 x generator
12 kw solar
1 x waterturbine

sg-25–sg-16-datasheet—new-2018—pdf-_-785962.pdf (179.4 KB)