Did I get any petrol?

With the budget on Wednesday, I wouldn't be surprised if it does... ;)

D

This time I was more suspicious and noted everything down at the Esso filling station.

Before filling my car computer said: 49 miles 29.5mpg After filling: 435 miles 29.5mpg Received mileage: 386 miles Received gallons: 386/29.5 = 13.0847 gallons = 59.44 litres.

Pump displayed 60.99 litres Shortfall = 1.55 litres @ 1.059 that £1.64!!!

Notice that an error in the mpg will have no effect on this calculation, since the car will just calculate a different mileage left corresponding to the volume received and the current estimate of mpg. That is; the mpg can be any number, it wouldn't matter. The car tank measures the volume and translates this to mileage.

However any error in your fuel guage will have an effect.

[...]

That is possible, but such an error would more likely be an offset error, but taking the difference eliminates any offset error. Next time I will try to fill the tank half full and see what happens. If the pump registers air at start instead of fuel, then the discrepancy would be constant rather than proportional.

And here is another clue while I have my calculator is out of the drawer:

49 miles at 29.5 mpg translates into 49/29.5*4.54609 = 7.55 litres. Pump litres received ... 60.99 litres

------------------------------------------------------------------- Total 68.54 litres.

The tank capacity of a 1993 Saab 9000 CSE is 66 litres!

And I always stop filling when to pump stops to leave room for air. Again, an error in mpg doesn't matter, but there could be a small offset error in the tank, but even then it is hard to reconcile this result.

Why? Car fuel tank guages aren't calibrated & tanks aren't a constant cross section.

Well report it to trading standards or get a large metal petrol tank, weigh it, fill it & measure the fuel temp, then weight it again.

Car tank not independently calibrated, if that is what you mean. Surely, the car manufacturer would think of cross sections and take that into account for the fuel gauge; that would just be a look-up table in the car computer. The car was at perfect horizontal ground at the pump station. I am not making any claims, just reporting my findings which makes me suspicious.

There's no conceivable way the estimate of fuel received can be accurate based just on the tank's sensor level reading. You only have to look at the way fuel gauges are non linear on most cars. Mine doesn't even start to move until I've used a gallon or so. It's a 55 litre tank and if the gauge were linear each 1/4 mark would indicate 13.75 litres had been used. In fact based on fillups at various times combined with other calculations the first

1/4 mark comes up after 16 litres has been used, the 1/2 way mark after 25 so that 1/4 only represents about 9 litres, the 3/4 mark after a further 12 litres and the last 17 litres is all in the final 1/4 of the gauge.

That means when on a long trip it takes ages for the gauge to hit the first

1/4 mark and I always think this looks like a good run for mpg and then the 1/2 tank mark comes up in no time and dashes my expectations again. If your computer can measure to even with a couple of litres I'd call that very good indeed but no reason to dispute a calibrated filling station pump.

However, the non-linearity that you see affects the middle range and not necessarily the top and bottom mark. The needle display inevitably has a safety margin to prevent that you run out of fuel. The computer readout has a much higher resolution than the needle display, though there could also be an offset as discussed. I don't know how this mechanism works, but I assume that it is an easy job for the Saab manufacturer to have a look-up table to correct for any non-linearity.

However you seem to be under the impression that Saab would have had a burning desire to make the fuel tank gauge incredibly accurate which would be a very difficult thing to do and rather pointless for most purchasers. Even tiny errors in the mounting position given that a fuel tank is a not a high precision machined shape would affect its reading. Yours appears to be within a couple of percent. That's extraordinarily good by any standards other than space shuttle requirements. However as I said it will still be far short of the accuracy of a petrol station pump.

I am more inclined to believe the pump than your car's fuel gauge!

So what about that the tank capacity of my car is 66 litres, the computer indicated the there were still 7.55 liter in the tank, and the pump said I received 60.99 litres. And I stopped pumping when the pump cut out, leaving air at the top. Something doesn't add up.

Given most guages are a needle that would be fairly pointless.

Not really, they'd have to allow for fuel temperature & the angle of the car.

Not pointless when you have a "mileage left" readout.

We are all agreed that something doesn't add up.

For some reason you don't seem to be grasping the point that everyone else thinks the reading from your fuel gauge is likely to be more inaccurate than the reading from the pump. It doesn't matter how many inconsistencies you can show between the two of them -- that they differ is not in dispute.

That's not very accurate either though. Unless you run it out you'll never know so they all have a safety margin.

A tilt of the car would of course make the reading unreliable, but then we would expect that and not rely on the reading. As far as I have been able to find out from quick browsing, the expansion coefficient for petrol is

0.0012 litres/Deg Celcius. That is; a 5 degree C increase for a 50 litres tank would produce 0.3 litres expansion; not very much. of