# reliability of consumption indicator

Still in the long term purpose of understanding the various factors of consumption of my car, I recently discovered that I have about 15-20% difference between the liter/km (and indeed, liters) given by the car computer vs the value I obtain by filling the tank quite to the same max level each time divided by km done since the last filling.

-> Where is the problem ?

On a few measures on more than 40l, I see about 7l of difference between liters used according to the computer vs the gas pump. So I don't think imprecision of my "filling at same level" could explain. Idem for evaporation (plus it's in the wrong order, since the computer see more liters consumed than what I do have to refill).

-> On which kind of sensors is based the consumption evaluation? tank level, or some accumulated flux at injection? Or more indirect? Are some values empirically estimated rather than measured, or accumulated with bad precision? Are my sensors faulty, or is it quite a common imprecision?

• Rather obviously the computer's algorithm and / or measurement tool is way out of calibration. Commented Aug 31, 2016 at 11:15

The measure at the pump should certainly be pretty accurate, at least in terms of volume as they typically use positive displacement type pumps (ie each cycle of the pump delivers a known volume) and are calibrated for the purposes of trading standards.

However one complication is that although fuel is sold by volume it is its mass which is the true indicator of how much energy is consumed. Tanks at filling stations are usually underground and so should be at a fairly constant temperature but the fuel tank of a vehicle can heat up and cool down a lot so the fuel expands and contracts, changing its volume.

If the computer is measuring fuel consumption via the flow through the injectors it is entirely plausible that the absolute value of volume flow-rate isn't very accurate, again because what the ECU really needs to know is the air/fuel ratio by mass. Here the volume flow-rate is a convenient measure but will actually be adjusted by feedback with other sensors so it's more interested in whether more or less fuel is required by the engine than the absolute quantity. Again it is likely that the fuel will be substantially warmer by the time it gets to the injectors than it was when dispensed at the filling station.

If it is measuring the level in the tank then it may just not be a very accurate sensor as it's main purpose is to give the driver a rough idea of how much fuel is left and changes in level and fuel sloshing around.

The two measures you have which should be pretty reliable are the volume dispensed at the pump and the car's odometer.

Ideally you should keep a running total of both ie when you fill up note the mileage and the volume of fuel added. That way your calculations are based on the most reliable data easily available and you should smooth out some of the rounding errors.

Ultimately volume is a convenient way of selling fuel but for very precise measurements of efficiency and fuel economy you really need to be dealing in mass.

• The UK regulations for fuel sale require errors in volume less than about 1% - see Appendix 1 or nmodoitonline.bis.gov.uk/fileuploads/legislation/…. The EU regulations should be similar. There is no way that temperature corrections etc will high enough to cause the 17% difference quoted by the OP. The simplest hypothesis is that the car's fuel measurement system has a fault, either in the hardware or the software. Commented Aug 31, 2016 at 11:28