Kia Forum banner

Non-linear Fuel Gauge

2K views 16 replies 9 participants last post by  ecanderson 
#1 ·
On a full tank of gas, my gauge barely moves after a 100 miles of driving. After 200 miles, it goes down about a quarter mark. Then it goes to near half way mark as it approaches 300 miles. After that, the needle drops rather quickly close to empty mark after about 450 miles.

I know most cars are designed this way, but it's a little extreme in my Sorento's case.

In terms of MPG, I'm happy with about 29 (manual calculation) I've been averaging so far over 10,000 miles. I do about 75% highway driving with half of it in rush hour traffic. The best range I've gotten was 520 miles, but the fuel light came on and had to fill up with just over 17 gallons of gas.
 
#4 ·
The shape of the tank makes it virtually impossible to have gauge follow actual consumption. You used to be able make adjustments by bending the float arm years ago - don't know if that's what could be done now or not. Like motorcycle speedometers reading fast, the inaccuracy at the bottom is to keep people from having to walk.
 
#6 ·
If the gauge is direct wired to the indicator and the data only passed on to the computer, that would be true. A decent design would pass the data entirely to the computer, and the computer would tailor the output to the gauge based upon the non-linear volume of the tank vs. sensor. It CAN be done better.
 
#9 ·
Just did a fillup
The Lie-O-Meter showed 16.9 MPG
Computation based on gallons and miles was 16.6 MPG
Pretty darn good.
I do force in as much gas as possible, if I had stopped when the pump did the first time would have probably been right on.
 
#13 ·
Thats a bad idea to overfill. You have to leave room for vapor expansion.

Actually it would not matter in the actual calculations as you pump in more gas and again overfill the next time. The dash will probably get even more optimistic than it is.
 
#10 · (Edited)
In reality the electrical controls of all modern vehicles know fuel injector usage in grams per/second and speed for TC/AWD/ABS/etc to fractional degrees. Any manufacture with programming knowledge, time and money could report instantaneous and average MPG/KPL down to 2-3 decimal point accuracy. Problem is, at what point is it economically feasible, and how many manufactures would add cost and that complexity, when +/- 10% accuracy is available at lower cost using an analog level sensor in weird (non linear) shaped fuel tank.
 
  • Like
Reactions: Sloppy
#15 ·
The g/s figure is only as accurate as the rail pressure behind the injector and whether any restriction has begun at the injectors over time. FISFR (which in itself can have as much as ~4% between injectors) x time only gives you an accurate volume reading where the pressure is a known fixed value (many vehicles spec a 5-8% spread between min/max) and the injectors are factory fresh and carefully selected. Beyond that, things get variable.

Given that variability, fuel trim measured at the pre-cat O2 sensors is what keeps the (mostly)closed loop working properly.
 
#12 ·
Yeah, this is annoying but.... you know in your head what to expect to it really poses no issue (especially when you have the miles to empty that seems pretty accurate.... at least when you are starting to get low on fuel)... and then of course there is a reserve on top of that so you would have to work really hard to run out of gas.
 
#16 ·
ecanderson, thanks, you bring up a good point, if the fuel g/s reading is not directly altered by the trims then the reading is only a relative number and is not a true reading. I've never really paid attention the fuel g/s to see if it moves with the trims on my other vehicles. I'm going back to check some of my other vehicle traces. My assumption was the g/s number would be adjusted inversely with the injector PW, and then used in the fueling formulas. But I can't validate that assumption. In any case I'm way off of the original posters topic.
 
#17 ·
In trying to maintain fuel trim for ALL reasons, the ECU will attempt to deliver more or less fuel, and the resulting +/or/- change in timing of the injectors would be picked as a +/or/- difference in the calculation of the g/s. However, that change in the inputs to the g/s calculation will make some of the same assumptions about fuel delivery per injector per unit of time that causes the variable components to contribute to a fuel trim 'error' that needs to be corrected in the first place. So yes, trim will most definitely come into play in a g/s calculation, but the relative error (due to some of the variables dictating the exact amount of fuel used in each cylinder) will remain in the final calculation. Guess the best way to say this is that on a relative basis, trim will drive the g/s up and down correctly, but with whatever % error exists in the original baseline value.
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top