When you purchase the fuel, it is spent. If the machine takes 100 gallons and you need it April 12, you load the 100 gallons on April 12 in your schedule at your estimated cost. There is no value tracking the fuel burn because there is nothing you can do with the performance variance. Once the fuel is loaded, it is gone. (I retract this entire answer, above; I was too narrowly focused.)
EDIT: My original answer was constrained by thinking only from like an earned value perspective, but after thinking about this more, and the comment below from the OP regarding his attempt at planning, I think I was looking at this through a paper towel holder.
For estimation, I would use the data I have already collected around work and duration for labor as an input into my fuel estimation. The manufacturer of the machine ought to have a rated fuel burn for its use, like a car's miles per gallon or a plane's gallons per hour. That datum is also an input to your estimation. If you have historical fuel burn values, then that is also an input.
So as an example: If in my work package I estimated 100, 180, 125 hours (optimistic, pessimistic, most likely) of work; and 10, 25, 13 days duration. For planning values, based on my risk analysis, I may choose 135 hours of work and 15 days of duration to load into my schedule.
In this particular package, the machine is expected to operate about 80% of the time. So only using this is a guiding principle, 80% of 135 hours might be the planning value I use for fuel usage, or 108 hours. I can reduce or add to that based on other risk factors.
Continuing this example, the machine might be rated at 35 gallons per hour and my historical data might show I typically get 30 to maybe as high as 38 gallons per hour. So for a planning value I might choose 33 gallons per hour to load some buffer. 33 x 108 is 3,564 gallons to be used in that package over the 15 days of work.
Price of fuel is also variable. I would apply the same logic to the price per gallon and choose the value that is best suited for risk purposes.
Loading the fuel into the schedule depends on how you are purchasing it. Are you purchasing all the fuel at once and storing it or as you go? If as you go, I would load a separate control account for materials and underneath this a package for fuel. I would schedule the fuel purchase when I would expect to purchase it, the quantity and the associated cost based on my estimation above. If you run out early or have more than expected at the next purchase, you would load the actuals just like labor and track your variances, which would then allow you to readjust your planning values if and when you replan your work down the line.
I hope this helps and good luck!