The problem inherent with GA fuel tanks is they tend to be flat and large as apposed to thin and tall and since planes are always moving around it becomes very difficult to measure the remaining fuel as it sloshs around.
The PA31 has an interesting gotcha if you have the optional nacelle tanks fitted (they bring about 1:20 – 1:30hr to the party). The tanks feed into the mains by way of an electric fuel pump. When you select the pump, you get a nice light to say that it is connected, not that it is necessarily working. If you then have to rely upon the trusty float gauges, you may be in for a surprise when you realise a pump hasn’t worked, an hour or so before you thought you would run out of fuel. I know of one pilot who spent an uncomfortable few days at an oil well support airstrip in the Middle East because of this. 
Since then we have changed our fuel tank management strategy such that we prove/disprove the feed from nacelle tanks by sticking on mains & nacelles prior to using the outers (they have about 1:45 in them).
Oh, there’s another gotcha where the nacelle tank may slowly drain into the main due to gravity. A puddle of AVGAS at morning A Check is a good indicator if you’ve been daft enough to refuel all tanks the day before.
Peter wrote:
except that the dodgy fuel systems tend to be in MEs
The “dodgiest” fuel system I have ever used was in a V-tail Bonanza. Tip tanks needing transfer to the main tanks before fuel could be used, and an aux tank that was only usable in flight, combined with a single fuel gauge and two switches (aux/main, left/right) and a fuel return line going into the left main tank only made provided hours of amusement.
I have almost complete faith in the totaliser. In the DA62 it’s more accurate than the pilot judging the rim of the fuel filler cap (probably within 1 or 2 litres on a 326 litre aircraft). On the PA31-350, we have a retrofitted JPI EDM system. That will be accurate to within 5-7 litres on a 896 litre aircraft.
Connect all that to your GPS and, with the correct training, there’s almost not excuse for fuel starvation.
The only accurate indication is the E for Empty.
Actually you still have a few gallons useable when the sight gauges in the wing roots of the Piper Super Cub show E.
The TB20GT also has capacitive sensors which are accurate to the thickness of the needle – probably would be within 2-3% if the gauge had markings on it other than colour bands
The problem is that while one operates a car by the fuel gauge, stopping for fuel when it gets low, a flight is planned for fuel and during flight you monitor the fuel burn versus progress along the route. For this continuous monitoring to work you need either finely graduated and accurate gauges, or a GPS-linked totaliser. In the old days (crap gauges and no totaliser) you assumed a certain fuel burn and you monitored progress by the time of arrival at waypoints (hence the old plog forms which had loads of columns for time and calculated fuel at each waypoint etc) which worked if the fuel burn was accurately known.
In reality almost nobody does any the above methods. In the GA scene most people say to themselves “I have 4hrs in the tanks so I can fly for 3hrs”. That is how one is trained in the PPL. It is an obvious recipe for running out as soon as something slightly unexpected happens. Even most pilots who have a totaliser don’t use it. Socata shipped several hundred TBs with the totaliser reading some 25% out (due to the transducer being mounted in the wrong place) and almost nobody noticed.
What really surprised me in that survey is how large a % are fuel related. Yes fuel planning is taught so badly in the PPL, and even if it was taught well it would be hard to execute because it is almost impossible to accurately run any of the along the route fuel monitoring methods.
Or on planes with capacitive fuel senders such as those from CIES on newer Cirrus models. Those are pretty accurate.
AdamFrisch wrote:
Whoever goes by fuel gauges in 40 year old aircraft, needs to watch out. They’re useless.
Let’s make that “Whoever goes by fuel gauges … needs to watch out.” Even on brand new SEP aircraft! The only accurate indication is the E for Empty.
Once you go SET it gets a bit better. I.e. the Cessna Caravan has the same type sensor, but four of them per tank, leading to an average reading on the gauge.
Peter wrote:
128 by fuel starvation
93 by fuel exhaustionWhat is the difference?
Simple. At least in US parlance:
starvation = fuel doesn’t get to the engine(s) for whatever reason, e.g. mismanagement (more likely on piston twins, as Timothy mentions), blockage, etc
exhaustion = no fuel left on board
Whoever goes by fuel gauges in 40 year old aircraft, needs to watch out. They’re useless. They need to keep track of usage, either the old fashioned way, or by using a totalizer.
That said, on the Aerostar, they were correct down to the gallon. I could see if the plane wasn’t flying level just by looking at the two wing tank gauges – that’s how accurate they were.
