Gas meter accuracy

I would have thought that gas like water meters rely on a rotating vane as the sensor, that any friction problem over time in the bearings would lead to under-recording, not over recording. Perhaps the impellor gets coated in grime overtime, allowing more force and so speed transmitted to the rotations.

I think the older type used a bellows, the newer ones pass an ultrasonic pulse through the cavity and either use doppler or time-of-flight differences between forwards and reverse direction to calculate flow speed.

In my previous life, I occasionally used flow meters for industrial plant. There were several that used the ultrasonic pulse method, strangely, the dirtier the water, the better the result, as the pulses had more targets to bounce off of. Pure clean water was hard to monitor accurately, and benefited from the old rotating vane technology.

A small source of error is the gas pressure regulator. It should be set to produce 20mBar at the output of the meter with a load. If it's actually higher, then you get more gas in every cubic metre or foot, and if it's lower, then you get less gas in every cubic metre or foot. (Static pressure with no load will be higher due to the way the regulators work.)

The regulators seem to drift over time. They can also become slow to respond, so a large boiler starting up produces a momentary dip in pressure, which in bad cases might cause another appliance to go out.

Temperature also changes the amount of gas in a cubic metre/foot, but that's allowed for on the gas bill - the calorific value is very slightly changed between summer and winter.