The net voltage will not affect the amount of energy that the meter reads if the amount of energy doesn't change. However, the net voltage will change the amount of energy used, depending on the characteristics of the load.
For a resistance load such as incandescent lights, the energy used will be proportional to the square of the voltage. A 10% increase in the voltage will cause a 21% increase in energy. (1.1²-1)·100
For thermostatically controlled heating loads, the energy will not change because with higher voltage. Even though the energy will increase with the square of the voltage while the heating element is on, the heating element will not be on as long because it will get hotter.
Motor loads generally will use about the same energy independent of voltage.
What is "net voltage"? Is something subtracted from the gross to get the net? If the question is simply regarding accuracy, ANSI (if applicable) allows accuracy to vary up to 0.1% for voltages of 90 and 110% of rated.
Suggestion: Changes in terms of statistical fluctuations will be converging to a statistical average, i.e. nominal voltage, e.g. 240V/120V 1phase 3wire system (often offered to residential consumers), will be statistically the same (up to adjustments on the voltage drops that influence statistical results).
My point is that if the nominal voltage (net voltage)decreases and the power drew by the load is constant the current will increase. And some meters use this current, as the only parameter, to move a wheel (using the Eddy Current and the electromagnetic force) which is the motor of the clock that shows the reading.
Industrial meters use PT and CT, and this values are passed to the meter and then the energy is determined.
Power(W)= V x I x cos &
Energy(Wh) = V x I x cos & x h
The meter has to register V and I to determined the exact energy drew. And using only the current, like the old residential meter, it will create a wrong reading. I know that the voltage has to stay close to the nominal value (240/120 V), but if it change to 100 V some load will work with this value and will draw more current. The meter using only the current like parameter will register more energy.
Still a problem with nomenclature here. Nominal or rated voltage is unchanging. Actual voltage does vary. The standard electro-mechanical house meter has voltage and current coils and measures energy very accurately. Never heard of an energy meter built with a current coil only. Such a meter would be highly inaccurate and would violate any standards I'm aware of.
(a larger pdf file)
for an application note:
A Tamper-Resistant Watt-Hour Energy Meter Based on the AD7751 and Two Current Sensors by Anthony Collins and William Koon
It states accuracy of kWhr meter.
