Best Answer
Complicated answer sorry : I will report some anomalies which I detected by setting the audio of a computer to listen to 1 second samples of mains Voltage and Current waveforms for up to a week at a time, and then to do some basic maths which can be written P(t)=I(t).V(t) to get local net Power waveform P(t) at the meter tails. Textbooks say that mains AC is 50Hz sinewave, and that fridge motors can be a few milliseconds late, describable by "power factor". Mains Voltage V(t) here is a regular wave not exactly like a 50Hz sine, with a peak to root mean square (rms) ratio of 1.37 instead of 1.41. My solar panels on the roof generate a current of similar shape to V(t), while fan heaters, toasters, ovens and 'plain' loads use an I(t) of similar shape. That is, power generated from the solar panels is in the right waveform to cancel plain resistive loads. Whenever I switch on a computer, a modern appliance such as a television, or an energy saver lightbulb, that uses a different waveform, which can be so bad that I made up the phrase "uglywave" to describe what a sample of general household use minus solar generation can look like. LED energy saver bulbs which I tested all draw "twospike" waveform.
In my house, power imported in a sensible waveform for the cooker and resistive loads does agree between my homemade unaccredited waveform metering, and import paid for according to the ordinary meter outside. Power imported for uglywave component of uses does not seem to agree, possibly because billing meters were designed to be "same as old spinning disk billing meters", and those are not quite right with uglywaveforms.
My opinion is that there is nothing wrong with your smart meter; it is doing what billing meters have always done, and there is nothing wrong with your solar display, that is showing you a correct P(t)=I(t).V(t) averaged over the 20ms cycle as could my contraption. That is, all of your discrepancy is from all billing meters doing something different to P(t)=I(t).V(t) except in the special case of perfect resistive loads such as cookers toasters immersion heaters and plain old lightbulbs. You can test this by switching off at the wall or RCD household circuit breaker all TV on standby, mobile phone chargers, and things like that, while you are doing a 30 minutes in the oven cooking job, and check for agreement then between your solar display and your billing meter. Half an hour of my oven uses nearly 1kWh, so that is the smallest testable job for this comparison.
So, does your billing meter make more sense if you remove off all of those newer electronic gadgets and look at only oven minus solar generation ?
If it is a long way off then perhaps you should ask questions at GroWatt.
