Wanted to give you a couple of realistic answers on a couple of your questions.
-
Sense Accuracy - For me Sense is very accurate, and should be for most people unless they have data dropouts or inaccurate meter readings (more on that in a bit). Sense should be within 1% - very specific analysis here. And detailed techniques for comparing here.
-
Utility Company Polling - the unfortunate truth is that most smart utility meters send every 15 min measurements back to the mother ship via an ad-hoc mesh network. If your meter is a lot of hops away from a collection point or has to go through some congested nodes (other meters that are weak links), you won’t always get accurate readings. Your utility will then “Calculate” your in between 15 min reads so that the aggregated usage (delivered and received in the case of solar), add up to the correct amount, but the distribution over time is likely to be very different. I have done some work with another Sense user who is on NV Energy (Nevada) and gets data from his utility that shows which 15 min reads are “Actual” and which ones are “Calculated”. He’s only seeing 15% of his reads as “Actual” and 85% are “Calculated”. The “Actual” reads are within 0.5% of Sense. The “Calculated” reads are all over the map. My parents have the same utility and also have solar and live about 10 minutes from this other Sense user, but are seeing 98% “Actual” reads. This user has a real axe to grind with his utility since the “Calculated” reads are shortchanging him in his TOU billing credits. More here.
-
Solar Production - On thing most solar users will see with Sense. Sense measured solar production comes up about 3-5% less than the inverter measurement. Yet the Sense net result lines up more accurately with utility net data (“Actual” reads that is), than using Sense total usage minus inverter solar
production. My theory is that the inverter value includes the energy that much be used to power the inverter, while Sense only see the net resulting produced power.