The Tesla roof was up and running for part of the day on August 28th. It has been operating for almost three months. The app does not have a date to date feature so I downloaded performance for each full month, September and October.
The charts show the total KWh the house used each day in the first column of data, the second column is the total solar generated kWh. The Power wall numbers indicate extra power generated and stored, the positive numbers show the days where more power was stored than used, the negative more used than generated. The grid kWh’s were used from the grid when positive and when negative excess power was sent to the grid.
The average home use over the month of September was 41 kWh. The average electricity generated was almost the same at 40.99 so it would seem that the solar roof provided 100% of use in September. But the power wall contribution is negative for the month at 1.93 because the batteries fill each day and use 47% of the stored energy when solar does not meet the house use. The negative draw of the power walls is exactly what the grid contributed for the month, 1.93 kWh.
We would expect October to produce less electricity since the sun is lower and the roof is on the North side of the house.
The average use was 37 kWh and the average production was 27 kWh. So in October we produced 70% of our electricity use and the grid provided almost 30%. The power walls drew 1.83 kWh more than the change, that 5% is included in the 11 kWh drawn from the grid. One day seems to be an anomaly with the house using 133 kWh. I have no idea what happened that day that created that large increase.
My theory was that the power walls would charge with the sun only and return that power after the sun set. But it appears that they also charge from the grid. If a storm threatens they do not use power in the house but save it for a potential outage. The amount of power the walls use each day is a little more than used to charge them so it appears at least in the fall they do not contribute to a lower use of the grid.
I put all these averages into a circle graph that shows the relationship.
As winter approaches the solar roof will produce less power. I expect the numbers will again begin to climb slowly after the winter solstice.