November 2026 Solar Report
- This report covers my solar array, two Powerwalls and an EV on Long Island, New York.
- The figures are my own records. Scripts compute them, fixed at 9 November 2026.
- Each page's footer names its sources.
One day from the record, dawn to dusk
- These are 15 days from my record, each picked by a rule its heading names. House use includes my EV's charging.
- The power, battery and temperature meters show the mean over the 10-minute window from that time, from my Powerwall gateway, Wall Connector and weather station.
- Rain says whether the station's gauge rose in that window; Rain today is the day's total. A dash means not recorded.
- Power bars run from 0 to 13 kW, temperature from 0 to 110°F and the day's rain from 0 to 2.78 in; the battery bar runs from empty to full.
| The year before solarBefore8 Nov 2023 – 7 Nov 20242023–24 | Year 1Year 18 Nov 2024 – 7 Nov 20252024–25 | Year 2Year 28 Nov 2025 – 7 Nov 20262025–26 | |
|---|---|---|---|
| Solar generation | no system | 16,927 kWh362 of 365 days | 15,173 kWh |
| Generated, as a share of household consumption¹ | no system | 88% | 71% |
| Drawn from the grid² | 15,414 kWh310 of 366 days | 9,300 kWh364 of 365 days | 10,909 kWh |
| Grid failures | no system | 4 | 12 |
| Electric bills | $3,854 | $855 | $1,699346 of 365 days billed |
| Solar loan | none | 10 of 12 payments, not totaled | $2,69512 payments |
- No system yet, so nothing measured.
- ² Before solar, the figure is the utility's metered usage, from the bills, the usage sheet and the daily meter.
- After, it is grid import as the Powerwall gateway measured it, not net of export, and before 19 May 2025 integrated from its power gauge.
- The loan is its own line, not added to the bills.
- Array
- 36 × 425 W panels, 15.3 kW, on four faces of a hip roof at 23°
- Storage
- 2 × Tesla Powerwall 3
- Energized
- 8 November 2024
- Record
- 729 days of daily totals, November 2024 to November 2026, 32,158 kWh generated
Results
1. Delivered output against rated capacity
- My array delivers 42% of what 36 panels rated 425 W would produce facing due south at 37° under a clear sky.
- Roof geometry accounts for 20.1 points of the shortfall.
- Conversion and shading losses account for 15.5.
- Weather, including panel temperature, accounts for the rest.
- I compare measured daily generation with a clear-sky plane-of-array model for the four roof faces.
- The conversion band is the shortfall on the five best days since the counters began.
- Those days are 16 October 2025, 6 October 2026, 18 October 2026, 22 October 2026 and 23 October 2026.
2. My credits barely reach the discounted overnight band
- The utility's optional overnight rate discounts 10pm–6am.
- It nets net-metering credits within bands, not across them.
- Over the last 430 days my array exported 24 kWh into that band.
- My house imported 6,403 kWh from it.
- So 350 kWh of credit would strand in bands with no use left to offset.
- The grid meter is split by band under each published rate.
- Federal holidays are not modeled.
3. Generation offsets most, not all, of household load
- Over the twelve months from 10 November 2025 to 9 November 2026 my array generated 71% of my household use.
- The hero's figure is solar year 2, a slightly different window.
- Over the 338 of those 365 days both grid counters recorded, my house was a net importer of 4,640 kWh.
- Over the 391 days of my EV's record, from 15 October 2025 to 9 November 2026, its charging was 20% concurrent with generation.
- Under flat-rate billing that does not matter: the bill is struck on net import.
- So each added kilowatt-hour of use is one imported.
- Counter series give load and grid flows. Generation spans both eras.
- Concurrency is measured in 15-minute bins.
4. Islanding duration exceeds utility outage duration
- My system islanded 17 times: 16 grid failures and 1 intentional off-grid test.
- They total 96 minutes.
- The grid failures account for 93 of those minutes.
- About 13 minutes of that is the utility's.
- The rest is the 300-second Enter Service delay that IEEE 1547 requires.
- The shortest grid failure lasted 299.9 seconds, within the 3 seconds of jitter in the gateway's edges.
- The gateway's grid-status changes, checked against grid power.
5. A nearby tree is measurable from generation data
- A tall tree stands west of my house.
- My array's generation at mirrored solar positions puts its upper edge at 35–40° elevation near azimuth 255°.
- An overhead photo puts its crown's center farther north, toward 290° and about 67 feet from the middle of the house.
- East and west output are compared at matched solar elevation and azimuth offset.
- No page charts this; the working is in the methodology.
Prior work
- I have published periodic write-ups of this installation at evan.wtf.
- Result 2 restates a mechanism already described in the one-year write-up.
- What this report adds is the measurement.
Method and limitations
- Each page is a single HTML file that holds its own data. No external resources load.
- Series are written on change, so a constant value produces no rows. Absence of data is not zero.
- The record has two eras. From 19 May 2025, a day's energy is the peak of the meter's cumulative counter for that day.
- Before then, a day's energy is the power gauge integrated by the trapezoid rule, never across gaps longer than 30 minutes.
- The EV's wall connector writes only while it charges, so each reading holds until the next. A gap longer than 30 minutes counts as no charging.
- The 729 days of daily totals span both eras. Days the recorder did not cover are left out, not counted as zero.
- The split of my EV's charging between sources is an estimate by proportional split, not a measurement. The EV page describes it.
- Federal holidays are not modeled in the time-of-use bands.
- This report describes one installation. It is not a product comparison and contains no recommendation.