On Tuesday, I had a good solar day, hitting 43.6 kwh on my 11.28 system when I checked before bedtime, or so I thought. Looking at numbers today, which will be a great day, I scrolled back in time and the Tuesday number now shows as 41.6. Have others seen the app retroactively change the solar production? Anybody know the cause?
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Changing Solar Numbers
- Thread starter dhquigley
- Start date
Yep - I have mentioned this before. It is always 3 days in the past, and it seems to drop in the morning (or overnight - I have not attempted to figure out exactly when) and get back to its correct value by the evening. (For reference, I got 27.9 kWh Tuesday, but the app showed 17.0 this morning and now shows 20.6.)
As a side note, since you are in the area. even though Wed was seemingly just as clear, I only got 27.0 kWh, which I think might partly have been due to the warmer temps. (There might also have been just a bit of haze.)
As a side note, since you are in the area. even though Wed was seemingly just as clear, I only got 27.0 kWh, which I think might partly have been due to the warmer temps. (There might also have been just a bit of haze.)
Really helpful, thanks. Noticed the same phenomenon. Think today could be a banner day if it stays clear.
Curious - have you observed this just in the Tesla app or in the API data as well?
same for me: Tuesday: 31.2 kWh, Wednesday: 29.3 kWh
Yeah - finally making up for a dreadful Jan/Feb production-wise, even compared to PVWatts estimates.
FYI, the totals also seem to affect weekly and monthly numbers (and probably yearly/lifetime.) So, if you ever want monthly numbers right when the new month starts, wait until the 4th, check at night, or remember to adjust for the low number.
It is only in the app for me - the API data that I pull is all instantaneous, and so it is not affected. That is actually one of the ways I noticed it was doing some sanity checks on data from the API and noting the discrepancy. Also, if I use the option in the app to download the data, that appears unaffected, so the bug appears to be specific to the calculations in the app, and really (at least for me) only ever affects 3 days ago. Why there is something special about how they handle 3 days ago, I have no clue. I could understand today or yesterday, but 3 days ago is just weird.
Good to know - thanks. I've switched to pulling historical data from Tesla's energy API ( State And Settings ) and have only been looking at the app for current day info. I had not noticed the 3-day bug in the historical data I've pulled over the last month or so.
The API - which the app likely also uses - requires a request for each day and returns a list of solar_power and grid_power values for each 15-minute interval (for me - no powerwalls). I have no idea what's special about 3 days, but maybe it's some kind of caching bug in the app where the intent is to immediately grab the 3 most recent days of data but they accidentally only get 2+ days.
Glad you said that about Jan and Feb. I was running pretty far below PVWatts (like 20-30% below) and beginning to think I had greatly overestimated (which I may have anyway). Seem to be more on pace this month.
In terms of the local area crew, really liked my solar roof crew, although the head of it just got promoted to quality. The initial roofing company was good, although it seems like they do not get every job. Come to think of it, the Powerwall team was also very good. In this area, your crew can come from Baltimore or Beltsville. Mine was from Baltimore. WJGJR is right though. The crews are fantastic. Communications less so, but I will say I feel like they have worked on that. I would suggest that once you find a responsive person, you tie yourself to them forever, as best you can. I will echo that everyone seems very concerned about getting this right, in ways that I have found surprising. Big fan.
In terms of the local area crew, really liked my solar roof crew, although the head of it just got promoted to quality. The initial roofing company was good, although it seems like they do not get every job. Come to think of it, the Powerwall team was also very good. In this area, your crew can come from Baltimore or Beltsville. Mine was from Baltimore. WJGJR is right though. The crews are fantastic. Communications less so, but I will say I feel like they have worked on that. I would suggest that once you find a responsive person, you tie yourself to them forever, as best you can. I will echo that everyone seems very concerned about getting this right, in ways that I have found surprising. Big fan.
I have noticed if I select all 4 icons for grid, house, powerwall, solar. And the. I go back by using the arrow keys a few days and forward it seems to refresh and give me good numbers. I am not checking it against the API but I know when it’s raining and it isn’t so I know it was off. It also doesn’t seem to update peak numbers until the peak has ended. I would attribute it to the app sleeping on a phone and not updating correctly. But if you cycle through a few days it does a refresh. YMMV...
Does anyone else see glitches in the Tesla app? About once a week, the data is wrong. I have a 7.6kW inverter so anything above 7.6 is impossible.
When I download the data, it matches with the graph so it is also wrong. Then I download SolarEdge data and it is correct and doesn't have the glitch.
The total for the day is accurate though.
When I download the data, it matches with the graph so it is also wrong. Then I download SolarEdge data and it is correct and doesn't have the glitch.
The total for the day is accurate though.
@wjgjr has mentioned in this and in other threads that he has noticed a recurring problem with data from 3 days back. I haven't noticed that myself, but I have noticed that something changed (or broke) with the API recently. I'm able to get good numbers from the API for March 14th on, but prior to that the API I'm using ( https://owner-api.teslamotors.com/api/1/energy_sites/<site-id>/calendar_history?kind=power&end_date=<date> ) only returns data for the first hour of each day. It used to return the data for each 15-minute interval of each day but now only returns the first 4 intervals for March 13th and earlier. I'm curious whether anyone else using the API has noticed this.
I think another user reported something like this (though I do not recall whether it was resolved.) Looking at your first image in particular, it is as if the graph mapped some of your production to the wrong time, causing a spike follwed by a dip (this is assuming no cloud/shading right at that time.) And the second graph is clearly well above your inverter's capacity. As I recall, for the other user, the spike happened to be right on the hour, so I speculated that possibly the app was somehow losing the minutes and assigning it to the hour bucket. (The app seems to show data at 5-minute increments, so it presumably does some binning of the data.) I think his was also consistent enough to be able to watch the next day at the correct time to see the behavior in real-time (which would let you see if the spike appears immediately or only a bit later at the time the dip appears.) In any event, if it continues to occur, you will probably need to talk to Tesla to try and debug it.
Today I figured out why I was getting bad numbers from the energy sites calendar history API. I had not properly set up my Python date range objects to include time zone information, so the switch to daylight savings time broke any queries for dates prior to the switch. I'm just a Python hack (spend most of my time with Java) - but it's still embarrassing that I created this problem.
