Powerwall and solar acting very sporadic in full sun - no charging

[AI1337Tech]

After a rather normal rain storm (batteries in garage, only solar inverter outside) the next morning the powewrwall was not charging when it normally does. During the first few hours, my 10.4 kwh system would spike to 10+kw (nowhere near solar peak time), then drop off completely.

The app showed my house using 5-8 kwh of power - when no one was home and usually uses <0.6 kwh and the power would sporadically go out to the grid throughout the day.

The app also shows very weird behavior with these gaps in time completely missing any data and the powerwall did not charge much all day.

After finally reaching someone at Tesla energy, they said it may be an internet connection thing, but NO, this has zilch to do with internet connectivity since I was able to reach every device on my network during the glitching.

They did push an update to 1.41.2 this morning after being stuck on 1.40.2 for a while, but the glitching still occurs today (but to a lesser degree). System has been flawless since installation 1 month ago.

 

[charlesj]

Feel your pain. May also want to go out and check your smart meter when you are home and watch your app at the same time so what is happening.
Does the battery charging at all? How full is it?

 

[Ulmo]

```````

After a rather normal rain storm (batteries in garage, only solar inverter outside) the next morning the powewrwall was not charging when it normally does. During the first few hours, my 10.4 kwh system would spike to 10+kw (nowhere near solar peak time), then drop off completely.

The app showed my house using 5-8 kwh of power - when no one was home and usually uses <0.6 kwh and the power would sporadically go out to the grid throughout the day.

The app also shows very weird behavior with these gaps in time completely missing any data and the powerwall did not charge much all day.

After finally reaching someone at Tesla energy, they said it may be an internet connection thing, but NO, this has zilch to do with internet connectivity since I was able to reach every device on my network during the glitching.

They did push an update to 1.41.2 this morning after being stuck on 1.40.2 for a while, but the glitching still occurs today (but to a lesser degree). System has been flawless since installation 1 month ago.

I agree that you should diagnose it more, but at first appearance, it looks like something is wrong. I have never seen that failure that you are showing. You should definitely get it fixed.

That is not what Internet failures look like; the PowerWall gateway buffers the data and uploads it when the Internet comes back.

I would tend toward saying it is a technical problem that needs to be fixed. After some more diagnosis like looking at the utility meter when it is happening (whether you learn anything or not), I'd say have them fix it.

 

[charlesj]

...

That is not what Internet failures look like; the PowerWall gateway buffers the data and uploads it when the Internet comes back.
....

Do you know how much it can buffer and upload when internet is back up? Days?

 

[AI1337Tech]

Feel your pain. May also want to go out and check your smart meter when you are home and watch your app at the same time so what is happening.
Does the battery charging at all? How full is it?

It’s usually kept at 5% for backups so drains to then. This weeks I kept it to roughly 20% since weather may have gotten worse. The whole day it only charged +2% when it’s at 100% by 3:30-4:00 ish.

after the sw update it reliably charges now like normal, but the weird gaps are always just before noon and just before 4:00 pm. Super weird! At night we are on battery till roughly 4 am when the grid takes over since my 3 drinks the power all night. I’ll set a recorder on the meter and see how it fares and see what happens at the gap times.

```````

I agree that you should diagnose it more, but at first appearance, it looks like something is wrong. I have never seen that failure that you are showing. You should definitely get it fixed.

That is not what Internet failures look like; the PowerWall gateway buffers the data and uploads it when the Internet comes back.

I would tend toward saying it is a technical problem that needs to be fixed. After some more diagnosis like looking at the utility meter when it is happening (whether you learn anything or not), I'd say have them fix it.

 

[AI1337Tech]

It’s usually kept at 5% for backups so drains to then. This weeks I kept it to roughly 20% since weather may have gotten worse. The whole day it only charged +2% when it’s at 100% by 3:30-4:00 ish.

after the sw update it reliably charges now like normal, but the weird gaps are always just before noon and just before 4:00 pm. Super weird! At night we are on battery till roughly 4 am when the grid takes over since my 3 drinks the power all night. I’ll set a recorder on the meter and see how it fares and see what happens at the gap times.

They will

Do you know how much it can buffer and upload when internet is back up? Days?

I will switch the systems network connection off for an hour and then have it come back up in 15 minute increments for a bit to see how the buffering behaves.

 

[AI1337Tech]

It’s usually kept at 5% for backups so drains to then. This weeks I kept it to roughly 20% since weather may have gotten worse. The whole day it only charged +2% when it’s at 100% by 3:30-4:00 ish.

after the sw update it reliably charges now like normal, but the weird gaps are always just before noon and just before 4:00 pm. Super weird! At night we are on battery till roughly 4 am when the grid takes over since my 3 drinks the power all night. I’ll set a recorder on the meter and see how it fares and see what happens at the gap times.

For sure. I noticed it was reading the grid connection and not showing the “updated x minutes ago” message when the app can’t connect, so the network was definitely good. My nagios server also showed nothing was down to the wan nor to the Powerwall gateways. Tesla said 1-4 business days they will have someone out here, but within an hour of hanging up they pushed the update. Since it’s been sunny with no rain I am not sure if a red herring may be at play since I dunno if the rain may have caused some issue. We shall see!

 

[Ulmo]

Do you know how much it can buffer and upload when internet is back up? Days?

At least, since that's how long Comcast was out in the last outage (almost 48 hours). However, it might be less, since I noticed the Internet came up for a few minutes a few times in that 48 hours (maybe the noise got low enough that my cable modem was able to punch through a signal on an unamplified long distance cable).

 

[bkp_duke]

OP - this is EXACTLY what my powerwalls did on their day of failure:
StormWatch - San Diego County

Very odd behavior in the morning before an outage for wind-related problems. Then I had a lot of issues getting them to come back online (granted, was working through the wife since I was on the road for business).

Tesla was useless in diagnosing the issue.

 

[AI1337Tech]

OP - this is EXACTLY what my powerwalls did on their day of failure:
StormWatch - San Diego County

Very odd behavior in the morning before an outage for wind-related problems. Then I had a lot of issues getting them to come back online (granted, was working through the wife since I was on the road for business).

Tesla was useless in diagnosing the issue.

The interesting thing about this is that we didn’t have any outages. The choppy graph issues didn’t occur until well after the rain ended, and no shutdown had to happen.

 

[JohnRatsey]

I see some similarities to the problem discussed here which appears to have been resolved by firmware 1.41.2. However, that problem appeared to only affect those able to do off-peak charging of the battery.

 

[AI1337Tech]

I see some similarities to the problem discussed here which appears to have been resolved by firmware 1.41.2. However, that problem appeared to only affect those able to do off-peak charging of the battery.

AWESOME thank you! Definitely wonky.

 

[PBBear]

I see some similarities to the problem discussed here which appears to have been resolved by firmware 1.41.2. However, that problem appeared to only affect those able to do off-peak charging of the battery.

FWIW, I think I've figured out what the algorithm is trying to do in the morning. Overnight, between midnight and whenever the sun rises around 7am-ish, the system runs the house on some mixture of battery and grid. (you can argue that during off-peak it shoudn't ever draw from battery, but thats a different issue). With 1.41.2, the system remembers how much has been drawn from the grid since midnight, and tries to pay the grid back as soon as sunlight exceeds consumption. Only when it has fed-in to the grid as much power as was drawn out since midnight, does the system then switch to start re-charging the battery. After that, it seems to be a random mix of feeding the grid, and charging the battery, until at the end of the day it tries to time the battery charging to just 'kiss' 100% by the time the sun sets and the house starts drawing down the battery again.

Check these extended stats from today...from WillisAve 7.370kW | Live Output

The grey line represents cumulative energy drawn from the grid. It rises from 0 at midnight up to ~6 kWh at 7:20am when the sun rises enough to have excess, then the system feeds back towards the grid (grey area pointing down) until at ~10:30am the grey line reaches precisely 0 kWh again, and the system jumps (briefly) to charging the battery a little.
I don't think its bugs - it seems to be an admirably complex dance of predictions warring with unpredictable variabilty of clouds. (today, the afternoon was less variable than the morning, and the system overshot the charging and reached 100% battery a little earlier than it might have planned!)

 

[AI1337Tech]

FWIW, I think I've figured out what the algorithm is trying to do in the morning. Overnight, between midnight and whenever the sun rises around 7am-ish, the system runs the house on some mixture of battery and grid. (you can argue that during off-peak it shoudn't ever draw from battery, but thats a different issue). With 1.41.2, the system remembers how much has been drawn from the grid since midnight, and tries to pay the grid back as soon as sunlight exceeds consumption. Only when it has fed-in to the grid as much power as was drawn out since midnight, does the system then switch to start re-charging the battery. After that, it seems to be a random mix of feeding the grid, and charging the battery, until at the end of the day it tries to time the battery charging to just 'kiss' 100% by the time the sun sets and the house starts drawing down the battery again.

Check these extended stats from today...from WillisAve 7.370kW | Live Output

View attachment 473033

The grey line represents cumulative energy drawn from the grid. It rises from 0 at midnight up to ~6 kWh at 7:20am when the sun rises enough to have excess, then the system feeds back towards the grid (grey area pointing down) until at ~10:30am the grey line reaches precisely 0 kWh again, and the system jumps (briefly) to charging the battery a little.
I don't think its bugs - it seems to be an admirably complex dance of predictions warring with unpredictable variabilty of clouds. (today, the afternoon was less variable than the morning, and the system overshot the charging and reached 100% battery a little earlier than it might have planned!)

We have ours set to "Self Powered" mode so it makes sense that it powers everything during the night (but I agree with the time shifting/time-based mode not doing so). Initially when the system was just dropping out and not charging at all, I first assumed it was warming the battery or something, which is the only time it would pull from the grid besides storm watch, but it didn't do this on other similarly cold days yet.

Since we don't have 1:1 net metering (only earning back .7kwh for what we send out.) I would have to send out ≈1.43 kwh to get credited for the 1kwh I sent out, so I avoid it as much as I can. The algorithm definitely behaves better under 1.41.2 than it did on previous releases, and that choppy graph stays sane as time continues. thanks for the post and link!

 

[eml2]

With 1.41.2, the system remembers how much has been drawn from the grid since midnight, and tries to pay the grid back as soon as sunlight exceeds consumption. Only when it has fed-in to the grid as much power as was drawn out since midnight, does the system then switch to start re-charging the battery.

I'm on 1.41.0. That would explain what I saw yesterday when my PW exported excess solar to grid instead of charging it, which was at 9% SoC at that time. I was on TBC-Balanced. Since I don't have 1:1 netmetering, I would rather have my PW charged instead of paying back off-peak usage. So, after 30 minutes of exporting, I switched back to Self-Powered.

I have a Model X and 3 that are charged overnight during off-peak. If that is the new algorithm for TBC, then potentially my excess solar will not be able to pay back the off-peak usage and never charge the PW. Since my export only is only worth 1/3 of off-peak rate, I would rather have the excess solar to charge PW so that it can cover my peak usage.

As some of us here in the forum suggested before, Tesla needs to give us another mode or just add a few on/standby schedule settings in Self-Powered mode.

 

[charlesj]

...

Since we don't have 1:1 net metering (only earning back .7kwh for what we send out.) I would have to send out ≈1.43 kwh to get credited for the 1kwh I sent out, so I avoid it as much as I can. The algorithm definitely behaves better under 1.41.2 than it did on previous releases, and that choppy graph stays sane as time continues. thanks for the post and link!

What a bummer. A 30 premium to the power company so you can have and use solar. They don't like competition and the regulators are on the company's side.
And, if your rates are like in California, tiered, they sell your excess at a premium on top of it.

 

[AI1337Tech]

Got an update from Tesla. They said that the Powerwall “brain” Was not “communicating” with the inverter, so put itself into protection mode since it couldn’t sense grid phase or inverter status. If it happens again, they will send someone out. Just had another storm so will see once morning time comes, and the panels wake up, what happens.

current firmware seems solid, but ultimate test will occur in the morning.

 

[jboy210]

Got an update from Tesla. They said that the Powerwall “brain” Was not “communicating” with the inverter, so put itself into protection mode since it couldn’t sense grid phase or inverter status. If it happens again, they will send someone out. Just had another storm so will see once morning time comes, and the panels wake up, what happens.

current firmware seems solid, but ultimate test will occur in the morning.

Best of luck and keep us posted.

 

[Ulmo]

Got an update from Tesla. They said that the Powerwall “brain” Was not “communicating” with the inverter, so put itself into protection mode since it couldn’t sense grid phase or inverter status. If it happens again, they will send someone out. Just had another storm so will see once morning time comes, and the panels wake up, what happens.

current firmware seems solid, but ultimate test will occur in the morning.

Something like that is what I suspected: an actual hardware fault of some sort. It could be something as simple as a loose wire or one little faulty part. Let us know how well their techs figure it out and fix it!

Whatever it is, I wish it was redundant and when one failed they'd just come by at their leisure and fix it while the redundant part was in place. I wonder how much that would cost extra. Oh well!

 

[PBBear]

Since we don't have 1:1 net metering (only earning back .7kwh for what we send out.) I would have to send out ≈1.43 kwh to get credited for the 1kwh I sent out, so I avoid it as much as I can.

FWIW, we don't have 1:1 net metering either. In fact, when I called the Tesla hotline (through to the US-based callcentre) and asked about the two TOU modes when they were first brought out, her response was 'You're in Australia? ah, probably those TOU modes won't make sense, since they are built for net metering (assumption, USA-style 1:1 net-metering) - you'll be better off in 'self-consumption' mode."

As it happens though, my feedin tariff (20c/kWh) is more than the off-peak overnight rate (15c/kWh), so after the unit has sucked from grid overnight, then paid the grid back in the first hours of sunlight, according to the bill I should be ahead 5c/kWh.

The way I look at it - by the end of the day, the battery is full and I've exported X to the grid. It doesn't really matter what order it does these two things, or whether it does in in many stages or few, or (as it seems to do now) both at the same time - 3.3 kW to the battery, the remainder to the grid - since here the feed-in rate is constant for the whole 24 hours. If it thinks its optimising for US-style net-metering where the value of feed-in energy varies with the consumption rate, good luck to it, even if it makes no difference here.