As many of you know, PG&E deliberately cut power to large parts of Northern California last week in a Public Safety Power Shutoff. I don't want to debate the various aspects of that action (there are plenty of other threads for that). I observed something unexpected, related to the AC frequency adjustments in this thread, with my Powerwalls during this incident (which lasted for about 36 hours in our case).
First, with Storm Watch active, the Powerwalls were charged to 100%. When the power cut out, the AC frequency went up to 65.2Hz. The effects this has on common UPSs is well known in this thread, but in the middle of the night we weren't using a lot of power and I calculated the UPSs would run out of battery before the Powerwall SOE got down to 96-97%, which is what I'd thought the frequency shifting threshold was. (Note that for my inverter, a Fronius IG Plus, only 60.5Hz is needed to make it shut down...this was confirmed by Fronius tech support and experimentally verified, ironically, a few hours before the first PSPS warnings went out.)
I found out during this outage that the point at which the Powerwalls would begin increasing the AC line frequency was more like 83%. I wound up charging my Model S at various times to soak up energy from the Powerwalls to keep them below this level, and we didn't observe any other problems for the remainder of the outage (the AC line frequency was nominal at 60 Hz).
I put in a phone call and an email to Tesla Energy support to figure out what's going on.
Today (Sunday 13 October) I tried to recreate this situation with an off-grid test...my intent was to make a video demonstrating this problem. Of course under these circumstances, it behaved as expected...while the AC frequency shift was still much higher than necessary (or desirable) for my setup, it went back to nominal below about 97% SOE. I had the Powerwalls do a couple of cycles around this point, letting them shut down and start up the inverter, and each time it was right about 96-97% SOE. This much better than how the system worked during the actual grid outage.
I'm a little confuzzled as to the difference in the threshold. The only difference I can think of is that Storm Watch was active during the actual grid outage, and not in the tests.
Also of note is that the Level 2 support person my technician was working with, was extremely reluctant to make any adjustments to the AC frequency to solve this problem. If they could just lower the the maximum adjusted frequency below 63 Hz, I'm pretty sure the inverter would still shut down as required, since we already have multiple instances of it doing this at a lower frequency (and Tesla knows exactly what inverter I have since it was originally a Solar City install).
I was trying to get someone at Tesla Energy to help resolve this during the outage, but that didn't happen. Still going to try to press them on this, even though the urgency has passed...I want things working better for future outages. For unrelated reasons, a couple of my existing UPSs need replacing, so I decided to try the Eaton UPSs mentioned upthread for better frequency tolerance.
Bruce.
PS. As a part of troubleshooting the Insteon gear, I called Insteon tech support to see what the frequency tolerance was. The uninformed but trying-to-be-helpful tech support rep would only say "50-60Hz", which was completely useless. Experimentally I'd seen flickering lights attached to SwitchLinc Dimmers at 66Hz but the constant flickering went away below about 64.5Hz. At around 63.5 Hz and below, I observed powerline communication working, although there was still flickering of lights during powerline communication.