New Powerwall 2 Installation and First Simulated Grid Outage Test

[Frankman60]

After a 16 month waiting period, our 2 PWs became operational yesterday! Of course I couldn’t wait to do my first simulated grid outage test which I did today. I wanted to share my results of the test so others can compare and contrast it to other forum member’s experiences. I am presently still running in Self-Powered mode with no option for TIme Based Control yet. How long does it normally take to get access to TBC and is there any way to speed that up?

When the simulated grid outage test began, the PWs were at 53 percent charged, the PV system was generating about 1 kW (lots of dark clouds this morning), and the test lasted about 10 minutes before I flipped the main breaker back on to reconnect to the grid.

1. Some LED lights flickered when the main breaker was turned off.
2. Clocks did NOT reset on kitchen appliances or any other AC powered clocks in the house.
3. No other electronic devices were affected that I observed.
4. I have several Cyberpower and APC UPS devices in the house and none of them intervened during the simulated outage.
5. I received a push notification of the outage on all of my Apple devices that have the Telsa app installed, including my Apple Watch.
6. I turned on the oven, toaster, and hair dryer to increase the house backup load to about 6 kWs with no problems except 6 below.
7. A Panasonic Inverter type microwave oven did not work the same during the outage. It seemed to run at a lower pace and it had a lower pitch hum when running. After the grid power was restored I turned it on again and it sounded normal. Does this have anything to do with the inverter technology in the oven? I also have another GE Profile built in microwave oven and that oven did not seem to run any different during the outage.
8. I flipped the main breaker back on to end the simulated outage and shortly thereafter received another push notification saying that the grid had been restored.

Should I also run another simulated outage test under different conditions to make sure everything is working as it should be working?

 

[Musterion]

am presently still running in Self-Powered mode with no option for TIme Based Control yet. How long does it normally take to get access to TBC and is there any way to speed that up?

24 hours (was for me and that was what manual quotes) or as much as a couple days according to reports here. TBC appears without software update after it monitors your usage pattern for a day or so. Maybe it will be speeded up if you don’t do a lot of grid failure tests during that period

 

[gnumeric]

Should I also run another simulated outage test under different conditions to make sure everything is working as it should be working?

Did your solar turn back on during the outage test? It should turn back on after a few minutes and power your home + charge your powerwalls. I have read posts from people whose solar wasn’t coming back on during outages.

 

[Frankman60]

I believe that the solar stayed on the entire time. I didn’t look at the Tesla app until about 3 minutes after I turned off the main breaker and at that time, both the Solar and Powerwalls were powering the house as shown in the Power Flow page of the app.

QUOTE="gnumeric, post: 3685579, member: 75496"]Did your solar turn back on during the outage test? It should turn back on after a few minutes and power your home + charge your powerwalls. I have read posts from people whose solar wasn’t coming back on during outages.[/QUOTE]

 

[Frankman60]

Yes, that might be true but simulating a grid outage is so much fun

24 hours (was for me and that was what manual quotes) or as much as a couple days according to reports here. TBC appears without software update after it monitors your usage pattern for a day or so. Maybe it will be speeded up if you don’t do a lot of grid failure tests during that period

 

[liuping]

Should I also run another simulated outage test under different conditions to make sure everything is working as it should be working?

There are four main cases to test:
1) solar present and charging the batteries, house running from grid
2) solar present but going to grid, the batteries are discharging to cover house usage
3) no solar is present, batteries are discharging
4) no solar present, batteries not discharging


I'd also test on a bright sunny day, with the batteries full (or near full) to start, to make sure the solar panels shutting down correctly.

 

[boaterva]

Be sure to test with batteries full and solar running. That’s the test to see what solar does to your UPS’s since it has to shut solar down with full batteries. Just had my PWs installed today and was doing some catch-up reading about the 63/66Hz issue. My batteries are at 27% tonight (started at around 10) and expect to be full this weekend for some testing! As liuping says but added point is what does it do to your UPS’s: do they shut down or turn off when power is restored after the small outage.

Also, when we cutover as part of post-install testing, it took 4 minutes for the solar inverter to restart (the countdown timer was actually 250 seconds, I believe). This is maybe for all the micro inverters to reset/sync?

 

[Frankman60]

Thanks for the input on the extra tests to run. I expected that my test would not cause PW to elevate the Hz and cause the UPSs to be unhappy. But I know to expect that when the batteries are near or at full charge and solar is producing. I think all of my UPS devices will intervene with the elevated Hz but I think they will eventually settle down after PWs lower the Hz.

 

[MorrisonHiker]

Thanks for the input on the extra tests to run. I expected that my test would not cause PW to elevate the Hz and cause the UPSs to be unhappy. But I know to expect that when the batteries are near or at full charge and solar is producing. I think all of my UPS devices will intervene with the elevated Hz but I think they will eventually settle down after PWs lower the Hz.

Another test would be to try it when the Powerwalls are down to 5%. I ran into the issue in March where the Powerwalls kept turning on solar but since it was cold, they were putting out a high frequency so the solar kept going on and off for nearly 12 hours while the grid was still down. I tried turning off various inverters/arrays but never did get things to work properly. Tesla sent someone out the next day and they did a firmware update which supposedly fixed the problem. They tested it twice since then and things worked fine at higher states of charge...but I haven't tested it again at 5% yet.

 

[Musterion]

Yes, that might be true but simulating a grid outage is so much fun

I forgot to ask: did Tesla (or your installer) do a simulated grid failure after installation was completed? Tesla had that on their to-do list and said they were looking mainly at interaction with solar (for this reason they wanted to do it before it got too dark, and they were running out of time on a long work day). They also tested correct orientation of CTs, app connection, and the transfer time off-grid and back on-grid.

The suggestions above are great and go beyond what they tested, and are on my list. Incredibly, I’ve already had 3 unplanned grid failures since install and one happened exactly during the county inspection! Inspector was highly amused. We have 2 PG&E full-day (7am-6pm) “planned outages” scheduled for next week so I will use those for testing.

 

[Frankman60]

No, the contractor (not Tesla) that installed my system did not simulate a grid outage to test its operation because it was late in the day when they brought it online and had not completely all of the work. Today they are finishing the work and perhaps will do a grid outage test today. The Powerflow depicted in the Tesla app is correct so I believe they got the CTs right. I already had an existing Neurio monitoring system for the solar and it is still working. Why have two monitoring systems? Because the original Neurio system also monetizes the electricity by allowing me to enter the cost per kWh during each TOU period.

I forgot to ask: did Tesla (or your installer) do a simulated grid failure after installation was completed? Tesla had that on their to-do list and said they were looking mainly at interaction with solar (for this reason they wanted to do it before it got too dark, and they were running out of time on a long work day). They also tested correct orientation of CTs, app connection, and the transfer time off-grid and back on-grid.

The suggestions above are great and go beyond what they tested, and are on my list. Incredibly, I’ve already had 3 unplanned grid failures since install and one happened exactly during the county inspection! Inspector was highly amused. We have 2 PG&E full-day (7am-6pm) “planned outages” scheduled for next week so I will use those for testing.

 

[MorrisonHiker]

2. Clocks did NOT reset on kitchen appliances or any other AC powered clocks in the house.

I forgot to mention earlier that you should check the clocks more closely. The longer the Powerwalls were powering the house during an extended grid outage, the faster our clocks ran (on the range and microwave, alarms clocks, etc.). We didn't notice it right away but after the extended outage, you could definitely see a discrepancy. Since the Powerwalls were outputting 66 Hz instead of 60 Hz, I think the clocks were running 10% faster, so after a 10 hour outage, the clocks were showing 11 hours had elapsed (or something close to that).

Supposedly the new 1.36.1 firmware fixes the 66 Hz issue so YMMV, depending on what firmware you have. We just got 1.36.1 this week but haven't put it to the test yet.

 

[Electricfan]

How long does it normally take to get access to TBC and is there any way to speed that up?

Did you ever get time base control? I got my powerwalls installed two days ago, and today the storm monitoring showed up but still no time based control.

I asked why it isn't available immediately and was given some mumbo jumbo about the powerwall monitoring my system. This makes no damn sense to me - if they are going to let me tell the powerwall when I want it to stop supplying power, why can't they do that on day one? Very frustrated. Today I did at least manage to set the reserve percentage up before 9pm, when my power cost drops, so I'm not discharging the battery when I don't want to. But I'm hoping time based control does this more easily than me setting the reserve percent every single night.

 

[Frankman60]

The test didn’t last long enough for me to see the clock inaccuracy but that is interesting. BTW my PWs have 1.35.2 FW and the frequency did not exceed 62.8 Hz during the test.

Supposedly the new 1.36.1 firmware fixes the 66 Hz issue so YMMV, depending on what firmware you have. We just got 1.36.1 this week but haven't put it to the test yet.

 

[Frankman60]

I got access to Time Based Control about 36 hours after the PWs started working.

Did you ever get time base control? I got my powerwalls installed two days ago, and today the storm monitoring showed up but still no time based control.

I asked why it isn't available immediately and was given some mumbo jumbo about the powerwall monitoring my system. This makes no damn sense to me - if they are going to let me tell the powerwall when I want it to stop supplying power, why can't they do that on day one? Very frustrated. Today I did at least manage to set the reserve percentage up before 9pm, when my power cost drops, so I'm not discharging the battery when I don't want to. But I'm hoping time based control does this more easily than me setting the reserve percent every s

 

[MorrisonHiker]

Did you ever get time base control? I got my powerwalls installed two days ago, and today the storm monitoring showed up but still no time based control.

I asked why it isn't available immediately and was given some mumbo jumbo about the powerwall monitoring my system. This makes no damn sense to me - if they are going to let me tell the powerwall when I want it to stop supplying power, why can't they do that on day one? Very frustrated. Today I did at least manage to set the reserve percentage up before 9pm, when my power cost drops, so I'm not discharging the battery when I don't want to. But I'm hoping time based control does this more easily than me setting the reserve percent every single night.

I believe one reason there there is a delay is because they need to ensure the CTs are installed and reporting correctly. With Backup Only or Self-Powered, I'm guessing the data reported by the CTs wouldn't matter as much. TBC is usually used because there are different rates charged throughout the day so maybe they want to be sure of the reporting before enabling TBC. With our system, they did find a few of the CTs had been configured incorrectly. They made some changes remotely and then changed things again the following day. A day or so later, TBC mode became available fairly soon once things looked right on their end. Hopefully you'll have TBC enabled soon.

 

[cwied]

I think the reason for the delay is because Tesla thinks that machine learning needs to be applied to this problem. The Powerwalls appear to use machine learning to try to predict consumption and production and adjust target state of charge values for the different rate periods in order to optimize the behavior. Ironically a simple, rule-based approach with a small amount of configuration by the user would be more effective than the current implementation.

 

[Musterion]

I think the reason for the delay is because Tesla thinks that machine learning needs to be applied to this problem. The Powerwalls appear to use machine learning to try to predict consumption and production and adjust target state of charge values for the different rate periods in order to optimize the behavior. Ironically a simple, rule-based approach with a small amount of configuration by the user would be more effective than the current implementation.

Do you have any evidence for this? If true I would presume the machine learning part would be done off-site on server side to take advantage of a global data set, which would require network communications back to gateway which you could monitor on your local network. I would say that if you are correct, it would explain why my setup on much older firmware (1.34) doesn’t exhibit most of the problems quoted in this forum on same and higher firmware revisions (although I have plenty of my own problems that others don’t seem to have). Thanks!

 

[cwied]

Do you have any evidence for this? If true I would presume the machine learning part would be done off-site on server side to take advantage of a global data set, which would require network communications back to gateway which you could monitor on your local network. I would say that if you are correct, it would explain why my setup on much older firmware (1.34) doesn’t exhibit most of the problems quoted in this forum on same and higher firmware revisions (although I have plenty of my own problems that others don’t seem to have). Thanks!

The only evidence I have is vague marketing literature (Powerwall Time-Based Control | Tesla Support) and things that Powerwall support has said as mentioned in this thread. I don't think the kind of machine learning the Powerwalls are doing (likely tuning heuristic parameters) would need to be done on the server. If I had to guess, it's probably just a feedback loop based on expected and actual SoC's at different time points.Since the tuning would be individual to each site, I don't think there's any particular reason to do it on the server.

 

[Electricfan]

I woke up this morning (6am) and the App had a new option - Advanced in big white letters, and then "Time based control" in tiny gray letters! Not sure why they do that with the fonts - why not just have "Time Based Control" in big white letters? Anyway, I set it up backwards right away, discovered my error about an hour later and fixed, and it seems to have worked so far today. I did notice that the system started drawing from the grid and solar panels, and did not draw from the powerwall for a brief period about 4pm. But as soon as the powerwall tech support put me on hold, it flipped to drawing from the powerwall and stopped drawing from the grid. No idea what that was about. The powerwall was 100% charged at that point. I'm eagerly awaiting 9pm, when it should stop drawing from the powerwall and start drawing from the grid again. We'll see how it goes.