PowerWalls didn't kick in for power failure due to "grid out of compliance"?

[tmc42]

I have three PowerWalls, a Gateway, and ~8 kw solar installed by Tesla in late 2020. The PowerWalls were at about 52%. I momentarily lost power and all the lights went out! Twice. First time was about a quarter second, the second time more than a second, and a brief flicker too. This used to happen a few times per year before I got my PowerWalls. Now, I am used to the PowerWalls kicking in when we lose power, and possibly seeing a ~1/60 second flicker, but not this time. The app shows three events (duration "a few seconds") in Backup History, but no errors.

Tesla looked at the logs and said "the grid was out of compliance" so the gateway couldn't switch to the PowerWalls. Pressing for more information, she said "the grid impedance was too high, 0.6 ohms, and it needed to be below 0.27 ohms". Basically, my whole-house backup that is supposed to take over when I lose power, can't, and Tesla is pointing their finger at the grid, saying the "grid was out of compliance" multiple times. I was told to contact my electric utility to resolve this. She said this is not very unusual, and happens a lot. Has anyone ever heard of this? And how do you resolve this so the lights and fridge stay on at the next power failure? (I'm expecting my utility, PG&E, to tell me to contact Tesla, and I don't want to be stuck in the middle).

 

[Vines]

I have three PowerWalls, a Gateway, and ~8 kw solar installed by Tesla in late 2020. The PowerWalls were at about 52%. I momentarily lost power and all the lights went out! Twice. First time was about a quarter second, the second time more than a second, and a brief flicker too. This used to happen a few times per year before I got my PowerWalls. Now, I am used to the PowerWalls kicking in when we lose power, and possibly seeing a ~1/60 second flicker, but not this time. The app shows three events (duration "a few seconds") in Backup History, but no errors.

Tesla looked at the logs and said "the grid was out of compliance" so the gateway couldn't switch to the PowerWalls. Pressing for more information, she said "the grid impedance was too high, 0.6 ohms, and it needed to be below 0.27 ohms". Basically, my whole-house backup that is supposed to take over when I lose power, can't, and Tesla is pointing their finger at the grid, saying the "grid was out of compliance" multiple times. I was told to contact my electric utility to resolve this. She said this is not very unusual, and happens a lot. Has anyone ever heard of this? And how do you resolve this so the lights and fridge stay on at the next power failure? (I'm expecting my utility, PG&E, to tell me to contact Tesla, and I don't want to be stuck in the middle).

Impedance issues can be traced to faulty utility connections at times but 0.6 ohms is way out of spec.

I would reach out to the utility company and report this issue. Also if possible get an impedance reading of your existing service condition. Your installer should have the tools for this, though it's not usually checked for less than 5 Powerwalls.

Your installer could set you in high impedance mode, not sure if they have already. It would be interesting to know what your impedance is under normal operating conditions.

 

[miimura]

This makes no sense to me. Why do the Powerwalls care about grid impedance when it's trying to transition to Island Mode? The grid is being disconnected.
When you want to talk about Powerwalls balancing grid flows in grid interactive mode, I can understand why the impedance matters.

 

[Vines]

This makes no sense to me. Why do the Powerwalls care about grid impedance when it's trying to transition to Island Mode? The grid is being disconnected.
When you want to talk about Powerwalls balancing grid flows in grid interactive mode, I can understand why the impedance matters.

You are correct of course that the rise in impedance alone wont prevent switchover and going off grid should have worked. If activated from the app it probably would have.

I just know that impedance issues will cause strange behavior. I have no idea why but impedance is something that we are checking and keeping it under the specified values from Tesla. A sudden rise in impedance could be a loose connection anywhere in the loop to the transformer and the system likely has safety interlocks that are intended to prevent thermal events if anomalies are detected.

 

[tmc42]

I followed the advice of Tesla Energy and Vines, and reported this to my utility. They are still dealing with getting other people’s power restored, so I told them this was low priority. They are sending someone out to investigate this on Friday.

I told the representative to please make sure that the work order says “high grid impedance” and that it measured 0.6 ohms, so they know why they are there. The representative was very nice, but needed me to spell “high”, “impedance “, and “ohms”.

 

[tmc42]

A PG&E (our electric utility) worker came to the house and immediately escalated this to PG&E's Power Quality Department. Someone from that department called an hour later and took the time to talk with me and answer my questions. Overall, good customer service from PG&E.

When I mentioned high-grid impedance he said they get these calls a lot from Tesla Powerwall customers and explained "Tesla is correct, we are not in conformance to Tesla's specifications". He added that they are in conformance to National Electrical Code (NEC), and they need to be under 25 Ω, and they are significantly under based on Tesla's logs. He said that if Tesla is reading 0.6 Ω that should be OK, even if Tesla says it is not. Tesla said for three Powerwalls it should be <0.27 Ω, a two-orders of magnitude difference between Tesla's specs and NEC.

He pointed me to Tesla's "Electrical Equipment Sizing and Overcurrent Protection" Powerwall documentation that shows Tesla's specs for Maximum Grid Impedance in Table 2. Tesla's requirement is 0.27 Ohms for three power walls, but he said since NEC allows for up to 25 Ω there is nothing for PG&E to do. He suggested contacting Tesla to make changes to settings for the Powerwalls to kick in. Basically, Tesla points a finger at the electric utility, and they point the finger at Tesla, leaving the customer stuck in the middle!

It is surprising that there are two orders of magnitude difference between Tesla's specifications and NEC. I also don't understand why the Powerwalls always kicked in in the past, but won't kick in now, and Tesla blames the electric utility.

 

[skepticcyclist]

I still don't understand how grid impedance would prevent the PW from turning on when they are disconnected from the grid during an outage

 

[tmc42]

It gets more confusing. Tesla emailed "Our team has reviewed your Powerwall system and based on the signals and the information provided the system functioned as intended". I don't understand how having the lights go out when we lose power is functioning as intended.

I called Tesla Powerwall support again, and the agent said the Powerwalls did not kick in because, allegedly, the house was drawing too much current when the power went out momentarily (a couple of times just before 8pm). She said this is a safety issue, and the 3x Powerwalls won't deliver more than 5kW total (I thought it was 5kW each) and their new analysis shows that at the time of the events the house was drawing more than 34 Amps. This makes no sense. The graph from the Tesla app shows the home never went close to 2 kW the entire evening. We had almost no load except the blowers from our gas heater turning on and off, and some LED lights. We weren't using Air Conditioning, we weren't charging our cars, not using our electric ovens — a very low load. She insisted we were drawing more than 5kW and that's why the Powerwalls never came on. I guess she was also telling me that the Tesla app's energy graph (attached) is completely fabricated too, and I can't trust it.

BTW, my wife set the Powerwall reserve to 52% because a storm was coming and she wanted enough energy in the Powerwalls. We had just enough sunlight to reach that reserve. So at the time we lost power, we had sufficient charge for the Powerwalls to come on, but they didn't.

I used the Tesla app's "Go Off-Grid" to further troubleshoot with the Tesla agent. But the lights didn't even flicker. The Powerwalls took over perfectly. I asked how we further troubleshoot this. She said we just did all that can be done to troubleshoot the issue, and the system is working correctly. I told her the Powerwalls not kicking in when we lost power is not working correctly. She said she couldn't help me any further.

Any suggestions for how to troubleshoot this? Has anyone else had Tesla PowerwallSupportNA analysis completely uncorrelated with reality and data from their own app?

 

[miimura]

I have also had a power outage and the Powerwalls failed to power the house. My house was also under 2kW draw at the time of the outage and I have two Powerwalls. The app said something about too much load. This was at the beginning of what would turn out to be a 6 day outage last March. Luckily I was able to turn off most of my load breakers and get the system to start by pushing Go Off-Grid on the PW web interface.

 

[power.saver]

While the Powerwall (Gateway really) is capable of detecting grid failures, it does so by monitoring the voltage and frequency upstream of the main contactor which is what's used to disconnect from the grid. Usually this work, but sometimes it does not.

By using the Go Off Grid function, you are first notifying the Gateway that you are about to disconnect, and it can get ready for this mode. And when it does open the main contactor, the disconnect is clean, meaning it totally disconnects from the grid and only your house loads are present. It is not a very good test of a true grid failure.

The next best test is to open your main breaker at the meter and before the Gateway. This doesn't give the Gateway any advance notification, but like the main contactor, it cleanly disconnects your house from the grid. I suspect if you do this test, the Powerwall will power your house, but you will see a flicker of the lights in the time it takes to recognize the power is out and to takeover the load.

When the grid goes down, it's not a clean disconnect. It might be clean enough that all you'll see is a flicker of the lights, but it could be much worse. The grid can fail in may ways, from a lineman removing a fuse to your transformer, to a tree falling on distribution lines blocks away, to the generating station going off-line. In all these scenarios, the power on the grid can be way out of spec for frequency and voltage, which can make it appear as a giant load on your Powerwalls before the Gateway has a chance to open the main contactor.

It is possible that the Powerwall started to try to supply this grid failure load and, even momentarily, exceed the 5kW per Powerwall limit. This sends a shutdown command to protect the inverters, and then you have no backup. This very brief surge would not show up on the graphs because it was so short, but it was recorded in a way that Tesla can see that it happened.

There's not much you can do about this. Grid failures are notorious for damaging electronics, and having the Powerwalls shutdown probably saved them from further damage.

What you should do is become familiar with restarting procedures, so when this happens, you know how to restart them. Disconnect all loads, use the Go Off Grid function, and slowly reconnecting usually works. Having a 12V battery and cable ready in case your Powerwalls need to be jump started after being fully discharge during an extended outage is also a good idea.

One preemptive measure you can take is to use the Go Off Grid function if you suspect the conditions might cause a grid outage. Once the main contactor is open, you are isolated from the grid. After the threat of a grid failure has past, you can restore the grid connection.

 

[offandonagain]

She said this is a safety issue, and the 3x Powerwalls won't deliver more than 5kW total (I thought it was 5kW each) and their new analysis shows that at the time of the events the house was drawing more than 34 Amps. This makes no sense. The graph from the Tesla app shows the home never went close to 2 kW the entire evening.

The 5kW limit doesn't make sense to me either, but I'd believe the 34A part -- note that the graphs in Tesla's app have 5 minute granularity, so the "power" values there are a bit deceiving. Your home load in particular seems bursty, so it's possible those peaks are a fair bit higher than 2kW. Tesla has better visibility internally; you can also get it with Powerwall-Dashboard or just by looking at the app during those peaks (the live data is ~5 second granularity).

 

[tmc42]

@power.saver, thanks for the detailed explanation. The only thing that does not make sense is that for the prior 30 events the Powerwalls took over. And I can buy the argument that during a grid failure the "grid can be way out of spec for frequency and voltage", and maybe this caused a high load and explains the Powerwalls failing to take over. But this happened twice in one night where I was literally in the dark. So the odds of the Powerwalls seeming to take over 100% of the time for three years (30 events), and failing twice in one night seems very low.

Speculation: Maybe the latest firmware is not as robust as it used to be for some edge cases on messy grid failures? Or maybe it is "an improvement" (more conservative, preventing high-loads for too many milliseconds). If it is a firmware regression where the Powerwalls fail more often when they should not, I'm sure Tesla can download enough logs, especially after storms, and do data analysis to see if this is the case.

@offandonagain, thanks for the pointer to Powerwall-Dashboard. I may try it out if I have some spare time.

 

[power.saver]

@power.saver, thanks for the detailed explanation. The only thing that does not make sense is that for the prior 30 events the Powerwalls took over. And I can buy the argument that during a grid failure the "grid can be way out of spec for frequency and voltage", and maybe this caused a high load and explains the Powerwalls failing to take over. But this happened twice in one night where I was literally in the dark. So the odds of the Powerwalls seeming to take over 100% of the time for three years (30 events), and failing twice in one night seems very low.

Speculation: Maybe the latest firmware is not as robust as it used to be for some edge cases on messy grid failures? Or maybe it is "an improvement" (more conservative, preventing high-loads for too many milliseconds). If it is a firmware regression where the Powerwalls fail more often when they should not, I'm sure Tesla can download enough logs, especially after storms, and do data analysis to see if this is the case.

@offandonagain, thanks for the pointer to Powerwall-Dashboard. I may try it out if I have some spare time.

Certainly firmware changes could affect how well they respond to grid failures. Since we don't know exactly how they do the detection, or what new shutdown protections they put in the code, this could be a factor. Or the previous grid events were not so severe that the grid load caused a shutdown. If the two recent events occurred during the past storms, there were a lot of outages caused by trees falling on wires, which have some of the worst effects on grid power as it is almost always not a clean disconnect. Without expensive power line monitoring equipment, it's very hard to say conclusively what happened during a grid failure.

 

[mongo]

@tmc42
Speculation: I think something is getting lost in the communication with the PoCo. The 25 Ohm spec is usually related to grounding system impedance where, if one ground rod does not achieve 25 ohms, a second must be installed (but at that point, there is no requirement for impedance). 25 Ohms can't be related to the incoming voltage path since just 10 Amps would yeild zero volts (unless on the high voltage side of the transformer, but that's should be out of scope to the Powerwalls).

0.6 Ohms sounds like an active power path impedance. This would be detected by the Powerwall as either a voltage vs current sag value across the 240V or, more likely, a differential between the two 120V split phases.
With one PW @ 5kW that's 21 Amps * 0.6 Ohm = 12.6V which is a 5% drop over 240V or 10% on a single phase.
Three Powerwalls tracking that 5%/10% requirement at three times the power (current) would need 1/3 the resistance.

I'd be checking for a neutral or phase conductors from the transformer to the meter that was failing. Or meter to panel to Powerwall.
If you have a couple multimeters or kill-a-watts or such, I'd check your phase voltages, especially if you can differentially load your service with a lot of 120V loads on the same phase.

 

[cali8484]

Tesla looked at the logs and said "the grid was out of compliance" so the gateway couldn't switch to the PowerWalls.

That makes no sense. It would make more sense if Tesla said the gateway switched to the PW's due to the log message. The grid impedance thing looks a like red herring.