Hmm, I see what your saying, and you may be right (although I hope you're wrong). My hope is that it would behave the same way as it does during a sudden grid outage (while solar production is cranking in combo with the batteries at 100%SOC), where it nearly instantaneously changes the frequency to shut the solar down. The kid in me wants to simulate an outage during a sunny day just to test this very scenario. But, alas, I guess I'll take the more prudent approach and try to avoid the situation for now and attempt to get more information from Tesla. You'd think, though, that if the Tesla engineers that approved our plan thought that the mismatch of solar and powerwalls would compromise their performance, or posed a danger to the equipment, that they'd have proposed a different solution.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
5 Powerwalls not working as expected in Maryland
- Thread starter MD38
- Start date
It is true that there is more to the PW frequency modulation than just battery SoC!
I used to think the same thing but my own testing showed otherwise. Check out this thread I started Utility outage simulation data dump.
The thread has more details but the quick version is I queried the Powerwall twice a second (every 500ms) to grab the Load Power (house usage), Solar Power (PV production), Battery Power (power supplied to/from the Powerwall), Site Power (power supplied to/from the utility) and Load Frequency. The SoC isn't on the chart but I did include the raw data if you'd like to plot yourself. I threw the main breaker 10 seconds in to simulate a grid outage then did some things around the house while testing.
In the chart it clearly shows that while the SoC is a large contributing factor (as shown by the orange line sloping down as the system discharges) at high home consumption the frequency drops substantially. The blue "load watts" line is the same as the yellow "battery watts" line and with no PV production or grid availability those will be the same. Those yellow spikes you see is when I was running the microwave - when the load is high the frequency drops!
The reason I believe this is how the Powerwall works, as opposed to a binary 62.5hz/60.0hz function, is because many inverters have the ability to scale back their total production based on the microgrid frequency. My Enphase IQ6s support this and I've configured them to ramp down production between 60.5hz and 61.4hz at a rate of 83%/hz. All this does is help smooth out the charge/discharge curve during times of sun and grid outages.
Last edited:
Just replied but he's not correct - it's based on SoC, production, and consumption at least and in some strange way.
And for sure, do your own test!
Last edited:
One would hope that the Powerwalls would also raise the frequency to force the solar offline when the charging rate approaches the limit. So, the islanded micro-grid would stay up, but it may not charge effectively during peak solar production if you have more solar AC kW than 5kW per Powerwall. Obviously, the cure is to turn off one or more solar inverters when the power goes out during the daytime.
Yes, I assume so too! The frequency function seems pretty complex - even with a quadratic solver and the data points I collected I wasn't able to accurately backsolve it. Clearly there are upper and lower limits too such as never go below 60hz, never above 62.5hz (which is configurable by Tesla). My inverters are rated for almost 7kw so I should be able to test it...in the summer
Oh man, I might just need to test it now-- for science! Monday or Tuesday should be sunny enough here to to reach over 5kw production per powerwall. I'll have to drain the powerwalls sufficiently the night before to ensure they have a low enough SOC by the time late morning brings high solar production. Being a stingy SOB, I'll want to limit the time that the solar array is down during an otherwise high production time.Just replied but he's not correct - it's based on SoC, production, and consumption at least and in some strange way.
And for sure, do your own test!
Thanks for all the additional info. Some of it is over my head, for sure. But it should be easy to understand if it works or not in practice: it works if solar is successfully shut down while remaining on battery power, and it doesn't work if everything shuts down, including the powerwalls. And if I see smoke, it REALLY doesn't work.
Alright, with some encouragement from gpez, I tested it out today, and everything worked as I had hoped and as gpez suspected. To summarize:
What I was testing: To determine whether or not the gateway would shut the solar down in a grid-outage scenario, where battery SOC was less than ~95%, and the solar production was higher than 5kw per powerwall (plus household consumption). We already know that the gateway shifts frequency to shut down the solar when there is a grid outage, and batteries are at 100%, but there was some question (at least to those of us who weighed in on this thread) as to whether the powerwalls might get shutdown too, since they can't handle a charge above 5kw per powerwall.
The test: Simulate a grid outage (by shutting off the main) during a time when solar production is higher than 5kw per powerwall, plus home consumption. To do this, I waited until shortly after noon today, and then shut off the main with battery power having been purposely depleted to about 61%. I also made sure that consumption was relatively low by turning off the heat pump, and making sure there weren't any other high draws. It has been a partly cloudy day today, with sun peaking out occasionally. Shortly before I did the test, the sun peaked out for a minute, and our solar was producing a little above 27kw as reported by the Tesla app. Clouds covered the sun again, but it looked promising that the sun would be back out soon, so I turned the main power off. Solar shut down for about 5 minutes when switching over to batter power, then came back on. So with the grid power off, battery power on, home consumption at about 1.5kw, maybe less, and the PV system producing, the full sun emerged again for about 20 seconds. I didn't get to see exactly how high the production reached, because the solar was shut down, but when it was sunny two days ago at the same time, it was producing over 28kw. So with 5 powerwalls, plus ~1.5kw consumption, there was a threshold of about 26.5kw. I'm guessing that that one burst of sunshine out the production over the threshold by 1 to 3 kw.
Results: The gateway adjusted the frequency in order to shut down the solar, but the powerwalls continued to supply power to the household (and outbuildings). I know that the frequency was adjusted because the error code on my SolarEdge inverters is code 64 "AC Frequency Too High (Max 1)".
So there you have it, folks, the powerwalls worked great, and I don't have to worry about high solar production frying my shiny new powerwalls during a grid outage.
Way cool, thanks for confirming - I hadn't actually tried this scenario myself but expected this result. Good to see empirical results.
If you haven't already may I suggest configuring your SolarEdge inverters to "ramp down" as the frequency increases to avoid the PowerWall limits during off grid sunny days? Check out the "P(F) Power Frequency" feature (https://www.solaredge.com/sites/default/files/application_note_power_control_configuration.pdf, page 6). My Enphase IQ6 microinverters call it "curtailment" or "ramp down" (https://enphase.com/sites/default/f...Considerations-AC-Coupling-Micros-Battery.pdf, page 6). The ultimate goal would be to get your inverters to ramp down production as the microgrid frequency increases, effectively reducing the production maximum to under the Powerwall's inverter limits.
Of course the trick would be to align your SolarEdge settings with how the Powerwall's frequency is configured. That may take some napkin/excel math or some field testing
Very interesting, I wasn't aware of that feature. Seems like that would be a great way to keep solar production at optimal levels when grid is down and batteries are charging. Thanks!Way cool, thanks for confirming - I hadn't actually tried this scenario myself but expected this result. Good to see empirical results.
If you haven't already may I suggest configuring your SolarEdge inverters to "ramp down" as the frequency increases to avoid the PowerWall limits during off grid sunny days? Check out the "P(F) Power Frequency" feature (https://www.solaredge.com/sites/default/files/application_note_power_control_configuration.pdf, page 6). My Enphase IQ6 microinverters call it "curtailment" or "ramp down" (https://enphase.com/sites/default/f...Considerations-AC-Coupling-Micros-Battery.pdf, page 6). The ultimate goal would be to get your inverters to ramp down production as the microgrid frequency increases, effectively reducing the production maximum to under the Powerwall's inverter limits.
Of course the trick would be to align your SolarEdge settings with how the Powerwall's frequency is configured. That may take some napkin/excel math or some field testing
Boatguy
Active Member
Earlier in this thread there was an assertion that during a grid failure the PWs would switch on for a minimum of 5min. FWIW, in my own testing with two new PWs, I switched off the main power for a couple of minutes, then back on and the app reported a "backup" of 2min. I see no indication they were on for 5min.
Hi @Boatguy - I stated that the IEEE standard is defaulted to 300s - your Powerwall could be configured differently depending on the interconnect rules set by your local utility or by your/Tesla's decision if the interconnect rules specify a range. My system is set to 300s based on my experience, clearly yours is different
From https://www.nerc.com/comm/OC_Reliab...rter-Based_Resource_Performance_Guideline.pdf (page 41):
Intentional Time Delay: Some inverters may use an intentional, programmed time delay to return to service following a trip. IEEE Std. 1547-2018 Requirement 4.10.3 requires an adjustable range of the minimum intentional delay between 0–600 seconds, with a default of 300 seconds (five minutes).
Note too that this is an "intentional time delay" - there is always some amount of time needed for the grid connected system to notice the utility connection has resumed, resync the microgrid with the utility, and reconnect.
Last edited:
Similar threads
