Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Long term regular use of home PowerWall 2 system with solar

This site may earn commission on affiliate links.
I use “self generation” load profile for my PowerWall system, since I want to shift my solar use to cover as much use as possible (generally speaking; time of use rate plan optimization would also be nice, as well as some grid services assistance to the utilities, when those features become available to home systems).

Since we have different types of seasons in California microclimates than in the East, for sake of discussion, I define summer as those times of year there is more solar output from our solar system, and winter as those times of year with less solar output.

Since my solar production is insufficient for our home needs in winter, I chose a minimum self use battery level which averaged around 50% state of charge for the day for winter (for the purpose of keeping longevity of the Lithium Ion batteries as long and good as easily possible), which meant a much higher minimum charge % than during summer. For us, that was somewhere between 30% and 35%.

This last week, solar output has substantially recovered from the winter mailaise, and I have down-adjusted the minimum state of charge to 20%, once again targeting 50% average state of charge.

Someday I’d like it if this were automated. For now, this is an easy way to adjust the charging every few months or so. I’m triggered to review it whenever I sense changes in solar conditions or decide to review the system.
 
I use self consumption, usually keeping 30-50% as backup depending on weather. But if I was given the option to keep the Powerwalls at 80% SOC, I might be inclined to leave them there just to avoid the efficiency losses and avoid cycling the batteries too much. As is, since we don't have that option, I reason that 30-50% backup reserve is a decent way to keep the batteries in the "sweet spot" while still having plenty of reserve during outages.

I've not made any attempt to measure the efficiency loss of the Powerwalls, but it sure seems like it's way higher than the stated 10%.
 
  • Helpful
Reactions: EVortex
I have solar, and use it in self consumption. I allow it to fully discharged to 0% if I know next day is going to be sunny. Being a lithium-ion battery, it doesn't really matter if it is discharged to 0% or 50%. In fact, it never actually discharged to 0% because the battery will be dead if that is the case. I trust PW's battery management system to take care of it.

The efficiency loss of my PW seems to be around 15%. I figure it out from the Tesla app. On a partial sunny day, I start my PW with 0%, charge up during the day, and use up all the charge before midnight. Then, get the kWh for "From Powerwall" and "To Powerwall" on that day. It comes up to around 15% loss.
 
I don't have a Powerwall (yet), but at a minimum I want to be able to do is this:

1. Set a minimum charge level (so that I always have some energy if the power goes out - you can already do this).
2. Instruct the Powerwall to only discharge during peak TOU periods.

All other times the Powerwall should charge from solar instead of exporting power to the grid.

Bonus points for the following:
1. Recharge back up to a specified percentage above the minimum during super-off-peak from the grid.
2. Disable charging of Powerwall during peak TOU periods - this would allow maximum export of solar for credit.

With some advanced predictive analytics, you could further tweak the above to maximize revenue.
 
I bought mine to use it, so I'm solar with self consumption and the backup charge set to 15%. However because of the warm days it hasn't been going below 50% before the sun kicks in again the next day. But in Dec and Jan it was regularly hitting 15%. I figure since the warranty covers for 10 years at 70% capacity, by that time who knows what type of battery will be out for the same price.
 
I've not made any attempt to measure the efficiency loss of the Powerwalls, but it sure seems like it's way higher than the stated 10%.

Powerwalls have two losses that may be commingled.

1) The stated round-trip efficiency of 90% (10% Loss).
2) Each Powerwall on average uses just over 1 kWH/day to maintain and condition the battery.

I think discussing regular use practices of PowerWall systems with solar at home would make sense.

In order of personal preference from the stated applications in the Tesla Powerwall FAQ.

1) Time-of-Use Load Shifting - Although the rate of return is debatable (especially without SGIP), it helps "pay" for the system. I already somewhat do this by setting self-powered mode during peak periods. I do look forward to a robust setting, once the software update is released.

2) Backup Power - I set the percentage set higher in winter, lower in summer to account for the difference in solar production. We live in an area with overhead utility lines that are vulnerable to weather. It is nice to have.

3) Home Energy Monitoring - A cool minor feature. I use it as an independent check on grid power I use/send to the utility. Solar production is also compared to the Enphase microinverters. On average, Tesla is reporting 1.7% more energy produced that what Enphase is reporting.

4) Self-Powering your Home - Used in as part of the Time-of-Use Load Shifting. Stay self powered during high peak periods and use energy from the grid when costs are lower. I like being green, but not the primary factor.
 
  • Like
  • Informative
Reactions: Ulmo and Shygar
I have 4+ kW solar and 2 PW2 units with a third coming in spring via the Tesla referral program. Average home usage is about 18 kWhr/day overall for a 2700 sq. ft. with air conditioning. Here in Atascadero, the summers are warm and the winters cool, ranging between a max of ~110 to lows in the teens. Usually much cooler/warmer than the max/min resulting in little air con or forced air heating.

My goal is to maximize self consumption of solar. Selling back power via NEM (1.0 for me) seems less useful than capturing the avoided costs in the first place. PG&E is paying ~3 c/kWhr vs selling at anywhere from ~12 to 48 c/kWhr depending on your rate plan. I've been able to minimize the amount of power sold back to PG&E by using the battery on my MS90D as a surge volume to dump power when I produce more than I can use as in summer when it's cool enough not to require air conditioning.

Winter operations: November through January I don't normally generate enough power for daily unless it's clear. Worst case has been about 5 kWh generation on a cloudy day. Careful use of electricity can reduce our daily use to about 10 kWh if necessary leaving about a 5 kWh deficit. Based on 2 years of data those minimum generation days rarely happen more than 2 days in a row so I keep at least 13.5 kWh (50% of my current system) reserve in the PWs during the winter to tide us over. Backup is a 3 kW inverter generator to run the refrigerator, freezer, internet equipment and charge battery sources for lighting and entertainment.

Summer operations: (May - September) Generally enough generation to avoid using any peak or partial peak power. EV charging occurs between 2300 and 0700. Since summer generation is almost guaranteed I keep my reserve as low as necessary (20%) to have enough battery to make it until actual solar generation meets morning startup loads at about 9 a.m. The added use of electricity by the freezer in the garage during the summer seems at least partially offset by the significant reduction in the use of lighting and space heating from the cooler months.

It's been interesting seeing all the posts about the desire for TOU capabilities with many folks making the assumption that TOU means they will be allowed to buy power cheap and sell it back during peak hours. As someone that worked in a power plant control room for nearly 20 years and interfaced with system grid operations I'm pretty sure that CALISO (and probably any other system operator) won't allow that. The system operator have enough on their hands controlling grid stability with all of the renewables they are being forced to take. As the industry news will bear out, renewables are great but they aren't very stable. Solar ebbs and flows not only with the time of day but also when clouds blow over. Wind outputs go up and down too. I just can't see how the system operators would willingly submit to more "must take" generation unless it was politically driven by clueless legislators.
I could see the system operators and the utilities allowing or even forcing off-peak charging from the grid even in conjunction with solar to reduce overall load on the system. In fact, that method of load shifting makes the most sense to me. On a large scale it would allow the storage of unused generation during low grid demand rather than curtailing or selling at a loss. This in turn, would maximize the efficiency of the grid we currently have.
In a cursory review of NEM 2.0 I noted that there seemed to be language requiring smart inverters. I assumed that smart inverters would be units that the system operator could control to act as a "virtual power plant" as is being touted in Australia. Having thousands of batteries available at their disposal to mitigate grid fluctuations would be a pretty smart thing. A system like that might significantly reduce the amount of spinning reserves that the system operators try to maintain in order to assure grid stability. In addition, being able to estimate how much battery power will be available in the upcoming hours, days, and weeks could significantly change the landscape of power trading and reduce overall energy costs (for the utilities, not you and me <g>)
However, I don't know how many folks would be willing to give up control of their battery storage. I suppose a lot would depend on whether the owner could control the amount of reserve to be untouched, max power draw, etc. A much more complex system than what the PW software seems capable now would be required. Say hello to Big Brother.

Bigtanuki
 
  • Love
Reactions: Ulmo
However, I don't know how many folks would be willing to give up control of their battery storage. I suppose a lot would depend on whether the owner could control the amount of reserve to be untouched, max power draw, etc. A much more complex system than what the PW software seems capable now would be required. Say hello to Big Brother.

Here's a quote from the powerwall purchase agreement:

10. Grid Services. Your System is capable of supporting the reliability of the electrical grid by providing services under programs offered by utilities or third parties. You grant us the right to access and use your System to provide these services. Before we do this, we will explain the terms of your participation in the relevant program and give you the opportunity to opt-out.

At least there's an opt-out!
 
  • Like
  • Informative
Reactions: Ulmo and GenSao
There was some discussion above about PowerWall 2 efficiency. I have one datapoint about that. Both Yesterday and Today show that I filled my PowerWalls to 100%. Yesterday's "From PowerWall" evening draw was 6.6kWh and Today's "To PowerWall" figure was 7.7kWh. So, the conversion losses to get back to 100% were 1.1/6.6 = 16.6%
 
  • Informative
Reactions: Ulmo
However, I don't know how many folks would be willing to give up control of their battery storage. I suppose a lot would depend on whether the owner could control the amount of reserve to be untouched, max power draw, etc. A much more complex system than what the PW software seems capable now would be required. Say hello to Big Brother.

Bigtanuki
I love most of your post. I have very little of it to add to or comment on. Just this: I think what you just said that I quoted, that a user should be able to set the amount of capacity to be made available for others vs. what is to be made available only for oneself, is essential to a complete settings system, and my bare minimum for sane control of the PowerWall. I would define "capacity" as kW in, kW out, kWh in, kWh out, % SOC range for them, etc. Although nominally they would be max settings (max kW out, max kW in, etc., lowest SOC % they could draw down to, highest SOC % they could charge up to, etc.), additional settings of minimums to others might be required for contractually created edge cases that would require a few fine tuned further settings available. But it's really not super complicated. Anybody who thought of the dynamics of this type of system decades ago realized all of this was required, and I know that includes me, and by today, anybody who looks into this issue with anything but the lowest of IQs. For instance, you, Bigtanuki, brought up basically everything in one simple post, so it can't be that hard for seasoned professionals actually in the field to properly handle it, provided they don't have obstructionist vested interests hindering proper conduct.

As @markb1 said, we already signed an agreement setting the legal permission for the utility to use our system for the grid, and to look at our use patterns. Now, PG&E and Tesla just have to set up the programming and settings. I really don't like how they're dragging their feet on this straightforward step, trying to make it sound and be as complicated as possible, to feed into their dragging their feet even more, in order to make it sound even more complicated, etc. It means that when they do implement, they won't iterate fast enough to debug it, and it will always be a quagmire since it will always be woefully wrong, when instead, it could be really clean and great (like the Australian Tesla battery).

Meanwhile, I'm extremely happy to avoid the utility as much as possible.
 
Last edited:
There was some discussion above about PowerWall 2 efficiency. I have one datapoint about that. Both Yesterday and Today show that I filled my PowerWalls to 100%. Yesterday's "From PowerWall" evening draw was 6.6kWh and Today's "To PowerWall" figure was 7.7kWh. So, the conversion losses to get back to 100% were 1.1/6.6 = 16.6%
Here's mine:

$ powerwall_lan_ip_number=10.999.999.999 # censored my ip#
$ curl -s http://${powerwall_lan_ip_number}/api/meters/aggregates | json_pp
{
"frequency" : {
"instant_power" : 0,
....
},
"load" : {
"instant_apparent_power" : 597.150035007353,
"instant_total_current" : 2.51734742100917,
"instant_reactive_power" : 3.4621286850206,
"instant_average_voltage" : 237.210006713867,
"i_a_current" : 0,
"i_b_current" : 0,
"energy_exported" : 0,
"instant_power" : 597.13999863872,
"energy_imported" : 5780477.42694444,
"last_communication_time" : "2018-02-10T17:07:16.089322353-08:00",
"i_c_current" : 0,
"frequency" : 60
},
"solar" : {
"instant_total_current" : 0,
"instant_average_voltage" : 237.17000579834,
"instant_reactive_power" : 161.819999694824,
"instant_apparent_power" : 686.514019083056,
"energy_exported" : 4493519.89404667,
"i_a_current" : 0,
"i_b_current" : 0,
"energy_imported" : 9070.34654667211,
"instant_power" : 667.169982910156,
"i_c_current" : 0,
"frequency" : 60,
"last_communication_time" : "2018-02-10T17:07:16.089493021-08:00"
},
"busway" : {
"instant_apparent_power" : 0,
...
},
"generator" : {
...
},
"site" : {
"energy_imported" : 2023063.98836899,
"instant_power" : 2.74000549316406,
"i_c_current" : 0,
"frequency" : 60,
"last_communication_time" : "2018-02-10T17:07:16.089322353-08:00",
"instant_average_voltage" : 237.210006713867,
"instant_total_current" : 0,
"instant_reactive_power" : -506.720001220703,
"instant_apparent_power" : 506.727409232234,
"energy_exported" : 265966.108924546,
"i_a_current" : 0,
"i_b_current" : 0
},
"battery" : {
"energy_exported" : 2153090,
"i_a_current" : 0,
"i_b_current" : 0,
"instant_average_voltage" : 237.1,
"instant_total_current" : 0.4,
"instant_reactive_power" : 350,
"instant_apparent_power" : 359.026461420325,
"i_c_current" : 0,
"frequency" : 59.983,
"last_communication_time" : "2018-02-10T17:07:16.090351025-08:00",
"energy_imported" : 2614160,
"instant_power" : -80
}
}​

I can see the battery energy_exported and energy_imported directly from my PowerWalls. It also notes the grid ("site"), home ("load"), solar. Note that I correct for the flow direction in my graphing at ulmo.solar, but direct from the PowerWall, the data is incorrectly inversed for many of the points, and can show many upside down graphs in the smartphone app as well; just be aware of having to reorient yourself for each measurement point depiction to make certain you understand each input and output direction. Anyway, for the battery, the listed "export" must be to the battery, and "import" must be from the battery. I wish they would just call it precisely "to" and "from" the entity on the other end of that measurement point and be done with it. Anyway: simple division:

2153090Wh/2614160Wh=.8236259%

That's around 18% loss throughout the year.

According to @GenSao, "2) Each Powerwall on average uses just over 1 kWH/day to maintain and condition the battery.", so for my approximately 201 days of having our PowerWalls, that's 201kWh for one or 402kWh for two, so:

(2,153,090Wh+201kWh)/2,614,160Wh=.9005148%

That's dead on 10% efficiency round trip using 1kWh daily to "maintain and condition the battery". Woops! I have two batteries, so the calculation becomes:

(2,153,090Wh+2*201kWh)/2,614,160Wh=.9774038%

That's 2.3% round trip efficiency with 2kWh to "maintain and condition" both my batteries. What gives?

That 1kWh daily use for my system, or, put more simply, ~41 watts for my two battery system (what is it for 1 or 3 batteries?), seems to be left out of the Tesla specs. Here's Tesla's "Technical Specifications":
  • Usable Capacity 13.5 kWh
  • Depth of Discharge 100%
  • Efficiency 90% round-trip
  • Power 7kW peak / 5kW continuous
  • Supported Applications Solar self-consumption Back-up power Time-of-use load shifting (coming soon) Off-grid capabilities (coming soon)
  • Warranty 10 years
  • Scalable Up to 10 Powerwalls
  • Operating Temperature -4°F to 122°F / -20°C to 50°C
  • Dimensions L x W x D: 44" x 29" x 5.5" (1150mm x 755mm x 155mm)
  • Weight 276 lbs / 125 kg
  • Installation Floor or wall mounted Indoor or outdoor
  • Certification North American and International Standards Grid code compliant
How does that compare to grid losses from Moss Landing to my home, overgeneration by Moss Landing, truck pollution and energy use to maintain that grid, the beauracracy for that grid, the energy for the metal for that grid, etc.? A quick Internet search suggests "grid losses" is around 5% (probably an industry deflated number). Another document for Australia says "1% to 8%". Wikipedia says "Transmission and distribution losses in the USA were estimated at 6.6% in 1997[22] and 6.5% in 2007.[22]". The utilities try to lowball it. So does Tesla, leaving out the constant drain.
 
Last edited:
  • Like
Reactions: Bigtanuki
I love most of your post. I have very little of it to add to or comment on. Just this: I think what you just said that I quoted, that a user should be able to set the amount of capacity to be made available for others vs. what is to be made available only for oneself, is essential to a complete settings system, and my bare minimum for sane control of the PowerWall. I would define "capacity" as kW in, kW out, kWh in, kWh out, % SOC range for them, etc. Although nominally they would be max settings (max kW out, max kW in, etc., lowest SOC % they could draw down to, highest SOC % they could charge up to, etc.), additional settings of minimums to others might be required for contractually created edge cases that would require a few fine tuned further settings available. But it's really not super complicated. Anybody who thought of the dynamics of this type of system decades ago realized all of this was required, and I know that includes me, and by today, anybody who looks into this issue with anything but the lowest of IQs. For instance, you, Bigtanuki, brought up basically everything in one simple post, so it can't be that hard for seasoned professionals actually in the field to properly handle it, provided they don't have obstructionist vested interests hindering proper conduct.

As @markb1 said, we already signed an agreement setting the legal permission for the utility to use our system for the grid, and to look at our use patterns. Now, PG&E and Tesla just have to set up the programming and settings. I really don't like how they're dragging their feet on this straightforward step, trying to make it sound and be as complicated as possible, to feed into their dragging their feet even more, in order to make it sound even more complicated, etc. It means that when they do implement, they won't iterate fast enough to debug it, and it will always be a quagmire since it will always be woefully wrong, when instead, it could be really clean and great (like the Australian Tesla battery).

Meanwhile, I'm extremely happy to avoid the utility as much as possible.
I agree on all counts. In addition to my career in the control room I spent the last few years at PG&E being a "business rep" for our SAP implementation. As the business rep I essentially translated customer requests into developer speak. That included writing specs, test plans, change requests and maintaining a semblance of scope control for customers that just "thought of something" they would like to add to their software request. The one phrase I learned to hate the most was "how hard could it be to write that code?". I learned that it may be straightforward to write a piece of code to do something but it's quite another to write a piece of code that (A) looks like the rest of the pre-existing code and (B) not break some innocuous piece of code somewhere else in the package. Testing was a bitch and we were building business software not something that could get you killed. It is impossible to test every piece of new software against every piece of existing software unless you have immense resources and you don't want to get software except perhaps annually.
With that in mind, I'm pretty patient about any software for the Powerwalls or my Model S because I don't want to risk my life or those around me because of an overlooked piece of code.
I would add, that it seems like Tesla is missing a bet by not drafting a few reasonably techie folks (like Ulmo) to do beta testing on software pre-release. I believe that's the method that most software vendors use. It would be complicated by the CPUC here in California because they have to have their fingers in EVERY pie but it sure does seem like something could be worked out. I would certainly be game to offer my time.
 
Testing was a bitch ... It is impossible to test every piece of new software against every piece of existing software unless you have immense resources and you don't want to get software except perhaps annually

We use regression testing - a set of tests that are run mechanically without user interaction. Those tests can be run daily (e.g. against a "Daily Build") and thus provide feedback to Programmers the day after new code has triggered a difference from previous behaviour (might be intended difference of course ... in which case that becomes the new Standard against which future tests are compared).

Of course this is only as good as our suite of tests ... but any new test we add, either to test a new feature, or as a reaction to a fault found "in the field", is part of the test suite for ever after that. We test our new features, but we rely on the regression testing to find any existing, unmodified, code which has been adversely effected by changes "elsewhere" in the code base.

We're a small company. It cost us a fair bit to build the test system in the first place, but it was also costing us a lot to both a) manually test everything (inevitably that was incomplete / inadequate) and b) to have the cost of failure in the field which entailed not only loss of confidence by customers but also "drop everything" for DEVs to fix it, which caused the delivery dates for next development schedule to go out of the window.

My expectation is that this sort of Daily Build and automated regression testing should be "normal" ... but clearly from the regression errors that I've seen in Tesla Infotainment system it ain't (can't speak about their safety critical code, although it does seems to me that that has had much better QA)
 
According to @GenSao, "2) Each Powerwall on average uses just over 1 kWH/day to maintain and condition the battery.", so for my approximately 201 days of having our PowerWalls, that's 201kWh for one or 402kWh for two, so:

I think I was being extra conservative and will have to revise my statement.

Recently (over President's Day weekend) I had the Powerwalls set to backup only mode for approximately 72 hours. The Powerwalls charged up the 1st day in the morning (from 84% to 100%) and lost capacity over time. They only recharged the morning of the 4th day. The Tesla app shows that 1.6 kWH was used that morning. So 1.6 kWH / 3 days / 2 batteries = 0.27 kWH / day / Powerwall.

Roughly 2% a day.

Of course the weather can play a part. The temps were:
17th, Hi = 70F, Low = 40F
18th, Hi = 58F, Low = 43F
19th, Hi = 54F, Low = 37F
20th, Hi = 54F
 
I think I was being extra conservative and will have to revise my statement.

Recently (over President's Day weekend) I had the Powerwalls set to backup only mode for approximately 72 hours. The Powerwalls charged up the 1st day in the morning (from 84% to 100%) and lost capacity over time. They only recharged the morning of the 4th day. The Tesla app shows that 1.6 kWH was used that morning. So 1.6 kWH / 3 days / 2 batteries = 0.27 kWH / day / Powerwall.

Roughly 2% a day.

Of course the weather can play a part. The temps were:
17th, Hi = 70F, Low = 40F
18th, Hi = 58F, Low = 43F
19th, Hi = 54F, Low = 37F
20th, Hi = 54F

As I like to say here in Atascadero - "Another crappy day in California!" Hard to believe it's February although we did see some snow flurries while out for our walk on Monday morning.
 
Sadly, seems to me, most government agencies tend toward extorsion. [charge fees for NO services]

Business tend toward monopoly - Peter Thiel has said publicly he prefers to invest in companies with monopoly advantages.
Amazon, FaceBook, Google most recent
Cable TV naturally and cell phone companies with contracts and must buy their cell phone.
Mainstream news media
You get the idea.
and monopolies tend to extorsion also, right?

Investor owned electric utilities get rate increases every two years mostly and are therefore about 30-40% more expensive than public run utilities. Corruption is a problem that needs fighting continuously - corruption never sleeps.

That is why it is so great to watch Elon conduct business in a "fair/honest" way as best he can.
</rant>

Sadly many utilities will want to extort customers vs co-operating for a more robust grid and cheaper electricity, right?

Good Luck, fight for us all, work with your friends, colleagues and neighbors.