I have at Tesla system with 3 batteries, and I argued during installation to include 8kw of panels situated on the west side of our roof against Tesla ‘s engineering. The panels would only by 72% efficient on the west side as opposed to 74% on the east side (catching the morning sun). But my modeling showed that we would exhaust the batteries in the evening due to the fact that my usage was higher in the late afternoons and it wasn’t offset by any generation of solar panels during those late afternoon hours.
After modeling scenarios based on historical usage PER HOUR, I was able to show that if we had enough solar generation during peak late afternoon hours, we would be able to ‘survive the night’ on batteries until morning solar generation resumed. This means my 14kw solar panels coupled with 3 batteries gets me completely off grid for 9 months out of the year. That’s not bad considering I get 7ft of snow during winter months and I am surrounded by very tall trees.
Optimize on hourly generation not daily, most solar companies use DAILY numbers without a clue on hourly usage. I currently get 0.08$ for every 1$ in electric production, so there is very little benefit in producing electricity when you don’t use it. Optimize your system based on your usage not on DAILY production. If electric companies would give me credit of say 0.90$ per 1$ then the equation changes, but electric companies would rather benefit from your overproduction, be careful as these systems are not cheap!
This exactly. We run a 100KW microgrid on Hispaniola, and most of our panels are oriented to maximise winter afternoon sun, or just pointed randomly at the sky. Random pointing gets us more power than all oriented 12 degrees south because the power we care about is in overcast conditions, especially high altitude overcast, and that is variable in intensity in different sky regions on a minute by minute basis.
Also 12 degrees south would put a mountain partially in view of the panels, and mountains don’t provide much light here. (I don’t mean the mountains would block the sun, just a band of otherwise visible sky)
When we do have high intensity light, the late afternoon is when we want it, and also when the sun is most off to one side of the sky.
My advice: over panel as much as you can. We can fully charge our batteries while running the farm and 6houses in three hours of full sunlight, so we still get plenty on overcast days, and even on the few darkest days we make about 70%. We have to supplement the solar about 60 days a year total, burning a total of 300 gallons of fuel over the year for a small farm and 6 houses.
There is no 1$ in electric production, unless you are selling and buying simultaneous at the exact same date and time, and from the same location. A fair price would be the current market price at the point in time, subtracted with the cost of operate transmission, facilitating the sale, and the predicted reduction in market price by increased supply. Depending on where you live the price difference between the average day of selling and average day of buying may very well be quite different. It is not uncommon in Europe to see negative market prices, where overproduction is not going to benefit the electric company at all.
The Powerwall system prioritizes filling up the batteries first. I assume it will take pretty much your entire solar day to fill your three batteries so why didn’t you choose the model that fills up your buckets as the first priority? I think the rule is “optimize on hourly storage” and the hourly production should follow that requirement. Doesn’t the 74% give you a bigger area under the production curve than the 72%?
Bear in mind that charging and discharging batteries has an efficiency penalty - perhaps 98% efficient for each. So 74 stored is worse than 72 used directly because 74 -> 72.52 stored -> 71.06 discharged.
And usually the efficiency is much worse than 98%.
Oh, and also batteries such as the tesla power wall can only be charged and discharged about 1000 times before they have lost a lot of capacity. So generating when you use also makes your batteries last much longer. You could think of this as a cost of battery depreciation per kWh stored.
That 1000 is for NMC batteries and its a 70% capacity. Still enough to be useful.
Also, there's a lot of factors that go into play. For example, this assumes the batteries are fully charged and discharged. If you do something smarter like going down to 40% and up to 80% then they end up being able to do a lot more cycles. In fact, battery age starts mattering more than the cycles.
But besides that, LFP batteries are currently being used in home battery storage (including powerwalls) because it's cheaper and it has 5000->10,000 cycles before dropping to 70% capacity.
Generally, though, I'd agree that having more generation throughout the day is better than having perfectly optimized generation.
> Oh, and also batteries such as the tesla power wall can only be charged and discharged about 1000 times before they have lost a lot of capacity.
Powerwall's cycle life is much better than 1000. The Powerwall warranty guarantees 70% capacity after 10 years of daily cycles (i.e. 3650 cycles). This means they expect the capacity to be substantially above 70%.
We posted an analysis of Powerwall capacity retention: https://www.netzero.energy/content/2025-02/powerwall-analysi...
> The Powerwall warranty guarantees 70% capacity after 10 years of daily cycles (i.e. 3650 cycles). This means they expect
... to have pulled some corporate restructure which leaves a bankrupt legal entity responsible for the warranty claims before they start costing any real money.
The Powerwall solar controller prioritizes the home before sending surplus to the grid. And in the home, the controller will send power direct to any running load first, then to the Powerwall battery. Any spillover then goes to the grid. It’s very dynamic. I would go with the algorithm to “capture the rain into my rain buckets the moment I can” because the rain could stop. Solar irradiance is unpredictable.
If you have time based billing you can also input that into the system and it's even more effective. For example, if you tell it that electricity is cheap from 9am - 5pm (peak solar) and expensive from 5pm - 9pm (peak residential demand) it will take your trending consumption and decide when your solar production isn't keeping up with foretasted demand and let you charge from the grid to at the cheap rate to make up the shortfall and minimize cost. It even factors in things like grid charging speed and total site usage limits, which are great given my 100amp panel.
The charging the battery from the grid on its own is interesting in spaces where the TOU between 4-7pm (or whatever yours is locally).
Here it is more than 3x, so if I can charge a battery and run off of that for those 3 hours, I am saving money.
And it's not that I can lose money, a charge in the battery doesn't become stale.
Don't forget to account for the additional battery wear from extra charging and discharging. However your cost saving probably exceeds the wear cost.
> Solar irradiance is unpredictable
Except in my case it is! I wanna do most of my charge at noon when I'll have far more than I can export. Export first, then charge!
That strategy make sense if you're goal is longevity for your first battery units. But with most early-adopter, often the attitude is to know you'll be upgrading somewhat frequently (3-5 years) due to rapid advances in the tech, until more advanced offerings come around that could least 10-20 years.
[dead]
Excuse what might be an ignorant question, but is the difference between 74% and 72% significant? At a big scale it certainly could be, but at the scale of a single family (or even small-ish multi-family) home I would think it wouldn't make a lot of difference, but I'd much appreciate a correction
I think the efficiency figure is the one the tesla engineers used. The one that he used is that he’d rather have optimal generation from 1-6pm than from 7-12am
> The Powerwall system prioritizes filling up the batteries first.
I think it depends on how you've configured it. My never charges the battery up from solar. I think it estimates how much power you'll need based on the configuration and charges accordingly. I've noticed on hot summer days mine will charge a bit in the morning then stop, then in the afternoon when the AC kicks on and off it will charge between AC cycles.
This is annoying approach. Mine is not Powerwall so I have to try to play with HAs scripts to only charge battery on excess solar - I have far more panels than I can export, and inverter (Deye) can only handle so much AC so excess DC needs to go into batteries.
Trouble is you don't really know how much excess power you've got until you crank up battery charge current.
We have a small roof with 1/3 of our panels facing east, 1/3 facing west, and 1/3 facing south. Given a sufficiently large roof, it would theoretically make sense to have all of the panels facing south.
However, due to the fact that PG&E keeps shifting our peak hours, we actually get more credit for producing in the afternoon, so when we expand our house, I'm planning on having all the panels (as much as possible) on the west-facing roof.
Also, we plan to install air conditioning at that time, so it will be helpful to be able to handle that peak demand.
> theoretically make sense to have all of the panels facing south
It depends what you are optimising for. East/west makes a lot of sense to optimise for morning and evening sun. Especially as during the middle of the day electricity prices usually drop due to excess solar.
> and I am surrounded by very tall trees.
https://www.suncalc.org works great for shade calculations, I was surprised when I checked tree shadows for different times a year.
Totally agree on the feed-in tariff reality check. Unless you're in a rare location with generous net metering, exporting surplus is basically a donation
Man your comment rings home so badly. Made a similar jump last year, and I was surprised we outperformed the predictions. Living in a cursed area (rainy + cloudy all the time), more than 50% of the year our electricity is 100% covered by our solar panel. 0 transportation (which is very high too).
It’s $0.08 for every $1 at retail rates, from the powerwalls, regardless of time or market prices?
Thank you for the example. I am always interested in real-life stories from other snow-laden users.
What do you use for heat?
Very tall trees, I presume.
Did you answer for A/C or is there a way to get heat from trees (without burning them)? (not a sarcastic question though I concede it probably sounds like it is)
I use both electric and Natural Gas heaters. Natural gas heaters are around 65/mo to run
Mostly agree, but there are also variables like feed in tariff (or delta between export and import) and mine just changed to 2 cents NZD. I've switched off hot water, battery and car charging scheduler because there is no point to save this anymore.
Plus modeling goes only so far when your behaviors change. I found it's incredibly hard to model this when there's so many variables.
Can your system work during a power outage? Most US installed ones up until about 5-10 years ago could not because they lack a proper automatic transfer switch certified to never back feed.
Yes, it’s one of the main reasons I purchased a system, I have had outages for longer than 2 weeks at a time .. PG&E
PSPSes? I lived in Paradise until 2020 and had to do the generator-filling and extension cords dance while not having internet for a week or 2 at a time several times per summer. Comcast/Xfinity's gear would last about 24 hours without power and then internet would go out. Contemplated off grid power, but it wouldn't been pointless and futile to sink any more money into an area that was fundamentally unsuitable for long-term occupation.
So, instead of living in a desertifying moonscape with jacked-up insurance costs, we bailed for around the 100th meridian west where there aren't forest fires or earthquakes. And no hurricanes or floods either. The most concerning things are hail and wind which can be more-or-less mitigated and/or insured for, while tornado risk here is on the order of 0.0001%/structure/year.