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

Tesla: we need more control over our PowerWalls

This site may earn commission on affiliate links.
The limit is because the utility company doesn't want people installing more solar than they 'need' so they limit customers to 120% of their previous year's usage.
That is another reason I am considering installing a hybrid inverter and adding more solar. Since that system can be programmed to not feed the grid I only need to get a building permit. No interconnection agreement and I can incrementally add panels and batteries until I have enough capacity to only pay the minimum bill.
 
  • Like
Reactions: abasile
Thanks Ulmo for the explanation - just to be clear you want to use grid energy when the network demand AND the price was low?

If you are asking Tesla for 'More Control' - what specifically would you like to see?
- Dispatch based on real-time demand (or GHG) signals from your utility
- Co-optimize for pricing and grid benefit


When I was referring to peak - I was referencing the peak pricing in your rate plan. It sounds like you want 2 different outcomes, which are not necessarily coincident.

1. Dispatch to reduce load on the grid regardless of pricing
Currently this is done by using self-powered mode, agreed this is not perfect as there is missed opportunity to support the grid during the evening ramp. Alternatively you could set your time based control periods to match CASIO forecast demand - the variable time and geographical nature of this demand signal makes is challenging but not impossible. At the CASIO demand level, you don't know what the marginal cost of generation is for the system to make a dispatch decision.

Local control will never achieve this in any meaningful way. Aggregation programs and virtual power plants are the only way to dispatch storage that is situated at the residential level that is guaranteed to have an impact on GHG or generation costs as only the utilities have the information to make a fully informed dispatch decision.

The network operators will pay you for the capacity to dispatch when they need it. This is what happens in the Green Mountain Power program in Vermont and other programs in Australia.


2. Dispatch to reduce your costs
This is done by setting time-based control to match your rate plan for peak pricing.
In an ideal world utilities would price energy as a accurate residential signal for the demand or even better greenhouse gasses - otherwise why have peak pricing, but you are right - that is not always the case.

In your example, you intervened in this operation to prioritize the first strategy - so questions becomes, why not just operate in the first mode all the time if that is your preference?



In my view - if you want to help the grid, operate in self-powered mode (or grid aligned TBC) until you can enroll in an aggregation programs. You setting your own strategies to try and co-optimize the two strategies will almost certainly result in loss of performance on both metrics.
 
After finally getting down to within 1% of my set reserve last night, my Powerwalls got up to 80% today. I still haven't figured out why it doesn't charge them all the way when on TBC. It could've easily charged the Powerwalls to 100% and started exporting to the grid over 2 hours early...but no...my system started sending all of my solar production to the grid even though peak doesn't start for another 3 hours. :confused: Before 1.25.0 and 1.26.0, I think it used to charge the Powerwalls all the way, if possible, before peak started.

If they could just give me a setting to indicate what hours I want to self-power, I'd be happy!

Did you run out of Powerwall energy during the peak period? If yes, then there is an issue stopping at 80%.

However - if your peak period is covered, then why put extra energy through the battery and incur efficiency losses? As long as Powerwall has enough energy to cover your predicted peak, you actually want to export gross solar to get NEM credits right? Thats the best economic outcome

I think the real issue is does Powerwall accurately forecast your required peak usage? I've seen a few examples where it fell short
 
Did you run out of Powerwall energy during the peak period? If yes, then there is an issue stopping at 80%.

However - if your peak period is covered, then why put extra energy through the battery and incur efficiency losses? As long as Powerwall has enough energy to cover your predicted peak, you actually want to export gross solar to get NEM credits right?

I think the real issue is does Powerwall accurately forecast your required peak usage? I've seen a few examples where it fell short
No. My peak period is only 4 hours long, from 2 pm until 6 pm. During the entire peak period, I only use about 3 to 4 kWh so the Powerwalls have no problem covering peak.

One complaint with the Balanced TBC mode is that it often prioritizes sending production back to the grid when I want the Powerwalls charged more. Early last week, I was down to the 25% reserve and we had several cloudy days in a row. While I have Storm Watch enabled, it never activated and never tried charging up the Powerwalls and they hovered between 25% and 45% for days. Yes, the Powerwalls were charged up enough to cover the short 4 hour peak period every day but if the power had gone out due to the storm, they wouldn't have been able to cover an extended grid outage. I don't mind some efficiency losses if it means I can go days without grid power in an emergency vs. only a few hours.

Other times, it will still prioritize sending to the grid and needlessly drain the Powerwalls. For example, if I try to charge a car at noon to take advantage of the max solar power generation, I've seen it sending 12+ kW to the grid while simultaneously drawing 10+ kW from the Powerwalls. I could understand logic during peak period since it would be trying to generate the most peak credits while using power that was stored in cheaper periods...but it does this hours before peak is even scheduled to start! I usually have to switch it over to self-powered so that it will actually charge the cars and charge the Powerwalls. Most days, I can do both and still have hours before peak starts...but since 1.25.0, it doesn't seem to care when my peak period actually starts.

At this very minute, it is actually working as I would expect. I'm generating 10 kW and sending 9 kW directly to the house to cover the household usage and charge a car. The Powerwalls are also being charged at over 1.7 kW and a small amount is being drawn from the grid. Peak doesn't start for me for another 4 hours but I predict that within 1 to 2 hours, it will start sending the 12 kW to the grid while pulling 9 kW or more from the batteries. :confused:

Untitled.png
 
  • Like
Reactions: abasile and eml2
you actually want to export gross solar to get NEM credits right?
For my TOU plan, I actually do not want to export any power to grid during peak because I don't have 1:1 net metering. The utility pays me less than half of what they charge for off-peak. I would like an option for not prioritizing export during peak. Self-powered is sort of working for me, but it doesn't support any time-based decision. e.g., for a certain cloudy day, I can see that just in self-powered mode it won't have enough to cover the whole peak (11a - 7p). Therefore, I would like to have an option of just charging it with all solar during part-peak (before peak). In the current self-powered mode, there is no concept of peak/part-peak/off-peak.
 
  • Like
Reactions: MorrisonHiker
No. My peak period is only 4 hours long, from 2 pm until 6 pm. During the entire peak period, I only use about 3 to 4 kWh so the Powerwalls have no problem covering peak.

One complaint with the Balanced TBC mode is that it often prioritizes sending production back to the grid when I want the Powerwalls charged more. Early last week, I was down to the 25% reserve and we had several cloudy days in a row. While I have Storm Watch enabled, it never activated and never tried charging up the Powerwalls and they hovered between 25% and 45% for days. Yes, the Powerwalls were charged up enough to cover the short 4 hour peak period every day but if the power had gone out due to the storm, they wouldn't have been able to cover an extended grid outage. I don't mind some efficiency losses if it means I can go days without grid power in an emergency vs. only a few hours.

Other times, it will still prioritize sending to the grid and needlessly drain the Powerwalls. For example, if I try to charge a car at noon to take advantage of the max solar power generation, I've seen it sending 12+ kW to the grid while simultaneously drawing 10+ kW from the Powerwalls. I could understand logic during peak period since it would be trying to generate the most peak credits while using power that was stored in cheaper periods...but it does this hours before peak is even scheduled to start! I usually have to switch it over to self-powered so that it will actually charge the cars and charge the Powerwalls. Most days, I can do both and still have hours before peak starts...but since 1.25.0, it doesn't seem to care when my peak period actually starts.

At this very minute, it is actually working as I would expect. I'm generating 10 kW and sending 9 kW directly to the house to cover the household usage and charge a car. The Powerwalls are also being charged at over 1.7 kW and a small amount is being drawn from the grid. Peak doesn't start for me for another 4 hours but I predict that within 1 to 2 hours, it will start sending the 12 kW to the grid while pulling 9 kW or more from the batteries. :confused:
View attachment 350596

It didn't happen as early as I expected today but when the Powerwalls were at about 70%, it switched over to send everything to the grid/house around 45 minutes before peak started. It's now showing that the Powerwalls are being drained, sending 200 W back to the grid and 700 W to the house. So not only is not charging up my Powerwalls all the way, but it's sending energy from the Powerwalls to the grid. :(
 

Attachments

  • Screenshot_20181107-135802_Tesla.jpg
    Screenshot_20181107-135802_Tesla.jpg
    81.4 KB · Views: 57
I just did it again. My PowerWalls were at 60% state of charge, which for me isn't enough for the use I was about to do to last through the evening, so moments ago I turned them off and started our electric dryer, and also my electric water heater and electric coffee maker are also running. Sun is shining in California, and I'm getting 3.7kW from my solar array, but using 8.1kW currently, making up the rest from 4.4kW from PG&E, which includes solar from across the street roof (and untold other customer side of the meter solar panels throughout our area) and about 8,376,000kW utility solar at this very moment in California spread across the tens of millions of people in California, which currently is around 1/3rd (33.19%) of 25,233,000kW total being used at this very moment from utilities in California. After the dryer and coffee maker are done in an hour, I'll turn the batteries back on to charge up a bit more in the afternoon. I'll be saving this energy to use this evening when the sun isn't shining and PG&E is using a lot more fossil fuels for everyone else.
Screen Shot 2018-11-07 at 13.19.01.png
In terms of the PG&E rate plans, I am saving this energy for the PG&E peak evening period. I don't know if right now the energy is at PG&E part peak (I think so) or PG&E full peak rate, and it doesn't matter, since I'll be offsetting PG&E peak rate evening use no matter what. I will suffer no or minimal financial loss, likely a financial gain, but more important, reduce fossil fuel demand. Some people think that I am increasing the fossil fuel demand in the day rather than the evening, but the fossil fuel demand in the evening is dirtier than the fossil fuel demand in the day. Someday, daytime solar power will be so plentiful that using daytime electricity will always be or almost always be cleaner than nighttime use because the daytime off-peak times will be entirely solar power.
 
Last edited:
It didn't happen as early as I expected today but when the Powerwalls were at about 70%, it switched over to send everything to the grid/house around 45 minutes before peak started. It's now showing that the Powerwalls are being drained, sending 200 W back to the grid and 700 W to the house. So not only is not charging up my Powerwalls all the way, but it's sending energy from the Powerwalls to the grid. :(
That's odd. I wonder why. If the PowerWalls are supplying your solar generated electricity to you all the way through the rest of the day and night and the next day until solar takes over again, maybe it's just because "it can" without giving up your self-generation solar use, and is offsetting some of the utility daytime fossil fuels as the use ramps up a tiny bit in the afternoon. Or maybe that's what it thinks it's doing. The purple "Net Demand" CAISO plot line just started upward about when you posted your post:
Screen Shot 2018-11-07 at 13.34.19.png
 
I suppose there could be some logic behind the scenes that we aren't aware of. If so, it would be nice if they could show us a message indicating that they intentionally aren't charging the PWs to 100% because of xyz or that they are exporting to the grid because of a request from the power company, etc.

As it is now, it just doesn't make any sense when it is charging or discharging the PWs or when it is sending power to the grid.
I look at it this way: Elon is not a full time Tesla CEO, so he doesn't micromanage the Energy division very well and they kind of are new to all this storage stuff, and are just learning how to optimize the charge and discharge timing and amounts, so they're getting their feet wet experimenting with their programming of our batteries. Once they learn enough, maybe they'll know what default programming to offer, and maybe they'll release its control more to us users. Since Elon likes to control what we do with our own equipment, I'm not holding out 100% certainty of that outcome. Either way, I hope Tesla learns more and more about optimal ways to program the batteries.
 
The issue is that the power companies are balking at buying their own power back at an increased rate. Someone needs to write some software to allow that control. Tesla should open source this to allow the consumer to modify. FWIK, Tesla’s hands are tied.

FWIW, there are third-party control systems being developed. Mine uses a third-party control system from Evergen, an Australian company spun-off from our government science research division CSIRO - Why Evergen Solar Batteries | Evergen

It may not be suitable for US use yet, but it works for Australian quirky plans and answers a lot of issues that people on this thread are having:
* you can give it the actual c/lkWh rates for consumption and feed-in as well as the TOU time bands, and feed-in rates can be different to consumption rates, so it can do proper cost-optimisation
* it looks ahead at the weather and tries to predict future generation and consumption patters using machine-learning
* it will stop the battery discharging overnight, and allow the house to draw from grid while rates are cheap, preserving the energy stored for use during the morning before the sun rises to take over
* it will charge the battery from off-peak cheap power overnight up to whatever % it calculates might be needed the next day. At the moment its summer here, and the battery isn't being charged at all, but during winter when the next day is forecast to be dismal, its been known to cause the PW2 to charge itself overnight from grid to 80%.
 
  • Informative
Reactions: MorrisonHiker
* you can give it the actual c/lkWh rates for consumption and feed-in as well as the TOU time bands, and feed-in rates can be different to consumption rates, so it can do proper cost-optimisation
* it looks ahead at the weather and tries to predict future generation and consumption patters using machine-learning
* it will stop the battery discharging overnight, and allow the house to draw from grid while rates are cheap, preserving the energy stored for use during the morning before the sun rises to take over
* it will charge the battery from off-peak cheap power overnight up to whatever % it calculates might be needed the next day. At the moment its summer here, and the battery isn't being charged at all, but during winter when the next day is forecast to be dismal, its been known to cause the PW2 to charge itself overnight from grid to 80%.
That's what I need and what I was expecting Tesla to have developed. Tesla's current implementation of Cost Saving is erratic. Today my PW drew 5.8 kWh of off-peak power to charge the battery and later on sent 2.3 kWh back to the grid as the battery was full. I get no money for what goes back to the grid.

See my PV Output graph and click on the right-most blob under the date to see the battery data. There are days when Cost Saving mode works OK and days when it doesn't and I get the impression that Tesla's software is not attempting to do any forecasting of solar output.
 
You can charge the PW from the grid if you setup it to backup and the reserve % is greater then the actual %.
It depends on where you live and if you have solar installed and how they program them.

In the Continental US, I don't think anyone has been able to charge from the grid like you've described if they have solar. We've had snow all day today. I switched my Powerwalls over to backup this morning when they were at 66% and my reserve is set to 100%. The Powerwalls haven't charged from the grid at all.
 
  • Informative
  • Like
Reactions: JES2 and eml2
We got about 8" of snow yesterday. I changed the PWs over to Backup mode when they were down to ~28 kWh so we'd be prepared for an extended outage. There was no production yesterday and today most of the panels are still covered with snow, so I'm only generating like 0.3 kW currently. I switched the PWs back to TBC and of course they started sending power directly to the grid outside of peak instead of charging the Powerwalls back up. o_O
 
That's what I need and what I was expecting Tesla to have developed. Tesla's current implementation of Cost Saving is erratic. Today my PW drew 5.8 kWh of off-peak power to charge the battery and later on sent 2.3 kWh back to the grid as the battery was full. I get no money for what goes back to the grid.

Tesla's app has no concept of actual rates - the TOU control works solely on time bands, and doesn't know about how different peak/shoulder/off-peak rates are from each other. Tesla's TOU control also assumes US-style net-metering, where you are paid back at the same rate as energy costs, so it makes economic sense to pump energy into and out of the grid, treating the grid like a battery.
That doesn't work here in Australia, where our feed-in tariff is much lower than consumption tariffs (in my case, 11.1c/kWh feed-in, compared to 23 - 54 c/kWh cost for consumption depending on time). This ignorance of the actual c/kWh rates means the Tesla control algorithms sometimes can't make good choices.
 
Tesla's app has no concept of actual rates - the TOU control works solely on time bands, and doesn't know about how different peak/shoulder/off-peak rates are from each other. Tesla's TOU control also assumes US-style net-metering, where you are paid back at the same rate as energy costs, so it makes economic sense to pump energy into and out of the grid, treating the grid like a battery.
That doesn't work here in Australia, where our feed-in tariff is much lower than consumption tariffs (in my case, 11.1c/kWh feed-in, compared to 23 - 54 c/kWh cost for consumption depending on time). This ignorance of the actual c/kWh rates means the Tesla control algorithms sometimes can't make good choices.
It's not even consistent throughout the US. For my utility company in Colorado, our rates are something like 8 cents/kWh off-peak, 13 cents/kWh shoulder and 18 cents/kWh peak...but we only get 1.5 cents if we send power to the grid. We opted to take kWh credits instead to offset our usage since those are earned at the same rate that they would cost us.
 
Last edited:
That doesn't work here in Australia, where our feed-in tariff is much lower than consumption tariffs (in my case, 11.1c/kWh feed-in, compared to 23 - 54 c/kWh cost for consumption depending on time). This ignorance of the actual c/kWh rates means the Tesla control algorithms sometimes can't make good choices.
I agree that lack of detailed data about the tariffs (peak / off-peak / feed-in - about 15p / 9p and 4p in my case) can't help with the decision making but it appears to me that ther control algorithms do not include forecast of the solar generation and this is currently the biggest short-coming. Consequently, tonight's off-peak charging appears to be based on what's happened during the past few days and doesn't allow for tomorrow's weather being different from today. This failing can be largely mitigated if Tesla provided the facility for a user-defined minimum charge level at the end of the off-peak period. The user can then adjust this based on past experience and the weather forecast.
 
  • Like
Reactions: PBBear