Strategies for Powerwalls and Utility Demand Plans

[destructure00]

Thanks. Really appreciate it. Yes, I've found that self-powered does a better job than TBC Cost-Saving as far as quickly ramping up to cover usage spikes. I've been using the Powerwall Manager to define Self-Powered mode schedules to match my utility peak periods and it has eliminated that last vestige of usage/demand I was still seeing during peak hours. It definitely adds some reliability risk though, since the PowerWall Manager is dependent on a 100% operational SmartThings cloud, Tesa cloud, and your local internet.

Are you still using the self-powered mode with varying reserve capacities instead of time/cost savings mode? After reading this thread I'm on the fence between the two once I'm up and running. I'm on APS, so their demand charge is based on an hour of peak demand instead of 30 mins. Seems like the quicker response of self-powered would be ideal, but I'd be worried that a single "lost" reserve change command would kill the whole month. At least in my case, Hubitat runs locally, so there's only one cloud service to worry about. But still dependent on my internet not going down for some reason, and Tesla's cloud receiving the message.

I'll have 11.34kw solar and 2 PWs when it's all finished, and flat 10.44 cents per kwh solar export price, regardless of time.

If I end up using time based/cost savings mode, definitely sounds like I should set up 3-8pm as a shoulder period rather than peak, thanks for posting your experience with that. I'm really surprised Tesla doesn't have some option for demand plans yet.

 

[Darwin]

Are you still using the self-powered mode with varying reserve capacities instead of time/cost savings mode? After reading this thread I'm on the fence between the two once I'm up and running. I'm on APS, so their demand charge is based on an hour of peak demand instead of 30 mins. Seems like the quicker response of self-powered would be ideal, but I'd be worried that a single "lost" reserve change command would kill the whole month. At least in my case, Hubitat runs locally, so there's only one cloud service to worry about. But still dependent on my internet not going down for some reason, and Tesla's cloud receiving the message.

Yes, I'm still using Self-Powered. I schedule reserve in the app to 0% before peak and then back to 100% at peak end. I probably should be more concerned about a missed command, but since I'm also running the demand manager software in the background, if I were to get a demand spike due to the Powerwall not being configured correctly, the demand software would turn off my AC, give me heads up with a phone alert, and flash an led red on a couple of smart switches throughout the house.

I'd still probably recommend going with TBC shoulder/Cost-Savings for most people on demand plans that have high $$ consequence of missed commands. Some users also schedule a redundant reserve % command just to be safe.

I implemented a simple error checking/re-try and alert capability on the commands a few months ago in the Powerwall Manager, so technically that should help if there is an issue. The error check code doesn't actually verify that the new state is set, but does look for command response errors back from Tesla. In practice, I haven't observed a missed command since I implemented the error checking code and haven't actually seen the safety code kick in. During the SRP two peak periods during winter, I was actually scheduling 8 commands per day. since I also schedule stormwatch to off during peak and then back on afterwards. A Powerwall stormwatch grid pull during peak would be outrageously costly for me with SRP demand rates.

I alternate every few months between running/testing the code on Hubitat and SmartThings. For the last few months I've been using Hubitat. As you touched on, with Hubitat, it's probably a little more important that your local internet connection is as close flawless as you can get it. I'd also recommend a hardwired ethernet connection for your Powerwall regardless if you can swing it.

If I end up using time based/cost savings mode, definitely sounds like I should set up 3-8pm as a shoulder period rather than peak, thanks for posting your experience with that. I'm really surprised Tesla doesn't have some option for demand plans yet.

Yes, if you end up using TBC/Cost-Saving, a shoulder period instead of a peak period is definitely much safer if you're concerned about your batteries lasting through 8pm. How many AC's do you have? With 11kw solar and 2 PWs you might be safe regardless, but I'm not seeing a huge benefit (but lots of risk) of using a peak vs shoulder period if APS isn't giving you anything extra for selling back your excess solar during peak vs off-peak.

 

[destructure00]

Thanks for the detailed response, sounds like your app is very thorough and well thought out, thanks for putting in the time and sharing with the rest of us. I'll probably start with the TBC/Cost savings mode and see what the small spikes do to my demand. My plan with APS has a $17.44/kw demand charge, but it's "a charge for the single highest hour of energy used", so maybe the small spikes, being short duration, won't add up to much. I guess I'll probably find out, and switch if they amount to too much. Do you have any demand at all using self-powered mode?

I only have one A/C unit, 135 LRA rating if that's important.

 

[Darwin]

Thanks for the detailed response, sounds like your app is very thorough and well thought out, thanks for putting in the time and sharing with the rest of us. I'll probably start with the TBC/Cost savings mode and see what the small spikes do to my demand. My plan with APS has a $17.44/kw demand charge, but it's "a charge for the single highest hour of energy used", so maybe the small spikes, being short duration, won't add up to much. I guess I'll probably find out, and switch if they amount to too much. Do you have any demand at all using self-powered mode?

I generally have zero demand each month now that I'm running in Self-Powered. Considering you have 2 Powerwalls, a single AC, and a one hour demand period, I'm guessing you'll get pretty close to zero demand even in TBC mode. Since SRP uses 30 minute demand periods, a single demand spike for me has twice the demand impact as a demand spike during an APS 1 hour demand period. For example, a single aberrant 1kW grid draw for 15 minutes would be a 500W demand hit for me with SRP, but would only be a 250W demand hit for you with APS.

On a side note, I am actually on the SRP E-15 Average demand plan which means even if I have a small ~500-800W demand spike, this bad day will usually get averaged down to zero at the end of the month considering most days in the month are zero max demand. I highly recommend the E-15 Average Demand plan over the E-27 plan for SRP PW users with a reasonable expectation of having zero demand for the month except for maybe a single aberrant demand spike now and again. Even though the E-15 demand rates are higher than the E-27 demand rate, this larger rate multiplied by the zero (usually for me) average demand is still zero. With E-15, you can rest easy if you accidentally blow demand on just one day a month.

 

[Rahul]

This schedule looks good to me for APS 3pm-8pm peak hours. You'll also want to add a return to 100% probably at 8pm. I'm really interested in how this works out from a demand shaving perspective. Good luck!

Hi @Darwin, I just set this up through the app but have a few questions:

• I currently set reserve to 0% and the PW2 charges off-peak. Since my standard reserve is 0%, do I really need to set it back to 100% at 8p thru SmartThings scheduling or should I leave it?
• For settings like "Mode to set", "Time-Based Control Strategy", where I don't want to change from default, should I choose "No Action" or should I pick an option?
• Will I run into any issues if I set a higher log option like debug?

Thanks!

 

[Darwin]

I currently set reserve to 0% and the PW2 charges off-peak. Since my standard reserve is 0%, do I really need to set it back to 100% at 8p thru SmartThings scheduling or should I leave it?

Since you don't have solar, you may be fine not ever setting the reserve to 100%. I admittedly am not very familiar with how exactly the Powerwall operates with grid only and no solar charging. With solar, I've found that the TBC mode "smarts" don't always work/predict well and I sometimes begin peak period with less than 100% battery.

For settings like "Mode to set", "Time-Based Control Strategy", where I don't want to change from default, should I choose "No Action" or should I pick an option?

You can either leave these unselected or choose "No Action" if you don't want the commands for these issued.

Will I run into any issues if I set a higher log option like debug?

Leaving the log output as Debug should be fine. I've never experienced any related performance issues.

 

[Rahul]

Since you don't have solar, you may be fine not ever setting the reserve to 100%. I admittedly am not very familiar with how exactly the Powerwall operates with grid only and no solar charging. With solar, I've found that the TBC mode "smarts" don't always work/predict well and I sometimes begin peak period with less than 100% battery.



You can either leave these unselected or choose "No Action" if you don't want the commands for these issued.



Leaving the log output as Debug should be fine. I've never experienced any related performance issues.

I got everything working and the commands are triggering as expected but something kind of weird happened at the first peak hour. The reserve was set to 80% but it drained down to 76-77% as my 1 AC unit was running. Around when the AC unit stopped, the PW2 started recharging during the peak time and I think it was trying to get back up to 80%. I did not expect it to try to do that during peak time even if below reserve threshold. I manually set it down to 70% but it kept charging. I know there can be a delay of commands but I'm pretty confident the 70% command had reached the PW2 because I killed the app and reloaded it which I think pulls from the API. The charging stopped after a little while but I definitely don't want the unit to recharge itself during peak time even if it goes below the reserve threshold. Any thoughts on this?

 

[blueice89]

We started a video channel on strategies and tips with powerwall and Tesla Energy gateway
Please subscribe , like and leave comments!
Gilbert Solar

 

[Darwin]

I got everything working and the commands are triggering as expected but something kind of weird happened at the first peak hour. The reserve was set to 80% but it drained down to 76-77% as my 1 AC unit was running. Around when the AC unit stopped, the PW2 started recharging during the peak time and I think it was trying to get back up to 80%. I did not expect it to try to do that during peak time even if below reserve threshold. I manually set it down to 70% but it kept charging. I know there can be a delay of commands but I'm pretty confident the 70% command had reached the PW2 because I killed the app and reloaded it which I think pulls from the API. The charging stopped after a little while but I definitely don't want the unit to recharge itself during peak time even if it goes below the reserve threshold. Any thoughts on this?

I have occasionally seen the PW drop a little below the set reserve level, but it's probably less of an issue with solar, since the PW will never draw from the grid for battery charging. The Tesla Powerwall firmware may need to handle grid only hysteresis situations like this a little better. I'm guessing this specific situation may not harm your overall demand situation all that much (aside from charge/discharge efficiencies) since you don't quite have the battery capacity to achieve zero demand anyway. You'll pull from the grid to recharge and cover the 80% vs 77% mistake, but it's just deferring the demand hit you should have had anyway. I may be missing something, but I believe APS does not track instantaneous max demand and only records the single highest hour of net energy use. I know this is how the SRP E-27 and E-15 demand rates work.

How did the rest of the peak period go?

If you want to view/monitor your demand usage in real-time and also check your peak demand hit for the day and month, you can also install and run the SmarthIngs/Hubitat Demand Manager Smart App in monitor only mode: The Demand Manager App for the SmartThings (and Hubitat) Hub - DarwinsDen.com. You'll just need to set your demand cycle to 1 hour and choose the Powerwall as your home energy meter device. You'll get better results if you also set your peak utility schedules in the Demand Manager app since the app polls the Powerwall more frequently during peak periods.

 

[Rahul]

I have occasionally seen the PW drop a little below the set reserve level, but it's probably less of an issue with solar, since the PW will never draw from the grid for battery charging. The Tesla Powerwall firmware may need to handle grid only hysteresis situations like this a little better. I'm guessing this specific situation may not harm your overall demand situation all that much (aside from charge/discharge efficiencies) since you don't quite have the battery capacity to achieve zero demand anyway. You'll pull from the grid to recharge and cover the 80% vs 77% mistake, but it's just deferring the demand hit you should have had anyway. I may be missing something, but I believe APS does not track instantaneous max demand and only records the single highest hour of net energy use. I know this is how the SRP E-27 and E-15 demand rates work.

How did the rest of the peak period go?

If you want to view/monitor your demand usage in real-time and also check your peak demand hit for the day and month, you can also install and run the SmarthIngs/Hubitat Demand Manager Smart App in monitor only mode: The Demand Manager App for the SmartThings (and Hubitat) Hub - DarwinsDen.com. You'll just need to set your demand cycle to 1 hour and choose the Powerwall as your home energy meter device. You'll get better results if you also set your peak utility schedules in the Demand Manager app since the app polls the Powerwall more frequently during peak periods.

Yeah I don't fully undersand how APS is calculating demand over the hour. This is from their tariff sheet:

The kW used to determine the demand charge above will be the Customer’s highest amount of demand (kW) averaged in a one-hour On-Peak period for the billing month.​

I read that as they average the demand over the hour.

I pulled usage data from the PW2:

Electric Peak Usage - 20200526

I averaged each hour and the max hourly average demand I had was 1.767kW which is not bad. Although I think it would've been <= 1kW had the PW2 try not to charge itself.

I'm kinda wondering if I'm just better off just rotating my A/C units on/off via thermostat schedules instead of adjusting % reserve on the hours. Thoughts?

I will check out your Demand Manager app.

Thanks!

 

[Darwin]

I pulled usage data from the PW2:

Electric Peak Usage - 20200526

Looking at the data for yesterday at least (I love looking at that stuff), overall, the single Powerwall handled your peak period demand as good or better than my expectations. As you saw, it looks like that peak demand for the day was 1.7kw, but would have been roughly 4.7kW without the Powerwall. Will be interesting to see what APS actually records for yesterday.

It looks like (roughly) during the worst case 5pm-6pm period, the Powerwall discharged to shave off 3kW of the 4.7kW demand - resulting in that 1.7 kW demand hit. That 3 kW overall discharge from 5pm-6pm is in line with your 20% reserve per hour program (13.5 kWh * 0.2 = 2.7 kWh). From the data, I just don't think the charge/discharge issue affected you much aside from the turnaround efficiency losses, since the grid re-charge draw was just averaged away with the over-enthusiastic Powewall discharge that occurred earlier in the hour. I suppose you may not be as lucky next time though if the extra charge draw occurs in a different hour demand period from the extra discharge.

You're right though. The best a single Powerwall 20% per hour demand shaving schedule can be expected to do is shave off ~2.7kW of demand in any given hour - which it looks like it did yesterday. So maybe $50 per month demand savings in the summer at $17/kW if it continues to work. You should also save $$ with the 13.5 kW per day load shifting - although that may not be as significant with APS. Rotating your AC units (and pre-cooling) has more potential for bigger demand reduction, but I know it can be exponentially more complicated. I think the optimum solution is a combination of the two, but depends on how much effort you're willing to put into it. It's good to have options!

 

[Rahul]

It's funny you mention you love looking at this stuff. I do too. I realize I'm putting more effort into this than the return I'm getting but that's not the point. It's for the sport of it! ;-)

I'll keep you posted.

Thanks!

 

[Rahul]

@Darwin

I just wanted to report back that your app is working pretty well for me. I've definitely got a complicated thermostat schedule but it keeps my demand down. Basically, I precool the whole house from 12-3p, bump up the downstairs temperature @ 3p, and keep the upstairs cooled where we work until 5p, and then swap when we go downstairs. Interestingly my usage per day went up 38% but my daily cost went down by 39% from last year.

 

[Darwin]

@Darwin
I just wanted to report back that your app is working pretty well for me. I've definitely got a complicated thermostat schedule but it keeps my demand down. Basically, I precool the whole house from 12-3p, bump up the downstairs temperature @ 3p, and keep the upstairs cooled where we work until 5p, and then swap when we go downstairs. Interestingly my usage per day went up 38% but my daily cost went down by 39% from last year.

That's great! A 39% daily cost reduction and ~5kW demand drop is nothing to sneeze at considering the 4 A/C units, single Powerwall and no solar. Are you still scheduling the hourly reserve % reductions with the app?

 

[Rahul]

That's great! A 39% daily cost reduction and ~5kW demand drop is nothing to sneeze at considering the 4 A/C units, single Powerwall and no solar. Are you still scheduling the hourly reserve % reductions with the app?

Yes that's exactly how I'm doing it:

3p - 80% reserve
4p - 60% reserve
5p - 40% reserve
6p - 20% reserve
7p - 0% reserve

When I first started with this schedule, I freaked out because if the PW2 sometimes went a little bit under and then started charging during on-peak hours. I definitely didn't want that to happen during peak. I spoke to TSLA and they said yeah it's a little weird but the box is trying to normalize load, whatever that means. Anyway, it seems to not charge for very long so my net is still way better.

 

[StoneDragon]

@Darwin --
First, thanks so much for all the work you've done on the controllers!
I got them loaded and configured the Powerwall controller last night. It's been raining since yesterday, so I was getting concerned by the end of our Demand Charge period.
Some background: I'm in a suburb of Kansas City (on the Kansas side), and Evergy (merger of KCPL and Westar) implemented a Demand Charge this time last year ($9/kW 5/16-9/15, and $3/kW 9/16-5/15, on the max usage at 15 min sampling intervals from 2-7pm M-F, except holidays). Shortly after that, I purchased a Honeywell T9 thermostat and added a 'Screw KCPL' setting from 2-7 that cuts back my AC during that time (I 'pre-charge' from noon til 2 with an extra crank of AC). But like many here, the straight Powerwall algorithms troubled me, as I would see it discharging before peak and not taking weather into account. I found this thread a while back, but got sidetracked by other issues. Anyway...

I have set it up to run as follows (these are M-F):
8:00am -- Backup only (I figure this will force a charge when needed instead of discharging)
1:55pm -- Self-powered/Cost Saving (don't pull from the grid during peak time; 5 min grace period, just in case)
7:05pm -- TBC/Cost Saving (back to my 'normal' operation post-peak. I still want some additional savings from the grid if we have the juice)

For all of these, I selected 'No Action' for the reserve %, thinking that means it'll use the settings in the Powerwall app (I have it at 40% currently, down from my winter setting of 65%). I'd like to have some reserve in case we have a power outage, but we've been very stable on that. <knocks on wood>
I am concerned, though, that as we apparently have another rainy day in store and my battery is currently at 42%, that there still may not be enough to push me through peak time. Does changing the reserve help in those instances? I was worried about the lag in Powerwall's update, and that it might cause a grid use spike when it switches over.
I'll try to report back when I have some data.
Thanks again.

 

[destructure00]

I generally have zero demand each month now that I'm running in Self-Powered. Considering you have 2 Powerwalls, a single AC, and a one hour demand period, I'm guessing you'll get pretty close to zero demand even in TBC mode. Since SRP uses 30 minute demand periods, a single demand spike for me has twice the demand impact as a demand spike during an APS 1 hour demand period. For example, a single aberrant 1kW grid draw for 15 minutes would be a 500W demand hit for me with SRP, but would only be a 250W demand hit for you with APS.

On a side note, I am actually on the SRP E-15 Average demand plan which means even if I have a small ~500-800W demand spike, this bad day will usually get averaged down to zero at the end of the month considering most days in the month are zero max demand. I highly recommend the E-15 Average Demand plan over the E-27 plan for SRP PW users with a reasonable expectation of having zero demand for the month except for maybe a single aberrant demand spike now and again. Even though the E-15 demand rates are higher than the E-27 demand rate, this larger rate multiplied by the zero (usually for me) average demand is still zero. With E-15, you can rest easy if you accidentally blow demand on just one day a month.

Finally got PTO this morning! Already realizing that I don't like the way that TBC works...in off-peak times, it prioritizes sending solar to power the house first, then charging the batteries second, and then exporting anything left over to grid. I want to prioritize charging the batteries first...between that and the small spikes and random discharges to "make room" for extra solar, I want to go a different direction.

From what I've read, it sounds like Backup Only mode and Self-Powered with 100% reserve should operate the same and prioritize charging the battery first, then powering the house, then exporting any excess. Is there any advantage to using your Powerwall manager to set one over the other, with the assumption that I'm going to set Self-Powered with 0% reserve during peak times?

Also, FWIW, I made a virtual switch in Hubitat that turns on at midnight on peak days and off at midnight on off-peak days. I have the APS holiday schedule integrated into the scheduler for this switch rather than following a strict Mon-Fri schedule, and want to use it to conditionally set the PW modes based on the combination of day of week and APS off-peak holidays. I didn't see a way to use a switch as a restriction in the PW manager, so for now I'm using Rule Machine to set the PW mode and reserve levels. Maybe something to consider integrating at some point.

 

[BrettS]

Finally got PTO this morning! Already realizing that I don't like the way that TBC works...in off-peak times, it prioritizes sending solar to power the house first, then charging the batteries second, and then exporting anything left over to grid. I want to prioritize charging the batteries first...between that and the small spikes and random discharges to "make room" for extra solar, I want to go a different direction.

I believe that TBC does have some intelligence built in, in that it will attempt to make sure your powerwalls are charged enough for you to make it through your peak time without drawing from the grid. However, that may mean that it doesn’t get the powerwalls up to 100% every day.

And I think there is some sense in this. Remember that energy is lost as it goes through the powerwalls. They are rated for 90% round trip efficiency, but some people here have said they they are seeing it as low as 80 or 85%. So it doesn’t really make sense to prioritize charging the powerwalls, assuming that they powerwalls do have enough of a charge for you to get through your peak times. It would be more efficient to allow the solar energy to power your home at the time it is generated, rather than lose 10% or more of that power when you take it out of the powerwalls later.

 

[BrettS]

From what I've read, it sounds like Backup Only mode and Self-Powered with 100% reserve should operate the same and prioritize charging the battery first, then powering the house, then exporting any excess. Is there any advantage to using your Powerwall manager to set one over the other, with the assumption that I'm going to set Self-Powered with 0% reserve during peak times?

Backup only and self powered with 100% reserve are essentially the same. But they will prioritize charging the powerwalls and if the powerwalls are at less than 100% the solar power will go to the powerwalls first, then the house, then the grid.

 

[destructure00]

And I think there is some sense in this. Remember that energy is lost as it goes through the powerwalls. They are rated for 90% round trip efficiency, but some people here have said they they are seeing it as low as 80 or 85%. So it doesn’t really make sense to prioritize charging the powerwalls, assuming that they powerwalls do have enough of a charge for you to get through your peak times. It would be more efficient to allow the solar energy to power your home at the time it is generated, rather than lose 10% or more of that power when you take it out of the powerwalls later.

The plan I'm on has cheap kwh rates for both off-peak ($0.05) and on-peak ($0.08). Where they get you is the peak demand charge...$17.50/kw during the on-peak hours. So I'm less worried about small losses and more worried about having the batteries 100% charged so I can make it through the 5 hour window running the A/C nearly full-time (I'm in Phoenix, it's 110+ outside, and it's monsoon season where we get afternoon storms that can block solar production and potentially knock out power after the sun goes down), and pull zero or as close to zero as possible from the grid. That's my motivation to prioritize charging the batteries first.

Once I get through the rest of the summer seeing how this system performs, I may re-evaluate the plan I'm on and see if I could do better on a different plan with higher kwh rates and lower or no demand charges, which would change my strategy. TBC/Cost Savings could be a better route then.