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

Strategies for Powerwalls and Utility Demand Plans

This site may earn commission on affiliate links.
That is not the behavior I want. It wish Tesla would keep things simple and implement scheduled self-consumption before jumping into estimating my needs tomorrow. My observations are that they are not taking local weather conditions into account (today and tomorrow) when deciding how much to discharge. My area is looking at 6 days of cloudy/rainy weather. I have seen the powerwall drain itself in similar situations. That is a bad move for someone on a demand plan.

In self-consumption mode, the batteries do not seriously begin to discharge until 4pm for cloudy days and 6:30pm for sunny days. With 6 days of clouds and rain, time based control could leave me without enough to traverse the demand period.

Got it. I have the same power company (PG&E) as the Tesla mothership so I'm not surprised that they configured things to work perfectly for my situation and not for people in other areas.

When I was trying to get my powerwall to behave, I came across this article Powerwall2 update – Time-of-Use enabled that includes a description of some custom code that a powerwall user wrote (BJReplay/PowerwallService) to get more control over how the powerwall functions. Parts of this may be specific to Australia and will probably require some customization to work in the US. But the description sounds kind of what I think you are trying to do:

"Broadly, the service sends your solar array details to an external solar-output estimation provider, who combines them with local weather forecasts, to return a tailored estimate of your future energy production to the service. You define your estimated usage and the amount of power your PW2 needs to retain, the TOU periods and tariffs you have and so on in a .config text file. Then, the service runs to calculate the amount it needs to save for the following day, taking in to account the estimated future solar generation, and it will use your Tesla cloud login to switch the PW2 modes from “self-powered” to “standby” or “backup” (charging)."
 
That is not the behavior I want. It wish Tesla would keep things simple and implement scheduled self-consumption before jumping into estimating my needs tomorrow. My observations are that they are not taking local weather conditions into account (today and tomorrow) when deciding how much to discharge. My area is looking at 6 days of cloudy/rainy weather. I have seen the powerwall drain itself in similar situations. That is a bad move for someone on a demand plan.

In self-consumption mode, the batteries do not seriously begin to discharge until 4pm for cloudy days and 6:30pm for sunny days. With 6 days of clouds and rain, time based control could leave me without enough to traverse the demand period.

I tried to dig a little deeper into your request and the two TBC policies Tesla provides. I really think what you need is a third policy and you cannot use one of the existing ones to achieve your goal.

Here is a table I made to get my head around things:
Screen Shot 2019-05-02 at 4.28.32 PM.png

You want the last case with the Powerwall charging up to 100% . I would imagine that there is a larger number of customers with such need but don't know a good way to get this across to Tesla.
 
  • Like
Reactions: JES2
maybe make the peak just the last 3 hours when you have no solar production. Try it on a day when you know you won't exceed your capacity to be sure though.

I added a two hour peak the end of the shoulder period (after solar stopped pruducing) and that didn't work very well on my trial last night. With the peak period added, the PW would not discharge at all during the shoulder period when the air conditioning came on. The shoulder period without any peak had been working almost flawlessly for demand management for the last two months, but I'm guessing with the added peak period, the PW outprioritized preceived needs of the peak period to the point that it wouldn't cover shoulder net usage. By my calculations there should have had plenty of reserve to cover both shoulder and peak. Fortunately the resultant demand hit didn't end up being too bad. I have software running continuously that is monitoring my home demand, and once it detected the demand hit, it bumped up the temperature on my AC, halting it. The software hadn't been needed since I had the PW installed, but came in handy this time. It's good to have as a backup.

Better is the enemy of good enough I guess. I'll stick with a single shoulder period for demand management purposes. The PW with a single shoulder period is really is working well for demand management me. The discharge in off-peak is a concern from an longevity/efficiency perspective, but I have to believe Tesla will address that at some point.

Another side note - it seems that if I log out of the web UI and then log back in, the time-based control app settings and reserve are fairly quick to update on the PW. I need to do more testing to verify this is actually occurring, but it does sound similar to the results @DanAman was getting by disconnecting the ethernet cable for a minute or so.
 
Last edited:
  • Like
Reactions: JES2
It seems clear that Tesla is not concerned with demand customers. Before I purchased the powerwalls I told Tesla my intended use case and they agreed that the powerwall would work fine. They told me that the ability to schedule changes was coming. When I was finally given access, it was clear this was not what I needed. I can script something basic to change the reserve and have it run via a cron job. I did not want to have to do this because I hoped the Tesla tools would improver.

I looked at the PowerwallService tool and it looks interesting. I don't want predictive charging and discharge. I only want to discharge during a specific time period. This is especially important in the winter because of the reduction in sunlight. Also, I don't have a windows machine to run the service. As I said, I like to keep things simple. I have net metering with a demand charge. To me, this means that I want to make sure I have enough energy stored to last through my 1pm - 8pm demand period without thinking about what appliances I use regardless of sunshine. My preference would be for solar to be used in the following way.

Charging (non-demand period)
  • Fill the battery as fast as possible in the morning
  • Excess solar used to satisfy household loads
  • Excess solar from the household goes to filling the battery
  • If the battery does not need it, send it to the grid

Demand period discharge
  • Solar used for household loads
  • Anything not covered by solar is supplied from the batteries
  • Anything not covered by solar and the batteries and comes from the grid

Demand period charging
  • Solar used to satisfy household loads
  • Excess solar from household loads sent to the batteries
  • If the battery does not need it, send it to the grid
If you have net metering, the grid is another battery. I want as much energy to take me through a long demand period because there could be days with little sunshine.

Screenshot_20190503-131053_resized.png


Got it. I have the same power company (PG&E) as the Tesla mothership so I'm not surprised that they configured things to work perfectly for my situation and not for people in other areas.

When I was trying to get my powerwall to behave, I came across this article Powerwall2 update – Time-of-Use enabled that includes a description of some custom code that a powerwall user wrote (BJReplay/PowerwallService) to get more control over how the powerwall functions. Parts of this may be specific to Australia and will probably require some customization to work in the US. But the description sounds kind of what I think you are trying to do:
 
  • Like
Reactions: JES2
I added a two hour peak the end of the shoulder period (after solar stopped pruducing) and that didn't work very well on my trial last night. With the peak period added, the PW would not discharge at all during the shoulder period when the air conditioning came on.

Well I'm sorry that didn't work out for you and hope it didn't bust your demand for this month. You are probably best staying on shoulder only until they come up with a better system for this type of tariff.
 
  • Like
Reactions: Darwin
I added a two hour peak the end of the shoulder period (after solar stopped pruducing) and that didn't work very well on my trial last night. With the peak period added, the PW would not discharge at all during the shoulder period when the air conditioning came on. The shoulder period without any peak had been working almost flawlessly for demand management for the last two months, but I'm guessing with the added peak period, the PW outprioritized preceived needs of the peak period to the point that it wouldn't cover shoulder net usage. By my calculations there should have had plenty of reserve to cover both shoulder and peak. Fortunately the resultant demand hit didn't end up being too bad. I have software running continuously that is monitoring my home demand, and once it detected the demand hit, it bumped up the temperature on my AC, halting it. The software hadn't been needed since I had the PW installed, but came in handy this time. It's good to have as a backup.

Better is the enemy of good enough I guess. I'll stick with a single shoulder period for demand management purposes. The PW with a single shoulder period is really is working well for demand management me. The discharge in off-peak is a concern from an longevity/efficiency perspective, but I have to believe Tesla will address that at some point.

Another side note - it seems that if I log out of the web UI and then log back in, the time-based control app settings and reserve are fairly quick to update on the PW. I need to do more testing to verify this is actually occurring, but it does sound similar to the results @DanAman was getting by disconnecting the ethernet cable for a minute or so.
Hey! I sent you a PM for a few questions, looking to getting Tesla solar and Powerwall here in SRP area but it’s daunting because of the possible demand charges.
 
My preference would be for solar to be used in the following way.

Charging (non-demand period)
  • Fill the battery as fast as possible in the morning
  • Excess solar used to satisfy household loads
  • Excess solar from the household goes to filling the battery
  • If the battery does not need it, send it to the grid

Demand period discharge
  • Solar used for household loads
  • Anything not covered by solar is supplied from the batteries
  • Anything not covered by solar and the batteries and comes from the grid

Demand period charging
  • Solar used to satisfy household loads
  • Excess solar from household loads sent to the batteries
  • If the battery does not need it, send it to the grid

This is exactly the behavior I'm getting with a shoulder period in cost saving (without a peak period), unless I'm misinterpreting.

Hopefully you'll have luck if you end up re-retrying over the weekend without the risk. If it works for you, I don't think you'll see demand as low as the 0.033kW you're getting, but I'm fine with my 0.2kW demand for a hands-off system. Most 30 minute demand periods the PW keeps demand close to zero during shoulder, but for some reason, every once in a while it takes a minute or two for it to kick in to cover surge use.

To add insult to injury, my demand with SRP is determined in 30 minute cycles - and exporting doesn't count for demand purposes. So if I use 2kW from the grid for 15 minutes and then export 2kW for 15 minutes, my demand is 1kW (2kW + 0kW)/2 instead of 0 (2kW - 2kW)/2. They make it really hard on you.

The battery discharge at night in off-peak is annoying and wasteful, but so far at least its predictive behavior for that has worked. It has never left me short on power for the next day, but then again, we don't have many cloudy days here. If I have concerns over a rainy days ahead, I suppose I could always bump up the reserve for the night.
 
Last edited:
I have software running continuously that is monitoring my home demand, and once it detected the demand hit, it bumped up the temperature on my AC, halting it. The software hadn't been needed since I had the PW installed, but came in handy this time. It's good to have as a backup.

What kind of software do you have that can adjust your AC unit, and what kind of thermostat is on that AC unit!? I’d like some more control over my AC but never heard of anything like this before.
 
What kind of software do you have that can adjust your AC unit, and what kind of thermostat is on that AC unit!? I’d like some more control over my AC but never heard of anything like this before.

I am using a Demand Management app for the SmartThings hub. I actually wrote it a while back to help manage my demand with solar, but its available free for anyone who wants to give it a try.

There's more information about the software, platform it runs on and required and optional devices at: DarwinsDen Demand Manager.
If you're interested in giving it a try, I'd be happy to help with any questions or issues. It's beta at best software with only a few users, so I wouldn't doubt you would experience some initial hiccups with it, but It is working pretty well for me.

The app requires a SmartThings home automation hub (about $70) and a SmartThings compatibe thermostat. It's been tested with Nest and Ecobee thermostats, but theoretically should work with almost any network enabled thermostat. It runs in the SmartThings cloud - which comes with some network reliabiliy risks, but I plan on porting it to run on a local hub at some point.

The app install will be pretty simple for those familiar with SmartThings custom apps. I use SmartThings with smart light switches and outlets extensively to help manage and shift my electric usage. The app currently requires an Aeotec Home Energy meter (about $95) or other meter that is compatibe with SmartThings. I should be able to modify the software to pull home usage directly off the PW gateway (and eliminate the need for the Aeotec meter for those with PW's) once I get a little time. Other indicator devices are optional.

A brief sumary of features (some more polished than others):
  • Monitor and display your (30 or 60 minute) demand usage data in real-time.
  • Automatically turn on and off devices in your home and adjust your home thermostat cooling temperature based on your demand goal and real-time demand projections.
  • Get notified when you’re approaching your demand goal limit via your mobile device or indicator devices that can be placed throughout your home.
  • Customize a 30 or 60-minute demand target depending on your utility plan.
  • Set a demand management schedule based on your utility’s peak usage hours.
  • Let your family know whether you are in a peak utility period via simple red/green indicator lights that can be placed at any power outlet.
  • Display the current power output of your solar array using LED level indicator devices.
  • Create optional virtual demand meters for display on dashboards such as ActionTiles™.
  • Pre-cool your home on scheduled peak utility days.
  • Monitor your solar array inverter production for improved predictive demand projections
 
Last edited:
  • Like
Reactions: tracksyde
Ahh, sounds like it is software tailored to your specific situation. So basically the AC portion is a more complex/custom IFTTT type solution that interfaces with an existing smart thermostat to adjust temperatures when certain conditions exist regarding peak demand usage (among other things)? Unfortunately none of this would apply for me since my electrical company charges a flat rate at all times (and I'm just lurking here since I have no solar or powerwall).

I did review your site and may look more into smartthings since expanding my home automation is something I am interested in, especially if it can run custom programs.

I don't want to hijack the thread here...carry on.
 
That's pretty much it. The software attempts do do other things, but the main focus is managing air conditioning in response to demand as AC is a huge power consumer and not easily load shifted into off-peak hours.

In a nutshell, the app will continuously project out your estimated 30 or 60 minute cycle demand. Based on your current demand projection, demand goals, and current thermostat conditions (and some other preference options) the app will either halt your AC (by bumping up and then back down your thermostat set-point to trigger the hysteresis check) or just bump up the set point.

It is designed to work best in conjunction with pre-cooling your home - and also provides some simple pre-cooling settings. Now that I have the PW, I'm not currently pre-cooling, and the thermostat controls only kick in if something goes wrong and the PW doesn't properly manage demand.

Aside from the AC controls, the app can turn off smart switches and devices during peak hours - or only when your demand goal is exceeded. There's several optional display indicator devices you can add as well for peak period indication, current solar production and current demand status. Again, it's all beta and likely has some issues, but I'm willing to help with fixes and make enhancements as time permits, if there are suggestions.
 
I experimented with time based control this weekend. As suggested I did not define a peak.

weekday-schedule.png
weekend-scedule.png



These are the balanced mode results, lots of discharging and pulling from the grid during the shoulder period. My reserve was set to 20%. The highest pull from the grid during the shoulder period was 1.8kW.

saturday-weirdo-flow.png
daturday-grid-battery-chart.png


On Sunday I switched to cost savings. these are the results. I did not record any of the undesired grid pulls, but there were some.

sunday-grid-battery-chart.png


Things looked good so I decided to try a real world run of cost savings mode with a reserve of 50%. The result was my demand went from 0.033kW to 0.384kW. I observed higher peaks in the chart. i will post the chart tomorrow after observing the off-peak batter discharge.
 
We had a powerwall installed three weeks ago and I had all sorts of software issues for the first couple of weeks. One thing I didn't realize was that it takes an hour or two for a setting entered into the app to make it to the powerwall. I didn't figure this out until I used a computer to directly connect to the powerwall gateway via wifi and look at the settings using the web-based powerwall UI. I also had a situation where things would seem to work after Tesla would visit and tinker with the powerwall. But then I would play with the app and the powerwall would start to misbehave and show obviously incorrect data in the app (solar energy was correct, but home consumption was shown as exactly matching solar and grid energy was shown as zero which didn't match the electric meter or common sense). I could also get things to behave again by hitting the reset button on the gateway and then re-entering the wifi info and other stuff through the web-based powerwall UI. I was able to get self-consumption mode to work as long as I didn't try to change the reserve setting in the app. I also turned off storm watch just in case that was messing things up. Eventually I gave the Advanced time-based control a try (with the cost saving option) and much to my surprise it worked perfectly! I haven't changed the settings in the app since then and the powerwall now does exactly what I want it to given our time-of-use plan.

In the morning before peak prices start, all the solar is being used to charge the battery and the power to run the house is coming from the grid. When the battery fills up then the excess solar is sold to the grid. Once we hit 2 pm when peak prices start, the powerwall switches to using the battery to run the house and sells ALL the solar to the grid. Once the sun goes down, the house is powered exclusively by the battery until 9 pm when the peak prices end. Then the house switches back to the grid.

In order to qualify for the federal tax credit and/or to make our local utility happy it seems like the powerwall is only able to charge from solar and can only use this stored energy to power the house rather than selling it back to the grid. Under these constraints, the powerwall seems to be behaving optimally.

I can't be completely sure if rebooting the gateway + leaving the setting on the app alone is what fixed the problem. I also sent lots of emails to Tesla technician who installed the system and he escalated things to his manager so it's possible Tesla did something remotely to fix things.

The TL;DR version is to try rebooting the powerwall gateway, connecting directly through wifi to the gateway, and leaving the settings on the app alone as tempting as it is to tinker with things.

Thanks for the info. I don't currently have solar and thinking to use the powerwall to store energy during off peak hours from the grid and to use the battery during on peak hours. Previously the option was just not there to have battery kick in on demand. Do you see that option right now in the settings right now?
 
I had a similar dilemma with solar only and that is how I ended up getting Powerwall 2’s. My public utility forces all solar customers on a net metering plan which charges roughly $0.15 per kWh between 12-9PM weekdays and $0.10 per kWh for weekends and holidays. If you make any excess for the billing period they credit you at wholesale rate. There is also a demand charge of $2 per kW between May and November and $1.75 per kW between November and May.

After experimenting there is just nothing in the Powerwall control settings that could possibly minimize charges. My experience has been for the most part good with the Powerwalls but I’ve always been annoyed by how punitive the rates are for this utility’s solar customers.

The later hot months here of July and August I usually deplete the battery at night since I generate slightly less than what the house consumes and I end up incurring about 3.5 kW demand. In the colder months such as December and January I usually run the furnace and I also don’t make enough solar to offset the usage through the day so I’ll usually hit demand of 1 to 2 kW. Spring I usually have way more energy than I need so I usually won’t have demand charges. But it is spring where this utility’s net metering is pretty much a scam. The only time that I really overproduce is spring and since the utility pays wholesale rate for any energy sold to them I get a big fat zero for this energy I put back on to the grid. Why you ask? California has an excess of energy during the spring, and specifically during the day so wholesale prices during the spring are usually zero or even negative. Thus the only time of the year I overproduce I get paid zero for it.

My only strategy has been to conserve during the months where I come up short (July, August, December, and January) and pretty much be wasteful during the excess production months (February through June). I’m essentially not getting paid for excess electricity produced during these months so I might as well use it.

My conservation basically is being careful not to use anything that is high powered when I’m on grid power: electric dryer, electric stove, etc.. In this case I try to get any clothes drying or cooking done before the battery depletes and that way the worse case scenario on late summer nights is the AC which is about 3.5 kW or during winter nights the furnace fan or a space heater which is about 1 kW each. I don’t charge my electric cars at home other than during the spring since I charge at work. This approach to demand requires careful planning especially when the whole family doesn’t think in terms of demand or for the most part may not care when appliances are ran.
 
Thanks for the info. I don't currently have solar and thinking to use the powerwall to store energy during off peak hours from the grid and to use the battery during on peak hours. Previously the option was just not there to have battery kick in on demand. Do you see that option right now in the settings right now?
Yes, if you don't have solar you can charge during off-peak and supply your house during Peak. This is enabled in Advanced - Time Based Control.
 
  • Informative
Reactions: tracksyde
Thanks for the info. I don't currently have solar and thinking to use the powerwall to store energy during off peak hours from the grid and to use the battery during on peak hours. Previously the option was just not there to have battery kick in on demand. Do you see that option right now in the settings right now?

There is no specific setting for demand plans. The "Self-powered" mode works well for demand plans. We pretty much use all the major appliances during the demand period except water heater and clothes dryer. This typically results in a demand of 0.015 kW to 0.033 kW. The only problem with this method is that it cannot be scheduled. My demand period in the summer is from 1:01pm - 8:01pm. I adjust the battery reserve from 100% at 11:50am to 30%, and to 100% at 8:02pm. I set phone alarms to remind me when to make the change.

The other option explored in this thread is to use the Advanced - Time-based control, cost savings mode. While not specifically designed for demand plans, it seems to come closest to what you would want. It schedules the starting/stopping of responding to demands. The downsides are that it is not as aggressive about responding to demand. It will pull from the grid to help while it ramps up the batteries. This results in a higher demand measurement 0.2 kW to 0.4 kW. The demand could be higher based on what turns on during the period. The other downside is that it discharges the batteries to your set reserve level after satisfying demand. The advanced modes wants to help you run your house from solar as much as possible.

If you have net metering in which energy sent to the grid is worth the same as energy received from the grid, then running off the batteries as much as possible is not as necessary. I have this kind of net metering with a side of demand charges, so I use the powerwall to ride out the demand period and act as a backup as a secondary feature.
 
For solar customers, my utility lowered the per kWh cost of electricity and added a demand charge. This may be why we get get 1:1 credits. A standard customer pays about $0.091 and a solar customer pays $0.065. The demand charge is $5.27/kW.

Self-powered works really well for shaving my demand down to almost nothing. My last billed demand charge was $0.30. It just requires managing the reserve if you want to guarantee you will have enough power for the next day. Winter is a bigger challenge because we produce 40% less power than spring and the demand period is from 6am - 10am. The heater pulls more power than the AC so that also drains the batteries faster. My best strategy for managing that was to lower the battery reserve at night before I go to sleep and have an alarm remind me to change the reserve to 100% at 10:02am. There should be enough sunlight to charge the batteries for the next day.

That's how the powerwall is helping me fight the demand charge.

I want to start another thread about what we powerwall demand customers want from Tesla so we can flood them with well thought out feature requests. Maybe tweet at Elon Musk with our requirements. I don't know how many utilities push demand plans onto solar customers.
 
  • Like
Reactions: Darwin
Things looked good so I decided to try a real world run of cost savings mode with a reserve of 50%. The result was my demand went from 0.033kW to 0.384kW. I observed higher peaks in the chart. i will post the chart tomorrow after observing the off-peak batter discharge.

I'm having a hard time seeing any net on-peak (or on-shoulder in this case) grid usage in the cost savings screenshot. Is this just due to the battery overlay or did the PW perform that well on this trial?

I've found that the PW is most days very good at keeping my demand at 0.050kW or less in shoulder / cost saving, but every few days it seems slow to respond to net usage. It seems to be the most sluggish responding in late afternoon after a full day of solar production and if the battery has been idle for awhile. At that time, if the AC kicks in when solar is just enough to cover nominal home usage, it can take several minutes before the battery will begin discharging to cover the added AC load.

Fortunately, my utility just started offering a new Average Demand Plan. With this new plan, you are charged at a higher demand rate, but your calculated demand is the average of each days' maximum 30 minute demand instead of the maximum demand for any 30 minute period for the month. This new plan seems to work much better with my very low, but somewhat erratic demand (0-200W) each day. My old max monthly demand was 200W, but I expect my average daily peak demand to be closer to 50W.