-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Time-Based Control?
- Thread starter powersaver
- Start date
-
- Tags
- Powerwall Tesla Energy
So, thought I'd share my experience so far with Tesla Time Based Control since I enabled it 11 days ago (Wed 4th April).
First up, some 'baseline' data.
I'm located in Sydney, Australia and my power retailer is Powershop (Ausgrid is the supplier).
My 'effective' rates after 21% discount are:
Using Tesla app 3.3.5 on Android, I have set my Tesla Powerwall battery to Time-Based Control (aka TBC or ToU), and the sub-setting at "Cost Saving" (as opposed to Balanced). The webpage definitions can be found here (note: this is the AU document as opposed to the US version of the document. If you review the Modes of Operation with Solar there IS a difference between the US and AU versions. US version can be found here.
Now, with all that out of the way, let's get on to some assumptions (please feel free to question or challenge in case I am wrong on any of these)!
Due to:
Firstly, my battery level data is all logged and graphed here. I find that this makes it a LOT simpler to track the results of what I observe the app doing. This is a skunkworks solution by a good friend of mine (author of the Powerwall Companion app on Windows 10) and is not available for sharing, so please don't ask. I'm also logging to PVOutput here and using the PVOutput Integration Service to grab my Powerwall data via a local WiFi connection to the battery which is a little flaky.. but is usually logged to extended data.
During the week, I found it worked exactly as I expected. It seemed to grid charge as soon as I hit off-peak (10pm) and fill the battery to around 40-50%, The next day, Solar generated power would continue to top-up during the day, while the house drew from Solar and Battery during Peak period. The battery was generally down to under 20% by the end of the Peak period. Given the significant variance in household power consumption, I'm pretty happy with that.
What I DON'T think is working correctly, is that on weekends, I was still seeing the battery grid charging, and then discharging during my Shoulder periods. As per assumptions listed above, this doesn't actually seem to be saving me money, in fact, I believe it is costing me! I feel that this is caused by the current version of the Tesla app only asking you to provide peak/off-peak time windows, but not your actual tariffs. Given the variety of tariffs available, and in particular, those with a FIT HIGHER than their Off-peak rate, and those with a larger differential between their Off-peak and Shoulder rates, there are quite a number of different scenarios that could play out.
During the week, I might switch over to the "Balanced" Time-based Control setting and see if it alleviates the weekend behaviour.
Note: Prior to enabling the Tesla Time-based Control, I did use Bernard's (aka user 1253 on Whirlpool and BJReplay on TMC Forums) PowerwallService for Windows and found it excellent. My inner geek would gladly still be using it if not for the fact that my system was enabled by Tesla for T.B.C quite early in the staged rollout.
Given the lack of transparency from Tesla about what factors they take into consideration (beyond what they have documented publicly in the aforementioned Modes of Operation With Solar), what assumptions are made, and what priorities or weightings are assigned to different pieces of data, plus the lack of good long-term monitoring and reporting tools, I still think there is great opportunity for a community developed solution. IMHO, it would need to:
Thanks for reading.
First up, some 'baseline' data.
I'm located in Sydney, Australia and my power retailer is Powershop (Ausgrid is the supplier).
My 'effective' rates after 21% discount are:
- Off-peak: 17.92 c/kWh
- Shoulder: 19.64 c/kWh
- Peak: 40.37 c/kWh
- Daily Supply Charge: 94.29c
- Solar Feed-in (F.I.T.): 12.8 c/kWh
- Weekdays: Peak 2pm-8pm, Shoulder 7am-2pm and 8pm-10pm, Off-peak 10pm - 7am
- Weekends: Shoulder 7am-10pm, Off-peak 10pm-7am
Using Tesla app 3.3.5 on Android, I have set my Tesla Powerwall battery to Time-Based Control (aka TBC or ToU), and the sub-setting at "Cost Saving" (as opposed to Balanced). The webpage definitions can be found here (note: this is the AU document as opposed to the US version of the document. If you review the Modes of Operation with Solar there IS a difference between the US and AU versions. US version can be found here.
Now, with all that out of the way, let's get on to some assumptions (please feel free to question or challenge in case I am wrong on any of these)!
Due to:
- My F.I.T. (12.8 c/kWh) being lower than my Off-peak (17.92 c/kWh), AND
- My Off-peak (17.92 c/kWh) and Shoulder (19.64 c/kWh) being so similar (only 9.16% difference)
- My Solar generated power to be stored and only fed to the grid if the battery is full
- DON'T charge battery from the grid (off-peak) on Friday and Saturday nights to discharge during Shoulder on Saturday and Sunday as the efficiency loss and battery wear and tear is higher than the 9% price difference.
- DO charge the battery from grid Sun-Thur nights (off-peak) but only as much as is required so that when we hit Peak period, we can entirely power the house from Solar with battery covering the delta if house consumption exceeds Solar generation through the entire Peak period.
Firstly, my battery level data is all logged and graphed here. I find that this makes it a LOT simpler to track the results of what I observe the app doing. This is a skunkworks solution by a good friend of mine (author of the Powerwall Companion app on Windows 10) and is not available for sharing, so please don't ask. I'm also logging to PVOutput here and using the PVOutput Integration Service to grab my Powerwall data via a local WiFi connection to the battery which is a little flaky.. but is usually logged to extended data.
During the week, I found it worked exactly as I expected. It seemed to grid charge as soon as I hit off-peak (10pm) and fill the battery to around 40-50%, The next day, Solar generated power would continue to top-up during the day, while the house drew from Solar and Battery during Peak period. The battery was generally down to under 20% by the end of the Peak period. Given the significant variance in household power consumption, I'm pretty happy with that.
What I DON'T think is working correctly, is that on weekends, I was still seeing the battery grid charging, and then discharging during my Shoulder periods. As per assumptions listed above, this doesn't actually seem to be saving me money, in fact, I believe it is costing me! I feel that this is caused by the current version of the Tesla app only asking you to provide peak/off-peak time windows, but not your actual tariffs. Given the variety of tariffs available, and in particular, those with a FIT HIGHER than their Off-peak rate, and those with a larger differential between their Off-peak and Shoulder rates, there are quite a number of different scenarios that could play out.
During the week, I might switch over to the "Balanced" Time-based Control setting and see if it alleviates the weekend behaviour.
Note: Prior to enabling the Tesla Time-based Control, I did use Bernard's (aka user 1253 on Whirlpool and BJReplay on TMC Forums) PowerwallService for Windows and found it excellent. My inner geek would gladly still be using it if not for the fact that my system was enabled by Tesla for T.B.C quite early in the staged rollout.
Given the lack of transparency from Tesla about what factors they take into consideration (beyond what they have documented publicly in the aforementioned Modes of Operation With Solar), what assumptions are made, and what priorities or weightings are assigned to different pieces of data, plus the lack of good long-term monitoring and reporting tools, I still think there is great opportunity for a community developed solution. IMHO, it would need to:
- Pull data from Tesla cloud (universally available to everyone), as well as via local URL (not everyone has local WiFi connectivity to their PW2)
- Provide long-term reporting capability
- Use actual tariff data AND ToU windows to transparently lay out the charging/discharging priorities
- Use Machine learning to:
- Learn power consumption habits of the household
- Predict weather and likely Solar PV production the following day/s to determine how much grid charging to conduct, and when
- Learn power consumption habits of the household
- Based on collected data, report on 'health' of the battery, and potentially even the Solar panels.
Thanks for reading.
that's a great write up of TBC.
My understanding is that the price schedule does not have to represent your exact tariff, but rather if there is economic benefit to load shift. By setting off-peak and shoulder during a weekend - Powerwall is just doing what you tell it and attempting to shift load between those two periods.
In your case - it seems to be there is no economic benefit to shift load on the weekend - so you should just set the weekend to 'all day off peak' and then Powerwall will know there is no need to do anything until Sunday night.
If you believe the efficiency losses are worse than the cost difference - then you shouldn't be putting any additional throughput on the battery and only avoiding the lower rate FiT by basically doing self consumption on the weekends. You're right though - if the FiT is higher than off peak - then I don't see a way to capture that in the current settings as you would want to discharge to gross home load during off peak (~5c/kWh benefit), but personally, I would prefer to keep that in case of a shortfall in peak the next day (~27c/kWh benefit).
My understanding is that the price schedule does not have to represent your exact tariff, but rather if there is economic benefit to load shift. By setting off-peak and shoulder during a weekend - Powerwall is just doing what you tell it and attempting to shift load between those two periods.
In your case - it seems to be there is no economic benefit to shift load on the weekend - so you should just set the weekend to 'all day off peak' and then Powerwall will know there is no need to do anything until Sunday night.
If you believe the efficiency losses are worse than the cost difference - then you shouldn't be putting any additional throughput on the battery and only avoiding the lower rate FiT by basically doing self consumption on the weekends. You're right though - if the FiT is higher than off peak - then I don't see a way to capture that in the current settings as you would want to discharge to gross home load during off peak (~5c/kWh benefit), but personally, I would prefer to keep that in case of a shortfall in peak the next day (~27c/kWh benefit).
Dborn
Confirmed
Hi. My setup is nearly identical to yours. Main differences are all panels are north and total is 6kw. I too am with powershop here in Sydney.
My main issue is that my system regularly starts discharging weekdays at around 4.30am, while still on offpeak . I have no idea why. There are no abnormal loads at that time.
Secondarily, I find the system often going partially to the grid even when solar is insufficient for the house, and even when the battery is not fully charged.
Can you explain the difference between balanced and cost saving? I really have difficulty understanding what the difference is.
That's a great post......
Great to know that there is a Windows app for PW2, I've tried it...... not bad at all.
I'm located in Sydney, Australia and my electricity retailer is AGL (Endeavour Energy is the distributor).
My 'effective' rates after 32% discount are:
- Off-peak: 11.97 c/kWh
- Shoulder: 23.92 c/kWh
- Peak: 28.42 c/kWh
- Daily Supply Charge: 97.9c
- Solar Feed-in (F.I.T.): 11.1 c/kWh
- Weekdays: Peak 1pm-8pm, Shoulder 7am-1pm and 8pm-10pm, Off-peak 10pm - 7am
- Weekends: Off-peak all weekend + non-business days (e.g., public holidays).
My situation is whether it's worthwhile discharing the PW2 on the weekends given that it's off-peak tariff all weekend and the parity between FiT and Off-Peak is very small (see above and factoring about 10-15% loss in storage). So, is it better to set the reserve on the PW2 to 100% all weekend and then Monday morning change to 0% reserve? That way I have a full battery to kick off Monday and pay off-peak rates for any net grid usage from Friday 10pm to Monday 7am?
I would probably only do this for Winter as solar production will decline and heating loads increase and therefore need to preserve battery for the 5 day peak/shoulder period during weekdays (and possibly top up overnight from off-peak for next day in case of bad weather).
In summer/autum/spring, I would probably just let the PW2 be in self-powered mode do its thing 7 days a week and I will end up with surplus export anyway.
I guess the trick is to balance the FiT over off-peak rates + losses in storing. Do we save the wear and tear of the battery being cycled or pay less profits for grid use to the retailer? FiT would need to cancel out the grid off-peak rates , so my fomula would be FiT in $ = 1.15 * Off-peak rates in $. So I would need to export about 1.24x to match the off-peak usage. (11.97 * 1.15 / 11.1) (Off peak $ * 1.15 divided by FiT $) with the aim that FiT > grid import. 1.15 is the 15% loss factor.
My thinking is that for my rates, avoiding charging to provide off peak energy, and running from grid for off peak is best:
Rates (all inclusive of GST):
So, for every kWh of PV that I use to charge my battery, I lose just over 1c discharging into off-peak, but pick up almost 8c discharging into peak.
By putting my PW2 into standby at the end of peak (by setting backup_reserve_percentage to the current state of charge), I preserve that charge for using in peak the next day, and, I also mean it is more likely that I'll spill that excess generation to the grid at feed in rates but still have a fully charged battery for the evening peak (because it's starting with a greater charge that it would if it were discharging into off peak).
In summer, it's almost certain to be full by 3pm, so any excess will feed in - potentially at 29c/kWh, though realistically at 11.3 until 1 July and 9.9 after 1 July because I can't see any retailer choosing to offer 29c/kWh yet.
Even when FIT drops to 9.9c, a stored kWh will cost me 9.9 * 1.15 = 11.38c/kWh - still more than the cost of off-peak from the grid.
So I'm better off "resting" my PW2 in standby overnight (and all weekend), paying for off-peak, and spilling excess for FIT. In the depths of winter, I'll probably charge around 7kWh per day, perhaps up to 10kWh for really cloudy days - but that will still save me (after round trip losses) 8c/kWh, so 56 to 80c per day.
Rates (all inclusive of GST):
- FIT: 11.3 c/kWh
- Peak: 20.66c/kWh from 7am to 11pm Monday to Friday only (I know, I know, cheaper than shoulder for @GKPW2)
- Off-Peak: 10.34c/kWh (all other times, including all weekend)
- FIT given up to provide 1kWh usable costs 12.995c/kWh (11.3c/kWh * 1.15 for the round trip loss)
- Off-Peak used to charge to provide 1kWh usable costs 11.891c/kWh (10.34c/kWh * 1.15)
So, for every kWh of PV that I use to charge my battery, I lose just over 1c discharging into off-peak, but pick up almost 8c discharging into peak.
By putting my PW2 into standby at the end of peak (by setting backup_reserve_percentage to the current state of charge), I preserve that charge for using in peak the next day, and, I also mean it is more likely that I'll spill that excess generation to the grid at feed in rates but still have a fully charged battery for the evening peak (because it's starting with a greater charge that it would if it were discharging into off peak).
In summer, it's almost certain to be full by 3pm, so any excess will feed in - potentially at 29c/kWh, though realistically at 11.3 until 1 July and 9.9 after 1 July because I can't see any retailer choosing to offer 29c/kWh yet.
Even when FIT drops to 9.9c, a stored kWh will cost me 9.9 * 1.15 = 11.38c/kWh - still more than the cost of off-peak from the grid.
So I'm better off "resting" my PW2 in standby overnight (and all weekend), paying for off-peak, and spilling excess for FIT. In the depths of winter, I'll probably charge around 7kWh per day, perhaps up to 10kWh for really cloudy days - but that will still save me (after round trip losses) 8c/kWh, so 56 to 80c per day.
You're right, I'll give that a go and see if it makes the battery behave the way I feel is most efficient. To some of my other points, this is one of the annoyances of the currently Tesla app's config of T.B.C. We should not have to 'mess with' settings to make the app behave the way it *should*. If the app had the ability to take the necessary inputs, it would make the correct decisions automatically (hopefully
)!Here's hoping Tesla keep iterating QUICKLY on T.B.C. in the app to better handle all of these different scenarios!
Until you get T.B.C - yes.
Once you get T.B.C., you'll get same behaviour, but without having to manually switch the settings on a Friday
You could also give BJReplay's Windows Service that I referenced a go to automate this process right now.
If you haven't got T.B.C yet from Tesla, then if your Tesla app is set to 'Self-powered' then your house will always first discharge your battery before going to the grid for power. If you do have T.B.C, then that is odd, it really shouldn't discharge during off-peak. Can you clarify whether you have T.B.C yet and what setting Balanced or Cost Saving?
You may need to give Tesla a call if nothing obvious pops up on a config level.
Again, this may come down to your Tesla settings (Backup-only vs Self-powered vs T.B.C). Please clarify by providing your settings.
So, the primary difference is: Balanced - it will not charge your battery from the grid. Cost Saving - it can.
The discharging policy (based on AUS document I linked) is identical.
Dborn
Confirmed
Hi Anthony, I do indeed have time based control and set to cost saving.
I did phone support and asked them about the early morning discharge. Their reply was that the system is still learning and to wait a month, and if still happening to call them again. They did look at my settings and fixed an error I had during weekend times, but other than that considered my settings correct.
Thanks for explaining the “balanced”.
I did phone support and asked them about the early morning discharge. Their reply was that the system is still learning and to wait a month, and if still happening to call them again. They did look at my settings and fixed an error I had during weekend times, but other than that considered my settings correct.
Thanks for explaining the “balanced”.
Actually, now I've had more time to think about this, setting my weekend to 'all day off-peak' may not make any difference.. (I'll still try it to validate).
If you review the Tesla Modes of Operation With Solar document, it will (or at least can) still discharge in Off-Peak, it just treats that as the lowest priority in its calculations!
On an encouraging note, today if the first cloudy, overcast, slightly rainy in ages and last night the PW2 charged from the grid during off-peak to 83%. This is 25% higher than it's ever done previously so hopefully that is a sign that the forecasting prediction of low solar generation today was pretty accurate.
It will be interesting to see how much power is left in the battery at the end of my Peak period this evening.
So far, I'm finding that it is quite conservative and I usually have around 20% remaining (conservative margin for error).
It will be interesting to see how much power is left in the battery at the end of my Peak period this evening.
So far, I'm finding that it is quite conservative and I usually have around 20% remaining (conservative margin for error).
It's interesting to contrast the operation modes in Australia to those in the US. The prohibition on charging from the grid and the net metering (which effectively means the feed-in tariff is equal to the currently active rate) change how the time-based control modes are implemented significantly.
I know. That's why I very deliberately called it out in my post. And I still can't work out how to get to that AUS page just navigating from the main support page.
Nope. Didn't make it. 8:30pm and battery empty, well short of 10pm peak period end.
If it is any consolation, solcast was forecasting a shape like this for today (where the lighter green trace was the original forecast, and the darker trace the forecast in the early hours of this morning, as off-peak ended. At 6:41am the forecast was 14kWh of PV for the day, and shortfall generation (from my target of 17.5kWh) was 3000 at midnight, and 3,500 at 6:41am.
They progressively reforecast today, until we ended up with this shape
So the last 24 hours makes no sense to me...
Starting from left..
10:00pm (start of my off-peak) - battery is at 10% and it starts charging from grid.
3:40am - from 10pm through to 3:40am it fills battery back to 100%
4:00am - it lets the house consume battery.. all the way down to 50%
8:00am - Solar starts to kick in meaningfully and battery actually takes some excess to charge (it was a pretty overcast day)
8:40pm - Battery is empty 1hr 20 mins before Peak period ends.
Now, I know this is not an exact science, but qhat doesn't make sense to me is that my overnight off-peak period ends at 7am by which time Tesla has let the battery discharge down to 70%. Given we ran short on battery power to get us through the afternoon/evening peak, and it is *normally* fairly conservative, it seems very odd that it let the house consume power from the battery during off-peak (i.e. before 7am) rather than holding 100% charge until end of off-peak and then letting the house start consuming...
Then again, I guess because the Australian 'rules' from Tesla, when using T.B.C (Balanced OR Cost Saving) is to allow discharge in Off-peak as the lowest priority, then I guess Tesla will say that this is 'expected'?
Utterly stupid in my mind.
This is pretty odd. I would think there should be never be any discharging from the PW during nighttime off-peak hours unless there's a grid outage, or there's excess solar energy stored in the battery and your feed-in tariff is lower than the off-peak rates. If any charging from the grid took place, then any internal "have we stored excess solar energy?" flag should be set to False.
Also, in general, I like the concept of the PW system "learning" the home's energy consumption and generation patterns, as the majority of PW owners will likely prefer to "set it and forget it" even if the outcome isn't totally optimal. However, for my personal purposes, I'd prefer a simple, dumb, rule-based system, with rules that I can control, subject to regulatory constraints.
Dborn
Confirmed
Have a look at Flat Out 4.840kW | Live Output (ignore last weekend 15th and 16th - I was working on things then) and see if the pattern is what you'd expect. If so, you know where to go: BJReplay/PowerwallService
17th was a great example - only got to 70% charge, but the pre-charge on the 16th meant that was enough to see through to off peak with 10%
Pre-charge from 11pm through to 1:35am then onto standby overnight on the 17th / 18th.
Standby mode overnight means that the PW goes into standby (setting backup_reserve_percentage to the current state of charge) unless there is a forecast shortfall the next day - in which case it charges sufficiently make up for that shortfall.
I still don't have TBC, and I will give it a go when it arrives, but I suspect I'll be turning it off and turning mine back on.
17th was a great example - only got to 70% charge, but the pre-charge on the 16th meant that was enough to see through to off peak with 10%
Pre-charge from 11pm through to 1:35am then onto standby overnight on the 17th / 18th.
Standby mode overnight means that the PW goes into standby (setting backup_reserve_percentage to the current state of charge) unless there is a forecast shortfall the next day - in which case it charges sufficiently make up for that shortfall.
I still don't have TBC, and I will give it a go when it arrives, but I suspect I'll be turning it off and turning mine back on.