Time-Based Control?

[cwied]

I'm actually on PG&E E-6, not EV-A, as I don't charge my EV enough to make it worthwhile to get the slight rate advantage in exchange for shifting the Peak period later. I forgot that EV-A has weekends as peak.

I agree that at least a one-day lookahead would be great.

I don't really see a lot of extra discharging during part-peak in practice. It helps that I have a little off-peak energy on Monday mornings that fills my Powerwalls back up after the Sunday part-peak discharge. That makes the Sunday discharge more sensible than it otherwise would be.

As to the SOC window: I agree that it doesn't matter which part you use. I would also prefer to err towards higher SOC in the winter. In the summer, having lower SOC during the week means there's a small arbitrage of the extra off-peak energy on the weekend.

 

[arnolddeleon]

I suspect that my system doesn't charge all the way to 100% on some days (I was in the process of setting up logging for this but I can't get to my powerwalls via the LAN, but that's a topic on another thread). On those days it still runs out of energy in the peak period. I appreciate the difficulty of the optimization problem that I'm presenting.

I have off peak solar available. The system is behaving perfectly as best as I can currently tell at this point during off peak solar time. It uses all the available PV to charge the batteries.

I also believe the system behaves perfectly well during the peak window. It basically discharges to match the load that I have. During solar production hours this means I export the PV production.

During the weekends I have no peak, I just have part peak. The system appears to also be perfect during this time, it treats it as peak.

Part peak during the weekday when there is no PV production appears to be handled correctly.

Now the really challenging window during weekdays, part peak during PV production. What should the system do? There are two starting conditions, batteries are fully charged or batteries are partially charged.
If it had perfect precognition and know how much energy is going be discharged during peak then then system can figure out if has enough energy for peak. If it doesn't then it should charge during the part peak. On the other hand if it already has enough energy for the peak period, it needs to predict how much energy is going to be consumed during the part peak shoulders. This can go wrong in two ways. If holds on to the off peak power and the second part peak ends. It discharges too early and discharges in the part peak what could have been peak.

About a week I was ready to declare success and give Tesla high marks for doing pretty well. Right now I'm a little steamed because I can't get the data that I need to support my claims about how well they're managing things.

 

[wwhitney]

On the other hand if it already has enough energy for the peak period, it needs to predict how much energy is going to be consumed during the part peak shoulders.

It should only discharge during part peak if your off-peak production exceeds your peak consumption. Otherwise it should just hold over the surplus energy for the following peak consumption period. Every part peak discharge -- part peak charge cycle is a waste of 10% of the energy, unless the initial part peak discharge was required to provide available space for an intervening off peak production period.

Cheers, Wayne

 

[abasile]

I agree that there's a roughly 10% efficiency penalty for self consumption and that it's hard to financially justify "zeroing out" grid usage during shoulder hours if on NEM 1.0. On the other hand, I figure that by self consuming, we are reducing the burden on the grid relative to the amount of renewables deployed, which is the whole point of the SGIP in California.

One thing I'm motivated to do is avoid contributing to the "duck curve", most notably the rapid ramp-up of fossil generation that has to occur practically every evening. For this reason, I've configured our Powerwalls to treat 2pm - 10pm as "peak" hours on weekdays, even though peak is in actuality from 2pm - 8pm on our rate plan. That way, we aren't burdening the grid with our home loads until after things have stabilized for the night and the very lowest rates are in place. We have the luxury of doing this because our Powerwalls store more than enough energy to get us through each day's peak period.

Something else I'd like to be able to do, but can't with the current Powerwall software, is delay our shoulder-rate charging until as close as possible to midday, i.e., around 11am or so. Midday tends to be when the grid is most challenged to absorb solar production, so it would be more helpful to the grid to be charging our Powerwalls at midday rather than right at 8am. Earlier in the morning, the grid would more likely benefit from our solar exports.

In an ideal world, the needs of the grid would be reflected in the TOU rate schedules. On the other hand, this would unfortunately tend to reduce the value of installing PV, so it's complicated.

Also, regarding what part of the SOC window gets used, every lithium ion battery chemistry that I'm aware of has better longevity when higher SOCs are used only occasionally and/or only for brief periods of time. I'll admit that we don't know enough about the design of the Powerwall 2 and its battery chemistry to predict the battery capacity loss over ten years with daily charging to "100%" versus "90%". My guess is that there's an SOC buffer at the top and that Tesla designed the system to not degrade excessively when left at "100%" for very long periods of time. However, it would certainly not hurt to avoid going all the way to 100% whenever a full charge is not needed, and this would undoubtedly increase the long term battery capacity retention by a non-zero amount. We just don't know by how much. Personally, I'd just leave it set at 80-90% all spring and summer.

 

[arnolddeleon]

It should only discharge during part peak if your off-peak production exceeds your peak consumption. Otherwise it should just hold over the surplus energy for the following peak consumption period. Every part peak discharge -- part peak charge cycle is a waste of 10% of the energy, unless the initial part peak discharge was required to provide available space for an intervening off peak production period.

Cheers, Wayne

Correct, that's a succinct way of putting it. I happen to be in situation where on some days this true and other it isn't. Specifically when air conditioning loads drop I expect to off peak production to exceed my peak usage. In that case there is benefit to discharging during part peak.

 

[Tevvy]

Sorry but I can’t read the last 18 pages.

Can someone give me a yes/no answer.

Can the powerwall 2 now be charged up overnight on cheap rate and then dispurse during the day to cut out the grid use for day (assuming day usage is <14KWh? And then do it all over again the next night?

I have no solar panels (and can’t get any)

 

[bishoppeak]

Sorry but I can’t read the last 18 pages.

Can someone give me a yes/no answer.

Can the powerwall 2 now be charged up overnight on cheap rate and then dispurse during the day to cut out the grid use for day (assuming day usage is <14KWh? And then do it all over again the next night?

I have no solar panels (and can’t get any)

yes

 

[cwied]

That should be qualified somewhat - yes with time-based control. Time-based control is still being rolled out in limited regions. I don't know if the UK is included yet. On the other hand, given the delivery delays, it may be more widely released by the time you'd to get the Powerwalls.

 

[sashton]

Yes, time based controls are available in the UK. I got it with 1.17.2 a couple of months ago still there in 1.20.x. Delivery delays in the UK are not so bad. My additional PWII ordered in April is due in September.
There is a slight price to pay when enabling TBC. It appears that when it is enabled the inverter or the processes which control the inverter do not appear to sample the power flows as regularly. This is demonstrated by higher marginal imports and exports while the battery is charging and discharging. I had times when this "error " was up to 300W. The problem appeared to be less after 1.20 but was still noticeable.
Since the existing PW here is usually full by midday I'm currently leaving it on "self-powered" and will revisit the setting when the generation reduces in the autumn.

 

[arnolddeleon]

I agree that there's a roughly 10% efficiency penalty for self consumption and that it's hard to financially justify "zeroing out" grid usage during shoulder hours if on NEM 1.0. On the other hand, I figure that by self consuming, we are reducing the burden on the grid relative to the amount of renewables deployed, which is the whole point of the SGIP in California.

One thing I'm motivated to do is avoid contributing to the "duck curve", most notably the rapid ramp-up of fossil generation that has to occur practically every evening. For this reason, I've configured our Powerwalls to treat 2pm - 10pm as "peak" hours on weekdays, even though peak is in actuality from 2pm - 8pm on our rate plan. That way, we aren't burdening the grid with our home loads until after things have stabilized for the night and the very lowest rates are in place. We have the luxury of doing this because our Powerwalls store more than enough energy to get us through each day's peak period.

Something else I'd like to be able to do, but can't with the current Powerwall software, is delay our shoulder-rate charging until as close as possible to midday, i.e., around 11am or so. Midday tends to be when the grid is most challenged to absorb solar production, so it would be more helpful to the grid to be charging our Powerwalls at midday rather than right at 8am. Earlier in the morning, the grid would more likely benefit from our solar exports.

In an ideal world, the needs of the grid would be reflected in the TOU rate schedules. On the other hand, this would unfortunately tend to reduce the value of installing PV, so it's complicated.

Also, regarding what part of the SOC window gets used, every lithium ion battery chemistry that I'm aware of has better longevity when higher SOCs are used only occasionally and/or only for brief periods of time. I'll admit that we don't know enough about the design of the Powerwall 2 and its battery chemistry to predict the battery capacity loss over ten years with daily charging to "100%" versus "90%". My guess is that there's an SOC buffer at the top and that Tesla designed the system to not degrade excessively when left at "100%" for very long periods of time. However, it would certainly not hurt to avoid going all the way to 100% whenever a full charge is not needed, and this would undoubtedly increase the long term battery capacity retention by a non-zero amount. We just don't know by how much. Personally, I'd just leave it set at 80-90% all spring and summer.

I commend your efforts to do the right thing. I think we get to what you are talking about, charging when energy is plentiful and discharging to reduce fossil fuel use, when we relinquish more direct control our systems and are operating as a virtual power plant (coordinated by Tesla or 3rd party if they allow it). The value of storage with solar will go up as the TOU rates come to reflect the need of the grid better.

 

[arnolddeleon]

I don't think cost-saving does this any more unless there is surplus power from a prior off-peak period. I do have off-peak power during the week, but even so I do not see a lot of discharging except on Mondays when the Powerwalls are full from the weekend. The big difference is that the Powerwalls don't charge to 100% during the week on cost-saving, unlike balanced mode. They seem to aim to charge just enough to get through the peak period.

The bug(s) around this appears to be have been fixed in cost savings mode. Over the last few (over four I think) weeks I haven't seen any clearly stupid errors like discharging during off peak, charging during part peak and and discharging a similar amount in part peak again (silly roundtrip). The only things I've seen are prediction errors that don't have trivial solutions. For the most part I think they're all errors in failing to maximize cost savings but they're still cost saving. I think they are basically of the following form; discharging off peak energy into part peak instead of peak and failing to charge in part peak when the energy could have been discharged into peak.

 

[BJReplay]

The bug(s) around this appears to be have been fixed in cost savings mode. Over the last few (over four I think) weeks I haven't seen any clearly stupid errors like discharging during off peak, charging during part peak and and discharging a similar amount in part peak again (silly roundtrip).

You're lucky!

I see discharge into off peak - my cycle (in winter) is charge during off peak to 100%, but it will often discharge 10% one to two hours before peak starts into off peak. Needless round trip, and insufficient charge to see me through the peak period.

 

[DelM]

Well, mine is still discharging during off-peak and then running out of charge during peak. It did it yesterday and is displaying the same behaviour today. Frustrating, as I end up paying for peak tariff units when I shouldn't need to. The other odd behaviour I have noticed recently is that I can no longer select "Balanced" mode: I can only switch between "Self Powered" and "Cost Saving".

 

[cwied]

@DelM - that graph looks very strange. What is your state of charge at the end of off-peak, and what is your backup reserve set to?

It appears like the Powerwall is charging from the grid during off-peak and then possibly changing its mind to try to make room for solar charging during your part-peak period. It also appears to be underestimating your peak usage. How long has it been running on cost-savings mode?

I assume you've reported this to Powerwall support. What did they say?

 

[DelM]

@cwied Yes, I have reported it to support. They kept advising that I have only just started using the cost-saving mode and that the Powerwall will learn with time. I accept that that could be part of the issue. However, what I can't accept is that the powerwall uses stored energy I have paid for during the off-peak period! Better to store it and run the risk of sending some free solar energy back to the grid later in the day. In my mind, that is one of the simplest rules the gateway should have. I am going to follow up my conversation by sending them the graphs. We'll see what happens. I am in the UK, so don't have the back-up option, so no minimum reserves set on the powerwall. The weather just turned

 

[DelM]

...for the worse here, so I decided to give cost-saving mode a go, instead of self-powered. (Sorry, clicked "Post" by accident!)

 

[Ulmo]

How do I access Time Based Control software in my Tesla PowerWall 2 system?

 

[MorrisonHiker]

How do I access Time Based Control software in my Tesla PowerWall 2 system?

What do you see if you choose the Customize option in the app?

 

[Dborn]

I see we now have storm detection in the app and it seems to work.!!
I too have some in appropriate discharge during off peak periods with some charging occurring off peak as I believe the app now gets forecast weather information.

 

[liplus]

I'm having Powerwall 2.0 + backup gateway + solar installed in January. My distributor limits exports to 5kW but I'm having 13kW PV installed (10kW of inverter capacity)

This means that in addition to "inverter clipping", I will have "export clipping" where (self consumption + 5kw export) is less than PV generated:

For this reason, I want to delay the charging of Powerwall 2.0 until ~10am. This means all early to mid-morning generation is self-consumed or exported. Around the time "export clipped" power should start to be available, I want to start charging the Powerwall.

I have a single tariff and never want to charge battery with grid power. I anticipate that I will be discharging the battery down to reserve level every night due to self-consumption.

I think the configuration I need is something like:

TBC (Balanced)
9pm - 10am Peak
10am - 9pm Offpeak

If I'm reading the thread and Modes of Operation correctly, this should:

- Never charge the powerwall during peak time (even when "excess solar" is available?)
- Begin charging with excess solar at 10am (possibly "any solar" if the Powerwall thinks it will need to?)
- All evening/night consumption will be offset by the Powerwall until it hits reserve power.

Does this sound correct?