Time-Based Control?

[miimura]

How easy is it to set this up? Are these 3rd party services like TeslaFi for the car?

You just need some kind of computer running 24/7 to pull the data out from the Powerwall. Some people use a Windows PC, some people use Raspberry Pi. I think most send it to PVoutput.org

Tesla Powerwall2 data to PVoutput?
PowerWall and switchgear API's

 

[arnolddeleon]

Your Yesterday chart matches my experience with Cost Saving.
- During Off-Peak All Solar goes to batteries
- During morning Part-Peak Surplus Solar goes to grid, evening Part-Peak net consumption comes from battery.
- During Peak All Solar goes to grid, batteries power home.

If you want Surplus Solar to go to the battery during Part-Peak, then use Balanced.

I don't mind the general ideas behind Cost Saving if they worked right. I unfortunately didn't take a capture of Monday which had the Powerwall discharging during the part peak and peak. The main issue I see with cost saving is correctly predicting how much peak usage there is going to be. It would be suboptimal if the system used battery power during the first part peak when it could have used it during peak. It looked the system was heading in the right direction when it no discharge during the first part peak, discharged during peak and then finished off the remaining sellable battery into the next part peak. But then we got today's no discharge, it's actually a little lame, it was idling (I forget the exact state name) but then later (don't know when it started) it went back to its stupid discharge mode to offset the load of my Outback inverters. So today is a negative day, the Powerwall simply wasted energy

 

[miimura]

I don't mind the general ideas behind Cost Saving if they worked right. I unfortunately didn't take a capture of Monday which had the Powerwall discharging during the part peak and peak. The main issue I see with cost saving is correctly predicting how much peak usage there is going to be. It would be suboptimal if the system used battery power during the first part peak when it could have used it during peak. It looked the system was heading in the right direction when it no discharge during the first part peak, discharged during peak and then finished off the remaining sellable battery into the next part peak. But then we got today's no discharge, it's actually a little lame, it was idling (I forget the exact state name) but then later (don't know when it started) it went back to its stupid discharge mode to offset the load of my Outback inverters. So today is a negative day, the Powerwall simply wasted energy

If you use Balanced, you won't have that problem. During Part-Peak it will finish charging the batteries with Surplus Solar and then when they're full, they will go into Standby and then Surplus Solar will be exported. The main difference is that Part-Peak export doesn't start until the batteries are full. This guarantees that you will have battery available during the Peak period, even on heavy A/C days. I think Cost Saving assumes that you always have enough generation during Off-Peak, but it appears that even you don't on E-6. PG&E EV rate on weekends works great on Cost Saving because there's no Part-Peak at all.

 

[Gear]

I think I just encountered the bug with the cost-savings mode. My powerwalls are sitting at 99% and didn't discharge during the peak. I believe @cwied has E-6 like I do.

Mine did some pretty weird stuff today too.



It stopped using energy from the Powerwalls during my peak period even though there was ~50% energy remaining and my reserve is all the way down at 25%. Then it even charged them up a little during peak. After the sun went down, it started using grid power during peak even though there was 55% charge remaining in the Powerwalls. That makes no sense to me.

Your Yesterday chart matches my experience with Cost Saving.
- During Off-Peak All Solar goes to batteries
- During morning Part-Peak Surplus Solar goes to grid, evening Part-Peak net consumption comes from battery.
- During Peak All Solar goes to grid, batteries power home.

If you want Surplus Solar to go to the battery during Part-Peak, then use Balanced.

View attachment 308039

Maybe I need to be using Balanced then. That seems far more straightforward and how I thought Cost Saving would work. The description for Balanced ends with "and after the sun goes down" so I figured that meant it would continue to use energy from the Powerwall all night. My off-peak power is only $0.13/kWh, so I'd rather use that to charge my car at night rather than the Powerwall. My shoulder period starts at 8am and ends at 10pm (peak from 2pm-8pm) so I get almost no off-peak charging, but a decent amount of shoulder charging.

 

[cwied]

@miimura - your description of Cost Saving doesn't quite match my experience. My off-peak behavior does match what you describe, but during part peak, the batteries will charge to about 80% to be able to handle the peak. My experience on one Monday makes me suspect that they would also discharge 80% during part-peak if the SoC starts at a higher level. I believe the 80% is derived from my backup reserve of 50%.

The main advantage of this algorithm vs. Balanced is that it will charge up over the weekend when there is more off-peak energy and use that to offset part-peak usage. The main problem is that it seems to be buggy and do something completely weird some days. I wouldn't be surprised if the programmers confused themselves with the definition of time periods that wrap around midnight.

Here is my chart from yesterday showing charging during the part-peak period. In case the picture doesn't come through, you can see the pvoutput chart for that day here: cwsm 6.300kW | Live Output Click on the last little square under the date to see the Powerwall statistics.

 

[arnolddeleon]

Hmm, I'm at 90%, plenty of solar, off peak and the PW is not charging. Is that how Cost Saving is suppose to work on the weekend (no peak, just part peak). Looks like another bug.

 

[arnolddeleon]

Hmm, I'm at 90%, plenty of solar, off peak and the PW is not charging. Is that how Cost Saving is suppose to work on the weekend (no peak, just part peak). Looks like another bug.

I thought of a valid reason not to charge, if it is using predictions then it could have decided that it has enough battery for the part peak load and so charging the battery today makes no sense (it's well above the reserve for power fail)

 

[cwied]

I don't think anybody knows for sure how cost saving is supposed to work (including the powerwall support rep I talked to), but I currently believe that it should charge up to 100% on off-peak power. I think it divides the usable battery charge into two portions. It will charge up to one threshold on part-peak and to the other on off-peak. The reason why it makes sense to charge up to 100% is that on Monday it will then have extra energy to discharge during the morning part-peak period. It could be smart enough to do that tomorrow instead of today, but I don't see why they would try to be so smart when a simple rule would have the same effect (plus it might be cloudy tomorrow).

I can confirm later today whether that's what actually happens with my system. Right now I'm only at 55% since cost saving worked yesterday.

One complicating factor is that I switch away from cost saving when OhmConnect gives me an OhmHour. There was an OhmHour last night, so I was on self-consumption from 8-9pm. I don't know if there's any state kept by the Powerwall software that gets wiped out on mode changes.

 

[cwied]

I just wanted to confirm that my Powerwalls charged to 100% using all solar today. I expect them to discharge at 5pm when my rate goes to part-peak.

 

[miimura]

The bottom line for me is that Balanced is consistent and predictable while Cost Saving frequently does things that I don't like.

 

[arnolddeleon]

The bottom line for me is that Balanced is consistent and predictable while Cost Saving frequently does things that I don't like.

I'll end up there if Tesla doesn't fix the bugs. For now I'm willing to try to help Tesla out (if they actually want to fix things). I haven't actually talked anyone that gives me the impression that they understand yet.

BTW my powerwalls are still at 90% today.

I've ordered a Raspberry Pi and I will setup PVoutput to get more data out of the Powerwalls. Right now I can't look at back to see what the SOC of the batteries were, I'm hoping I can do that with PVoutput or some other work that someone has already done so I don't have to figure it out.

 

[arnolddeleon]

I'll end up there if Tesla doesn't fix the bugs. For now I'm willing to try to help Tesla out (if they actually want to fix things). I haven't actually talked anyone that gives me the impression that they understand yet.

BTW my powerwalls are still at 90% today.

I've ordered a Raspberry Pi and I will setup PVoutput to get more data out of the Powerwalls. Right now I can't look at back to see what the SOC of the batteries were, I'm hoping I can do that with PVoutput or some other work that someone has already done so I don't have to figure it out.

PW has started charging with part peak

 

[cwied]

@arnolddeleon - that's exactly what my system did last week. I've added half-hour of peak at the end of the part-peak period and it seems to be doing the right thing. I also reported the issue to Powerwall support, though, so they may have tweaked something too. I don't think the artificial peak period will do any harm, so I'm going to try with this programming for a while. With the buggy set up, my Powerwalls started discharging at midnight. You may want to watch for that.

@miimura - I'm not questioning your decision at all. I just think that Powerwall support gave an answer that's not completely correct and that some of the undesirable behavior is due to bugs rather than design. The EV rates have enough off-peak hours on the weekend that there is some benefit from maintaining a lower SoC during the week and filling that up on weekends. It's not a big gain, though, so it may not be worth having less reserve for an outage during the week.

 

[arnolddeleon]

@arnolddeleon - that's exactly what my system did last week. I've added half-hour of peak at the end of the part-peak period and it seems to be doing the right thing. I also reported the issue to Powerwall support, though, so they may have tweaked something too. I don't think the artificial peak period will do any harm, so I'm going to try with this programming for a while. With the buggy set up, my Powerwalls started discharging at midnight. You may want to watch for that.

Yep, it just started discharging. That's pretty silly, charged during the part peak and it is now discharging during the off peak.

 

[JohnRatsey]

I've ordered a Raspberry Pi and I will setup PVoutput to get more data out of the Powerwalls. Right now I can't look at back to see what the SOC of the batteries were, I'm hoping I can do that with PVoutput or some other work that someone has already done so I don't have to figure it out.

Just type in the gateway's local network IP address on a phone or computer and you will get a display showing the power flow and battery charge level. Note that the charge level shown here isn't exactly the same as on the Tesla app which shows zero at 5% battery charge and linearly changes as higher charge levels (eg 80% actual charge is 79% in the app). If you haven't done so already, it's useful to assign the gateway a fixed IP address on the router. However, you'll need a logging script if you want the monitor the change in the SoC. Two data sources on the gateway are <gateway IP address>/api/meters/aggregates and <gateway IP address>/api/system_status/soe.

In case you haven't come across this excellent explanation of the Powerwall's modes, it's Powerwall2 update – Time-of-Use enabled .

 

[Az_Rael]

So how are Balanced and Cost Saving supposed to work? I am sitting in Balanced right now after reading the last few posts about bugs with Cost Saving, and because I don’t have a lot of generation in Off-Peak.

Today/ last night my Powerwall charged up at the end of my solar curve in the shoulder time. But then it held it until midnight (off-peak) instead of discharging during the evening. Today it started charging briefly during off-peak solar generation, then stopped and let the solar pump the cheap energy back to the grid (?). Now it’s charging the rest of the way on my shoulder rate solar.

 

[arnolddeleon]

I'm on a call with support for over an hour now.

Clues from support they have costs associated with the bands. According to them at midnight last night my cost went from $0.10 to $0.60 which would explain the system discharging. In the weekend there is no part peak (she wasn't entirely sure about this). I wonder if this is way @cwied workaround "fixes" things. The UI and the logic may not be matching.

 

[cwied]

I suspect they have a bug with the algorithm when one of the rate levels is missing. I'd be interested to see if it works if you add a fake peak into the schedule. That's what I did, and it's working for me this weekend.

It's interesting that Balanced has the same issue. I had the impression it was working better.

 

[arnolddeleon]

I suspect they have a bug with the algorithm when one of the rate levels is missing. I'd be interested to see if it works if you add a fake peak into the schedule. That's what I did, and it's working for me this weekend.

It's interesting that Balanced has the same issue. I had the impression it was working better.

I agree, the PW didn't start charging this morning until 10:00 am? Why 10:00 am unless you are confused on what the bands truly are. I suspect the code trying to figure the beginning and end of the various bands is all knotted up. Of the the damn cute UI doesn't help things.

How did you define your schedules initially? For the weekend I have it as "peak none" Off peak: 8:00 pm to 5:00 pm. For week I ave peak 1 - 7:00 pm, off peak 9:00 pm to 10:00 am.

I'm going to let the system blunder through for now so the support team has something to look at tomorrow.

I also think I finally explained the issue that I'm been having with the powerwall discharging when it's not supposed to. The magic phrase was it was discharging past the reserve limit. They kept trying to tell me it was all about trying to offset its own usage or the solar usage. This never made any sense to me when the powerwall was already at or below the reserve limit. Why just not let the grid handle it?

 

[cwied]

I had weekends at Peak - none, off-peak 7pm to 5pm (I don't discharge after 7pm because I use OhmConnect during that time). During the week, I have peak 1pm-7pm, off-peak 7pm to 10am. The only place I can figure the 10am time came from is from tomorrow, since there's a transition at that time. This seems to be similar to Friday's behavior where the 5pm transition from Saturday seemed to have an impact.

I've never observed my Powerwalls discharge past the reserve limit, though. They always stop when they hit the reserve. My reserve is currently set to 50%.