New Powerwall experiences

[JES2]

I have my reserve set to 35% and never see them discharge in off-peak . They are usually charged up to between 80% and 100% but I haven't figured out why the max charge percentage varies. I think it tries to charge just enough to cover usage down to 35%.

I have my reserve set to 10%, per Tesla L2 instructions. I don't think I have ever seen the PW2s try to discharge power below the 30% threshold, but I have not had that scenario in the last week to observe. What is really odd is that regardless of the amount of PW2 power the system begins with in off-peak, it won't start discharging until a couple of hours into off-peak. It seems to have some timer that tells it to wait a few hours before discharging in off-peak.

 

[simonog]

I have a 4 PW2 configuration which has now been installed and running for almost a year. I am seeing far more satisfactory behaviour than many of you.

With 7 hours of off-peak power overnight I use cost saving mode all the time. My only gripe with that is that on my meter the time window is GMT and therefore when the clocks change twice a year I have to remember to edit the PW settings in the app - I could do with an option which doesn't assume the time window changes.

I very rarely see the battery fail to prepare itself well for the next day. All the behaviour is that it is watching some weather forecast for number of hours of sunshine the next day, so the only times it has been caught out has been when our British weather differs markedly from the detailed forecast (which happens from time to time ).

The other thing I notice is that the battery does seem to be doing some self-care: some days it will charge to 100% and not quite run down to zero; other days it will run down to zero (often reaching that exactly when the cost switches to low price - impressive forecasting). I think this is careful battery management to maximise battery life.

The one feature which for me, as a lucky Model S owner, is missing is on a day with lots of sun once the battery is 100% full, the excess goes into the grid. Once I get into the low price 7-hour window (with obviously no solar generation) the car starts to charge on its timer. So I gift the grid units during the day, and pull them back overnight when I have to pay for drawing power. It would be wonderful if the PWs could talk to the car / its charger in some way to take any power that would otherwise have gone into the grid.

 

[simonog]

All my empirical evidence is that the PW2s use a machine intelligence system to work out how much to charge from off peak grid power, and some form of weather forecast. Which forecast is a mystery, though after a few weeks sorting itself out (training?) it has been very consistent in performance if the forecast behaves as expected. I think all that is overlaid by battery care intelligence which would explain some of the variations.

As a quick update: last night my PW2's discharged all through the off-peak cost period. Today, the forecast is for enough sun that I believe the battery will fully charge and once again, even with charging the Roadster yesterday, I will use no grid power on or off peak.

 

[JohnRatsey]

I very rarely see the battery fail to prepare itself well for the next day. All the behaviour is that it is watching some weather forecast for number of hours of sunshine the next day, so the only times it has been caught out has been when our British weather differs markedly from the detailed forecast (which happens from time to time ).

The other thing I notice is that the battery does seem to be doing some self-care: some days it will charge to 100% and not quite run down to zero; other days it will run down to zero (often reaching that exactly when the cost switches to low price - impressive forecasting). I think this is careful battery management to maximise battery life.

I'm intrigued that your PWs are showing some intelligent operation of off-peak charging. My PW assumes that the solar generation trend of the past few days will continue. For example, last week the output was: Sunday 15.2kWh, Monday 10.2kWh, Tuesday 6.8kWh, Wednesday 5.2kWh, Thursday 4.9kWh, and Friday 3.7kWh. My household demand is less than the PW capacity so there can be carry-over of power from a sunny day. The PW did an hour of off-peak charging early Thursday morning and about 2 1/2 hours early Friday which meant that the battery was still at 47% at the start of Saturday morning's off-peak period. However, instead of deciding that this was an adequate level of charge it decided to fully charge the battery with the result that much of Saturday's 15.7kWh solar generation went into the grid. I don't call that "intelligent". And that sunshine wasn't a surprise to anyone who had looked at a weather forecast.

 

[JES2]

All my empirical evidence is that the PW2s use a machine intelligence system to work out how much to charge from off peak grid power, and some form of weather forecast. Which forecast is a mystery, though after a few weeks sorting itself out (training?) it has been very consistent in performance if the forecast behaves as expected. I think all that is overlaid by battery care intelligence which would explain some of the variations.

As a quick update: last night my PW2's discharged all through the off-peak cost period. Today, the forecast is for enough sun that I believe the battery will fully charge and once again, even with charging the Roadster yesterday, I will use no grid power on or off peak.

I'm fine with the PW2s using some type of forecasting and calculating based on your usability statistics, plus some type of weather forecasting. However, there should be a base (or option) that simply allows you to not discharge at all during off-peak (or whatever you want to call it.)

In our state, it makes no sense to discharge during off-peak. We have so much wind, that there is a great abundance of it at night. We need the electricity during the day, due to the heat. In my case, it makes no sense for either the state or me personally for the system to work the way it us currently working. It should not be difficult to allow AI type intelligence or just a simple option of on/off during the non-peak period.

 

[simonog]

@Jes: that's a clear case for needing manual control or a simple API so someone can create the equivalent of TeslaFi for the PW2. I would say, however, that I only ever see it discharge in the low cost period when it is likely to get enough solar during the day that the remaining battery capacity plus the solar it predicted for the day mean it take no peak power either. In other words it is truly being lowest cost - first ensuring no peak power usage if it can, even if that means charging during off peak; and secondly charging at off peak only as much as it needs to do to get through the next peak period.

@JohnRatsey: what version of software are you running? I wonder whether it's either that or that you have changed the parameters a lot and perhaps it recalibrates whenever that happens? I also have enough battery capacity for a normal house day - almost exactly. a heavy car charge exceeds that of course.

 

[JohnRatsey]

@simonog: My PW has been on 1.32 for a month or more and it's been in Cost saving mode since October. However, while I have fiddled around with the off-peak duration setting it's been unchanged for a month as one of the Tesla support people showed some interest in my complaints. If there is any forecasting which takes the weather into consideration then I would expect it to be done at the server level. Perhaps it is a feature which has been enabled for you but not me (but Tesla support should have picked that up).

I'm deliberating on whether to revert to Balanced mode for the summer months. If not, I'll curtail the potential for over-zealous off-peak charging by reducing the declared off-peak period (to 1 hour duration?). Last summer the battery, without any off-peak charging, hit empty during a cloudy period in late April and was then OK until 6th October. However, the one certainty about British weather is that it is unpredictable more than a few days ahead.

 

[abasile]

I'm fine with the PW2s using some type of forecasting and calculating based on your usability statistics, plus some type of weather forecasting. However, there should be a base (or option) that simply allows you to not discharge at all during off-peak (or whatever you want to call it.)

Personally, I'd like more manual control! Our daily usage is highly variable, and cloud and snow cover can vary greatly around here.

 

[JCUK]

Tesla have informed me that they do not at the moment use weather forecasting, instead they use recent generation. The powerwall will try to save enough energy in offpeak to cover your normal usage until it expects solar generation to take over. The problem occurs when the weather changes.
I have been on self power for two weeks, and adapt my daily usage accordingly, I have not taken any energy from the grid in that period, this is possible because the days are getting longer and we have had decent weather. Prior to that I was very frustrated as it would often do the exact opposite of what was required.

 

[simonog]

What confuses me about that statement from Tesla is that I have several instances when after some poor days for solar generation, our PW will not charge at night at all and the next day fills completely from a sunny day. Indeed, we have started to use the PW overnight charge as a check on the weather forecast we normally look at!
If I can talk to someone in core engineering in the US I'll see if I can get any more explanation. Judging by the confusion we see when the car software goes through a period of change, clarity of message in the company can be very inconsistent....

 

[bmah]

I'm seeing some odd behavior in the Power Flow view on the iOS app. PowerWalls are running 1.34.2, configured for Cost Saving with 30% reserve, currently in the partial-peak period with partial cloudiness.

Typically I'd expect the solar panel output to go straight into the Powerwalls, with the house drawing from the grid, so that the batteries can feed the house during peak time. Most of the time (as I'm watching the display) it does this. However:

1. Sometimes, all but about 0.1 or 0.2 kW will go to the Powerwall, and the rest will go to my Home. This lasts for a few seconds. I wonder if this is some kind of rounding error?

2. I'll sometimes see the solar output going to my home and backfeeding the grid for a few seconds, with nothing going to or from the battery. I'm not sure why it should do this. Again this will last only for a few seconds.

Anyone else see these momentary deviations from expected behavior? Any ideas about why?

Thanks,

Bruce.

 

[miimura]

The behavior you are seeing is common, but I would not say they are expected, desirable, nor logical. The only one that has a good explanation is when the solar output is swinging wildly during partly cloudy conditions, the Powerwall charging power is not updated as fast as the solar output is changing. So, you will frequently see extra power going into the house from the solar or even into the grid from the solar when it could be going into the battery. However, I have noticed that it will quickly drop the charging power so that it prevents the Powerwalls from charging from the grid when the solar generation suddenly falls off.

 

[simonog]

I'm seeing some odd behavior in the Power Flow view on the iOS app. PowerWalls are running 1.34.2, configured for Cost Saving with 30% reserve, currently in the partial-peak period with partial cloudiness.

Typically I'd expect the solar panel output to go straight into the Powerwalls, with the house drawing from the grid, so that the batteries can feed the house during peak time. Most of the time (as I'm watching the display) it does this. However:

1. Sometimes, all but about 0.1 or 0.2 kW will go to the Powerwall, and the rest will go to my Home. This lasts for a few seconds. I wonder if this is some kind of rounding error?

2. I'll sometimes see the solar output going to my home and backfeeding the grid for a few seconds, with nothing going to or from the battery. I'm not sure why it should do this. Again this will last only for a few seconds.

Anyone else see these momentary deviations from expected behavior? Any ideas about why?

Thanks,

Bruce.

Bruce
It is a regulatory requirement for any local generation to stay in synchronisation with the grid supply. I believe that is why you will see a tiny amount briefly go into the grid or come back out.

Simon

 

[bmah]

It is a regulatory requirement for any local generation to stay in synchronisation with the grid supply. I believe that is why you will see a tiny amount briefly go into the grid or come back out.

Thanks, that makes sense.

Bruce.

 

[JES2]

What confuses me about that statement from Tesla is that I have several instances when after some poor days for solar generation, our PW will not charge at night at all and the next day fills completely from a sunny day. Indeed, we have started to use the PW overnight charge as a check on the weather forecast we normally look at!
If I can talk to someone in core engineering in the US I'll see if I can get any more explanation. Judging by the confusion we see when the car software goes through a period of change, clarity of message in the company can be very inconsistent....

Tesla (L1) called me up and confirmed that my Powerwall2s were not working as they should be. They agreed that it was discharging when it shouldn't (off-peak). L2 is now looking at it to see why this is occurring. And the story continues...

 

[NinjaVece]

I'm seeing some odd behavior in the Power Flow view on the iOS app. PowerWalls are running 1.34.2, configured for Cost Saving with 30% reserve, currently in the partial-peak period with partial cloudiness.

Typically I'd expect the solar panel output to go straight into the Powerwalls, with the house drawing from the grid, so that the batteries can feed the house during peak time. Most of the time (as I'm watching the display) it does this. However:

1. Sometimes, all but about 0.1 or 0.2 kW will go to the Powerwall, and the rest will go to my Home. This lasts for a few seconds. I wonder if this is some kind of rounding error?

2. I'll sometimes see the solar output going to my home and backfeeding the grid for a few seconds, with nothing going to or from the battery. I'm not sure why it should do this. Again this will last only for a few seconds.

Anyone else see these momentary deviations from expected behavior? Any ideas about why?

Thanks,

Bruce.

Bruce, 1.34.2 has some issues - see my recent thread on the 1.34.2 release.

 

[NinjaVece]

Tesla (L1) called me up and confirmed that my Powerwall2s were not working as they should be. They agreed that it was discharging when it shouldn't (off-peak). L2 is now looking at it to see why this is occurring. And the story continues...

Jes2, are you on 1.34,2 firmware? Definite issues with that one - see my recent thread on the 1.34.2 release.

 

[JES2]

Jes2, are you on 1.34,2 firmware? Definite issues with that one - see my recent thread on the 1.34.2 release.

1.32.

Level 2 called me after I talked with Level 1. I spoke with Caleb. He quickly assured me that the Powerwalls were working as designed (by destaging at various times and amounts during off-peak.) I asked him various questions on why or how it decides to destage and it was like talking to a parrot. (Works as designed, works as designed,...) with no capability of explaining any detail or reasoning. Frustrating. Horrible customer relations. I hope he is not representative of the product group. Looks like I'll have to start monitoring the reserve on cloudy days.

 

[NinjaVece]

1.32.

Level 2 called me after I talked with Level 1. I spoke with Caleb. He quickly assured me that the Powerwalls were working as designed (by destaging at various times and amounts during off-peak.) I asked him various questions on why or how it decides to destage and it was like talking to a parrot. (Works as designed, works as designed,...) with no capability of explaining any detail or reasoning. Frustrating. Horrible customer relations. I hope he is not representative of the product group. Looks like I'll have to start monitoring the reserve on cloudy days.

I had no issues with1.33 - you may want to get them to update you to that release...

 

[oleboyredw]

So, I had a puzzling new experience in the last week or so...

On 23rd March at roughly 14:00 comms with the Powerwall ceased, I tried the app, the web interface from a browser and the PVOuput code failed. Roughly two hours later it just sprung to life again.

During the outage on 23rd March I wrote to Tesla EU tech support to point out I could not communicate with the Powerwall and how should I proceed (diagnosis, resetting etc).

On 25th March I got a generic response that looks like a cut and paste from a document.

On 30th March at about 19:27 I wrote back stating it access returned after two hours or so and am concerned about unexplained outages (a life in IT).

Overnight on 30th/31st March UK reset clocks to summer time.

On 1st April I got an email from support stating "The reason why it started working again might be related to the change from winter to summer time." As these emails were in a chain of responses the whole interaction was available to all so I'm baffled by the explanation!

Can anybody enlighten me on how the clock change caused the Powerwall to return? Have I bought a Time machine??