Fun with Powerwall 2 stats...

[sberinger]

Just got my bill
22 Aug - 20 Nov ( 91 days )

Solar generation : 8009.03 KW
- Self consumption : 2245 KW ( saving $506 )
- Grid Export : 5674 KW ( -$709 )

Consumption from grid
- Peak Consumption : 1116 KW : $251
- Dedicated Circuit Consumption (Controlled Load 1) : 963 KW $85


Supply Charge : $85

Bill : $283.47 refund

Total Savings from solar : $1215
- self consumption : $506
- FIT : $709

 

[EcoCloudIT]

Just got my bill
22 Aug - 20 Nov ( 91 days )

Solar generation : 8009.03 KW
- Self consumption : 2245 KW ( saving $506 )
- Grid Export : 5674 KW ( -$709 )

Consumption from grid
- Peak Consumption : 1116 KW : $251
- Dedicated Circuit Consumption (Controlled Load 1) : 963 KW $85


Supply Charge : $85

Bill : $283.47 refund

Total Savings from solar : $1215
- self consumption : $506
- FIT : $709

Sorry maybe you've said before however what sized solar and whereabouts in Australia are you to be generting 88KW's per day of enerage during mostly winter months?!

 

[sberinger]

Sorry maybe you've said before however what sized solar and whereabouts in Australia are you to be generting 88KW's per day of enerage during mostly winter months?!

I have 21.6 KW system
roof faces East / West ( 50 / 50 )
I am just north of sydney
I average around 72KW / day over the whole year
In July 2019 I generated 1466KW (average 47KW / day )

 

[Blue heaven]

In Western Australia the 7 cents feed in tariff is capped to a 5kw inverter, the high install rate is a combination of very low purchase price ( sometimes less than $3000 for a 5kw system ) and guaranteed sunny skies. I've heard a rumor that a second suburban Tesla power pack has been installed south of Perth, I think a few dozen extra power packs per year will be required spread across the suburbs to get a grip on proceedings-
In one of Australia's sunniest states, the huge rise of solar is jeopardising the entire power grid

 

[Vostok]

OK, it’s been a while since I posted to this thread after promising to do a bunch of “what ifs”.

The reason is I discovered a flaw in my analysis spreadsheet when I changed my PW2 from operating in self-consumption mode to time-based control (cost saving mode) as an experiment.

In TBC mode, the PW2 does two things which it rarely if ever does in self-consumption mode:

• Solar will export to the grid, even if the battery is not full. It is not obvious when PW2 will choose to do this.
• The grid will charge the battery (it only ever does this at offpeak times)

I never wrote the algorithms in my s/sheet with these possibilities coded in, so once I went to TBC mode it was a case of good data into my spreadsheet but garbage out. So it was back to first principles to re-write it without making implicit assumptions about how the power flows operated.

Turned out that was a much harder problem than I anticipated. It seriously did my head in. The key is to break apart the four bits of data we have (S = solar generation, H = power used by house, B = battery charge or discharge and G = grid draw or export) into the underlying power flows between each of these. There are 7 possible flows:

• S to H, B or G (S can go to just one of these, or any two, or all three simultaneously)
• G to H or B (to either or both)
• B to H or G (ditto. I found that on rare occasions, PW2 would discharge to the grid!)

The task is to calculate these 7 flows based solely on the 4 bits of input data. What was hard was working out the conditional logic which governs each flow. Took weeks to finally crack it (#stayathome turned out to be helpful here ). Once the underlying flows are obtained, it is possible to manipulate and redirect them any which way you like to work out what would happen if the PW2 algorithm did something different.

So my first task is to work out whether TBC Cost Saving mode actually did what it says on the tin - did it save me money compared to Self-Consumption mode? I am not convinced it did. But with my new s/sheet, I can redirect the flows in TBC mode to what they would have been in Self-Consumption mode. And this will be exact. No guesswork.

Hopefully I’ll have the answer in a few days and I’ll report back

 

[Hairyman]

@Vostok Are you on time-of-use metering? I'm using the maximum cost saving setting on my PW2 since I moved to TOU and get a much cheaper rate between 0000-0400 and much more expensive between 1400-2000. In fact my 0000-0400 rate is lower than my FiT. The numbers for me are looking pretty nice.

 

[jamesgo]

I've had our PW2 from 2017 and it's been running fine for most of the time, we just got the 1.45.2 update (haven't noticed anything changed yet).

@Hairyman which state/retailer are you with?

I'm in Vic with AGL, signed up for their BYO Battery Bonus ($280 in the first year) & Peak Energy Rewards (that promo just ended but was around $10 per event).

 

[Vostok]

@Vostok Are you on time-of-use metering? I'm using the maximum cost saving setting on my PW2 since I moved to TOU and get a much cheaper rate between 0000-0400 and much more expensive between 1400-2000. In fact my 0000-0400 rate is lower than my FiT. The numbers for me are looking pretty nice.

Yes, I am on ToU. My FIT is higher than my offpeak tariff by 6 ¢/kWh - and that’s even after paying for 100% Green Power . Like you, I have eye-watering peak rates (54.9 ¢/kWh).

So superficially it seems I would save money by charging the battery using offpeak, do more export, and try to avoid the Peak like the plague. But the TBC settings only indicate when the different ToU rates take effect - it has no idea what the price relativities are between offpeak, peak etc or what my FIT is. It doesn’t know my FIT is higher than my offpeak. Without that information, I cannot see how the algorithm can cost optimise properly. It might get lucky. Or not.

What TBC has done well is we almost never draw grid power during the Peak tariff period. In fact from 1 Feb to 31 March, we drew a grand total of 1.3 kWh from the grid at peak rates!

Also during this period, the battery has been charged by the grid to the tune of 115 kWh (48 kWh of this has been during offpeak, 66 kWh during shoulder, and somewhat bizarrely, 1 kWh during peak). It seems the highest rate at which the grid generally charges the battery is 3.2 kW, but there are a few instances higher than that.

 

[Hairyman]

I've had our PW2 from 2017 and it's been running fine for most of the time, we just got the 1.45.2 update (haven't noticed anything changed yet).

@Hairyman which state/retailer are you with?

I'm in Vic with AGL, signed up for their BYO Battery Bonus ($280 in the first year) & Peak Energy Rewards (that promo just ended but was around $10 per event).

Powershop in NSW on the electric vehicle contract

 

[Vostok]

Drum roll... I have processed the data and have the answer...

In my case, Time-Based Contol (TBC) "Cost-Saving" mode only just beat Self-Powered Mode (SPM) - but the gap was not significant ($3). Here's the results comparing the 2 months I had with TBC with what those same 2 months would have cost under alternative algorithms:

Power flows:

Costs:

Looking at these numbers, my summary of TBC's characteristics:

• It favours grid export over keeping the battery full
  
• It does everything it can to minimise peak grid consumption
  
• It uses the grid to charge the battery (which SPM almost never does) and it does that at both offpeak and shoulder times (shoulder period grid charging was 58% of the total).
  
• It results in lower self-consumption - about 60% in TBC mode compared to over 80% in SPM mode. If high self-consumption percentages are a badge of honour, then TBC is probably not for you.

"Vostok" is my idea for cost-saving - SPM at all times, supplemented with the grid charging the battery only at offpeak times, so that it is 100% by 7am when the shoulder period starts. According to my s/sheet, it beats both TBC and SPM by about 10%:

• It greatly reduces the amount of Shoulder grid power consumed.
• It results in more grid export, because the battery is full every morning.
• Generates income by buying offpeak electricity then selling it later at a higher price. This only works if your FIT is higher than the offpeak price (including any GreenPower charges).
• It significantly increases the total amount of grid power consumed, but also significantly increases the amount of solar exported. You may have a philosophical objection to using the grid in this way.
  
• Similar to TBC, self-consumption percentage is reduced compared to SPM.

These results are based solely on my solar generation and consumption patterns, so they may not apply for different patterns. But I think the insight is still interesting:

• TBC mode only just won with the two rateplans with the highest FITs of 21c / kWh. Every other rateplan I tried with lower FiTs were worse under TBC mode, and SPM won - sometimes by a lot. Without a high FIT, TBC may leave you worse off.
  
• TBC doesn't know about GreenPower charges - a fixed cost per kWh regardless of whether that kWh is peak, shoulder or offpeak. With a fixed GreenPower charge, TBC may also leave you worse off.
• If I turn GreenPower off, TBC wins under all the rateplans I tried, but why would a solar/PW2 owner not also buy 100% GreenPower ? It would be a violation of the mission!
  
• TBC doesn't know about the relative prices between shoulder, peak and offpeak, nor what FIT you get - it must have inbuilt assumptions which cannot be changed in the App UI. So TBC is trying to optimise with incomplete information.

I have now turned TBC off and have reverted to SPM.

 

[Vostok]

I've crunched the full-year numbers for my solar/PW2 installation with some interesting results.

I have relatively small system compared to some here (5.84 kW), but that was the maximum I could fit on my roof using the highest efficiency panels available at the time (365W). I started collecting the Tesla 5-minute data on 1 July 2019, and so now have a full year or 105,406 lines of data in my spreadsheet.

Here's the high-level stats:

• Total solar generation: 8.06 MWh
  
• Total home consumption: 8.23 MWh (so I fell 2% short of solar matching my consumption )
• Total solar export: 2.23 MWh
  
• Grid consumption: 2.82 MWh (only 6.5% of this was at peak rates, 26.1% shoulder, and 67.4% offpeak)
• Net grid consumption: 0.59 MWh (grid consumption minus solar export)
• Solar self-consumption: 67.7% (so fell quite a bit short of my hoped-for 75 to 80%, my 2 months experiment of TBC wrecked it)
• Most solar generated in a day: 36.0 kWh
• Least solar generated in a day: 1.1 kWh

My off-grid-ness:

• Longest continuous time off-grid: 2.82 days
• Longest continuous time on-grid: 1.00 days
• Zero grid draw 75.0% of the time

Some battery stats:

• Battery at minimum level: 25.5% of the time
• Battery at maximum level: 9.5% of the time
• Battery round-trip efficiency: 93.5%
• Number of battery cycles: 236

And a few charts:

This one shows the cumulative net position over the year of solar generation minus house consumption - positive means more solar has been generated over the time period than consumed. I was in the black until 21 June, when it went negative due to a few chilly nights and using the ducted air-con. What a power hog. As Maxwell Smart would say, "missed it by that much"



This one shows solar generation over the year compared to what was predicted by the model at PVWatts Calculator. It's pretty good. The system generated 93.6% of what was predicted. I had 4 good months of production well above the model prediction but Dec and Jan production was smashed by extensive bushfire smoke around Sydney, only to be followed up in Feb by weeks of rain. If it wasn't for that, production probably would have exceeded the model and my net position for the year would have been positive.



Finally, the distribution of solar production as a percentage of the maximum possible on a given day (i.e. if it was sunny all the time, "sun factor"). Overall, the "sun factor" for the year was 66%. 50% sunny days or better occurred 71% of the time, 90% sunny days or better occurred 23% of the time. Only 3% if the time was the sun factor less than 10%.



I hope I haven't sent you to sleep!

 

[jamesgo]

So whilst we're talking KW Energy... my ODBII adaptor arrived.

Model 3, LR, I've driven 7618 KM, DC Charge Total 1140kwh, AC Charge Total 745kwh, Regen Total 780kwh

Most of my charging is done at a local 50kw DC charger, not much supercharging, hence i've still got >1000km about to expire.

 

[paulp]

I've crunched the full-year numbers for my solar/PW2 installation with some interesting results.

I have relatively small system compared to some here (5.84 kW), but that was the maximum I could fit on my roof using the highest efficiency panels available at the time (365W). I started collecting the Tesla 5-minute data on 1 July 2019, and so now have a full year or 105,406 lines of data in my spreadsheet.

Here's the high-level stats:

• Total solar generation: 8.06 MWh
  
• Total home consumption: 8.23 MWh (so I fell 2% short of solar matching my consumption )
• Total solar export: 2.23 MWh
  
• Grid consumption: 2.82 MWh (only 6.5% of this was at peak rates, 26.1% shoulder, and 67.4% offpeak)
• Net grid consumption: 0.59 MWh (grid consumption minus solar export)
• Solar self-consumption: 67.7% (so fell quite a bit short of my hoped-for 75 to 80%, my 2 months experiment of TBC wrecked it)
• Most solar generated in a day: 36.0 kWh
• Least solar generated in a day: 1.1 kWh

My off-grid-ness:

• Longest continuous time off-grid: 2.82 days
• Longest continuous time on-grid: 1.00 days
• Zero grid draw 75.0% of the time

Some battery stats:

• Battery at minimum level: 25.5% of the time
• Battery at maximum level: 9.5% of the time
• Battery round-trip efficiency: 93.5%
• Number of battery cycles: 236

And a few charts:

This one shows the cumulative net position over the year of solar generation minus house consumption - positive means more solar has been generated over the time period than consumed. I was in the black until 21 June, when it went negative due to a few chilly nights and using the ducted air-con. What a power hog. As Maxwell Smart would say, "missed it by that much"

View attachment 560347

This one shows solar generation over the year compared to what was predicted by the model at PVWatts Calculator. It's pretty good. The system generated 93.6% of what was predicted. I had 4 good months of production well above the model prediction but Dec and Jan production was smashed by extensive bushfire smoke around Sydney, only to be followed up in Feb by weeks of rain. If it wasn't for that, production probably would have exceeded the model and my net position for the year would have been positive.

View attachment 560348

Finally, the distribution of solar production as a percentage of the maximum possible on a given day (i.e. if it was sunny all the time, "sun factor"). Overall, the "sun factor" for the year was 66%. 50% sunny days or better occurred 71% of the time, 90% sunny days or better occurred 23% of the time. Only 3% if the time was the sun factor less than 10%.

View attachment 560349

I hope I haven't sent you to sleep!

Given you have exported more than your grid consumption, have you looked at whether an addition battery would be viable? When I decided on 3 powerwalls my calcs suggested I would fill between 3 and 4 in winter, 3 in autumn/spring, and more than 4 in summer. I went with the autumn / spring number of 3.
I’m now collecting data to see if my assumptions are validated.

 

[Vostok]

Given you have exported more than your grid consumption, have you looked at whether an addition battery would be viable? When I decided on 3 powerwalls my calcs suggested I would fill between 3 and 4 in winter, 3 in autumn/spring, and more than 4 in summer. I went with the autumn / spring number of 3. I’m now collecting data to see if my assumptions are validated.

Funny you should ask that, here's one I prepared earlier...

The chart below is a 'what-if' analysis of what my off-grid time would be if I added more and more PW2 (or indeed installed half or 10% of one). Note the x-axis is logarithmic. The rather surprising result is that it would take 50 Powerwalls to allow me to go completely off-grid.



Now you might say that can't possibly be right, but it turns out it is. Looking at my cumulative net power position chart from my earlier post, the most I was in the black was about 800 kWh. It would take 59 PW2 to store all that 'excess' power so that I could draw down on it without going negative. Applying the actual statistics of my generation and consumption, I don't need quite that many, but the answer still comes out surprisingly high.

Note this was done a few months ago when my net generation was still in the black. It no longer is, so even an infinite number of powerwalls would not get me off-grid now.

To specifically answer your question (at the time I did this analysis), 0 x PW2 battery gave me 36.2% of time off-grid, 1 x PW2 gave 81.4% off-grid, but 2 x PW2 only improves that to 88.5%. So the first battery provides an extra 45.2% of off-grid time (165 days) but a second battery only provides an extra 7.1% or 26 days of off-grid time. Every battery after that adds smaller and smaller amounts of off-grid time.

Given the economics of the first battery was marginal enough (10.1 year payback) the second battery payback time would be in the order of 64 years, and a third battery payback would be nearly 200 years.

Now all of this is specific to my installation and location - PV array size, capacity, solar generation statistics, pattern of household consumption etc. Everyone's curve would be different. But I think the insight here that the economics of additional batteries becomes ridiculously bad extremely quickly would still hold true.

 

[paulp]

Funny you should ask that, here's one I prepared earlier...

The chart below is a 'what-if' analysis of what my off-grid time would be if I added more and more PW2 (or indeed installed half or 10% of one). Note the x-axis is logarithmic. The rather surprising result is that it would take 50 Powerwalls to allow me to go completely off-grid.

View attachment 564786

Now you might say that can't possibly be right, but it turns out it is. Looking at my cumulative net power position chart from my earlier post, the most I was in the black was about 800 kWh. It would take 59 PW2 to store all that 'excess' power so that I could draw down on it without going negative. Applying the actual statistics of my generation and consumption, I don't need quite that many, but the answer still comes out surprisingly high.

Note this was done a few months ago when my net generation was still in the black. It no longer is, so even an infinite number of powerwalls would not get me off-grid now.

To specifically answer your question (at the time I did this analysis), 0 x PW2 battery gave me 36.2% of time off-grid, 1 x PW2 gave 81.4% off-grid, but 2 x PW2 only improves that to 88.5%. So the first battery provides an extra 45.2% of off-grid time (165 days) but a second battery only provides an extra 7.1% or 26 days of off-grid time. Every battery after that adds smaller and smaller amounts of off-grid time.

Given the economics of the first battery was marginal enough (10.1 year payback) the second battery payback time would be in the order of 64 years, and a third battery payback would be nearly 200 years.

Now all of this is specific to my installation and location - PV array size, capacity, solar generation statistics, pattern of household consumption etc. Everyone's curve would be different. But I think the insight here that the economics of additional batteries becomes ridiculously bad extremely quickly would still hold true.

Thanks, your two methods are a very good way of validating and expressing it, especially the 3rd last paragraph. I’ll look at a similar approach to mine now.

 

[miimura]

It would be cheaper to overbuild the solar than the batteries. Curtailing consumption on low production days would also help a lot.

 

[Vostok]

It would be cheaper to overbuild the solar than the batteries. Curtailing consumption on low production days would also help a lot.

My next plan for a “what if” is to do a 3D plot of battery capacity vs solar array size vs self-powered percentage and see what that shows.

It’s still WIP but the few scenarios I’ve done so far show that increasing solar PV generation is way more economic than increasing battery capacity, by a significant margin. So totally aligned with your comment. It also shows that it can make sense to increase the size of the battery in proportion to the size of the PV array, but at a ratio lower than 1, i.e. doubling PV array size doesn’t mean you should double battery size.

In my case, it was not possible to “overbuild” my solar, otherwise I would have done. My roof has as many solar panels on it that can physically fit. I could add some more on the south-facing roof plane, but they would generate meaningful electricity for only about 4 months per year.

And yes, since getting solar/PW2 we have become much more focussed on demand management, shifting consumption to when the sun is shining as much as possible. Interesting how one’s mindset changes when you have “skin in the game”. But still, there’s a limit to what is possible there without putting on the hairshirt.

 

[EcoCloudIT]

My next plan for a “what if” is to do a 3D plot of battery capacity vs solar array size vs self-powered percentage and see what that shows.

It’s still WIP but the few scenarios I’ve done so far show that increasing solar PV generation is way more economic than increasing battery capacity, by a significant margin. So totally aligned with your comment. It also shows that it can make sense to increase the size of the battery in proportion to the size of the PV array, but at a ratio lower than 1, i.e. doubling PV array size doesn’t mean you should double battery size.

In my case, it was not possible to “overbuild” my solar, otherwise I would have done. My roof has as many solar panels on it that can physically fit. I could add some more on the south-facing roof plane, but they would generate meaningful electricity for only about 4 months per year.

And yes, since getting solar/PW2 we have become much more focussed on demand management, shifting consumption to when the sun is shining as much as possible. Interesting how one’s mindset changes when you have “skin in the game”. But still, there’s a limit to what is possible there without putting on the hairshirt.

Agreed from a cost/ROI perspective, I have the 1x PowerWall but by Tesla's metric I should have 1x more (if not 2x) since I am running around a 12KW system....however when I am not off grid (battery at 3%) and drawing from the grid my solar is large enough that the FIT offsets and grid draw (maybe not on the day but the next sunny day etc.)....generally this Autumn/Winter I am off setting all grid draw events 2 to 1 and making another 8.5kw per day (again over an average period of say a month).....so technically from a cost perspective off-grid…..this includes plugging in one of the cars when we are getting 7-8kw’s of power generation (in the middle of Winter, get more than this in Summer of course)….obviously SOH makes it easy to always have 250km’s+ in the cars….would probably struggle under normal circumstances so I guess COVID-19 has some benefits

 

[paulp]

It would be cheaper to overbuild the solar than the batteries. Curtailing consumption on low production days would also help a lot.

If there is no space capacity then you cannot add more solar

 

[paulp]

My next plan for a “what if” is to do a 3D plot of battery capacity vs solar array size vs self-powered percentage and see what that shows.

It’s still WIP but the few scenarios I’ve done so far show that increasing solar PV generation is way more economic than increasing battery capacity, by a significant margin. So totally aligned with your comment. It also shows that it can make sense to increase the size of the battery in proportion to the size of the PV array, but at a ratio lower than 1, i.e. doubling PV array size doesn’t mean you should double battery size.

In my case, it was not possible to “overbuild” my solar, otherwise I would have done. My roof has as many solar panels on it that can physically fit. I could add some more on the south-facing roof plane, but they would generate meaningful electricity for only about 4 months per year.

And yes, since getting solar/PW2 we have become much more focussed on demand management, shifting consumption to when the sun is shining as much as possible. Interesting how one’s mindset changes when you have “skin in the game”. But still, there’s a limit to what is possible there without putting on the hairshirt.

Someone I know has done south facing 17 degree panels in Adelaide. In winter the southern panels produce near the same as his north panels. Less production in summer, but they produce more in the early morning and late arvi in summer than the north panels.