Can a 7kWh and a 10kWh PowerWall be used together in a single installation?

[ecarfan]

@schueppert, thanks for your comments, and that does seem like a reasonable approach to buy two 7kWh units instead of a 10 and a 7 though it would have slightly less total backup capacity (14 instead of 17).

I am not convinced that Elon said the use case I posed was not currently supported. He was not completely clear on that, in my opinion.

 

[Ampster]

As @Electracity pointed out two Power walls in series is 700v. I don't know of an inverter that is designed for that. Optimizers are in series in order to get the string voltage up to 350v.

I do agree that the best use case is to parallel two 7k power walls and set your inverter to only use half (7k) for cycling and save the other half for backup. Since the Power walls are in parallel, the inverter would draw evenly from each one.

 

[Ampster]

........

I am not convinced that Elon said the use case I posed was not currently supported. He was not completely clear on that, in my opinion.

Until someone or some inverter installation manual shows us how , I am completely clear that it has not been demonstrated to work. As I mentioned earlier, I could make it work with some high amperage relays and some logic, but why bother? I would much rather have the flexibility for example to use 10kWhrs of 14kWhr paired Powerwalls. The likelihood of a need for backup at the end of the cycle is the same as the likelihood of needing backup at the beginning of a daily cycle. In the first case I would have 4kWhr left for backup and in the second case I would have 14khr available for backup. That is a more optimum use of resources in my mind.

 

[schueppert]

How could powerwalls be connected in series?

As I say, I think you will find it works in exactly the same way as the SolarEdge optimizers. The SE optimizers are connected in series and the output from the string is a constant 350V volts regardless of how many optimizers are in the string.

 

[electracity]

As I say, I think you will find it works in exactly the same way as the SolarEdge optimizers. The SE optimizers are connected in series and the output from the string is a constant 350V volts regardless of how many optimizers are in the string.

Microinverters work in parallel. There would be no advantage of microinverters if they were wired in series.

 

[Ampster]

Yes, microinverters are in paralell so you get increased amps with constant voltage. In the case of optimizers they are in series because the Panel connected to the optimizer puts out 40 to 60 volts. So you need them in series to get to 350 volts. There are string sizing tools that determine the number of optimizer/panel combinations.
We know that each power wall runs at 350 volts. There would be no logical reason to put a power wall in series unless you wanted 700v.

 

[schueppert]

There are significant technical issues with connecting batteries in parallel. It's certainly is not an architecture that scales to utility level. Tesla has confirmed PW contains a DC-DC converter and it is hard to see why if it is just a dumb 350V battery as you suggest. And technical details provided by SolarEdge suggest a solution based on, and very similar to, their optimizer architecture.

Regarding the original question of combining 7 kWh and 10 kWh packs. If PW is just a dumb 350V battery connected in parallel then clearly that combination would not be possible since there would be no way for the inverter to draw power from one pack and not another. The fact that Musk at least paused about the answer hints that we are dealing with a more complex architecture that can be controlled by software.

 

[Cosmacelf]

Anyone know anyone over at SolarEdge? Maybe they can tell us. I doubt we can get more info from Tesla...

 

[Ampster]

There are significant technical issues with connecting batteries in parallel. It's certainly is not an architecture that scales to utility level. Tesla has confirmed PW contains a DC-DC converter and it is hard to see why if it is just a dumb 350V battery as you suggest. And technical details provided by SolarEdge suggest a solution based on, and very similar to, their optimizer architecture.

Regarding the original question of combining 7 kWh and 10 kWh packs. If PW is just a dumb 350V battery connected in parallel then clearly that combination would not be possible since there would be no way for the inverter to draw power from one pack and not another. The fact that Musk at least paused about the answer hints that we are dealing with a more complex architecture that can be controlled by software.

Tesla has apparently overcome many of those technical difficulties since their pack consists of a number of small cells in parallel and series. WK057 has strung together a number of Tesla Bricks (not sure of the correct term) in parallel on his thread here:
Plan: Off grid solar with a Model S battery pack at the heart

I never said the Powerwall was a dumb battery. It contains a DC to DC converter and a battery management system. I believe that the battery management system in each PowerWall is what would allow them to be connected in parallel. I have to admit, I haven't seen the SolarEdge technical details that suggest the similarity to their Optimizers. In the meantime I will take a look or if you have a link that would be helpful. I would guess that the integration of the Optimizers, Inverter and PowerWall includes the communication to the PowerWall that we know takes place between the Optimizers and the inverter.

I agree the Powerwall would not scale up to the utility level but Tesla has already installed Powerpacks that appear to scale up pretty easily. Once an installation was designed it appears to be a simple matter of connecting more powerpacks in parallel to get more capacity. To do that in series with Powerpacks would mean a different inverter input voltage each time you added more capacity. In the thread referenced above, WK057 has 6 or 8 inverter/pack combinations in parallel. The inverters communicate and deliver power as needed by the minigrid. In essence each inverter comes on in parallel.

I initially understood that the purpose of the DC to DC converter is to enable a solar string to charge the PowerWall. You pointed out in another thread that the Optimizer has the capability to regulate voltage and it may be the communication between the optimizer and the DC to DC converter that controls the charging.

 

[schueppert]

The WH057 thread is probably the most awesome discussion on this board. What he has done in amazing. But I'm not convinced his battery is going to last 10 years. He has basically removed/disabled the sophisticated Tesla battery management system and replaced it with off the shelf charge controllers. It works, but it's not ideal.

There are some interesting comments from SolarEdge in an interview here: Solar InterviewsSolarServer. I stand corrected because while this article does describe a close relationship between the battery and optimizer solutions, he states the batteries connect in parallel rather than in series as is the case with the optimizers.

 

[dhanson865]

The WH057 thread is probably the most awesome discussion on this board. What he has done in amazing. But I'm not convinced his battery is going to last 10 years. He has basically removed/disabled the sophisticated Tesla battery management system and replaced it with off the shelf charge controllers. It works, but it's not ideal.

He didn't just do off the shelf, he wrote or will write his own controller code. I think you are underestimating his skill and the level of detail he went through and is continuing to persue on that project.


Dec 2014

Not as of yet, but I will in phase 2 of the project when I get my custom BMS system completed.

Jan 2015

I'm in the process of prototyping my custom BMS. Unfortunately I need to take a short break from this project to finish up some other work this week.

mar 2015

I will be adding the cooling loop once I complete my custom BMS setup, though.

Mar 2015

I will be monitoring every single 74-cell set with my custom BMS (> 200 of them) once I have time to finalize the design and get some PCBs printed.

 

[schueppert]

I think you are underestimating his skill and the level of detail he went through and is continuing to persue on that project.

Not at all, I'm a fan. But the quotes you provided support what I said: his existing implementation is based on off-the-shelf charge controllers, he knows this is a problem, he is building his own BMS, this is not a trivial task.

 

[electracity]

There are significant technical issues with connecting batteries in parallel. It's certainly is not an architecture that scales to utility level. Tesla has confirmed PW contains a DC-DC converter and it is hard to see why if it is just a dumb 350V battery as you suggest. And technical details provided by SolarEdge suggest a solution based on, and very similar to, their optimizer architecture.
...........

The details will be interesting. Powerwall is announced to work with inverters besides solaredge. So whatever their solution is to multiple powerwalls it can't really on solaredge specific functionality.

- - - Updated - - -

Yes, microinverters are in paralell so you get increased amps with constant voltage. In the case of optimizers they are in series because the Panel connected to the optimizer puts out 40 to 60 volts. So you need them in series to get to 350 volts. There are string sizing tools that determine the number of optimizer/panel combinations.
We know that each power wall runs at 350 volts. There would be no logical reason to put a power wall in series unless you wanted 700v.

Sorry, I read too quickly and saw "microinverter".

 

[Ampster]

Even though we have hijacked the OP'S thread I think the collective knowledge has benefitted. Still the unknown is how the PowerWall will communicate to the Inverter/optimizer string. I presume it will be through the same PowerLine communication that the optimizers use. I hope that is an open architecture so others are encouraged to innovate in that space.

 

[schueppert]

Even though we have hijacked the OP'S thread I think the collective knowledge has benefitted. Still the unknown is how the PowerWall will communicate to the Inverter/optimizer string. I presume it will be through the same PowerLine communication that the optimizers use. I hope that is an open architecture so others are encouraged to innovate in that space.

SolarEdge are using powerline communication for their "Storedge" product. It's certainly possible Tesla is using this for PW also. But my guess is they will probably use RS485 instead. Most inverters, including SE and Fronius, already have a port for this. And the benefit of avoiding an extra wire is not so great from inverter to battery as it is for inverter to optimizer.

 

[Ampster]

SolarEdge are using powerline communication for their "Storedge" product. It's certainly possible Tesla is using this for PW also. But my guess is they will probably use RS485 instead. Most inverters, including SE and Fronius, already have a port for this. And the benefit of avoiding an extra wire is not so great from inverter to battery as it is for inverter to optimizer.

Excellent point. And the benefit in the big picture is that if some of that high level communication can be communicated to the larger grid we could move in the direction of a more intelligent grid.

Imagine that instead of buiding a peaker plant, your utility implemented communication with all tied in battery banks out there. Then instead of paying high rates for a few megawatts of capacity for 15 minutes the utility could instantaneously draw that power from the batteries in the distributed intelligent grid. Later at an off peak time those batteryies could get recharged.
BTW thanks for the link to the Solar Server StoreEdge article. Very informative and great insight into where Tesla Energy and Solar Edge are heading.

 

[schueppert]

Imagine that instead of buiding a peaker plant, your utility implemented communication with all tied in battery banks out there. Then instead of paying high rates for a few megawatts of capacity for 15 minutes the utility could instantaneously draw that power from the batteries in the distributed intelligent grid. Later at an off peak time those batteryies could get recharged.

Cool thoughts. Quite a bit of smarts needed for the inverter to determine when to store excess power in the batteries and when to export to the gird. As OP question illustrates, power management needs are likely to be varied and complex. Export algorithm may depend on some or all of: time of day, anticipated solar production, battery SOC etc.. Utility signals would create interesting options.

 

[dhanson865]

Cool thoughts. Quite a bit of smarts needed for the inverter to determine when to store excess power in the batteries and when to export to the gird. As OP question illustrates, power management needs are likely to be varied and complex. Export algorithm may depend on some or all of: time of day, anticipated solar production, battery SOC etc.. Utility signals would create interesting options.

I'd expect you would want something resembling the user interface for an ECObee thermostat web app so you could set your max and min desired for several time periods in a day and several days in a week with no limit on how many time periods you can configure.

Add your requests/scheduling on top of all the other factors you mentioned (production, SOC, utility requests).

 

[Ampster]

...... Quite a bit of smarts needed for the inverter to determine when to store excess power in the batteries and when to export to the gird. As OP question illustrates, power management needs are likely to be varied and complex. Export algorithm may depend on some or all of: time of day, anticipated solar production, battery SOC etc.. Utility signals would create interesting options.

As I have discovered with the Outback Radian inverter a lot those options are enabled in the firmware or the Mate3 interface. There are also Web based interfaces that I haven't explored yet. Currently most of the hybrid systems like these operate in the low voltage (12-48) DC voltage range. I hope we see a convergence of the software options available in the future on both low voltage and high voltage systems.