AC coupled battery system

[Oldgreybeard]

Hi all, just posted on the intro bit, used to be on the other place years ago, and have been lurking there but not posting for years.

I built a passive house a few years ago, with the whole of one side of the roof clad with in-roof PV instead of slates, two strings with a total of 6.25 kWp, running to a 6 kW inverter. I had intended to install a battery system at the time, but the costs were just too high, and with the overall low energy use of the house (it is all-electric - I don't believe in burning stuff locally) there wasn't that much potential to save money just by storing excess PV generation.

I already have a "heat battery", an electrically heated phase change thermal store, that provides instant hot water and stores between 9 kWh and 10 kWh of heat. That is mainly heated by excess PV generation, and only boosted by the grid at night, during the off-peak period, if there's not enough sun during the day. The heating is an ASHP that also runs at night in winter, and uses the insulated concrete ground floor as a storage heater (that sits on 300mm of foam insulation and has UFH pipes cast inside it). We run on Economy 7, rather than one of the shorter period TOU tariffs, as we really need the 7 hour cheap rate slot, not least because both of our cars are EVs, also both always charged overnight. We have a 100 A fused main supply, and the EV chargers are now on an automatic priority board, so only one at a time can draw power, so there's plenty of headroom at night for battery charging. The DNO granted me an "up to 10 kW export" permit when I applied for the G59 consent for the PV, so I should have no problems from them.

I've just switched to a supplier that's offering a two year fixed term tariff, with the 7 hour overnight slot being only 7p/kWh. That's so cheap that it seems rude not to try and use as much of it as we can, and try to reduce, perhaps even remove, all our peak rate usage. Hence this post, as the time has come to fit batteries,

I ran cables in via ducts to a "battery house", when I built the house, as I always assumed I'd do this one day. So, the plan is to install 3 off Pylontech US3000 battery packs, to give around 10.5 kWh of storage (at least initially) and I think it looks as if the SoFar ME3000SP inverter/charger will meet my requirement, if set to Time of Use mode. Be useful to confirm this though, and understand what quirks it may have. A friend has a Tesla Powerwall, and that seems to have a mind of its own, and that's something that does concern me a bit.

What I want to be able to do is charge up the battery pack from excess PV, after the hot water heat battery has charged. If there is still spare capacity after that, then I want to charge the battery up overnight, during the off-peak period. Ideally I want to prevent grid usage during the peak rate completely if I can. Our worst case daily (peak rate period) electricity use, excluding the EV charging (which is always at night anyway) is around 7 kWh, excluding any offset from PV generation during the day. It rarely peaks above about 3 kW, and then only for a short time when the oven or hob is first turned on, so I think the 3 kW output from the ME3000Sp should be OK. I'm reluctant to over-size the inverter, because that probably increases the phantom power usage a bit, so the sweet spot is probably around 3 kW I think.

At this stage I'm really looking to see if my ideas make sense, and whether anyone with experience with the SoFar ME3000SP can see a glaring problem that makes it a non-starter for our requirement. Be really grateful for any comments or observations, especially as experience is worth far more than a millions words of Chinglish in a manual . . .

[Stinsy]

Hi OGB,

Your plan seems to be a good one!

A few points:
- Most people prioritise charging the batteries over heating water. The energy in the batteries can be used for anything (including heating water) whereas putting it through a resistive heater is 2nd only to allowing it to spill into the grid. The easiest way to achieve this is to program the "diverter" attached to the hot water heating element to target 100W (or whatever) export and program your battery system to target 0W export.
- I have no personal experience with that inverter. However it is very popular, I've not heard of it being unreliable, and I believe it can do what you need.
- You can get EV charging points that intelligently monitor you house total demand so as to ensure you don't go over your main fuse rating. You should be able to charge 2x EVs simultaneously on a 100A fuse because that still leaves over 8kW for everything else. Even if you did have a 12kW electric shower or several 3kW loads on simultaneously, they tend only to pull full power for a short period and the aforementioned intelligent charging points can vary the charge rate of the EV during those periods.
- I have a similar setup, I use the Octopus Go 2030-0030, 5p, cheap rate period. Here is yesterdays chart:



I'll explain the chart:
- At the start of the chart you can see where the batteries are programmed not to discharge during the cheap period so my demand (orange) is matched by the power I'm pulling from the grid (red).
- From 0030 onwards the battery discharge (blue) matches demand and grid reverts to zero.
- Then when the solar (green) comes online the battery is charging/discharging to keep the grid power at zero. The jagged grid is because the induction hob was on all day (cooking batches of bolognese/chilli for the freezer). For some reason the grid usage is never flat while the hob is on but I've learned to ignore the few W in either direction as insignificant.
- Then at 2030 the batteries started charging and we ran the dishwasher (no EV connected yesterday). It was a sunny day so the batteries were already at 62% when charging started so the charging period was brief.
- Yesterday according to Octopus I used 0.69kWh of "peak" electricity and 8.11kWh of "off-peak".
- Today is overcast and the washing machine is on so unless the sun comes out this afternoon we won't get all the way through to 2030 on batteries.

[Oldgreybeard]

Thanks for the reply.

Perhaps I should have made it clear that I don't heat water directly, but store heat in a very efficient phase change thermal store, that uses a sodium acetate compound to store heat very efficiently, by exploiting the energy needed to make it change phase. This is more efficient than the double losses through a battery charge and discharge cycle, that will typically be between 5% and 10% charging loss (assuming a switched mode charger efficiency of between 90% and 95%) and another 5% to 10% discharge loss through the inverter switched mode losses, that will be around the same magnitude. That means that charging the battery, only to then discharge it into our form of water heating, would be less efficient.

The total losses in our heat battery are around 600 Wh/day, for a hot water usage that's typically around 8 kWh per day, so about 7.5%. If an inverter/charger was, say, 95% efficient both ways (and I don't think any of them are this good) then we'd have 10% round trip losses just from the battery, plus whatever losses we had in the water heating system. There would have to be water heating losses, as I don't think an inverter capable of supplying the ~15 kW needed to deliver a modest hot water flow instantaneously makes sense, either economically or in terms of phantom losses, given the mean power requirement. Our heat battery can deliver around 30 kW on demand, as it "flash heats" water when required, by triggering the release of stored heat from the phase change from liquid to solid in the sodium acetate mixture.

For this reason, I must prioritise charging the thermal battery over charging the battery storage system, as we always need hot water, and it's more efficient to charge the thermal battery, due to it's lower round trip losses.

I should also perhaps mention that I built a monitoring system into the house, so there are around 15 microcontrollers dotted around the place, all synchronised by a GPS master clock in the loft, that collect data from around 30 sensors, ranging from temperature, humidity and CO2 at various points, to the power used by various circuits. Data are logged every 6 minutes, and can be used to control things like the car charge points. I actually stopped faffing around trying to use excess PV to charge the cars as it's more hassle than it's worth. The algorithm I used was to charge the car whenever it was cheaper to do so using a mix of grid and PV generation than it was to charge overnight at the cheap rate. This was a work around because of the 6 A lower limit in IEC61851, and also because the car OBCs get horribly inefficient down at 6 A, with very high losses, as most of the switched mode charging losses are fixed, so as charge current reduces efficiency plummets.

In the year or so I was using data from the house monitoring system to control my car charging I saved a grand total of just over £11, so although sort of worth doing, given that it meant having the car plugged in all the time I thought that hassle just wasn't worth the small saving. With this new, pretty cheap, E7 rate, the savings will be a lot greater, as before we were paying around 11p/kWh for off-peak, and now we'll be paying only 7p.

The reason for the priority switch on the charge points is because of the peak demand, not the continuous demand. There can be very short periods when the heat pump is running flat out at start up, the heat battery is boost charging because there's been no solar that day, and our borehole pump will be running for 20 minute or so backwashing the sand filter. That can push the peak load up, albeit only for a short time, but as we never need to charge both cars at the same time (or haven't ever done so far) it's no hassle to just let the priority board do its thing and make sure there's always headroom. This also keeps the DNO happy, as when I notified them about the second charge point (a mandatory requirement) they did stipulate that I needed to include load management, as they weren't happy with a 64 A load for car charging.

I've been tracking Go (and Agile, although that now looks pretty dire) for some time, but the off-peak period is just too short. There are times in winter when we need 6 or 7 hours of charge in the floor slab, during very cold spells, and my car has a pretty big battery pack, and having ageing relatives I always like to be sure that I have a couple of hundred miles of range available every morning, in case we need to rush off. The other issue is that we cannot have a smart meter here, as there's no signal. They've been out and done a survey, and nothing can be done, apparently. We're too far away from our nearest neighbour for the mesh system to work, plus there's no signal where they are, either, so at the moment they can't have a smart meter either. TBH I'm not that bothered, as E7 works very well for us, and I designed the house and it's systems around using E7, especially the heating system, so as long as we can still get a decent E7 tariff we'll be OK.

I'm reasonably sure the SoFar inverter/charger will do what I want to do, but I've been caught out a few times by Chinese stuff not having the clearest of manuals - not uncommon for things to get mixed up during translation, it seems. What might be interesting is how my (home made) excess PV diverter system interacts with the sensing the SoFar uses. At the moment I use an "energy bucket" algorithm, that uses the inherent built-in leeway in the meter to shuttle half a Wh back and forth, as a way of transferring as much energy to the heat battery as possible, without triggering the meter to read an import. I have a small "leak" built in to the "energy bucket" that ensures it always errs slightly on the side of export, but only by enough to overcome any tendency to digitally drift during very low excess PV generation periods.

If the SoFar samples it's CT faster than the ~16k samples/sec that I sample the current and voltage at the incoming supply, then there's a chance that the battery charging may take priority over the heat battery charging. I'm hoping that this isn't the case though, as if the SoFar samples more slowly then things should be OK. I think I may be able to get around this if it happens, by using another of the time of use slots to stop the battery charging for, say, the first few hours each morning. Not ideal, but probably workable. What I don't want to happen is have the battery charge the heat battery, as that would be really wasteful, and probably impact on the energy we have available to offset peak period usage. It's a pity that details like this aren't clearer in the documentation for these things, really.

[Tinbum]

It sounds as though you have most things covered and the knowledge and ability to adapt to whatever way you go. I'm presuming you use an Arduino for your energy diversion so you should be able to adapt that sketch slightly if it were not to work as you want.

I'm installing US3000C batteries at the moment and using an Arduino to get them to talk to the SMA inverter (with settings adapted) and Eltek Chargers for night time charging. Most things are possible and their's loads of information on the net.

[Oldgreybeard]

That sounds an interesting approach. I'm not using Arduinos in the house, but PICs, but I like the idea of having total control myself, rather than rely on whatever functionality has been built in to things. I have spotted on Github that someone has managed to control one of these inverter/chargers by opening another port to their (optional) wifi connection, so gaining access in parallel with the Chinese control server they normally connect to.

My plan is to try and keep the system free from any connections to China if I can, simply as I don't much like the idea of opening access to a Chinese server from my LAN, and don't want the faff of setting up a separate WLAN just for the inverter/charger. It does look as if the inverter/charger may be able to be controlled via either CANbus or RS485, although I can't find any complete/reliable documentation on this. There's mention of some sort of test mode, that allows external control, and this does open up the possibility of being able to completely customise the system to do exactly as I need, without using any of the built-in programmed functions.

Your approach of using a separate charger and inverter offers a lot more flexibility, thanks for planting the seed of another idea - I shall now spend (waste, my wife would call it) another afternoon on seeing what possibilities this option may have.

[Tinbum]

I used to use the energy bucket system to control my loads but I changed to 3 phase and found it wouldn't work with my new meter. I then changed over to a frequency system.

The Pylon CANbus documentation is available on the internet though it may be slightly out of date. I seem to have one piece of data coming from the batteries that isn't mentioned in it.

[nowty]

Not much to add as you seem to know your stuff.

I have used a Sofar ME3000Sp and "Time of Use" mode on them will give you what you want in terms of charging batteries on the cheap rate. In terms of interfering with your home brew diverter, as long as you can set a minimum export value, then the Sofar should work as I do this with my Immersun diverter.

I still have the Sofar ME3000Sp but now only use it in winter as a supplementary battery charger as I now use a 8kW Sunny Island as the battery inverter. There have been some very cheap ones coming up on ebay as some installers are persuading folk to move to the Lux battery inverters as there is some form of integration with Octopus Agile tariffs.

Is either of your EVs a Tesla ?, because if it is you could join Octopus Intelligence which I am planning to do. Gives a fixed 6 hour 5p slot, plus additional 5p slots if the wholesale price is low enough.

[Oldgreybeard]

Many thanks for that useful info about the ME3000SP, especially the bit about setting the minimum export value. I'd spotted the ~2 year old used ME3000's coming up on eBay, and wondered what that was about, too.

I used to own a Tesla Model 3 LR, but frankly got fed up with it, for a host of reasons, including umpteen visits to the service centre getting manufacturing defects sorted, so I sold it earlier this year. I bought it in November 2019, though, and so was one of those unfortunates that got hit with the first batch of really crap cars from Fremont, I gather the newer MIC cars are a lot better built.

Octopus is out for us, both because we can't have a smart meter, (no prospect of a signal of any sort here for it to use to connect to DCC), but mainly because I had a REALLY bad time with Octopus a few years ago, ended up having to go to the regulator to get hundreds of pounds they owed me for incorrect billing back. Not my favourite supplier, TBH. Although I'm sure they've probably improved since then, I still have a bit of that "one bitten, twice shy) feeling left. Never did recover all our costs from them, either, in the end I just wrote off the last couple of hundred pounds, as life's to short to trace scoundrels forever.

Do you happen to know if the Lux Power inverters are better than the SoFar ones, aside from the integration with Octopus Agile? I have looked at them, and there doesn't seem to be much to choose between them, apart from Lux offering a slightly higher output, IIRC (not sure we really need that). TBH, I find it hard to see any benefit in Agile now, the price has been sky high for ages, probably the most expensive tariff going lately.

I've switched to Good Energy, as they have an EV tariff that offers 7p/kWh with an E7 meter, that suits us well. Bit expensive during the peak rate, and the standing charge is higher than I'd like, but overall it should work out a lot cheaper if I can get the battery system running as I want. I think it's going to get tougher for those of us that are never likely to get smart meters to find good tariffs before long. Might mean I have to look to see if we can go off-grid completely if that happens. We're already off-grid for water and sewage, so the only utilities we have are electricity and a phone wire. Can't get rid of the phone wire, for the same reason we can't have a smart meter, no signals here, which is a PITA, as broadband isn't much better here than dial up used to be.

[Stinsy]

How much room do you have for a ground-mount array? Going off grid for electricity isn’t out of the question but I’d suggest the following:

- An additional 12kWp of panels (minimum).
- A generator (you will need it in the darkest mid-winter days).
- An alternative source of heat (eg wood burner/bottle gas boiler).

I know some of those go against your principles but you needn’t run the generator/ aux heat source very frequently, however on the darkest midwinter days you will need them. On the other hand the grid is a great way of delivering clean electricity from a wind turbine in the North Sea or Cornwall to your house…

[Oldgreybeard]

How much room do you have for a ground-mount array? Going off grid for electricity isn’t out of the question but I’d suggest the following:

- An additional 12kWp of panels (minimum).
- A generator (you will need it in the darkest mid-winter days).
- An alternative source of heat (eg wood burner/bottle gas boiler).

I know some of those go against your principles but you needn’t run the generator/ aux heat source very frequently, however on the darkest midwinter days you will need them. On the other hand the grid is a great way of delivering clean electricity from a wind turbine in the North Sea or Cornwall to your house…

Sadly none, we had a real struggle to get consent to fit the in-roof panels, as we're in an AONB and a conservation area, adjacent to a listed building , just to make things even harder. We do have a possibility of putting a wind turbine up on the ridge behind us, and have been exploring putting a tower up there to give the few of us in this valley with crap broadband a mobile signal. So far we've had no specific objection to the idea of putting a tower up there, although we've not mentioned the dreaded "wind turbine" words yet, and they often seem to bring out the worst in the NIMBYs. We had enough trouble with NIMBYs when we were getting planning consent - the real problem being that it seemed many people can't read plans and visualise what something will really look like.

Our energy needs are pretty modest, our worst case heating power requirement is only about 1.5 kW when it's -10°C outside, and it's never got down that low since we've been here. The snag with that is that even the very smallest canal boat sized wood burner would overheat the house very quickly, and no one makes one that can deliver just a couple of hundred watts of heat, which is all we need in winter, normally. If we really had to burn something, then LPG would make the most sense, perhaps a small LPG water heating unit that could heat up the UFH in the slab. Something caravan size would be more than adequate, I think, and could probably be run off bottles, given the relatively low usage.

Realistically I think we are stuck with the grid, and as you rightly say, the grid is pretty clean, and getting cleaner all the time, so seems the least polluting energy source available. It's just that when you've gone down the rabbit hole of reducing CO₂ emissions to the point where the net emissions from our house and both cars is negative, there's this strong subconscious desire to do just a bit more, and try and cut free from the dependence on the grid. Not really rational, but then so much about the things we chose to do isn't really rational, anyway. Lots of people around here think I'm already a bit of a nutter for building a passive house, that has no foundations and just sits on 300mm of polystyrene foam laid on the ground!