How can I check the actual capacity of my battery ? (Calibration ?)

[mustjust]

This forum continues to be really useful so I wonder if someone help me with this conundrum ? I have a newish system which did not do well in the cold weather since Dec but is now working more or less as expected. My problem is the apparent capacity of my 2off HV2600s is less than 3.7kWh instead of 5.2kWh, and I realise this may be faulty reporting of the SoC, poor operation of the inverter or duff batteries, or a combination of all 3 !

This is how I know there's a capacity question - I have done the following on several of the last few (sunny) days with a conventional setup that includes a separate generation meter between the inverter and my consumer unit:
1) Record the export energy after the sun is down along with the reported SoC (and any energy import)
2) Repeat the records in the following morning before the sun shines whilst the battery has fed the house (and confirming no battery charging from the grid by observing no change to the import total)
3) Calculate the overnight exported energy (in kWh) and the change in SoC in %.
4) Divide the energy delivered by the change in SoC and multiply by 100 to get the notional 100% capacity.

I have done this several times with reasonably large changes to the SoC (60-75%) and the result is a consistent 3.65 kWh.

I can see this calculation does not allow for conversion losses in the inverter and my system is typically only drawing 150W or so between occasional short bursts nearer 1kW which I can imagine is inefficient for the inverter. Also my batteries are in an outbuilding so have been at around 10deg C during this period.

However, I am surprised if these factors would account for a 30% loss in battery capacity of new kit.

I would be please to know if anyone explain or suggest a way to calibrate the system's measurement of SoC ? Thanks.

PS I see an answer to this would resolve Billsfan's problem also posted today.

[Dave Foster]

I was told as a young engineer that the only way to find out how much capacity a battery has is to fully charge it, and then discharge it and measure the actual power it gives up.

The way to get an approximation of the battery capacity is to look at the amount of daily battery discharge in kwh and divide it by the change of SoC over that period (less any charge they receive from solar over the measured period) but unless you measure across the entire 100%->10% range the BMS SoC reporting and calibration will be at best vague.

The battery discharge/charge figures need to be taken off the website daily statistics as this is the measured real energy the batteries have given up and will include any losses in dc>ac conversion and keeping the inverter awake - it’s a difficult sum to perform I do it using a home assistant computer but it still varies throughout the day.

Also don’t forget the usable battery capacity is 5.2kwh less your minsoc setting so at 10% maximum usable is 4.68kwh

Temperature also has it’s affects at this time of year, the batteries are rated 100% at 25C ambient and will fall to approx 85% of that at 0C

Having said all of that, battery chemistry and characteristics of LifePo4 appears more a dark art - I have an 18.2kwh pack and on average get 15.4kwh usable, my personal best is 16kwh so still a little behind where I would wish them to be.

[mustjust]

Thanks, Dave - your detailed knowledge and experience are very helpful. I have left it a day or so to see if there are any other responses to my post. I can tell this is a tricky one.
I too have received the same advice in the past to assess battery capacity, and I have done exactly what you described in my tests, so last night I tracked from 94% SoC to 14% this morning and my generation meter which has the advantage of being built to a high measurement accuracy, recorded exactly 3kWh of export (and no import).
The temperature of the battery was around 10deg C, so I can imagine my realistic 100% capacity was down by 10% to 4.6kWh
If I focus as you do on usable capacity then I should be seeing about 4.1kWh of capacity between 100% and 10% SoC - what I am seeing is under 3.4kWh

This discrepancy suggests something else is wrong does it not ? Your figures suggest you might be losing 6% or so of your rated capacity. I seem to be losing more like 17%.

Any further thoughts on more precise measurements that don't rely on the "snapshot" recording of power by the inverter but have accurate integrations of the energy over time (as per the generation meter) would be most welcome,

[Dave Foster]

Is the generation meter you are using on the inverters AC output ?, I agree a meter is far more precise than a CT clamp.

If it is on the AC output it's a difficult metric as you are looking at the precise output of the inverter and not the energy given up by the batteries, the FoxEssCloud website will give you battery discharge power in kwh but it is difficult to do the calculation manually whilst load, solar input and charging is changing.

I use a home assistant that calculates the near real time losses at any point in time, this includes the 60~70W the inverter needs to stay operational which it takes these from the DC side (battery/solar), not the AC output - this is the current graph with some solar coming in as well

When you are doing your 94% to 14% soc calculation do you take any losses into account?, over time they will become more significant - it's probably safe to say overnight there may is a typical average of 120Wh lost so if the 80% change in SoC is (say) over 10 hours, you would need to account for the 1.2kWh which would come from the battery.

Of all the things in your system, solar generation, grid usage, export and in particular battery health are such key parameters and I would strongly recommend the installation of a Home Assistant as even when connected to the FoxESScloud it gives you access to many more sensors and even with 5 minute granularity of the cloud it's easier to work out solar generation, battery use, capacity etc.. in almost real time. - you can see more here. viewtopic.php?t=18

[mustjust]

That's really helpful, Dave - you have identified the key issue (once again). I am grateful for the clarification.

For the benefit of any other readers interested in this thread: yes, my generation meter sits on the supply to/from the inverter - between the AC isolator for the H1-5.0-E and the consumer unit feeding it.

Similarly, I recognised my calculation ignored the inverter conversion losses. Since it is fed by the battery and I am only recording the energy output from the inverter, I can't measure those losses. I was mostly relying on the quoted numbers in the datasheet which refer to 97% efficiency for Battery to AC, though I appreciate the caveat of a spec. at "full load". see below:

Your assessment of 60-70W just to run the inverter is 3x more than I was reckoning from the figures I see on the power figures. Typically the PV power is 20W or so greater than the delivered power to the battery plus the house. I must look more closely when the house load is met from the battery but only 100-200W. You are correct that my overnight energy demand is mostly below 150W - with your figure of 60W to simply run the inverter, then I am easily losing 30% of the battery energy just to keep my house "alive" when the sun is not shining. That is a lot and means my original assessment that a 5.2kWh battery should be sufficient may be seriously out.

I'd love to hear the experience on this score from others - it seems the real world performance of a battery system may be well below the rated figures (e.g. 20%) and this should be better understood by the salesmen, installers as well as us users.

Finally, I have been preparing for Home Assistant and will probably get there before too long. I am less excited as I read HA will not be able to set parameters on my H1 inverter and so make decent use of smart tarrifs with more intelligent decision making. I will be pleased to have more accurate real-time data as to how the system is working, but I am more interested to be able to exploit it to deliver genuine cost savings.

Overall then, I am going to pursue the available smart tarrifs as a priority - Octopus Flux looks extremely interesting.