Battery drain increase but load was the same

So I noticed something unusual this morning when checking my batteries, we had load shedding between 12am and 2am, my SOC was set to 35% as its been for a while, my usage is typically 500w to 700w in the middle of the night depending on if we have a fan on in my sons room.

I noticed that between 12:11 and 1:09 ±1 hour the batteries drained about 7% the load was ±750w but between 1:30 and 1:46 ±16 minutes the batteries drained 7% the load ±770w over a much shorter space of time, the load didn’t increase by 4 times (which was the draining increase) so just wondering what it could be?

They are Pylontech batteries, about 3 years old and SOH is 94%

1 Like

Those look like home assistant graphs. Maybe zoom in the graph to cut out the peaks and compare again eg 12:00 to 1:45 so that you can actually see the consumption. What metric did you use for power? does it include the power factor? What inverter are you using?

How many batteries?

To me it looks like some of them might have reached their cutoff SoC and you are left with about ⅓rd of the battery bank carrying the entire load. Were they charged to 100% prior to this or did they float for a couple of cycles without reaching 100%?

But that is just guesswork based on the little info available.

They are HA graphs I’ve done the graphs from 12:00 to 01:45 and the consumption hovers between around 780w and 750w so very little variance.

Sorry not sure what you mean by what power metric I used?

Its a Victron Multiplus 2 5kva inverter.

I have 3 x Pylontech US3000C’s, in summer they get up to 100% pretty much everyday and then sit fully charged for 4-6 hours so aren’t being hammered. Battery temperatures range from 28c to 34c and I have cooling fans for my inverter and MPPT and these run at around 28c to 45c.

I’ve gone back about 10 days and it seems like its only had this sharper drop off from around 24% the last 2 early morning loadshedding sessions and I’ve had it set to a lower SOC of 30% about a week ago and it dropped down at a consistent rate even when it got to 24% so its a bit puzzling.

Yes I wasn’t very clear. I wanted to know what you used to obtain the load value. Is it a clamp meter? Does the meter include a power factor? I was thinking that maybe the value shown might not be accurate when there is a power factor involved.
How is the SOC reported? Is this the BMS reporting? does it do some guesswork? How does the system combine the values of 3 batteries? Is it perhaps a smart shunt of your victron system? I would not trust the BMS value, they tend to guess and are not very accurate when it comes to SOC

I’m not using a clamp meter, all of the inverter and even battery data I’m getting from the Victron modbus register and pull this into Home Assistant and then use this for dozens of power management automations.

As to your questions I unfortunately do not know enough about how the Victron gets this data and averages or estimates things to answer them.

I do understand that there is some guess work being done by the inverter kind of like when the batteries sit at 88% for an unusually long amount of time before hitting 100% but on these two occasions the discharge rate has changed and is disproportionate to the load hence my question.

I’ll set the inverter to a higher SOC for the next early morning loadshedding session and see if I can spot some sort of pattern.

Perhaps limit inverter power to 500W or whatever your average-ish draw is over the night and see if the SoC runs down smoothly?

I think _a_a_a has it right, one bank suddlenly drops its SOC due lonq low cell. Maybe track the lowest cell to see this.

Agree it seems the most plausible explanation.

How would I track the cells of the battery? As mentioned I’m pulling all battery data from the Victron modbus register and not getting data from the Pylontech BMS directly.

I think Pylontech has a #modules blocking charge #modules blocking discharge on the GX device? Checking that (not sure if stored on VRM) might help.

I had a look and it doesn’t seem to send that info to the VRM but might be wrong.

Not all batteries report the modules online/offline. I’m not sure what firmware version is needed to see this.