As per the title - someone has screwed up the Peukert exponent algorithm in the BMV 712.
I'll state from the outset - I do not need a lecture on what the exponent means. I know what it means. I've been using it with my bank of Trojans at my present home for nearly 15 years, and my previous cabin for 20 years before that. I've used many other monitors tthat accurately de-rate the bank capacity based on instantaneous load. I know how this should work.
And the new Victron equipment I have paid thousands for in my recent RV build, does not.
For context, I have LFP bank rated at 400Ah @ C20, 385 @ C8. Both of these I have personally tested at STC and used the C20 value as the baseline capacity in the BMV settings, along with the corresponding Pe of 1.04, . The typical, expected behaviour then, would be that the capacity used for deriving instantaneous relative SoC should adjust under load to match these values.
Which is to say:
- With a 20A load (corresponding to a C20 discharge), the SoC should be derived on a capacity of 400Ah. So if 100Ah is discharged, and a 20a load present, this should result in a 75% SoC displayed and a 15hr runtime, based on the C20 rating of 400Ah. ( [400-100]/400 = .75 ).
- Equally, a 48.13A load (for a C8 disharge) when applied at the same 100ah discharged volume should result in a 74% SoC based on the C8 capacity ( 385-100 / 385 = .74), and obviously in-between values should be extrapolated
However, this is not what occurs.
Despite the manual explicitly stating the SoC value " is compensated for both the Peukert efficiency and charge efficiency.", I have never seen the SoC change with variation in instantaneous load. Even with changes in load as as high as 1C, the SoC remains the same (save for decreasing with Ah consumed as time progresses). This is not correct behaviour - relative SoC should be immediately lowered based on instantaneous load and Peukert exponent, to reflect the reduced total capacity available under heavy loads.
The "time remaining" value DOES adjust - but often incorrectly, either over or under-estimating with no pattern I can make sense of . Time to go averaging is set to 0m, and I am testing with constant loads, so this cannot be put down to variance in averaging calculations.
Every more bizarre, in the process of trying to troubleshoot this, I noticed that changing Pe in fact has the OPPOSITE to intended effect! For example, increasing the Pe to the max 1.5 value (simulating a high resistance, 'lossy' bank) somehow results in the SoC INCREASING!! This occurs even when under a load greater that the C20 rate, so it doesn't make sense as 'overrating' the explaining it as overrating capacity for loads lower than C20 (eg extrapolating to the C100 capacity). Indeed, when actually reducing the load to something like C100 (or increasing it further, for that matter) the SoC still remains exactly the same - again indicating no actual capacity adjustment is occurring based on load/Pe at that moment.
Conversely, lowering the Pe to 1 - for an 'ideal' bank that should simply used a fixed C20 400Ah as the reference regardless of load, somehow lowers the SoC! It's as if the capacity of the battery now HAS been shrunk, despite explicitly applying a Pe value that should disable this! This results in bizarre situations where in one instance, 47Ah taken from the 400Ah total capacity was shown somehow as 57% SoC remaining.
I have tried resetting the unit and SoC from known full charge to no change. This behaviour is totally at odds with not only the (very limited and simplistic) explanation provided in the documentation, but also with the very principle of how the exponent should work. It also appears this bug has been present for some YEARS:
https://community.victronenergy.com/questions/73478/smartshunt-and-peukert-coefficent-doubts.html
https://diysolarforum.com/threads/peukert-exponent-vs-soc.54769/
This is utterly unacceptable. I've already had to return both an Orion and a Smartsolar under warranty due to hardware/design defects (a whole other issue I wont get into in this post) and now I find sloppy programming has crippled basic monitoring functions that you'd find on a $50 ebay monitor.
The vendor has told me 'take it up with Victron' and refuses to warranty it with the outright incorrect assertion that "Lithium batteries don't have a Peukert thingy". But apparently Victron doesn't have any sort of actual technical help line (a huge letdown in its own right), which leaves me no choice but to air it publicly.
This needs to be fixed immediately. I didn't pay for this sort of sloppy execution, and frankly I've just about had it with this entire dogs breakfast of a company.
PS - I would provide images of these bizzare results, but apparently this forum is as buggy as your products and any attempt to upload any image of any kinds simply spits a "Parsing Response Failed" error. Well Done.