Battery target charge level discussion

psiloc

Just wanted to split this out from the solar forecast topic so as not to continue hijacking it :)

@admin I was just curious if you had made a decision regarding changes to the battery charge target calculation?

Last night based on a solar forecast of 9.7kWh my battery was charged to 80%.

According to the spreadsheet with the new formula you provided (https://docs.google.com/spreadsheets/d/1Iz3dzsideWWcNt2ZEcZuRnukAHLXbqVDopYhBfq3Ovg/edit?usp=sharing) it would have charged to 43%, which would have been better and made more sense to me.

I'm not intending to pressure you to make changes to the formula that you're unsure about, I'm just interested to hear what your thoughts or concerns are?

psiloc

Today my battery was fully charged by 12:00, obviously exactly halfway through the day, so a 43% target instead of an 80% target would have worked out perfectly

admin

psiloc See the new formula under it 'With adjustment multiplier for target KWh', which doesn't have diminishing (adjusted) returns for correction factors as the delta nears 0.

https://docs.google.com/spreadsheets/d/1Iz3dzsideWWcNt2ZEcZuRnukAHLXbqVDopYhBfq3Ovg/edit?usp=sharing

We'll apply an initial correction factor of 20% or so, so a multiplier of 1.2 - but will provide a user setting which will default to 1 but can be adjusted to correct from there on.

We're aiming to roll this out today.

SilverArt

admin We're aiming to roll this out today.

Looking forward to seeing the effect it has.

psiloc

Looks amazing! Thanks so much for looking into this.

Both formulas look good to me, there's not really much in it when feeding it the parameters from my setup, so I'll defer to your good judgement there.

I look forward to seeing how it works in the days to come then!

admin

psiloc Thanks for pressing us to look into it! 😁

admin

This has been rolled out now. Advanced settings page is available too.

In addition, we also made a slight tweak for solar yield calculation for low sun angle, typically in winter seasons. Looks like it tended to overestimate mostly.

All these changes are now in v0.3.265.

psiloc

Woo! Looking forward to sharing my results from tonight

psiloc

Looks fantastic guys!

I don't know if it's important, but the spreadsheet formula calculated it at 68% (same multiplier of 1.00). Obviously that difference is barely worth mentioning but I thought I would mention it in case it's not intended.

admin

psiloc Thanks, yeah we did spot one small issue with the formula, and will get a fix out.

Important to note though that you should look at the bottom numbers in the spreadsheet and apply the 1.2 factor - which is an initially fixed constant. But can be further adjusted by the user multiplier. If you find it's over, use 0.9 for instance.

var targetKWhBattery = (batteryKWhCurrent + batteryKWhTargetDelta) * 1.2 * targetKWhUserMultiplier

I would be careful making tweaks every day to your own multiplier though, and gather some more data first.

psiloc

Ah, so there is a hard coded multiplier of 1.2, and then the user's multiplier setting is subsequently applied to that. That makes perfect sense - setting the multiplier to 1.2 in the spreadsheet results in a target of 75%.

Yep I agree and will leave things alone for a while now and only make further tweaks when I have a bigger dataset. It looks to have predicted everything perfectly today though.

Thanks again for your work on this!