search

springfall2008 / batpred

Home battery prediction and charging automation for Home Assistant, supporting many inverter types
https://springfall2008.github.io/batpred/
114 stars 39 forks source link

Best SOC Keep Not Being Maintained #907

Open nbullus opened 5 months ago

nbullus commented 5 months ago

Describe the bug The Best SOC Keep setting is meant to keep the battery SOC above the user specified threshold (1.4kWh - 10% SOC in my example). On numerous occasions I have seen the SOC drop below this threshold and down as low as my 5% SOC Reserve. In the attached example plan, the SOC is forecast to drop below my 10% threshold between 22:00 and 23:30.

Expected behavior Predbat should not intentionally plan to take the SOC below the Best SOC Keep threshold in future slots. The SOC should only drop below this threshold in the current slot if either the demand is higher than forecast and/or solar is lower than forecast.

If the SOC does drop below the threshold due to unexpected changes in load/PV, it should restore the battery level back to the Best SOC Keep at the next Charge slot opportunity.

In the attached screenshot the Discharge Slot at 16:00 should have discharged down to 30% rather then 24% to prevent the SOC forecast to drop to 6% later in the plan.

Predbat version

7.6.10

Environment details

Screenshots

image

image

Log file

Predbat Log.txt

gcoan commented 5 months ago

Worth you having a look at the documentation for best_soc_keep and best_soc_min https://springfall2008.github.io/batpred/customisation/#battery-margins-and-metrics-options

best_soc_keep is not a hard battery minimum level, as explained in the documentation its ignored for the forthcoming 4 hours of the plan, I think this might be what's happening.

I am personally seeing predbat doing extra discharges a bit too early (on the fixed agile 15p outgoing tariff) that result in the Soc reducing to empty and unprofitable imports #861. This might be a different symptom of this.

nbullus commented 5 months ago

The issue is not with slots in the next 4 hours, but slots which are 20+ hours away.

If it plans to idle the SOC down to 6% based on the forecast demand, and the demand ends up increasing in this period, it would start pulling from grid instead of using the extra 5% SOC which was meant to be kept for this very purpose.