batpred Predbat plan using PV values significantly below forecast

Describe the bug For tomorrow, my solar PV energy output is predicted to be 40 kWh with little variance between the various Solcast forecasts. The graph below shows sensor.predbat_pv_today and sensor.predbat_pv_tomorrow:

However, Predbat's plan doesn't seem to reflect this:

The "PV kWh" column totals only 24.4 kWh
The 10% values shown for PV kWh with the Plan debug on are higher, at 30.1 kWh
The peak for either PV value is 3.37 kWh at 11:30, when Solcast's pv_estimate is 4.693 kWh
If the PV values are meant to represent "PV in excess of load", meaning that the load should be added to give the total PV output, they are still significantly lower than the forecast (but I don't believe this is how Predbat represents the PV values)

I have disabled switch.predbat_metric_pv_calibration_enable in case there was some strange calibration data, with no obvious impact.

I have inspected the Solcast data in case it was somehow mangled by the 429 issues today, but it appears to be crrect.

Expected behaviour

PredBat's solar energy total is within the range of the Solcast forecasts
The 10% PV output is more pessimistic than the 50% PV output

Predbat version

v8.27.23

Environment details

SigEnergy inverter @ 12 kW max, battery @ 27 kWh with max charge rate 12 kW, 15.5 kWp solar
Standard HAOS install

Screenshots Predbat plan PV values:

Log file predbat(1).log

Predbat debug yaml file

predbat_debug_22_10_00.yaml.txt

Predbat plan

predbat_plan(1).html

Solcast debug output

config_entry-solcast_solar-01K9ZG3D7GWP1HEX1GCEBRYGEC.json

Nov 16 '25 22:11 DFEvans

Your logfile shows that predbat is getting the right forecast data from solcast integration:

2025-11-16 22:02:35.716591: Using Solcast integration from inside HA for solar forecast
2025-11-16 22:02:35.716728: PV Data for pv_forecast_today total 10.02 kWh
2025-11-16 22:02:35.716786: PV Data for pv_forecast_tomorrow total 40.69 kWh
2025-11-16 22:02:35.716834: PV Data for pv_forecast_d3 total 9.41 kWh
2025-11-16 22:02:35.716892: PV Data for pv_forecast_d4 total 14.9 kWh
2025-11-16 22:02:35.716925: PV Forecast today adds up to 20.03 kWh and total sensors add up to 10.02 kWh, factor is 2.0
2025-11-16 22:02:35.741243: PV Calibration: Fetching PV data for calibration
2025-11-16 22:02:35.811100: PV Calibration: Past day 3 had 0.0 kWh of PV forecast data and actual 0.0 kWh
2025-11-16 22:02:35.811175: PV Calibration: Past day 2 had 3.77 kWh of PV forecast data and actual 0.7 kWh
2025-11-16 22:02:35.811200: PV Calibration: Past day 1 had 3.46 kWh of PV forecast data and actual 3.71 kWh
2025-11-16 22:02:35.811223: PV Calibration: Worst day scaling factor 0.5 best day scaling factor 1.07
2025-11-16 22:02:35.811518: PV Calibration: PV production: 4.41 kWh, Total forecast: 7.22 kWh adjustment 0.6098x slot adjustments {0: 1.0, 30: 1.0, 60: 1.0, 90: 1.0, 120: 1.0, 150: 1.0, 180: 1.0, 210: 1.0, 240: 1.0, 270: 1.0, 300: 1.0, 330: 1.0, 360: 1.0, 390: 1.0, 420: 1.0, 450: 1.0, 480: 0.5, 510: 0.5, 540: 0.7203362431296475, 570: 0.9097560975609756, 600: 0.7572692793931732, 630: 0.5, 660: 0.5, 690: 0.6566523605150214, 720: 0.5, 750: 0.5, 780: 0.5, 810: 0.62808671065033, 840: 0.6432523454414242, 870: 0.5079479768786127, 900: 0.5158878504672897, 930: 1.1238223418573352, 960: 2.0, 990: 1.0, 1020: 1.0, 1050: 1.0, 1080: 1.0, 1110: 1.0, 1140: 1.0, 1170: 1.0, 1200: 1.0, 1230: 1.0, 1260: 1.0, 1290: 1.0, 1320: 1.0, 1350: 1.0, 1380: 1.0, 1410: 1.0} max_kwh 9999 divide_by 2.0
2025-11-16 22:02:35.823370: PV Calibration: Using calibrated PV data
2025-11-16 22:02:35.831180: PV Forecast for today is 10.02 (0.97 10% 40.91 90% 5.89 calib) kWh and left today is 0.0 (0.0 10% 0.0 90% 0.0 calib) kWh
2025-11-16 22:02:35.864070: PV Forecast for day tomorrow is 40.69 (30.11 10% 44.78 90% 24.46 CL) kWh
2025-11-16 22:02:35.881925: PV Forecast for day d2 is 9.41 (3.56 10% 25.22 90% 5.66 CL) kWh
2025-11-16 22:02:35.901659: PV Forecast for day d3 is 14.9 (2.51 10% 42.98 90% 8.77 CL) kWh

I'm running an older version of Predbat, 8.25.14, and am not seeing this behaviour. For tomorrow, solcast is giving me a forecast of 17.2kWh, and sensor.predbat_pv_tomorrow is giving 17,17kWh with similarly aligned PV10 values

Looks like it is a problem in a newer version of predbat, can you revert to an older version to confirm this?

Nov 16 '25 22:11 gcoan

Having come back to it this morning, the solar values are looking much more sensible - so sadly I'm unable to confirm if downgrading version would have fixed it, but it's at least not always broken if it is a new bug.

I'll keep an eye on it and see if it reoccurs, and if so, check whether a downgrade fixes it (and a re-upgrade causes it to return).

My assumption when trying to debug was that the re-plan that occurs after changing the config should reflect that new config immediately, but let me know if it's better to do a full Predbat restart or even a HA restart in such cases.

Nov 17 '25 08:11 DFEvans

My assumption when trying to debug was that the re-plan that occurs after changing the config should reflect that new config immediately, but let me know if it's better to do a full Predbat restart or even a HA restart in such cases

predbat should replan every 10 minutes or if the config is changed

and per earlier comment, PV is not net of load, it is adjusted for PV10/50 weighting, but its raw PV you should see in the sensor/plan

Nov 17 '25 08:11 gcoan

To keep the thread updated, I have not seen this issue reoccur in the last few days, having stayed on the newer Predbat version. My current gut feeling is that it was some stale data somewhere in the system that came out of the Solcast issues.

Feel free to close if you'd like; I'll keep an eye on it.

Nov 19 '25 18:11 DFEvans