solar clipping example

[pookey]

What happened

This is an example to go with #305 - hopefully the debug data shows that clipping.

typically when I'm at home on a sunny day, I turn BESS off and force export till about 2pm, when I turn BESS Back on - this is to avoid clipping. however it would be great if the optimiser was aware of this.

Debug bundle

Please drag-and-drop bess-debug-2026-07-14-215425.md from your Downloads folder into the editor below before submitting.
The file contains your version, settings, recent logs, and the raw HA discovery dump that the maintainers need to reproduce the issue.

bess-debug-2026-07-14-215506.md

[pookey]

The plan this morning looks good, it's exporting all morning to allow for battery capacity later to avoid clipping.

bess-debug-2026-07-15-090409.md

[pookey]

Apparently I've reached a level of laziness now where I just ask the robots to post GH comments.....

Anyway, the below is a summary - today it performed nearly perfectly, which is significantly better than a whole day of clipping. I'm quite happy with 'nearly perfect'! I'll see how it performs over a few more sunny days.

BELOW IS ALL ROBOT 🤖

End-of-day follow-up (2026-07-15) — first full sunny day running the PR #305 feature live (fork build 9.8.1.1 = upstream 9.8.1 + inverter AC cap, inverter_ac_capacity_kw: 5.0, margin 0.05, Growatt MIN 5000TL-XH, 10 kWh battery).

TL;DR: zero clipped solar on a 40.5 kWh day whose DC peak hit 7.2 kW against the 5 kW inverter — and the optimizer reproduced, on its own, exactly the manual routine this issue describes (force-export all morning, then soak up the midday peak).

What the schedule did

• 00:00–11:00 — battery deliberately run down from 6.1 → 1.0 kWh (export + covering house load), creating ~9 kWh of headroom ahead of the forecast above-cap peak.
• Late morning — the new HOLD behaviour is visible in the data: repeated SOLAR_EXPORT periods with a near-empty battery (SOE ~1.0–1.3 kWh), exporting surplus instead of passively charging. Pre-feat: inverter AC output cap — solar-clipping-aware optimization #305 this state didn't exist — load_first would have been quietly filling the battery all morning.
• 12:15–15:00 — SOLAR_STORAGE across the above-cap window: battery charged 1.3 → 10.0 kWh, absorbing the DC production the AC stage couldn't deliver. Peak solar was 7.2 kW at 13:00.
• 15:00 — battery reached 100% just as the peak subsided, then held for the evening; discharge now covering the house through the 0.30–0.42 GBP/kWh evening periods.

Per-period clippedSolar is 0.00 kWh for every slot of the day.

The peak window (actuals, 15-min slots)

Time	Solar (kW)	To battery (kWh)	SOE (kWh)	Intent
12:15	5.6	0.8	1.3 → 2.1	SOLAR_STORAGE
12:30	5.6	1.0	2.1 → 3.1	SOLAR_STORAGE
12:45	6.4	1.2	3.1 → 4.4	SOLAR_STORAGE
13:00	7.2	1.4	4.4 → 5.7	SOLAR_STORAGE
13:30	6.0	0.7	6.2 → 6.9	SOLAR_STORAGE
13:45	6.4	1.2	6.9 → 8.1	SOLAR_STORAGE
14:00	6.4	1.1	8.1 → 9.1	SOLAR_STORAGE
15:00	5.2	0.0	9.7 → 10.0	(full)

Summing the above-cap slots: 2.65 kWh of above-cap DC went into the battery instead of being lost; had the battery been full through the peak (the old failure mode), ~3.5 kWh would have clipped.

Day totals

• Solar: 40.5 kWh, house: 24.4 kWh, exported: 17.1 kWh, imported: 0.8 kWh
• Battery: 9.0 kWh charged (99% from solar) / 9.3 kWh discharged, 0.9 cycles
• Clipped solar: 0.0 kWh

One nuance worth noting: the battery filled at 15:00 while solar stayed marginally above the cap (5.2–5.6 kW) until ~16:15, but house load plus export absorbed it — measured flows show no clipping in those slots either. The model's 5% margin (planning against 4.75 kW) is presumably what pulled the fill slightly ahead of the true tail of the peak.

This was previously a manual daily chore (BESS off + force export until ~14:00). Today it was fully hands-off. 🎉

[pookey]

Closing as #305 will fix.