1Loss waterfall — where the energy went
From theoretical (measured POA × capacity) down to actual export. Green-bordered = recoverable; grey = structural/inherent.
reference / subtotal
recoverable loss
structural loss
temperature gain / actual
2Daily energy & performance ratio
Bars = daily export (MWh). Line = PR (%, POA-referenced). Red bars = outage; amber = sensor glitch; dim = weather-limited.
3Inverter imbalance
6 inverters, nominally equal. The “.1” units lag badly → DC-side problem.
4Tracker fleet (173 TCUs)
Each square = one P4Q tracker controller, coloured by mean elevation deviation.
OK <2°
2–10°
10–30°
>30° (stuck)
5MVPS / combiner-box (DC input) status
3 inverter stations, 2 inverters + 13 combiner boxes each. Boxes coloured by measured DC-input current on clear days. This is the disconnected-CB view requested.
connected / normal
degraded (low current)
disconnected (no current)
6Plant single-line — where the faults are
7The May 17–19 outage
~2-day total stop. May 18 logged zero data all day (full SCADA/plant blackout). Root cause not in telemetry — no May O&M report.
8Cross-validation & PR method
Independent checks (all passed)
| Check | Result |
|---|
| VCOM export vs GPM meter (May 30/31) | identical |
| Meter vs Σ inverters (month) | +0.03% |
| Stuck trackers: May VCOM ∩ April SCADA | |
| Grid clipping in May | none (peak 6.35<9 MW) |
PR methodology
PR = Egrid / [(GPOA/1000) × Pnom], Pnom=11,048.7 kWp,
irradiance = POA (G_M0 = POA CIT2). Temp-corrected with β=−0.34 %/°C (CS7N-655MB-AG),
Tmod from sensor.
Measured PR ≈ 65% vs design budget ≈ 89% → ~24 pt real underperformance.
Cold month (+5.2% temp gain) means a healthy plant should exceed ~88%.
Caveat: POA CIT2 vs CIT3 diverge 25–47%; CIT3 implies PR>100% (invalid) so CIT2 used.
GHI sensors read ~10× low. No plant PR report exists; platform PR channels were dead.
9Fix list — recoverable vs structural
| # | Issue | ~MWh/mo | Type | Action |
|---|
| 1 | DC disconnection + inverter imbalance (MVPS1.1/2.1/3.2; MVPS2 CBs) | ~130 | recoverable | Reconnect/repair CBs & strings on weak inverters |
| 2 | Tracker fleet — 58/173 misaligned, ~31 stuck | ~70 | recoverable | Mechanical tracker repair (motors/gearboxes) |
| 3 | Outage May 17–19 | ~40 | recoverable | Find root cause; protection/availability |
| 4 | Intrinsic system losses (~12%) | ~115 | structural | Inherent design |
| 5 | Soiling | n/a | mixed | Install soiling sensor; cleaning cadence |
≈190–240 MWh/month recoverable (≈30–38% uplift on May output) — fixing trackers, reconnecting DC sections, and resolving the outage, independent of weather.