PV underperformance: causes and proof

Important: One “bad month” is not underperformance. What matters is a clean classification including weather, availability and plausible expected values.

What is underperformance?

Underperformance means the system repeatedly falls below plausible expected values over a relevant period. Expected values can be: yield forecast, year-on-year comparison, reference plants or model-based expectations (with weather correction).

Typical causes (that are actually common)

• Shading / soiling: temporary, seasonal or permanent (trees, structures, leaves).
• String/design errors: wrong string lengths, MPP range, mixed module fields.
• Inverter/MPP tracker: misconfiguration, failures, derating, temperature issues.
• Degradation / module faults: PID, hotspots, microcracks, LID—depending on module type and environment.
• DC/AC losses: bad connectors, contact resistance, wrong cable routing.
• Availability: faults, shutdowns, fuses, grid issues.

What data do you need for a first classification?

• Yield (daily/monthly), ideally per inverter/tracker
• Faults/error logs (inverter)
• System data (kWp, orientation/tilt, module/inverter type)
• String plan/layout (if available)
• Info on shading/site changes

Systematic approach (saves time and money)

1. Plausibility check: Compare weather, availability, periods.
2. Segment: Which inverters/trackers/strings drop off?
3. Build hypotheses: e.g., shading vs design vs defect.
4. On-site check: Visual inspection (cables, connectors, module fields, hotspot indicators).
5. Measurement/verification: Targeted electrical tests where data is abnormal.
6. Assessment: Cause + yield impact + action plan.

When does it become provable?

When you don’t just show deviations, but explain them technically and substantiate them traceably. That matters for warranty, insurance or disputes.

FAQ

From when do we speak of underperformance?

When yields repeatedly fall below plausible expected values over a relevant period and weather/availability have been accounted for.

What data is needed for an initial assessment?

Yield data, system data, inverters/strings, site, and information on shading/faults.

Can monitoring alone prove the cause?

Monitoring shows deviation, but doesn’t always prove the cause. Robust conclusions often require document review and measurements.