Why your solar panels produce less than their rating
A 585 W panel will, on an ordinary sunny day, deliver something closer to 420 W. Nobody is lying to you. The rating describes a laboratory test your roof almost never reproduces, and the gap between the two is made up of eight losses that are individually small and collectively large.
What the rating actually means
Panel ratings are measured at Standard Test Conditions: irradiance of 1,000 W/m², a cell temperature of exactly 25 °C, and a defined atmospheric spectrum. The contradiction is built in — anywhere receiving 1,000 W/m² of sunlight is heating the panel far above 25 °C. STC is a bench comparison standard, not a prediction.
Some datasheets also quote NOCT or NMOT figures, measured at 800 W/m² and 20 °C ambient with a 1 m/s breeze. Those are much closer to reality and typically land 25–28% below the STC number. If a datasheet shows both, the NOCT figure is the more honest one.
Temperature — usually the biggest single loss
Crystalline silicon loses output as it heats. The coefficient is on every datasheet, written as something like
-0.34 %/°C, and it applies to every degree above 25 °C.
A reasonable rule is that a roof-mounted panel in full sun runs about 25 °C above ambient. So:
| Air temperature | Approx. cell temperature | Loss at -0.4 %/°C |
|---|---|---|
| 15 °C | 40 °C | 6% |
| 25 °C | 50 °C | 10% |
| 35 °C | 60 °C | 14% |
| 45 °C | 70 °C | 18% |
This is why a cool, bright spring day often out-produces a blazing July afternoon, and why hot countries see lower specific yields than their irradiance alone would suggest.
What you can do: leave an air gap. Panels flush against a hot roof with no ventilation run 10–15 °C hotter than panels on a frame with 10 cm of clearance. That gap is worth several percent a year and costs almost nothing at installation. Panels with a lower temperature coefficient (-0.29 %/°C rather than -0.40) are genuinely worth a small premium in hot climates.
Soiling — the cheapest loss to fix
Dust, sand, pollen, bird droppings, salt spray and industrial film all block light. Datasheets assume 2%. Reality in a dry, dusty region with infrequent rain is 10–25% between washes, and the curve is steep in the first few weeks after cleaning.
Bird droppings and leaf litter are worse than their area suggests, because a fully opaque patch on one cell can shut down an entire sub-string through the bypass diodes. A spread of fine dust costs you proportionally; a single hard shadow costs you disproportionately.
What you can do: wash monthly in dusty regions, quarterly elsewhere. Early morning with plain water and a soft brush — never cold water on hot glass, and never a pressure washer against the seals. In most dusty climates a wash schedule returns more energy per unit spent than any equipment upgrade.
Shading — small shadow, large consequence
Panels in a string are wired in series, so the current through all of them is set by the weakest one. Shade one panel and, without mitigation, you pull the whole string down towards its output. Bypass diodes limit the damage to a third of a panel at a time, but the effect is still far out of proportion to the shaded area.
The usual culprits are not trees. They are water tanks, parapet walls, a neighbour's new floor, satellite dishes, chimneys and the array's own rows shading each other on a flat roof in winter.
What you can do: walk the roof at 9 am and 4 pm in winter, not at noon in summer — winter is when the sun is lowest and shadows are longest. Where shade is unavoidable, per-panel optimisers or microinverters confine the loss to the affected panel. Group panels that share a shade pattern onto the same string.
Inverter conversion
Turning DC into AC costs 2–5%. Modern inverters peak above 97% efficiency, but peak efficiency occurs somewhere in the middle of their range — a grossly oversized inverter spends its mornings and evenings running inefficiently at very low load. This is another quiet argument for a DC/AC ratio above 1.
Wiring and mismatch
DC cable runs lose 1–3% to resistance; the AC side loses another 1%. Long runs from a far corner of a flat roof to a distant inverter are where this goes wrong. The fix is thicker cable or a shorter run, and it is far cheaper to decide at installation than to retrofit.
Mismatch accounts for another 1–3%. No two panels are identical even from the same batch, and a string performs like its weakest member. It grows over time as panels age at slightly different rates.
Degradation
Panels lose output permanently as they age: typically 1–3% in the first year as the cells stabilise, then 0.4–0.6% a year after that. A 25-year warranty usually guarantees 80–87% of the original rating at year 25. This is slow, predictable and already accounted for in any payback calculation worth trusting — including the one on this site.
Adding it up
| Loss | Good install | Typical hot, dusty install |
|---|---|---|
| Temperature | 8% | 15% |
| Soiling | 2% | 8% |
| Shading | 0% | 3% |
| Inverter | 3% | 4% |
| DC + AC wiring | 2% | 3% |
| Mismatch | 1% | 2% |
| Other (spectrum, reflection, downtime) | 2% | 3% |
| Performance ratio | ~0.83 | ~0.67 |
That spread — 0.83 against 0.67 — is a 24% difference in annual generation between two installations with identical panels on identical roofs. Nearly all of it comes down to decisions made at installation and a cleaning schedule.
Is my system underperforming, or is this just normal?
Compare your specific yield: annual kWh divided by installed kWp. Then look for a pattern:
- Down 5–10% across the board, gradually worsening — soiling. Wash it and watch the curve.
- A clean notch at the same time each day — shading. Something is casting a shadow; go and find it.
- A flat ceiling on sunny days — inverter clipping. Expected if your DC/AC ratio is high.
- One string consistently below its twin — a fault, a bad connection, or a failed bypass diode. Call the installer.
- Worst in summer, best in spring — temperature. Normal, and worse if the panels have no air gap.
Run your coordinates through the calculator to get the specific yield your location should support, then compare it against what your inverter app is reporting. A gap of 5% is noise. A gap of 25% is a problem worth chasing.