Why measure soiling?
After sunlight itself, soiling is the biggest single factor in PV yield (IEA PVPS 2022). Yet production data can’t show it. SCADA records output, and a dip in output could be dust, cloud, clipping or a fault. A soiling measurement separates dust from everything else. It tells you how fast your site soils, how much a clean recovers, and when the energy you’re losing is worth more than the next wash.
It is also part of the monitoring standard. IEC 61724-1:2021 requires soiling measurement for Class A monitoring systems (IEA PVPS 2025).
How soiling is measured
| Method | How it works | Typical uncertainty |
|---|---|---|
| Power-based reference pair | Two matched modules side by side, one cleaned on a schedule and one left to soil; compares maximum power | 1–2% |
| Current-based reference pair | The same pair, compared on short-circuit current | 3–5% |
| Reference-cell pair | Two small reference cells, one cleaned and one soiled | 4–7% |
| Optical sensor | Estimates the light lost to dust on a small glass window; no cleaned reference needed | 4–7% in 2018 reviews; DustIQ’s maker states 1% + 0.1 × reading after site calibration |
The best set-ups reach about ±1% (IEA PVPS 2022, citing Dunn et al. 2018); the DustIQ figure is from its 2025 manual. The right choice depends on the decision you need to make, your monitoring class and your budget.
Reference pairs, the classic soiling station, need one discipline: the clean device must really be cleaned, on schedule. If it isn’t, the ratio drifts towards 1 and hides the loss. Optical sensors skip the cleaned reference, but their readings still need checking against the modules around them.
The numbers that matter
Three terms carry the subject; soiling ratio explained covers them in more depth.
- Soiling ratio is soiled output divided by clean output. Soiling loss is 1 minus the ratio, so a ratio of 0.97 means a 3% loss (IEA PVPS 2022).
- Soiling rate is how fast the ratio falls, usually per day. It is fitted over a dry spell of at least 14 days without rain or cleaning (IEA PVPS 2022).
- Cleaning trigger is the point at which you clean. We prefer a cost-based trigger, set in your cleaning strategy.
What we supply and do
How many sensors, and where? IEA PVPS (2022) recommends more than one soiling sensor on plants larger than 5 MW that expect more than 2% soiling loss a year, because soiling can vary 1.5–2× across a single site. We place sensors where the soiling differs: downwind of a dust source, near roads, and in the plant’s cleanest block for contrast. Each sensor sits in the plane of the modules, on the tracker if the plant tracks.
- Choose the method for your plant size, monitoring class and budget.
- Supply the hardware: soiling sensors and soiling stations.
- Install and commission them, and set the cleaning schedule for any reference device.
- Check the data and fit soiling rates through each dry spell.
- Turn it into action: soiling ratio, soiling rate and cleaning triggers your O&M team can act on, adjusted with them as the seasons change.
What field trials show
Chemitek’s published case studies show what a soiling measurement adds, and what a trial looks like without one.
- Without soiling meters, the soiling is judged by eye. The Vietnam and Santiago Solar cases compared coated strings with water-cleaned strings on energy, but neither plant had soiling meters, so the soiling itself was only assessed visually.
- One position doesn’t speak for the whole site. At a Scatec-operated plant in Honduras, the test string sat nearer a road and collected much more dust than the two control strings (Case study: Chemitek (Portugal), with Scatec). Sensors belong where the soiling differs. See the Honduras case.
- An instrumented trial gives the clearest answer. At Green Energy Park in Morocco, five identical strings each had their own inverter and DC monitoring, with a meteorological station on site. By the end of the dry season, the uncleaned string had fallen 6.94% behind the water-cleaned one (Case study: Chemitek (Portugal), with Green Energy Park). See the Morocco trial.
A soiling sensor adds what an energy comparison can’t: how quickly the glass soils again after each clean, which is what sets the cleaning interval.
