Why Most Indian Plants Still Wash on a Schedule That Ignores Their Dust Curve

Most plants aren't losing energy because nobody cares about cleaning. They're losing it because someone set a schedule years ago and nobody tested it against the plant's actual dust.

October 17, 2026. By News Bureau

Tags:

articles

Most utility-scale plants in India clean their modules on a calendar. Weekly, fortnightly, monthly. Ask where that number came from and it's usually the O&M contract, or the tanker schedule.

Dust doesn't care about the contract.

Soiling builds on a curve, and the shape of that curve depends on the site, the season and what the wind did that week. A fixed schedule gets it wrong in both directions. Some weeks you wash panels that were barely dirty. Other weeks the panels sit dirty right when losses climb fastest.

Every Site has its Own Curve

Published studies and industry discussions report wide ranges for soiling loss in India, depending on region, tilt and O&M practice. I'd treat any national figure as a reason to measure your own plant, not as a planning number.

Arid belts get fine mineral dust that settles steadily and then jumps after a windy day. Plants next to farmland get agricultural dust that follows the crop cycle; harvesting, tilling and stubble burning can leave heavier, stickier deposits. Coastal plants get salt. A salt film behaves nothing like loose dust and often won't come off with a light pass.

Seasons change the picture again. The monsoon does some of the cleaning for you, though it can leave mud streaks behind. Then the dry months arrive and the curve gets steep. A weekly wash that was fine in August can be well behind by March.

Two public case studies from my own company show how uneven this gets. We published them and the sites reported the details, so treat them as field observations, not independent data. At a 334 MW plant on the edge of the Thar in Rajasthan, the write-up reports that soiling was uneven, that downwind rows suffered most, and that performance often dropped before dust was visible to the eye. At a 200 MW plant in Agar, Madhya Pradesh, it describes sharp pre-monsoon soiling swings and fine dust that came back quickly after dry spells. A calendar catches neither.

Measure First, Then Set the Cadence

You don't need fancy equipment to find your dust curve. You need discipline, and one full dry season.

There are two ways to do it. A soiling station logs the ratio between a clean and a soiling reference cell or module. The cheaper route, and the one I'd suggest most plants start with, is a clean versus dirty reference string. Pick two comparable strings on the same inverter, same orientation, no shading difference. Clean one on a strict, frequent routine. Leave the other on your existing schedule, or uncleaned for a set period. Compare their output every day.

Before you start, check that the two strings match when both are clean. Watch for mismatch and inverter or tracker faults. Then keep a daily log. It can be this simple:

Field	What to record
Date and window	Same fixed time window each day, for example 11:00 to 14:00
Irradiance	Plane-of-array reading for the window, if available
Module temperature	Back-of-module or met station value
Clean string energy	kWh for the window
Dirty string energy	kWh for the window
Soiling ratio	Dirty divided by clean, normalised to the clean-state baseline
Days since last clean	For each string
Weather	Rain, dust storm, high wind, dew or fog
Cleaning done	Which string, method, time
Anomalies	Inverter trips, curtailment, shading, string faults
Initials	Who recorded it

Don't skip the weather and anomaly columns. When the data jumps for no obvious reason, that's where you'll find the reason.

A Worked Breakeven Example

The numbers below are illustrative round numbers chosen for easy arithmetic. They are an example only, not Taypro data and not typical values for any site.

Say you have:

A plant producing 500 MWh per day when clean.
Soiling that cuts output by 0.2 percent per day since the last clean. So you lose about 1 MWh on day one, 2 MWh on day two, and so on.
An all-in cost per cleaning cycle (labour, water or energy, supervision, wear) equal to 8 MWh of generation at your tariff.

If soiling builds roughly linearly, the energy you've lost since the last clean grows with the square of the days: about half of (daily output × soiling rate × days²). The cheapest moment to clean is when that lost energy equals the cost of one cycle. Here, 0.5 × 500 × 0.002 × days² = 8. That gives 4 days.

Double the soiling rate to 0.4 percent a day, as can happen in the dry season, and the interval drops under 3 days. Cut it to 0.05 percent in the monsoon and it stretches to 8 days. The interval also moves with the square root of cycle cost, so a cycle four times as expensive only doubles it.

Real life is messier. Soiling is rarely linear, rain resets the curve and no clean is perfect. Use the formula to frame the decision and your log to supply the real rate, month by month.

Where Manual Washing Still Wins

There's no single right method. I'll say this plainly, even though I build cleaning robots: robots are not the answer for every site.

Small rooftops often don't justify automation. If your reference string shows low soiling, occasional cleaning may be all you need. Salt films and bird droppings can need water and a person with a cloth, because a light dry pass won't fully deal with them. Plenty of plants run a mix: a routine method for everyday dust, targeted washing for the stubborn stuff.

Dry, automated cycles make the most sense at utility-scale plants with steady dust, long rows and scarce or costly water. A lower cost per cycle shortens the breakeven interval, so cheap, repeatable cycles let you clean more often. Running them at night means no lost production. Night has its own catch, though. Heavy dew can turn dust into smears, so any automated programme should skip or delay a cycle when modules are wet.

What to Put in Your RFQ

Manual, semi-automatic or robotic, this is what I'd insist on.

Ask for PR by block, not plant averages. Block-level performance ratio before and after cleaning tells you whether the method works where soiling is worst. Averages hide bad blocks.

Pilot on a dirty block. Use one of your worst blocks, next to a reference block, long enough to see the weather change.

Require logs of rows cleaned. Which rows, when, and which cycles were skipped and why. If a provider can't show this, you can't match cleaning to generation.

Don't accept a single uplift percentage. There is no universal figure. Results depend on the site, the dust and the season. Ask how gains will be measured on your plant, against what baseline, over what period.

Check the practical details. Water needs, behaviour in wind and dew, fit with your structures or trackers, and acceptance by your module manufacturer.

The Short Version

Most plants aren't losing energy because nobody cares about cleaning. They're losing it because someone set a schedule years ago and nobody tested it against the plant's actual dust. Measure for one dry season. Then let your own numbers decide how often you clean.

- Yogesh Kudale, Co-founder and CEO, Taypro