Category: Guides

The short answer

Yes, but the water quality decides whether it helps or hurts. Below roughly 10 parts per million (ppm) total dissolved solids (TDS), water evaporates off panel glass without leaving a visible residue. US municipal tap water commonly tests at 100–500 ppm — the EPA’s own secondary (aesthetic) guideline caps recommended levels at 500 ppm — and European municipal supplies vary by local water hardness in a broadly similar range. That is high enough that a handful of wash cycles will leave mineral spots, and high enough that repeated cycles turn those spots into bonded scale that is harder to remove than the dust you were trying to clean off in the first place.

Whether tap water is safe to use on your site is a question you answer with a TDS meter, not a blanket rule. The rest of this article covers the threshold, why the damage compounds, and what to do about it if your water tests too high.

This article covers water quality only. For how often to clean by climate, see our companion guide, How Often Should You Clean Solar Panels? Water quality decides whether each individual wash helps or hurts; frequency decides how many times you do it. The two questions compound — using untreated hard water on a site that needs quarterly cleaning does four times the damage of the same water on a site that needs annual cleaning.

1. What TDS measures, and why it is the number that matters

Total dissolved solids is exactly what it sounds like: the combined mass of everything dissolved in the water — calcium, magnesium, silica, sodium, bicarbonates, and trace minerals — expressed in parts per million. It is a broader measure than “hardness,” which refers specifically to calcium and magnesium. A water sample can be low in hardness and still carry enough dissolved silica or sodium to spot glass on evaporation, which is why TDS, not hardness, is the number to track for cleaning water.

TDS is cheap to check. A handheld TDS meter (the same type used in aquarium and drinking-water testing) gives a reading in seconds and is standard equipment for any crew doing water-fed cleaning. Testing your actual supply matters because TDS varies significantly by region and even by season within the same municipal system — a published national average tells you little about the water coming out of your specific tap.

2. The TDS thresholds that determine whether tap water will spot

TDS reading	What happens as the water dries on glass
0–5 ppm	Spot-free in virtually all conditions, including direct sun
6–10 ppm	Spot-free in shade or moderate humidity; risk rises in direct sun
10–20 ppm	Visible mineral spotting likely
20+ ppm	Spotting close to certain — panels typically need to be wiped dry by hand to avoid marks

This 10 ppm figure originates in the spot-free rinse industry (window cleaning, vehicle detailing) rather than a solar-specific laboratory standard, but solar water-quality guidance cites the same threshold, and the underlying chemistry is identical — dissolved minerals left behind when water evaporates. Treat it as a well-corroborated benchmark, not a hard physical law: your own site’s outcome depends on your water’s specific mineral mix, glass temperature, and drying speed, which is why testing your own supply is worth the five minutes it takes.

3. Why hard water scale is worse than ordinary dust

Dust is loose and mechanical — it sits on the glass and, in principle, wipes or rinses off. Mineral scale is different: it forms when the water itself changes state. As a droplet evaporates, everything that was dissolved in it stays behind as a solid deposit, chemically bonded to the glass rather than resting on top of it.

Dissolved mineral	Deposit it forms	Why it matters
Calcium (as calcium carbonate)	Chalky white scale	Builds up in layers with repeated wetting and drying
Magnesium (as magnesium hydroxide)	Fine white film	Often co-deposits with calcium; contributes to haze
Calcium sulfate	Denser, harder scale	More resistant to plain water or mild rinsing than carbonate scale
Dissolved silica	Glassy, tenacious film	The hardest of the four to remove; resists dilute acid and typical descalers

The practical consequence is a compounding cycle rather than a one-time mark: a roughened, mineral-etched surface holds the next layer of dust and the next round of mineral residue more readily than smooth glass, so a panel that has been washed repeatedly with untreated hard water tends to soil faster between cleanings, not just look worse after each one.

We have not found a published, solar-specific timeline for how quickly light spotting becomes bonded scale — it depends on cycle count, UV exposure, glass temperature, and the specific mineral mix in your water, and we would rather say that plainly than invent a number. What is well documented is the direction: each wash with untreated hard water adds another mineral layer, and each layer is harder to remove than the last.

A factory-applied anti-reflective (AR) coating cannot be re-applied in the field once it is stripped or etched, which is why aggressive removal — scraping, strong acid, abrasive pads — is a last resort rather than routine maintenance. The lower-cost fix is upstream: treat the water so the deposit does not form, rather than remove it after it has bonded.

4. Comparing water treatment options: filtration, softening, RO, and DI

Four methods show up repeatedly in cleaning-water discussions, and they are not interchangeable. The most common misconception is that a water softener solves the spotting problem — it does not.

Method	What it actually removes	Stops spotting?
Sediment / carbon filtration	Particulates, chlorine, organics	No — dissolved minerals pass through unchanged
Water softener (ion exchange)	Swaps calcium and magnesium for sodium; does not remove dissolved solids	No — TDS stays roughly the same, so sodium salts still spot on evaporation
Reverse osmosis (RO)	Rejects roughly 90–99% of dissolved solids	Yes, for most source water
Deionization (DI), used after RO	Residual trace ions RO leaves behind	Yes — the most reliable spot-free result, best used as a polishing stage rather than a first pass

If your site already runs softened water for other purposes — irrigation, equipment cooling — that does not solve the cleaning-water problem. Softening protects plumbing from scale by swapping ion type; it does not lower total dissolved solids, so softened water still leaves spots. RO is the method that actually cuts TDS at volume, and DI is worth adding afterward as a final polish rather than running raw tap water through a DI resin bed directly, which exhausts the resin far faster than feeding it pre-treated water.

For sites that need volume, our A40 mobile RO system is built for this: 300–500 L/h output, rated for inlet water up to 1,000 ppm TDS and 10 NTU turbidity. The product page does not publish an output TDS figure, so treat the 90–99% reduction above as the general RO benchmark rather than an A40-specific guarantee, and verify against your own feed water if you need a precise number.

RO membranes and DI resin are consumables — their service life depends on your feed water chemistry and throughput volume, not a fixed calendar interval. We are not publishing a replacement schedule here because we have not confirmed one against your specific water quality; check manufacturer guidance for your actual TDS and turbidity readings.

5. Water use: metered spray vs. hose washing

Treating water is only worth it if the treated water actually reaches the glass instead of running off during an open hose-down. This is where equipment matters as much as water chemistry.

Method	Typical water use	Basis
Manual hose / pressure washing	Roughly 0.1–1.2 gal (0.4–4.5 L) per panel per wash	Published estimates vary widely by technique and panel size — treat as an order-of-magnitude range, not a precise figure
R1 crawler robot, metered spray	Up to 1.84 L/min at the nozzle, calibrated per pass	Manufacturer spec — a per-minute rate, not directly convertible to per-panel use without site-specific dwell time

Those two figures are not on the same basis — one is volume per panel, the other is flow rate per minute — and we are not going to force a conversion that would require assumptions about panel size and dwell time we have not verified. What the comparison does establish directly: hose washing sprays continuously and lets runoff carry most of that water off the array unused, while a metered crawler system doses water at a fixed rate as it passes, so a much higher share of what comes out of the nozzle actually contacts the glass.

Vendor case studies in the robotic-cleaning industry commonly report 75–90% lower total water use for metered/automated systems versus manual hose washing. We flag that range as vendor-reported rather than independently audited — treat it as directional, not a guarantee for your site. The practical implication for water treatment economics is straightforward either way: the less water you spray and lose to runoff, the less treated water your RO/DI system has to produce per panel, which is why treatment and metered application are usually worth planning together rather than separately. See the R2 crawler robot for the AC-powered, continuous-duty version of the same metered-spray approach.

6. When tap water is probably fine — and when it is not

Probably fine:

• Your own tap water tests below roughly 10 ppm TDS — uncommon, but it happens with some soft-water municipal supplies and rainwater-fed systems.
• You are doing an occasional light rinse that gets wiped dry before it air-dries — rare in any commercial O&M schedule, more realistic for a small residential array cleaned by hand.
• You already run RO or DI water for another purpose on site and can route it to cleaning at no extra cost.

Treat the water first:

• Your tested TDS is above roughly 10 ppm — this covers most municipal and well-water sources in the US and Europe.
• The site is cleaned more than a few times per year, since spotting damage compounds with each cycle (see section 3).
• The site is in a hot, high-irradiance climate, where water evaporates faster and mineral concentration on the glass surface happens before the deposit can be rinsed away.
• The site operates under a performance-guarantee or O&M contract where cumulative optical degradation from scale is a measurable liability.

For guidance on how often your specific climate actually requires cleaning — which changes how much this compounding effect matters — see How Often Should You Clean Solar Panels?

Frequently asked questions

Can I use tap water to clean solar panels?

Yes, if it tests below roughly 10 ppm TDS. Most municipal tap water in the US and Europe measures 50–400 ppm, which is high enough to leave mineral spots after a handful of washes and, over repeated cycles, to form bonded scale. Test your specific supply with a handheld TDS meter rather than assuming.

What TDS level is safe for cleaning solar panels?

Below 5 ppm is spot-free in essentially all conditions, including direct sun. 6–10 ppm is spot-free in shade or moderate humidity but riskier in full sun. Above 10 ppm, visible spotting becomes likely, and above 20 ppm it is close to certain unless the panel is wiped dry by hand.

Does a water softener stop water spots on solar panels?

No. A softener exchanges calcium and magnesium for sodium through ion exchange, which prevents scale in plumbing but does not lower total dissolved solids. The water still spots on evaporation — just with sodium deposits instead of calcium and magnesium ones.

Is reverse osmosis necessary for solar panel cleaning?

For any site cleaned more than occasionally, yes, if the local tap water tests above roughly 10 ppm TDS. RO removes roughly 90–99% of dissolved solids and is the standard bulk treatment method; deionization is commonly added afterward as a polishing stage for a fully spot-free result.

Can hard water permanently damage solar panels?

Repeated cleaning with untreated hard water can etch or degrade the anti-reflective coating on the glass, which is a factory-applied layer that cannot be re-applied in the field once damaged. The mineral deposits themselves also roughen the surface, which accelerates future soiling. The damage is cumulative and difficult to reverse, which is why treating the water is cheaper than remediating the glass later.

How do I test my water’s TDS before cleaning panels?

A handheld TDS meter, the same inexpensive type used for aquarium and drinking-water testing, gives a reading in seconds. Test the actual water source you plan to use for cleaning, since TDS varies by region and can shift seasonally within the same municipal supply.

Sources

1. Basidewt — How Does a Spotless Water System Work? A Complete Technical Guide (0–5 / 6–10 / 10–20 / 20+ ppm spot-free tiers)
2. Solaris Hydrobotics — Wet vs Dry Solar Panel Cleaning (solar-specific water-quality guidance; hard water reduces panel efficiency)
3. AmPac Water — TDS Levels in Drinking Water: Complete Guide (US municipal tap water 100–500 ppm; EPA secondary standard)
4. SpotFreeRinse.com.au — DI Water vs RO vs Water Softener vs Distilled (softening does not reduce TDS; RO removes 90–99% of dissolved solids)
5. DIY Solar Forum — What Cleaner to Use to Rid Panels of Hard Water Stains (factory-applied AR coating cannot be recoated in the field)

The short answer

Sometimes — and the deciding variables are the soiling rate on your site, what a kilowatt-hour is worth to you, and the installed price you are quoted per kWp, more than plant size alone. For a ground-mounted plant selling power at about $0.035/kWh, a fully automated cleaning system breaks even over 10 years at roughly $22–34 per kWp under moderate-to-high soiling. Peer-reviewed literature puts the cost of such systems at about $12–42 per kWp. Those ranges overlap, so payback runs from about 1.2 years (very high soiling, low-end price) to no payback at all (low soiling, high-end price).

Plant size matters mainly through the installed price per kWp, because a robot deployment has costs that do not shrink with the plant. Section 4 therefore turns plant size into a budget: the most a site of that size can spend and still break even.

Scope and disclosure. This article covers the purchase decision for cleaning robots. Cleaning frequency is covered in How Often Should You Clean Solar Panels?, method selection in Dry vs Wet Solar Panel Cleaning, and water quality in Can You Clean Solar Panels With Tap Water?. WaterSprayElephant manufactures PV cleaning robots, so the analysis below is built from third-party sources and an open formula (Section 2) that you can check without relying on us. It contains no WaterSprayElephant prices.

1. Why published payback figures disagree

Search for cleaning-robot payback and the figures run from under a year to several years. The spread is less a disagreement about robots than a difference in what each source discloses.

Source	What it reports	What it discloses
Robot vendor ROI page (Solaris Hydrobotics)	Payback of 12–24 months for utility-scale, 18–36 months for commercial and industrial, and 3–4 years for residential; one 500 kW rooftop case at 1.8 years.	Robot cost, power price and site soiling rates are not stated.
Robot vendor illustrative 25 MW comparison (Taypro)	In its own illustration, the robot fleet’s annual cleaning O&M is higher than manual wet cleaning; the case rests on water saved and energy recovered. It notes that small fixed-tilt sites under 5 MW with mild soiling and cheap water may not clear robot capex.	Tariff stated; robot capex and soiling loss in percent not stated; labelled “illustrative only”.
Peer-reviewed assessment (Ilse et al., Joule, 2019)	Computes the maximum investment that gives a positive net present value over 10 years at a 5% discount rate, then compares it with reported robot costs. For an 80–100% cut in soiling rate, the average allowed investment is about $13–41 per kWp, against reported fully automated system costs of about $12–42 per kWp.	Method, equations and input assumptions published; utility-scale plants with an assumed power price of about 3 cents per kWh.

Read together: at the average utility-scale conditions in the peer-reviewed paper, what a robot can justify and what robots cost land in the same place. That is a break-even, not a 12–36 month payback. A short payback is possible, but it describes a specific site where soiling is heavier, power is worth more, or the price sits below the literature range — and a claim like that needs that site’s numbers.

Currency and unit conversions. Ilse et al. and the European study cited below report cleaning and robot costs in euros per m² of module area (or per kW). We convert at an assumed 1.10 USD per EUR and 21.4% module efficiency (0.214 kWp/m², the efficiency used in the European study) and show only dollars. The conversion is ours, not the papers’, and the 2019 cost figures predate recent changes in hardware and labour costs.

2. The payback formula and where each input comes from

The calculation has three parts. Every input is either sourced or flagged as our assumption.

1. Value at stake — what soiling and cleaning cost per kWp per year under your best manual schedule. Soiling loss = Y × p × SR × (n + 1) / 2, where Y is specific yield (kWh/kWp·yr), p is the value of a kWh, SR is the soiling rate (fraction of output lost per day) and n is days between cleanings. Cleaning spend = u × 365 / n, where u is the cost of one cleaning per kWp. The cost-minimising interval is n = √(730 × u / (Y × p × SR)), the form derived by Ilse et al. We floor it at one cleaning per year.
2. Value recovered — value at stake minus the soiling that remains when a robot passes daily, which is one day of accumulation (1 × SR). The daily pass reflects the literature figure of more than 95% soiling-rate reduction for fully automated systems.
3. Payback and break-even — simple payback = C / (B − r × C), where C is the installed quote per kWp, B is value recovered per year and r is annual running cost as a fraction of C. The break-even quote over 10 years at 5% is 8.11 × B / (1 + 8.11 × r). By construction, a simple payback of 8.1 years is exactly the break-even point.

Input	Value used	Source or basis
Specific yield Y	1,400 kWh/kWp·yr	NREL O&M cost-model example (10 MW tracking plant, Denver). Adjust to your site.
Power value p, ground-mount	$0.035/kWh	LBNL Utility-Scale Solar, 2024 Edition: newly signed PPAs average $35/MWh levelized (2023 dollars).
Power value p, rooftop	$0.145/kWh	EIA Electric Power Monthly: US commercial average retail price 14.53 ¢/kWh, July 2026. Assumes every kWh offsets retail power; exported kWh earn less.
Cost per manual cleaning u, ground-mount	$1.0 per kWp	IEA PVPS Task 13 modelling example (0.2 euro per m² per cleaning, about $1.03 per kWp when converted). The European study reports about $0.3–4.7 per kWp across countries. NREL’s 10 MW model example carries $41,410 per year for cleaning and vegetation management together, about $4.1 per kW.
Cost per manual cleaning u, rooftop	$4.0 per kWp	About four times the ground-mount figure. Ilse et al. report rooftop cleaning at 3–8 times the cost of ground-mounted PV.
Soiling rate SR	0.01 / 0.04 / 0.10 / 0.30 % per day (about 0.3 / 1.2 / 3.0 / 9.0 % per month)	IEA PVPS Task 13 tabulates measured US monthly rates from 0% (natural desert, winter) to 11.5% (dry agricultural, autumn). The four levels span that range; they are scenarios, not regional claims.
Robot installed cost C	$12–42 per kWp	Ilse et al., Table 2: fully automated cleaning systems, converted from euros per m². A 2019 literature range, not a quote.
Running cost r	5% of C per year	Our assumption, not a published figure. Covers batteries, brushes, service and supervision. Section 3 shows the sensitivity.
Evaluation period and discount rate	10 years, 5%	Ilse et al.

Two limits of the model. It treats soiling as accumulating linearly between cleanings, which ignores rain; where rain cleans well, the real value at stake is lower. A European study of measured soiling found annual loss averaging 0.9% across Europe if rain is assumed to clean perfectly, and 5.3% if rain removes only 10% of accumulated soiling. A range that wide is why measuring your own site matters more than any table here.

3. Ground-mounted plants: break-even quotes and payback

Table 1 applies the formula to a ground-mounted plant with power worth $0.035/kWh and a manual cleaning cost of $1 per kWp per pass, at four soiling rates. The last two columns show payback if the installed price is at the low ($12/kWp) or high ($42/kWp) end of the literature range.

Soiling level	Soiling + cleaning cost at stake ($/kWp·yr)	Break-even quote, 10 years at 5% ($/kWp)	Quote for a 5-year payback ($/kWp)	Payback at $12/kWp	Payback at $42/kWp
Low: 0.01% per day (0.3% per month); best manual schedule 1.0 cleanings per year (floor)	$1.90	$11	$8	9.3 years	No payback
Moderate: 0.04% per day (1.2% per month); best manual schedule 1.9 cleanings per year	$3.79	$22	$15	3.8 years	>20 years
High: 0.10% per day (3.0% per month); best manual schedule 3.0 cleanings per year	$6.01	$34	$24	2.2 years	10.9 years
Very high: 0.30% per day (9.0% per month); best manual schedule 5.2 cleanings per year	$10.43	$59	$41	1.2 years	5.1 years

Under the best manual schedule, mean annual soiling loss in these four cases is 1.8%, 3.9%, 6.2% and 10.7%. With daily robot passes it stays near one day of accumulation, 0.01–0.30%. Modelled results, not measurements.

• Very high soiling (9.0% per month): 1.2 years at the low end of the price range and 5.1 years at the high end. This is the only row where payback stays near or under five years across the whole range.
• High soiling (3.0% per month): 2.2 years at $12/kWp and 10.9 years at $42/kWp. A 5-year payback needs a quote of about $24/kWp or less.
• Moderate soiling (1.2% per month): 3.8 years at $12/kWp, but >20 years at $42/kWp. A 5-year payback needs about $15/kWp or less.
• Low soiling (0.3% per month): the best manual schedule is one cleaning a year and the value at stake is $1.90/kWp·yr. Break-even is about $11/kWp, below the low end of the literature range, so no price in that range pays back within 8.1 years.

Running cost moves these results more than most buyers expect. At the high soiling rate, raising running cost from 0% to 5% to 10% of the quote per year moves the break-even quote from $48 to $34 to $27 per kWp, and payback at $42/kWp from 7.1 years to 10.9 years to >20 years. The 5% figure is our assumption; replace it with the running cost stated in the quote you receive.

As a cross-check, the model’s break-even quotes for moderate to high soiling, $22–34 per kWp, fall inside the $13–41 per kWp that Ilse et al. report as the average allowed investment for an 80–100% soiling-rate reduction.

4. Plant size: what budget does your site support?

Per kWp, the economics above do not depend on plant size. What size changes is the installed price per kWp. A deployment has costs that do not shrink with the plant: at least one robot or one row system, commissioning, training and, for fixed-row systems, a row-layout and end-structure assessment. For a small plant the useful question is therefore not a payback figure but the total budget it can support. Table 2 multiplies the break-even quotes by plant size at moderate and high soiling.

Plant	Soiling + cleaning cost at stake per year (moderate soiling)	Break-even budget, moderate soiling (1.2% per month)	Break-even budget, high soiling (3.0% per month)
Residential rooftop, 8 kWp	$120	$710	$1,100
Small commercial rooftop, 250 kWp	$3,900	$22,000	$35,000
Large commercial, 2 MWp, retail-priced power	$31,000	$180,000	$280,000
Ground-mount, 2 MWp, PPA-priced power	$7,600	$44,000	$69,000
Utility-scale, 50 MWp, PPA-priced power	$190,000	$1,100,000	$1,700,000

How to read Table 2. Compare it with an all-in quote: hardware, commissioning, training and site works, plus 10 years of running cost at the 5% assumption. If the quote exceeds the budget, the purchase loses money over 10 years at a 5% discount rate in that scenario. Rooftop rows use the retail power value ($0.145/kWh) and a cleaning cost four times the ground-mount figure; ground-mount rows use the PPA-level value. Figures are rounded to two significant digits. A contractor spreading one machine across many customers is a different calculation: the machine earns service revenue instead of recovering one site’s energy.

Two readings follow. For an 8 kWp home system the value at stake is about $120–200 per year, and any all-in robot quote above roughly $710–1,100 loses money over 10 years in this model. At 250 kWp the budget is about $22,000–35,000; whether one deployment fits inside that depends on the quote. From 2 MWp upward the budget grows in proportion to plant size, fixed deployment costs matter less per kWp, and the soiling rate and the value of a kWh take over as the deciding inputs.

5. Which robot type matches which economics

The daily-pass assumption in Table 1 applies to fully automated systems. Remote-controlled crawler robots change the economics differently, and each class needs its own test.

Equipment class	Published specifications	Where the money comes from	Savings ceiling
Remote-controlled crawler robots: R1 (battery) and R2 (mains)	R1: wet or dry cleaning, working angle up to 20°, module-gap crossing up to 20 cm, published runtime of 6–8 hours on dual lithium batteries. R2: mains power (stable site power of 100–250 V), crossing up to 25 cm, working angle up to 20°.	A lower cost per pass, and more passes with the same crew. The IEA PVPS report classes portable robots that are moved from row to row as semi-automatic, so a crew is still part of each pass.	Your current manual cleaning spend plus the energy value of any extra passes. On a ground-mount plant, cleaning spend under the best manual schedule is about $1.9–5.2 per kWp per year across the moderate-to-very-high rows of Table 1, roughly half of the cost at stake.
Fixed-row automated systems: R20 (hanging autonomous robot) and R20A (shuttle that moves robots between rows)	R20: scheduled cleaning through cloud, mobile-app or timer control; designed for uniform fixed ground-mounted rows, with row length, gradient and module dislocation inside engineered limits; a row-layout and end-structure assessment is required before quotation. R20A: moves compatible hanging robots between engineered rows, reducing the need for one robot per row, as part of an integrated cleaning-and-transfer project.	Scheduled passes without a crew — the mechanism behind the more-than-95% soiling-rate reduction the literature attributes to fully automated systems. This is the class Table 1 models.	Value at stake (Table 1) minus residual soiling and running cost.

We make these products, so read this section as a summary of published specifications, not as an endorsement. We have not published prices or measured payback figures for them in this article; run the formula against a project quote. Both crawler robots also run wet or dry, which ties the economics back to soil type and water cost — see the method comparison linked at the top of this article.

6. When a robot is the wrong purchase

• Your best manual schedule is about one cleaning a year. That is the low-soiling row of Table 1: $1.90 per kWp per year at stake on a ground-mount plant, below what any price in the literature range can recover.
• Rain does most of your cleaning. The European soiling study’s 0.9% to 5.3% range depends entirely on how well rain cleans. Measure before you buy; do not borrow a regional average.
• The array is outside the robot’s published limits. Our crawler robots are designed for working angles of about 15–20°; R1 crosses module gaps up to 20 cm and R2 up to 25 cm; fixed-row systems need uniform rows within engineered limits. A robot that cannot traverse your array has no payback.
• Cleaning labour is already cheap. The European study reports per-cleaning costs from about $0.3 to $4.7 per kWp across countries. At the low end, there is little spend for a robot to avoid.
• The plant is small and the quote is not. Compare the quote with the budget in Table 2 before requesting a site visit.

Our Romania rooftop project documents a double-head cleaning robot on a commercial rooftop. It is a project narrative rather than a payback calculation, so it is not an input to the model here.

7. Run your own numbers

1. Measure your soiling rate. Four to eight weeks of readings in a dry period give a rate in % per day. An optical soiling sensor such as our A60 PV Dust Measuring Instrument reports PV-glass cleanliness (soiling ratio), not energy loss, so convert the rate to energy with your own irradiance-normalised data; the validation methods are in section 6 of the cleaning-frequency guide.
2. Find your cost per cleaning per kWp from your O&M provider’s invoices: total cost of a pass divided by plant capacity in kWp.
3. Compute the value at stake with the formulas in Section 2. They are short enough for a spreadsheet.
4. Request all-in quotes with hardware, commissioning, training and running cost listed separately.
5. Test the result against the 8.1-year threshold, then re-run it with the soiling rate 50% lower and the running cost five points higher. If the answer survives both, it is robust.

After any purchase, verify the result with a same-day before-and-after reading rather than comparing two different days, because weather alone can swing daily output enough to hide a cleaning gain.

Frequently asked questions

Are solar panel cleaning robots worth it?

It depends on three numbers: your soiling rate, the value of a kWh, and the installed quote per kWp. For a ground-mounted plant at $0.035/kWh, fully automated systems break even over 10 years at about $22–34 per kWp under moderate-to-high soiling, inside the $12–42 per kWp literature range. They do not break even under low soiling, and they do across the whole range under very high soiling (break-even about $59 per kWp).

How long is the payback period for a solar panel cleaning robot?

In the model for a ground-mounted plant, payback runs 1.2 years to 5.1 years at very high soiling, 2.2 years to 10.9 years at high soiling, 3.8 years to >20 years at moderate soiling, and 9.3 years to no payback at low soiling, from the low to the high end of the literature price range. Robot vendors publish 12–36 months for utility and commercial sites without disclosing robot cost, power price or soiling rate.

How much does a solar panel cleaning robot cost?

A 2019 peer-reviewed assessment reports fully automated cleaning systems at about $12–42 per kWp of module capacity (converted from euros per m²). We do not publish WaterSprayElephant prices in this article; the R20 product page states that a row-layout and end-structure assessment is required before quotation.

Is a cleaning robot worth it for a home solar system?

In this model, the value at stake for an 8 kWp rooftop is about $120–200 per year at moderate-to-high soiling, and the 10-year break-even budget is roughly $710–1,100. An all-in quote above that loses money, so periodic manual cleaning is the option a robot has to beat.

Do cleaning robots recover more energy than manual cleaning?

In the model, daily robot passes hold average soiling loss near one day of accumulation (0.01–0.30%), against 1.8–10.7% for the best manual schedule. Real fleets miss days to weather, faults and battery cycles, so treat the robot figure as an upper bound.

Does a cleaning robot pay back faster in a desert?

Higher soiling raises the value at stake and shortens payback: at $42/kWp, payback in Table 1 falls from 10.9 years at high soiling to 5.1 years at very high soiling. Dust type, dew and wind also change how well a given robot performs, so measure the site rather than assuming the region.

What running costs should I include?

Batteries or mains supply, brushes, service visits, spare parts and supervision. The model assumes 5% of the quote per year, which is our assumption. At 10% per year, payback at $42/kWp under high soiling goes from 10.9 years to >20 years.

Sources

1. Ilse et al. — Techno-Economic Assessment of Soiling Losses and Mitigation Strategies for Solar Power Generation, Joule 3 (2019) (soiling-loss model, cost-minimising cleaning interval, 10-year / 5% allowed-investment method, fully automated cleaning cost range, rooftop cleaning cost multiple)
2. Fernández Solas et al. — Photovoltaic soiling loss in Europe: Geographical distribution and cleaning recommendations, Renewable Energy 239 (2025) (European annual soiling loss with and without rain cleaning, per-cleaning cost range, 21.4% module efficiency)
3. IEA PVPS Task 13 — Soiling Losses: Impact on the Performance of Photovoltaic Power Plants (T13-21:2022) (US monthly soiling rates, per-cleaning cost example, semi-automatic robot classification, annual manual cleaning recommendation)
4. NREL — Model of Operation-and-Maintenance Costs for Photovoltaic Systems (NREL/TP-5C00-74840, 2020) (specific yield example)
5. Berkeley Lab — Utility-Scale Solar, 2024 Edition (average levelized PPA price)
6. US EIA — Electric Power Monthly, Table 5.6.A (commercial average retail price, July 2026)
7. Solaris Hydrobotics — The ROI of Autonomous Solar Panel Cleaning (vendor-published payback ranges)
8. Taypro — Benefits of Cleaning Robots on Indian Utility Solar Plants (vendor illustrative 25 MW comparison)

Solar glass is tough, but it’s not scratch-proof — and a scratched panel loses output permanently. The good news: damage almost always comes from how panels are cleaned, not from cleaning itself. Follow a few rules and you get clean glass with zero risk. Here they are.

1. Use a soft, non-abrasive brush

Never use stiff industrial bristles, scouring pads or anything abrasive. Purpose-made solar brushes use soft synthetic filaments set into a smooth plate. Match the bristle to the job: softer nylon for gentle dry-dust sweeping, and a firmer PBT for wet cleaning of stubborn dirt — firm enough to lift grime, still safe on glass.

2. Don’t dry-scrub stuck-on dirt

The single biggest cause of scratches is pushing dry, gritty dirt or dried droppings across the glass. If it won’t sweep off dry, add water — water floats the grit away so the brush glides instead of grinding. Save dry cleaning for loose, powdery dust.

3. Watch your water quality

Hard tap water dries into mineral spots that look like dirt and tempt more scrubbing. On sensitive sites, feed the brush with reverse-osmosis (RO) water for a spot-free finish and less rework.

4. Skip the pressure washer

High-pressure jets can force water past the frame seals and stress the cells. Gentle, brush-fed water at low pressure cleans just as well without the risk.

5. Avoid extreme temperatures

Don’t clean panels that are scorching hot in midday sun — cold water on hot glass causes thermal stress. Early morning or evening is safer and more effective.

How our equipment is built for this

Our machines are designed around exactly these rules: soft nylon brushes for dry dust and firmer PBT for wet work, tufted (not woven) brush plates that won’t trap grit, water-fed heads that keep the glass wet, and insulated, non-conductive handles for safe work near live panels. Our crawler robot even carries both brush types for dry and wet cleaning.

Want help choosing a method and brush that’s safe for your panels? Tell us about your site and we’ll spec it.

In deserts and high-dust regions, soiling isn’t a nuisance — it’s one of the biggest losses on the plant. Panels can lose a quarter or more of their output between rains, and the dust piles back on within days. The challenge is cleaning often enough, with little or no water, and without scratching the glass. Here’s what actually works.

Why dusty regions are different

• Soiling is severe and fast — with little rain to rinse the glass, dust rebuilds quickly, so losses climb through the whole dry season.
• Water is scarce and precious — trucking clean water to a remote site is expensive, so waterless or low-water methods win.
• Fine, abrasive dust — desert sand is gritty; dragging it across the glass with the wrong brush risks micro-scratches.

What works in the desert

• Dry cleaning first — for loose, powdery dust, a roller or a robot sweeps it off with no water at all, using soft bristles that lift dust without grinding it in.
• Water only when needed — for baked-on grime or bird droppings, spot-clean with water and a harder PBT brush. Feeding the brush with RO (reverse-osmosis) water leaves a spot-free finish, because there are no dissolved minerals to dry on the glass — important where hard groundwater is the only supply.
• Clean on data, not the calendar — because soiling swings so much, an optical soiling sensor tells you exactly when a clean pays for itself, so you don’t waste water and labour cleaning too early.
• Off-grid ready — lithium and dual-power machines run where there’s no mains at the row.

Built for these conditions

Our equipment is rated for −20 to 50 °C operation, offers soft nylon brushes for gentle dry sweeping and harder PBT for wet work, and pairs with RO water treatment for spot-free results. It’s the combination high-dust markets across the Middle East and beyond actually need.

Cleaning a site in a hot, dusty or water-scarce region? Tell us your conditions and we’ll spec a low-water cleaning setup for it.

When you buy solar cleaning equipment, you’re really choosing who stands behind it — the company that built it, or a trader who bought it to resell. Both can sell you a machine. Only one can spec it to your site, ship you genuine spares and answer for it when something needs fixing. Here’s how to tell them apart, and why it matters for a purchase you’ll rely on for years.

How to tell who you’re dealing with

Ask one direct question: “Do you manufacture this, or resell it?” A manufacturer can talk about brush materials, motor ratings, battery cycles and tolerances without checking with someone else. A pure trader usually can’t — because they didn’t build it.

What you get from the manufacturer

• Real customization — pole length, brush material, power version, plug and voltage set to your market and site, not a fixed catalogue.
• Genuine spare parts — the exact brushes, batteries, chargers and hoses that fit, for years — not a guess after the trader has moved on to another product.
• One accountable source — design, production, testing and after-sales under one roof, so warranty and support don’t get lost in a chain of resellers.
• Honest advice — a maker of the full range can tell you when you don’t need the expensive machine; a trader is limited to whatever they happen to stock.

Where a trader can still make sense

A local distributor who holds stock, handles import and offers on-the-ground service is genuinely useful — as long as they’re backed by a real manufacturer for parts and warranty. The setup to avoid is a middleman with no factory behind them: markup without accountability.

Our position

We’re a manufacturer that builds the full range of solar cleaning equipment in-house and sells direct to overseas projects — robots, electric brushes and manual tools, all from one source. That’s why we can match the machine to your site, ship the right plug and voltage, and stand behind it long after the sale. More about who we are, or tell us about your project.

The honest answer is: it depends on the machine and your site — so the better question isn’t “what does it cost,” it’s “how fast does it pay back.” For most commercial and utility sites, mechanised cleaning earns its price back through the labour it saves and the generation it recovers. Here’s how to think about both.

What drives the price

• Type of machine — a long-reach electric brush is a fraction of the cost of a crawler robot; a robot costs more but replaces far more labour.
• Reach and width — longer poles, wider brushes and larger robots cost more but cover more per hour.
• Power version — mains, lithium or dual-power; battery packs add cost but unlock off-grid and remote sites.
• Brush and customization — brush material, plug and voltage for your country, and any OEM/branding.

Because these vary site to site, we quote per project rather than list a single price — it’s the only way to give you an accurate number instead of a misleading one.

What drives the payback

The price is only half the equation. The return comes from three places:

• Labour saved — usually the biggest line. A wide electric brush lets one person cover far more area per hour than a hand tool; a robot covers the most and runs with minimal supervision.
• Yield recovered — clean panels generate more. Where soiling is heavy, the recovered output alone can justify regular cleaning (see how much soiling really costs).
• Safety and risk — keeping crews off high roofs removes fall risk and the costs that come with it.

On larger, dustier or higher sites, those returns add up quickly — which is why the machine that looks more expensive up front is often the cheaper choice over a year of cleaning.

Get a real number for your site

Tell us your site size, module tilt, power situation and location, and we’ll recommend a machine and give you an accurate quote — usually within 24 hours. Request a quote →

The short answer

Cleaning frequency is a function of local soiling rate, not a fixed calendar interval. For most of the United States and Europe:

Climate zone	Annual soiling loss (uncleaned)	Recommended cleaning
Humid, frequent rain — Pacific Northwest, UK, Ireland, Netherlands, Denmark	1–3%	0–1× per year, spot-clean only
Temperate / moderate — Northeast & Midwest US, Germany, France, Poland, Czechia	3–6%	1–2× per year
Dry or agricultural — Texas, California Central Valley, Spain, Italy, Portugal, Greece	4–8%	2–4× per year (roughly quarterly)
Desert / high dust — Southwest US, Almería, Sicily, Cyprus, interior Türkiye	8–25%	Quarterly to monthly

If your site falls in the lower half of that table, the question stops being “how often should I clean?” and becomes “how do I afford to clean that often?” That is an equipment and labour question, not a scheduling one — see section 5 below.

This article covers cadence only. For a general maintenance framework — soiling mechanics, method selection, and inspection routines — see our earlier guide, Practical Tips for Solar Panel Cleaning and Maintenance. That article is a broad best-practices overview; this one goes deep on scheduling and cost, and is intended to be read alongside it.

1. Why frequency is a climate question, not a calendar question

“Twice a year” is the industry’s default rule of thumb. It is an average across all climates, which makes it wrong in both directions for most individual sites.

Soiling accumulation is driven by five factors:

• Atmospheric dust load. Arid and semi-arid regions accumulate dust several times faster than humid ones.
• Rain frequency. Rain removes loose dust and little else — see section 3.
• Local sources. Agriculture, construction, unpaved roads, industrial emissions, and coastal salt spray all add residue that rain will not lift.
• Module tilt. Low tilt retains dust longer. Heavy soiling concentrates along the lower edge of each row.
• Dew and humidity. Overnight condensation cements dust into a film. This is the mechanism behind the cemented soiling discussed in section 2.

Two sites in the same country can legitimately require 0.5 and 4 cleanings per year — a spread of roughly 8×. That is why we recommend measuring rather than scheduling by habit, and why any single-number frequency recommendation should be treated with suspicion unless it states the climate it applies to.

2. Soiling rates by region

United States

Region	Monthly loss	Annual (uncleaned)
Pacific Northwest	0.05–0.15%	1–2%
Northeast US	0.08–0.20%	1–3%
Midwest US	0.15–0.35%	2–4%
Texas / Oklahoma	0.20–0.50%	3–6%
California Central Valley	0.25–0.65%	3–8%
Industrial / Rust Belt	0.30–0.80%	4–10%
Southwest US desert	0.45–1.20%	5–15%

Europe

Region	Estimated monthly loss	Estimated annual (uncleaned)
UK, Ireland, Netherlands, Denmark	0.05–0.15%	1–3%
Germany, France, Belgium, Poland, Czechia	0.10–0.30%	2–5%
Portugal, Spain, Italy, Greece, southern France	0.25–0.70%	3–8%
Almería, Sicily, Cyprus, interior Türkiye	0.60–1.80%	8–20%

Methodological note — read this before citing the European figures. The US monthly figures are drawn from published soiling and cleaning-economics datasets (see Sources). The European figures are climate-analogue estimates, produced by mapping each region onto the closest measured climate band, because site-level soiling data is far less publicly available in Europe than in the US. We publish them as estimates, not measurements, and we will replace them with measured values as we obtain them. If you operate in Europe and have measured data, we would like to hear from you.

Reference points outside these markets

Region	Monthly loss	Annual (uncleaned)
Northern India	0.70–1.50%	8–18%
Middle East / North Africa	0.80–2.50%	10–25%

Reconciling the ranges you will see elsewhere

Published soiling-loss figures vary widely — you will see 2%, you will see 30%, and both may be correct. The apparent contradiction comes from three different measurement bases:

1. Time-based (the tables above) — loss accumulated over a period under local conditions.
2. Density-based (the table below) — loss as a function of how much dust is actually on the glass.
3. Event-based — loss measured immediately after a discrete event such as a dust storm.

Dust density	Power loss
0.1 mg/cm²	2–3%
0.5 mg/cm²	6–9%
1.0 mg/cm²	11–15%
2.0 mg/cm²	18–24%
5.0+ mg/cm²	30–45%

The low single-digit percentages describe lightly soiled arrays in temperate climates. The 20–30% figures that circulate for “dirty panels” describe heavy, prolonged accumulation or post-storm conditions. Both are describing real states — they are just not the same state. When you compare sources, check which basis is being used before concluding that one of them is wrong.

Secondary thermal effect. Heavily soiled modules run 7–12 °C hotter than clean ones, which adds a further 3–5% efficiency loss on top of the optical loss. Soiling costs you twice: less light reaches the cell, and the cell converts what does arrive less efficiently.

3. Why rain is not a cleaning strategy

Rain removes an estimated 40–75% of loose dust, and almost none of the residue that drives the larger losses:

• Pollen — adhesive and seasonal; largely rain-resistant once dried.
• Bird droppings — acidic, and capable of etching the anti-reflective coating if left in place.
• Soot and diesel particulate — oily; binds to glass.
• Cemented dust — dust that has cycled repeatedly through dew and drying, forming a chemically bonded film rather than a loose layer. Cemented soiling in hyper-arid regions has been measured at up to 9.8% annual energy loss.

Rain also creates a failure mode of its own. On a partially soiled array, rain can wash dust into streaks and concentrate it along the lower edge of each row. Because that produces non-uniform soiling, the result is not merely a proportional output loss — it is cell mismatch and, in the worst case, hot spots.

The “self-cleaning array” idea holds only in climates with frequent, heavy rain and low dust load — roughly the first row of the table in section 2. Everywhere else, rain is a partial credit at best, and a reason to inspect rather than a reason to skip cleaning.

4. How to tell whether your own array needs cleaning

Three methods, in increasing order of usefulness.

Method 1 — Visual inspection

Cheapest and least reliable. You can see bird droppings and heavy dust. You cannot see the 3–5% film that accumulates before soiling becomes obvious — by which point you have already lost weeks of production.

Method 2 — Output comparison

Compare current production against a known-clean baseline at equivalent irradiance and temperature. Reliable only when irradiance is normalised. See the warning in section 6 — this is the single most common source of bad cleaning data.

Method 3 — In-line soiling sensor

An optical-scattering sensor mounted at module plane continuously reports how clean the glass is, and can trigger cleaning automatically when a threshold is crossed. This is the only method that scales: it removes guesswork, produces a time-stamped record, and lets you clean on condition rather than on a calendar.

One clarification that matters, because this class of instrument is widely misdescribed: a soiling sensor measures PV-glass cleanliness — it does not measure energy loss. Energy loss also depends on irradiance, module behaviour, and system conditions. What the sensor gives you is a direct, repeatable reading of the variable you can actually act on. Our A60 PV Dust Measuring Instrument works on this principle: optical scattering, measurement range 100%–50%, accuracy 1% across 100%–90% and 4% across 80%–50%, stability 1% per year, response time under one second, RS485 Modbus with optional 4G, IP67, no moving parts, and no on-site calibration.

Whatever instrument you use, validate its readings against the site plan before you rely on them, and treat the number as a cleanliness indicator rather than a revenue figure.

5. Where manual cleaning stops being economic

Cleaning economics are size-dependent, and the crossover is sharper than most operators expect:

Segment	Cleaning cost basis	Break-even soiling level
Residential, DIY	~$60/yr	~2% — usually worth doing
Residential, professional crew	~$300/yr	~8% — rarely worth it
Commercial, professional crew	$0.04–0.06/W/yr	15–25%
Utility-scale, robotic	below truck-based O&M	standard practice in high-dust regions

The pattern is clear. For residential systems in moderate climates, professional cleaning rarely pays for itself — labour cost per panel is too high relative to the energy recovered. As arrays grow larger and soiling gets heavier, the arithmetic flips hard:

• Utility-scale waterless cleaning robots vs. manual labour: payback in 1–3 years
• Residential and small-commercial robots: 8–15 years — usually not justified

The practical consequence: once your required frequency passes roughly quarterly, manual cleaning becomes both expensive and difficult to schedule, and the bottleneck shifts from whether to clean to how to clean that often with the crew you have. At that point the question is an equipment question. We cover the selection logic in our solution paths by site type, and the hardware in the R1 crawler robot and R2 crawler robot pages.

All figures above are USD and are drawn from published cleaning-economics datasets (see Sources). They are benchmarks for order-of-magnitude planning, not quotations. Labour rates, water cost, and access difficulty vary enough by region that any real cleaning decision should be costed against your own site.

6. How to validate a cleaning result

Most published “we cleaned the array and output rose by X%” claims are not measurable as stated, because they compare two different days. Two consecutive days can differ by 20–40% in yield on weather alone. Attribute that to cleaning and you have built a case on noise.

To produce a result that survives scrutiny, use one of the following, in descending order of strength:

1. Same-day before/after. Read output in the morning, clean, read again in the afternoon at comparable irradiance. Irradiance conditions are largely controlled, so the comparison is defensible.
2. Multi-day averaging. Average three to five days before cleaning and three to five days after. Weather noise is reduced, though not eliminated.
3. Irradiance-normalised comparison. If you have GHI or a reference cell, normalise both periods and express the result as a performance ratio change. This is the most rigorous option and the one that supports a published claim.
4. Soiling-sensor readings. An optical soiling sensor gives a direct before/after cleanliness figure, independent of weather. This does not by itself demonstrate an energy gain, but it removes all ambiguity about whether the array actually got cleaner.

Two further cautions:

• Check your chart axes. Monitoring dashboards auto-scale the Y-axis. A before/after pair of charts with different axis maxima will visually exaggerate or understate the difference. Fix the scale before publishing.
• State the conditions. A percentage gain is meaningless without irradiance, ambient temperature, and soiling level. Publish the conditions alongside the number.

We apply this methodology to our own project reporting, including the rooftop installation documented in our Romania project case.

Frequently asked questions

How often should solar panels be cleaned in the US?

It depends on region. The Pacific Northwest needs 0–1 cleanings per year. The Southwest desert needs quarterly to monthly. Texas, California’s Central Valley, and the industrial Midwest fall between those at 2–4 times per year.

Can I just rely on rain to clean my solar panels?

No. Rain removes roughly 40–75% of loose dust and almost none of the pollen, bird droppings, soot, or cemented dust that cause the larger losses. In low-rainfall or high-dust regions, rain is not a substitute for cleaning.

Do solar panels really lose that much power from dirt?

It depends on how the loss is measured. Time-based losses range from 1–3% annually in humid climates to 10–25% in desert regions. Density-based losses range from 2–3% at 0.1 mg/cm² to 30–45% above 5 mg/cm². Post-dust-storm losses can reach 40–60%. Heavily soiled modules also run 7–12 °C hotter, adding 3–5% further loss.

How do I know when my panels actually need cleaning?

An in-line soiling sensor is the only method that scales, because it measures glass cleanliness continuously and can trigger cleaning on condition. Visual inspection misses anything below roughly 5% loss, and comparing output across different days is unreliable unless irradiance is normalised.

Is professional cleaning worth it for a home system?

Usually not in moderate climates. Residential professional cleaning costs roughly $300 per year against a break-even soiling level of about 8%, which most residential sites in the US and Europe never reach. DIY cleaning at around $60 per year breaks even at about 2%, which is why it is usually the better option at residential scale.

Does a soiling sensor tell me how much energy I am losing?

No. A soiling sensor measures PV-glass cleanliness, or soiling ratio. Energy loss also depends on irradiance, module behaviour, and system conditions. The sensor tells you how dirty the glass is — which is the variable you can act on — not what the dirt is costing you in revenue.

Should I clean my panels in winter?

Lower sun angles and snow cover change the calculation, and cleaning in freezing conditions introduces its own risks. In most of Europe and the northern US, winter is a low-yield period; the highest-value cleaning windows are late spring and late summer, after pollen and after peak dust season. Snow should be removed by other means — never by scraping.

Sources

1. ScienceDirect — Techno-Economic Assessment of Soiling Losses and Mitigation Strategies for Solar Power Generation (regional soiling rates, cost benchmarks, payback periods)
2. ScienceDirect — Impact of dust accumulation on PV and CSP systems: experimental analysis and simulation insights in semi-arid climate
3. Green Building Africa — Cemented soiling of solar panels in hyper-arid regions can cause annual energy losses of up to 9.8%
4. Solaris Hydrobotics — Water Quality for Solar Panel Cleaning (TDS thresholds; used in the companion article on cleaning water)

The short answer

Dry vs. wet is not a philosophy to pick once and apply everywhere — it is a question you answer by looking at what is actually on the glass. Loose surface dust and wind-blown sand brush off dry without much trouble. Bird droppings, pollen, soot film, and dust that has been cemented on by dew or a light rain will not come off dry at all, and dragging a brush across them dry mostly smears the residue around rather than lifting it.

The method mismatch cuts both ways. Use wet cleaning where dry would do and you spend water you did not need to spend. Use dry cleaning where wet is actually required, and on top of leaving soiling behind, you risk a second problem: grit dragged dry across glass behaves like loose sandpaper, and the resulting micro-scratches in the anti-reflective coating are permanent. Section 3 has the numbers on that.

This article covers method selection and execution: dry vs. wet, and how to run each one safely. It does not re-cover cleaning frequency (see How Often Should You Clean Solar Panels?) or water quality (see Can You Clean Solar Panels With Tap Water?) — those are separate decisions that this article builds on in Section 4.

1. Match the method to what is actually on the glass

The deciding factor is soil type, not site preference. A cleaning-equipment vendor comparison of wet and dry methods puts it plainly: wet cleaning “excels” at bird droppings, sticky pollen, and industrial residues, while dry methods “may struggle with sticky residues or hardened bird droppings.” Loose, uniform dust is the one category where dry genuinely competes.

Soil type	Dry cleaning	Wet cleaning
Loose dust / wind-blown sand	Removes effectively	Also effective, but uses water you may not need
Pollen / light organic film	Struggles — residue resists dry brushing	Removes effectively
Bird droppings	Struggles — tends to smear rather than lift	Required — water softens and lifts it
Soot / industrial residue	Struggles	Required
Rain-cemented or dew-cemented dust	Cannot lift once bonded to the glass	Required, may need longer dwell time

If your site only ever sees loose dust — dry, low-humidity climates with no airborne organics — dry cleaning is a legitimate option, not a compromise. Everywhere else, water is doing real chemical work (softening and lifting a bonded residue), not just rinsing.

2. Dry vs wet, side by side

Dimension	Dry cleaning	Wet cleaning
Water use	Zero	Manual hose/pole: roughly 0.1–0.5 gal (0.4–1.9 L) per panel. Metered or robotic application can cut that substantially — see note below.
Soiling it actually removes	Loose dust and sand only	Full range, including sticky and organic residue (see §1)
Glass-surface risk	Elevated without water as a lubricant — see §3 for the numbers	Lower in principle, but shifts the risk to pressure and panel temperature — see §4
Best-fit scenario	Arid/desert sites, water-restricted or freezing locations, uniform loose dust	Heavy soiling, sticky/organic residue, salt-spray coastal sites, routine maintenance where water is available

One vendor field comparison reports wet cleaning achieving roughly 2–5 percentage points better restoration of original output than dry methods on comparable soiling, and metered/robotic water application cutting consumption by up to ~90% versus manual hose washing. Both figures come from a single cleaning-equipment vendor’s own comparison rather than an independently audited study — treat them as directional, not a guarantee for your site.

3. The hidden cost of dry cleaning: micro-abrasion

Water does more in a wet clean than carry dirt away — it lubricates. Without it, any grit trapped between a brush and the glass gets dragged rather than floated, and dragged grit behaves like sandpaper. A brush manufacturer’s own guidance is blunt about the consequence for sandy sites: “dry brushing should be avoided entirely, as it grinds particles into the glass surface.” Their recommended alternative is to rinse dust off the glass before any brush makes contact, rather than pushing a dry tool across it.

The damage this causes is to the anti-reflective coating (ARC), a factory-applied layer that cannot be re-applied in the field once it is scratched or stripped — so unlike a layer of dust, it is not reversible by cleaning again.

Wire brushes and scouring pads are the most common cause of avoidable scratching; the same source recommends soft materials and warns against abrasive tools on ARC-coated glass. If your site does need dry cleaning on loose dust, that guidance — soft rotary brush, never wire bristle or an abrasive pad — is the control that actually matters, more than the dry-vs-wet choice itself.

Our G4 roller-brush machine and R1 crawler robot both run in wet or dry mode on the same unit with a soft roller brush rather than a wire or abrasive tool. That does not make dry mode risk-free on gritty sand — it still depends on how much loose grit is on the glass — but it avoids the specific wire-brush and scouring-pad failure mode above.

4. If you clean wet, control these three variables

Water quality

Above roughly 10 ppm total dissolved solids, water leaves mineral spots on evaporation and, over repeated washes, bonds into scale that is harder to remove than the dust you started with. Most US and European municipal tap water tests well above that threshold. This is covered in full, with TDS tiers and treatment options, in Can You Clean Solar Panels With Tap Water? — if you are setting up a wet-cleaning program, read that before you read the rest of this section.

Water pressure

Module manufacturers cap wash pressure well below what a typical pressure washer puts out, and the limits are specific enough to quote directly:

Manufacturer	Maximum pressure
Trina Solar	700 kPa (~101 psi), front glass
JA Solar (ARC glass)	<690 kPa (100 psi); soft tools only
First Solar	35 bar (500 psi) at the nozzle

At Watersprayelephant, we offer a wide range of water pump products to meet the varying maximum allowable washing pressure requirements of different panels.

Panel temperature

Direct afternoon sun can bring a panel’s surface to roughly 60°C. Pouring noticeably cooler water onto glass at that temperature creates a rapid, localized temperature differential — the classic thermal-shock mechanism, the same reason a cold drinking glass can crack under hot water and vice versa — and it can stress or crack the glass.

Worth stating plainly, because it is easy to get backwards: the documented risk is cold water hitting a hot panel, not hot water hitting a hot panel. “Avoid washing with hot water on a hot panel” is not the mechanism at work here — the mismatch that causes thermal shock is between the panel’s temperature and the water’s temperature, and on a sun-heated panel the water is almost always the cooler side of that gap.

The practical fix is timing, not water temperature management: clean in the early morning or on an overcast day, when the panel has not yet heated up, rather than mid-afternoon. This control applies on top of water quality and pressure — all three are independent checks, not substitutes for each other.

5. A decision framework for your site

Putting Sections 1–4 together as a single path from soil type to execution controls:

1. Identify what is actually on the glass: loose dust/sand, or something sticky, organic, or cemented.
2. Loose dust/sand, water available, site not freezing: either method works; wet typically recovers a few extra percentage points of output (§2).
3. Loose dust/sand, no water available, water-restricted, or freezing conditions: dry cleaning with a soft rotary brush.
4. Bird droppings, pollen, soot, or cemented grime, in any water situation: wet cleaning is required (§1).
5. Whichever branch you land on, apply the matching controls before you start — soft-brush-only for dry (§3), or water quality / pressure / panel temperature for wet (§4).

Frequently asked questions

Can you clean solar panels without water?

Yes, for loose dust and sand — a soft rotary brush removes it effectively dry. It will not remove bird droppings, pollen, soot film, or rain-cemented dust; those need water to soften and lift. Dry cleaning also carries a documented risk of scratching the anti-reflective coating if grit gets dragged across dry glass; see the next question.

Is dry cleaning bad for solar panels?

It can be, depending on tool and soil type. Industry-reported estimates put moderate scratching of the anti-reflective coating at roughly 2–5% efficiency loss, and severe scratching at up to 15% — both permanent, since the coating cannot be restored once damaged. The main cause is dragging grit or using the wrong tool (wire brushes, scouring pads) dry; a soft rotary brush used only on loose, non-abrasive dust carries substantially lower risk.

Can you pressure wash solar panels?

Only within limits that are much lower than a typical pressure washer’s output, and the limit is set by your panel manufacturer, not by general advice. Trina Solar specifies a maximum of 700 kPa (~101 psi) on front glass; JA Solar specifies under 690 kPa (100 psi) for ARC-coated glass; First Solar caps nozzle pressure at 35 bar (500 psi). Check your specific panel brand before using any powered sprayer, and use a soft brush head rather than a bare jet.

Will cold water crack a hot solar panel?

Yes — and the mechanism runs opposite to what is sometimes assumed: it is cooler water meeting sun-heated glass that causes thermal shock, not hot water on a hot panel. Panel surfaces can reach roughly 60°C in direct afternoon sun; water applied at that point is almost always the cooler side of the gap, and the resulting temperature differential can stress or crack the glass. Clean during early morning or overcast conditions instead of mid-afternoon.

Which WaterSprayElephant products support both dry and wet cleaning?

The G4 roller-brush machine and the R1 crawler robot both operate in wet or dry mode on the same unit. Switching to dry mode does not remove the soil-type logic in Section 1 or the scratch-risk considerations in Section 3 — it just means you are no longer spending water on soiling that did not need it.

Sources

1. Solaris Hydrobotics — Comparing Wet vs Dry Solar Panel Cleaning Technologies (soil-type fit, water-use figures, vendor-reported restoration comparison)
2. Fortex Brush — The Brush for Solar Panel Cleaning Dilemma (moderate/severe scratching efficiency-loss estimates, wire-brush and dry-brushing guidance)
3. The Green Watt — How to Clean Solar Panels (panel surface temperature, thermal-shock mechanism)
4. Window Washing (South Africa) — Can You Pressure Wash Solar Panels? (Trina Solar / JA Solar / First Solar manufacturer pressure limits)

Short answer: use a robot when the array is large, high or unsafe to walk, and use long-reach electric brushes when sites are smaller, scattered or need spot cleaning. Most operators end up with both — a robot for the open field and brushes for the edges. Here’s how to decide.

Where manual (long-reach brush) wins

• Small and residential systems — for a rooftop or a 100–500 m² array, a single-head brush is light, cheap and precise.
• Edges, odd shapes and obstacles — brushes reach the rows, corners and gaps a robot can’t.
• Off-grid and remote sites — lithium versions run all day with no mains power, so there’s nothing to plug in.
• Mixed throughput needs — a double-head or roller widens the sweep to 0.7 m for medium and larger sites without stepping up to a robot.

Where a robot wins

• Large area, fast — a crawler robot covers far more per hour than any hand tool, so labour per m² drops sharply.
• Height and safety — it walks the array on its own, keeping crews off high or steep roofs; anti-drop edge detection stops it at gaps.
• Repeatable, scheduled cleaning — consistent coverage every cycle, less dependent on how tired the crew is.

The real deciding factor: labour cost

On most sites the biggest cleaning expense isn’t water or equipment — it’s people and the time they spend. That’s why the comparison usually comes down to area per operator-hour. A hand brush covers a modest area; a wide electric brush covers more with the same person; a robot covers the most and can run with minimal supervision. The larger and dirtier the site, the more that difference compounds over a year of cleaning cycles.

A simple way to choose

• Up to ~500 m², rooftop or scattered: single-head electric brush.
• ~500–6,000 m², routine soiling: double-head or roller brush.
• Large, high or hard-to-reach arrays: crawler robot, with brushes for the edges.
• Utility-scale ground farms: a robot built for long rows, backed by brushes.

Not sure which line you fall on? Send us your site size, module tilt and location and we’ll recommend a matched setup — we build the full range, so the advice isn’t biased toward one machine.

Dirty panels don’t fail loudly — they just quietly generate less every day. Depending on where your array sits, soiling can shave anywhere from a few percent to well over 20% off annual output, and in dusty or desert regions the losses climb fastest during exactly the sunniest, highest-value months.

What “soiling” actually is

Soiling is the build-up of dust, sand, pollen, salt, industrial fallout and bird droppings on the glass. It blocks light before it ever reaches the cells. Two things make it worse: low rainfall (nothing rinses the glass) and low tilt angle (dirt sits instead of sliding off). That’s why ground-mount farms in arid climates and low-slope commercial rooftops tend to suffer most.

How much output does it cost?

Field studies and site data across the industry generally point to a wide range:

• Temperate, rainy regions: often ~2–7% average annual loss — rain does a lot of the cleaning for you.
• Dry, dusty or agricultural regions: commonly 10–25% and rising through the dry season.
• Desert and heavy-industrial sites: can exceed 25–30% at peak if left uncleaned, with daily loss rates that add up fast.

The exact figure depends on your climate, tilt, nearby dust sources and rain pattern — which is why measuring soiling on your own site (rather than guessing) is increasingly common on utility-scale projects.

When does cleaning pay for itself?

The economics are simple: if recovered generation over a cleaning cycle is worth more than the labour, water and equipment to clean, cleaning pays. On larger or dustier sites that math usually favours regular, scheduled cleaning — and it favours methods that cut the biggest cost, which is almost always labour.

That’s the case for mechanised cleaning. A long-reach electric brush lets one person cover far more area per hour than a hand tool, and a crawler robot removes people from the array entirely on large or hard-to-reach roofs.

The takeaway

Soiling is a recurring tax on your yield, heaviest where sunlight is most valuable. The question isn’t whether to clean — it’s how to clean often enough, cheaply enough and safely enough to stay ahead of it. If you’d like help sizing the right method for your site, tell us your site size, tilt and location and we’ll point you to the right configuration.