Commercial solar O&M: what a maintenance scope covers

Commercial solar O&M (operations and maintenance) covers monitoring, scheduled inspection and testing, repairs, reporting, and who handles the roof and inverters. Two national benchmarks from 2025 put the annual cost at about $20 and $40 per kilowatt of direct-current (DC) capacity without storage, and neither explains the gap.^1,2

What should a commercial O&M scope list?

A scope should name every major component, the task done on each, how often, who performs it, and what report follows. The National Renewable Energy Laboratory (NREL) says to itemize measures by inverter type, mount type and roof or ground location instead of using a generic checklist.^3,4

NREL, Sandia National Laboratories and the SunSpec Alliance wrote that guide with an industry working group in December 2018. The U.S. Department of Energy’s Federal Energy Management Program (FEMP) still tells owners to follow it.^3,5 Its sample schedule for a commercial rooftop array shows how line items read. The sample is a template with blanks, so read it for wording, not for a requirement or a price.⁶

Corrective work sits outside that table. NREL says most contracts set a fixed cost for plant operations, standard preventive maintenance and an agreed response time, then add fees for corrective maintenance and non-covered services.⁷ The sample schedule lists a dispatch fee per alarm, with the amount left blank.⁶

Size changes the scope. For systems above 500 kilowatts, automated monitoring with alerts that trigger corrective work is a key consideration.⁸ For small commercial systems, on-site inspection falls to the party buying the power (the offtaker), and small size often rules out automated monitoring.⁸

Soleil Energy’s monitoring and O&M page lists performance monitoring, preventive maintenance of mounting, connections, inverters and combiner equipment, fault diagnosis and repair, fleet monitoring on one view, and production reports. It says Soleil services systems it did not install and agrees response times for each site. The utility-scale solar page lists long-term O&M once a plant is in commercial operation.⁹ With any provider, get test methods, performance commitments and response times in writing.

These tasks happen on live DC circuits. Testing inside combiner boxes calls for arc-flash protective equipment, and for some designs the hazard under NFPA 70E (the National Fire Protection Association’s Standard for Electrical Safety in the Workplace) may be too high to test live.¹⁰ NREL’s provider checklist asks for a licensed electrical contractor with NFPA 70E experience.¹¹

Why isn’t monitoring enough on its own?

Monitoring has built-in inaccuracies, so owners commonly add a second check of the DC array. Field electrical tests, including I-V curve traces, find defects that already cause significant loss. Infrared imaging can reveal hot spots, but it shows symptoms that crews must then diagnose on site.^10,12

Imaging conditions belong in the scope. NREL’s guide reproduces a 2016 vendor recommendation of irradiance of at least 600 watts per square meter during a flight, with irradiance varying by less than 100 and temperature by less than 5 degrees Celsius. Aerial imaging, it adds, can replace manual electrical testing in annual preventive maintenance and can also support commissioning and end-of-warranty inspections.¹²

Manual testing means opening combiner boxes and inverter enclosures. They must be resealed afterward, with gasket and screw torque, to keep the original enclosure rating.¹⁰ A bidder should be able to say how it documents that step.

What reports and records should the contract name?

A contract should name each report, its frequency and the data behind it. A federal request-for-proposal template ties performance-report frequency to array size, from biannual below 250 kilowatts to monthly above 3,000 kilowatts.⁵ NREL’s sample operations schedule adds daily data checks and reports against the committed availability.¹³

The table comes from a DOE FEMP page (last modified May 20, 2026) that quotes a federal template for government-owned systems. The page prints “biannually” without saying whether it means twice a year or every two years, so check the template before you copy the word into a contract.⁵

The sample operations schedule shows what a monitoring line item can contain. It lists daily data acquisition, a daily check of communications with escalation to the service provider, and reports comparing actual and expected production, actual availability against the committed level, and actual performance ratio against the committed level. It also lists the maintenance done and planned, plus incidents with resolution times.¹³

Records matter as much as reports. A provider has to know what equipment was installed, and every later change, before starting work. Keep warranties and the terms that keep them valid, past corrective work, insurance policies and proof of ownership on file.¹⁴

What does commercial solar O&M cost per year?

Public benchmarks run from about $20 to $40 per kilowatt DC per year for a commercial array without storage, depending on the source.^1,2 The National Laboratory of the Rockies’ 2025 baseline lists $20 in 2023 dollars.^2,15 A Department of Energy (DOE) 2025 benchmark for a 250-kilowatt rooftop system lists $40 in 2024 dollars.¹

Don’t read these rows as competing quotes. The third uses AC capacity, a narrower scope and plants above 5 megawatts.¹⁶ The first two use DC capacity and different dollar years.

Berkeley Lab also reports the trend for large plants: the median fell from about $40 per kilowatt AC per year in 2012 to about $11 in 2024.¹⁶

The DOE model behind the second row lists its parts: cleaning with vegetation and pest control, inspection of modules and electrical panels, repair labor for components outside warranty, replacement parts outside warranty, property tax, insurance, and asset management with security. The data file also notes a range of $34 to $46 for the no-storage case.¹⁷

Neither the baseline nor the benchmark explains why $20 and $40 differ, and both describe costs averaged over the system’s life.^1,2 The documents do hint at what moves a bid. The baseline says the wide range in reported prices comes partly from maintenance practices, that not every practice applies to every system, and that part quality and construction matter.²

Roofs and sites add their own costs. Rooftop repairs run higher on complex layouts, on steep roofs that need more safety equipment, and on damaged roof decks.¹⁸ For larger systems, cleaning, snow removal and tree trimming belong in the plan according to the site environment, such as dust, and the owner’s performance goals.⁸ When you compare bids, ask what the annual fee includes and what sits in a separate reserve.

O&M is one input to a project’s return. For the others, see the guides on commercial solar installation cost in Maryland and what commercial solar earns over its life.

What is a cost per kilowatt-hour for?

A cost per kilowatt-hour turns a yearly O&M bill into a cost per unit of energy, which lets you compare projects of different sizes. NREL says it climbs late in a system’s life as production falls, failures rise and inflation lifts unit costs.²⁰ For one array’s budget, use dollars per year or per kilowatt DC, not per kilowatt-hour.

How are availability and performance commitments written?

An availability or performance commitment ties a remedy to a measured result such as time operating, energy delivered or a performance ratio. NREL’s guide says each indicator must be defined in detail, usually against an independent yield report from commissioning, with events outside the provider’s control excluded.^21,7

Such a commitment is sometimes required when a third party owns the system, or when the O&M provider is not well known or lacks significant financial backing. Third parties, O&M contractors and insurers also offer these terms. Limits often come from the commissioning yield report. Because accuracy falls over time, NREL advises using the same model each year with measured weather.⁷

Two public reference points show the range. A site-wide operating contract at a military base, reproduced in NREL’s guide, required output of at least 80 percent of rated output, corrected for solar and temperature conditions, tested annually.²² FEMP separately says a performance ratio of 85 percent and availability of 95 percent should be achievable, which is an agency expectation, not contract language.⁵ Treat the first as sample wording and the second as a yardstick, not as terms any provider offers.

On response time, the 2018 guide says the contract should specify it, and that it “typically will be 10 days or less” for non-safety corrective work. Safety faults should be addressed as soon as possible, even when the recovered revenue is small.²³

Is paying for O&M worth it?

The federal data shows that arrays underperform and that long outages cost output. It can’t say what a maintenance contract recovers. In a 2021 federal review of 75 systems, measured output averaged about 75 percent of modeled output.²⁴ Availability averaged 95 percent, and production while running averaged 79 percent of the model.²⁴

• Average
• Median

• Availability95.1 percent98.0 percent
• Performance ratio (measured against modeled, while available)78.6 percent79.0 percent
• Energy ratio (all hours, measured against modeled)74.6 percent76.0 percent

View the data as a table

Measure in the federal sample	Average	Median
Availability	95.1 percent	98.0 percent
Performance ratio (measured against modeled, while available)	78.6 percent	79.0 percent
Energy ratio (all hours, measured against modeled)	74.6 percent	76.0 percent

The 75 systems average 410 kilowatts and were installed from 2011 to 2020.²⁴ Long, repair-delayed outages pull average availability below the median of 98 percent.²⁴ The report judges a performance ratio of 85 percent probably achievable through optimal O&M, “especially addressing the underperforming minority of systems.”²⁴ It also finds that arrays not evaluated through a routine O&M program often underperform, and that some sites lost several months of generation waiting for inverter repairs.²⁴

Those are site-staff observations, not a measure of any contract’s effect. The report calls full model output unrealistic and says 100 percent availability would be prohibitively expensive, so part of any gap to a model isn’t recoverable by maintenance.²⁴

To test a bid, divide the annual fee by the value of one year’s output. The result is the share of production the contract has to protect to break even. Use the rate the building pays for the energy the array offsets, not an all-in average rate. In the illustrative example above, a $5,000 bid and a $10,000 bid differ by $5,000 a year, and the owner needs the provider to protect at least that much extra output to justify the higher fee.¹⁹

The answer shifts with the situation:

Who handles the roof, the inverter and the utility?

NREL’s guide says roofs under warranty need annual preventive roof maintenance, and it recommends that the solar O&M provider and the roof maintenance provider meet so each understands the other’s duties.¹⁸ Three handoffs need an owner: the roof, the inverters and the utility.

Roof. A roof warranty may require an overburden waiver, an agreement to remove the array if the roofer needs access. Removal and reinstallation is costly and loses production. Settle who pays before it comes up.¹⁸

Inverters. Ten-year inverter warranties were commonly available when NREL’s 2018 guide was written, and the guide calls inverter failure one of the most frequent causes of performance loss, citing a 2010 Electric Power Research Institute report.²⁵ One 924-kilowatt carport system sat down for 2 years because the owner had no funds ready to replace a failed inverter; the guide recommends a reserve account or standby financing.²⁶ Decide before a failure who pays and who orders the part.

Utility. If a manufacturer makes up lost module power by adding modules, the larger nameplate capacity would generally trigger a new interconnection agreement and permitting, even though output wouldn’t rise.²⁵ Maryland also has a rule for replacement inverters. As of October 25, 2026, Code of Maryland Regulations (COMAR) 20.50.09.06(O) says a small generator facility requiring an inverter that submits an interconnection request after January 1, 2024 shall use a smart inverter with a utility-required settings profile. A facility may replace an existing inverter with a similar spare bought before that date, and the requirement doesn’t apply to replacements under manufacturer warranty.²⁷ The rule is silent on whether a planned swap needs notice to the utility, so get the utility’s answer in writing before the work. A Public Service Commission proposal to amend the chapter was published September 18, 2026, and the state’s table of pending proposals lists none as adopted as of October 25, 2026.²⁷ For the full process, see commercial solar interconnection in Maryland.

Before you sign: ten checks

Get the scope in writing, line by line. NREL’s guide lists the items a plan should carry: equipment and warranties, preventive tasks, alert procedures, availability and performance definitions, spare parts and an O&M budget that includes a reserve for major repairs.³

1. List every asset. Make, model, serial number and location of each inverter and combiner, tied to its warranty.³
2. Match tasks to warranties. Which preventive tasks keep each manufacturer’s warranty valid?^3,7
3. Name who monitors alerts. Who reviews alarms, how are faults ranked, and does a response mean a remote check or a site visit?²³
4. Define each committed measure. The formula, the data source, the model and the excluded events.^21,7
5. Find the exclusions. What the fixed fee leaves out, such as corrective labor, parts, dispatch fees and module washing.^7,6
6. Check who diagnoses faults. Root-cause analysis comes from qualified staff or from software, and it costs less when done before a crew is dispatched. Which does the bidder use?²⁸
7. Check qualifications. The provider checklist in NREL’s guide includes a licensed electrical contractor, experience with NFPA 70E and the solar sections of the National Electrical Code, qualified thermal-imaging staff and an equipment calibration program.¹¹
8. Check insurance and finances. Confirm current business insurance and ask for solvency references, because a failing provider hurts service quality.¹¹
9. Settle the data. Who owns the monitoring data, can you export raw data, and where does the maintenance log live?^3,29
10. Fund the reserve. Agree who pays when a major component fails, and where the money sits.²⁶

Where do these figures stop being useful?

The cost figures in this guide are national and dated, so they can’t price one building. The benchmarks are modeled, the guide is from 2018, and the federal systems are government-owned. Use them to build questions for bidders, then replace them with written quotes for your site.

• Not a local quote. None of the cost figures is a Maryland, Washington, DC or Northern Virginia price. The DOE benchmark is a national school rooftop, the baseline is modeled, and the Berkeley Lab figure covers plants above 5 megawatts AC.^1,2,16
• Different dollar years. The $20 figure is in 2023 dollars and the $40 figure in 2024 dollars.^15,1 The figures aren’t converted between years here.
• A federal fleet. The 75 systems are government-owned, average 410 kilowatts, and are compared with a model.²⁴ They aren’t private rooftops in this region.
• Storage and utility scale. Battery O&M and plants above 5 megawatts AC follow different plans and costs.^16,8

Your next step. Write your own scope from the table at the top, price it with at least two bidders, and use the public benchmarks only to spot a number that’s far out of line. If you own a commercial array and want its current scope reviewed, Request a system assessment or call (301) 953-4140.

This guide is general information, not legal, tax, or financial advice.

Sources (29)

1. U.S. Department of Energy, “Solar Photovoltaic System Cost Benchmarks,” energy.gov, published October 29, 2024, last modified September 2, 2026; Table 2 (2025Q1 PV-only commercial benchmark) and Table 1 (with storage); read October 25, 2026. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸

2. National Laboratory of the Rockies, “Commercial PV,” 2025 Annual Technology Baseline, atb.nlr.gov, operations and maintenance section; 2025 edition, read October 25, 2026. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸

3. National Renewable Energy Laboratory and partners, “Best Practices for Operation and Maintenance of Photovoltaic and Energy Storage Systems; 3rd Edition,” NREL/TP-7A40-73822, docs.nlr.gov, December 2018, sections 6.2 and 6.2.1 (printed pages 45 to 46); read October 25, 2026. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵ ↩¹⁶ ↩¹⁷ ↩¹⁸ ↩¹⁹ ↩²⁰ ↩²¹ ↩²² ↩²³ ↩²⁴

4. The National Renewable Energy Laboratory (NREL) published the 2018 guide under that name. The laboratory’s website and data now appear under the name National Laboratory of the Rockies at nlr.gov; current site, read October 25, 2026. ↩

5. U.S. Department of Energy, Federal Energy Management Program, “Life Cycle of Photovoltaic Systems: Operate and Maintain an Existing Photovoltaic System,” energy.gov, published August 25, 2025, page last modified May 20, 2026; read October 25, 2026. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸

6. Same source as ³, Appendix E, sample schedule of services for commercial rooftop installations (printed pages 126 to 128); December 2018. ↩ ↩² ↩³

7. Same source as ³, section 10.1 (printed pages 91 to 92); December 2018. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷

8. Same source as ³, sections 6.3.1.1, 6.3.2, 6.3.3 and 6.3.4 (printed pages 47 to 48); December 2018. ↩ ↩² ↩³ ↩⁴ ↩⁵

9. Soleil Energy, monitoring and O&M and utility-scale solar pages, current pages, read October 25, 2026. ↩

10. Same source as ³, sections 9.1 and 9.1.1 (printed pages 74 to 75); December 2018. ↩ ↩² ↩³ ↩⁴ ↩⁵

11. Same source as ³, sections 8.2.1, 8.3 and 8.4 (printed pages 70 to 72); December 2018. ↩ ↩² ↩³

12. Same source as ³, section 9.1.2 (printed pages 76 to 77); December 2018. ↩ ↩² ↩³

13. Same source as ³, Appendix E, example description of operations services (printed pages 131 to 135); December 2018. ↩ ↩²

14. Same source as ³, section 6.4 (printed page 49); December 2018. ↩

15. National Laboratory of the Rockies, Annual Technology Baseline, 2025 edition overview, atb.nlr.gov, “Monetary values are in 2023 USD”; current page, read October 25, 2026. ↩ ↩² ↩³

16. Lawrence Berkeley National Laboratory, “Utility-Scale Solar, 2025 Data Update,” slides, eta-publications.lbl.gov, October 2025, slides 30 and 35; read October 25, 2026. ↩ ↩² ↩³ ↩⁴ ↩⁵

17. U.S. Department of Energy and national laboratories, 2025Q1 PV System Cost Benchmark Model, commercial data file “CPV 2025Q1,” data.nlr.gov, Q1 2025 data; operations and maintenance entries; file downloaded October 25, 2026. ↩

18. Same source as ³, section 5.3 (printed pages 15 to 17); December 2018. ↩ ↩² ↩³

19. Illustrative calculation, run October 2026, with the inputs sourced at ¹ and ². ↩ ↩²

20. Same source as ³, section 6.8 (printed page 60) and section 11 (printed page 95); December 2018. ↩

21. Same source as ³, section 6.1.1 (printed pages 38 to 40); December 2018. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶

22. Same source as ³, example Performance Work Statement, Appendix E (printed page 126); December 2018. ↩

23. Same source as ³, section 6.7 (printed page 55); December 2018. ↩ ↩²

24. U.S. Department of Energy, Federal Energy Management Program, “Understanding Solar Photovoltaic System Performance: An Assessment of 75 Federal Photovoltaic Systems,” energy.gov, December 2021, executive summary and Table ES-1; read October 25, 2026. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹²

25. Same source as ³, sections 5.2 and 6.7.1 (printed pages 13 and 57 to 58); December 2018. ↩ ↩² ↩³

26. Same source as ³, sections 6.2.1, 6.7 and 11 (printed pages 46, 55 and 95); December 2018. ↩ ↩²

27. Maryland Division of State Documents, “COMAR 20.50.09.06 General Requirements,” regs.maryland.gov, paragraphs O(1) and O(2), read October 25, 2026 (the page shows no effective date). The Division’s Table of Pending Proposals, read October 25, 2026 (latest entries are from the October 16, 2026 Register), states that none of the listed proposals has been adopted and lists a Public Service Commission proposal to amend COMAR 20.50.09, published in the Maryland Register, volume 53, issue 19, September 18, 2026 (notice 26-143-P, issue text), which marks paragraphs O(1) through O(11) of regulation .06 as unchanged and adds a new paragraph O(12) on site-specific volt-watt settings; issue text read October 25, 2026. ↩ ↩²

28. Same source as ³, section 8.1 (printed page 70); December 2018. ↩

29. Same source as ³, sections 6.2.1, 9 (introduction) and 9.5 (printed pages 46, 74 and 85); December 2018. ↩