What maintenance does a commercial solar system actually need?

O&M scope is set by contract, not by a regulated standard, so it varies between providers and you have to read what is in and what is out. That said, a commercial O&M arrangement typically includes a recognisable set of activities:

• Remote performance monitoring and alerting. Continuous tracking of generation against the expected yield, with alerts when a string drops out, an inverter faults or output sags.
• Scheduled preventive inspections. Periodic visual checks, thermographic or infrared scanning of modules and connections to find hot spots, and electrical testing.
• Inverter servicing and fault rectification. Firmware, fans and filters where applicable, and putting right faults the monitoring flags.
• Panel cleaning. Scheduled or as-needed, set by how the site soils rather than a fixed calendar.
• Vegetation and structural checks. Keeping growth off ground-mount rows and confirming the mounting and fixings remain sound.
• Reactive call-outs with a defined response time, and reporting.

Where contracts differ most is on the expensive items: whether spare parts, a full inverter replacement and a performance or availability guarantee are inside the fee or charged on top. Check exactly which of those are in scope before comparing one quote with another, because a cheaper headline fee can simply mean the big-ticket items sit outside it.

Knowing what an O&M arrangement usually covers is one thing. Judging the specific contract in front of you is another. The same three letters cover everything from an annual photograph of the inverter to a properly scoped programme of inspection, monitoring and agreed response, which is why two agreements at the same rate can be worth wildly different amounts to you. The label tells you almost nothing. The clauses tell you everything.

The scope clause is where that difference sits, and a vague one is where disputes are born. A clause that promises regular maintenance and monitoring in general terms has told you nothing you can hold anyone to. A sound scope names the work, names how often it happens, and names what lands on your desk afterwards. On an asset over 50 kWp, read the scope for each of these by name and treat an absence as a question rather than an assumption.

• Monitoring, and who is contracted to act on it. What is watched, how alerts are raised, and whose job it is to respond. Monitoring that nobody is obliged to act on is a dashboard rather than a service.
• Inspection, at a stated frequency. A physical site visit at a named interval covering the panels, cabling, isolators, inverters, earthing and any battery units, because those faults are found by hand rather than through a portal.
• Electrical testing, referenced to a standard. The periodic verification the asset needs to stay safe and compliant, written against the standard it is carried out to rather than left implied.
• Cleaning, assessed rather than assumed. Set against how your roof actually soils and how it is accessed, so nobody is billing to clean a roof the rain keeps clean.
• Fault response. What happens when something breaks, on what terms, and at whose cost.
• Reporting. A written record after each visit, so the asset is documented and you can see what was done.

Every O&M contract blends two kinds of work, and the split between them is how you read what you are actually buying. Preventive maintenance is the planned work that stops faults happening: the scheduled inspections, the electrical testing, the settings checks, the cleaning where it earns its place. It is predictable, it sits on a calendar, and it is what a fixed annual fee normally covers.

Corrective maintenance is the work that puts a fault right once it has occurred: diagnosing a failed inverter, replacing a damaged isolator, tracing a string that has dropped offline. This is where contracts diverge sharply. Some include corrective labour up to a stated limit and charge parts and major component replacement separately. Others include a diagnostic visit and quote the repair once the cause is known. Neither approach is wrong, though you need to know which one you have signed, because an agreement that looks cheap on the preventive line can carry an open-ended corrective bill behind it.

The question to put to any provider is a plain one: what does the fee cover, and what is charged on top. A sound contract answers it in writing, and the answer is worth having before the first fault rather than after it.

Two clauses separate a contract you can manage an asset against from one you simply pay. The first is reporting. After each scheduled visit you should receive a written report recording the condition of the system, the generation against what was expected, anything found, and a prioritised list of what needs doing and how soon. That report is what lets a finance director see the truth about the asset, and it doubles as the evidence of maintenance that insurers and landlords often ask for.

The second is response. A serious contract sets out what happens when a fault is raised: how it is triaged, the order in which work is prioritised, and the route from alert to attendance. Read those terms as commitments agreed for your asset and written into your agreement, rather than a promise lifted off a marketing page. Treat a headline uptime percentage or a fixed response time advertised before anyone has seen your site with caution, because a figure that has not been scoped to your asset may not survive contact with it. What matters is that the response terms are named, agreed and yours.

Two more clauses decide what the agreement is really worth. The first is the term. Read the initial period, the renewal basis and the notice you have to give to end it together, rather than the headline duration on its own, because an attractive rate locked to an auto-renewing multi-year term with a long notice window is a different commitment from the same rate on an annual one. Visit cadence belongs in the same read: the inspection frequency should be named and fixed to your asset rather than left to the provider's discretion, since the periodic testing and the warranty conditions both depend on the work happening on a stated schedule.

Warranty obligations are the part buyers most often miss. Manufacturer warranties on panels, inverters and batteries usually carry conditions: maintenance carried out and recorded, settings kept within stated limits, faults reported within a window. A contract that does not keep those conditions met can quietly cost you a warranty claim years later, when a component fails and the paperwork behind it will not support the claim. That is the argument for a scope cut to your asset rather than dropped in from a template, and it is why the cleanest moment to put a contract in place is at handover, so the maintenance the warranty depends on starts the day the system is signed off. For an existing or inherited system there is no waiting for an anniversary; the contract starts once the scope is settled. How that scope is built for a system we maintain is set out on our operations and maintenance page.

The panels and the mounting structure are commonly rated for 25 to 30 years or more. The inverter is not. String and central inverters are the shortest-lived major component, typically rated for around 10 to 15 years, with standard product warranties commonly running 10 to 12 years and extendable, at a cost, to roughly 20 to 25 years. Financial models for a solar asset routinely budget for at least one full inverter replacement over the project life, commonly modelled somewhere around year 10 to 15.

The honest framing for a finance director is that this is the single largest planned mid-life capital cost on the system, and it should be sinking-funded rather than treated as a surprise. The exact timing and cost depend on the specific inverter, how hard it is run and how the contract handles replacement, so we do not put a fixed year or figure on a live page. The principle is fixed: budget for the replacement from the start so it is a planned event, not a shock to the model. The structural survey and the design set which inverter goes in; the O&M plan sets how its end of life is funded.

Continuous monitoring is the backbone of commercial O&M, because it is what turns maintenance from reactive break-fix into early intervention. Inverter-level monitoring is the minimum; string-level or module and optimiser-level monitoring localises a problem faster. The standard metrics are the performance ratio and expected-versus-actual yield, and they flag the failure modes that otherwise go unseen: a string that has silently dropped out, an inverter fault, a soiling-driven slide in output, or degradation running faster than it should.

That last point matters because not all yield loss is a fault. Panels degrade slowly and predictably. Analysis of large fleets of installed systems by the US National Renewable Energy Laboratory puts the median module degradation at around 0.5% per year, with premium n-type cells often lower, so a panel degrading at that median rate is still producing around 88% of its original output at year 25. That is the gentle, expected decline that 25-year performance warranties are written around. The job of monitoring is to separate that slow, normal fade from the sudden or abnormal underperformance that signals a real problem, and to surface it while it is cheap to fix. If you suspect your existing system is underperforming, our performance audit covers how that is diagnosed.

Monitoring and the inverter are the line items a finance director budgets for. The condition of the array itself is what an engineer finds when they get on a roof that nobody has walked across since the last visit, and it is worth knowing what that is, because it is the part that separates a real maintenance plan from a monitoring contract with a cleaning add-on. A few faults turn up on commercial roofs again and again.

Soiling that the roof will not self-clean. This is the slow erosion of yield that nobody notices happening. International PV-reliability research from the IEA-PVPS programme puts global soiling losses at roughly 3 to 5% of annual energy production, and UK-specific work from Loughborough University has found it can cut output by more than 5% here. A shallow-pitch array does not shed dirt the way a steeper roof does, because the rain does not sheet across it hard enough to carry the soiling off, so output drifts down slowly without ever registering as a fault. Coastal sites are worse again: salt and more airborne soiling dirty the glass faster and the rain clears less of it, and agricultural or industrial dust and heavy pollen push the same way. UK rain does a share of the cleaning for you, which is why there is no universal schedule worth publishing: the interval is set by how your site actually soils, read off the monitoring trend rather than off a calendar. The mechanics of commercial cleaning are covered in the commercial solar cleaning guide.

Bird fouling and broken modules. Droppings and dead birds soil the glass and shade the cells beneath them, and a shaded cell in a working string runs hot, which over time damages the panel. Modules also crack and shatter, from impact, hail, thermal cycling and the microcracks that spread from it, and a damaged panel can keep generating while it underperforms and admits water. A visit clears the fouling, inspects for the hot-spot damage it leaves, and identifies the cracked modules for replacement. Because a single failed panel rarely announces itself, a periodic performance audit is what traces the lost yield back to the module behind it.

Everything above assumes you know what you have. Plenty of commercial buyers do not, because the array arrived with the building, or the firm that installed it has since disappeared and taken the handover pack with it. That is a different starting point from a system you commissioned yourself, and it changes the order of the work: the audit comes before the contract, because nobody can price the care of an asset whose condition is unknown.

The practical first step is an as-built audit and condition survey. That means establishing what was installed and its age, confirming whether monitoring exists and whether you can get credentialled access to it, running a performance baseline against expected yield, infrared-scanning for hot spots and failed bypass diodes, checking the inverter age against its replacement horizon, and verifying the grid-connection paperwork, including the network operator records behind the G99 connection agreement and any export registration. Older systems commonly turn up lost monitoring access, an out-of-warranty inverter near the end of its life, and degraded or mismatched strings. Those findings define the remediation scope, which is settled before an ongoing maintenance arrangement is priced. A commercial system over 50 kWp sits outside the domestic MCS scheme, so this audit, rather than a certificate, is what tells you the real condition of the asset. Where the original installer has gone altogether, our system takeover page sets out how an orphaned array is adopted and carried forward.