Solar AMC & Monitoring: Soiling, Degradation, DC Faults & Performance Ratio

A solar plant looks like a static asset. It isn't. It loses output the moment it's commissioned, and the rate of loss is almost entirely a function of how it's maintained. If you already own a system, this is the article you wish your installer had handed you on day one.

Soiling: the biggest avoidable loss on Indian rooftops

Soiling is the accumulation of dust, bird droppings, pollen and pollution film on the panel glass. Across most of India:

• Typical soiling loss without cleaning: 0.6-1.2% per week in dry months, and higher in the dust belt across Rajasthan, Gujarat, Delhi-NCR and the Indo-Gangetic plain
• Peak loss spikes after a dust storm or a long dry spell: a single film of fine dust can cut output 8-15% overnight
• Bird droppings are the worst - they cause localised hotspots, not just shading

Cleaning cycle that actually works in Indian conditions:

• Every 21-30 days through the dry season (roughly Nov-May in most states)
• Every 45-60 days in monsoon (rain does partial cleaning, but leaves streak deposits)
• Tighten to every 10-15 days near cement plants, stone crushers, highways and coal belts
• Demineralised or RO water only. Hard borewell water leaves a calcium film worse than the dust it removes.
• Soft brush, top-down, early morning or evening. Never mid-day on hot glass (thermal shock cracks cells).

A disciplined cleaning programme recovers 3-6% of annual generation. On a 30 kWp plant at ₹10/unit, that's ₹13,000-₹27,000 a year, easily - usually more than the AMC fee itself.

Degradation: the loss you can't avoid (but should monitor)

Two clocks tick the moment the plant is commissioned:

Panel degradation

• Year 1: 2-2.5% (first-year light-induced + LeTID losses, baked into the warranty)
• Years 2-25: 0.4-0.55% per year linear for Tier-1 mono PERC; 0.3-0.4% per year for TOPCon
• End-of-warranty (Year 25) output: typically guaranteed at 84-87% of nameplate

Inverter degradation / failure

• String inverters: MTBF ~10-12 years. Plan a replacement somewhere in years 10-13.
• Capacitor ageing is the most common failure mode - manifests as derating, then nuisance trips, then hard fault
• Firmware matters: outdated firmware causes phantom anti-islanding trips after grid disturbances on the DISCOM side
• Always keep a spare inverter of the same model on the AMC contract for fleets of 5+ plants

A well-maintained 30 kWp plant should still be producing 88-90% of its first-year output at the 10-year mark. If you're below that, something is wrong that cleaning won't fix.

DC-side faults: the ones that hide in plain sight

The DC side (panels → combiner → inverter input) is responsible for the majority of unexplained output loss. The usual suspects:

• MC4 connector corrosion / loose crimps - common after monsoons and on coastal sites; causes intermittent string dropouts and arc-fault risk
• Hotspots from partial shading, bird droppings, or cell-level micro-cracks - show up as visible burns on thermal imaging
• PID (Potential Induced Degradation) - voltage-stress related, more common on older non-PID-free panels and ungrounded systems, and accelerated by high humidity
• Diode failure in junction boxes - silently kills a third of a panel's output without tripping the inverter
• Earth faults on the DC array - inverter logs them; many plant operators ignore the log

What catches these:

• Annual IV-curve tracing on every string
• Annual thermal imaging of all panels, junction boxes and MC4 connectors
• Monthly string-current comparison in monitoring (one outlier = one site visit)

Performance Ratio (PR) is the only metric that matters

Everything else - units generated, ₹ saved, panel cleanliness - is a derived number. The single metric that tells you whether your plant is healthy is the Performance Ratio:

PR = Actual generation / (Plant capacity × Plane-of-array irradiance × time)

It strips out weather and tells you what fraction of the theoretically available energy you actually captured.

Benchmarks for a typical Indian rooftop plant:

• New plant, Year 1, clean: PR ≥ 78-80%
• Year 5, well-maintained: PR ≥ 76%
• Year 10, well-maintained: PR ≥ 73-74%
• Anything below 70% on a clean, sunny day = active fault

Hot-climate sites (Rajasthan, Vidarbha, interior Telangana and Andhra) run 1-2 points lower on thermal derating alone; that's expected, not a fault.

Most monitoring platforms calculate PR daily and monthly. If yours doesn't, change platforms - or insist your AMC provider reports it. "We generated X units this month" without a PR is meaningless: a bad month of weather can look the same as a half-failed inverter.

What a real AMC contract should include

• Scheduled cleaning to the cycle above, with photo evidence each visit
• Monthly generation + PR report in plain language
• Annual IV-curve test and thermal imaging
• Annual earthing, SPD, LA and DCDB/ACDB inspection (pre-monsoon)
• Inverter firmware updates as released
• Response SLAs in writing: remote ack < 4 hours; site visit for inverter fault < 24 hours; inverter swap from spare < 72 hours
• Generation guarantee or PR guarantee, with compensation terms if missed

If your installer can't commit to those in writing, you don't have an AMC, you have a phone number.