When One Inverter Fails, How Much of Your Plant Fails With It?

Two 1 MW rooftop plants, same modules, same city. Plant A runs twenty string inverters of 50 kW; Plant B runs one central unit. In year four, an inverter fails in each. Plant A loses 5% of generation for the week the spare takes to arrive. Plant B loses 100% — a week of the entire plant's output riding on one service call.

Neither architecture is "right". But the trade-offs are structural, they are set on day one, and they surface at exactly the moments — failures, expansions, year-12 replacements — when changing course is most expensive.

The three architectures, honestly compared

The Indian C&I reality: string architecture has eaten most of the rooftop market precisely because availability and serviceability dominate economics at rooftop scale — a shelf spare and a two-hour swap beat any service SLA. Central units still win on very large uniform arrays where their per-kW cost and simpler DC field pay for proper on-site service arrangements.

Selection questions that actually decide it

• What does a week of downtime cost you? Multiply your plant’s daily generation value by a realistic service lead time — that number, weighed against the architecture, is the availability argument in rupees.
• Is the roof uniform? Multiple orientations, parapet shadows, future rooftop equipment — each pushes toward more MPPTs (string) or, in extremes, module-level electronics.
• Grid compliance is non-negotiable either way: whatever you choose must meet your DISCOM’s interconnection and protection requirements — anti-islanding, ride-through and power-quality behaviour per the applicable standards your state enforces. This is a checklist item at approval, not a preference.
• DC/AC ratio and clipping: modern designs deliberately oversize the array relative to inverter capacity; sensible ratios trade tiny midday clipping for better capacity utilisation. Ask to see the clipping estimate in the yield simulation, not just the ratio.
• Who stocks the spare? An EPC promising “48-hour replacement” should be asked: from which warehouse, and is it in the contract?

The year-12 question nobody prices on day one

Inverters are the plant’s wear item — design lives of 10–15 years against modules warranted for 25. Whatever you install today will likely be replaced once within the plant’s life:

• String units: replacement is granular and budgetable — a few units a year as they age out, at whatever the market then charges (budget in real terms; treat vendor “extended warranty” pricing as the ceiling, not the plan).
• Central units: one large, lumpy capex event — and a compatibility question if the model family is discontinued. Confirm the frame footprint and DC field compatibility of plausible successors.
• Either way, write the replacement into the financial model — a plant IRR that omits it is fiction.

Get the plant’s basic electrical sizing right first — panel count, inverter rating and DC/AC ratio — with our Solar Panel & Inverter calculator, then pressure-test the economics (including that year-12 line) in the full Solar Financial Model.

What we do differently

As an EPC, our solar practice is architecture-agnostic — we install string and central plants both, so the recommendation follows your roof, your downtime economics and your DISCOM’s rulebook, not a distributorship. Designs ship with string-level monitoring, contractual spare arrangements, and a financial model that already contains the inverter’s second life.

The three takeaways

• Choose architecture by failure blast-radius and service reality, not brochure efficiency decimals.
• Uniform roofs widen your options; messy roofs demand MPPT granularity.
• The year-12 replacement is part of the price — any model that hides it is flattering the IRR.

Comparing two solar offers right now? Book a Free Project Blueprint & Statutory Approvals Roadmap or call +91 70099 85817 — we’ll show you where the two BOQs will behave differently in year four.