What One Hour of Downtime Actually Costs an Indian Factory (₹)

A stopped hour costs far more than the repair bill. Add four things running at once: the margin on parts you didn't make, the wages of operators standing idle, the fixed overhead that keeps burning, and — if you're sold out — the orders you can't recover. For a mid-size Indian line the honest number usually lands between ₹5,000 and ₹15,000 per hour (roughly US$50–160 at ~₹95/US$), and almost none of it appears as a single line in any report. Below are indicative figures by factory type and a formula you can plug your own numbers into.

Key points

• One stopped hour = four costs at once, not just the maintenance ticket: lost contribution margin + idle labour + fixed overhead + unrecoverable orders.
• Bill lost output at contribution margin (price − variable cost), not full price — otherwise you overstate the loss, because you never bought the material for a part you never made.
• Indicative range for a mid-size Indian plant: ~₹5,000–15,000/hour, dominated almost everywhere by lost margin, not the idle labour everyone watches.
• One question swings the number 5–10×: can you catch up? Sold out at your bottleneck → full margin gone for good. Can recover on overtime → you only pay the overtime premium plus overhead.
• The stopwatch part is cheap; counting every stop is the hard part. Micro-stops and slow running — the biggest bucket — never reach a manual log, so the annual total is always understated.
• The global headline figures (Siemens: ~US$1.4 trillion/year across the world's 500 largest firms; up to US$2.3 million/hour for an auto line) describe giants — use them for direction, then compute your rupee number.

The four things a stopped hour costs

When a machine stops, most owners picture the repair. That's the cheapest part. The real hourly loss is four costs running in parallel:

Downtime cost per hour = (lost good units × contribution margin) + idle labour + fixed overhead + unrecoverable orders

1. Lost contribution margin. Every good unit not made is margin gone — sale price minus the variable cost (material, consumables, per-unit power) you didn't spend. Use margin, not price, or your number will read high. This is almost always the biggest of the four.
2. Idle labour. Operators, helpers and the line supervisor are paid whether the line runs or not. In an Indian plant this is the smallest line — cheap labour is exactly why owners wave the whole stop away.
3. Fixed overhead that doesn't pause. Rent, machine depreciation, supervision, interest, and the demand charge on your sanctioned kVA keep accruing while nothing ships. A slowed machine draws less energy, but the fixed capacity charge doesn't move.
4. Unrecoverable orders. This one is situational and can dwarf the other three: an OEM line-stop penalty, premium air freight to recover a schedule, a perishable batch spoiled, a pharma batch lost to a validation break. It applies only when the stop touches a customer commitment or a time-sensitive product.

Before you trust any total, answer the one question every quick calculator skips — can you catch up? If you're sold out or the stopped machine is your bottleneck, the parts are gone for good and full margin applies. If you can recover next week on overtime, the real cost is the overtime premium plus overhead, not full margin. If the stopped machine is a non-bottleneck that never starves the constraint, an hour of it can cost close to ₹0. Decide which case you're in first — the deeper method is in the real cost of factory downtime and how to measure it.

₹ per hour by factory type (indicative)

The figures below are illustrative September 2026 estimates for a mid-size plant with one machine cell or line stopped, in the sold-out / bottleneck case (full margin). They exist so you can see the shape and rough magnitude — plug in your own margin, crew and overhead before quoting any number. All assume ~₹95/US$ (USD/INR ~₹94.8, 8 Sep 2026, Trading Economics).

Two things jump out of the table. First, the ranking is driven almost entirely by lost margin and fixed overhead — the idle-labour column, the one owners actually watch, is the smallest everywhere. That's the trap in Indian plants: labour is cheap, so a stop feels cheap, while the expensive part (unsold margin) is the part nobody is looking at. Second, the situational add-on column is often the largest number of all — an OEM penalty or a spoiled batch can exceed a whole shift of on-floor loss — but it only fires when the stop touches a commitment.

Labour is loaded from India's Annual Survey of Industries average factory wage of about ₹563/day in 2021–22 (CEDA, Ashoka University, on ASI data), grossed up for statutory contributions and supervision. The fixed-overhead line includes the sanctioned-kVA demand charge — large industrial energy in a state like Uttar Pradesh runs around ₹6.1–7.1/kVAh for FY2026 (Mercom on the UPERC tariff order) — which is why a foundry, holding a furnace hot with nothing to forge, tops the table. Idle-but-energised machines still pull power; see energy waste from idle machines.

Want your own figure instead of ours? Put your margin, crew and overhead into the downtime cost calculator and it returns your ₹/hour in seconds.

Annualise it — the number that makes the case

One hour looks survivable. The annual total is what changes minds. Take the auto-components cell above at ~₹12,400/hour. Run two shifts, 26 days a month (~416 planned hours), and lose a conservative 10% to downtime — breakdowns, changeover overruns, material waits and micro-stops. That's ~42 lost hours a month:

• 42 × ₹12,400 ≈ ₹5.2 lakh/month ≈ ₹62 lakh/year (~US$65k) — on one cell.

Even if your real inputs are half of these, the point holds: downtime is a seven-figure annual line item that never appears as one line. For context on why the global headline numbers (Siemens' ~US$1.4 trillion/year, up to US$2.3 million/hour for an auto line — Siemens/Senseye, True Cost of Downtime 2024) don't transfer to a Rajkot foundry, see factory downtime, theft & accident statistics for India.

What a stopped hour shows — and what it hides

Here's the honest limit on all of the above: you can only cost what you counted. And most plants count only a fraction of what stops.

The loud breakdowns — the ones that make it into the shift book — are usually the smallest bucket. Micro-stops and slow running are the biggest and the least recorded. That's how a plant swears it has "a couple of breakdowns a month" and still runs at 55% OEE. If you're measuring downtime as a ratio rather than in rupees, OEE explained for Indian factories walks through the same losses as a percentage.

What you need to measure to get a real number

To turn the formula into your number, you need four inputs and one honest count:

1. Ideal good output per hour for the cell or line — its demonstrated best rate, not the nameplate.
2. Contribution margin per unit — sale price minus variable cost.
3. Crew and fully-loaded labour rate for that cell.
4. Overhead allocated to it per hour, including the fixed demand charge.
5. A two-week honest count of all stops — micro-stops and slow running included, not just breakdowns.

The first four you already know or can find in an afternoon. The fifth is where every estimate goes wrong, because a person cannot stand at every cell with a stopwatch. A camera can watch one continuously — an unmanned station, a cell idle while an operator hunts for a trolley, a changeover that overran, a machine cycling below rate — and timestamp every transition. Those are exactly the Availability and Performance losses a manual log drops. Most plants already bought cameras for theft and security, so this is a sunk asset you can re-read for downtime with no new capital. The mechanics are in spotting hidden productivity loss with the cameras you already have.

This is the wedge Mama is built around: record a short phone walkthrough of your floor, get back a camera placement plan for the zones where downtime actually accrues, then read those feeds into a plain-language efficiency summary — so the invisible 40-minute changeover starts carrying the ₹/hour tag from the formula above.

Do this today

1. Pick your worst cell or line and note its ideal output/hour and margin/unit.
2. Decide the fork — sold out (full margin) or can catch up (overtime premium).
3. Drop the numbers into the downtime calculator to get your ₹/hour.
4. Count every stop for two weeks, then annualise. That figure is your business case.

The measurement is cheap. Not measuring is the expensive part.

FAQ

How much does one hour of downtime cost an Indian factory? For a mid-size plant, an indicative ₹5,000–15,000 per hour (~US$50–160 at ~₹95/US$) in the sold-out case, depending on margin, crew and overhead. The number is dominated by lost contribution margin, not idle labour. A foundry or forging shop tends to sit at the top because a furnace held hot burns fixed energy with nothing to show; a small plastics cell sits lower. Compute your own with the four-part formula rather than borrowing a figure.

What is the formula for the cost of downtime per hour? Cost per hour = (lost good units × contribution margin per unit) + idle labour per hour + fixed overhead per hour + unrecoverable orders. Use contribution margin (price minus variable cost), not full price, and add the fourth term only when the stop touches a customer commitment or a perishable/validated product.

Why is lost margin bigger than idle labour if labour is cheap? Precisely because labour is cheap. Six operators at ₹120/hour cost ₹720 while idle; the 120 unsold parts at ₹85 margin cost ₹10,200 — about 14× more. The cheap labour is why the stop gets waved off, and the expensive part, the margin, is the part nobody is looking at.

Does the repair or maintenance cost count? Yes, but it's usually the smallest piece. Add the actual repair spend on top of the four components — plus scrap and warm-up rejects at restart. The point of costing downtime by the hour is to show that the on-floor loss dwarfs the repair bill that owners fixate on.

Can I measure my true hourly cost without an expensive MES? Partly. The four inputs (output rate, margin, labour, overhead) you already have. The hard part — an honest, continuous count of every stop including micro-stops — can come from a camera watching machine state, provided its sightline to a running-state cue is clear. That's a practical start before any full MES project.