Somewhere on your plant, right now, a hole the size of a grain of rice is hissing. You have walked past it so many times you no longer hear it. This lesson does one thing: it turns that hiss into rupees per year, so it can compete for attention with everything else on the maintenance list — and win.
The physics: every leak is a little rocket nozzle
Above roughly 1 kg/cm²g, steam escaping through a hole reaches the speed of sound in the hole itself — the flow "chokes". Choked flow has a beautiful property: it no longer matters what is downstream. The leak rate depends only on the area of the hole and the absolute pressure behind it. (The same physics, pointed at a diffuser instead of the sky, powers the thermocompressor deep-dive — a leak is recovery running in reverse.)
The working estimate, descended from Napier's nineteenth-century orifice experiments:
Real leaks are ragged slots, not drilled orifices, so treat results as estimates — good to a few tens of percent, which is all a priority list needs.
The price list
| Hole Ø | at 3.5 kg/cm²g | at 7 kg/cm²g | at 10.5 kg/cm²g | ₹/year at 10.5* |
|---|---|---|---|---|
| 1 mm | 1.8 kg/h | 3.1 kg/h | 4.4 kg/h | ₹19,000 |
| 2 mm | 7.1 kg/h | 12.4 kg/h | 17.8 kg/h | ₹78,000 |
| 3 mm | 15.9 kg/h | 27.9 kg/h | 40 kg/h | ₹1.75 lakh |
| 5 mm | 44 kg/h | 78 kg/h | 111 kg/h | ₹4.9 lakh |
*7,300 running hours, steam at ₹600/tonne fuel-only cost (the practice plant's number — the true-cost lesson argues it should be higher, which makes every figure here conservative).
Now the multiplier: a mid-size plant does not have one leak. A first-ever leak survey on an unaudited plant typically tags dozens — glands, flanges, unions, corroded drips — plus the leaks you cannot hear at all: steam traps failed open, each one a permanent 2–4 mm leak into the condensate line. Traps get their own lesson, and their own survey discipline.
How to survey a plant for leaks
- Walk at pressure, walk quiet. Survey during running hours (leaks close when lines cool) but during a lull — lunch, shift change — when the plant is audible.
- Ears first, then eyes. Hissing finds most of it. Visible plumes at glands, flanges and unions find more. An ultrasonic gun (₹30–80k) finds the rest and works in noisy plants.
- Tag and size on the spot. Estimate every find against the table above — a 2 mm equivalent is a fair default for an audible gland leak. Photograph, tag, log: location, line pressure, estimated Ø.
- Rank by rupees, fix in order. The table converts the tag list to a priority list. Most fixes are gaskets, gland packing and union re-seats — maintenance consumables against lakhs.
- Repeat twice a year. Leak populations regrow; the second survey is always faster.
- Choked flow: loss ≈ 0.5 × area(mm²) × bar(a). Pressure and hole area — nothing else matters.
- A 3 mm leak on a 10.5 kg/cm²g header ≈ ₹1.75 lakh/year at fuel-only value.
- Failed-open traps are silent leaks — survey them with the same seriousness.
- Survey → tag → price → rank → fix → repeat. Twice a year.
The choked-flow rule this lesson runs on predates the aeroplane: it comes out of 1860s-era experiments on steam escaping through orifices, tabulated by engineers who wanted to size safety valves for boilers that kept exploding. The same physics was later pointed downward instead of outward — and became the rocket nozzle. Your leaking gland, the boiler safety valve and a launch vehicle all obey the same one-line equation.
FAQ
Why doesn't the leak rate depend on the air outside the hole?
Because above roughly 1 kg/cm²g the flow chokes: the steam reaches the speed of sound inside the hole itself, and downstream conditions can no longer talk back upstream. From then on only two things set the loss — the area of the hole and the absolute pressure behind it. That is what makes the one-line estimate possible.
How do I size a leak with no instruments at all?
Use the working estimate: loss (kg/h) ≈ 0.5 × hole area (mm²) × absolute pressure (bar a). For an audible gland leak, a 2 mm equivalent is a fair default; a visible plume deserves 3 mm or more. Real leaks are ragged slots, so treat the answer as good to a few tens of percent — which is all a priority list needs.
Is chasing leaks worth it when steam costs "only" ₹600 a tonne?
One 3 mm gland leak on the 10.5 kg/cm²g header is 292 t of steam a year — about ₹1.75 lakh at fuel-only value, against a repack costing a few thousand. And ₹600/t is the conservative number: add water and treatment for the make-up and the figure passes ₹2 lakh. F1 argues the true rate is higher still.
When is the best time to walk a leak survey?
During running hours, at a lull — lunch or shift change. The plant must be at pressure, because leaks close as lines cool; and it must be quiet enough to hear the hissing that finds most of the list. Ears first, then eyes for plumes; an ultrasonic gun covers the plants that are never quiet.
What about the leaks I cannot hear?
The biggest ones are silent by design: steam traps failed open, each a permanent 2–4 mm leak discharging invisibly into the condensate line. They never hiss at you, which is why traps get their own survey discipline in G2. An ultrasonic gun also finds gasket weeps that ears miss.
Check yourself
Five quick questions on this lesson. No marks, no records — the score is for you.
