Compressed air leaks: finding them is easy, costing them is the point

Every engineer knows the system leaks. Most have a rough idea where. What almost nobody has is a number, and without a number the repair list competes for budget against things that do have numbers - and loses, every year, indefinitely.

Why leak surveys get ignored

A conventional survey produces a list: forty tagged locations, a severity rating, maybe a photograph. It goes to the person who controls the maintenance budget, who has no way to compare "forty leaks, some severe" against a pump that needs replacing and a guard rail that failed inspection.

So it gets deferred. Next year someone does another survey and produces a similar list, frequently including most of the same leaks.

What acoustic imaging changes

An acoustic camera uses an array of microphones to locate ultrasonic sound and overlays it on a visual image. You see the leak as a bright spot on the picture of the pipe, from a distance, with the plant running.

That in itself is a productivity improvement over walking the line with an ultrasonic probe. But the change that matters commercially is what modern instruments do next: estimate the leak rate and convert it into an annual cost and a CO₂ figure.

The output stops being a maintenance list and becomes a financial case, ranked, with the biggest leaks at the top. That is a document that gets approved.

Why compressed air specifically

Because it is the utility everyone treats as free. It is not - it is electricity converted at fairly poor efficiency into stored pressure, and a substantial share of the air a typical system generates never does any work. The leaks are distributed, individually small, and collectively expensive, which is exactly the profile that survives unexamined for years.

On a vessel it also has a second-order effect: compressors that run more than they should, wearing faster and consuming fuel to do it.

Beyond compressed air

The same instruments cover more ground than most people initially buy them for:

• Gas leaks, including on live plant where shutting down to investigate is not an option.
• Vacuum leaks, which are otherwise genuinely difficult to locate.
• Partial discharge in electrical equipment, a different failure mode that happens to produce a detectable acoustic signature.
• Flange and valve integrity checks as part of a routine round.

For operators with methane obligations, the regulatory driver has become an additional reason rather than the only one - worth reading alongside what the EU methane regulation covers, and what it does not.

Running a survey that actually works

A survey that becomes a programme

• Survey at normal operating load
• Locate and tag every leak
• Estimate each leak rate
• Annual cost and CO₂ per leak
• Rank by value, then fix
• Re-survey

↻ Step 6 feeds step 1: the re-survey proves the spend and becomes the next baseline.

• Survey under normal operating conditions. A quiet plant at low demand under-represents the problem.
• Cover the whole distribution system, not just the plant room. Leaks cluster at the point of use - couplings, hoses, tools.
• Record position properly. A leak nobody can find again has not been found.
• Re-survey after the fixes. This is the step that turns it into a programme instead of an event, and it is the one that proves the spend.

A note on expectations

Leak rate estimates are estimates. They depend on the instrument's model, the distance, the pressure and the geometry of the leak, and they are good enough to rank and to justify - not to invoice against to two decimal places. Present them as what they are and they will hold up. Present them as metered measurements and someone will eventually take them apart.