Compressed air monitoring and leak detection

TL;DR — Electricity, water and effluent get meters. Compressed air almost never does, which is why it is usually the most wasteful system in the plant: a fifth to a third of what the compressor makes escapes through leaks, and the header runs a bar or two higher than anything on the floor needs. Four signals — compressor kW, header flow, pressure at the far end, and dryer dew point — turn all of that into numbers. The single most useful one is kWh per Nm³, and the cheapest leak audit in existence is a flow meter left logging over a weekend. On the Highveld there is a fifth problem nobody mentions: at Johannesburg’s altitude a compressor draws air at about 82 kPa instead of 101 kPa, so a machine rated at sea level delivers roughly a fifth less mass of air than its badge implies.

Why compressed air is the expensive one

Compressed air feels free because it comes out of a pipe. It is not. Compression is thermodynamically poor: most of the electrical energy that goes into a screw compressor leaves as heat, and only a modest fraction arrives at the tool as useful work. That makes air, per unit of work delivered, the dearest utility on the site — which is exactly why leaving it unmeasured is expensive in a way that leaving, say, lighting unmeasured is not.

The waste is also structural rather than accidental. Two patterns show up in field reporting again and again. The first is oversizing: as one industry source in our research corpus puts it — directional, not audited — “oversized systems can waste an estimated 10% to 30% of energy before the compressor even reaches productive loading.” A compressor specified for the plant someone hoped to build spends its life part-loaded and short-cycling, drawing meaningful power while producing nothing. The second is leakage: “leaks alone can waste up to 30% of compressed air output.” A third source names the whole set in one line: “excess pressure, distribution losses, inefficient control, and unrecovered heat.”

Every one of those four is invisible without instrumentation, and every one of them is continuous. Unlike a breakdown, none of it announces itself.

The Highveld problem: your compressor is already derated

This one is specific to where most South African manufacturing actually sits, and it is routinely missed at specification time. A compressor is a volumetric machine — it swallows a fixed volume per revolution. What matters to your process is the mass of air in that volume, and mass depends on inlet pressure.

At sea level the standard atmosphere is about 101 kPa. Johannesburg sits at roughly 1,750 m, where it is about 82 kPa — near enough to 81% of sea-level pressure. So the same machine, on the same day, drawing the same volume, moves close to 19% less air by mass on the Reef than it would in Durban. A unit sized from a sea-level catalogue figure will be short on the Highveld, and the usual response — wind the pressure up until the tools behave — converts a sizing error into a permanent energy penalty.

The same arithmetic applies to anything specified in free air delivery: pneumatic conveying, blow-off, air knives, bag houses. If you are in Gauteng, Mpumalanga’s highveld, the Free State or the Northern Cape’s higher ground, check whether your rated figures were quoted at altitude or at sea level. It is a question worth asking before the order, not after.

The arithmetic in the section above, drawn to scale. The machine is identical; the air is not — and the same correction applies to anything else quoted in free air delivery, from blow-off to bag houses.

What a leak actually costs

Leaks are the easiest thing in a compressed air system to ignore, because a hiss is not a fault code. The arithmetic below is deliberately simple so you can redo it with your own numbers: flow figures are indicative for a sharp-edged orifice at 7 bar gauge, energy is at 0.11 kWh per Nm³ (a reasonable figure for a well-maintained screw compressor at that pressure — an ageing system is worse), and the annual column assumes 6,000 running hours, roughly a two-shift operation.

Leak (equivalent hole)	Indicative flow at 7 bar	Compressor power to feed it	Per year at 6,000 h
1 mm — a weeping fitting	~1.2 L/s	~0.5 kW	~2,900 kWh
3 mm — a failed quick-coupler	~11 L/s	~4.4 kW	~26,000 kWh
5 mm — a cracked hose tail	~30 L/s	~11.9 kW	~71,000 kWh
An open blow-off left running	often >30 L/s	>12 kW	>71,000 kWh

Multiply the last column by your own tariff to get rand. The reason to keep it in kWh is that it stays true when the tariff changes, and it is the number your own energy records can confirm. Note the shape of the table rather than any single row: leak cost rises with the square of the hole, so one neglected 5 mm split costs more than two dozen weeping fittings. Plants that run a leak programme and find nothing dramatic are usually looking for the wrong size of problem.

The four things worth measuring

You do not need a full audit rig to get useful numbers. Four signals cover the system, and three of them are cheap.

Signal	How it is picked up	What it tells you
Compressor power	CT clamps or a three-phase meter on the supply	Total kWh, and the loaded / unloaded split. A machine unloaded more than about a third of the running day is oversized or fighting a control problem.
Header flow	Thermal mass flow meter in the main, downstream of the dryer	Nm³/h actually consumed — and, combined with power, the kWh/Nm³ that is the system’s real efficiency figure.
Pressure — three points	Transmitters at discharge, after filters/dryer, and at the furthest machine	Differential across the filters is filter loading. Discharge-to-far-end drop is the distribution network. A large drop is why someone raised the set point.
Dew point	Dew-point transmitter after the dryer	Dryer health. Rising dew point precedes water in the lines, which shows up as valve and tool failures long before anyone blames the dryer.

Of these, kWh per Nm³ is the one to put on a board. It is a single number that captures leaks, pressure, control strategy and compressor condition at once, it is comparable month to month, and it degrades visibly as a system ages. Everything else is diagnosis after that number moves.

The weekend test: measure your leak rate without a leak detector

Ultrasonic leak detection is a good tool, but it is a survey — a day of walking, and a number that is out of date by the next shutdown. There is a cheaper measurement that gives you the total, and it needs nothing but a logging flow meter.

Pick a genuine non-production window — a Sunday, or the back half of a long shutdown.
Leave the compressor and the ring main live, exactly as they normally sit overnight.
Log header flow for several hours.

With nothing consuming air, whatever the meter reads is your leak rate. Put it over your production-hours average and you have leakage as a percentage of output — the figure that tells you whether a leak programme is worth running and, afterwards, whether it worked. Repeat it quarterly and it becomes a trend rather than an anecdote. This is also the one measurement that survives an argument with a supplier, because it is your meter, your system and your weekend.

The same logic applies to the control question. If the loaded/unloaded split from the power signal shows the compressor cycling every couple of minutes, the receiver is too small or the control band too tight — both fixable, neither visible without logging.

Pressure: the cheapest saving in the room

Header pressure tends to drift upward over a plant’s life. A tool misbehaves, someone raises the set point, nobody lowers it again. The penalty is roughly 6–7% more compressor energy for every extra bar, and it compounds: higher pressure also pushes more air through every leak and every open blow-off, so the artificial demand grows with it.

The three-point pressure measurement above is what makes the correction safe. Once you can see what the furthest machine actually receives, you can lower the set point in steps and watch whether anything on the floor notices — instead of guessing and being blamed for the next quality problem. In most plants there is at least half a bar available, and the measurement costs less than the first month of the saving.

What load shedding does to a compressed air system

An outage does more to compressed air than simply stopping it. The system blows down while the plant is dark, so the restart is not a resumption — it is a full re-pressurisation of the receiver and the entire ring main, at maximum load, at exactly the moment every other motor on site is also restarting. That combination is how a compressor ends up contributing to a monthly maximum-demand peak it has no business setting.

It also matters for what you can prove afterwards. If your air monitoring lives only in a cloud dashboard fed by a gateway with no local buffer, the outage erases the most interesting part of the record — the restart. Monitoring that buffers at the edge and backfills when power and connectivity return keeps the restart profile, which is where both the demand-charge argument and the compressor-condition argument live. That buffering requirement is the same one we describe in will your IoT sensors survive load shedding, and it applies here for the same reason.

What this means for a South African plant

Meter the air before you buy anything else. A flow meter and a power clamp on the compressor will tell you more in a month than a walk-around audit tells you in a day — and they keep telling you.
Get your kWh/Nm³ and hold it. It is the one compressed-air number worth reporting alongside OEE, and the one that shows a system decaying.
Run the weekend test quarterly. Leakage as a percentage of output is cheap to measure and impossible to argue with.
Check whether your ratings were quoted at altitude. On the Highveld a sea-level figure overstates delivered mass flow by roughly a fifth.
Walk the set point down, with the far-end gauge visible. Six to seven percent per bar, and it is free.
Buffer at the edge. If the restart is missing from the record, so is your evidence.

Frequently asked questions

How much compressed air does a typical plant lose to leaks?

Field reporting consistently puts unmanaged systems in the range of a fifth to a third of compressor output, and one industry source in our research base states that leaks alone can waste up to 30% — directional, not audited. The only figure that matters is your own, and you can measure it in one weekend: log header flow during a genuine non-production window and whatever the meter reads is your leak rate.

What does a 3 mm compressed air leak cost?

Indicatively, a 3 mm equivalent hole at 7 bar passes around 11 L/s, which takes roughly 4.4 kW of compressor power to feed — about 26,000 kWh a year on a 6,000-hour two-shift operation. Multiply by your own tariff for the rand figure. Because flow scales with the square of the hole diameter, one 5 mm split costs more than two dozen weeping fittings.

What should I measure on a compressed air system?

Four signals: compressor power (kW, and the loaded versus unloaded split), header flow in Nm³/h downstream of the dryer, pressure at three points (discharge, after the filters, and at the furthest machine), and dew point after the dryer. Power divided by flow gives kWh per Nm³ — the single number that captures leaks, pressure, control strategy and compressor condition at once.

Does altitude affect compressor capacity in South Africa?

Yes, and it is routinely missed. A compressor swallows a fixed volume per revolution, so what it delivers depends on inlet air density. At Johannesburg’s roughly 1,750 m the atmosphere is about 82 kPa against 101 kPa at sea level — so the same machine moves close to 19% less air by mass on the Highveld. Check whether a quoted free air delivery figure was stated at altitude before you size from it.

How much does lowering header pressure save?

Roughly 6–7% of compressor energy per bar, plus a second saving because lower pressure pushes less air through every leak and open blow-off. The safe way to claim it is to fit a pressure transmitter at the furthest machine first, then walk the set point down in steps while watching what the floor actually receives — rather than lowering it blind and being blamed for the next quality problem.

Put a number on the air.

Tell us what your compressor house looks like. We’ll scope flow, power and pressure monitoring that gives you kWh/Nm³, a measured leak rate, and a restart profile that survives load shedding — on the compressors you already own.

Book a Discovery Call See OEE management →

A note on the numbers

Leak flows are indicative values for a sharp-edged orifice at 7 bar gauge; energy is converted at 0.11 kWh per Nm³, a reasonable figure for a well-maintained screw compressor at that pressure and optimistic for an ageing one; annual figures assume 6,000 running hours. The assumptions are stated so you can redo the arithmetic with your own meter readings, tariff and running hours — which is the only version of these numbers worth acting on. Quotes are reproduced verbatim from a 2026 research base of public engineering discussion, review platforms and industry reporting, anonymised; statistics inside quotes are attributed as reported by their sources, not as our own measurements. The altitude figures follow from the standard atmosphere at 1,750 m.

Compressed air: the utility nobody meters, and what it costs you