Electricity has a meter on the wall and an invoice every month. Compressed air has neither, so a plant can lose a quarter of its compressor energy through holes it has stopped hearing. Every figure in this note can be measured in an afternoon with a pressure gauge and a watch.
The most useful half hour we spend in a customer plant is the one before the shift starts. The machines are off, the hall is quiet, and the compressor is still running. Somebody usually explains that it keeps the system topped up overnight. What it is actually doing is filling holes: a cracked push-in fitting on a third-floor machine, a drain valve that somebody wedged open years ago because it was blocking, two burst-proof hoses that were never disconnected after a trial. In a plant with no leak programme, 20 to 30 per cent of compressed air production goes to places nobody intended, and because air arrives without a meter or an invoice, it is the one utility a plant can waste indefinitely without noticing.
This note is the arithmetic we use. None of it needs instrumentation beyond a pressure gauge, a stopwatch and the nameplate of the compressor, and the numbers are large enough that the first survey usually pays for a year of maintenance.
What air costs, before anything is wasted
A modern fixed-speed screw compressor delivering around 7 bar produces roughly ten cubic metres of free air a minute for every 55 kilowatts it draws. That works out at about 0.1 kilowatt hours per cubic metre, a figure worth memorising, because it converts any flow straight into money. One litre per second of air is 3.6 cubic metres an hour, so a continuous litre per second costs about 0.36 kilowatts. Over 8,000 operating hours that is 2,900 kilowatt hours a year: at ten cents a unit, roughly 290 euros for one litre per second, running quietly and forever.
Air is also an expensive way to move energy. Of the electrical power that enters a compressor, something like 10 to 15 per cent reaches the point of use as useful work; the rest leaves as heat in the compressor room. That is not a reason to abandon pneumatics, which are robust, cheap to install and ideal for short, hard, repetitive motions. It is a reason to stop using air where an electric actuator, a spring or a well-shaped piece of metal would do the job, and in particular to stop using it to blow things dry or clean when a fan or a brush would do.
What a hole costs
A hole one millimetre across at 7 bar passes about a litre of free air per second. By the figure above, that single hole is 290 euros a year. Few plants have one. A one-millimetre hole is also smaller than most of what we find: a loose push-in fitting leaks more, a perished hose far more, and a 3 mm hole passes roughly nine times the flow of a 1 mm one, because the flow goes with area.
The honest way to find out how much a plant leaks is to measure it rather than estimate it, and the measurement takes twenty minutes on a Sunday. Close every machine isolation valve so that only the distribution system is live, let the compressor fill the system to its normal pressure, stop it, and time the fall. Multiply the system volume, receiver plus piping, by the pressure drop in bar, and divide by the time. A three cubic metre system falling from 7.0 to 6.0 bar in six minutes has lost three cubic metres of free air in 360 seconds, which is 8.3 litres per second: about three kilowatts, about 2,400 euros a year, from a plant that is standing still.
Leaks are found with an ultrasonic detector, which costs less than the air it saves in the first month and lets one person survey a hall during production. Without one, the same work can be done during a shutdown by ear and by a brush of soapy water at every fitting. What matters more than the method is what happens afterwards: a found leak that is not tagged, listed and scheduled is a leak that is still there next year. We tag them, number them, write the repair into the next planned stop, and re-measure the system drop afterwards, because that is the proof.
The seven per cent rule, and why plants run too high
Compressor power falls by roughly seven per cent for every bar of pressure you stop making. A plant running at 7.5 bar because one old machine needs 6.5 bar at its inlet is paying that premium on every cubic metre it produces, all day, for one machine.
The reason is almost always pressure drop rather than a genuine requirement. A dirty filter element costs half a bar. An undersized hose and a quick coupling at the machine can cost another half. Twenty metres of 3/4 inch pipe carrying ten litres per second costs a few tenths. The total is taken from the machine, so somebody turns the compressor up, and the whole plant pays for one bottleneck. A sensible budget is no more than 0.1 to 0.3 bar from the receiver to the point of use, and when a machine is starved the right question is which element in that chain is eating the pressure, not how high the compressor can be set.
Three habits keep that budget. Pipe the plant as a ring rather than as dead-end branches, so air reaches a machine from both directions and the effective length halves. Size the branch for the flow it will actually carry, which usually means one size larger than seems necessary and costs almost nothing while the building is open. And log the differential pressure across filters instead of changing elements by the calendar: a filter is replaced when it has cost 0.2 bar, not when a year has passed.
How much a machine really drinks
Pneumatic consumption is arithmetic too, and it belongs in a machine specification. A cylinder of 50 mm bore and 200 mm stroke sweeps 0.39 litres. At 6 bar gauge, which is 7 bar absolute, that is about 2.75 normal litres of free air for the extend stroke. The return stroke moves the annular side, a little less, around 2.3 litres, so one complete cycle is roughly 5 normal litres plus whatever the tubing and the valve dead volume add. At twenty cycles a minute that is 100 litres a minute, 1.7 litres per second, about 0.6 kilowatts, or some 500 euros a year on two shifts for one cylinder.
That number changes decisions. Shortening the stroke to what the job needs, choosing a 40 mm bore where the force allows, and shortening the tubing between the valve and the cylinder all come straight off it. Mounting valves on the actuator rather than in a cabinet six metres away removes air that is compressed and vented every single cycle without ever doing work. And on cylinders that only need force in one direction, a return spring or gravity removes half the consumption outright.
Blow-off is where the real money goes. An open 4 mm tube at 6 bar passes something like fifteen to twenty litres a second, which is six or seven kilowatts of compressor power running continuously, and it is usually aimed at a part that could be cleared mechanically. If air really is needed, an engineered nozzle with entrained ambient air does the same job on roughly half the flow, and a solenoid that only opens when a part is present removes the rest. Cleaning yourself or a machine with an air gun is the most expensive broom ever invented.
Dry air, and the drains that leak on purpose
Air leaves a compressor saturated and warm. As it cools in the pipework, water comes out of it, and water in a pneumatic system corrodes pipe from the inside, washes grease out of cylinders, and makes valves stick in a way that reads as an electrical fault. The specification that matters is pressure dew point: a refrigerant dryer gives about +3 degrees, which is fine for a heated building, and a desiccant dryer gives −40 degrees, which is what an outdoor line or an unheated store needs. Dew point is worth logging at the far end of the system, not at the dryer outlet, because the far end is where the water actually falls out.
Every drain in the system is a potential permanent leak. Timed solenoid drains open for a few seconds regardless of how much condensate there is, which on a dry day is pure loss; zero-loss drains with a float open only when there is water, cost more, and pay for themselves on any system of size. The drain somebody has wedged open with a cable tie because it was blocking is the single most expensive object in most compressor rooms.
The compressor room itself
Two facts about the machines are worth knowing. A fixed-speed screw compressor that is unloaded still draws something like a quarter to a third of its full-load power, so a plant whose demand varies all day and whose compressors run load-unload is paying heavily for air it does not take; this is the case where a variable-speed machine, or a properly staged pair of fixed-speed machines with one trim unit, repays its price quickly. And the heat the compressor rejects is roughly all of the electrical energy it consumed: a 55 kilowatt machine is a 50 kilowatt heater. Ducting that heat into the hall in winter, or into process water, recovers a significant part of the bill and costs a duct.
Intake air deserves one sentence of its own: it should come from the coolest, cleanest place available. Every three degrees of intake temperature is worth about one per cent of capacity, and a compressor inhaling its own exhaust in a closed room is working against itself.
What belongs in a routine, and in a specification
For a plant, the routine is short: a leak survey every quarter with tagged, numbered findings and a repair date; the system drop test twice a year as the measure of whether the programme is working; pressure logged at the worst machine rather than at the compressor; filter differential pressure logged; drains checked; dew point checked at the far end. None of that needs a consultant, and all of it fits on one page of a maintenance plan.
For a machine we build, air consumption is part of what we state and test: normal litres per cycle, the pressure required at the machine inlet, the air quality class the machine needs, and the consumption measured during the acceptance run rather than calculated. A plant that knows what each machine drinks can size the next compressor from data instead of from the supplier catalogue, which is the difference between a utility that is managed and one that is merely paid for.
— GANI Lifecycle engineering team