What an Inappropriate Use of Compressed Air Looks Like

Why this matters

Every list of "inappropriate uses" you will find names the same applications by category, and a shop that walks in and condemns an application by category is guessing. Some of those applications are genuinely indefensible and some are perfectly reasonable at the duty they actually run. The difference is a measurement that takes a Saturday morning and no instruments the shop does not already own, and it produces a number nobody in the building has: what fraction of their compressor is turning to work nobody asked for.

Before the walk: an open compressed air tube is loud enough to matter, and standing next to several of them for an hour is a noise exposure. Wear hearing protection under a program meeting 29 CFR 1910.95 for any period in the compressor room or beside an open discharge. Never direct a stream at anyone's skin or clothing, including your own, and do not put a hand in front of an open outlet to feel the flow; air entering through a break in the skin can enter the bloodstream. Where a branch has to be isolated by breaking a connection rather than closing a valve, isolate upstream, lock and tag under 29 CFR 1910.147, vent downstream and confirm zero on a gauge open to the line first.

The word "inappropriate" is doing no work until you attach a measurement

An application is not inappropriate because of its category. It is inappropriate when both of these are true at once:

• Its flow is continuous or near-continuous, so it draws a compressor's output around the clock rather than in bursts, and
• the work it does could be done by a device drawing a fraction of that flow, or by no air at all.

Either half alone is not a finding. A high-flow burst that fires twice a shift costs almost nothing in a year. A continuous low-flow use with no alternative is simply a load. It is the intersection, and the only way to locate it is to isolate the use and see what the compressor stops doing.

The reason this matters more than category lists is direction. Compressed air is a poor way to move heat, a poor way to move liquid and a poor way to make vacuum, in the sense that each of those costs several times the electrical input of a purpose-built device. But "several times a small number" is still a small number when the duty cycle is 2 percent, and a plant that rips out a well-controlled venturi and installs a vacuum pump has spent capital to add a machine to the maintenance list.

The method, and the one condition it depends on

With a fixed-speed compressor on load and unload control, the fraction of time it spends loaded is a usable proxy for flow, because its loaded output is essentially constant. Multiply the loaded fraction by the compressor's rated delivered flow at the pressure it is actually holding and you have scfm. That condition is the whole method: it does not work on a variable-speed machine, on a modulating machine, or on a multi-compressor header where a second unit can start. For those, take the flow off a meter instead.

One honest caveat on the proxy, in the direction that matters. Every unload blows the sump down and every reload re-pressurises it, and that air is counted as loaded time while never reaching the header. So at low loaded fractions with frequent cycling the proxy reads slightly high. It overstates demand rather than understating it, which is the safe direction for a finding you are going to put in front of a customer, and it is one more reason to state the loaded fraction alongside the derived flow rather than instead of it.

Run the walk with the process off. That removes welding fume, machining coolant mist and moving machinery from the exposure list, and it gives you a plant whose only air demand is the things somebody left on.

The walk: one Saturday, three uses

The plant is a light fabrication shop. One fixed-speed rotary screw on load and unload, rated 100 scfm delivered at 100 psig. A stopwatch, a notebook, and ten-minute observation windows, each covering several full load and unload cycles.

Baseline. All production equipment off, all point-of-use valves closed. Loaded 18 percent of the window. That is 18 scfm, and with nothing running it is the leak population plus whatever nobody managed to close. The leak-survey articles in this library own what to do with that number; here it is the floor everything else is measured against.

Use 1, the weld fixture cooling tube. A quarter-inch copper line, no nozzle, aimed at a fixture to keep it cool enough to handle between parts. Left on continuously because its only isolation is the branch valve up in the rack, which is why nobody had ever closed it: shutting it takes a ladder. Closed at the rack for the baseline and opened again, the loaded fraction goes from 18 to 46 percent, an adder of 28 percentage points, which is 28 scfm.

Twenty-eight scfm, continuous, against a leak population of 18 scfm. One open tube is drawing more than every leak in the building put together, and it is 28 percent of a 100 scfm compressor. The work it does is move heat out of a steel fixture, which a small electric fan does for a small fraction of the electrical input, or which a fixture with a different thermal mass avoids needing at all. Both halves of the test are met: continuous, and replaceable by something drawing a fraction. Condemned.

Use 2, the sump pump. An air-operated double diaphragm pump on a coolant sump, running whether or not there is anything to move, because its air supply is on a manual valve somebody opened in the spring. Open it and loaded goes from 18 to 39 percent, an adder of 21 percentage points, which is 21 scfm.

Here the category list would condemn the pump. The measurement condemns the control. An air-operated diaphragm pump is a reasonable choice on a sump with solids, a corrosive fluid or a location where you do not want a motor, and its cost is only intolerable because it runs continuously against a discharge that is mostly closed and a sump that is mostly empty. Interlock it to a level switch so it runs on demand and the same pump becomes a low-duty load. Condemned as installed, not as a technology.

Use 3, the venturi vacuum generator. A pick-and-place head makes its vacuum from compressed air through a venturi. Open it and loaded goes from 18 to 25 percent, an adder of 7 percentage points, which is 7 scfm, and that is with the generator held on continuously as it sits on the bench.

In production it does not sit on continuously. The generator is valved with the pick cycle and a check valve holds vacuum on the cup while the head travels, so the generator runs about 20 percent of the cycle. Seven scfm at a 20 percent duty is an average of 1.4 scfm. Against 18 scfm of leaks it is not the finding, and against the alternative, which is a dedicated vacuum pump running continuously plus a vacuum line to a moving head plus one more machine to maintain, the case for change does not close. Not condemned.

What the numbers said about the plant

Continuous demand with the plant not producing anything: 18 scfm of leaks, 28 scfm of fixture cooling, 21 scfm of sump pumping. That is 67 scfm, of which 49 scfm is the two condemned uses, or 73 percent of everything the compressor was doing on a Saturday. Against the compressor's 100 scfm rating, those two uses alone were consuming 49 percent of its capacity to do work no one had asked for since spring.

Note what is not in that total. The venturi's 7 scfm is not in it, because in production it is 1.4 scfm and on a Saturday it is zero. Adding a peak flow into a continuous-demand total is the most common way this walk gets misreported, and it is the same confusion of a peak with an average that the demand-register article in this library is built to prevent.

Confirmation came the following Saturday. The fixture tube was valved and left closed, the sump pump was interlocked to a float switch, and the baseline window was repeated with the same stopwatch and the same ten-minute window: loaded 19 percent, against the 18 percent measured before any of the three uses were opened. The 49 scfm was gone and the leak population had not moved, which is what you want to see, because a leak population that moves between two Saturdays means one of your valves is not actually closed.

The definition worth keeping, and what flips each call

The portable version: an inappropriate use is one whose flow does not stop when the work stops. That single sentence catches the fixture tube, catches the sump pump as installed, and correctly acquits the venturi, and it does it without knowing anything about the category the application belongs to.

Three conditions flip these specific calls, and each is a real one to check before you write the report.

The fixture tube's call flips if the fixture must be cooled in a location where an electric fan cannot go, an area classified for the presence of flammable vapour being the usual case, and where the cooling is genuinely needed rather than habitual. Then the honest finding is not "remove it" but "put an engineered nozzle and a valve on it", which cuts the flow substantially at the same cooling effect and stops it between parts.

The sump pump's call flips if the sump genuinely runs full continuously, which makes the pump a continuous duty by requirement rather than by neglect. Then the question moves from control to technology and it is a real comparison, not an easy one.

The venturi's call flips if anyone valves it on and leaves it on, which turns 1.4 scfm into 7 scfm and puts it above a third of the leak population. That is the failure mode to warn the customer about, because it is one operator habit away and it leaves no trace.

The failure mode of skipping the measurement entirely: the shop condemns the venturi because it is on every category list, quotes a vacuum pump, leaves the fixture tube and the sump pump running because neither is on any list under a name the shop recognised, and the plant's continuous demand falls by 1.4 scfm out of 67. The customer paid for a machine and got a 2 percent improvement, and the two real findings are still running.

References

• 29 CFR 1910.95, occupational noise exposure, general industry: the basis for hearing protection during extended time beside an open air discharge or in the compressor room.
• 29 CFR 1910.147, control of hazardous energy, general industry: the basis for the isolate, lock, tag and verified-zero sequence where a branch must be isolated by breaking a connection.
• See related: Why Blowing Things Off With Compressed Air Is a Regulated Act; How to Put a Number on What Leaks Are Costing; The Demand Events That Size a System Nobody Measured; What a Compressor Control Strategy Is Choosing Between.