What Pressure Drop Through Distribution Actually Costs

Why this matters

Ask what a 15 psi distribution loss costs and almost everyone answers with the 15 psi. That is the one thing it does not cost, because you never had that pressure to spend in the first place; you paid to make it. What the loss actually buys you is a permanently raised setpoint, and a raised setpoint does two separate things: it makes every cubic foot more expensive, and it makes the plant consume more cubic feet. The second effect is the one that ends in a capital purchase, because it shows up at the compressor as a demand the plant appears to genuinely have.

Before you fit a gauge to measure any of this

Fitting a test gauge opens a pressurised path. Close the upstream isolation valve, open the vent, confirm the section gauge reads zero with the vent still open, and lock and tag that isolation under 29 CFR 1910.147 before a wrench touches the fitting. A compressed air line opened while still pressurised ejects the fitting and whips the hose, which is a struck-by injury from a system everyone in the building treats as harmless.

Never clear a port, a line or a bench with a blow gun. 29 CFR 1910.242(b) permits compressed air for cleaning only when reduced to less than 30 psi and then only with effective chip guarding and personal protective equipment. Air rooms commonly reach the 85 dBA eight-hour time-weighted average action level, at which point hearing protection and the rest of 29 CFR 1910.95 apply.

If any part of this work involves moving a compressor's discharge setpoint, the ceiling is not a preference. Never set discharge above the lowest pressure rating in the system, never above the set pressure of the receiver's safety valve, and never adjust, block or gag a safety valve to make a setting fit.

What the loss is not

It is not the pressure you lost. Pressure in a pipe is not inventory. The 15 psi did not go anywhere you can recover it from; it was converted to heat by friction the moment it was made.

It is not a cost paid by the tool that suffers it. The tool at the end of the bad run is one point of use. The setpoint it forces is carried by every cubic foot the plant makes, for every user, including the bench sitting on a ten-foot drop with a clean coupler.

It is not paid only when the offending tool runs. The setpoint sits where the worst point of use put it, all day and all night. A loss that only appears during a two-hour peak is charged to twenty-four hours of production.

It is not fixed by a bigger compressor, which is the conclusion the plant reaches on its own and which the rest of this article is about. That failure and the piping scaling behind it belong to the pipe-sizing article; what follows is what the drop does to demand, which that article does not cover.

The loop

Here is the sequence, and every step in it is something a reasonable person concludes from a true reading.

A run develops a large loss, either because it was always undersized or because the plant grew onto it. A tool at the end of it starves. Somebody raises the discharge setpoint until the tool works, which is the correct immediate fix and the whole plant now runs at the higher pressure.

Every unregulated opening in the building is now fed at that higher pressure. Leaks, open-blow applications, worn cylinders, air motors, anything without a regulator between it and the header: all of them pass more air, because an opening venting to atmosphere passes air in proportion to the absolute pressure upstream of it. Nobody used anything differently. The plant's measured demand rose anyway.

That measured demand is real air leaving a real compressor, so it reads at the machine as a capacity problem. The next capital request is for a larger compressor. The larger machine holds the same setpoint, delivers the same starved tool the same pressure, and runs at a lower load fraction, which for most control strategies is the expensive end of its curve.

Nothing in that chain is a mistake in isolation. The chain is the mistake, and the entry point is a length of pipe.

Give the extra demand its own name and its own number

Call it what a system audit calls it: artificial demand, meaning air consumed by unregulated end uses purely because they are fed above the pressure they need. It is not leakage, though leakage is part of it, and it is not waste in the sense of something broken. It is the plant's real consumption at the pressure the plant is really running.

It has one governing relationship and it is the same one that governs leaks. Above roughly 13 psig any opening to atmosphere is choked, so its mass flow is proportional to the absolute upstream pressure. Change the setpoint and every unregulated flow in the building changes by the ratio of the two absolute pressures.

The lever that removes artificial demand without touching the setpoint is a regulator at the point of use, set to what that use needs. That converts an unregulated opening into a regulated one, which stops responding to header pressure. It does nothing for leaks upstream of the regulator, which is why the two levers are complementary rather than alternatives.

Worked example: a 200 scfm plant and an 18 psi run

A machine shop's worst point of use needs 85 psig at the tool. Measured with one gauge moved between two ports at the afternoon peak, the loss from the receiver to that tool is 18 psi, so the header has to hold 103 psig for the tool to see 85, and with the control band the discharge setpoint sits at 105 psig.

Compressor output is 200 scfm. A no-production window with every tool valved off measured unregulated flow, leaks and open-blow together, at 30 percent of output, which is 60 scfm at 105 psig. That leaves 140 scfm of productive demand.

The proposed fix cuts the loss from 18 psi to 6 psi, which lets the setpoint come down to 93 psig: 85 needed at the tool, plus 6 psi of loss, plus the same control band.

Currency one: each cubic foot gets cheaper. Compression work rises with the logarithm of the pressure ratio, so the cost of one psi depends on where you already are. Near a 100 psig discharge from an atmospheric intake, one psi is on the order of four tenths to half of one percent of input power, and that coefficient does not travel far from that discharge. Take it at 0.45 percent: 12 psi off the setpoint is 5.4 percent less input power per cubic foot, for a machine running loaded. A machine spending significant time unloaded has a control-strategy question sitting on top of this.

Currency two: there are fewer cubic feet. 105 psig is 119.7 psia and 93 psig is 107.7 psia, a ratio of 0.900. Unregulated flow falls by 10.0 percent, from 60 scfm to 54.0 scfm. Productive demand is unchanged at 140, so total flow falls from 200 to 194 scfm, which is 97.0 percent of what it was.

Do not add these two. They are different kinds of quantity: 5.4 percent is a saving per cubic foot and 3.0 percent is a reduction in cubic feet. They multiply. 0.946 x 0.970 = 0.918, so the plant draws about 8.2 percent less input for exactly the same productive work. Adding them would have given 8.4 percent, which is close enough to look right and is arrived at wrongly, and the error grows as the two terms grow.

What the loop would have cost instead. Before the repipe, this plant's demand reads 200 scfm against a 200 scfm machine, which is a machine with no headroom and a starving tool, and that is a completely defensible case for a larger compressor. Sixty of those 200 scfm were unregulated and 6.0 of them existed only because of the drop. A larger machine would have delivered 85 psig to that tool by holding 103 psig at the header, exactly as before, and the 18 psi would still be there.

The failure mode to catch in your own work. A repipe that reduces the loss and leaves the setpoint where it was captures nothing at all. Both currencies are realised through the setpoint and only through the setpoint. The saving is booked on the day someone turns the pressure down, and it comes down only as far as the new worst point of use still tolerates, which is a fresh measurement rather than an assumption.

Budget the allowance across segments, not as one number

A total loss figure tells you what you are paying and nothing about what to fix. Split it before you measure: air room to receiver, the treatment train, the main, the branch, the drop with its filter-regulator-lubricator and shutoff, and the hose and coupler at the end. Each segment gets a share of the total allowance, and the total is what your setpoint has to carry.

Where a total distribution loss allowance is specified as a percentage of operating pressure, take that figure from the piping system manufacturer's design data or from the design standard your project specification invokes, in the edition it names, rather than from a remembered number. Filters and dryers are the segment most often budgeted at the wrong figure, because the clean rating gets used where the element's change-out differential is the one your setpoint has to carry; read it off the manufacturer's data.

The allocation is what turns a survey into a work order. A plant with a 6 psi allowance on its drops and 14 psi measured on one of them knows which twenty feet to work on. A plant with a single 18 psi total knows only that it has a problem.

How to verify you got this right

Take the two currencies separately and check each against something you can observe.

For the setpoint, confirm the reduction actually happened and held. A setpoint that was lowered and then quietly raised again a week later, because one tool complained, has returned the whole saving, and the tool that complained is the next measurement.

For the demand, repeat the no-production window at the new setpoint and confirm the unregulated flow fell close to the ratio you predicted. If it fell by materially more than the pressure ratio explains, something was switched off during one of the two windows and the comparison is not clean. If it did not fall at all, the flow you measured was not unregulated, which usually means a regulator downstream of it was already holding the pressure and the artificial demand term was smaller than the survey assumed.

References

• 29 CFR 1910.147, control of hazardous energy, for isolating and verifying depressurisation before fitting a test gauge
• 29 CFR 1910.242(b), compressed air used for cleaning, and 29 CFR 1910.95, occupational noise exposure, including the 85 dBA eight-hour time-weighted average action level
• Piping system manufacturer design data and the project's own specified design standard, in the edition it names, for distribution loss allowances; filter and dryer manufacturer data for change-out differential rather than clean rating
• See related: Why Pipe Size Decides More Than Compressor Size; How to Measure Pressure Drop Across a Compressed Air System; How to Put a Number on What Leaks Are Costing