What Storage Buys You That Horsepower Cannot

Why this matters

A shop with a pressure complaint gets told to buy a bigger compressor about nine times out of ten, and a large share of the time the compressor was never the problem. The plant has a demand that arrives faster than any compressor can respond to, and what it is short of is stored air, not horsepower. Storage is the cheapest thing in the room and the least understood, so it is the thing that gets skipped when the system is built and the thing nobody thinks to check when it starts sagging. Knowing exactly what storage can and cannot do lets you tell a customer "your compressor is fine, your receiver is in the wrong place and it is too small for the event that is hurting you" and then prove it with a gauge.

Before you touch a receiver: it is a pressure vessel holding energy that does not go away when the compressor stops. Isolate it at its inlet and outlet valves, lock and tag the isolation and the compressor's disconnect under 29 CFR 1910.147, open the drain and the manual vent, and confirm zero on a gauge that is open to the vessel before you break any joint or remove any fitting. A receiver you assumed was empty because the compressor was off is the classic struck-by event in this trade.

The one relationship storage runs on

Air stored in a vessel is not the vessel's volume. It is the extra air packed in above atmospheric pressure, and you only get to use the part of it that sits above the lowest pressure your fussiest machine will tolerate.

Free air held per psi of allowable pressure drop, at sea level:

• Receiver volume in cubic feet, divided by 14.7 psia.
• A US gallon is 0.1337 cubic feet, so divide gallons by 7.48 to get cubic feet.
• At altitude, substitute your local barometric pressure for 14.7. At about 5,000 feet the standard atmosphere is near 12.2 psia, so the same vessel holds roughly 20 percent more free air per psi than the sea-level figure suggests, because a psi of gauge pressure is a larger fraction of the local absolute pressure.

Allowable drop is not the gap between your compressor's cut-out and cut-in. It is the gap between the pressure the header actually holds when nothing unusual is happening and the lowest inlet pressure the worst-affected machine will accept, less whatever the distribution path eats at the flow the event pulls. That last subtraction is where most of the band disappears, and a sibling article in this library on what pressure drop through distribution actually costs owns that calculation.

The three things storage genuinely buys

It converts a flow problem into a pressure problem you can survive. A demand event that exceeds the compressor's output does not have to be met by the compressor at the instant it happens. It can be met out of the vessel, and the price is pressure, which decays at a rate you can compute rather than collapsing.

It buys the compressor time to respond. No compressor answers instantly. A load/unload machine has to close its inlet valve and rebuild sump pressure. A machine that was fully stopped has to start, and the number of starts per hour is limited by the motor. A variable-speed machine ramps. Every one of those responses is measured in seconds, and storage is the only thing covering the gap.

It stops the control from thrashing. Frequent load/unload cycling wastes the blowdown air on every unload and wears the inlet valve and the motor. More storage means fewer, longer cycles at the same average demand. The compressor control article in this library owns why the control behaves that way.

The negative space: four things storage does not buy

This is the part that gets sold wrong.

It does not buy sustained capacity. Against a demand that exceeds compressor output continuously, storage sets how many minutes you have before the pressure hits the floor, and nothing else. Double the receiver and you double the minutes. You never reach a receiver size that solves a continuous deficit, because the arithmetic is a fixed quantity of air divided by a rate that never stops.

It does not buy dryness. A receiver does drop out bulk liquid water and it is a useful place to put a drain, but it does not lower the pressure dew point of the air leaving it. Air that leaves the receiver saturated at the receiver's temperature will condense again in the first cold run of pipe. The dryer does that job, and the pressure dew point article owns the specification.

It does not buy coverage against leaks. A leak is a continuous demand, which puts it in the first category. Storage will make a leaky system's pressure sag more slowly during the shift and will not change how much air the leaks take. What it can do is indirect: it lets you hold the same ride-through at a lower average pressure, and a fixed hole passes mass in proportion to absolute upstream pressure, so the reduction storage permits does cut the leak load, by roughly 8 percent for 10 psi off a 105 psig setpoint. That is a second-order benefit of the pressure change, not a leak fix.

It does not fix a restriction between the storage and the demand. Air in a receiver on the far side of an undersized dryer, a plugged filter or a long small-bore run is not available at the rate the event needs it. Storage is only worth its volume if the path from it to the load can pass the flow.

Worked example: one event, then the same event every forty seconds

A fabrication cell has an air-operated clamp bank that fires as a group. The complaint is that the last clamp in the sequence is slow, and only when the cell runs its heavy part.

Take the numbers off the plant, not off a catalogue. The receiver is 240 gallons, which is 240 divided by 7.48, or 32.1 cubic feet. At sea level that vessel holds 32.1 divided by 14.7, which is 2.18 standard cubic feet of free air for every psi you are willing to lose. The header normally sits at 100 psig, the clamp bank's valve manifold is rated to work down to 90 psig, and at the flow the event pulls the run into the cell measures 0 psi of loss because it is short and generously sized. So the allowable drop is 10 psi and the usable stored air is 10 times 2.18, or 21.8 standard cubic feet.

The event: a logging gauge on the receiver shows pressure falling from 100 to 90.8 psig over 20 seconds each time the cell indexes. That is 9.2 psi out of a 10 psi band, spent in 20 seconds.

Turn the decay into a flow. 9.2 psi times 2.18 standard cubic feet per psi is 20.1 standard cubic feet delivered out of storage in 20 seconds, which is 20.1 divided by 20 seconds, or 1.0 standard cubic foot per second, or 60 scfm. That 60 scfm is the amount by which the cell's demand exceeded what the compressor was putting in during those 20 seconds. It is not the cell's total demand, and calling it that is the most common misreading of this measurement.

One event is survivable: 9.2 psi of a 10 psi band, and the last clamp gets 90.8 psig, just above its 90 psig floor. That matches the complaint being marginal rather than total.

Now run it as it actually runs. The cell indexes every 40 seconds. The event takes 20 seconds, leaving 20 seconds of recovery. To be back at 100 psig before the next index, the compressor has to push 20.1 standard cubic feet into the vessel in 20 seconds, which is the same 60 scfm of surplus above the cell's baseline draw. The compressor does not have 60 scfm of surplus, which is exactly why storage had to cover the event in the first place. So the vessel starts the second event lower than the first, the third lower still, and within a handful of cycles the pressure has walked down to the floor and the clamp misses. That walk-down is the signature: not a sudden failure, a slow decline over the first few minutes of a production run, then a stable low pressure.

The same 240 gallon vessel against a continuous deficit tells the other half of the story. Suppose instead that the plant simply uses 10 scfm more than the compressor makes, all shift, no events at all. The 21.8 standard cubic feet of usable storage divided by 10 scfm is 2.18 minutes. Two minutes and change is the entire contribution of that vessel. Quadruple it to 960 gallons and you get under nine minutes. There is no vessel size that answers a continuous 10 scfm shortfall, and this is the arithmetic to put in front of an owner who has been quoted a bigger receiver for a leak problem.

What would flip the recommendation: if the logged decay had shown the receiver holding within a psi or two while the pressure at the clamp manifold collapsed, storage would be irrelevant and the fault would be the path, not the volume. Reading only one gauge is what makes that mistake possible, and the article on deciding whether the problem is supply or demand owns that split.

Where the receiver has to sit for any of this to be true

Storage upstream of the dryer and filters is control storage. It smooths the compressor's cycling and it feeds the dryer at a steady rate, which is what the dryer was rated to see. Storage downstream of the dryer and filters, close to the load, is demand storage, and it is the only storage that can answer a fast event, because the air in it does not have to cross the dryer and filter pressure drop on its way out.

A plant with one large receiver in the compressor room and a fast event at the far end of the building has control storage and no demand storage, and it will behave in the worked example's second pattern no matter how large that room receiver is. Adding a modest vessel local to the event, fed through a line sized for the recovery flow rather than the peak flow, usually beats adding volume in the room.

How to verify you got this right

Log receiver pressure and the pressure at the affected machine on the same time base, through at least ten cycles of the event, with the plant running normally rather than on a quiet Saturday. Then check three things against each other.

First, convert the receiver's decay slope to a flow using your own vessel's standard cubic feet per psi, as above, and confirm the number is physically plausible for the equipment that fired. A number far larger than the cell's rated consumption means something else fired at the same time and your event is not the event you thought.

Second, confirm the recovery. Pressure that returns to the same starting value before the next event means storage is adequate for that event. Pressure that starts each event lower than the last means you have a rate problem, not a volume problem, and more storage buys cycles, not a fix.

Third, check that the machine's actual inlet pressure, not the header pressure, stayed above its stated minimum. If the header held and the machine did not, stop working on storage.

References

• 29 CFR 1910.147, control of hazardous energy, general industry: the basis for the isolation, lock, tag and verified-zero sequence before opening a receiver or any component holding stored pressure.
• ASME Boiler and Pressure Vessel Code Section VIII Division 1, in the edition the vessel was built and stamped to, which is the reading that applies to a vessel already in service; for one being installed, 29 CFR 1910.169(a)(1) incorporates Section VIII by reference and is the federal floor. It binds through the vessel's own stamping and through your state's boiler and pressure vessel program where that state has adopted the Code: the basis for treating a receiver as a code vessel with a required relief device rather than as a tank.
• See related: How to Work Out Whether a Receiver Is Sized for the Demand; What a Receiver Holds in Stored Energy Terms; What Pressure Drop Through Distribution Actually Costs; How to Decide Whether the Problem Is Supply or Demand.