The Compressor That Could Not Keep Up With No New Equipment Added

Why this matters

"It used to keep up and now it does not, and we have not added anything" is one of the most common compressed air calls, and the default answer, sell them a bigger compressor, is wrong more often than it is right. Nothing was added, so either the machine is delivering less than it used to or the plant is consuming more without anyone installing a machine. Those two have completely different repairs and the same symptom, and the record on the wall usually separates them before anyone puts a hand on the equipment.

This case is reconstructed backwards from what the plant had written down. The gaps in that record turned out to be as informative as the entries.

Stop the two things people do first

Before diagnosis, two actions, because both are already in progress by the time you arrive.

Nobody raises the setpoint again. The plant will have done this already and will offer to do it once more while you watch. Raising discharge pressure raises the compressor's power draw and simultaneously raises the flow out of every leak and every unregulated point of use, because those are fixed restrictions and their flow scales with upstream absolute pressure. Both effects push the same way, so the raise buys a few minutes and makes the underlying problem larger. Get agreement to leave the setpoint where it is until the trace is done.

Nobody opens anything on the pressurised side. An air receiver holds stored energy that does not dissipate when the motor stops, and the compressor room is where people reach for a drain valve to "see if there is water in it." Any opening of a drain, filter bowl, joint or the receiver itself waits until the compressor's disconnect is locked open under 29 CFR 1910.147, the isolation valve is closed, the leg is bled, and the receiver's indicating gauge, which 29 CFR 1910.169 requires it to carry, is confirmed at zero. Work inside the starter enclosure forks to 29 CFR 1910.333(b)(2), not 1910.147, with dead proved by the live-dead-live sequence of NFPA 70E-2021, 120.5 where your employer's electrical safety program adopts it. And wear hearing protection from the door: a compressor room commonly sits at or above the eight-hour time-weighted average of 85 dBA at which 29 CFR 1910.95 attaches its hearing conservation requirements.

What the plant had written down

One lubricated rotary screw on load/no-load control, a receiver, a refrigerated dryer, one shift plus occasional Saturday work. The record consisted of four things:

• A commissioning sheet from four years ago with a pump-up time, a discharge pressure and a note of the control band
• The compressor's own hour counters: total run hours and loaded hours, read annually and written in the log
• A maintenance log, mostly filter and lubricant changes
• Two undated notes reading "raised pressure" and one dated note reading "raised pressure to 118"

The counters were the useful part. Commissioning year: 3,800 run hours, 2,100 of them loaded, so the machine was loaded 55 percent of its running time. Most recent year: 4,050 run hours, 3,500 loaded, so 86 percent loaded. On a load/no-load machine, air delivered is loaded hours times capacity, so loaded hours rose by a factor of 3,500 divided by 2,100, which is 1.67.

That is the whole finding stated as a question: what accounts for 67 percent more loaded time on a plant that added nothing?

Candidate one: the machine is delivering less

A machine that has lost capacity needs more loaded hours to deliver the same air, so capacity loss and demand growth show up identically in the counter. Separating them needs a direct capacity measurement, and pump-up time is the one the commissioning sheet already supports.

Pump-up time is the time to fill the system from atmospheric to a marked pressure with the plant isolated. Commissioning recorded 4.2 minutes. Repeating it with the same gauge, the same two pressure marks and the plant isolated gave 4.6 minutes, so the fill is about 10 percent slower and delivered capacity is down roughly 10 percent.

Two notes on that comparison, both about what kind of error is in it. The gauge offset is a fixed systematic error, and because both endpoints of both tests were read on the same gauge at the same two marks, that offset shifts both tests the same way and largely cancels in the ratio. What does not cancel is intake air temperature, which was not recorded at commissioning: a positive displacement machine moves a fixed volume per revolution, so warmer intake air means less mass delivered per minute, and a compressor room that has got hotter over four years will read slower on this test with nothing wrong inside the machine. That is a real and separate finding, and this library covers it under intake conditions.

Take the 10 percent as an upper bound on capacity loss, then. If capacity is down 10 percent, the loaded hours needed to deliver air A become A divided by 0.9 times capacity. Working the ratio: 1.67 equals the air ratio divided by 0.90, so the air ratio is 1.50. Demand is up 50 percent, and the capacity loss accounts for the remaining part of the 67 percent rise in loaded hours.

Fifty percent more air, nothing installed.

Candidate two: demand that exists when nobody is working

The commissioning sheet had one entry nobody had ever looked at again: a weekend loaded-hour reading. Over a 48-hour non-production weekend at commissioning, the machine logged 5.8 loaded hours. Repeating the same reading over a recent 48-hour weekend gave 14.1 loaded hours.

As a fraction of rated delivery, the non-production floor went from 5.8 divided by 48, which is 12 percent, to 14.1 divided by 48, which is 29 percent. Seventeen points of the compressor's rated delivery are now going to something that runs when the plant is empty.

Put that against the production week. The machine is loaded 86 percent of its running time, so on a base of loaded time, 29 divided by 86 is about 34 percent: roughly a third of the compressor's loaded hours are now serving loss rather than work. At commissioning the same ratio was 12 divided by 55, about 22 percent.

A leak survey found the bulk of it. This library covers turning that into a load figure under leak costing and the traverse itself under the survey article, so it is enough to say here that the traverse tagged 34 leaks here and the largest population was point-of-use couplers and hose whips.

Candidate three: the setpoint notes

The three "raised pressure" entries were the part of the record that explained why the problem accelerated rather than crept.

Every fixed restriction in the system, meaning every leak and every unregulated point of use, is passing choked flow at plant pressures, so its mass flow rises in direct proportion to upstream absolute pressure. Raising discharge pressure therefore raises the loss flow, which lowers pressure, which invites another raise. The plant had walked itself up that loop three times over four years, and the last note put the setpoint at 118 against a commissioning value in the low 100s.

The gap in the record is the finding here. Two of the three raises were undated, so nobody could put them on a timeline against the loaded-hour counters, and that is precisely the comparison that would have shown the plant what it was doing. A dated one-line entry for every setpoint change costs nothing and would have made this a fifteen-minute call.

What actually fixed it

The order mattered, because doing these in the wrong sequence makes the plant's pressure worse before it gets better.

1. Repair leaks first. The largest group needed no more than a drop valve closed, the leg bled through the coupler and the local gauge confirmed at zero before the joint was broken. This raises available pressure, which creates the headroom for step 3.
2. Regulate the unregulated points. Several drops fed tools and blow guns straight off header pressure. A point-of-use regulator set to what the device actually needs takes those out of the group whose flow scales with header pressure.
3. Walk the setpoint back down in steps. Drop it, run a full shift, and watch the pressure at the worst point of use, not at the compressor. The critical machine's inlet is what decides whether the step holds.
4. Address the transients that had driven the raises. Two of the three raises traced to a short, large draw that pulled the header down for under a minute. That is a storage question, not a compressor question, and this library covers it under storage and under pressure and flow control.

How they confirmed it

Not with a pressure gauge, because pressure was the symptom the plant had been chasing for four years and it can be made to look fine by raising the setpoint.

They confirmed it on the counter. Over the month after the repairs, loaded hours ran at a rate that annualises near 2,600 against the 3,500 of the previous year, with the setpoint 15 lower than where it had been. The weekend reading, taken by the same technician with the same marks so the systematic part of the comparison cancels, fell from 14.1 loaded hours to 7.9 over 48 hours, or about 16 percent of rated delivery against the previous 29.

The remaining gap against the commissioning figure of 12 percent is the 9 branch leaks that were deferred to the next shutdown plus whatever the traverse did not find, and stating it that way is more useful to the customer than declaring the system fixed.

The generalisable part

A compressor that cannot keep up with nothing added is a loss problem until the counters say otherwise, and the counters are usually already on the machine. Read loaded hours before you read pressure. Pressure tells you the header is low; loaded hours tell you how much air is actually being made, and that is the number that separates a machine problem from a plant problem.

References

• 29 CFR 1910.147, lockout and control of stored energy before opening the receiver or the pressurised side; 29 CFR 1910.169 for the receiver's required drain and indicating gauge; 29 CFR 1910.333(b)(2) for work inside the starter enclosure, with NFPA 70E-2021, 120.5 for the live-dead-live sequence where your employer's electrical safety program adopts it
• 29 CFR 1910.95, hearing conservation requirements attaching at an eight-hour time-weighted average of 85 dBA
• Manufacturer documentation for the compressor's rated delivery, control band and pump-up procedure
• See related: How to Decide Whether the Problem Is Supply or Demand; What Intake Conditions Do to Compressor Performance; How to Put a Number on What Leaks Are Costing; What a Pressure Flow Controller Is Actually Doing