Why a Compressor Room Needs More Ventilation Than It Has
Why this matters
Compressor rooms get chosen for what they are convenient for: a spare closet, a mezzanine corner, the end of a maintenance bay. Nobody sizes them, and the ventilation gets whatever louvre and fan the general contractor had. Then the plant grows, the compressor runs a higher duty, and one August the machine starts shutting down in the afternoon. The room did not change. The heat did, and the ventilation was never sized against heat in the first place. This is a finding a shop can produce with a thermometer, a power meter and one calculation, and it turns a compressor complaint into a building fix that is usually cheaper than the machine.
A room in this condition is a heat exposure before it is a diagnostic problem. Limit how long anyone stays in it, work in pairs, drink water, and take readings from outside the door on a probe with a lead wherever the reading allows it. Wear hearing protection under a program meeting 29 CFR 1910.95 for the whole time you are in there, and stay off the discharge piping between the airend and the aftercooler, which runs hot enough to burn on contact.
The requirement is a heat balance, not a room size
Room volume does not enter this calculation anywhere. A large room simply takes longer to reach the same steady temperature as a small one. What sets the steady temperature is the balance between heat put in and heat carried out, and heat is carried out by air moving through.
The airflow needed to carry a given heat load at a given temperature rise, for standard air at about 0.075 pounds per cubic foot, which is sea level at ordinary temperatures:
Airflow in cubic feet per minute equals the heat in BTU per hour, divided by 1.08 times the temperature rise in degrees Fahrenheit.
The 1.08 constant is derived at that standard density and it shrinks roughly in proportion to density, so at about 5,000 feet, where the standard atmosphere is near 12.2 psia against 14.7 at sea level, use roughly 0.90 rather than 1.08 or the calculation runs about a fifth optimistic. It also drifts downward as the air itself gets hot, which is another reason not to design for a large rise.
Two more figures the arithmetic below depends on. One kW is 3,412 BTU per hour. And for an air-cooled package, essentially all of the electrical input to that package appears as heat in the room; the heat recovery article in this library owns that claim and the split between the cooling air stream and the casing.
What sets the rise you are allowed
This is the step everyone skips, and it is why rooms get ventilated to a comfortable-sounding number and still fail.
The allowable temperature rise is not a preference. It is the package's maximum permitted ambient or intake temperature, from the manufacturer's data, minus the outdoor design dry bulb temperature for the site. Whatever is left over is the rise the ventilation is allowed to produce. On a mild site with a machine rated for a high ambient the allowance is generous. On a hot site with a machine rated conservatively, the allowance can be single digits, and single digits is a very large airflow.
Two consequences follow, and the second is the one that saves projects.
First, the required airflow rises steeply as the allowable rise falls, because the rise is in the denominator. Halving the allowable rise doubles the airflow, which usually means the louvre free area doubles too, and free area is where retrofits die.
Second, when the required airflow is impossible for the room, you stop trying to remove the heat and start preventing it from entering. Ducting the package's cooling-air discharge straight through the wall takes the majority of the heat out before it ever mixes with room air, and the ventilation then only has to handle the casing and radiant losses left behind.
The case: shutdowns every August
A single air-cooled rotary screw in a walled-off room at the end of a maintenance bay, sea level. Complaint: high-temperature shutdowns between roughly 2 and 4 in the afternoon, in August only, no faults in winter, no faults in the morning. The machine had already had a cooler cleaning and an oil change, and both were done correctly.
The seasonal and time-of-day pattern is the tell. A fouled cooler, low oil or a failed thermostatic valve does not know what month it is. A fault that appears only when outdoor temperature peaks and disappears when it drops is a fault whose driver is outside the machine, so the room got measured before anything else came off the package.
What the room measured. At 3 in the afternoon, outdoor dry bulb 92 F. Air at the package's intake, measured in the stream with the probe secured, 128 F. That is a 36 F rise across the room. A logging power meter on the package over that hot window returned an average input of 34 kW, against a measured full-load input of 40 kW.
Work out what the ventilation is actually doing. 34 kW times 3,412 is 116,008 BTU per hour going into the room during that window. Rearranging the balance, effective airflow equals 116,008 divided by 1.08 times 36, which is 116,008 divided by 38.88, or 2,984 cubic feet per minute. Call it 3,000.
The wall fan is rated 2,000 cubic feet per minute. The room is therefore moving about half again as much air as the fan alone would explain, and the rest is infiltration through the door and around the louvre. That gap is worth knowing: it means the fan is not underperforming, and a fan-only fix has less headroom than its rating suggests.
Work out what it needs to do. The package's data states a maximum permitted ambient of 104 F, which is illustrative here and must come off your own machine. The site's outdoor design dry bulb is 95 F. Allowable rise is 104 minus 95, which is 9 F, and the calculation has to be run at full-load heat rather than at the hot-window average, because the machine can and does run at full load. So the required airflow is 136,480 divided by 1.08 times 9, which is 136,480 divided by 9.72, or 14,041 cubic feet per minute. Call it 14,000.
Fourteen thousand against three thousand is a factor of 4.7. There is no fan that goes in that wall, and if there were, the louvre free area to feed it does not exist either. This is the point where ventilation stops being the answer.
Why the fix was a duct, not a bigger fan
Take the majority of the heat out at the source. The package's own data gives the fraction of input energy leaving in the cooling air stream, as against casing and radiant losses; on this machine that figure is 75 percent, the same illustrative basis used in the heat recovery article and the same number you must confirm on your own package. Duct that discharge straight through the wall and the heat left in the room is 25 percent of 136,480, which is 34,120 BTU per hour.
Now re-run the requirement at the same 9 F allowable rise: 34,120 divided by 1.08 times 9 is 3,510 cubic feet per minute. The room already moves about 3,000, though that figure was inferred with all the heat still in the room, so re-measure the balance once the duct is in rather than assuming the infiltration term survives the change. A modestly larger fan and a cleaned, correctly sized louvre closes it, and the whole job is duct, fan and louvre rather than a compressor.
Two conditions would change that recommendation. If the package's fan cannot carry the duct's resistance, the duct reduces cooling airflow through the machine and makes the problem worse rather than better; get the allowable external static pressure from the manufacturer before drawing the run, and where it is exceeded, add a booster fan interlocked to the compressor rather than lengthening the package fan's work. And if the discharge duct is fitted with a winter damper that returns warm air into the building, which is worth doing, that damper failing in the return position in August recreates the original fault exactly, so it gets a position indication and an August check.
The air path that has to exist
Where the openings sit matters as much as how big they are. A room can move the calculated airflow and still fail if the hot air finds its way back to the package's intake instead of leaving.
louvre low, package cooling
far wall air ducted out
| |
v v
+------------------------------------------------+
| cool air enters low |
| |
| +---------------+ |
| | compressor | hot air leaves |
| ----> | package | high, through |
| +---------------+ the wall |
| |
| cool air crosses the floor before it is heated |
+------------------------------------------------+
The failure this arrangement prevents is short-circuiting: an inlet and an outlet on the same wall, or a package sitting between them by a couple of feet, moves air from one opening to the other without ever reaching the machine, and the measured airflow will look correct while the intake temperature stays high. The intake conditions article in this library covers what that intake temperature then does to the machine's delivered capacity, which is a second cost on top of the shutdowns.
Confirming it, and the hazard the fix creates
Confirm by repeating the same measurement in the same window: outdoor dry bulb, intake temperature in the stream, and average input power over the window. Rise falling to within the allowable figure at the same or higher outdoor temperature is the proof. Rise that improves while outdoor temperature also dropped proves nothing, which is why the outdoor reading is recorded every time.
The hazard the fix creates is worth more attention than the fix. A larger exhaust fan depressurises the room. If any fuel-fired appliance shares that room or the space that feeds it, a water heater, a unit heater, a boiler with an atmospheric vent, depressurisation can reverse its draft and spill combustion products, including carbon monoxide, into an occupied building. So: size and provide a dedicated makeup air path for the TOTAL exhaust, the wall fan plus the package's own cooling airflow now leaving through the duct, which on an air-cooled package is usually the larger of the two, rather than relying on infiltration. Take the package's cooling airflow from the manufacturer's data; it is the number that sets the louvre, and where a fuel-fired appliance shares the space, verify the appliance's draft with the new fan running and every other exhaust in the building running, before the job is called complete. Fit a carbon monoxide alarm in the room and treat any alarm as an evacuation, not a nuisance.
Electrical work on the fan's branch circuit is a different isolation from work on the compressor. De-energise and lock the fan circuit and verify absence of voltage under 29 CFR 1910.333(b)(2), using the live-dead-live proving sequence in NFPA 70E-2021, 120.5, in the edition your employer's electrical safety program has adopted. Mechanical and stored-energy isolation of the compressor package itself, including relieving sump and receiver pressure, is 29 CFR 1910.147, which expressly excludes electrical work on utilization installations at (a)(1)(ii)(C). They are two separate procedures on the same job and both get done.
References
- 29 CFR 1910.95, occupational noise exposure, general industry: the basis for hearing protection while working in a compressor room.
- 29 CFR 1910.333(b)(2), general industry electrical safe work practices, for de-energising and verifying the ventilation fan's branch circuit; and 29 CFR 1910.147, control of hazardous energy, for mechanical and stored-energy isolation of the compressor package, which 1910.147 covers and electrical utilization work does not.
- NFPA 70E-2021, 120.5, in the edition your employer's electrical safety program has adopted, which binds through that program and through the employer's own written procedures: the basis for the live-dead-live proving sequence.
- Compressor package manufacturer's technical data, for the maximum permitted ambient temperature, the fraction of input energy leaving in the cooling air stream, and the fan's allowable external static pressure.
- See related: What Intake Conditions Do to Compressor Performance; What Heat Recovery From a Compressor Can Realistically Deliver.