Troubleshooting

Costly Fan and Blower Mistakes and How to Catch Them

The number is off because of density, unit, and system-curve errors that keep repeating in fan work. Here is each mistake by symptom, root cause, and the fix.

Symptom: the fan hits catalog CFM on the test stand but moves 15 to 20 percent less air in the field. Root cause is almost always uncorrected density. Catalog curves are rated at standard air, 0.075 lb per cubic foot at 70 F and sea level. At 400 F the density drops to about 0.046 lb per cubic foot, and at 5,000 feet elevation you lose another 17 percent. The fix: correct static pressure and power by the density ratio before you trust any point. Run the Airflow CFM and Static Pressure Power calculators at actual conditions, not standard, or you will chronically undersize.

Symptom: the motor trips on overload at startup even though it ran fine on the bench. The classic error is sizing the motor for operating density instead of cold-start density. A fan handling 600 F gas draws far more power when the system is cold and the air is dense, because brake horsepower scales directly with density. Air at 70 F is roughly 2.9 times denser than air at 600 F, so a motor sized for hot running can see nearly 3 times the load at ambient start. Fix: size the motor with Motor Sizing at the worst-case cold condition, then add a 10 to 15 percent service factor.

Symptom: measured static pressure is double the calculated value and the fan is buried on its curve. Root cause is missed or lowballed system resistance, usually dirty filters, transitions, and duct fittings counted as clean, straight runs. A 90 degree elbow can add 0.15 to 0.30 inches of water column, and a loaded filter easily triples its clean drop. Add every fitting loss coefficient and design to the dirty-filter condition, not clean. Undercounting resistance by 0.5 inches wc on a system rated at 2 inches wc pushes the operating point 25 percent off and starves the process.

Symptom: CFM readings swing wildly and never match the fan law prediction when you change speed. Two unit errors dominate. First, mixing inches of water with inches of mercury, a factor of 13.6, blows up any static pressure figure. Second, forgetting that CFM scales linearly with RPM while pressure scales with the square and power with the cube. Drop speed 20 percent and pressure falls 36 percent, power falls 49 percent. Teams that scale pressure linearly overestimate available lift badly. Lock units to inches wc and CFM, and apply the cube law before promising an energy saving from a slowdown.

Symptom: a new impeller vibrates at 0.30 in per second and throws the bearing life estimate. Root cause is specifying a balance grade looser than the duty needs, or balancing at the wrong plane count. An ISO 1940 G6.3 grade on a 1,800 RPM wheel allows a residual unbalance that a G2.5 process would reject. Use the Impeller Balance Tolerance calculator to convert grade and RPM into permissible unbalance in gram-millimeters, then verify the wheel is under 0.10 in per second on the stand. A single-plane balance on a wide wheel leaves a couple imbalance that no static check will catch.

Symptom: the quoted noise level passes on paper but the installed fan measures 8 to 10 dBA over spec. The common miss is treating sound power as sound pressure and ignoring the additive nature of decibels. Doubling the number of identical fans adds 3 dBA, not zero, and moving from 10 feet to 5 feet raises pressure by about 6 dBA. Estimate with the Noise Estimate calculator using sound power levels per octave band, then correct for distance and reflective surfaces. Skipping the distance correction is why field readings routinely embarrass a data sheet.

Symptom: the housing part comes off the press brake short on material and scrap climbs past 12 percent. Root cause is a blank layout that ignores bend allowance and nesting geometry. Each 90 degree bend in 16 gauge steel consumes roughly 0.4 to 0.5 inches of flat length depending on inside radius, and a layout that does not account for it produces undersized flanges. Run the Sheet Metal Housing Yield calculator with the real k-factor, near 0.44 for cold-rolled steel, before releasing the nest. A 3 percent yield gain on a 5,000-unit run is thousands of pounds of steel recovered.

Symptom: the test cell backs up and throughput falls below plan even though each test is quick. The overlooked variable is cell recovery and stabilization time, not just the measured run. A fan performance test may take 6 minutes of data but 4 minutes to load, seal, and stabilize airflow, so real cell time is 10 minutes and capacity is 40 percent lower than the naive estimate. Model it with the Test Cell Capacity calculator using door-to-door time, and align the line with the Assembly Takt calculator so you never feed the cell faster than it can clear.

Published 2026-07-01.