How to Check Whether a Change Put the Drive Outside Its Range
Why this matters
A drive is not one rated thing. It is a stack of separately rated things - a wheel or impeller, a shaft, bearings, a coupling or belt set, a motor, a controller - and each carries its own limit, published by a different manufacturer, derived under its own condition. A change that stays comfortably inside one of those limits can walk straight past another, and the reason this keeps happening is that the limit people check is the one they can see on a nameplate, while the limit that fails is usually the one that lives in a table nobody opened. The output of this procedure is a sheet with one row per component. The sheet is the deliverable, and the discipline it enforces is that every row gets its own verdict.
Before you change a speed setting
A wheel or impeller has a maximum safe speed, and exceeding it is a burst hazard, not a wear problem. Do not raise a speed limit, change a sheave ratio, or lift a frequency ceiling until you have the wheel's maximum safe speed from the manufacturer's data for that wheel class and construction, at the actual airstream or fluid temperature, because those ratings derate with temperature. If you cannot obtain it, the change stops there. Any test run at the new speed is made with the cabinet closed and all guards fitted, with everyone out of the plane of rotation and clear of the discharge.
Opening a variable-frequency drive is energized electrical work under 29 CFR 1910.333(b)(2), not 29 CFR 1910.147, which excludes exposure to electrical hazards from work on conductors and equipment in electric utilization installations at (a)(1)(ii)(C). The drive's DC bus capacitors hold a lethal charge after the supply is removed, so wait the manufacturer's stated discharge time, then prove absence of voltage on the bus using the live-dead-live sequence in NFPA 70E-2021, 120.5, in arc-rated PPE and voltage-rated gloves. For any mechanical work on the fan, coupling or bearings, lock and tag the energy-isolating device under 29 CFR 1910.147 and watch the wheel until it has fully coasted to a stop.
Step 1: Write one row per component before you evaluate anything
List every component in the power path and every component the power path acts on. For a direct-drive fan that is the wheel, the shaft, the shaft bearings, the motor, the drive controller, and the system the fan discharges into. For a belt-driven machine add the belts, both sheaves and the arc of contact. Write them all down before you start looking anything up, because a row you never wrote is a check you will never fail.
Step 2: For each row, record the limit, its source, and the condition it was stated under
Three fields, and the third one is the one that gets skipped:
- The limit as a number with its unit.
- The source: which manufacturer's document or which standard it came from. "I remember it being around..." is not a source and does not go on the sheet.
- The condition it was derived under: at what temperature, at what lubrication method, at what mounting, on what side of base speed. A limiting speed quoted for oil lubrication is not the limit for a grease-packed bearing, and a maximum safe wheel speed quoted at ambient is not the limit in a heated airstream.
Step 3: Compute the post-change value for each row from the change itself
Do not measure first. Compute what the change implies for each row from the physical relationships, so you know which rows to go and measure and what you expect to find. Measuring first invites you to accept whatever the machine is doing as evidence that it can do it.
Step 4: Give every row its own verdict, and do not let one row vote for another
Pass, fail, or unknown. Unknown is a legitimate verdict and it stops the job exactly as a fail does. A row cannot inherit a verdict from a neighbour: the motor being inside its rating tells you nothing about the wheel, the wheel being inside its rating tells you nothing about the bearings, and all three being fine tells you nothing about whether the controller can deliver the torque at the new speed.
The sheet, filled in: raising a fan's frequency ceiling from 60 Hz to 66 Hz
The complaint is not enough air. The proposed fix is to raise the variable-frequency drive's maximum output frequency from 60 Hz to 66 Hz on a direct-drive plenum supply fan, a 10 percent speed increase. Nothing in the system is being changed, so the fan relationships apply.
| Row | Limit and source | Condition it holds under | After the change | Verdict |
|---|---|---|---|---|
| Wheel maximum safe speed | Fan manufacturer's class rating for that wheel | Derates with airstream temperature and wheel construction | 1.10 times the previous rpm | Must be confirmed against the manufacturer's figure at the actual airstream temperature; unknown until it is |
| Shaft power against motor rating | Motor nameplate rating | At rated voltage, rated frequency, nameplate ambient | Power goes as speed cubed: 1.10 cubed is 1.33, and 0.78 of rating becomes 1.04 of rating | Fail on a drive-fed motor, where service factor is generally 1.0 unless the motor is inverter-duty rated |
| Motor torque available above base speed | Motor and drive data | Above base speed the drive cannot hold the volts-per-hertz ratio, so flux falls | Available torque falls roughly as 1 over speed: 0.91 of rated. Demanded fan torque goes as speed squared: 0.78 times 1.21 is 0.94 of rated | Fail, and it fails before any thermal limit does |
| Motor cooling | Motor manufacturer | Shaft-mounted fan on a totally-enclosed fan-cooled motor moves more air as speed rises | Improves slightly above base speed | Pass, and worth recording as a pass so nobody re-litigates it. The cooling limit on this motor lives at the low-speed end, not here |
| Bearing limiting speed | Bearing catalog, speed factor of mean diameter times rpm | Differs by lubrication method; grease-lubricated limits are lower than oil | 1.10 times rpm, and grease life shortens with speed | Check against the catalog limit and shorten the relubrication interval accordingly |
| Shaft and rotor critical speed | Fan manufacturer's rated speed range | First bending critical for that shaft and wheel assembly | Operating speed moves 10 percent closer to whatever lies above it | Check; the manufacturer's stated safe operating range is the source, not a calculation you make on site |
| Residual unbalance force | Balance grade the wheel was accepted to | Unbalance force goes as the square of speed for a given residual unbalance | 1.10 squared is 1.21, so 21 percent more force from the same residual unbalance | Check the wheel's balance grade; a wheel that was marginal at 60 Hz is worse here |
| System pressure | Duct or piping pressure class and any relief | Same system, speed change only | Pressure goes as speed squared: 1.21, so 21 percent more | Check against construction class and any pressure-relief setting |
| Controller output current | Drive continuous current rating and its overload profile | The drive's own time-limited overload allowance, typically a stated percentage for a stated number of seconds | Rises with the motor's current at the new load | Check against the drive's continuous rating, not its peak |
Where this change actually dies. Not on the thermal row, which is where people look. It dies on the torque row, and the arithmetic is worth following because it is not intuitive.
Below base speed a drive holds the volts-per-hertz ratio and the motor keeps roughly constant torque capability. Above base speed the output voltage is capped by the supply, so as frequency rises the flux falls and available torque falls with it, approximately as 1 over speed. At 1.10 times base speed, available torque is about 0.91 of rated.
Meanwhile the load is a fan, whose torque demand rises with the square of speed. At the previous setting the fan drew 0.78 of the motor's rating, so at 1.10 speed the demanded torque is 0.78 times 1.10 squared, which is 0.78 times 1.21, about 0.94 of rated torque.
Demand at 0.94 against availability at 0.91. The motor cannot hold 66 Hz at that load, and it will not fail loudly: it will simply not reach the setpoint, or it will reach it and fall back under any additional resistance, and the airflow complaint will persist while everyone believes the change was made.
One hedge belongs in the same breath as that 0.91. Treating available torque as falling exactly as 1 over speed is the constant-horsepower approximation, and the motor's breakdown torque above base speed falls faster than that, roughly as the square of speed, so the real usable margin degrades sooner than the constant-power line suggests. The conclusion is not marginal; it is worse than the line implies.
The row that would have been a safety event. The torque row costs a wasted afternoon. The wheel maximum-safe-speed row is the one that puts metal through a cabinet, and it is the row most likely to be skipped because a fan wheel does not carry a nameplate the way a motor does. If that figure is not available for the wheel at the actual airstream temperature, the verdict is unknown and the change does not happen, regardless of how every other row scored.
What passing one row does not tell you. In this sheet the motor cooling row genuinely passed and the thermal question people would have asked first was never the constraint. If the sheet had been a single overall judgment instead of nine independent rows, that pass would have carried the change forward. A per-row verdict is what stops one comfortable answer from covering for three uncomfortable ones.
How to verify you got this right
- Confirm every row has a source, not a recollection. A sheet with the source column blank is a list of guesses in table form.
- Confirm every condition field is filled, and that it matches the machine in front of you: lubrication method, airstream temperature, which side of base speed, mounting orientation.
- After any approved change, re-measure the rows you predicted, at thermal steady state rather than immediately: motor current, bearing temperature rise over ambient, and the paired system readings. Predicted and measured should agree in direction and roughly in magnitude, and a disagreement means one of your conditions was wrong.
- Write the revert criterion before you make the change, as a number: the reading at which you put the setting back. Deciding that afterwards, with the customer standing there, produces a different answer than deciding it beforehand.
- File the sheet with the equipment record. The next person to be asked for more air needs to know this was already tried and which row stopped it, or they will try it again.
References
- AMCA published fan performance and rating practice, for fan-law conditions and wheel class and maximum safe speed ratings
- NEMA MG 1, Motors and Generators, for service factor conditions and motor behaviour on adjustable-frequency supplies
- 29 CFR 1910.333(b)(2), electrically safe work practices for work on drives and controllers; NFPA 70E-2021, 120.5, for the live-dead-live verification sequence
- 29 CFR 1910.147, control of hazardous energy, for mechanical isolation and stored rotational energy on wheel, bearing and coupling work
- Manufacturer bearing catalogs for limiting speed by lubrication method, and drive documentation for continuous current rating and overload profile
- See related: How a Fan Law Changes What the Drive Has to Deliver; What a Service Factor Actually Buys You; How to Decide Whether a Drive Is Undersized or Misapplied