Why a Belt Slips and What It Costs

Why this matters

A slipping belt almost never announces itself. It does not squeal on a steady load, it does not throw a fault, and the machine keeps running. What it does is settle quietly into a new equilibrium where the drive turns slower than the sheaves say it should, and everything downstream is a little short forever. The following case is worth walking because the belt was not the fault, the tension was correct, and replacing either one would have produced a callback inside a month.

Isolate before your hands go in

Every hands-on step below happens with the motor locked out and tagged at its disconnect. A blower wheel holds real rotational energy and will windmill on duct airflow with the power off, which makes it stored mechanical energy under 29 CFR 1910.147: verify zero rotation by eye and block the wheel or close and secure the damper before reaching past the guard. The two readings taken with the drive running are taken from outside a closed guard, which stays on under 29 CFR 1910.219 in general industry and 29 CFR 1926.300(b) on a construction site. In a fan room at or above the 85 dBA eight-hour action level, hearing protection under a program per 29 CFR 1910.95 applies to the time you spend standing there taking readings.

The signal

A rooftop unit's supply fan, belt driven, in service six years. The complaint was thin air at the far end of the building, arriving about three weeks after a contractor had rebalanced the system. Nothing had been serviced on the unit itself.

Drive geometry, measured with the unit locked out: driver pitch diameter 5.0 in, blower pitch diameter 10.0 in, motor nameplate 1750 rpm. Predicted blower speed is 1750 x 5.0 / 10.0 = 875 rpm.

Optical tachometer on the blower shaft, taken through the guard's sight opening with the guard closed: 812 rpm. That is 63 rpm below the nameplate-referenced 875, or 7.2 percent, and every figure below is computed against that same 875 reference. The motor shaft read 1757 rpm at the same moment, above its 1750 nameplate full-load speed, so the shortfall was in the drive, not in the motor.

Creep is not slip, and the number separates them

Two different things get called slip and only one of them is a fault.

Elastic creep is unavoidable. The belt enters the driver sheave at tight-side tension and leaves at slack-side tension, so it is physically shorter on the way out than on the way in, and it contracts as it travels around the arc. That contraction shows up as the belt crawling backwards relative to the sheave surface. It is a function of the tension difference and the belt's stiffness, it exists on every correctly running V-belt drive, and on a properly loaded drive it is normally well under 1 percent of speed. You cannot design it out and you should not chase it.

Gross slip is different in kind. It happens when the torque demanded exceeds what friction can hold, so the whole belt slides against the sheave face rather than deforming with it. It is not proportional to load in any gentle way: below the friction limit it does not happen at all, and above it the belt lets go.

7.2 percent is far outside the creep band. That reading, on its own, said gross slip and said the drive had run out of friction capacity.

The three obvious candidates, and why each died

Tension low. Checked by force-deflection against the belt maker's published figure for the section, small sheave diameter and belt speed. It was inside the specified window. This is the reading that sends most techs down the wrong road, because it eliminates the assumed cause without suggesting another one.

Contamination. Sheave grooves and belt sidewalls were dry, no oil film, no glaze, no rubber dust bank on the guard floor beyond normal. A wipe with a clean rag changed nothing on a repeat tach reading.

Motor not delivering. Motor shaft at 1757 rpm under load against a 1750 nameplate, running current inside nameplate full-load amps. Nameplate speed is the speed at rated load, and induction slip rises with load, so a motor turning faster than nameplate is carrying less than rated load, not straining. The motor was doing its job. Note that the amp reading was taken with a clamp meter at an accessible point that did not require opening an energized enclosure; opening one to meter would be energized work under 29 CFR 1910.333(a)(1), requiring justification and arc-flash protection per NFPA 70E.

Sheave groove wear was checked and ruled out separately, with a groove gauge, drive locked out: belt tops sat flush with the rims and the flanks were straight. That failure mode has its own article and was not this one.

What was left was the load

Nothing had been taken away from the drive's capacity. So the demand on it must have risen, and the timing pointed at the balancing visit. The zone damper serving the large interior zone nearest the unit had been opened two positions and left there. That zone was not the complaint; the far end was, and it lost twice, once to the fan slowing and again to a nearer branch that now took a larger share of what was left.

Here is the mechanism. Opening a damper moves the fan's operating point along its curve toward higher flow, and a centrifugal fan absorbs more shaft power at higher flow. Required torque rose. Friction capacity did not, because tension and wrap angle had not changed. Once demanded torque crossed the friction limit the belt began to slide, the blower slowed, and slowing reduced the torque the blower demanded, until demand fell back to exactly the friction limit and stayed there. That is why a slipping drive does not stall or scream: it finds a stable slipping speed and holds it, indefinitely, quietly.

That equilibrium is also a measurement. On a fixed system curve, fan torque scales with the square of speed, and the damper had been in its new position for both readings, so comparing 812 rpm against 875 rpm on the same curve is legitimate. The torque needed to hold 875 rpm is therefore (875 / 812)^2 = 1.16 times the torque the drive can currently transmit. The drive was 16 percent short of capacity, and the slip itself handed over that number.

What the slip actually cost

Along that same fixed system curve, with speed at 812 against a design 875, the speed ratio is 0.928:

• Air delivered scales with speed, so 7.2 percent less than design. That is the customer complaint, in one line.
• Static pressure scales with the square, 0.928^2 = 0.861, so about 14 percent low. On a system with any meaningful duct resistance, that is where the far-end diffusers stop throwing.
• Shaft power absorbed by the fan scales with the cube, 0.928^3 = 0.799, so the blower is taking about 80 percent of its design power.

Now account for where the motor's output went. Power into the belt is net pull times the driver surface speed; power out is the same net pull times the driven surface speed. The gap between them is the slip fraction, 7.2 percent, and it does not vanish. Motor output was 0.799 / 0.928 = 0.861 of design shaft power, the fan received 0.799, and the difference, about 6 percent of design shaft power, went into the contact patch as heat. The motor was running easier than at design while the building got less air and a small heater ran continuously inside the guard.

That heat is what ends the belt. A commonly used rule of thumb from steady-state oven aging of elastomers is that service life roughly halves for each 18 F rise in bulk temperature. That rule describes uniform bulk aging, not the local surface scorching of a sliding contact, which is faster and shows as a glazed, hardened sidewall. So the belt was going to fail, and the failure would have been read as a bad belt.

The fix, and what the alternative fix would have cost

The drive needed 16 percent more torque capacity. Two routes, and the difference matters.

More tension. Net pull scales with the tension difference, so covering the shortfall at the same tension ratio means about 16 percent more tension on both strands, and where belt pull dominates the bearing load, as it does on most belted fan and motor shafts, the bearing load rises by that same 16 percent. Bearing life falls as roughly the cube of load for ball bearings, so 1 / 1.16^3 = 0.64: about a third of the bearing life gone, on both the motor and the blower, to recover air the drive was never sized for. It also runs the belt at a higher mean tension for its whole life. It works, and it is the wrong first move.

More drive. Increasing the small sheave's pitch diameter raises the effective radius, so the same torque needs less net pull, and it simultaneously increases wrap angle on the small sheave because the diameter difference shrinks. Both effects add capacity without adding shaft load. The trap is doing it to the driver alone: enlarging the driver raises blower speed, and the motor torque the drive has to carry rises with the cube of the driver diameter while the capacity that diameter buys rises only in proportion to it, so you lose ground. The ratio has to be held by moving both sheaves, or capacity added by going to a wider section or an additional groove.

What would flip this recommendation: if the balancing change was itself wrong and the zone was never meant to see that flow, the correct fix is neither, it is putting the damper back and re-balancing. Always establish whether the new load is legitimate before you buy capacity to serve it.

How they confirmed it held

Same optical tachometer, same guard-closed position, after the drive change: blower at 871 rpm against a recomputed prediction of 874 from the new pitch diameters. A 0.3 percent gap sits inside creep and inside the tachometer's own resolution, so the drive was no longer slipping. The belt's running surface temperature was checked with a non-contact infrared thermometer aimed through the guard opening rather than by hand, and it settled near the sheave and enclosure temperature instead of sitting well above it. Air readings at the far diffusers came back on the next visit.

The habit worth keeping from this: tach the driven shaft on every belt drive you have the guard open near, compute what it should be from the pitch diameters, and write both numbers down. A drive that has quietly given up 7 percent will never tell you any other way.

References

• 29 CFR 1910.147 for mechanical isolation and stored rotational energy; 29 CFR 1910.333(a)(1) and NFPA 70E-2021 for energized-work justification and protection
• 29 CFR 1910.219 (general industry) and 29 CFR 1926.300(b) (construction) for guarding of belts and sheaves; 29 CFR 1910.95 for the occupational noise program and its 85 dBA action level
• Belt manufacturer engineering data for force-deflection tension values by section, sheave diameter and belt speed
• See related: How a Belt Drive Transmits Torque; What Over-Tensioning Does to the Bearings Either Side; What a Worn Sheave Groove Does to a New Belt