How to Tension a Belt Without Guessing

Why this matters

The thumb test has a resolution of roughly "tight" and "not tight," and it is wrong in both directions on the same drive depending on who does it. Force-deflection tensioning replaces the thumb with five recorded values, and the recording is the point: a tension you cannot reproduce next visit is a tension you cannot tell has changed. The single most useful thing to understand about the method is that the maker publishes two numbers for the same belt, a new-belt figure and a used-belt figure, and they are not two opinions about the same target. The new-belt number is a value you set once at installation. The used-belt number is a floor you measure against for the rest of the drive's life.

Before anything: isolate, and watch the tensioner

Lock out and tag the motor at its disconnect. A fan wheel or flywheel stores rotational energy after the power is off, which is stored mechanical energy under 29 CFR 1910.147, so verify zero rotation by eye and block the wheel or close and secure the damper before a hand goes past the guard. Where the drive has a spring-loaded idler or a spring-assisted motor base, release it only with the tool the design provides and stand clear of the arm's swing path; prying one loose with a bar puts the full stored energy through whatever the bar slips off. Opening an electrical enclosure to reach the disconnect is electrical work under 29 CFR 1910.333(b)(2), proved live-dead-live per NFPA 70E-2021, 120.5. The guard goes back on before restart under 29 CFR 1910.219 in general industry or 29 CFR 1926.300(b) on a construction site. If you are coming back for the run-in re-check, let the sheaves and belt cool or handle them with gloves rated for the surface temperature you measure, because a drive that has been running under load is hot enough to burn through thin work gloves.

The card you are filling in

Everything below produces one of five values. Write them on the job, not from memory afterwards.

1. Get the span length, not the center distance

The span is the free length of belt between the two tangent points, and it is shorter than the center distance. Compute it from what you can measure:

span = square root of (center distance squared - ((D - d) / 2) squared)

with D the large pitch diameter and d the small, all in the same unit. Skip this and use center distance instead and your deflection target is too large, which sets the belt tighter than the table intended. On drives with close diameters the two are nearly identical and the shortcut is harmless; on a high-ratio drive it is not, and you cannot tell which case you are in without doing the arithmetic.

2. Turn the span into a deflection distance

The convention is 1/64 in of deflection per inch of span, which is why span length matters more than any other input. That number is not arbitrary. Push a span of tension T sideways at its midpoint by a distance that is small relative to the span, and the transverse force needed is about 4 x T x deflection / span. Substitute a deflection of span divided by 64 and the span cancels: the force is T divided by 16. So the whole convention exists to make one deflection ratio correspond to one fixed fraction of strand tension, on any drive, at any span. The relation assumes both strands are at equal tension with the drive at rest, treats the belt as a flexible tension member, and ignores its bending stiffness, so read it as the reason the method works rather than as a substitute for the table.

The useful consequence: strand tension is about 16 times the deflection force. That gives you a way to sanity-check any figure before you set it.

3. Read both forces off the maker's table

The table is keyed to belt section, the pitch diameter of the small sheave, and belt speed. Belt speed in ft/min is pi x small sheave pitch diameter in inches x its rpm, divided by 12. All three inputs move the answer, and a figure remembered from a different drive is a figure for a different drive. Take the new-belt column and the used-belt column both, because you need the second one at the re-check and on every future service visit.

4. Measure it, on each belt, at more than one position

Use a tension tester that reads applied force, with its two markers set to the span length and the deflection distance. Apply the force at the exact midpoint of the span, square to the belt, on the outside face. Read the force at the moment the deflection marker reaches the span marker.

Then bar the drive over one full revolution by hand, gripping the sheave rim and keeping fingers clear of the point where the belt enters the groove, and repeat at a different belt position. Belts vary slightly around their own length, and a single reading can be taken at the loosest or tightest point on the loop. On a multi-belt drive, measure every belt: the reading is per belt, and a spread between belts is a load-sharing problem the total tension cannot tell you about.

5. Adjust, then re-measure alignment

Every tension change moves the motor, and moving the motor moves alignment. Check alignment after the final tension adjustment, not before it, or you have verified a position the drive is no longer in.

6. Come back after run-in

Belt makers commonly call for a re-check after a run-in period in the range of the first 24 to 48 hours of operation, and the interval for your section is the one they publish. This is the step that decides the tension the drive actually lives at, because seating into the grooves and the belt's initial permanent set both happen early and both drop tension. At the re-check you are measuring against the used-belt figure, not the new-belt figure.

Worked example: a two-belt pump drive

Motor nameplate 7.5 hp at 1750 rpm. Driver pitch diameter 4.0 in, driven 10.0 in, center distance 20.0 in, two B-section belts.

Span. The square root of (20.0 squared - ((10.0 - 4.0) / 2) squared) = the square root of (400 - 9) = 19.8 in.

Deflection. 19.8 / 64 = 0.31 in, so set the markers at 5/16 in.

Belt speed. pi x 4.0 x 1750 / 12 = about 1,830 ft/min.

Table figures. Say the maker's table for a B section on a 4.0 in small sheave near 1,830 ft/min returns 8.0 lb new and 5.5 lb used. Use your own table; these are placeholders for the arithmetic.

Sanity check against the load. Motor torque at nameplate is 63,025 x 7.5 / 1750 = 270 lb-in, so the total net pull at the 2.0 in driver radius is 135 lb, or 67.5 lb per belt across two belts. Apply an illustrative service factor of 1.2 for this machine class and duty and the peak per belt is 81 lb. At a design tension ratio of 5 that means a slack side of 81 / 4 = 20.3 lb and a tight side of 101.3 lb per belt, so the mean strand tension is about 61 lb. Divide by 16 and the load implies a deflection force of about 3.8 lb per belt.

That is roughly half the table's new-belt figure, and the gap is information rather than an error. The table is keyed to what a B belt at that diameter and speed is rated to carry, not to what this drive is asking of it, so a drive running well inside the section's capacity will always compute lower than the table. Set to the table. The margin between the two numbers is the reserve that covers the start, the cold morning and the process upset, and a tech who "computes" his way down to 3.8 lb has deleted it. What the gap does tell you is that this drive is generously sized, which is worth knowing the next time someone asks whether it can take more load.

Measured on arrival. Belt A 4.5 lb, belt B 4.4 lb. Both are below even the 5.5 lb used-belt floor, so this drive was under-tensioned before anyone touched it, and the two readings agreeing tells you it is a tension problem and not a matching problem.

After adjustment. Belt A 8.0 lb, belt B 7.6 lb, both taken at two belt positions. The spread is 0.4 lb on 7.8, about 5 percent, which is small enough that the two belts will share load. A spread of tens of percent is a matched-set problem and no amount of adjustment fixes it, because both belts sit at the same center distance and their tensions are set by their own lengths.

At the 30-hour re-check. Belt A 6.2 lb, belt B 6.0 lb. Tension has fallen about 22 percent from where it was set, which is normal seating and set, and both readings are above the 5.5 lb used-belt floor. Nothing to do. This is the case people get wrong in both directions: one tech re-tensions back to 8.0 lb because that is the number he wrote down, over-tensioning a bedded-in drive; another skips the visit entirely and never learns whether it fell past the floor. The two-column table exists to settle exactly this, and the answer here was to record the numbers and close the guard.

What changes the method

A synchronous (toothed) belt is not tensioned by this logic at all. It does not rely on friction, so its tension exists to keep the teeth in mesh and control tooth entry, and the maker's procedure and figures are specific to the tooth profile. Do not carry a V-belt deflection number onto one.

An automatic or spring-loaded tensioner sets its own force, and your job is to confirm the arm is inside its published travel range rather than to measure deflection. An arm at the end of its travel is telling you the belt has stretched past the tensioner's authority, which is a replacement, not an adjustment.

A drive with less than about 120 degrees of wrap on the small sheave has little friction margin, and tensioning it to the table is treating a geometry problem with strand load. Fix the geometry.

References

• 29 CFR 1910.147 for mechanical isolation and stored rotational energy; 29 CFR 1910.333(b)(2) and NFPA 70E-2021, 120.5 for work at an electrical enclosure and live-dead-live proving
• 29 CFR 1910.219 (general industry) and 29 CFR 1926.300(b) (construction) for guarding of belts, pulleys and sheaves
• Belt manufacturer engineering data for new-belt and used-belt deflection forces by section, small-sheave pitch diameter and belt speed, service factors, and run-in re-tension intervals
• See related: What Belt Tension Actually Controls; Why Matched Sets Matter on Multi-Belt Drives; How to Check Sheave Alignment