Why Preload and Not Friction Holds a Joint Together

Why this matters

A bracket that keeps walking on its slotted holes gets treated as a friction problem, and the fixes follow: rough up the mating faces, add a lock washer, put a grippy compound in there, use a bigger bolt. Sometimes one of them buys a season. The reason they mostly fail is that friction is not a thing you install in a joint. It is a number the joint produces, and the only term in it that a tech sets is clamp force. Preload is the input. Friction is the output. If you fix the output you have fixed a symptom of the input.

Before touching a bracket, hanger, or mount on any driven equipment: lock and tag the energy source and verify the machine cannot start before a wrench goes near it, because a fan or pulley that coasts or restarts while a mount is loose will take a hand with it. That duty for mechanical and stored energy is 29 CFR 1910.147 in general industry, and the electrical half of an isolation, where you are opening a disconnect or working on the conductors, is 29 CFR 1910.333(b)(2). Set an independent support under any equipment its mount is carrying before you slacken a bolt, and stay out from under the piece rather than beside it.

The call: a blower mount that walked twice

A shop was on its third visit to the same air handler. The blower assembly sat on a steel sub-base, four bolts through slotted holes so the belt could be tensioned. Each visit the assembly had crept, the belt had gone slack, and the mounting holes were showing bright witness marks where paint had been scrubbed off the base.

The first visit re-set the position and tightened the bolts hard by hand. The second added split lock washers under the nuts. The third is where somebody asked what number they were actually working against.

The number the joint was working against

A joint loaded sideways resists by friction across the faces that touch, called the faying surfaces. The resistance is the clamp force pressing them together, times a slip coefficient for that surface condition, times the number of faying surfaces in the stack. Here there is one faying surface, the underside of the mount foot against the top of the sub-base.

Slip coefficients are properties of the surface pair, and the published values that field people meet most come from structural steel practice, where clean mill-scale steel is treated at about 0.30 and blast-cleaned steel at about 0.50. Those are surface classes defined for structural faying surfaces in the RCSC specification, in whatever edition the AISC specification and the governing building code adopted by the authority having jurisdiction bring in, and they bind a field-service shop only through a contract or a permitted structural scope. For equipment mounting they are a useful order of magnitude, not a specification. A painted surface is lower and, with an unqualified coating, can be very much lower.

Take the bolts as being installed to 8,000 lbf of preload each. Four bolts is 32,000 lbf of clamp. Against clean steel at 0.30:

• Slip resistance = 32,000 x 0.30 x 1 = 9,600 lbf
• Belt pull trying to drag the mount, call it 6,000 lbf at the tension that assembly runs
• Ratio: 9,600 divided by 6,000, about 1.6 to 1

At those numbers the joint does not move. It never should have needed a second visit.

What was actually installed

Nobody had measured the preload. The bolts had been tightened by feel, on a plated fastener, with paint under two of the four heads. Assume the joint achieved 45 percent of the 8,000 lbf target, which is well inside the spread that feel-tightening produces on plated hardware.

• Actual clamp = 0.45 x 32,000 = 14,400 lbf
• Slip resistance = 14,400 x 0.30 = 4,320 lbf
• Against a 6,000 lbf drag, that is a ratio of 0.72 to 1

The joint was not marginal. It was arithmetically guaranteed to slip, and it slipped every time, and every fix so far had been aimed at the coefficient.

Why fixing the coefficient could not win

Run the two levers side by side, which is the whole point of doing this in numbers.

Blasting the surface moves the coefficient from 0.30 to 0.50, a lever of about 1.67, and that is close to the ceiling: surface preparation has a top end and you are already near it once the steel is bare and clean. Restoring preload moves clamp from 45 percent to 100 percent of target, a lever of about 2.22, and it costs a torque wrench rather than a surface treatment on an installed machine.

More to the point, the blast-only row lands at 1.20. It holds on the day. It has no margin for the preload loss that every joint takes over its first heat cycles, and preload was already the thing this joint was bad at. That is the shape of a fix that works for one season and brings the truck back.

If you do dress a faying surface, the act itself carries a hazard the mount does not: abrading steel, primer or paint releases an airborne dust, and the control is respiratory, not a glove. Blasting or grinding a substrate that contains crystalline silica puts respirable silica in the air, controlled under 29 CFR 1910.1053 in general industry, and coatings on older structures can carry lead, controlled under 29 CFR 1910.1025. Use local exhaust or wet methods and a respirator selected and fit-tested under a written program meeting 29 CFR 1910.134, and do not substitute a nuisance dust mask for it.

Why the lock washers did nothing

The split lock washer added on visit two is worth its own line, because it is the most common wrong answer to this exact symptom. A split washer is a spring that flattens well below the preload of a properly tightened bolt, and once flat it is a washer. It adds nothing to clamp force, so it adds nothing to slip resistance. It was never going to change a single number in that table.

Where the joint had actually been slipping, the split washer also could not help, because slipping is a movement of the members and the washer sits between the nut and the member. What resists movement of the members is the squeeze across the faces. There is no path by which a washer under a nut generates that.

What happens after it slips, which is the part that compounds

Slip is not a soft failure. Once the faces move, three things follow in order.

• The hole clearance closes and the bolt shank comes into bearing. A standard hole for a half-inch bolt carries about a sixteenth of an inch of clearance; a slotted hole carries much more, which is why the belt-tensioning slots that made assembly easy also let this mount travel far. That travel is where the tension went.
• The faying surfaces fret. The bright marks the tech saw are not cosmetic. Fretting removes material, which removes grip length, which removes preload, which lowers the slip resistance further. The joint gets worse at resisting the thing that is happening to it.
• The transverse movement starts unwinding the fasteners. Side-to-side motion at the faying surface is the mechanism that rotates a nut loose, which is a different failure with a different fix and belongs to a sibling article. It arrives as a consequence of the slip, not as an independent problem.

Read the direction of that chain in both directions before trusting it. Raise clamp force and slip resistance rises linearly with it, fretting stops because there is no relative motion, and the loosening mechanism has no motion to feed on. Drop clamp force and every one of those runs the other way, which is exactly the escalation this mount showed across three visits.

The fix, and the one thing that was not a number

The mount went back with the paint cleaned off the two bearing faces under the heads, matched fasteners from one box, and a torque value applied with a wrench against the stated dry condition, in a crossing pattern in rising passes. The slots stayed, because belt tensioning needs them.

The part that was not a number: the belt-tension procedure got a line saying the mount bolts are torqued to value after the belt is tensioned and again after the first run-in period, in that order. Every previous visit had tensioned the belt by levering against a mount that was already tightened, which loads the joint sideways at exactly the moment nobody is checking clamp. The procedure had been building the failure it was called back for.

Checking you got this right

• Say out loud which surfaces are your faying surfaces and how many there are, before any arithmetic. A double-shear joint has two and a single lap has one, and the factor of two is the easiest thing in this calculation to get wrong.
• Confirm the slip coefficient you are assuming matches the surface actually there. A clean-steel number applied to a painted or galvanized face is optimistic in the direction that makes a joint look safe.
• Check whether the joint has already slipped before you calculate anything. Witness marks, fretting rouge, an elongated hole, or a bolt shank with a bright band on one side all mean the joint went through slip, and your job is preload, not a coefficient.
• Ask what loads the joint sideways during service work as well as during operation. Levering, belt tensioning and prying are transverse loads applied at the worst possible moment.
• Re-check preload after the first full run-in or heat cycle. A joint whose slip margin was 1.2 to 1 on the day does not have the margin to lose anything.

References

• 29 CFR 1910.147, control of hazardous energy, for isolating driven equipment before working on its mounts
• 29 CFR 1910.333(b)(2), for de-energizing and locking out circuits, which 1910.147 excludes at (a)(1)(ii)(C)
• 29 CFR 1910.1053 respirable crystalline silica and 29 CFR 1910.1025 lead, general industry, where a faying surface is abraded; respirator selection and fit testing under 29 CFR 1910.134
• RCSC Specification for Structural Joints Using High-Strength Bolts, for faying-surface classes and their slip coefficients, binding where the AISC specification and the adopted building code or a contract bring it in
• See related: What a Bolted Joint Is Actually Holding; Why a Bolted Joint Loosens; What a Washer Is For and When It Is Not