What a Bolted Joint Is Actually Holding

Why this matters

Ask a tech what a bolt is doing and you will usually hear that it holds the two parts together against the load. That is the wrong mental model, and it is wrong in a way that costs bolts. In a properly clamped joint the bolt is not the load path. The clamped members are. The bolt's job is to keep them pressed together hard enough that they can be the load path, and the bolt itself only picks up a small share of whatever the machine throws at the joint. Get that backwards and you will size, tighten, and diagnose bolts against a load they were never carrying.

Before you loosen a fastener on any joint that clamps a pressurised part, holds a spring or a lifted weight, or carries a conductor: isolate the energy source and relieve it to zero at the joint before the first fastener moves, verifying pressure on a gauge at the joint rather than at the source. For mechanical and stored energy that duty sits at 29 CFR 1910.147 in general industry; for a conductor or a panel it is 29 CFR 1910.333(b)(2), because 1910.147 excludes exposure to electrical hazards from work on conductors and equipment in electric utilization installations at (a)(1)(ii)(C). Where a support or hanger bolt is holding weight, set an independent support under the load before the last fastener comes out and stand out from under the piece, not beside it.

The three forces in play, named separately

• Preload is the tension you install in the bolt at assembly. Nothing external created it. It is stored the moment you stop tightening.
• Clamp force is the equal and opposite squeeze pushing the two members together. With no external load, clamp force equals preload.
• External load is what the machine applies later: pressure trying to push a flange apart, a fan pulling on its bracket, weight hanging off a hanger.

The interesting question is what happens to those three when the external load arrives, and the answer is not "the bolt takes it."

The gate: how the load splits

When an external tensile load pulls on a clamped joint, the bolt stretches a little further and the clamped members, already compressed, relax a little. Both move by the same amount, because they are still touching. So the load divides between adding tension to the bolt and removing compression from the joint, in proportion to how stiff each one is.

The share that reaches the bolt is the load factor. Call it C.

• Bolt tension = preload + (C x external load)
• Remaining clamp = preload - ((1 - C) x external load)

C is a geometry number, not a material constant. For a metal-to-metal joint with a machined bearing face, no gasket, and a grip length of roughly two to four bolt diameters, published load factors commonly sit around 0.1 to 0.2, because a short stack of solid steel is far stiffer than the thin bolt threading through it. Put a soft compressible gasket into that same joint and the members get much less stiff relative to the bolt, and C climbs toward 0.5 and beyond. Those are typical published ranges for those two geometries; where a joint is engineered, the drawing or the manufacturer owns the number.

That single gate produces two opposite outcomes, and both are worth carrying.

Case one: the hard flange

A machined steel flange, metal to metal, spotfaced under every head, grip about three bolt diameters. Take the load factor at 0.15. Each bolt is installed at 10,000 lbf of preload. In service, each bolt's share of the applied pull is 2,000 lbf, cycling from zero to 2,000 lbf as the machine loads and unloads.

• Bolt tension in service: 10,000 + (0.15 x 2,000) = 10,300 lbf. The bolt gained 300 lbf.
• Remaining clamp: 10,000 - (0.85 x 2,000) = 8,300 lbf. The joint lost 1,700 lbf of squeeze.
• Cross-check: 8,300 lbf of clamp plus 2,000 lbf of external equals the 10,300 lbf in the bolt. It closes.

Read that. Eighty-five percent of the applied load never entered the bolt at all. It was paid for out of clamp force. A tech who assumes the bolt carries the full 2,000 lbf would predict a bolt at 12,000 lbf, which is 17 percent high, and would draw the wrong conclusion about why it failed.

Now find where this joint runs out. Clamp reaches zero when 0.85 x external = 10,000, so at about 11,765 lbf of external load per bolt the members lift apart. Above that the load-sharing stops entirely and the bolt takes every additional pound. The failure is not gradual.

Case two: the same bolt through a soft gasket

Same bolt, same 10,000 lbf preload, same 2,000 lbf external load, but now a compressible gasket sits in the joint and drops the load factor to 0.5.

• Bolt tension: 10,000 + (0.5 x 2,000) = 11,000 lbf. The bolt gained 1,000 lbf.
• Remaining clamp: 10,000 - (0.5 x 2,000) = 9,000 lbf.
• Separation point: 0.5 x external = 10,000, so about 20,000 lbf per bolt.

Line the two up and the result is not the one most people expect. The soft joint holds its clamp better and separates at nearly twice the load. What it does instead is feed the bolt more than three times as much of the cycle: the bolt's alternating load is 300 lbf in the hard joint against 1,000 lbf in the soft one, a ratio of about 3.3 to 1 for identical external loading.

That is the fatigue trade, and it points one way at each end of the range. Push the load factor toward zero, with a stiff joint and a long thin bolt, and the bolt is almost blind to the cycling load but the joint gives up its clamp early and separates sooner. Push it toward one, with a soft joint and a short stubby bolt, and separation gets far away but the bolt lives the whole load cycle in its own tension range. A bolt that broke in a gasketed joint after a long run of service, with no sign of over-tightening, is very often a load-factor problem, not a torque problem.

What separation actually does

The moment clamp reaches zero, three things change at once and none of them are recoverable by tightening harder later.

• The bolt becomes the entire load path. Every additional pound is bolt tension.
• The alternating stress jumps by the reciprocal of the load factor. In case one that is a factor of about 6.7 the instant the joint opens, because the bolt goes from seeing 15 percent of the cycle to 100 percent of it.
• Any sideways component is now unresisted. With the faces apart there is no friction between them, so the joint can shift, and a joint that shifts under transverse motion is the classic route to a nut rotating loose. That mechanism belongs to a sibling article, which is where to go for the fix.

This is why "the bolt broke, use a stronger bolt" is so often the wrong repair. A stronger bolt of the same size in the same holes has almost the same stiffness, so C barely moves, and the joint separates at almost the same external load. What changed the fatigue picture was the separation, and a higher grade only helps if the extra preload it can carry raises the separation point.

What this changes about how you tighten

Preload is the only term in the separation equation you control at assembly. Neither the load factor nor the external load is yours to set in the field. So:

• A joint that separates is under-preloaded, not under-strength, unless the bolt is yielding, which is a different diagnosis with a different signature.
• Preload targets sit high on purpose. Structural pretensioned bolting works to a minimum pretension of roughly 70 percent of the bolt's specified minimum tensile strength, per the RCSC specification, which reaches a field-service shop only where the AISC specification and the building code adopted by the authority having jurisdiction bring it in, or where a contract does. That is not a general-purpose rule for equipment bolting, but the principle behind it is: you want preload well above the highest external load the joint will see, so the joint never opens.
• Losing preload after assembly does not need a loose bolt. Embedment, gasket creep and thermal relaxation all cut preload with the fastener never rotating. Anything that takes preload out is walking the joint toward its separation point.
• A gasketed joint deserves its re-torque. The soft member that gave you the friendly separation number is also the one that creeps, and it creeps most in the first heat cycle.

Checking you got this right

• Ask which member is stiffer before you ask anything about the bolt. If a soft gasket, a rubber isolator, a plastic spacer, or a stack of thin sheet is in the grip, assume the bolt is seeing a large share of the cycling load and treat it as a fatigue-sensitive joint.
• Compare the joint's likely peak external load per bolt against the installed preload. If they are within the same order of magnitude, the joint is living near separation and preload is the variable to fix.
• Look for the evidence of separation rather than assuming it: fretting rouge or polished witness marks between the mating faces, a gasket with a burnished ring on only part of its width, or paint cracked in a line along the parting face.
• On a joint that broke a bolt, check whether the fracture surface shows progressive beach marks over most of its area with a small final tear. That is a fatigue break, which means the joint was cycling the bolt, which means either it separated or the load factor is high. A single-event overload break has a different surface and points at assembly, not at the load path.
• After re-assembling a gasketed joint, plan the re-check for after the first full heat cycle rather than on the day. The cold test tells you the assembly holds; the post-cycle check tells you whether the creep took your margin.

References

• 29 CFR 1910.147, the control of hazardous energy, for relieving and restraining stored mechanical energy before a joint is opened in general industry
• 29 CFR 1910.333(b)(2), for de-energizing and locking out electric circuits, which 1910.147 excludes at (a)(1)(ii)(C)
• RCSC Specification for Structural Joints Using High-Strength Bolts, which reaches a project through the AISC specification as adopted by the governing building code or through the contract, for pretension targets in structural bolting
• Fastener manufacturer and equipment manufacturer documentation for the joint's specified preload, grip length, and any re-torque interval
• See related: Why Preload and Not Friction Holds a Joint Together; Why a Bolted Joint Loosens; What Torque Actually Controls and What It Does Not