Why Reusing a Fastener Is Sometimes Fine and Sometimes Not

Why this matters

A reused fastener rarely fails while you are standing there. It backs off over a few weeks and the joint weeps, or it lets go on the first hard thermal cycle after you leave, and the callback lands on a repair that was otherwise correct. Shops get this wrong because they inspect the fastener, when the real question is about the installation the fastener already went through. Threads that look perfect tell you nothing about whether the bolt is still elastic, whether its locking feature still locks, or whether the friction it was specified against still exists.

Before you break the joint loose

A fastened joint holds energy, and the act of loosening it is what releases the energy. Before a wrench touches anything:

• If the joint holds pressure, isolate it, relieve it through a vent or drain, and confirm zero at a gauge on the side you are opening, not upstream of the isolation valve. Isolate and lock under 29 CFR 1910.147 for general industry mechanical and stored energy.
• If the joint holds weight (a hanger, a bracket, a motor mount, a flange on a riser), support the load independently before the last two fasteners come out, and stand out of the fall path rather than under it.
• If the joint holds a spring (an isolator, a belt tensioner, a damper return), the spring extends when the fastener releases. Compress and restrain it first, and keep your face and hands off the axis it travels along.
• If the fastener is an electrical termination, de-energize, lock out under 29 CFR 1910.333(b)(2), and prove dead with the live-dead-live sequence in NFPA 70E-2021, 120.5, using the same meter on a known source before and after.
• If a seized fastener needs heat, set a fire watch and clear combustibles first (29 CFR 1910.252 in general industry, 29 CFR 1926.352 in construction), and note that heating a plated fastener puts metal in the air: zinc plating releases zinc oxide fume, and a cadmium-plated fastener releases cadmium oxide, which is acutely dangerous at low concentrations. Ventilate to outdoors or use local exhaust, and where cadmium plating is possible use respiratory protection under a written program per 29 CFR 1910.134, with cadmium exposure itself governed by 29 CFR 1910.1027 in general industry and 29 CFR 1926.1127 in construction. A glove does nothing about a fume.

The first installation may have spent something you cannot see

Four properties get consumed by tightening, and each one independently disqualifies reuse.

Elastic stretch. A bolt is a spring. Tightened inside its elastic range it returns to its original length when you loosen it, and it can do that many times. Tightened past yield it stays longer than it started, and the stretch you get on the second installation starts from a shorter remaining elastic range. Torque-to-yield fasteners are deliberately taken past that point because doing so gives tighter preload control than torque alone; the cost is that they are single use by design.

A prevailing-torque locking feature. A nylon insert deforms around the male thread the first time and is cut and heat-cycled thereafter. A distorted-thread lock nut yields its own thread geometry against the bolt. Both work by a resistance that exists before the nut seats, and both lose a measurable share of that resistance per cycle. Manufacturer data for these fasteners generally states a minimum prevailing torque after a stated number of on-off cycles, which is the number that owns the reuse decision.

A friction-controlling or sacrificial coating. A dry-film lubricant, a wax, a zinc flake finish or a plain zinc plate all set the friction coefficient the original torque figure was derived against. Removing a joint scrapes and burnishes that coating unevenly. The bolt is fine, the coating is not, and the torque spec no longer means what it meant.

An integral sealing element. A bonded washer, an O-ring under a head, a captive gasket on a banjo fitting: these seal by permanent deformation on first assembly.

The tightening instruction tells you which case you are in

Read the installation instruction, not the fastener:

What deliberately stays off the checklist

The items below get used as reuse criteria in the field and none of them decide anything.

Visual thread inspection. It catches gross damage, galling and corrosion, all of which are disqualifying, so keep doing it. It cannot see yield. A bolt stretched permanently looks identical to one that was not.

"It took the torque fine." A yielded bolt still reaches a torque reading, because torque is mostly a friction measurement. For an ordinary steel bolt in a plain unlubricated joint, the commonly cited breakdown puts roughly half of applied torque into under-head friction and about 40 percent into thread friction, leaving on the order of 10 percent doing the stretching. Reaching a number tells you the friction was normal. A sibling article owns that derivation in full.

A hardness test. Hardness tracks the material's strength grade, which is not what changed. The bolt did not get softer, it got longer.

Chasing the threads with a die. This removes material and changes the surface, so it changes both the effective thread engagement and the friction the torque spec assumed. Use it to clear debris from a damaged thread you have already decided to discard, not to rehabilitate one you intend to install.

Cleaning and oiling a dry-specified fastener. This is the reuse mistake with the largest consequence, because it looks like care. If the original torque figure was derived for a particular coating and you install the bolt clean and oiled, the friction share falls and a larger share of the same torque goes into stretch. The preload moves roughly inversely with the joint's nut factor K, the lumped constant in T = K x d x F. K is a property of the bolt, the coating, the lubricant and the bearing surface together rather than of the bolt alone, and a published K without its lubrication and finish condition attached is not usable. Halving the friction on a half-inch coarse thread does not double the preload, it raises it by around 80 percent, and that is enough to take a correctly torqued fastener past yield at the specified number.

The measurement that does decide it

Free length. Measure a new fastener from the same lot as your reference, then measure each removed fastener over the same two features with the same instrument. Any permanent growth beyond the repeatability of your tool means that fastener went past yield, whatever the instruction said.

This only works if you measure the reference. Comparing a removed bolt to a catalog nominal length measures the manufacturing tolerance, not the stretch.

Worked example: eight studs off a rebuilt assembly

A shop pulls eight studs from a bolted assembly during a rebuild. The service instruction reads "torque to the stated value" with no angle step, so on paper these are elastic, reusable fasteners. The crew wants to reuse them because the replacements are on order.

Set up the measurement. A new stud from the same box is the reference and measures 3.150 in over the full shank. The digital caliper repeats to 0.001 in on a gauge block, so anything inside 0.002 in is measurement noise and anything beyond it is real.

Six of the eight repeat within 0.002 in of the reference, which is 0.06 percent of the reference length and inside the noise floor. Two do not: one reads 3.157 in and one reads 3.161 in. Against the 3.150 in reference that is 0.007 in and 0.011 in of permanent growth, or 0.22 percent and 0.35 percent.

So two of the eight, a quarter of the set, went past yield on an installation that was never supposed to take them there. Six did not.

The decision on those two is easy: they are discarded. The useful part is the question the split raises. Eight fasteners on one assembly saw the same instruction, so a result that lands on two of them and not the other six is not a bolt problem, it is an installation or a service-condition problem. Two candidates fit: those two positions run hotter or carry more of the joint's external load, or whoever last assembled it pulled a click wrench past the click on the last two and did not feel the difference.

Walking the assembly settles it. Both long studs come from the same end, the end that carries the mounted accessory, which means those positions see external load added on top of preload. That changes the corrective action from "buy better studs" to "the accessory bracket needs its own support so its load stops adding into two studs," and it changes what the shop watches on the next service.

What it costs to get this wrong: the six good studs go back in and the two stretched ones go back in beside them, all eight get the same torque. The two stretched ones reach roughly the same tension at that torque, because torque is a friction measurement and does not know what the bolt has left. What they do not have is margin: they yield further during tightening, elongate again, and shed that tension over the first cycles. Those two are the joint's soft spots. The assembly holds through the shop test and loosens in service, and the return trip costs more hours than the whole rebuild's fastener count would have.

Where the answer flips

• Access is expensive. If getting back to the fastener means pulling the equipment again, replace it regardless of what the measurement says. The decision is not about the fastener, it is about the cost ratio between a fastener and a second teardown.
• The fastener is a lifting, pressure-boundary or structural element. The manufacturer's service data or the engineer of record owns the answer, not a field measurement. Reuse only where their document says reuse.
• The fastener came out of a matched set installed and tightened as a set. Replace the set, not the one.
• Small screws into sheet metal or plastic. The screw is almost never the wear item; the hole is. See the sibling article on what a tapped hole in soft metal can hold.
• You cannot identify the grade. A fastener with no head marking, out of a mixed bin, is not a reuse question. It is an unknown, and an unknown does not go into a joint you are responsible for.

Checking you got this right

Three checks, on the joint you just closed:

1. Mark and re-check. Put a paint or marker witness line across the fastener, the washer and the joint face after final tightening. It costs seconds and turns the next visit's inspection into a glance. Rotation relative to the joint means it moved.
2. Re-check after the first thermal cycle, not on a calendar. Bring the equipment to operating temperature, let it cool fully, then check. A joint that lost preload to embedment or to differential expansion shows it here and nowhere earlier. Do this with the equipment stopped and isolated under 29 CFR 1910.147, not running, and never re-torque a hot joint by hand: the parts are at service temperature and the assembly is still holding its full load.
3. Record what you installed. New fasteners or reused, and for reused ones the measured free length. The next tech's reuse decision is only as good as your record of how many cycles these have seen.

References

• 29 CFR 1910.147, control of hazardous energy for general-industry mechanical and stored energy; 29 CFR 1910.333(b)(2) where the fastener is an electrical termination
• NFPA 70E-2021, 120.5, in the edition your authority having jurisdiction or your employer's electrical safety program has adopted, for the live-dead-live proving sequence
• 29 CFR 1910.252 (general industry) and 29 CFR 1926.352 (construction) for fire prevention during heat application; 29 CFR 1910.134 for respiratory protection programs; 29 CFR 1910.1027 and 29 CFR 1926.1127 for cadmium
• Manufacturer service data for torque method, prevailing-torque cycle limits and discard criteria, which owns any specific number
• See related: What Torque Actually Controls and What It Does Not; What a Tapped Hole in Soft Metal Can Actually Hold