The broken bolt had a clean fracture surface. No thread stripping, no yielding at the shank. It had snapped in the first thread of the nut engagement, which is where fatigue fractures always happen. The bolt was M16 Grade 8.8, torqued to 210 Nm, holding a gearbox foot on a vibrating screen. It broke after eleven months. The customer’s first reaction was that the bolt was too small. That wasn’t it. A fatigue fracture says nothing about static strength. It says the bolt was carrying a cyclic load it couldn’t survive, and the preload wasn’t doing its job.
What a properly loaded bolt actually feels
A bolted joint works because the bolt is stretched. The preload clamps the joint faces together. The external cyclic load — the screen’s vibration — is carried by the friction between the joint faces, not by the bolt. The bolt only feels a small fraction of the external load, and the amplitude of the alternating stress stays tiny. That’s the whole design point. If the joint stays clamped, a bolt can survive millions of cycles with a fatigue strength around 50-60 MPa at 10 million cycles, which is low but acceptable for M16.
When the preload is too low, or the joint faces settle and the clamp force relaxes, the external load starts to pull the joint open. The bolt begins to feel the full cyclic load. The alternating stress jumps from maybe 20 MPa to 100 MPa. That’s above the fatigue limit for a rolled-thread bolt. The crack starts in the first thread root, grows slowly for months, and the bolt snaps without warning. The fracture surface shows the classic two zones: a smooth, rubbed area where the crack grew, and a rough final tear. The customer saw the rough zone and assumed overloading. It was fatigue.
The three things that killed this joint
The vibrating screen was bolted with spring washers. That was the first problem. Spring washers are for static joints that need to stay tight against small creep. On a vibrating screen they do the opposite of their job: they add flexibility, let the joint breathe, and the clamp load relaxes over time. The second problem was the tightening method. The bolts were torqued to 210 Nm with a torque wrench, but the joint surfaces were painted and had been re-used. The friction under the head varied, so the actual clamp load was scattered. Some bolts carried 80 kN of clamp, others only 40 kN. The low ones did all the flexing. The third problem was the joint faces. They were as-machined, but after eleven months of vibration the surface peaks had worn down. Clamp load relaxed another 10-15%.
The fix
The spring washers went in the bin. The bolts were replaced with new Grade 8.8 zinc-plated bolts, and the joint was tightened by the turn-of-nut method: snug the bolt to about 60 Nm, then turn the nut 90 degrees. That produces a consistent clamp load regardless of friction scatter. The joint faces were cleaned to bare metal with a wire wheel, and a thin layer of threadlocker was applied to the bolt thread. The threadlocker isn’t for locking — the preload locks the joint. It’s there to fill the thread clearance and stop micro-movement in the threads themselves, which is a separate fatigue source.
The result: no more breakage. The screen ran two years on the same bolts. The maintenance schedule went from “replace broken bolt every month” to “check torque on six bolts every quarter.” The torque check was the real maintenance item. If the preload relaxes again, the torque reading drops and the bolt gets re-tightened before a crack can start.
The fatigue numbers worth remembering
A rolled-thread bolt at 10 million cycles has a fatigue strength of roughly 50 MPa if the joint stays tight, and roughly 20 MPa if the joint is loose. That’s the whole story. A loose bolt carries five times the alternating stress of a tight one. The fatigue strength of the thread is set by the root radius of the thread, which is why rolled threads (the thread form is pressed, not cut) are about 30% stronger in fatigue than cut threads. A cold-rolled thread has compressive residual stress at the root, and a cut thread has machining marks that act as crack starters.
For a vibrating machine, the rules are short. Preload the joint to 70-80% of yield, use a method that controls clamp load (turn-of-nut or hydraulic tensioning), keep the joint faces clean and dry, and inspect torque on a schedule. None of this is exotic. It’s just the difference between a bolt that acts as a clamp and a bolt that acts as a pin.
The broken M16 wasn’t undersized. It was a joint that had stopped clamping. Fatigue fracture at the first thread means the cyclic load was reaching the bolt instead of being absorbed by friction. Clean faces, proper preload, no spring washers, and a torque check schedule fix it. A bolt that stays tight doesn’t fatigue.
