Fastener Failure Modes: Which Fatigue, Creep and SCC Controls the Evidence Supports

Start with the hardness grade, not the strength class

Strength and hardness drive susceptibility before any other variable. As steel gets harder it becomes less ductile and less tough, so the stress at which absorbed hydrogen causes delayed cracking falls ⁴. That is why the specified hardness band, not the tensile class on the head, is the first thing to fix in your specification. Two publishers put the significant threshold at about 39 HRC (380 HV). Below it, well-made fasteners normally tolerate hydrogen without delayed loss of strength, assuming controlled manufacture and adequate steel quality ⁴. Above it, susceptibility rises sharply ⁴. ISO 898-1 carries its own cautionary statement that class 12.9 fasteners are at risk of stress corrosion cracking ⁵. The two classes sit on opposite sides of that line. Class 12.9 is specified at 39-44 HRC (385-435 HV), which one supplier describes as the extreme-risk zone with virtually no tolerance for atomic hydrogen ¹⁰. Class 10.9 at 32-39 HRC is below the threshold and is cited as routinely safe, including when galvanized ^{4, 2}. Treat 39 HRC as a reported threshold, not an acceptance limit. The critical value shifts by as much as ±1.0 HRC with chemistry, tempering temperature and microstructure. Some standards set lower limits of 31-35 HRC to avoid internal hydrogen embrittlement ⁴. If your application sits near the line, ask the supplier for the measured hardness distribution of the lot, not the class nominal.

Pick a coating route that matches the grade's hydrogen risk

Coating choice is where a good grade decision gets undone. Electroplating introduces hydrogen during the process, so an electroplated finish on a susceptible grade adds risk the base metal did not have. Zinc-nickel is electroplated and must be baked after plating to reduce that risk ¹⁷. Non-electrolytic systems avoid the plating bath altogether. Zinc flake, sherardising and mechanical galvanising are the routes to call out for susceptible grades. A drawing that says only "zinc plated" defaults to electroplating ⁹. Zinc flake is applied by dip-spin or spray and cured rather than plated. One supplier positions it for hydrogen-risk reduction and high corrosion targets on high-strength fasteners ²². The trade-off is not free. Zinc-nickel gives a more metallic plated finish with good dimensional consistency on threads and good electrical conductivity, which matters where thread fit or conductivity is tight ²². If you specify it anyway, write the post-plating bake into the purchase order rather than assuming it. Neither route removes hydrogen risk outright, so the coating callout and the hardness callout have to be read together.

Set a preload window and a tightening method that holds it

Preload fails the joint at both ends. Too little and the joint loosens under vibration, the bolt carries the full cyclic load and fatigue follows. Too much and the bolt yields or breaks, and the sustained tensile stress can facilitate stress corrosion cracking ⁷. Over-tightening pushes the bolt closer to or above its hydrogen-embrittlement threshold stress ¹¹. The supported control is a tightening method that lands preload in the window and can be verified: calibrated torque, torque-angle, or stretch control ^{23, 11}. Stretch-controlled methods are recommended where reliability matters most ¹⁴. Whichever you name, the friction assumption behind the torque figure has to be stated. Lubricant on the threads changes the coefficient and calls for a torque reduction ¹⁵. Do not assume a stronger bolt buys fatigue life at the same torque. One source states that a stronger material at the same preload does not extend joint fatigue life and is more brittle ⁷. The sources here give no target preload or torque value, so the number has to come from your joint calculation and the supplier's torque-tension data.

Make thread root and head fillet an inspection item

Cracks start where stress concentrates. In screws and bolts that means the root of the first engaged thread or the fillet radius under the head ¹. Sharp radii, surface discontinuities, pits and thread laps at the root act as local concentrators. They can start a crack in a material that is already susceptible ¹. That turns geometry into a specification item rather than a drawing default. Generous fillet radii at the head-to-shank transition, large thread root radii and smooth surface finishes all reduce the peak stresses that drive hydrogen migration ⁹. Where the process allows it, thread rolling before heat treatment is cited as a way to minimise residual stress in the finished thread. Fatigue life is named as the exception ⁹. For your RFQ, the practical question is whether the supplier controls these features or merely draws them. Ask for the radii on the actual production drawing and for the thread process sequence relative to heat treatment. The sources give no acceptance criteria for laps or surface defects, so agree the inspection basis separately.

Hot or corrosive service needs data, not a higher grade

A high strength class is not a temperature rating. In high-temperature or pressure service the starting point is a code-specified bolting grade. The checks that decide the choice are allowable stress, strength retention, relaxation or creep, nut pairing and environment ⁸. Class 12.9 in particular is not a temperature rating ⁸. Creep and stress relaxation show up as preload loss over time. High-temperature and corrosive conditions cause preload to decay. The recommended response is high-temperature or coated bolts plus regular inspection ²³. That is a design-life question, not a grade question, and it is the one place where the evidence here is thinnest. If your joint runs hot, ask the supplier for relaxation or creep data and allowable stress at your service temperature over your design life. No source in this set publishes a creep rate, a relaxation curve or an allowable stress. The specification has to rest on data you obtain rather than on the class.

Choose the galvanizing standard and plan the post-bath test

Hot-dip galvanizing adds a thermal cycle, and that cycle is the risk. Where the bath temperature exceeds the fastener's tempering or stress-relief temperature, ASTM F2329 requires the purchaser to arrange mechanical testing after galvanizing ⁶. The responsibility sits on your side of the order, so build the test into the PO rather than assuming standard bath temperatures are harmless ⁶. Fasteners in the 25-39 HRC band are not normally embrittled by galvanizing, and class 10.9 and A354 BD bolts are routinely galvanized ². That is the band your hardness decision should already have put you in. Between the two standards, F2329 adds mandatory adhesion testing, higher thickness minimums for small fasteners and for A394 bolts, and fuller documentation. It does so at a reported cost premium over A153 ¹². A153 remains adequate for general non-critical hardware ¹². Name F2329 where the joint is critical or the grade is high strength. Confirm your galvanizer can run the required testing before you release the order ⁶.

Pre-RFQ checks before you freeze the specification

Each item below covers a figure the sources in this set do not supply. Getting them stated in writing before the specification is frozen is what keeps the failure-mode controls from resting on assumptions.

• Ask for the measured hardness distribution of the lot, not the class nominal — the 39 HRC threshold scatters by ±1.0 HRC ⁴.
• Confirm the coating route is non-electrolytic, or that a post-plating bake is written into the PO for electroplated finishes ^{9, 17}.
• Require the torque-tension data behind any torque figure, including the friction coefficient assumed for lubricated threads ¹⁵.
• Ask which tightening method will be used and how preload is verified — calibrated torque, torque-angle or stretch ²³.
• For safety-critical high-strength assemblies, plan a relaxation period of 24-48 hours before the final torque check ¹¹.
• Request radii and thread process sequence on the production drawing, not just the sample ⁹.
• For hot service, request relaxation or creep data and allowable stress at your service temperature and design life ⁸.

Where the sources disagree on galvanizing high-strength bolts

Galvanizing practice for high-strength bolts is not settled the same way in every region, and the split is visible in the standards themselves. The rows below keep both positions in view so you can decide which one your project's bolting standard follows.

Why the thread root decides fatigue life, and what rolling, radius, and preload do about it

Fatigue cracks start where stress concentrates, so the thread root, the first load-bearing thread, and the head-shank fillet are the sites that matter. The corpus quantifies the thread-root stress concentration, the rolled-versus-cut difference, the stiffness-driven share of cyclic load the bolt actually sees, and the environments that drive stress corrosion cracking.

The thread root carries the fatigue penalty. One engineering reference puts the elastic stress concentration factor at the root at Kt = 3.5 to 4.5, which is why steel-bolt endurance limits quoted at the thread root run only about 7 to 12 percent of tensile strength for rolled-thread alloy steel bolts (engicalchub.com, citing published references) — roughly 60 to 100 MPa for a class 8.8 bolt by the same source's example. Thread form moves this number: UN tolerance allows either sharp cut-thread corners or a rolled-style root radius, and the UNR designation specifies a root radius of 11 to 14 percent of pitch while the UNJ/MJ structural form, defined by MIL-S-8879C, enlarges it to 15 to 18 percent to cut concentration further (designjudges.com). One trade-off applies only where the bigger radius is used: UNJ bolts are incompatible with standard UN nuts.

Rolling versus cutting is the largest single lever. Thread rolling cold-works the root and leaves compressive residual stress, and one guide attributes roughly 2 to 3 times the fatigue life to rolled threads compared with cut threads (engicalchub.com); another reference states the same mechanism — compressive residual stress at the root — without a multiplier (mechcodex.com). Grade upgrades help less than buyers expect: because very high-strength steels grow more notch-sensitive, the net endurance-limit gain from 8.8 to 12.9 is only about 46 percent for rolled threads per one published table (mechcodex.com), and a fastener manufacturer warns that a stronger bolt at the same torque can fail faster, not slower (nord-lock.com). Practically, fatigue cracks initiate at the first loaded thread or under the head, and one source calls insufficient preload the leading cause of bolt fatigue failures (fasten.one).

Preload and stiffness decide how much of the cyclic load the bolt ever sees. As long as the joint stays clamped, most of the external load relieves the clamp instead of adding bolt tension — one worked example with a load factor C = 0.25 converts a ±5 kN load swing into only about ±8 MPa of alternating stress, versus roughly 40 to 60 MPa endurance for a rolled-thread 8.8 bolt; if preload halves and the joint nears separation, the alternating stress climbs toward ±32 MPa and the margin disappears (mechcodex.com). That is why one source targets 70 to 85 percent of proof load as nominal preload, reserving margin for torque scatter (mechcodex.com). Corrosion and heat close the loop: chloride SCC affects austenitic stainless under tensile stress, and keeping chloride below 10 ppm sharply cuts its probability (whatispiping.com); too much preload can itself facilitate SCC (nord-lock.com); and at temperature, stress relaxation — creep under sustained high stress — bleeds preload over time (boltscience.com).

The corpus gives no sourced percentage for how much of the total bolt load the first load-bearing thread carries (the qualitative claim that it carries most of the load appears in one segment but without a figure), and no creep-relaxation onset temperature thresholds for specific fastener grades; both are left unquantified rather than estimated.

FAQ

Related pages

• Surface finishes product page
• Hydrogen Embrittlement in High-Strength Bolts: Which Grades Need Controls, Which Coatings and Baking Schedules Are Acceptable, and What to Require Before the RFQ
• Fastener Materials for Continuous Service Above 400°C: What the Evidence Settles and What the Code Must Supply
• What Bolt Property Classes 8.8, 10.9 and 12.9 Encode — and When High Strength Stops Being the Right Choice
• Through-Hardened vs Case-Hardened Fasteners: Hardness Bands, Embrittlement Controls and Receipt Evidence