Home›Technical Guide›Fastener fatigue life

Failure & Quality

Fastener Fatigue Life: S-N Curves and Preload

📅 2026-10-13📖 ~13 min read📚 Based on VDI 2230, ISO 1099-1 and Palmgren-Miner

#Fatigue#Preload#VDI 2230#Wind energy

⚡ Key takeaways

Ninety percent of bolt failures are fatigue, not insufficient static strength. The fracture face is flat with beach marks: the crack grew slowly under cyclic load.
Higher preload gives longer fatigue life. It sounds backwards, but preload lets the clamped members share the external load, which reduces the stress amplitude in the bolt.
Fatigue life is set by stress amplitude, not mean stress. The S-N curve is the core engineering tool.
VDI 2230 gives the fatigue safety factor: S_F ≥ 1.2 for dynamic load.
Stress concentrates at the thread root and the under-head fillet. Those are where cracks start.
Thread rolling and shot peening raise fatigue life two to three times. Premium bolts always get them.
Do not solve fatigue by going up a bolt size. A stiffer bolt takes a larger share of the load and can fail sooner.

📑 Contents
1. A wind turbine bolt failure
2. Fatigue versus static strength
3. The S-N curve
4. Why preload extends life
5. Where stress concentrates
6. VDI 2230 fatigue check
7. Design measures that work
8. Decision tree and FAQ

1. A wind turbine bolt failure

In 2018 a wind farm in northern China suffered batch failures of blade root bolts after three years in service. The 2 MW turbine used 48 M36×4 class 10.9 bolts per blade root. The fracture faces were flat with clear beach marks: classic fatigue.

The investigation found:

Installation torque was insufficient, leaving preload at about 60 % of design.
Blade oscillation loaded the bolts directly, giving a large stress amplitude.
Three years produced roughly 10⁸ cycles, reaching the fatigue limit on the S-N curve.

The remedy was to recalculate preload to VDI 2230, retighten every bolt to 90 % of yield and fit rolled-thread bolts. Five years later there had been no further failures.

Fatigue is not sudden: it is insufficient preload plus cyclic load accumulating over years. This guide follows VDI 2230, ISO 1099-1 and Palmgren-Miner damage accumulation.

2. Fatigue versus static strength

Dimension	Static failure	Fatigue failure
Load	One excessive load	Repeated small load, 10⁴–10⁸ cycles
Fracture face	Plastic deformation, necking	Flat, beach marks
Location	The most highly stressed section	Stress concentration, thread root
Warning	Visible deformation	None; it breaks without warning
Design method	Yield check	S-N curve plus damage accumulation

3. The S-N curve

The S-N (Wöhler) curve plots stress amplitude σ_a against cycles to failure:

Higher amplitude means shorter life.
Steel shows a fatigue limit: below it, life is effectively infinite beyond about 10⁷ cycles.
Non-ferrous metals often show no clear limit.

Class	Fatigue limit amplitude (MPa)	Note
4.6	About 40–50	General
8.8	About 50–60	Common
10.9	About 60–80	High strength
12.9	About 80–100	Very high strength
Rolled thread	+30–50 %	Surface strengthening

4. Why preload extends life

This is the most counter-intuitive point in joint design. With an external load F:

Bolt stiffness C_b, member stiffness C_m.
The extra load carried by the bolt is F × C_b / (C_b + C_m).
The higher the preload F₀, the longer the members stay clamped and the more of F they absorb.

With adequate preload the bolt stress amplitude is F × C_b/(C_b+C_m), far below F. With insufficient preload the joint opens, the bolt takes all of F, and life collapses.

💡 VDI 2230 practice: preload to 70–90 % of the bolt yield strength. Insufficient preload is the number one cause of fatigue failure.

5. Where stress concentrates

Ninety percent of bolt fatigue cracks start in three places:

Thread root at the first engaged thread, stress concentration factor 3–4.
Under-head fillet, factor 2–3.
Abrupt section changes, such as an undercut.

Countermeasures: a larger under-head radius, rolled threads and no undercut.

6. VDI 2230 fatigue check

VDI 2230 defines the fatigue safety factor:

S_F = σ_a,allow / σ_a,max ≥ 1.2 (dynamic load)

Here σ_a,allow comes from the S-N curve and σ_a,max is the working stress amplitude.

7. Design measures that work

Measure	Life improvement	Cost
Raise preload to 90 % of yield	2–3×	None
Rolled rather than cut thread	1.5–2×	Low
Larger under-head radius	1.3–1.5×	Low
Shot peening or thread burnishing	2–3×	Medium
Increasing bolt diameter	May be worse	High
Reducing bolt stiffness (hollow bolt)	1.5×	High

⚠️ Counter-intuitive: a larger diameter raises C_b, so the fraction C_b/(C_b+C_m) rises and the bolt sees more of the external load. In fatigue-limited joints a smaller bolt with high preload often outlasts a larger bolt with low preload.

8. Decision tree and FAQ

Step 1 — Static or dynamic load?

Static → a yield check is enough. Dynamic (vibration, oscillation, cycling) → fatigue assessment.

Step 2 — Is preload adequate?

Below 70 % of yield → raise it first. It is the cheapest life extension available.

Step 3 — Where is the concentration?

Thread root → rolled thread. Under the head → larger radius.

Step 4 — Is S_F ≥ 1.2?

If not → raise preload, roll the thread, shot peen. Do not simply increase the diameter.

FAQ

Q. Is a broken bolt a quality problem? Not necessarily. Ninety percent of cases trace to design or installation: low preload or stress concentration.

Q. Is 10.9 better in fatigue than 8.8? Not necessarily. Higher strength means greater notch sensitivity, so the fatigue limit improves only modestly.

Q. How long can a bolt last? Steel has a fatigue limit around 10⁷ cycles, so infinite-life design is possible. Non-ferrous metals need Palmgren-Miner damage accumulation.

Q. Why does my bolt always break at the first thread? The first engaged thread carries 30–40 % of the load and is the most highly stressed point. That is normal, not a defect.

🛠️ Engineering support: Toprecision supplies class 8.8, 10.9 and 12.9 high-strength bolts with rolled threads and optimised under-head fillets, and can support a VDI 2230 fatigue check.
👉 Engineering services ｜ 👉 Request a technical quote