Fastener Failure Analysis Process: From Field Sampling to Root-Cause Location
Executive Summary
Failure analysis is not "put the broken part under the microscope"—it is a complete, ordered chain with evidence preserved at every step. This article gives the standard fastener failure-analysis workflow: field information collection, fracture protection and sampling, macroscopic and microscopic inspection, material and process re-verification, simulation and root-cause determination, plus a ready-to-use analysis-planning checklist.
Table of Contents
Reaching for the SEM First Is the Most Common Opening Mistake
A batch of connecting bolts broke; an engineer sent the broken parts to the lab for SEM, and the conclusion was "fatigue fracture." After the report was submitted, the process department raised the tightening torque by one step, but the problem remained. The reason is clear: the analysis only reached "failure mechanism," not "root cause."
Fatigue is a mechanism, not a cause. The real cause could be insufficient joint-face stiffness leading to excessive Φ, preload consumed by embedding loss, or an underestimated transverse load at design. This information is not on the fracture—it is in field records and assembly data.
The correct starting point of failure analysis is not the microscope, but complete collection of field information.
Stage 1: Field Information Collection
This step determines whether the whole analysis can close the loop, yet it is often skipped. The information to collect falls into five categories:
| Category | Specific Content | Why It Matters |
|---|---|---|
| Failure phenomenon | Number of fractures, location distribution, time of occurrence, service duration, whether accompanied by abnormal noise or leakage | Count and distribution point to systemic issues vs. isolated defects |
| Assembly records | Tightening torque setpoint and actual value, tool calibration records, tightening sequence, whether graded loading was used | Distinguishes over-tightening from under-tightening |
| Part batch | Fastener batch number, supplier, coating process, incoming-inspection data | Locates batch-level deviation |
| Service conditions | Estimated load type and amplitude, temperature, media, vibration spectrum, maintenance records | Baseline input for fatigue and corrosion analysis |
| Intact parts from same batch | Unfailed bolts of the same batch, other bolts at the same location | Establishes a comparison baseline |
Stage 2: Fracture Protection and Sampling
Protection
- Never touch, clean, or pickle the fracture; pack it separately in a clean paper bag or plastic bag.
- Pack the two fracture halves separately and mark their correspondence to avoid secondary collision damage to matching surfaces.
- Record the original position of the broken part in the joint (which side, which bolt, which orientation).
- If corrosion-product analysis is needed, store at low temperature and dry to prevent further rusting.
Sampling plan
| Specimen | Purpose | Sampling Notes |
|---|---|---|
| Fracture itself | Macro morphology, SEM, EDS analysis | Keep original state; macro before micro |
| Longitudinal section near fracture | Metallography: microstructure, decarburization, crack path | Include fracture and thread profile; observe secondary cracks |
| Head-to-shank transition section | Check flow lines, laps, inclusions | Verify forming quality |
| Intact same-batch part | Performance retest, comparison baseline | Randomly sample no fewer than 3 pieces |
| Clamped-part thread section | Stripping analysis, engagement length | Preserve thread morphology and mating state |
Stage 3: Inspection and Analysis
Macro inspection (naked eye + low magnification)
- Fracture location: first loaded thread tooth / shank / head transition / thread run-out.
- Whether there is necking, obvious bending, and the angle of the fracture to the axis.
- Cross-section zoning and proportions (fatigue source, propagation zone, final-fracture zone).
- Wear, debris, discoloration and corrosion marks on threads and bearing surface.
Micro inspection (SEM)
Confirm the fracture mechanism: dimples (ductile) / fatigue striations (fatigue) / intergranular rock-candy (hydrogen embrittlement or stress corrosion) / cleavage (brittle). Meanwhile use EDS to analyze the composition of corrosion products and inclusions on the fracture.
Material and process re-verification
| Item | Method | Criterion |
|---|---|---|
| Chemical composition | Spectrometric analysis | Does it meet grade requirements |
| Hardness | HV / HRC, multiple points and layers at surface, core and thread root | Is it within the property-class range |
| Tensile strength and proof load | Tensile test per GB/T 3098 series | Does it meet class requirements |
| Metallographic structure | Longitudinal-section metallography | Decarburization, overheating, abnormal structure, non-metallic inclusions |
| Decarburized layer | Cross-section metallography at thread root | Is full and partial decarburization depth within limits |
| Coating | Magnetic thickness measurement, salt spray, adhesion | Do film thickness and adhesion meet spec |
| Hydrogen content | Hydrogen measurement when needed | Compare with unplated parts of the same size |
| Friction performance | Torque-clamp-force test per GB/T 16823.3 | Is K within the design assumed range |
Stage 4: Root-Cause Location and Reproduction Verification
Organize the information from the previous three steps into a causal chain, tracing back level by level from mechanism to cause:
- Mechanism layer: what failure does the fracture tell us (fatigue, hydrogen embrittlement, overload…).
- Condition layer: what conditions are needed for this mechanism (stress amplitude over limit, hydrogen ingress, load over design…).
- Cause layer: why these conditions hold (Φ calculated too low, insufficient baking, wrong torque setting, coating batch change…).
- System layer: why it happened (design review did not cover it, missing procurement technical requirements, process parameters not frozen, supplier change without notice…).
Usually the true root cause lies in the third or fourth layer, while reports often stop at the first or second. To reach the third layer, assembly data, batch data and friction-performance data must be aligned with the fracture conclusion.
Reproduction verification
A root-cause conclusion must be reproducible or at least data-supported:
- Hydrogen embrittlement: run the pre-load test per GB/T 3098.17, using specimens from the same batch and surface-treatment state to verify whether delayed fracture occurs.
- Fatigue: calculate stress amplitude and compare with material fatigue limit; if needed, reproduce on a vibration table.
- Friction anomaly: run the torque-clamp-force test on the same batch to reproduce K deviation.
- Stripping: re-measure engagement length and run a pull-off test to confirm the failure mode.
One criterion: if the proposed root cause cannot explain "why only this batch / this location failed," the root cause has not been found. Scale and distribution are the most powerful verification tools.
Failure-Analysis Planning Checklist
| Stage | Key Action | Deliverable | Common Omission |
|---|---|---|---|
| Field collection | Failure distribution, assembly records, batch number, service conditions, retain intact parts | Field information sheet | Not retaining intact parts and assembly records |
| Fracture protection | Separate packaging, no cleaning, mark position | Sampling and protection record | Touching the fracture; mixing parts and collisions |
| Macro inspection | Fracture location, necking, cross-section zoning, wear marks | Low-magnification photos and reading conclusion | Cutting specimens directly without macro recording |
| Micro inspection | SEM morphology and EDS | Micro photos and mechanism determination | Only producing photos without interpreting morphology |
| Material re-verification | Composition, hardness, mechanical properties, metallography, coating | Retest report | Only testing hardness; skipping metallography and decarburization |
| Root-cause location | Four-level traceback: mechanism—condition—cause—system | Root-cause conclusion and causal chain | Concluding at the mechanism layer |
| Reproduction verification | Pre-load test, stress-amplitude calculation, torque-clamp-force test | Verification report | Conclusion with no data support |
| Measures and closure | Countermeasures, owners, deadlines, effectiveness tracking | Corrective actions and closure evidence | Report only; no follow-up on measures |
Conclusion
The value of failure analysis is not producing a morphology description, but stringing the four layers of mechanism, condition, cause and system into a reproducible, verifiable and closable chain. The fracture gives the mechanism; data gives the conditions and causes; the gap at the institutional layer is the root cause.
The order cannot be reversed: collect field information first, then protect the broken parts, then inspect macroscopically, microscopically and materially level by level, and finally return to data for root-cause location and reproduction verification. Skip the first step and go straight to the microscope, and what you get is only a pretty report that cannot solve the problem.
