Bolt Preload Loss in Wind Turbines: Causes and Troubleshooting

Preload loss is the leading cause of in-service bolt failures in wind turbines. A bolt that was correctly torqued at installation may have lost 20–40% of its clamping force within weeks through a combination of embedding relaxation, cyclic loading, and corrosion. Identifying which mechanism is at work determines the correct remedy.

§ 01  Why Preload Is the Critical Variable

A bolted joint in a wind turbine relies on clamping force — the compressive load that holds flanges together — not on the bolt's tensile or shear strength alone. Preload is the bolt tension deliberately induced during tightening, typically 70% of proof load for grade 10.9 structural bolts per EN 14399 and VDI 2230.

Sufficient preload provides three protections:

• Joint separation resistance: As long as the clamping force exceeds the separating load, the flange interfaces remain in contact. The bolt sees mostly static tension rather than fatigue cycling.
• Fretting and slip prevention: A fully preloaded joint transfers shear through friction. A loose joint allows micro-slip at the interface, which rapidly erodes mating surfaces (fretting) and introduces impact loading into the bolt shank.
• Fatigue life extension: The stress amplitude seen by a preloaded bolt under variable loads is a small fraction of the mean stress, keeping it in the long-life regime. A bolt with low preload sees the full load range as stress amplitude, dramatically reducing fatigue life.

For a tower flange bolt, the target preload might be 800 kN. A 30% loss to 560 kN may still be within the acceptable range — or it may not, depending on the wind load profile. The key point is that the design preload is a minimum, not a nominal.

§ 02  Causes of Preload Loss

Embedding relaxation is normal and expected: microscopic asperities on thread flanks, nut bearing faces, and washer surfaces plastically deform under load, compressing slightly and reducing bolt elongation. Standard practice (VDI 2230, EN 1090) accounts for this by specifying a re-tightening check 24–72 hours after initial assembly for critical joints.

Cyclic loading loosening (Junker effect) is the most serious mechanism. Under transverse dynamic loads, the thread engagement experiences cyclic slip, unwinding the nut incrementally. This is distinct from simple relaxation: the nut physically rotates. Indicators are fretting marks on the washer face, rust-coloured debris at the nut, and rotation marks on the bolt head. Prevention requires thread-locking strategies rather than higher torque alone.

Coating creep applies particularly to hot-dip galvanised (HDG) joints: the zinc layer under the nut bearing face is softer than the steel beneath and compresses gradually under sustained clamping load. EN 14399-4 and -8 specify HDG assemblies with an allowance for this; assemblies not designed for HDG under-estimate the preload loss.

§ 03  Diagnosing Preload Loss

The diagnostic approach depends on whether preload can be measured directly or only inferred:

1. Torque audit: Apply a calibrated torque wrench at the specified installation torque and observe whether the nut rotates. If it rotates, the bolt has lost clamping force below the torque-corresponding load. If it does not rotate, preload may still be acceptable (torque-angle audits are more reliable). Record which bolts failed and their position in the joint — clustering indicates cyclic loading rather than uniform relaxation.
2. Ultrasonic bolt measurement: Ultrasound measures actual bolt elongation and thus direct preload, independent of friction variability. The most accurate non-destructive method. Requires access to the bolt end face and a baseline measurement from installation or a known reference length. Used on critical flange bolts (monopile-to-transition-piece, blade root) where torque-method uncertainty is unacceptable.
3. Rotation-angle inspection: For torque-plus-angle tightened bolts (common in structural flanges), check alignment marks applied at installation. Any mark divergence indicates loosening. Rotation of more than 5° from the installation mark warrants immediate investigation.
4. Visual and tactile inspection: Fretting debris (red-brown oxide powder) around nut bearing faces is a strong indicator of joint slip. Movement marks, micro-cracks in paint or coating at the flange line, and water ingress into the joint cavity suggest a joint that has opened under load.

§ 04  Prevention by Design and Specification

Most preload loss problems are preventable at the specification stage:

• Specify HV assemblies (EN 14399) rather than standard hex bolts: EN 14399 assemblies are calibrated as a system (bolt + nut + washer from the same lot). The controlled K-factor (nut factor) reduces torque-to-preload scatter from ±30% to ±10%, meaning the actual achieved preload is closer to the design target.
• Use the correct tightening method for the application: Torque-only tightening is sensitive to lubrication variability. Torque-plus-angle or direct-tension indicators (DTI washers) provide better preload accuracy. Hydraulic tensioning (used on large monopile flange bolts ≥ M72) applies controlled tension directly, bypassing thread friction entirely.
• Specify appropriate thread locking for dynamic joints: For bolts subject to transverse vibration (nacelle base frame, cable clamp rail mounts), Nord-Lock washers, locking flange nuts, or prevailing-torque nuts prevent Junker-effect loosening. Do not use standard flat washers alone on vibrating joints.
• Account for coating thickness in flange design: HDG coating adds 50–85 µm per surface; on a tower flange with six mating surfaces (two flanges + two washers each side), this is 300–500 µm of additional compressed material subject to creep. Use HDG-specific bolt assemblies with extended grip length or account for creep in the design preload calculation.
• Define and enforce re-tightening schedules: IEC 61400-1 and DNVGL-ST-0262 require post-installation re-tightening checks. The timing (typically 100–200 operating hours or 2–4 weeks for tower flanges, 6–12 months for blade root bolts) should be in the site-specific maintenance plan, not left to judgment.

§ 05  Re-Tightening Practice

When re-tightening is required, the procedure must be controlled to avoid over-stressing the bolt:

• Do not simply re-apply the original installation torque: After embedding relaxation, applying the same torque to a relaxed bolt increases its tension. If the original target was 70% of proof load and the bolt has relaxed to 55%, re-applying installation torque could bring it to 80–85%, approaching proof load with reduced fatigue margin.
• Use a retightening torque reduced from the installation value: Typically 5–10% lower than installation torque for the first retightening. For hydraulically tensioned joints, reduce the target load by the expected embedding relaxation (quantified by the designer per VDI 2230).
• Retighten in a cross-pattern sequence: Retightening one bolt to full load before others creates uneven flange deflection that loosens adjacent bolts. Retighten in two passes: first to approximately 50% of target torque in cross sequence, then to full value.
• Replace any bolt showing signs of corrosion on the thread or bearing face: A corroded bolt's friction behaviour is unpredictable — the torque-preload relationship is invalid. Replacement is the only safe remedy.
• Document every retightening event: Record bolt identification, torque or angle applied, tool calibration reference, and any anomalies observed. This forms the maintenance traceability required by certification bodies and is essential for root-cause analysis if a bolt failure occurs.

For wind turbine bolts subject to offshore conditions, see also salt spray and cyclic load degradation in offshore environments.