Why Track Bolts Loosen
A bolted joint loosens when transverse vibration causes the thread flanks to slide and the preload to decay: each cycle of sliding consumes a little of the elastic stretch that holds the joint together. Track bolts face this continuously, because every passing wheel bends the fishplate and shakes the fastening. The design response has three layers: choose a material and thread form that resists sliding, add locking elements that maintain preload, and protect the fastener so corrosion does not turn a tight bolt into a seized or slack one. Each track scenario needs a different combination of these layers.
Turnout Bolts: High-Strength Anti-Loosening Design
Turnout areas subject bolts to the harshest duty: switching impacts, high-frequency vibration and heavy lateral loads. The standard design starts with 42CrMo alloy steel, quenched and tempered to a tensile strength of at least 1080 MPa, yield strength of at least 930 MPa and hardness of 32-36 HRC, material properties verified to GB/T 3077-2015. The thread is a fine thread, with pitch reduced from 3 mm to 2 mm, which increases the self-locking angle effect and enlarges the thread contact area so vibration energy is dissipated over more surface. Locking is provided by a double-nut arrangement with a lock nut torqued to about 80% of the main-nut preload, plus a disc-spring washer whose elastic deflection compensates for preload decay as the joint beds in. The bolt surface is nitrided to a layer depth of at least 0.15 mm and hardness above 900 HV for wear and corrosion resistance, and the head is a hexagonal flange design that spreads the bearing load over the fishplate so the head cannot sink into the plate.
Pre-Embedded Bolts for Ballastless Track
Ballastless track bolts are cast into the concrete slab, so the design problem is anchoring stability rather than head access. Embedment depth is held at 200-220 mm, and a positioning jig keeps the bolt verticality deviation within 0.5° so the fastening does not sit on a tilted stud. Three annular ribs, about 5 mm high and 8 mm wide, are forged on the shank to grip the concrete and resist pull-out; the surface is derusted to Sa2.5 and coated with an epoxy anchoring adhesive at least 2 mm thick before casting, giving a design anchoring force of at least 120 kN. Preload is applied in three stages, 50% of the design value first, then 80% after 24 hours and 100% after 72 hours, which lets the concrete around the embedment settle without cracking. The exposed shank is protected by hot-dip galvanizing at least 120 µm thick with a sealant fill so moisture cannot reach the steel.
Ballasted Track: Corrosion and Loosening Solved Together
On ballasted track the bolt head is exposed to ballast abrasion, moisture and chemicals. The proven combination is a weather-resistant steel grade such as Q450NQR1, whose copper and chromium content cuts the corrosion rate to about 0.03 mm/year in humid service, roughly 70% lower than plain carbon steel, combined with an anaerobic thread-locking adhesive that cures in the thread gap and forms a rigid bond, and a self-locking thread form with a reduced flank angle of about 30° for higher self-locking resistance than a standard 60° thread. Corrosion protection is doubled: hot-dip galvanizing at least 100 µm thick plus a zinc-flake coating at least 8 µm thick, which together withstand more than 2000 hours in the salt-spray test (ISO 9227 / GB/T 10125). Installation torque for bolts on 60 kg/m rail joints is controlled at 400-450 N·m with a calibrated torque wrench, and a torque audit is repeated every three months to catch decay early. A polyethylene cap over the bolt head keeps rainwater and ballast fines out of the threads.
Dynamic Testing and Batch Acceptance
Anti-loosening performance is proven by vibration testing, not by marketing claims. Bolts are assembled on a fixture that reproduces the rail-and-fishplate joint at working preload and vibrated at 10-50 Hz with 0.5-1 mm amplitude for 100 hours; torque is read at 10-hour intervals and the attenuation rate after 100 hours must be 5% or less. An impact test applies 30 t axle-load impacts 100,000 times and checks that the thread shows no slipping or deformation and the locking devices remain engaged. Field monitoring on a live line over three months should keep torque fluctuation within ±10% of the set value. Batch acceptance follows GB/T 3077-2015 for material with strength deviation within ±5%, ISO 898-1 / GB/T 3098.1 for mechanical properties, and the salt-spray criterion of no more than 5% rusted surface area with coating adhesion at least 5 MPa. Twenty bolts per batch are sampled, five per test group, with double sampling on any failure before a batch is rejected.
Frequently Asked Questions
Why are fine threads used on turnout bolts?
A 2 mm pitch thread has a smaller lead angle and a larger contact area per turn than the standard 3 mm pitch, so it resists vibration-induced sliding better, at the cost of slightly slower assembly.
Can the three-stage preload cycle be skipped on ballastless track?
No. Staged preload lets the concrete around the embedment and the adhesive settle progressively; full preload at once can crack the slab and reduce the anchoring force below the 120 kN design value.
What is the difference between a lock nut and a disc-spring washer?
A lock nut adds friction at the top of the joint and must be torqued to a defined share of the main preload; a disc-spring washer stores elastic deflection that re-applies preload as the joint beds in. They work best together.
Is a zinc-flake coating as protective as hot-dip galvanizing?
They protect differently. Hot-dip galvanizing gives a thick sacrificial layer, while a zinc-flake coating is thin, uniform and compatible with threaded surfaces; using both, as specified for ballasted-track bolts, combines thickness with thread tolerance.
Why is torque audited every three months?
Preload decay is fastest in the first months of service as surfaces bed in and the adhesive cures. A three-month audit catches the decay curve early, when re-torquing restores full preload; later audits can be spaced out once the curve flattens.