Steel bridge members rarely fail suddenly from fatigue. A crack initiates at a detail where stress concentrates, extends a small amount with each heavy vehicle, and continues for years while remaining far too small to threaten the structure. The failure occurs when it finally reaches a size the remaining section cannot tolerate. That long history is written into the fracture surface, and reading it back is what allows an investigation to say when the crack was large enough to have been found — which is usually the question that matters more than the fact that it was there.
Fatigue is driven by stress range, not peak stress
A member can carry loads far below its yield strength indefinitely in a static sense and still accumulate fatigue damage, because what drives crack growth is the magnitude of the stress cycle rather than its absolute value. A bridge member sees a cycle every time a heavy vehicle crosses. Over decades that becomes millions of cycles, and details that appear entirely adequate under a static check can be governed by fatigue instead.
The detail category is the design's own prediction
Bridge design specifications classify welded and bolted details into categories by their fatigue resistance, with the poorer categories covering geometries that concentrate stress sharply — cover plate terminations, transverse stiffeners, and welded attachments across a tension flange. The category assigned to a detail, together with the stress range and cycle count, is the design's own statement of how long that detail should last. Comparing that prediction against when the crack actually appeared is a direct test of whether the detail was correctly categorised.
What the fracture surface records
A fatigue fracture has a characteristic appearance: a smooth, progressively marked region where the crack grew slowly, and a rougher final region where the remaining section failed rapidly. Beach marks within the slow-growth region record changes in loading or environment, and they radiate from the origin. Locating that origin identifies the initiating feature — a weld toe, a tack weld, a drilled hole, a corrosion pit — and measuring the slow-growth region establishes how large the crack was when the member finally gave way.
Sizing the crack at earlier points in time
Fracture mechanics relates crack growth per cycle to the stress range and the crack size, so the growth history can be integrated backwards. Combined with traffic data — weigh-in-motion records, traffic counts, permit histories — this supports an estimate of how large the crack was at earlier dates. The estimate carries real uncertainty and depends on assumptions about the load spectrum, and those assumptions should be stated rather than absorbed into a single number.
Detectability is the other half of the question
A crack of a given size at a given date is only meaningful alongside what inspection would have been capable of finding. Visual inspection finds cracks that have opened enough to see, often only once they are substantial or have produced rust staining. Magnetic particle and dye penetrant testing find surface-breaking cracks far smaller. Ultrasonic testing finds embedded flaws that no surface method reaches. What was actually performed at each inspection, and to what procedure, determines whether a crack of the estimated size should have been reported.
Redundancy determines the consequence
Whether a cracked member threatens the structure depends on whether load can redistribute. A member whose failure would bring down a span carries a different significance, and a different inspection regime, from one where adjacent members can pick up the load. Establishing how the structure was classified in this respect — and whether that classification matched the as-built condition — frequently explains why a given detail was or was not receiving the scrutiny it needed.
Corrosion and fatigue compound each other
Section loss raises the stress range in what remains, accelerating fatigue, while corrosion pits act as initiation sites in their own right. Locations where drainage failed, joints leaked, or debris held moisture against steel are therefore where both mechanisms concentrate. A crack originating at a corrosion pit under a leaking deck joint implicates maintenance as much as design, and separating those contributions requires the maintenance record alongside the metallurgy.
The inspection file is primary evidence
Biennial inspection reports, condition ratings and their trend over time, prior findings at or near the location, any fracture-critical inspection procedures, and the resolution of previously noted defects all bear directly on the question of notice. A defect noted years earlier and not resolved carries a very different weight from one that appears in no report. These records are formal and retained, which makes them among the more reliable sources in this kind of matter.
What to preserve
The fractured member with both fracture faces protected from mechanical damage and corrosion, and without cutting through the fracture surface during removal. Adjacent members and connections, since the same detail exists elsewhere and undamaged examples establish the as-built condition. Traffic and permit records for the crossing, and the complete inspection and maintenance file before routine retention limits are reached.
This article is general technical orientation, not a failure analysis, an engineering opinion, or advice on any specific matter. Determining the cause of a particular incident requires hands-on examination by a credentialed expert.