01Executive summary
Two celebrated bridges and one destroyed one. The collapse taught the profession more than either success, and it is still widely taught wrongly.
The Sydney Harbour Bridge opened in 1932, a 503-metre two-hinged steel arch. The Golden Gate Bridge opened in 1937 with a 1,280-metre suspended main span, then the longest in the world. The Tacoma Narrows Bridge opened in July 1940 and destroyed itself in November of the same year in a wind of roughly 68 km/h — far below its design wind speed. The mechanism was aeroelastic flutter, and the profession acquired a wind engineering discipline as a direct result.
02Sydney Harbour Bridge: erection as the design driver
A steel arch across a busy deep-water harbour cannot be built on falsework from below. The erection method therefore determines the structural form, not the other way round. Each half arch was built out as a cantilever from its abutment, restrained by cables anchored into rock tunnels behind, until the two halves met at mid-span. Only after closure was the deck hung from the arch.
This produces a structure that experiences two entirely different sets of actions. During erection each half is a cantilever with large hogging moments and enormous horizontal restraint loads at the anchorage. In service it is a two-hinged arch carrying predominantly compression with the deck suspended below. Members must be adequate for both, and for every intermediate stage.
The temporary condition often governs
Erection stresses, lifting cases, transport loads and partially completed states routinely exceed service conditions. A design that checks only the completed structure is incomplete, and this remains a common source of construction failure.
Closure requires geometry control
Two cantilevers must meet within tolerance in three dimensions, at a temperature that changes their length by measurable amounts. Closure was timed to thermal conditions and adjusted by controlled release of the restraining cables.
Credit for the design of the Sydney Harbour Bridge is genuinely disputed. John Bradfield was the chief engineer of the New South Wales Public Works Department and drove the project from conception through delivery, and the general arrangement and specification originated on his side. Ralph Freeman, of consulting engineers to the contractor Dorman Long, prepared the detailed structural design of the arch. Both parties asserted primacy, publicly and at length. This series does not attempt to resolve it: the accurate statement is that Bradfield led the project and set its form and requirements, and Freeman performed the detailed arch design, and the two functions were both necessary.
03Golden Gate: suspension at the limit
A suspension bridge is elegantly simple in principle. The main cables carry tension, the towers carry compression to the foundations, and the deck hangs from the cables. Because the cables work in pure tension the span is limited chiefly by the cable’s ability to carry its own weight, which is why suspension bridges dominate the very longest spans.
The main cables were spun in place from individual galvanised wires drawn back and forth across the span by travelling wheels, then compacted and wrapped. Spinning in place rather than lifting a prefabricated cable is the only practical route at this scale, and it means the cable is constructed to a geometry that changes as its own weight accumulates — requiring continuous survey adjustment.
Safety as an engineered system
The project ran a formal safety programme unusual for the period: hard hats were required, a safety net was strung beneath the deck, and workers were removed for unsafe practice. The net caught a number of men who would otherwise have died. It is an early and clear instance of designing the temporary works for human protection to the same seriousness as the permanent structure, and of the hierarchy of control — where the hazard cannot be eliminated, engineer a barrier rather than relying on individual care.
04Tacoma Narrows: what actually happened
The commonly repeated explanation is that wind produced periodic vortex shedding whose frequency matched a natural frequency of the bridge, causing resonance. This is not what happened, and the distinction is important because the two mechanisms require different countermeasures.
At Tacoma Narrows the deck was a shallow solid-web plate girder section — a bluff body with very low torsional stiffness and very low aerodynamic damping. As it twisted, the angle of attack of the wind on the deck changed, generating a torsional moment that reinforced the twist. The motion was self-excited. The bridge did not fail because the wind found its frequency; it failed because above a certain wind speed the system extracted energy from the flow faster than it could dissipate it.
If the problem were resonance, detuning would fix it — change the mass or stiffness to move the natural frequency away from the driving frequency. That would not have saved this bridge, because flutter has no driving frequency to avoid. The remedies for flutter are aerodynamic: change the section so it does not generate the destabilising moment. Modern long-span decks use shallow streamlined box sections, open grating in the deck to break the pressure differential, and wind-tunnel testing of section models and full aeroelastic models as a standard step. Getting the mechanism right determines which family of solutions is even relevant.
| Phenomenon | Mechanism | Typical countermeasure |
|---|---|---|
| Vortex shedding | Alternating vortices produce periodic cross-wind force at a speed-dependent frequency | Helical strakes, spoilers, added damping, mass or stiffness change |
| Galloping | Self-excited across-wind instability of a bluff section, often when iced | Section shape modification, damping devices, ice management |
| Torsional flutter | Twist alters angle of attack, generating a reinforcing moment | Streamlined section, deck venting, torsional stiffness increase |
| Buffeting | Random turbulence in the oncoming flow forces the structure broadband | Damping, fatigue design of details, serviceability limits on motion |
Approach to slenderness
The trajectory is not a retreat from slenderness. It is the replacement of brute depth with understanding: the modern box section is nearly as shallow as the failed girder but behaves completely differently, because its shape and torsional stiffness were designed against a known mechanism rather than assumed adequate.
05Takeaways for current practice
- Name the mechanism before selecting the remedy. Resonance and flutter look similar in a video and demand opposite responses.
- Check every construction stage, not only the finished structure. The temporary condition frequently governs member sizing and stability.
- Extrapolating a successful form is where long-span design fails. Each increase in slenderness moved further from validated experience without a corresponding increase in analysis.
- Engineer the barrier rather than relying on care. The Golden Gate safety net saved lives that instruction and vigilance would not have.
- Attribute honestly. Large projects have contested credit because they genuinely require several distinct contributions. Naming what each party did is more accurate than naming one designer.
Australian references relevant to this part include AS 5100 for bridge design, AS 1170.2 for wind actions, and AS 4100 for steel structures. Cited by number for orientation only — verify currency before use.