Most accounts of the Akashi Kaikyo Bridge lead with the record: a 1,991-metre main span, the longest suspension span in the world for nearly a quarter of a century. The more instructive number is the one that was never designed — the extra metre.
The event
On 17 January 1995 the Great Hanshin earthquake struck with an epicentre roughly 20 kilometres west of Kobe — close enough to the Akashi Strait that the bridge site experienced the event directly. The crossing was under construction. Both main towers were complete, and work on the main cable system was already well advanced. The suspended deck was not yet erected.
Ground movement displaced the foundations. When the site was resurveyed, the distance between the towers had increased. The main span, designed at 1,990 metres, now measured approximately one metre more. The bridge that opened in 1998 has a main span of 1,991 metres because the ground decided so, not because an engineer specified it.
A one-metre change across a two-kilometre span is a rounding error in percentage terms and a serious problem in fabrication terms.
Cable geometry — planned versus revised
Tap to enlarge
Diagram: Original — EE&HL Network 2026. Schematic only; tower displacement and cable geometry not to scale.
Why the timing mattered
Published seismic assessments of the bridge note that it escaped serious structural damage, and attribute this in part to the stage of construction. The suspended deck — the mass that would later attract the bulk of inertial load in a seismic event — had not yet been erected. What stood in the strait that morning was substantially lighter than the completed structure would be.
That is a favourable accident of sequence, not a design achievement. The same earthquake arriving two years later would have found a very different structure standing.
Why a survey change propagates
The interesting part is not that the span grew. It is what a changed tower position does to everything derived from it.
Cable geometry on a suspension bridge of this scale is not a field-adjustable parameter. The catenary profile, the sag at midspan, the angle at which each strand leaves the saddle at the tower top, and the total wire count required to reach the designed cable cross-section are all calculated from a fixed set of tower positions. Change those positions and the derivation has to be repeated from the actual survey rather than the planned one.
That change does not stop at the cables. Hanger lengths connecting the main cables to the stiffening girder are derived from the cable profile, which is itself derived from the tower positions. A revised tower survey is not a local correction — it runs downstream through the design.
The scale makes this consequential rather than administrative. Each main cable would reach 1,120 millimetres in diameter, spun across a strait carrying roughly 1,400 vessel movements a day, with tidal currents and persistent wind, on a programme that could not simply pause maritime traffic. The spinning wheel follows a defined path; the positions at which wires are clamped into bundles and the sag profile they form are set by geometry locked in before spinning begins. A one-metre change in tower separation is not a rounding error in that context. It is a change of input.
Published accounts record that the cable configuration was revised to reflect the post-earthquake tower positions, and that some stiffening truss members were adjusted in length to accommodate the change. The programme continued. The bridge opened in April 1998, within its original completion window.
The commercial reading
Two things are worth separating here.
No project team schedules around a magnitude 7.3 earthquake. What the team did control was how much fabricated geometry was already committed at the moment the ground moved. Deck steel not yet cut to final dimensions is a variable that can still be adjusted. Deck steel already fabricated and delivered is a sunk cost that has to be reworked or replaced.
Absorbing a dimensional change through member lengths and assembly positions keeps the problem inside fabrication tolerance management. Escalating it into a design revision moves it into approvals, re-analysis and programme. The first is a production problem. The second is a contractual one.
The bridge that stands in the Akashi Strait was not built to the approved design geometry. It was built to a cable geometry derived from a survey taken after construction was already well advanced. Those are two different documents. The earthquake became an input to the final design rather than a disruption to it — and that outcome required survey precision, the engineering authority to revise an approved design, and a programme able to hold the spinning operation while the revision was worked through. None of those conditions is automatic.
Frequently Asked Questions
The Great Hanshin earthquake of 17 January 1995 struck with its epicentre close to the construction site. The bridge was still being built. Ground movement displaced the foundations, and the main span had to be increased by approximately one metre, from a design length of 1,990 metres to the 1,991 metres built.
The cable configuration was revised to reflect the post-earthquake tower positions, and published accounts record that some stiffening truss members were adjusted in length to accommodate the change. Because cable geometry is derived from tower positions, and hanger lengths are in turn derived from the cable profile, a revised survey propagates downstream through the design rather than being a local correction.
Published seismic assessments attribute this in part to the stage of construction. The suspended deck had not yet been erected, so the mass that would later attract inertial load was not present. The structure standing at the time was substantially lighter than the completed bridge.
It shows that the commercial consequence of an external event depends on where a project sits in its erection sequence when the event occurs. The same displacement at a later stage, with deck steel already fabricated to a fixed geometry, would have been a materially more expensive problem.
Sources: Honshu-Shikoku Bridge Expressway Co. seismic design and retrofit studies (Okuda, Fukunaga & Endo, 2009); proceedings of the World Conference on Earthquake Engineering on the seismic behaviour of the Akashi Kaikyo Bridge during the Hyogo-ken Nanbu earthquake; Structurae; Wikipedia. Figures are reported as published. Published tower heights differ between sources depending on whether the figure is measured above sea level or above foundation level.