A new bridge across the river Dee in Chester was needed for the Chester & Holyhead Railway, a project planned in the 1840s for the expanding railway system. It was built using cast iron girders, each of which was made of three very large castings dovetailed together. Each girder was strengthened by wrought iron bars along the length. It was finished in September 1846, and opened for local traffic after approval by the first Railway Inspector, General Charles Pasley. However, on 24 May 1847, a local train to Ruabon fell through the bridge. The accident resulted in five deaths (three passengers, the train guard, and the locomotive fireman) and nine serious injuries.

The bridge had been designed by Robert Stephenson, and he was accused of negligence by a local inquest. Although strong in compression, cast iron was known to be brittle in tension or bending, yet on the day of the accident the bridge deck was covered with track ballast to prevent the oak beams supporting the track from catching fire. Ironically, Stephenson took this precaution because of a recent fire on the Great Western Railway at Uxbridge, London, where Isambard Kingdom Brunel's bridge caught fire and collapsed.

One of the first major inquiries conducted by the newly formed Railway Inspectorate was of the Dee bridge disaster. The lead investigator was Captain Simmons of the Royal Engineers, and his report suggested that repeated flexing of the girder weakened it substantially. He examined the broken parts of the main girder, and confirmed that the girder had broken in two places, the first break occurring at the centre. He tested the remaining girders by driving a locomotive across them, and found that they deflected by several inches under the moving load. He concluded that the design was basically flawed, and that the wrought iron trusses fixed to the girders did not reinforce the girders at all, which was a conclusion also reached by the jury at the inquest. Stephenson's design had depended on the wrought iron trusses to strengthen the final structures, but they were anchored on the cast iron girders themselves, and so became deformed with any strain on the bridge.

Stephenson maintained that the locomotive derailed whilst crossing the bridge, and the impact force against the girder caused it to fracture. However, eye witnesses said that they saw the girder break first, and the locomotive and tender remained on the track at the far side of the bridge. Indeed, the driver raced on to the next station to warn of the accident and prevent any traffic using the line. He then came back on the other side and drove to Chester where he gave a similar warning. The accident occurred a few hours after the track had been ballasted. When the locomotive reached the final girder, it cracked in the middle, allowing all the carriages to fall into the river Dee 50 feet (15 m) below.

The extra load of ballast undoubtedly helped cause the accident. The design of the bridge was seriously flawed, although different authors have emphasised different causes. Lewis and Gagg state that failure occurred in tension at the bottom of the girders, exacerbated by stress concentrations. Henry Petroski notes that the wrought iron bars would tend to exacerbate compression in the beams, and as they are eccentric they increased the tendency towards failure by lateral torsional buckling. The suggestion does not explain the brittle cracking however. It is more likely that the beam cracked by fatigue from a sharp corner in the lower flange by repeated flexing of the girder. William Fairbairn had warned Stephenson of the problem of cast iron girders only a few months before construction of the bridge at a meeting at the Institution of Civil Engineers in London, but his advice was ignored. There had been several building failures involving such girders which Fairbairn had investigated, and had found them to be flawed. He thought the design itself was poor because the trusses could not reinforce the girders, being attached directly to their ends.

A subsequent Royal Commission (which reported in 1849) condemned the design and the use of trussed cast iron in railway bridges. There were a number of other failures of cast-iron railway under-bridges in subsequent years, such as at the Wooton bridge collapse and the Bull bridge accident. Other failures occurred in the Staplehurst rail crash, the Inverythan crash and the Norwood Junction crash. All the structures used un-trussed cast iron girders, and generally failed from blow holes or other casting defects within the bulk material, and so completely hidden from external view.

The Norwood accident in 1891 led to a review of all similar structures by Sir John Fowler, who recommended their replacement. Cast-iron had been used very successfully in The Crystal Palace of 1851 and the Crumlin Viaduct in South Wales (built in 1857), but the first Tay Rail Bridge of 1878 failed catastrophically due to its poor use of the material, putting the cast iron lugs on the columns into tension. The Tay Bridge disaster stimulated engineers to use steel, as achieved in the Forth Railway Bridge of 1890.

The Dee bridge was later rebuilt using wrought iron after several more failed attempts to use cast iron by Stephenson

Photos: Colin Smith / River Dee Railway Bridge;  John S Turner (cc-by-sa/2.0); Martin Clark, CC BY-SA 2.0 <https://creativecommons.org/licenses/by-sa/2.0>, via Wikimedia Commons