The Clyde Arc in Glasgow is a relatively young bridge, only having opened to traffic 2006. It received awards for excellence in construction and civil engineering. So it came as a big surprise when the connection on one of the 14 tension bars supporting the deck from the bridge's bowstring arch failed on 14 January.
Contractor Nuttall and designer Halcrow promptly began an investigation into this failure and discovered a second fault – a stress fracture in another of its the connections (News last week).
The Clyde Arc is a tied arch bridge where outward-directed horizontal forces of the arch are borne by the bridge deck, rather than the ground or bridge foundations.
The deck is hung from the arch using 14 inclined tension bars; seven supporting one side and seven supporting the other, over the 96m central span across the River Clyde. The tension bars are connected to the arch using twin lug pin joint connections manufactured by Macalloy.
There was no construction work going on at the time of the first failure, there wasn't even any traffic on the bridge.
The failure is all the more surprising because the bridge is not visibly pushing the boundaries of engineering despite its elegent appearance.
"It's not a radical design as far as the cable system is concerned, but it is interesting in its structural form," says Gifford director Ian Hunt. "I'd be surprised if design was the issue."
ICE Glasgow and West of Scotland regional chairman Gordon Pomphrey agrees.
"While it is a modern design and was built using up-to-date methods, it is not so cutting edge that the security of the bridge should be in doubt," he says.
So how could connections on such a structure fail?
Modern bridges incorporating tension bar support systems generally use high strength steel, which has a relatively high carbon content. This minimises the quantity of steel needed and the number of tie bars.
"There's a compromise between strength and ductility," says Mott MacDonald materials and corrosion engineering technical director Paul Lambert. "The higher the carbon content, the stronger the steel and you need less of it, but you lose ductility. The material is more likely to fail in a brittle failure.
"The size of the defect needed to initiate the crack gets smaller and smaller. With a ductile material, if it receives a knock or a bit chips out of it, the stresses will yield and redistribute. In those circumstances [with brittle materials] they just fail."
There are a number of different mechanisms which can cause failure of a brittle material.
"Stress corrosion cracking is specific to certain metals with high carbon in certain environments where water and high stresses are present," explains Lambert. He says that if a susceptible alloy is used, failure will occur if a combination of stress and environmental factors come into play.
Other mechanisms that can cause a brittle failure of high strength steels include corrosion fatigue and hydrogen cracking.
A source from one of Macalloy's competitors believes that the Clyde Arc connectors could have suffered a brittle failure which had its roots in the fabrication process.
"It looked to be a brittle failure which could have been caused by inadequate heat treatment during manufacture."
But if there was a fabrication issue, should this have been picked up in the production process? Engineers often have to deal with insitu
construction defects but factory made components usually come with some guarantees.
"All manufacturers purchase from a foundry and they specify the product," says the source.
"They should then be testing to make sure [they get what they specify]."
When a steel bridge collapses, fatigue is often an issue. When the I-35W bridge in Minnesota collapsed in August last year, it was fatigue that caused under-designed gusset plates to yield. This is unlikely to have been be the problem on such a young bridge.
"When engineers see metal crack, they often think of fatigue," says Lambert. "It comes from cyclic loading, where a defect is caused to generate and then propagate. However it seems unlikely on a new structure, because it won't have been through the necessary number of cycles."
"It's very early in its history for fatigue issues. It looks to be a tensile failure even though shear has been talked about," agrees Hunt.
Thermal issues are another possibility. On the Clyde Arc, the slender 100mm diameter 35m tension bars could be affected by temperature change.
"The bar is not particularly flexible and if the sockets [connections] are not aligned properly and temperature changes change the shape, the two wings of the socket may not be picking up equal load," says independent consultant Jolyon Gill.
"You would normally overdesign the sockets so they wouldn't fail like that."
Thankfully no one was on the bridge at the time that the tension bar came down, but it does raise the issue of what caused the connection to go at that particular time.
"Nothing strikes me as being out of the ordinary," says Ramboll Whitbybird director, Mark Whitby. "It is curious it failed unloaded."
Materials consultant Sandberg is examining the pieces of the fallen tension bar to try and find out exactly why the failure occurred.
Given the rarity of such incidents, perhaps the second occurrence on the same bridge points to a specific problem with a particular batch of castings or the way in the connection elements are used on the Arc, which should provide some degree of reassurance for other structures that use tension systems.
"Stress corrosion cracking is specific to certain metals with high carbon in certain environments where water and high stresses are present"