At Foster Associates' new Hong Kong international airport on Chek Lap Kok island a pair of canopies covers the spaces between the main terminal building and the two vast container storage systems (part of the airport's 26,000m2 cargo handling facility known as Superterminal 1). The 200m-long canopies 35m above the css Roadway, as it's known, provide all-weather cover for vehicles transferring cargo from the airport apron to either the main terminal or a storage system. The Hong Kong building authorities said the road could also be used as an access route for fire-fighting vehicles. The trade-off was that the canopies should not fail during a fire or, more specifically, that their structure should have a two-hour fire rating and sprinklers on the under side.
The worst scenario envisaged was a 7MW fire in a dolly truck below ground level involving fire impingement, that is with flames engulfing the structural members of the canopy. The likelihood of this happening to a structure 35m up was fairly remote and at first Arup Fire argued that fire protection was not really necessary even when alternative scenarios such as fire in either of the adjacent buildings or one of the high level link bridges were taken into account.
But regulations are regulations and the Arup Fire team's task became either to conform conventionally, or come up with an intelligent fire- engineering solution. A conventional solution - using passive methods such as physical fire protection and perhaps intumescent paint - posed too many real-life maintenance problems.
The canopies had been designed as long shallow barrel vaults with orange- segment-like chunks removed every 4.8 metres. The supporting structure is a double truss with a straight common bottom chord with two inclined arches above connected at the ends and joined to the bottom chord by regularly spaced straight tubes inclined to meet at the bottom of the open V. The sloping straight tubes are supports for glazing and vents on either side. The roof between adjacent double trusses is supported by conventional steel purlins.
The fire engineers suddenly saw the solution: fill the structural members with pumped circulating water which would also feed oversized sprinkler heads fixed into the underside of the bottom chord of the trusses. Happily the Foster office had already designed the trusses of the canopy using steel hollow sections. All that was needed was to make sure that the waterways were clear, especially at junctions. That could be done by welding sockets on to main tubes and drilling out the connecting hole using the socket as a jig.
That looked like a neat solution. And it wasn't out of the blue. There were precedents. Back in the 1970s Arup itself had designed the water- filled tubular external lattice structure of the City of London's Bush Lane House and there are contemporary examples in France and Germany. The forerunner of such structures had been put up some time before: the tower for us Steel in Pittsburgh, with its water-filled columns.
The first patent for the idea was taken out nearly a century earlier in 1884 in the us, and the original idea remained more or less unchanged. The conventional approach to fire safety assumes that various forms of insulation will delay the transmission of heat to the steel and thus delay its collapse so that people can evacuate the building. The alternative and perfectly sound approach argues that the heat of a fire impinging on steel can be dissipated by constantly circulating water inside the tubular members.
Quite a lot of research had already been carried out on the idea, notably gvl Bond's constrado report of 1975 ( now re-issued by the sci as P038). To ensure Arup Fire's solution would work in real life as well as in theory, the team persuaded its client to have a full-size section of the canopy structure assembled and tested at Faverdale Technology Centre in Darlington. Here it set up the sample water-filled structure (fixed 2m rather than 35m above ground), lit the butane heaters and powered up a 7MW fire with the flames lapping the trusses. It set up several improbable worst-case scenarios including not turning on the sprinklers (which would, as in real life, have put out the fire) and diverting the water away from the hot spots. The team had calculated that the steel would reach temperatures around 200degreesC, well below its failure temperature. In fact the test generated steel temperatures of only 46degreesC. With this in mind the team called in the Loss Prevention Council and repeated the test. When the fire was put out, the truss was cool enough to touch.
Although sprinklers had at first seemed a bit superfluous it turned out that they helped the whole system to work automatically. First, the water pressure created by the circulating pump meant that pressure would be maintained at each sprinkler. Second, and perhaps more important, when a fire broke out the glasses in a group of sprinklers around the seat of the fire would break and the sprinklers would activate. Because they were using the cooling water in the tubes, new and cool water would be drawn to the hot spot, thus self-regulating the cooling of the structure.
One of the misconceptions about sprinklers is that they rust up and don't work when needed. The team did some work in cutting sections of pipe from quite elderly existing installations and in no cases came up with anything like a serious problem. It seems to be the case that when the sprinkler water has become deoxygenated during preliminary rusting, which happens quite quickly, there is no possibility of further oxidation. The conclusion was that there would be no need for rust inhibitors, further reducing the maintenance budget.