|
Shell Logic
A funicular shape – a thin shell in pure compression - may have appeared to be the ideal structural solution for Duxford's roof. There were problems, however. When planes are hung from the roof they induce bending stresses, so that a thin shell must become thick (structurally deep). Moreover, a funicular shape would be a double-curved shell with its radii changing constantly in two directions. A large number of components would be needed to build it. Acceptable perhaps for the Sagrada Familia, but what Duxford's designers needed was a geometry approximating to the pure shell but uniform enough in curvature to be made from a small variety of repetitive components.
Arup's solution for the geometry was to work from a torus – a ring-donut shape. The illustration sows the Duxford shell form abstracted from the outer edge of the torus. While pure torus does not offer exactly rectangular plan components of two constant radii, it comes close. By standardizing on a close approximation, with limited different radii, it was possible to create the concrete roof panels using only six different sets of shell components – an impressive reduction in variety given that there are 924 precast panels. (Blocking pieces were needed in the sets of moulds to deal with the edge of the shell where panels meet the edge beam obliquely).
The concrete shell is in two layers 100mm thick, 900mm apart. Inverted T-beams were stitched together to form the inner skin and the structural webs. There are 274 of them, each weighing 12.5 tonne. Then 650 flat panels were stitched to the webs and to one another to make the outer surface. Over these were laid insulation and a single-layer membrane from Sarnafil.
TEMPERING ENVIRONMENT
Building in concrete was not a foregone conclusion. Indeed, a steel shell would have been slightly cheaper for the same structural performance. Concrete was favoured in part for its ability to help control the internal environment. The thermal capacity of the concrete roof and floor greatly helps to even out diurnal temperature changes. This is important for the exhibits because they are sensitive to humidity change. Stable thermal conditions help reduce condensation risk. Because of this, and because there is little moisture in the main display space, limited mechanical dehumidification was needed to keep relative humidity to a maximum of 50 percent.
In the small exhibition space winter heating and summer cooling keep temperatures to 22±2°C in summer and 17±2°C in winter, with relative humidity also limited here to 50 per cent.
These standards have some of the feel of a transition between indoors and outdoors, and the main exhibition space is even more like outdoors, its temperature free running. The client decided that because people are moving round the museum site in the open, from building to building, this would be acceptable. The predictable temperature conditions for the unheated uncooled display space are a summer temperature of 2°C below outdoors and in winter 2-3°C above outdoors, around freezing at worst.
Services engineer Roger Preston and Partners says the summer performance is being achieved. The trouble is that while the outdoor design maximum was 28°C (indoors 26°C), the actual outdoor temperature at Duxford for several weeks this summer was up to 33°C. Visitors complained. The client is now considering some partial cooling to reduce the worst summer peak temperatures. (Opening lights at the top of the glazing could spill some of the heat built up. Unfortunately they would also admit air with higher humidity.)
The south-east-facing glazing and the perimeter daylight slot are expected to provide adequate ilumination during daylight hours. The museum is fitted with 46 metal-halide floodlights (2kW) which are directed on to the concrete soffit for evening use.
|
