Form-finding for such shells pushes existing computer software to its limit. In fact, it is necessary to produce both physical and computer models. Measurements are taken from the physical model at some stages to inform the computer model. The physical model is also an opportunity to rehearse the construction of the actual grid shell.
The choice of laths for the grid is still under discussion. They must be thin enough to bend to the required curvature while also providing adequate strength. The solution is a double-layer grid of 50 x 30mm timbers. Testing suggested oak would be ideal: not least because it is not prone to sudden, brittle failure. However, if the timber is sourced from thinnings rather than mature trees, chestnut may perform better. The costs for both woods is similar.
Which ever wood is chosen will be used in its green state, though the small size of the sections means that the timber should be relatively dry when installed. Laths are cut from the tree in 2-3m lengths and finger- jointed to about 12m in the factory. Once delivered to site, scarf jointing or some simple alternative will provide the required lengths of up to 40m.
It is hoped that all wood can be sourced locally to the museum. Timbers may have to be sent elsewhere to be laminated (for the floor beams) and fire treated (for the laths).
All this is some time away: following the hoped-for go ahead in October, there will be four to five months of detailed design work and around 12 months of construction.