Design Museum Pavilion · form & structural system

Peter, Margaret & Hattie
Looking down on the site from the Museum offices
Play of shadow and light under the model

Yesterday I had a really productive meeting with Peter and Hattie to talk through the possibilities of the proposed double-curved structure – a ‘hyperbolic paraboloid’ form similar to that used on the iconic roof of the former Commonwealth Institute. We met on the site, with a model, which was a useful way to think about the design physically and spatially. Peter quickly explained the two main challenges posed by the way I had modelled the structure – firstly the lack of supports at the two high points which would lead to global overturning, and secondly the lack of triangulation within the surface of the form, which would lead to distortion. I had used a series of ribs all running in the same direction, with no perpendicular cross-members, which demonstrated very beautifully the way in which a curved form is achieved through straight elements.  Whilst we all agreed that this is a more elegant solution, and creates a uncluttered visual pattern as you view the structure from different angles, the concern is achieving adequate strength in each rib without increasing the dimensions too much. This is particularly true of timber, which might end up becoming very chunky.

The size of the structure is determined by the need to span the existing benches (6.2m at the widest point) and achieve a minimum height of 2m at the center of the pavilion. This leads to a height of 4m at either end and members which are about 8m long – which is quite a large span for timber. One potential option is to have two sets of cables, one running above the structure, and the other running in the opposite direction underneath, essentially breaking the span of each member into three shorter pieces. These could be notched into the timber and have the lighting integrated into them, so that they’re read as an integral part of the structure. Another option is to use timber for the perimeter pieces, and a cable net solution instead of the ribs. This would avoid the complication of a series of individual timber joints at the junction between the perimeter pieces and the ribs, which would each meet at a slightly different angle. It might also be easier to re-assemble, as timber frames always require a certain amount of nudging to fit into place and can end up a bit battered and full of screw holes when dis-assembled. Given the structure will be built as part of a workshop with students (essentially unskilled labour), making sure the construction process is as easy and efficient as possible is really important.

The other crucial thing to resolve is the foundations for the two lowest points of the structure where it meets the ground, and to tether to the two highest points. The area immediately around the benches is paved with new granite sets, and the sloped banks around the site are made of freshly laid soil which probably isn’t very well compacted. Given this landscaping is already complete and we want to keep on-site activity to a minimum (both during assembly and disassembly) pouring concrete pad foundations isn’t a viable option. Instead we discussed using something ‘as-found’ like precast concrete bollards, which the pavilion could be post-fixed to with a steel footing. This would also lift the structure up off the ground, protecting it from water damage and increasing the height underneath the edges of the structure, which is quite low at the moment and makes some of the benches hard to use. For the tethering of the high points, we could possibly just have one support at the rear… but this, along with many other questions, will have to be explored through computer modelling – which Peter’s team will do next week. Alongside this, I’ll be making more physical models to understand the design implications of each solution, and developing options for the cladding, which will be independent of the structural frame, allowing it to be changed and adapted throughout the installation.