Glass is such a good structural medium that in the 16th century Galileo used it to test hypotheses, as did Griffith in the 20th century. But it is only recently that techniques have been developed to attach two pieces of glass together. Pieces of wood can be joined with wood pins hammered into holes that distort both the pin and the hole to ensure a perfect fit. The fit between the pin and hole needs to be 0.5 per cent of the diameter, making tolerance difficult without distortion. For steel, red-hot rivets are similarly hammered into holes in plates. Joining glass to glass, however, necessitates a variation on this theme.
Wood and steel both have an atomic structure that means that when its bond is broken the nuclei behind that act as a reinforcement. In such ductile materials, the reinforcing action spreads out until the pressure is accommodated. In brittle material like glass, however, the nuclei line up, allowing the bonds to break, concentrating the pressure on the break until the material fails.
Developed in steelwork so that bolts could be dropped into oversized holes for a degree of tolerance, friction bolts apply large loads to the face of the plates to stop them sliding against each other. Friction bolting first crossed over into glass in around 1960, primarily to bolt glass mullions to the substructure. Bolt groups, applying loads to the face of glass, could carry quite high loads while providing tolerance, but they needed the help of unsightly flitch plates.
At Parc La Villette, engineer Peter Rice, of RFR, extended the notion of the glass friction joint by chamfering the countersunk hole and clamping the glass sheet to the background with a bush bearing (the RFR joint). This joint, which provides a flush surface, is a satisfactory solution for bolting glass to steel frames and armatures, but its essential asymmetry (fixing from the chamfer side) makes it difficult to use for bolting together, say, two glass beams. But by 1995, Tim Macfarlane, of Dewhurst Macfarlane, put two RFR joints back-to-back in one sheet, and showed this joint carried considerable load, even shearing a bolt in an early test.
Mild steel bolts rust and expand to create uneven pressure points on glass holes, which can shatter them. Poor understanding of the nature of glass, which concentrates rather than disperses the load, meant that the early generation of roughly drilled holes left potential bond-failure pressure points. By using the right grade of stainless steel, this danger is removed. Similarly, joints are further strengthened by using tempered sheets of glass.
Although Griffith had worked out in the 1930s that the theoretical tensile strength of glass is 14,000N/mm 2, this figure is considerably reduced by the action of the molecules in the air during manufacture, breaking the bonds as the glass cools and reducing its strength in annealed flat building glass to approximately 50N/mm 2. Tempering improves the strength to around 160N/mm 2. The new generation of jointing methods, of glass to glass, would not exist without the developments in the use of stainless steel and tempered glass.
Working closely with US architect Bohlin Cywinski Jackson, Eckersley O'Callaghan has developed the structural use of these materials and joints in an all-glass staircase and bridge that forms the core design feature of Apple computer stores across the world.
In New York, the overall clear span of the treads is 2.45m and the loading is around 4.5kN/m 2. The tread is as thin as possible while safely resisting the forces imposed in use, or even if damaged. Tread deflection has been kept within a comfortable range in order to keep the natural frequency above the advisable threshold of 5Hz.
There were two key reasons why an all-annealed glass tread was used. Firstly, it could be polished after the laminating process, and, secondly, should the top sheet be clipped by foot traffic, the tread could remain in place for a reasonable amount of time without needing a replacement, whereas a tempered glass sheet that failed would compromise the aesthetic. The use of a Sentry Glass Plus interlayer, instead of a regular PVB interlayer, allowed a greater transfer of shear between layers of glass under both short-term and long-term loading conditions. The result of this was a significant reduction in deflections experienced for the same loading scenarios as PVB.
A primary concern was the desire for a method of fixing the glass treads to the vertical glass support walls that would integrate the support with the glass itself. This was achieved by experimenting with the lamination of a metal insert within the laminates, discrete to the location where support would be provided by a bolt through to the glass wall. The metal insert, now known as a 'puck', is generally semi-circular in plan and set within the third layer of the tread laminate. The puck's shape eases the build-up of local stresses around the metal insert both during the lamination process and from the transfer of load in use.
Tests with stainless steel and aluminium pucks were not promising - cracking occurred due to an incompatibility between the glass and the metal insert during the cooling and contraction period. Titanium was chosen as a material with the necessary load transfer capacity and with closer thermal expansion and conduction properties to those of glass. The titanium pucks are laminated shy of the glass edge to allow for polishing.
The Los Angeles stair design has treads that are supported between glass walls that are, in turn, hung from the second floor, thus appearing to levitate clear of the floor from which the stair rises. Oversized holes taking the connections were filled with Hilti HIT HY50 - a liquid material that cures after time - enabling all lamination tolerance and hole-quality issues to be ironed out, ensuring a more predictable behaviour of load transfer through the hole in the glass.
The techniques used in laminating material into glass in such a manner as to allow direct connection of the adjacent structural glass elements is a first on such a large-scale commercial project. More forward-thinking, client-sponsored research and development projects are needed to take the material even further.
Brian Eckersley and James O'Callaghan are directors of Eckersley O'Callaghan Structural Design. Email: info@eckersleyocallaghan. com