NotesThe architect's competition scheme for the roof was that of four main pairs of surfaces for the Major Hall. The proposals for the Minor Hall varied slightly. Each surface was a triangle in elevation, with boundaries formed by curves in space geometrically undefined. In cross-section, a pair of surfaces (or shells) formed a gothic arch. The main shells were connected to each other by a further series of surfaces termed 'side shells', also geometrically undefined.
The terminology used to describe the roof structure has grown with its development but is, strictly speaking, misleading. The term 'shell' stemmed from early hopes that membrane action would suffice to support the roof structure.
The structural implications of Utzon's design were first discussed with him at the first interview after he had won the competition. There were difficulties with Utzon's design that had to be resolved, thus forging a close collaboration between architect and engineer. The chief difficulties, as perceived by Ove Arup and G. J (Jack) Zunz, a senior partner in Over Arup and Partners, London, were:
(1) The interplay of surfaces made an assessment of structural feasibility by normal approximations difficult and of dubious value,
(2) The scale of the structure was misleading. The size of the site and the scale of the harbour and the bridge tended to diminish the building's apparent dimensions,
(3) Not only were the roof shapes geometrically undefined, but external and internal finishes had yet to be chosen, the auditoria ceilings and their acoustic requirements had not been formulated nor were the size and details of the stages and machinery available.
Several solutions to these difficulties were examined in the early stages of the collaboration between Utzon and Arup. Some ideas 'floated' then were the use of non-pointed arches, doubly curved shells covering each hall, and a single roof without discontinuities over both halls. These solutions were regarded as simple structural forms, and crucially, the solutions were meant to eliminate some of the large bending moments inherent in Utzon's roof shapes.
As said, the roof was initially determined by architectural, rather than engineering considerations. It was apparent early in the project, that any major engineering deviation from Utzon's proposal would destroy the sculptural quality of his design. Nevertheless, there was a firm commitment from Arup to finding a structural solution that would retain the roof profile and silhouettes initially conceived by Utzon.
A two-phased programme of architectural and engineering designs for the roof occurred between 1957 and 1963. The first phase was from 1957 to 1961, and the second was from 1961 to 1963. In the first phase, an enormous amount of analytical work and 'model tests' were directed towards finding a solution to the 'roof problem'. In structural engineering terms, the problems to be solved were:
(1) A geometric discipline had to be imposed on the surfaces in a way which would provide adequate clearances for the stage towers, balconies and auditoria roofs, all of which were unknown in any detail (refer to spherical model of roof).
(2) The roof structure had to be proved stable under all possible loads and without undue distortion under normal conditions.(3) The wind loads on the curved surface were unknown and had to be established by wind tunnel tests (refer to large timber model of the Major Hall, Opera House).
(4) A construction method had to be evolved having regard to structure, cladding (tiles) and the variable geometry of any staging system.
It was clear in these early days that to achieve a solution, to make it possible to build the structure, extensive use of electronic digital computers were necessary, to cope with the massive number of geometric problems and the complexity of the analytical work. Utzon was not opposed to geometric discipline. Early geometric systems embodied a system of parabolas and this greatly improved the appearance compared with the original free shapes of the roof. Crucially, the introduction of a geometric discipline, gradually rationalized the design and construction of the entire project, and made possible the factory production of geometrically similar elements.
The second phase, from 1961 to 1963, saw the design of the roof and associated analyses shift from an ellipsoid scheme to a final spherical scheme.
In its simplest structural form, each main shell was regarded as made from a number of individual tied arches (i.e. the ribs). The two legs of each arch lie in different planes, neither of which is vertical. The plane containing one leg of an arch is a mirror image of the plane of the other leg. To make structural sense of an arch of this form it must be supported at the ridge.
Arup engineers had to overcome many structural difficulties, especially loading conditions. These included: (a) the dead weight of the shells; (b) wind wall loads; (c) auditorium ceiling; (d) wind; (e) temperature; (f) creep; (g) stresses due to the erection procedure. The timber model and the loading condition of 'wind' is important for this collection.
At the time of the Opera House project, it was known (through meterological records) that wind speeds in Australia had reached 185 km/h. Tolerances and allowances had to be calculated for the wind-pressure distribution that would blow over the roof of the Sydney Opera House. Estimates of the wind-pressure distribution over the shells for various wind directions were obtained from wind-tunnel tests at Southampton University and at the National Physical Laboratory [NPL] (England).
The model used was of solid wood representing the roof of the Major Hall and part of the base to a scale of 1:100. A number of sub-surface ducts was formed in each shell, parallel to the ridge, and each duct was connected at its lower end to a manometer. By opening these ducts to the air at points successively closer to the lower end, pressure readings for many points could be obtained from each manometer. To form the ducts, nylon wires were laid in 3.2mm diameter grooves cut in the surface of the shell. The grooves were then filled with resin and the wires withdrawn.
The wind tunnel at the University had a closed working section 1.22m (square) and the model occupied a large proportion of the stream area, particularly when at right angles to the wind. However, concerned was expressed that the high 'blockage factor' of the model would lead to an over-estimate of the leeward suction of the roof.
Further experimental tests were undertaken at the National Physical Laboratory, using a 2.74 x 2.13m wind tunnel. The NPL tests were carried out by Whitbread and Packer (1962) of the Aerodynamics Division of NPL. A series of pressure contours were obtained for Shell A2 when the experiment 'determined' a wind direction from the critical north-east direction.
A typical difference between the University and NPL tests was that for the wind blowing due east. The Southampton tests indicated a range of suction on the leeward face, which were higher than the NPL tests, when the wind blow was set at 161 km/h. On the other hand, the windward pressure distribution was unchanged between the two tests. Suction differences, which were recorded by the engineers at Southampton and NPL respectively, were important for the design of the tile lids and their fixing. To examine the airflow over the shells in the NPL tests, tufts were attached to the surface of the model. Observation of these tufts indicated that the flow separation was confined to the sharp edges (the ridges) of the model. It was understood by Arup engineers that the test results gave pressure distributions, which were adequate for the purpose of roof design.
Des Barrett, Curator Engineering Design.
The curator, D. Barrett, has been assisted greatly in writing this section from conversations with Over Arup engineers Dr John Nutt, Jim Forrest, and Robert O'Hea. A more detailed exposition of this section can be found in Arup, O, and Zunz, J. (1973). 'Sydney Opera House', The Arup Journal, 4-21. See also, Whitbread, R.E. and Packer, M.A. (1962). 'Wind Pressure Measurements on a Model of the Proposed new Sydney Opera House', National Physical Laboratory (Aeronautics Division), Report 1049.
Film of wind tunnel testing of Sydney Opera House model at National Physical Laboratory in 1961 https://www.youtube.com/watch?v=XLKEPDIMtm0