Professor Kourosh Kayvani, Aurecon's Global Director of Excellence and Expertise, explores the increasingly important role of advanced structural modelling and analysis technology in achieving optimum performance based design outcomes.
Architectural history will record the early years of the 21st Century as the beginning of a quantum leap in landmark building design.
According to the Council on Tall Buildings and Urban Habitat, some 38 buildings have now reached heights of 300 m or more, meeting the definition of “super-tall” structures. The council estimates that a further 76 super-tall buildings are currently in construction. For six years, Taiwan’s Taipei 101 Tower was the tallest building in the world at 508m. In January 2010, it was humbled by the opening of Dubai’s Burj Khalifa, at 828 m and consisting of 162 stories.
While super-tall buildings are the most obvious example of previous design records being continuously smashed, this is not the only area of extraordinary innovation. For years, some of the world’s most spectacular long span roof structures remained among its earliest, such as the 73 m train shed at St. Pancras Station in London, built in 1868, and the 114 m Galerie des Machines for the Paris Exhibition of 1889. Yet, in the 21st Century, roofs have tripled in span and found spectacular geometric forms, with marvels such as Wembley Stadium with its single structural span of 315 m in its arch, and the bubble-shaped biodomes of the Eden Project.
But, what’s driving this sudden explosion of innovation? While architects and engineers have always sought to push the envelope – with “deconstructivist” architects, such as Frank Gehry and Daniel Libskind, bringing unpredictable shapes and a sense of controlled chaos into the urban environment – these latest designs appear to tear the envelope apart and scatter it to the four winds. For an industry that has always been restrained by incremental development – looking at what worked before and improving on that – what has flung the door to creativity wide open?
While new materials have played their part, a major driver is technology. For years traditional structural engineering was constrained by codified design rules that were simplified to reduce time in analysis. While this was cost effective, the rules were somewhat inflexible. Of course, not everyone played by the rules. Felix Candela and Luigi Nervi used their ingenious approach and deep understanding of structural behaviour to flout convention with their marvellous concrete shell designs. However, for many, lack of ability to analyse complex structural forms under complex loading condition has been a key limiting factor in translating structural imagination to reality. But the opportunity is now finally here. With modern computational tools even the most complex analysis can be done in reasonable time, allowing engineers to test ever more versatile designs.
This new wave of design-enabling technology includes computation simulations techniques using non-linear finite element analysis (FEA) and the like. Non-linear FEA allows engineers to analyse complex structural forms by closely simulating their behaviour and testing for stresses and displacements. For example, when designing high performance and complex steel structures, advanced FEA software allows increasingly realistic modelling in 3D, accounting for complexities arising from geometry and nonlinear effects such as large deformations; plasticity; friction, sliding, and ‘gap openings’; and residual stresses in the steelwork.
By building and analysing a simulated model, engineers can overcome the unknown, creating an unconventional design that pushes the boundaries of form and lightness, while still achieving its core objectives of strength, durability and serviceability. Thus, a building can be tested and de-risked before it is built, without compromising its technical integrity.
Non-linear FEA is far from new. For decades, many industries such as aeronautical, automotive and manufacturing, with a vested interest in avoiding the expense and time involved in testing physical prototypes, have used FEA models, speeding time to market and making better, safer products.
That the construction industry appears to have lagged behind is down to simple economics. The companies that first made use of non-linear FEA techniques were using them to design products that would be sold in their hundreds of thousands – if not millions. A building will only be constructed once.
While super-computing remained the prerogative of the super-wealthy, no single building could afford the luxury of testing the unknown. Its architects and engineers were forced to build from past experience, because no safety code would allow the construction of a building design that could not be tested.
But, in the 21st Century, dramatic increases in computing power and the development of intuitive interfaces have changed the game, making advanced modelling tools such as non-linear FEA software available to anyone with a PC – and allowing the increasingly realistic modelling of innovative structures that simply would not have been built in a previous century.
For example, without computer modelling analysis, the iconic arch floating above London’s Wembley Stadium would have remained a designer’s dream.
Weighing in around 1,500 tonnes, Wembley’s giant, yet slim looking, arch is a 7 m diameter lattice shell structure with a span of 315 m and a height of 135 m. The long span nature of the arch required it to be designed to be as efficient and lightweight as possible. At the same time, the architectural intent of the arch required it to be as slender as possible.
The structural implication of this slenderness is a unique level of interaction between the overall buckling of the arch and the local buckling of the individual chords.
To achieve this lightweight, yet safe, design, extensive nonlinear buckling analyses were needed to consider the destabilising effects of expected geometrical imperfections in the arch in its erected form. This required the design team to develop novel incremental analysis techniques to account for tension stiffening and softening of the cable net formed by the forestays and backstays, in restraining the arch at the onset of buckling. Local and global geometrical imperfections in the members of the lattice arch were introduced into the analysis by modelling a modified geometry of the arch consistent with the eigen-vector for the buckling mode shape under consideration. This allowed the overall safety factor of the arch to be determined with a high level of accuracy.
FEA-based design was also used to determine the final thickness of the connection plates for the 500 or so nodal connections in the arch. In this procedure, 3D models of the arch nodes were generated on a CAD platform and imported into the FE analysis software. The objects were then connected and auto-meshed with an optimum level of refinement, striking a balance between computer run time and the precision of the results. Design criteria were developed, based on first principles and calibrated against safety indices of the steel design standard (BS5950). Finally, these criteria were applied to assess the analysed stresses and strains, to ascertain the adequacy and efficiency of the connection design.
In Adelaide’s rapidly changing architectural landscape, the elegant, slender, and lightweight designs of diagrid steel shell roofs over the redeveloped Western stand of Adelaide Oval and the spectacular new dome canopy at the entry of the Adelaide Entertainment Centre would have remained on paper without these exciting new design tools.
While, advanced structural analysis techniques such as non-linear FEA are potentially very powerful tools for performance based design, there are significant challenges involved in their effective application. This requires high level skills and knowledge in both structural behaviour as well as the characteristics and limitations of the software. Some of these challenges, as they relate to steel structures, are discussed in a recent paper delivered at the joint IABSE-FIP conference on Codes in Structural Engineering (K. Kayvani, May 2010, Dubrovnik, Croatia). This paper discusses effective modelling techniques and ways of interpreting the modelling results. The aim is to meet required safety margins of the codes of practice without being overly conservative. For example, while localised regions of yielding in ductile steel material may be allowed, limits are put to the levels of tensile plastic strain and the extent of the compressive plastic zones accounting for residual stresses due to welding, and initial imperfections that could lead to local buckling of steel plates.