Irregularly shaped building designs with surfaces curving in two directions (“double curved”), and also known as Free Form, Blob or liquid architecture, have gained renewed interest in the last decade due to the then emerging availability and user-friendliness of computerized design tools in the 1990's. These tools were introduced in the domain of architectural design through a technology transfer from film-, car and aeroplane industry. Whereas architects explored the tools' formal consequences and proposed building designs with them, the disciplines involved in the technical realisation of such building designs underwent little change, as they focussed on digitising their existing working methods. This has resulted in structural designs which, despite fulfilling the functional and structural requirements, are not perceived as fully satisfying by the engineering disciplines. Notable shortcomings are the non-conforming of structural shape and architectural shape (e.g. straight beams in a curved building), the structure's inability to anticipate the locally required load bearing capacity, and finally the low level of systemisation, which for instance impedes re-using details in a customised manner. This observed lacking defined the primary research question of this thesis: What are the most appropriate structural schemes, structural systems and structural designs for free form (parts of) buildings? “Appropriate” is defined here as the highest degrees of systematisation, formal freedom and material efficiency, while “schemes”, “systems” and “structural designs” are three levels of abstraction, from merely abstract to highly specified. Through these transitory steps, solutions to be developed may initially be material-independent, thus keeping the range of applications wide. To answer the research question, three domains out of numerous influential parameters have been targeted specifically:
1. The definition of the structure's geometry;
2. The applied material and how it is brought into shape;
3. The structural action through which loads are transferred.
A study of precedent projects, structures and systems addressing sub-solutions on these three domains has resulted in the definition of search directions for structures that meet the requirements. The three most important potential solutions that have contributed to the final three systems were:
1. The use of planar elements, whose contours in plan are not constrained by manufacturing techniques;
2. The usage of developable surfaces. These surfaces can be unrolled to a plane without tearing or wrinkling, thus allowing use of commonly available sheetlike materials;
3. The use of partially controlled techniques to form materials. This means that to for instance curve a flat panel, only the position of a few points should be secured, while the rest of the panel settles itself in between. This greatly simplifies the fabrication of such panels.
Several variables per potential solution, as well as multiple values that this variable can take, were defined. By varying on them, the potential solutions, and their combinations, were systematically explored. As the potential solutions are inherently related to one or more aspects of geometry, structural action or the processing of a material class, the 23 highly abstract structural schemes that resulted, were technically viable. The next step comprised the development of three structural schemes into structural systems, namely the systems: 1. Planar members with end-connections, comprising a polygonal network of custom-cut timber members of variable height, while connections are kept simple by joining only two members at a time; 2. Delta Ribs, comprising a network of curving and twisting members. Its members are triangular in cross section, and are composed of custom-cut steel sheets that are rolled into a single curved shape; 3. 3D components, comprising prefabricated structural elements that are both structure and building envelope. They are based on a curved frame following the same principle as the Delta Ribs system, while the space in between is filled in with a structurally-active layer, which fixates the frame underneath. All systems consist of medium-sized elements that allow the system to locally anticipate the building shape. By doing so it has been demonstrated how the geometrical setup of the schematic setup are propagated to a system design which comprises a specific material and a method to process it, continually maintaining the system's technical viability. To assess the ability to apply the system in a systematic manner, and validate its versatility in an irregular shaped building design, each system has been implemented on an imaginary building design. The application in a specific context resulted in findings on the systems' limitations. The Delta Rib-system has been developed further into three full-scale prototypes, made of sheet metal. Building the prototypes was a test of the underlying system. Although the anticipated primary principles of the system were confirmed, notably the implications of the steel sheets' thickness on the dimensional accuracy necessitated a further specification of the system. This study demonstrates how structures consisting of systematically generated components fulfil the needs for structures appropriate for free form building designs. The systems resulting from this research give unprecedented freedom of shaping, while maintaining rational fabrication standards. This can be implemented through parametric modelling, in which systems adapt themselves to the local geometry, as well as to local structural needs. Technically speaking, structures of irregularly shaped buildings no longer need to be constructed as 2-dimensional frames that slice through a building with total disregard of the building's geometry.