This process of growth under stress enables the remarkable versatility of natural fibre composites. The selective deposition of new material at the position and in the direction where it is needed is driven by the forces the organism experiences. Nature has derived multifaceted and numerous patterns of structural and functional fibre architectures in response to a specific set of mechanical conditions and requirements (Neville 1993). Investigating these remarkably resourceful and performative fibre structures has been a continuous concern of the Emergent Technologies and Design research programme. In the following paragraphs a project will be presented that takes this work a step further by not only investigating the organisation and layout of fibres, but also developing and employing a computational process of growth under stress
for man-made fibre composite structures. Christina Doumpioti’s MArch dissertation (2007) aimed at developing a generic shapefinding and fibre-path generation method for tow-steered composite structures by transferring the underlying principles of natural adaptive growth into a computational design process. Tow steering is a manufacturing technique for high-end composite structures and is currently used in aerospace engineering and sailing technology. This production technology allows fabricating large-scale fibre composite structures, whereby each laminate is produced by combining layers of different fibre orientations, materiality and thickness. Most importantly, tow-steering technology allows for laying each fibre along an individual, digitally defined path. The fibre tows are fed off spools through a tensioning system to the tow placement head, which travels computer numerically controlled along a very thin layer of fibre mats laid onto a mould. The thin mats provide the base surface for individually laid fibres deposited by the fibre head. Compared to more conventional fibre lay-up technologies, tow steering provides for a production process of highly specific fibrous organisations. This manufacturing possibility
enables the conception of large-scale, architectural fibre systems with highly differentiated fibre layouts that correspond with the structural and functional requirements of the system. A computational process of adaptive growth was developed in order to exploit this manufacturing potential. This process consists of two interrelated sub-processes: one generates the particular shape of the overall system, and the other derives the fibre layout as a fibre reference pass for tow-steering manufacturing. Both subprocesses are interlinked within an ontogenetic process, which is informed by external forces and environmental influences acting on the system. In contrast to evolutionary adaptation taking place across generations of individuals, ontogenesis is the process of individual adaptive growth, in which stress acts as one of the main growth-promoting agents. As ontogenetic processes are always environment specific, the vehicle of a specific design project was used to develop and test the generic computational tool. This project’s design intention is bridging between two existing buildings by constructing a long-span monocoque fibre composite shell that functions as a passageway as well as an exhibition space. Thus the objective of the
system’s structural development is improving the load-bearing behaviour, minimising strain energy, levelling the magnitude of stress across the system and achieving a reduction of weight, while meeting the essential criteria of directional strength and stiffness. At the same time the fibre organisation has to respond to programmatic requirements as well as environmental influences. The process of computational growth is based on transferring tow key principles of natural adaptive growth. In the process of generating the overall shape, points act as morphogen cells. Driven by an iterative algorithmic procedure, they self-organise into a particular pattern of point distribution. This serves as a base for defining a surface, from which new nodes are extracted that act as fibroblast cells during the fibre path generation process. Triggered by stress concentrations, they generate fibres in the direction of principal stresses in order to achieve a stress levelling across the entire system. The ontogenetic process of the composite bridge structure is initialised by a first cycle of shape generation. Within a simulated environment of forces and other influences, the design domain is defined as a cylindrical geometry spanning 10 metres between two supports. Within this geometrically set search space, points are initially randomly distributed. Driven by a Delaunay algorithm, a tessellation
is derived by connecting the neighbouring points, while avoiding any intersection of edges between the vertices. For each vertices point a specific load and support condition is specified according to its location within the overall system. Subsequently the resulting structure is evaluated through a finite element analysis. For the key objective function of stress and strain levelling, this analysis calculates an outcome value for the defined set of material properties, for example density and elasticity. In the following step the vertices points with the lowest values of stress begin to act like attractors triggering surrounding points to migrate towards them, whereas the rate of attraction proportionally relates to the rate of proximity. In search of equilibrium of forces, the algorithm iterates through various cycles of structural analysis and point reconfiguration in accordance to the point values obtained each time until a relatively equal distribution of stress is reached. For the fibre path generation, the generated overall shape is further analysed. By means of finite element analysis, the stress type, direction and magnitude are investigated. After defining a stress threshold value, the nodes displaying the highest stress concentrations are marked and defined as agents that organise the fibre structure between the nodes so that the fibres are laid in the direction of the largest principal stresses. They aim at maximising the system’s load-bearing capacity through a fibre
arrangement that at the same time minimises transverse stresses and shear between the fibres. The resulting structure displays a differentiated fibre layout in which each fibre’s location and direction is directly related to the forces acting on it. In addition to the structural requirements, another critical factor was included in the computational process of deriving the system’s morphology. In order to achieve locally differentiated levels of porosity of the composite skin, the generation process is further elaborated by including environmental simulation cycles in dialogue with the structural analysis. Accordingly, the location and distribution of apertures is derived through a twofold process. First a structural analysis of the fibre path system allows for identifying skin areas with the least stress concentrations, that is to say areas where material can be removed without having a major impact on the overall load-bearing behaviour. Simultaneously a fibre volume fraction analysis indicates areas of little fibre density. Where the two areas overlap, there are possible locations for surface openings. In order to select which of these locations should subsequently be used to create openings, solar exposure analysis was employed to investigate the distribution and magnitude of incident solar radiation and light transmission to the system’s interior. Furthermore, computational fluid dynamics enabled testing the impact of prevailing winds on the interior airflow patterns in relation to different distributions of surface openings. This iterative, computationally driven negotiation of design criteria such as structure, space, light and ventilation results in a highly differentiated fibre composite structure integrating a wide range of performance criteria. The computational design process is also directly informed by the manufacturing method of steered fibre lay-up. The splines derived during the fibre path generation process can be directly employed to establish the related machine code for laying up the fibres. According to this information the applicator head of the tow placement machine traverses the surface and applies the fibres where needed. Approaching a surface area assigned to form an opening, the
fibre paths are locally altered so that the fibres are not disrupted or cut. Rather than terminating abruptly, as is common in other fibre lay-up techniques, here the fibres follow the contours of the openings, which allows the forces to ‘flow’ around voids. In addition, the fibre layout on both ends of the prototype structure is further differentiated. Here the fibres are not only oriented according to the distribution of forces, but also to prevent delaminating at the edges. Thus the transversal tows are multiplied on both ends of the structure. The entire bridge structure of the prototype proposal can be prefabricated off site. A mould with the particular surface articulation can be constructed from several pieces with a removable core. This allows for dismantling and removing the mould after the structure is manufactured through superimposed fibre courses laid up by the tow-steering application head. The finished one-piece monocoque shell is lightweight and easy to transport as long as the maximum size of the related loading space is considered. In summary, this research demonstrates the enormous potential of combining the versatility of fibrous systems with new computational design and computer-controlled manufacturing processes. An interrelated development of material, structure and form in concert with novel design methods moves one step closer to the higher level functionality displayed by natural systems. In this process differentiation emerges through the intricate reciprocity between material make-up and environmental forces and influences that result in a structural articulation that cannot be reduced to its load-bearing capacity, but rather provides a robust and multifaceted range of performative capacities.