The AA Membrane canopy was designed by the EmTech Programme 2006/2007, in collaboration with Buro Happold, one of London’s leading engineering companies. The project was completed for the end-of-year Projects Review at the Architectural Association School of Architecture with the intention to remain for a lifespan of two years, and to function as a canopy for the roof terrace of the School. It aimed to operate both as a design research exploration of material and construction experimentation as well as a commissioned project for the School’s most socially interactive open space, with a restricted design and manufacturing period of only seven weeks.
For research purposes, the structure served as an attempt to test the notion of design and construction in a collaborative and integrated manner: collaborative as it involved both architects and engineers who worked very closely in all matters of design decisions and integrated as it focused on the continuous feedback and interrelation of various (and sometimes divergent) parameters such as material thresholds, assembly logics, performative potentials and budget constraints.
The canopy was a highly differentiated cantilevered structure resting on just three points and comprised of fabric membranes and steel members. The individual lengths and bending angles of the steel rods, which are the compressive elements of the canopy, define the global geometry, while the membrane elements act in tension and allow permeability. The overall morphology consisted of 150 membranes and 650 geometrically different steel elements.
The component differentiation and emergent global morphology was driven mostly by spatial and environmental inputs, which negotiated different factors: pressure differentials caused by wind loading, sun-shading and rain protection, structural lightness and the maintenance of the northwest view towards Ron Herron’s Imagination Building. The notion of differentiation was inspired by biological systems, and in particular cellular arrangements, whose differentiation is dependent on each cell’s position and role within the overall system. Likewise in the canopy system each component was derived from the hyperbolic paraboloid geometry, but depending on the relative position of its four anchor points settled into variant geometrical configurations.
Porosity between components was calibrated by maintaining overall curvature continuity while achieving overlapping between each component to allow for rainwater drainage. This was dependent on two main design factors: the depth of vertical compression members and their angles in relation to the local normal of the base surface. While the depths of the vertical elements were adjusted for structural or shading purposes, the logic of the overlap needed to be preserved. Therefore, although from above the surface can be perceived as a singular surface, from all other views it is seen as a highly porous system allowing for light penetration and a number of framed views.
Parametric modelling, developed using Generative Components software, underlined the entire design process by facilitating a seamless interdisciplinary exchange between the architects and engineers, enhancing the integrity of the design. The associate modelling software enabled a significant level of control over an intensely complex structure through a hierarchical build-up of parametric relationships in tandem with certain control mechanisms. The model was continually updated using interpolated data from the engineering analyses regarding global geometric strategy, local and global population densities, force vector paths and structural depths. For example, in cases where the engineers indicated, after structural analysis, that a certain part of the structure exhibited an excessive magnitude of stresses, the architecture was seamlessly adjusted in multiple ways to better accommodate the stress distribution along the surface: [i] by increasing the sectional dimensions of the local compression elements, [ii] by altering the overall form to reduce the extent of the cantilever, [iii] by maximising local component density and thus creating less surface exposure, [iv] by increasing structural depth caused by an increase in the height of vertical elements or [v] by altering the orientation of the compression elements streamlined to match the stress trajectories, thus better encompassing the stress flow.
The chosen material assembly along with the geometric behaviour, structural requirements and manufacturing restrictions became the variables of the associative model, leading to a variation of possible solutions, all of which complied with the set range of constraints. The final form therefore was derived not from a formalistic exploration but from the negotiation of different design criteria that allowed for the generation of a robust and multi-performative system. Iterative analyses through different software, along with a series of physical tests drove the ‘fitness ranking’ of each phenotypic result and helped for the identification of the most coherent solution in terms of the design objectives and assembly logic. CFD analysis, in feedback with environmental simulation analysis, demonstrated the system’s interaction with environmental inputs and aided in its development and calibration.
The final digital model was translated into manufacturing data so that the membranes could be laser-cut at the London College of Fashion, utilising their expertise and available machinery; steel elements were fabricated manually by students using the Architectural Association’s workshop. The establishment of the associative parametric framework proved to be highly indispensable for such a differentiated and precision-demanding structure; the component-based scheme permitted a high level of prefabrication and manageability through a repetitive manufacturing process. Data related to the dimensions, position and joint angles of each designated element was exported to a master document for fabrication while each member was described with a scripted labelling system that allowed easy identification during manufacturing and assembly.
This collaborative approach enabled a thorough exploration of various design outcomes in a bottom-up evolutionary process resulting in an integral design of interrelated formation and materialisation. The project was approached as design research and exploration, identifying the potentials and challenges stemming from a multi-variable, cross-disciplinary design framework through computational morphogenesis. The difficulties that arose by trying to encompass all the different parameters became the point of departure for further projects to follow; future research would benefit from the coherent fitness ranking of different design objectives.