AA Agendas 8: Nine Problems in the Form of a Pavilion by AA School

Architectural Association London AA Agendas No. 8

NINE PROBLEMS IN THE FORM OF A PAVILION Edited by Alan Dempsey and Yusuke Obuchi

Contents

AA Agendas No. 8

4 6 8

Introduction by Brett Steele Process and Discipline in Contemporary Design Practice by Alan Dempsey On Making Unexpected Outcomes by Yusuke Obuchi

Problem One: A Brief

20 30

Problem Two: Enclosure The DRL Ten Pavilion within the Framework of Parametricism by Patrik Schumacher

Problem Three: Modelling

48 60

Problem Four: Structure Beyond Problem Solving by AKT: Hanif Kara, Reuben Brambleby and Jugatx Ansotegui

Problem Five: Material

76 86

Problem Six: Scale Interview with Wolfgang Rieder

92 98

Problem Seven: Fabrication The [C]Space Pavilion: An Experiment in Digital Craft by Alvin Huang

102

Problem Eight: Assembly

122 132

Problem Nine: Design Research Oneness by Charles Jencks

136 138 144

The Team Construction Diary by JoĂŁo Bravo da Costa Acknowledgements

Nine Problems in the Form of a Pavilion Edited by Alan Dempsey and Yusuke Obuchi AA Agendas Series Editor: Brett Steele AA Managing Editor: Thomas Weaver AA Publications Editor: Pamela Johnston AA Art Director: Zak Kyes Design: Claire McManus Editorial Assistants: Clare Barrett and Mark Campbell AA Publications are produced through the AA Print Studio. aaprintstudio.net Printed in Belgium by Cassochrome ISBN 978-1-902902-82-1 ÂŠ 2010 Architectural Association and the Authors. No part of this book may be reproduced in any manner whatsoever without written permission from the publisher, except in the context of reviews. To order a catalogue of AA Publications or specific titles visit aaschool.ac.uk/publications or email publications@aaschool.ac.uk AA Publications 36 Bedford Square London WC1B 3ES T + 44 (0)20 7887 4021 F + 44 (0)20 7414 0783 Cover photo by Valerie Bennett

PROCESS AND DISCIPLINE IN CONTEMPORARY DESIGN PRACTICE Alan Dempsey It is always a challenge to write a book about a single architectural work, especially if you have been directly involved in its design and construction. Since its completion, the pavilion has been extensively published in both print and digital media. For this book to offer any new insights, a certain detachment had to be achieved. Rather than being merely an annotated précis of the pavilion, this book had to become a project in itself – a means of objectively assessing the strengths and weaknesses of the realised project. I hope that the two-year lapse between the writing of this introduction and the opening of the pavilion has allowed a sufficient interval for some critical evaluation. As architecture is such an intensely collaborative process we have attempted to ensure that this book reflects the multiple voices involved in the project’s realisation. The perspectives of all of the main protagonists are presented through short essays, interviews and a diary mapping out the different stages of the project. As a prototype, the pavilion was bound to encounter new and unforeseeable problems during its design and production. To reflect this, the central section of the book is structured as a series of chapters that, while broadly chronological, each address a particular problem encountered by the team. These problems represent significant moments of crisis: while their resolution was not always entirely successful, they illustrate a process of design innovation that contributed to the success of the completed structure. The nine problems that follow can be related in different ways to the infiltration of digital design tools and environments into contemporary design practice. They are not intended to describe a comprehensive ontology, but should instead be understood as markers in a more open-ended exploration that will no doubt lead to the identification of other issues as computation and construction become increasingly integrated. Although specific to the pavilion, I believe these problems can also be applied more generally to contemporary design practice and education. Within this context, the problems of Enclosure, Modelling, Material and Scale are encountered internally within the process of design itself, as it is transformed by the development of technologies and techniques. These chapters try to show that despite a transformation of our design tools, it is still not possible to account fully for the limits of precision, dimensional tolerance and material resistance. At the same time, the problems of Brief, Structure, Fabrication, Assembly and Design Research are more externally orientated, and address the shifting boundaries of our discipline in relation to the other participants with whom we collaborate: clients, consultants, manufacturers and ultimately a wider public audience. Our use of new platforms of communication has raised new problems, and we have set out to show that the translation of information across these boundaries often encounters significant interference – a fact that is not generally acknowledged in discussions on digital design. There can be no doubt that existing models of architectural education and practice have been disrupted by the spread of computers into design offices. This transformation can be seen as part of a general technological development, creating the capacity to visualise large amounts of information in dynamic, accessible and intuitive ways. These visualisations are like interactive maps where discrete pieces of information are connected to create a dynamic system where the value of any one variable is no longer stable but continually recalculated according to the status of the other variables it is associated with.

Originally developed in other design disciplines, this associative, or parametric, capability enables new kinds of generative systems that can create complex yet coherent assemblies and spaces. Yet within architecture there has been a tendency to assume that these parametric processes are a design agenda in themselves. In my view this leads to the collapse of any ensuing discussion into formal preoccupation. It is true that parametric computation can engender particular formal tendencies, such as the coordination of large numbers of elements, serialisation and differentiation at the level of component and organisational patterning. It is also entirely valid to explore these potentials as a way of expanding our spatial repertoires: despite its physical modesty, the pavilion was intended as just such an exploration – a manifestation of the design research agenda of the AADRL, which had investigated emerging digital design and communication platforms in architecture over the preceding ten years. However, we must remember that parametric computation is an operating environment for designers and not a style to be adopted. Working parametrically suggests a way out of the limitations of a discourse on style that has contributed to the marginalisation of architectural practice since the failure of modernism. If we take this opportunity to discard obsolete institutionalised codes and retool our discipline we can take greater control over the projects we design by extending our reach from the earliest stages of briefing through to the fabrication floor. This extension of our role has the potential to reinvigorate our discipline and reinstate architects as the nexus of the project team. But I do not propose a return to the master-builder ideal of the past. I believe that represents a nostalgic wish for a long-defunct ideal of architectural craft tied to an intuitive knowledge of working directly with materials and easily imaginable spatial strategies. Architecture today is infinitely more complex both in its technical sophistication and in the number of infrastructural systems that a building or city must incorporate. Most of these systems are designed within other domains of expertise, so we can only hope to have a rudimentary understanding of them. Parametrics facilitates our creative engagement with this information without us succumbing to the megalomanic desire to have an allencompassing understanding or control of it. When approached this way, the use of computation and parametric environments offers our discipline the greatest expansion of its creative design space since the industrial revolution and the mass production of building materials. The manner in which we each choose to explore this new creative space remains more open than ever, and the pavilion should be understood as just one instantantiation, by Alvin and myself, of a new architectural domain. But if this book serves a purpose it is to highlight that this new creative domain is not the smooth space that is often described in literature on digital design and fabrication. It is a space where one encounters rough resistance and contaminating noise, and the success of a project will still come down to how we creatively engage with the inevitable disruption in the exchange between material and information.

ON MAKING UNEXPECTED OUTCOMES Yusuke Obuchi A few days before the opening of the DRL Ten exhibition in February 2008, the tenth anniversary of the Design Research Laboratory, I met with AA Director, Brett Steele, for a quick update on the progress of the pavilion which was scheduled to open with the exhibition. From his office window we looked down at the pavilion’s construction site, scattered with loose construction materials, where a dozen students were busy working. At this point the site had become a public spectacle, provoking reactions from people of all sorts, every day from early in the morning till late in the evening. It was a real-time exhibition of DRL students making a pavilion right in front of the AA’s 230-year-old Georgian building on Bedford Square. Despite a strong sense of optimism among students and staff, from our vantage point it was clear that the pavilion was not going to meet its deadline and we needed to come up with an alternative strategy to deal with the widely anticipated opening event. By this time, my position as co-director of the DRL programme had been transformed into construction supervisor, with Brett’s office becoming a temporary DRL Situation Room where problems and strategic planning were discussed while we witnessed the slowly-growing pavilion. It was during this tense, anxious and physically exhausting period that the idea of this book and its title, ‘Nine Problems in the Form of a Pavilion’, was proposed by Brett as an idea to turn the countless problems into something inventive and imaginative. The design problems for the DRL Ten pavilion had in fact begun to emerge the day DRL co-directors and invited jurors selected the winning design by DRL alumni, Alan Dempsey and Alvin Huang. Dempsey and Huang faced an extremely tight schedule to further develop their original competition design and to produce a set of CAD files that could be fed straight into an industrial-scale water-jet cutter to cut over three tons of 13mm fibre-reinforced concrete panels worth over €200,000 in two and a half months. Their design progress was reviewed weekly, often at 8 o’clock in the morning, at the office of Adams Kara Taylor, 15 minutes’ walk from the DRL studio, by a team of structural engineers led by Hanif Kara. One would assume in the era of advanced digital technology, powered by computational modelling, structural simulations and algorithmic optimisation capabilities, that the structural problems hidden in the design of a 10 x 5 x 5m pavilion would not be an issue. On the contrary, we discovered the limitations of digital technology to solve highly complex structural problems. This was partly due to the time constraints, but primarily due to the intricacy and speculative nature of the pavilion’s structural logic which was designed with limited knowledge of the properties of the principal construction material: 13mm fibre-reinforced concrete panels. The creative use of the concrete panel to construct a pavilion in a public space exposed the fundamental conflict between the material’s performance and the designers’ will to overcome structural constraints. Since the growth of mass media in the early twentieth century, the publication of architectural projects has played a significant role in shaping the culture and discourse of architectural education and the profession. Today it is common practice for successful architectural projects to be documented and turned into a monograph accompanied by essays, photographs and computer-generated images. Some of these publications are educational in nature while others are designed as promotional devices. Built projects

shown in these books are well composed in near perfect conditions, with any problems air-brushed out. This book, on the contrary, is less glamorous and glossy and more like a manual. As part of AA Publications’ Agendas series, it is about basic architectural problems in the age of the computational paradigm, where the boundary between matter and information becomes ambiguous, posing the question as to how architecture and the built environment can merge with the global network and information flow. What we have put forward as an agenda here is the necessity for generating and exploring new design problems. These problems are not merely skills-based, but require the ability to see through the complexity of the interconnected, dynamic properties of design systems that architects face today. This is an exciting time for architectural education to be forward-looking and to capitalise on the unprecedented potential afforded by the integration of computational processes and human environments which are becoming ever more individual and selforganised. Students are empowered not only with computers, software and fabrication equipment, but also with the use of common platforms shared by academics, professionals and industrial practices, enabling them to reach out and start connecting dots on the matrix of information patterns and turn them into the knowledge that architects require in the twenty-first century. The complexity of emergent, global and environmental issues at all scales can no longer be ignored, yet to deal with it is a daunting task that no architect through a single project can possibly tackle alone. My rule of thumb is to start anywhere you can. However small the project, the latent value of architectural experimentation cannot be underestimated. In his conviction for an educational system based on learning by doing, Richard Buckminster Fuller famously claimed, ‘there is no such thing as a failed experiment, only experiments with unexpected outcomes’. The success story of the DRL pavilion must be measured along with a series of failed attempts to overcome the problems rarely explored by practitioners. The task for future architectural education is to create an environment where unexpected outcomes, in Fuller’s words, become meaningful and also beautiful.

THE PROBLEM OF A BRIEF Despite repeated claims to the contrary, architecture is essentially a problem-solving exercise. A design usually starts out as a response to a set of very specific criteria, with variables such as location, use, cost, durability, performance and specific materials or construction practices presented to the architect as a set of problems that require a solution. These variables collectively constitute a brief. The problem for the architects is that they tend largely to be arbitrary and indifferent to whether or not the best solution coincides with the designer’s interests or ambition in creating an architectural agenda which extends beyond the individual project. If we regard ourselves as visual artists, architects have perhaps the least control over the realisation of our work and the direction of our careers because of the capricious nature of briefs. In this respect, there is little difference between practice and academia because in both contexts the architectural enquiry begins with the brief. An architectural brief must outline the ambitions and constraints required for the realisation of a design outcome. However, unlike a legal document, which answers questions with yes or no, right or wrong, the function of a good architectural brief is to allow multiple possibilities within a given framework. In other words, a good brief requires both imagination and clarity allowing for different outcomes depending on how it is approached by the prior interests of the designers. Occasionally, however, a chance alignment of beneficial conditions and compatible interests creates an opportunity to produce a project with qualities that extend it beyond a mere solution and imbue it with a more openended possibility. Such projects can expand existing solution-spaces, promote architectural agendas, and contribute to a broader architectural discourse. And, needless to say, such projects are experimental by nature and can enable genuine design research.

Within the Architectural Association there is a distinct history of constructing experimental events, structures and pavilions over the past 30 years. This history of radical spatial and material invention runs through the projects realised by students and staff including: Frei Otto & Buro Happold; John Hedjuk; Bernard Tschumi & dRMM. While some of these projects are naturally more successful than others, they all demonstrate the common ambition of design experimentation. The brief for the DRL Ten pavilion competition shared this ambition, seeking entries that provided more than a solution to a given set of criteria. The brief called for a design that demonstrated the spatial and material research undertaken by the Design Research Lab over the past decade and synthesised it into a new concrete form. The sheer ambition of the brief set it apart from previous structures undertaken at the school – the pavilion explicitly requested innovation in the architectural design, structural performance, user engagement and manufacturing processes. To constrain matters further, it had to be constructed in fibreC, a flat, dense, fibre-reinforced cement panel that had only been used as a non-structural cladding panel system. Moreover, the project needed to be designed and built within a twelve-week construction programme. Two factors that made the project viable: the collective talents of the team at the DRL and AKT engineers, who collaborated in realising the pavilion; and the extraordinary conviction of Rieder, the Austrian manufacturer of fibreC who remained committed to meeting the design challenges. Although the brief for the DRL Ten pavilion included a number of potentially overwhelming constraints, it also represented an exciting opportunity in which matched the ambitions of the participants. It was an opportunity that any graduate would dream of getting.

Mary Miss, Urban Pavilion, 1987. Photo Hélène Binet

Coop Himmelb(l)au, Open House, Malibu, part reconstruction, Bedford Square, 1988. Photo Hélène Binet

THE PROBLEM OF MODELLING Unlimited freedom is often an obstacle to creative innovation. However, dramatic advances in digital design over the last decade have offered that freedom to architects with an unprecedented sense of design precision. New computational capabilities have led to an explosion in generative experimentation. The reintroduction of animation into this design process has further facilitated a discourse that prioritises the iterative over the singular. How then do we evaluate these models to understand if there is genuine design intelligence in the seemingly complex? How do we evaluate one version against another? And, most importantly, how do we communicate the performance of models across different disciplines and digital platforms? The solution to modelling lies in reformulating the concept of the design prototype. The use of the prototype in architecture is nothing new, but the prototype – as we understand it – has two important qualities. Firstly, it is more of an instrument than an artefact. It is not designed as an original from which all subsequent copies will stem, but rather as the first term of an evolutionary process. Secondly, the prototype can be digital. A digital prototype is a fully parametric model – driven by rule-based geometric constraints or specific performance-driven relations. Rethinking the concept of prototype introduces constraint-based rigour into design development, allowing us to evaluate design intelligence in an objective way. It also facilitates an exchange between information and material that is genuinely new. These challenges were all addressed in the design of the DRL Ten pavilion. From the outset the design team included architects, engineers, European- and American-based computation specialists, and manufacturers. The challenge lay in developing the concept proposals into a feasible project and completing the

documentation for its fabrication within a six-week period. In order to achieve this, we had to create a digital prototype that provided a shared platform for collaboration. Model constraints were carefully established to control possible geometric variables and allow the team to assess structural performance and make corresponding adjustments and behavioural predictions. There were significant limitations in our iterative design models, however, as they were never truly parametric in the sense outlined above. The variables did constrain the model’s adjustment but the model itself was too complex to drive that adjustment and automatically update itself. Instead it had to be manually rebuilt through an exhausting process of redesign and optimisation. Transferring our geometric prototypes to other consultant platforms proved equally challenging. An engineering model has to capture very different attributes to an architectural one and this disjunction means that a model, which is valuable in one environment is useless in another, and the engineering team needed to take our design information and overlay additional variables such as material properties, joint stiffness, profile orientation and cross-sectional area. With literally thousands of joints and components, they were forced to automate the process through customised software programming that enabled them to translate our sixteen prototypes into one hundred models for further structural analysis. If the modelling process for the pavilion demonstrated anything, it is that we need to find a way to resolve the disciplinary differences that are being inscribed into software programmes, while also retaining our specific expertise.

Initial sketches exploring the assembly of multiple elements

Initial sketches exploring form and enclosure

Over successive iterations an increasing degree of geometric constraint was introduced into the digital model.

Study models showing options with two (top) and three (bottom) splice locations: the lower option was selected for further development.

THE PROBLEM OF STRUCTURE Structural design involves the search for equilibrium. The engineerâ&#x20AC;&#x2122;s task is to find the stable conditions that effectively use the appropriate materials to ensure all forces are safely transferred through to the ground. Previously this has been achieved through a reductive top-down analysis that assumes uniformity in the system wherever possible. Although such uniformity makes it easier to identify a structural solution, this comes at the expense of introducing redundancy through excess material usage and limiting the range of potential solutions to those that can be described in linear equations. To overcome this, complex structures have historically been analysed with carefully calibrated physical models. More recently computational procedures such as Finite Element Analysis (FEM) have allowed engineers to evaluate the performance of complex structures through virtual tools. Such analyses work iteratively by introducing differentiation and optimising the distribution of material to required areas. As sophisticated as these tools may be, they still operate a top-down mode of analysis that works best for structures that observe a linear behaviour. In other words, for this kind of analysis to work, the engineers need to know in advance what the equilibrium conditions of a structure are likely to be. Running the analysis then tells them how far a structure will deflect from these conditions, with too great a variation invalidating the calculation process. Complex structures often exhibit non-linear behaviour, however, and in order to design these structures engineers need new tools that can perform a bottom-up analysis that examines the properties and connections of individual members, allowing them to produce a solution that cannot be predicted in advance. The structural design of the pavilion required exactly this kind of non-linear approach. If the enclosure had been a series of continuous arches or a compression shell, it would have offered the engineers a

straightforward structural design process through form-finding and material analysis. Instead, it comprised a discontinuous network of components and the resulting shell form was a complex transfer lattice where an alteration in one area affected the performance of another. The engineering team at AKT had to address this problem. At our first meeting they compared the structure of the pavilion to that of a bicycle chain, with each link working independently while also affecting the performance of the whole. In short, there was no way for them to predict the optimal solution from which to begin their detailed analysis. In contrast to this bottomup approach, the architectural design had been defined from the top-down. Trying to reconcile these two demands created some tension during the design process in which an engineering solution was not always acceptable as an architectural proposal. Given the tight timeframe, AKT worked within the constraints of the competition scheme and looked at multiple design options. We soon discovered that creating localised zones of higher strength within the structure by adding extra material had no benefit, as stresses would always be redistributed through the weaker areas of the lattice. Adding reinforcement to one area caused stress failures in adjacent areas as loads quickly exceeded the limits surrounding the reinforcement. As it became apparent that fibreC alone would not be capable of meeting the ambitions of the project, AKT adopted a brute force approach to their analysis by testing multiple engineering models to find a viable solution. In six weeks they translated 16 geometric design models into over 100 FEM analysis simulations that explored everything from triple layers of fibreC profiles, to tension cables, to plywood stiffening boxes. At the last moment three rings of 15â&#x20AC;&#x2030;mm-thick steel plate were introduced into the centre of the structure to provide lateral restraint. These rings were connected with short cross-profiles and knitted with the concrete profiles. This solution was not necessarily the most efficient, but it was the most stable within the timeframe available.

Early sketch of forces active in lattice and resulting distortion

Geometry of lattice before and after loading

BEYOND PROLEM SOLVING AKT: Hanif Kara, Reuben Brambleby and Jugatx Ansotegui From the position of current structural engineering practice, the project for the DRL Ten pavilion offered an opportunity to unite construction, manufacture, design and education in the post-digital era. One potential lay in the different criteria under which the winning scheme had been judged by the AA, AKT and Rieder – posing the broader question of how architectural value is ‘measured’ and why ‘technical’ issues can undermine architectural aspirations. Another question lay in what constitutes (re)search in the disciplines of Architecture/Engineering, in the manufacturing of materials and the processes that lead to the building. The winning scheme provided a valuable case study of the challenges made by digital interventions in representation, technical modelling, manufacturing and assembly. As the technical servant with two masters – ‘the designers’ and ‘the constructors’ – AKT enjoyed the chance to participate in a discussion on the role digital interventions play in the industry. We believe that such discussions enliven the debate between intellectual and less intellectual activities – a disjuncture we believe has eroded the role of the architect and engendered suspicions between clients and designers. The AA employed the concept of a pavilion as a vehicle for implementing a didactic process that has the potential to synthesise design teaching, manufacture and construction. While the pavilion has historically provided a balance between the freedom of an aesthetic piece and the rigour of constructing a public work, the numerous constraints of the DRL Ten pavilion necessitated a rigorously developed design and the assistance of a laterallythinking structural engineer. As engineers, AKT prefer to work with architects – and the design team in general – from the first design phases of a project, enabling architectural and structural design processes to act as mutual stimuli in moving forward. During these early stages architectural proposals are fluid and there is scope for them to be influenced, inspired and shaped by technological concerns. Once a shared design strategy has been identified, the structural designer must then respond with a rigorously engineered solution that is costed and communicated in a clear manner to keep the design on track. The nature of the DRL Ten project – with its short programme and predetermined material – did not allow such an early involvement and it was not possible to compare design options, material possibilities and geometric configurations between the architect and engineer to find an optimum solution. Instead the geometries had already been determined by the winning design proposal, which was being developed by the AA and Rieder, with fibreC already selected as the construction material. One of AKT’s principal challenges, then, was to find a structural solution that was compatible with the architectural grid of planar ribs. It soon became apparent that understanding these structural constraints and how fibreC would respond to the forces generated by the pavilion’s geometry and construction would be essential. A series of mechanical tests on both fibreC and the connection rubber gaskets was proposed, which was undertaken by AKT staff in conjunction with Aachen University. Traditionally, one of the structural engineer’s most important design tools is the understanding of a specific material’s behaviour. At AKT, we view any opportunity to work with new materials as not to be missed and we collaborate with specialist manufacturers to stay abreast of developments. The use of fibreC in the pavilion was uniquely challenging as it was developed as a panelised cladding material. As such, the critical loading typically

comes from the wind blowing perpendicular to the surface of the panel, creating out-ofplane bending. The orientation of the ribs within the pavilion meant that in-plane bending of the ribs would be significant and act in combination with minor axis bending and axial compression. A significant proportion of the mechanical tests carried out were designed to determine how fibreC behaves when subjected to in-plane bending and how this behaviour would vary depending on the orientation of the ribs in relation to the orientation of the original panel. Additional tests also had to be carried out that assessed the behaviour of the bolted connections and the interaction of fibreC with the rubber gaskets. The design of the pavilion meant that these gaskets acted as the only connection between the crossing ribs, so they performed a primary structural function in terms of both vertical load and lateral stability. Although rubber is a homogeneous material with isotropic mechanical properties, it was necessary to mechanically test the gaskets and gasket-to-rib connections to validate the volumetric FE model we had used to predict the behaviour of the individual gasket connections. Once these material behaviours were understood, appropriate properties could be assigned to the connecting elements in the 3D model we employed to predict the global behaviour of the pavilion. AKT developed a model in Rhino and carried out the structural analysis in Sofistik. Even though we had not been involved at the early stages of the design, our subsequent involvement meant we could develop a bespoke geometry-model to structural-model computational interface that enabled a fast turnaround between geometric definition, structural analysis and design, giving the feedback which influenced the next design iteration. This interface allowed the essential structural elements – such as gasket connections and differing rib thicknesses – to be translated from Rhino to the structural analysis model. This standard process – whereby an architect first defines base geometries and the engineer then operates on this model to create the structural analysis information – means the engineer is excluded from defining the structural solution in parallel with the architectural design. Therefore when engineers develop their analysis models it usually implies a significant amount of reverse engineering of the geometry before the forward engineering can commence. We believe architects and engineers must collaborate fully in this aspect of the design process in order to develop an improved understanding of how the geometric and analytical models depend on one another – a conversation that must occur during the early design stages, when the geometry is most fluid and the development between one iteration and the next the greatest. Although the use of digital design and manufacturing meant the DRL Ten pavilion project was completed on time, it also seems clear that the vast majority of projects waste the opportunity to streamline the delivery process through the poor collaboration between architect and engineer at the early stages. Often there is a perception that the architect’s creative edge could be dulled by the ‘interference’ of engineers during this phase, but it should also be acknowledged that the lack of discussion at this point often severely impedes a project in the long run. As engineers, the lesson of the DRL Ten pavilion was a reaffirmation of the importance of teamwork and collaboration – both between and within disciplines – in the realisation of the final project.

THE PROBLEM OF FABRICATION Traditionally the construction process involved a series of on-site operations that transformed raw materials into a completed project. As architecture became more technically complex through industrialised mass production, the concept of prefabrication – in which components were fabricated off-site and then brought into the construction process – emerged. In both cases the construction procedures were similar, however, in that the architect provided construction drawings and the builder interpreted this information and employed the appropriate craft skills or industrial resources to realise the project. The fundamental problem with this process lies in its interpretive nature, where the gap between the architect’s intentions and what the builder understands means that the quality of the resulting architecture is overly dependent on the effectiveness of representational technique and the perceptiveness of the fabricator. The emergence of professional and legal codes for architects in the nineteenth century was, in part, a response to the architect’s separation from the building process. The more recent spread of Computer Assisted Manufacturing has created a revolutionary alternative to this established methodology. Contrary to Fordist models of production, digital fabrication allows design information to be translated into material reality with an unprecedented degree of control. To extend the WYSIWYG acronym, what is drawn is literally what is manufactured. This process also permits a high degree of component variation without incurring significant additional costs. With these potential benefits, mass production processes are being steadily upgraded to enable mass customisation. But in order to achieve the level of precision required by CAM processes, schematic design ideas are now more informed and construction data is required at an earlier stage in the design process. This change means fabricators must

be involved from the outset so the project can be simulated and adjusted in serial versions before it is actually constructed. The DRL Ten pavilion offered a remarkable opportunity to test such ideas and realise a structure through digitally controlled manufacturing processes. All of the production information was issued as 2D CAD files, which contained over three kilometres of cutting paths for fibreC and mild steel. In a three-week period, each fabricator used our CAD files to drive their computer-controlled cutting machines to produce over 900 pieces of concrete and steel along with 12,000 EPDM rubber components. The logistics of the enormous number of components meant that each part had to be identified and indexed in the factory using a five-digit foil sticker that located it in relation to the other pieces to which it was connected. However, the compressed programme did not allow enough time to build a complete clashdetection model to check the fit of all components, which meant the checking had to be done manually. With such complexity, mistakes were made and it was only on site that a few missed notches in the transition from steel to concrete were discovered and remedied in the traditional manner using an angle-grinder. Digital fabrication allows a precision that enables components made from different materials, fabrication techniques, and produced in global locations to be fitted together like a finely made puzzle. This transformation of the design to production process also raises significant questions for the future of our discipline. In return for greater design versatility, we as designers must accept greater responsibility in ensuring the fidelity of design information that will require a closer cross-disciplinary collaboration to ensure the realities of production are integrated into our design thinking.

A profile for the lower wall of the pavilion is taken off the CNC waterjet cutting bed. Photo Rasmus Norlander

PROBLEM EIGHT: ASSEMBLY

104

Once the material arrived on site it took some time to arrange it in sequence.

Early sketch option exploring assembly armature.

The painted steel frame arriving at Bedford Square.

Step 1: A sand blinding and fibreC mat is laid to form a level base.

105

THE PROBLEM OF DESIGN RESEARCH Architectural design research does not really exist. By design research I mean a sustained investigation into the process of design and the enquiry into what we can do with novel technological developments that may then become assimilated into our discipline and in time even transform it. One obvious development involves new digital tools that enable us to design and communicate our work. Design research can also include the investigation of architectural problems through applied design with the aim of producing innovative approaches to the social, political and environmental challenges we face. There is a potential contradiction between ‘design’ and ‘research’, however, with the design process tending to be solution-driven for a given set of variables and more or less closed to new inputs. At all levels it favours precedent in providing a proven approach to reducing the risk of failure. By contrast, genuine research processes are problem-driven and open to any new information that could affect the outcome. In research, the risk of failure is always present and must be actively sought. It is this contradiction that makes architectural research so rare in practice and difficult in academia. Architectural practice attempts to alleviate risk when seeking solutions to client briefs, with innovation tending to be limited to specifically defined areas. After all, architecture is expensive and affects people’s lives when it goes wrong. In a comparable manner, most academic institutions also tend to be solution-driven, restricting themselves to training students through well-understood professional models that reflect the needs of industry both now and in the near future. Of course, exceptions to this exist in the form of a small group of architectural practices and schools that undertake a sustained kind of applied architectural research and experimentation. These

practices – with willing clients – accept the risk of failure in order to pursue innovation. Certain academic environments also accept this risk, and we were fortunate that the AA School is one of them. To quote a recent school prospectus: ‘the school’s aim is not to teach architecture as it is already known, but to create the conditions for new forms of enquiry that will transform architecture in ways not yet fully realised’. The ambition for the DRL Ten pavilion was to deliberately play with this conflict between research and design. By definition the project had to be experimental in experience, form and material configuration. But the finished structure also had to be safe for public access and usable. In retrospect, it is shocking how often the design team were faced with abject failure. Throughout the project’s development, we were unsure the project would work at all and – without the comfort of a Plan B – the conflation of research, design and assembly deadlines meant the outcome would be all or nothing. In addition, all of the problems we had encountered made us unsure as to whether the pavilion would behave according to the engineering simulations or be safe for occupation. In the end it was only through building the pavilion that we proved it could work – providing, through all these uncertainties, the best model of design and research we could hope for. Finally, as the preceding pages show, the design process for the pavilion was neither systematic nor always incremental in its progress. Under constant pressure to meet deadlines, it was messy and nonlinear in its evolution. Each problem that arose led to crises, which sometimes resulted in dramatic transformation of the project. In order to lay claim to an ability to carry out applied design research, we must at the very least demonstrate a capacity to be systematic in the way we document and learn from our experiments, so that we can go beyond isolated problem-solving and create a closer collaboration with both consultants and manufacturers by sharing design intelligence. This book attempts to achieve just that for the DRL Ten pavilion. The pavilion finally opens to the public in March 2008. Photo James Brittain

122

123

The multiple levels and slopes of the thick ground base enabled many different occupation scenarios. Photo James Brittain

128

129