Chapter 4

Architectural Design Proposal

4-1. Introduction

This chapter introduces the philosophy of Programmable Architecture (PA). The developed architectural model, a flexible kinetic-tensegrity structure, provides one approach to PA but the number of possible solutions are countless. The author does not think that there is one optimal solution for PA but rather it needs to combine local decisions for each of its functions or programmes. Thus, the proposed solution should not be taken as the best answer for PA, neither in terms of component design nor with regards to its global design in relation to the site. Rather than searching for the best design operating in its optimal state, the principle of PA offers an alternative design strategy. This system should be seen as a dynamic architectural product that is able to answer different user requirements deploying different solutions in each circumstance where it is implemented (built/introduced). It has the capacity to be developed using a variety of locally available materials, at different scales and within varying budgets.

4-2. Space Reconfiguration

As an assumption, it is set that every activity which contributes to all the actions which make up daily life, is determined by a set of environmental conditions. From this hypothesis, reductionistic environmental parametricism is now introduced. If this hypothesis is to some degree true, activities become easier to control by adjusting environmental parameters. It becomes possible to reconfigure activities through environmental reductionism. For example in this figure, a space (5m x 5m) in the park is bisected (sectioned), and its environmental conditions such as light, air quality and flow, temperature and sound are analyzed layer by layer. Some environmental layers such as light are measured using scalar units such as the illumination level, while other layers such as the air flow layer use vectors to describe wind direction and turbulence. When this data is analyzed and displayed as a grid over the site, it becomes possible to reconfigure the space through adjusting the overlaid environmental factors.

Fig 4-2,1: The Idea of Environmental Reductionism (by Author)

This is a diagram to explain the possibility of transferring activity by the environmental reductionism idea.

4-3. Electrical Controlled - Cybernetic Architecture

The proposed cybernetic architectural system is not an incredibly complex system, especially when compared to existing machines such as cars, computers or even a 17th century steam engine. At its heart is the Feedback Loop. The Feedback Loop system is defined as a re-insertion of information where one system’s resultant information (output) is inserted into another system’s processes (input). Cybernetics (Wiener, 1961) is the word used to convey the idea of comparing physical machines with biological organisms in regards to how their behaviour and metabolism are controlled. Both possess sensors that feed information into a decision-making process which then regulates the output, or behaviour. In the early development of Cybernetics, which began with Norbert Wiener, three key aspects of the process were identified:

•coordination (the system which controls the machine itself);

•manipulation (the human input that corrects the error);

•the governing factors (the system which includes a group of people and the controlling dynamics).

Wiener treated this process as a feedback system and the information flows were identified either as an input or an output, which was ground-breaking at the time (Wiener, 1961). This proposal will explore the development of a model for an urban metabolism with the design and construction of intelligent buildings and urban blocks that can modulate their surface properties and spatial configurations in response to external stimuli (such as light, sound, wind etc.) as well as climatic and ecological changes coupled with economically driven programmatic changes.

Fig 4-3,1: The Diagram of Electrical Circuits (by Auther)

Fig 4-3,1: The Diagram of Electrical Circuits (by Auther)

4-4. Flexible Structure: Kinetic Tensegrity Component (Drawing)

Why use tensegrity components? Tensegrity-tensional integrity-developed by Gomez-Jauregui has a basic structural principle that uses isolated components in compression inside a net of continuous tension, in such a way that the compressed members (usually bars or struts) do not touch each other and the pre-stressed tensioned members (usually cables or tendons) delineate the system spatially(Gomez-Jauregui., 2010) . A tensegrity structure has a number of unique points.

・Firstly, loading members are either in pure compression or in pure tension, meaning the structure will only fail if the cables yield or the rods buckle.

・Secondly, preloading or tensional pre-stress allows cables to be rigid in tension.

・Thirdly, mechanical stability allows the members to remain in tension/compression as stress on the structure increases.

Because of these patterns, no structural member experiences a bending moment. This can produce exceptionally rigid structures for their mass and for the cross section of the components.

In addition, tensegrity structural systems have an affinity for axial actuators such as Bio-metal(R) as noted below. As Harvard physician and scientist Donald E. Ingber pointed out in his article (Ingber, 1998) ; “The tension- bearing members in these structures —whether Fuller’s domes or Snelson’s sculptures —map out the shortest paths between adjacent members (and are therefore, by definition, arranged geodesically). Tensional forces naturally transmit themselves over the shortest distance between two points, so the members of a tensegrity structure are precisely positioned to best withstand stress. For this reason, tensegrity structures offer a maximum amount of strength.”

Hence, tensegrity’s lightweight characteristics make it suitable for adaptive architecture according to Korkmaz et al (Korkmaz, 2011) . Eleanor Hartley further points out that visual transparency contributes an important aesthetic quality to these structures. (Hartley, 2009) Theoretically, there is no limitation to the size of a tensegrity structure. Cities could be covered with geodesic domes.

fig 4-4,1: Kinetic Surface in Msc thesis at Bartlett UCL, 2008 (by Author)

This is the electric controlled fabric on the thesis (Programmable-Kinetic-fabric for Architecture, 2008). The components are not yet tensegrity but use a somewhat tensile system. The deformation was really limited both in the point of direction, and amount of movement.

Fig 4-4,2: The Early Model of Tensegrity Surface, 2010, by Author

This is the very early physical model of tensegrity surface, made with bamboo sticks and rubber bands. Individual components, made by simple tensegrity, can stand themselves but the surface could not keep their shape against gravity.

During the development of the component based system proposed in figure 4-4,2, the author paid special attention to the development of the physical model. In the beginning of the experiments the components had binary behaviour – flexible or rigid. There was no space for an intermediate stage between the two behaviours. The malleability of the model became controllable over time and the final component can exhibit different levels of behaviour from flexible to rigid. The adaptive behaviour (in terms of rigidity of the component) was achieved by using different types of springs. The component would change according to the desired performance. The membrane, for instance, was evaluated according to the amount of shadow it could provide. The structure was evaluated according to the degree to which it is able to change its shape.

Fig 4-4,3 Self Stand - Recover Tensegrity (by Author and Y.Komiyama)

In contrast the designed tensegrity system has a single component that after local deformation at the component level can recover its initial shape. There is no pre- differentiated morphology; the differentiation occurs through electrical controlled inputs. The minimum requirement for the component is that it have change-ability, and self-recovery functions in its shape. As demonstrated on the fig this architectural system consists of a 2D array of the same component and it can be industrially mass produced.

Fig 4-4,4 , The 1st Generation Model Parts at Kinetica Exhibition, 2010 (by Author and Shari*3)

This Picture shows every part of one tensegrity component. In this generation, still the number of parts are too much, so to make one components takes time and man's energy.

Fig 4-4,5 : Responsive Kinetic Tensegrity roof patch (by Author and Shari*3)

This model is electrically controlled, dynamic responsive model. The system is light sensitive, when it detects light the structure starts to deform, each component has its own sensor, so the movement is organic and smooth. This model was exhibited at Kinetica ArtFair2012.

Fig 4-4, 6 , Manual Control Kinetic Tensegrity Roof Patch (by Author and Shari*3)

This picture shows one of the alternative control methods of tensegrity structure. Basically this structure is working automatically detecting light with sensors, but also allowed to control human input through mobile. Here Android mobile is used to control wirelessly.

Fig 4-4, 7 ,:The 2nd generation Model at SFF Exhibition, 2011 (by Author and Shari*3)

This picture shows the second generation model of this system. This is made with laser cut MDF and fishing wire and bio metal. First whole model which has an electrical system with Arduino though still the scale is small.

Fig 4-4, 8 : The 2nd generation Model and its Movement (by Author and Shari*3)

This diagram shows the ideal deformation of this model.

Fig 4-4,9 : 3rd Generation, Physical model and it's Parts Fabrication

Reviewing the previous model, the simplified model is designed. This makes easy assembling of the model. Also quick production. Recent development of rapid prototyping, especially in this case laser cutter, the planned parts can fabricate easily, quickly and precisely. In the future development 3D printing methodology is also worth to use.

Fig 4-4,10 : 3rd Generation, Physical Model

This picture shows a simplified model, and attached membranes. Models are scaled up to almost 300mm as one component. It is fold-able, easy to carry, and light weight structure.

Fig 4-4,10 : 4th Generation, Physical Model

This picture shows the final model, scaled up to almost 700mm as one component. There is no rule but this could be one to one scale model. The second picture shows the metal parts at the end of the suspension. These parts enable us to change components hardness/flexibility. Depending on that, components can adapt actuator's power.

Fig 4-4,11 : Drawings of the Components (Active Shading System)

With the changing degree of the membrane, components can shade the light. So this component can adapt to required illumination level. These components work as a both environmental shade and structure. These components have a changeable membrane inside for shading and air transparency. The components are made by carbon (or aluminium) pipe, Its depth will be changed depending on the part. If Internal stress becomes high, it will be amax 4000mm on the top.

Fig 4-4,12 : Drawings of the Components (Integrated Skin)

Top of these components are covered with inflated ETFE. This material is flexible and printable, so it becomes possible the top can generate solar energy power.

fig 4-5,1: Image Perspective Okayama Competition (K.Hotta and Shari)

The proposal was for a single large fabric structure covering the huge plaza (100m x 45m) with a soft kinetic roof creating one large square. Normally the building space and the surrounding space are treated separately. This large roof consists of a number of tensegrity components each with its own unique character.

A real world proposal is shown here to further illustrate the possibilities of PA. This proposal for a LRT (Light Rail Transit) station in Japan was for a competition held in 2010 in Okayama prefecture. This architectural paradigm diverges from the traditional architectural planning method. In Modern architecture functionalism was looking for an optimization of the space. The design needed to fulfill the maximum architectural function, whether a bedroom or toilet among others. In other words, the space was designed for a static and specific function. For example, when an architect designed a 'bedroom', he/she thought: it is better to have carpet than wooden pavement as for functional reasons silence is needed; a big window is not needed since the room is mainly used to sleep; the toilet should be close to the bedroom but the entrance hall which is a noisier place should be far away from this room, etc... Following such practical and functional decisions, the 'bedroom' functioned as required – a silent and dark place where one can sleep. For each programmatic function there were a set of parameters that one needed to control from the outset.

However the proposed PA, suggests that these relationships between function and the place should not be fixed, as outlined by the functionalist line of reasoning. More functional flexibility was allowed as the function of the space can change both in a cyclical fashion as well as over the lifespan of the structure, e.g. a bedroom can be a living room during different times of the day.

From the user’s point of view, the individual selects such 'floating' functional spaces by himself rather than having it imposed by architectural planning. As functions and places are not connected, users have to search for comfortable spaces or even they have to create their favorite spaces by manipulating the existing architectural systems. Users have an active position in deciding what is happening in each space according to their requirements.

The proposed PA roof system allows the user to begin to manipulate the functions within these spaces.

Fig 4-5,2: Screenshots of the movie of the OkayamaLRT(K.Hotta and Shari)

This roof can be a filter for the natural environment thus varying the inner spatial character of the station square. A computer program, built with ‘Processing’, allowed the development of new geometrical decisions. By modeling a spring in the script, we proposed a new method for determining the geometry of the roof in response to the desired floor plan. It could realize antithetic concepts, simultaneously providing natural but sophisticated planning like a biological system.

Fig 4-5,3: Plan and Elevation of the OkayamaLRT(K.Hotta and Shari)

By modeling a spring in the script, we proposed a new method for determining the geometry of the roof in response to the desired floor plan. It could realize antithetic concepts, simultaneously providing natural but sophisticated planning like a biological system.

Fig4-5,4:Section of the Okayama LRT (Station Mode)

The roof canopy maintains the certain height to be able to provide the enough space for the LRT train.

Fig.4-5,5: Section of the Okayama LRT (Gathering Mode)

The roof canopy provides a big amount of space in the middle of the canopy. This affords a gathering activity underneath.

Fig.4-5,6:Section of the Okayama LRT (Scattering Mode)

The roof canopy provides several small spaces for different use. People utilize the space for individual activity.

Fig.4-5,7: Plan of the Okayama LRT

Fig 4-5,8: Cut Section of one leg of the Okayama LRT

4-6 . The Three Different Scales: Local- Regional -Global

Fig 4-6,1: Hierarchical System Diagram


In the proposed PA system, there is a clear hierarchy, but this hierarchy is not defined by the physical form. As shown in the diagram (fig4-6,1) the suggested fabric has a single hierarchy, as individual components are repeated as much as one wants. The above mentioned 'Local-Regional-Global' hierarchy is defined in the virtual design model. This hierarchy cannot be seen in the object’s static representation but can be recognized in the dynamic model (ie. digital simulation of the dynamic behavior). As Tschumi (Tschumi, 1994) or HiroshHara (Hara, 1987) declared "Architecture is not an object but an event". This is true not only with the proposed architectural system which is a dynamic-kinetic system but also in traditional architecture which was static, yet the surrounding environment and human action were constantly evolving. The dynamic quality of architecture is not only a property of the kinetic systems but includes traditional architecture.

In this diagram(fig 4-6,1), Architecture is briefly described as a system surrounded by various ecological systems.

Within PA, the architectural systems can be defined as a device which connects (mediates between) the environment and humans. When this system is placed in a dynamic-ecological world, various reciprocal actions can happen. The designer has the task of creating and planning how humans interact and engage with the environment. Depending on the place where this architectural system is placed the level and type of interaction would be different. For example positioned in the desert, it’s interaction with a sand storm would be more important in terms of performance, whereas if it is located on the surface of the moon, air pressure or solar radiation would be crucial performance criteria. These differences would be incorporated into the above definition of the system.

4-7 . The Compromise System Between Global-Local as Democracy-Socialism


Here the two principles of PA are introduced in an architectural system (framework).

The first principle prioritizes the ‘Local’ or human action i.e. the capacity of humans to manipulate the system locally. The second principle prioritizes the ‘Global’ where the global environment is generated through external environmental input from sensors. The 'Region' level of the hierarchy mediates between these two principles . In addition to negotiated interactions between the ‘Global’ and ‘Local’ levels there are also exceptional situations where one overrides the other. For example, if there is an accident and an ambulance has to go under this big roof, the architecture cannot wait for a mediated slow decision, but it needs a quicker response. Human input can override global 20 changes functioning as temporary administrator. The opposite happens, for example when there is an earthquake. The architectural system does not have time to respond to thousands of user’s movement and consequent inputs but rather an emergency shape is generated as a response to an unexpected input. In this case the system just ignores a single user’s input.

The relationship between the two principles depends heavily on the dynamic capacity of the system. A key aspect of this dynamic system is the time axis, allowing temporal inputs and outputs to be empowered and to interact with each other. The actual algorithm used to implement this time axis will be explained in chapter 6 and 7, but the importance of the time frame is now briefly explained via illustration. For this example, the system is responding to strong wind at 10am so it attempts to minimize its surface area and thus its air resistance to avoid collapse. However, after one hour at 11am, with the sun rising to its highest point, a second requirement is revealed, and the system has to provide maximum shade so it needs to reconfigure itself. The situation becomes more complex if two requirements are present for instance if the strong wind hasn’t stopped at 11 am. The building has to withstand the strong wind, but also has to provide maximum shade at the same time. The system has to make complex decisions where one environmental requirement is in conflict with others. This kind of conflict can become more extreme when human input is added.

For example, the shape is optimized for strong wind at 10 am, but also a large domed space is needed underneath the kinetic roof for a lunch party with 100 people. The wind resistance shape and the dome request from human input will clash. The architectural system has to mediate and make a decision each moment along the time axis. These types of situations happen all the time, because architecture- both system and building – exist simultaneously on the environmental and social ecological realms that change with 'time'. The process of meditation is the most important issue here.

There is discussion, the political comments might fit well here. “democracy” & “socialism” are models for manipulating conflicting demands. A building with human input brings this in.

fig 4-7,1: System Diagram in Ecology (Drawn by K.Hotta)

4-8 . The Relation between Physical Model and Constructive Model

Two realms coexist in the PA, one architecture is the physical built structure the other is the virtual system. The connection between the two worlds is not simple (according to Braitenberg, the method of connection could produce even intelligence) and the relationship between the two is inseparable. This research is an effort to separate the two parts. The physical system is composed of an MDF structure(the bones of the system), a membrane made of elastic fabric, shape memory alloy and the electrical wiring to the Arduino microcontroller (Banzi et al., 2005~). The virtual model visible on the computer screen, runs on the computer program 'Processing' using a physical engine called the 'constructive model'. It represents the virtual part of the system. Lipson H.(Lipson and Pollack, 2000) from Cornell university has similar vision for a dual model system. He called the two parts ‘robot’ and ‘awareness’. This hybrid model idea is not novel in engineering and complex system science where simulations are used. The model which will be made in the real world is called the 'physical model', and the one which is used for simulation is called the 'constructive model’.

In the traditional architectural field, the relation between a 'physical model' and a 'constructive model' corresponds to the relationship between a building and its blueprint. However, with the recent computational enhancement of the 'constructive model', the new field is now evolving from there. For example, the blueprint was no more than an 'idea' of the building. The computational model in PA, however, is not just an idea. Thanks to computer’s program and physical engines, it works as an autonomous system which works beyond human ideas and planning. Furthermore, if this robotic 'physical model' and 'constructive model' communicate as they do in PA, the definition of and boundary between real and virtual are going to change.

For example when the structure needs to change, the constructive model executes the calculations in the computer, and when the optimum answer is attained it will be output to the physical model reducing the interaction time to a moment when the output is passed to the physical model. Then the physical model- the big roof-implements the necessary deformation. This effectively saves time and energy as it does not execute physically until a good answer is calculated in the constructive model. This difference in nature between physical model and simulated constructive model can be used in many situations to allow complex scenarios to be implemented with a minimum of time and energy.



Fig 4-8 ,1 The correspondence between simulation (construction model) and robot (physical model)

Those pictures are from the video of the experiment, which connects a simulated model and a physical model. The connection between the two models is through Ardino, precisely speaking, Grasshopper.

Firefly - Arduino - Serial USB.

4-9 . Conclusion

The methodology for the meditation of conflict is the most important issue here. Even though the physical design may take an alternative approach to the proposed tensegrity design this conflict between one global input (environment input A) and other global input(environmental input B), or a global input(environment input) and a local input (input from a user), or even between one local input (input from person A) and other local input(input from person B) might still happen. The successful resolution of these conflicting inputs is a key aspect of PA.

The use of PA with its physical and virtual models and its three levels of hierarchy allow it to mediate the multiple inputs from users and environment to create a functionally flexible environment for its users.