You are on page 1of 148
ae ens eu asa Kinetic Architecture Modeling design and behavior Carlo Rousseeuw Thesis voorgedragen tot het behalen van de graad van Master in de ingenieurswetenschappen: Architectuur Bouwtechnische Optie Promotor: Prof. Vande Moere Assessor: Prof. Van Broeck Begeleiders: Prof. Boeykens Academiejaar 2010 - 2011 tou LEUVEN Kinetic Architecture Modeling design and behavior Carlo Rousseeuw Thesis voorgedragen tot het behalen van de graad van Master in de ingenieurswetenschappen: Architectuur Bouwtechnische Optie Promotor: Prof. Vande Moere Assessor: Prof. Van Broeck Begeleiders: Prof. Boeykens Academiejaar 2010 - 2011 © Copyright by KULowan ‘Zonder voorafgaande schrifelifke toestemming van rowed de promator(en) als de anteur(s) is overemen, kapiéren, gebruiken of realiseren van dere ultgave of gedeelten ervan verboden. Voor aanvragen tot of informatie ivan. het ovememen en/of gebruik en/of realisatie van gededten uit deze publicatie, wend u tot de KU Lauven, Faculteit Ingeniaurswotenschappen ~ Kastoolpark Arenberg 1, B-3001 Hovertoo (Belgid), Telefvon +32-16-32 13 50 & Fax, 432-1632 1988, Voorafgaande schriftdijke toestemming van de promotor{en) is eveneens vereist voor het amnwenden van de in dit afstudeerwork beschroven (originols) methoden, produces, schakelingen en programma’s voor industrical of commercieel nut en voor de inzending van deze publicatie ter deelname aan wetenschappelifke prifen of wedsijden. © Copyright by K.U.Lewen ‘Without written permission of the prometors and the authors itis forbidden to reproduce or adapt in any form or by ‘any means any pait of this publication, Requests for obtaining the right to reproduce or utilize parts of this publication should be addressed to KU.Leuven, Faculty of Engineering ~ Kasteelpark Areberg 1, B-3001 Heverlee (Belgi@). Telefoon +32-16-32 13 50 & Fax. 432-1632 19 88, ‘A written permission of the promotor is also required to use the methods, products, schematics and programs described in this work for industrial or commercial use, and for submitting this publicationin scientific contests. Preface The idea for this research came to me when I was first introduced to the notion of Interactive Architecture: a new kind of architecture, previously unknown to me, but which grabbed my attention. Various projects in the current landscape intrigued me and caused me to dive deeper into this new and upcoming field. ‘The main focus for me was to gain insights, information and to be able to pass this on to others too. This focus sometimes led to quantity over quality, which I am not ashamed to say. It is the intent of this research to draw the bigger picture and introduce this new skillset to all who want to know. It was never the intention to lure people into a narrow mindset, only to introduce them to this new sandbox where they can play themselves. Another difficulty for me personally was the absence of previous local research and the possible prejudices that others might have towards this research. This research did not take form as fast as I would have liked and was not easy to complete but the supervising professors were always positive and guiding, which allowed me to keep on going. A thanks goes out to people who inspired me, family and friends who supported me and helping hands that guided this research into the right direction. Carlo Rousseeuw Table of contents Preface . . ili Table of content: Abstract... List of Figures. List of Tables Nomenclature. Chapter 1: Introduction. Interactive Architecture... Physical Counterpart. Intelligence Counterpart... Practical Knowledge.. Motivation... Objectives... Significance. Chapter 2: Background... 2.1 Barlier Works. AE 2.2 Literature study... 12 2.2.1 Timeline... a 2.2.2 Project Location...... . 16 2.2.3 Project Mechanism 16 2.2.4 Project Typologies 218 2.2.5 Project Application Kinetics. ie 18 2.2.6 Project Timespan., 9 2.2.7 Project Structural behavior. -20 2.2.8 Project Intelligence... a 2.2.9 Project Sensor values... a 21 2.2.10 Web Based Literature... Chapter 3: Methodology & Results... 3.0 Methodology ... 23 3.0.1 The Simulation Software, Grasshopper 23 3.0.2 The microcontroller, Arduino... 26 3.1 Simulating design 3.1.1 Simulations, Case Studies, 3.1.2 Numerical Validation... 3.1.2.1 Default Scenario Results 3.1.2.2 Actuated Scenario Results... 3.2 Simulating behavior 3.2.1 The Simulation Software. a 44 3.2.2 Linking Data Methods . 45, 3.23 Simulations. r 3 . 46 3.2.4 Emotive behavior. 3.3 Design Issues..... 3.3.1 Joints... 33.2.1 2D Joint.. 56 33.2.2 3D Joint. 58 3.21 Members nnnnnnnninnmnnnnninnninnnnnnnunnininmnnnnnnnsnnnnnnncnnse OL 33.2 Cladding. se 5 62 33.3 Actuator design 67 33.5 Prototype Design ma 33.5.1 Introduction... 78 Flexible Skin... Biomimicry... 3.3.5.2 Results Design. Polyp Truss Cladding. Behavior .... Chapter 4: Evaluation & Discussion... 4.1 Evaluation & Discussion Simulating Design. 4.1.1 Sinmulation Case Studies... 90 4.1.2 Numerical validation... 92 4.2 Simulating behavior 94 4.3 Design Issues... 96 43.1 Joints. 43.2.1 2D Joint. 43.4 Cladding... 43.5 Actuator design... 43.6 Prototype Design Design & Polyp : ps 102 Truss & Cladding, 104 Behavior... . z 106 (CHAptee SY COMELLMOT sasnasmiaenicanismtaieitit J. canes ED 5.1 Simulating Design.. 108 5.2 Simulating Behavior... 110 5.3 Practical Issues aa 53.2 Cladding... 12 5.3.3 Actuator Design, 113 53.4 Prototype. : 1s Appendices... 18 Appendix A... spree summa TG Appendix B... Bibliography. Fiche Masterproef. vi Abstract ‘The way we use and experience objects in our daily lives is constantly being improved with increasing user-interactivity. From our cars which are filled with sensors to enrich our driving experience to the automated shading of our windows which disappears when we need to get out of bed in the morning. Architecture today on the other hand is static, its structural form does not interact with its users or its changing environmental factors. Instead of shielding the inhabitants from these factors, these factors can be responded to and interacted with to change the inhabitant's perception of this new space, Interactive Architecture, In the design of interactive architecture with structural kinetic changes, Kinetic Architecture, the simulation of a structure and its behavior plays a valuable role in its overall design and production. Being able to connect a wide range of sensor data with this design-software we cross the bridge necessary for completely simulating interactive architecture, which in turn has an effect on the final design. Recent developments and community efforts in plugins for clrawing software like Grasshopper for Rhinoceros have given us these abilities. Every project is unique by its own context and usage and therefore unique by its means to interact. This thesis simulates different existing structures in the current landscapes and tests the scope of current simulation packages and their use to designers with regards to Kinetic Architecture. Also the intelligence which controls this Kinetic Architecture and the different kinds of data streams are addressed together in the context of the simulation software. Besides the research in terms of simulation, this thesis also discusses practical issues of Kinetic Structures ina general way before building a working prototype. ‘This research will act as a catalyst to provide architects with the necessary skillset to develop and design interactive architecture but also to provide a mutual goal for other disciplines like robotics and material engineers to form and research different end products with enhanced user interactivity which could be used in this new breed of Interactive Architecture. vii List of Figures Figure 1: Burble at the Singapore Biennale (Haque 2006) sass Figure 2: Interactive Wall (Festo Corporate 2009). oo ene Figure 3: Tesselate (Lab[au] 2010) Figure 4: Fabric Dome (Hoberman 1997) Figure 5: Pneumatic Muscle. Figure 6: Visualizing Wifi Strength, Immaterials (Arnall et al. 2011) conch Figure 7: Pixelskin 2.0 (Orangevoid nd.). Figure 8; Flock of birds (National Geographic n.d... Figure 9: High Tech Teamwork of swarm robots (National Geographic n.d.) Figure 10: Ball Joint, 3D-Print.. Figure 11: Stiff Cladding Connection. Figure 12: Ersting Warehouse Gate, Santiago Calatrava, 1983 (Tzonis & Lefaivre 1997) cu-11 Figure 13: Scale Model, Santiago Calatrava (Tzonis & Lefaivre 1997) Figure 14: Interactive Architecture Categorization + Dissertation. Figure 15: Time versus Book Contents.. Figure 16: Location versus Book. Figure 17: Mechanism versus Book eT Figure 1 Typology Categorization... - cna TB Figure 19: Application Categorization een snnennee AD Figure 20: Load Bearing categorization 20 Figure 21: Intelligence Categorizatio 21 Figure 22: Sensor Values Categorization Figure 23: Interactive Architecture dot org, Wordle (Glynn 2005)... viii Figure 24: Actuators, Spring Implementation. Figure 25: Truss member, Spring Implementation . Figure 26: Kangaroo Components in Grasshoppei Figure 27; Grasshopper Environment. sna sven 25 Figure 28 Testing the Arduino, LED-bar. Figure 29: Scale model Type 3 (D'Estree Sterk 2003). Figure 30: Simulation Actuated Tensegrity Type 1. Figure 31: SimulationActuated Tensegrity Type 2 Figure 32: Simulation Actuated Tensegrity Type 3 ee cone BO Figure 33: WhoWhatWhendir, Flexible Tower (Kilian et al. 2006). Figure 34: Muscle Tower Il, Hyberbody (Qosterhuis 2000), Figure 35: Simulation Flexible Tower... Figure 36: TESSEL (Lab[au] 2010).. Figure 37: Robotic Membrane (Orangevoid n.d. Figure 38: Simul ion Robotic Membrane Figure 3% Excerpt from dissertation, Responsive Actuated Truss (Merali & Long 2009) -.......38 Figure 40: Simulation Actuated Responsive Trus: Figure 41: Simulation Kinetic Circk Figure 42: Expanding Geodesic Dome (Hoberman 1997) Figure 43: Strata Module(Adaptive Building Initiative 2006) ss 88 Figure 44: Simulation of the Expanding Geodesic Dome. Figure 45: Adding a color scale in Grasshopper. Figure 46: Simulation Grashopper, SKN, Not Actuated, Deformation Scale 1:1. Figure 47: Simulation ANSYS, 5KN, Not Actuated, Deformation Scale 1:1 Figure 48: Simulation Grasshopper, OKN, Actuated, Deformation Scale 1:1... Figure 49: Simulation ANSYS, OKN, Actuated, Deformation Scale 1:1. Figure 50: Arduino Send /Receive implementation in Grasshopper Figure 51: Manipulating and visualizing Sensor Data Grasshopper. Figure 52: Optimization, Kinetic Arch towards point, Figure 53: IR sensor connected to the Arduino. Figure 54: Manipulating & visualizing Sensor Data in Grasshopper. Figure 55: Push sensitive sensor connected to the Arduino. ses sone AB Figure 56: Implementing Sensor Dat, Figure 57: Optimization, Kinetic Arch minimizing stresses. Figure 58: Pachube receive implementation in Grasshopper. Figure 59; Pachube, Live sensor streaming and stream information Figure 60: Fiducial implementation in Grasshopper on 52 Figure 61: Sensors. Figure 62: Grasshopper/Kinect Sensor (Andy Payne et al. 2010) Figure 63: Dune 4.0 Maastunnel(Roosegaarde 2011), Figure 64: Laser-cutting and engraving a sheet of MDF. Figure 65: 2D Turning Joint, with eccentricit Figure 66: 2D Turning Joint, without eccentri Figure 67: 2D Joint Unstable... Figure 68: Intersecting 2D joints, Snap-Fi Figure 69: Parametrical Model Universal Joint 2... Figure 70: 3D-print Universal Joint 2 Figure 71: Cardboard Space frame; Ring Pass, Delft (Octatube 2010)... pen Figure 72: Parametrical Ball Join Figure 73: Section Parametrical Ball Joint Figure74: 3D Printed Ball Joint.. Figure 75: 3D printed Ball Joint, Section. Figure 76: Truss Member... Figure 77: Truss Member Figure 78: Textile membrane, Unstretched Figure 79: Textile membrane, Stretched. Figure 80: Kinetic Box, Default Scenario Figure 81: Kinetic Box, Scenario 1. Figure 82: Kinetic Box, Scenario 2... Figure 83: Expansion joint Wooden cladding o stn Figure 84: Ball joint (Kejia Industry n.d). Figure 85: Living Glass (The Living n.d.) Figure 86: Xeromax Envelope (Future Cities Lab 2010). Figure 67: Rectangle, Diagonal Actuation. Figure 8% Muscle Wire Actuation. ee ees OB Figure 89: Screw Linear Actuatoi Figure 90: Crankshaft Linear Actuator Figure 91: Gear-Pinion Linear Actuator Figure 92: Muscle Project (Festo Corporate 2009)... Figure 93: McKibben Principle (Daerden & Lefeber na.) Figure 94: Test Setup. Figure95: McKibben air muscle setup... a7 Figure 96: 3/2 Air Valve Festo.. Figure 97: Arduino Controller. Figure 98: Elongation/Original Length [%].. Figure 99: Underwater Polyps (National Geographic nd.) cst cena Figure 100: Axonometric View Prototype. Figure 101; Prototype, 3D Sketch. Figure 102: Scotch Yoke Mechanism (Mechanisms 101 n.d.) Figure 103: Polyp Actuation, 3D Sketch Figure 104: Polyp Actuator. Figure 105: Polyp, Upper node connection... Figure 106: 2D Simulation, Design 1. Figure 107: 3D simulation, Design 1. xi Figure 108; Figure 109: Figure 110: Figure 111: Figure 112: Figure 113: Figure 114: Figure 115: Figure 116: Figure 117: Figure 118: Figure 119: Figuur 120: Figure 121: Figure 122: Figure 123: Figure 124: Figure 125: Figure 126: Figure 127: Figure 128: Figure 129: Figure 130: Figure 131: Figure 132: Figure 133: Figure 134: Figure 135: 2D simulation, Design 2.. 3D simulation, Design 2. Truss assembly. Lasercut Vacuum Forming Molds. o stn ‘Vacuum Former with mold Vacuum forming result with high mold.. Cladding, Upper View.. Cladding, Lower View. Built-In Opto Resistor. Built-In Piezo Element Built-in IR Sensor... Prototype Side-View. Prototype Side- View. Prototype Perspective View. se 89 Prototype Upper View. Simulation Flexible Tower... Simulation of the Expanding Geodesic Dome. 91 Simulation Grasshopper, OKN, Actuated, Deformation Scale 1:1 .. ‘Simulation ANSYS, OKN, Actuated, Deformation Scale 1: Optimizatin, Kinetic arch moving towards points. ss soso 2D Turning Joint, without eccentricity. 2D unstable joint. 3D-print Universal Joint 2... 3D printed Ball Joint, Section Truss Member. Kinetic Box, Scenario 2 Expansion joint Wooden claddit 99 McKibben air muscle setup 100 xi Figure 136: Elongation/Original Length [%] 101 Figure 137: Sketches Polyp, Design 1 102 Figure 138: Polyp Actuator... 103 Figure 139: Prototype Perspective View... . ss seven 103 Figure 140: Cladding Hinge, Truss member connection. 104 Figure 141: Vacuum Forming results with high mold. 105 Figure 142: Arduino powered by USB and 9Y battery. 107 Figure 143: Simulation Grasshopper, OkN, Actuated, Deformation Scale 1:1 109 Figure 144: Simulation ANSYS, ON, Actuated, Deformation Scale 1-1... 109 Figure 145: Central Intelligence versus Swarm intelligence, Simulation 1. 110 Figure 146: 3D printed Multiple member Ball joint an Figure 147: Kinetic Box, Scenario 2. 112 Figure 148: Expansion joint Wooden cladding. 12 Figure 149: Linear actuators, Shrink Rate. 113 Figure 150: Air muscle, Elongation rate [%] 114 Figure 151: Prototype, Perspective view... 115 Figure 152: Prototype Upper View. 116 Figure 153: Prototype Side view 116 Figure 154: 3D Truss Simulation. 117 vi List of Tables Table 1: Legend Time versus Book Contents 15 Table 2: Legend Books... cod ‘Table3: Error Margin %, Default SKN Table 4: Error margin %, Default 10 KN Table 5:Error Margin %, ON Table 6:Error Margin %, SEN nen AB Table7: Intelligent- versus Swarm Behavior. Table 8: Comparison Different iteration processes... Tabel 9: Linear Actuators, Shrink Rate. 69 Table 10: Muscle Elongation [%]... Table 11: Polyps Behavioral Scheme, Table 15: Error Margin %, Default SkN Table 16:Error Margin %, SKN ‘Tabel 17: Optimization results versus swarm implementation, Simulation 1 Tabel 18: Optimization results, Simulation 2. Tabel 19: Shrink Rates of different actuators. 100 Table 20: Muscle Elongation [%] 101 Table 21: Error Margin % Default SKN Table 22:Error Margin %, SkN sn ss OO xiv Nomenclature IA Interactive Architecture KA Kinetic Architecture FE Finite Elements KDG Kinetic Design Group Force ‘Young's Elasticity modulus Displacement inx Direction Displacement iny Direction (Elasticity) Stiffness H_ Grasshopper FF Firefly KG Kangaroo ar< mn Chapter 1: Introduction Interactive Architecture Architecture today is monotone and static. Imagine architecture however to be alive, to be able to partake in a discussion with its inhabitants or the environment in which it has been placed. These inhabitants and environmental factors, like sunlight and ‘wind, are dynamic: they are not static or monotone and they deserve to be acknowledged and interacted with. When reacting and interacting with these factors, architecture changes the inhabitant’s perception of space and lets them live in symbiosis with architecture rather than only inhabiting architecture. This new kind of architecture has to be dynamic, responsive and interactive. “One way to begin exploring the dynamics is through rethinking architecture beyond conventional static and single-function spatial design.” (Fox & Kemp 2009) Let us for example think of a pavilion, a pavilion that can change its shell form to automatically use the best form for minimizing the displacements or stresses in its structure. A pavilion that can brace itself for the incoming impact of an earthquake or a pavilion that breathes, ventilates, and catches renewable resources for its inhabitants and their current activity. Buckminster Fuller even coined this as “Ephemeralization’(Fox & Kemp 2009), being able to build a stronger form with minimal material using active measures, similar to the human body where a fixed amount of muscles and bones can provide various stances for various positions and actions. ‘Perhaps the most applicable research to draw upon in designing intelligent systems lies in an area of study called active control research, which focuses on the use of active control to modify the structural behavior in a building” (Fox & Kemp 2009) Usman Haque, founder of Haque Design + Research, specializes in the design and research of these interactive architecture systems. Together with his entourage he designed and constructed various interactive installations en various scales, including Burble, a massive installation consisting of multiple balloons embedded with lights and infrared receivers being able to receive signals from a simple television remote used by the people who came to see the installation. With regards to IA, Usman Haque states: “Such system must utilize a definition of interaction as circular, or they are merely “reacting” and not “interacting’. As people interact with architecture they should not be thought of as “users” but instead as “participants’.” (Fox & Kemp 2009) Figure 1: Burble at the Singapore Biennale (Haque 2006) Research involving IA can be approached through different viewpoints, such as the sociological aspects, change in social behavior and the perception of space but also in amore practical way of building, modeling and simulating IA design to gain practical knowledge. ‘The Burble by Usman Haque is an example of researching sociological aspects and social behavior through [A. His Burble and Sky Ear projects made a variety of people leave their homes and point their television remote control in the sky to interact with this new kind of architecture, This research however addresses the more practical issues such as simulating TA, building joints and building a prototype. This is a common research method where building and experimenting on a small scale gains practical insight but is not introduced asa social experiment. For example the interactive wall created by the Festo Corporation or the Tesselate project by lab[aul] are both installations ina controlled environment with minimal documentation involving their effect on social interaction. “The physical architecture can be used to include or exclude people from one another, to facilitate, dissipate, or focus crowds of people. In this way, in the realm of the physical world, interactive public spaces can have a profound space effect on social interactions."(Fox & Kemp 2009) Figure 2: Interactive Wall [Festo Corporate 2009) Figure 3; Tesselate (Labfau] 2010) To guide the reader in this research, IA is cut up into two counterparts: “The current landscape of interactive landscape or interactive space is built upon the convergence of embedded computation (intelligence) and a physical counterpart{kinetics).” (Fox & Kemp 2009) Physical Counterpart ‘The physical counterpart nowadays is that of kinetics. We therefore utilize the name Kinetic Architecture (KA) to depict IA that utilizes kinetics as a physical counterpart, Kinetic structures can be categorized by their “ways” and “means’: (Fox & Kemp 2009) The “ways” ofa kinetic structure can be the various methods of spatial change in both size and shape like folding, expanding or sliding, The Hoberman dome for example can expand its structure using a scissor mechanism and the Ernsting, Warehouse gate by Santiago Calatrava opens with a linkage mechanism, ‘The “means” of a kinetic structure is formulated as the driving mechanism(s) behind the “ways” of the kinetic structure. These are also named actuators in the scope of this thesis. Actuators can be mechanical, pneumatically, naturally, chemically or magnetically driven. The pneumatic muscle researched in this thesis is an example of a pneumatically driven actuator. Both categories are necessary when talking about the kinetic counterpart. This thesis simulates different kinetic mechanisms, ways, during a case-study, to find the scope of present tools but also fabricates linear actuators to gain insight into the different means of kinetic structures. Figure 4: Fabric Dome (Hoberman 1997), Figure 5: Pneumatic Muscle + Linkage Mechanism: “A series of rigid links connected with joints to form a closed chain.” (Wikipedia 2011) Intelligence Counterpart ‘The second counterpart of LA is intelligence under the form of embedded computation. The intelligence of a structure is the central nervous system that operates our KA. “A kinetic environment without the computation is like a body without @ brain: incapable of moving."(Fox & Kemp 2009) Embedded computation is the term that depicts the numerous microcontrollers oF electronic components that are embedded, implemented, in today’s electronic devices and are able to do only a couple of dedicated functions. In IA these are microcontrollers that enable the building to receive, process and act upon incoming data. ‘The intelligence of IA is defined as the programming, that under the influence of data controls the actuation of the kinetic counterpart. These streams of data can vary from pragmatic data to humanistic data. Under pragmatic data fall values such as the daylight level, wind speed ora wide variety of data that is tangible in numbers. Projects such as “Painting Wifi’, which measures WIFI strength in the field, or "Pixelskin” by Orangevoid, which is an interactive shading reacting to daylight levels, are examples of using pragmatic data. Human behavior and emotions however cannot be measured in numbers. These are called humanistic data, where pragmatic sensors are already available ona wide scale, Sensors which can accurately sense human movement or emotion however are not yet available on a large scale. This research will address both data types and how. to implement them in the simulation of KA, Figure 6: Visualizing Wifi Strength, Immaterials (Arnall et al. 2011) Figure 7: Pixelskin 2.0 [Orangevoid n.d.) In this research there is a difference between swarm intelligence and central intelligence. Both of these are addressed here and compared to each other using a basic numerical comparison. ‘The term swarm intelligence is used in this thesis when a structure is comprised out of different cells. Cells which have their own sensors but can also only actuate their own actuators, meaning their sensory environment is equal to their actuating environment. Projects such as the “Pixelskin” solar shading are comprised of these cells which act on the specific value given to them by their ewn sensor. Swarm. intelligence is also directly linked to the behavior science of swarms like bees, ants and flocks of birds. Central intelligence is the opposite of swarm intelligence and works much like the ‘human brain. When a hair particle senses vibration our brain is alerted which allows us to use our other muscles to act. Hence the sensory environment is not the same as the actuating environment. Physically this means all of the sensor data from the cells or even off-site data can be used to actuate the entire structure using a coordinated act. A practical example can bea structure where minimal stresses are minimalized further. This structure not only measures the force applied to a node ‘but collects this data to analyze and actuate accordingly with its entire structure, “The benefit ofan active sustainable system is that it can intelligently combine the resources of a number of systems so that when working together, the individual elements or systems achieve more than the sum of their parts."(Fox & Kemp 2009) Figure Practical Knowledge “Architects need not become specialists in this area but should clearly understand the potentials of how this new area of design could impact and/or enhance the projects they are designing.” (Fox & Kemp 2009) ‘This research will use a bottom-down approach where an ideal kinetic structure is broken down into various parts such as cladding, connecting members, actuators, etc. and researched as such. At first ina general way before using this knowledge to physically built and assemble a prototype of a conceptual design. Techniques used in this thesis are 3D printing, laser-cutting, computer numerical control (CNC) milling and vacuum forming. It is the writer's opinion that these tools ‘will keep improving during the upcoming years, so that they will eventually be used in the production of building components suitable for carrying the calculated loads. This however is not the case at the moment of this research. Figure 10: Ball Joint, 30-Print Figure 11: Stiff Cladding Connection Motivation “The use of tools with real-time feedback for prototyping behaviors can greatly influence the overall process of design and have a profound effect upon the final end product.” (Fox & Kemp 2009) ‘Trying to visualize a KA design, the designer needs the tools to simulate, calculate and animate the movement of its structure. The 2D or 3D drawing software commonly used are not yet ready to generate this for him. They deliver, just like our current perception of architecture, static drawings. Design tools for kinetic systems do exist in specialized software packages (Fotiadow 2007), designed for mechanical engineers and animators in the media or entertainment sector. They utilize “skeleton” tools to simulate their design based on stiff members and actuators that bring movement in the system. But even when software that can handle skeleton animation is available we lack easy-to-use real- time calculation of stresses and displacements in the structure during its animation. ‘What we gain in simulation of the kinetics we lose in structural insight of the global structure. Does the kinetic structure for example decrease its internal forces by moving in that specific stance or does it increase them ? Even when talking about smaller scale kinetic structures without simulation tools a thorough knowledge and structural insight is needed to notice problematic points in the entire structure. “The integration of computational tools, such as 3D modeling software for real-time simulation and actual physical testing into the process of designing also allows designers to confront and anticipate many of the issues that occur when building at full scale.” (Fox & Kemp 2009) Designing and constructing IA also involves a specific skillset in microcontrollers, sensors and actuators to make the design come to life. This skillset is not part of the basic curriculum of an architect. But without a basic notion of these skills, the architect cannot construct KA. These simulation tools and specific skills do not exist because KA exists but because they will complement each other to produce better (and parametrical) KA when introduced in an early phase of design. “When the tools evolve with the design, the heuristics are facilitated by the tools, and not necessarily limited by their parameters. The design processes associated with interactive systems design are constantly evolving and are fostered by the consequent development of new tools.”(Fox & Kemp 2009) Objectives ‘The aim of this research consists of researching today’s simulation tools to simulate the physical counterpart, researching the various sensors and actuators while also linking them to our simulation and gaining practical knowledge in this field by addressing different general solutions and constructing a working prototype. When researching the chosen simulation software, its working range is found by implementing different case-studies and documenting where the simulation software experiences problems. The simulation software is also validated by comparing it numerically to a common FE software package. ‘Various sensors and actuators are researched to gain practical knowledge on how to operate and implement them but are also linked to the simulation software to test its range again in this aspect. Design issues when physically building KA are addressed using a bottom down approach involving different general aspects like actuation, cladding and nodes, but also by building a prototype. Significance “Architects are eager to embrace technology that can increase optimization through adaptation with respect both to the environment and user needs, yet they must learn to recognize the interdisciplinary needs that such technologies have ensnared.” (Fox & Kemp 2009) ‘This research is going to act as a catalyst to inspire new designs or research in the multidisciplinary field of lA. Itwill provide the necessary skills for architects together with means to begin simulating and designing KA. But will also provide the concept of 1A in the form of Ephmeralization to structural engineers who can produce further calculations and can thus develop lighter and stronger structures using active control measurements. Engineers studying the indoor building climate can develop strategies that form the basis for architects and their designs for interactive fagades?, 2 Interactive Facades: Interactive components embedded in a larger structure (facade) interacting with building physics behavior and appearance. These include sunlight, temperature, ventilation and also appearance in all kind of forms. Mechanical, chemical and electrical engineers can also find inspiration for new mechanisms, smart materials or electronics to upgrade current versions of IA, like actuators that are powered by the sun or textile membranes that can change color like a chameleon. This common goal is the complementary inspiration for both parties for designing and discovering new applications. Even social sciences can observe prototypes of [A ina social context to see what kind of impact it has on human and social behavior, which in return forms the basis for architectural critique and the branching out of [A in different directions, Not only can they observe direct confrontation between the built environment and their users but ‘they can also interpret the vast stream of pragmatic and humanistic data that will soon be available from every corner of this new sentient city, comprised of Interactive Architecture. 10 Chapter 2: Background 2.1 Earlier Works ‘The concept of Kinetic Architecture is a young concept. Less thana handful of architects have incorporated KA in their own oeuvre before 1990. The most prominent figure of them is Santiago Calatrava (Calatrava nd.). He has designed a couple of kinetic projects which contain linkage systems with a set of non-bearing beams where the Kinetics could still be calculated by hand. For example his Ernsting Warehouse Gate and the design for the Milwaukee Art Museum. It is the writer's opinion that this is directly linked to the absence of more complex simulation tools at that time. Besides the independent work of Calatrava, a research group at the Massachusetts Institute of Technology was commissioned around 1995. The Kinetic Design Group (Fox 1995a) categorized different interactive structures that had been conceived up to then by their mechanism(s) and is thus a great way of capturing most of the earlier works. Figure 13: Scale Model, Santiago Calatrava (Tzonis & Lefaivre 1987), un 2.2 Literature study From the earlier works up to 1990 there was still no large scale interest in interactive architecture. ‘The background works of William J. Mitchell (Mitchell n.d.) were the only ones published until the 21" century. These paved the way for IA by introducing the bigger picture regarding the changing social roles thanks to upcoming phenomena at that time such as the internet and the digital realm. Around 2000 different architects and architectural firms began to experiment with and document IA: Anew generation who now had the means and skills to do so. The magazine iA for example was published by Kas Oosterhuis who is still the current director of the Hyperbody workgroup (Oosterhuis 2000) at the Technical University of Delft. These magazines therefore consist of many student projects and prototypes together with parts of theoretical theory. Around 2009 an elaborate work of Michael Fox and Miles Kemp titled “Interactive Architecture” was published. Michael Fox, founder of the Kinetic Design Group at MIT can be considered as the person with the largest historical experience in the field and makes the book “Interactive Architecture” (Fox & Kemp 2009) a prominent piece in this literature study as well as the entire field of 1A at the moment of writing. 2 ‘The following books regarding IA were taken into account while conducting this literature study: 1 W.J. Mitchell, City of Bits, Space, Place and the Infobahn, MIT Press, 1996, pp.24 (Mitchell 1996) W.J. Mitchell, e-topia, Urban Life ~ But not as we know it, MIT Press, 2000, pp.184 (Mitchell 2000) W.J. Mitchell, ME++, The Cyborg Self and the Networked City, MIT Press 2004, pp.259 (Mitchell 2004) K. Oosterhuis; X.Xia, iA n°l -Interactive Architecture, Jap Sam Books, 2007, pp.96 (Oosterhuis & Xia 2007) K. Oosterhuis; X.Xia, iA n°2 -Interactive Architecture, Jap Sam Books, 2008, pp-112 (Oosterhuis & Xia 2008) M-Fox; M.Kemp, Interactive Architecture, Princeton Architectural Press, 2009, PP.225 (Fox & Kemp 2009) ‘The following magazines have been read but in the opinion of the writer they do not fit inside the category of KA since most of the projects in these magazines have no Kinetic counterpart in their interactive design. Nonetheless these projects are part of ‘the current interactive project landscape and form a source of inspiration and Imowledge. 7. K. Oosterhuis; XXia, iA n°3 - Emotive Styling, Jap Sam Books, 2010, pp.128 (Oosterhuis & Xia 2010) 8 Lucy Bullivant, 4Dspace - Interactive Architecture, Academy Press, 2005, pp.128 (Bullivant 2005) 9, Lucy Bullivant, 4Dsocial - Interactive Design En nments, Academy Press, 2007, pp.127 (Bullivant 2007) 13 Figure 14: Interactive Architecture Categorization + Dissertation ‘The entire written 1A landscape has been split up into different categories by the writer. Abstracting their quantity, each of these categories contribute to the domain of 1A. Each book has been mapped on its categorization in relationship to the other categories, most fitting for its content in the opinion of the writer. The contents of this dissertation have also been categorized to visualize the content in relationship to the existing landscape. ‘This study also states that simulation tools are still not yet widely documented or actively used. Except for some projects in the magazine iA, KA and IA in general are not simulated. in current projects. The majority of the project landscape arise from a practical knowledge and heuristic production methods of building prototypes and scale models. “The majority of the projects in the above books have been documented? and categorized in the following sections. By doing this, the research compares the books based on their vision on the current IA landscape/projects, > Appendix A: Project Landscape Raw Data 14 2.2.1 Timeline ‘To visualize the changing roles of these categories inside the IA landscape each book ‘was roughly divided by the writer into percentages and plotted ona timeline. The stim of the categories at any one time is therefore 100% and thus will not take into account the amount of literature at any one time. It is obvious to see that after the trilogy of W. J. Mitchell the overall background disappears and the IA theory takes over together with a boom of prototypes and simulations in the project landscape. Another remark is the small rise in simulation projects rather than real projects and prototypes, meaning that the simulation packages are relatively new and have made their way into some specific works. = : i Fe 35 he rsa intr : Background Prototypes 1A Theory Real Life Projects ‘Simulations Table J: Legend Time versus Book Contents 15 2.2.2 Project Location Figure 16: Location versus Book TA n°1(Oosterhuis & Xia 2007) Interactive Architecture(Fox & Kemp 2009) IA n°2 (Oasterhuis & Xia 2008) Kinetic Design Group Matrix(Fox 1998a) Table 2: Legend Books Categorizing the examples found in literature per book and geographic location it is noticeable that the magazine iA mainly depicts national work. The book Interactive Architecture has a wider scope, not only showing national projects but conceptual projects in Western Europe as well. ‘The earlier works, as categorized by the KDG, originated from the home countries of prominent figures like Santiago Calatrava and Otto Frei. Although the size of their respective pie charts tells us that newer projects are more present in quantity. 2.2.3 Project Mechanism Based on a categorization in mechanisms made by the KDG at MIT (Fox 1995b), all of, ‘the aforementioned projects are put into groups while abstracting their absolute value but showing their respective book ina stacked column. Keeping in mind the same color code is used for the respective books as the geographic location. Itis shown that the KDG has made and documented a variety of mechanisms. Newer projects tend to avoid the earlier linkage systems or Nurnberg scissors mechanisms ‘which are interesting but only have a limited scope of usability and a predetermined end- and start point, These mechanisms are the same categories as the aforementioned “means” of the Kinetic structures. The “ways” have not been documented since the amount of categories was too high, only proving the enormous amounts of actuators available, 16 me ane " Prsanane" Petgwnat "Terie Tass Setagenst— 2ocmonny SetTad ner offers g § g § g 3 g a 3 5 SLA XX 9 eK SL Figure 17: Mechanism versus Book 2.2.4 Project Typologies In the book Interactive Architecture (Fox & Kemap 2009) the authors elaborate the Kinetical counterpart in different typologies. The following categorization puts the majority of projects found in the literature into one of three types: Deployable, Dynamic or Embedded. Deployable structures are conceived as structures with a small setup time. These include military, non military and crisis relief tents: very kinetic systems during setup but the end position of the buildings is still static. Dynamic systems are systems that are interactive and have a physical counterpart, ‘but who are still only a piece of a bigger system. These include solar shading, ventilation systems, etc. Embedded systems are like dynamic systems but form an integral part of the structure and space. It can be seen that the newer projects tend towards dynamic and embedded structures instead of the earlier deployable structures. Destorae Figure 18: Typology Categorization 2.2.5 Project Application Kinetics Another categorization also coined by the authors of Interactive Architecture(Fox & Kemp 2009), is that of the application of kinetic systems. What is a project used for ? Contextual Adaptability includes structures reacting to our changing environment, solar position, wind velocity on the scale of our building. ‘Multi-function design functions on an interior scale: Positioning of movable dividing walls or multi-function furniture. Spatial optimization seeks to optimize the structure to its usage. Whether a basketball game is being played in a space or our offices are placed there, the space will optimize acoustics and lighting to our changing needs. 18 5 a os NN Figure 19: Application Categorization 2.2.6 Project Timespan When dividing each project by the speed of their movement, there was no difference to be seen. Every project’s timespan was direct, implying minutes. This means there has been very few or even no projects or designers prototyping this bigger picture in their work, The concept of time is not yet widely documented and taken into account in the current IA project landscape. Itis for this reason that the writer explicitly takes into account three levels of interactivity in time: Direct, Medium and Long timespans. Direct interactivity involves stimuli like daylight levels, sound levels and the presence of objects or people. A building can for example change its skin, thickening it where there is a lot of noise to maintain an optimal space for the occupant to read its book. Or it can even change the shape of windows so that the space is equally lit and no glare or overheating will occur. Stimuli over weeks can be phenomena such as changing wind patterns or geological conditions like rising water levels, These invoke a change in the applied loads on the structure and will therefore change the structure's optimal form versus internal stresses and displacements. Over months more humanistic values can change. Like the average amount of visitors ina certain restaurant so it can adjust his spatial configuration. Or playgrounds can change their size according to the amount of children between the age of 3-15 which are staying in the vicinity. Itis the writer’s opinion that different levels of interactivity cannot always be incorporated in each other on a technical scale. For example the inhabitant's need for daylight and ventilation cannot coincide with optimal structural behavior. “Perhaps the mast important goal of an interactive system today should be to act as a moderator responding to change between human needs and external environmental conditions.”(Fox & Kemp 2009) 19 2.2.7 Project Structural behavior ‘This thesis made a division in various timespans of LA. By doing so the writer implies that these levels cannot be put together ina single working mechanisna. Different levels react to different data, therefore the actuation cannot couple these two levels. ‘This categorization however also implies that for example interactive facades should not be able to be load bearing; structures interacting with applied load should be. High structural ability depicts kinetic structures which are load bearing and able to carty loads ona large scale, These include dome roofs or moving floors. No structural capacity means the project is probably combined with small scale interactivity like interactive facades or installations, Asmall structural capacity implies a normal resistance to environmental factors like ‘wind, snow and rain. Those structures could be high load bearing if they were reinforced enough. = ie oe We WN KRONOS ‘ot Tom Number of Recast igure 20: Load Bearing categorization 20 2.2.8 Project Intelligence Figure 21: Intelligence Categorization The first part of [A in the definition of the book Interactive Architecture was that of intelligence. Intelligence can follow different paths to achieve its goal. Central {intelligence implies a central microcontroller and extensive programming whereas swarm behavior implies local microcontrollers with simple programming. Examining all of the literature projects we notice an equal distribution of 8-6-10 as depicted in the above Venn Diagram. 2.2.9 Project Sensor values Categorizing the above projects by their sensor values we have two extremes. Pragmatic values are values like temperature, proximity or light level. Flumanistic values are values that are able to sense a persor’s behavior, mood or activity and act accordingly. We notice that pragmatic sensors are widely available but humanistic sensors are not yet available on a large scale. It is the writer’s opinion however that this will change drastically in the next decade. The scale of the distribution is 18-4-1. Figure 22: Sensor Values Categorization 2 2.2.10 Web-Based Literature Because of the nature and age of IA and KA, a lot of the literature can also be found through various website's, These can provide a database containing different projects or are from architectural practices researching 1A. Most of them are addressed through the bibliography and illustrations. However the other part of the web-based literature is not discussed in this literature study, because of its large extent. The exception is the following visualization, which was made based on the website “Interactive Architecture dot org’, an independent blog describing various projects in the current IA landscape. The visualization shows the relationship of keywords of all the posts and arranges them by quantity. Surprisingly enough keywords such as “research” and “students” have an active role in this curriculum, proving the youth of Interactive Architecture at the time of writing. de but wou bh Hoque! Eherion = ee nstrotarts SG Agere! comings Eason =ateseerch nes “Landon NOK Ko sage sn pos Figure 23: Interactive Architecture dot org, Wordle (Glynn 2005) 2 Chapter 3: Methodology & Results 3.0 Methodology Before talking about the different packages of the methodology this thesis will first elaborate on the used simulation software, Grasshopper, and the used Arduino microcontroller: 3.0.1 The Simulation Software, Grasshopper Research in simulation software led to a growing user community involving the add on “Grasshopper"(Davidson nd.) for Rhinoceros 4 SR8. The grasshopper add-on gives us a visual coding environment where parametrical design is the main topic. Rhinoceros is not developed for skeleton structures though. The community groups have developed pieces of code, visualized as “components” in Grasshopper, which allow us to simulate real physical behavior, After reading a comparison thesis on commercial skeleton tools (Fotiadou 2007), together with the documentation on Rhinocerus + Grasshopper (Davidson nd..), Kangaroo (Piker 2011b) and Firefly (Andy & Johnson 2010), this research has taken Grasshopper as primary simulation tool This is mainly because of the available components, Firefly and Kangaroo, which allow us to manipulate data and simulate kinetics, the two basic needs in this 1A research, Kangaroo is a physics engine based on the use of a particle-spring system. For a better understanding of the system and its implementation we refer to the Kangaroo Manual (Piker 2011c). But this basically means that in this thesis structural nodes will be modeled as particles and interconnecting beamss or truss members will be modeled as a “spring” connection between two particles. These springs have a certain stiffness defined by following formulas of general structural engineering, in accordance with their compatible units: EA Fekedx with k= 23 Linear actuators, an integral part of KA, are however being implemented ina different way than the stiff beams, Actuators are implemented as springs with a variable rest length. Variable length which is linked to the stroke of the specific actuator that could be used in reality. The result of a successful simulation will have adjustable sliders which can be used to control the stroke of every single actuator, just like in reality. When we adjust these parameters we adjust the length of the actuator and simulate the kinetics of the structure, which this research is looking for Figure 24: Actuators, Spring Implementation Figure 25: Truse member, Spring implementation = K Figure 26: Kangaroo Components in Grasshopper 24 3.0.2 The microcontroller, Arduino For receiving and sending digital and analog signals this research uses an Arduino Duemilove microcontroller, The Arduino was developed as part of a student project developing Open-Source hardware. It can be programmed using the Arduino programming language based on Processing, The Arduino has the ability to receive analog and digital signals but is also able to send digital signals to actuators like lights or motors. Different models differ in the amount of ports as well as having an Ethernet connection port. The latter is not implemented onthe Duemilove and thus not researched used in this thesis. “Arduino is an open-source electronics prototyping platform based on flexible, easy-to use hardware and software. It's intended for artists, designers, hobbyists, and anyone interested in creating interactive objects or environments.” (Arduino 2005) Testing the Arduino, LED-bar 26 To tackle the subject spoken of in the introduction it has been broken down into three packages: 1, Simulating Design ‘The first package is a compilation of different case studies. These involve practical cases which can be found in the current KA landscape. An information-orientated sampling has been used to find different atypical designs to test the abilities of the simulation software, Grasshopper, This package will also contain a comparison study between Grasshopper and a Finite Elements-software package, ANSYS, for numerical validation. 2. Simulating Behavior ‘The following package will implement different low cost, highly available sensors in some of the previous case studies. This again to test the abilities of the simulation software, Grasshopper and to find the scope ofits ability to link the physical sensory environment to the simulation. Not only the pragmatic but also humanistic sensors will be addressed here, together with the numerical differences of swarm and central intelligence. 3. Practical Design Issues In the practical design package, an ideal kinetic structure will be broken down into its different parts like cladding, connecting members, actuators, etc. Different general solutions will first be analyzed and produced in a practical study of building and using, The end of this package will contain a working prototype of a conceptual KA design. 7 3.1 Simulating design 3.1.1 Simulations, Case Studies 1. ORAMBRA, Tristan D’Estree Sterk a. Introduction The Office for Robotic Architectural Media & Bureau for Responsive Architecture, or in short ORAMBRA was founded by Tristan D’Estree Sterk, This small bureau is interested in “rethinking the art of construction alongside the emergence of responsive systems”.(D'Estree Sterk 2003) In one of his first publications he sets up a basic actuated tensegrity module (type 1) able to multiply in different fashions (type 2&3), Actuators are shown in his illustration as elastic members. Figure 29: Scale model Type 3 (D’Estree Sterk 2003) 28 b. Simulations 29 Figure 31: SimulationActuated Tensegrity Type 2 Figure 82: Simulation Actuated Tensegrity Type 3 30 2. Hyberbody, TUDelft, Kas Oosterhuis a. Introduction “Hyperbody is a research group at the Faculty of Architecture at the Delft University of Technology directed by prof. ir. Kas Oosterhuis. The goals set for the group's research are to explore techniques and methods for designing and building non-standard, virtual and interactive architectures. Cutting edge techniques and methods are taught and applied by researchers and students."(Oosterhuis 2000) During their work they collected multiple examples and student projects. The one this research will simulate is the first Muscle Tower. Stiff interconnecting crystals, ‘were held together by outer actuators. Actuators were pneumatic muscles that were made available by the Festo Corporation. A similar project was realized under supervision of Axel Killian et al, that goes by the name of WhoWhatWhenAir. Figure 33: WhoWhatWhenAir, Flexible Tower (Kilian et al. 2006) Figure 34: Muscle Tower Il, Hyberbody (Oosterhuis 2000) 31 b. Simulation Figure 35: Simulation Flexible Tower 32 3. ORANGEVOID, Robotic Membrane a. Introduction Orangevoid is an conceptual architectural firm specializing in 1A, interactive facades and robotic membranes. One of their remarkable prototypes is that of a robotic triangulated membrane that can move due to pneumatic actuators attached to it. Without declaring the sensory inputs, only the kinetic counterpart, this design is merely technical. However TESSEL is an interactive installation also consisting of a triangulated membrane with the same appearance but actuated by retractable cables attached to ‘the ceiling. This installation moves according to sound waves that a couple of sensors pick up on the membrane, Figure 36; TESSEL (La Figure 37: Robotic Membrane (Orangevold n.d.) 33 b. Simulation Figure 38: Simulation Robotic Membrane 34 4. Actuated Truss a. Introduction Acollaboration at the University of Toronto, between an architect and an aerospace engineer, came up with an responsive truss (Merali & Long 2009) consisting of an upper beam connected by hinged truss members and steel tensioned cables. Each triangle pointed dowmwards is a single cell. Each cell has embedded force sensors which can measure pressure in the upper beam. Depending on the internal stress the triangle elongates, stretching the steel cables but adding more reaction upwards force and bending resistance to the beam, The beam will thus adjust its height to the forces applied to it, The usable height of the space below is therefore constantly optimal. (a teat sist 1) Sar ees emp Figure 39: Excerpt from dissertation, Responsive Actuated Truss (Merall & Long 2009] 35 b. Simulation Figure 40; Simulation Actuated Responsive Truss 36 5. Kinetic Circle a. Introduction Designed by the writer himself, a Kinetic Circle that can change its height by rotating the lower circle, Based on the mechanism of an aperture this mechanism will heighten its upper circle instead of closing it towards the center. The only member that has not been modeled is the column that has to be placed in the middle of the circle to avoid the upper circle moving from the center. b. Simulation Figure 41: Simulation Kinetic Circle 7 6. Nurnberg scissors, Hoberman, a. Introduction Chuck Hoberman, Architect and founder of Hoberman Associates, is the inventor of the expanding Geodesic Dome. The dome, which has even been changed into a child’s toy, is one of the first in the era of KA based on the Nurnberg Scissor mechanism. Besides the expanding dome the Hoberman Associates have developed different expanding structures like the helicoids and hypars. A sister program called the Adaptable Building Initiative or ABI has also made different kinetic models including, their patented Strata System and different adaptable glass fritting’s. ea 2 Figure 43: Strat uilding Initiative 2006) Module( Adaptive 38 Figure 44; Simulation of the Expanding Geodesic Dome 39 7. Results Models compiled from nodes, beams and anchor points which are constant in space, can be simulated. Bending can also be accurately incorporated in Kangaroo by adding bending stiffness ona node and two of its connecting members. Different structural nodes like hinges, trusses or beams can thus be modeled with this option. Figure 2: Bending validation Kangaroo (Piker 2011a) Issu |: Models involving rolling anchor points, meaning anchor points on a line or surface, are not yet implemented. However Daniel Piker, the creator of Kangaroo, tells the community that these features will be available in the next version of Kangaroo expected late 2011. ‘The geodesic dome for example cannot be properly simulated. Rolling guides are supposed to stay on a certain line and are unable to move from it. The example of the Kinetic circle was able to define a rolling guide upwards. This however by introducing a new force calculated by its displacement from two perpendicular surfaces, cutting in the center point. A very complicated solution whereas it can be easily modeled by an anchor point by line if available. Figure 3: Geodesic Dome Simulation Issue 1-2: When introducing the exact stiffness in Kangaroo, the iteration engine cannot handle this high, very stiff, value. Therefore a smaller time step of the iteration has to be chosen which can be camouflaged by increasing the sub iterations that happen between plotting the structure in the Kangaroo engine. More information about the calculations of the engine can be found in the Kangaroo manual (Piker 20110) 3.1.2 Numerical Validation Grasshopper is validated through the simulation of a 2D truss confined at its ends. In this validation the displacements of all the nodes will be compared to an FE calculation made in ANSYS. The differences in displacement will be calculated and their percentages are shown next to each other for comparison. Both models have the same material properties and dimensions. Young's Modulus, E, equals the standard construction steel value of 200000MPa. The cross section of all the truss members is equal to 2500mm?, consistent with a square section of 50mm. Issue 1-3: This validation is based on a linear calculation in ANSYS. In other words ‘we assume that the superposition of different load cases will lead to the same end results. It is possible to account for non-linear behavior of the construction but is not needed in these stages of the validation. As mentioned before KA describes the truss members as springs. Springs that follow Hooke's Law. Since we know the stiffness using the same parameters for calculations in ANSYS, we can simply divide the member's strain in the simulation with their appropriate stiffness to find the internal forces. These forces are projected on a color scale and previewed on the structure. Figure 45: Adding a color scale in Grasshopper aL 3. 1 Default Scenario Results ‘The default scenario is that of no actuation and is used to validate the Grasshopper as a means to calculate displacements and internal forces ina normal situation, SKN and 10KN loads were applied to the upper nodes. tt aaabeer LA, ININ \ Figure 47: Simulation ANSYS, SKN, Not Actuated, Deformation Seale 1:1 9 1 16 Pred — bass 1 00 2 00 3 oa 4 a 5 6” 0,0 7 8 1 9 10 ni” ur 2 iw 5 16 v7 | __39. 0,0 20 00 2 i) 2” 00 00 ‘Table 3: Error Margin %, Default SkN Table : Error margin %, Default 10 KN 3.1.2.2 Actuated Scenario Results ‘The actuated scenario will shorten the 4* vertical truss member to half ofits length and apply OKN and SKN loads on the upper nodes for validating the kinetics and the Kinetics respectively together with applied loads. Figure 48: Simulation Grasshopper, OkN, Actuated, Deformation Scale 1:1 Le“ |, Figure 49: Simulation ANSYS, OkN, Actuated, Deformation Seale 1:1 ‘Table S:Error Margin 5, OkN Table 6:Error Margin %, 5kN 3.2 Simulating behavior 3.2.1 The Simulation Software ‘Thanks to the plugin, Firefly, Grasshopper can input and ourput data from the Arduino microcontroller. By doing so this plugin is the missing link for observing our interactive architecture with a real sensory environment without physically constructing it Itis even possible to compare the physical with the simulated model ona small scale for validation before constructing the entire design, With these final steps the simulation of KA is complete and the physical design can be bi Figure 50: Arduino Send/Receive implementation in Grasshopper 44 3.2.2 Linking Data Methods 1. Swarm Intelligence ‘Swarm behavior, as aforementioned, can be seen as the simplest but also the fastest way to control intelligence. Sensor data is interpreted by a single cell, which it acts ‘upon according to its programming, Therefore swarm behavior entails that the sensory environment and the actuation environment are the same. ‘Swarm behavior is implemented with ease in Grasshopper. The raw data values of sensors can be remapped to the range of the actuator’s stroke, Functions such as smoothing over a number of intervals to avoid sudden spikes as well as dampened ‘wave functions can be implemented to simulate a swaying movement responding to a certain threshold of applied pressure or acoustic waves. If only remapping the sensor data is no longer enough, Grasshopper supports basic mathematical operations to manipulate the data. Ifthe data is manipulated in a more drastic way than in mathematical operations, this thesis refers to this intelligence as central intelligence. Figure 51: Manipulating and visu izing Sensor Data in Grasshopper 2. Central Intelligence This research considers two methods of implementing central intelligence: Programming in an external language and implementing in Grasshopper or using the standard Galapagos Evo! Programming is possible in C++ or VB, While programming intelligence the designer can anticipate various scenarios and implement them to his choosing. This however requires time and programming skills from the designer, but can be implemented in Grasshopper as a component to act with the incoming data. Another way to simulate a certain actions to use an optimization solver. Meaning introducing an end value which can be minimized or maximized. These values could for example be the sum of internal forces multiplied by its beam lengths to minimize the internal stresses. Or the distance between a point and a structure to minimize, grow towards, or maximize, grow away from. The optimization solver Galapagos will then optimize each parameter, actuator stroke, to minimize or maximize the end value. 45 3.2.3 Simulations 1, Kinetic Arch expanding towards a specific point. Imagine a kinetic arch which is a 2D section of an actuated tensegrity half pipe . This arch has perpendicular actuators in his structure which can shrink their length by up 10 20%, A point is designated on the exterior of the arch to visualize a certain point of interest to the arch. A point where the arch can grow towards or stay away from. The Grasshopper environment can calculate the distance between the point and the surface of the arch. This point canalso be the starting position of a specific node to ‘measure and minimize its displacements. Galapagos, an multi parameter optimization solver, built into Grasshopper, can iteratively find the right elongation for every actuator to minimize, or maximize the distance between the point and the arch, In this observation are the results of 3 evolutionary optimizations and one comparison swarm method. The compared method is a swarm behavior calculating the distance to each division of the arch and remapping those numbers to the interval ofactuator movement, eg. [0;-0,2}. The swarm method is direct and can be performed without a central microcontroller performing a cumbersome iterative calculation. Figure 52: Optimization, Kinetic Arch towards point Infrared sensors are able to measure a certain distance away from them and can thus be implemented directly from the sensor in the simulation, Like aforementioned the data can be smoothed, meaning taking an average of a couple of previous values, to avoid unwanted spikes. oe 0.565594 0.565594 0.565595 0.565692 Difference ° ° 0 986-05, Table 7: Intelligent- versus Swarm Behavior Issue 2-1: These compared results show that an intelligent design can actuate more accurately than a swarm method. However the long calculation method and the physical means to connect every sensory input to a central microcontroller gives it a disadvantage compared to the swarm method. It can also be seen that various actuator combinations offer an optimal solution. Therefore the end results are not predetermined but depend on the proceedings of 7 2. Actuated Truss ‘The responsive actuated truss as simulated in the previous chapter can be linked toa pragmatic sensor value like that of a push sensitive sensor. In this example the sensor outputs a value between 0 and 1024, which is normal for analog sensors. Just like the infrared sensors the values are smoothed and mapped to fit in the actuator stroke range. These inputs are directly inputted like the parametrical sliders were in Grasshopper earlier. The following illustration used the value for the right ‘most cell in the truss. Simulating 5 cells also implies that there need to be 5 independent sensor values. Current Arduino microcontroller boards have only 6 analog slots and 14 or more digital inputs. This practical problem can be solved by using shift registers to use only three input slots fora range of @ independent inputs. Figure 55: Push sensitive sensor connected to the Arduino, Figure 56: Implementing Sensor Data 3. Kinetic Arch minimizing internal stresses The same kinetic arch as spoken of in the first simulation can also minimize its internal forces by defining the sum of the products of internal force and length of each beam, Galapagos can then minimize this sum to calculate its optimal form. For this simulation a downward force has been placed on the third node from the left. Using this downward force and structure, Kangaroo calculates the internal forces and displacements, The calculated forces and lengths will then be used in the product sum. In reality the applied forces are not known in time. Strain gauges would therefore be applied to every beam in the structure, providing the structure with real time ‘monitoring of the strain and internal forces in its members. These values would then. be used to make it possible to optimize its structure using its programming, intelligence. Issue 2-2: If we compare three different iteration processes we see that the exact actuator values accur in every process. However this does not prove that every iterative process delivers the same results, as shown in issue 1. FL 0.000033, 0.000033 0.00003: Table 8: Comparison Different iteration processes 49 4, Remarks a Sensors An enormous amount of low cost sensors are available as an electrical component. ‘Some can even be recuperated like shown in the informal handbook of Usman Haque and Adam Somial-Fischer: “Low-tech sensors and actuators". (Haque & Somlai- Fischer 2005) Sensors worth mentioning that have been used during this research but haver‘t been implanted in a working simulation are: Flex-bend resistor, Optical resistor, tilt sensor and RFID antenna. The RFID antenna cannot be directly implemented in GH without ‘the necessary programming, whereas all the others can be directly remapped. b. Actuators ‘The add-on Firefly for Grasshopper can also output values to servo's, motor or other actuators from GH. This way GH can even be used to syne the simulation model to a physical model. This is useful for example when working ona prototype or when comparing the models to each other for research or validating the prototype before producing on large scale. 50 c. Protocols Other remarkable functions are implemented in Firefly, like the ability to link GH to Pachube (Haque n.d.). Pachube was created by Usman Haque and offers a place to upload and download data streams. For this thesis a temperature and light sensor ‘were set up to upload data to Pachube, From the Pachube portal the streams of data were downloaded into Grasshopper. This protocol can be used in various situations varying from a single sensor on top of a building that controls the entire facade or lets actuators respond to data from another continent for various design reasons, Figure 58: Pachube receive implementation in Grasshopper 1 asaootene, arte S240 ter Figure 59: Pachube, Live sensor streaming and stream information SL Firefly also introduces the fiducial protocol in GH. This protocol was developed and first used by a team (reacT1Vision nd.) which was introducing a recognition tool for ‘use in multi touch tables. Using this protocol a standard camera can notice different standard shapes and even the direction they are pointed in, This can form the basis for a simple tracking algorithm for use in IA. Linking these shapes to different people makes this protocol a humanistic sensor: Coordinates from the movement of different people can thus be used for actuating the structure using swarm or central intelligence. Figure 60: Fiducial implementation in Grasshoppé 52 3.2.4 Emotive behavior a Humanistic Sensors ‘The sensors available today are merely analog or digital devices who can output pragmatic values like applied pressure, daylight level, etc. Itis the writer’s opinion however that seeing the current pace of events, sensors will be designed that can decide on what activity we are conducting or how we are feeling by looking at our stance or facial expression: humanistic values. RFID components are already able to sense information stored in passive RFID tags ‘which are getting smaller and which can even be implanted sub-dermally, under our skin, addressing some key issues such as privacy though. Figure 61: Sensor Microsoft launched its “Kinect” sensor this year. A motion capture device for console gaming, that makes use of a controller obsolete. The Kinect is a small 3D scanner that can capture our skeleton be deciding a couple of key points in our stance. These points can be linked to our mood or activity. At the time of writing third party experiments are being setup to send Kinect data to Grasshopper using the universal OSC protocol Figure 62: Grasshopper/Kinect Sensor (Andy Payne et al. 2010) 53 b. Behavior Another topic is the actuation ofa design. We take it for granted that IA can follow pragmatic or humanistic data as it comes along per second, minute or hour. But does it have to? Keeping in mind that there are different levels of interactivity, they can all have different behaviors. Behaviors as discussed here are the structure’s reaction speed, accuracy and logic. People react and experience space differently when the structure moves gently with a swaying move or suddenly shakes to its end position. People can startle or even. empathize with the kinetic structure, The place receives its own character and behavior and can alter the perception of space. For example the Dune 4.0 project by Studio Roosegaarde changes the space of a sterile pedestrian tunnel into a living, responsive environment. Accuracy, as depicted in our comparison between intelligent and swarm behavior, can be a design issue, Maybe we want a building that does not do as well as bigger, stronger and more expensive buildings. Or maybe increased accuracy just is not necessary to successfully complete its basic goals, hence swarm versus central intelligence. Logic can be introduced when more accessible versions of learning artificial Intelligence make their way to the public, This is not the case at the time of wri Figure 63: Dune 4.0 Maastunnel(Roosegaarde 2011) 54 3.3 Design Issues Simulating a model of your design is a great way of getting to know the global structure and its behavior, Only an abstract level of detail is necessary to do so. It is. possible to detail a simulation model to the level of construction. This is however very time consuming and without the use of parametrical design almost un-existing for large scale applications. Anew era of CNC milling, 3D printing and laser cutting however has come to our doorstep. Parts with different lengths, but the same characteristics, can now easily be produced parametrically to be installed in a final design as a puzzle. These new technologies together with existing means should be incorporated in our designs and should be taken into account in the primary design phases. Figure 64: Laser-cutting and engraving 2 sheet of MOF 55 33.1 Joints Joints are defined as the nodes between members. A joint couples different members in different directions and allows different degrees of freedom for each separate member, Since different push and pull forces come into contact with the joints, they have to be developed and created to be able to withstand them with minimal weak points that might cause failure. In general it can be said that nodes only moving in the 2D plane, involving two or more members, are relatively easy to produce. Nodes connecting members in3D space require, even with only two members, an enormous amount of extra design and production time, 33.2.1 2D Joint ‘The connection point of the node is calculated by the intersection point of the connecting members. Using a transverse pin different 2D members can be attached ‘to each other. This connection also resists movement out of the plane due to its connection. Issue 3-1: This obvious solution however introduces eccentricity, which creates movement out of the 2D plane if the structure has large forces working onit. To avoid this the member can be connected with a symmetrical end point, which avoids the problem of eccentricity. Limiting factors of these nodes are the thickness of the node and the strength and length of its connecting pin. Figure 66: 20 Turning Joint, without eccentricity 56 Issue 3-2: When designing these 2D joints their intersection line always has to fall in the same point. Nodes designed in an earlier stage as depicted in the lower figure are thus not correct, Even when we connect multiple members the node is still unstable and cannot be used as such. igure 67: 2D Joint Unstable 3D nodes comprised of two 2D joints are also possible. The movement of the connected members is however only in those two planes, which is an integral part of the single 2D joint. Meaning this 3D node is not useful in applications where the connected members move out of the two intersecting planes. Because the connection does not fall in the intersecting point there need to be at least three connecting members to stabilize the node in space, instead of the unstable 2D node aforementioned. The model produced in this research consisted out of parts which were laser-cut and could be inserted into each other forming a perpendicular structure. The connection was modeled with a snap-fit! design so that the two perpendicular Parts could not be pulled apart before first releasing the snap-fit. * Snap-Fit Design: Using a hook-like mechanism parts can be locked into position. The design of a snap-fit involves extensive calculations in material strength and applied forces. 7 33.2.2 3D Joint Design of 3D joints depends greatly on the number of attached truss members, The basic connection of only two members has been used in many applications such as the transmission of boats and cars. Issue 3-3: The universal joint or Cardan-joint is made of two symmetrical pins connected with a middle cross connecting both. This node is thus able to move intwo directions and even able to transmit a rotating motion in the member in another direction. During this research a parametrical model was designed to 3D print these connections. The first model that was printed was not correctly designed. There was no structure ‘holding the middle pins in its location, causing it to fall out of place but also break during creation. This however has been adjusted by adding a circle in the inner cross. ina second model. eS Figure 70: 3D-print Universal Joint 2 58 Issue 3-4: A ball joint is another method linking two members able to rotate in the 3D plane. An obvious limitation when using 1 ball is the inability to connect more than two members. A patented node system for space trusses, that goes by the name “Tubal”, consists of a metal sphere where threaded ball joints attached to the beam can be implemented to connect space truss members. Figure 71: Cardboard Space frame; Ring Past, Delft (Octatube 2010) In the scope of this research the writer designed a similar parametrical ball joint only ‘with internal ball joints in every cardinal direction, which makes the ball joint attached to the beam obsolete. Parameters are the general dimensions as well as the placement of the ball joint, With these parameters the angle of freedom is also calculated and can be taken into account. ‘This design is not better in the economical sense of material use but has a wider angle of movement, developed especially for KA, and cannot be compared to the node system for static space trusses, which only require a certain degree of freedom consistent with its calculations. This node is obviously limited in strength to the specifications of the 3D printer. However it is the writer’s opinion that these printers will be able to use plastic of higher strength and density to even print ready-to-use nodes for real structural projects in the next few decades. Figure 72: Parametrical Ball Joint 59 Figure 73: Section Figure74: 3D Printed Ball Joint Figure 75: 3D printed Ball Joint, Section A3D printer works by layering hot plastic on top of each other to make a model. ‘When stiffening structure is needed the 3D printer can print a dissolvable plastic ‘which later can be removed ina NaOH bath. This ability was used in this model to implement ball joints in the sphere without having to use any other holding structure. The ball pins are thus able to handle tension and pull forces. 60 3.2.1 Members Issue 3-5: When constructing a truss member or beam in.a 2D plane the beam has no need for stiffness in the perpendicular plane. However when talking about 3D trusses or beams they should have stiffness in both directions, meaning height in both directions, considering the moment of inertia of beams. Froma basic mechanical insight we learn that the bending moment implied in the structure is maximal in the middle of the member whereas it is flexibly connected to the nodes. More height is therefore needed in the middle of the members rather than in the endpoints. For this research parts were laser-cut from plastic to be inserted into each other using an exact fit method®. The result is a bidirectional stiff beam, ideal for 3D applications. Its connection however is still only in one 2D-plane. Figure 77: Truss Member * Exact-fit method: The method of laser-cutting connecting parts with attention to the thickness of. the laser-cut line of 0.1mm, Interconnecting pieces will then fit perfectly and provide a certain strength to the connection (thicknesses change according to the specific model), oL 33.2 Cladding When talking about viable spaces we cannot abstract the need for a suitable insulation and airtightness that is available in today’s performing skins. Whereas new materials and flexible insulations are making their way into large scale applications, connection details and proper specific sealants seem to be lagging behind, However it is the writer's opinion that these materials will follow when more practical experience is harvested with regards to cladding systems on kinetic structures. 1. Textile membranes Textile membranes, which are an obvious choice for waterproofing our kinetic structure, can be insulated. Tensotherm (Birdair nd.) for example is flexible composite insulated membrane composed out of a layer of Nanogel insulation protected by two layers of PTFE fiberglass textile. It has a remarkable low heat transmission, strong resistance to impact damage and a very low own weight Negative points might still be the inability to withstand vandalism or cutting, Textile membranes can also be used as a means to manipulate light levels. A stretched ‘membrane will be more transparent than an un-stretched membrane. For example when a structure expands its surface, the membrane will be stretched, thus letting more light into the space. Figure 78: Textile membrane, Unstretched Figure 79: Textile membrane, Stretched 2 2. Stiff Cladding Issue 3-6: Stiff cladding like wooden or metal parts can also be constructed, However when working in a 3D plane this cladding has to be constructed isostatic. When a cladding material moves out of his plane, as a result of the structural frame deforming, it introduces hyperstatic forces in the panels when they are simply fixed toa structure. The new connection is similar to an expansion joint for metal cladding systems which are influenced by their heat expansion coefficient. ‘To research these phenomena a kinetic box was simulated consisting of one actuator in each horizontal face of the box. Scenario 1 shows what happens when the cladding plane is moved and scenario 2 shows us what happens when the cladding materials move out of the plan but stay connected to both upper and lower points on the frame. 63 Scenario 1 shows us that stiff cladding made of vertical or horizontal strips can have the ability to follow the deforming surface when it moves in its own plane. Suitable rubber fittings should be applied between two strips which are able to move along their line and still apply sufficient closing pressure, In scenario 2 the strips were fixed to the lower frame member: Instantly can be seen. that the cladding strips have moved upwards, in this example the middle strip moves 18,9mm, or 2% of its length, upwards when the actuator in the other face shrinks 30% of its own length. Figure 80: Kinetic Box, Default Scenario z Figure 81: Kinetic Box, Scenario 1 Figure 82: Kinetic Box, Scenario 2 64 Itis thus important that connections are made which are able to withstand these movements during the actuation of the entire structure. A model has been made to simulate this connection, When operating this model the movement of the cladding was possible in the same plane, However this joint still resisted movement of the upper beam out of the plane. This due to the straight connection of the transverse pin. This can be solved by replacing the traverse pin by a knuckle joint. A knuckle joint is a ball joint with attached bolt to apply a nut to fasten this cladding. This also implies that the cladding has to have moving space between the plane of the structure and the plane of the cladding to allow this movement, -_ a 65 3. Interactive dadding Interactive claddings follow the same upper principles whether the cladding is stiff or flexible. Different examples are already available in the current [A landscape, mostly without discussing the attachment to a lead carrying structure. This connection is however an important aspect which involves a lot of practical knowledge like aforementioned. This is however obsolete if the load carrying structure itself is static. Projects such as Living Glass, which opens creases in its cladding as a response to proximity, or the xeromax envelope, which is an elaborate conceptual design for a living, self-replicating skin, are projects where this is not fully elaborated in the writer's opinion. igure 86: Xeromax Envelope (Future Citi 66 3.3.3 Actuator design Actuators in this thesis are mostly designed as linear motion mechanisms, meaning beams that can contract or expand their own length to actuate the entire structural mechanism. This is mainly sufficient when describing kinetic space truss mechanisms, but does not imply that other actuators must do the same. Actuators have a wide scope of driving mechanisms involving pneumatic, electrical or even chemical processes. A defining characteristic of these actuators is what torque they can resist and how long they can shrink or elongate their length. Figure 87: Rectangle, Diagonal Actuation o7 L. Muscle Wire ‘Muscle wires that go by the production names Stabinol, Flexinol and Nitinol are ‘wires based on a Nickel alloy which can contract up to 3-5% of its length when heated or conducting electricity. Issue 3-6: These wires should however be avoided most of the time. A muscle wire reacts to heat and its use in for example an interactive facade will actuate the muscle wire thanks to the sun without programming, This could be implemented as a passive actuation but is mainly uncontrollable by other means. Muscle wires also have a pull strength of multiple times their own weight which amakes them useful in controlled smalll scale applications. However not in large scale outdoor and large applied forces application such as building skins and other structural building concepts. ‘The physical environmental features and the small absolute pull strength make the use of these wires thus non-existing in the field of KA. Figure 88: Muscle Wire Actuation 2. Rotational Actuator Rotational actuators are already widely available because of the nature of the generation of motion in electrical mechanisms, Researched models are small-scale servo and stepper motors controlled and fed by the Arduino microcontroller. This circular motion however can be changed into linear motion thanks to various mechanisms. The most popular have been researched, including a crankshaft mechanism, a screw mechanism and a gear-pinion system. These basic mechanisms are also widely available in different shapes and stroke lengths available through process engineering manufacturers. Thanks to this research it became very obvious that every mechanisms has its own limiting torque and elongation rate. 8 Issue 3-7: A screw and gear-pinion actuator can theoretically shrink by up to 50% of their total length. In practice though, all actuators depend on their practical construction. In this practical research the rates were respectively 28% and 48%. crankshaft mechanism can theoretically expand a value equal to 2 times the diameter of the rotation lever: There is however need for a guide that stiffens and allows the movement of the expanding arm and therefore limits the total elongation of the actuator: The elongation of the test model is equal to 23%, For even larger actuation, possibilities are telescopically linear actuators or a windup method of a string, which are not researched in this thesis. The torque is the force multiplied with the lever which is installed on the servomotor There is a large difference in price and available torque of servomotors. The right motor therefore has to be calculated on the applied load or practically researched. For these experiments motors where used with the same characteristics, which allows us to compare elongation rates. Figure 89: Serew Lin Figure 91: Gear-Pinion Linear Actuator ere Crankshaft Gear Pinion Screw Pneum: abel 9: Lin Shrink Rate [%] muscle 1 F Actuators, Shrink Rate 69 3. Hydraulic & Pneumatic actuators ‘The final group consists of pneumatic and hydraulic actuators. These are actuators which can contract due to an air pressure or compressed liquid. These actuators: however are able to withstand great pressure of tension forces, which, in the writer’s opinion, makes them ideal for KA. Again, mechanical process engineering firms have large lines containing both types in all shapes and forms. This thesis will only research one example based on the Pneumatic muscle of the Festo Corporation, igure 92: Muscle Project (Festo Corporate 2003) After reading a dissertation on the analysis (Daerden & Lefeber nd.) involving different pneumatic muscle it was quite obvious that these muscles were not the only types available. Different types of muscles have their own characteristics, negative and positive aspects. For this research the McKibben type air muscle was chosen The McKilsben type is the most widespread muscle, that has an easy installation and still has performing characteristics. Pleated Air Muscles (PAM's) have been researched at the University of Brussels on a theoretical and practical level (Van Mele 2008) Issue 3-8; The McKibben air muscle consists ofa supple inner tube that does not resist deformation while under pressure. The material tsed in this research is silicone rubber. A braided sieving will however be put on top of the inner tube which ightens when stretched. This will keep the inner tube from deforming locally and failing under even a small air pressure of 3 bar. The total air muscle is a composite where the braided sieving is ideal for resisting pull forces. The total pull resistance ‘before snapping is enormous (dependent on type), making it ideal for structural applications. /~ breed sleeving / & inner tabe ES este atest pressurized igure 93: McKibben Principle (Daerden & Lefeber n.d.) 70 ‘The McKibben air muscle has been researched using a basic setup to apply loads on the muscle, The elongation of the muscle was documented with every step of 0.5 bar air pressure. Different weights were connected to the end of the muscle ranging from Lkg to Skg, The silicone tubing used had a total diameter of 15mm and a wall thickness of 2mm, Figure 94: Test Setup, FigureS5: Mekibben air muscle setup mn ‘The muscle is controlled by a 3/2 way Festo air valve which connects the muscle to either the pressured air tank or the outside ait; This valve was steered by a solenoid valve on 12Y, which only has a proper working range of 2-8 bar, which makes the test results only available from that point on. Using a Huntington array asa relay the Arduino microcontroller was able control the valve using a pressure knob or timer. ‘The maximum pressure range was practically chosen at 5 bar. Since connections are weak points in the setup, they could suddenly break without notice. There was also only a slight change noticeable in the change of length after around 5 bar, Test results on the next page show us the elongation (positive) or shrink (negative) relative to its original length. The “Total Range” row shows us the actuation range of each load case. Results vary between 9-11% of actuation between its stretched and actuated stance. Figure 96: 3/2 Air Valve Festo Figure 97: Arduino Controller 2 Elongation[/] vs. Air Pressure[bar] 1908 106 | 108 102 1,00 098 | 0.96 094 092 | o 0,00 0,50 1,00 1,50 2,00 2,50 3,00 3,50 4.00 4,50 5,00 Air Pressure [bar] ekg eis aks asks ets Figure 98: Elongation/Original Length [%] SOOM sa: = 6 Total Range 8,78 Table 10: Muscle Elongation [%] 3 33.5 Prototype Design 3.3.5. Introduction Flexible Skin As a summarization of practical knowledge this research ends with the design and creation of an Interactive Kinetic Structure. This prototype does not focus on documenting the social interaction with the structure, hence its social aspects, but documents its design process, production process and problematic issues. The prime conceptual idea behind the prototype is that of a flexible skin. A skin which is able to define architectural space on its own. A skin that responds to ‘pragmatic data such as sunlight, proximity and touch. By doing so the inhabitant is no longer shielded from the outside, passersby or sunlight by a brick wall but perceives the space in.an interactive way, constantly changing in time. While existing prototypes in the field of 1A do not specifically mention the different levels of interactivity involving timespan, this research does. This research states that there are three different levels of interactivity: Direct, Weeks and Months. Most of the projects in the current landscape only involve direct interactivity. This prototype skin is comprised of the two first layers of interactivity, meaning direct interactivity such as daylight entrance and ventilation as well as defining its own behavior like shyness and aggressiveness based upon these pragmatic values such as proximity and touch. ‘The second layer of interactivity, meaning the structural layer in this case, can form itself into an optimal space according to minimal internal stresses or spatial optimization, Both layers are aesthetically woven into each other unlike projects in the current landscape which mainly introduce either structural or direct interactivity. Issue 3-8: Another fact that this research specifically mentions is the overall inability to incorporate two different levels while they are linked together with only one actuator. Meaning one actuator cannot be responsible for two levels of interactivity. For example, our need for daylight and ventilation does not coincide with our need for an optimal structure with minimal displacements. This insight only came after ‘the practical issues during the actuator design. 74 Biomimicry “Biomimicry or biomimetics is the examination of nature, its models, systems, processes, «and elements to emulate or take inspiration from in order to solve human problems. The terms biomimicry and biomimetics come from the Greek words bios, meaning life, and mimesis, meaning to imitate, Other terms often used are bionics, bio-inspiration, and biognosis.” (Wikipedia 2011b) ‘The design was inspired by nature in the sense that the designer was intrigued by corals and underwater polyps. Seemingly monotone and static these polyps can change their shape and form under different conditions. Polyps can retract when touched by an unknown entity or can open up to catch sustenance. This like our flexible skin that also has to catch sustenance like daylight and ventilation but also intrigues people with its interactivity and response to touch or proximity. It is for this reason that the design of the actuator is based upon the form of a polyp and delivers the primary layer of interactivity. 75 3.3.5.2 Results Design ‘The axonometric view shows the design split up into the two different layers. The polyps, layer 1, and the truss, layer 2. Both of the layers are independently actuated. ‘The truss actuation is not physically built but will depict a certain snapshot of its, actual behavior: Besides the polyps and truss a cladding will be fitted to the outer surface to provide protection but also to serve as a touch sensor. ‘The cladding together with the outer polyp arms will be colored black. The entire structure of polyps and truss members are made out of Plexi-Glass of 2mm, laser-cut into its form. When the polyps are opened they will allow daylight and/or ventilation. Daylight disperses aver the transparent surfaces, space. Figure 100: Axonometric View Prototype juminating the inner Layer 1, Polyps Direct interactivity: Daylight, ventilation Actuated by electrical servo Motor Layer 2, Truss Medium Interactivity: Optimal structural form Actuated by linear actuators (Not modeled in this prototype) 76 Polyp ‘The design of the polyp actuator uses the “Scotch Yoke’, a cam mechanism, to transform the rotational actuation of the servo motor to a linear motion of polyp arms moving in and out of the skin, This mechanism exists out of two guides and a middle beam, “yoke’, which is attached to the electric motor As shown in the lower figure the yoke transforms into four separate polyp arms which allow them to move when moved out of the skin. Each arm is held out of its normal configuration by transparent strings, mimicking the appearance of an underwater polyp, ‘Simulation of this movement is possible but not necessary because of the minor difficulty of the system and since there are no external loads placed on the mechanism. The outer shell of the actuator provides the necessary stiffness for use in ‘the truss as a member. The shape of the outer membrane follows the interior mechanisms as well as the optimal structural form for a truss member, meaning the largest bending moment occurring in the middle of the member. 4pm ip igure 102: Scotch Yoke Mechanism (Mechanisms 101 n.d.) Figure 103: Polyp Actuation, 3D Sketch 78 Polyp, Upper node connection Issue 3-9: During construction of the entire prototype it became obvious that the polyps connection to the truss members was not sufficient to provide pull strength. Making the structure unstable and body parts falling out of their appropriate slide. To provide them with the necessary pull strength, all of the body parts were connected through plastic wire, Making the node connection able to withstand pull forces and act as presumed in the design, This is mainly due to the fact that the exact-fit method used for laser-cutting the parts was not accurately calculated to provide the necessary pull resistance by shear. 79 ‘Truss Members of the truss are reinforced in both directions for the occurring bending moment, which is maximal in the middle of the truss member. This joint design, ‘which is fairly easy to produce, was the limiting factor in the search for a valid truss system, Truss members could only move in the 2D plane they have been connected in, Issue 3-10: One of the first truss designs was a 2D simulation where the actuator still actuated the polyp and the truss and where a series of rectangles would provide movement of the skin, However when simulating the structure in 3D it became obvious that nodes and intercormecting members would move out of the plane they ‘were connected. The construction with 2D joints, is therefore not possible. Meaning nodes would have to be of the 3D printed ball joint type mentioned before. Because of the unknown material strength and cost of the 3D print, this research chose to search for an alternative design where truss members would only move in the plane they are connected in. Figure 106: 2D Simulation, Design 1 Figure 107: 3D simulation, Design 1 By simplifying the truss to a space truss of rectangles and diagonals a solution was found, The diagonals placed in each rectangle of the space truss can be actuated. When the actuator elongates the rectangle will deform and will heighten the upper connected node while shrinking in the other cardinal direction, For this prototype a grid consisting of nine nodes (polyps) will be created and simulated. To move one node of the grid in the upper direction surrounding diagonals must be pointed towards it. When all surrounding actuators are given the same elongation the point will move vertically upwards. A small shrinkage of the skin will occur in the other directions to allow the movement of the node. Figure 108: 2D simulation, Design 2 aL Diagonals can be placed ina pattern so that every couple oftwo nodes can deliver upward and downward movement, Following figures represent a simulation made on agrid of nine nodes ina specific stance, This stance will also be applied in the final prototype, by using stiff diagonals with different lengths and not by actuators. Figure 109: 3D simulation, Design 2 Figure 110: Truss assembly 82 Cladding ‘The cladding for the prototype is intended to shield sunlight from the inner space, increasing the effect of the opening polyps, and to be as lightweight as possible. ‘Therefore a cladding material of painted plastic was chosen, However to provide the cladding with a large stiffness but at the same time maintain a low weight of its own, this research chose parts which were vacuum formed! to create stiffer cladding sheets through their three dimensional form. These claddings were then painted black to provide the light shielding effect. Another design requirement of the cladding was that it could be used in the chosen truss rectangle truss system and that if one node of the rectangle should move upwards that the interlaying cladding could cope with the movement, Based on seemingly random patterns occurring in nature the designer sketched a random continuous intersecting line. This line was then transformed into a pattern ‘which was laser-cut onto a sheet of Medium Density Fiberboard (MDF) of 6mm. Figure 111: Lasercut Vacuum Forming Molds “Vacuum forming, commonly known as vacuuforming, is a simplified version of, thermoforming, whereby a sheet of plastic is heated to a forming temperature, stretched onto or into a single-surface mold (BrE, mould), and held against the mold by applying vacuum between the mold surface and the sheet. (Wikipedia 2011¢) 83 For the vacuum forming process, an operating box was produced consisting of suction holes on the top, based on the dimensions of the plastic sheets, and an insertion point for a vacuum cleaner, providing the vacuum needed to form the plastic. Figure 112: Vacuum Former with mold Issue 3-11; Vacuum forming would also have been used for the polyp arms. The ‘mold, that has been CNC-milled out of a piece of cellulose, however was too high for the plastic causing it to loosen from its frame and annihilate the vacuum effect. The plastic dimensions are to be chosen in accordance with the height of the mold. Figure 113: Vacuum forming result with high mold 84 Issue ‘he dimensions of the vacuum box were limited to the surface of the laser-cutter, meaning 600x300mm. This vacuum box therefore limited the size of the plastic sheets used to 250x250mm and thus also limiting the cladding to 1908&190mm. If larger vacuum formers are available the cladding and thus dimensions of the truss are possible. This is not the case however during the design of this prototype. Issue3-13: The pattern of the cladding, each of the four rectangles, was triangulated, meaning cut over both of its diagonals. This would allow movement of each connected node, while still maintaining cladding integrity. Cladding seams are connected to each other using a plastic one directional hinge. 85 Behavior Various sensors were implemented into the flexible skin. Piezo elements capable of measuring impact or shocks were glued to the cladding, transforming the cladding to a touch sensitive surface. One light sensor was attached to measure daylight levels, so that the polyps could respond accordingly. Since the 9 nodes of the prototype lie close to each other there is no need for installing 9 independent sensors. However in reality these are supposed to be installed. Each of the polyps then reacts to its specific sensor, transforming the flexible skin into a visualization of the incident solar radiation. Finally two infrared sensors were also implemented on the surface of the cladding, measuring the proximity of objects towards the skin. The polyps can thus react and interact with passersby or could shy away from them to maintain inner space privacy. The same principle goes for the implementation of multiple infrared proximity sensors so that the polyps will act more autonomously and inat a greater resolution. As a means of prototyping the polyps furthest away from each other will be driven by a separate infrared sensor. And the middle polyps will be driven by an averaged value of the 2 infrared sensors. All of the programming” was done in Arduino environment based on Processing. Ere Eee ight Polyps Open Polyps will catch daylight dispersing it over its structure. Light Polyps Close Polyps will block daylight to intervene in its inner space. Polyps will shy away from interested people, opening again over Touch Polyps Close time. Polyps will be eager to respond to interested people. Conversing Touch Polyps Open_ with the person in front of it. Proximity Polyps Close Polyps will shy away from passerby's. Proximity Polyps Open Polyps will affirm passerby's, luring them closer Table 11: Polyps Behavioral Scheme B: Behavior Arduino code ” Appendi 86 Figuur 120: Prototype Side-View View Figure 121: Prototype Perspective Figure 122: Prototype Upper View Chapter 4: Evaluation & Discussion 4.1 Evaluation & Discussion Simulating Design 4.1.1 Simulation Case Studies ‘The simulation of kinetic beam structures is possible with the Grasshopper environment. The component Kangaroo uses a spring based model where nodes are considered as discrete points and beams are modeled as springs with the appropriate stiffness. Actuators are springs with adjustable rest length. This research however did not implement a change in spring stiffness involving actuators. When the rest length of an actuator changes, its stiffness is supposed to change according to the stiffness formula E*A/L, when abstracting the actuator as a truss member with constant cross section. An actuator could also have a different Stiffness behavior since its cross section is not constant over its length. Further research could analyze different actuators and their stiffness behavior over its stroke, to implement in Grasshopper as an accurate stiffness function Kinetic structures only made of truss members, actuators and anchor points fixed in time and space can be implemented, when the basic skillset involving Grasshopper and Kangaroo is available. Bending canalso be implemented on two coupled lines to simulate hinged nodes. However Kangaroo lacks the necessary documentation explaining the used units as bending strength. The Grasshopper environment together with Kangaroo is not a commercial package. It is also not intended to be a theoretical calculation package. Therefore understanding Grasshopper only for the use of displacement calculation is not meant as a basis for further research. This package however provides the simulation environment with easy to use kinematics where the user can have total, visual, control over its design. Further research can however take a deeper look into the exact programming of Kangaroo and improve or add components where necessary. 90 ‘The main problem when using Kangaroo is its current inability to introduce rolling guides, meaning anchor points attached to a specific line or surface. These rolling guides form the basis for numerous kinetic structures, meaning a part of kinetic structures can still not be simulated, Kangaroo is continuously being developed as we speak. It has been announced that implementation of the rolling guide is to be introduced in the next release of Kangaroo and should prove to be a big improvement in this package's simulation capacity. In the meantime simulating rolling guides by variable pull forces towards intersecting planes isan elaborate and inaccurate solution. Figure 123: Simulation Flexible Tower Figure 124: Simulation of the Expanding Geodesic Dome 91 4.1.2 Numerical validation Kangaroo is quite accurate for simulating non-dynamic structures. A comparison study in displacements, comparing Kangaroo and ANSYS, shows that the averaged error margin of Kangaroo is only 0,1% of the ANSYS result. ‘When simulating a dynamic structure, meaning an actuated structure, Kangaroo is not as accurate based on the comparison results. Displacements found by Kangaroo have error margins of 316% maximum. However this does not prove that Kangaroo is not suitable for the simulation of KA. In ANSYS the actuator was modeled as a truss member with an initial strain. In Grasshopper it was modeled as a spring with variable rest length. Changing stiffness and other factors could play a part in these end results, as discussed before. Non-linear analysis in ANSYS and further research would be conclusive, Since the difference in results are small in absolute value this research however states that Kangaroo is a valuable early simulation package and will improve or justify its results in further research. ‘Table 13:Error Margin %, SKN 92 ‘When analyzing the internal stresses of the different calculations it can be seen that the results in ANSYS calculate tension in the upper left corner, where Kangaroo calculates neutral or positive internal forces. This result is caused by the way actuation is implemented in each simulation software: Kangaroo simulates a spring with variable rest length and a relatively large stiffness, whereas actuation in ANSYS is inserted as a different constant set. The initial strain constant of the specific member is inserted into the simulation program. The simulation by Kangaroo can be seen as a non-linear solution because of the iterative process ofits simulation engine. A non-linear solution will produce an applied bending moment thanks to the angle which the truss members has been placed in. A bending moment will imply two opposite fixed end moments like shown below. The simulation by ANSYS will keep the initial strain constant during the entire simulation, meaning no extra strain will occur other than the initial strain. This is also not the case in reality, which Kangaroo simulates better by only changing its rest length but still being able to elongate according to its stiffness. Figure 125: Simulation Grasshopper, OkN, Actuated, Deformation Scale 1:1 Figure 126: Simulation ANSYS, OkN, Actuated, Deformation Scale 1:1 93 4.2 Simulating behavior Using the Arduino microcontroller and the component Firefly In the Grasshopper environment, sensor data can be read directly from the board and used in the simulation of KA. The actuation of the kinetic structure is only based on the different variable lengths of its actuators. Based on the incoming data these lengths vary. Directly remapping the data values to the stroke range of the actuator can be easily implemented in the Grasshopper environment using the Firefly components. ‘Meaning pragmatic sensor values can be directly linked to the actuation of different actuators. This method is named Swarm Intelligence. Humanistic data cannot be stored in numbers and thus cannot be implemented in the above method. Central intelligence on the other hand is a coordinated actuation of all of the actuators based ona couple of sensor values not necessarily linked to each ‘respective cell Central intelligence can be implemented using either the integrated Galapagos Optimization Solver and /or programming in external language C++ and. vB. ‘The optimization solver can be used when optimizing a specific value such as displacement, distance or internal forces, Humanistic data such as skeleton recognition or RDIF readings can only be implemented by programming in an external language. igure 127: Optimizatin, Kinetic arch moving towards points 94 Besides central intelligence being more accurate than swarm intelligence, which was found in the compared results of a basic simulation, the end actuation is also dependent on the iterative process of the optimization solver. This was not visible in the first simulation, minimizing the distance towards a point, but was in the second simulation when minimizing the internal forces. This problem will always occur when using evolutionary solvers and their iterative process. However results show that different actuation end results give the same end value. Problems are likely to arise when more complex simulations will lead to different end results with major differences the end result. Therefore optimization solvers are to be handled with care and researched when implementing in a final design. 1 2 3| 4 s| > 6 ° ° of ns = = on 3a = oe) 4 ob 3,25 9 o ° 0,00 o 0.565598 0.565594 0.56554 0.565682 Difference ° o 0 886-05, swarm implementation, Simulation 1 Edintelligent1 Bed intolligent2 > SED 0,000033 0,000033 0,000033, ‘Tabel 15: Optimization results, Simulation 2 Grasshopper is not only interesting for the simulation of kinetics, this free and open coding environment makes way to different protocols that can be used in combination with upcoming humanistic as well as pragmatic sensors. For further details involving subjects such as the emotive layer and behavior we refer to the chapter methodology and results in question. It is the writer’s opinion that Grasshopper provides and will continue to provide a valuable environment for the simulation and design of Interactive Architecture. 95 4.3 Design Issues 43.1 Joints 45 12D Joint In general it can be stated that modeling 2D joints moving in their own plane are easier to build than a 3D node connecting members in different directions in 3D space and allowing the necessary degree of freedom. A node combining two 2D joints is possible when at least three members connect with the node, fixing it in space, and when the truss members do not move out of their 2D plane. ‘When designing 2D joints the limiting factors are the strength of the transverse connecting pin and the height of the combined connecting members at the intersection point. To avoid eccentricity in the 2D structure, causing it to break the joint or move out of the 2D plane, a symmetrical attachment has to be used. ‘Symmetrical connections will provide a balanced node without eccentricity. Nodes should be designed to be stable, meaning 2D joints always need to be connected in the intersecting point of the truss members. Even when three members are connected to such an unstable node, it becomes unstable and unable to be placed. ina kinetic structure. Figure 128: 2D Turning Joint, without eccentricity Figure 129: 20 unstable joint 96 43.2.2 3D Joint Different 3D nodes were modeled in this research. When discussing these nodes there is a substantial difference between the connection between two members, ‘which can be modeled ideally with an intersection point, and the connection between more than two members. When connecting more than two members which are able to move ima large angle relative to the node this research states that this is only possible when using a multiple ball joint node. ‘The universal joint or Cardan-joint consists of two symmetrical pins providing excellent behavior without eccentricity and still placing the node center directly at ‘the intersection point of the two connecting members. During the 3D printing of this node the first model broke at the center cross. This was ‘mainly due to the lack of mass for strength and stiffness but also because of the characteristics of the 3D printer itself. 3D printing works by layering plastic string on top of each other. In the perpendicular direction of the layers the model is relatively fragile, which caused it to slide/break from its form. Solutions are strengthening the ‘model but also printing the model in accordance to its layering, avoiding crucial points in the perpendicular direction of the printer. Figure 130: 3D-print Universal Joint 2 7 ‘The concept ball joint was designed by the writer: [t was formed by placing ball joints in cardinal directions ofa larger ball-form node. When this node is connected by more than three truss members it is fixed in3D space and able to fulfill its role in Kinetic Architecture. Also the dimension of the total node has to be multiple times smaller than the order of length of the truss members. Further research and FE-analysis of the ball subjected to push and pull forces will show crucial breaking points in the model, providing the design more optimal forms. A crucial point would be the outer rim which keeps the ball joint in place when pull forces are introduced. Thanks to multiple bail joints placed in this outer rim, the mass is minimal at those points and could lead to failure/breaking of the outer layer. Push {forces do not seem to cause a problem since sufficient mass is available in the body of the node. Figure 131: 3D printed Ball Joint, Section 43.3 Members ‘Truss members used in a 3D structure, connected by a 2D or 3D joint, have to be strengthened in perpendicular directions, to provide stiffness in all directions. These parts can be easily constructed by inserting laser-cut parts into each other. By enlarging interconnecting parts by half of the cutting-line thickness, they fit into each other providing a suitable connection without extra material. This method was also used when producing the final prototype. Figure 132: Truss Memb 98 43.4 Cladding Cladding of kinetic frame structures is a necessary step for making our newly designed spaces viable. At the time of writing, elastic cladding such as membranes can already be insulated, even providing different densities when stretched to allow more or less light into the space. Except for the connection details of membranes their implementation is relatively easy by using stiffened metal rings in reinforced parts of the membrane. Stiff parts however can also be tsed as cladding and could also be insulated like sandwich panels consisting of two layers of metal glued to 2 stiff insulation. However ‘when designing a cladding system ona kinetic structure, in which a cladding plane ‘moves out of its original plane, necessary steps have to be taken: Knuckle joints, embedded in the upper or lower truss element which attach the cladding isostatically allow the necessary degree of freedom. The necessary freedom of the connection point has to be researched and simulated beforehand. The lower figure depicts the isostatic connection of the stiff cladding. Knuckle joints however have not been implemented in the physical test but have the same physical appearance of a nut and bolt, only attached to a ball joint built in the beam to which the cladding is connected. The exact degree of freedom should be simulated using the necessary software. 3 igure 133: Kinetic Box, Seen: Figure 134: Expansion, Wooden cladding 99 43.5 Actuator design Actuators are available in all sorts of types, driving mechanisms or driving forces. In this thesis linear actuators were researched, whereas this does not imply that all KA has to be actuated by linear actuators, The implementation of rolling guides in the form of lines and planes will allow more comprehensive structures. Muscle wire, the smallest in the family of linear actuators, is not suitable for KA because of its relatively small strength, unsuitable for building loads, and its, dependence on physical factors such as external heating. Rotational actuators are widely available in different sizes, available strength/torque and speed. A number of basic mechanisms to transform the rotational movement into linear movement were introduced and built to measure the practical elongation rate of these mechanisms. Different mechanisms were used but further research could improve or design new mechanisms which allow bigger actuation rates where needed. Pneumatic and hydraulic muscles are the final types of linear actuators, proving to be very easy to produce and capable of carrying relatively large loads in tension. This ability is thanks to the setup of the muscle. The McKibben air muscle consists of a silicone tubing which deforms under air pressure and a strong braided outer sleeving providing tension strength. Obviously the muscle is unable to resist push forces, and therefore cannot be used as such in a kinetic structure. The braided sleeving is also necessary in the setup of the McKibben air muscle to avoid local deformations of the silicone tubing causing the muscle to fail Despite this strength the muscle can only contract up to 10% of its original length, whereas rotational/ linear actuators can contract up to 48% and can resist push and ull forces according to the strength of the electrical motor. Eimer Gear Pinion Screw 23) Pneumatic muscle 11 Tabel Figure 135: McKibben air muscle setup 100 By looking at the test results of the McKibben air muscle this research recognizes the fact that the air muscle has a variable elongation rate according to the applied load. There even ‘Seems to be a maximum value at 30N. This effect was not examined further but is likely to be caused by different material strengths and properties lke strenath of the specific braided sleeve and silicone tubing. Additionally parameters such as the wall thickness, length or diameter of the silicone tubing used could lead to these findings. Further research, documentation and examination of material strength should provide conclusive insights. Elongation[/] vs. Air Pressure[bar] tks mks a 3ks asks se tkg 0,00 0,50 1,00 1,50 2,00 2,50 3,00 3,50 4,00 4,50 5,00 Air Pressure (bar] Figure 136: Elongation/Original Length [%] ‘Table 17: Muscle Elongation [%] 101 43.6 Prototype Design Design & Polyp The first conceptual design was a single actuator that could actuate the first layer of. interactivity. This first layer was defined as a structural frame together with the polyps and still needing the support of a load-bearing structure (layer 2) which was actuated and controlled by other actuators and data, When designing the actuator the designer quickly came across different physical complications and issues which could not be overcome. ‘The first actuator form also did not perform optimally regarding internal forces. Its physical design of elongating at one end and staying fixed at the other end made its form not optimal involving bending moment introduced at the middle and at the connection point at the connection between layer one and layer two, which were designed separately in this preliminary idea. Besides the physical design, the electrical motor that should have been implemented ‘was also not strong enough to carry the introduced loads of the entire skin structure. Instead of over-dimensioning one electrical motor, the method of dividing the layer into two structurally separate layers was the obvious choice. ‘This was the confirmation but also the practical insight that flowed out of the design: that different layers of interactivity could not be actuated by the same actuators. Layers can be aesthetically interwoven but cannot be actuated by the same mechanism. Figure 137: Sketches Polyp, Design 1 102 ‘The second and final design used this insight and decoupled both of the layers. The second layer now provides structural stability and the polyps provides the first layer of interactivity, allowing daylight entrance and/or ventilation, Both layers are interwoven but actuators work completely independently and can also be dimensioned only on their layer. Servo's with low torque could then be used for the polyps, on which no extemal loads are being placed. The shell of the polyps acts as a truss member which provides stability and necessary strength towards the middle for the occurring bending moment and normal forces. The separate polyp arms are supposed to disperse from the center when fully extended. This was supposed to be possible by attaching plastic strings to the ends to ‘work as a spring holding them back. This implementation proved to be very difficult and caused the arms to break when tensioned to hard. Another possibility is to design the polyp itself to spread the polyps arms when fully extended, by laser- cutting the holes for the arms at the yoke closer to each other. Figure 138: Polyp Actuator Figure 139: Prototype Perspective View 103 ‘Truss & Cladding When designing a structure which is kinetic, a suitable concept for the cladding is needed in the earlier phases of the design, The design of the cladding will have implications on the structural frame below. One basic insight was gained in this research. Which is that only triangulated surfaces are able to move together with nodal displacements and can thus forma basic idea to begin designing structural cladded frames. In the production of this prototype the cladding was designed after the structural frame. Like shown in the prototype the rectangular cladding pieces, based on the rectangular frame, were cut up after their production. Meaning the triangulation that was necessary for nodal movement was not thought of in the design phases. By using this insight to redevelop the same flexible skin, the use of a triangulated frame would have been more optimal. Cladding surfaces could also not be attached to nodal points. The preliminary concept was to attach the cladding surfaces to the nodes and not to the truss members with the use of simple hinges which are able to move in one direction, However when one of two neighboring nodes (triangulation) changes its height, the node in question and its connecting cladding moves in a direction other than that made possible by the hinge. It is for this reason that the connection of the cladding moved to the truss members instead of the nodes. 104 Vacuum forming is a great way of introducing 3D form into a simple plastic sheet. This stiffens the plastic sheet, combining strength and low own weight. When using the vacuum forming process, the setup needs to be of sufficient dimensions to provide cladding with the appropriate dimensions. Cladding which is had internal seams lead to weak points in the cladding design. Also molds which are used in vacuum forming should be chosen wisely. A higher ‘mold will utilize more plastic, causing the plastic to loosen from its frame, eliminating the vacuum effect underneath. When large, meaning high, molds are used in the vacuum forming process it best that a setup is used where an movable heating element is used to place over the vacuum former. This way the vacuum forming process will be slower but successful in using a higher mold. The original plate aiso has to be dimensioned to allow the higher deformation. Figure 141: Vacuum Forming results with high mold 105 Behavior Behavior was programmed using the Arduino programming language®, which is based on Processing, During development and end use it became obvious that smoothing methods have to be applied on incoming sensor data, Pragmatic data such as proximity, delivered by an IR (beam) sensor, introduces, thanks to the nature of its sensor design, a particular noise onto the data stream. This has to be smoothed out to overcome the effect of “twitching” actuation. The used smoothing function sums up an amount of measurements with a delay and divides them by their amount. Causing the data stream to be flattened out to reduce major spikes. The servo motors used (Hextronik HX550) moved in an arbitrary direction every time the programming was uploaded to the Arduino board. This twitch caused the servo to move out of the range, enabled by the mechanism, causing failure of the lever connecting the motor with the mechanism This however was overcome when moving the servo to a known in-range position at the time of declaration in the programming. The used Motor library used in the Arduino programming environment is also only able to actuate 8 servo motors. B: Behavior Arduino code * Append 106 When uploading the first tests of the behavior to the Arduino board, the servo motors began to twitch. After further research the Arduino programming began to restart following the actuation of the servo motors. This was due to the fact that the Arduino board connected to a USB port was not able to supply the necessary power to actuate eight servo motors. An additional 9V battery was attached to the Arduino board to successfully actuate every servo motor: This was thus not caused by faulty programming. Accurate measurements of the energy consumption showed that 8 servo motors during actuation could consume up to 2000 mA. The USB port powering the Arduino board was only delivering about 500 mA. The extra battery or external power supply with the necessary specifications was able to deliver that amount of current. Figure 142: Arduino powered by USB and SV battery 107 Chapter 5: Conclusion ‘This research concludes with the summarization of different results and practical insights involving Kinetic Architecture, modeling design and behavior. 5.1 Simulating Design ‘Kangaroo and Grasshopper are rather new software packages and not yet fully incorporated in the use of architecture design. Like stated in the literature study, evenin the specific field of Interactive Architecture, simulation packages are not, widely used. Simulation tools do not exist because Interactive Architecture exist but parametrical tools which are continuously developed together with new projects are no longer bound to that specific design. New tools therefore stimulate and positively reinforce new structures during their design. In this research a case study research showed that Kangaroo, the used component for simulating physics, is a promising tool, When designing structures, which are anchored by points, Kangaroo is able to simulate kinetics, a primary demand. The biggest issue involving this component is the current inability to implement rolling gutides or anchor points. Meaning other than anchor points fixed in space, Kangaroo is not yet able to implement them, Kangaroo is still being developed at the time of writing however and these components have been announced. Further research could be the design of multiple component packages based on a specific design to implement and publish for the use in Grasshopper Involving accuracy, it can be concluded that Kangaroo is able to accurately simulate structural behavior of non-actuated structures. Averaged error margins in comparison to ANSYS, a Finite Elements package, are as low as 0,1%. Error margins of the comparison between actuated structures however are as high as 316,2%. This research is not able to accurately argument this large error margin. It states however that ANSYS is not designed to actuate members during its calculations. An initial strain was applied to the beam in ANSYS which was the only possible method. Compared to Kangaroo however where actuation of a beam is implemented by adjusting the rest length of the stiff spring depicting the beam. 108 ‘The methods of implementing the actuation together with the linear calculation by ANSYS and a non-linear, iterative, calculation method by Kangaroo are the main issues to be further researched to validate the numerical accuracy of the simulation package Kangaroo and account for the numerical mismatch. ‘Table 18: Error Margin %, Default SEN ie 19:Error Margin %, SkN Figure 143: Simulation Grasshopper, OkN, Actuated, Deformation Seale 1:1 Ze 17 of —S1e__Nis. Figure 144: Simulation ANSYS, OkN, Actuated, Deformation Seale 4:1 109 5.2 Simulating Behavior. Besides being able to simulate kinetics, Grasshopper has the ability to introduce real. life sensor data to the programming environment thanks to the component Firefly. Being able to implement these streams of data, the designer can begin to design the intelligence which is a crucial part in the translation of data into actuation of the Kinetic structure. Basic techniques which are already implemented in Grasshopper are the remapping function and the Optimization solver, Galapagos. The remapping function allows to easily remap the value in the sensor range to the actuation range. The sensor range is dependent on the specific sensor and post processing, The actuation range is specified by the used actuator driving the structure, Remapping functions however ead to a swarm like intelligence not taking the whole structure into account. ‘The Galapagos Optimization solver however is able to optimize a single value, ‘maximally or minimally, based on every actuator length. Techniques which prove to be valuable in further designs and research are position towards a point (displacement or attractor) and the sum product of beam length and internal force. Structures could be actuated to minimize internal forces or minimize displacements. Again further implementation and programming of components specifically for architectural design and data manipulation is a possibility. This research also remarks the fact that the optimization solver is an evolutionary. type solver, meaning every iteration process will lead to another actuation. However seeing compared results the end value is always the same, leading to an optimal actuation, independent from the chosen actuation values. ra Se = SITE) e 0.565594 0.565594 0.565594 __0,565692 Difference ° ° 0986-05, Figure 145: Central intelligence versus Swarm intelligence, Simulation 1 110 5.3 Practical Issues This research acknowledges the use of new fabrication techniques such as 3D printing and laser cutting, And states that seeing the overall fabrication process of current project landscape and its own prototype, that these techniques forma valuable source of inspiration and require a thorough knowledge to successfully construct a design. 5.3.1 Joints A variety of joints was first researched involving the difference between 2D and 3D joints. To conclude this thesis states that nodes in a 2D plane are easy to fabricate to account for eccentricity and degrees of freedom. 3D joints however are relatively difficult to construct. ABD joint connecting 2 members is mainly due to the universal joint mechanism or ball joint. The optimal node point then still lies in the intersecting point and allows a large degree of freedom. The requirement for the intersection point is necessary because otherwise the node becomes unstable and cannot be implemented in a Kinetic structure. When connecting more than 2 members however the ideal node point cannot be built while maintaining a large degree of freedom. This research designed a node incorporated with a ball joint per attached truss members. This allows a large degree of freedom. The intersection point does not fall in the same point, making the node unstable. However when attaching a minimum of 3 truss members in $D space this node is still stable and can be used ina kinetic structure, Attaching 1 or 2 members makes the node unusable for use in kinetic structures. ‘This ball joint has not been researched using a Finite Elements calculation involving its resistance to push and pull forces. This is however an important point in nodal design, to recognize and analyze weak points in the structure. The developed ball joint could be examined in further research acknowledging its weak points and adjusting the node design maintaining a large degree of freedom. Figure 146: 30 printed Multiple member Ball joint a 5.3.2 Cladding Cladding is an important part of making our interactive spaces viable. Systems include flexible membranes and stiff materials, Flexible membranes are not thoroughly researched in this research but states that membranes have the ability to introduce sunlight into spaces using their stretch density and the fact that they can insulted using Nano gel technology embedded in a composite membrane. Further tools or insights could be taken care of in further research. Stiff cladding like composite panels are already widely available. These could also be used in Kinetic Architecture, However when attached to a kinetic structure, the cladding has to be attached isostatic, When attached hyperstatic movernent of the structure out of the cladding plane will cause the attached cladding or connection to fail and break. Further research into proper sealants for the air tightness of these panels as well as insulation research will prove valuable in improving the viability of Interactive Space. z : Kinetic Box, Scenario 2 Figure 148: Expansion joint Wooden cladding 112 5.3.3 Actuator Design Actuators are the means of our Kinetic Architecture. According to the literature study a variety of actuators, which not even be categorized because of their numbers, are available. All of these with different characteristics or limiting factors. This research looked into muscle wires, rotational actuators and pneumatically driven actuators. Muscle wires were researched but discarded for the use in Kinetic Architecture because of their limited strength and large dependence on environmental factors such as temperature, Muscle wires therefore did not fit in the use for Kinetic Architecture, Rotational actuators, in this research small servo motors, are available in different applications. To transform the rotational actuation into linear actuation for our Kinetic structures different possibilities or mechanisms occur. This research has built, 3 different mechanisms with rotational actuators: a rack pinion system, a crankshaft mechanism and a screw mechanism. This research states that these actuators are limited by their available torque, depending on the type, and their elongation/shrink rate (stroke). Since actuators used were of the same torque strength, different shrink rates could be calculated and compared to each other. Further research in cooperation with mechanical engineers could provide new insights involving actuators and their behavior in real scale applications. Screw 28) Pneumatic muscle Zz 11) Figure 149: Linear actuators, Shrink Rate 113 For this research a McKibben air muscle was also fabricated to test its elongation rate versus applied load. When introducing a variable pressure range of 0 to 5 bar the muscle contracted, Different total elongation rates were found, depicting the accuracy of the setup or other issues that might influence the end results. ‘These issues are the strength of the braided sleeve, a component of the muscle, as well as the dimensions of the silicone tubing. Further research could research this specific air muscle or design and validate different new mechanisms for linear or rotational actuation. Druk(Bar) Bl 1 kg 0,00 0,35 0,50, 1,00 1,50 a Bask Figure 150: Air muscle, Elongation rate [%] 118 5.3.4 Prototype ‘The produced prototype is inspired by the current landscape of IA, KA and foremost nature. Starting from a general idea ofa structural skin being able to form its own shape and inner space this research designed a prototype. A prototype based on the typology of underwater polyps residing in corals this prototype combines direct interactivity, response to daylight, proximity and vibration, with a layer of structural stability and interactivity on medium long timeframes, ‘The first layer of interactivity, directly linked to the biomimicry of underwater polyps, offers the flexible skin with means of introducing light and ventilation in its inner space. Besides these physical needs of this inner space, the polyps are able to detect proximity or touch, passing it through the skin making its inhabitants aware of presence. Inhabitants are no longer shielded from outdoor factors but are now made aware of them using this Interactive Architecture. The production of the polyps made way for insights in making as well as designing actuators. Structural internal forces were taken into account of the polyps which produced a simple but efficient and elegant actuator design. 15 » Figure 153: Prototype Side view Even if the structural layer was not implemented in this prototype the idea remains ‘the same. A structural frame is able to react to sets of data to deform according to spatial needs or structural optimization. The basic structure consisting of a ‘rectangular frame was simulated and fabricated. Actuators with different lengths were fabricated as members with a fixed length. By doing this, the prototype did not show its full potential but does show its true use and meaning of different layers of interactivity in a real timeframe. ‘The second layer is based on slower structural change. Changes in wind patterns due to new buildings or the blossom of trees in the vicinity will cause the environmental applied forces to shift. Altering the conditions of this interactive equation. The structure will then respond according to its intelligence ané change its shape ina period of days or weeks and not minutes or seconds. ‘The design and construction of this prototype was conceived as a response to the current Interactive and Kinetic architecture landscape where structural interactivity and direct interactivity follow their own independent path. This prototype suggest that these two different layers can be woven into each other and do not have to be strictly divided as in the current project landscape. Further research and designs can develop more elaborate and more intelligent structures which extend these notions into the field of practical, usable and built architecture, 17 Appendices 118 Appendix A err] Bearing Err crc) ce ETC’ Pcs us cad roy Poe Jaartal rene ors ro structural capacity Bending atnode Both Both Bending Motor Direct M Building Embedde d Activity Climate) 2008 David Fisher Arab Emirates rotating tower Architect ure structural capacity Linkage None- Physics None Weight Direct S- interior Embedde d Design (Interior) 2008 Spectacul ar United States ery Module Architect ure structural capacity Linkage Both Pragmatic Bending Motor Direct Dynamic Systems Activi Climate) 2008 ‘Associate 5 United States Emergent Surface Architect ure: structural ‘capacity Bending at node Head Pragmatic Bending Motor Direct uM Building Dynamic systems Activity imate) 2003 Gang Architects United States Starlight ‘Theater Architect ure structural capacity Truss Head Pragmatic Musdes uM Building Embedde d Activity Climate) 2006 Group mir United States ‘WhoWha tWhenAir Architect ure structural capacity Truss Swarm Pragmatic « Cilinders Direct ue ing Dynamic systems Activity imate) 2008 Group wir United States e Skylichts Architect ure structural capacity ‘Antagonis t Swarm Pragmatic sma, Dynamic Systems Activity Climate) 2008 Orange void United Kingdom Pixel Skin 2 | architect ure structural capacity Linkage Pheumatic Citinders, Direct M- Building Embedded Activity Climate) 2008 ‘Anshuman, Orangevoid United Kingdom Bs seek. Interactive Architecture structural capacity Linkage © Cilinders Direct Mm Building Dynamic ‘Systems. Activity Climate) 2008 WHITEvoi d Germany iare Fl Architect ure 119 structural capacity Postitioni ne Head Both c Cilinders Direct S- interior Dynamic Systems Design (interior) 2007 Pezshkpo United States Environm ent Architect ure structural capacity Both Various Various Direct Belgium rable House Architect ure structural capacity linkage Pragmatic Bending Motor Direct mM Building Dynamic Systems Activity Climate) 2007 Morphosi United States Federal Building Architect ure structural capacity Linkage Both Pragmatic Motor Direct mM Building Dynamic Systems 2009 Foster+Pa rtners United Kingdom of justive Shading Architect ure structural capacity Linkage Swarm Pragmatic SMA Direct S- interior Embedde States Glass, ure structural capacity Head Humanist Dynamic Systems Activity Climate) 2008 ‘Smart Studio Germany RemoteH ome Architect ure structural capacity linkage Swarm Pragmatic Servo Direct S- interior Dynamic Systems Activity Climate) 2005 rde studios Netherlan ds 40 Pixel Architect ure structural structural capacity capacity linkage Linkage Swarm Swarm Pragmatic Both SMA, Fans Muscles Direct Direct S-interior Dynamic Systems Activity ate) 2005 2008 Frederic Philip fy Beesley United Germany States Hylozoic Aperture Grounds Architect Architect ure ure structural capacity Pheumati c Both Both Preumati e Direct S-interior Dynamic Systems Optimizat jon 2010 Michael Fox United States Bubbles Architect, ure: structural capacity Antagonis t Swarm Pragmatic light Direct S- interior Dynamic Systems Activity Climate) 2008 rde Sstudios Netherlan, ds Dune 4.0 Architect ure 120 structural capacity Preumati « Head Pragmatic Muscles Direct S- interior Embedde d Optimizat ion 2003 Oosterhui Netherlan ds Musde w number 1 structural capacity Postitioni 8 Head Pragmatic Muscles Direct S-interior Embedde d Optimizat ion 2005 Oosterhui Netherlan ds Reconfigu red tw number 1 structural capacity Postitioni ne Head Pragmatic Muscles Direct S-interior Embedde d Optimizat ion 2005 Oosterhui Netherlan. ds Muscle Body mw number 2. structural | structural capacity capacity Truss Truss Head Head Pragmatic Pragmatic Muscles Musdes Direct Direct mM ue Building Embedde Embedde 2007 2008 Oosterhui Oosterhui Netherlan Netherlan ds ds Musde Muscle Tower Tower 2 rn w number 2 number 2 structural capacity Postitioni ng Swarm Pragmatic Muscles Direct S-interior Embedde d Optimizat 2007 Oosterhui Netherlan ds Muscle Space rn number 2 structural capacity Truss Head Pragmatic Muscles Direet S-interior Embedde d Activity vate) 2004 Festo Netherlan ds Interactiv e Wall w number 2 structural capacity linkage Head Pragmatic Building Embedde 4 Activity Climate) 1995 Santiago Calatrava Spain Plaza & Fountain Design Group structural capacity Antagonis t Head Pragmatic Direct Deployabl e Activity Climate} 1976 Otto Frei Germany structural capacity Nurnberg, scissors Head Pragmatic Direct ue Building Embedde d Activity ate) 1998 Associate United States Hoberma Sphere Design Group structural capacity Postitioni ng Head Pragmatic Direct uM Building Embedde d Activity Climate) 1976 Otto Frei Germany Hoechst Stadium Design Group a2. structural structural capacity capacity Bending atnode Linkage Head Head Pragmatic Pragmatic Direct Direct uM. mM Building Building Embedde Embedde d Activity Climate) TL Happold United Spain States Music Pavilion Design Group Group structural capacity Numberg scissors Head Pragmatic Direct mM Building Embedde d Activity Climate) 1992 K Pallares, United Kingdom Deployab! e Sche Design Group structural capacity Antagonis t Head Pragmatic Direct S- interior Dynamic systems Optimizat ion 2004 Group mir United States tic wall Design Group structural capacity Bending at node Head Pragmatic Santiago Calatrava Switzerta nd Floati Pavilion Design Group structural capacity Numberg scissors Head Pragmatic ing Embedde d Activity Climate) 1998 Associate 5 United States Iris Dome Design Group structural capacity Linkage Head Pragmatic Direct mM Building Embedde d Activity Climate) 1976 Otto Frei France etic Canopies Design Group structural capacity Bending atnode Head Pragmatic Embedde d Activity Climate] 1995 Piano Renzo United States BM Pavilion Design Group 122 Appendix B Jude #include // Declaration! Servo myservol; // create servo object to control a servo //amaximum of eight servo objects can be created Servo myserve2; // create servo object to control a servo //a maximum of eight servo objects can be created Servo myservo3; // create servo object to control a servo //a maximum of eight servo objects can be creat Servo myservo4; // create servo object to control a servo //a maximum of eight servo objects can be created Servo myservoS; // create servo object to control a servo //a maximum of eight servo objects can be created Servo myservo6; // create servo object to control a servo // a maximum of eight servo objects can be creat Servo myservo7; // create servo object to control a servo //a maximum of eight servo objects can be created Servo myservo8; // create servo object to control a servo //a maximum of eight servo objects can be created int sensorPinLight = 0; int sensorPinlR = 2; int sensorPinKnock = 5; int sensorPinButton = 13; int pos= 30; boolean Knocking = false; int counter Button Knop = Button(sensorPinButton,PULLDOWN); 123 // Initialization. void setup() { myservoLattach(12); // attaches the servo on pin 12 to the servo object myservoLwrite(pos); myservo2.attach(3); // attaches the servo on pin 3 to the servo object myservo2.write(pos); myservo3.attach(4); // attaches the servo on pin 4 to the servo object myservo3.write(pos); myservo4.attach(5); // attaches the servo on pin 5 to the servo object myservo4.write(pos); myservoS.attach(6); // attaches the servo on pin 6 to the servo object myservoS.write(pos); myservo6attach(7); // attaches the servo on pin? to the servo object myservo6.write(pos); myservo7.attach(8}; // attaches the servo on pin 8 to the servo object, myservo7write(pos); myservo8.attach(10); // attaches the servo on pin 10 to the servo object miyservo8.write(pos); pinMode(sensorPinLight, INPUT}; pinMode(sensorPiniR, INPUT); SeriaLbegin(9600); // prints title with ending line break Serial_printin("Program Starting"); 128 Joop! void loop() { int ValueLight = map(irRead (sensorP inLight, 10),0,900,30,160); //NalueLight = map(analogRead (sensor? inLight),0,900,30,160); float distance = 12343.85 * pow(irRead (sensorP inIR,10),-1.15); //Moat distance = 12343.85 * pow(analogRead(sensorP infR),-1.15); //Moat distance = irRead (sensorPinIR,10); int ValuelR = map(distance,0,100,30,160); int ValueKnock = irRead(sensorPinKnock,10); if (ValueKnock >= 50) {Knocking = true;} ValueLight = constrain(ValueLight, 30, 164 ValuelR = constrain(ValuelR, 30, 160); if (Knop.uniquePress()){ counter + +; if(counter == 3){count Serial printin("Program:"); Serial printin(counter); t if (counter == 0){ Serial printin(ValueLight); ‘WriteAll(ValueLight); + if (counter == 1) { Serial printin(ValuelR); //Serial.printin (distance); ‘WriteAll(ValuelR); } if (counter ==2){ Serial printin(ValueKnock)}; if (Knocking == true) {WriteAll(170); delay (10000); Sweepall(170,30); Knocking = false;} + delay(100); + int irRead(int readPin, int amount) t int haliPeriod = 13; //one period at 38.5khZ is aproximately 26 microseconds int cycles = amount; //26 microseconds * 38 is more or less 1 millisecond inti; int total = 0; for (i { int interval = analogRead(readPin); total = total + interval, delay(halfPeriod); } return (total cycles); q void WriteAll(int value) { myservoLwrite(value); J jAelay(1000}; myservo2.write(value); //Aelay(1000); myservo3.write(value); J /delay(1000); myservo4.write(value)}; //Aelay(1000); myservo5.write(value); //Aelay(1000); myservo6.write(value}; //Aelay(1000); myservo7.write(value}; //Aelay(1000); myservo8.write(value}; / Aelay(1000); } void SweepAll(int start, int einde) { if (start > einde) { for( int value = start; value > einde; value--) // goes from 0 degrees to 180 degrees { ‘J/ in steps of 1 degree Writeall(value); // tell servo to go to position in variable ‘pos’ delay(15); // waits Sms for the servo to reach the position H if (einde > start) { for( int value = start; value < einde; value! +) // goes from 0 degrees to 180 degrees { // in steps of 1 degree Writeall(value); //tell servo to go to position in variable ‘pos’ delay(15); // waits 15ms for the servo to reach the position M 126 Bibliography Adaptive Building Initiative, 2006, Adaptive Building Initiative. Strara. Available at: http:/www.adaptivebuildings. com/strata-surface html [Accessed June 6, 2011} Andy Payne, Matt Bell & Elise Elsacker, 2010. Kinect & Grasshopper. Home. Available at: http //ghkinect blogspot. com/201 1/03/kinect-grasshopper-htm! [Accessed May 20, 2011] Andy, P. & Johnson, J., 2010. Firefly. Home. Available http://www firefly experiments. con [Accessed May 7, 2011]. Arduino, 2005, Arduino. Homepage. Available at; http:/Avww.arduino.ce/ [Accessed May 6, 2011] Amall, T., Knutsen, J. & Martinussen, E,S., 2011, Immaterials: Light painting WiFi ‘on vimeo. Available at: http://vimeo.com/20412632 [Accessed May 2 2011]. Birdair, Birdair Tensotherm. Homepage. Available at htip:/vww.tensothermroofing, conv [Accessed May 23, 2011] Bullivant, L., 2007. ddsocial: Interactive Design Environments, Wiley Bullivant, L., 2005. 4dspace: Interactive Architecture 1st ed., Academy Press. Calatrava, S., Santiago Calatrava : Homepage. Available at: hitp:/Avww.calatrava.com/# [Accessed May 27, 2011] D'Estree Sterk, T., 2003. The Office for Robotic Architectural Media & Bureau for Responsive Architecture. Homepage. Available at: hutp:/Avww.orambra.cony [Accessed May 2, 2011]. Daerden, F. & Lefeber, D., Pneumatic Artifical Muscles: actuators for robotics and ‘automation, Vrije Universiteit Brussel. Davidson, S., Grasshopper - generative modelling for Rhino. Home. Available at http:/Avww.grasshopper3d.com/ [Accessed May 2, 2011] 127 Festo Corporate, 2009. Festo Corporate. Interactive Wall. Available at http://www festo.com/cms/en_corp/9776 htm [Accessed May 17, 2011] Fotiadou, A., 2007. Analysis of Design Support for Kinetic Structures, TU Vienna, Fox, M., 1995a, Kinetic Design Group. Homepage. Available at: tp:/www-robotecture.convkdg/project.html [Accessed May 8, 2011]. Fox, M., 1995b. Kinetic Design Group. Kinetic Matrix. Available at http://www robotecture.comv/kdg/Matrix/matrix.html [Accessed May 6, 2011], Fox, M. & Kemp, M., 2009. interactive Architecture 1st ed., Princeton Architectural Press. Future Cities Lab, 2010. future cities lab. Xeromax Envelop(e). Available at: http:/Avww future-cities-lab.net/index/?cat=71 [Accessed May 28, 2011] Haque, U., 2006. haque :: design + research, Burble. Available at: http://www haque.co.uk/burble php [Accessed May 17, 2011] Haque, U., Pachube - data infrastructure for the Internet of Things. Homepage. Available at: http:/Avww.pachube.cony [Accessed May 27, 2011] Haque, U. & Somlai-Fischer, A., 2005, Lowtech Sensors and Actuators. Available at: http://lowtech. propositions. org.uk’ [Accessed May 27, 2011]. Hoberman, C., 1997, Hoberman Associates, Expanding Fabric Dome. Available at: http:/Avww hoberman, com/portfolio/pompidou, php 2myNum=7Smytext-Ex panding+Fabric+ Domedemyrollovertext=%3Cu%3EExpanding+Fabric+Do me%3C%2Fu%SEScategory=&projectname=Expanding+Fabrie*Dome [Accessed May 17, 2011} Kejia Industry, KEJIA INDUSTRY. Bull Joint. Available at: http://www kejiaplasticmould.com/display.asp?id=31 [Accessed May 28, 2011]. Kilian, A. et al., 2006. WhoWhatWhenlR. Available at: htp:/‘musclesfrombrussels. blogspot. comy [Accessed May 7, 2011] Lab[au}, 2010, LAbfau]. Tessalate. Available at; hitp://lab-au,conv#/projects/tessel/ [Accessed May 17, 2011] ‘Mechanisms 101, Mechanisms 101. Scotch Yoke. Available at: http:/Avww.mekanizmalar.com/scotch_yoke.html [Accessed May 28, 2011]. 128 Van Mele, T., 2008. Scissor-Hinged Membrane Structures - a system for retractable roofs, Vrije Universiteit Brussel. Available at hitp:/Awww.vub.ac.be/ARCH/ae-lab/people.php *researcher=Van%20Mele [Accessed May 28, 2011]. Merali, R. & Long, D., 2009. Actuated Responsive Truss, University of Toronto. Mitchell, W.J., 1996. City of Bits: Space, Place, and the Infobahn, The MIT Press. Mitchell, W-J., 2000. e-topia: Urban Life, Jim-But Not As We Know It, The MIT Press. Mitchell, W.J., 2004. Me++: The Cyborg Self and the Networked City, The MIT Press. Mitchell, W.I., William J. Mitchell, Available at: http://web media mit.edu/~wjm/ [Accessed May 8, 2011], National Geographic, National Geographic. Swarm Intelligence. Available at: hitp://ngm.nationalgeographic.cony/2007/07/swarms/swarms-photography [Accessed May 20, 201 1a]. National Geographic, National Geographic. Coral Photos, Coral Wallpapers. Available at: hitp://photography.nationalgeographic.com/photography/photos/pattems- coral/#/coral-polyps-henry_1387_600x450,jpg [Accessed May 28, 201 1b] Octatube, 2010, Octatube Intemational bv. Projects. Available at: http:/Avww. octatube.nl/index php%id=5676 [Accessed May 23, 2011] Oosterhuis, K., 2000. Hyperbody. Homepage. Available at: http:/Avww-hyperbody.nl/ [Accessed May 7, 2011]. Oosterhuis, K. & Xia, X., 2007. Interactive Architecture: No. 1, Ram Distribution. Oosterhuis, K& Xia, X., 2008. Interactive Architecture: No. 2, Episode Publishers. Oosterhuis, K. & Xia, X., 2010. Interactive Architecture: No. 3, Ram Distribution. Orangevoid, ORANGEVOID. Homepage. Available at: http:/Avww. orangevoid.com/ [Accessed May 7, 2011] Piker, D., 2011a. Kangaroo bending validation at Vimeo, Available at http://vimeo.com/20287194 [Accessed May 12, 2011] Piker, D., 2011b. Kangaroo Group. Homepage. Available at: hitp:/Avww.grasshopper3d.com/group/kangaroo [Accessed May 2, 2011] Piker, D., 2011c, Kangaroo Manual, Available at: 129 https://docs google.com! View?id=ddpv99dx_44f88¢75th [Accessed May 2, 2011]. reacTIVision, reacTIVision. Homepage. Available at: hitp://reactivision. sourceforge net/ [Accessed May 30, 2011]. Roosegaarde, D., 2011, Studio Roosegaarde. Projects. Available at: htp:/www.studioroosegaarde.net/projects/tdune [Accessed May O11) The Living, The Living New York. Living Glass. Available at: http:/Avww.thelivingnewyork. conv/Ia/Ig1 6.htm [Accessed May 20, 2011] ‘Tzonis, A. & Lefaivre, L., 1997. Santiago Calatrava's Creative Process 1st ed., Birkhauser Architecture. Wikipedia, 2011a. Wikipedia, the free encyclopedia. Linkage (mechanical), Available at: http //en.wikipedia.org/wiki/Linkage (mechanical) [Accessed May 23, 2011]. Wikipedia, 2011b. Wikipedia, the free encyclopedia, Biomimicry. Available at: http://en.wikipedia.org/wiki/Biomimicry [Accessed May 28, 2011] Wikipedia, 2011. Wikipedia, the free encyclopedia. Vacuum forming. Available at: http: //en.wikipedia.org/wiki/Vacuum_forming [Accessed May 28, 2011] 130 KULeuven Faculteit Ingenieurswetenschappen 2010-2011 Fiche Masterproef Student: Carlo Rousseeuw Titel: Kinetische Architectuur: Modelleren van ontwerp en gedrag Engelse titel: Kinetic Architecture: Modelling design and behavior ube: 72 Proefschrift voorgedragen tot het behalen van de graad van Master in de ingenieurswetenschappen: Architectuur - Bouwtechnische Optie Promotor(en: Prof, Andrew Vande Moere Assessoren: Prof. Leo Van Broeck Begeleiders: Prof. Stefaan Boeykens Korte inhoud: ‘The way we use and experience objects in our daily lives is constantly being improved with increasing user-interactivity. From our cars which are filled with sensors to enrich our driving experience to the automated shading of our windows which disappears when we need to get out of bed in the morning. Architecture today. on the other hand is static, its structural form does not interact with its users or its, changing environmental factors, Instead of shielding the inhabitants from these factors, these factors can be responded to and interacted with to change the inhabitant’s perception of this new space, Interactive Architecture. In the design of interactive architecture with structural kinetic changes, Kinetic Architecture, the simulation of a structure and its behavior plays a valuable role in its overall design and production. Being able to connect a wide range of sensor data with this design-software we cross the bridge necessary for completely simulating, interactive architecture, which in turn has an effect on the final design. Recent developments and community efforts in plugins for drawing software like Grasshopper for Rhinoceros have given us these abilities. 131 Every project is unique by its own context and usage and therefore unique by its means to interact. This thesis simulates different existing structures in the current landscapes and tests the scope of current simulation packages and their use to designers with regards to Kinetic Architecture. Also the intelligence which controls this Kinetic Architecture and the different kinds of data streams are addressed together in the context of the simulation software. Besides the research in terms of simulation, this thesis also discusses practical issues of Kinetic Structures ina general way before building a working prototype. This research will act as a catalyst to provide architects with the necessary skillset to develop and design interactive architecture but also to provide a mutual goal for other disciplines like robotics and material engineers to form and research different end products with enhanced user interactivity which could be used in this new breed of Interactive Architecture. 132

You might also like