Page 1

- Jim Rhoné - Master Project Thesis - Research . 10 - A Degree Of Freedom - JUNE14 - ENSAPM - Digital Knowledge -

A Degree of freedom Applied Research in Embedded Kinetics Material Systems and Control for Active Architecture

june14

Master Project Thesis . R10

Digital Knowledge Department

jim rhone

- Ecole Nationale Supérieure d’Architecture Paris-Malaquais - Supervisor : Philippe Morel Assistant : Pierre Cutellic -


ABSTRACT & INTRODUCTION

02

MATERIAL

06

FABRICATION PROCESS

KINETIC STRUCTURE

SIMULATION & CONTROL

POTENTIALS

COMPOSITION

10

PHYSICAL PROPERTIES

12

ELECTRICAL ACTIVATION

16

ROBOTIZED OPTIMIZATION

21

COMPONENTS DEVELOPMENT

28

PROCESS & AUTOMATION

32

WEIGHT TESTS

40

STRUCTURAL CONSIDERATIONS

43

UNIDIRECTIONAL KINETIC PROTOTYPE

52

MULTIDIRECTIONAL KINETIC PROTOTYPE

58

3D SCANNING & SIMULATION

66

FEEDBACK SYSTEMS

68

BALANCE CONTROL

72

DIGITAL FABRICATION COMPETITION

80

RESEARCH POTENTIALS

81

18

38

64

78


a degree of freefom

prologue

More recently, innovations in the field of materiality introduced the notion of responsiveness, which characterizes the property of the so-called «smart materials» to answer significantly to a stimulus, with the possibility of being controlled, measured, predicted and recorded. This notion thus led to the ideas of programmable matter and «responsive architecture», coined by the American computer scientist Nicholas Negroponte in the 1970’s to define a type of architecture that has the ability to alter its form in response to changing conditions. This evolution is not merely relative to size or motion but concerns energy and the transformation of spatial forms and material substances.

ABSTRACT

This idea enables to draw the distinction between movable and active structures. Movable structures are structures that can perform some kind of motion when an external actuator is activated. They are the buildings that «rotate, swivel, and pivot» (Chad Randl, Revolving Architecture : A History of Buildings That Rotate, Swivel, and Pivot, Princeton Architectural Press, may 2008, p. 12) and are related to a mechanistic vision of what kinetic architecture is.

Since the end of the XIXth century, architects’ spatial desires of flexibility, modularity, adaptability, modulation, variation and interactivity have progressively been envisaged via mechanic, kinetic and dynamic logics.

For instance, Gregg Lynn’s RV (Room Vehicle) prototype house, exposed at the 23rd International Biennale Interieur 2012 in Kortrijk (Belgium), is a 1/5th scale model of a residence that increases the living space by rotating in two axes on a robotic base.

The expression “Kinetic architecture” was coined by William Zuk and Roger H. Clark in the early seventies when dynamic spatial design problems were explored in mechanical systems : “form may change very slowly by evolution, moderately fast by the process of growth and decay, and very fast by internal muscular, hydraulic, or pneumatic action” (ZUK William & CLARK Roger H., Van Nostrand Reinhold Ed., New York, 1970).

Although very recent, this project remains, as far as motion is concerned, extremely mechanistic in its conception, and finally doesn’t differ that much from the aerial rotating house that the French illustrator

2


ABSTRACT

Albert Robida imagined and drew in 1883 in his book «Le Vingtième siècle» («Maison tournante aérienne»).

erhuis’ (Hyperbody TU Delf) speculative and transformable building project which is able to change shape and content in real-time and would define, according to him, «a new ecology between people and things, not as a static output, but as a dynamic one».

With active structures, the actuating principle is supposed to be much more integrated in the structure itself, even in the material system that composes it.

Indeed, in the video showing this building, the skin, as one huge membrane, transforms, but without any added mechanism, the motion seems to come from the skin itself. Is this pure speculation ?

Despite this scientific progress, the conception of kinetic responsive architecture still remains deeply anchored in a very conventional and mechanistic approach that belongs to the last century and mainly consists in using external machines, mechanisms or robots to move and transform structures.

Oosterhuis has already made various experiments addressing new approaches, testing for instance pneumatic muscles as an architectural membrane to respond to various spatial conditions.

For instance, the Aegis Hyposurface is an ornamental faceted metallic surface that has potential to deform physically as a real time response to electronic stimuli from the environment (movement, sound, light, etc.).

Using the analogy of cars headlights, he explains that embedded systems are probably the key for innovation in the field of kinetic architecture. This change has already happened in car industry, starting from external headlights for old cars and leading to fully integrated ones for recent ones; so why not in architecture ?

Developed in 1999 by Mark Goulthorpe and the dECOi office along with a large multi-disciplinary team of architects, engineers, mathematicians and computer programmers, ‘Aegis’ is driven by a bed of 896 pneumatic pistons, enabling the formation of dynamic ‘terrains’ generated as real-time calculations.

Considering this statement as a starting point, this research investigates the application of embedded form-changing material systems in order to minimise the use of intricate and high-tech mechanistic joints, actuators and control for responsive and active architecture. It focuses on the development of Soft Frame Electroactive Polymers and aims at designing and producing an experimental and advanced prototype system whose kinetics digital control offers new possibilities to envisage what we build as «machines for living».

Although enabling a very accurate control of these systems, this use of external mechanisms makes very challenging their application in architecture but also raises a new question : is there an alternative approach for responsive kinetic architecture, which actively responds to stimuli without using intricate mechanical systems ? This same question is at stake in Kas Oost-

3


a degree of freefom

prologue

INTRODUCTION

4


INTRODUCTION

Being given the choice last semester for the P9 studio «introduction to robotics», lead by P. Morel and F. Agid, to choose a topic related to digital fabrication, Martin Genet and I decided to orient our project around these ideas of kinetics, reactivity and interactivity in the field of materiality. Establishing a direct link with my thesis «Fabriquer pour concevoir, retour à la matérialité» (Fabricate to Conceive, Back to Materiality), this P9 project aimed at being an applied research focused on materiality, simulation, control and digital fabrication. With the clear will to get a dynamism embedded within the studied material itself, we quickly recentred our topic on smart materials and decided to investigate the possibilities of Soft Frame Electroactive Polymers (SFEAPs) in the kinetic architectural field.

Technology (EMPA) ), to the experimentation of new assembly approaches, and to the development of a simulation and control system. In the continuity of this 6-month project, my master thesis is an experimental research aiming at developing an architectural application of SFEAPs thanks to the knowledge gathered so far. The starting objectives were divided into four parts : - Prototyping : Developing the fabrication process of the prototypes including optimization, robotization and automation in order to improve their properties and increase their size. - Structure : Investigate the possibility of SFEAPs application in structures and mechanisms with the setting-up of a reliable material system.

Electroactive polymers are not new, and their study started in the 1880’s, when the German scientist Wilhelm Roentgen (who discovered X-Rays in 1901) created an experience in order to measure the effect of an electrical current on the mechanical properties of a rubber band. Since the early 1990s, materials scientists and engineers have been developing electroactive polymers for use as sensors, actuators, and artificial muscles.

- Simulation and control : Exploring and defining a digital simulation, control and feedback method enabling the mastering of the prototypes’ kinetics. - Architectural application : Designing an architectural project that combines the three latter : the full-scale prototype, the structural analysis, its simulation and control.

Necessitating very accurate and sophisticated production methods and tools, we essentially dedicated the P9 semester to mastering the fabrication process of SFEAPs (which was inspired by the one developed for the project “Shape-shift”, resulting from a collaboration between the chair for Computer Aided Architectural Design (ETHZ) and the Swiss Federal Laboratories for Materials Science and

5


M AT E -

R


W H AT A R E SOFT FRAME ELECTROACTIVE P O LY M E R S ?

COMPOSITION PHYSICAL PROPERTIES

RIAL

ELECTRICAL A C T I V AT I O N

-


a degree of freefom

material

Electroactive polymers are light and flexible organic components able to react to an electric stimulation by a change in their morphology. Historically, the first hard electroactive materials were discovered in 1880 by French scientists Pierre and Jacques Curie, with piezoelectricity. Electroactive polymers’ study started in the 1880’s, when the German scientist Wilhelm Roentgen (who discovered X-Rays in 1901) created an experience in order to measure the effect of an electrical current on the mechanical properties of a rubber band. The first chemically stimulated polymers were obtained by Katchalsky in 1949, but it is only since the 1990’s that electrically stimulated polymers have been explored. Nowadays, medicine and aeronautics constitute their main fields of application.

W H AT ARE SFEAPS ?

The electroactive polymers classification is divided into two categories: the ionic group and the electronic group, whose main properties are exposed in the following charts (Fig. 1 & Fig. 2) Based on high deformation criteria, the research was refocused on dielectric elastomers that belong to the electronic group; and more precisely on one of its possible application known as Soft Frame Electroactive Polymers (SFEAP). As far as physics are concerned, a dielectric material is an electrical insulator that can be polarized by an applied electric field. When a dielectric material is placed in an electric field, electric charges do not flow through the material as they do in a conductor, but only slightly shift from their average equilibrium positions, causing dielectric polarization. Because of dielectric polarization, positive charges are displaced

8


WHAT ARE SFEAPS ?

NAME

ELECTRONIC GROUP

MATERIAL EXAMPLES

material having, at an elementary domain scale, a permanent dipolar momentum (ex: piezoelectricity, electrostriction)

. piezoelectricity : PZT, PVDF . electrostriction : P(VDF-TrFE), P(VDFTrFE-CTFE)

electrets

material having a non null electrical polarization after an exposition to a high electrical field or a charge injection

. PVDF, Polypropylen (PP) or charged PTFE

dielectric polymers

material working with the principle of a variable capacity : electrostatic effect between charges

. acrylate : VHB 4910 from 3M . silicone : HS3 from Dow Corning

grafted electrostrictive elastomer

ferroelectric material electrostrictive type with the particularity to be composed of a flexible matrix and of a grafted polymer under cristal form

. matrix : copolymer of the chlorotrifluorethylen or trifluoroethylen . graft : P(VDF-TrFE)

electroactive paper

paper type material composed of a lot of particules (natural fiber) forming a network and combining piezoelctric properties with ionic migration

. cellulose layer between two metal electrodes (gold, platinum...)

ferroelectric polymers

IONIC GROUP

PHYSICAL PROPERTIES

electroviscoelastic elastomer

material composed of a silicone elastomer and with a polar phase. It behaves like an electrorheological fluid

liquid cristal elastomer

material having piezoelectric properties and that deforms as a reaction to a thermic gradient (Joule effect for an electrical activation)

ionic gel

material becoming dense (contraction) or inflated (stretching) when transfering from an acid environment to an alcalin one

. polyacrylonitryle

ionic composite (IPMC)

material that curves when reacting to an electrical field by ionic migration inside a selective ion membrane

. selective membrane : Nafion or Flemion . electrode : platinum, gold

ionic conductive polymer

material subject to oxydoreduction reactions that induce volume variations

. CP : polypyrrole, polyaniline

carbon nanotubes

material whose electronic balance between the nanotube and the electrolyte is modified by charges injection, resulting in dimension changes

electrorheological fluid

material subject to particules migrations, that modify the rheological properties of the fluid, like viscosity

. LID 3354 from ER Fluid Developments Ltd.

Fig. 1 . Electroactive polymers classification (Claire JEAN-MISTRAL - 2008)

9


a degree of freefom

GROUP Electronic Group

Ionic Group

material

ADVANTAGES

DRAWBACKS

. important generated force . short response time . works in ambient conditions . high life-span

. high electrical field for activation (from 20 to 150 MV/m) . unique deformation direction for quadratic coupling

. important displacements . low electrical field supply . deformation direction according to the tension polarity

. slow response time . low generated force . specific use conditions (humidity, ...) . low electromechanic coupling

Fig. 2 . Main characteristics of the two EAP groups . (Claire JEAN-MISTRAL - 2008)

toward the field and negative charges shift in the opposite direction. This creates an internal electric field that reduces the overall field within the dielectric material itself.

SFEAPs are composed of five main components stacked up like various layers (fig. 3) : an elastic membrane, a conductive layer or soft electrodes, an insulation layer (silicone layer), a soft frame and two electrical plugs (+ & -).

While the term insulator implies low electrical conduction, dielectric typically means materials with a high polarizability. The latter is expressed by a number called the relative permittivity, which is itself determined by the electric susceptibility, which is a measure of how easily a dielectric material polarizes in response to an electric field. The term insulator is generally used to indicate electrical obstruction while the term dielectric is used to indicate the energy storing capacity of the material, by means of polarization.

The pre-stretched elastomer membrane is coated with soft electrodes on each side and then constrained in a soft frame. The conductive layers are connected to the positive and negative plugs for future activation. As far as materials are concerned, the elastomer used for the research was the VHB4910 from 3M. In its initial use, it is a double-sided adhesive with the property of being very elastic (stretch factor reaching up to 500%). The electrodes applied on this membrane have to be easily spread on the membrane and have to form a continuous layer even when the membrane is stretched and then released. Therefore, the electrodes or conductive layers used were either carbon grease or carbon powder. The carbon grease from MG Chemicals was applied with a brush, and the carbon powder (Ketjenblack) from Akzo Nobel was spread out with a soft stamp.

COMPOSITION

10


COMPOSITION

4

5

3 2 1 2 3 4

5

Fig. 3 . Axonometric view (regular and splitted) of an SFEAP / 1. elastic membrane / 2. soft electrodes / 3. insulation layers / 4. soft frames / 5. electrical plugs. (JR & MG - 2014)

11


a degree of freefom

material

PHYSICAL PROPERTIES

When a high voltage and low intensity current is passing through an SFEAP, the latter starts acting like a capacitor in an electric circuit, which means it accumulates energy in the form of an electrostatic field between its two electrodes. As a result, the membrane is crushed up in its normal direction. The kinetic principle of an SFEAP can be explained according to three different scales (Fig. 4). At a macroscale and in its deactivated state, the soft frame is pre-constrained by the elasticity of the pre-stretched membrane (VHB), and SFEAP morphology is directly linked to the initial frame geometry. When the SFEAP is electrically activated, the electrode particles polarize. The membrane is crushed up and tends to extend, leading to the release of the constraint applied to the soft frame. As a result, the latter, previously constrained, bends back to its original shape.

Fig. 4. Hierarchical breakdown of the

1. Macroscale (element size ranging

SFEAPs kinetic principle.

between 5 and 100 cm diameter). Illustra-

(JR & MG - 2014)

tion of the constraint release (constraint

Columns from left to right : macroscale /

exerted by the elastic membrane on the

microscale / nanoscale

soft frame).

12


PHYSICAL PROPERTIES

Deactivated state (0 Volts)

Activation (0 to 5 KVolts)

Activated state (5 KVolts)

2. Microscale (film thickness ranging

3. Nanoscale (carbon black particles

between 25 and 100 Âľm). Illustration of the

diameter ranging between 15 and 300nm).

resultant elongation in the membrane plane

Illustration of the electrodes polarization on

caused by the crushing, whose vector is

each side of the membrane.

normal to the elastomer surface.

(Jim RhonĂŠ & Martin Genet - 2014)

13


a degree of freefom

material

the ground of the power supply and to a combination of a capacitor and a transistor that eventually acts as a digital potentiometer. It is then possible to make the initial tension vary between 0V and 12V by modifying the digital output directly set on the computer instead of doing it manually on the power supply.

ELECTRICAL A C T I V AT I O N

This part of the circuit is connected to a high voltage converter (EMCO G 50), which converts the low digitally controlled tension that ranges between 0V and 12V to a high tension that ranges between 0V and 5KV, according to the following graph.

As far as control is concerned, SFEAP activation constitutes one of the main interests of these smart materials. Indeed, their electroactivity enable a digitally controlled activation. The circuit used in the research is the following (Fig. 8 & Fig. 9) : A power supply delivers a constant current which can be set between 0V and 12V. From the computer, with the use of Grasshopper and Firefly plug-in, an output signal that can vary between 0 and 255 is sent to a PWM pin on the Arduino Uno chip. PWM, which stands for Pulse Width Modulation, is a technique for getting analogue results with digital means.

Fig. 7. Typical input vs. output voltages (G50 data sheet - EMCO High Voltage Corporation - 2013)

The high voltage converter is connected to the SFEAP, whose activation can be modulated according to the value of the delivered tension. When the tension is close to 0V, the SFEAP is in its deactivated state, and progressively activates when the tension increases, till it reaches its fully activated state around 5KV. A first resistor is placed between the converter and the SFEAP in order to avoid the formation of electric arcs, and another one is used as a discharging resistor when used when the SFEAP deactivates.

Digital control is used to create a square wave, a signal switched between on and off. This on-off pattern can simulate voltages in between full on (5 Volts) and off (0 Volts) by changing the portion of the time the signal spends on, known as pulse width, versus the time that the signal spends off. It uses an 8-bit integer value to control the digital signal, and therefore, any numeric value between 0 and 255 can be used to control the PWM signal on the pin. The Arduino Uno chip is directly linked to

16


ELECTRICAL ACTIVATION

+ -

.00

12

Fig. 8 & Fig. 9. 2D and 3D circuits for a digitally controlled SFEAP activation. (JR & MG - 2013)

17


FA

B R CA

T I O


-

B RI-

HOW TO IMPROVE SFEAP’S PRODUCTION PROCESS ?

ROBOTIZED O P T I M I Z AT I O N COMPONENTS DEVELOPMENT

ON

PROCESS & A U T O M AT I O N

-


a degree of freefom

fabrication

SFEAPs’ fabrication process is extremely challenging and demanding in time. Necessitating very accurate and sophisticated production methods and tools, we dedicated a large amount of time of the P9 semester to master the fabrication process of SFEAPs (which was inspired by the one developed for the project «Shapeshift», resulting from a collaboration between the chair for Computer Aided Architectural Design (ETHZ) and the Swiss Federal Laboratories for Materials Science and Technology (EMPA) ). Eventually, the fabrication process Martin Genet and I reached last semester was entirely manual and only enabled the production of small and non insulated prototypes.

HOW TO IMPROVE SFEAPS PRODUCTION PROCESS ?

With a will to enter a pre-industrialization phase, three new objectives were set in terms of fabrication : 1. Improving the prototypes’ quality, resistance, insulation and life span by optimizing its components. This part has been focused on the conductive and insulating layers. They are indeed very difficult to apply uniformly with a manual process and are often responsible of electrical instability and short-circuit. 2. Increasing the prototypes’ size by developing its components. This part has been focused on the soft frame, whose material and dimensions choice are deciding factors for the size of the whole prototype. 3. Automating the whole production process using a F2F (File to Factory) strategy in order to generate all the technical drawings used for the production from one main file.

20


ROBOTIZED OPTIMIZATION

order to get a fluid substrate, less viscous than the silicone alone, a solvent called toluol (liquid) was added in the mixture. It prevents the polymerization of the silicone before vanishing totally after few hours. Once the mixture set, the tool connector and the frame support fabricated, this part of the process required the definition of a toolpath (Fig. 11), made with HAL plug-in, and that the robot would follow when spraying on the membrane. It ended with the essential calibration of the whole environment (end effector + membrane support) in order to adapt the speed of the robot and the distance to the membrane in order to get the thinnest layer possible.

ROBOTIZED S P R AY O P T I M I Z AT I O N

This optimization need started from the observation that applying the conductive and insulating layers (electrodes and silicon) manually lacks accuracy and efficiency.

For different technical issues, the toolpath has evolved a lot.

The considered solution was to develop a robotized process in order to spray the latter directly on the elastic membrane. As a result, this phase would take advantage of the accuracy of a robot toolpath to apply these layers uniformly and regularly.

The first difficulty was avoiding singularities in the toolpath. A singularity occurs when the robot moves along the toolpath and its rotating axes get aligned. The consequence is that the robot stops and thus cannot end the path. This first implied to change the toolpath from a linear logic (Fig. 11) to a circular one (Fig. 12), so that the first axis would rotate more than the others, especially the fifth and sixth axis, most often causing singularities.

Developing a spraying process with a robot implies two steps : 1. the definition of the substrate to spray 2. the fabrication of the tool connected to the robot’s end-effector. In order to compose the substrate, two factors had to be taken into consideration: the viscosity of the mixture, which is essential to obtain good results when spraying, and its conductivity.

During the first tries with the spray, a second issue appeared, due to the air gun used for spraying. Spraying vertically was indeed creating drops at the extremity of the air gun that would fall on the membrane. Once again, the whole logic of the toolpath had to be rethought, opting this time for a horizontal direction of the spray (Fig. 13 & Fig. 14). The whole process could then be tested (Fig. 15).

Before even using the robot, few tests were made in that perspective (Fig. 10). The ingredients’ ratios were modified for each new test in order to establish the best recipe for the future robot application. In

21


a degree of freefom

fabrication

mixture 1

mixture 2

silicone : 33% diluent : 26% lead : 4% solvent : 37%

silicone : 47% diluent : 21% lead : 3% solvent : 29%

mixture 3

mixture 4

silicone : 57% diluent : 16% lead : 2,5% solvent : 24,5%

silicone : 67% diluent : 10% lead : 6% solvent : 16%

mixture 5

mixture 6

silicone : 52% diluent : 11% lead : 7% solvent : 30%

silicone: 61% diluent : 8% lead : 10% solvent : 21%

Fig. 10. Spraying tests to define the best components ratio for the mixture balance. (JR & MG - 2014)

22


ROBOTIZED OPTIMIZATION

Fig. 11. horizontal and linear toolpath /

Fig. 12. horizontal and circular toolpath

problem : singularities

/ problem : substrate dripping

Fig. 13. vertical and linear toolpath /

Fig. 14. vertical and horizontal toolpath /

required : support fabrication

geometry taken into consideration

Fig. 11, Fig. 12, Fig. 13 & Fig. 14 . Successive tool and toolpath adaptations during the robotized spraying process calibration. (JR & MG - 2014)

23


a degree of freefom

fabrication

Fig. 15. Robotized spray process. LRTB : mixing toluol . mixing silicone . mixing carbon grease . connecting the air-gun to the end effector . loading the mixture in the air-gun tank . spraying, zoom x1 . spraying, zoom x2 . spraying, zoom x4 . removing the stencils before applying the soft frames on the clean surface. (JR & MG - 2014)

24


ROBOTIZED OPTIMIZATION

25


ROBOTIZED SPRAY PROTOTYPE

Fig. 16. Spraying the soft electrodes and the silicone coating with the robot gives an homogeneous surface that was impossible to obtain by a manual process.

1 polycarbonate frame thickness 0.75mm width 12mm

2 elastic membrane carbon grease electrode silicone coating


1

2


a degree of freefom

fabrication

new fabrication process had to be developed in order the test the efficiency of this material.

SOFT FRAME DEVELOPMENT

During the JEC, three companies accepted to sponsor us for these prototypes. Hinddostan Technical Fabrics, based in India, offered us a roll of 200g/sqm bi-axial carbon fibre. TeXtreme, based in Sweden provided us with a roll of 60g/sqm bi-axial -Ultra Light Composites- carbon fibre, supposed to be really adapted to our application. Finally, as far as resin is concerned, Sicomin sent us two different kinds of resin : a soft one -SR 8160 / SD 815 Bx- and a harder one - SR 8200 -.

The material used so far for the soft frame was a polycarbonate sheet of 0.75mm thickness. But when trying to produce bigger prototypes, the encountered issue was that this material was not adapted anymore because it was too heavy and too soft.

The objective was to realise a very thin, flexible and lightweight frame by progressive lamination of a carbon fibre layer and a resin layer.

Indeed, when increasing the size of the SFEAP, the pre-constraint applied on the frame increases too, necessitating a wider or thicker frame to prevent the latter from too much deformation in the deactivated state. As a result, the frame becomes heavier, and proportionally too heavy to enable any actuation when the SFEAP is activated.

The hard resin must be applied in the neutral fibre to ensure the frame reactivity, whereas the soft one has to be applied in the extreme fibres to enable its softness without folds. In order to save material, carbon fibre bands had to be cut from the roll and then assembled during the lamination. Being a manual process, this may have caused some thickness variations where the bands are superposed. In the future, this process could be fully industrialized, avoiding at the same time avoiding this lack of accuracy.

A short investigation was lead to find materials whose properties would fit to bigger prototypes : lightweight, soft but reactive and resistant. After meeting some industrials during the JEC Composites fair 2014, carbon fibre appeared to be an adapted material for this application.

Indeed, the company Mateduc Composite that we met during the JEC, produces very thin and flexible sheets of carbon (0.3mm0.5mm-0.7mm) that can be water-jet cut and welded if necessary.

But to get the wanted properties from carbon fibre, the latter has to be mixed with epoxy resin, which solidifies the fibres when polymerizing. This meant a whole

28


SOFT FRAME DEVELOPMENT

Fig. 17. LRTB : carbon fiber cutting . epoxy resin mixing (soft and hard) . lamination process (soft epoxy) . lamination process (carbon fiber) . lamination process (hard epoxy resin) . 2-days polymerization to reach the optimal properties . demolding from the aluminium support . cutting and sanding the frame edges . application on the membrane.

29


CARBON FIBRE FRAME PROTOTYPE

Fig. 18. element diameter : 450mm carbon fibre type : 200g/sqm electrode type: carbon powder insulation : silicone coating

1 carbon fibre frame thickness 0.5mm width 45mm

2 elastic membrane carbon powder silicone coating

2


1


a degree of freefom

fabrication

frame

PROCESS & A U T O M AT I O N

electrode As far as the efficiency of the process is concerned, the automation of the technical 2D drawings generation was required to avoid repeating the same task each time the geometry of the prototype would change.

file to factory

Therefore, the creation of a Grasshopper definition enabled the generation of all the drawings needed for the production, and this from a simple base geometry (Fig. 19).

export file

For instance, for an hexagonal SFEAP, setting the radius of the base geometry enables the generation of the frame drawing which is directly exported in a .DXF file that can be sent to the company Mateduc Composites to water-jet cut the carbon sheets. The definition also provides the masks drawings for the electrodes and silicone applications, which are exported in .DXF files that will be used to lasercut the masks. Finally, the definition generates an optimized toolpath that can be encoded (RAPID) and then directly streamed to the ABB robot in order to spray the silicone on the membrane taking the SFEAP geometry into consideration (Fig. 20).

frame drawing

Fig. 19. Automation logic diagram. From a simple base geometry to 2D technical drawings for production. (JR & MG - 2014)

32


PROCESS & AUTOMATION

carbon carbon black black electrodes electrodes

ctrodes lectrodes masks masks

silicone silicone layer layer

silicone silicone masks masks

lasercut lasercut

robotised robotised spray spray

masks masks drawing drawing

toolpath toolpath generation generation

export port files files

base base geometry geometry

33


a degree of freefom

fabrication

base geometry

R

S

base hexagon radius : R optional stretching : S

Fig. 20. Technical drawings generated by the Grasshopper definition from the base geometry. (JR & MG - 2014)

34


PROCESS & AUTOMATION

electrodes’ masks drawing

toolpath generation

W

Oi L Oo C

frame drawing inside , outside offsets : Oi , Oo connections spots : C witnesses : W

silicone’s mask drawing

spacing of the toolpath : L speed of the robot : 60mm/sec

J

F F Oo

f f

Es

separately drawn joints : J electrodes’ slits : Es frame width , fillet radius : F , f

frame width , fillet radius : F , f outside offset : Oo

35

36


KINE

T STRUC-

T U


-

T I C

A N A S S E M B LY O F ELEMENTS ABLE TO BEAR LOADS AND TO CHANGE MORPHOLOGY

WEIGHT TESTS

URE

STRUCTURAL C O N S I D E R AT I O N S KINETIC PROTOTYPES

-


a degree of freefom

kinetic structure

Tackling the issue of structure with SFEAPs first raises the question of stability. Indeed, the frame softness, though being one of the key properties to enable large deformations, is also what makes so difficult envisaging SFEAPs as structures. A first step while broaching these structural considerations is the measure of how much weight an SFEAP can bear. An experiment dispositive was then fabricated to test the performances of SFEAPs (Fig. 21).

WEIGHT TESTS

TESTED ELEMENT PROPERTIES : Frame geometry : hexagon Frame type : double frame Frame thickness : 0,75mm Frame width : 9mm Interior frame radius : 97mm Total weight : 13,2g

EXPERIMENTAL PROTOCOL TEST #1 : An SFEAP is maintained vertically with a metal attach at its bottom. A weight (increasing progressively) is then applied at its top. The prototype is finally activated to check if it is able to bend back under the given weight (Fig. 22).

RESULTS TEST #1 : When the weight applied reaches 21g, e.g. 160% of the prototype’s weight, the latter bends so much in its deactivated state that the force produced by its high voltage activation is not enough to bend it back, leading to no movement.

40


WEIGHT TESTS

1

2 4 3

6

5

Fig. 21. Experimental dispositive and components used for the weight tests 1. measure stand / 2. power supply / 3. metal support / 4. hexagonal SFEAP / 5. plugs / 6. weights 1 and 20g (JR & MG - 2014)

41


a degree of freefom

kinetic structure

Fig. 22. Successive steps of the test #1. (JR & MG - 2014)

Fig. 23. Successive steps of the test #2. (JR & MG - 2014)

42


STRUCTURAL CONSIDERATIONS

As far as structure is concerned, this limit weight of « non-return » is really low and does not even reach twice as the weight of the prototype.

STRUCTURAL C O N S I D E R AT I O N S

HYPOTHESIS : The weight of «non-return» is that low because the response of the prototype to this weight is a too strong curvature due to its soft frame. Reducing the deformation of the prototype in its deactivated state when a weight is applied at its top could be a way of increasing the limit weight supported by the prototype.

FIRST ASSEMBLY TESTS From these weight tests results, few SFEAP assemblies were realised to check if some rigidity could be obtained using the elements’ morphology and elasticity (Fig. 24).

EXPERIMENTAL PROTOCOL TEST #2 : The element is maintained vertically from its both sides, with a metal attach at its bottom, and a thin thread hanging from the pivoting pole at its top. A weight (increasing progressively) is then applied at its top. The prototype is finally activated to check if it is able to bend back under a given weight (Fig. 23).

RESULTS TEST #2 : With this experimental set-up, the prototype managed to bend back almost totally up to a weight of 254g applied at its top. Of course this result does not mean that it is able to support twenty times its own weight, but rather that if its curvature is blocked to a certain radius when a weight is applied, it is still able to produce a movement. And this even when more important weights are applied, which is the kind of configuration looked for in order to develop a structural principle.

Fig. 24. First assembly tests as a double membrane using pop rivets for joints. (JR & MG - 2014)

43


a degree of freefom

kinetic structure

The first attempts were realised with a double-membrane logic in order to create stronger nodes and give more thickness to the system. Pop rivets were used to ensure

the joints between elements. The results from this assembly strategy did not provide enough rigidity to the system and the latter was therefore quickly abandoned.

ANGLE X

ANGLE Y

ANGLE X’

ANGLE Y’

Fig. 25. Evaluation of an element’s angular deformation : for each element, curvatures (in XZ and YZ planes) are evaluated at deactivated (angleS X & Y)and activated states. (JR & MG - 2014)

44


STRUCTURAL CONSIDERATIONS

SFEAP TYPE SELECTION

the remaining ones differing from their electrodes type and their frame width.

After these first attempts, the assessment was made that investigating SFEAPs’ possible structural performances also required to select a unique type of SFEAP according to their behaviour and deformation. Indeed, according to the fabrication process and the components used to produce SFEAPs, their properties can differ, and it was therefore necessary to choose one in order to eliminate varying parameters and thus facilitate results analysis for the rest of the research.

In order to make this comparison, the different SFEAPS were analysed in their activated and deactivated states. From the recorded movement, an angular deformation range was determined, measuring it directly on a side picture (Fig. 25). According to the comparison of the deformation range between the different prototypes that were produced at this time of the research (carbon grease, carbon powder, single frame, double frame, variation of the frame width, etc.), the double frame and carbon powder layered prototype (PD) was chosen for the future prototypes to come (Fig. 26 & Fig. 27 within the following pages).

In order to make this selection, three criteria were taken into account : deformation during activation, fatigue and efficiency of the fabrication process. The efficiency issue eliminated elements produced with a silicone coating, because this process, though robotized, is still too demanding in time to be implemented in an assembly-line process.

HYBRIDIZATION In order to validate or reject the obtained results with the weight tests #2, the objective was to combine SFEAPs with a stronger element able to limit their curvature when a weight is applied on, but without preventing their movement when activated.

The fatigue issue eliminated multi-layered elements. Multi-layered SFEAPs are very interesting for their deformation. Indeed, the force developed by an SFEAP is linked to the volume of the elastic membrane. Basically, the force of an SFEAP increases with its number of layers.

Hybridize the assembly using a complementary system that would give rigidity to the whole therefore appeared to be a worth considering solution. But again, the challenging part was designing a skeleton that would rigidify the system and at the same time tolerate the SFEAP movement when activated. In order to avoid disturbing the SFEAP dynamic behaviour when activated, the skeleton had to remain very light.

However, fatigue tests realised on these elements proved that their life span was shorter than the one of regular elements. This process was therefore aborted during the research, but will for sure be an important domain of investigation for the future development of the project. Eventually, the deformation criterion allowed to select one type of SFEAPs among

45


a degree of freefom

kinetic structure

180°

160°

110°

90°

60° 10° 0° PS

GS

48

PM

GM

PL

GL

PD


GL

STRUCTURAL CONSIDERATIONS

Fig. 27. Angular deformation range comparison. the bigger the box is, the larger will be the deformation during the activation. (JR & MG - 2014)

range angle allowing the use of an element as an actuator in a kinetic structure

activated state

X Y

PD

GD

P G S M L D

passive state

carbon powder electrode carbon grease electrode 13mm wide frame 16mm wide frame 19mm wide frame 8mm wide frame (doubled)

49


a degree of freefom

kinetic structure

CARBON RODS RUNNERS The first experimentation consisted in using thin carbon rods as runners, with a displacement range enabled by two blockers. Attached to the SFEAPs in strategic points, they would allow their movement and then block the system when extremities of the runners are reach (Fig. 28).

t=0

The advantage of such a system is that SFEAPs thus cannot reach their «nonreturn» state anymore, no matter how much weight is applied on.

t=1

More over, the distance between the two blocking elements can be set, meaning that the deformation range between the deactivated and activated states can be totally controlled. t=2

t=3 Fig. 29. Activation of the kinetic prototype

However, in the system’s deactivated state, the carbon rod blocker is longer than the membrane and does not behave like an integrated element anymore, making the assembly of more elements quite challenging (Fig. 29). In addition to the fact that this system was not coherent structurally, this main drawback imposed to develop a new solution.

Fig. 28. Kinetic module with carbon rod runners. (JR & MG - 2014)

50


STRUCTURAL CONSIDERATIONS

TENSEGRITY Considering the elasticity of an SFEAP, and the fact that it behaves better in tension than in compression, the next step was focused on designing a system working in tensegrity. Most of the carbon rods were then replaced by threads with the objective to constrain SFEAPs with one degree of freedom left.

t=0

This much simpler solutions appeared to work better than the previous one. But still, two main problems were noticed : Firstly, according to the SFEAPs state, threads can be fully tensed and ensure their structural role, or in the opposite state fully released and then totally useless. The second issue faced with this experiment was a twisting effect, directly

t=1

t=2

t=3 Fig. 31 Activation of the kinetic prototype

related to SFEAPs behaviour when they are assembled together. Indeed, since the fabrication process developed so far is not accurate enough to produce exactly similar elements, the tension of the membrane can slightly differ from one element to another. When two SFEAP are assembled, one takes advantage over the other, resulting in this twist effect that is impossible to control with threads only.

Fig. 30. Kinetic module with tensegrity (JR & MG - 2014)

51


a degree of freefom

kinetic structure

y

EMBEDDED KINETICS SYSTEM The necessity of adding external elements that work in compression in order to avoid the whole system to bore imposed to develop a new solution. This time, instead of thinking the SFEAPs assembly separately from the skeleton, the conception strategy changed and SFEAPs were considered as the skeleton actuators. Giving up the idea of constraining SFEAPs so that one degree of freedom would be left for the actuation, the skeleton of one module has now one degree of freedom and needs SFEAPs (working as antagonist muscles) to be stable. It is then by activating the latter that the whole module can transform.

x

Fig. 32. Module’s static diagram / assuming that the membranes’ pre-stretching enable the system’s balance. (JR & MG - 2014)

SFEAPs (blue) are attached to the skeleton extremities and stabilize the module in a deactivated state (without taking the softness of the latter into consideration). In order to think the system as kinetic, the movement resulting from SFEAP activation was represented as follows (Fig. 33), before being integrated in the previous diagram.

UNIDIRECTIONAL KINETIC PROTOTYPE

y

To understand the kinetics of this hybrid system, the latter was first simplified and represented as a simple 2D static diagram (Fig. 32). x

The carbon skeleton (orange) of the module is split into two parts than can rotate around the same axis, resulting into one degree of freedom. Two identical

Kinematic diagram of the element activation (soft frame hexagonal prototype)

Fig. 33. SFEAP’s kinetic diagram / combination of two bars linked by a turning pair.

52


UNIDIRECTIONAL KINETIC PROTOTYPE

y

This led to the following kinetic diagram of the whole module (Fig. 34 & Fig. 35). Testing this principle was then enabled by the fabrication of a physical prototype. This prototype is composed of a carbon rods skeleton (2mm diameter) and four hexagonal SFEAPs. It is divided into two modules and each module contains part of the skeleton that can rotate around an axis. On these parts, two antagonist SFEAPs are attached symmetrically between these rotating parts. The nodes of the skeleton have been 3D printed, allowing accuracy in the structure but also the production of rotating joints for the movement of the system (Fig. 36).

x

Fig. 34. Module’s kinetic diagram / deactivated state. (JR & MG - 2014)

y

x

Fig. 35. Module’s kinetic diagram /

Fig. 36. Rotating joints 3D printed for the

activated state. (JR & MG - 2014)

prototype. (JR & MG - 2014)

53


a degree of freefom

kinetic structure

54


UNIDIRECTIONAL PROTOTYPE

Fig. 38. Side view chronophotography of the unidirectional kinetic prototype / activation of one of the same antoganist SFEAP for each module (JR & MG - 2014)

Fig. 39. Top view of the unidirectional kinetic prototype / deactivated state (JR & MG - 2014) Fig. 40. (on the left) Picture of the whole prototype (JR & MG - 2014)

55


UNIDIRECTIONAL KINETIC PROTOTYPE

Fig. 41. prototype : 2 modules actuators : hexagonal SFEAPs electrode type: carbon powder frame type: doubled

1 carbon rods (skeleton) diameter : 2mm

2 3D printed rotating joint printing : SLS (Selective Laser Sintering)


1

2


a degree of freefom

kinetic structure

But still, the axes of rotation are all parallel and prevent the latter to move in all directions. As a result, the new objective was to develop a new kinetic prototype enabling much more freedom in its movement.

M U LT I D I R E C T I O N A L KINETIC PROTOTYPE

Therefore, the same rotation principle with one degree of freedom per module was kept, but instead of stacking the modules identically (which led to parallel rotation axis), the aim is to stack them with a 90째 rotation from one to another.

The unidirectional prototype confirmed that the envisaged principle was working as expected. Indeed, when activating, the antagonist SFEAPs create a momentum around the rotation axis that connects the two skeleton parts of the module together, leading to the rotation of the upper part of the module. Moreover, the results obtained with the 3D printed rotation joints seem to be accurate : the module rotates till it reaches the maximum angle which is set by the space let between the two parts of the rotating joint.

The first modelizations that were made revealed the impossibility of doing so with hexagonal SFEAP. Indeed, to reach the goal, this geometry implied to produce a very intricate skeleton that was not coherent anymore with the type of prototype and structure envisaged. The reason for this is that a hexagon has too many sides to preserve the simple vertical stacking and add at the same time the rotation of 90째. Decision was made to quit this geometry in order to try new modelizations with square SFEAPs.

Of course the resulting movement is not as important as the one observed when activating a simple SFEAP, but the objective is to stack few of these modules so that the angles of rotation are added and finally create a large deformation. Indeed, compared to the first assembly and hybridization tests that were previously made, weight is not an issue anymore. Modules can thus be superposed to form larger prototypes.

With this geometry, the 90째 rotation appeared possible while keeping a simple vertical stacking logic of the modules. Eventually, an adapted skeleton with new rotating joints was conceived. This skeleton, entirely triangulated was way more efficient and using less carbon rods. As far as kinetics and material systems are concerned, the whole design of this new soft kinetic structure appeared to be an improvement and an optimization of the previous prototype. It was therefore decided to produce it with more modules.

With this prototype, the idea of a soft kinetic structure started making much more sense.

58


MULTIDIRECTIONAL PROTOTYPE

z

z

y

y

x

x

z

z

y

y

x

x

Fig. 42. 3D kinetic diagram of the multidirectional prototype. LRTB : deactivated state / Y rotation axis / X rotation axis / X+Y rotation axis. (JR & MG - 2014)

59


a degree of freefom

kinetic structure

60


MULTIDIRECTIONAL PROTOTYPE

Fig. 43. Side view (a) chronophotography of the multidirectional kinetic prototype / activation of one of the same antagonist SFEAP for each module (JR & MG - 2014)

Fig. 44. Side view (b) chronophotography of the multidirectional kinetic prototype / activation of one of the same antoganist SFEAP for each module (JR & MG - 2014) Fig. 45. (on the left) Top view of the unidirectional kinetic prototype / deactivated state (JR & MG - 2014)

61


MULTIDIRECTIONAL KINETIC PROTOTYPE

Fig. 46. prototype : 6 modules actuators : square SFEAPs electrode type: carbon powder frame type: doubled

1 3D printed rotating joint printing : SLS

2 3D printed joint + detail driving the cables : plugs are integrated inside the structure

1


2


SIMU L AT &CON T


-

ION

A LIVE STRUCTURA L A N A LY S I S E N A BLING A BALANCE CONTROL OF THE PROTOTYPE

3D SCANNING & S I M U L AT I O N FEEDBACK SYSTEMS

ROL

BALANCE CONTROL

-


a degree of freefom

simulation & control

of an SFEAP. The tool used for the scans is a Kinect and the software enabling the generation of a mesh from scans interpretations was KScan 3D. To be workable in Rhino 3D, the mesh had to be exported with Geomagic.

3D SCANNING AND S I M U L AT I O N

In order to get an accurate mesh, scans from different angles have to be realised and then reoriented and aligned on the software. It was therefore necessary to make a support for the Kinect that would be able to rotate around a fixed element. The same measurement stand used for the weight tests was converted into a scan support in order to be able to scan systematically with the same angles (Fig. 47). The first tries revealed that it was possible to obtain quite accurate surfaces and meshes (Fig. 48), at least enough to be able to recompose the SFEAP movement (Fig. 49).

As far as SFEAP activation is concerned, the state reach so far was the generation of a Grasshopper definition allowing a digital and accurate command, compared to the manual setting of the power supply. In order to envisage this digital command in a relevant way, consisting in using this command for the control of the soft kinetic structure, the very first step was the understanding of an SFEAP movement and its digital simulation. The first simulation attempt were based on an empiric process, measuring an element with different degrees of activation and trying to recompose it geometrically on Rhino 3D and Grasshopper. The behaviour of the elastic membrane constrained by the soft frame is very close to the minimal surfaces geometric principle. The problem was the lack of accuracy coming from the simulation of the frame.

Fig. 48. Mesh of an hexagonal SFEAP scanned geometry (JR & MG - 2014)

It was therefore decided to quit this first option to try a 3D scanning procedure. 3D scanning an element in different degrees of activation would indeed allow for recomposing very precisely the movement

Fig. 49. Recomposition of the movement with the various meshes (JR & MG - 2014)

66


3D SCAN & SIMULATION

5

2

1 4 3 Fig. 47. Experimental dispositive and components used for the scans / 1. KScan 3D software / 2. SFEAP / 3. Rotating axis / 4. Kinect support / 5. Kinect. (JR & MG - 2014)

This movement simulation via 3D scans was supposed to be used for different purposes.

the hybridized systems, SFEAP are linked to a carbon skeleton which has only one degree of freedom per module. The possible movement of the structure is thus mathematically known and the movement of the SFEAPs does not need to be simulated anymore.

Firstly, it was a way to understand better SFEAPs’ behaviour when activated. It was also meant to help for conception and representation issues as it allowed to visualize and measure an SFEAP in every degrees of activation. Finally the main objective was to use this simulation to generalize it to a system and be able to control it (with a kind of predictive method).

From this point, the new objective to reach was the live control of the kinetic prototypes and not only a command as done before. It is no more the SFEAP morphology that matters, but the behaviour of the whole structure. As a result, instead to be interested in how an SFEAP activates, the issue is to be able to know if an SFEAP activates and actuates the module how it should be. This logic requires feedbacks.

However, the hybridization turning point in the prototype conception led to the abort of this simulation logic. Indeed, with

67


a degree of freefom

simulation & control

effect tricky, and it is necessary to analyse the system as a whole. A consequence of this is that the behaviour of feedback systems is often counter intuitive, and it is therefore necessary to resort to formal methods to understand them.” (Feedback Systems, ALSTRÖM, MURRAY, 2012)

FEEDBACK SYSTEMS

A feedback, whereby the system is in constant transformation, poses a set of logical interactions by which information is reinvested into the system to provide a constant re-balancing and re-calibration of its functioning state, and means a cyclical organization of its causality logics.

WHAT IS A FEEDBACK (Fig. 50) ? “A dynamic system is a system whose behaviour changes over time, often in response to external stimulation or forcing. The term feedback refers to a situation in which two (or more) dynamic systems are connected together such that each system influences the other and their dynamics are thus strongly coupled. Simple causal reasoning about a feedback system is difficult because the first system influences the second and the second system influences the first, leading to a circular argument. This makes reasoning based on cause and

FEEDBACK AND CONTROL Controlling the movement and the behaviour of the prototypes with a live structural analysis that would take loads distribution into consideration seems inappropriate and too complex. Indeed the kinetic prototypes are soft structures and implementing a softness factor in the structural analysis

Fig. 50. Simple feedback scheme . Ludwig von Bertalanffy 1968 - (reprinted from original publication. George Braziller)

68


FEEDBACK SYSTEMS

would require a very intricate procedure compared to a hard mechanical system whose components rigidity is reliable.

initial release will typically bring the resistance value down to approximately +10% of its initial resting value. Resistive value continues to decay to its nominal resting value.

In order to compensate for physical inaccuracies and to be able to develop an effective control of the prototype, the latter has to integrate a feedback system. Therefore, the use of sensors directly placed on the prototype itself could be a conceivable solution.

Those electrical effects disturb the received data from the structure’s deformation, and drive measures to be relatively inaccurate or wavering, leading the feedback to become “shaky”. Furthermore, bending sensors’ elasticity is usually too low, which would apply external forces within the concerned DOF. The choice of stretching sensors to compute the structure’s dynamic deformation now appears to be inadequate, as they would influence the movement by themselves.

STRETCHING SENSORS A stretch sensor is a polymer component that changes resistance when stretched. An un-stretched sensor has a nominal resistance of 1000 ohms per linear inch. As the stretch sensor is stretched the resistance gradually increases. When the sensor is stretched to 50 %, its resistance will approximately double to 2.0 Kohms per inch. The sensor is a flexible and elastic cylindrical cord (2mm in diameter), with spade or ring electrical terminals at each end. Recommended operating range is 4050% elongation for repeatable operation.

BENDING SENSORS Bending sensors can be used to detect vibration, humidity, motion, impact, and airflow. The sensor consists of a plastic film printed with a special carbon ink. The resistance of this ink increases the more

Taking measurements of the stretch sensor uses the same methodology as taking resistive measurements of a variable resistor. The terminal ends of the sensor are connected to a VOM meter set to measure ohms. Nevertheless, the stretch sensor has a few resistive artefacts. When stretched into position and released, the resistance may increase slightly upon release, before decaying to its resting resistive value (Fig. 51). The decay of the resistive value to its resting value takes place over time. The

Fig. 51. Stretching sensor resistance variation. (Images Scientific Instruments Inc.)

69


a degree of freefom

simulation & control

it is bent. This provides non-mechanical reliability in electronic sensing and actuator technology. If they’re un-flexed, the resistance is about ~10 K-ohms. When flexed all the way the resistance reaches ~20 K-ohms.

sensed value to its corresponding meant measure. This creates a curve of tendency that helps to match any given number from the sensor to an approximate measure (Fig. 52 & Fig. 53).

CALIBRATION SENSORS & IMPLEMENTATION Electronic sensors deliver a numerical value within an operational range from 0 to 1023 specific to Arduino. The calibration of the sensors then involves empiric tests to translate the abstract output of the sensor into analogue measures.

In a preliminary stage, the implementation of stretch sensors or bending sensors appeared to be an affordable and relatively simple solution.

This translation is a discrete reconstruction of meaning. It implies to draw a certain amount of connections between one

The objective was to measure the angular deformation of the structure’s modules (Fig. 54), i.e. the angle of rotation around each rotation axis (1 DOF). By linking each

Fig. 52. Stretching sensor calibration

Fig. 53. Bending sensor calibration

(JR & MG - 2014)

(JR & MG - 2014)

70


FEEDBACK SYSTEMS

Fig. 54. Bending sensor implementation on the kinetic prototype : the carbon skeleton angle variation is recorded with the aid of two bending sensors on each module. (JR & MG - 2014)

moving part of the skeleton to the one connected by this rotation axis, with a stretching or a bending sensor, it would then be possible to evaluate a distance or a curvature radius that would lead to the angular value of each local deformation.

To conclude, both of those sensors are not accurate enough. They rather act as qualitative sensors than quantitative, and therefore are not suitable for live measuring. Plus, as the analysis of the movement needs precision, the choice of sensor has to be reconsidered. An other inconvenient of those types of sensors is the lack of integration of the system, which could decrease even more its reliability.

FEEDBACK PERSPECTIVE Using the sensor to measure absolute values entails, before each use, the calibration of the range – stable and extremes states – of values it will operate on. Unfortunately, values are drifting slowly in time.

In fact the integration of the sensor in the structure itself is completely conceivable as SEAP elements could potentially stand as their own deformation sensors. The SFEAP elements would combine both the function of actuator and sensor, and allow for a virtually complete reliability of the feedback system. Besides, SFEAPs could even optionally act as generator.

Despite the relative accuracy and localized stability of the signal, as bending sensors are primarily designed for detecting relative change, the outcome remains uncertain.

71


a degree of freefom

simulation & control

However, considering the high level of the technology involved, the best option to be considered here seems to lie in the possibilities offered by Inertial Measurement Units (IMU).

BALANCE CONTROL

INERTIAL MEASUREMENT UNITS An IMU works by detecting the current rate of acceleration using one or more accelerometers, and detects changes in rotational attributes like pitch, roll and yaw using one or more gyroscopes. Some also include a magnetometer, mostly to assist calibration against orientation drift. This allows better performance for dynamic orientation calculation in Attitude and heading reference systems (AHRS), processing systems based on IMUs, which calculate the relative orientation in space.

value comprised within a range from 0 to 255, which is then translated into PWM (Pulse-Width Modulation), and inserted as a command of activation of an SFEAP. To be able to control the change of shape of the elements instead of just the voltage, another transcription of values is needed. Indeed, the deformation is not proportional to the applied voltage. In order to obtain the calibration needed, the following functions are elaborated: - Percentage of deformation by applied voltage, 0-12V as the initial voltage Ui, and

Aiming at controlling the kinematics of the structure, it is necessary here to refer to Attitude Control logics. Attitude Control is the exercise of control over the orientation of an object with respect to an inertial frame of reference or another entity. Controlling vehicle attitude requires sensors to measure vehicle orientation, actuators to apply the torques needed to re-orient the vehicle to a desired attitude, and algorithms to command the actuators based on sensor measurements of the current attitude and specification of a desired attitude.

COMMANDS DEFINITION & GRAPHS The command sent to the Arduino via Grasshopper and Firefly, is a numerical

Fig. 55. Percentage of deformation by applied voltage. (JR & MG - 2014)

72


BALANCE CONTROL

SENSORS DATA INFLUENCE Fig. 56. Percentage of deformation by time of activation and deactivation. (JR & MG - 2014)

0-5kV as the converted voltage Uc (Fig. 55) - Percentage of deformation by time at the moment of activation, and deactivation (Fig. 56). Again, after reordering data, a tendency curve is drawn to serve later as a mathematical function.

FEEDBACK CONTROL SYSTEM By comparing the sensors’ data and the command calibrations, the role of the algorithm will be to recompute the adapted command (Fig. 57 to Fig. 62).

digital model

sensors datas

physical model

desired morphology

comparison with sensors datas

(re)compute adapted commands

clock

Fig. 57. Feedback based control diagram (JR & MG -2014)

73


ACTIVATION STATEMENT

a degree of freefom

simulation & control

-16°

A

16°

B

A

objDef = -100% A = 0% B = 100%

B

objDef = 0% A = 0% B = 0%

A

B

objDef = 100% A = 100% B = 0%

desired deformation 0°

A

B

READ COMMAND

READ SENSOR

PROCESS

UPDATE & SEND

objDef = 50% A = 50% B = 0%

speed = 0°/sec recDef = 50% compAngle = 0°

A = keep A B = keep B

A = 50% B = 0%

desired deformation 8° received deformation

A

B

READ COMMAND

READ SENSOR

PROCESS

UPDATE & SEND

objDef = 50% A = 0% B = 0%

speed = 0°/sec recDef = 0% compAngle = +8°

A = give objDef B = keep B

A = 50% B = 0%

desired deformation received deformation

TBLR : Fig. 58. Activation statement, differentiated commands according to the desired deformation / Fig. 59, Fig. 60, Fig. 61, Fig. 62. Updating commands scenarii. (JR & MG - 2014)

74


BALANCE CONTROL

A

B

READ COMMAND

READ SENSOR

PROCESS

UPDATE & SEND

objDef = 50% A = 50% B = 0%

speed = 2°/sec recDef = 25% compAngle = +4°

A = keep A B = keep B

A = 50% B = 0%

desired deformation load 4° received deformation on A > on B

A

B

READ COMMAND

READ SENSOR

PROCESS

UPDATE & SEND

objDef = 50% A = 50% B = 0%

speed = 0°/sec recDef = 25% compAngle = +4°

A = A + 25% B = keep B

A = 75% B = 0%

desired deformation received deformation

Fig. 63. Bending sensors feedback, comparison of the desired and received models upon activation command and in the perspective of the balance control. (JR & MG - 2014)

75


a degree of freefom

simulation & control

EXECUTION FEEDBACK: CAUSALITY PROBLEMS AND DYNAMIC LOGICS Writing the program code can rely on establishing state machines, which help to expose the conditional sequence of logic rules (Fig. 64, Fig. 65, Fig. 66). This method is organizing the “if � condition with yes or no questions, which are represented by the diamond shapes: if the answer is yes the code continues with the down arrow, and if it is no with the right arrow. Rectangles are statements, i.e. they contain the actual command once you have defined the condition it has to apply for.

Fig. 65. State Machine 2a in the case the objective is too far left. (JR & MG - 2014)

76


BALANCE CONTROL

Fig. 64. Main State Machine, main conditions statements. (JR & MG - 2014)

Fig. 66. State Machine 2b in the case the objective is too far right. (JR & MG - 2014)

77


PO TIALS

T


T EN-

ROBOTS BUILDING BUILDINGS OR ROBOTBUILDINGS ?

D I G I TA L F A B R I C AT I O N COMPETITION RESEARCH PERSPECTIVES & POTENTIALS

-


a degree of freefom

potentials

Bauhaus Building of 1927. This pavilion came naturally as an experimentation answering to the question: can we build with Soft Frame Electroactive Polymers?

D I G I TA L F A B R I C AT I O N COMPETITION

Momentum is a kinetic pavilion with no additional mechanisms, no pistons, and its dynamism comes from the pavilion’s shell itself. The kinetic pavilion offers a large range of spatial variations, adapting its shape for diverse uses (Fig. 67 & Fig. 68) such as student events , lectures, or open air cinema. (Fig. 69 & Fig. 70)

As part of the experimental process, the hypothesis proposed had to be brought to a higher conceptual level. For that purpose, the Digital Fabrication Competition organized by the Dessau International Architecture Graduate School, provided an appropriate platform for the research to be confronted to a series of constraints (context, programme, size restriction).

It is composed of two parts: a low wooden deck from which emerges a very light structure split into two curving parts and conceived within the following system: each structural arc is composed of repeated hybrid modules composed of two hexagonal antagonist SFEAPs attached to a carbon rod skeleton, whose particularity is a remaining degree of freedom allowing a rotation around one of its symmetry axis and the variation of the arc’s curvature radius.

The task was to design the Summer Pavilion 2014 and present a buildable proposal for an innovative temporary structure that would provide shelter, shade, and seating for a range of events on a square, inside the campus across the famous Gropius

As a result, the electrical activation of one of the SFEAP tends to create a momentum

Fig. 67. Side view of the pavillion «Momentum» (JR & MG - 2014)

80


DIGITAL FABRICATION COMPETITION

3.50m

3.70m

5.00m

Fig. 68. Axonometric view of the pavillion ÂŤMomentumÂť (JR & MG - 2014)

around this same axis, which can therefore rotate. The balance of these structural arcs is managed by the automated control of activation and deactivation of these antagonists SFEAPs, working as muscles. The price oh the pavilion was also at stake in the competition. According to the material price of each component, we estimate the price of one module to eighty euros, so 5670 euros for the 72 modules of the pavilion.

RESEARCH PERSPECTIVES & POTENTIALS

PROS & CONS Technically, it was proved earlier the feasibility of such a structure. Although the joints linking the structure to the flexible textile and the deck were not perfectly considered yet, the kinetic pavilion project draws a promising opening towards a soft kinetic architecture.

A large part of this project has been dedicated to the development of a soft and light kinetic prototype. Considering the whole evolution and the progress that has been made since the beginning of the research, the idea is now to analyse the results in order to highlight the potentials

81


a degree of freefom

potentials

of this material system and extract the main stakes that the latter raises.

hypostatic and would not be stable without the attached antagonist membranes, which are therefore not only actuators but also part of the structure itself.

OBJECTIVES AND RESULTS 3. Controlled Movement A first step consists in comparing the final results to the initial objectives that were set for this applied research.

As far as architectural applications are concerned, controlling the movement of the system was set as one of the main conditions. The latter is one of the reasons that guided the research towards SFEAPs that are electrically activated.The progressive development of an electrical circuit and a Grasshopper definition allowed an important progress in this domain. The movement and balance of the system are now digitally activated and controlled. The implementation of a feedback logic, that besides still requires some investigations, is the key for a full kinematic balance control.

1. No Hard and External Active Mechanism Conceiving a kinematic system that would not require a hard and external active mechanism was the first challenge. The latter directly oriented the research towards the use of smart materials and represented an important condition allowing the investigation of kinematics within the field of structure and architecture. From this point of view, the prototype is a hybrid system composed of an active membrane that actuates a skeleton composed of passive articulations. Therefore, the actuator is no more still and external but integrated into the material system itself.

ADVANTAGES 1. Kinetic Efficiency

2. Embedded Dynamic System

The designed and fabricated prototype is a fully dynamic system in the sense that the passive structure and the active membrane are merged into one same unit.

The second objective of a totally embedded dynamism has been studied through the analysis of the role ensured by the active membrane and the structural skeleton.

In kinematic structures where a hard mechanism actuates membranes, skins or surfaces, the main issue is the limit in the movement itself, imposed by the necessity of a fixed primary structure. The project Aegis Hyposurface, already mentioned in the thesis, is the example of such a system. Although the faceted surface is dynamic, the structural «wall» that actuates pistons cannot transform, thus restraining the

A key to reach this challenge was to envisage the interdependency between these parts and eventually make them work as one unit. Indeed, the complementarity and correlativity between the «skin» and the «structure» led to the merge between the actuator and the actuated parts into one same dynamic system. The skeleton is

84


RESEARCH PERSPECTIVES & POTENTIALS

maximal deformation of a facet to the length of a piston. As a result, the combination of multiple 2D translations enables a dynamic surface effect, which is directly constrained by the actuator properties.

capacities define the role of the latter. Most of the time, this question thus does not have to be taken into consideration. As soon as kinetics enters the field of architecture, there seems to be a need for ÂŤclosedÂť system hiding the mechanism responsible for its dynamism. Actuating a continuous surface thus becomes the main goal to reach. However, in the related projects that have been developed so far, discretization of a main surface into a multitude of smaller surfaces or facets is the most commonly used solution, leading to a continuity effect rather than a real and physical one.

As far as movement is concerned, the multidirectional kinetic prototype considerably increases the liberty of evolution in a 3D space. Indeed, as there is no more distinction between what belongs to the actuator and the actuated part, the system does not require any primary structure holding a hard mechanism. Detached from these constraints, the system is independent and can evolve much more freely.

This strategy actually highlights the main challenge for kinetic systems when the actuator and actuated parts are two separate parts. Indeed, the difficulty for such systems is ensuring that the surface or the connections between surfaces can tolerate the movement of the actuator to maintain continuity in the system.

Moreover, the limit of its possible deformation is not directly set by the deformation range of one membrane, but by the combination of the deformation range of all the membranes of the prototype, thus increasing the directions and points that can be reach in a 3D environment. The system, instead of being partitioned and discretized, is continuous.

The particularity of the developed prototype is that the membranes are the actuators and are totally integrated in the system. With this configuration, continuity is not an issue anymore as the membranes, acting as an outer skin, create the movement. Though presenting various membranes, the latter is not discretized.

2. Continuity In the field of kinetic structure and architecture, the issue of the surface, of the skin is central. Indeed, being a visible / acoustic / physical frontier between some 3D environments, the movement of a skin or surface is obviously at stake when dealing with kinetic systems. Behind these notions, the main question that is raised is actually the one of continuity.

On the contrary, all the membranes form a continuous and dynamic skin thanks to their assembly logic and actuation properties. Moreover, the actuator being a surface totally integrated in the dynamic system, there is no distinction between the technical/hidden and the functional/visible faces of the project.

For common industrial robots or machines, the mechanism itself is the main focus as the end-effectors’ properties and

85


a degree of freefom

potentials

As a result, instead of necessitating intricate connections or nodes strategies to enable the formation of ÂŤclosedÂť systems tolerating movement, the inherent surfacic continuity of the SFEAP membranes, acting at the same time as actuators and thus allowing movement, is a key point in the implementation of this prototype in the kinetic architecture field.

The problem is that these components are expensive and that in an optimal situation, one component would be needed for each membrane to enable more possibilities in the prototype deformation. In the future, a solution could be the use of hightech components allowing high voltage. The whole electrical circuit could then be inverted, placing the high voltage converter right after the power supply, or even directly using a high voltage generator.

DRAWBACKS 2. Fragility and Life Span 1. High Voltage Necessity The elastic membranes that enter in the composition of SFEAPs make the use of these material systems quite challenging when thinking about architectural applications. Indeed, once stretched to 400%, they get really thin and very sensitive to any impact or perturbation, that often lead to a break.

In order to be activated and used as actuators, SFEAPs need high voltage. As far as security is concerned, this high voltage is a problem and requires an insulation layer on top of the soft electrodes. The robotized silicone spray process developed during the research could be a solution to this issue, but would still need some improvements. Indeed, the membrane is so thin that any additional layer lead to a change of its behaviour when activated. As a result, to limit the influence of the insulation layer on the SFEAP deformation, the silicone used has to be very elastic, but at the same time resistant when applied in thin film.

Taking into account this stretching factor of the membrane and the way soft electrodes are applied on the latter, SFEAPs cannot support long and constant high voltage activation that often ends in a short circuit. At the level of this research, this considerably limits their life span, but also the fields of possible applications. However, envisaging collaboration with material engineers would probably improve the performances of this material system.

This high voltage necessity also raises the issue of differentiated control. Indeed to get 5KV with a 12V power supply, high voltage converters are needed. In the electrical circuit developed for the research, the digitally control allowed by the Arduino intervenes before the conversion, meaning that for each differentiated activation, one converter is required.

3. Manual Fabrication Process At this stage, the fabrication of the prototype is based on a process which is essentially manual. This obviously leads to a lack of accuracy, meaning that the elements are far from being identical. Of

86


RESEARCH PERSPECTIVES & POTENTIALS

course, developing an entire robotized process would have taken too much time this semester, and it was definitely not he aim of the research. Still, the use of the robot and the reflexion around the improvement of the fabrication process was aimed at proving that even the most challenging steps of the process can be automated.

to a considerable improvement of the strength produced by the activation of an SFEAP.

POTENTIALS At this stage of the research, the evaluation of the prototype’s performances allows to envisage two main application potentials.

Actually, if this prototype was to be developed and industrialized, the whole process could already be robotized. For the membrane, film-stretching machines exist and could stretch the latter to the exact factor. For accuracy and regularity, the conductive and insulating layers could be sprayed with a robotized process that would probably be quite close to the one developed during the research.

Kinetic Model As far as objectives are concerned, this applied research, following an experimental process, was first aimed at proposing new ways of envisaging kinetics in architecture rather than finalizing an industrialized and totally reliable prototype. Although the research did not lead to a prototype structurally efficient in the way that it cannot bear heavy elements, the latter is still based on assembly and nodes logics that make it truly architectural. From this point of view the prototype proposes new solutions for integrating dynamism in architectural systems. Getting rid of the common mechanistic approach, this research led to a material system that differs from others investigations made in the field of soft kinetic structures, as pneumatic systems for instance, in the sense that it merges the actuator and actuated parts into one same system.

Finally, the carbon frame manufacturing could be done by water-jet cutting thin sheets of fibre carbon (mixed with epoxy resin), which is a process that some companies met during the JEC fair (Mateduc Composites) use as an industrialized process. 4. Low Strength Low strength is an inherent drawback of SFEAPs. SFEAPs are indeed capable of large deformations, but provide a very low strength when activated. This property limits the use of SFEAPs and especially in the field of structure. However, the strength developed by an SFEAP is related to the volume of its membrane. Material engineers like Patrick Lochmatter have already proved that multi-layers assembly process can increase a lot the performance of an EAP. As a result, further investigations in this domain might lead

This prototype could thus be envisaged as an experimental kinetic model that starts bringing answers to the issues raised by the implementation of kinetics in architecture, as continuity, joints and connections. Considering the prototype as it is now, which means using SFEAPs with their current properties (large deformation

87


a degree of freefom

potentials

and low strength), the latter could be used for investigations in the field of movement and dynamism applied to architecture.

Coming back to Kas Oosterhuis Trans-port project evoked at the beginning of the thesis, the transformable skin of the building does not seem impossible anymore.

Till today, movement in the domain of architecture shells has mostly been answering the need of responsiveness via local and discrete dynamic elements as paneling logic for example.

And taking an even more radical step, soft kinetic structures could be applied to a larger architectonic field and not only reduced to locally adaptive skins, bringing closer the idea of robot-buildings. From earthquake-resistant technologies to space deformation, the fields of application appear to be limitless.

On the contrary, global kinetics allows for the design of really adaptive and evolutive systems as continuous entities. Therefore this study arises here new solutions enabling this shift.

Considering the potential of the kinetic prototypes that were designed and produced, the idea is now to take a much more radical turn by envisaging and conceiving active and soft kinetic structures for a new architecture. Leaving behind the mechanical limitations, the combined stakes of those dynamics and adaptable behaviours indeed reveal another degree of freedom...

 A Machine for living  Most of the prototype’s drawbacks are related to technical and material aspects and could probably be solved in a further study including for instance collaboration with material engineers. Only the low strength issue, inherent to the current SFEAP technology, still prevents an application dedicated to a structural purpose. One could make the hypothesis that in the direct continuity of nowadays material innovations, the membranes, that are part of the EAP composition and largely responsible of this issue, are going to be optimized so that their resistance to traction and the force developed when activated are tremendously increased, which could also imply the invention of a new similar technology. Once this issue is solved, it seems conceivable to apply such systems to an architectural scale, using the latter as a combination of skin and structure.

88


RESEARCH PERSPECTIVES & POTENTIALS

89


a degree of freefom

BIBLIOGRAPHY

90


ELECTROACTIVE POLYMERS

SHEIL (Bob), “Protoarchitecture Between the Analogue and the Digital”, in Architectural Design – Protoarchitecture : Analogue and Digital Hybrids, vol. 78 - issue 4 (juillet - août 2008), directed by Bob Sheil, John Wiley & Sons, London 2008

BAR-COHEN (Yoseph), Electroactive polymer (EAP) actuators as artificial muscles : reality, potential, and challenges, SPIE Press, 2001. LOCHMATTER (Patrick), Development of a Shell-like Electroactive Polymer (EAP) Actuator, PHD Thesis (ETH Zürich), 2007.

WEST (Mark), “With Matter“, in Architectural Design – Protoarchitecture : Analogue and Digital Hybrids, vol. 78 - issue 4 (juillet - août 2008), directed by Bob Sheil, John Wiley & Sons, London, 2008

JEAN-MISTRAL (Claire), Récupération d’énergie mécanique par polymères électro-actifs pour microsystèmes autonomes communicants. PHD Thesis, (Université Joseph Fourier Grenoble 1), 2008.

KINETIC ARCHITECTURE

KHAN (Omar), Elasticity - the case for elastic materials for kinetic and responsive architecture. Proceedings of the UbiComp ‘09, Florida: ACM Press. 2009

BROWN, (Gary), Freedom and transience of space (Techno-nomads and transformers). dans Transportable Environments 2, eds. R. Kronenburg, J. Lim and Y. C. Wong, 3-13. London: Spon Press. 2003

PROTOTYPE AND EXPERIMENTATION

NEGROPONTE (Nicholas), Soft Architecture Machines, The MIT Press, 1970

BRUCKERMANN (Olivier) for p.art, “Digital Efficiency vs. Physical Necessity“, in From Control to Design :Parametric/Algorithmic Architecture, directed by Tomoko Sakamoto & Albert Ferré, Barcelona, Actar-D, 2008

OOSTERHUIS (Kas) Hyperbody, First Decade of Interactive Architecture. Jam Sam Books, Heijningen, 2012 RANDL (Chad), Revolving Architecture: A History of Buildings That Rotate, Swivel, and Pivot. Princeton Architectural Press, 2008

KHOO (Chin Koi), BURRY (Jane), BURRY (Mark), “Soft Responsive Kinetic System, An elastic transformable architectural skin for climatic and visual control”, in Integration through Computation: Proceedings of the 31st Annual Conference of the Association for Computer Aided Design in Architecture (ACADIA), Edited by Joshua Taron, Vera Parlac, Branko Kolarevic and Jason Johnson. ACADIA. Calgary/Banff, Canada: The University of Calgary, 2011

ROBIDA (Albert), Le Vingtième siècle, Georges Decaux, Paris, 1883. STERK (Tristan d’Estrée), “Shape control in responsive architecture structures - Current reasons & challenges”. Proceedings of the 4th World Conference on Structural Control and Monitoring (4WCSCM):1-8. San Diego, 2006

91


a degree of freefom

ZUK (William), CLARK (Roger H.), Kinetic Architecture, Van Nostrand Reinhold Ed., New York, 1970

FEEDBACK & CONTROL SYSTEMS ASTRÖM (Karl Johan), MURRAY (Richard M.), Feedback Systems, An Introduction for Scientists and engineers. (Princeton University Press 12 avr. 2010 – 424 pages) Version v2.11b (28 September 2012) electronic edition available from http://www. cds.caltech.edu/~murray/amwiki

MATERIAL SYSTEMS MENGES (Achim), “Integral Formation and Materialisation – Computational Form and Material Gestalt“, dans Architectural Design Reader – Computational Design Thinking, directed by Achim Menges et Sean Ahlquist, John Wiley & Sons, London, 2011

MARGOLUS (Norman) TOFFOLI (Tommaso), Programmable Matter : Concepts and Realization, in Physica D47 p.263-272. (Etats-Unis : MIT Laboratory for Computer Science, Cambridge, MA, 02139) 1991

SIMULATION VON BERTALANFFY (Ludwig), General System theory: Foundations, Development, Applications. George Braziller, New York, 1968 revised edition 1976

FISCHER (Al), “Engineering Integration – Real-time Approaches to Performative Computational Design“, in Architectural Design – Material Computation, profile n°216 (mars – avril 2012), John Wiley & Sons, London, 2012 STANTON (Christian J.), Material Feedback in Digital Design Tools, PHD Thesis, Department of Architecture and Engineering Systems Division, Massachusetts Institute of Technology, 71 p. extracted on november 18th, 2013 from the website < http:// dspace.mit.edu/> ; septembre 2009 VARENNE (Franck), Epistémologie des modèles et des simulations : tour d’horizon et tendances [online], Symposium “Les modèles : possibilités et limites“ organised by Société française de Physique (Bibliothèque Nationale de France, Paris, december 10th, 2008), extracted on july 26th, 2013 from the website < http://hal. archives-ouvertes.fr> ; Paris, 2008

92


a degree of freefom

ACKNOWLEDGEMENTS

94


DIGITAL FABRICATION COMPETITION

My sincere thanks go to all those who have contributed to the completion of this research project. They include in particular : • Martin Genet, for being my partner during this one-year research. • Philippe Morel, Pierre Cutellic and Christian Girard, for carefully driving this work and giving helpful inputs. • TeXtreme company for providing us with ultra-light carbon fibre rolls. • Hindoostan Technical Fabrics company for providing us with carbon fibre rolls. • Sicomin company for providing us with hard and soft epoxy resins. • ENSAPM for its financial support. • and finally Marie-Capucine Crosnier, Oldouz Moslemian and Catherine Bourguignon for helping on various aspects of the project.

95


- Jim Rhoné - Master Project Thesis - Research . 10 - A Degree Of Freedom - JUNE14 - ENSAPM - Digital Knowledge -

A Degree of freedom Applied Research in Embedded Kinetics Material Systems and Control for Active Architecture

june14

Master Project Thesis . R10

Digital Knowledge Department

jim rhone

- Ecole Nationale Supérieure d’Architecture Paris-Malaquais - Supervisor : Philippe Morel Assistant : Pierre Cutellic -

A Degree Of Freedom - Jim Rhoné & Martin Genet - Architecture Thesis Project  

Since the end of the XIXth century, architects' spatial desires of flexibility, adaptability, modulation, variation and interactivity have p...