Booklet: Graded Territories / Composite Territories (2011)

Composite Territories

Smart Geometry 2012, Troy, USA, Material Intensities CITA: Centre for IT and Architecture Royal Danish Academy of Fine Arts, School of Architecture RPI: Rensselaer Polytechnic Institute

http://cita.karch.dk

http://www.smart geometry.org

Composite Territories Smart Geometry 2012 Troy

Collaborators, Participants, Sponsors Collaborators: Paul Nicholas, Martin Tamke, Mette Ramsgard Thomsen, Hauke Jungjohann, Ivan Markov, Jan Knippers, Andrius Vilcinskas Participants: Felecia Davis, Kyuseon Hong , Benay Gursoy, Lucille Ynosencio, Michael Drobnik, David Mans, Thomas Wortmann, Nicholas Waissbluth, Moa Karolina Carlsson, DeWitt Godfrey, Christos Constantinou, Kieran Martin, Kateri Knapp Sponsors: Smart Geometry, Sofistik, Polystrand The workshop ‘Graded Territories’ was run as part of SmartGeometry 2012, at Rensselaer Polytechnic Institute (RPI). The four day workshop developed around a collaboration with Knippers Helbig Engineers and access to RPI’s material testing facilities. Through the process of designing and building a bending-active structure, the workshop explored how mechanical testing, and the measures generated through testing, could inform the modeling of composite behaviour

when linked to time-based simulation methods.

The aim for Graded Territories was to extend thinking about processes for measuring, specifying and embedding of forming information for variable stiffness composite structures. Mechanical testing brings to this problem an improved understanding of what happens when forming information is directly embedded into composite materials,

while engineering software provides a means to simulate the behaviour of materially differentiated assemblies. In feeding the simulation process with information drawn directly from mechanical testing and measurement, the process developed for this installation was able to link transformations at the scale of element with that of the structure, and to reveal some of the subtlety around the influence that a material’s history has upon its current behaviour.

Introduction The workshop Graded Territories took place at the 2012 SmartGeometry Conference in Troy, USA. The conference, titled Material Intesities, invited selected workshops that answered the following call:

enclosing physically sealed spaces. Spaces and boundaries are abundant with vibration, fluctuating intensities, shifting gradients and flows. The materials that define them are in a continual state of becoming: a dance of energy and information.

Material Intensities Imagine the design space of architecture was no longer at the scale of rooms, walls and atria, but that of cells, grains and vapour droplets. Rather than the flow of people, services, or construction schedules, the focus becomes the flow of light, vapour, molecular vibrations and growth schedules: design from the inside out. The sg2012 challenge, Material Intensities, is intended to dissolve our notion of the built environment as inert constructions

Harnessing Material Potential Material potential is defined by multiple properties: acoustical, chemical, electrical, environmental, magnetic, manufacturing, mechanical, optical, radiological, sensorial, and thermal.

The challenge for sg2012 Material Intensities is to consider material economy when creating environments, micro-climates and contexts congenial for social interaction, activities and organisation. How can multiple properties of the same material be harnessed? This challenge calls for design innovation and

dialogue between disciplines and responsibilities.

Simulation, Energy, Environment sg2010 Working Prototypes strove to emancipate digital design from the hard drive by moving from the virtual to the actual in wrestling with the tangible world of physical fabrication. sg2011 Building the Invisible focused on informing digital design with real world data. sg2012 Material Intensities strives to energise our digital prototypes and infuse them with material behaviour. They gain the potential to become rich simulations informed by the material dynamics, chemical composition, energy flows, force fields and environmental conditions that feed back into the design process. http://cita.karch.dk

Composite Territories Smart Geometry 2012 Troy

Composite Materials As Case Most architectural structures have been compressionbased and the shape of the structure, as opposed to the calculation of material properties and stresses, has been the primary factor governing stability. The process of designing the correct geometry involved careful reference to guarded systems of geometric compositional rules, developed around long experience of particular materials.

Large scale models were a key tool against which this structural understanding could be represented, tested and proven - if the model was stable then the building would be too (Heyman 1998).

In response to the development of new materials and their application within new structures across the 18th and

20th centuries, this geometric understanding was replaced by an approach based on the calculation of material properties, loads and deflections. An accompanying interest emerged for specifying materials based on their mechanical properties and behaviour under loading, and of materials and structures based on properties other than compression. Examples include the tensile form-finding and thin shell projects of Otto, Isler and Candela (Isler 2008). These structures are termed formactive, meaning that their form and the forces applied to it are interdependent –they differ from previous types of structure in that they are form-found, based on material behavior and structural characteristics (Menges 2010).

Today, fibre-reinforced composites extend these material

understandings by providing a means to program performance through design. Composites are made from the combination of two of more component materials, most typically a fibre and a polymer matrix. The mechanical properties of a composite, for example the mechanical relationship between form and force that it exhibits, are directly dependent upon the organization of these components within the material. While research into composites to date has largely concentrated on one area of performance, the optimisation of resistance to deformation, a second and less explored way of understanding composite performance is that of designing material to exhibit a controlled deformation.

Bending Active Structures Bending-active structures derive their form from the erection process, during which it they are elastically deformed. Where traditional building are considered to have failed if they bend, bending active structures utilize a material’s capacity to bend elastically to generate 3D curved geometries from initially straight or planar 2D building elements. Their design is made possible by simulation techniques that link form-finding and analysis, and the prediction and steering of bending is a central part of the design process.

only takes place if the cantilevers are relatively long. Simulation approaches are based on the relatively recent ability to iterate within FE simulation, analyzing the structure according to third-order theory (material and geometrical non-linearities are considered) and an iterative equation solver for convergence.

The SG2012 design focused on the instrumentalisation of elastic deformation where shape is only acquired through the bending of flat strips – either under self-weight or the compressive loading of its end positions. The benefits lie in Prediction is made difficult by the fact that GFRP has great potential the system’s simplicity where material properties determine for large deformations since the stiffness is low and the strength is the geometry while joints are simple and standardised. The high. Additionally, the GFRP elements used for Graded Territories design objective was to develop a lightweight, resource efficient are very light, and bending in the cantilever structure under gravity structure. http://cita.karch.dk

Composite Territories Smart Geometry 2012 Troy

Pre workshop testing: programming and recording behaviour

http://cita.karch.dk

Composite Territories Smart Geometry 2012 Troy

Varying element stiffness material & form

3D scan

Varying structural stiffness geometry & form

3D scan

http://cita.karch.dk

Composite Territories Smart Geometry 2012 Troy

6 LAYERS

200

4 LAYERS

10 LAYERS 8 LAYERS

100

100

200

300

400

making

testing

form& program

local specification

3D scan

assembly

fabrication

advanced simulation & iteration

Process Overview

500

600

7

Design geometry

Design The design of the Graded Territories installation deliberately reduced the level of structural and formal complexity so as to isolate bending. Within the structure, shape is only acquired through the bending of flat strips which have one of two distinct conditions. The first of these is the cantilever. In this condition, typically combining five sub elements of different bending stiffness into a total length of 2.5m, bending is induced by the self-weight of the structure. The second condition is a fixed compression arch, where the strip is restrained at both ends. This kind of strip typically combined 10 sub elements and had a total length of 5m.

As with Composite Territories, the overall shape of the installation was designed in a top down manner, based on

an experiential understanding of the design space occupied by the interaction of restraint conditions, bending stiffness and self-weight. This research was developed in the lead-up to the workshop, and utilised 3D scanning as a means for measuring the geometries that resulted from varying configurations. In order to link the material behavior with design intent, the target -based approach used in the Composite Territories installation was re-used. In this approach, a shape is firstly defined, then a material specification is made based on measurements gained from material testing. The target shape is then approached as closely as possible via simulation, within the framework and limitations of material grading and the possibilities of the overall system.

3D simulation

http://cita.karch.dk

Composite Territories Smart Geometry 2012 Troy

300

0.9mm 3 layers

1.2mm 4 layers

1.5mm 5 layers

1.8mm 6 layers

2.1mm 7 layers

2.4mm 8 layers

200

100

10

20

30

40

50

60

70

80

90

100

Load (newtons)

Testing An Instron Universal Loading Testing Frame was used to load-test material samples of each layering configuration used within the structure. Because the design comprised of independent cantilever strips and arches with no interaction between them, the properties of interest were stiffness in bending and to the lesser extent stiffness in tension.

Dog-bone samples were used for tensile tests and strips for bending tests. Both types of test were carried out in the direction along the fibers, and multiple tests were performed to obtain average values. The samples were tested in deformation-controlled settings where specimens were tested up to failure in tension and up to reasonable deformations in three point bending test.

This process of mechanical testing revealed behaviours

that were overlooked in the Composite Territories project.

While the tension tests revealed linear behavior far beyond loads experienced in the final structure, the cantilever and four-point bending test bending tests revealed highly nonlinear bending stiffness. This was especially apparent for specimens fabricated with only a small number of layers. As with Composite Territories, heat and pressure were used to consolidate variable numbers of layers of pre impregnated composite material, and what the tests revealed was that during the process of consolidation, the material properties of the assembly were changed. The underlying reasons for this change were related to the length of time in the press, the level of pressure applied, or discrepancies in local material properties that were more apparent with the increased use of thinner

elements.

http://cita.karch.dk

Composite Territories Smart Geometry 2012 Troy

Local specification In graded material systems, the relationship between material differentiation and shape under loading is very sensitive.

Within the simulation process, the starting point is very important. For these reasons, a specification process needs to precede simulation, but to still consider load, deflection and target in order to determine a specific bending stiffness.

One computational approach to estimate the stiffness to be specified is to feed back the stress-strain curves generated from material testing into the design process. An algorithm was written to iterate over each member within the structure, assessing its loading condition (the load of those elements it supports), the deflection that is required to match the underlying ‘target geometry’, and through reference to a lookup table containing the load-deflection relationship generated

during material testing assign a number of layers. Once the load and required deflection is calculated, the closest loaddeflection relationship determines the layer assignment for that beam element. While an iterative simulation process capable of optimizing material organization would render this step unnecessary, such an approach would also be prohibitively time expensive. Instead, this approach provides an achievable method of quickly specifying bending stiffness by increasing the information from material testing directly incorporated within the design process.

Advanced Simulation & Iteration In order to get a more complete understanding of the materials bending the data from the parametric modeling environment where translated to the Finite Element Analysis (FEA) software SofistikTM. Here not only the element based system used so far had to be changed into a node based one but moreover the FEA needed the initial flat state of the strip geometry – where the model so far represented the design target. A python script provided the FEA with all necessary parameters as node coordinates, conditions for connections and support at each node, material properties (E-modulus, specific weight) and specifications for all potential combinations of beam dimension and material. Finally the programming of the load cases within FEA was automatically generated by the http://cita.karch.dk

Composite Territories Smart Geometry 2012 Troy

Stills from Sofistik Simulation Process

Fabrication using a Water Jet Cutter

http://cita.karch.dk

Composite Territories Smart Geometry 2012 Troy

Structure 1 beams

Structure 2 beams

http://cita.karch.dk

Composite Territories Smart Geometry 2012 Troy

3D Scanning of Structure Design geometry

3D scan

Demonstrator The design was limited to two distinct structural conditions to isolate the property of bending. The first was a cantilever, restrained at one end. In this condition bending was induced by the self-weight of the structure. A cantilevering strip typically combined 5 sub elements of potentially different bending stiffness, and had a total length of 2.5m. The second condition was a fixed-in compression arch - a strip restrained at both ends, which typically combined 10 sub elements and had a total length of 5m. This reduction of structural and formal complexity allowed the workshop participants to perform an FE analysis on their own.

In addition to the specification of stiffness, this installation explored other physical properties such as colour. Thermochromatic inks were embedded within the composite elements, which then changed their colour in response to temperature.

http://cita.karch.dk

Composite Territories Smart Geometry 2012 Troy