Fib rou s Co ns t r u ct i o n s Wo r ks h o p , D e c e m b e r 2 0 11
C omposite Territories CITA / http://cita.karch.dk / Afd 8/ Afd 2
December 2011 Participants: Paul Nicholas, Martin Tamke, Phil Ayres, Matthew Gilbert & Students from Departments 2 and 8 Supported by: Det Frie Forskningsråd (FKK) – Postdoctoral Grant ‘Designing Material, Materialising Design’ for Paul Nicholas 2011-2012
CITA
Paul Nicholas- Variable Stiffness Monocomposite
Norbert Palz, Additive Fabrication of Heterogenous Materials
Graded Materials This workshop explored how architects might simultaneously be designers of material as well as form. Today, material innovation is increasingly occuring around the possibility to meet performance criteria via the design of material, rather than solely through geometric or mechanical solutions. Fibre reinforced composites in particular allow materials to be designed for specific properties and specific contexts, and for their properties to be graded in the same way as many natural materials.
But the idea that a material property such as bending might be desired, rather than minimised, is new to architecture. So is the idea that materials might be incorporated into the form–making process, rather than form being imposed upon them. The object of this exhibition is then to rediscover the capacity of a material to bend, not as failure or a shift from the normal but rather as something that can be tailored to specific design concerns.
Sean Hanna- Variable Modular Material
Fib rou s Co ns t r u ct i o n s Wo r ks h o p , D e c e m b e r 2 0 11
Composite wind turbine blade
Fibre-reinforced Composites Composites, which represent some of the oldest building materials as well as the most modern, combine multiple materials to create a new material with properties beyond that of its components. They emerge from two basic ideas. The first of these is that, if a material does not exist, it can instead be made through the combination of two or more component materials. Secondly, that the properties of this new material, for example the mechanical relationship between form and force that it exhibits, is dependent upon the organization of these components within the material.
Through the design of the composite’s substructure, it is possible to control load transfer and mechanical deformation in bending, flexure, tension and shear. By altering the relationship between form, substructure and loading across a composite, the same material or structure can become variegated and possess specific properties in specific places and directions. As designed synthetic materials, composites form the basis for an expanded architectural practice, but their use implies the navigation of an unknown space, outside the properties of more familiar materials and design processes.
Controlling material properties within a composite through varying fibre orientation
CITA
Charles Darwin- engraving showing the variable properties of a climbing plant
Research Questions The workshop explored how the design of material can expand the architectural design space. Today, material innovation is increasingly occurring around the possibility to meet performance criteria via the design of material, rather than solely through geometric or mechanical solutions. As designed synthetic materials, composites form the basis for this expanded architectural practice, but their use implies a more active role for material in the design of form, outside that of more familiar practices. Within this workshop we sought to simultaneously design the material and the object. Taking a point of departure in the material properties of glass fibre reinforced
thermoplastics (GFRP), we explored the possibility to specify our own material properties, ie. bending. The workshop focused on the detailed design and fabrication of a structure in which each element is designed to have its own specific bending behaviour. Instead of imposing form onto material, this approach provided a way to specify a global form through the local design of material - to program the bending behaviour of 2D elements so that they support a complex 3D shape.
Guiding research questions were: How do we design for the highly variable and specific material performance that composites afford? How does this new practice lead to new requirements for digital tools? How do we employ and parametricise modeling processes that are multi-scale? How do we link emperical testing with digital specification?
Fib rou s Co ns t r u ct i o n s Wo r ks h o p , D e c e m b e r 2 0 11
This text explains the graphic. This text explains the graphic. This text explains the graphic. This text explains the graphic. This text explains the graphic. This text explains the graphic. This text explains the graphic. This text explains the graphic. This text explains the graphic. This text explains the graphic.This text explains the graphic.
This text explains the graphic. This text explains the graphic. This text explains the graphic. This text explains the graphic. This text explains the graphic. This text explains the graphic. This text explains the graphic. This text explains the graphic. This text explains the graphic.
1:1 Demonstrators built during workshop
Introductory Workshop1 This workshop sought to devise and develop design intent, detail and construction of a graded material system. The material used was glass fibre reinforced polypropylene tape. The end goal was to build speculative scale models, full scale prototypes and a test demonstrator. The vast majority of building objects surrounding us are made of compositions of homogenous materials. Systems that require different types of behaviour are typically described through the assemblage of parts, each fulfilling one particular aspect of performance, ie. hinges, joints, struts, sheets. But imagine if these systems of linking and joining were replaced by a differentiation in material and
shape? For example, how might material properties change as needed for the different areas of a chair or the functioning of a window or door? This idea was the departure point for speculation about areas of application.
1:1 Demonstrators from workshop
CITA
Workshop2
Fib rou s Co ns t r u ct i o n s Wo r ks h o p , D e c e m b e r 2 0 11
Consolidated Composite Element
CONSOLIDATION PROCESS Application of heat and pressure
PRE-PREG LAYUP Stack of single sheets
Material testing
CITA
Number of layers of tape
10
Bending under load
Layers
8
6
4
2
7
14
27
147
310
14
24
63
215
333
23
47
114
280
380
33
84
172
293
67
146
254
50
Load Load (grams)
100
200
400
800
Deflection amount (mm) 82
1000
333
Each beam is composed of a variable number of composite strips
Fib rou s Co ns t r u ct i o n s Wo r ks h o p , D e c e m b e r 2 0 11
Testing larger section of structure against arcs derived from varying radii
Material testing In contrast to more traditional building materials, a very slight increase in the thickness of a composite can have a very large impact on its stiffness. Each beam element is made from multiple layers of tape, each layer being 200 microns thick. Empirical testing was used to establish the relationship between load, number of tape layers and deflection, both on single beams as well as more developed topologies. The results for single beams were compiled as a matrix (left), which at a later stage is connected back to a digital model.
CITA
Sketch design of shape in gallery space
Form & Program An underlying form was developed specifically for the gallery space, taking spatial considerations, circulation paths, human scale and ideas about the exhibition of additional information into account.
Fib rou s Co ns t r u ct i o n s Wo r ks h o p , D e c e m b e r 2 0 11
Growth seqence
Pattern Growth A pattern was grown over the underlying form. A bespoke branching tool was developed in grasshopper, alongside conceptual sketches of how such a tool should perform and the effects that it might achieve. This tool allowed for precise control over the length and branching angle, as well as the the joining of separate branches.
Sketch designs of pattern logic on surface
Sketch designs of pattern logic on surface, Visualisation of load carried by each beam
CITA A2
A1
B2
D
HIGHER 77
100
59
B1 40
LOWER
12
L
Diagrammatic representation of beam specification code
Determining the amount of deflection required to match the underlying form
Property Matrix
Deflection
6 LAYERS
200
4 LAYERS
16 LAYERS
10 LAYERS 8 LAYERS
20 LAYERS
100
30 LAYERS
100
200
300
400
500
600
700
800
900
Load
1000
1100
1200
1300
1400
1500
1600
1700
Fib rou s Co ns t r u ct i o n s Wo r ks h o p , D e c e m b e r 2 0 11
Geometric base
Result of specification process
Structural performance
Relating Micro & Macro Performance Specifying the stiffness of each element means considering two interconnected design problems. At the scale of the individual beam element, the problem is to specify the right number of layers so that the beam deflects by a particular amount under loading. At the scale of the structure, the problem is to achieve strength and stability given a specific restraint condition, while minimising material use. The number of layers specified for each beam was calculated locally, using an iterative algorithm that connected the 3D model directly to the material testing. The algorithm assessed the particular loading condition of each beam, the deflection that it should make to best match
the underlying surface at that point, and accessed the property matrix to find the closest match from the testing process. The pseudocode is as follows (refer to diagram on opposite page): For all beams If A1 and A2 are assigned layers: Calculate their weight, plus that of the beams they support Divide that weight by number of supporting elements (B1 and B2) Calculate the required bending deflection for B1 Assign layers to B1 using nearest curve on the property matrix End if Next
Fib rou s Co ns t r u ct i o n s Wo r ks h o p , D e c e m b e r 2 0 11
Final distribution of beams throughout structure
Fib rou s Co ns t r u ct i o n s Wo r ks h o p , D e c e m b e r 2 0 11
Flow of force through the structure
Prior to optimisation process
After optimisation process
Taking Structure into Account An approach to layer specification that only considers the micro condition cannot take advantage of macro topological conditions. If only the initial specification process is used, the structural specification tends to ‘max out’ at 30 layers. To refine the specification process, the data gathered from empirical testing was also used to calibrate a Karamba FE model, which included material definitions and beam thicknesses. Using this model, initial layer specification is modified via an understanding of loadpaths within the structure, but only where there is no impact on the bending performance. This results in greatly reduced beam thicknesses.
4 6 1mm
8 10 16 20 30 7.5mm
4 6 1mm
8 10 16 20 30 7.5mm
Reduction in beam thickness: change in distribution
CITA
Connection Detail; Pouring plaster to make the footings
Connections & Detailing By overlapping material and using a system of bolts and washers, we ensured that bending only occurs within the beams, and not at the nodes. Using bolts kept the weight of the connection detail to a minimum (together, the elements of a node weigh just 6g). Wall connection details and plaster-cast footings were also developed to situate and stablise the structure in the gallery space.
Fib rou s Co ns t r u ct i o n s Wo r ks h o p , D e c e m b e r 2 0 11
Automating varied beam connections directly from the 3D model
Fabrication Information and Construction All information required for assembly was transferred directly from the 3D model to each beam via a template. By assessing each beam and its neighbouring beams, we were able to create a common intersection shape (below right) which informed the geometry of each end. Printed templates, which specified neighbouring elements, orientation and fixing positions facilitated assembly without any further information being needed.
Automating beam templates
CITA
Image credits: Anders Ingvartsen
Image credits: Anders Ingvartsen
Fib rou s Co ns t r u ct i o n s Wo r ks h o p , D e c e m b e r 2 0 11
Image credits: Anders Ingvartsen
Image credits: Anders Ingvartsen
Fib rou s Co ns t r u ct i o n s Wo r ks h o p , D e c e m b e r 2 0 11
Image credits: Anders Ingvartsen
Image credits: Anders Ingvartsen
Fib rou s Co ns t r u ct i o n s Wo r ks h o p , D e c e m b e r 2 0 11
Image credits: Anders Ingvartsen
Fib rou s Co ns t r u ct i o n s Wo r ks h o p , D e c e m b e r 2 0 11
Image credits: Anders Ingvartsen
Fib rou s Co ns t r u ct i o n s Wo r ks h o p , D e c e m b e r 2 0 11
Image credits: Anders Ingvartsen
Fib rou s Co ns t r u ct i o n s Wo r ks h o p , D e c e m b e r 2 0 11
3D scan Comparison of initial specification geometry (blue), predicted geometry from Karamba (red), and as-built structure (grey)