Digital Design - Module 02 Semester 1, 2018 Saran Kim
904662 Joel Collins + Studio 15
Reading: Kolerevic B. 2003. Architecture in the Digital Age
Kolarevic described three fundamental types of fabrication techniques in the reading. Outline the three techniques and discuss the potential of Computer Numeric Controlled fabrication with parametric modelling.
For the fabrication of three-dimensional objects, Kolarevic discusses subtractive, additive and formative fabrication techniques. Subtractive fabrication involves the removal of material from the volume. The number of axes a milling machine has determines its flexibility and limitations to the object form it can produce. Additive fabrication is the technique of incrementally layering material to form a solid. There are various materials such as liquid polymer, and curing processes based on light, heat or chemicals (p.36). Formative fabrication involves the deformation or reshaping of material forms through the application of mechanical forces, heat or steam. Compared to the subtractive and additive fabrication, formative fabrication could be employed on a large-scale work with precision using the numerically-controlled system for pins and bending of materials. Computer Numeric Controlled fabrication provides the accuracy of the measurements and forms of the materials which are crucial for constructing buildings designed with parametric modelling.
Surface script The script of two surfaces starts with constructing a cube and extracting edges using DeBrep command. Four curves are selected using List Item command, divided into five parts and points are extracted. Then lines are drawn using Line command and the surface is made by lofting lines. I used Surface Domain Number to create a 5 x 5 grid on each surface. In order to create the variations to the height and directions of panelling units, I used Offset Grid command to lift the grid off the surface and Curve Attractoion command to move the location of points. Another method I used to move the grid structure was the sequence of List Length -> Series -> Graph Mapper -> Bounds -> Remap Numbers -> Z vector -> Move. This allowed individual points to be moved by different units in Z direction. The moved grid, original grid and the panelling unit become inputs for 3D Morph command. Four panelling unit designs were chosen and applied onto surfaces.
Week Four Panels & Waffle
I selected surfaces with two distinctive panelling unit designs: design 1 and design 4 (refer to page 3). While design 1 features geometries predominantly formed by triangular shapes, design 4 consists of a series of triangular prisms which vertices attracted in different directions.
Due to the structural properties of two surfaces, the waffle script introduced in the workshop could not be applied in producing my waffle. Since one surface was touching the ground level only by end points, I had to create a triangular base to stabilise the waffle structure. As a result of creating the base, the waffle structure became more complex; the final iteration consisted of two types of fins for one surface, two more types for the other surface and one type of horizontal plates for holding all together.
In selecting panelling units for two surfaces, I had to consider their ability to unroll them without overlap of sides. I explored how to unroll units by exploding, and modifying the locations of the unjoined edges of the source geometry (unselecting edges and joining other edges). This influenced the way panelling units are developed - unjoined edges are always separated on the unrolled surface and this affected whether unrolled surfaces are overlapped or not. I found that some panelling units could not be unrolled when Weaverbird frame command was used, which in reverse informed the ideal properties of panelling units. After the experimentation, I found that design 1 and design 4 could be unrolled without any major overlap. Especially, pairing up two horizontally adjacent panelling units of design 1 led to reduction of the number of developed pieces, which increased the efficiency of model making process.
Panelling unit (design 4) Panelling unit (design 1)
Waffle (1mm mountboard)
Initially I had the folding lines of unrolled surfaces on the etch layer of the laser cutting template. However, etching on the ivory card did not allow me to fold in both directions with crisp edges. I then took the advice on the use of dashed cut lines for folding lines, which helped me create the physical model. In regards to nesting, I attempted to remove duplicated or overlapped lines by using Make2D command. Since panelling units had irregular shapes, I tried to minimise the wasted space by clustering in groups.
Yellow lines showing unjoined edges - different iterations by chaning edges to be unjoined.
For scripting task 2 model, I learnt from task 1 that it is important to keep groups of commands arranged by design iterations in order to make it easy to look back afterwards. Firstly, I created a cube with partitions that create nine cubes. Then I inserted different attractor commands such as Curve Attraction, Point Attraction and Random Attraction in order to manipulate the internal structure of the cube. Then I used Volume command to find centroids, which locations are then manipulated by combinations of Curve and Point Attraction. The geometry is located on centroids using Orient and XY Plane command. For producing boolean interations, I utilised Scale / Scale NU command with remapped distance between reference points and centroids. After creating design 3, I found that the scale of geometry units were not intersecting with the cube surfaces enough, so I decided to increase the scale to make them interact with the volume.
I selected design 4 as it had a dynamic relationship with the original cubic volume. The boolean units for design 4 were rotated twice in Z direction and X direction using the radian input of the distance between a reference point and centroids, which gave unique characteristics to individual boolean units. This portion of the subtracted volume was chosen as it showed interesting relationships with the trimmed sections.I decided to trim the cube in the triangular prism form as it has a connection with the form of the boolean unit made of triangular surfaces. While one of three surfaces has smooth solid space (hidden in the isometric view), two other surfaces (visible in isometric view) consist of dynamic relationship between different forms of voids and carved surfaces. The vaiations of the volumes of voids give a rythmic movement. The complexity of the negative volume is emphasised by the angular edges. The voids are interlocked, which allows the flow of air in terms of the permeability. In conveying the form of the volume in the isometric view, I decided to highlight three visible sides in three different shades for the clarity of the form.
Week Six Task 01
Task 01 Matrix For the forms of the surfaces, I selected 1.3 design as it showed more unique relationship between two surfaces compared to other designs. This design features one surface facing up and the other surface facing down, which would be interesting in terms of their interaction with light and shadow. The selection of panelling grid was based on the dynamic changes of the offset grid. I chose 2.2 design which had the great vertical differences. I selected two different three-dimensional panelling units for surfaces, in order to explore the aspects of juxtaposition and contrast between two different structures, characteristics and the spatial qualities.
Week Six Task 02
Task 02 Matrix For the grid manipulation, I explored different attractor commands. I found the internal structure formed by Random Attraction command very interesting as it created the complexity to the relationship between angles of the internal edges. The distribution of centroids was influenced by combination of Attraction commands. 2.3 design iteration was nicely spread over the cube while having clusters that do not retain the original alignment of the centroid grid. The unit transformation involved changing the scale of individual units and rotating them using the remapped distance between a reference point and centroids. I selected 3.4 design as it shows the sense of movement in individual units due to the variations of rotation and scale.
Final Isometric Views
Steps involved in the Task 1 panelling process
Offset Grid command to lift the grid off the panel surface. With the set curve geometry, I used Curve Attraction command to manipulate locations of points of the grid.
The exploration of the panelling unit. The screenshot above shows one of designs I found too complex for the task 1 model. Using Mesh command made panelling unit developable. I also used MeshtoNURB command to allow Make2D command to take place.
Experimenting with Cellulate 3D Grid command. The form created between two grids consists of rectangular prisms, which do not offer a depth of unique elements.
Experimenting with Morph 3D command. The option to set Mesh geometry into PatternObjects input provides flexibility to the design of the surface.
Experimenting with Weaverbirdâ€™s Picture Frame command. The scale of the picture frame varied by the distance between a reference point and the grid points.
The basis of the panelling consisted of Surface Domain Number -> Offset Grid -> (Curve) Attraction -> Morph 3D command with Geometry input and two Grid inputs.
Waffle making process
Contouring two surfaces. I experimented with different vectors (X, Y, Z) to find the most suitable waffle structure. I found that Y vector gave the structure that provide relatively even spans.
Moving the contoured lines by 10 units in Y direction and using Loft command to create surfaces. However, it started to twist and did not create uniform fins.
Using Orient command with YZ Plane instead of Move command with Y vector created parallel lofted fins with consistent width.
Some outputs of lofted surfaces had data tree structure that gave more elements than needed so I used Cull Index command to reduce the number.
I placed XY Planes at the intersections of fins and created a rectangular prisms using Rectangle command.
The orientation of rectangular prisms had to be corresponding to the direction of fins. The length of the rectangular prism had to be adjusted for each group of intersections in order to make sure that they are touching the edge of fins.
Script for the waffle structure
Script for the waffle structure Panelling units that were too complex for Task 1
Making of the panelling unit for Design 1. A square lifted by Gumball and rotated by 45 degrees.
I arranged unrolled surfaces in the grid form. I labelled with alphabet and numbers to identify each of them.
Steps involved in the Task 2 Boolean process
Using multiple Point Attraction commands in a sequence in order to give more variation to the structure of the centroids.
Using Ghosted view to examine the relationship between 15 x 15 x 15 cube and the manipulated Boolean units inside the cube.
I created point and surface geometries from triangular surface (made in Rhino), extruded to form pyramids and combined them together using Solid Union command.
I utilised Orient command to position centroid of the Boolean unit onto XY planes of centroids of the nine cubes.
I baked the cube and Boolean units to manipulate in Rhino. I used BooleanUnion command to merge all units together, then BooleanDifference command to subtract them from the volume.
I created a large rectangular prism for spliting the subtracted volume into one eighth of the original volume.
Model making process
The first attempt with etched folding lines did not work as they did not fold in both directions.
After lasercutting the waffle fins. Peeling off masking tap had to be done carefully to prevent tearing.
Before removing fins off the base mountboard, I marked each fin with identification numbers.
Completed waffle structure.
Removing unrolled surfaces off the Ivory card.
Folding the unrolled surface according to the digital model and sticking tabs together.
Model making process
I made units in rows, however I found that they curl up which made assembly step very difficult.
Folding the unrolled surfaces of the second panel. Initially I was using UHU but I found it less effective than PVA.
Connecting the second panel together by sticking corresponding tabs.
Attaching panels onto the waffle structure. I stuck tabs that align with the fins of the waffle.
Putting tabs together individual units to form two surfaces using PVA glue.
I trimmed sections of tabs to create nodes for waffle fins to interconnect. I referred to the digital model to make sure that the panel is correctly located onto the waffle.
Task 1 Final model
Task 2 Final model