AIR

ALGORITHMIC SKETCHBOOK

ARCHITECTURE DESIGN STUDIO AIR SEMESTER 1 2014 HENRY CHHEN 586676

TASK 1: LOFTING Creating curves on Rhino

Loft created through the use of grasshopper

2

GRASSHOPPER TREE: BASIC LOFT

A simple pathway showing how to loft curves. Each curve was allocated to its own curve definition on grasshopper. Then each definition was grouped and numbers, This reminds me of maths, how variables are added to the formulat to generate an outcome. In this sense, a design is made instead.

3

LOFTING IN DIFFERENT MODES The behavior of each loft type is evident and relevant to its name. Normal, tight and uniformed look very similar. Straight is very flat. Developable is the most interesting of all, due to its more static shape and shiny surface.

4

Normal

Straight

Tight

Uniformed

Loose

Developable

LIVE LOFTING: ALTERNATING CURVES Curve

Loft

5

Each curved was altered by change its control point, and due to grasshopper, with the change in curve, the loft was respond as well. Regardless of the complexity or large change, lofts were successfully made.

6

TASK 2

Conversion from points to lines to surfaces to curves and back to points 1: Creating points Points were first created in rhino, then registered each as a separate point in grasshopper

2: Converting points to lines The points were then connected to its corresponding neighbor to create a basic shape using the point to point line tool

7

3: Creating a surface from the lines

The surface was created using the patch tool. This tool is extremely flexible as it can create straight sided surfaces with any proper input.Thus all the lines were connected to the patch toolâ€™s C (curve), to generate the surface that follows the lines boundary.

4: Converting surface back to curves

The breb wireframe tool was then used to convert the boundary of the surface into curves. The cuves have been created, but despite having the lines from the previous steps hidden, the curves were of the same colour, thus making the picture look like nothing had changed.

8

5: Convert curves to points

From the curves, the divide curve tool is used to create points running along the curves. This tool allows users to create as many points along the curve as they want by inputting the desired variable into its N (count - number of segment).

9

EXPERIMENTATION Lofting contour lines

This piece was purely a mistake, but the resultant was great. From doing what I assumed a simple task on grasshopper, I had a general idea of what the end result would look like, but I had a miscalculation and accidently created a beautiful mess. My initial plan was to create a simple surface, make contours on it ,project a copy of those contours in the x/y plane and have those two contours loft together. Instead, the contours lofted with itself, and lofted with each other, which made this surprising result.

10

11

TASK 3

Create an interesting surface pattern. The making of the pattern 1: Outline

A simple line square was first made on rhino, and then registered as a curve on grasshopper. A copy of this curve was projected in the Z direction with the value of 1. Then with the first square, and its projected version, they were loft together to make this outside wall.

2: Top surface

With the projected curve, it was scaled down by a factor of 0.5, and lofted to create the top surface.

12

3: Second box

With the scaled down curve, it was moved up by one unit in the z direction. Again it was lofted to create the smaller walls. The curve that moved up was also scaled down by a factor of 0.5, and lofted to make the top surface.

4: Third and final box

As with the above, curves were moved up, scaled and lofted to create the top box. Thus the pattern looked like a simple pyramid.

13

5: Final Pattern

To make the pattern look more interesting, the top and second top surfaces were disabled to grasshopper, to make gaps to look through in the pattern.

14

TASK 3

Create an interesting surface pattern 1: Creating the surface

The curves were created in rhino, registered in grasshopper and then lofted. 2: Creating points on surface 4: Third and final box

The surface was first divided using the divide surface tool on grasshopper. Domain2 and divide domain2 were added to create domain segments. The input of U and V on both divide domain2 and divide surface were both 10 on the number slider. The N value (normals) of the divide surface tool, was flattened and multiplied by a factor of 2 (to increase the variable), and plugged into the deconstruct

15

The deconstruct vectors was again multiplied to greater a greater variance, then added to the bounding box, it create the box layouts on the surface. The height and amount of box tacked is dependent on the points placement in the xy plane. Final pattern looked decent but it wasnâ€™t what I imagines. The pattern protrusions werenâ€™t high enough, and it was too wide, hence the scale was off. I need to divide the surface into small points, stack the pattern box higher, and increase the multiplications of the variables.

16

17

TASK 3

Final pattern

The pattern was finally added to the bounding box to create the surface. There wasnâ€™t a large variance in height as I would have hoped, but altering the parameters to make it higher, looked worst. I believe I need to divide the surface into more points first before adding the pattern, so that the patterns height would look proportionate to itâ€™s width.

18

EXPERIMENTATION

Curved transformation and grid shell For this experiment I decided to combined the things I learnt from the videos transform menu and creating a grid shell. I ideally I wanted to create a grid shell that had long skinny column protrusion of various heights. So using what I learnt, I made the dome, divide the surface into various surface boxes and used a skinny column as the geometry of the box morph.

I meet the limitation of my desktop computer. When I tried to divide the surface even more, and increase the change in height in the columns, my computer lagged greatly. Thus I reached the limitation of what my computer can do. It wasnâ€™t want I imagined it to be, but my goal was far too complex to create.

The pattern wasnâ€™t what I desired it to be. The columns were short and fat, opposite to what I wanted. So I made the surface be divided into much smaller points, which somewhat fixed the issue to a small extent. This is where

19

20

FRACTAL TECTRAHEDRA 1: Creating the triangle base

The triangle was created by projecting the polygon tool in the xy plane, and giving it the required 3 corners (segments).

2: 3D triangular pyramid

To make the triangular prism, the height was found using the pythagoras theorem with the formula tool. Then the points of the corners were extruded to the found height to create the 3D prism.

21

FRACTAL TECTRAHEDRA 3: Creating the facets

The facets were produced by converting the triangular prism into a berp, scaling it into thirds, and using those smaller triangles to trim the overall object.

4: Facet form on corner.

This basically duplicated that previous pathway. Essentially, the form was exploded, the corners of that form had triangles in each of its own corner. Then the trim tool was used to remove the corners of the corners to create the smaller facets on the large facet.

22

23

EVALUATING FIELDS 1: Curve generation

Curves made on rhino and defined in grasshopper

2: Points generation

7 points generated on each curve through the usage of the divide curve tool. 3: Convert to point charges

Point charge tool is used to create an attraction field around each points.

24

4: Circles around points

Circles were made around points. The field line tool is put in as well

5: Circle division

The circles are then divided into more points

6: Field generated

The attraction fields and circle divisions are connected to the field line tool. The attraction fields essentially guide the lines generated through the circle divisions without touching other lines cause by itself and other attraction fields

25

26

A small experiment where the circles around each points, were divided into 100 points

27

EVALUATING FIELDS

28

Continuing on from evaluating fields, the lines are dropped due to the behavior of the graph mapper. To a degree, the lines follow the nature of the graph inside the graph mapper.

29

GRAPH CONTROLLER 1: Circle divisions

This division of circles is create by first making a graph mapper and using the brezier graph. This graph is then used as the radius input for a circle generator. As the graph have different x and y values, this translates as different radiuses, thus generating different circles. The curves of these circles are then divide to generate points.

2: Voronoi

Voronoi is applied to the division of points on each circle.

30

3: Voronoi pattern

The circle curves are then divided into a larger odd number to generate the voronoi pattern

4: Finalized voronoi pattern

To generate a more interesting response, every secondary point in the circle curve divisions are culled

31

MODEL DATA TREE 1: List of points

A list of points were made through the construct tool where it is placed in a series

2: Y values

A series of Y values were added. This adjusted the hight of each points

3: Holistic

Holistic applied which cross referenced the x and y values, distributing points along its range.

32

4: Point list

An alternative to holistic is to graft either the x or y series. Point list was applies to the see the numerical response when grafting either x or y

5: Parameter view

Essentially shows how the values are organized.

6: Tree statistic Another approach to analyzing the data tree statistics

33

TREE STATISTICS AND VISUALIZATION

Essentially showing how to visualize the data points on a sphere by dividing its surface and tagging those points

34

35

DRIFT WOOD 1: Offset

2: Moving offset curves

3: Brep intersection

36

Using the pre-made model, the curve underneath was offset in a series, creating those numerous curves of increasing scales around it

The series of offset curves were then extruded in the z direction and adjusted to cover the model

Brep intersection was used between the model and the offset curves to generate this series of curves

4: Surface split

Surface split between the brep intersection and offset curves. This generates vertical surfaces that follow the contour of the brep intersections

The unnecessary surfaces generated from the surface split were manually deleted to create this driftwood model

37

DRIFT WOOD FRAMES 1: Curve division

The curves underneath the driftwood are divided into 30 points. 2: Division plane

Planes were made between the corresponding points on the two curves

3: Lines in plane

The lines within the planes are made through brep to brep intersections. The intersection is between the plane and the brep model

38

3: Points around the plane

The plane boundary is covered in points resulting from finding the end points from moving the brep intersection

4: Plane morph to desired shape

Poly lines are connected through the end points making the overall shape

39

30 ITERATIONS

CONICAL

OPEN CONICAL LOW HEIGHT MEDIUM DENSITY

OPEN CONICAL MEDIUM HEIGHT LOW - MEDIUM DENSITY

OPEN CONICAL LOW HEIGHT MEDIUM DENSITY

OPEN CONICAL HIGH HEIGHT MEDIUM DENSITY 3D POPULATE

CLOSED CONICAL MEDIUM HEIGHT LOW DENSITY HIGH OVERLAP

SPHERES

CLOSED SPHERES MEDIUM DENSITY

OPEN SPHERES LOW DENSITY

CLOSED LARGE SPHERES LOW DENSITY LARGE OVERLAP

CLOSED LARGE SPHERES LOW DENSITY LARGE OVERLAP

40

OPEN SPHERES HIGH DENSITY HIGH OVERLAP

CLOSED SPHERES HIGH DENSITY 3D POPULATE

CLOSED LARGE SPHERES LOW DENSITY QUARTER OVERLAP

OPEN CONICAL MEDIUM HEIGHT LOW DENSITY HIGH OVERLAP

OPEN PATTERNED CONICAL MEDIUM HEIGHT LOW DENSITY

CLOSED LARGE SPHERES OPEN SPHERES LOW DENSITY HIGH DENSITY SMALL OVERLAP

OPEN LARGE SPHERES VERY HIGH DENSITY HIGH OVERLAP

OPEN LARGE SPHERES LOW DENSITY

CLOSED LARGE SPHERES LOW DENSITY LARGE OVERLAP

41

CYLINDERS

LOW HEIGHT LOW DENSITY

LOW HEIGHT LOW - MIDDLE DENSITY

VARIOUS HEIGHTS LOW - MIDDLE DENSITY LARGE OVERLAPS

SPHERE AND CONE LOW HEIGHT LOW DENSITY

CYLINDER AND CONE HIGH HEIGHT MEDIUM DENSITY NO OVERLAP

LOW HEIGHT VERY HIGH DENSITY

HYBRIDS

SPHERE AND CYLINDER LOW HEIGHT LOW DENSITY

42

SPHERE AND CONE LOW HEIGHT LOW DENSITY NO OVERLAP

DOUBLE CONES LOW HEIGHT MEDIUM DENSITY

The 30 iterations were created by using the Voltradom definition. Initially the numbers parameters were changed, but as we grew more confident were started to change the actual definition itself. For instance, we changed the cone tool, to a sphere and cylinder. In addition we were able to combined different shape tools together, so the definition didnâ€™t only focused on one gemoetry

DOUBLE CONES LOW HEIGHT MEDIUM DENSITY

43

44

Form generation through trial and error Random forms were developed by creating various curves and lofting them in Rhino. The features we liked, we further worked on. The inital idea was to have a cave like form, thus forms that behaved like a tunnel were explored

45

SITE TASK MODEL 1: Boundary creation

For this task I first traced the boundary on Rhino, and defined that boundary as a curve on grasshopper

2: Curve creation

With the boundary, I then created curves on rhino, and defined them on grasshopper. With these curves, I wanted to then loft them and use that surface as the canvas for pattern creation.

46

3: Pattern generation Originally, I lofted the surface on grasshopper, and then applied populate 3d points. I wanted to used what I learnt in tesselation and create an interesting pattern on the surface. However, when I generated the formula on grasshopper, no pattern was formed despite having no error message on any of the grasshopper components. After many trials of diagnosing my problems, I decided to generate a pattern in a different way. Thus I decided to use my previous models grasshopper formula to see what would happen. This was highly experimental, and I didn’t know what the outcome who look like, so I used as simple cube to serve as the ‘berp’ for this formula. The pattern generated transcended any of my expectations and create this high complex and beautiful model. The massive array of extruded squares in various heights is where this model derives it’s complexity from. I did not expect a simple cube to create something in this level. To further push the bounds of my computer, I made this model translucent, in order to see the various arrangements, and rendered that on rhino. Due to complexity of the model, and being translucent, it took nearly 2 hours to render one image.

47

48

49

Reverse engineering Atmosphereic Tessellation by Chris Knapp and Jonathan Nelson is the piece my group is following for the reverse engineering assignment. From looking at the pattern, it is essentially triangles laid out in tessellation with 3d projections within them. Thus from looking at a singular pattern, we could tell the pattern was made from voronoi to get that split from the centre look. From that understanding, we knew that there were 3 points in the triangle to get the 3 3D projections. Therefore we started on first making a singular triangular pattern, then array those patterns on a surface.

http://www.detail-online.com/daily/atmospheric-tessellationwellington-new-zealand-12712/

https://www.flickr.com/photos/nicolesphotos/13554544603

50

1: 1st attempt at pattern

First created a triangle, scaled it, and lofted between each one to create the outer boundary. Then the inner triangle was surface divided to create the 3 inner points. The points were not in the best position, and we could move it, but we continued to experiment on it. With the 3 points we attempted to add to the voronoi tool, which rejected the points. Lots of hours were spent trying to concert these tree points to proper points to be attempted, but to no avail

2: pattern creation

Lots of trials were used to the 3 points from surface division, and use a triangle as a boundary for voronoi, which all failed. Turns out voronoi only accepts squares and rectangles. Finally figured out that we could use region intersection to basically crop the square boundary of the voronoi into a triangle. Division of surface was rejected to create points, so we manually created the points. These manual points was accepted by the vornoil tool. With this, we scaled the voronoi pattern to create the inner boundaries. Then everything was moved and lofted.

51

3: Attempt at surface We experimented with the triangulation tool. It put triangles on a random surface we made, but it also put triangles on the sides as â€˜negativeâ€™ walls. After not knowing how to get rid of these walls, we stopped testing it.

4: Bounding box surface attempt

Boundary box was used next as a surface potential. Despite making a interesting pattern, it did not achieve what we hoped. The gaps in between were too large, thus not tessellation, the triangles were not interlocking each other as desired, and it overly stretched the pattern at certain points.

52

5: Surface triangulation

We went back to surface triangulation, and generated that on a 2d rectangle. Through graft tree, we were able to single out each triangle, and then applied our pattern to it. Again we met the difficult to generating 3 points in each triangle to serve as the reference to the voronoi, which then makes the pattern. This was the only step left to finish, but we couldnâ€™t figure it out. Various commands were used, which all failed. Voronoi failed to recognize our points.

53

6: Final pattern generation of surface

The central points were finally created by first generating a centre point in each triangle (polygon centre), and moving these points in the 2 desired direction using the move and plane (direction) command. With these 3 points, the voronoi tool was able to connect, thus successfully creating the 3d pattern on the surface. However two issue did arise from this. Firstly, since the points were commanded to move at certain planes, it could only work on the x and y plane. Therefore it could only be used in a flat 2d horizontal plane. 3d surfaces, curves, and wavy surfaces cannot be used. The other problem is the patch tool. It seem to only generate the surfaces on one triangle

54

55

REVERSE ENGINEERING DEFINITIONS

PATTERNED SURFACES

56

S WITH ALL PARAMETERS ARE A FACTOR OF 1

The original reversed engineering definition pose too little flexibility in generating large amounts of patterns so two others were made with different surface divisions The 50 iterations in the next pages are the products of altering primarily the scale parameter. The more wild and messy patterns were made where changes were applied to v values of the surface division tool.

57

50 ITERATIONS TRIANGULAR SUBDIVIDE

TOP SCALE FACTOR: 1 BOTTOM SCALE FACTOR: 1 Z FACTOR : 1 PATCH DISABLED

TOP SCALE FACTOR: 1 BOTTOM SCALE FACTOR: 1.56 Z FACTOR : 1 PATCH DISABLED

TOP SCALE FACTOR: 1.560 BOTTOM SCALE FACTOR: 0.162 Z FACTOR : 1 PATCH DISABLED

TOP SCALE FACTOR: 1.560 BOTTOM SCALE FACTOR: 1.635 Z FACTOR : 1 PATCH DISABLED

TOP SCALE FACTOR: 0.533 BOTTOM SCALE FACTOR: 1.635 Z FACTOR : 1 PATCH DISABLED

TOP SCALE FACTOR: 0.241 BOTTOM SCALE FACTOR: 0.769 Z FACTOR : 4 PATCH ENABLED U DIVISION: 24 V DIVISION: 25 LOFT BY EXTRUDE

TOP SCALE FACTOR: 2.000 BOTTOM SCALE FACTOR: 0.908 Z FACTOR : 2 PATCH ENABLED U DIVISION: 1 V DIVISION: 5

TOP SCALE FACTOR: 1.0 BOTTOM SCALE FACTOR: 0.9 Z FACTOR : 2 PATCH ENABLED U DIVISION: 2 V DIVISION: 2

TOP SCALE FACTOR: 1 BOTTOM SCALE FACTOR: 1 Z FACTOR : 1 PATCH ENABLED

TOP SCALE FACTOR: 3.400 BOTTOM SCALE FACTOR: 0.264 Z FACTOR : 1 PATCH ENABLED

TOP SCALE FACTOR: 0.097 BOTTOM SCALE FACTOR: 0.854 Z FACTOR : 15 PATCH DISABLED

TOP SCALE FACTOR: 0.4147 BOTTOM SCALE FACTOR: 1.316 Z FACTOR : 6 PATCH ENABLED

TOP SCALE FACTOR: 1.48 BOTTOM SCALE FACTOR: 0.80 Z FACTOR : 3 PATCH DISABLED

TOP SCALE FACTOR: 1.0 BOTTOM SCALE FACTOR: 0.7 Z FACTOR : 1 PATCH ENABLED U DIVISION: 8 V DIVISION: 20 PARAMETER (T): 0.8

TOP SCALE FACTOR: 7.0 BOTTOM SCALE FACTOR: 1.3 Z FACTOR : 3 PATCH DISABLED U DIVISION: 10 V DIVISION: 15 PARAMETER (T): 0.9

TOP SCALE FACTOR: 1.3 BOTTOM SCALE FACTOR: 0.3 Z FACTOR : 3 PATCH ENABLED U DIVISION: 5 V DIVISION: 10

TOP SCALE FACTOR: 0.600 BOTTOM SCALE FACTOR: 0.869 Z FACTOR : 4 PATCH ENABLED U DIVISION: 13 V DIVISION: 15 PARAMETER (T): 0.1

TOP SCALE FACTOR: 0.1 BOTTOM SCALE FACTOR: 0.3 Z FACTOR : 2 PATCH DISABLED U DIVISION: 13 V DIVISION: 15 PARAMETER (T): 0.1 AND 0.3

HEXAGON

58

TOP SCALE FACTOR: 1.548 BOTTOM SCALE FACTOR: 0.769 Z FACTOR : 2 PATCH ENABLED

TOP SCALE FACTOR: 1.548 BOTTOM SCALE FACTOR: 0.769 Z FACTOR : 2 PATCH DISABLED

TOP SCALE FACTOR: 0.928 BOTTOM SCALE FACTOR: 0.769 Z FACTOR : 1 PATCH DISABLED

TOP SCALE FACTOR: 0.928 BOTTOM SCALE FACTOR: 0.769 Z FACTOR : 1 PATCH ENABLED

TOP SCALE FACTOR: 0.928 BOTTOM SCALE FACTOR: 0.769 Z FACTOR : 1 PATCH ENABLED

TOP SCALE FACTOR: 1.000 BOTTOM SCALE FACTOR: 0.908 Z FACTOR : 1 PATCH ENABLED

TOP SCALE FACTOR: 0.680 BOTTOM SCALE FACTOR: 0.807 Z FACTOR : 4 PATCH ENABLED

TOP SCALE FACTOR: 1.555 BOTTOM SCALE FACTOR: 0.807 Z FACTOR : 2 PATCH DISABLED

TOP SCALE FACTOR: 1.555 BOTTOM SCALE FACTOR: 0.807 Z FACTOR : 2 PATCH ENABLED

TOP SCALE FACTOR: 0.6 BOTTOM SCALE FACTOR: 0.9 Z FACTOR : 5 PATCH ENABLED U DIVISION: 8 V DIVISION: 10

TOP SCALE FACTOR: 0.6 BOTTOM SCALE FACTOR: 0.9 Z FACTOR : 5 PATCH DISABLED U DIVISION: 8 V DIVISION: 10 PARAMETER (T): 0.75

TOP SCALE FACTOR: -0.6 BOTTOM SCALE FACTOR: 0.9 Z FACTOR : 6 PATCH ENABLED U DIVISION: 6 V DIVISION: 8 PARAMETER (T): 0.75

59

CONSTANT QUAD SUBDIVIDE

TOP SCALE FACTOR: 1 BOTTOM SCALE FACTOR: 1 Z FACTOR : 1 PATCH ENABLED

TOP SCALE FACTOR: 0.853 BOTTOM SCALE FACTOR: 0.964 Z FACTOR : 6 PATCH DISABLED SUBDIVIDE: 2

TOP SCALE FACTOR: 2.300 BOTTOM SCALE FACTOR: 0.410 Z FACTOR : 1 PATCH DISABLED

TOP SCALE FACTOR: 0.795 BOTTOM SCALE FACTOR: 0.914 Z FACTOR : 4 PATCH ENABLED

TOP SCALE FACTOR: 0.795 BOTTOM SCALE FACTOR: 0.914 Z FACTOR : 2 PATCH ENABLED SUBDIVIDE: 2

TOP SCALE FACTOR: 0.427 BOTTOM SCALE FACTOR: 0.583 Z FACTOR : 2 PATCH ENABLED SUBDIVIDE: 2

TRIANGULAR PANELS TOP SCALE FACTOR: 0.488 BOTTOM SCALE FACTOR: 0.846 Z FACTOR : 2 PATCH DISABLED U DIVISION: 1 V DIVISION: 3

TRIANGULAR PANELS TOP SCALE FACTOR: 0.488 BOTTOM SCALE FACTOR: 0.846 Z FACTOR : 2 PATCH DISABLED U DIVISION: 1 V DIVISION: 3

TOP SCALE FACTOR: 0.189 BOTTOM SCALE FACTOR: 0.964 Z FACTOR : 6 PATCH DISABLED

TOP SCALE FACTOR: 1.0 BOTTOM SCALE FACTOR: 0.8 Z FACTOR : 2 PATCH ENABLED

TRIANGULAR PANELS RevSrf 3: REVERSE UV TOP SCALE FACTOR: 0.3 BOTTOM SCALE FACTOR: 0.9 Z FACTOR : 7 PATCH ENABLED U DIVISION: 2 V DIVISION: 1 SKEWED QUADS (T): 0

SUBDIVIDE QUAD TOP SCALE FACTOR: 0.3 BOTTOM SCALE FACTOR: 0.9 Z FACTOR : 7 PATCH ENABLED U DIVISION: 3 V DIVISION: 3 SKEWED QUADS (T): 0

CHANGING DEFINITIONS

TRIANGULAR PANELS TOP SCALE FACTOR: 1.0 BOTTOM SCALE FACTOR: 0.7 Z FACTOR : 2 X FACTOR: 6 PATCH ENABLED U DIVISION: 1 V DIVISION: 3 PARAMETER (T): 0.8

60

TOP SCALE FACTOR: -0.432 BOTTOM SCALE FACTOR: 2.000 Z FACTOR : 8 PATCH DISABLED

TOP SCALE FACTOR: 0.539 BOTTOM SCALE FACTOR: 0.899 Z FACTOR : 3 PATCH ENABLED

SUBDIVIDE QUAD RANDOM QUAD PANELS: 1 TOP SCALE FACTOR: 0.3 BOTTOM SCALE FACTOR: 0.9 Z FACTOR : 7 PATCH ENABLED U DIVISION: 2 V DIVISION: 1

SUBDIVIDE QUAD RevSrf 3: REVERSE UV TOP SCALE FACTOR: 0.8 BOTTOM SCALE FACTOR: 0.9 Z FACTOR : 7 PATCH ENABLED U DIVISION: 6 V DIVISION: 2 SKEWED QUADS (T): 0

TOP SCALE FACTOR: 1.225 BOTTOM SCALE FACTOR: 0.130 Z FACTOR : 4 PATCH ENABLED

TOP SCALE FACTOR: 0.325 BOTTOM SCALE FACTOR: 0.900 Z FACTOR : 6 PATCH ENABLED

TOP SCALE FACTOR: 0.450 BOTTOM SCALE FACTOR: 0.371 Z FACTOR : 5 PATCH ENABLED

61

Surface generation A existing definition called bubble form finding on the Karamba example page was used to create the surface of the pavilion. In essence, a boundary was defined with the area divided by a voronoi. Load applying forces were applied onto points of intersection, causing these points to move down, thus making the canopies.

62

63

Sun analysis with ladybug

Series 1

Series 2: Changed the boundary fro ommendations and to make the form

1

6

2

7

3

8

4

9

5

Solar analysis legend

Zero solar potential

64

Maximum solar potential

om straight to curved under tutorâ€™s recm more interesting

Series 3: Changed the orientation from east to west, to north to south, to respond better to the site and have greater access to sea water

10

11

12

13

14

15

65

66

Sun analysis Ladybug was used to find the amount of solar radition was exposured to the surface, and then the surface was improved accordling and retested again. Ladybug uses actual sun data from the site, thus the results would be quite accurate.

67

Shadow analysis with ladybug Ladybuy was again used to analyses the shadow of the patteren shapes and the final design form. From the shadow analysis we determined which is the best shape to use based on the least amount of shadow it cast on its self and its surroundings.

68

69

Structure analysis with Karamba

1cm in diameter

2cm in diameter

5cm in diameter

6cm in diameter

70

3cm in diameter

4cm in diameter

7cm in diameter

8cm in diameter

Karamba was used to study the stress applied to the structure. By apply the calculated load of the total pods onto the structure, the stress is express as colours along the structure. White is the best colour to achieve as it means no stress is being experienced along that member. As such, the diameter of the circular hollow section was slowly increased until the structure is shown to have no stress at all, with is at 8cm. Structure analysis stress legend

Excessive upwards stress

No stress

Excessive downwards stress

71

Structure creation

72

The structure was created using pipes rather than the structural elements given by Karamba. Reason being is that the pipes are the quickest and easiest to make whilst giving the same look as those created using karamba. Furthermore the Karamba generated structure had more complications to it thus slowing the computer when it was being used.

73

Pattern creation

74

The pattern was created using the same hexagon pattern generation formula made previously. It essentially uses hexagon divide to layout a hexagon grid on the surface, which is then scaled, moved and lofted to create the patterns.

75

Template for laser cutter

76

Template for 3D printer

The surface was first worked on to ensure there were no overlaps or gaps. Then it was extruded 3mm ups. The model was then converted to a mesh file and then further deeply analysised to ensure there were no collusion nor mistakes before the print.

77