Performative Parametrics

51.
Geometry subtracted with mass 1
Geometry subtracted with mass 2
Geometry subtracted with mass 3
Geometry subtracted with mass 4
2000 Points
3000 Points
4000 Points
5000 Points


03APERIODIC VORONOI
BRICK SYSTEM (deduction + gradient)
52.
DEVELOPED
Defining Interior & Exterior + Voronoi Gradient
This script heavily focused on differentiating the brick system finishes - internal brick face containing a smooth finish, whereas external contain- ing a textured/geometric finish for crystal to latch on to (later on in the process). Brick gradient was also tested, whereby bricks would go from small to large, or from large to small, in a specified direction; x,y,z.
1.
First part sets out the boundaries in which 3D Voronoi is generated.
Custom Mass Dimensions
The internal brep mass geometry is offset and used to create an area in which voronoi cells would be selected and kept. The slider controls the distance of the brep offset, which also represents the thickness of the selection area (wall).
Creating Mass
Populating mass with points
Culling points to make point gradient
Offset Brep to create voronoi selection area
Using new offset brep solid difference it between box
Select points that fall inside the new brep box and cull them
Using the remaining centre- points select the voronoi cells
2.
A set brep geometry, which represents an ‘inter- nal’ space mass, is used to boolean difference the generated voronoi - this creates a smooth internal voronoi face.
5.
The points which populate 3D region, are culled out random- ly to create a gradient of points in the Z direction. Using the Graph Mapper the gradient can be mannually controlled.
A predefined brep containing a mass that intersects the selected voronoi cells is used to cull out the intersecting centre points, thus creating an opening.
*Script Resource: https://www.youtube.com/watch?v=eiDJsdN- vsNQ
Part 2 of this script was heavily influenced by the video linked above. It demonstrated and enabled me to construct a gradient of random points.
Custom Brep
Solid difference between brep & box
Creating 3D Voronoi
Set Custom Brep (to be used for openings)
Cull the centre-points that fall inside the brep
Select remaining voronoi cells
3.
4.
Geometry 1
Geometry 2
Geometry 3


53.
Geometry 1 Covered in Voronoi Cell Bricks
Plan
Elevation
Geometry 2 Covered in Voronoi Cell Bricks
Geometry 3 Covered in Voronoi Cell Bricks
Plan
Plan
Elevation
Elevation
Voronoi Brick Gradient
Voronoi Brick Gradient
Voronoi Brick Gradient


04SIMULATING SALT
CRYSTAL GROWTH (random)
Attempting to simulate salt crystal growth 3.
54.
After attempting some real-life experiments in growing salt crystals and seeing and understanding their growth patterns and geometry, I have set out to recreate it in the virtual world. The random set of points represent randomly latched salt particles, which manifest themselves into grown crystals of various sizes and geometries that follow the same growth lan- guage and habits of those in the real-world.
The generated boxes are then deconstructed. The in- dividual faces are used to create a new set out surface boundaries along which another set of random points can be created and used to generate smaller boxes.
20 Points
40 Points 60 Points 80 Points
1.
Generating a set of random points on a specified surface. Using those points, a series of random sized boxes are con- structed on the points.
Setting Surface
Generating Points
Generating Boxes (crystals)
Deconstruct brep & select faces
Extruding Surface
Boolean Difference
2.
Populate new faces with points
Generate additional set of boxes
Creating a surface extrusion to use as a ‘boolean difference’ for the generated crystals.
Deconstruct brep & select faces
Populate new faces with points
Generate additional set of boxes
4.
The same process as 3 is repeated to add an extra level of complexity and detail. Essentially this part of this script can be repeated to get the desired level of complexty/detail of the generated crystals.


20 ‘Crystals’
40 ‘Crystals’
55.
60 ‘Crystals’
80 ‘Crystals’
Crystal Complexity 0
Crystal Complexity 1
Crystal Complexity 2
Crystal Complexity 3


05SIMULATING SALT
CRYSTAL GROWTH (regional)
Attempting to simulate regional salt crystal growth
After my initial attempt to simulate crystal growth of random salt particles that may randomly latch onto a surface and begin to grow, I wanted
to experiment by using a single ‘catalyst’ point which manifests a single regional growth on a surface.
1.
A surface for the ‘crystals’ to grow on is predefined. In this script a starting point of growth is also predefined through a point placement. The point is then used as a centre point for a circular boundary, in which the next set of points is generated in, and so on indefinitely.
56.
Repeat 2
Repeat 4
Repeat 6
Repeat 8
Setting surface & starting point
Generating boundary shape
creating bounding surface
Populating random points
Generating boundary shape for new points
Creating bounding surface for new points
Extruding Surface
Generating Boxes (crystals) from points
Boolean Difference using extruded shape
Join final shape (crystals)
3.
4.
A loop which records data is used to record the ‘crystal’ growth. The loop is also used as a catalyst of growth, as the loop is repeated, more and more set of points get generated which in return create more and more ‘crystal’ shapes along the predefined surface area.
2.
In this part, the populated points create their own circular boundary, however as the points get pop- ulated and the boundaries grow outwards, they are trimmed using the main surface, thus meaning that no points fall off the predefined surface area and growth is constrained to the surface.
Just like in the previous script, a set of randomly sized boxes (crystals) are created on a set of populated points and a trimmed with the predefined surface.


2 Loop Repeats
2 Loop Repeats
57.
4 Loop Repeats
4 Loop Repeats
6 Loop Repeats
6 Loop Repeats
8 Loop Repeats
8 Loop Repeats


05SIMULATING SALT
CRYSTAL GROWTH (regional spread)
58.
DEVELOPED
Refining regional salt crystal growth simulation
After having a first attempt at simulating salt crystal growth in the previous script I found my script structure to be inefficient as the growth areas over- lapped and thus resulted in a lot of ‘duplicates’. In my second attempt (below) I refined the script to perform without overlapping crystal growth areas. This resulted in a script which is able to cover a larger surface area without increasing the file size. Additionally it simulates a slightly more interesting growth path.
Set No. of Paths
Setting surface & starting point
Generating ‘Growth’ starting points
Set No. of Loops (Path/Growth Lenght)
1.
A surface for the ‘crystals’ to grow on is prede- fined. In this script a starting point of growth is also predefined through a point placement. The point is then used to develop a regional growth path.
Like in previous scripts; detailed salt crystals are generated along path.
2.
Growth path exploded
Drawing a line back to centre-points allows to set initial direction of ’growth’
Vertices are used to generate boxes
‘Growth’ direction is randoly selected on each loop, thus creating a unique ‘growth’ path along which salt crystals form and populate the surface. Number of loops and groth paths can be controlled via the sliders. More loops = longer path, More paths = density of crystals
Creating direction controlling shape
A random point on arc is selected
The process is looped and recorded
3.
The ‘growth’ path line is extracted from the loop, joined and exploded. The vertices are used to construct a first set of boxes that will later represent the salt crystals.
1 Path
1 Loop
5 Paths
5 Loops
10 Paths
10 Loops
5.
15 Paths
15 Loops
20 Paths
20 Loops
Initial boxes are exploded
Randomly Populate Faces w/ point
Join final shape (crystals)
4.
The initial surface on which the salt crystals are growing on, is extruded and used to boolean difference the crystals.
Extruding Surface
Solid Difference using extruded shape


59.
1 Loop Repeat
(10 growth paths)
5 Loop Repeats
(10 growth paths)
10 Loop Repeats
(10 growth paths)
15 Loop Repeats
(10 growth paths)
20 Loop Repeats
(10 growth paths)
1 Loop Repeat
(10 growth paths)
5 Loop Repeats
(10 growth paths)
10 Loop Repeats
(10 growth paths)
15 Loop Repeats
(10 growth paths)
20 Loop Repeats
(10 growth paths)
1 Loop Repeat
(5 growth paths)
5 Loop Repeats
(5 growth paths)
10 Loop Repeats
(5 growth paths)
15 Loop Repeats
(5 growth paths)
20 Loop Repeats
(5 growth paths)


05SIMULATING SALT
CRYSTAL GROWTH (curved surface)
60.
DEVELOPED
Growing crystals on curved surfaces
As all previous simulations ran on planar surfaces, however the designed brick system had curved and planar surfaces that went in all directions. This improvement of the previous script focused on adapting the 2D growth path, and mapping it onto a geometric surface. The result, as displayed below, allowed the ‘crystals’ to grow on a geometric surface that is not planar.
Set No. of Paths
Setting surface & starting point
Pull point to surface
Drawing a line back to centre-points allows to set initial direction of ’growth’
Pull curve to surface
Growth path exploded
Vertices are used to generate boxes
Set No. of Loops (Path/Growth Lenght)
1.
A surface for the ‘crystals’ to grow on is predefined. In this script a starting point of growth is also predefined through a point placement, which in this script is pulled back to the pre- defined surface. The point is then used to develop a regional growth path.
5.
2.
‘Growth’ direction is randoly selected on each loop, thus creating a unique ‘growth’ path along which salt crystals form and populate the surface. Number of loops and groth paths can be controlled via the sliders. More loops = longer path, More paths = density of crystals
Creating direction controlling shape
A random point on arc is selected
The process is looped and recorded
Offsetting Solid Surfce
3.
The ‘growth’ path line is extracted from the loop, joined and pulled bak to the surface before the path is then exploded. The vertices are used to construct a first set of boxes that will later represent the salt crystals.
Surface 1
Surface 2
Surface 3
Surface 4
Like in previous scripts; detailed salt crystals are generated along path.
Initial boxes are exploded
4.
Randomly Populate Faces w/ point
Join final shape (crystals)
The initial surface on which the salt crys- tals are growing on, is offset (solid) and used to boolean difference the crystals.
Solid Difference using extruded shape


61.
1 Loop Repeat
(5 growth paths)
5 Loop Repeats
(5 growth paths)
10 Loop Repeats
(5 growth paths)
20 Loop Repeats
(5 growth paths)
Elevation
Custom Surface 1
(10 growth paths)
Elevation
Custom Surface 2
(10 growth paths)
Elevation
Custom Surface 3
(10 growth paths)
Elevation
Custom Surface 4
(10 growth paths)
Elevation
Elevation
Plan
Elevation


06APERIODIC VORONOI BRICKS & CRYSTAL GROWTH
Merging the two studies
In the last study the two explorations; 1. Aperiodic Voronoi Brick system, and 2. Salt Crystal Growth on surface are tested to see how the slat crystals would be able to grow on the voronoi brick system. For the test a simple arched massing was used, and the random crystal growth was applied. In addition the crystal formation has also been improved and finalised in this script.
1.
A square box of 3D voronoi cells is generated from which the ‘interior’ space mass is subtracted. This cre- ates the opposing finish between interior and exterior.
3.
The 3D Voronoi Brick Wall is exploded and the faces are measured against one another. If there are two overlapping faces then it means that it is a ‘clothed’ face. If it is a single value, then it means that the face is ‘naked’ (outer facing). Using this method, the outer face of the wall is selected and joined as a single surface for the salt crystals to grow on.
62.
Setting custom mass
Creating 3D Voronoi
Solid Difference
External Geometry
2.
The set mass geometry is offset by the desired ‘wall thickness’ which generates an area in which ‘clothed’ voronoi cells can be selected. This creates the rough external finish to which the salt crystals can latch onto and begin growth.
Offsetting geometry & Solid Difference
External Wall Geometry
Initial Salt Crystal
Detailed Salt Crystal
4.
Finding ‘Naked’ Surfaces
‘Naked’ Surface Geometry
Lastly, the salt crystal growth is applied to the outer face. In this example a ‘random growth’ version was used, however it was improved by further developing the crystal formation. With the introduction of 3D Rotation component, each crystal particle was able to obtain a unique orientation, which can be seen in the real-life salt crystals.


50 Crystals
Simplified
100 Crystals
Simplified
50 Crystals
Detailed
100 Crystals
Detailed
63.
Improved Crystal Formation - Detail
Geometry
Improved Crystal Formation - Detail
Representation
Improved Crystal Formation - Detail x2
Representation
Improved Crystal Formation - Detail x4
Representation




PART B - stage ii
Applying the Developed System
Following the development of the salt brick system and the crystal growth simulation, a test was required to see how it could be used to form an architectural piece. The last chapter of the report looks at the develop- ment of a salt brick/crystal pavilion, situated on the salt flats of Salar De Uyuni, Bolivia.


01PART B ii PROJECT BRIEF + ASPIRATIONS
Design Strategy
In an attempt to design a medium sized pavilion and to test the devel- oped design methods of mass deduction, firstly a general massing of the pavilion will need to be generated. In order to do that, the technique of selecting ‘clothed’ 3D voronoi cells, that is present throughout the brick system studies, will be applied. However the scale of the 3D voronoi cells will be at a larger architectural scale, so that the generated geometry is able to accommodate a fully functioning space where multiple people can enjoy.
A simple design layout should also be integrated in the massing, thus meaning that the desired pathway of the pavilion should be used to generate the voronoi pavilion massing. Opening masses should also be considered at this stage as they would be used to generate the pavilions’ access openings and openings that would provide light into the space inside.
Design Methods
The chosen design methods that I wanted to test in an architectural appli- cation were; 1. Textured Salt Bricks on both interior/exterior faces, which can be achieved by either mass deduction/skeletal construction methods studied in pages 46 & 48 of this report.
2. Uniquely formed openings for access & light, which can be achieved by setting up a brep that intersects the voronoi salt brick system and either solid differences or culls the centre points of voronoi bricks that fall within the brep region - methods studied in page 50 of this report.
3. Random Salt Crystal growth pattern on the exterior face, which can be simulated by extracting the ‘outer’ face of the brick system and generat- ing a set of random points, which would then form the salt crystals. The intention would be to apply a ‘medium’ amount of crystals without entirely covering all the external faces, as I aim to perform a visual analysis of the pavilion and see how well the salt brick system responds to the context.
Desired Outcome
As the final outcome for this part I would like to produce a pavilion locat- ed in Salar de Uyuni, Bolivia salt flats that is entirely made out of the salt brick and salt crystal system developed in the previous section. The aim of the pavilion will be to test the salt brick system concept, it’s appearance (how well the structure fits the context), and the methods of construction that were developed as part of the studies.
The aim is to produce a medium sized conceptual pavilion, which would allow for multiple people to enter and experience the unique space inside. The pavilion is to have openings to allow light and access, and potentially some seating inside.
The pavilion is to be a single storey structure, although the structure itself could potentially be larger to create double height spaces inside. The massing and general scale of the structure should refer closely to the precedents of ‘Grotto’ by Aranda\Lasch and the ‘ICD-ITKE’ Research Pavilion.
66.
1. Brick texture on both interior/exterior faces
2. Uniquely formed openings for access & light.
3. Random Salt Crystal growth pattern on the exterior face


Grotto by Aranda\Lasch
Bloomberg Pavilion by Akihisa Hirata
Tafoni at Pebble Beach
Tafoni at Pebble Beach
67.
Primitives by Aranda\Lasch
ICD-ITKE Research Pavilion
Voussoir Cloud / IwamotoScott Architecture + Buro Happold


02VORONOI MASSING & DIVIDING SEGMENTS
Generating Pavilion’s Massing
The first step in creating the pavilion is to generate the massing of the structure/internal spaces. For this, the same method of 3D voronoi cell selection will be used, however a predefined generic mass layout should be set first. This generic layout can be as simple as a collection of room/ space resembling boxes/geometries.
1.
A large rectangle box is created with a series of smaller 3mx3mx3m sub-dividing boxes. The large box is to be filled with large 3D voronoi cells.
68.
Controls for sub-dividing grid
Controlling Number of Floor Levels
Controlling Mesh Offset Distance
3.
2.
Sub-dividing boxes can be selected to begin massing the generic internal layout.
4.
Once a series of sub-dividing boxes are selected, and the generic internal layout mass is baked fully, it can then be set as the ‘brep’ input which will perform an analysis on the entire list of large 3D voronoi cells. The cells that fall within the brep region are selected and thus a 3D Voronoi mass of the pavilion is created.
Setting custom mass
Sub-dividing mass
Selecting Sub-division
Building up generic mass
Scale of 3D voronoi cell is defined through mathematical equation which takes the grid cell size and divides it by a factor of 10. The result is then multiplied by the desired amount of floor levels. This ensures that the size of voronoi cells stay consistent even if there is a change in the desired level amount. Once scaling is defined, the script culls the ‘naked’ 3D voronoi cells - as previously explained in page 44.
Populating mass with points Generating 3D Voronoi Cells
Exploding 3D Voronoi Cells
Culling Voronoi Cells
Setting baked generic massing as brep
Voronoi cells inside the generic mass brep
Voronoi Massing as brep


General Mass
Sub-diving Mass
Selecting Generic Layout
69.
Access
Internal Circulation
Access
Converted into Voronoi Massing
Creating Dividing Segments
Access


03VORONOI BRICK APPLICA- TION & CRYSTAL GROWTH
Applying Brick System to Generated Mass
The second part of the script focused primarily on applying the developed voronoi brick system onto the generated massing from the previous script. In order to simplify and make it more manageable for the computer, the pavilion was divided into 5 segments, each one containing 10,000 3D voronoi cells. This allowed the computer to generate the voronoi brick system without crashing.
3.
The 3D Voronoi Brick Wall is exploded and the faces are measured against one another. If there are two overlapping faces then it means that it is a ‘clothed’ face. If it is a single value, then it means that the face is ‘naked’ (outer facing). Using this method, the outer face of the wall is selected and joined as a single surface for the salt crystals to grow on.
70.
Setting part of 3D Voronoi Internal Massing
2.
Lastly, the salt crystal growth is applied to the outer face. In this example a ‘random growth’ version was used, however it was improved by further developing the crystal formation. With the introduction of 3D Rotation component, each crystal particle was able to obtain a unique orientation, which can be seen in the real-life salt crystals.
Scaling Original Brep
Selecting voronoi centre points that fall in the scaled brep region
Selecting voronoi centre points that fall in the original brep region
Selecting voronoi centre points that fall in the offset region but not the original
Selected voronoi centre points returned geometry
1.
The part of internal mass geometry is used to subtract from the ‘pool’ of 3D voronoi cells that fill one of the 5 set segments.
Finding ‘Naked’ Surfaces
‘Naked’ Surface Geometry
Setting 1 of 5 segments
Populating segment with points
Generating 3D Voronoi Cells within the segment
Solid difference between voronoi segment & Internal Massing brep
The set brep of internal mass is scaled up. The points within the scaled brep are selected and culled with the points in the original sized brep, thus leaving the centre points of voronois cell that fall in this created area between scaled and unscaled breps (creating wall thickness - larger scale equals to thicker walls, offset surface can also be used).
Offset | Original
Offset | Original
Offset | Original
4.
Initial Salt Crystals
Detailed Salt Crystals


Populating Segment 1.
Populating Segment 2.
Populating Segment 3.
71.
Populating Segment 4.
Populating Segment 5.
Final Pavilion + Crystal Simulation


04SALT BRICK/CRYSTAL PAVILION PARAMETRIC PROPOSAL
Final Study Conclusion
The final pavilion proposal showcases the potential use of the voronoi brick system to construct an architectural form. After experimenting with different process of construction (skeletal selection, mass deduction, gra- dient) the one I chose to explore in this pavilion proposal was the mass deduction. This means that a 3D geometry of the pavilion’s overall mass had to be designed before the application of the bricks could have been applied. Once generated, the geometry was then ‘sunken/dipped’ into
a ‘pool’ of 3D voronoi cells and used to cull/solid difference the voronoi cells, thus removing the voronoi cells from the interior spaces of the mass.
The second part to this method was creating a ‘wall thickness’ to enclose the interior spaces. This was achieved by scaling/offsetting the internal mass by a desired wall thickness number and using the region in-between the offset and the original internal mass as an area to select the 3D vor- onoi cells. Openings in the pavilion have been created in a similar logic. A mass which represents an opening has been created, and placed to intersect the 3D voronoi brick walls. The voronoi cell centre points that fall within the opening mass region are culled, thus creating an opening.
Due to the scale of the model, several 3D Voronoi cell ‘pools’ had to be created, as one large pool would have been too complex for the comput- er to handle. It is therefore that this pavilion was divided into 5 sections (as demonstrated previously) in order to compute smaller parts at a time and make it more manageable for the computer. Using this method,
an even larger pavilion/structure can be created, although it will mean repeating more dividing sections, and the increase in file size as more complex geometry is generated.
The crystals for this pavilion were used using the ‘random growth’ script which populated the outer surfaces of the bricks, but not the interior spac- es. Although there is theoretically nothing from stopping the crystals grow inside the pavilion, the developed brick system specifically requires a layer of crystals to grow on the outer surface to provide extra protection from the rain. This is intended so that the salt crystals are dissolved first before the rain is able to reach the salt bricks themselves and start to dissolve them.
For further development I would have liked to look at further development of the voronoi brick design - potentially reducing the salt volume used within the brick itself, whist trying to maintain the structural strength and reducing the overall weight of the building. Additionally, I would have also liked to investigate and designed a potential sub-structure (from stainless steel), which would have provided additional support to the bricks, and allowed an increase in the scale of the overall building.
Overall, I believe that the final proposal demonstrates well the possibility of building structures on the Bolivia’s Salar De Uyuni (salt flats) without the structure looking too out of context. The construction method of this potential building type would be the most controversial part, as instead of the conventional method where a bricklayer is used to lay modular bricks, this type of structure would require a large scale industrial 3D printer to 3D print unique salt bricks already on site and in their custom positions.
Due to the uniqueness of each brick cell, the final pavilion composition looks less man made, and more organic/natural, resembling a carved out cave or a rocky mountain. The openings of the pavilion also contrib- ute to the look of a natural cave, as they appear to be carved/corroded away from the overall mass.
I believe I was able to accomplish what I initially set out to achieve, and in addition create an array of scripts that focus on different techniques/ways of constructing, whilst using this developed salt brick system. In my studio project, I aim to continue and advance this salt brick/salt crystal growth system to suit large scale industrial factory buildings, as it is my studios preliminary focus.
72.
Pavilion Massing - Final Outcome


73.


74.

75.


76.

77.


END OF REPORT
Martynas Kasiulevicius
January 2018, University of Westminster