Karamba Model

Currently I would like to work with Karamba for my graduation project, so therefore I volunteered to do the structural testing of the current form so I could experiment and start to understand Karamba.

Due to the material behavior script (latex behavior) not being completed, I experimented with the form in Karamba if it was only beams. If the latex script cannot be completed, the beam only structure is something we will need to continue with.

Original Form

The first division method I tested with Karamba was scripted by Nariddh. He created a script that divided the surface based on the ‘start’ and ‘end’ sliders within the script. These values would determine the number of panels up the length of one side.

After the surface had been divided, I used the ‘Brep Edges’ and ‘Remove Duplicate lines’ components to get each divided line. This is then feed straight into the ‘line’ input of the ‘line to beam’ component.

From this step, I needed to construct the model using the ‘assemble’ component.


ELEMENT: Line to beam

SUPPORT *: Support> Points on Ground, Support> Intersecting Points

LOAD: Load > Gravity, Load> Material Weight

CROSS SECTION: Cross Sections> width&height = 5cmx5cm

MATERIAL: Material Section>Wood

* The supports are the ground points and off ground points. You can’t put these points into the same support Component as they have different flexibility. For example, the points on the ground must have zero flexibility, hence all conditions are locked.

You need to divide the list into ground lines and non-ground lines, you can do this using the list item tool. Then you need to find the end points of each line with the ‘end points’ component. Due to this script finding the points to every end, you need to remove the duplicate points before inputting this into supports. *

From here, I analysed the model and then made it so it could be view in rhino by previewing only the ‘Beam View’ component.

I then used a script written by Dominic to find the overall displacement.

To run Galapagos, I chose the fitness to be the number value being produced out of the last component (AxB — R). The Genetic input was the ‘start’ and ‘end’ sliders that were created by Nariddh as explained earlier. I then set the optimization to find the lowest value possible over a two hour run.

This result achieved a displacement of 0.07. I had aimed to the displacement lower than this, around 0.01.

After consulting the group, we decided that this design had far too many panels. Not only would it increase cost, it would not give the visual affect we were aiming for. Sol then suggested testing triangular panels, which may improve the amount of panels and displacement. Another issue we discussed is the amount of over hangs in this pavilion. I believed this was the reason for the amount of panels and displacement percentage. Nissie then volunteered to remove as much over hang without damaging the aesthetics of the design.

Updated Design
Blue lines are the old design, Red are the updated design

As shown, Nissie took a fair amount of over hang off the top. Nariddh then brought to my attention that his script only works with quad panels. So I ran the new design through a simple ‘lunch box’ division. I was then told that the two beam options were 42x42 cm, or 70x70cm. I ran the optimization twice in aim of finding the best beam size for the design.

Left: 42x42cm — Displacement: 0.12 Right: 70x70cm — Displacement : 0.03

From this optimization, it is obvious that the most structurally sound is the 70x70cm beams, however, the panel size is so small and everyone was not happy with this.

I then decided to run a quad panel with the new design, however I set a limit on how small the panels could be, to keep the aesthetics of the design.

Left: 42x42cm — Displacement: 0.37 Right: 70x70cm — Displacement : 0.09

As a result, the class decided to update the whole pavilion so it is more stable with not so many panels. The plan is that Dominic, Simon and Nissie try to get the latex behavior working in kangaroo, and also update the design of the pavilion. From here, they will pass it on to me so I can test it in Karamba.

* In the presentation, Alessandra noted a pattern in the structural model.

There are faint red diagonal lines throughout the pavilion (red=compression, blue=tension. )This pattern is something I should look into this week to try work out what it is and if this is providing details into what needs to be updated in the pavilion.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.