Radiolarian Egg Pavilion

This is my contribution, done on March for an ongoing research project for a pavilion to hold a sculpture at the Barcelona campus of UIC, where the ESARQ is.

Like the other Radiolarian project, it’s conceived and directed by Alberto T. Estevez, head of the Genetic Architectures research line. It consists of an ovoid-shaped structure also designed as an interpretation of bone structures of Radiolarians. Its design develops a series of irregular hexagonal cells that are repeated in a tessellation of a surface and generate volumes that, following the logic of bone structures, are thickened at nodes forming spikes and leaving rounded holes.

The non-standard geometry outcomes from programming and modeling methods using RhinoScript and the Rhino3DE’s Math Plug-in.

Differently from the previous project, a more generative approach could be made this time. One in which almost all form generation was part of the development of the script from the beginning. Instead of modeling hexagons and circles and then input them to the script, only the base surface and a couple of numerical values were needed as input data and the rest was integrated into the script. I had the opportunity to implement an interesting algorithm to set the general configuration of the structure: the honeycomb algorithm, with which a difficult and inaccurate modeling technique was avoided.

The other good thing of developing scripts form the beginning, is that an efficient feedback process was had in terms of time spent. In a period of two weeks of work, a big amount of temporary output structures were generated, also producing lots of interesting blocks of code that can be useful for future projects.

On the second image, a series of these temporary structures are shown. Step 4 exhibits cells formed from longitudinal section curves; the sharp borders suggest ‘primitive’ spikes. On step 8, using my ShiftArray function, border tangencies are met, without spikes. Step 13 implements the current construction of spikes, the sine wave pattern for holes scaling and a random component. On step 15, randomizing was aborted. In turn, some execution efficiency issues were met.

I want to thank Universitat Internacional de Catalunya for supporting this research and acknowledge Alberto’s initiative. Thanks also to Pablo Baquero and Daniel Wunsch for their pertinent observations.