Design Computation : March 2016

ARCH 655 _ PROJECT 1

Nancy Al-Assaf

Dragonfly by Tom Wiscombe (SCI-Arc, Los Angeles, 2007)

This project studies the form and structure of the dragonfly wings and explores the methods to compute the complexity of this structure in Rhino and Grasshopper.

As a beginning, in nature, the dragonfly wing is unique in its formal variation. It’s an example to how form in nature responses to structural forces. Its morphology cannot be reduced to a single pattern, but it is multiple pattering systems interacting according to various force flow and material behavior.

Dragonfly Syntax ( source: http://www.australiandesignreview.com/features/659-an-imperative-of-survival/2)

In architecture, several projects by Tom Wiscombe utilize this concept. In this study, I am going to focus of Dragonfly- an installation at SCI-Arc, Los Angeles, 2007 by Tom Wiscombe. According to the architect:

Dragonfly Project by Tom Wiscombe ( http://projects.tomwiscombe.com/filter/installation/DRAGONFLY)

Project Description

“Dragonfly wings consist of both honeycomb patterns which are flexible and exhibit membrane behavior and ladder-type patterns which are stiff and exhibit beam-like behavior. These patterns are characterized by their rule-based interaction in terms of cell density, cell shape, and cell depth, as well as other parameters affecting overall wing performance, such as out-of-plane pleating behavior and material distribution. A composite of distributed and linear structural formations, the dragonfly wings are fields of continuous variation and adaptation evolving toward overall robustness.

Dragonfly Project by Tom Wiscombe ( http://projects.tomwiscombe.com/filter/installation/DRAGONFLY)

In this installation, dragonfly morphology and syntax are employed biomimetically rather than biomorphically, that is in terms of formal and behavioral logics rather than pure aesthetics or pure engineering. We know that that dragonfly wings in nature are generated by evolutionary processes involving aerodynamics, lightness, mechanical properties, composite performance, the smooth accumulation of organic material, and the active flow of dragonfly blood. Dragonfly is governed by a different set of parameters including gravity and seismic loads, specific support locations and quality of those supports, flat material increments, and buckling failure, differences which lead to an unpredictable hybrid morphology. Seen in a larger context, this project contributes to the recent contemporary discourse on cellularity in architecture as a departure from pure cellularity toward a tectonic based on emerging structural hierarchies within cellular aggregations.”

[ source: http://projects.tomwiscombe.com/DRAGONFLY]

Modeling in Rhino and Grasshopper

First method:

The first method is more a form making approach. It relies on an implicit understanding to the morphology of dragonfly structure. It aims to understand the relationship between points’ location/pattern and the voronoi structure of these points.

The logic:

Start with a rectangular pattern points in rhino
Then divide them into two groups (the center / the boundaries) in grasshopper
Create a Voronoi structure in grasshopper
Go back to Rhino and change the location of the points at the center ( x, y )
The Voronoi structure updates automatically in rhino

Dragonfly structure in Rhino / Grasshopper : pull points manually ( source: author)

Then :

Apply forces (Kangaroo) on the strings of the Voronoi structure
To get more accurate result : divide the strings into two different groups : the rectangular pattern with higher stiffness and the honeycomb patterns with lower stiffness and higher flexibility

Dragonfly structure in Rhino / Grasshopper : two patterns with different stiffness ( source: author)

Second method:

The second method relies heavily on form-finding approach and on analysis as a generative feedback to the algorithm.

The logic:

· Start with a generic rectangular Mesh in Rhino or Grasshopper

· Deform that mesh in Grasshopper using Kangaroo according to the required design intent (the shape depends highly on the UForce value and direction, the location of AnchorPoints)

· After getting the required shape use that mesh as an input to the next stage: the Curvature analysis, but first :

o Convert the mesh into surface

o Deconstruct the surface to get the: UV coordinates of the points, the Surface and the Vertices on that surface.

o Then apply the curvature analysis:

§ Create a numeric domain for the analyzed data and colorize the mesh according to that domain

§ Divide the points on the surface according to the curvature value into two groups ( take the average AND Dispatch the list into two targets lists around that average.

§ For the points with low curvature value we need to keep them in the same original pattern BUT for the point with higher curvature value we need to change that pattern into hexagonal cells OR a random arrangement of points will give similar result ( this conclusion came from the first method and the study of points’ arrangement) .