The goal was to have a parametric materialization that could change density in response to different design factors. We were inspired to create a surface made from weaving curvilinear agents that could respond to global design demands. We looked to the 2014-2015 ITECH pavilion as an inspiration for what was possible. However, instead of focusing on structural performance, we decided our material density should be an environmental control, admitting variable light for different programmatic requirements. Our weaving boid agents would form a curvilinear pattern for a large building facade, changing their density to allow more or less light in certain areas, depending on the programmatic requirements of the building. The purpose is to have the building program influence the morphology of the facade. Different programs have different lighting requirements. Areas with computers might require less light, whereas communal spaces, reading areas, or break areas need more light. The facade reacts to this building program by becoming less dense in this area, thus admitting more light. To achieve lesser density, the facade moves around the area in a curvilinear fashion. Thus, the fenestration of the facade is directly responding to the environmental requirements of the occupants inside.
The goal was to have a parametric materialization that could change density in response to different design factors. We were inspired to create a surface made from weaving curvilinear agents that could respond to global design demands. We looked to the 2014-2015 ITECH pavilion as an inspiration for what was possible. However, instead of focusing on structural performance, we decided our material density should be an environmental control, admitting variable light for different programmatic requirements. Our weaving boid agents would form a curvilinear pattern for a large building facade, changing their density to allow more or less light in certain areas, depending on the programmatic requirements of the building.
To realize this concept, we needed to create a boid swarm behavior where the swarm could become less dense at certain areas of interest. We created a couple of new Boid behaviors: attractor curve, repeller points, and edge containment. Combined, these simple behaviors form a system fulfilling our performance criteria. The first behavior we made is an attractor curve. In essence, this is the opposite of a curve containment. The agents move towards the nearest point on the attractor curve. They are always attracted to the curve, regardless of their proximity to the curve. This creates a type of global migration behavior from one side of the facade to the other. Secondly, we created repeller points. These are scripted in a very similar manner to attractor points, but with a negative value. The agents move away from the nearest repeller point, always influenced by one of the repellers. After the Boid System runs, we had to do some Post Processing. We made a component that will track the history of the agent movements and generate a polyline. Each agent creates its own discrete polyline. Lastly, we used a simple sweep operation to give the lines some dimension.
When all these behaviors are synthesized into one, the combined outcome is a curvilinear facade reminiscent of organic materials. There is a global migration from one side of the mesh to the other, with the agents curving around certain areas of interest. It takes on a natural look, while still maintaining a very planned and orderly approach to where the fenestrations should occur. It’s organic, yet engineered.
We made several iterations on our design. Our goal was to explore the potentials of openness and curve continuity. First, how much is it appropriate to completely open up the facade. Second, we wanted to eliminate curves overlapping each-other and have them continuous the whole height of the facade. Through this rigour, we discovered that less repeller points generates smooth continuous sweeping behavior, but more repeller points generates a more organic, novel, and higher curvature behavior. We settled on a value between the two extremes; this gave us something aesthetically novel, yet clean and without self-conflicts.
These simple agent behaviors were able to give us the aesthetic results we were looking for. With just a couple simple behaviors, we had control over the agent movement, and we were able to create a myriad of design options. If we are to continue researching this topic, the next step would be to have this become more of a structural design concept. The agents should be informed by outputs from Karamba.