An architectural form-finding experiment
Award: 2021 ArchiModel Winner
Course: Arch 2854
Timeline: Aug 2019 - Dec 2019
Partner: Alex Poff
Role: Architect
Architectural Form & Diagrammatic Analysis
Lars Spuybroek's Figuring-Configuring Studio introduced a design methodology that derives architectural form through means of pattern finding and making. In the first phase of the studio, we explored form through the branching configurations created by thread models dipped in water, an experiment inspired by Frei Otto's soap bubbles. The interactions that emerged from the experiments configured together according to specific rules that we uncovered through diagrammatic analysis.
Catalog System
Figures & Configurations
To understand the complex branching structures, we created a named catalog for the various figures and configurations found in the experimental thread models. Figures are the smallest possible interactions between threads while configurations are conglomerates of figures.
Phase One
Branching Thread Experiments
We created a series of four experimental thread models to analyze the branching structures created by wet threads clinging together. With each iteration, we altered variables including the grid type, density, height, and existence of split levels in the models to increase the variation of configurations. The thread type, slack, and dipping technique were kept consistent throughout all the experiments.
Thread Model A
Simple & Basic
To kick off what would be a long series of experimental models, we wanted to start with a simple base model and use that as a starting point to gather data. Our first thread model uses a simple 1" square grid with 10" long threads. We then analyzed how the the threads grouped together and dissected specific thread structures.
Thread Model A'
Square to Isosceles Triangle Grid
The 1" square grid yielded too little interactions, so we added a string at the center of each square in Model A' to create an isosceles triangle grid. The increase in density increased the number of interactions between strings.
Thread Model B
Density Grid & Addition of Split Levels
As a test to determine the optimal grid density, we used a density gradient across the grid of Model B. From there, we would be able to discern the best density for creating the most variation in interactions. Also, adding another level allows us to analyze interactions between strings of different lengths that may create more unique figures and configurations. For each experiment, we varied the height of the split level to further determine the ideal split level height.
Thread Model C
Equilateral Triangle Grid & Height Increase
Since the strings are equally spaced apart in an equilateral triangle grid, each string has an equal chance of interacting with each of its partners. This creates more variation in groupings, and it allows the strings to self-organize without the influence of uneven spacing. In order to observe more complex interactions between the split levels, we increased the overall height of the machine from 10" to 16".
Cambridge, MA
Only after these form finding experiments were we given the site for the building in Cambridge, MA as part of the Harvard campus on the intersection of Brattle St and Appian Way. We identified pedestrian paths and access to open areas as drivers to the location of entrances and orientation. While the studio did not emphasize site considerations, we were cognizant of the site conditions as we moved forward into the analog models.
Phase Two
Branching Analog Models
Using our matrices of figures and configurations along with the branch analyses, we abstracted the branching patterns to create a series of analog models that explored spatial conditions. The goal was to modify the system to achieve a greater variety of spatial experiences. To do so, we studied the voids created by the branching arrangements within each analog model.
Analog Model A
Abstraction & Rubber Bands
After feedback from our midterm critique, we decided to use rubber bands to physically model our abstracted branching patterns to maintain the clean angles in our diagrams. The first analog model corresponds to the last iteration of our thread models with the inclusion of a singular split level and the equilateral triangle grid. We also experimented by altering the density of the branches vertically above and below the split level.
Analog Model B
Multiple Split Levels & Branching Logic
The addition of another split level allows us to arrange interactions between bands that span three different lengths. We also repositioned the double and triple height spaces to a more central location to feed into the surrounding areas. We imagined that this space could translate into an atrium and circulation core for the building. In addition, the branching logic for Analog Model A lacked a clear organizing system, so Analog Model B forms branches from clusters of three bands on the top and bottom.
Analog Model C
Shifted Atrium & Central Branch
With our site in mind, we shifted the atrium space to create a clear entrance and façade at the southern corner of the building for pedestrian access. As the levels ascend, the opening twists across the building to create interesting views and varied height spaces. Inside that rotating atrium space, we designed a large central branch to anchor the rotations.
Finally, the programming of our building was revealed to be a library. Our spatial system was flexible enough to accommodate the new program, and we calculated the square footage necessary for each space and aligned suitable adjacencies between programs across the floors.
Branched Column Structure
From our material experiments and analog models, we abstracted the geometric logic and branching formations from the diagrams to formulate the column structures of our library. As the branches configure together, they attach side-by-side to create planar-like surfaces that act as partitions on some floors. The largest branch structure that fills the atrium space twists as it ascends, a trait of this system that we found in a material model. It creates a sense of movement and transformation while still remaining a structural element. To emphasize this twisting motion and provide a more dynamic spatial experience, the atrium openings twist as they rise through the building providing a range of views throughout. finally, a central staircase wraps around the main branch to provide circulation around the architectural feature.
"The jury looked favorably upon the experimental nature of this project. The scaled model can be a great vehicle to study and simulate a structural design. The use of inverted tensile forces in both design and model-making processes was highly commendable." - ArchiHacks