Courtyard Cluster Housing
Columbia GSAPP | Generative Design | Spring 2023
Chung-Ying Hor | Ting-Wei Shih | Weiheng Zhao | Joe Mihanovic | Sixue Chen
Special Thanks: Haojun Wang | Ziyao Gao
Introduction
Addressing the growing need for sustainable and community-oriented living spaces, this design concept seeks to foster a harmonious balance between private and communal areas. By segregating the spaces, each individual room benefits from natural light streaming in from all four facades. Furthermore, the site is divided into a series of small courtyards organized based on their proximity to specific rooms.
In this architectural typology, the courtyards are regarded as outdoor rooms, with designated primary ones allocated for specific functions, such as outdoor dining areas or compact basketball courts. These shared spaces encourage social interaction and promote a sense of community among residents. Consequently, our aim is to devise a layout that optimizes not only the adjacency between interior rooms but also the connectivity between select interior spaces and their corresponding courtyards, fostering both ecological and social sustainability.
Inspiration
The innovative Moriyama House, designed by renowned architectural firm SANAA, serves as our inspiration for creating the “Courtyard Cluster Housing” concept.
Methodologies
We have designed a configuration wherein 3 to 5 rooms encircle a courtyard, forming a cohesive cluster. Three such clusters are strategically placed within the site to create an optimal layout that enhances the overall living experience.
Environment Set:
a. Site Dimension: 144 ft x 90 ft
b. A Park Located in the Northside
c. Main Street Located on the Southside
Fixed Inputs:
a. Rooms set for each of the 3 courtyards(3–5 Existing Rectangles)
Variable Inputs:
a. Courtyards (Area / Width / Location on Site)
b. Room (Location around Courtyards)
Design Space:
Step 1: Randomly set courtyard planes’ dimension
Step 2: Randomly locate existing rooms to surround each courtyard
Step 3: Correct rooms’ overlapping
Python Script for Checking Overlapping:
Step 4: Randomly Put three clusters in the site
Step 5: Correct clusters’ Overlapping
Step 6: Analyze Discover/Galapagos
Output Metric:
a. Privacy: Courtyard View to Street (Minimum)
b. View: Courtyard View to Park (Maximum)
c. Adjacency: Distances between Offices (Minimum)
d. Adjacency: Distance from Offices to Main Street (Minimum)
e. Scattered: Distances between Rooms (Maximum)
Result
The Courtyard Cluster Housing concept is designed for architects, residents, and real estate developers seeking an innovative approach to community living. The core value of this product lies in the optimization of the layout, with scattered clusters of rooms strategically arranged around courtyards. This arrangement focuses on factors such as adjacency, views, and daylight to enhance the overall living experience.
To deliver this value, a minimum of three sets of courtyards, each encircled by their surrounding rooms, is required, along with a suitable site for development. This unique layout fosters an environment that encourages interaction among residents while providing private spaces for rest and relaxation.
According to the six results(#1966, #1951, #1939, #1834, #1628, #1419):
a. The park facing the north side has no rooms obstructing the view.
b. The offices are all situated close to the main southern streets, blocking the street view to enhance privacy.
c. To ensure the comfort of all courtyards, the three courtyards must be close to narrower rectangles.
According to the best two results(#1951 & #1628):
a. The arrangement of the rooms tends to be L-shaped to avoid overlapping rooms on a limited site.
b. Maximize the use of open space on the site.
c. The smallest cluster is placed in the center of the site to maximize the view of the left and right courtyards towards the northern park and also to allow the three offices near the south to create an independent small garden that can serve as the main entrance.
Conclusion
We believe that adopting the bottom-up approach of placing rooms in clusters on the site is more in line with expectations compared to using the “subdivision” top-down method.
In “subdivision,” after drawing the first line, no space will be placed on the line of the subdivision, and the boundaries of the clusters will be clearly divided, which may result in missing some interlocking possibilities.
Reflection: In the current implementation of our system, we employ a bottom-up approach, wherein we randomly place options in the process of generating potential solutions. While this method may yield some valid outcomes, it also inadvertently gives rise to a considerable number of invalid choices. These invalid options can negatively impact the overall efficiency and effectiveness of the system, as users must sift through these undesirable results to find more viable solutions.