Generative Wall
A study exploring versatile wall designs through generating and rotating openings in the wall.
Team Members: Malvina Mathioudaki, Qiwei Sun, Vinay Agrawal
a. Introduction
This project explores a design of a wall that could be used in various circumstances. The walls are generated through rotating multiple openings in the wall. The rotations could be horizontal or vertical, and could be along any vertical or horizontal axis within the openings, and at any angle.
This project is inspired by STOREFRONT FOR ART AND ARCHITECTURE project by Steven Holl Architects, and thinking of further possibilities and usage of this design prototype.
This project defines three circumstances, exhibition, outdoor dining and duty-free shelves. In each circumstance, certain restrictions are applied to make designs suitable. The design strategy includes a group of random parameters and could generate a great number of designs. Discover is then used to optimize the designs towards different goals according to the circumstances.
b. Methodology
i.)Design Space Model
The designs are generated on a wall with a preset dimension of 10 feet * 45 feet. The generation of the designs of the walls mainly includes two parts. Firstly, The wall is divided into multiple rectangular parts, from each of which the openings are generated. Secondly, the openings are rotated to make up the use of the wall.
The Division of the wall is conducted through a recursive process. At each step, a random part of the wall is divided in a random direction, vertically or horizontally, and at a random position. This process is repeated until a preset number of rectangular parts are generated. The openings are generated from each divided part of the wall. The boundary of the part is offset inside for 1 feet to make space between openings. Then the aspect ratio of the rectangular openings are evaluated, and the openings with aspect ratio larger than 5, which indicates they are too thin or too long, are discarded.
The rotation of the wall could be in both horizontal and vertical directions. A random parameter is applied if the direction is not fixed by the circumstance. Then the rotation axis is placed within the range of each opening. Another parameter is used to define the position of the rotation axis. The axis is constrained so that the openings are not rotated around the very edge of the openings. Finally, the degree of rotation is set to make the wall design.
After the design is made, it is evaluated under three different circumstances. i) for the exhibition, the area from which people could see the exhibition is evaluated. In this circumstance, all openings are rotated around a vertical axis, and the maximum distance of seeing the exhibition is set to 15 feets. Circles are drawn from the edge of the openings, and the area of a combined region is evaluated. The optimal design is expected to provide maximum area.
ii) for outdoor dining, the number of seats provided is evaluated. In this circumstance, each opening is rotated around a horizontal axis and by 90 degrees. An opening is considered to create a place for dining only if its height is between 2.5 and 3.5 feet. Then a seat is generated every 2.5 feet along the combined edge of the qualified openings. The optimal design is expected to provide most seats.
iii) for duty-free shops, the maximum area for showing goods is evaluated. In this circumstance, the openings are rotated both horizontally and vertically, and by any angle. The height range of showing goods is here considered from 3 feet to 6 feet, and the parts of openings that fit into that range are counted. The optimal design is considered to provide maximum space for showing.
In all three cases, the number of qualified openings is considered another performance metric to generate designs with more openings.
ii.)Input Parameters
Division Direction — Categorical — Boolean: Identifies the direction of division when dividing the wall into rectangular parts. (True = Vertical, False = Horizontal)
Division Position — Continuous — Float: Identifies the position of division when dividing the wall into rectangular parts.
Rotation Direction — Categorical — Boolean: Identifies the direction of rotation of each opening. (True = Vertical, False = Horizontal)
Rotation Position — Continuous — Float: Identifies the position of rotation axis
Rotation Degree — Continuous — Float: Identifies the degree of rotation
iii.)Performance metrics
(For Exhibition): Area of watching — Maximize
(For Outdoor Dining): Number of Seats — Maximize
(For Duty-free Shops): Area of showing — Maximize
(For all three cases): Number of Openings — Maximize
c. Results
The optimization is run on Discover with 35 designs for each generation and for 25 generations. The optimization is finished separately for each circumstance.
1)Exhibition
According to the optimization result, the options with maximum area tend to have less number of openings. An option with 14 openings was chosen as the optimal design.
2)Outdoor dining
According to the optimization result, the options with maximum outdoor dining seats tend to have less number of openings. An option with 11 openings was chosen as the optimal design.
3) Duty-free shops
According to the optimization result, the options with maximum showing space tend to have less number of openings. An option with 13 openings was chosen as the optimal design.
d. Conclusion
This project developed a strategy of generating wall designs for different uses. A total of five input parameters are used to generate random options. So the generation of optimal designs need more tryouts, on which generative design really helped a lot. The vision is to make a wall versatile through making openings and rotation of the openings. A random sense is expected to make the design more interesting. As a result, generative design methods are applied to generate random design options, from which the optimal options are selected. In the circumstances, designs with maximum usage tend to have fewer openings. So, the number of openings is set as another optimization goal to make more openings in the designs for more versatile results.
The project explored the possibilities of creating space through rotating openings of a wall in three different circumstances. The circumstances have different restrictions and generate different optimization results, while the third circumstance had no restrictions on rotations. More restriction options and optimization goals could be further explored to generate different designs towards different functions and under different circumstances.