A Generative Design Approach to Outdoor Gathering Space

Alonso L Ortega
Generative Design Course
6 min readDec 20, 2021

--

Dodge Plaza Redesign

Ata Gun Aksu, Ece Cetin, Max Cai, Zhanhao Fan, Alonso L Ortega

Columbia University’s Morningside campus offers grass and hardscape areas that can be used for a variety of outdoor or tented events. The existing plaza layouts are organized more generically and have not been designed to allow for flexibility of uses. Our approach is to introduce singularity to the design of the plaza by optimizing the interior layout for circulation, designing for solar radiation, and introducing plan distribution of four program types. These programs would include a large event, dining, workstation, and a leisure area that is manageable through the use of a multipurpose seating device. Through this, we are aiming to activate the plaza as a dynamic point on campus. The modularity of the canopy design and the generative design approach makes it possible for the project to be incorporated into other outdoor event spaces on campus such as Ancell Plaza which also lacked a clear approach.

Dodge Plaza is located near the upper campus of Columbia University and is a large, paved area surrounded by landscape. With a maximum occupancy of 700 individuals, it offers plenty of space to support large and small activities to the students. The design approach that is usually taken in this area is to tie down a large shed point tent and scatter around enough chairs at the time of the gathering. The interior circulation is busy and there is no clear path as to where the main entrances are located or where one should follow to get to their group sessions. This limited and over-complicated solution can be distributed through generative design that responds effectively to these problems. It is thus that we are taking on two major goals for this project. Our priority will be to maximize the distribution of four programs while optimizing a clear and efficient circulation path to the main entrances. Our second priority will be to respond to a more dynamic and permanent shading device.

Figure 1. Dodge Plaza (existing) adjacent to the Dodge Building and Lewisohn.

Bringing better outdoor public seating experiences has been an ongoing problem for the University’s Administration. Ancell Plaza, for example, is an ongoing project initiated by the Common Space Initiative. The organization completed surveys and completed an iterative design process that offered a low budget and minimal architectural intervention solution. From this organization’s study, it was found that lighting, seating, shade, food/coffee, and greenery were among the top elements that people wanted to improve in the plaza. This project’s solution will differ from The Common Space’s approach in that generative design will be a major decision-maker for the outcome of our interior layouts. But our goals to provide more vibrant, cultivating social activities will be the same.

We aim to design better outdoor public seating at Dodge Plaza by efficiently maximizing spatial conditions through modularization and ensuring green space and walkways are unobstructed. This project has been an ongoing problem for the University’s Administration as the traditional design approach has not taken into consideration a more customized approach that uses quantitative data to mitigate proximities and walking distances, as well as natural lighting.

To be able to achieve our objective, our process must include a multi-criteria optimization that starts at defining our programmed areas. Type A is prescribed for large events and has been given a diameter of 10 meters. Type B has been programmed as a dining/gathering space with a diameter of 5 meters. Type C is a smaller program for group work labeled workstation, this has a diameter of 3.6 meters. Lastly, Type D is single-use labeled leisure and has a diameter of 2 meters.

Figure 2. 4 Types of spaces designed for the plaza. Left to right; Type A-Large Event, Type B-Dining, Type C-Work Station, Type D-Leisure

The next step will be to define the dimension of the available space for Dodge Plaza. This is drawn as a parameter in grasshopper and using a generative design method deployed to create the shortest paths to multiple main entrances of the plaza while optimizing the usable space with the four programs and having a one-meter buffer that will serve as a walking path. Discover, a grasshopper plugin is used to maximize the number of these spaces distributed within the plaza, while ensuring the shortest walkways. After the layout optimization, the modular canopies are distributed in the plaza in response to our goals of lowering solar radiation in the program spaces and maximizing them at the walkways. To achieve this goal, the column heights under the canopies will use an optimization that will be adjusted at varying heights and angles.

Figure 3. Initial sketch for layout organization and shortest walk for Dodge Plaza.
Figure 4. From left to right; Grid for shortest walk calculation, optimized space layout, shortest walks from access points, the paved circulation, the canopies.
Figure 5. Four types of spaces with furniture layout; large event space, dining space, workstation, and leisure space.

When performing the shortest distance optimization, we were expecting something similar. However, we were intrigued by the outcome chosen with how the four zones were allocated. We noticed that Discover seemed to have an easier time implementing these sequences while the larger areas were closer together. This mitigating of space helps us at a schematic step of design, which the designer could then implement the floor plans for these circle areas.

Figure 6. Multi-image composite output during optimization and most optimal output.

We performed the grasshopper/python optimization with the following settings in discover:

Number of designs per generation: 20

Number of generations: 20

Mutation rate: .01

Figure 7. Screenshot of Discover Platform.

For the canopy design, we first began with a direct sun hours analysis of the site which helped us run an optimization of the least amount of direct sun hours for the opening location of the design. With the most optimal output of least solar radiation to the interior layout, it resulted in our final adjustment of the deflection and we were able to combine this with the interior layout.

Figure 8. Multi-image composite output during optimization

We performed the grasshopper/python optimization with the following settings in discover:

Number of designs per generation: 20

Number of generations: 20

Mutation rate: .05

Figure 9. Screenshot of Discover Platform.
Figure 10. Optimal canopy design.

Overall, the project was successful in its deliberation to create an optimal design that is permanent and has the flexibility to host different events. There could be further development of the type of furniture used or a deeper investigation of a modular seating device that could serve various activities. In all, the application of using shortest distances and the ability to have collision data that optimizes space is a tool that offers multiple solutions that should be considered in the way we think about movement in the design field since most practical solutions today tend to be the most inefficient and not how people get from point A to B.

--

--