/w=1080,quality=80,fit=scale-down)
/w=3840,quality=80,fit=scale-down)
Step 1 - Generative Design Framework
A very brief description of the design decisions from Step 1 following the Generative Design Framework.
- Design Decision 1: What is the optimal location of cabins in a plot of land in order to allow for privacy and minimize distance to points of interest (e.g. fire pit or bathroom)?
- Design Variables: degree rotation, u, v coordinates, obstacles
- Evaluators: privacy score, cost to build paths, distance to points of interest
- Most Important Tradeoffs to Consider: privacy vs. proximity, spread out vs. clustered
- Design Decision 2: What is the optimal location and orientation of shading panels on a building façade in order to maximize shade in the summer and provide sun in the winter based on use of the rooms?
- Design Variables: vertical vs horizontal, width of panels, height of panels, frequency of panels
- Evaluators: unobstructed views, cost, amount of shading/sunlight
- Most Important Tradeoffs to Consider: cost vs. performance, orientation/placement vs. building aesthetics
- Design Decision 3: What is the optimal shape of a 3-part building with three different shapes?
- Design Variables: u, v coordinates, building height, building material, building shape
- Evaluators: views, cost to build, floor area, time to build
- Most Important Tradeoffs to Consider: views vs. time to build, shapes vs. cost
Step 2 - Generative Design Study
- I decided to explore Design Decision 3 from above. Making it more of a real-life application, I assumed I was exploring initial design options for a new children’s science museum and that the structure would be a 3-part building made by combining a cuboid, a cylinder, and a cone. The goal for this assessment was to explore different combinations of the building shapes (modifying location and height of each, independently) and evaluate the best option for minimizing cost to build, minimize additional time required (compared to a traditional rectangular building), maximize floor area, and maximize glazing area (children LOVE looking out windows!).
The generative design takes in two kinds of inputs, variables (which are modified during the evaluation to explore different combinations) and constants (don’t change from one design iteration to the next). These are shown below.
/w=640,quality=80,fit=scale-down)
/w=640,quality=80,fit=scale-down)
Next, the graph manipulates the geometry of the three building parts (cuboid, cylinder, and cone) by exploring different combinations in their location (x, y coordinates), as well as their height. Since it is a playful science museum, each section of the building is colored with a different vibrant color.
/w=3840,quality=80,fit=scale-down)
Next, the graph performs four evaluations (the first two are done within the same group of nodes):
1) Compute the total exterior wall surface area of the combined buildings to assess how much surface area could be dedicated to glazing, providing views for the curious children to explore.
2) Compute the total floor area by creating planes at 10 ft vertical intervals (height of one floor) and intersecting with the solid resulting from the building mass. The larger the total floor area, the better since it means there is more space to set up exhibits.
3) Cost to build, which is computed by assuming that the cost to build a unit area of building space (1 sq ft) increases every 10 stories.
4) Time to build, which is computed by assessing how much of the exterior of the building corresponds to each of the three main geometric shapes and uses a factor for each to calculate the amount of time it would take to procure and construct relative to a standard rectangular building. This scaling is based on the fact that curved and slanted facade components (such as those needed to build a cylinder or a cone) require a greater lead time to procure and also take longer to install on site.
/w=3840,quality=80,fit=scale-down)
Finally, the graph results in four outputs, which are used as evaluators to compare different design alternatives.
/w=750,quality=80,fit=scale-down)
Step 3 - Generative Design Study Results
It was fantastic to see the creative design combinations that the computer came up with.
Upon evaluation, it was interesting to visualize views (window area) plotted against the time to build score (normalized against the assumed time to build of a rectangular building of equivalent surface area). In what is somewhat to be expected, the plot suggests that the more window area there is, the longer it takes to build…until a certain point at which the trend changes. This is likely due to there being a threshold beyond which one shape gets too large and therefore making another shape larger actually offers more added window area per unit of construction time. What’s more, when “cost to build” is superimposed to the plot (represented by the size of each data point) it is evident that cost to build coincides with the peak in the plot before the trend begins to reverse. With these insights, we can better decide what characteristics we want to prioritize in the building’s design and we can decipher that within a cluster of designs, the overall design benefits from the cylinder and the cone sharing a surfaces.
/w=3840,quality=80,fit=scale-down)
/w=3840,quality=80,fit=scale-down)