Part 1: Generative Design Framework
Design Decision 1: Constructing a net-zero building
Design Variables
- Renewable energy (solar panels) and green roofs
- Building envelope: shades and overhangs to reduce direct sun rays, and high levels of insulation
- Energy efficiency measures: energy control systems, ventilation, and HVAC
- Life cycle assessment of the building
Evaluators
- Maximize the area available for solar panels/green roofs: including lower walls, walls of a particular orientation (to maximize insolation), and roof space
- Maximize solar insolation potential: The more natural lighting the building receives, the less energy needs to be
- Optimize window-to-wall ratio (range constraint): Divide the number of total exterior wall area but the total area of all window openings in the building
- Minimize embodied carbon: sum of the estimated carbon intensity of each material used multiplied by volume
Trade-offs
- Solar insolation potential vs energy intensity: Maximizing solar insolation potential could result in greater energy costs during warm periods like summer or heatwaves. During cold periods, it will help to reduce the net load though.
- Construction cost vs operating cost: Net-zero components of the building have high upfront costs of installation, such as solar panels and green roofs, but will pay off in the long run due to lower energy costs.
- Green features vs rentable area: Optimizing for net-zero goals may mean that less floor area can be rented for commercial/residential or other purposes as they have to be set aside for emissions reduction/ efficiency purposes.
Design Decision 2: Constructing a skyscraper in an earthquake-prone region
Design Variables
- Height of Building and Floor Height
- Material Type and Cost: Earthquake-resistant construction materials, e.g. ductile steel buildings, may cost more. There is a possibility of utilizing different materials for different parts of the building for optimizing cost and minimizing trade-offs on seismic resistance.
- Seismic Design: follow earthquake regulations in design (e.g. density, building capacity, open areas for evacuation)
Evaluators
- Minimize construction cost: Sum of both total floor area (depends on the height of building and building geometry) multiplied by cost rate and weighted by material cost for different components (e.g. lower vs upper floors, or core vs accessory components)
- Minimize the amount of deformation: Proxies of earthquake damage can be used, standardized on the expected earthquake intensity, e.g. inclination rate/ building slant, displacement of the building
- Maximize access to safety features: area of open spaces for evacuation, and average time to escape the building (e.g. greater floor area on lower floors than upper floors = lower average time)
Trade-offs
- Construction cost vs Height vs Deformation: Taller buildings tend to sway more when there are seismic shocks, which may make them less vulnerable to earthquake damage/ deformation (less stiff). However, taller buildings cost more.
- Aesthetic quality vs Seismic resistance: e.g. interlocking steel frameworks may improve seismic resistance at the expense of the building’s aesthetic features.
Design Decision 3: Constructing a multi-family residential affordable housing
Design Variables
- Ease of pre-fabrication: repetitive structures that can be quickly installed for fast turnover
- Reliable, sturdy materials
- Energy efficiency
Evaluators
- Minimize Construction cost: based on total floor area and material cost
- Optimize view quality (range constraint): lines of sight for varying building shapes, influenced by the number of units and windows, as well as the height of the building against surrounding buildings
- Optimize rent (range constraint): maximizing floor area against constraints (available space, zoning requirements) and multiplying it by rental rate
Trade-offs
- Construction cost and time vs building aesthetics: The aesthetic design of the building may take a backseat in favour of reducing construction cost and time — a simple design that suits prefabrication and modular construction will be most efficient.
- View Quality vs functionality of the building: Not everyone will have great uninterrupted views/lines of sight as the building will seek to maximize floor area for more units with sufficient living space.
- Energy efficiency/operating cost and construction cost: Adding energy-efficient features to the building will cost more initially, but may pay off in the long run. This should be compared to the rental rates to understand if the initial costs are worth it.
Part 2: Generative Design Study
I chose to study Design Decision 1 - a net-zero building. Inspired by the stacked/blocked building design, where different roofs levels can be used for urban farms, vegetation or solar panels (as seen in the image above), I used the multi-box template as the starting point of my model. This was a good model to use for several reasons:
- able to maximize roof space for solar panels
- able to iteratively test different relative positions of the blocks to maximize solar insolation potential, e.g. more wall area facing the sun
- able to adjust the height of each tower and total volume of the building to minimize construction cost while accounting for the above green features
Design Variables
- Cuboid height
- Cuboid relative X-location
- Cuboid relative Y-location
- Project location
- Construction cost of walls
- Construction cost of roofs
4 Evaluators
- Total volume
- Cumulative solar insolation potential on walls
- Total roof area
- Total construction cost
Instead of three cuboids, I added a fourth cuboid just to increase the variety of results, and created the following input parameters that can be adjusted (height, X-location and Y-location). To ensure that each cuboid tower would not converge into just one block, I set the range of each parameter such that the buildings overlapped but not too significantly.
I used a Data.Remember node for both Site Location (San Francisco, CA) and Sun Settings Start Date Time and End Date Time to capture the output of these node and cache the results in the .dyn file when the graph is saved. This is therefore an input variable that can be changed in a new Generative Design study run, but not within each run.
Cost metrics were also included as input parameters for subsequent cost evaluation. I differentiated construction cost per area for walls and roof, but set them as constants in the Generative Design study.
Net-Zero Building Features
First, to optimize the roof area for solar panels and walls for green features, I created a group of nodes to select them separately for further analysis. The PolySurface.BySolid and PolySurface.Surfaces nodes extracted the surfaces from the tower solid and the surfaces were filtered by calculating their normals (perpendicular if greater than 0, and therefore a roof surface). The respective surfaces are colorized differently as well.
The volume, roof area, and wall area can be calculated directly from the Solid, Roof Surfaces and Wall Surfaces respectively. I decided to maximize the volume of the building.
Next, to calculate the cumulative solar insolation potential on the walls, I used the SolarAnalysis.Analyze node with a grid spacing of 8. This feature is maximized to improve the emissions reduction potential of green walls.
Lastly, to calculate the construction costs, I multiply the input costs by the wall area and roof area respectively and aggregate them. This will be minimized in the Generative Design study.
There are 4 outputs: total volume (maximized), total roof area (maximized), total wall insolation (maximized), total cost (minimized).
Trade-offs
The main trade-off that this design study seeks to address is the tension between minimizing construction cost and maximizing green building features. Green building features like solar panels and green walls are important to attain a net-zero building by contributing to emissions reduction, but they are expensive and the developer will require cost control as well.
The input parameters allow the relative locations of the four towers/cuboids to be varied, meaning that the tallest towers may not give the largest green wall or solar panel surfaces. This is interesting because it allows me to identify which of the cuboid’s parameter (e.g. height, X/Y-location) is most important n optimization.
Volume was used as a substitute for more precise measures, e.g. total rental potential of the net-zero building. I felt it was necessary to compare volume against surface area, cumulative solar insolation potential, and cost to ensure that a range constraint can be set, i.e. the final volume of the building must fall within an expected range. This will help to eliminate oddly-shaped design outcomes from the generative design study that may optimize other features, but are too big or small and inappropriate for actual construction.
Although the time-sensitive issue of maximizing cumulative solar potential (i.e. during summer and heatwaves) has been discussed above, it was not considered as a significant trade-off to make in this design study as the green walls are prioritized, and their design may be able to incorporate insulation and building heat management.
Part 3: Generative Design Study Results
Inputs tested (each of the 4 cuboids):
- Tower height
- Tower’s relative X-location
- Tower’s relative Y-location
Inputs set as constant:
- Project location: San Francisco, CA
- Cost of wall per area = $40
- Cost of roof per area = $70
Outputs:
- Total volume (maximized)
- Total roof area (maximized)
- Total cumulative solar potential on walls (maximized)
- Total construction cost (minimized)
Discussion of Results
Total roof area increases rather linearly as total volume increases, which is expected as there will be less overlaps between the cuboids in the design. However, this does not depend too much on the height of the tallest tower (C1), i.e. a range of heights for the tallest tower can give similar roof areas and volumes. The tallest tower’s height may be a constraint in certain locations based on zoning requirements, but the developer may also want to increase the height for greater rental revenue (more floors + greater rental value on upper floors).
For a range of heights - C1 around 170 to 180 feet and C2 around 100 to 110 feet, there is a small variance in total wall insolation, suggesting that wall insolation correlates closely with tower height. However, with greater variance in the total cost and total volume, it suggests that there is a mixture of effects with other variables with less clear relationships.
Referring to the yellowish-green dots, the scatterplot above shows that total volume can be maximized for the designs with similar C1 heights and total wall insolation.
Maximizing the total wall insolation will increase emissions reductions and potentially reduce operating costs of the buildings in future through green walls. As the scatterplot shows, there is no direct relationship with tower heights.
Following this, I decided to focus on parallel coordinates graphs to look at the trade-offs for different features:
This shows that the maximum total roof area is highly correlated with total volume, total wall insolation, and total cost. This is a good example of how different designs (relative positions of all 4 cuboids) can give similar evaluative features, which allows the developer to consider even more factors outside what is modelled in this generative design study, such as technical feasibility, construction time, and internal design.
When trying to maximize total wall insolation, there is a greater variance amongst total roof area and total cost, which is consistent with how cost is influenced by both wall and roof area. This suggests that a net-zero building cannot have the best of both worlds, i.e. maximize both wall insolation for green walls and roof area for solar panels, or the total cost would be too large. The developer would need to prioritize one green building feature over another in order to maintain cost control while ensuring maximum emission reductions possible.
For a middle range of construction cost at around $3,000,000, it seems that maximizing wall area is preferred over roof area. However, further information would be needed with regards to the long-term viability of solar panels vs green roofs, e.g. greater maintenance costs for green roofs, additional utility derived from the aesthetic value of green walls, or the reduction in energy costs due to solar energy generation.
Possible Design to Recommend
Based on largely median values for all 4 evaluators (without weighting like in Module 6), the design below could be considered by the developer.
