- What problem are we trying to solve
- Given site constraints, eliminate poor building configurations based on excesssive structural cost
- What are we not trying to solve
- Architectural programming
- Formfinding in the architectural sense
- What skillsets do we have at our disposal
- Clifton - presentation, expertise in drawing buildings with math (at the macro level), other engineers who can do front end development if needed
- Anthonie - geometric visualization in Rhino & Grashopper, parametric models for optimization, structural psuedo-design & analysis
- Luke - .Net as it relates to geometry, generate variations, set up any databases, front-end angular app dev
- Mathias - Machine Learning, database queries, Python, AWS, Jupyter notebook
- What is our technology stack
- Rhino/ Grasshopper
- SQL Lite
- Jupyter
- Python
- Determine randomizer parameters and ranges
- Unit mix (3 variations)
- Tolerance
- Unit layout (4 variations)
- Typologies (6 variations)
- Dimensional constraints
- Vary depending on the typology
- Limit to multiple of 6 feet
- Seismicity (Sds - 0.5, 1.0, 1.5)
- number of stories (2 - 5)
- Drift limit (0.15, 0.20, 0.25)
- Unit mix (3 variations)
- Determine performance metrics (both binary and quantitative)
- Total structural shear wall cost
- Do all walls meet upper limit on force
- Does drift meet upper limit
- Determine database table schema
- Consider the impact of rho
- Consider a nonlinear wall-cost model
- Luke
- Set up git repo
- Rigid analysis & wall evaluation widget
- Database setup
- What database format?
- Variation generator
- Anthonie
- Unit layout generation (from CAD layer - could be a rectangle or something more complex) to get close to target unit mix
- Mathias
- Set up Jupyter notebook
- Consider different Machine Learning models
- Consider AWS for training the model
- Set up an EC2 instance (use a bucket if it needs to)
- Clifton
- Visuals, graphics
- Business case
- Presentation (Powerpoint)
- Market data