Suggesting Design Directions: Early Examples of Simulation-Based Guidance for Common Model Types

Simulation data can inform early design, but for many design spaces, there is a need for distilling large amounts of performance data into guidance that can support creative, interactive design. Automated computational processes could help support this activity. This paper demonstrates a procedure for finding suggested design directions based on automatic simulations of design options for three common model types. The workflow involves reading in a static geometry and basic simulation information, automatically generating dummy variables, running a series of simulations, applying data analysis to find a direction for improvement, and then returning that direction to the user. This process is demonstrated on a 2D truss for reducing weight, a 3D surface structure for increasing stiffness, and an urban neighborhood concept for increasing PV potential. These simulations reveal one possible foundation for a future design system that intelligently suggests areas of performance improvement for an initially fixed truss, shell, or urban form.