SMART LIVING LUNCH | Structural Design Space Exploration through the generation of variant networks of bars in static equilibrium, using grammar rules and interactive genetic algorithms

Thumbnail

Event details

Date and time 22.06.2021 11:0012:00  
Category Conferences - Seminars
In this Smart Living Lunch titled “Structural Design Space Exploration (DSE) through the generation of variant networks of bars in static equilibrium, using grammar rules (GR) and interactive genetic algorithms (IGA)”, Ioannis Mirtsopoulos (SXL, EPFL) will present selected results from his PhD thesis. Follow the online presentation on 22 June 2021 from 11:00 to 12:00. Link on demand.

Architectural design only represents a tiny subset of the design space and it is often constrained by premature design fixation and lack of diversity and/or creativity. The exploration of variant structural forms is constrained by the lack of computational tools that unveil structures in static equilibrium, which are liberated from precedent structural typologies. The generation of bespoke networks of bars in static equilibrium during the early conceptual design is crucial for the design of unconventional structural forms.

Ioannis Mirtsopolous has developed a design pipeline free of numerical design variables, that implements a rule-based design approach to generate new, unexpected, but valid, (structural) networks of bars. The grammar rule is applied incrementally, blindly regarding the design domain and specific structural typologies, and builds the network’s topology impulsively, namely transforms it, while aiming at the transition from a state in interim static equilibrium to one that is in global equilibrium. Its syntax builds on vector-based graphic statics, embeds structural awareness and ensures the network’s static equilibrium throughout the transformations. The grammar rule’s exploratory power is amplified through its fusion with interactive genetic algorithms. Rather than pre-defining the properties of the successive grammar rules, variant design features ruling the incremental transformations are applied, while searching for optimized structural networks. At the end of each transformation (step), the generated (intermediate) design variants are compared based on qualitative (i.e. aesthetics) or quantitative (i.e. static action, number of joints, number of members etc.) criteria. The developed computational tool operates in the parametric environment of Grasshopper.