Computational Optimization
Sampler of Various Projects
Drawn from a number of different courses, assignments and projects, these examples demonstrate different approaches to optimization.
Whilst most of the examples are based on structural analysis, the tools of linear regression, genetic algorithms, and multi-objective optimization can be brought to bear on just about any design problem with parametric conditions and complex goals.
Multi-Objective Optimization (MOO)
For the Tumbling Tumblers, I ran the base parametric model and functional targets for the design through a multi-objective optimization program. Whereas I previously could only design one or two iterations of the balancing containers, this allowed me to generate thousands of iterations to target for an optimal balance of 1) volume of liquid contained 2) ratio of mass of vessel to the liquid 3) how much the tumblers tilted and moved. The mass sampling allows for human input as well, for aesthetic selection along the pareto curves.
Michell / Prager Truss
Parametric model processed with Particle Swarm Optimization (PSO) to generate different solutions for an optimized truss (cantilever). Objectives were calculated with Karamba, a finite element analysis model, with variables such as the number of structural members, and the base dimension of the truss, being altered with each solution.
Active Bending
Accurate form finding and structural evaluation of form-active structures using Grasshopper (parametric visual code), Kangaroo (spring-based physics simulator), K2 Engineering (Kangaroo add-on for active bending), and Karamba (finite element analysis). This work subsequently informed form finding for the bamboo structures in the Wang Shu Design Build project.