Evolutionary problem solving mimics the theory of evolution employing the same trial-and-error methods that nature uses in order to arrive at an optimized result. When automated for specific parameters and results, this technique becomes an effective way to computationally drive controlled results within the iterative design process – allowing designers to produce optimized parameters resulting in a form, graphic or piece of data that best meets design criteria. In this post we walk you through the process of using Galapagos, an evolutionary solver for Rhino/ Grasshopper, and show an example of how this method can be tied in with analysis tools to optimize form based on energy data.
David Rutten, a developer with McNeel & Associates, created a tool called Galapagos which facilitates this process within grasshopper(graphical algorithm editor). He covers the basics of the program and potential uses for Galapagos in his 2010 lecture Computing Architectural Concepts at the Architectural Association in London, and in a blog post, Evolutionary Principles Applied to Problem Solving.
The video below shows an experiment illustrating the entire Galapagos process: a simple, parametric form is modeled and run through the genetic algorithm searching for the “fittest” 3 dimensional form (the form that yields the lowest amount of solar radiation). In this example, the inputs are two curves that are lofted to form a faceted surface open on the top and bottom – the “genome”. The parameters, controlled by the sliders, manipulate the form – sending each iteration to Ecotect, an environmental simulator, where it is evaluated for average solar radiation within a specified environment (process for exporting forms to Ecotect is outlined here). These values are then sent back to Grasshopper, where they inform the algorithm to focus on conditions that favor low solar radiation; thereby, each new generation draws from only the fittest instances of the generation before, creating at each pass a more-optimized result. The final form is an inverted cone – all planes tilt away from the sun’s path and provide shading across the overall form.. .
As the designer, you control the initial formal inputs (in this case the choice of two lofted polygons), which parameters are manipulated, which results the program values as “fit” and which generation best meets your desired level of refinement. For example, the program is then re-run, favoring results that maximize solar radiation exposure. The result is the opposite: a faceted cone with its planes tilted in, maximizing sun exposure.example file * to run this file you will need Rhino3d, Grasshopper3d and Gheco GH