Advanced manufacturing techniques and new materials are unleashing a new era for engineers to create new breeds of high-performance designs. Fully exploiting these opportunities is very complex for human designers alone, as it is often difficult for our minds to imagine the intricate shapes and materials that embody the best design for a given purpose, such as the best possible shape and material for a wind turbine blade. Generative design tools are a new class of computer programs that uses algorithms and artificial intelligence to help engineer solve these challenging problems, by virtually experimenting with thousands of designs to discover the best ones. To give a generative design system a way to distinguish good designs from bad designs, powerful computers simulate a variety of natural phenomena in a virtual environment. Designs meant to be used in environments where high-pressure and/or high-speed fluid flow is present are especially difficult for generative design to discover, as the accurate simulating the turbulent, chaotic flow of water or other liquid or gas requires enormous computational power, and thus it only allows generative design to be used for objects of very small scale. We propose using modern artificial intelligence techniques to radically improve the speed at which we can obtain accurate simulation of environments where high-speed flow is present, and therefore enable vastly improved, generatively designed objects, that are meant to be used in such environments.

Industry Partner(s):Autodesk Inc.

Academic Institution:University of Toronto

Academic Researcher: Richard Peltier

Focus Areas: Advanced Manufacturing

Platforms: Cloud, Parallel CPU