Objective: Develop and implement into a production environment, genetic algorithms(GA) capable of generating optimised shapes for steel transmission towers.
key challenges: -How to adapt the often conflicting geometry restrictions of design codes with the organic shapes returned by Genetic algorithms? How to do this without loosing the material efficiency gains of the GA? -How to implement this into the work flow of a company?
Results: Developed a bespoke GA that used computer vision to make spacial relationships between bar elements to develop shapes acceptable to the building codes while returning design with 10 %+ material savings when compared to structures designed by humans.
The final software was deployed in a structural design company and is now in use.