While looking at the issue of generative design, I came across a post from Chad Jackson on his site, Liefcycle Insights. He discusses generative design technology with Autodesk in his post. He also expands his discussion to look at the business opportunities, the benefits to design, and how he feels the technology will impact engineering.
Mechanical design is ripe for automation and design optimization - two key benefits of a generative design approach
Please take a look at Chad's post. He walks through the basic idea behind generative design and he also talks about the many precedents where software performs complicated design processes today - such as in ASIC design. Finally, he produces his own analysis of the technology and it's well worth a look
Mechanical design is ripe for automation and design optimization - two key benefits of a generative design approach. We already have simulation tools which provide design exploration features and provide optimized results to choose from. Rules-based approaches also apply to Design for Manufacturing tools like those from Geometric.
What lies behind generative design are algorithms and computing power. Basically, engineers determine the design objectives and set the constraining conditions. Then, as with simulation and design exploration, algorithms and processing power are thrown at the problem. The engineer can then select among the design choices provided.
The joy in this process for designers includes removing / automating much of the grunt-work. Theoretically, engineers will be able to expend their efforts at a higher level and better leverage their expertise. Chad Jackson envisions mechanical engineers having multiple design "agents" running simultaneously - the engineer focusing his time on objectives and constraints for his design projects. And he sees this added efficiency as a boon to overloaded engineers.
From a business perspective, Chad notes that generative design tools will help companies move beyond the 'first feasible' approach. I agree completely that the combination of automating complex design processes combined with constraint-based optimizations will allow an engineer to produce better, more efficient designs.
Hod Lipson, professor at Cornell and Columbia Universities, has actively researched this area for years. He states that his research focuses on two challenges in engineering: can we design machines that can design other machines and can we make machines that can make other machines? His book, 'Fabricated, the New World of 3D Printing' was released in 2012 by Wiley Press. While machines designing machines is not a reality today, the generative design techniques needed for this are finding their way into commercial products today such as Autodesk Within.
Optimization software used for design exploration and simulation can be applied to problems in architecture.
Generative Design is not limited to mechanical fields. There is interest in architecture as well. Michael Bergin at the Architecture Research Lab in Berkely discussed the issues around optimizing architectural design. Interestingly, he notes examples where simulation tools for design exploration and optimization have been applied to architecture.
The organic nature of generative designs seems natural to me. Nature is our most prolific teacher and researchers such as Lipson and many others recognize this. The fuel for future algorithms and optimizations in generative design are surely to come out of further research on how nature createsmany of the wonders around us. Just as tools for design optimization from mechanical simulation have been applied in fields such as architecture, it seems clear that advances in algorithms for generative design will cross-populate and enrich the automation and optimization of designing our world across a range of products large and small be they manufactured or be they built.