SP18 Generative Design - Week 7 - GD case studies

Editor's Notes

• #2: The Living – an applied research and design consulting group operating within the research organization of Autodesk
• #4: For this presentation, I’d like to focus on our use of a different kind of technology, which we call generative design. I’ll show you three projects where we used the generative design framework not only to design novel solutions for our clients, but also to help us focus in on the goals and problems our clients had.
• #5: The first project is the Bionic Partition. This was the result of a collaboration between us and the innovation team at Airbus. The goal of the collaboration was to explore how new design methods we were researching in our group could be combined with research at Airbus in additive manufacturing could be combined to create the next generation of components. As first experiment we chose the A320 partition.
• #23: New flagship office space for Autodesk in the MaRS discovery district in Toronto
• #24: Two floors and ground floor space in existing new construction. Accommodate many programs and create an urban presence for the company. Both traditional office programs (desks, conferences rooms, etc.) as well as new spaces such as workshops, VR prototyping space, and maker space.
• #25: Adapt the generative design methods we developed for bionic partition to solve architecture problem. The most difficult part here is the evaluation – straight forward for mechanical engineering problems but difficult for architecture – can architectural goals even be quantified?
• #26: Generative design workflow can be embedded in a larger set of technologies for architectural design – working with other groups to realize other pieces using the MaRS project as a prototype
• #27: Generative design does not replace traditional design process – embedded within it. Here we use our expertise and training as designers to figure out where to insert generative design.
• #28: Before generative design – need a concept. Learn from way organization takes place in cities as a result of both formal and informal pressures Adopt the idea of neighborhoods to organize clusters of teams and work areas within the office space.
• #29: Now need to create model which can generate different design solutions. First need to specify the constraints, for example the existing footprint.
• #30: As well as the program requirements and the people who will be using the space.
• #31: Next we create the parametric model or ‘design space’ which can create different design solutions.
• #32: Next part is the evaluation. Here we explored whether we could use generative design to capture some of the individual, qualitative goals of the work space, and to make the final design fit the individual desires and goals of the future occupants. First we used survey to gather data from individual employees and teams who would be moving into the space.
• #33: Then, working closely with stakeholders from the office, we designed six individual goals for the design.
• #34: We then developed custom models and algorithms to simulate different environmental phenomenon in the space and calculate a numerical measure for each goal. This would allow the computer to automatically evaluate the performance of each generated design according to the 6 goals.
• #40: The ran the genetic algorithm to find the set of optimal designs
• #42: Chosen design
• #43: Design refinement – traditional methods
• #44: Unexpected results
• #45: Under construction and post-occupancy
• #66: Van Wijnen – manufacturer of pre-fab homes. Have a lot of expertise in home construction and manufacturing methods. Partnered with us to develop high-tech methods for designing communities with the homes. Allowed us to apply our generative design methods to a larger scale. Same problem as with architecture – how to quantify what is important. Some easy (environmental, solar exposure) some difficult (felling of community, experience, etc.)
• #68: Constraints – site conditions
• #69: As before started with a concept – bringing the iconic street network organization from several Dutch cities to organize the small site.
• #70: Develop geometry system – evolution of model developed for MaRS – based on mesh subdivision
• #71: Placement of residential units on generated site.
• #72: Goals – developed through workshop with clients.
• #74: For each goal we can look at the range of values for different design options
• #79: Project still in progress, still developing the model and have not run the optimization yet. But as we develop the model we create and analyze design sets (called DOE) to see the range of possibilities of the model.
• #82: And look at good designs according to the goals, and see if they match our intuition for good urban design. This also helps us communicate with the client – use results to discuss what we mean by good design and adapt our model and goal metrics to help us produce better designs. Not technology for technology sake. Also not technology replacing design. Using emerging technology to augment the design process to allow us to create better designs. This actually is more demanding and requires more design expertise. The promise is better, more novel, and high performing design solutions beyond what humans could design alone.