Workshop 4: Design Optimization and Exploration

Workshop 4: Design Optimization and Exploration 2018-09-13T13:52:36+00:00


A single parametric model generates an infinite number of concrete designs. Which are “best”? Which are “good”? Which are “interesting”?

The tools we use to answer such questions are optimization and exploration. Optimization takes a model and varies its inputs to discover designs that improve or are “best” with respect to specified outputs. Exploration enables designers to move through a space of designs to find “good” or “interesting” ones. Optimization and exploration are complementary: both are needed in an effective design workflow.

This workshop focuses on two things. First, it focuses on how designers work with multi-criteria optimization. There is both art and science in setting up, running and interpreting optimizations. Workshop participants will gain hands-on experience with using an optimization engine within a parametric modeller. Second, it will demonstrate concepts and techniques for design exploration using an on-line gallery system that can, in principle, work with any parametric modeller.

The optimization part of the workshop will use Dynamo and its internal multi-criteria optimization engine. The exploration part will combine the research projects Fractal and Refinery from Autodesk and Design Gallery from Simon Fraser University.

The multi-criteria optimization we will use is based on the non-dominated sorting genetic algorithm, which produces a set of multiple incomparable design alternatives. This means that designers get more than one option and must explore to find the option(s) they prefer.

The exploration system provides a general approach to interaction with design alternatives. While extant parametric modelers theoretically define allow exploration of alternatives, their interfaces generally provide access to designs serially. Research has proved that designers prefer to work with alternatives, and such interfaces are not well suited to this activity. The goal is design interfaces to support alternative explorations in parallel, which is more aligned with designers’ creative processes.  to change this near-universal feature of parametric interfaces to support exploration using multiple alternatives. SFU researchers have built a prototype gallery system on a web browser that supports saving alternatives from three graph-based parametric modeling tools. Users can retrieve alternatives from the gallery, share them with others, and combine them to generate more alternatives.  Autodesk researchers have built Fractal and Refinery, systems for exploring spaces of designs generated by optimization and simulation systems.

To get the most out of the workshop, participants should have a basic to intermediate understanding of visual scripting including Dynamo. Bring your questions and your challenges. Participants have to bring their own Laptop and have Revit 2018.3 (or 2019), and Dynamo 2.1 for Revit or Dynamo Sandbox 2.1 installed prior to the Workshop.

– to access Autodesk Software for Students and Educators go to the DynamoBim project
– for Participants who want to take a look upfront into Dynamo you can access the Dynamo Primer

Required Infrastructure of the participants: Laptop with minimum CPU I5, Ram 8GB, comfortable to run Revit 2018.3 (or 2019) and Dynamo 2.1 for Revit or Dynamo Sandbox 2.1.

Expected Skill Level: Well-versed in either Grasshopper, Dynamo or Generative Components

Keywords: Architectural geometry, Divergence, Convergence, Design Exploration


Subhajit Das, Arefin Mohiuddin


Subhajit Das is a Ph.D. student in Computer Science at Georgia Tech, USA working at the intersection of Interactive Machine Learning, Optimizations and UI Technologies. His current research interests explore the realm of interactive systems helping non-data scientists to build machine learning models while doing exploratory data analysis. His recent works have been in the domain of building novel applications in data analytics implementing machine learning algorithms. With a diverse background in design and technology, Subhajit Das worked in the domain of computational geometry and design automation after completing MArch. in Design Computing from the University of Pennsylvania, USA. His work has been published in diverse platforms including KDD IDEA Workshop, IEEEVis, Ecaade and Acadia conferences.

Arefin Mohiuddin is a PhD student in the School of Interactive Arts & Technology at Simon Fraser University. He holds a Bachelor of Architecture from Jadavpur University, and has worked as a User Experience Engineer in a fast paced and innovative design start-up. His interest lies at the intersection of design and technology, and is passionate about building tools for designers. His research is on cognitive and HCI issues in computational design, and the design of tools to support creativity and augment the users’ design process using parametric design systems.