When you think of generative design, you probably envision this:
When Autodesk first developed generative design technology, it was used in the product design and manufacturing industry. It was a way for a designer to specify a set of requirements and have the software generate a set of designs that meet that criteria. The requirements often depicted a quest for optimization with a set of forces that define where an object needs to be strong and where it could get away with being lightweight. Though generative design is still used in this fashion, it's actually a multi-variable problem solver. Given a set of requirements, the variables, generative design computes various combinations and evaluates them regarding those variables. So it's no surprise that generative design has made the leap from product design and manufacturing to Architecture, Engineering, and Construction (AEC).
On May 30, I attended a webinar where this very topic was covered. As described in the webinar:
Generative design is the process of defining high-level goals and constraints and using the power of computation to automatically explore a wide design space and identify the best design options. This workshop will give participants hands-on experience in using the generative design workflow for space planning in architecture. The workshop will start by describing a computational model for space-planning based on subdivision and show how the model can be built in Autodesk Dynamo using existing functions and custom Python scripts. We will then describe the setup of several performance criteria including environmental and programmatic analysis. Once the model is set up, we will use an in-house optimization tool developed at The Living to automatically search the model for the best-performing design solutions. Finally, we will explore the data set of explored designs using a custom web-based data visualization platform.
Our instructors for the webinar were Danil Nagy and Lorenzo Villaggi, my colleagues from The Living that is also a part of the Office of the CTO. The Living, a first-of-its-kind Autodesk Studio, explores the future by building full-scale functioning prototypes today. Their projects apply generative design, biology, and new materials to real built projects in the context of technology, culture, and the environment.
I captured some screenshots during the webinar. Though I am oversimplifying it, the basis for generative design in space planning involves:
Design strategies and goals are expressed as simple functions, often using recursion, and object-oriented programming:
This programming can be done in Python whose integration is supported by Dynamo:
The requirements for a space impact how the space is subdivided, since there are many ways to subdivide a space. Furthermore, a space can be divided into multiple spaces, and then the resulting spaces can also be subdivided (recursion). Generative design allows space planners to explore thousands of possible subdivisions. For example, here is a space divided into 4 rooms, numbered 0 through 3:
The subdivision process can be modeled as: how many zones (i.e., rooms), if a space is split horizontally or vertically, and how much area each room should have:
Danil and Lorenzo ran through a test case using a typical floor plan:
For the sake of simplicity, they only modeled two requirements: desired aspect ratio of each room and desired adjacency of rooms (i.e., what rooms are next to each other). In practice, lots of requirements can be modeled and considered simultaneously:
The requirements were captured as simple data objects in Python code:
Using Dynamo and Python to subdivide the space recursively, floor plans are automatically generated, evaluated (i.e., scored), and evolved. Each solution is captured as an SVG file:
Based on their scores, the generated floor plans are graphed along an axis that shows how much the rooms match their desired aspect ratios and how adjacent the rooms are to their desired rooms. Blue dots depict floor plans that were generated early in the process. Red dots depict floor plans that were generated later in the process, i.e., refinements of the original plans generated. The other colors represent intermediate floor plans between those two ends:
The best floor plan (Design #14103) is in the lower left. The first floor plan generated (Design #171) is in the upper right. The floor planner can consider those two plans plus everything in between. Pink areas of a floor plan depict poor adjacency scores whereas white areas depict good adjacency scores.
Generative design allows the designer to consider more possibilities. The designer can select the best design instead of the first design that works. In space planning, an aspect like structural performance may be easy to calculate but is often less important than intangibles like the feeling or comfort level of a space. Generative design provides a framework to allow space planners to scientifically evaluate those elements.
Thanks, Danil and Lorenzo.
Based on this webinar, I was able to:
- Gather technical and theoretical knowledge about generative design and its application for architectural space planning.
- Get experience with computational design and modeling in Autodesk Dynamo.
- Learn about integrating Python scripts for custom computational design workflows in Dynamo.
- Obtain experience in performing architecture-specific simulation and analysis including environmental analysis and daylighting.
- Learn about the application of Evolutionary Algorithms for exploring computational design spaces and discovering optimal solutions based on a set of goals and constraints.
You can purchase a replay of the webinar for yourself.
|To access the recording:
The Living team members have applied generative design on a variety of real-world projects:
Autodesk makes software for people who make things. If you've ever driven a high-performance car, admired a towering skyscraper, used a smartphone, or watched a great film, chances are you've experienced what millions of Autodesk customers are doing with our software. Autodesk gives you the power to make anything.
Autodesk has always been an automation company, and today more than ever that means helping people make more things, better things, with less; more and better in terms of increasing efficiency, performance, quality, and innovation; less in terms of time, resources, and negative impacts (e.g., social, environmental). Generative design is a key element of our mission. We want to make the computer an assistant to the design process rather than a documentation tool. Generating designs and allowing a designer to pick one automates the design process itself. This can be done for a high-performance car, smartphone, or towering skyscraper. I guess that applying it to movie making is on the horizon.
Generative design is alive in the lab.