Keeping it real: When Generative Design works (Part 1)

Posted on May 2, 2019 by Franck Murat and Vincent Carignan
Category: Technical

Keeping it real: When Generative Design works (Part 1)

We’re all biased. We are biased simply saying we believe we are all biased!

It is completely understandable that as an industry, we are excited about the possibilities of generative design. But, the excitement surrounding generative design tools favours our biased interpretations of the potential and limits of the principle. Consider as proof that the very definition of generative design does not correspond to the reality of current solutions that are closer to optioneering or topology optimization without the autonomy associated with generative art for instance.

In this multi-part series, we will explore the nuances of generative design by attempting to document the obvious benefits, but equally the disadvantages and limitations. This article will focus on the benefits within a set of constraints. The good, if you will, in our series on the good, the bad and the ugly of generative design.

Knowing our own limits

The decision-making process can be limited by several factors on projects, including economic and temporal factors. If you ask a designer why they chose a particular system as part of a project, they might say that it was the best for this project ... but that’s because it was already proven as successful in part of a previous project. The construction industry is renowned for trying to create unique projects every time, so how can one solution be perfectly adapted to two completely unique projects?

It’s also hard to imagine, for example, a mechanical engineer who is able to efficiently to explore all the options that may be applicable, under a mandate with a defined budget, and to make a comprehensive comparative analysis. Similarly, for lack of resources, designers are often forced to reduce their analysis to a fraction of the possibilities available to them, often by exploring variations of the same concept. The selection of the solution will often be limited to the first solution that seems to correspond to the needs of the project, which is of course biased by the previous experience of the designers.

Enter Generative Design

Easier access to digital tools, however, allows designers to challenge traditional project boundaries and attempt to truly optimize their concepts. Optimization and generative tools increasingly being explored by practitioners, researchers, and programmers.

The principle is simple: provide an optimization algorithm with a series of constraints (generation or optimization for example) and produce a set quantity of options that respect these constraints. The designer can then navigate through the list of proposed options, weight them and analyze them to select one that best meets the requirements.

The efficiency of keeping it simple

The more precisely the problem is defined (clear constraints, a determined number of variables), the more interesting and effective this approach becomes . For example, space allocation optimization issues are ideal for this type of application, such as creating a grid for parking or allocating bin areas for a storage building.

Figure 1: Area allocation for planning lockers in a storage building. Source : testfit.io

WeWork is renowned for developing tools with this laser-focused philosophy. For example, their automatic desktop positioning tool generates possible options to configure a desktop, according to the constraints they defined. The designers still analyze, according to more subjective criteria or exceptional orders that could not be taken into account by the tool, the options created to select the most appropriate ones. Again, these nested tools can be very effective in contexts where problems are well defined with clear variables . They become part of the overall process. And like other possible automatisms, the more often the problem that we try to solve is presented to us, the more interesting it is to spend time trying to automate the resolution.

The design of MEP systems is also well-suited to this kind of problem-solving approach. The choice of the heating / cooling system supplying the distribution can be optimized using a tool for generating and analyzing options, if only to compare the different performance of the options (possibly over their entire life cycle). Tools of this kind have been around for some time, but it seems that their recent and wider dissemination follows the current leap in the digitization of the construction industry.

Will robots take my job?

Generative design, or the principles supporting it, cannot realistically replace designers. It is a tool, among others, which supports designers’ decisions in specific situations. Subjective analysis is still necessary in order to separate and weight the importance of the suggested concepts. However, it remains a powerful design tool when properly understood and controlled. Just as the production of multiple foam different models can do, it can feed a designer's creative process and / or document a multitude of potential concepts, but there’s still a lot of design work needed downstream to bring those concepts into reality.

Next up: exploring the limits of generative design

Overall, the possibilities offered by the partnership between the computing power of generative tools and human analysis are within the AEC industry's reach. The limits are still important, however, and identifying them is essential to any effort to integrate this kind of tool into its routine. The next post in this series will address these limitations. In the meantime, do not hesitate to contact us so that we can support you in your efforts to explore and exploit this technology.


Franck Murat
BIM/VDC Technical Director


Vincent Carignan
Chargé - Développement de marchés

By clicking Sign Up, I agree that I have read and accepted the Privacy Policy.