by Krishna Rao

This series, which acknowledges the importance of computation and technology in today’s practice, will explore the architectural opportunities afforded by the concepts and strategies of Generative Design (also known as Computation Design or Parametric Design) followed by those of Performative Design. Architectural design has gone from hand drawing to 2D CAD to explicit 3D modeling, and of course, to BIM. Generated digital content is all around us, and it seems that designers are starting to realize the potential power of generated forms as opposed to explicitly modeled forms as evidenced by the complex forms of many new projects in the Middle East and Asia in particular.

I will post a few articles on Generative Design before getting into Performative Design where I will more fully describe the nuances and advantages of each.

Why am I writing about Generative Design? Processes like Generative Design (GD) and Performative Design (PD) affect the earliest stages of a design effort. Both GD and PD are paradigm shifts compared to more traditional, normative design processes. I believe that it is important that we, as a profession, explore and understand GD, PD and BIM in order to get a complete picture of the digital design ecosystem.

If you are still reading this entry you might be interested in some definitions. 

Generative Design
There isn’t a definition that really captures the essence of GD, so I chose to use this quote “Generative design is not about designing a building, It’s about designing the system that designs a building” from Lars Hesselgren, one of the founding members of the Smart Geometry Group, a non-profit organization of AEC professionals interested in using computational and parametric approaches to design that encourages collaboration between practice, academia, and research. 

Performative Design
According Dr. Andrew Marsh, creator of the widely used Building Performance Analysis application EcoTect, Performative Design basically considers the wide array of building performance issues simultaneously with all other aspects that shape a design from the earliest, most formative conceptual stages. It does not, as the name might suggest, myopically focus on meeting performance criteria through the manipulation of form. Instead, it requires synthesizing performance and form when the design concept is still sufficiently plastic and pliable so it can be shaped by these considerations as much as possible. Performative Design processes are increasingly viable because the tools for simulating and analysing buildings are becoming increasingly fast, flexible, and simpler to use. This means that designers can use these programs themselves, instead of relying on specialists, to apply even the simplest block models. This, in turn, allows them to effectively integrate their results into the early designs and pursue a whole range of design ideas, rigorously test and accept or reject them very early in a project.

In other words, think of using Ecotect to analyze the heat gain on the façade of a building. That is an example of performative design.

Features
To further illustrate the point; see image below. Let me try and explain the process behind this image.

Generative Design is significantly different because you do not begin by creating a form; you start by defining the rules that shape a form or even by defining an individual component of the eventual form. The process might be something like:

• Define component
• Capture relationships
• Define constraints
• Generate form
• Iterate by varying any of the first three inputs

The defining component can be a simple geometry, but the resulting form can be many times more powerful and complex, for example the complex structures found in ant hills and bee hives result from simple repeating components. By harnessing the computational prowess of generative algorithms we can design and rationalize forms that human minds typically cannot conjure, Gaudi and Nervi as examples of exceptions. Despite the complexity of the resulting forms, design iteration is quick and the feedback is instantaneous. This allows designers to test a number of “what-if” scenarios rapidly and removes the time limitation of explicit modeling and accelerates interaction with resulting form.

Examples
A complex project produced by the GD process is Herzog and de Meuron’s widely-known Beijing National Stadium (Bird’s Nest). I can imagine that the project’s interwoven structural system may have been so challenging and time consuming to model explicitly that it might not have been feasible using traditional 3D computer modeling. Instead, a generative design solution may have enabled the designers to refine their solution by iterating multiple versions, at least more efficiently.

At FXFOWLE, we have used the same process on one of our Middle East projects for the first time. In this case, the façade was a saw-tooth module arrayed along multiple curves while stepping in and out at different floors. By using generative design ideas, we were able to study slightly different versions of the façade in a short amount of time. I will explain this process in detail in a subsequent post. The saw-tooth module was the individual component and had a certain relationship to the curves along the façade; the number of panels and their justification and height along each curve was constrained. This resulted in a form that was iterated by varying the (a) module, (b) relationship to the curve, and (c) constraints along the curve. See images below.

Tools
Some of the software that can be used in a GD process are Mc Neel’s Grasshopper (a Rhino plug-in), Generative Components by Bentley (a Microstation plug-in) & Revit’s Conceptual Massing environment. Apart from these tools, scripting routines such as VB, C#, and Python can be used independently or in conjunction with the above software. Other tools are available, but for the most part I will focus on Grasshopper and Revit since these are within my expertise at FXFOWLE.

I would like to end this post with a quote from Henry Ford “If I’d asked my customers what they wanted, they’d have said a faster horse.”

CAD = faster horse!