Computational Morphogenesis: Leveraging Proceduralism to Unlock Temporal Design
Interview with David Burpee
Ahead of his presentation at CDFAM NYC 2025, we spoke with David Burpee, Senior Computational Designer at Brooks Running and co-instructor at the University of Washington to discuss his work and what he will be discussing at the event.
In this interview, Burpee introduces his work in computational morphogenesis and how procedural simulation is being applied to footwear design, dynamic tooling, and emerging material systems.
He also reflects on his research into engineered living materials through the NSF-funded ELM program, and the broader implications of designing with processes and materials that change over time.
Could you start by introducing yourself, your design practice, and what you will be presenting at CDFAM on computational morphogenesis?
My name is David Burpee and I am currently a Senior Computational Designer on the Brooks Running Innovation Group.
For the past couple of years however I have had my own Computational Design practice where I have worked with brands inside a variety of different industries. I also co-teach two Master of Architecture classes at the University of Washington here in Seattle as part of an NSF Grant on Engineered Living Materials or ELMs.
My presentation at CDFAM will explore the usage of dynamic simulation as a design tool, spanning from my early interest in creative programming, through to more advanced utilization of procedural tools like Houdini to generate performance-driven tooling geometry in footwear applications. I will also touch on future explorations as part of my teaching and research through the ELM program and how these dynamic simulation environments can more closely model, simulate, and validate how these living materials can be used in the future.
You describe most current computational design outputs as static. How does incorporating temporal change into your workflow alter the design process compared to conventional methods, and what are the benefits?
At one point or another I think every Computational Designer has “felt the magic.” They build their first real script, graph, or program that solves a real problem, provides a real design solution.
Behind the scenes they move sliders, adjust variables, tweak graphs until everything is just right, and then they freeze the output and go present their design. But where does that magic go? Is it in the design, the process, or the script itself?
Leveraging temporal design provides a two-fold benefit: First, it externalizes that magical process that hooked so many of us on the flexibility and adaptability of Computational Design.
Secondly, working in this way provides the framework for designing with this new classification of materials that will continue to grow, change, and morph over time even after they are manufactured.
No longer do I need to explain the logic of a complex algorithm to a stakeholder, describe the variability of outputs to a manager, or elucidate the boundary conditions of a design system to an engineer. If someone can literally watch a design grow, change, and morph over time, the system is self-evident.
What software platforms and procedural modeling techniques are you using to integrate these time-based variables into product design?
A lot of my research over the past few years has been advancing my ability to design physical products in a software that was never intended to do so. I deliver more and more of my projects using a software called Houdini, originally a procedural software environment mostly used for film and VFX.
I am not the first nor the only person to explore this path, however I do believe that I have one of the larger catalogs of products in the market (or soon to be in the market) that were designed, modeled and simulated in Houdini.
Within Houdini I build a lot of custom tools and setups that are more geared towards physical products compared to something that is purely digital. I have created automated geometry processing pipelines, procedural lattice generation tools, a variety of custom particle and point simulation systems written in a programming language called Vex, and a host of other smaller helper functions for analyzing parts, dimensional tolerances, and other things.
A lot of my own interest lies within natural systems, and using algorithms that mimic or model certain behaviors observed in nature. Occasionally someone has already ported an algorithm or simulation technique over to Houdini, but I also build my own setups from referencing an academic research paper or other external reference. Even in the former case, the bigger challenge tends to be the actual application of the algorithm that can meet the limitations of physical product, such that all design outputs can be valid with respect to dimensional tolerances, manufacturing constraints like draft angle, pull direction, beam thickness, and material properties.
Given the variable nature of material properties in 3D-printed parts, and the difficulty in validating them fully, how can we approach testing and validation for Engineered Living Materials whose performance changes over time?
As I am sure many of us are intimately familiar, even validating a static 3D-printed lattice structure or complex physical part can be fairly complicated.
The complexity gets an order of magnitude greater when we are now expecting the material properties to change, sometimes drastically over time. For instance, dimensionally the difference between a hydrated and non-hydrated state of a living material can be greater than 100%. For this complex problem space it is my belief that a hierarchical approach is needed, analogous to the atomic model.
There is fundamental research (the atom level) that is being done and will continue to be done in order to build knowledge with primary behaviors of Engineered Living Materials.
Building upon this research is prototypical work that might combine a few behaviors to create a more complex system (the molecule level) that can be validated experimentally in the lab using physical testing, observation and measurement. Computational algorithms and other advanced computing techniques can be used to tune and adjust the system based on these results. This is where I believe the bulk of the work to be done lies.
Further in the future with any luck we can begin to design and build at the compound level, that is commercialized and scalable design for solving specific problems with engineering, design and manufacturing.
What do you believe to be the biggest technical challenges in modeling and simulating dynamic materials, and how might you speculate on addressing them in the future to move closer to commercialization?
It is important to note that I am still fairly early in the research, and some of this is still highly speculative. That being said, one of the greatest challenges is that there is really no tool or even model that exists to combine the various domains of knowledge and application needed in order to design this way.
Despite my belief that tools like Houdini are a powerful and dynamic environment that is more suited to this type of work versus say a typical CAD software, it still has its limitations, primarily in scope.
There is no hardcore FEA or engineering validation for instance. Even if there was, it is likely that closing the loop to validate these systems would rely heavily on physical prototyping and validation. There are perhaps some techniques leveraging things like computer vision or machine learning that can assist in wrangling the complexity of this problem but fundamentally I believe the solution still lies in a research-driven approach.
An example is that I was researching how to model capillary action through complex structures consisting of micro-fluidic channels. There are equations such as Jurin’s law and Washburn’s equation that describe this phenomenon, so at first glance it might seem like building a programming solution towards modeling even a simplified version of the behavior is relatively feasible.
Digging deeper though you discover that it requires a simulation that is “multi-physics” or requires multiple physical modeling computing methods to compute, and is impacted by gravity, temperature, atmospheric pressure, surface tension of a liquid, smoothness of the internal surface, and other variables.
Without even getting into the feasibility of printing complex capillary structures which is its own bag of worms. All of a sudden a fundamental building block towards modeling living behaviors is actually a very sticky problem. Certainly outside of the scope of something I can tackle in a few, free weekends.
…a fundamental building block towards modeling living behaviors is actually a very sticky problem. Certainly outside of the scope of something I can tackle in a few, free weekends.
As such we have to be specific in our scope and what we tackle, and be strategic about how we identify problems that are both high value and also technically feasible given the current paradigms of digital tools.
Luckily there is a large body of fundamental research that is ongoing at UW and other universities, and there are still many areas to explore given the early nature of commercialized development in this space.
What do you hope to share with, and learn from, the CDFAM community about designing for products and materials that evolve over time?
Firstly I am just looking to share my approach to Computational Design, and how integrating procedural tools has allowed for a more dynamic design environment. Particularly in connecting simulated or dynamic behaviors from design, visualization, and validation through to production and manufacturing.
Second I want more people to be aware of this material classification because I frankly think they are not only fascinating and beautiful, but also potentially incredibly powerful.
We talk a lot about architected or programmable materials in this space, but again, most of that is in reference to a material that is ultimately static. We have an opportunity to now quite literally expand another dimension of programmable materials to incorporate dynamic behaviors, and I think the potential for application in industry is extremely high.
I am hoping to also learn how others might be tackling the challenge of validating dynamic systems that change over time, because it is a very non-trivial problem space.
To learn more about David’s perspective of exploring and exposing the fourth dimension, and to connect with others advancing approaches computational design, AI, and machine learning in engineering and architecture, join us at CDFAM NYC 2025, October 29–30 at Newlab in Brooklyn.
