Modelling in parallel: from physical to digital and back again


When Martin Melioranski from 3D Lab asked me to conduct a parametric design workshop to 4th year architecture students at the Estonian Academy of Arts, we decided to change the usual approach of teaching parametrics using solely computerised tools. Instead, we wanted to introduce physical prototyping in parallel to digital modelling and asked participating students to build several paper and cardboard prototypes in parallel to using Grasshopper in Rhino 5. Our objective was to achieve a smooth two-way transition between the two modelling methods and use both the digital and the analogue computation in order to marry form-finding with fabrication.

All too often digital modelling precedes physical prototyping and fabrication constraints are typically considered late in the process making the whole design and production cycle to stretch uncontrollably. We really wanted to avoid this scenario since we only had one week at our disposal. Add to this that most of students did not have any prior experience with the software, we could not really afford to waste any time learning all the lessons the hard way.

After a short introduction to Grasshopper, students were asked to sketch their ideas and build the first cardboard or paper prototype before moving into the realm of parametric modelling. The first concept model was supposed to serve as a reference for creating the digital one. And it did indeed help the students to materialise their ideas early on and circumvent some typical pitfalls. From this point on, students were encouraged to continuously test the whole process and build physical prototypes whenever they made a radical change in their parametric model.

The biggest group of students chose to build a model of a gridshell based on an irregular pentagonal footprint (see the Slideshare of their presentation below). The gridshell was conceived as a minimal surface stretched over five arches drawn on the polygon edges and the structural laths were created to follow geodesic lines on this surface between each corner and the division points on the edge arches. Their digital modelling approach appeared to be fairly straightforward, but the hardest part was assembling the whole structure. At some point it even looked like the whole model was going to fail. But in the end it was only the edge arches that remained slightly deformed and everything else fitted nicely together. It proved that cardboard is not the best material for building gridshells as some of the geodesic lines on double-curved surfaces bend in two directions.

The second group of students looked into modelling an asymmetrical tower structure built on a triangular footprint. Their initial paper prototypes focused on finding the right type of joints between the folded structural elements on each side. Constantly moving between the physical and digital they managed to find an elegant solution where each side of the tower were made of only two folded elements.  Unfortunately they found out that the structure was not stable enough to be built out of paper. They switched to cardboard but then the folds started to tear under the weight of material. Eventually they fixed the torn folds with additional paper patches and managed to find a stable configuration.

The smallest group of students of only two members created a paper model of arched canopy. After a short period of experimentation they found a structure that was sufficiently strong while the parametric model remained fairly flexible. The most surprising effect was the emergence of ‘gothic’ arches – the paper prototype seemed to naturally form a fold at the highest points of the arches creating an unwanted yet great-looking sharp apices.

Despite the tight time frame all groups succeeded to produce cardboard and paper models of non-standard structure and geometry. I can only conclude that the learning curve seemed to be of the right steepness. My personal feeling is that the results could not have been achieved if the modelling process had not been done in parallel. Naturally, mistakes did happen but most of these were discovered and fixed on the spot. The adopted approach forced students repeatedly go through the whole process from design to fabrication and helped avoiding the common mistake of over-polishing some aspects of the design and ignoring others.

The parallel modelling approach seemed to be particularly suitable for producing results within a short span of time. While the digital parametric models often suffer from the lack of real-world constraints, the physical prototyping serves as a testing method for the manufacturability and stability of the structure. But physical prototyping alone is limited in terms of manipulating the form and perhaps requires too much effort for creative exploration.

Students who participated in the workshop:
Riin Kersalu, Gert Guriev, Kerstin Kivila, Taavi Lõoke, Liisi Aomets, Silver Liiberg, Kersti Miller, Marilin Tuulemets, Sven Luik, Suzanne Marrier.


Grasshopper, Parametric design, Workshop

Leave a Reply