Engineering and form finding research often occur in parallel with architectural research. In our increasingly digital world, architects have been exploring how technologies such as modeling, scripting, and digital fabrication should affect our built environment. Architects such as Fabio Gramazio and Matthias Kohler, who have studied the effects of the digital world on the material world, observe that:
“…data and material can no longer be interpreted as a mere complement but rather as an inherent condition and thus an essential expression of architecture in the digital age. A digital materiality is emerging, where the interplay between data and material is seen then, in a new light, as an interdependent structuring of architecture and its material manifestations.” 
Furthermore, in their book Digital Materiality in Architecture (2008), they claim that:
“…digital orders intensify the particularities of material properties. Materials do not appear primarily as a texture or surface, but are exposed and experienced in their whole depth and plasticity.”
Inspired by the works of artists such as Hans Haacke (Blue Sail) and Shinji Ohmaki (Liminal Air Space-Time), Olek Niewiarowski of the Form Finding Lab explored these interactions between the digital and the physical by taking a seminar at the School of Architecture with Ryan Luke Johns.
Together with architecture students François Sabourin and Benjamin Vanmuysen, Olek created a physical set-up to explore an interactive “force finding” method with fabric under varying air pressures. The design agenda behind this was two-fold: the first part pertains to Gramazio and Kohler’s “sensuality of the digital” and sets to explore the possibility of enhancing perceptibility of a medium through digital intelligence. The group chose to work with air pressure, an invisible material, that when informed with digital order can be revealed and made plastic. The second intention is to explore an “analog” process of “inverse-form finding” or “force-finding,” where the objective is to use material actuation, sensing, and control to find the best fitting forces for a given shape not through analytical means, but through constant trial and error.
Keeping in mind this agenda, the group developed a prototype of control system that allows a suspended sheet to be manipulated using air pressure. The objective for this system was to reproduce positions as defined by a user through interaction with the sheet.
The system, integrated in a wooden frame, is composed of five elements: 1) a square piece of shear fabric, with corners constrained in the x-y directions and unconstrained in the z direction; 2) five 120 mm high-performance computer fans, fixed to the frame and located under the sheet; 3) a computer running a Processing script connected to 4) an Arduino Mega microcontroller and 5) a motion capture device (Kinect V2) situated over the sheet.
The five elements combine to form an information loop. The sheet is read by the motion capture device, which transmits data to the computer. The Processing script calculates fan speeds using proprietary algorithms and then transmits instructions to the Arduino board to modify those fan speeds. The information loop closes when the fans transmit their speed settings materially to the sheet via air pressure. This is a closed cycle until someone intervenes.
One of the project’s goals is to have direct tactile engagement with the material. The interaction with the user is one of indirect modeling. Rather than acting on a medium to produce a particular form, the user selects a moment to be reproduced by the system. The visual interface of the system gives the user access to the logic behind the fan control. The system displays an image of the sheet as seen by the motion capture device. It colors the sheet with a gradient from green to red, green indicating that the sheet is currently within a satisfying range of the saved sheet state and red indicating that it is either too low or too high compared to the saved sheet state.
The Processing script written for this system evaluates the heights of all points. For the purposes of this project, the data was pared down to the number of variables that could be controlled. Since the extent to which the sheet state can be controlled is related to the five different fan speeds, ultimately, only five variables were extracted from the point cloud. These five variables are the heights of the sheet directly above each fan.
In turn, these variables are used to control the fan speeds using different algorithms. Some of the algorithms used are illustrated graphically below.
This work effectively allows for indirect interaction between user and air pressure. While the floating sheet is the most perceptible element, the true medium of the project is the air affecting the sheet. When someone acts on the sheet and saves a sheet state as a target, she is effectively changing the air flow by initiating a force finding process.
This project embeds digital rules to control air pressure, enabling the system to transform the invisible material of air into an intensified medium that can be read and manually modeled. Thus, the hyper rationality of programming actually enhances the sensuality of a material. While most augmented materiality workflows have engaged with the extensive properties of visible materials, this project introduces a framework for workflows that deal with the intensive properties of invisible materials.
In addition, the project presented here experiments with what could be loosely termed inverse form-finding. In most form-finding processes, known forces are used to determine forms. However, the work presented here pursues an inverse process, by which known forms are used to determine forces, namely air pressures. In architecture, such inverse form-finding processes have not been as thoroughly investigated as their counterparts.
Such a process obviates analytical computation, and thus exemplifies what architects call “material computation.” For a given shape, one could use mathematical analysis or computer simulation to find the forces that might produce that shape. This project, however, uses simple rules to physically explore possibilities and slowly converge towards a solution that may reproduce the initial shape.
Author: Olek Niewiarowski, with François Saboruin and Benjamin Vanmuysen contributing