Andrew Delle Bovi (website), Grant Herron (website), Mark Knutson (website), John LarmorShan Sutherland
University of Michigan, Taubman School of Architecture and Urban Planning
Thesis Project – Novel Constructions
Faculty: Glenn Wilcox (website), Wesley McGee (website)


While civilization has extracted resources from the earth for millennia, much of the energy expended in acquiring these materials still remains embedded therein. Of these materials, aluminum, now ubiquitous, was once more rare and valuable than gold. Because aluminum can be cast at relatively low temperatures, using simple equipment and inexpensive mold-making materials, it is an ideal material for a small scale industrial process where no hierarchical separation exists between designer, craftsman, and producer. Additionally, recasting existing metal requires only 3% of the energy used in smelting, and can be executed at a small scale in the likes of atelier studios and design schools.

With a self-constructed foundry, we explored a design process highly dependent on tangible physical restraints, without utterly eschewing digital influences. The range of physical processes involved in the creation of a single aluminum artifact includes: wood frame constructs, sand cast molds, robotic burnishing, robotic scratching/milling, pouring molten aluminum, and post-production work. By employing the 7-axis robot at the intersection between primordial elements (sand and clay) and cutting-edge software, we navigated the gap between the digital and the analog, while still allowing for creative expression.

Following 2 months of research, a global form was generated in Rhino with considerations made based on our findings: the physical limitations of our team (ability to lift x lbs. etc.), economical limits, time, etc… The form needed to be structurally dynamic, formally interesting, logistically challenging and pose broad material demands—requiring the aluminum to perform as both a linear and surficial element. The aluminum components would be leveraged to perform as efficient structural members while simultaneously providing a medium for emergent aesthetic expression and calculated process failure.

A fluted, flaring column was selected. The form was analyzed and sectioned to achieve efficiencies in production and assembly (max part size=3’x4’x2′ or roughly a 400lb. mold). Surfaces were modeled in Rhino via Grasshopper, and corresponding burnish toolpaths were generated using a second set of Grasshopper scripts. The digital surface geometry was translated into the sand mold via a specially developed burnishing operation utilizing a robot and custom made tooling. A python-based, algorithmic code (in continued development at the University of Michigan) was used in translating toolpaths from rhino to the .src code used by the robots. Processing scripts were adapted to generate tool paths along a generative fractal geometry for the etching and milling operations.

Of these, primary tool paths were selected and optimized in Rhino via Grasshopper to define the boundaries of each part. Another Grasshopper script was developed to translate these linear toolpaths into ordered-reference planes that would dictate the complex, interlocking edges of each piece. Secondary tool paths were identified and optimized based on the technical behavior of molten aluminum within a sand mold. These are milled into each mold using a robot and specialized tooling, and form the distribution network for distributing aluminum throughout each part (sprue system). Tertiary tool paths are selected based on aesthetics and continuity and etched into each mold utilizing a robot and specialized tooling. These fine geometries are intentionally machined at the limit of the sand mold to perform optimally.

The calculated failure at this level will produce the pseudo-organic character of the artifact. The final form shown in the photographs is comprised of 12 parts at a 15% failure rate. It was completed in 10 days and assembled in 11 hours. It is 14′-0″ tall and weighs only 80lbs. The project was successful in producing an artifact fabricated with high degrees of precision, allowing for the rapid assembly of an extremely complex form, while emerging from that same fabrication process with an entirely unexpected, and unreproducible, aesthetic.

Our research is intended to serve as a precursor to future creative manufacturing processes. Where others might intervene in the age-old process of sand casting to increase its efficiency, reliability, or economy, our intent was to produce novel effects. As we reconciled such paradoxes as the aesthetically perfect vs. imperfect, and the designed vs. the emergent, we came upon some truly unique results. Such discoveries suggest a new possible role of the architect/designer: that of the “digital craftsman” as well as the “chaperone of natural processes.”


↑ Robot burnishes a surface into a green sand mold ↑ Mixing sodium silicate sand ↑ Using C02 to cure the sodium silicate sand ↑ Calibrating a custom tool head for the KUKA robot ↑ Mounting the tool for use ↑ Operating the KUKA robot ↑ KUKA performs swarf cuts at 1:1 scale ↑ Close up of swarf cut maneuvers ↑ Firing up the foundry ↑ Charging the crucible ↑ Removing the molten aluminum ↑ Pouring the molten aluminum ↑ Pouring the molten aluminum ↑ Pour site ↑ Conceptual study of entire column ↑ Early texture and swarf study ↑ Close up of burnishing pattern in cast aluminum and imperfections in the pour ↑ Four detail studies of the final piece ↑ The final twisting column. 1/3 of the conceptual piece was realized

Leave a Reply