Rob|Arch 2018 Workshop 6: Sub-additive 3D printing of optimized double curved concrete lattice thin shell structures
Sasa Zivkovic (Cornell University, Assistant Professor), Martin Fields Miller (Cornell University, Visiting Critic), Christopher A. Battaglia (Ball State University, 2018–19 Design Innovation Fellow, Assistant Teaching Professor)
This year as part of Rob|Arch2018 I had the privilege of participating in a workshop exploring possibilities in printing concrete at full scale using non linear (not 2D layered) 3D printed tool-paths, developed at RCL, the Cornell Robotic Construction Laboratory
BVN have been exploring and adopting Additive Manufacturing technologies for a number of years now, both in the form of traditional 3D printing as well as coreless robotic deposition of carbon fibre tow in collaboration with The University of Sydney. This workshop presented an excellent opportunity to get hands on experience both in robotics 3D printing as well as using concrete as a medium.
To quote from the workshop outline:
[The] workshop investigates new means of architectural scale concrete 3D printing to fabricate robust doubly curved geometries and construct structurally optimized thin shell structures. Sub-Additive Manufacturing explores a three dimensional deposition of concrete printing material on a mechanically shaped substructure of reusable aggregate. Concrete material is extruded with a robotic arm using three dimensionally curved deposition paths on top of the shaped gravel. By printing in three dimensions, the time consuming process of corbeling is replaced by spatial concrete arching. Creating curvature in concrete with this method increases the speed over typical 3D printing practices, can produce double curved geometries which are structurally optimized, does not require form-work during assembly of shell components, and eliminates waste material. From nozzle design and material feedback to digital form finding processes and implementation of advanced fabrication technologies, this workshop introduces a series of key concepts and methods radically advancing concrete 3D printing at full scale.
Day 1 Workshop
After a brief introductory talk, we formed into groups, (My group consisted of Elisa Lublasser, Nicole Yim, Scheller Gabor and me) and were assigned an initial provocation — our group was to design a cantilever using the process.
We started with hands on exercises using fine aggregate and hot glue, to both understand how the tooling paths necessary to us a custom end effector (manufactured from OSB) to shape a base of aggregate, as well as understand the print path required to deposit concrete.
We then moved on to create a model in Rhino of our proposed bed of aggregate and an approximate toolpath to simulate and test on the Kr120 — R2900 which was set up in the workshop space.
Repose of aggregate (45 degrees) was considered, as well as the immediate context of the work (we had to make sure we didn’t clash with other groups), and reach of the robot (limited to 2900mm).
Day 2
Day 2 began with a lot of shoveling! We crafted a pile of rough aggregate (pumice stone) approximately to the dimensions of the finished piece.
After some quick edits to the shaping path we begun the passes to shape the base form-work. The initial roughing passes (3 in total) were executed at a 50mm stepover.
After the form-work was roughly sculpted, a thick layer of granite aggregate was laid over the top, and 2 finishing passes were executed (alternating paths sides with a 30mm stepover)
The major challenge was ensuring that we got adequate ‘scoop’ to ensure that the depression we created matched our digital model (reasonably closely).
This being complete, it was time to fit the print nozzle and mix the concrete!
The concrete was mixed to the below recipe. I was most excited to see this, as really, getting the mix right is one of the most difficult things to do!
1 bag Portland Cement
1 bag Sand
1 bag Mason’s Mix
10g Nylon Fibre (to add rigidity)
Superplasticiser and Retarder (to give a working time of 30–40 minutes)
Once the concrete was mixed we were ready to print!
To execute the print smoothly we had to ensure that we were interpolating at a constant velocity of 90m/s (as opposed the the default interpolation by distance when linear moves are generated). This ensured that the movement of the nozzle over the piece remained consistent. Each layer was 10mm deep
Day 3
Day 3 was spent iterating. Now being familiar with the basics of the system the group was able to were able to revise and optimise toolpaths for a slightly slimmer extrusion nozzle and a new mix of concrete:
1 bag Portland Cement
1 bag Sand
10g Nylon Fibre (to add rigidity)
Superplasticiser and Retarder (to give a working time of 30–40 minutes)
Masons mix was eliminated, giving a longer open time to the concrete at the sacrifice of curing time — these new prints would take longer to cure, but as these were the finals we were less concerned by that.
The above prints were executed, with all groups demonstrating an improvement in control and print quality.
At the end of the day, a larger piece was also executed demonstrating an optimized model which was generated to reflect force lines in a doubly curved surface, bringing a further level of development to the print pathing.
Reflections
The workshop presented an excellent opportunity to get a good hands on feel for the process of designing and robotically executing 3D prints in concrete.
I was struck immediately with the reality of the material. This is concrete guys! Work is messy, reasonably uncontrolled and always constrained both by the working time of concrete, and the variability which comes with every mix. Scales are big. Work is fast. Everything is heavy. There’s a risk of blowout from the pumping that needs to be monitored.
This has an upside — the setup is reasonably straightforward. End effectors are simple to produce from found objects and standard industrial parts, characterized more by their strength than sophistication in design, meaning that this setup would be pretty straightforward to replicate in any decent workshop.
More work is required. These prints are currently un-reinforced, which is required both to resist tension but also to achieve good bonding between layers. This future work is outlined the workshop leaders paper presented at the conference entitled ‘Sub-Additive 3D Printing of Optimised Double Curved Lattice Structures’.
This work can be seen in context with the work being done by NCCR Digital Fabrication and ETH Zurich’s Dfabhaus at NEST pictured right. This structurally optimized post tensioned slab, executed on 3D printed sand form-work clearly suggests that a rethink of how we execute form-work is required if we want to be able to execute designs which use less concrete in a more efficient manner.
The system developed by the Workshop 6 leaders shows promise in this sense. Concrete is able to be deposited quickly where it is needed, on a re-usable bed, of an affordable, scalable yet controllable nature, which shows a promising economy.
Overall, a fantastic experience! Many thanks to the Workshop leaders, the workshop team-mates, as well as to BVN for providing the opportunity for this experience.
For more images from Rob|Arch 2018 as well as insights into some of BVN’s collaborative explorations into digital fabrication, additive manufacturing and robotics, follow our Instagram account or visit our website.
