Content uploaded by Zach Seibold
Author content
All content in this area was uploaded by Zach Seibold on Dec 13, 2018
Content may be subject to copyright.
JOURNAL OF
THE INTERNATIONAL ASSOCIATION
FOR SHELL AND SPATIAL
STRUCTURES
FORMERLY BULLETIN OF THE INTERNATIONAL ASSOCIATION FOR SHELL AND SPATIAL STRUCTURES
P
r
o
f. D. h-C Eng .E. TORROJA, founde
r
international association
for shell and spatial structures
PORTADA Y CONTRA.indd 1 28/12/04 07:06:0928/12/04 07:06:09
V
ol. 55
(
2014
)
No. 4
D
ece
m
be
r
n
. 1
82
I
SS
N: 1028-365X
December n. 182
Vol. 55 (2014) No. 4
CUB journal 182.ai 1 07/01/15 16:51CUB journal 182.ai 1 07/01/15 16:51
VOL. 55 (2014) No. 4 contents
n. 182 December
Annual Letter from the President
R. Motro
Memorial Statement
195
Memorial to
Vladimir Shugaev
J. Abel, P. Eremeev and S. Pellegrino
Tsuboi Proceedings Award Paper for 2013
199
Parapluie - Ultra Thin Concrete Shell Made of UHPC
by Activating Membrane Effects
P. Eisenbach, R. Vasudevan, M. Grohmann, K. Bollinger and S. Hauser
201
Hangai Prize Papers for 2014
Form Finding and Optimization: Getting the Best of Both Worlds
to Design Lightweight Structures
B. Descamps and R.F. Coelho
213
Upcoming Events 222
F
ormulas for the
D
erivation of
N
ode
C
oordinates of
A
nnular
C
rossed
Cable-Truss Structure in a Pre-Stressed State
R.-J. Liu, S.-D. Xue, G.-J. Sun and X.-Y. Li
223
Critical
B
uckling
L
oad and
N
onlinear
P
ost
-
B
uckling
R
esponse
of Guyed Towers
G. Margariti and C.J. Gantes
229
Reviewers of Papers 236
Robotic Fabrication of Components for Ceramic Shell Structures
Z. Seibold, M. Singh, L.-L. Tseng and Y. Wang
237
Technical Papers
On the Robustness of Cable Supported Structures, a Theoretical
and Experimental Study
H. Wu, J. Ye, B.-Q. Gao, Y.-L. Shan and C. Zhang
243
COVER: Figure from paper by B. Descamps and R.F. Coelho
IASS Secretariat: CEDEX-Laboratorio Central de Estructuras y Materiales
Alfonso XII, 3; 28014 Madrid, Spain
Tel: 34 91 3357409; Fax: 34 91 3357422; iass@cedex.es; http://www.iass-structures.org
SODEGRAF ISSN: 1028-365X Depósito legal: M. 1444-1960
Journal
JOURNAL OF THE INTERNATIONAL ASSOCIATION FOR SHELL AND SPATIAL STRUCTURES: J. IASS
Copyright © 2014 by Zach Seibold, Malika Singh, Ling-Li Tseng, Yingyi Wang.
Published by the International Association for Shell and Spatial Structures (IASS) with permission.
237
ROBOTIC FABRICATION OF COMPONENTS FOR CERAMIC
SHELL STRUCTURES
ZACH SEIBOLD
1
, MALIKA SINGH
2
, LING-LI TSENG
3
, YINGYI WANG
4
Master of Design Studies Candidates
Harvard University Graduate School of Design
48 Quincy Street, Cambridge, MA 02138
1
zseibold@gsd.harvard.edu,
2
mdsingh@gsd.harvard.edu,
3
ltseng@gsd.harvard.edu,
4
ywang@gsd.harvard,edu
Editor’s Note: The first author of this paper is one of the four winners of the 2014 Hangai Prize, awarded for outstanding
papers that are submitted for presentation and publication at the annual IASS Symposium by younger members of the
Association (under 30 years old). It is re-published here with permission of the editors of the proceedings of the IASS-SLTE
2014 Symposium: “Shells, Membranes and Spatial Structures: Footprints” held in September 2014 in Brasilia, Brazil.
ABSTRACT
This research investigates the assembly of funicular shell structures using a single layer of flat ceramic tiles.
The objective is to synthesize recent advances in structural prediction software with existing means and
methods of on-site assembly. The primary area of investigation is at the scale of the tectonic unit - most
specifically how introduction of geometric intelligence at the scale of the unit can simplify the assembly of
forms that are difficult to realize in the context of modern construction. The project simulates an industrial
production scenario in which components for a given shell structure can be fabricated using a wire cutter-
equipped 6-axis robotic arm. It aims to increase the adaptability and applicability of ceramic shell
structures.
Keywords: Robotics, Fabrication, Ceramics, Construction Systems
1. INTRODUCTION
Recent advances in structural prediction software
have facilitated the design of increasingly free-form
funicular structures. While it is possible to realize
these complex forms using standardized modules
and traditional construction techniques, the process
is not without its drawbacks. Variation within
masonry vaulting systems usually requires variation
in assembly tolerances between standardized
modules, or the on-site modification of modules.
The need for a skilled labor force, multiple
overlapping layers of masonry units, and the
implementation of expensive temporary formwork
also contributes to making the realization of these
forms technically difficult and cost-prohibitive in
the context of the modern construction industry.
Reducing these complexities could increase the
popularity and applicability of this building system.
Simultaneously, the development of robotic
fabrication systems for ceramic materials has
permitted mass-customization at the scale of the
ceramic module. An opportunity exists to shift the
moment at which shell structures are resolved into
individual components away from the building site
to a controlled industrial production environment.
2. RESEARCH CONTEXT
The construction technique developed by Rafael
Guastavino eliminated much of the formwork
historically required by masonry vaulting systems.
The material innovations of Eladio Dieste
“transformed our perception of brick as a traditional
material associated with heavy, vertical
construction elements into one that allows for
extreme thinness and long spans” (Bechthold [4]).
The hanging-cloth models and full-scale
constructions of Heniz Isler hinted at an infinite
number of formal possibilities for compression-only
shell structures (Chilton [8]). It is now possible to
fully describe, model and understand the geometry
of free-form funicular structures. While advances
in software have facilitated the design of
increasingly complex, irregular forms, the
techniques used in their fabrication and assembly
have not made equal gains. It remains difficult to
resolve complex curvature into construction
elements.
Vol. 55 (2014) No. 4 December n. 182
238
The following research is situated within a context
of several other projects that explore the
relationship between free-form compression only
structures and construction technique. Catalan
vaulting – also known as Timbrel or Guastavino
vaulting – is the technique that many of these
projects seek to improve. While this technique is
formally flexible and possesses a high strength to
weight ratio, it also requires multiple layers of tiles
adhered by a bonding agent, elaborate
formwork/guidework and skilled labor. Our
research found that recent projects that have
explored this area of study have resolved the
challenges of constructing non-developable surfaces
with two distinct approaches.
2.1. Novel Processes
One approach has been to design novel assembly
processes using existing building materials and
methods. In these projects, surface curvature is
resolved by varying the assembly tolerances
between modules. Shear resistance is
accommodated for using the traditional Catalan
method: adhering two or more layers of tiles laid at
differing orientations with a bonding agent.
Standardized modules are cut on site to
accommodate for edge conditions. In the Free-form
Catalan Thin-Tile Vaults project (Beorkem [5]), the
BLOCK Research Group makes use of a novel
scaffolding system and form-finding process for
compression only shells. Architects Marta
Domènech and David López López recently made
use of this construction system in the Bricktopia
project (Map13 [1]). In these projects, the
construction innovation is more explicitly related to
the design of a system of scaffolding and guidework
that adapts a traditional construction technique to
complex form. While the overall form of the shell,
the formwork used in construction, and even the
aggregation patterns of the tiles make use of digital
modeling and fabrication techniques, irregular
surface geometry is still accommodated for via on-
site manual modification of standardized modules.
2.2. Novel Components
The second approach has been to design a module
that accommodates for double-curvature in its
overall form. In general terms, this technique has
the advantage of being able to produce systems with
precise assembly tolerances and intelligent part
geometries that simplify assembly. A potential
drawback emerges when these specialized
components become non-developable themselves.
The raw materials available for the manufacture of
these custom modules are usually available as some
form of cuboid. It follows that the dimensions of
the raw material needed to manufacture a given
component must be at least the size of the
orthogonal extrema of that component. When the
interior and exterior faces of these modules are non-
planar they require a greater amount of machining.
Barring a method of reusing the offcuts from these
processes, manufacturing building components with
complex curvature on all sides inherently generates
more material waste than an orthogonal component
of equal size. A module with surface curvature on
all sides can also complicate shipping and material
handling. BLOCK’s Hyperbody (Feringa, et al. [9])
project introduces complexity at the scale of the
tectonic unit, and introduces holes at unit
intersections to reduce the number of complex
joints and provide daylighting, however it is only
realized in high-density foam. Andreani, et als.
Flowing Matter project introduces the concept of
the volumetric seam to traditional construction
techniques (Andreani, et al. [3]). Re:VAULT by
Supermanoeuvre (Kaczynski [10]) adapts joint
complexity to a masonry unit but the proposed unit
fabrication process generates a relatively large
amount of waste relative to traditional techniques
and necessitates machining surface curvature onto
each tectonic unit.
2.3. Research Opportunities
In short, it is possible to construct complex, double-
curvature surfaces in a number of ways. The
research presented attempts to capitalize on the
advantages of each method. Highly customized
assembly modules that follow a given surface
geometry with high fidelity are able to simplify the
assembly process, increase assembly tolerances,
and introduce other performative characteristics.
Standardized, flat modules are well suited to the
production capabilities of the building industry.
The research team began this project with the
following research questions in mind:
1. What performative gains can customization and
complexity at the scale of the tectonic unit afford at
the scale of the overall form?
JOURNAL OF THE INTERNATIONAL ASSOCIATION FOR SHELL AND SPATIAL STRUCTURES: J. IASS
239
2. How can customization and complexity at the
scale of the tectonic unit simplify and expedite
construction techniques?
3. What particular advantages can the use of
ceramics have on the manufacturing process?
3. OBJECTIVES
This project seeks to synthesize select aspects of the
projects described above to create a construction
system that can reduce the need for skilled labor
and complex formwork while adding a minimal
amount of material waste and complexity to the
manufacturing process. The variables manipulated
by the research team include the following: the size
and shape of the assembly modules, and the type of
bonding agent used, the allowable assembly
tolerance between modules.
3.1. Methodology
The research team proposes a new methodology for
the fabrication and construction of unreinforced
thin-shell structures using a single layer of flat
ceramic tiles. The objective of the proposed
production scenario is to reduce technical
complexity during the assembly process by adding
formal complexity to the tectonic unit. More
specifically, the research team posits that the
introduction of variable ruled surfaces at the contact
faces of each unit can impact assembly at the scale
of the construction system. The implementation of
variable ruled surfaces at the seams between units
provides increased shear resistance much in the
same way that multiple layers of tiles with a
bonding agent in between do in Catalan vaulting
systems. By using triangular panels, the proposed
construction system is able to approximate dual-
curvature surfaces with planar components. The
project uses prefabricated concrete tiles and a water
jet equipped 6-axis robotic arm to approximate a
low to medium-volume industrial production
scenario in which extruded clay slabs can be wire
cut to the specified geometries.
3.2. Proposed Production Scenario
As with most industrially produced masonry
systems, the proposed system employs off site
production of individual units. The production
scenario approximated is one in which an industrial
clay extruder is paired with a wire cutter equipped
6-axis robotic arm. Other tooling options that the
research team considered were the use of a rotating
drill on either leather-hard or fired clay. This
option was abandoned due to the inherent messiness
of using rotating tools for material removal and the
prospect of tool wear in the case of using fired clay
slabs. The research team also attempted a test to
waterjet cut leather-hard clay slabs, but found that
the unfired clay was affected by the water to a high
degree and that the tiles were too brittle for
handling in their unfired state.
Figure 1: Proposed industrial production sequence
The research team speculates that it is feasible to
create a low to medium-volume production scenario
in which an extruded clay body is cut during the
plastic or leather-hard phase. A major advantage of
this process is that the portions of the original
extrusion that are not used in the shape of each unit
module can be recycled and used to create the next
round of clay tiles. These tiles would be able to lay
flat during cutting, firing and transportation.
Additionally, if variables such as the moisture
content of the clay, drying time and firing schedules
were systematized and controlled, the shrinkage of
each module could be predicted and accommodated
for in the generation of the initial tool paths.
4. PROTOTYPES
The research team developed two prototypes to test
different aspects of the proposed construction
system. A full-scale prototype shell was developed
to test the potential of the assembly system. For the
construction of the prototype, the research team
chose to simulate the proposed ceramic production
process using industrially produced concrete pavers
and a water jet-equipped robotic arm. The use of
concrete pavers allowed the research team to avoid
many of the challenges encountered in earlier clay
tile studies – including irregularities in the slab
production and firing process and the differential
shrinkage of tiles. The waterjet was chosen to
approximate an industrial wire cutter because of
Vol. 55 (2014) No. 4 December n. 182
240
their similar geometric capabilities – most notably
the ability to produce variable ruled surfaces
through the entire depth of a given material.
4.1. Prototype Shell
Figure 2: Photograph of completed shell prototype
The initial surface geometry for the prototype shell
was generated using Rhinoceros 5 and the Thrust
Network Analysis tools available in the
RhinoVAULT plugin (Block et al. [7]). From the
initial funicular geometry, the research team
generated a triangular grid with a spacing dictated
by the size of the stock concrete tiles used in the
shell construction. Once the surface was divided
into individual triangles based on this pattern, and
extruded to the thickness of our raw material, the
team mapped a specific oscillating curve to each
vertex. This curve was scaled based on the specific
edge lengths of each surface. The specific
oscillations of each curve were confined to the
maximum achievable by the waterjet cutter –
roughly 45 degrees from vertical. Each module was
then oriented on a model of the concrete tile stock.
Wireframe geometry that described the edges of
each module was used to define toolpaths for the
robot. Cutting paths were generated in Mastercam
X7, then verified and checked for collisions in
Robotmaster. The research team produced several
studies to determine the optimal cutting speed for
the waterjet on the given material.
The size of the prototype was determined by
requirements for transportation to and from the
available fabrication and assembly spaces. While
the inherent geometric logic of the modules helps to
specify the intended assembly sequence, the 24
modules used in the construction of the prototype
used a hand-applied numeric coding system to help
expedite the assembly process. In more elaborate
installations, this logic could be combined with an
assembly drawing and automatically applied
numbering system to facilitate onsite assembly. To
ensure stability during transportation to and from
the assembly site, a small amount of polyurethane
adhesive was used to secure the tiles to one another.
Further research is required to determine how much
– if any – adhesive would be required to assemble
the shell structure in a more permanent location.
Four custom-designed springers contained the
thrust of the shell. Each springer supported one of
the shells “feet” at a direction normal to the thrust at
the ground condition. In an effort to elevate the
shell for display purposes and provide ample space
for formwork, the springers supported the shell at a
height of 12” above the ground plane.
Figure 3: Photograph of wire cut tectonic units after firing
The prototype required the design of a proprietary
formwork system for use during assembly. The
geometry of the formwork components was
generated from the original Rhinoceros model of
the prototype and designed in such a way that each
tile could be at least partially supported during
assembly. A temporary 1-ton capacity jack was
used as a central support for the shell. This allowed
the research team to accommodate for any
unforeseen discrepancies in assembly tolerance by
raising and lowering the formwork as needed. This
method also permitted the team to de-center the
shell gradually, and to easily re-install the
formwork during transit.
JOURNAL OF THE INTERNATIONAL ASSOCIATION FOR SHELL AND SPATIAL STRUCTURES: J. IASS
241
Figure 4:
Completed shell and diagram of assembly sequence
The formwork system is tailored specifically to the
particular shape of the prototype; more study is
required to generate a more general set of
characteristics and rules for future constructions.
Larger constructions may benefit from a mix of
standardized infill formwork and customized CNC-
produced final formwork.
4.2. Prototype Tiles
In an effort to gain an understanding of the potential
for clay tile shrinkage and deformation, the research
team also developed a process for producing several
prototype tiles out of earthenware clay. The
research team produced a number of 1.5 inch thick
clay slabs as an approximation of industrially
extruded slabs. In order to accurately mimic the
shape of the tiles used in the concrete shell
prototype, the geometry of the top and bottom face
of each module was laser cut from acrylic sheets.
These pairs of laser cut sheets we used as guides to
simulate a robotic wire cutting process.
Registration holes in each acrylic plate were used to
control the relative position of each pair of plates.
Four adjacent tiles identical in shape and size to the
concrete pavers were produced. While leather-hard,
these tiles were able to fit together with a high
degree of precision. Upon firing, the research team
recorded shrinkage of 4-5% of the tiles total size.
The tiles shrunk uniformly enough to still interlock
with a reasonable degree of precision relative to
their overall size.
5. FUTURE RESEARCH
The project proposes to strengthen the relationship
between contemporary digital form-finding
techniques and the process of assembly by infusing
an existing method of production with
computational design techniques and computer-
aided manufacturing. The proposed tectonic system
provides a simplified assembly method and
increased structural performance through the design
of the joints between modules. The implementation
of variable ruled surfaces at the seams between
units provides increased shear resistance much in
the same way that multiple layers of tiles do in
Catalan, etc. vaulting systems.
While the waterjet-cut concrete paver prototype
exists as a proof-of-concept for the tectonic and
formal goals of the project, combining the existing
process of clay extrusion with robotic wire cutting
could adapt this process to be a very low-waste
method of producing double-curvature shells. The
wire cutting system would allow for the low to
medium-volume production of individually
customized ceramic modules. Future work on this
project can continue on a number of aspects,
including; identifying the relationship between
surface curvature and unit size, refining the swarf
geometry generated at the contact faces, exploring
the swarf geometry as it relates to the force diagram
of a given shell structure and developing a more
universal approach to a formwork system.
Vol. 55 (2014) No. 4 December n. 182
242
ACKNOWLEDGEMENT
This research was conducted under the guidance of
instructors Nathan King and Rachel Vroman during
the course: Material Systems: Digital Design,
Fabrication, and Research Methods at the Harvard
University Graduate School of Design; Cambridge
MA; Fall 2013
Research supported by ASCER Tile of Spain;
Harvard University Graduate School of Design,
Design Robotics Group; the Office for the Arts at
Harvard, Ceramics Program; and the Harvard
Graduate School of Design, Fabrication Laboratory.
REFERENCES
[1] Vaulted brick pavilion in barcelona by
Map13. 2013 [cited 11/23 2013].
Available from
http://www.dezeen.com/2013/11/26/bricktopi
a-vaulted-brick-pavilion-barcelona-map13/
(accessed 11/26).
[2] Anderson, Stanford. Eladio dieste:
Innovation in structural art. New York:
Princeton Architectural Press, 2004.
[3] Andreani, Stefano, et al. Flowing matter:
Robotic fabrication of complex ceramic
systems. Available from
research.gsd.harvard.edu/drg/files/2012/08/Fl
owing-Matter.pdf (accessed 11/13/2013).
[4] Bechthold, Martin. Innovative surface
structures: Technologies and applications.
1st ed. New York: Taylor and Francis, 2008.
pp #42.
[5] Beorkrem, Christopher. Freeform Catalan
thin-tile vaults. In Material strategies in
digital fabrication, 2012. pp. #142-#147.
[6] Block, Philippe. Efficient and expressive
thin-tile vaulting using cardboard formwork.
[cited September 10 2013]. Available from
http://my.arch.ethz.ch/pblock/downloads/IAB
SE-IASS2011_Davis-Rippmann-Pawlofski-
Block.pdf.
[7] Block, Philippe, and Ochsendorf, John.
Thrust network analysis: A new methodology
for three-dimensional equilibrium. Journal of
the International Association for Shell and
Spatial Structures, 2007, 48 (3): 1-7.
[8] Chilton, John. Heinz Isler’s infinite
spectrum: Form-finding in design.
Architectural Design, 2010, 80 (4): 64-71.
[9] Feringa, J., and Rippmann, M. MSc2 studio
at hyperbody TU delft. 2012 [cited 10/21
2013]. Available from
http://www.block.arch.ethz.ch/brg/teaching/
msc2-studio-at-hyperbody-tu-delft.
[10] Kaczynski, Maciej, McGee, Wes and
Pigram, Dave. Re:VAULT. Available from
http://supermanoeuvre.com/re-vault/.
