Content uploaded by Oliver Tessmann
Author content
All content in this area was uploaded by Oliver Tessmann on Dec 21, 2019
Content may be subject to copyright.
2019
SimAUD
Edited by:
Siobhan Rockcastle
Tarek Rakha
Carlos Cerezo Davila
Dimitris Papanikolaou
Tea Zakula
2019 Proceedings of the
Symposium on Simulation for
Architecture & Urban Design
Georgia Tech, College of Design, School of Architecture, Atlanta, GA, USA
April 07-09, 2019
10
th
ANNIVERSARY EDITION
191
SimAUD 2019 April 07-09 Atlanta, Georgia
© 2019 Society for Modeling & Simulation International (SCS)
Rotoform - Realization of Hollow Construction
Elements Through Roto-Forming with
Hyper-Elastic Membrane Formwork
Oliver Tessmann and Samim Mehdizadeh
TU Darmstadt, Digital Design Unit (DDU)
Darmstadt, Germany
{Tessmann, mehdizadeh}@dg.tu-darmstadt.de
ABSTRACT
The paper presents a digital process chain for modeling,
simulating and fabricating rotationally molded,
individualized hollow concrete components using
material-efficient and geometrically flexible formwork
systems made from hyperelastic membranes. The hollow
concrete components are to be used as prefabricated
components for architectural constructions. The inner
cavity can be efficient in different ways: To save weight
and material, for subsequent filling with other materials
(insulating, climate regulating, water heating circulation
etc.) or as permanent formwork for solid, reinforced
structural components that are poured with concrete.
Rotoforming concrete significantly reduces the hydrostatic
pressure within a formwork and therefore unlocks
completely new possibilities for material-efficient and
geometrically flexible formwork systems.
Author Keywords
Complex concrete structures; Casting; Dynamic casting;
Membrane formwork; Rotoforming; Minimal surface;
Computational design; Simulation; Material behavior;
Additive Fabrication.
1 INTRODUCTION
Concrete is one of the most widely used building materials.
Mark Wigley conceives of concrete as “the single biggest
form of evidence of our species’s existence” on planet earth
[16] If the material is everywhere it is inevitable to enhance
concrete performance i.e. respond to the socio-economic
need for a diverse living environment that consumes less
material and energy while adapting to various local
contexts. Given the ubiquitous use of concrete even minor
improvements have a huge impact.
In the construction industry prefab-concrete elements are
still bound to a repetitive and serial logic of production.
Customized and site specific building parts on the other
hand come with high production costs and material-
intensive formwork systems.
Computational design allows the simple creation of
geometric differentiation. Digital fabrication offers a series
of adequate materialization procedures. Data flows fluently
from models of ideation and exploration to data for
fabrication. This process chain has been tested and
established in the timber industry [17]. If, however, it is
transferred to concrete structures, a contradiction arises:
The promise of concrete taking every possible shape comes
with the price of a formwork that supports the material
during the process of curing.
This research seeks to bridge the gap that emerges between
the possibilities offered by computational design and
robotic fabrication and the geometric constraints of
conventional formwork systems. This goal is achieved by
developing a process chain from digital modeling, physical
formfinding, material and process simulation and robotic
fabrication. Through migrating the rotomoulding
technology into the process of concreting we could reduce
the hydrostatic pressure of liquid concrete significantly
which allows for a completely new range of lightweight,
hyperelastic, compostable membranes as concrete
formwork.
2 RESEARCH CONTEXT
Research in the context of fabric formwork, dynamic
formwork systems, rotomolding and robotic fabrication is
relevant for this project. Fabric formwork is not new .The
technology appears in different eras and contexts of the
20th century. The majority of the work is based on
craftsmanship. Since the production process, the textiles
and the entire formwork setup have a huge impact on the
resulting form designers don’t design through drawing but
rather by experimentation with scale models and 1:1
prototypes. Veenendaal et al. propose a taxonomy of
different textile formwork systems [15]. Within this
taxonomy our hyperelastic membrane falls into the
category of bi-axial mechanical prestressed formworks.
Computational formfinding - the simulation of external
192
forces impacting on a material system - is only recently
migrated into the realm of fabric formwork. The method is
known for finding the geometry of form-active structures.
In the context of fabric formwork it becomes a construction
method [14]. But more important, tools like Kangaroo
allow for simulation within an architectural design
environment. Thus simulation allows exploring the design
potential of fabric formwork before physically making it.
The Block Research Group at the ETH Zurich combines
tailored fabric formwork with prestressed cable nets as
underlying falsework. Concrete is sprayed onto the
formwork in thin layer. The cable net deforms under the
weight of the sprayed concrete into the shape that is
designed by computational methods developed by the
researchers. The research thus developed a computational
design process for fabric formwork [13]. As the range of
possible forms resulting from the use of fabric formwork is
limited to shapes that emerge by fabric being exposed to
hydrostatic pressure, other research trajectories explore
ways to incrementally improve existing formwork. The
increase of material efficiency through the use of recycling
and re-shaping formwork material such as wax [8] or water
[10] is explored in various research projects. Geometric
freedom is furthermore achieved through flexible
mechanical-kinetic systems such as dynamically
reconfigurable double-curved molding surface shaped by
an array of actuators [5]. Another example of a dynamic
formwork system is the Smart Dynamic Casting (SDC)
project by Gramazio Kohler at the ETH Zurich. SDC is
based on the concept of slipforming in which concrete is
poured into a continuously moving form. The procedure
allows for a continuous and gradual change of the cross
section of the cast element by shaping the concrete during
curing through the subtle movement of the formwork by a
robot [6] The project exemplifies the importance of
merging design intent, digital fabrication processes and
material science into one coherent process. The MARS
pavilion by Sarafian, Culver and Lewis exemplifies the use
of robots in combination with fabric formwork. The system
allows fabricating branching concrete structure cast into
adjustable fabric formwork. Robots guarantee the exact
position of tailored fabric formwork sleeves that are
subsequently assembled into a dome-like lattice structure.
The aim was to find a cost competitive way to fabricate
parametrically designed concrete structures [9]. Martin
Bechthold and Jonathan King from Harvard GSD mass
customize concrete objects through robotically orienting a
mold while the material cures. The project was presented
as a workshop at the Robotics in Architecture 2012
Conference in Graz and Vienna.
None of the mentioned projects addresses the reduction of
hydrostatic pressure that we regard as a key concept to
unlock a completely new range of material efficient
formwork materials. This is achieved by migrating the
rotomoulding technology into the realm of concrete
processing. Roto-forming is a production process in which
a liquid material is poured into a mold. The amount of
material is sufficient to adhere to the wall of the slowly
rotating formwork, but not enough to fill the entire mold.
The manufacturing process is used in the plastics industry
for the production of hollow objects such as water tanks,
barrels, kayaks, plastic furniture etc. Here massive steel
molds are heated to melt the plastic. Al-Dawery et al
migrate rotomoulding from the plastic industry into the
field of ceramics [1]. Empirical design and prototyping
research on the use of hyperelastic membranes as
formwork within a DIY rotomoulding process have first
been developed by Thomas Vailly and Itay Ohaly for the
production of small-scale design objects. The latex
membranes allowed the production of different shapes
without the need of previous tailoring [12].
Figure 1. A lightweight hollow concrete (UHPC) object
rotoformed in a prestressed latex membrane.
3 METHOD
3.1 Material System
Rotoforming is conceived of as a material system in which
form, material, structure and its synthesis (materialization,
fabrication and assembly) are regarded as integral and
closely linked elements.
Figure 2. RotoForm Material System
193
Computational tools and techniques, as part of this system,
allow notating and instrumentalizing the intricate
interactions between form, material, structure and
environment within the architectural design process.
Simulating material systems within digital generative
models utilizes computation beyond formal and
geometrical design schemes. The notion of the model shifts
from representation of objects towards the abstraction of a
process and the prediction of behavior [3]. The material
system approach was used in this project to revisit and
challenge conventional formwork systems. Instead of
incrementally improving existing formworks we
reconsidered the entire process of concreting and identified
membranes as formwork material. Besides minimal
material consumption these membranes can be tensioned
in a wide range of forms without previous tailoring.
We reduced the consumption of both concrete and
formwork material and at the same time expanded the
design potentials for our built environment. This could be
achieved through rotoforming concrete.
3.2 The Simulation-Based Design Tool of RotoForm
Fabric formwork has no significant tradition in the building
industry as it is very different from conventional
formwork. A rigid formwork is a technological means to
transcend geometry envisioned by the architect into matter.
Fabric formwork, in contrast, becomes part of the design
process as its material performance and boundary
conditions have a significant impact on the resulting shape
[14]. The forms that emerge when hydrostatic pressure acts
on a fabric or a membrane that is prestressed, can hardly be
captured by 2D drawings or even 3D models. Hence,
physical models, small-scale or 1:1, have been the tools for
designers when working with fabric formwork [7]. Making
these physical models and prototypes is a craft that requires
a different skill set than the production of drawings or
digital models. The craft includes experience and tacit
knowledge that is not easy to standardize. As Remo
Pedreschi describes, the previously separated roles of
builder and the designer merge into one:
"The role of the builder or maker of fabric cast concrete
involves both the deconstruction of the object into a
sequence of steps and the continual re-evaluation and
adjustment of the form during the assembly and casting
process. The design develops during the making." [2]
As comprehensible as this coalescence may sound, it is
however also responsible for the absence of fabric
formwork in the construction industry. Designs that only
unfold during making cannot be represented in the
conventional artifacts that designers produce. Furthermore
design decisions are required during the making, which
means that the design phase does not stop with the
production of representations such as drawings and
models. Designers have to be involved into the
materialization.
Against this background we sought to develop RotoForm
into a material system consisting of digital and physical
components (see Figure 1) that are fluently combined, but
also clearly sequenced. Designing with rotoformed
elements should be possible without the need for physical
prototyping but through the use of simulation-based design
tools and methods. Design tools incorporate material
behavior under the impact of external forces to overcome
mere geometric representation. Thus we use Kangaroo in
Grasshopper to simulate the prestressing of membranes.
These tools are accessible for designers and well integrated
into the architectural design environment. The simulation
accomplishes both: It is a technical necessity for the
subsequent fabrication, but it also contributes to the
representation of the design proposal. The formal potential
of the material system is visualized. At the same time its
geometric limitations and material constraints are
displayed. Digital simulation is not meant to replace
prototyping and physical modeling but should rather
complement these activities.
4 PROCESS
4.1 Design Process
In this research we conceive of rotoformed elements as
nodal connections for irregular space-frame structures.
The form of the nodes is a result of a form-finding process
in which a particle-spring model is used to find the minimal
surface that emerges between all rods intersecting at one
point (node) of the space frame. The topology, described
by the center-line model, is complemented by a low-
polygon solid that approximates the dimensions and
orientation of the node (see Figure 3).
Figure 3. Intersecting centerlines, relaxed mesh of nodal
geometry minimal surface.
The mesh resolution is increased for the formfinding
simulation using the Grasshopper add-on tool Weaverbird
and Meshmachine. Tension forces along the center-lines
induce prestressing into the mesh. Mesh edges act as
springs and the vertices are exposed to the forces. The
particle spring model generates a relaxed mesh
approximating the minimal surface that emerges when a
hyper-elastic membrane is pre-stressed. Commonly-
available software packages, such as Rhinoceros and
Grasshopper, Kangaroo are used in order to make the
digital process accessible for architects. The same software
194
packages are also used as a means of direct-communication
with a UR 10 Universal Robots and a turning table.
The simulated form of the nodes is subsequently used to
calculate its volume and surface area, two important
parameters to the rotoforming process. Based on the
surface area and the aspired wall thickness of the hollow
element, the amount of cast material is calculated.
4.2 Manufacturing Process
To manufacture the digitally designed and simulated nodes
the digital geometry is translated into a prestressed latex
membrane kept in place by a spherical falsework (see
Figure 7). All steps including their methods and tools are
described in the following paragraphs.
Figure 4. Pretentioned the hyper-elastic membrane formwork in
the spherical falsework
The adaptive spherical falsework
The pre-stressed membrane is fixed to a spherical
falsework that acts as the boundary resisting the large
anchoring forces (see Figure 4). The adaptive falsework is
designed to allow for the generation of different membrane
shapes by changing the position of the tension anchors.
Two aluminium plates form the poles of the sphere. The
poles are tied together by a series of median arcs (between
6 to 10 depending on geometry). The arcs carry clamps that
act as anchoring points and take the loads from tensioning
the membrane. 3D printed elements connect the rims to the
poles. They can slide in a notch to change the location of
the meridian arcs. A screw allows tightening or loosening
the connector for proper placement. Clamps that slide
along the meridian arcs are holding the anchor rods that
define the location and direction for tensioning the
membrane.
Robotic placement
A robot translates the digital geometry into the physical
setup. In this setup, a UR 10 six axis robot and a turning
table are used to place a rod in the correct position and
orientation in relation to the sphere. The turntable rotation
is controlled via a combination of an Arduino single board
micro controller and Funken, a serial protocol toolkit for
interactive prototyping [11]. The robotically
positioned rod is fixed to the clamps. After all rods are
placed, the membrane formwork is placed inside
the falsework. The anchor rods penetrate the membrane
and connect it to the falsework. Steel and rubber washers
connect the rods to the membrane and transfer the stresses
of the subsequent tensioning.
Figure 5. digital data extraction and Robotic Placement
Pre-tensioning membranes
Membranes are flexible, non-rigid structures that transfer
loads through tension. They require fixed ends or rigid
linear boundaries that withstand the horizontal forces
inherent in every form-active system. Their bearing
mechanism relies on the material form. Form coincides
with the flow of stresses that are equalized or harmonized
along the surface. Loads are dispersed in the direction of
resultant forces without any shear [4] In this project we
used latex membranes as membrane formwork within the
rotoforming process. Filled with air, these balloons take a
spherical shape due to internal pressure that acts
perpendicular to the membrane surface. Filled with liquid
concrete the combination of gravity and hydrostatic
pressure generates drop-like shapes. Thus these external
forces during production would not allow generating any
other form. We therefore sought to minimize this effect
through the reduction of cast material in the formwork and
through prestressing the membrane formwork (see Figure
6). A series of tension-inducing anchor points are
connected to the membrane. In order to achieve a
harmonized stress distribution and avoid wrinkles in the
membrane the anchor points need to be placed in a way that
tension creates curvature in all areas of the membrane.
Figure 6. Pre-tensioning membranes
The Rotoforming machine
The rotoforming machine consists of a frame that rotates
around a horizontal spindle powered by an electric motor.
The frame carries a vertical spindle to which the spherical
formwork is connected (see Figure 7). A belt and 90 degree
tapered gear wheels transmit the rotational motion.
195
Figure 7: The digital-physical robotic-aided process. From left to right: Robotic placement of rods on spherical scaffold, Prestressed
membrane formwork. Rotoforming machine.
The movement of the two spindles needs to be aperiodic to
make sure that the formwork is fully rotated and all its
regions pass the lowest point as the liquid material flows
downwards.
The formwork slowly rotates to disperse the liquid cast
material to the membrane. In contrast to spun concrete
parts, the material is not allocated through centrifugal
forces that tend to stratify the material. The material rather
adheres to the membrane surface yielding high quality
surfaces.
Cast material
During rotoforming the liquid material is subject to a
constant change of shape. The impacting loads are not
static. In the early stage of curing the material furthermore
deforms according to the geometry of the membrane. For
this research, the material needed to be adjusted in such a
way that it could absorb the quasi-dynamic load at different
phases of rotation in different layer thicknesses without
cracking. Component sizes, rotational speeds and
production speeds as well as the composition of the
formwork material have a decisive influence on the
viscosity requirements. Mass inertia, adhesive forces on
surfaces and hardening processes must be controlled in
combination with layer thickness formation and shrinkage
cracking.
Two different materials were tested in the project: An ultra
high performance concrete (UHPC) and an acrylic/plaster
composite. The composite was used for testing the entire
process. Its short curing time allows for a fast production
of rotoformed elements. However, the main goal is the
production of rotoformed concrete elements. Increased
hydrostatic pressure and longer curing periods of concrete
pose additional challenges to the process. First concrete
prototypes were manufactured in collaboration with the
concrete company Gtecz (see Figure 1).
Resulting hollow body component
The materialized object is a lightweight hollow component
with approx. 3% of the weight of a solid component in
similar size (180 mm radius and 7 mm shell thickness). The
resulting surface is smooth with gradually changing
curvature continuously blending all directions of the
surface. The cast-in anchor rods serve as connection
between two components or between nodes and rods. The
cast nodes were subsequently 3d scanned in order to
compare the cast object to the empty prestressed
membranes in the spherical falsework and to the simulated
minimal surface of the particle-spring model. The
comparison between object and membrane proof that
hydrostatic pressure has no impact on defining the shape of
the membrane. The simulated form deviates from the form
of the cast object. As the current role of simulation is
limited to provide a better formal approximation in the
early design stage the current precision is satisfactory.
Figure 8. The Section of the resulting demonstrator with hollow
body nodal component.
5 CONCLUSION / OUTLOOK
Within our research we could prove through a novel
fabrication process and the resulting prototypes that
rotoforming concrete in membrane formwork is possible
and leads to material-efficient formworks that allow for
geometric differentiation without the need of tailoring the
formwork. Furthermore, the process yields hollow concrete
objects with reduced weight and material consumptions.
196
Figure 9. The large scale demonstrator of space truss system at
AAG Conference 2018 with hollow body concrete nodal
component
By adding rotational movement to the process of
concreting and thereby reducing hydrostatic pressure we
are able to reconsider the established palette of formwork
materials towards more lightweight and efficient materials.
Besides the important aspect of saving resources these
materials also simplify the de-molding of concrete objects
and generate high surface qualities. The shape of the
minimal surface of the prestressed membrane is perfectly
mirrored in the concrete object. The tools and methods
developed for the process yield geometric precision of the
complex forms. A prototype of a series of six
interconnected nodes and a rotoformed base plate
demonstrate the validity of the material system (see Figure
8 and 9).
Hydrostatic pressure is no longer the form-defining force
in this membrane formwork system. However, the need for
an even of harmonic stress distribution in the membrane
create novel constraints for the range of possible forms.
One hypothesis of this research was that such a tool and
method would allow designers to explore the formal
potentials of membrane formwork and make them part of
their design. The hypothesis was tested in a series of
workshops with students, researchers and professional
designers.
We could observe how the teams implemented material
performance into their aesthetic approaches: One example
is the design of extra tension-spikes for Harmonizing the
stress distribution in the membrane requires similar
curvature in all regions of the membrane. A lack of tension
in the membrane leads to wrinkles and reduced capacity to
withstand the liquid material.
Figure 10. The resulting demonstrator with hollow body
concrete nodal component.
A constraint that may be in conflict with the location and
orientation of space frame rods that tension the membrane.
When testing the system with students in a rotoforming
workshop design team added extra tension-spikes that were
independent from the space frame rods, in order to generate
the necessary pretension in the system. The integration of
simulation of the material system into the early stages of
the design process in which the shape is of importance for
designers generated novel material-appropriate but also
aesthetic design solutions.
The roto-formed elements are significantly lighter than
massively cast elements and can thus contribute to more
lightweight constructions that consume less material. The
reduced weight of a rotoformed facade element allows for
more lightweight substructures. The effect thus propagates
through the entire construction.
Rotoforming comes with challenges that requires future
research in the following fields:
The aim of letting a concrete cure while it is being moved
is a conceptual contradiction that can only be solved by the
careful design of movements coordinates with concrete
recipes that allow for the curing under these delicate
circumstances. More data needs to be collected and
procedures require standardization to be able to reliably
reproduce results of similar quality.
Curing generates heat which is currently collected within
the closed membrane system. The heat expands the
formwork and can lead to a delimitation of the concrete
from the formwork and in the worst case to a collapse of
the hollow element during rotation. Casting subsequent
layers of material is necessary to not deform the delicate
197
membrane and the first layers of concrete but requires a
cumbersome process of filling the material into the
formwork without destroying the previous layer of
material.
The above mentioned challenges will be addressed in the
ongoing research together with questions of enlarging the
range of possible forms and Morphologies through
variations of the membrane formwork system and its
falsework/boundary condition, the integration of
reinforcement and mounting elements and the variation of
wall thickness through a differentiated and controlled
rotational movement.
AKNOWLEDGMENTS
This research greatly benefited from a series of workshops
in which participants tested and prototyped. Their feedback
and experiences informed our work. Special thanks to the
participants og the workshop at the AAG conference 2018
in Chalmers: Johan Dahlberg, Deena ElMahdy, Eftixis
Efthimiou, Felix Graf, Fabio Scotto, Franz Theobald,
Athanasios Vagias, Yuwei Zhang. First tests with
rotoforming concrete have been conducted with the
support of G.tecz/Gregor Zimmermann. The Machines has
been built with the technical supports of Mirko Feick/PTU,
Marecel Bicolay/ VKM, TU Darmstadt, Alexander Stefas,
Andrea Rossi and Felix Graf by DDU of TU Darmstadt.
REFERENCES
1. Al-Dawery, I., Binner, J., Tari, G., Jackson, P.,
Murphy, W., & Kearns, M. Rotary moulding of
ceramic hollow wares. Journal of the European
Ceramic Society, 29(5), (2009), 887-891.
2. Chandler, A., Pedreschi, R. (editied by). Fabric
Formwork. London: RIBA Publishing. London UK,
2007
3. Hensel, M., Menges, A. (eds.), Morpho-Ecologies:
Towards a Discourse of Heterogeneous Space in
Architecture, AA Publications, London. 2006
4. Engel, H. Structural Systems, Hatje Cantz Verlag
2007
5. Kristensen , Mathias Kraemmergaard; JEPSEN,
Christian Raun. Flexible mat for providing a
dynamically reconfigurable double-curved moulding
surface in a mould. U.S. Patent Nr. 9,168,678, 2015.
6. Lloret, E., Shahab, A. R., Linus, M., Flatt, R. J.,
Gramazio, F., Kohler, M., & Langenberg, S.
Complex concrete structures: merging existing
casting techniques with digital fabrication.
Computer-Aided Design, 60, (2015). 40-49.
7. MANELIUS, A.-M. Fabric Formwork.
Investigations into Formwork Tectonics and
Stereogeneity in Architectural Constructions. Ph.D.
Thesis, Royal Danish Academy of Fine Arts Schools
of Architecture, Design and Conservation, School of
Architecture, Denmark, 2012.
8. Oesterle, S., Vansteenkiste, A., & Mirjan, A. Zero
Waste Free-Form Formwork. in Second International
Conference on Flexible Formwork, ICFF. CICM and
University of Bath, Dept. of Architecture and Civil
Engineering, Bath (2012) 258–267
9. Sarafian, J., Culver, R., Lewis, Trevor S. Robotic
Formwork in the MARS Pavilion: Towards The
Creation of Programmable Matter. In ACADIA
2017: DISCIPLINES & DISRUPTION [Proceedings
of the 37th ACADIA, Cambridge, MA 2-4, (2017)
pp. 522- 53
10. Sitnikov, V. Ice Formwork for High-Performance
Concrete: A Model of Lean Production for
Prefabricated Concrete Industry. Structures, Elsevier,
2018
11. Stefas, A., Rossi, A. and Tessmann, O. Funken -
Serial Protocol Toolkit for Interactive Prototyping.
In: Computing for a better tomorrow - Proceedings of
the 36th eCAADe Conference - Volume 2, Lodz
University of Technology, Lodz, Poland, 19-21
September (2018), 177-186
12. Vailly, T. , Ohaly, I. The creative Factory
http://www.vailly.com/projects/the-creative-
factory. As of 15. April 2015
13. Van Mele T., Méndez Echenagucia T., Pigram D.,
Liew A. and Block P.A prototype of a thin, textile-
reinforced concrete shell built using a novel, ultra-
lightweight, flexible formwork system,structure,1
,(2018), 50 - 53
14. Veenendaal D. and Block P.Computational form
finding for fabric formworks: an overview and
discussion,Proceedings of the 2nd international
conference on flexible formwork,Ohr, J. et al.
(editors), Bath, UK, (2012), 368-378,
15. Veenendaal, D.; Block, P.; West, M. History and
overview of fabric formwork: using fabrics for
concrete casting. Structural Concrete 12, 3 (2011),
164 – 177
16. Wigley, M. Foreword, in M. Bell and C. Buckley
(eds), Solid States (Columbia Books on Architecture,
Engineering, and Materials), Princeton Architectural
Press, 2010
17. Willmann, J., Knauss, M., Bonwetsch, T.,
Apolinarska, A. A., Gramazio, F., & Kohler, M.
Robotic timber construction—Expanding additive
fabrication to new dimensions. Automation in
construction, 61, (2016) 16-23
