Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

961778

research-article2020
SPS0010.1177/0956059920961778International Journal of Space StructuresGil Pérez et al.

Article

International Journal of Space Structures


Structural design, optimization and 1­–13
© The Author(s) 2020

detailing of the BUGA fibre pavilion Article reuse guidelines:


sagepub.com/journals-permissions
DOI: 10.1177/0956059920961778
https://doi.org/10.1177/0956059920961778
journals.sagepub.com/home/sps

Marta Gil Pérez , Bas Rongen, Valentin Koslowski


and Jan Knippers

Abstract
The BUGA fibre pavilion built in April 2019 at the Bundesgartenschau in Heilbronn, Germany, is the most recent
coreless fibre winding research pavilion developed from the collaboration between ICD/ITKE at the University of
Stuttgart. The research goal is to create lightweight and high-performance lattice composite structures through robotic
fabrication. The pavilion is composed of 60 carbon and glass fibre components, and is covered by a prestressed ethylene
tetrafluoroethylene (ETFE) membrane. Each of the components is hollow in section and bone-like in shape. They are
joined through steel connectors at the intersecting nodes where the membrane is also supported through steel poles.
The components are fabricated by coreless filament winding (CFW), a technique where fibre filaments impregnated
with resin are wound freely between two rotating scaffolds by a robotic arm. This novel structural system constitutes
a challenge for the designer when proving and documenting the load-carrying capacity of the design. This paper outlines
and elaborates on the core methods and workflows followed for the structural design, optimization and detailing of the
BUGA fibre pavilion.

Keywords
structural design, FRP structure, lightweight dome structure, coreless filament winding, structural optimization

Introduction winding as described by La Magna et al.3 and Reichert


et al.4 This technique allows the placement of fibre fila-
Carbon Fibre Reinforced Polymer (CFRP) has a high ments between scaffold frames without the geometric lim-
strength to weight ratio, with a specific strength almost 10 itation of a core. The shape of the structure gradually
times higher than structural steel.1 It also presents the addi- emerges from the interaction of these freely spanning
tional benefit that its fabrication through the placement of fibres as they are progressively hooked by a robotic arm
fibres into a resin matrix allows for material optimization around winding anchor pins at the scaffold frames.
and efficiency.2 These and other advantages as low thermal Among the pavilions developed by the two institutes,
expansion, and high fatigue and corrosion resistance,1 two may be cited as particularly significant design prece-
offer a great opportunity to use CFRP in architecture. dents for the realisation of the BUGA fibre pavilion. In both,
However, the current lack of structural codes and stand- modular fibrous morphologies resulting by winding glass
ards for this type of material triggers a need for tedious and carbon fibres between two frames were investigated.
full-scale test procedures to prove the load capacity and
safety of the structural system.
The BUGA (Bundesgartenschau) Fibre pavilion is the
Institute of Building Structures and Structural Design (ITKE), University
latest in a series of collaborative research projects between of Stuttgart, Stuttgart, Baden-Württemberg, Germany
the two institutes at the University of Stuttgart; the Institute
for Computational Design (ICD) and the Institute of Corresponding author:
Marta Gil Pérez, Institute of Building Structures and Structural Design
Building Structures and Structural Design (ITKE). The (ITKE), University of Stuttgart, Keplerstraße 11, Stuttgart, Baden-
research aim is to generate lightweight lattice composite Württemberg 70174, Germany.
structures through the process of robotic coreless filament Email: m.gil-perez@itke.uni-stuttgart.de
2 International Journal of Space Structures 00(0)

Figure 1.  Modular fibre morphologies. ICD/ITKE research pavilion 2013-14 (left) and Elytra Filament Pavilion at the Victoria and
Albert Museum in London 2016 (right) © ICD/ITKE University of Stuttgart.

Figure 2.  BUGA fibre pavilion structure with internal illumination © ICD/ITKE University of Stuttgart.

Such antecedents are shown in Figure 1: the ICD/ITKE membrane. A more detailed description of the BUGA fibre
research pavilion 2013-145 and the Elytra Filament Pavilion pavilion architectural design is given by Dambrosio et al.7
at the Victoria and Albert Museum in London 2016;6 the The design elements that composed the pavilion are
latest being the first pavilion which had to undergo thorough illustrated in Figure 3. Steel brackets are used for the con-
structural testing to prove structural safety and reliability. nection between components. A steel pole is additionally
integrated between the brackets at the centre of each node,
to facilitate the attachment of an ETFE membrane. These
Pavilion design overview steel poles transfer the forces from the membrane into the
The BUGA fibre pavilion (Figure 2) was constructed for composite components and are designed as low and high
the National Garden Exhibition (Bundesgartenschau 2019) points following the requirement of the ETFE membrane
in April of 2019 at Heilbronn, Germany. The pavilion is a and prestressed cables design. The components touch the
lightweight dome composed of 60 elongated hollow bone- ground through a foundation detail consisting of a steel
like carbon and glass fibre components. The dome spans plate mounted on a concrete cylinder and fastened via
23 m and has a height of 6.8 m. The transparent skin is anchor bolts. A total of 6 unique-shape fibre components
composed of prestressed cables and a tensioned ETFE are shown in Figure 3 right. Components from C1A to C4
Gil Pérez et al. 3

Figure 3.  BUGA fibre pavilion design elements. Pavilion global design including composite components, ETFE membrane,
membrane poles, connection brackets and foundation detail (left); and composite component geometry types (right).

are mirrored to produce a fifth of the dome, of which the


geometry is repeated to complete the 60 components.

Fibre composite components


Each of the fibre composite components is fabricated by a
robot arm that places the fibre filaments after being
impregnated by epoxy resin between two frame scaffolds
mounted on a rotating axis.8 The resulting hollow shape is
defined by the two frames and the anticlastic lattice sur-
face produced by the fibres while winding. The frame pro-
file geometry is contained in planes corresponding with
the intersection nodes of the dome where component to
component is connected. All these planar curves have a
fixed height of 710 mm and a width of either 600 mm or
910 mm depending on their location. Since the angles
between these planes change depending on the node, the
resulting cross-section of the components is elliptical in
shape, being the height always defined by the shorter
diameter. The robotically fabricated fibre layup of each of
the components is described in detail by Zechmeister
et al.9 Due to the relatively low price of the glass fibres in
comparison to the carbon fibres and the higher strength Figure 4.  BUGA fibre component robotic fabrication winding
and elastic modulus of carbon fibres,10 the two materials stages. Stage 1 when the glass fibre lattice is placed (up) and
are used with different functions in a multi-stage winding stage 2 when the carbon fibre reinforcement is added on top
process as shown in Figure 4. Firstly, the glass fibre lattice of the glass fibres (down) © ICD/ITKE University of Stuttgart.
is placed to generate the shape of the components. This
layer is normally regular and contains the minimum fibre filaments are wound as the component reinforce-
amount of material to produce a surface that serves as scaf- ment. The specific winding layup for the carbon fibre con-
fold for the winding of the carbon fibres. Secondly, carbon stitutes the loadbearing structure of the pavilion and it is
4 International Journal of Space Structures 00(0)

Figure 5.  Modelling strategy, global simplified model of composite components with ETFE membrane (left) and component carbon
fibre layup beam model (right).

structurally designed to withstand the specific loading sce- strategy is divided into a simplified global model and a
nario of the pavilion. The cross-section of the bundles is detailed component fibre layup model. This dual approach
defined by the number of rovings used during fabrication was utilized to analyse the structure at different scales. All
together with the total amount of passes or layers. In the the load cases are firstly modelled within the global FEM
case of these components, 6 fibre bundles of 48k filaments which is analysed for the corresponding load combinations.
were used with a total of 6 passes. This produced a cross- The results acquired for each of the simplified beam ele-
section of approximately 15 mm to 20 mm in diameter ments representing the six different types of components
which changes to a flatter profile where the curvature (C1 to C6) are then filtered and classified as input for the
becomes pronounced. various fibre layup models (Figure 5 right). This detailed
model, representing beam elements for every carbon fibre
bundle, is used to calibrate material quantity and the fibre
Structural design
layup pattern that can withstand the buckling induced by
The structural design of coreless filament wound struc- the global model’s established forces and moments. Due to
tures builds upon the approaches and experiences gained the unconventionality of these approaches, the final struc-
from previously developed projects such as the Elytra tural design verification is given by the full-scale tests.
Pavilion in London11 and the ITKE/ICD Research Pavilion The global FEM model incorporates the form-finding
2016 to 2017.12 Various types of modelling techniques are of the ETFE membrane with prestressed steel cables. The
used to represent either the surface of the composite com- conceptual shape of the membrane was designed by the
ponents with shell elements or the fibre layup with beam architects’ team. Several options were studied for the sup-
elements. The assumptions made for each project are based porting cable structure and corresponding cutting pattern.
on engineering experience and frequently need to be sup- As a result, a radial cable grid was chosen as it proved to
ported by testing as demonstrated by Koslowski et al.11 have the simplest details and provided freedom to maxi-
and Solly et al.12 The structural design requires to be tai- mize cable sags and reduce the prestress on the dome. The
lored for each project typology, modelled at different lev- form-finding of the membrane was carried out using
els of refinement, and compared for structural assessment. RhinoMembrane in Grasshopper3D, from which the
The BUGA fibre pavilion structural design continues the resulting shape was subsequently implemented in the
research developed at the ITKE adapting the modelling Sofistik FE global model. The membrane cables are
strategies to the project needs. Finally, for structural verifi- attached to the steel poles that transfer the forces into the
cation full-scale testing was performed as described in the composite components simplified as beam elements. To
next section. realize this simplification, an equivalent hollow circular
cross-section is chosen with a diameter of 800 mm and a
wall thickness of 3 mm. The area of this cross-section rep-
Structural modelling strategy resents the comparable material amount of the component
The pavilion’s dome-structure was analysed with the finite when distributed homogenously. The stiffness of the beam-
element method (FEM) via the Sofistik FEM software. As like components was assumed empirically due to prior
shown in Figure 5, the pavilion structural modelling experience and later adjusted according to the full-scale
Gil Pérez et al. 5

Figure 6.  FEM global model main single load cases including membrane prestress (P), temperature change (T), soil settlements
(ST) and three cases for the unbalanced wind load (UWL1, UWL2, UWL3).

test results. This approach allows investigating the behav- equilibrium shape while becoming a preloaded state for the
iour of the global structure for all the loading resulted from pavilion. Temperature change is only applied to the mem-
external actions such as wind and settlement as well as the bers directly exposed to the sun with differentiation between
prestressed cables that keep the ETFE membrane suffi- dark and bright surfaces. Soil settlements constitute another
ciently tensioned to avoid wrinkling. Additionally, the set of load cases where a displacement is induced in one of
simplification of the composite components into single the foundations while half of that value is induced to the
beam elements also facilitated the sorting of main envel- adjacent ones. This process is repeated for the different
oped forces for each component type. This set of results foundation locations of the dome. In the case of wind loads,
was used for the analysis of the component fibre layup two strategies were investigated following the guidance of
models with its continuous, iterative update during the EN199113 (part 1-4 Wind actions). The first applying air-
design process. flow over canopy roofs, assuming the wind can fully run
through the pavilion; and the second one applying wind
loads for domes with circular base, assuming the pavilion
Load cases behaves as a closed dome. The set of unbalanced wind load
All load cases and load combinations are modelled as per cases produced from the last one resulted to be the most
EN1991.13 Most relevant single load cases are shown in conservative for the design. A total of three scenarios were
Figure 6 with their main loading values. The prestress of the modelled by varying the inflection point between positive
ETFE and cables is modelled to form-find the membrane and negative load as shown in Figure 6.
6 International Journal of Space Structures 00(0)

Figure 7.  Maximum compression force distribution for the initial membrane prestress in LC2 (left) and maximum compression
forces in components C1, C2, C5, C6 produced in LC2 (centre) and in C3, C4 found in LC3 (right).

Global structure behaviour On one hand, the unbalanced load case 2 in combina-
tion with the temperature change and settlements produced
To understand the structural behaviour of the pavilion, the the maximum compression in the pavilion, which is found
individual components were firstly studied. As the compo- in both components C1 and C6. On the other hand, the
nents are composed of a fibre composite lattice structure, unbalanced case 3 results in maximum compression at
load combinations that can produce the buckling of the component C3 and C4. This behaviour is directly related to
surfaces are easily identified as crucial cases for the failure the lateral deformation of the dome which also reaches its
of the component. Therefore, high compression forces and maximum while loaded with the LC3 as illustrated in
bending moments, which increase the compression force Figure 8 (left). Due to this reason, peak bending moments
at the surface level produced the lower buckling factors in both directions are also found in this load case. For the
during this study. Torsion moments resulting from the out-of-plane moments (My), higher values are found at the
global model were minimal in relative with bending bottom components in the opposite side of the wind direc-
moments. These, as well as shear forces, did not induce tion due to the deformation and constraints given by the
large effects at the component level but were highly con- foundation support (Figure 8 centre). In the case of the in-
sidered for the design of the connections. When comparing plane bending moments (Mz), peak values appear in all
the bending moments, out-of-plane moments (perpendicu- components which geometry is against the wind direction
lar to the surface of the dome) produce larger compression (Figure 8 right).
forces at the surface level than in-plane moments (parallel
to the surface of the dome) due to the smaller lever arm
resulting from the elliptical shape of the component cross Geometrical optimization
section. From the global FEM analysis, maximum forces Global geometric optimization was performed to increase
and bending moments were found in load combinations the structural performance of the dome. The objective of
which included the unbalanced load cases 2 and 3 (UWL2 this optimization was to reduce the maximum connection
and UWL3). Figure 7 represents the compression load dis- forces at the nodes, while at the same time rationalizing the
tribution for the most relevant load combinations which geometry to simplify fabrication. For the structural optimi-
have been numbered as shown in Figure 7 for convenience zation, the geometry of the components was considered.
of the reader. In LC1, the pre-existing compression forces The components present an oval-liked section having a
due to the membrane prestress is shown. These compres- much wider diameter in-plane than out-of-plane. Therefore,
sion forces are transferred from the membrane and cables out-of-plane moments result in higher compression forces
through the steel poles at the component’s node intersec- at the connection level. By minimizing these bending
tion. In LC2 and LC3, combinations that include the unbal- moments, the absolute maximum connection forces are
anced wind load 2 and 3 respectively, the maximum also minimized.
compression forces can be read for the different compo- The optimization was performed with a parametric
nent types. Comparing the three cases, it is noticeable that model within the Computer-Aided Engineering (CAE)
a great proportion of compression is given by the prestress environment of Rhino3D and Grasshopper3D. The geom-
of the membrane. etry was simplified as a wireframe but also considering the
Gil Pérez et al. 7

Figure 8.  Deformed shape (left) and bending moment distribution for LC3, out-of-plane moments (My) (centre) and in-plane
moments (Mz) (right).

Figure 9.  Geometrical optimization of the pavilion.16 Force induction from the membrane structure to the components through
the steel pole detail (left), abstraction of pole forces from the global model into the optimization model (centre), and geometry
deviations mapped after optimization (right).

exact location of the winding pins to compute its respec- were computed and post-processed with grasshopper3D to
tive connection forces. This was mainly achieved by output the absolute maximum connection forces. In paral-
reducing the complex oval shape of the component pro- lel, the inclination of the winding formwork towards the
files to rhomboids, with their four vertexes based in each centre axis was evaluated for each component type ena-
principle bending axis. The parametric model of this wire- bling the categorization of components into fabrication
frame geometry allowed each of the nodes of the overall groups. For the geometrical rationalization and structural
grid to move in space and detach from the starting refer- optimization the evolutionary solver, or genetic algorithm
ence sphere. Additionally, the orientations of each of the (GA) that is part of Galapagos15 and operates within
nodes were also allowed to rotate changing the connection Grasshopper3D was used. When running the genetic algo-
plane location whilst maintaining symmetry. rithm, both the absolute maximum connection forces and
The finite element model used for the optimization was the number of fabrication groups were minimized.
built with Karamba3D14 using this simplified parametric The final geometry as a centreline model can be seen in
geometry as input. In this model, the façade membrane Figure 9 (right) where the deviations are mapped in col-
was omitted and replaced by the internal forces in the ours with respect to the starting geometry, of which the
membrane poles (Figure 9 left) to perform a simplified lin- nodal points lie on a sphere. In addition, Figure 10 illus-
ear analysis. These internal forces were exported from the trates the section comparison between the original geom-
global FEM described in the previous section and trans- etry and the optimized geometry used in the final design.
ferred to the linear analysis model as point loads (Figure 9 Extended information on the geometric degrees of free-
centre). During the analysis, the most relevant load cases dom and modelling approach is given by Rongen et al.16
8 International Journal of Space Structures 00(0)

Figure 11 illustrates the internal forces and bending geometry (Figures 8 and 11 right) without producing
moments for LC3 for the initial geometry (before optimi- additional compression at the connection level. The opti-
zation). These diagrams can be compared with Figure 7 mization helped to decrease the maximum connection
(right) for normal forces and Figure 8 (middle and right) forces by 27% and reduced the geometrical deviations
for bending moments of the optimized geometry used for 62% by clustering geometries in groups, and conse-
the design. Here, it can be seen that while the overall quently simplifying fabrication.
maximum compression redistributed but with a similar
magnitude, diagrams for the bending moments varied
Global buckling sensitivity analysis
greatly. By adjusting the spatial location of the intersect-
ing nodes with respect to the ideal sphere as shown in The connection stiffness between members was unknown
Figure 10, the maximum out-of-plane bending moments during the design stage of the pavilion. Changes in stiff-
of most of the nodes are reduced to their minimum ness at the intersection nodes influence the out of plane
(Figures 8 and 11 centre). As a consequence, in-plane stiffness of the dome structure, and therefore, the snap
bending moments increase in comparison to the initial through buckling behaviour. A sensitivity study was per-
formed in order to prove whether the influence of this
parameter on the global stability could be neglected. The
finite element model used for this study was a simplifica-
tion without the membrane and cable structure, where
single point loads were applied replacing the maximum
compression forces resulting from the cladding under
wind loads.
In this model, cases were computed with varying stiff-
ness of component-to-component joints. On top of the
resulting load cases, nonlinear geometric buckling eigen-
values were computed to give an indication of the resist-
ance against local snap-through of the nodes for the
different joint stiffnesses. Figure 12 shows the first buck-
ling eigenmode shapes of nodes 1 to 5. The sensitivity
analysis proved that the global structural system of the
dome was not sensitive to elastic snap-through buckling
instabilities due to variations of stiffness at the joints.

Figure 10.  Section of the original geometry which lays on a Component fibre layup structural analysis
spherical surface (top) and section of the optimized geometry
used in the final design where the component centrelines have The fibre bundles of each component are modelled and
been slightly shifted from its original position (bottom). analysed for the set of resulting forces found from the

Figure 11.  Initial geometry normal forces (left) and bending moment distribution for LC3, out-of-plane moments (My) (centre)
and in-plane moments (Mz) (right).
Gil Pérez et al. 9

Figure 12.  Component intersection node types and location (top-left) and first buckling eigenmode of nodes 1 to 5 under
compression forces induce from the membrane poles.

Figure 13.  First buckling eigenmode of a component fibre layup model (left) and a node fibre layup model (right).

global FEM model. As a simplified and conservative The stability of the building system is also assessed at
assumption, only the carbon fibre bundles are modelled. In the components’ node level to investigate the possibility of
this way, the carbon fibres are considered as the loadbear- buckling due to the forces coming from the membrane
ing structure neglecting the effects of the thin and regular poles at the centre of each intersection (Figure 13 right).
glass fibre lattice. These models gave an initial estimation on the amount of
Under compression loads, when the fibres are not fully material needed for the components, the values for which
aligned to the direction of the forces, presumably buckling were later corroborated by the full-scale structural tests.
will precede the failure of the fibre structure.17 For this
reason, the component models are evaluated for buckling
Testing process
and the required bundle thickness is estimated for stability.
Figure 13 left shows the first buckling eigenmode of one To compensate for the lack of engineering standards for
of the component fibre layup models. composites, a multi-scale experimental testing scheme,
10 International Journal of Space Structures 00(0)

Figure 14.  Destructive testing prototypes overview. Full component test (left), representation of component intersection node
test (right).

including mechanical and fire tests, was developed. Component connection design
Additionally, scanning electron microscope (SEM) scans
of randomly chosen CFRP samples were examined to For the design of the steel brackets, a computational tool
ensure consistent quality within the entire pavilion. was developed to calculate the alignment angle of the
The main full-scale structural tests investigate the fibre bundles to the edge condition, parallel to the loading
behaviour of a single component under different loading direction. This design was the result of a set of small spec-
conditions (Figure 14 left) and the interaction of various imens to investigate the load transfer between steel con-
components at the node level (Figure 14 right). Prototypes nections and the fibrous body representing the boundary
were chosen based on the geometry and loading behaviour conditions of the real joint.19 The study showed that align-
from the global model. The objective of the full-scale tests ing the fibres at the connection level to the geometry of
was to withstand maximum loading extracted from the the component can increase the capacity of the joint up to
global FEM model but also accounting for an additional 8 times in comparison to previous coreless wound pro-
material factor (γM) calculated using existing guidelines.18 jects connection design. Although the joint system for the
The full-scale structural tests were later compared with the BUGA Fibre Pavilion increased the design complexity
FEM models and these were calibrated accordingly to com- and overall weight of the structure, it greatly improved the
plete the structural checks. In addition, non-destructive structural performance of the composite components.
structural tests for each of the component geometries was Figure 15 left illustrates the type of connectors and its
planned to ensure quality control of the produced structural alignment with the fibres while Figure 15 right shows the
elements. These tests were done during the fabrication implementation of these steel brackets in the design of the
phase of randomly chosen components that were later pavilion.
assembled as part of the pavilion. Detailed explanation of
the testing process and the results is not the scope of this Membrane poles
paper, but all tests were performed for the set of enveloped
forces resulting from the global FEM analysis with satis- The interface between skin and components is a set of high
factory results. and low steel hollow poles that are connected directly to
the steel brackets at the intersection of each node. The dif-
Connection interfaces design and ferent heights of these poles are designed to maximize the
anticlastic curvature of membrane and cables when ten-
detailing sioned. The high-pole detail incorporates a ball bearing
Since the pavilion is composed of modular elements and between the pole and a cross-like plate connected to the
the integration of an ETFE membrane roof was required brackets (Figure 16 top). This is used to mitigate bending
for the design, the detailing of connections and interfaces moments at the components’ intersection nodes by allow-
with the ground and skin play a crucial role in maintaining ing rotation. In the case of the low poles where bending
the dome’s integrity. This section gives an overview of moments are almost negligible, the poles are directly
each of these details. welded to the cross-like steel plate acting as a clamped
Gil Pérez et al. 11

Figure 15.  Components connection detail. Typical connection detail where fibres are connected through sleeves and bolts to
a bent steel plate which allows the component-component connection using prestress bolts and types of connectors (left)17 and
assembled connection of four components using this system (right) © ICD/ITKE University of Stuttgart.

connection. At the top of each pole, different configura-


tions of welded fins incorporate the steel cables of the skin
(Figure 16 bottom).

Foundation design
The foundation was planned in two construction stages to
help the assembly process of the pavilion. Below ground,
a conventional flat foundation slab was designed. After
this slab was constructed, a prefabricated steel footing
was brought on site and filled with concrete. Threaded
bars were subsequently placed within the welded anchors
of the prefabricated steel pieces. Using these bars as refer-
ence, an elliptic-shaped steel plate was aligned and subse-
quently fixed via on-site welding. Afterwards, the
composite components could be attached to the founda-
tion with a special variant of the previously described
connection system.
The difference in precision between the flat conven-
tional foundation and the prefabricated components that
could neither be adjusted during the erection process, led
to a risk of misalignment on-site. Consequently, this spe-
cial detail was designed to compensate such geometric
inaccuracies by catering for a high degree of tolerance.
Figure 17 shows the foundation geometry in top view and
axonometry, and also illustrates the attachment of the
ETFE membrane to the foundation via tension cables and
custom steel brackets.
Figure 16.  Membrane high-pole detail showing the ball
bearing and steel plates connected at the components Discussion
intersection node (top) and connection of membrane, pole
and component of the built structure (bottom) © ICD/ITKE This paper describes the procedures followed for the
University of Stuttgart. structural design, optimization and detailing of the BUGA
12 International Journal of Space Structures 00(0)

Figure 17.  Foundation detail; top view highlighting the interface between ETFE membrane and ground (left) and axonometric
highlighting the interface between composite component and ground (right).

fibre pavilion. The project showcases the potential of the ORCID iD


robotically fabricated technique of coreless filament Marta Gil Pérez https://orcid.org/0000-0002-9365-5318
winding by being the largest pavilion of its type con-
structed in a public national exhibition. However, it also References
shows the difficulties and structural design uncertainties 1. Fitzer E. Technical status and future prospects of carbon
that such system presents by the lack of structural codes fibres and their application in composites with polymer
and standards. This results in a project-specific and tai- matrix (CFRPs). In: Fitzer E (ed) Carbon fibres and their
lored structural design approach and a complex experi- composites. Berlin, Heidelberg: Springer, 1985, pp. 3–45.
mental testing scheme, the exact procedures of which 2. Harris B. Engineering composite materials. London: IOM,
become necessary to prove the load capacity of the system 1999.
and obtain building permits. Nevertheless, the BUGA 3. La Magna R, Waimer F and Knippers J. Coreless winding
fibre pavilion constitutes a step forward in the research and assembled core – Novel fabrication approaches for FRP
made by the collaboration of the two institutes, ICD and based components in building construction. Constr Build
ITKE, at the University of Stuttgart. Mater 2016; 127: 1009–1016.
4. Reichert S, Schwinn T, La Magna R, et al. Fibrous struc-
tures: an integrative approach to design computation, simu-
Acknowledgements lation and fabrication for lightweight, glass and carbon fibre
The authors would like to express their gratitude towards their composite structures in architecture based on biomimetic
fellow investigators, Prof. Achim Menges, Serban Bodea, design principles. Comput Aided Des 2014; 52: 27–39.
Niccolò Dambrosio, Monika Goebel, Christoph Zechmeister of 5. Parascho S, Knippers J, Dörstelmann M, et al. Modular
the Institute for Computational Design and Construction (ICD), fibrous morphologies: computational design, simulation
University of Stuttgart. The authors would also like thank their and fabrication of differentiated fibre composite building
project partners FibR GmbH and PFEIFER GmbH and the components. In: Block P, Knippers J, Mitra N, et al. (eds)
Bundesgartenschau Heilbronn 2019 GmbH. Advances in Architectural Geometry 2014. Cham: Springer
International Publishing, 2015, pp. 29–45.
Declaration of conflicting interests 6. Prado M, Dörstelmann M, Solly J, et al. Elytra filament
pavilion: robotic filament winding for structural composite
The author(s) declared no potential conflicts of interest with respect
building systems. In: Fabricate 2017: rethinking design and
to the research, authorship, and/or publication of this article.
construction. JSTOR, Stuttgart, Germany: UCLPress, 2017,
pp. 224–231.
Funding 7. Dambrosio N, Zechmeister C, Bodea S, et al. Buga fibre
The author(s) disclosed receipt of the following financial support pavilion: towards an architectural application of novel fiber
for the research, authorship, and/or publication of this article: The composite building systems. In: Bieg K, Briscoe D, Odom C
work presented in this paper was partially supported by the (eds) Acadia 2019: Ubiquity and autonomy, proceedings of
Deutsche Forschungsgemeinschaft (DFG, German Research the 39th annual conference of the association for computer
Foundation) under Germany’s Excellence Strategy – EXC 2120/1 aided design in architecture. Texas: Acadia Publishing
– 390831618. Company, 24–26 October 2019, pp. 140–149.
Gil Pérez et al. 13

8. Bodea S, Dambrosio N, Zechmeister C, et al. BUGA fibre 14. Preisinger C. Linking structure and parametric geometry.
pavilion: towards robotically-fabricated composite building Archit Des 2013; 83: 110–113.
structures. In: Burry J, Sabin J, Sheil B, et al. (eds) Fabricate 15. Rutten D. Galapagos: on the logic and limitations of generic
2020 making resilient architecture. London: UCL Press, solvers. Archit Des 2013; 83: 132–135.
2020, pp. 234–245. 16. Rongen B, Koslowski V, Gil Pérez M, et al. Structural
9. Zechmeister C, Bodea S, Dambrosio N, et al. Design for optimisation and rationalisation of the BUGA fibre com-
long-span core-less wound, structural composite building posite dome. In: Lazaro C, Bletzinger K, Onate E (eds)
elements. In: Gengnagel C, Baverel O, Burry J, et al. (eds) Proceedings of IASS annual symposia 2019: form and force.
Impact: Design with all senses. Proceedings of Design mod- Barcelona: International Association for Shell & Spatial
elling symposium. Berlin: Springer International Publishing, Structures (IASS) 2019, pp. 1859–1866.
2019, pp. 401–415. 17. Knops M. Analysis of failure in fiber polymer laminates:
10. Knippers J, Cremers J, Gabler M, et al. Construction manual the theory of Alfred Puck. Germany: Springer Science &
for polymers + membranes: materials, semi-finished prod- Business Media, 2008.
ucts, form finding, design. Basel: Walter de Gruyter, 2012. 18. Ascione L, Caron J-F, Godonou P, et al. Prospect for new
11. Koslowski V, Solly J and Knippers J. Structural design meth- guidance in the design of FRP: Support to the implemen-
ods of component based lattice composites for the Elytra tation, harmonization and further development of the
Pavilion. In: Bögle A, Grohmann M (eds) Proceedings of Eurocodes. Italy: Publications Office of the European
the IASS annual symposium: interfaces: architecture engi- Union, 2016.
neering science. Hamburg: HafenCity University Hamburg 19. Gil Pérez M, Dambrosio N, Rongen B, et al. Structural
and International Association of Shell & Spatial Structures optimization of coreless filament wound components con-
(IASS) 2017. nection system through orientation of anchor points in the
12. Solly J, Früh N, Saffarian S, et al. Structural design of a lat- winding frames. In: Lazaro C, Bletzinger K, Onate E (eds)
tice composite cantilever. Structures 2019; 18: 28–40. Proceedings of IASS annual symposia 2019: form and force.
13. CEN Europen Committee for Normalization. EN 1991
Barcelona: International Association for Shell & Spatial
Eurocode 1: actions on structures. Brussels: CEN, 2002. Structures (IASS), 2019, pp. 1381–1388.

You might also like