You are on page 1of 7

An overview of using physical models in shell structures form

finding from the past to the present

Fatemeh Baygi1,* Azade Sargazi2


1
M.A. Architecture, Faculty of Art , Tarbiat Modares University ,Tehran , Iran
2
M.A. Islamic Azad University Gorgan, Iran

ABSTRACT

This paper discusses existing physical form finding methods for shell structures. It focuses use of
physical models in the last 100 years, followed by a brief description of some projects where physical scale
modelling played an active role in the design process and then compares methods with each other. For a shell
structure, the geometric shape plays an important role in its structural efficiency, which means they form
according to the load they must carry for the best resistance, therefore structure and architecture designing are
integrated in these kinds of structures. Therefore, to obtain a structurally efficient shell, its shape should
depend on the flow of forces, and vice versa, its design requires a process of form finding. The aim of this
research is to survey physical models that have been developed by researchers, engineers, architects, or
contractors over the past several decades. This paper will demonstrate how use of physical models led to
substantial improvements in the design of shell structures from the past to the present.

Keywords: shell structure, physical model, form finding.

1. INTRODUCTION
Shell structures will always have a role for architecture and engineering. More so than any other
structural system, shells have the ability to create eye-catching forms, to provide freedom for design
exploration and to resist loads efficiently. Shell structures are constructed systems described by three-
dimensional curved surfaces, in which one dimension is significantly smaller compared to the other two. They
are form-passive and resist external loads predominantly through membrane stresses. For even the most highly
constrained geometry, an infinite number of solutions are available to the shell designer. But each different
shell geometry has advantages and disadvantages, and all shells are not equal. How then does the designer find
shell forms that are inherently structural? Shell designers can invent forms by taking inspiration from nature,
by innovating from precedent structures, or by exploring various form finding possibilities (Adriaenssens et al,
2014).
The definition of form finding is: Finding an (optimal) shape of a [form-active structure] that is in (or
approximates) a state of static equilibrium (Lewis, 2003). Form finding is finding an appropriate architectural
and structural shape (Coenders and Bosia, 2006). a structural optimization process which uses the nodal
coordinates as variables (Basso and Del Grosso, 2011). Adapted from Adriaenssens et al (2014) Form finding
is a forward process in which parameters are explicitly/directly controlled to find an ‘optimal’ geometry of a
structure which is in static equilibrium with a design loading.

1.1 Form finding methodology


In the history of form finding of constrained structures, including masonry constructions and thin
concrete shells, three approaches have been developed: physical model, theoretical analysis on specific
geometries, and numerical solutions using computers (Tamai, 2005). Although, Computational models for
form finding may be a numerical simulation of the physical model or they might use imaginary properties that
could not be simulated physically (Adriaenssens et al, 2014).
Until the advent of computer technology in architectural design, the physical model study had been the
main and the most effective approach to generate and evaluate so-called natural forms of structures (Tamai,
2005) and this models based on the “form follows force” principle were widely used to design or construct
shells. As form-finding means, physical models are made of small pieces of flexible membranes or fabrics
subject to certain loads and boundary conditions, and after evaluating and scaling them, efficient shapes of
shells can be obtained. As moulds for construction, referring to pneumatic physical models in this paper, they

http://cau.icnf.ir
are made of flexible membranes but with real sizes, and afterward they would be covered with building
materials. Based on their different manufacturing methods, these physical models can be divided into three
groups (Li et al, 2017).

1.1.1 hanging models


The engineer and scientist Robert Hooke (1635–1703) worked with Christopher Wren (1632–1723)
during the design of St Paul’s Cathedral in London (Adriaenssens et al, 2014). In 1676 Hooke noted that the
equilibrium form of an arch was the same (although inverted) as a hanging chain comprising the same weights
as the arch (Addis, 2013). One of Wren’s sketches for the 33m diameter dome shows the shape of a chain
suspended over a cross section of the building (Fig.1).This simple model was the earliest known use of a
physical model being used to help determine the form of a structure (Adriaenssens et al, 2014).
Hooke's idea directly led to the use of hanging models for designing and calculating structures (Huerta,
2006). It reduces the form finding process to a natural process leading automatically to a bending free result
with pure compression, ideally suited for no or low tension material like masonry or concrete (Ramm and wall,
2004).

Fig. 1. St Paul’s Cathedral, London. Sketch of Hooke’s catenary superimposed on a proposed section
of the dome (Adriaenssens et al, 2014)

Giovanni Poleni (1682-1761) used Hooke’s principle in the 1740s, this time using a chain with different
sized weights representing the voussoirs, to help demonstrate that the 100 year old dome of St Peter’s
Cathedral in Rome was stable despite some radial cracks (Heyman, 1996). he divided the dome into slices and
hung 32 unequal weights, and then showed that the hanging chain could fit within the section of the arch and
was therefore safe(Fig.2) (schenk, 2009).

Fig. 2. Poleni's drawing of Hooke's analogy between an arch and a hanging chain, and his analysis of
the Dome of St. Peter's in Rome (Block et al, 2006)

In Germany, in the 19th century, Heinrich Hübsch carries out some research on uses hanging models
(Graefe 1985). He used hanging string models to determine the weights of voussoirs needed to achieve the
desired shape of an arch or vault shape (Fig.3)(Graefe, 1983).

http://cau.icnf.ir
Fig. 3. Investigation of form and construction of various vaults using a hanging chain by Heinrich
(Graefe, 1983).

During the second half of the nineteenth century, a number of books recommended the use of hanging
models to establish the best geometry for arches and vaults. In the 1890 edition of Ungewitter’s classic book on
Gothic construction, Karl Mohrmann specifically recommended the use of three-dimensional hanging models.
In the 1890s, Friedrich Gösling (1837–1899) also used both two and three dimensional models for some of his
designs (Fig.4) (Addis,2013).

Fig. 4. Form-finding of arches and vaults using hanging chains by Friedrich Gösling (Addis, 2013)
Rather better known is the work of the Catalan architect Antoni Gaudí (1852-1926) who used both two
and three dimensional hanging models, made with strings and bags of sand to help establish the forms of
arches and vaults for several of his masonry buildings. The most well-known of these models was for The
Church of Colònia Guell (Fig.5) (Addis,2013).

Fig. 5. Reproduction of Gaudí’s hanging model for The Church of Colònia Guell, Barcelona
(Addis,2013).

http://cau.icnf.ir
He used similar models for other projects that were not built. Gaudí used the results of his model tests to
complement his use of both statical calculations and graphical statical methods to determine the forms of the
treelike columns and vaults.The use of parametric equations can be seen in many aspects of Gaudí’s
architecture, but it is better illustrated through his use of his hanging chain model (Fig.6). Gaudí used this
principle to design the Colònia Güell chapel, creating an upside-down model of the chapel using ropes loaded
with weights. This way, because of “Hooke’s law”, the ropes will be always arranged in a way that results in
pure compression. Thus, through changing the parameters in the parametric model, Gaudí could generate other
versions of the Colònia Güell chapel and be sure that the resultant structure would be under pure compressive
stress (Makert and Alves, 2016).

Fig. 6. Gaudí’s hanging model and inverted photographs used to render forms (Maher and Burry,
2003)
Heinz Isler (1926-2009) was the last of the great concrete shell builders of the 20th century and, like
many before him, brought his own unique approach to the challenge. He took into three dimensions Hooke’s
premise about the inverted catenary using various sheet and membrane materials to make hanging models,
which he then scaled up to reproduce their funicular geometry at full size (Addis,2013).
Hanging cloth reversed Isler has commented that he first realised the potential for form-finding with a
hanging cloth when he noticed the shape of jute fabric (used wet to aid the curing of concrete) draped over
steel reinforcing mesh (Chilton, 2000). Early experiments with the technique, using a wet sheet suspended
from four poles on a freezing winter evening, are illustrated in his paper “New Shapes for Shells” (Fig.7)(Isler
1961).
For form-finding of his shells Isler loaded the fabric surface with a plaster of his own formula
developed to maximize mouldability when wet whilst maintaining a constant thickness on the curved fabric
and minimizing cracking of the drying surface. However, Isler also realized the shortcomings of this technique,
due to the influence of fabric weave and its orientation relative to the boundaries. To minimize these effects,
for more accurate modelling of the hanging surface, Isler employed a selected highquality latex rubber
membrane which has consistent isotropic properties.
The idea was to hang an elastic surface between supports and allow gravity to shape the form of a
catenary surface in tension. If inverted this creates a form in pure compression under its self-weight and
provides the stiffness to resist local instability.
Heinz Isler experimented with thin structures of ice over many years. Taking advantage of the cold
Swiss winters and the extensive grounds around his house in Zuzwil he made a range of frozen shapes using
just water and fabric (Fig.8)(Chilton, 2012).

Fig. 7. The hanging and inverted membrane (Heinz Isler Archive)

http://cau.icnf.ir
Fig. 8. When combined they create an arched grotto(Heinz Isler Archive)

1.1.2 tension models


These models, which are made of soap film or gauze, aim to find the equilibrium shape of a minimal
surface between preset boundaries. They represent a highly significant tool for exploring the shapes of tent
constructions. However, tension models were also used for form-finding of shells (Li et al, 2017).The scope of
frei otto investigation includes soap-film experiments for producing minimal surfaces, pressure-loaded vault
forms by reversing tension-loaded suspended forms, optimized path systems, pneumatic structures, and
branched constructions. soap film model of Frei Otto was used for understanding the shape of a conoid form
for a tensile structure which he used in his design of inverted umbrellas. All points on this surface have equal
tension because of the modeling technique used, hence no wrinkling of the membrane and even stresses
throughout.(Fig.9) The lines on the surface were projected using a wire grid. In tandem with this form-finding
modeling design process, Otto had an amazing skill of being able to see the design issues from a 10,000-m
perspective and slowly hone in on a particular design. He would layer the complexity into the design with each
additional step, segueing from geometry into detail and back out again.

Fig. 8.
Frei Otto applied this kind of model in the design of the Stuttgart train station, demonstrating the formal
and structural novelty which derives from experiments with minimal surfaces (Li et al, 2017). Musmeci was a
pioneer in use of form finding techniques for the design of structural surfaces. He was able to derive
structurally efficient shapes for a wide series of bridges using an iterative process of physical models,
analytical formulations and simple finite-difference simulations. Nervi and Morandi were leaders in
experimenting with pre-stressed concrete structures, but they tended to stick with more traditional arch bridge
and shell forms (Trovalusci & Tinelli 2013). Comparatively, Musmeci extended his focus to using these
materials to build previously undefined continuous shell forms, determined by the stresses and strains
themselves. For design development of the Basento Viaduct, Musmeci explored the potential of soap film, a
thin layer of liquid between air, to start to define the three dimensional geometry of the lower section of the
bridge that connects to the foundations. A soap film is a minimal surface with identical surface tension in all
directions under a specific set of boundary conditions. This geometry minimizes its surface and its energy
while being constrained to the wire frame.( Magrone et al, 2016)

http://cau.icnf.ir
By exploring the possibilities of form finding for these minimal surfaces, he sets his focus away
from defined, calculable geometries and towards a simple definition of boundary conditions, from which
a form is self generated by the use of textile or soap film models.
His philosophy, “The form is the unknown, not the inner stresses” (Musmeci 1979), becomes
manifested in his form finding experiments. These new forms are based on textile membranes or soap film
models and are related to phenomena of natural sciences (Nicoletti 1999). During the prospering years in post
war Italy and in the shadow of the prominent Italian engineers Pier Luigi Nervi and Riccardo Morandi,
Musmeci researches concrete surface structures. He calls for new approaches to form finding, particularly in
respect to moldable concrete: “The form is the unknown, not the inner stresses”(Ingold and Rinke,2015).

1.1.3 pnematic models


For these models, the soap film or a piece of membrane (air tight or allowing very little air through) is
blown in a certain shape possibly with a closed preset boundary, and the overpressure inside then forms an
equilibrium shape. Pneumatic models can be adopted to determine the efficient shape of shells, and they can
also be used as moulds for construction (Li et al, 2017).
Isler’s experiments with pneumatic forms date back to 1954 when he noticed the shape of the plumped-
up pillow on his bed. The continuously curved surface of the pillow made him realize that physical models
were the key to solving the problem of organic, free-form structures.
Isler’s simple yet elegant solution comprised a rectangular wooden frame placed on a wooden
baseboard, with the latter having a small hole drilled into it. Isler then placed a pliable rubber sheet between the
baseboard and the frame, with its perimeter restrained by clamping the two wooden boards together. A sealed
flexible pipe and a hand pump could then be used to inflate the rubber membrane to produce a double-curved
synclastic surface.8 This ‘technical pillow’ provided Isler with the opportunity to experiment with this form,
and a similar apparatus was also used by Balz when designing his home.
As molds for construction, pneumatic models were extensively used by Bini to design and construct
reinforced concrete thin-shells (Fig.9), and were also applied by Kokawa to design and construct ice-shells
(Fig.10). From the above introduction, it can be observed that each group of physical models represents a
typical type of static force equilibrium which obeys the “form follows force” principle. However, all these
three groups of physical models maintain a pure tension state subject to certain loads and boundary conditions.
When they are used as shapes of shells after some required measures (e.g. inversion, scaling proportionally,
section design) and construction, the shells will maintain a high structural efficiency (Li et al, 2017).

Fig.9. One Binishell in Ku-ring-gai High School(Li et al,2017)

Fig.10. An Ice Dome at Tomamu in Hokkaido, Japan(Li et al,2017)

CONCLUSION
Form-finding of shells generated from three kinds of physical models, which are hanging models,
tension models and pneumatic models is introduced in this paper. This paper give a brief overview of some of
the possibilities created by physical models in the design of shell structures (Fig.11)

http://cau.icnf.ir
Fig.10. Evolution of physical models of shell form finding
REFERENCES

[1] Addis, Bill, (2013).'Toys that save millions' - A history of using physical models in structural design,
Structural Engineer.
[2] Adriaenssens S, Block P, Veenendaal D, et al. (2014) Shell structures for architecture: form-finding and
optimization. London: Routledge.
[3] Chilton, J.(2000) The Engineer’s Contribution to Contemporary Architecture – Heinz Isler, London,
Thomas Telford Ltd, pp. 35-47.
[4] Chilton J., (2000). Heinz Isler: Engineer’s Contribution to Contemporary Architecture, Thomas Telford
Ltd.
[5] Chilton, J.C., (2009). 39 etc...: Heinz Isler’s infinite spectrum of new shapes for shells. In Lázaro,
ADC., ed. Evolution and Trends in Design, Analysis and Construction of Shell and Spatial Structures,
Editorial de la UPV (Universidad Politécnica de Valencia), Valencia, Spain, pp51-62.
[6] Graefe R. (1986) ‘Zur Formgebung von Bögen und Gewölben’. Architectura - Zeitschrift für Geschichte
der Baukunst, 16, pp. 50-67.
[7] Heyman, J.: (1998) Structural Analysis: A Historical Approach, Cambridge, Cambridge University Press,
pp. 79-80
[8] Huerta, S. (2003). “El calculo de estructuras en la obra de Gaudí.” Ingeniera Civil, 130, pp. 121-133.
[9] Huerta, S. (2006). Structural Design in the Work of Gaudi, Architectural Science Review. 49(4), 324-39.
[10] Isler H.(1961) New Shapes for Shells, Bulletin of the International Association for Shell Structures;
123-130.
[11] Lewis, W.J. (2003). Tension structures. Formand Behaviour. Thomas Telford,London.
[12] Li, Qingpeng , Su Yan, Wu Yue, Borgart Andrew, Rots Jan G.( 2017). Form-finding of shell structures
generated from physical models, International Journal of Space Structures, Vol. 32(1) 11 –33.
[13] Magrone, P, Tomasello, G, Adriaenssens, S, Gabriele, S, Varano, V.(2016). Revisiting the form finding
techniques of Sergio Musmeci: The bridge over the Basento River, Structures and Architecture. Cruz
(Ed), Taylor & Francis Group, London, 543-550.
[14] Ramm E.(2011). Heinz Isler Shells - The Priority of Form, Journal of the International Association for
Shell and Spatial Structures, IASS, Madrid, Vol. 52 (3), No. 169, pp.143-154.
[15] Ramm E., Wall W.A. (2004). Shell Structures. International Journal for Numerical Methods in
Engineering. 60(1), 381-427.
[16] Schek HJ. The force density method for form-finding and computations of general networks. Comput
Method Appl M 1974; 3: 115–134.
[17] Tamai, hiroki (2005). Development Of Form Finding Of Tensile Structures: An Architectural
Approach, Illinois Institute of Technology.
[18] Trovalusci, P., Tinelli, A. (2010). Structural Optimization Vs. Shape Design. International Symposium
on the Tectonics in Architecture: between Aesthetics and Ethics, ICSA.

http://cau.icnf.ir

You might also like