See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/341179278

A New Form-finding Method for Shell Structures Based on BESO Algorithm

Conference Paper · October 2019

CITATIONS READS

3 1,066

5 authors, including:

Xin Yan Ding Wen Bao

Chinese Academy of Sciences RMIT University
6 PUBLICATIONS   6 CITATIONS    13 PUBLICATIONS   6 CITATIONS   

SEE PROFILE SEE PROFILE

Yi Min Xie
RMIT University
432 PUBLICATIONS   16,205 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Optimal Design of Energy Absorption Structures with Origami Patterns View project

Auxetic metamaterials and structures: design, fabrication, numerical and experimental investigation View project

All content following this page was uploaded by Ding Wen Bao on 12 November 2020.

The user has requested enhancement of the downloaded file.

 Article published in

Proceedings of IASS Annual Symposium 2019 (17), 1-8

A new form-finding method for shell structures based on BESO

algorithm
Xin YANa,b, Ding Wen BAOa, Kun CAIa, Yu Fang ZHOUb,c, Yi Min XIEa,d,*
a,
* Centre for Innovative Structures and Materials, School of Engineering, RMIT University,
Melbourne, 3001, Australia
[email protected]
b
Centre of Architecture Research and Design, University of Chinese Academy of Sciences,
Beijing, 100190, China
c
School of Architecture, Central Academy of Fine Arts
Beijing, 100102, China
d
XIE Archi-Structure Design, Shanghai, 200092, China

Abstract
Due to high efficiency and elegant geometry, shell structures are widely used in architectural and
structural designs. This paper presents a new numerical form-finding method for the design of shell
structures based on the bi-directional evolutionary structural optimization (BESO) algorithm. This
technique can perform topology optimization to find new forms of shell structures with high structural
performance, elegant geometries, and other features that are of interest to architects and engineers.
Diverse designs from the optimization process as a result of using different parameters in BESO are also
discussed. Finally, the innovative results produced by the BESO method are compared with a well-
known shell structure, the Palazzetto dello Sport, designed by Nervi.
Keywords: BESO, shell structure, topology optimization, form-finding

1. Introduction
With the fast development of computational technics, shell structure with complex geometry has made
an appeal to architects because of its characteristic of high structural efficiency and more elegant shape
potentials. Many engineers and architects have engaged into researches of shell form-finding methods,
e.g. force density method [1] , thrust network analysis [2], dynamic analysis [3], particle spring system
[4] and so on. Instead of concerning the continuum shell surface, topology optimization method pays
more attention to topology relationship of shell surface, which means that the functional cavities, e.g.
windows and air outlets, can be considered as well as structural performance in the form-finding process.
And the result is not only a beautiful pattern but also a structural design diagram for us to clarify where
the frameworks should be denser and thicker or vice versa.
The aim of topology optimization is to find a form with the maximal structural performance under a
certain constraints. Several topology optimization methods have been developed over the past three
decades, such as evolutionary structural optimization (ESO) and bi-directional evolutionary structural
optimization (BESO) [5,6], the homogenization method [7,8], the solid isotropic material with pena-
lization (SIMP) method [9,10] and the level-set method [11,12]. There was one recent study which
applied the homogenization method and obtained interesting topologies of shell domes [13]. In the
present work, we develop a new form-finding method for shell structures based on the BESO algorithm
by taking advantage of its simplicity and effectiveness.
2. Methodology
To realize the shell structure design, BESO method is introduced to optimize the shell structurally. In
the BESO method, the basic problem can be described mathematically as follows:
N
1 1
min C ( X )  U T KU 
2 2
x
i 1
p T
i ui ki ui (1)

N
Subject to: V ( X )  x v
i 1
i i V* (2)

in which C, U, K, 𝑉 ∗ are the objective function (compliance), displacement vector, global stiffness
matrix and the objective volumn, respectively. The terms 𝑣𝑖 , 𝑥𝑖 , 𝒌𝑖 and 𝒖𝑖 are the volumn, design
variable, stiffness matrix and nodal displacement vector for i th element. Particularly, it’s necessary to
point out that there are two alternative values for 𝑥𝑖 in BESO, which are 1 for solid element or prescribed
value 𝑥𝑚𝑖𝑛 for void element.
For stiffness optimization, the sensitivity 𝛼𝑖 for i th element, which is the criterion for desgin variable
𝑥𝑖 , can be calculated as the gradient of compliance with respect to the design variable [14],
C ( X ) 1
  pxip1uiT ki ui
xi 2 (3)

 1 T
u ku when xi  1
1 C  2 i i i
i     p1 (4)
p xi  xmin T
ui ki ui when xi  xmin
 2
And when the penalty coefficient p tends to infinity, the sensitivity number 𝛼𝑖 becomes the one in hard-
kill BESO method[14],

1 T
1 C  u ku when xi  1
i    2 i i i (5)
p xi 
0 when xi  xmin  0
The above sensitivity is usually modified to solve the mesh-dependent problem [15,16], using a filtering
scheme:
N

w  ij j

~i  j 1
N (6)
w
j 1
ij

wij  max(0, rf  d ij ) (7)

in which 𝑑𝑖𝑗 is the distance between the center of the j th element and the i th element and 𝑟𝑓 , 𝛼𝑗 is the
filter radius and the original sensitivity of the j th element.
To achieve a convergent solution, another historical average of 𝛼̅𝑖 in different iterations is introduced
[17]:

2
 ~i( n )  ~i( n1)
i  (8)
2
In the BESO method, the element sensitivities are ranked in each iteration to determine a threshold with
target volumn of next iteration, 𝑉 (𝑛) , which is defined based on the current volumn 𝑉 (𝑛−1) and the
evolutionary ratio 𝛿.

V ( n )  V ( n1) (1   ) (9)
The threshold can be used to evaluate if the element shall be changed in a such way that if one solid
element’s sensitivity is lower than the threshold, its design variable will be switch from 1 to 𝑥𝑚𝑖𝑛 , and
the design variable of a void element will be changed from 𝑥𝑚𝑖𝑛 to 1 as well if its sensitivity is higher
than the threshold.

3. Shell form-finding based on BESO method

In this work, a shell structure composed of shell elements is introduced (Fig. 1). The objective of the
problem is to find the optimal topology of the structure withstanding its self-weight. The whole
structure’s radius in the xy plane is 30 m, and the distance from the top point to xy plane is 4.5 m. The
thickness of shell element is 0.3 m. The following material properies are assumed: Young’s modulus
200 GPa, Poisson’s ratio 0.3 and the desity 7800 kg/m3. Considering symmetric geometry and boundary
conditions of the structure, only a one-eighth part with an angle of 45o is analysed, and 3 points at the
bottom boundary are fixed (Fig. 1). The total number of mesh nodes and elements are 16,396 and 32,230,
respectively.

Figure 1: The shell model and boundary condition

To achieve diverse designs from the topology optimization process, several controlled trials with
different BESO parameters are set up. The parameters for BESO algorithm are as evolution ratio (ER)
= 6%, 4%, 2%, filter radius (FR) of 30 mm (3 times of the element size) and volume fraction (VF) =
30%. The results are given below (Figs. 2-4).

3
 Figure 2: The BESO result of shell structure under self-weight load (ER =6%, FR =30 mm, VF =30%)

Figure 3: The BESO result of shell structure under self-weight load (ER =4%, FR =30 mm, VF =30%)

Figure 4: The BESO result of shell structure under self-weight load (ER =2%, FR =30 mm, VF =30%)

From the above BESO results, although the forms vary slightly, the differences of structural stiffness
among them are just within 0.4%, which means all of the structures share the almost same structural
perfomance.

4
4. A numerical example compared with Palazzetto dello Sport
The Palazzetto dello Sport, designed by Pier Luigi Nervi and Annibale Vitellozzi [18], is a concrete
shell characterized by its sixty-meter diameter, 36 Y-shape struts and the elegant interior ribbings with
Michell truss pattern (Fig. 5). In this part, self-weight and torsion are mainly taken into consideration in
the BESO process for the reason that they are the main load cases for shell structures.

Figure 5: The design drawings of Palazzetto dello Sport

(Picture source: http://www.archidiap.com/opera/palazzetto-dello-sport/)

From the BESO result of the model with the only load of self-weight in z direction, the Y-shape struts
and central hole with the similar size as Nervi’s design are also generated in BESO result (Fig. 6). In
Nervi’s design, the strucutres between the interior ribbings (the red line in Fig. 6) and the bottom
supports (the blue line in Fig. 6) are redesigned as a transitional structure to amplify the number of fixed
points from 36 to 108.

Figure 6: The original design condition and the BESO result of shell structure under self-weight

However, there is a significant difference of the interior ribbings’ pattern between Nervi’s design and
the BESO result. This is because that the load of torsion has not been introduced. Another calculation is
implented on the central part (Fig. 7). The load is a torsion on the boundary of the top hole. The number
of fixed points and the volume fraction for BESO are set as 36 and 40% respectively becuse of the

5
limitation of element number in FEA. It is clearly seen that the optimized design from the BESO method
has a similar pattern to Nervi’s design.

Figure 7: The original design condition and the BESO result of shell structure under torsional load

Furthermore, a new structure for this model can be obtained with BESO to exert these two load cases at
the same time. The relative weight for these two load cases are set equally. The BESO parameters are
evolution ratio (ER) =2%, filter radius (FR) of 30 mm (3 times of the element size), volume fraction
(VF) = 40%.

Figure 8: The original design condition and the BESO result of shell structure with multiple load cases

5. Conclusion
Most traditional form-finding methods for shell structures usually focus on shape rather than topology
optimization. In this work, a new BESO method is developed to help engineers and architects to perform
conceptual design of shells with complex geometries and high structural performance. Using this

6
method, it is also possible to obtain diverse designs of shell structures with similar structural perfor-
mance. Different BESO parameters and different loading conditions are effective tools for form-finding
in the early stage of the conceptual design, and designer could get many inspirations from these results.
It is worth noting that the form-finding method developed in this study can be applied to structures other
than shells.

Acknowledgements
The authors would like to thank several colleagues whose support helped fulfill the research project
described in this paper:
Mr Yulin Xiong and Mr Yunzhen He (Centre for Innovative Structures and Materials, RMIT)

References
[1] H. J. Schek, “The force density method for form finding and computation of general networks,”
Computer Methods in Applied Mechanics and Engineering, vol. 3, no. 1, pp. 115–134, 1974.
[2] P. Block and J. Ochesendorf, “Thrust network analysis: A new methodology for 3D equillibriu-
m,” Journal of the International Association for Shell and Spatial Structures, vol. 48, no. 155,
pp. 1–7, 2007.
[3] M. R. Barnes, S. Adriaenssens, and M. Krupka, “A novel torsion/bending element for dynamic
relaxation modeling,” Computers & Structures, vol. 119, pp. 60–67, 2013.
[4] A. Kilian and J. Ochsendorf, “Particle-Spring Systems for Structural Form Finding,” Journal of
the International Association for Shell and Spatial Structures, vol. 148, no. 2, p. 77, 2005.
[5] Y. M. Xie and G. P. Steven, “A simple evolutionary procedure for structural optimization,”
Compurers & Structures, vol. 49, no. 5, pp. 885–896, 1993.
[6] X. Huang and Y. M. Xie, Evolutionary Topology Optimization of Continuum Structures,
Methods and Applications. Chichester: Wiley, 2010.
[7] G. Carotenuto, L. Nicolais, B. Martorana, and P. Perlo, “Optimal shape design as a material
distribution problem,” Structural Optimization, vol. 28, no. 12, pp. 384–388, 2003.
[8] Martin Philip Bendsoe and Noboru Kikuchi, “Generating optimal topologies in structural design
using a homogenization method,” Computer Methods in Applied Mechanics and Engineering,
vol. 71, pp. 197–224, 1988.
[9] M. P. Bendsøe and O. Sigmund, Topology Optimization: Theory, Methods, and Applications.
Berlin: Springer, 2003.
[10] M. P. Bendsøe and O. Sigmund, “Material interpolation schemes in topology optimization,”
Archive of Applied Mechanics, vol. 69, no. 9–10, pp. 635–654, 1999.
[11] S. Y. Wang, K. M. Lim, B. C. Khoo, and M. Y. Wang, “An extended level set method for shape
and topology optimization,” Journal of Computational Physics, vol. 221, no. 1, pp. 395–421,
2007.
[12] M. Yulin and W. Xiaoming, “A level set method for structural topology optimization and its
applications,” Advances in Engineering software, vol. 35, no. 7, pp. 415–441, 2004.
[13] S. Adriaenssens, P. Block, D. Veenendaal and C. Williams (eds.), Shell Structures for Archite-
cture: Form Finding and Optimization. London: Routledge, 2014.
[14] X. Huang and Y. M. Xie, “A further review of ESO type methods for topology optimization,”
Structural and Multidisciplinary Optimization, vol. 41, no. 5, pp. 671–683, 2010.
[15] O. Sigmund and J. Petersson, “Numerical instabilities in topology optimization: A survey on
procedures dealing with checkerboards, mesh-dependencies and local minima,” Structural
Optimization, vol. 16, no. 1, pp. 68–75, 1998.

7
 [16] O. Sigmund, “Morphology-based black and white filters for topology optimization,” Structural
and Multidisciplinary Optimization, vol. 33, no. 4–5, pp. 401–424, 2007.
[17] X. Huang and Y. M. Xie, “Convergent and mesh-independent solutions for the bi-directional
evolutionary structural optimization method,” Finite Elements in Analysis and Design, vol. 43,
no. 14, pp. 1039–1049, 2007.
[18] T. Iori and S. Poretti, “Pier Luigi Nervi’s works for the 1960 Rome Olympics,” Actas del Cuarto
Congreso Nacional de Historia de la Construcción, pp. 27–29, 2005.

View publication stats