Research Paper

Original Paper
Forma, 15, 113119, 2000
Basic Consideration of Structures with Fractal Properties

and Their Mechanical Characteristics
Naoko KISHIMOTO* and M. C. N ATORI
The Institute of Space and Astronautical Science, 3-1-1 Yoshinodai,
Sagamihara, Kanagawa 229-0022, Japan
E-mail: [email protected]
(Received January 7, 2000; Accepted March 8, 2000)
Keywords: Space Structures, Self-Similarity, Fractal
Abstract. The purpose of this study is to clarify the mechanical characteristics of

structures with fractal properties. This clarification is necessary, in order to apply fractal
notion to artificial structures. Very few studies have been made from the viewpoint of
such attempts. First, fractal systems are formulated for engineering applications, and an
example of 3 dimensional structures with fractal properties is presented. Then the model
and method of the analysis are introduced. Results of the analysis indicate that both how
to change forms against applied forces and how to divide their forces reflect geometrical
shapes. Especially in the case of the structures with fractal properties, the levels of
internal forces are not much different from the regular structures. And it can be said that
the structures with fractal properties can absorb stress unbalance caused by the change of
local stiffness.
1. Introduction
There are many shapes in the natural world. They look so complex that we could not
express them by several simple functions. But after B. B. Mandelbrot devised the notion
of fractal (MANDELBROT, 1977), we are able to express many natural shapes by some
simple algorithms. From that time, hard- and soft-ware of computer systems have been
improved day by day. And various researches on how to construct fractal shapes or how to
imitate natural creatures have been developed (B ARNSLEY, 1988). Recently fractal notion
has been used for image compression, analysis on non-linear dynamics, and so on. But they
are not sufficient in order to study why we can see fractal shapes widely in the natural world
and to study how to apply fractal notion to our actual engineering structures, because they
usually apply fractal notion only as some measures of shapes.
Fractal is geometrically defined by its self-similarity or self-affinity, which means
that one partial shape can be mapped to the whole, or vice versa. The fractal systems
*Doctoral candidate of the University of Tokyo, Graduate School of Engineering.

113
114
N. K ISHIMOTO and M. C. NATORI
provide the following useful characters for the actual engineering structures.
1. Fractal systems can be expressed with some simple algorithms not with complex
functions.
2. Fractal systems consist of some unit elements, and form some stratified systems.
The former is effective especially for programming on computers, and the latter may be
useful to construct large structures. In this paper, behaviors of the structures with fractal
properties are analyzed based on these characters of fractal systems.
2. Definition of Fractal Systems
First of all, it is helpful to formulate fractal systems. It has a connection with how to
construct fractal systems. Here we adopt the method using generator. Let one unit consist
of n elements, P1, P2, , Pn. Using assembling rule A, a first generation G1 (equal to the first
unit, called a generator) is denoted by G1= A(P1, P2, , Pn). Next generation G2 consists
of n G1s assembled according to the rule A. Also a certain generation consists of n previous
generations according to the same rule A. This formulation can be expressed as follows:
G1 = A( P1 , P2 , L, Pn )
(
= A( A
G 2 = A GP11 , GP1 2 , L, GP1n

P1
( P1, P2 ,L, Pn ),L, AP ( P1, P2 ,L, Pn ))
G = A GPk11 , GPk21 , L, GPkn1

k
( (
= A AP1 GPk1 2 , L, GPkn 2 , L, APn GPk1 2 , L, GPkn 2

=
(1)
))
k 1
where GPn
is a (k1)-th generation Gk1 on an element Pn of a k-th generation Gk, and Apn
1
is assembling rule on an element Pn. It should be added that fractal systems are limited in
this paper.
1. Fractal systems have strict the self-similarity, which means that a part of some
generation takes the similar shape to the previous generation.
2. The number of mappings is finite.
3. Self-contact, self-intersection and self-overlap are allowed.
The first is useful to construct the structures made of same units. The second is inevitable
to the actual structures. The third is also inevitable to avoid unstable states.
3. 3-D Spatial Structures with Fractal Properties
3.1. Construction of structures with fractal properties
According to the definition stated above, one of famous fractals, which is well known
as a Koch curve, can be constructed using generators. In this case, elements of a generator
Basic Consideration of Structures with Fractal Properties and Their Mechanical Characteristics
115
are 1-d (1-dimensional) lines and a generator itself is a curve in 2-d plane. In general the
total length of all elements becomes longer according to the number of generation. But let
the total length of each generation be the same. It can be seen that one line is folded like
a string (Fig. 1 (a)). Other kinds of Koch curves, which correspond to some different shapes
of generators, are shown in Fig. 1 (b). Fractal dimension of each curve is appended in the
figure. In brief, it can represent how complex the fractal shape is. As the fractal dimension
of Koch curve becomes larger, it can be seen that one line is folded like a string in the
different way from in Fig. 1 (a).
3.2. 3-D Spatial structure with fractal properties
Considering the possibility of actual structures with fractal properties, we propose an
example of 3-d spatial structures constructed by 3-d generators with 2-d elements (Fig. 2).
In this case an element of a generator is 2-d regular triangle, and a generator consists of the
6 elements. Some generator is on each element of the next generation. Clearly many edges
overlap each other. 3-d spatial truss structures with fractal properties can be constructed
considering that all edges are some kind of truss members. Besides it forms a part of
octahedral trusses which are conventional truss structures (NATORI et al., 1985).
Fig. 1. 2-d Koch curves with constant total lengths.
Fig. 2. 3-d spatial structure with fractal properties.
116
4. Mechanical Analysis on Structures with Fractal Properties

4.1. Method and models for analysis
Mechanical analysis is discussed in this section. Sub-structuring methods are suitable
for analysis on structures with fractal properties, since these structures consist of stratified
sub-structures. In this research the direct flexibility method (P ARK and FELIPPA, 1998) is
used, which is not only one of sub-structuring methods but also suitable for parallel
computing. Parallel computing was not used this time, since models are so simple.
The model of the structure with fractal properites is the 2-d truss shaped SierpinskiGasket made by regular triangles (Fig. 3). The model shown in Fig. 3 (a) is the forth
generation shape of Sierpinski-Gasket, and we call this model the Sierpinski-Gasket truss.
In contrast, the complete planer truss shown in Fig. 3 (b) is analyzed by the same method.
The quantities of elements are assumed to be those of steel shown in Table1. And three
cases of applied forces and four cases of boundary conditions are considered (Fig. 4). Let
all applied forces be 1N step loads.
4.2. Results of the analysis
The representative results are shown in the following figures for the cases a and c, and
cases 3 and 4. The results of other cases have been shown in KISHIMOTO (1999). The
displacements of nodes are shown in Fig. 5. The scale indicates the unit value of arrows.
The unit length of arrows is equal to the length of an element of a triangle in each figure.
It can be seen that displacements of the structures with fractal properties are larger than
displacements of the complete planer truss. But both levels are not so much different that
Fig. 3. Models of a Sierpinski-Gasket truss and a complete planer truss.
Table 1. Quantities of models.
Young modulus; E
2.16 1011 N/m2
Section area of element; A
22 106 m2
0.1 m
7.86 103 kg/m3
Length of element; l
Density of element;
117
they can be expressed on the same scale.

The distributions of internal forces of truss members are shown in Fig. 6. Levels of
internal forces are not so different among these models, in spite that Sierpinski-Gasket truss
has fewer elements. It should be noted that some part of Sierpinski-Gasket truss (circled
part) shows the similar force distribution even under different cases.
Fig. 4. Applied forces and boundary conditions.
Fig. 5. Displacements; Case 3c and Case 4a.
Fig. 6. Internal forces; Case 3c and Case 4a.
118
4.3. Results of the models with random errors on elements lengths

Errors of elements lengths seem to be avoidable in actual structures. Random errors
with normal distribution N(0, 0.052) are assumed on all elements. The differences of
displacements of nodes between the models with no error and those with random errors are
shown in Fig. 7. The differences of Sierpinski-Gasket trusses are larger than those of
complete planar trusses. But they can be denoted on the same scale.
The differences of internal forces among these models are shown in Fig. 8. The
differences of complete planar trusses are 100 times larger than those of Sierpinski-Gasket
trusses. And they can not be denoted on the same scale.
4.4. Consideration to the results of analysis
The whole stiffness of Sierpinski-Gasket trusses is less than that of complete planar
trusses. Displacements from initial to equilibrium positions of the former trusses are larger
than those of the latter. But there is no so much difference on levels of internal forces
between two models. Levels of internal forces are important to design the elements of
structures. Because Sierpinski-Gasket trusses have fewer elements, it can lighten the total
weight that levels of internal forces of both models are nearly the same. This characteristic
comes from their gaps; it is the most typical feature of fractal.
Other feature of fractal is the scaling, which means that its characteristics are
independent of its size. From this feature the way to divide applied forces is different
between the model with fractal properties and the model of conventional trusses. In
Fig. 7. Differences of displacements between the models with no errors and those with random errors.
Fig. 8. Differences of internal forces between the models with no errors and those with random errors.
119
Sierpinski-Gasket trusses the elements transmitting forces are limited (Fig. 6). This
characteristic seems to come from its self-similarity. And it may be suitable for decentralized
control for its shape, because the elements under control can be limited.
The degree of stability of complete planar trusses is far more than that of SierpinskiGasket trusses. Sierpinski-Gasket trusses are almost statically determinate. From these
results, it can be said that they easily absorb stress unbalance under a few changes of
deformations. This properties remarkably appears for the models with random errors and
with local defects (K ISHIMOTO, 1999).
5. Conclusion
The characteristics of planar truss structures with fractal properties are clarified.
They have some ability of averaging stress unbalances, and their internal force distributions
reflect the self-similarity nature. This might be a first attempt to analyze the mechanical
behavior of structures with fractal properties and a first step to apply useful characters of
the natural world to our artifacts.
REFERENCES
B ARNSLEY, M. F. (1988) Fractals Everywhere, Achademic Press.
KISHIMOTO , N. (1999) A study on the construction and the mechanical characters of structures with fractal
properties, Master Thesis, The Univ. of Tokyo (Dept. Aeronautics and Astronautics).
M ANDELBROT, B. B. (1977) The Fractal Geometry of Nature, W. H. Freeman and Company.
NATORI , M. and M IURA, K (1985) Deployable structures for space applications, AIAA85-0727, AIAA/ASME/
ASCE/AHS 26th Structures, Structural Dynamics and Materials Conf., Orlands.
P ARK, K. C. and F ELIPPA, C. A. (1998) A variational framework for solution method developments in structural
mechanics, Trans. of the ASME, Vol. 65.

Research Paper

Uploaded by

Copyright:

Available Formats

Research Paper

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Research Paper

Uploaded by

Copyright:

Available Formats

Original Paper

Forma, 15, 113119, 2000

Basic Consideration of Structures with Fractal Properties

(Received January 7, 2000; Accepted March 8, 2000)

Keywords: Space Structures, Self-Similarity, Fractal

Abstract. The purpose of this study is to clarify the mechanical characteristics of

*Doctoral candidate of the University of Tokyo, Graduate School of Engineering.

N. K ISHIMOTO and M. C. NATORI

G 2 = A GP11 , GP1 2 , L, GP1n

( P1, P2 ,L, Pn ),L, AP ( P1, P2 ,L, Pn ))

G = A GPk11 , GPk21 , L, GPkn1

= A AP1 GPk1 2 , L, GPkn 2 , L, APn GPk1 2 , L, GPkn 2

Fig. 1. 2-d Koch curves with constant total lengths.

Fig. 2. 3-d spatial structure with fractal properties.

N. K ISHIMOTO and M. C. NATORI

4. Mechanical Analysis on Structures with Fractal Properties

Fig. 3. Models of a Sierpinski-Gasket truss and a complete planer truss.

Table 1. Quantities of models.

2.16 1011 N/m2

Section area of element; A

they can be expressed on the same scale.

Fig. 4. Applied forces and boundary conditions.

Fig. 5. Displacements; Case 3c and Case 4a.

Fig. 6. Internal forces; Case 3c and Case 4a.

N. K ISHIMOTO and M. C. NATORI

4.3. Results of the models with random errors on elements lengths

You might also like