Parametric Design Process of Facade Elements with
Characteristics of Fractal Geometry: Development,
Evaluation and Application
XXIV International Conference
of the Iberoamerican Society
of Digital Graphics
Medellín | Colombia
Pedro Oscar Pizzetti Mariano
Universidade Federal de Santa Catarina | Brasil | pedro.pm@hotmail.com
Alice Theresinha Cybis Pereira
Universidade Federal de Santa Catarina | Brasil | alice.cybis.pereira@ufsc.br
Abstract
This article deals with the use of the characteristics of fractal linear geometry and the different
knowledge related to the application of its patterns in architectural elements, considering their
compositional potentialities. For this, a theoretical and practical trajectory was developed, such
as understanding the characteristics of fractal geometry and the existing software and methods
for its reproduction. As a result, a parametric process was developed that allows to recreate
fractal patterns in architectural elements, making it possible to identify the necessary steps for
their elaboration, identifying the potentialities of their use and the skills necessary to reproduce
similar parametric processes.
Keywords: Fractal geometry; Dimension D; Parametric process; Design Process.
INTRODUCTION
This article deals with the use of the characteristics of linear
fractal geometry applied in patterns of architectural
elements. And it seeks to analyze the parametric process
that makes this action possible, identify the stages of its
elaboration, describe the potential of its use, and the skills
necessary to reproduce similar processes. For this, a
theoretical and practical trajectory was developed. Who’s
theoretical was analyzed the fractal patterns, their
characteristics and applications, in addition to identifying
software and processes that were used to reproduce the
fractal characteristics. In the practical part, parametric
programming was developed and the result was analyzed.
Using the patterns of nature during the design process can
be an opportunity to approximate the construction,
structure and geometry of the ornamentation and the user's
scale. An example of this kind of connection between
geometry, technology, and nature is the use of fractal
components and rules through computer software that
allow these patterns to be transformed into architectural
ornaments. One of the possibilities for the insertion of
fractal patterns is the use of computational design and
tools, which allowed changes in the way architecture is
conceived
(SALINGAROS;
MEHAFFY,
2007;
SALINGAROS, 2003).
The possibility of generating patterns, establishing
parameters and rules, directly influenced the composition
and studies involving fractal geometry. Mandelbrot (2004)
argues that theories of fractal geometry largely inspired the
conception of twentieth-century architecture and
generative design.
Mandelbrot (1977) states that a fractal is a structure that
has invariance, regardless of the scale at which it is
observed and is always linked to the same pattern or
shape, keeping its original structure identical. The
development of a fractal and its mathematical patterns are
derived from four main characteristics: self-similarity;
infinite complexity; irregularity or roughness; and a
nonintegrin dimension.
This last fractal characteristic, a nonintegral dimension or
dimension D, described by Mandelbrot (1977), can identify
and measure various fractals. Many fractals do not contain
easily identifiable patterns, and the method described by
Mandelbrot, box counting, is a viable procedure. Spehar et
al. (2003) describe that the fractal dimension or D can be
determined by comparing the number of squares occupied
in a mesh with different scales, where "D" is a value
between 1.0 and 2.0.
This feature is also associated with providing a positive
psychological reading of the space around us. Taylor
(2006) and Joye (2007) comment on a set of possible
reasons that associate fractals with positive human
interpretations. And unconsciously this may have
influenced that the fractal geometry is present in
architecture in their general context or in different details
and volumes.
The use of recent digital tools and methods influences
many professionals in various design situations, allowing,
among other things, the development of sun protection
elements with different and complex shapes. In some
cases, these concepts are inspired by nature to develop
their shapes and structures. The use of fractal geometry or
its features, combined with the use of parametric
algorithms, can assist in the creation and development of
these complex concepts and shapes. Parametric
processes also have the potential to encompass innovative
concepts through the use of modeling and simulation.
(MENGES, 2012; RIAM; ASAYAMA, 2016).
To apply the rules, characteristics and iterations of fractals,
different software and methods can be used. In the
research, we opted for the use of visual programming
software and plug-ins. To model generative processes
143
within the field of architecture and design, one of the
software that is accepted for the development of the
algorithms is the Grasshopper plug-in, a graphic editor
connected to the Rhinoceros 3D modeler. It provides a
variety of mathematical and geometric operations and
commands and can change the built digital model.
(KOLAREVIC, 2005).
Considering these concepts, this paper, which is part of a
master's dissertation, investigates the development of a
process to develop, organize and evaluate a composition
of facade elements with characteristics of linear fractal
geometry.
VISUAL PROGRAMMING
DEVELOPMENT
24th CONFERENCE OF THE IBEROAMERICAN SOCIETY OF DIGITAL GRAPHICS
The present study shows the construction of a parametric
process that involves different methods and concepts,
resulting in a compilation of unified procedures in the
development of a process to develop, organize and
evaluate a composition of facade elements with
characteristics of linear fractal geometry. To achieve this
goal, a series of theoretical and practical actions were
proposed to broaden the knowledge of the subject, acquire
skills to manipulate the tools and interpret their results. The
theoretical stage identified the different subjects for
understanding the process and manipulating the necessary
software. And the practical step was developed through the
development of a parametric and generative design
process, carried out through a visual programming
platform.
For the development of the digital process, the Rhinoceros3D 5 modeling software was used. Within this software, the
MII (incremental interactive model) was used as a method
to build the process described by Silva and Videira (2001).
This method allowed each of the steps developed in the
visual programming to act independently, without requiring
the complete completion of the previous step, for starting
next. The elaboration of the model and of each of its parts
was also performed in the Grasshopper visual
programming software and assisted by the Hoopsnake and
Diva plug-ins.
that is parametrically adjustable and susceptible to the
insolation consequent to its surrounding urban
environment, that also is parametrically modifiable.
For the research was used lot number seven in the central
block (number five) to insert the test building, the site has
the dimensions of 37m x 23.05m and was directed to the
North. The three-dimensional model to the test was molded
from a central point on the chosen area, being possible to
change different parameters as: width of the tested; length;
floor height; width of slabs; and number of floors.
The composition of the urban environment has some
limitations, such as: it is only possible to manipulate and
modify nine blocks (3x3); one can modify its general
dimensions, but orthogonally, without changing the
perpendicular angles or using curved drawings; the
dimensions of the blocks are generated from the division of
the terrain and the sizing of the infrastructure. In the figure
1 and 2 presents the two environments used for research
with the insertion of the construction in the middle of the
center block.
Figure 1: Parametrically adjusted urban environment to have
level relief and constructions below 35m, seeking to maintain the
central block of the urban environment provided with daylight.
The whole process was subdivided into different stages,
each of which also presents its divisions. The first stage of
the process is divided into two parts, the assembly of the
environment and the construction of the building. The two
parts contain a series of steps that result in a threedimensional environment with nine parameterized blocks,
with constructions, urban infrastructure and a space where
a testing building will be inserted and in which the facade
elements will be applied.
At the first step, a three-dimensional urban environment
was developed to simulate an environment suitable for
parametrical modifications. With this is crated a set of nine
orthogonal urban blocks is formed, where it is possible to
manipulate the topography, the number of lots per block
(individually), the infrastructure (pathways), and the
different characteristics of the constructions and scales. In
the second stage of the development of the environment, a
construction is modeled and inserted which will receive the
compositions of the panels, at the previously designated
site. In the end of this steps, a building model is created
Figure 2: Parametrically adjusted urban environment to have
uneven relief and constructions above 30m, seeking to increase
the area of the central area of the model.
In the sequence the facade elements with fractal
characteristics, that composed the main façade of the
modeled building were developed. This stage of the
process is divided into three parts: the design of panels with
elements of fractal geometry; the analysis and selection of
144
these panels and their respective iterations; and the
optimization of the main elements.
The first stage of the confection and modeling of the panels
with fractals is divided into two parts: the development and
iteration rule for the façade elements; and the twodimensional modeling of these elements. For the
development of each of the panels were used ten rules of
Linear fractals, such as Square of Gasket; Jerusalem
Cross; Sierpinski carpet; Cantor Set; Sierpinski Triangle;
Minkowski's curve; Peano's curve; Dragon Curve; Fractal
Tree; and Koch Island. Two examples of those rules can
be observed by figure 3 and 4, which presents the
formation pattern for the fractal Cantor Set and Minkowski
Curve (MARIANO; PEREIRA; VAZ, 2018).
Figure 3: Example of application of fractal rules for the Cantor
Set.
Figure 4: Example of application of fractal rules for the
Minkowski Curve.
In the second step, the rules for generating the fractals
were inserted into standard panels of 3m x 3m, and iterated
four times, to generate five distinct panels, each with a
different level of iteration. Thus, five different panels were
generated for each type of fractal, totaling 50 panels. Each
group of panels has its first facade element completely
opaque, with no subtraction, and as iterations undergo,
each rule subtracts a certain area. The Hoopsnake plug-in
was used for modeling those patterns, which allowed
iteration of the rule of each fractal pattern over and over
again. As an example of the construction of these patterns
the figures 5 and 6 presents the result of the modeling of
the facade elements to the Cantor Set and Minkowski until
its fourth iteration (MARIANO; PEREIRA; VAZ, 2018).
Figure 5: Panels with the Cantor Set, with percentage of the cast
area for each iteration.
to identify the results of the luminous behavior of each
panel group and to verify the different iterations of each
pattern. This evaluation of the daylight behavior, allowed to
identify and selected the patterns with the best daylight
performance (MARIANO; PEREIRA; VAZ, 2018).
To analyze the light results of each panel, a virtual
environment was modeled with the following dimensions:
6m deep, 6m wide and 3m high. The model was adjusted
so that the facade could be composed of a total of 8 panels,
each of them as the same dimension (1.5 x 1.5m). The
indexes, used to verify the light behavior of the
environment, followed the models of the IES (Illumination
Engineering society), and three indexes were measured:
Useful Daylight Illuminance (UDI), SDA (spacial daylight
autonomy) and ASE (annual sunlight exposure). Dynamic
simulations were done using the DIVA as a plug-in for
Grasshopper
and
using
the
BRA_FLORIANOPOLIS838990_SWERA climatic archive
was used, referring to the city of Florianópolis, SC, Brazil.
Each composition was evaluated twice in different
positions. In total, 100 simulations were run, ten for each
fractal geometry pattern (MARIANO; PEREIRA; VAZ,
2018).
And the panels that presented better daylight performance
through the daylight selection capability, in different
proportions for each iteration, were the elements
developed with the fractal rules: Sierpinski Carpet,
Minkowski Curve, Dragon Curve and Singer Set. These
elements have demonstrated that they have different light
filtrations for each iteration (MARIANO; PEREIRA; VAZ,
2018).
The panels that presented better luminous performance
through the daylight selection capability, in different
proportions for each iteration, were the elements
developed with the fractal rules: Sierpinski Carpet,
Minkowski Curve, Dragon Curve and Singer Set. These
elements have demonstrated that they have different light
filtrations for each iteration. (MARIANO; PEREIRA; VAZ,
2018).
Afterwards, the four panels identified in the previous step
received lighting control elements, and had their sizing
adjusted to evenly distribute daylight in the indoor
environment. This stage is divided in two others: the
development and modeling of the components; and its
simulation in a controlled environment through a generative
process, with the Galapagos component.
The first part of this process step is the modeling and
angulation of the different thicknesses that will complement
the panels. Each panel was complemented by a unique
composition of the process, since the different
characteristics of each one of the fractals need specific
alterations so that its parameters are modified. The
alterations comprise the opening angle of the shelves and
their respective length. Those characteristics can be
observed by the figure 7.
Figure 6: Panels with the Minkowski Curve, with percentage of
the cast area for each iteration.
The panels modeled with fractal geometry designs were
submitted to different daylights tests, whose objective was
145
totaling 64 facade elements per floor, with 1.13m wide and
1,125m high (figure 8).
Figure 8: Result of the division of facades
Figure 7: Different views of possible changes in the thickness of
the facade element (open or closed opening angle and extended
or short thickness)
24th CONFERENCE OF THE IBEROAMERICAN SOCIETY OF DIGITAL GRAPHICS
When they are ready, the shelves will be optimized through
the Galapagos component. The optimizations refer to the
distribution of internal light in the environment (UDI) and the
amount of radiation that passes through each of the solar
protection elements.
In order to use the Galapagos component, it was necessary
to model a simulation environment similar to the previous
step, with the dimensions of six meters by six meters and
three meters of height (6m x 6m x 3m). The North facade
was divided into a mesh of three by six, allowing the
insertion of 18 elements equal of one meter by one meter.
The simulations, more time, were performed through the
DIVA and the climatic file used again was the
BRA_FLORIANOPOLIS838990_SWERA.
Each of these spaces is tuned to receive one of the five
iterations of a particular fractal pattern, developed in the
previous steps. The fractals are updated to occupy one of
the positions of the matrix of each floor, automatically
adjusting its size (width, length and depth) to the division
made before, and positioning itself, initially, from the
reading of the radiation in the facade.
For the analysis of radiation, the DIVA plug-in was again
used, evaluating, in this case, the average radiation
incident on each floor. The panels are organized by dividing
their different iterations according to the intensity of the
radiation incident on the facade, placing the panels with
more iterations (more subtraction of the shape) in the
quadrants where the radiation indices are smaller, and the
panels with less iterations (less subtraction in shape) in the
divisions with higher incidence of radiation. Figure 9 shows
an example of this sequence of actions. It is possible to
control the positioning of the different iterations by means
of a value window. These windows, in turn, make it possible
to adjust the intervals that each type of panel occupy.
To equalize the values of the daylight indices, in order to
optimize the results, the recurring responses obtained from
the simulations were numeralized. In this process, the
result of the simulations is transformed into a rational
number between 0 and 1.0. As the Galápagos can only
evaluate a quantitative response, the two resulting values
had to be equivalent, assigning a percentage of 50% for
each one. Thus, the Galápagos could obtain the maximum
value of 1.0 and the simulations of IDUs would reach 0.5
and those of amount of radiation 0.5. For each result does
not exceed the desirable value, UDI 75% and amount of
radiation 55%, wrongly balancing the values, when the
result exceeds the ideal values, it is replaced by 0,
decreasing its final value.
After the organization of the panels, the main curves that
compose them are mapped and extracted, this selection is
made to contemplate the count of the fractal dimension of
the composition. This mapping of the curves that make up
the facade elements is automatic and updated to each type
of fractal used or the composition developed.
In the third part of the process, the union of the two previous
stages takes place. This stage is divided into two phases,
the organization and division of the facade and the insertion
and control of these elements. And it starts with the division
and organization of the spaces that will be filled by the
façade elements developed. Firstly, the facade of the teste
building is divided, individualizing the facade of each floor,
then the facade of the floors is divided into four equal parts.
Next, each of these four spaces is divided again by an
array, allowing the adjustment of the quantity and size of
the facade elements. For this research it was decided to
subdivide each part into sixteen smaller slices (4x4),
Finally, the fractal dimension of the composition of the
panels is verified through box-counting. For this part of the
process, the curves that make up the panels are
superimposed on a mesh that fills the space used for the
facade elements. The process is performed three times
and averaged in a four by four (4x4), eight by eight (8x8)
matrix and a twelve by twelve (12x12) matrix. With this
Counting it is possible to identify if the composition has a
fractal dimension. If the result of the box-counting is upper
of 1.0 the compositions are fractal, and if the values are
closer or between of 1.3 and 1.5 the composition possible
bring positive responses to human perception.
Figure 9: Result of the fill sequence.
146
Figure 11: Structure of the process and its logical sequence.
RESULTS
As a result of the parametric process, elements,
arrangements and simulations were created, able to
develop, potentiate and organize different geometric
elements with geometric characteristics of linear fractals.
The elements created by this process can receive different
attributions when used in façades, becoming an instrument
of identity and composition that looks for possible positive
psychological responses as well as element of shading and
distribution of internal natural light.
During the process verification step, the four different types
of panels were organized with the same compositional
arrangement. It is possible to verify which of the four fractal
patterns present the fractal dimension indices closer to the
range of 1.3D to 1.5D. The results of these organizations
can be observed in Table 1.
Table 1: Results of the fractal dimension.
Fractal
type
D. Fractal
Minkowski
curve
1.08D
Cantor Set
1.12D
Dragon
Curve
1.14D
The numerical interval, used for this first organization, was
adjusted to seek a fractal dimension greater than 1.3D and
to allow the visual contemplation of the environment from
the internal environment. However, in the results of this first
verification (fractal dimension), it can be seen that even
with a larger range for the values, which correspond to the
panels with more openings (more iterated), the filling of the
spaces by box-counting was not enough so as to reach a
size greater than 1,3D.
Evaluating these organizations, it was observed that
different composition intervals are necessary for each
fractal pattern, since their differentiated characteristics and
patterns make their results unique, having a fractal
dimension dependent on the rule and the design that is
chosen. To compare the results in terms of composition, an
image with the two arrangements developed was
humanized, representing the end result of the process
(figure 10).
Sierpinski
Rug
1.25D
147
which if not used, may compromise the interpretation of the
indexes or do not approve its use.
To develop geometries, which present features of fractal
geometry, with different architectural purposes, it is
necessary to deepen the knowledge about the discipline
that will guide its creation and evaluation. So that the other
steps of the process (Organization and Simulation) have
results without errors.
Figure 10: Humanization of two compositions made with the
parametric process, arranged by different radiation indices that
affect the façade.
Also were identify and characterized the steps to develop
the parametric process capable of creating, organizing and
evaluating elements of facade with fractal characteristics,
four different stages were identified in chronological form,
named as: environment; choice; organization; and
simulation. Figure 11 represents the general structure of
the parametric process.
24th CONFERENCE OF THE IBEROAMERICAN SOCIETY OF DIGITAL GRAPHICS
CONCLUSION
With the parametric process was possible to identify some
conclusions. The first is the identification of the steps of the
process resulted in the identification of four major stages
that compose it: Environment, Choice, Organization and
Simulation. Each stage was developed to complete a
singular goal, allowing different changes not to influence
the continuity of the whole. The first step performed
(Environment) enables you to create parametric
environment and building. The second stage (Choice)
develops panels by means of fractal rules, allows to choose
the panels with greater potential, and to add elements and
to evaluate them by means of a genetic algorithm. The next
step (Organization) joins the previous two steps, in addition
to the façade elements. And the last phase (Simulation),
verifies this organization resulting from the complete
process.
This sequence of stages presents different assumptions for
other uses. They have also been shown to be a coherent
sequence for the development of methods and processes
involving fractal geometry, using other features and
patterns of fractal geometry with other purposes like
structure, comfort, etc. Or in the construction of different
parametric processes, involving various forms, uses and
indices (thermal, acoustic, sensorial, physical, etc.).
The potential of the use of fractal geometry as a basis for
the design of different sunscreens allows controlling the
behavior of natural light within the internal environment
through its different iterations. The different adjustments
made in the step of choosing also allowed to extend the
potential of this geometry in the construction of solar
protectors, allowing to develop variations for fractal design,
to create more iterations and to adjust the shelves of light.
Another conclusion reached in the research is about the
development and handling of the process, considering that
it requires a series of specific knowledge (such as:
programming, logical processes, simulation and others),
The work presented limitations, in different stages, related
to the processing capacity of the hardware used to make
the simulations. During the use of Galapagos, the time of
the simulations limited, in some types of fractals, the
number of families generated by the plug-in. In counting the
box counting, the time to count the filled spaces also
prevented the development of a greater number of
compositions. The delay in obtaining the different results of
the process, such as the internal simulations and the fractal
dimension, resulted in difficulties in data collection. At
different times, the process needed to be repeated a few
times for the reading of the indices and figures to be
accurate.
REFERENCIAS
JOYE, Yannick. Fractal Architecture Could Be Good for You.
Nexus Network Journal, [s.l.], v. 9, n. 2, p.311-320, out. 2007.
Springer Nature. http:/dx.doi.org/10.1007/s00004-007-0045-y.
KOLAREVIC, B. Architecture in the Digital Age: Design and
Manufacturing. [s.1.]: Taylor & Francis Group, 2005. 320 p.
MANDELBROT, Benoit B. The Fractal Geometry of Nature. Nova
Iorque: W. H. Freeman and Company, 1977. 480 p.
MANDELBROT, Benoit. Fractals and Chaos: The Mandelbrot Set
and Beyond. [s.1.]: Springer, 2004. 308 p.
MARIANO, Pedro Oscar Pizzetti; PEREIRA, Alice Theresinha
Cybis; VAZ, Carlos Verzola. Avaliação luminosa de elementos
de fachada com características fractais. Parc Pesquisa em
Arquitetura e Construção, [s. l.], v. 9, n. 1, p.3-18, 31 mar.
2018.
Universidade
Estadual
de
Campinas.
http://dx.doi.org/10.20396/parc.v9i1.8650255.
MENGES, Achim. Biomimetic design processes in architecture:
morphogenetic and evolutionary computational design.
Bioinspiration & Biomimetics, [s. l.], v. 7, n. 1, p.1-11, 16 fev.
2012. IOP Publishing. http://dx.doi.org/10.1088/17483182/7/1/015003.
RIAM, Iasef Md; ASAYAMA, Shuichi. Computational Design of a
nature-inspired architectural structure using the concepts of
self-similar and random fractals. Automation in Construction,
[s.l.], v. 66, p.43-58, jun. 2016. Elsevier BV.
http://dx.doi.org/10.1016/j.autcon.2016.03.010.
SALINGAROS, N. A. (2003). The sensory value of ornament.
Communication & Cognition, 36 (3-4), 331-351.
SALINGAROS, Nikos A.; MEHAFFY, Michel W. A Theory of
Architecture. [s.l.]: Intercollegiate Studies Inst, 2007. 278 p.
SILVA, Alberto Manuel Rodrigues da; VIDEIRA, Carlos Alberto
Escaleira. UML - Metodologias e Ferramentas CASE:
Linguagem de Modelação UML, Metodologias e Ferramentas
CASEM na Concepção e Desenvolvimento de Software.
Portugal: Edições Centro Atlântico, 2001. 578 p.
SPEHAR, Branka et al. Universal aesthetic of fractals. Computers
& Graphics, [s.l.], v. 27, n. 5, p.813-820, out. 2003. Elsevier
BV. http://dx.doi.org/10.1016/s0097-8493(03)00154-7.
TAYLOR, R.p. Reduction of Physiological Stress Using Fractal Art
and Architecture. Leonardo, [s. l.], v. 39, n. 3, p.245-251, jun.
2006.
MIT
Press
Journals.
http://dx.doi.org/10.1162/leon.2006.39.3.245.
148