Parametric Design: Theoretical Development

and Algorithmic Foundation for Design

Generation in Architecture

Ning Gu, Rongrong Yu, and Peiman Amini Behbahani

Contents
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Generative Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Common Characteristics of Generative Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Main Generative Design Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Parametric Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Historical Review of Parametric Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Parametric Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Reshaping Architectural Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Impact on Architectural Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Limitations of Parametric Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Cross-References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

Abstract
This chapter presents the theoretical foundation of parametric design for design
generation in architecture. Parametric design has been increasingly applied to
architectural design in recent years. It is essentially a digital design method,
which can be characterized by rule-algorithmic design and multiple-solution
generation. Parametric design originates from generative design, which is a
typical computational design approach based on rules or algorithms (e.g., in
generative grammars or evolutionary systems). This chapter starts with a critical

N. Gu () · P. A. Behbahani
School of Art, Architecture and Design, University of South Australia, Adelaide, Australia
e-mail: [email protected]; [email protected]
R. Yu
School of Engineering and Built Environment, Griffith University, Southport, Australia
e-mail: [email protected]

© Springer International Publishing AG, part of Springer Nature 2018 1

B. Sriraman (ed.), Handbook of the Mathematics of the Arts and Sciences,
https://doi.org/10.1007/978-3-319-70658-0_8-1
2 N. Gu et al.

review of generative design, followed by the background, history, and theory of

parametric design, including various fundamental concepts and applications that
underpin parametric design, and concludes with a discussion of the impact of
parametric design on architecture.

Keywords
Parametric design · Architectural design · Generative design · Mathematics ·
Algorithm

Introduction

Generative design is a general term commonly used to describe a number of com-

putational methods that aim to automate the whole, or a part, of the design process.
Their main purpose has been to assist human designers to explore the design space
with computational means (Herr and Kvan 2007). This is usually realized through
rapid prototyping, powered by computers (Chien 1998). Furthermore, because the
conceptual frameworks and design opportunities they offer are very distinct from
the traditional approaches to design, it is arguable that generative systems have also
created new paradigms or tools for design thinking (Oxman 2006). Shape grammar
is one of the typical generative design methods (see chapter  “Shape Grammars: A
Key Generative Design Algorithm”). As discussed in that chapter, shape grammars,
their generative nature, as well as formal approaches to design representation
and generation, have influenced the emergence and development of contemporary
computational design methods and tools, including parametric design, which is the
focus of this chapter. Deriving from generative design principles, parametric design
has distinctive features and has been widely adopted in leading architectural design
practices.
This chapter is structured into three main sections. The first section summarizes
generative design as the theoretical background for parametric design; the section
introduces generic features of generative systems and reviews key examples of
generative design methods. The next section “Parametric Design” focuses on para-
metric design including its historical development as well as its main components,
applications, and implementation. The section “Reshaping Architectural Design”
discusses the impact of parametric design and how parametric design has reshaped
architecture and is followed by a concise conclusion.

Generative Design

As defined above, generative design refers to a collection of computational methods

or algorithms used to explore the design space for generating solutions. For
this purpose, algorithmic procedures are formally defined and applied to produce
numerous design alternatives, of which the most suitable would be selected (Herr
and Kvan 2007). This definition has been systematically used since the 1990s in the
Parametric Design: Theoretical Development and Algorithmic Foundation. . . 3

computational design domains, although most generative design systems (such as

shape grammars) had been developed decades earlier.
The term generative appeared much earlier in the relevant fields of design and
visual art, at least since 1965, as reported by Boden and Edmonds (2007). It has
gradually evolved to shift or extend its meaning from a description of design
mechanics (as computer-generated design) to a more specific reference to design
paradigms capable of supporting design thinking (Eckert et al. 1999).
Tracing back to its origin, generative design initially emerged in the 1950s–
1960s, inspired by natural phenomena. For example, evolutionary systems or
genetic algorithms were inspired by the processes of biological evolution (Holland
1975), cellular automata were modeled based on biological growth (von Neumann
1951), and design grammars had synergies with the formulation of natural languages
(Stiny and Gips 1972).

Common Characteristics of Generative Design

Although developed for a variety of purposes, generative design methods share some
common characteristics. As noted, generative design systems rely on formal algo-
rithms to describe and generate designs and are mostly associated with computation.
This chapter defines three common characteristics of generative design: algorithm,
ideation (for numerous design alternatives), and computation.
Algorithms in most generative methods, are realized as a collection of generative
rules, an interpretation made by Oxman (1990) and Boden and Edmonds (2007).
Others describe similar concepts as build commands (Hornby and Pollack 2001),
transformation rules (Gero and Kazakov 1996), and so on. Generative rules are
responsible for changing the whole or a part of the design to a new one. Rules
differ by the mechanisms of generation they adopt, which refer to the different ways
rules alter the design. For example, in shape grammars the generative mechanism
in a shape rule can be used for either replacement or modification (Knight 2003).
Replacement is the substitution of the design or its part(s) with another; it can also
include addition (replacing void with an element) and subtraction (replacing an
element with void). On the other hand, modification is used to change the scale,
orientation and direction, or other numerical properties (e.g., in parametric shape
grammars) of the design. Through the application of the rules, design generation is
then arranged in sequence (Herr and Kvan 2007). This sequential and recursive
process selects a combination of rules to process the design generation in a
structured manner, from an initial state of the design to the final outcome. Design
alternatives emerge when different combinations of rules are selected and applied.
These sequences are not always explicitly defined and may emerge during their
applications. For example, in shape grammars, rule sequencing is made possible
in real time by the recursive matching of existing shapes from the design with
the left-hand side (LHS) shapes of the rule set (Maher 1990). Another more
advanced type of sequencing is through the cyclic reproduction of design (Eckert
et al. 1999). In this case, rules that have directed the generation of the previous
4 N. Gu et al.

stage will undergo further round(s) of application, but only applied to the highest
potential designs and alternatives identified during the process. Such a quality
is quintessential to evolutionary design, which will be introduced in the section
“Evolutionary Systems.”
Ideation (for numerous alternatives), or design divergence, refers to the capability
of generative design methods to create multiple instances that satisfy the design
requirements; usually this is enabled by the selection of multiple rules and their
different sequencing for application during design generation. For example, in shape
grammars, the decidability of LHS shapes (e.g., multiple possibilities in recognizing
and matching LHS shapes) can contribute to the divergence of the design results
(Knight 2003). Other factors can also play a role in enabling design divergence. For
example, in parametric shape grammars, the range of parameters may allow multiple
definitions of design by adjusting the values of those parameters (Knight 2003).
The computational characteristic refers to the fact that generative design has a
formal structure comprising visual or mathematical properties for design generation
in a systematic way. Because of the increasing complexity and intensity of the
design and the generation processes, research has extended the notion of “computa-
tion” to the use of computers as well (Boden and Edmonds 2007). In some domains,
for example, “computer art” and “generative art” have been used interchangeably
deriving from this context.

Main Generative Design Systems

Generative design systems have been developed since the 1960s. Many of them
share common origins or features, which are useful for categorizing different sys-
tems. For example, Fischer and Herr (2001) characterized generative systems into
four general types: emergent systems (e.g., cellular automata), generative grammars,
algorithmic generation (e.g., parametric design), and algorithmic reproduction (e.g.,
genetic algorithms). While such categorization makes sense in abstracting and
distinguishing common generative features, the terminologies themselves may not
be very precise. For example, “algorithmic generation” can be a very broad term
and arguably applicable to all generative systems. The term “emergent systems”
suggests support for design emergence and is also applicable to a number of
systems. A different approach (Singh and Gu 2011) does not explicitly develop
more general categorizations, but acknowledges that different generative systems
can have both differences and overlaps between each other. This section reviews four
categories, by considering both their differences and similarities and therefore does
not mutually exclude one from another. These categories are: generative grammars,
evolutionary systems, self-organized emergent systems, and associative generation.

Generative Grammars
As discussed in chapter  “Shape Grammars: A Key Generative Design Algorithm”,
generative grammars are based on transformational rules which can be applied
recursively to develop the final shape. This notion was originated from linguistics
Parametric Design: Theoretical Development and Algorithmic Foundation. . . 5

but has evolved into different forms for different unique purposes. For example,
visual grammars such as shape grammars were inspired by Chomsky’s idea of
generative grammars (Stiny and Gips 1972), meanwhile graph grammars were
initially developed in computer science and are especially suitable for computer
implementations (Chakrabarti et al. 2011), and L-systems were based on another
linguistic concept – string grammars – developed in the 1950s (Singh and Gu 2011).
In recent years research has combined these approaches to enhance generative
capabilities, for example, the combinations of shape grammars and graph grammars
(Grasl and Economou 2013; Lee et al. 2016) have been demonstrated to address both
visuospatial and syntactic issues in architecture. Generative grammars are one of the
earliest generative design systems to be used in architectural design, for example,
the shape grammar of Palladian villas by Stiny and Mitchell (1978). On the other
hand, graph grammars and L-systems were applied to architectural design much
later, having been in use since the 2000s (e.g., Parish and Müller 2001).
The original generative mechanisms of rules in grammars are mostly for replace-
ment not modification (Knight 2003). Nevertheless, modifications later became
possible in grammars and significantly enhanced their generative power, especially
with the emergence of parametric shapes and parametric shape grammars.
Generative grammars are sequential due to the nature of rule application. The
sequencing of rules for application is determined by matching different existing
elements of the design against LHS shapes of the rule set for rule selection
(Stiny and Gips 1972). The divergent generation of design alternatives could be
fulfilled by different possibilities in such rule selection and application. While
all grammars are computational, only graph grammars and L-systems have been
realized as computer implementations to a greater extent. As discussed in chapter
 “Shape Grammars: A Key Generative Design Algorithm”, during shape detection,
shape grammars allow designers to freely decompose and recompose shapes. This
feature is called shape emergence, which regards shapes as not being predefined but
emerged. To appropriately handle shape emergence has been difficult in computer
implementations because of the finite representations of design being restricted by
the computer memory (Krish 2011; Tching et al. 2016), and therefore it has become
an ongoing challenge for visual grammars such as shape grammars.

Evolutionary Systems
As the term suggests, evolutionary systems in design were derived from evolu-
tionary biology, especially with the simplified notion of “survival of the fittest”
(Holland 1975). In summary, an evolutionary system defines a recursive process
of design reproduction, in which each organism reproduces (slightly) divergent
offspring. Only the “fittest” among them would survive to trigger reiteration of the
reproduction and selection cycle (Fasoulaki 2007). In an evolutionary system, the
“fitness” of the generated design instances is evaluated using certain fitness criteria
to select the most appropriate ones to continue the generation and selection cycle,
until a satisfactory outcome is achieved. Because of such a basis in evolutionary
biology, these generative design systems are also called genetic algorithms (GA).
6 N. Gu et al.

Unlike most other generative design systems, there are no inherent generative
rules in evolutionary systems. Design alternatives can be produced by different
means, including other generative methods (e.g., shape grammars, L-systems, etc.).
The sequential characteristic of generation reflects the cycle of generation and
selection, rather than the order of specific rule selection and application. Divergence
in evolutionary design is also supported by the generation and selection cycle,
and the evolutionary process is usually sizable in terms of both depth (number
of stages in design generation) and breadth (number of alternatives per stage). In
most existing evolutionary design systems, computers have been used to assist in
managing the generation and selection cycle. However, in cases where the selection
criteria are more subjective (e.g., aesthetics), human designers in particular have
been used for intervention.
Generative design systems or genetic algorithms have emerged from the very
beginning of computational design (Holland 1968). However, their explicit applica-
tions in architecture only began in the 1990s (Jo and Gero 1998). A possible reason
may be the complications of formulating fitness criteria that address rather complex
architectural design issues or considerations and the challenges for formulating them
using abstract computer representations.
The concepts of generation and selection, or synthesis and evaluation, have
always been an important part of design theories and methodologies, and they are
also the fundamental basis of evolutionary design systems.

Emergent and Self-Organized Systems

The concept of emergence here is defined as the explicit outcome of implicit
representation (Gero 1996). Therefore, emergent systems refer to generative design
systems whose outcome emerges out of self-organized components. A common
approach to this emergence is a collection of self-organizing agents, who shape
the final form by interacting with each other and with their surroundings. Cellular
automata and swarm intelligence are the two typical examples of emergent and self-
organized systems (Singh and Gu 2011).
Cellular automata systems were inspired by biology through an interest in
simulating biological growth (von Neumann 1951). With its formal establishment
in the 1980s (Wolfram 1986), it attracted the attention of designers. In practice, a
cellular automata system is made of a grid (or structured geometry) whose cells can
have different properties or states (Sarkar 2000). A rule or set of rules, inherent to
the cells, can direct the alteration of states responding to their neighboring cells or
time intervals (Batty 1997b). As time passes, the cells automatically change their
properties or states, and the corresponding design combinations emerge on the grid
as the outcome. Because of this automation process of the cells, the term cellular
automata is therefore used for such generative design systems.
Considering the rather abstract nature of the cells, the generative rules (i.e.,
for state alteration) can have both replacement and modification as the generative
mechanism. The automation process is sequential as state alteration proceeds and
spreads cell by cell. Design divergence in cellular automata can be achieved either
Parametric Design: Theoretical Development and Algorithmic Foundation. . . 7

by recording the collective states of cells at different times or enabling parameters

and variations that can affect the automation rules (Batty 1997a).
Swarm intelligence concerns the design of complex and intelligent systems
based on collective behaviors of simple individual agents who work collaboratively
(Blum and Li 2008). Similar to the cells in cellular automata, agents in swarm
intelligence behave collectively based on their inherent rules, as well as interactions
with each other and their surroundings. However unlike cellular automata, their
design outcomes are not bound to the defined grid (Singh and Gu 2011), therefore
their behaviors can be more complex than those of the alteration of states in
cellular automata. Swarm intelligence originated in the study and simulation of
insect swarms, and later has been extended to other animals and humans who can
undertake more complex cognitive activities (Garnier et al. 2007).
Similar to cellular automata, the agents in swarm intelligence usually operate
based on time intervals, which define the sequential characteristic of design
generation using swarm intelligence. The divergence of the design outcomes may
be realized by various means such as varying (or even randomizing) the numbers
and types of agents, their surroundings, time intervals, etc.

Associative Generation
In this type of generative system, the design and its alternatives are generated by
firstly defining the association between different components that will make up
the design, and then by assigning and alternating the values of different properties
between each other. Unlike grammars, the design components are not explicitly
transformed, but any changes made to their properties will subsequently lead to the
generation of new design instances by adjusting other properties due to the defined
association.
Parametric design is one of the most typical generative systems and the latest
development based on associative generation or associative geometry. Parametric
design is the focus of this chapter and will be more extensively introduced in the
second half of the chapter. Parametric, as a term or method, originated from the
mathematical notion of parametric equations, with its design application initially
seen in the works by Italian architect Luigi Moretti in the 1940s (Frazer 2016).
The essential idea of parametric design is that the values of different variables
change, usually associatively, based on different input parameters. The concept
of parametric design has also been adopted to enhance other generative design
systems, for example, parametric shape grammars. In theory, parameters and
variables can be of any types or values, including those related to visual qualities,
while in practice they are usually numerical, which make them ideal for computer
implementation. Divergence may be supported by either adjusting the parameters
and variables, or modifying and redefining the association between them. The
sequential characteristic here is largely reflected in iteration or rapid prototyping,
where a large number of design alternatives can be very efficiently produced by
the abovementioned adjustment, modification, or redefinition, which can then be
explored and reviewed (often with the intervention of human designers) to perfect
the design generation.
8 N. Gu et al.

Parametric Design

As described above, the concept of parametric design originated from generative

design. Deriving from those core generative design principles, parametric design
has since evolved into one of the most widely applied generative design systems in
practice and in education. This section firstly introduces the historical background
and development of parametric design. The main content of the section describes
the theoretical foundation of parametric design, including various fundamental
concepts, as well as their application and implementation.

Historical Review of Parametric Design

Origin of Parametric Design

As noted, parametric design is a specific type of system for generative design. The
generative process of parametric design focuses on the control of variations and
constraints. The variation control of geometry can be traced back to the Church
of Colònia Güell designed by Antonio Gaudi. He created a model in which the
ceiling shape could be altered by adjusting the position of the weight or the length
of the string. This association of the ceiling shape with the position of the weight
and the length of the string in the system, demonstrates the fundamental concept
of parametric design and is the initial application of it in architecture, although
without explicit computational techniques. Based on this concept, Hillyard and
Braid (1978) proposed a system for computer-aided mechanical design that allowed
the specification of geometric constraints, restricted within a certain range. This
system represents one of the first attempts to theorize variations and constraints
in computational design systems. A few years later, Light and Gossard (1982)
furthered the development and presented what was called variation geometry or
variation design. Their work provided geometrical representations with enhanced
mathematical and geometrical modeling tools to support the design process. These
parametric modeling tools were then utilized widely in aerospace, ship construction,
and product design industries. From the late 1980s, Frank Gehry, a leading figure in
contemporary architecture, used CATIA as the main platform for the documentation
of his building designs. As a component of this platform, a parametric package (for
modeling NURBs surfacing) formally entered commercial architectural practices.
Though most of Gehry’s designs only used parametric techniques in the documen-
tation stage, at around that time his style suggested new architectonic possibilities
and triggered intellectual debates about the style and the role of the computer.
Without parametric tools, these dynamic, open-ended, and complex designs would
have been difficult to achieve. Later, the consulting firm Gehry Technologies (GT),
standardized this parametric approach into Digital Project (DP) to specifically
handle complex architectural designs. Today GT has expanded their services to a
wide range of computational design areas.
Parametric Design: Theoretical Development and Algorithmic Foundation. . . 9

Development of Parametric Design

From the mid-1990s, the number of architectural practices and academic orga-
nizations that explored and adopted parametric design in architecture increased
significantly. During that period, software such as GenerativeComponents™ (GC),
Grasshopper, and Processing were developed and evolved rapidly with enhanced
features. By the year 2000, the applications of parametric techniques in building
design had matured, and a growing number of landmark buildings using parametric
design were proposed and constructed. The leading practices using parametric
design in the architecture industry at the time included Foster+Partners, Zaha Hadid
Architects, UNStudio, KPF, AA Emtech, and SPAN. Designers were able to use
different parametric design tools to produce and control free-form architecture. This
was a major breakthrough for many of the leading architectural practices shifting
into designing through parametric modeling.
In research and education, AA in London, LAAC in Spain, and Hyperbody at
Delft University of Technology, MIT, and Columbia University have since become
cradles for the next generation of parametric designers. Meanwhile, international
computational design conferences including ACADIA, ASCAAD, CAADRIA,
eCAADe, SIGraDi, and CAAD Futures have an increasing number of publications
on parametric design, an important topic that has since been established in these
research forums. In more recent times, there have been many other conferences and
workshops that continue the effort in promoting parametric design.

Parametricism
Over the past 20 years, parametric design has gradually played an increasingly
important role in architectural design, especially among leading design practices.
Some scholars have argued that it has become a design style or movement that is
replacing Modernism (Schumacher 2009). In the 2008 Venice biennial exhibition of
architecture and titled “Out There: Architecture Beyond Building,” the term para-
metricism was first raised by Patrik Schumacher and later more comprehensively
described as a combination of design concepts that provide a new and complex
order based on key principles of differentiation and correlation (Schumacher 2009).
During the 2011 ACADIA conference, the term then became the main conference
theme.
The overarching implications of parametricism and its related concepts are
still largely underexplored in a broader industry context, and the approaches to
apply parametricism can also vary. For example, architects can maintain modernist
aesthetics in buildings but only use parametric modeling to address different
levels of complexity or to optimize the overall design. Such approaches can be
seen in projects like Soho Shang Du in Beijing. It is considered to be a finely
designed building achieved through parametric tools, but with a rather conventional
appearance.
10 N. Gu et al.

Parametric Design

Key Concepts and Terminologies

Parametric design. Parametric design focuses on the representation and control of
the relationships between objects in a computational design model. It supports the
creation, management, and organization of complex designs (Woodbury et al. 2007).
Using parametric design tools, designers can create rules according to the aesthetics
and functional or performance-related requirements of a design. A parameter is a
value or measurement of a variable that can be altered or changed. Each object in
a parametric system may have certain rules embedded, and when one parameter
changes, other parameters will adapt automatically (Ostwald 2012). By changing
various parameters, particular instances can be created from a potentially infinite
range of possibilities (Kolarevic 2003).
From the notions above, parametric design can be understood to have the
following characteristics. Firstly, these systems are structured based on design
parameters. Eastman (2008) suggests that in parametric design, key variables are
often defined by parameters (Eastman 2008). While most parameters are related
to geometrical modeling, others can be connected to functional or performance-
related requirements. Secondly, there are relationships between design variables.
As Cárdenas (2007) noted, parametric design establishes the relationships between
modeling components defined by constraints, while Abdelsalam (2009) claims that
the relationships are maintained by variations. In most cases, both constraints and
variations in combination help to define the relationships between the modeling
components. The entire parametric system will update when certain variables are
changed, because the defined relationships between them will lead to changes in
other variables. However, a parametric system may become “over-constrained” if
there is no effective control over the defined relationships (Burry 2003). Thirdly,
the rule-based algorithms ensure that the parametric design process is flexible but
controllable. By making rules, designers can produce variations that result in “fully
organized controllable building forms” (Abdelsalam 2009). Fourthly, parametric
design provides a formal process from which multiple design solutions can be
developed simultaneously and efficiently. Both Hernandez (2006) and Karle and
Kelly (2011) emphasize that to develop parallel ideas is one of the main advantages
in parametric design.
In summary, parametric design is a dynamic, rule-based design process con-
trolled by variations and constraints, in which multiple design solutions can be
developed in parallel. It is especially effective in generating complex forms and
optimizing multiple design solutions. As a result, parametric design is usually
utilized in complex building form generation and fabrication, structural and energy
optimization, and other design iteration and prototyping tasks. As a new digital
design method, parametric design is quite different from traditional CAD/CAM
because of these rule-based algorithmic characteristics.
Parametric variations. In a parametric design system, variations are controlled
by changing the values of parameters and constraints without necessarily altering
Parametric Design: Theoretical Development and Algorithmic Foundation. . . 11

Fig. 1 Simplified examples of parametric variations generated in the same system

the original structure of the model (see Fig. 1 for different variations of a primitive
geometry generated by the same system). Variations can be single or multiple,
independent from or interrelated with each other. Karle and Kelly (2011) argue
that as a new design method, parametric design does not push designers into
generating the right design solution but, rather, into asking the right question,
which can then be answered with multiple solutions. As a result, the selection
process of the generated variations is an important step in parametric design. Prior
to selecting the appropriate variation(s), a typical workflow in parametric design
also involves identifying the design problem(s) and developing a series of rule sets
with associative design variables, which can support a dynamic and flexible design
process enabling the emergence of variations.
Design constraints have a significant role in directing the entire parametric
process. They can be used by designers to describe and generate a range of
variations and to define and control the unique characteristics of each. Constraints
set limits on parameters and control the possibilities for the range of variations.
In parametric design, constraint satisfaction is important in the decision-making
process, connecting individual factors with the overall design outcomes. There
are two basic forms of constraint – geometric and dimensional (Monedero 2000).
Geometric constraints are properties that control how geometrical entities relate to
each other. Dimensional constraints are properties that can be assigned to a singular
value, relating the geometry to a numerical value that fixes its behavior until it
is changed or removed. An optimal parametric design process often requires the
computational model to be well balanced, which is neither under-constrained nor
over-constrained (Monedero 2000).
Parametric modeling. Parametric modeling refers to specific techniques that
“create design spaces and geometric dependencies within a model” (Gane and
Haymaker 2009). It provides a formal descriptive and generative design framework
(through parameters and their associative relationships), by which designers are able
to change the input values to generate and optimize the design with variations.
The most significant advantage of parametric modeling is that it allows changes
to be made to the parameters at any stage of the design process (Monedero 2000).
Different parametric modeling techniques have been developed for visual purposes
(i.e., form finding), as well as for other functional or performance-related purposes.
Figure 2 demonstrates some typical parametric modeling techniques for form
finding, for example, repetition (Fig. 2, left) and subdivision (Fig. 2, right); other
similar types of techniques can include tiling, recursion, weaving, etc.
12 N. Gu et al.

Fig. 2 Examples of parametric modeling techniques for form finding: repetition (left) and
subdivision (right)

Research further synthesizes and formalizes essential parametric modeling

techniques into “design patterns” (Woodbury et al. 2007). With these patterns, more
standardized approaches can be adopted in parametric design to address particular
problems. The advantage of these general patterns in parametric design is that
“[p]atterns are a way to identify successful general strategies that exemplify a key
concept in a memorable fashion that can easily be taught” (Woodbury et al. 2007, p.
229). Parametric design patterns have since been proposed as abstract and reusable
tools in parametric design. It is suggested that by using and learning these general
patterns, architects and students alike are able to master parametric design more
efficiently and skillfully (Woodbury 2010) through abstraction and standardization.

Application
Parametric design and form finding – Parametric design tools can capture and
explore critical relationships between design intentions and geometries. Designers
interact with the tools through rule algorithms to capture and manipulate these
relationships, as well as relationships among different design elements. For the
purpose of form finding, parametric design can be especially useful to facilitate the
modeling process for complex geometries, and the integration of parametric tools
in design can also enhance flexibility and control during the process (Fischer et al.
2003). To explore form finding in parametric design, some studies have focused on
geometrical modeling methods. For example, Hnizda (2009) suggested that there
can be at least two different geometrical modeling methods in parametric design –
object extraction and transformation – each of which can be used to explore the
relationships between formal aesthetics and functional properties. Others, such as
Baerlecken et al. (2010), explored the issues through problem definition in the
early conceptual stage. To a large extent, the latter form-finding approach through
problem definition is dependent on variation settings (Baerlecken et al. 2010),
and through parametric variations they have explored functional and structural
properties resulting from sunshading and to derive the final design form.
One of the most significant form-finding exercises using parametric design is the
“Sagrada Familia” project. Gaudi’s Sagrada Familia cathedral is characterized by
curved sculptural surfaces that follow specific sets of principles. These embedded
rules in Gaudi’s design provide perfect opportunities for further design and analysis
Parametric Design: Theoretical Development and Algorithmic Foundation. . . 13

Fig. 3 Selected examples of new garden plans generated using a parametric design system (Yu
et al. 2015b)

using parametric tools (Roberto and Hernandez 2004). These rather simple geomet-
ric rules and procedures can produce a rich formal language. Since the early 2000s,
Professor Mark Burry (2003) and his team have developed and applied parametric
systems for modeling complex geometries in a major case study of Gaudi’s Sagrada
Familia. The research later extended to aspects of fabrication and construction
aiming to complete the unfinished masterpiece. Their series of developments has
provided important evidence of the effectiveness of parametric design for designing
and analyzing complex geometries through rule algorithms.
Extending into landscape design, Yu et al. (2015b) applied parametric design to
produce new garden plans that are consistent with the style of Chinese traditional
private gardens, replicating selected socio-spatial characteristics of those heritage
gardens. The socio-spatial characteristics of the original gardens were analyzed
employing various space syntax techniques, and the resultant mathematical mea-
surements were then mapped into parameters and constraints in the system. Through
the defined rule algorithms, the parametric system can generate new garden plans
with similar spatial structures and connectivity values. Figure 3 shows selected
examples of the newly generated garden plans using the developed system, on three
different given sites.
14 N. Gu et al.

Parametric design and design performance – As discussed earlier, beyond

forms and aesthetics, parametric design tools can also respond to functional
or performance-related requirements. Such applications are mostly for structural
analysis and building performance optimization (i.e., sustainability).
When using parametric tools for structural design, Maher and Burry (2003)
compared parametric structural analysis with traditional approaches in a cross-
disciplinary collaboration between architects and structural engineers. The combi-
nation of structural analysis and parametric design not only enhanced the design
process and result but also provided new platforms for collaboration across dis-
ciplines. In a similar study linking architectural design to structural optimization,
Holzer et al. (2007) investigated geometry generation with structural analysis and
optimization, which shows that parametric systems (supported by parameters, rules,
and constraints) are effective for generating a variety of solutions for structural
design. Most of these works have been focused on analyzing the volume of structural
materials. ETABS is one of the popular structural analysis tools whose data can be
imported into parametric design software. For instance, Almusharaf and Elnimeiri
(2010) studied structural performance in high-rise building design by utilizing
ETABS in the parametric environment of Grasshopper. A design scenario was
presented in their study where instant feedback of structural performance could
be provided during the parametric process to assist decision making and design
generation.
One of the advantages of a parametric design system is that multiple analyses
can be established and conducted in parallel within the same model to support
more comprehensive optimization. Besides structural analysis, building perfor-
mance, especially in terms of sustainability, is another important application of
parametric design. For instance, “multi-parametric façade elements” were proposed
and examined by Schlueter and Thesseling (2008). In their study, the performance
analysis of the façade offers instant assessment (in terms of the solar gain in
different façade forms) during the design process, so that designers can better
improve energy performance while generating design forms. In another study, a
formal framework that combines parametric modeling with the “performance-based
design” (PBD) paradigm was proposed (Bernal 2011), aiming to standardize the
parametric applications of building performance and providing real-time feedback
on the building performance index during the design process.
Collaboration in parametric design – With the continuing evolution and
growing popularity of parametric design, researchers have begun to explore design
collaboration using parametric systems, including cross-disciplinary and distant
scenarios. As discussed above, collaboration between disciplines using parametric
systems, for example, in structural design and analysis, can critically optimize
solutions using interdisciplinary knowledge. Cross-disciplinary collaboration is
considered to be especially beneficial in the early design stage, and the use of
parametric design in the early design stage can effectively assist multidisciplinary
teams in problem finding, which in turn directs the generation and selection of
variations during the parametric process (Gane and Haymaker 2009).
Parametric Design: Theoretical Development and Algorithmic Foundation. . . 15

In terms of distant collaboration, Burry and Holzer (2009) explored the potential
for sharing parametric models in a version control platform, which could be used by
design teams located in different geographical locations. Rajus et al. (2010) used the
participant observation method to study distant collaboration in parametric design.
In their study, 18 participants were asked to perform different design tasks using the
parametric software GenerativeComponents™ (GC). The study developed different
controls in collaboration and applied them in the experiments. The results show that
moderately controlled collaboration can enhance the team performance and user
satisfaction in parametric design.

Implementation Software
To date, quite a number of software packages have been developed and adopted
for parametric design in practice. Most software supports free-form modeling, but
also has scripting plug-ins that enable designers to create rule algorithms more
directly and freely. Some of these software packages are briefly introduced in
this section. Rhino + Grasshopper has been one of the most commonly used
parametric tools, especially in architectural design, while Digital Project (DP)
and GenerativeComponents™ (GC) are more tailored for large-scale projects with
complex parametric and geometric associations.
Computer Aided Three-dimensional Interactive Application (CATIA) was first
developed in 1977 by French aircraft manufacturer Dassault Aviation. Thereafter,
it was applied in aerospace, automotive, shipbuilding, and other industries for its
capabilities in controlling and manipulating complex geometries, as well as its
supports for manufacturing accuracy. Multiple versions of CATIA have since been
developed to perfect the commercial uses, and at the same time it has gradually
entered the architectural field. Based on CATIA 5.0, Gehry Technologies developed
Digital Project (DP), which specifically aimed to service architectural design. DP
has been known as a very powerful parametric software package that can effectively
handle complex parametric as well as geometric associations, which makes it ideal
for large complex parametric design projects.
In 2005, GenerativeComponents™ (GC) was developed by Robert Aish for
Bentley Systems. GC implements parametric concepts across the entire design
project life cycle, from the early conceptual phase to the final documentation. In
addition, GC has been integrated with Building Information Modeling (BIM) as
well as other analysis and simulation platforms for design evaluation and optimiza-
tion. Such integration can potentially make parametric design more targeted and
realistic, effectively linking design conceptualization with production, fabrication,
and construction.
Rhinoceros (Rhino) is a stand-alone, NURBS-based 3D modeling tool. It was
developed by Robert McNeel and has since been widely applied in a range of
domains, including architecture, industrial design, jewelry design, automotive, as
well as marine design. Grasshopper, on the other hand, is a rule-algorithm editor
with a graphic interface, which can be integrated into Rhino as a scripting plug-in.
It is structured with specific definition files that link to the main parametric model
in Rhino. Usually Grasshopper is used as a generative tool rather than a modifier
16 N. Gu et al.

during the parametric design process. Compared to other parametric software, Rhino
+ Grasshopper has been much more widely adopted in both practice and education
at the present time. This is because of its ease of operation as a visual programming
tool (supported by the graphic rule-algorithm editor) as well as its relatively low
cost.
Scripting uses computer programming languages such as Java and Visual Basic
(VB) to directly interpret and execute commands. It can establish and control
parameters and translate design intents into codes that are easily identified by
a computer. Consequently, designers (with fluent scripting skills) can define and
control the logics and rules for parametric design in a way that allows them more
freedom than any other tools. Common scripting languages include Python, VB,
and Ruby, while scripting tools include Python Script, RhinoScript, Processing, and
CADscript.
There are also several design analysis tools, which are often used together
with parametric software. For example, Ecotect is used for analyzing energy
performance, while ETABS is used for structural analysis. These analysis tools
are capable of exporting analysis data into parametric software to direct design
generation and optimization.

Reshaping Architectural Design

Impact on Architectural Design

Designers’ activities change in a parametric design environment compared to those

in a traditional CAD environment. Firstly, designers design rules and define logical
relationships of the design rather than only modeling geometries. One of the biggest
differences between parametric design and traditional computer-aided design is
that rule sets become basic design elements and procedures in parametric design
(Abdelsalam 2009). While building parametric models, designers set variations,
design data flows, and define and adjust rules as well as the values of parameters.
They are not only required to think about a particular building design but also the
rule sets that generate and underpin the design. Additionally, through the control
of these logical relationships, parametric design ensures many more possibilities
for the generated solutions (Hernandez 2006; Karle and Kelly 2011). Secondly,
architects can easily trace back through the parametric design process and are free
to make changes in any steps of the process. During parametric design, through the
defined rule algorithms, systems are clearly differentiated and correlated, and design
activities are clearly communicated with each other within the process (Schumacher
2009). Therefore, designers are able to go back to any step along the process to
change parameters or revise rules in order to modify the design process for different
purposes or to trial different variations. This level of flexibility allows architects
to keep the design “open.” Thirdly, a large number of design alternatives can be
developed in parallel. Traditionally, designers consider a very small number of
alternatives because of time or memory limitations (Woodbury and Burrow 2006).
Parametric Design: Theoretical Development and Algorithmic Foundation. . . 17

Fig. 4 A large number of form variations generated in a parametric design system

Therefore, as Akın (2001) has argued, design solutions are not usually optimal, but
only satisfactory, according to a preset level of aspiration. In parametric design, once
the rule algorithms are established, a large number of design alternatives may be
generated (see Fig. 4, for instance). This rapid design prototyping can significantly
expand the design possibilities and widen the designer’s thinking. Further, designers
do not need to predetermine or fixate on any solutions at early stages (Hernandez
2006; Holland 2011; Karle and Kelly 2011), which allows the maximum potential
to be maintained in the process, until the final design solution is achieved and
evaluated.
Parametric design offers not only a new computational design tool but also a new
way of design thinking. Generally, design is an abduction process in which both the
final artifacts and their behaviors are designed or defined (Dorst 2011). In parametric
design, Yu et al. (2015a) suggested that designers’ behavioral patterns shift between
two levels of activities – the design-knowledge level and the rule-algorithm level
(Fig. 5). During a typical parametric design process, there are design activities that
occur on both levels. At the design-knowledge level, architects make use of their
design knowledge, including, for example, how to adapt a building to the site, how
to shape the way people use the building, and how to satisfy other requirements of
their clients. At the rule-algorithm level, designers apply design knowledge through
the operations of various parametric design features, including defining the rules
and their logical relationships in the design, choosing the parameters suitable for
a particular purpose, and importing external data into the system to direct and
evaluate the generation. During the parametric design process, designers progress
by applying specialist knowledge, and in some parts (namely, at the rule-algorithm
level), they apply design knowledge indirectly through rule algorithms. As observed
18 N. Gu et al.

Fig. 5 Two types of design

spaces in a parametric design
environment

by Yu et al. (2015a), designers often start from the design knowledge level, then
as the design proceeds their activities at the rule algorithm level can rise, while
those at the design knowledge level can decrease but do not cease. This implies that,
although rule algorithms in parametric design are the focus and offer many advan-
tages, such as the support for design flexibility (Fischer et al. 2003) and a capacity
to deal with design complexity (Leach 2008), nevertheless design knowledge still
appears to be essential during parametric design. These preliminary findings by
Yu et al. (2015a) have implications for the practice and teaching of parametric
design in architecture. Although training for programming/scripting skills is of
course important, training for design thinking is also vital. As designers substitute
rule algorithm for design knowledge during parametric design, understanding how
design knowledge is expressed in rule algorithms in parametric design is important.
It has been noted that, in terms of high-level design thinking, designers’ cognitive
behavior does not significantly vary because of the computational tools adopted (Yu
et al. 2013).

Limitations of Parametric Design

Parametric design has been widely practiced in recent years, and its characteristics,
such as design flexibility and controllability, have rapidly escalated its popularity
(Barrios 2005; Fischer et al. 2005; Salim and Burry 2010). However, a number
of limitations have been reported. Firstly, parametric design requires relatively
higher acquisition and implementation costs. Some large-scale and more complex
tasks also require additional and specialized computing power. Therefore it can be
difficult for small-scale design firms to adopt parametric design. Secondly, a typical
challenge of using parametric design is how to select appropriate variations. Peña
and Parshall (2001) have argued that problem finding is the most important part in
a design process, and only by defining problems appropriately would problems be
solved. In parametric design, defining variations is critical to representing design
concepts. Variations support design flexibility and the possibility to develop ideas
in parallel. Designers benefit in having multiple variations, and this only makes
Parametric Design: Theoretical Development and Algorithmic Foundation. . . 19

sense when the system can adequately and appropriately capture and represent the
problems. For example, some design factors such as Energy Conservation Index
(ECI), budget, architectural building codes, and structural requirements can be quite
easily quantified and integrated into the parametric system. However, other factors
such as aesthetics, historical context, and social interaction are difficult to measure
and represent in the system. As a result, architects’ critical thinking (by applying the
relevant design knowledge) is still important even with the use of parametric design
tools. Thirdly, parametric design without careful implementation can sometimes
lead to excessive or unnecessary complexity. Over-pursuing forms and appearances
can make some parametric design projects appear to be superficial and lacking
contextualization. After all, for architecture, the complex form should come from
the complex needs and behaviors of people rather than the pure urge for “bizarre”
appearances (Leach 2008).

Conclusion

As one of the most frequently applied computational and generative design systems,
parametric design plays an important role, especially in various leading architectural
practices. This chapter has presented the theoretical development of parametric
design, discussing its connection to generative design in general, together with a
historical review and a basic introduction of the key concepts and applications.
Parametric design has significant impact on architecture. Firstly, parametric
design provides a formal framework and computational means for generating and
managing complex and dynamic building forms that have never been easily and
adequately achieved (by traditional CAD tools). This has opened up intellectual
debates about new aesthetics, styles, and movements in contemporary architecture.
Secondly, parametric design promotes collaboration. With parametric design tools,
as well as the integration of other specialization tools for performance analysis, we
have witnessed increasing and more integrated teamwork between architects and
other experts for design optimization. Thirdly, through defining rule algorithms,
parametric design adds and controls rationality in architectural design, which can
assist designers in making more informed decisions to achieve more superior
outcomes.
From the designer’s perspective, parametric design tools provide architects with
opportunities for designing at both the design knowledge and rule algorithm levels,
which opens up many new possibilities, as discussed throughout the chapter. At the
same time, new challenges also emerge. First and foremost, the role of architects
is changing, such that they need to be both architects and programmers/scripters.
The fact that architects can directly encode designs as computer languages can be
both efficient and constraining. Constantly switching between activities of design
knowledge and rule algorithm during parametric design can also affect an architect’s
design cognition, which should be carefully considered when designing parametric
systems and facilitating them in practice and in education.
20 N. Gu et al.

Cross-References

 Shape Grammars: A Key Generative Design Algorithm

References
Abdelsalam M (2009) The use of the smart geometry through various design processes: using the
programming platform (parametric features) and generative components. Proceedings of the
Arab Society for Computer Aided Architectural Design (ASCAAD 2009), Manama, Kingdom
of Bahrain, pp 297–304
Akın Ö (2001) Variants in design cognition. In: Eastman C, Newstetter W, McCracken M
(eds) Design knowing and learning: cognition in design education. Elsevier Science, Oxford,
pp 105–124
Almusharaf AM, Elnimeiri M (2010) A performance-based design approach for early tall building
form development. Proceedings of the Arab Society for Computer Aided Architectural Design
(ASCAAD 2010), Fez, Morocco, pp 39–50
Baerlecken D, Martin M, Judith R, Arne K (2010) Integrative parametric form-finding processes.
Proceedings of the 15th international conference on computer aided architectural design
research in Asia (CAADRIA), Hong Kong, pp 303–312
Barrios C (2005) Transformations on parametric design models. Proceedings of the 11th inter-
national conference on computer aided architectural design futures (CAAD Future), Vienna,
Austria, pp 393–400
Batty M (1997a) Cellular automata and urban form: a primer. J Am Plan Assoc 63(2):266–274
Batty M (1997b) Urban systems as cellular automata. Environ Planning B Plan Des 24:159–164
Bernal M (2011) Analysis model for incremental precision along design stages. Proceedings
of the 16th international conference on computer aided architectural design research in
Asia(CAADRIA), Newcastle, Australia, pp 19–18
Blum C, Li X (2008) Swarm intelligence in optimization. In: Blum C, Merkle D (eds) Swarm
Intelligence. Springer, Berlin, pp 43–85
Boden M, Edmonds E (2007) What is generative art. Digit Creat 20(1):21–46
Burry M (2003) Between intuition and process: parametric design and rapid prototyping. In:
Kolarevic B (ed) Architecture in the digital age – design and manufacturing. Spon Press, New
York/London, pp 149–162
Burry J, Holzer D (2009) Sharing design space: remote concurrent shared parametric modeling.
Proceedings of 27th eCAADe conference, Istanbul, Turkey, pp 333–340
Cárdenas CA (2007) Modeling strategies: parametric design for fabrication in architectural
practice. Dissertation, Harvard University
Chakrabarti A, Shea K, Stone R, Cagan J, Campbell M, Hernandez NV, Wood KL (2011)
Computer-based design synthesis research: an overview. J Comput Inf Sci Eng 11:021003–
021012
Chien SF (1998) Supporting information navigation in generative design systems. Dissertation,
Carnegie Melon University
Dorst K (2011) The core of ‘design thinking’ and its application. Des Stud 32(6):521–532
Eastman CM (ed) (2008) BIM handbook: a guide to building information modeling for owners,
managers, designers, engineers and contractors. Wiley, Hoboken
Eckert C, Kelly I, Stacey M (1999) Interactive generative systems for conceptual design: an
empirical perspective. Artif Intell Eng Des Anal Manuf 13:303–329
Fasoulaki E (2007) Genetic algorithms in architecture: a necessity or a trend? Proceedings of 10th
generative art international conference. Milan, Italy
Fischer T, Herr C (2001) Teaching generative design. Proceedings of 4th generative art interna-
tional conferene. Milan, Italy
Parametric Design: Theoretical Development and Algorithmic Foundation. . . 21

Fischer T, Burry M, Frazer J (2003) Triangulation of generative form for parametric design and
rapid prototyping. Proceedings of 21th eCAADe conference, Graz, Austria, pp 441–448
Fischer T, Burry M, Frazer J (2005) Triangulation of generative form for parametric design and
rapid prototyping. Autom Constr 14(2):233–240
Frazer J (2016) Parametric computation: history and future. Archit Des 86(2):18–23
Gane V, Haymaker J (2009) Design scenarios: methodology for requirements driven parametric
modeling of high-rises. Proceedings of the 9th international conference (CONVR 2009),
Sydney, Australia, pp 79–90
Garnier S, Gautrais J, Theraulaz G (2007) The biological principles of swarm intelligence. Swarm
Intell 1:3–31
Gero JS (1996) Creativity, emergence and evolution in design. Knowl-Based Syst 9(7):435–448
Gero JS, Kazakov V (1996) An exploration-based evolutionary model of generative design process.
Comput Aided Civ Infrastruct Eng 11(3):211–218
Grasl T, Economou A (2013) From topologies to shapes: parametric shape grammars implemented
by graphs. Environ Plan B: Plan Des 40:905–922
Hernandez CRB (2006) Thinking parametric design: introducing parametric Gaudi. Des Stud
27(3):309–324
Herr C, Kvan T (2007) Adapting cellular automata to support the architectural design process.
Autom Constr 16(2007):61–69
Hillyard RC, Braid IC (1978) Analysis of dimensions and tolerances in computer-aided mechanical
design. Comput Aided Des 10(3):161–166
Hnizda M (2009) Systems-thinking: formalization of parametric process. Proceedings of the Arab
Society for Computer Aided Architectural Design (ASCAAD 2009), Manama, Kingdom of
Bahrain, pp 215–223
Holland J (1968) Hierarchical descriptions, universal spaces and adaptive systems: technical report.
University of Michigan, Ann Arbor
Holland J (ed) (1975) Adaptation in natural and artificial systems. An introductory analysis with
application to biology, control, and artificial intelligence. University of Michigan Press, Ann
Arbor
Holland N (2011) Inform form perform. Proceedings of ACADIA regional 2011 conference, pp
131–140
Holzer D, Hough R, Burry M (2007) Parametric design and structural optimisation for early design
exploration. Int J Archit Comput 5(4):625–643
Hornby GS, Pollack J (2001) The advantages of generative grammatical encodings for physical
design, 2001 IE congress on evolutionary computation. IEEE, Seoul
Jo J, Gero JS (1998) Space layout planning using an evolutionary approach. Artif Intell Eng
12(3):149–162
Karle D, Kelly B (2011) Parametric thinking. In: Proceedings of ACADIA regional 2011
conference, pp 109–113
Knight T (2003) Computing with emergence. Environ Plan B Plan Des 30:125–156
Kolarevic B (ed) (2003) Architecture in the digital age: design and manufacturing. Spon Press,
New York
Krish S (2011) A practical generative design method. Comput Aided Des 43:88–100
Leach N (ed) (2008) (Im)material processes – new digital techniques for architecture. China
Architecture & Building Press, Beijing
Lee JH, Ostwald M, Gu N (2016) A justified plan graph (JPG) grammar approach to identifying
spatial design patterns in an architectural style. Environ Plann B: Urban Analytics City Sci
45(1):67–89
Light R, Gossard D (1982) Modification of geometric models through variational geometry.
Comput Aided Des 14(4):209–214
Maher ML (1990) Process models for design synthesis. AI Mag 11:49–58
Maher A, Burry M (2003) The parametric bridge: connecting digital design techniques in
architecture and engineering. Proceedings of the 2003 annual conference of the association for
computer aided design in architecture, Indianapolis, Indiana, pp 39–47
22 N. Gu et al.

Monedero J (2000) Parametric design: a review and some experiences. Autom Constr 9(4):
369–377
Ostwald M (2012) Systems and enablers: modeling the impact of contemporary computational
methods and technologies on the design process. In: Gu N, Wang X (eds) Computational
design methods and technologies: applications in CAD, CAM and CAE education. IGI Global,
Pennsylvania, pp 1–17
Oxman R (1990) Design shells: a formalism for prototype refinement in knowledge-based design
systems. Artif Intell Eng 5(1):2–8
Oxman R (2006) Theory and design in the first digital age. Des Stud 27(3):229–265
Parish Y, Müller P (2001) Procedural modeling of cities. Proceedings of the 28th annual conference
on computer graphics and interactive techniques. ACM, New York, pp 301–308
Peña W, Parshall S (eds) (2001) Problem seeking: an architectural programming primer. Wiley,
New York
Rajus V S, Woodbury R, Erhan H, Riecke B E, Mueller V (2010) Collaboration in parametric
design: analyzing user interaction during information sharing. Proceedings of the 30th annual
conference of the Association for Computer Aided Design in Architecture (ACADIA), New
York, pp 320–326
Roberto C, Hernandez B (2004) Parametric Gaudi. Proceedings of the 8th Iberoamerican congress
of digital graphics (SIGraDi 2004), Porte Alegre, Brasil, pp 213–215
Salim F, Burry J (2010) Software openness: evaluating parameters of parametric modeling tools
to support creativity and multidisciplinary design integration. Proceedings of computational
science and its applications (ICCSA 2010), Berlin and Heidelberg, Germany, pp 483–497
Sarkar P (2000) A brief history of cellular automata. ACM Comput Surv 32(1):80–107
Schlueter A, Thesseling F (2008) Balancing design and performance in building retrofitting: a
case study based on parametric modeling. Proceedings of the 28th annual conference of the
Association for Computer Aided Design in Architecture (ACADIA) pp 214–221
Schumacher P (2009) Parametricism – a new global style for architecture and urban design. AD
Archit Des – Digi Cities 79(4):14–23
Singh V, Gu N (2011) Towards an integrated generative design framework. Des Stud 33:185–207
Stiny G, Gips J (1972) Shape grammars and the generative specification of painting and sculpture.
Inf Process 71:1460–1465
Stiny G, Mitchell WJ (1978) The Palladian grammar. Environ Plan B 5:5–18
Tching J, Reis J, Paio A (2016) A cognitive walkthrough towards an interface model for shape
grammar implementations. Comput Sci Inf Technol 4:92–119
Von Neumann J (1951) The general and logical theory of automata. In: Taub AH (ed) John von
Neumann, collected works. Pergammon Press, New York, pp 280–326
Wolfram S (1986) Random sequence generation by cellular automata. Adv Appl Math 7:123–169
Woodbury R (ed) (2010) Elements of parametric design. Routledge, New York
Woodbury RF, Burrow AL (2006) Whither design space? Artificial intelligence for engineering
design. Anal Manuf 20(2):63–82
Woodbury R, Aish R, Kilian A (2007) Some patterns for parametric modeling. Proceedings of the
27th annual conference of the association for computer aided design in architecture Halifax,
Nova Scotia, pp 222–229
Yu R, Gero JS, Gu N (2013) Impact of using rule algorithms on designers’ behavior in a parametric
design environment: preliminary results from a pilot study. Proceedings of the 15th International
conference on computer aided architectural design futures (CAAD FUTURES 2013), Shanghai,
China, pp 13–22
Yu R, Gu N, Ostwald M, Gero J (2015a) Empirical support for problem-solution co-evolution in a
parametric design environment. Artif Intell Eng Des Anal Manuf (AIEDAM) 29(01):33–44
Yu R, Ostwald M, Gu N (2015b) Parametrically generating new instances of traditional chinese
private gardens that replicate selected socio-spatial and aesthetic properties. Nexus Netw J
17(3):807–829