Thesis

Recursive Relational Urban Design
by
Dessen Hillman
Bachelor of Architecture
Pratt Institute, 201 2
SUBMITTED TO THE DEPARTMENT OF ARCHITECTURE IN PARTIAL FULFILLMENT OF THE

REQUIREMENTS FORTHE DEGREE OF
MASTER OF SCIENCE IN ARCHITECTURE STUDIES:ARCHITECTURE AND URBANISM
AT THE
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
JUNE 2014
MASSACHUSETTS INSGMJTE
OF TECHNOLOGY
2014 Dessen Hillman.All rights reserved.

The author hereby grants to MIT permission to reproduce
and to distribute publicly paper and electronic
copies of this thesis document in whole or in part
in any medium now or hereafter created.
Signature of Author:
JUL 0 1 2014
LIBRARIES
_Signature redacted
Department of Architecture
May 22, 2014

Certified by:
Signature redacted_
Signature redacted
Michael Dennis
Professor of Architecture
Thesis Supervisor
Accepted by:
Takehiko Nagakura
Associate Professor of Design and Computation
Chair of the Department Committee on Graduate Students
by
Dessen Hillman
Thesis advisor: Michael Dennis
Title: Professor of Architecture
Thesis reader: Brent Ryan
Professor
of Urban Design and Public Policy
Title:Assistant
Thesis reader:Takehiko Nagakura
Title:Associate Professor of Design and Computation

by
Dessen Hillman
Submitted to the Department of Architecture

on May 22, 2014 in Partial Fulfillment of the
Requirements for the Degree of Master of Science in
Architecture Studies: Architecture and Urbanism
ABSTRACT
This thesis proposes a methodology for the act of urban design that is recursive and centered around
explicit relational operations, enabled by taking advantage of computation and parametric techniques. It
contains iterative experiments aimed to explore and discover the feasibility and potential of computational
incremental urban design
The initial idea for this thesis emerged as two urban design conventions are challenged.
The first is the teleological masterplan. Masterplans take a long time to be implemented, causing the majority of them to be only partially implemented. In addition, as the early parts of the design are seeing completion of built development, their surrounding context would have changed and developed as well, rendering
the rest of the initial design to be obsolete and out of context, which requires a new design to be created.
The second is a more recent norm: the fact that contemporary designers use generative computation
techniques often to generate some form of a masterplan. Sadly, most of the outcomes produce less coherent and intentional designs than what a conventional urban design approach would. Granted, each individual is entitled to his/her own belief on good urban form, but many urban design schemes produced today
by computer and parametric techniques are residues of interest and passion for the tools and techniques
themselves. Many computation-based urban schemes today, including this thesis, are still early explorations,
but I hope to take a step towards bringing our views on computation techniques away from digital obsession and towards a more pragmatic use.
This thesis is a response to my speculation that there are confusions between urban design and architecture at the urban scale. Unlike architecture, urban design cannot afford to take a single set of ideas that
aims towards idea clarity, which typically ends up with having a thing as an organizing datum in a single
design act, whether it's an axis, a mega structure, an open space, a topography map, etc.This approach is
too one-dimensional, regardless of how complex the designer claims his/her project is.
Thesis Supervisor: Michael Dennis
Title: Professor of Architecture
TABLE OF CONTENTS
1. Challenging conventions
i.Teleological masterplans
ii.The use of generative computation techniques in practice
2. Urban design vs. architecture at the urban scale
3. Precedents and similar thoughts (incremental and computational)
i. Early thoughts in computational and incremental urbanism
ii. Recent and contemporary thoughts in computational and incremental urbanism
4. Experiments and reports
i.The basic idea (components and relationships)
ii. Site selection
iii. Operations
iv. Experiments
5. Conclusions
6. Bibliography, sources, and references
1. CHALLENGING CONVENTIONS
i.Teleological masterplans
The majority of urban masterplans are either never implemented or only implemented
partially. Unlike architecture,
urban design schemes span over
decades in implementation and
development, even in a supportive economic and political
climate. Assuming uninterrupted
developments, urban plans will
still take a long time to implement that by the time the early
phases of implementations are
finished, the surrounding city
would have changed to a point
where the rest of the design is
rendered obsolete.
An example of this is the various masterplans incomplete
implementation in downtown
Providence between 1960-2000,
carefully studied and documented by Brent Ryan (Assoc. Prof. of
Urban Design and Public Policy
in MIT's Dept. of Urban Studies
and Planning as of 2014) in "Incomplete and Incremental Plan
Implementation in Downtown
Providence, Rhode Island, 19602000", published in the Journal of
Planning History,Vol. 5 No. 1 2006.
ED
>~
lL
F-
RJ.ds
Highw
1i
~C5N
9T
9ys
.,
1.Downtown Providence in 1958 (Journal of Planning HistoryVol.5 No. 1, 41)
~JI/
air
1D72
~j
-N
A-j\
1/
toi
Oip7\'
\
K-
Fxistia
VI
=7
j~ !
A.p
Zi7,
N
7Zp-T
T-~H~h
3_\
R,
Sp-
t\JL
p\w7/
li.2. Various proposed masterplans for downtown Providence between I 960-2000 (Journal of Planning History,Vol.5 No. I,
45,48,52,55)
In this study, Ryan looked at seven masterplans proposed for downtown Providence from 1960-2000.
The study concluded by stating,"This study concluded that Providence's downtown plan implementation,
successful as it may have been, was highly incomplete, representing at best the implementation of only a
portion of the propositions made by the plans. Second, the study concluded that the city's downtown plan
implementation was an incremental process where partially realized plan ideas could be appended by later
plans and where each plan's proposals confronted an increasing number of realized elements from previous
plans" (Journal of Planning History,Vol.5 No.1, 60).
Ryan's study gives a glimpse of a real case of unintended incremental planning and implementation due to
the unfeasibility of teleological masterplanning. He points out that despite Providence's often-lauded effort
and success on downtown planning implementation,"less than 25 percent of the individual ideas that had
been proposed in Providence's downtown plans since 1960 had been implemented even in part" (Journal
of Planning History,Vol.5 No. I, 37). Of the seven plans proposed for downtown Providence, only four actually experienced any implementation. Interestingly, the plans proposed overtime became more modest and
conservative, one after another, which can perhaps be an indication of planners and designers noticing past
trends in which large ambitious plans often face no implementation. Since modest and conservative plans
are more attainable, they may have decided not to waste time coming up with "big ideas" and instead focus
on what is immediately at hand.
Ultimately Ryan raised the following key questions to his own study: "Why were certain portions of the
plans realized but not others? Were some plan ideas easier to implement than others, or was implementation the result of serendipity?" (Journal of Planning History,Vol.5 No. 1,58).These questions, however, are
impossible to answer without looking deeply into the historical accounts of the city in terms of political
relationships, economic climate, policy changes, and designs altogether, which are beyond the intended
scope of both his study and this thesis.
One of the most critical issues with teleological masterplans is perhaps the fact that if implemented completely, the parts that are developed at the later stages are obsolete. Since masterplans take decades to
implement, they are usually developed in phases. By the time the early phases are done, the city around
the development site would have changed significantly, rendering the rest of the plan to be out of context.
We may argue that today, cities can be built much faster than even ten years ago. What used to take a
century now only takes a decade, which might mean there are more masterplans implemented through
completion, negating my argument that teleological masterplans are unfeasible.
It is important to clarify that this thesis does not challenge the existence of the masterplan. Rather, it proposes a different method at arriving to one, accepting the need for a finished masterplan to be given to
the client, developer, and contractor.With the speed of development we have today, it is even more critical
that we think about the masterplan differently. Otherwise, we may very well end up with a bunch of plan
Voisins all over the world.
ii.The use of generative computation techniques in practice

In many contemporary architecture and urban design practices, designers use computers and computation
techniques often to generate some form of a masterplan. Sadly, most of the outcomes are in fact much less
legible than a Nolli plan and produce less coherent and intentional designs than what a more conventional
urban design method would. Granted, each individual is entitled to his/her own belief on good urban form,
but many urban design schemes produced today by computer and parametric techniques are residues of
interest and passion for the tools and techniques themselves. Many computation-based urban schemes
today, including this thesis, are still early explorations, but I hope to take a step towards bringing our views
on computation techniques away from digital obsession and towards a more pragmatic use.
One of the most prominent figures in today's parametric design is undoubtedly Patrick Schumacher, whose
name is almost synonimous with the word "parametric" in today's design society. He sees parametric design as a style, which he refers to as Parametricism. For him, Parametricism is the next big architectural
paradigm after Post-modernism and Deconstructivism. His approach to parametric design is not based
on pragmatism. Schumacher accepts that his methodologies are considered to be avant-garde. In his essay "Parametricism as Style - Parametricist Manifesto" (2008), he writes, "The key issues that avant-garde
architecture and urbanism should be addressing can be summarized in the slogan: organizing and articulating the increased complexity of post-fordist society.The task is to develop an architectural and urban
repertoire that is geared up to create complex, polycentric urban and architectural fields which are densely
layered and continuously differentiated."
',Po
I ii. I. ZHA's Kartal-Pendik Masterplan (Parametricism -A New Global Style for Architecture and Urban Design)
10
One of the most important

point that Schumacher makes in
his approach to parametric urban design is what he refers to
as "deep relationality", which he
explains is aimed "to integrate
the building morphology - all
the way to the detailed tectonic
articulation and the interior organization. Parametric Urbanism
might involve parametric accentuation, parametric
figuration,
and parametric responsiveness

as registers to fulfill its ambition of deep relationality" (Parametricism - A New Global Style
for Architecture and Urban Design).
It is clear that Schumacher wants
to
emphasize
his parametric
implementation through unmistakably striking visual representation.The drawback here is the

emphasis on the pictorial quality
of an aerial-view of a city or a
city plan over the performance
and the actual quality of the city.
He approaches urban design the
same way he approaches architecture
design. Unfortunately,
unlike architecture, urban design

cannot afford to prioritize visual
and formal clarity over function
and performance.
Schumacher has a very strong
belief on how architects must
I ii.2. Schumacher's monotonous, uniform, and totalitarion parametric urbanism.

(Parametricism -A New Global Style for Architecture and Urban Design)
II
dictate what architecture is to non-architects. In an interview titled "On Parametricism" published on his
website (patrikschumacher.com), he compares architecture to science in a way that scientists prescribe
proper scientific knowledge to the world, and thus "architects - and only architects - who determine
through their collective architectural discourse what is good, appropriate contemporary architecture."
This view and understanding of architecture carry over to urban design, which is a critical error common
to individuals with an architectural background who become involved in urban design.
As mentioned before, Schumacher's primary agenda is visual clarity of parametric design, which means he
has, in his mind, deployed a pre-conception of the visual characteristics of parametric design. He mentions
that late modernist architects have used parametric modeling to "inconspicuously absorb complexity",
whereas he, as a parametricist, "pushes in the opposite direction and aims for a maximal emphasis on conspicuous differentiation and the visual amplification differentiating logics". For him, parametric design aesthetics "is the elegance of ordered complexity and the sense of seamless fluidity, akin to natural systems"
(Parametricism -A New Global Style forArchitecture and Urban Design). Ironically, Schumacher's criticism
of late modernist architects is only supported by his own version of guiding parametric method to achieve
his own personal visual agenda which, although happens to be different that that of the late modernists, is
still a subjective and personal endeavor nevertheless.
This is not to say that Schumacher is oblivious to the real essence of parametric design. He wrote,"Parametric variations trigger 'gestalt-catastrophes', i.e. the quantitative modification of these parameters trigger qualitative shifts in the perceived configuration. Beyond object parameters, ambient parameters and
observer parameters have to be integrated into the parametric system" (Parametricism - A New Global
Style for Architecture and Urban Design).
However, Schumacher chooses to focus on the visual changes obtained from tweaking various parameters
within a parametric system. He believes that parametric design can address complex urban issues and "articulate them with all their rich differentiations and relevant associations". For him, one of the most significant advantages of designing parametrically is that "the danger of overriding real-life richness is minimized
because variety and adaptiveness are written into the very genetic make-up of parametricism". Ironically,
while claiming to avoid dumbing down complexity and diversity, his design outcomes are strikingly monotonous and uniform, which is primarily caused by his ambition for "deep relationality" as he attempts to
package every aspect of urbanism into one parametric system.
12
2. URBAN DESIGN VS.ARCHITECTURE ATTHE URBAN SCALE
I suspect the term "urban design" is quite ambiguous for many. Many architects who are involved in urban
design projects often approach them as if they are architecture, unknowingly or not. In Urban Design as
Public Policy: Practical Methods for Improving Cities (1974), Jonathan Barnett wrote,
" A city planner, it seemed to us, was someone who was primarily concerned with the allocation of resources according to projections of future need.Allocating funds for a capital
budget is a series of planning decisions, because it involves determinations of need for, as
an example, a new school in a particular district, and balancing off that need against that
of other areas.
Architects, on the other hand, design buildings.They prepare a set of contract documents
so that the building, let us say a school, can be constructed, and they take legal responsibility for the process.
There is a substantial middle ground between these professions, and each has some claim
to it, but neither fills it very well." (Barnett,
186)
Unlike architecture, urban design cannot afford to take a single set of ideas that aims towards idea clarity,
which typically ends up with having a "thing" as an organizing datum in a single design act, whether it is an
axis, a mega structure, infrastructure (a term so loosely used these days), an open space, a topography map,
etc.This approach is too one-dimensional, regardless of how complex or well-organized the designer claim
his/her project is.
For years, many architects have gained interest in the issues of urbanism.Through architecture, many perceived urban issues are internalized within one holistic structure.There are many early iconic examples of
this attitude including Le Corbusier's Unite d'Habitation, Paul Rudolph's Lower Manhattan Expressway, and
the theoretical works of archigram.This attitude of internalizing urbanism within large architecture persists
until today.Architecture competitions such as eVolo is infested with "vertical city" schemes. Contemporary
design practices such as MVRDV and BIG are not exempt from these proposals either. For at least a century now, many architects cannot seem to move away from this idea that urban issues are somehow similar
to architecture issues and therefore the two can be combined and approached as one.
Architects are taught and trained to strive for idea clarity in a design scheme. Urban design, if approached
this way, becomes too focused on the completed state of the design scheme. Designers often describe
their urban design schemes the same way they describe architecture design schemes.A common scenario
2.! . Unite d'Habitation (Photo by Paul Koslowski, 1997. fondationlecorbusierfr)
13
14
2.2. Paul Rudolph: Lower Manhattan Expressway (US Library of Congress: loc.gov)
might sound something like this: "The idea of this design is about the separation of programs. The residentials
are clustered around the site with a commercial core in the center of each cluster. In the middle, there is a main
street that connects to the rest of the city and branches off to each neighborhood." In reality, it is not possible to
describe urban areas like downtown Boston,Tokyo, or Paris as being organized by one set of simple ideas.
This organizational strategy that stemmed out of architectural discourse is undoubtedly still commonly
practiced today in many professional urban design offices.
This attitude that mixes urbanism with architecture has resulted in a few common strategies in urban
design. One example that has become a common punching bag for designers and planners as a disastrous
urban design strategy is Le Corbusier's Plan Voisin for Paris (figure 2.1 ).This is a prime example where an
architect acts as a totalitarian, imposing his subjective and very personal opinion onto a city.Although this
project in particular was never realized, its influence was so powerful that it manifested into reality through
many other developments, such as the Barbican estates in London,Ten Eyck housing in Willamsburg (New
York), and the Empire State Plaza in Albany (NewYork).
We may categorize this urban design strategy that proliferated from architectural modernism as one that
simply aggregates a series of buildings across a city.This approach to urban design results in a scheme
that looks like socialist housing blocks, which often carries negative connotations with them for multiple
reasons.
2.3. Le Corbusier's Plan Voisin (fondationlecorbusier.fr)
15
16
2.4. From top: Ten Eyck housing, Empire State Plaza, Barbican Estates
17
Another common strategy towards urban design by architects

is through the use of large holistic
architectural intervention. This
includes Peter Eisenman's Penn
Station project and Morphosis'
World Trade Center design proposal. When one architectural
design agenda, which includes a
consistent design strategy and
visual aesthetics, governs an urban design project, we arrive at
projects like these.This strategy
believes in the power of architecture as the unifying element
in a city that brings multiple elements together into one cohesive design scheme.
The most recent implementation of architectural thinking into
urban design utilizing contemporary digital tools and techniques
are shown by projects such as
Xi'An
International
Horticul-
tural Expo by Plasma studio,
GroundLab, and LAUR Studio

(figure 2.4).These projects often
result in "incidental" but strong
and distinct form and overall
plan configuration. Often closely
related to landscape urbanism,
they are usually influenced by a
set of data/information that affect the shape of the design in
one way or another. This set
of information
2.5. Top: Eisenman's Penn Station design proposal (New York's Pennsylvania Stations, 192) and Morphosis'World Trade Center design proposal (morphosis. corn)
commonly in-
cludes things such as sunlight
18
2.6. Xian International Horticultural Expo (plasmastudio.com)

information, flood zones map, transportation, circulation (vehicular and pedestrian), topography map, and
population density, among many others.
We don't see too many of these designs implemented today. Much like architecture, these large scale
designs rely heavily on their full completion in order for them to function properly as intended. As our
technology advances, our construction becomes faster and more sophisticated, allowing more of these
projects to be built. If we keep addressing our urban issues as if they are architecture, we may very well
end up with a world full of theme cities.Thus, we cannot afford to think of urban design holistically, as if it
is one large object, one overarching idea, or one consistent design manifestation.
I suspect that this architectural approach to urban design was signified and proliferated by famous modernist architects. Le Corbusier and Louis Kahn thought of cities as if they are architecture. In his sketch, Kahn
explicitly wrote," The plan of the city is like the plan of the house", comparing streets to rooms and the sky
as ceiling. This attitude towards urban design assumes that a single grand idea can support an entire urban
scheme as it does architecture. It believes that a city's structures and systems can be conceived and drawn
by a single person as the genius mastermind.
2.7. Xi'an International Horticultural Expo. From top: 2005, 2009,2011, 2014.
19
20
9110 f
414* V,
J-
ftP*
&,e
2.8. Kahn's sketch of city to house comparison (Louis I. Kahn Collection, University of Pennsylvania and the Pennsylvania
Historical and Museum Commission)
To some extent, the components of a city are somewhat comparable to those of architecture. However,
when an urban design is ideologically treated as architecture, it often ends up unrealistic, impractical, and
downright ridiculous.This trend of architects generating urban designs has prevailed until today.As a result,
designers often manifest their personal design ideologies through urban design the same way they do in
architecture design. Modernist architects' urban design schemes revolved around one or a few ideas based
on certain observations of the city. For example, Kahn designed a plan for Philadelphia solely based on
traffic patterns and speed. Le Corbusier desired to separate different modes of transportation so they do
not clash with each other.
pV
2.9 Le Corbusier's Ville Contemporaine: Contemporary City for three million people (fondationlecorbusierfr)
In Urban Design as Public Policy: Practical Methods for Improving Cities (1974), Jonathan Barnett wrote,
"Architects and planners have inherited some funny ideas about themselves as the keepers
of the sacred flame of culture and the guardians of society's conscience.There has been
a tradition that a true professional, and, certainly, a true artist, should not be too closely
involved in the day-to-day process of government, or politics, or real eastate development. Instead, he has sent his instructions to the policy makers as manifestos or visionary
drawings, and, not suprisingly, the policy makers usually find them impossibly idealistic and
irrelevant to the problem at hand" (Barnett, 6)
21
22
3. PRECEDENTS AND SIMILARTHOUGHTS (INCREMENTALAND COMPUTATIONAL)
i. Early thoughts in computational and incremental urbanism

The idea of approaching urban
design computationally or incrementally is nothing new. From
descriptive to generative rulebased urban design to incremental design and implementation, there are multiple designs
and ideas in the past that have
attempted to pursue an alternative to the masterplan. So much
so that phrases like "death of the
masterplan" has become a clich6.
Centuries ago, in Ten Books of
Architecture, Vitruvius wrote "a
logic of a set of performance
criteria and their parameters to
be balanced in the city. Environmental orientation, exposure, altitude, climate, and flora are set
out as the basis for locating the
city to create optimal conditions
for health. Military defenses,
roads, and sea ports are also initial considerations in setting out
the position and height of defenses, and the location and design of the buildings and spaces
of the city." (Verebes, 68)
Early computation based urbanism focused on describing cities.
123456
123
k12345\
123456
34s
3i. 1. Semi-Lattice structure (top) vs. Tree structure (bottom)

(Alexander, "A City Is Not A Tree")
23
In 1965, Christopher Alexander
suggested that natural cities,

2
those that have "arisen more or
less spontaneously over many,

many years", have the structure
of a semi-lattice while artifically
A2av
A,-..
&...3 d
da-d
an8(2
designed cities, those that "have

\
been deliberately created by designers and planners", have the

structure of a tree. (Alexander,
"A City Is NotATree")
In
1986, Michael Batty and Paul
Longley in the University of

Wales Institute of Science and
Z,
Technology published a study on

fractals as a way to describe urban boundaries. Derived through
mathematics and computation,
fractals are shaped through a series of mathematical operations
that result in a large complex
patter based on simple patterns
aggregated through
a certain
rule/algorithm.
Batty
explained
that
urban
boundaries are typically defined

through a sruvery. He wrote,"
Typically, allotments and other
urban fringe land uses
were
excluded, villages linked to the
urban area by ribbon develop-
12Miles
ment were included, man-made

N
alterations to rivers and coast

were included, while large landed estates which subsequently
3i.2. Fractal-based description of urban form

(Batty, I/ -12)
become part of the urban fabric
24
~~iii
were only included if development had
surrounded them."
Ni/
/
(Batty, 9)
N
Frustrated and unsatisfied with

the inadequacy of methods to
define urban boundaries, Batty
suggested that fractals can provide a more accurate and natural
look at urban boundaries.
The most directly related precedent to this thesis perhaps
comes from Christopher Alex-
Map ofthe
completed project
ander's A New Theory of Urban

Design (1987). In the beginning
of the book,Alexander wrote,
"Urban design has a sense of dilettantism:as if the problem could
be solved on a visual level, as an
aesthetic matter. However, at
least the phrase "urban design"
does somehow conjure up the
sense of the city as a complex
thing which must be dealt with
in three dimensions, not two.
We
have therefore used the
phrase urban design in the title
C -/
--
4U
of this book, since it seems to

us that urban design, of all existing disciplines, is the one which
comes closest to accepting responsibility for the city's wholeness.
But we propose a discipline of
3i.3. Completed urban plan (AlexanderA New Theory of Urban Design, 106)
25
A
I
3i.4. Completed model (Alexander,A New Theory of Urban Design, 107)
V
W)
............
.....
U:'h~~i
NE
IZ~~LEE
~'
1L~~4Ir' I\4
~U>;?:~
"-i
-U
NE
1111 ~ET~ HE
I'
I'
',',
27
AFUE~~
Ni-
--- --- --
O\OK#9*T
P-197
ZN -
If
3i.5 Alexander's urban design steps overlaid to better show progression.

(Edited from A New Theory of Urban Design, 104-230)
28
urban design which is different, entirely, from the one known today.We believe that the task of creating
wholeness in the city can only be dealt with a process. It cannot be solved by design alone, but only when
the process by which the city gets its form is fundamentally changed." (Alexander, A New Theory of Urban
Design, 3)
Alexander's primary objective described in this book is to create "wholeness", which he describes as the
"feeling of organicness" of a city, similar to the way he described a "natural" city in "A City Is Not A Tree".
Alexander's design process, guided by "a series of seven rules", attempts to simulate an imaginary process
of city development over a five-year period. Similar to other theories related to urban growth, this one
seems to also propose a logic for urban growth that both guide and predict urban developments. Of the
seven rules, rule #3 is the most obvious example of this, which serves as "an answer to the fundamental
question: What shall we build in any given place, where a project is to be undertaken.This question does
not ask how it is organized, how it is designed, what character the architecture has... but simply the most
fundamental question of all: What is it? What is going to be there?" (A New Theory of Urban Design, 51-53)
Based on the rules he designed, we can see that Alexander's theory of rules is accompanied by a personal
view on what a city should be.Alexander, who is against the machine-like method of urban design, which
he claims is often "a result of considered, channeled, information", ends up projecting a classical renaissance
city to the site.Acknowledging this, he recognizes the drawbacks of his piecemeal design approach.
However, Alexander points out perhaps the most important thing anyone should keep in mind while designing a city: the fact that the formation of cities is a result of multiple interests that often do not correlate
with each other.The irony here is that the rules presented in his theory come as a set of policies that the
citizens must abide by, similar to a planning policy, ultimately dictating the subsequent urban forms. Did
Alexander simply create a set of planning policy and played a simulation game of what could happen if these
policies are implemented in a city? The rules in the policy he created are meant to guide developments.
Any developer may choose to build as long as the development falls within the constraints of the rules to
maintain "wholeness". In this case,"wholeness" is subjectively, and perhaps conservatively, defined as having
similar qualities to those of an old European city.
ii. Recent and contemporary thoughts in computational and incremental urbanism

In recent years, architects, designers, and planners began to gain interest in informal low-income poor
settlements. Some are drawn for altruistic reasons, proposing design solutions that can give better living
standards for those who live in poverty. Others are drawn to the seemingly random and complex ordering
growth logic of these settlements, claiming them to embody bottom-up urban design that possesses the
ideal natural growth pattern of a city (decoding the DNA of an urban growth). Naturally, individuals
7~K
740
77
29
-7
7
3ii. I SuperKampung (MVRDV, 356-35 7)
30
interested in computational, incremental, and parametric urban

design are attracted to informal
settlements. One recent study
on this topic was published in a
book called The Vertical Village by

MVRDV.
The study looks at East Asian
cities and refer to their origins
as "urban villages that are built
up of small scale, informal, often
'light' architecture" (MVRDV, 6)
in response to what they call the
relentless "Block Attack", a common phenomenon in Asian cities
where massive slab and block
towers are erected with repetitive housing units and floor plans
V, N
without any cultural context.

One of the projects in this study
is titled "SuperKampung" (figure
3ii. 1), which focuses on bottomup
incremental
developments
through fictional scenarios. The

study looks at Jakarta's Kampungs
(villages/small
e1,9
settle-
ments) and assumes an optimis-
"'X
tic scenario where "there will

be improvements by the central government in the coming
years" (MVRDV, 300). Here, the

study does not propose a design
scheme or an urban plan. Rather,
it experiments with incremental
growth by assigning each neighborhood a unique identity and
3i'.2 Patchwork Urbanism (Design After Decline, 219-22)
31I
imagining a scenario that plays out over the period of ten years. Each scenario is essentially the result of a
group of designers daydreaming, pondering, and going bananas at what each neighborhood might become
with few simple assumed elements in each neighborhood. It is as much an authoritarive design approach
as Le Corbusier's, cloaked in a haze of informal settlements. For what its worth, it is a comic book about
a few poor settlements in Jakarta.
On the other side of incremental urbanism, Brent Ryan, through the study of urban planning and policies of
post-industrial American cities, proposed an urban phenomenon called "patchwork urbanism". He wrote,
"The irregular levels of density, habitation, building stock, and open space in the shrinking
city are the precise opposite of the level, homogeneous urbanism of the historical industrial city that can still be seen in intact areas of older cities.The new urban patter resulting
from shrinkage may be called patchwork urbanism. The patchwork is dynamic, shifting and
changing over time as abandonment, demolition, and new development each make their
mark.
Patchwork urbanism implies that urban design, rather than acting homegeneously across
urban space in the same manner as zoning, acts in a strongly differential mode, influencing
certain areas of the city much more than others.This mode of urban design is not unique
to the shrinking city; market development is generally far more concentrated in some
parts than in others, so parts of cities have historically changed at a greater rate while others changed slowly or not at all." (Design After Decline, 213-214)
Based on this, we may imagine an urban design strategy that is more incremental and patch-like.The top
image of figure 3ii.2 shows a typical shrinking-city neighborhood before its decline.These neighborhoods
are often "moderate-density developments with one- and two-family apartments, with commercial buildings on major streets" (Design After Decline, 219). The middle image shows the neighborhood in decline,
assuming "a familiar appearance of vacant lots, abandoned houses, and scattered, still-inhabited buildings"
(Design After Decline, 220).The bottom image shows how a "progressive urban design could reconstruct
such neighborhoods with new housing that impoved upon the old, with new public and semipublic open
spaces, and with new street patterns" (Design After Decline, 221).
Both examples are equally fictional, though Ryan's tale is told with assumptions grounded in historical and
current trends, making it a more serious study.Though not explicit, both examples imply that the final
results shown are at least possible realities. I realized telling a story or a scenario of a city's development
would be misleading in delivering the arguments of my thesis as it implies a type of prediction. Instead, the
scenarios in this thesis are examples of design processes that designers would go through while designing
an urban scheme using the proposed methodology.They are used as stages for experiments discussed in
greater detail in the following chapter.
32
4. EXPERIMENTS AND REPORTS
i.The basic idea (components and relationships)

As the title suggests, the design methodology proposed in this thesis is recursive by nature. It focuses on
the components in a city and their relationships to each other, contained within an operation.An operation
contains multiple components and define their relationships. Multiple operations perform together to
ultimately create an urban design scheme.
Components are objective. They are elements that exist within a city, such as buildings, building facades,
parcel dimensions, sun exposure, building shadows, open spaces, parks, waterfronts, etc.The list is almost
endless. It is up to the designer how many components he/she wants to consider during the design process. Generally, more components require more complex relationships and operations. However, it doesn't
mean designers need to include as many components as possible during the design process. A good designer knows which components are important for the project and which are irrelevant.
4i. 1. Some of the components taken into consideration in this thesis: (from left) sunlight, building mass, large open space,
building scales and parcel sizes
Relationships, at least at this point, are subjective. One component may have multiple relationships with
other components based on the operation that contains it. Relationships define how each component affects others within an operation. It is up to the designer to define these relationships based on what kind
of design scheme he/she wants to create.
For example, as shown in figure 4i.2, the top two diagrams show two components, in this case building
mass and building shadows, with two different relationships.The left diagram shows a relationship where
the base of each building cannot be covered by any building shadows, resulting in a design scheme where
the building masses are spaced further apart from each other, exposing each of their street levels to receive
33
4i.2. Different relationships defined by different designers based on different design agenda
direct sunlight exposure.The right diagram shows a different relationship of the same two components,
where building shadows are allowed to cover only the base of other buildings, resulting in a scheme where
direct sun exposure cannot reach the ground spaces of each building.
Another example of this is shown by the bottom two diagrams of figure 4i.2. In this case, the two components being considered are surrounding (existing) building sizes and new building sizes along with their
placements.The diagram on the left shows a relationship where the designer decides to place new buildings of similar sizes with their immediate surrounding context, creating a similar condition with what
34
already exists on site.The diagram on the right shows a relationship where the designer decides to do
the exact opposite, placing large new buildings around existing smaller buildings and small new buildings
around existing larger buildings. Depending on what kind of city the designer wishes to design, different
relationships will result in different urban design schemes.
ii. Site selection

In order to test the proposed urban design methodology, I have chosen a site in Greenpoint, Brooklyn.The
site was simply selected based on Mayor Bloomberg's PlaNYC 2030, a long-term planning agenda to anticipate "a growing population, aging infrastructure, a changing climate, and an evolving economy" (PIaNYC,
3). It designates various areas of New York City for various future developments. Based on this plan, parts
of Greenpoint are planned for increased residential capacity in the form of new neighborhoods centered
around residential and commercial developments.
Recent. Planned, and Potential Initiatives

to Increase Capacity for Residential Growth
City-initiated Rezonings Sinc e 2002

PROMOTESRESIDENTIAL
M
*AREAS
OF OPPOiTUM
PROMOTES
BOTH DEVELOPMENT
ARAuWRN MtE EKSIEWAY SIATJON
I COMERCIAL DEVELOPMENT
NEIG4BORHOOD PRESERVAJ 1ON

AND PRESERVATI
-.Aar-
"
2,t7
4ii. 1. PlaNYC 2030 potential initiatives and zoning maps (PlaNYC, 22-23)
PlaNYC provides concrete program agenda for the site which allows the thesis to focus on experiments
without having to worry too much about site programming and analysis.The site selection is not critical
to the successes and failures of the project, but selected area in Greenpoint provides a variety of urban
scenarios that include a waterfront, a relatively regular urban fabric, large industrial and post-industrial
buildings, low-rise residences, and commercial activities.
4ii.2. Site: Greenpoint and its waterfront
35
36
iii. Operations
Each design scheme produced in this thesis is a result of a series of operations that are constantly tweaked
and developed as more experiments are conducted in order to find out proper relationships between
components and, more importantly, how tweaking relationships explicitly affects the urban form. Each operation contain multiple components and specific relationships between them.
There are a few operations I created during this thesis, each of which plays a specific role in the creation
of an urban design scheme. Each of them was tested on a generic rectangular site shown in figure 4iii. 1.
The first operation is called RoadNetwork (figure 4iii.2). It deals with the formation of the urban fabric. It is
a way to create circulation system and deal with an oversized site by splitting it up. It works by connecting
existing roads that end at opposing edges of the site. In this operation, a site is identified. In this case, it is
recognized as one oversized block indicated by the red color (figure 4iii.2 top).The threshod for tolerable
block size is determined by the designer. Depending on the number, this initial size may be either appropriate or oversized. If it is oversized, we proceed with the RoadNetwork operation.
4iii. /. Generic site to test various operations
37
After recognizing the site to be

oversized, it identifies existing
roads that lead to and end at
the site's edges. It then connects
the ends of these roads that are
located at the opposing side of
the site (figure 4iii.2 middle).This
can be tricky if the site's shape is
irregular as it becomes ambiguous as to which sides are opposite of which.
Now that the site is split, the
color indicator turns green, indicating the resulting split blocks
are now within the tolerable defined size and they are no longer
oversized. Finally, new city blocks
are created, generating a new urban fabric.
The second operation is called
BuildingHeights. It deals with the

sizes of new to-be-developed
buildings as well as suggesting a
build order based on surrounding context's density.The "recursive" part of the proposed methodology is demonstrated clearly
here.
The operation
is divided into
two sub-operations.
The first is called CatalystSequence (figure 4iii.3).This operation makes sense when used for
developing large buildings that
4iii.2. RoadNetwork operation
38
can be a catalyst of urban activity. It works by taking the least

dense area around the site and
suggests that it should be developed first during the design
process. The grayscale field (figure 4iii.3 top) shows the surrounding density value of each
parcel in the site. In this case, the
darkest parcels have the lowest surrounding density value.
The designer decides how many
buildings he/she wants to place
on each sequence. Ideally, one
building at a time would produce the most accurate result
as it will affect the density field
the least in one sequence. In
this case, I decided to place five
buildings per sequence.
Because the method is recursive,
once the first design sequence is
done, we must re-evaluate the
entire site again based on what
we just did. The bottom image
of figure 4iii.3 shows a different field of surrounding density,
which takes into account our
own design moves, affecting all
of the sequences ahead.
The second
called
sub-operation
GrowthSequence
is
(figure
4iii.4). It is the exact reverse of

CatalystSequence. This
opera-
tion makes sense when used for

extending an existing urban
4iii.3. BuildingHeights operation (CatalystSequence)
39
environment with smaller buildings. The idea is to build where

there are a lot of urban activities
already since smaller buildings
do not have the same capacity to
house a lot of people and generate urban life on their own.
The third operation is called
OpenSpace. It deals with how a
large open space, such as a waterfront or a large park, affects
the design sequence as well as
the size of the buildings. This
operation works together with
BuildingHeights if a large open
space is present at the site. In
the example shown here (figure 4iii.5), we assume an active
and successful waterfront public amenity, strong enough that
it makes sense to design along
the water earlier in the design
process. It affects building's size
based on how far the building is
to the nearest edge of the water.
The diagram on the top shows a
field of density, similar to those
in BuildingHeights operation that
suggests the design sequence.
This field, however, is affected by
the waterfront edge significantly,
placing greater priority on the
parcels along the water regardless of them being away from
existing buildings. The designer
must define the level of success
4iii.4. BuildingHeights operation (GrowthSequence)
40
of the waterfront. Depending on

how he/she defines it, the waterfront may not be strong enough
to pull priority towards the parcels along the water. It may not
affect the design sequence at all.
It may place even greater priority on parcels away from the
waterfront in the case where
the waterfront is completely undesirable for development and
public activities.
The fourth operation is called
SunShadow
(figure
4iii.6
and
4iii.7). It deals with large and

tall buildings to ensure that the
subsequent buildings in the design process receive ample daylight.The designer decides what
time frame he/she wants to take
into consideration in this operation. This decision can be influenced by the program of the
site (morning sun for residential,
afternoon/evening sun for parks
and playgrounds, etc.).
It works by removing parcels
that are covered by shadows
of buildings in each sequence,
ensuring that no buildings are
placed there.The base sequence
is generated by BuildingHeights
operation.
Figure 4iii.6 shows the operation
using the morning sun (9:00 AM)
4iii.5. OpenSpace operation
41I
as a component whereas 4iii.7 shows the operation using the noon sun (12:00 PM). In these examples, the
relationships remain the same, but the components change. By using different components, we arrive at a
different design scheme.
The version of the operation shown here has a major flaw.As shown by figure 4iii.7, because the operation only removes shaded parcels on each sequence, parcels that are not shaded remain even if developing
them will cast shadows over previously developed buildings.This problem was fixed in later versions and
operations.
The fifth operation is called BuildingTowers (figure 4iii.8).This operation stemmed from the previous operation, SunShadow.Just like it,the sequence here is also generated by BuildingHeights operation (surrounding
density field). It is used to create a tower-based neighborhood complex. It works by taking parcels selected
by BuildingHeights and creating a low-rise tower base (primary), whose height can either be determined
by BuildingHeights or manually entered by the designer. It then takes the parcels adjacent to it, checks their
density values, and creates extensions of the tower base (secondary) where the values are high enough. It
also creates a tower on top of the primary tower base, whose height can also be determined by BuildingHeights or manually enterred by the designer.
BuildingTowers includes a very important feature that SunShadow does not.Whenever a parcel is selected,
it checks whether placing a tower there will cast shadow over those created in previous sequences. If it
does not, the operation will place a tower base, check for tower base extensions, and place a tower. If, however, it casts shadows over previously created buildings, it will not do any of the former steps and eliminate
the parcel from future considerations.These parcels are highlighted in green in figure 4iii.8.
42
4iii.6. SunShadow operation (9:00 AM)
4iii. 7. SunShadow operation (12:00 PM)
43
J0
554
77
X 74/J.
_24i1
Z
71,7
,4-4el,
J~
"0
27
-A
I-w
'-.
2<Z//
4iii.8. BuildingTowers operation
44
iv. Experiments
The ultimate goal of the experiments is to demonstrate that
the proposed methodology can
be adopted as a standard of urban design practice. Thus, the
first experiments were only
concerned with making an implementable urban plan. From
this point on, each experiment
will be referred as Combined-X,
where X is a number that identifies the iterative version of the
experiment (i.e. Combined- 1,
Combined-2, etc.). Each of the
Combined experiments is conducted on the same exact site
in order to reduce as much unnecessary variables as possible
in the experiments.
It is important to keep in mind
that the overall design agenda of
each experiment is to create a
new mixed-use neighborhood
that consists primarily of residentials, which is derived from
PlaNYC's planning agenda.
The images on the left (figure
4iv.I) show sequences of the
first Combined experiment
(Combined-I), which utilizes
two operations: RoadNetwork
and BuildingHeights.There are a
few assumptions made when the
experiments began.The first
-mem
45
_-
.O.Il
memo
-memo
4iv I. Combined- I experiment sequences
46
is that new blocks and parcels

structure are similar with the
ones that already exist around
the site.The second is that new
buildings have similar sizes and
heights with those that already
exist around the site. These assumptions are made as part of
my own personal design agenda.
The experiment could have been
conducted with a different (but
consistent throughout the various Combined versions) agenda
and the overall goals would still
remain.
Combined- I showed a degree of
success in generating an urban
fabric with a network of roads.
However, in addition to ignoring
the waterfront, the experiment
was way too slow because each
sequence of the recursive process only calculates one parcel
at a time.
Combined-2 (figure 4iv.2) is the

most time consuming experiment in this entire thesis.There
are 247 recursive sequences in
this experiment because each
sequence only calculates one
parcel at a time.The purpose of
this experiment was to simply
see what kind of design scheme
the methodology might produce
with just two simple operations.
During Combined-2, settings
--snow
mum,
4iv.2. Combined-2 experiment sequences
47
48
within each operation had to be

tweaked every now and then in
order to produce a feasible and
implementable result. The experiment was inconsistent because the operations failed to
produce even the simplest acceptable urban design.
Combined-3
4iv.3)
(figure
is
the first experiment that shows

promise and acceptable urban
plan. Unlike the previous two,
Combined-3 calculates five buildings at a time in each sequence,
speeding up the process of completing the entire plan. However,
the operations involved in this
version ended up creating many
sharp triangular urban blocks
that obstruct the view corridors
from the city towards the water.
(figure 4iv.6, bottom) This was
the result of seeking the nearest
path from the existing road ends
towards the waterfront's edge
at the expense of other types of
the neighborhood's connections
to the waterfront.
In addition, this version introduces a set of bars and graphs
that show three
information
constantly changing at each sequence: surrounding
buildings
volume, new buildings volume,

and new buildings GFA.This feature was originally introduced
4iv.3. Combined-3 final urban plan
4iv4. Combined-3 experiment sequences
49
50
4iv.5. Combined-3 final axonometric

with the urban designer in mind. If a designer is to use this method in the process of designing, it should
take advantage of computer's calculations and constantly update the designer so he can always check how
much of the site context he's taking into account in his analysis in addition to the amount of real estate
he's creating with the design. This was thought to be helpful when a client gives a required GFA for an
urban scheme. Believed to be somewhat irrelevant and distracting from the true purpose of the thesis, this
feature was removed in the later versions.
Combined-4 (figure 4iv.7) is the first experiment that successfully results in an acceptable urban plan. Each
sequence calculates eight buildings at a time. In the images on the right, the red buildings indicate the newly
calculated buildings on each sequence.The lines radiating outwards from each of them point to the existing
buildings that are being taken into consideration. In all of the experiments, the early sequences rely on the
existing buildings as information sources.As the sequence progresses, buildings that were created during
the design process begin to replace them due to the recursive nature of the methodology.
4iv.6. Combined-3 bird's eye and street view
51
52
Combined-4 results in an urban

scheme that is simply an extension of the existing urban fabric,
both in plan and building sizes. It
preserves view corridors of the
neighborhood behind the site by
extending streets as straightly as
possible while still creating the
nearest waterfront edge. However, it still disregards the value
of the waterfront and how it affects the development along it.
Multiple experiments stemmed
from Combined-4, with
each
iteration improving and adding

operations to the previous one.
Combined-4-1 (figure 4iv.8) saw
the introduction of the OpenSpace operation, affecting the volume of buildings along the waterfront.
The application of OpenSpace
in a specific site such as this
one required the operation to
be tweaked from the generic
context it was created in. In this
case, each parcel's distance to
the nearest waterfront edge is
checked. If the distance is over a
certain number (defined by the
designer), the parcel should not
be affected by the OpenSpace
operation. This is to prevent a
diminishing effect of the waterfront and sets a threshold instead for how many parcels
4iv.7 Combined-4 experiment sequences
53
54
should be affected by the waterfront. This is based on the idea

that a parcel that is two blocks
away from the waterfront and
another one that is ten blocks
away can be considered as being
equally unaffected by the waterfront.
I~-
In addition to the introduction

of OpenSpace, the graph feature also saw an improvement.
The "new buildings GFA" chart
is color coded to show which
E1~
newly created building in the se-
quence has which new GFA.

As shown by the last sequence
in figure 4iv.9, the parcels immediately in front of the watefront

result in taller buildings. The issue here is that this resulted in a
rather large continuous wall-like
surface created by the facades of
each of these highrises.
ml
Combined-4-2 saw the introduction of two things: SunShadow operation and a new version
of parceling that created two
different parcel structures. So
far, all of the experiments have
only been using a generic New
York City parcel that mimics the
parcel structure of the existing
neighborhood around the site.
In order to break the continuous highrise facades wall along
4iv.8. Combined-4- I final urban plan
4iv.9. Combined-4- / experiment sequences
55
56
the waterfront, Combined-4-2

specifies a different parcel structure for different building typologies. In this case, the generic
New York parcel structure is
designated for low-rise developments (less than seven stories)
while a hypothetical grid is laid
out on larger blocks to create a
highrise apartment complex.The
large blocks perform as apartment complexes with a few towers and a series of open spaces
in the middle.
As the design method is recursive, some internal streets are
created through the New York
generic parcel blocks, connecting open spaces in the various
large apartment blocks (figure
4iv.10). In figure 4iv. 1I, the first
three
images show the first
blocks to be developed are the

large
apartment
blocks. The
fourth (bottom left) and fifth images (top right) show the New
York generic parcel blocks are
split with through streets running in them, connecting the
large apartment blocks created
in the previous sequences.
The main problem with this ver-
-~~
sion is that some of apartment

towers are so staggered, failing
to create acceptable open spaces in the large apartment blocks.
4iv. 0. Combined-4-2 final urban plan
4ivI I. Combined-4-2 experiment sequences
57
58
This is believed to be caused

by the sun angle setting in SunShadow operation used in this
experiment.
Thus, Combined-4-2a tests a
different sun angle to see if the
resulting overall urban scheme
would improve. Other than the
sun angle, everything else remains the same with the settings
of Combined-4-2.
The result is very similar to that
of Combined-4-2, but the open
spaces and tower placements
are significantly improved. The
open spaces are more continuous and less fragmented (figure
4iv. 12 vs 4iv. 10).
Combined-4-2a resulted in an
acceptable urban plan with a set
of open spaces connected by a
set of streets (public and semipublic), a mix of parcel structures and building typologies
with highrise towers near the
waterfront, and streets that connect the existing neighborhood
directly towards the waterfront
that simultaneously preserve visual connection over the water.
0~
a.
U'
Lu
-d
U-
So far, all of the sequences do

not represent an actual development sequence happening in reality.The sequence is a way
4iv. 12. Combined-4-2a final urban plan
4iv. 13. Combined-4-2a experiment sequences
59
60
Vp
...
..
..
4iv.14. Combined-4-2a final axonometric
61I
62
for designers to approach urban

design. It represents the design
process proposed in this thesis.
By thinking about it sequentially
and recursively, we are not attached with the absolute form
of the final urban plan.
Through Combined-4-3, I imagine a scenario where 4-2a design

scheme is being implemented
and developed. Over the course
of the development, the neighborhood around the site changes and develops as well. By the
~-
time the early phases of our

site's development
reaches a
certain point (figure 4iv.16, images 2-3), the rest of the plan is
now less appropriate since the
surrounding neighborhood is no
longer the same neighborhood
as the one when Combined-4-2a
was created.
ER\--
\rn
Lruu
Thus, Combined-4-3 shows an

alternative plan that maintains
the same set of relationships
about the city, taking into account the change in the surrounding neighborhood. Though
mostly similar, there are a few
noticable differences
between
I1Pm.
the plan of Combined-4-3 and

Combined-4-2a, mainly in the
southern generic New York parcel blocks, where most of the
neighborhood changes occured.
4iv. 15. Combined-4-3 final urban plan
4iv. 16. Combined-4-3 experiment sequences
63
64
~I
i~t
u-u"'
*p
~\
\~
A'
I
\I rJ.
ii~'Pu I
Emil
IE~z~\u
IbM
iL~3
4iv. I 7. Combined-5 final urban plan
A11
65
66
..........
-~
67
,~ -~
~-
A~
~.
-z~.
~i.
ij
~'v 4
4iv 19. Combined-5 final axonometric
.4
68
At this point, the methodology

has been proven to be feasible
in generating an acceptable and
"5i0
somewhat complex urban plan.

If we evalute the results of 4-2,
4-2a, and 4-3, we cannot describe
them as one single idea the way
architects often describe their
urban
design schemes. These
schemes are derived from a set

of components
and relation-
ships that remained consistent

throughout the design process,
performing recursively by calculating changing variables in each
sequence.
I proceeded to extend 4-2a inland, away from the waterfront,
simply to see what I would get.In
doing so through Combined-5,
I discovered more weaknesses
of the previous operations and
experiments. The final
urban
plan (figure 4iv. 17) shows a very

checkerboard-like
urban
pat-
tern, creating bad scenarios for

highrise buildings and inaccessible open spaces. In addition, the
highrises are too close to each
other (figure 4iv.19), meaning
the SunShadow operation was
inadequate in creating proper
apartment
tower
complexes
4V,
14
when used in a larger scale than

the previous experiments.
At this point, I re-evaluated my
4iv.20. Open spaces from the fixed plan of Combined-5
69
(j~W
~
___
"7,'
p
4gltNos
4 11iU
~7
UU\\I
I-P
aaBk-
fl1
'UIi
~Q/
f-k
4iv.2 1.Fixed urban plan from Combined-5
70
entire experiment process. The

computational and parametric
aspect of my thesis was never
intended to produce a system
that can create a flawless urban
scheme where every nook and
cranny of the scheme is solved.
Rather, it was meant to assist
designers in the process of urban design.Accepting this, I took
the result of Combined-5 and
quickly fixed the errors in the
plan in order to create a working urban plan (figure 4iv.21 vs.
4iv.17). If we evaluate the thesis
this way, then the experiments
can be considered successful in
achieving their goal.
However, one curiosity arose.
With this method of urban design, is it possible to come up
with a design scheme that includes a large object in it (e.g.
New York's Central Park)? Since
the method is recursive and sequential, it seems that it is impossible.
In the next experiment (Combined-6), I tried exactly just that.
I overlaid a piece of New York's
Central Park on the site and designed recursively with the operations I've created to see what
kind of issues and design I would
encounter. I imagined an urban
design project that has a
71I
72
Idsto
FU
F-low
L ---J
F
'Ii
Vj
Uj
El
4iv.23. Combined-6 final urban plan
F- w
woo
73
W-
.1
I~
is
4iv.24. Fixed urban plan from Combined-6
00
-9
9-
0-
C.'
SM-
C7/4~
'AAl
51
A"
75
large park as one of the client's

or city's requirements
Unfortunately, the large object
in this case was placed in the
beginning of the design
pro-
cess. This meant Combined-6

is the same with the previous
experiments, except that it has
a large open space in the middle (OpenSpace operation now
works along the park's edge in
addition to the waterfront edge).
This experiment did not try to
use a recursive and incremental
design method to create a large
park.At this point, it is uncertain
whether that is possible or not.
Regardless, the result of Combined-6 is satisfactory.The urban
scheme contains streets that
connect the existing neighborhood and the newly designed
neighborhood
to
the
park's
paths and waterfront edge, highrises along the waterfront and

park edges, mixed parcel structures with multiple building typologies, and multiple networks
of open spaces connected to
each other as well as to the waterfront and park by public and
semi-public streets.
We can imagine a much more
elaborate park design that has
much more complex edges with
L!T
Z,
PRar
~Li~j~
......................
..........
7774k
f/
~'~'I'
................
.....
.
(7IJI
\~r
x~f
~Pz
A>
'17,
/a
A-
CD
77
-j
i jr r
K>
#)riv7
--
4iv26. Combined-7 experiment sequences
78
4iv.27. Combined-7 final axonometric
79
80
J
IT
II
4iv.28. Combined-7 city park space detailed axonometric
81I
82
4i 29. Combined-7 waterfront promenade detailed axonometric

multiple park entrances.This would not diminish the effectiveness of the design methodology since it relies
on components and relationships, not form. Regardless of how complex the shape may be, a park will still
have the same components that were used in Combined-6: edges and entrances (vehicular and pedestrian),
We can even insert a less conventional public amenity such as the Highline (New York) and the design
methodology would still work just fine.The Highline has edges and a series of entrances spread along the
park.We may treat the edges differently than the way we treat the edges of a conventional park, but that is
the beauty of the design method.The components remain the same but the relationships are up to designers
to decide.
So far, all of the experiments have been more or less about creating a new neighborhood that resembles
its context in terms of blocks and parcels structure while creating new public spaces; a mixture of high,
medium, and low density developments; and a mixture of potential building typologies based on footprint
area, building heights, and location. So far, the experiments have stayed very close to PlaNYC's agenda, assuming the city wants to expand existing neighborhoods without disrupting too much of the current city
structure.
-v
4iv.30. Combined-7 small public square detailed axonometric
83
84
I decided to experiment with a completely different scenario in Combined-7. In this case, I imagine a large
developer that owns the entire site, has strong political and economic leverages, and therefore has more
control over what gets developed on the site.The client wants to create a new neighborhood identity
with a series of highrise apartment towers, offices, large retail complexes at the street level, a variety of
parks, and a strong waterfront identity.The client has a vision that this neighborhood is to become a new
urban center in the area with high population and strong public life through busy streets and a high level
of commercial activities.
With that in mind, Combined-7 saw the introduction of BuildingTowers operation, which is a more sophisticated way of creating highrise building complexes than what I have been using so far. Combined-7 results
in an urban scheme with streets that continue and connect surrounding existing streets into various spaces
within the site; a variety of open spaces that consist of parks of various sizes, locations, and contexts as
well as building plazas located in front of some of the large tower complexes; a small mixture of parcel
structures and building typologies; and a public waterfront recreational space.
The resulting urban plan is more complex than any of the previous one. Since I was not concerned with
strict continuation of existing streets in this scheme, I overlaid a rectangular grid (figure 4iv.26, image 2) on
the site, simply cropped by the site boundary.After BuildingTowers is completed (no more parcels can be
developed without having new towers cast or covered by shadows on/of those previously designed), there
are many leftover undeveloped spaces that are too wide and large to become streets and public spaces.
These leftover spaces are split into smaller blocks using RoadNetwork, the way the previous experiments
were done, and became smaller parcels with smaller developments (4iv.26, image I
0).This is a way to fill
up the leftover spaces with a second type of urban fabric and development.
Unfortunately, the rectangular grid proved to be too rigid for this scheme. Even though each tower base
complex has a unique configuration, all of the buildings individually have exactly the same size and shape.A
quick look at the urban plan reveals the inherent arbitrary grid that was imposed onto the site. Perhaps an
operation that creates a more appropriate grid should be devised as the base for running BuildingTowers.
Ultimately, each design scheme beginning in Combined-2 successfully created a variety of open and closed
spaces that can be appropriated as public and private.These spaces are most resolved in Combined-7 (figures 4iv.28, 29, and 30). Figure 4ivi.28 shows a long park space. Figure 4iv.29 shows a series of parks and
paved plazas along a promenade that steps down towards the waterfront edge. Figure 4iv.30 shows a small
square at the center of a building cluster along with privately owned public spaces that are delineated by
road blocks.
85
86
87
S. CONCLUSIONS
The most important aspect of this thesis is not so much based on the degree of completion and precision of the components, operations, and experiments set up throughout the thesis. It is much more about
getting us designers to focus on relationships and preventing us from attempting to materialize our own
preoccupations through an urban form, which is often too simplistic, visual-based, and one-dimensional for
urban design projects by providing an option that takes advantage of computational parametric tools.To
reiterate, urban design, especially in larger scales, cannot afford to have subjective design ideas projected
to it.Architecture is much more forgiving as it is small enough to become a vessel of an individual's interests and passion. By focusing on a city's components and their relationships, I think it is more possible to
propose actual, not just seemingly, complex and multi-dimensional urban schemes through layering of rules
and keeping track of what each of them does and how each of them affects the others.This thesis is the
antithesis of the desire to create the next canonical urban scheme.
One of the weaknesses of this methodology is that it relies heavily on context information in the form of
abstract or physical data.The richer the context, the more complex the design schemes can potentially
become. If this is to be used in a town planning project where a city or a large developer is seeking to
establish a satellite city away from the city center in the middle of a vast agricultural land, the methodology
will suffer as there is not as much information. In such scenarios, we will have to work with information
derived from the forces of nature such as topography, sun exposure, or wind direction in the absence of
density, road networks, public transit, existing open spaces, etc.
As it stands, the components, operations, and experiments I have set up in this thesis are simplistic and
primitive. For the purpose of this thesis, it has been my intention to keep them simple, controlled, and
contained in order to really understand what each component or relationship does, focusing on the design
outcomes. I realize by imbuing this work with words such as computation, digital, and parametric, many
complex urban information and analyses come to mind, including topics such as thermal comfort, air quality, mass transit connectivity, carbon footprint, noise, health, crime rate, and pollution. At this point, I am
uncertain of how, and if, any of these information should affect the process of urban design. I believe if my
method is developed further with more components and operations layered into the system, the outcome
will become that much more complex and sophisticated, creating responsible and well-informed design
schemes that otherwise would have been very difficult to achieve through the mind of individual designers.
Recursive Relational
88
Urban Design
6. BIBLIOGRAPHY, SOURCESAND REFERENCES
NYC Department of City Planning. PlaNYC Full Report (April 2011)

Alexander, Christopher, and Hajo Neis.A New Theory of Urban Design. Oxford University Press. NewYork,
1987. Print.
Alexander, Christopher."A City Is NotATree." Design. London: Council of Industrial Design, No 206,
1966. Reprint.
Ballon, Hillary. New York's Pennsylvania Stations. W.WNorton & Company. New York, 2002. Print.
Barnett, Jonathan. Urban Design As Public Policy; Practical Methods for Improving Cities. New York: Architectural Record Books, 1974. Print.
Batty, Michael, and Paul Longley."Fractal-Based Description of Urban Form". Papers in Planning Research
no. 95. Department of Town Planning, University of Wales Institute of Science and Technology, Colum
Drive, Cardiff, CF I 3EU,Wales, UK, 1986. Print.

Hill, Dan. "The Adaptive City." City of Sound. 7 Sep 2008.Web. 29 Apr 2014 (cityofsound.com)
Lehnerer,Alex. Grand Urban Rules. Rotterdam: 010 Publishers, 2009. Print.
Parish, I.H.Yoav, and Pascal Muller."Procedural Modeling of Cities." SIGGRAPH '01 Proceedings of the 28th
annual conference on Computer graphics and interactive techniques, 301-308.ACM. New York, 200 1.Print.
Ryan, Brent. Design After Decline: How America Rebuilds Shrinking Cities. University of Pennsylvania Press.
Philadelphia, 2012. Print.

Ryan, Brent."Incomplete and Incremental Plan Implementation in Downtown Providence, Rhode Island,
1960-2000." Journal of Planning History- 35-64,Vol. 5, No. 1. 2006. Print.

Schumacher, Patrik."Free Market Urbanism: Urbanism Beyond Planning." Patrik Schumacher. 2012.Web. 29
Apr 2014
Schumacher, Patrik."On Parametricism." T +A (Time + Architecture) 201215, Digital Fabrication. 2012. Print.
Schumacher, Patrik."My Kind of Town:The Parametric Jungle." Architecture TodayAT227. London, 2012.
Print.
Schumacher, Patrik."Parametricism - A New Global Style for Architecture and Urban Design." AD Architectural Design - Digital Cities,Vol. 79, No. 4, Novermber/December 2009. Print
89
Schumacher, Patrik. "Parametricism as Style - Parametricist Manifesto" Patrik Schumacher. 2008.Web. 29
Apr 2014
Verebes,Tom. Masterplanning The Adaptive City: Computational Urbanism in The Twenty-first Century. Rout-
ledge. New York, 2014

Thesis

Uploaded by

Copyright:

Available Formats

Thesis

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Thesis

Uploaded by

Copyright:

Available Formats

Recursive Relational Urban Design

SUBMITTED TO THE DEPARTMENT OF ARCHITECTURE IN PARTIAL FULFILLMENT OF THE

2014 Dessen Hillman.All rights reserved.

May 22, 2014

Recursive Relational Urban Design

Recursive Relational Urban Design

Submitted to the Department of Architecture

Recursive Relational Urban Design

Recursive Relational Urban Design

1.Downtown Providence in 1958 (Journal of Planning HistoryVol.5 No. 1, 41)

Recursive Relational Urban Design

Recursive Relational Urban Design

Recursive Relational Urban Design

ii.The use of generative computation techniques in practice

Recursive Relational Urban Design

One of the most important

and parametric responsiveness

implementation through unmistakably striking visual representation.The drawback here is the

unlike architecture, urban design

I ii.2. Schumacher's monotonous, uniform, and totalitarion parametric urbanism.

Recursive Relational Urban Design

Recursive Relational Urban Design

2. URBAN DESIGN VS.ARCHITECTURE ATTHE URBAN SCALE

Recursive Relational Urban Design

2.! . Unite d'Habitation (Photo by Paul Koslowski, 1997. fondationlecorbusierfr)

Recursive Relational Urban Design

Recursive Relational Urban Design

2.3. Le Corbusier's Plan Voisin (fondationlecorbusier.fr)

Recursive Relational Urban Design

Recursive Relational Urban Design

Another common strategy towards urban design by architects

tural Expo by Plasma studio,

GroundLab, and LAUR Studio

cludes things such as sunlight

Recursive Relational Urban Design

2.6. Xian International Horticultural Expo (plasmastudio.com)

Recursive Relational Urban Design

Recursive Relational Urban Design

Recursive Relational Urban Design

Recursive Relational Urban Design

3. PRECEDENTS AND SIMILARTHOUGHTS (INCREMENTALAND COMPUTATIONAL)

i. Early thoughts in computational and incremental urbanism

3i. 1. Semi-Lattice structure (top) vs. Tree structure (bottom)

Recursive Relational Urban Design

In 1965, Christopher Alexander

suggested that natural cities,

those that have "arisen more or

less spontaneously over many,

designed cities, those that "have

been deliberately created by designers and planners", have the

1986, Michael Batty and Paul

Longley in the University of

Technology published a study on

boundaries are typically defined

excluded, villages linked to the

urban area by ribbon develop-

ment were included, man-made

alterations to rivers and coast

3i.2. Fractal-based description of urban form

become part of the urban fabric