Download - Marshall Pres

Transcript
  • 7/26/2019 Marshall Pres

    1/42

    CUPUM/ECiD Joint Workshop 'Design out of Complexity' - UCL, 2 July 2005

    STEPHEN MARSHALLThe Young Foundation

    Bartlett School of Planning, UCL

  • 7/26/2019 Marshall Pres

    2/42

    The Urban Design Agenda

    Neo-traditional urbanism now in favour Traditional patterns now seen as models

    But these patterns were not planned

    How to plan or design these?

  • 7/26/2019 Marshall Pres

    3/42

    Random

    (Mandelbrot)

    Chaotic

    (Keeble)

  • 7/26/2019 Marshall Pres

    4/42

    Characteristic Structure

    Two senses:

    Typical distinctive character

    Likely

  • 7/26/2019 Marshall Pres

    5/42

    Characteristic Structure

    Typical distinctive character

    Means the quintessential street patternshape

    Not like a gridiron

  • 7/26/2019 Marshall Pres

    6/42

    Characteristic Structure

    Typical distinctive character

    Means the quintessential street patternshape

    Not like a gridiron Not like a tree

  • 7/26/2019 Marshall Pres

    7/42

    Characteristic Structure

    Typical distinctive character

    Means the quintessential street patternshape

    Not like a gridiron Not like a tree But like nothing otherthan a street pattern

  • 7/26/2019 Marshall Pres

    8/42

    Characteristic structure

    of street patterns

    a mixture of short and long routes, and more and less

    connective routes; some differentiation of routes by depth, but overall not

    too great a depth;

    three-way junctions are typically in the majority, butlikelihood of at least some crossroads and culs-de-sac;

    a medium or semi-griddy level of connectivity, with arelative connectivity () of around 0.35-0.45;

    a relatively high degree of irregularity and complexity,with complexity ()

    typically in the range 0.35 to 0.6.

  • 7/26/2019 Marshall Pres

    9/42

    DESIGN

    EVOLUTION

    TOP-DOWN

    Pattern selection

    Patternsgenerated by

    programs

    BOTTOM-UP

    Different kinds of program

  • 7/26/2019 Marshall Pres

    10/42

    Kit of identical parts

  • 7/26/2019 Marshall Pres

    11/42

    Kit of identical parts

    (i) generation 1

  • 7/26/2019 Marshall Pres

    12/42

    (i) generation 1

    (ii) generation 2

    Kit of identical parts

  • 7/26/2019 Marshall Pres

    13/42

    (i) generation 1

    (ii) generation 2

    Kit of identical parts

    or

    (iii) generation 3

  • 7/26/2019 Marshall Pres

    14/42

    The T-tree program

    _|_ _|_ _|_ _|_ _|_ _|_ _|_ _|_

    1) Each constituent element is identical.

    2) Each element is added to form structure one at atime.

    3) Each new element occupies a position on thestructure that is chosen at random.

    4) Each new element joins the existing structure atonly one of its ends, to form a 3-way connection (T-junction), such that the overall structure is a T-tree.

  • 7/26/2019 Marshall Pres

    15/42

  • 7/26/2019 Marshall Pres

    16/42

  • 7/26/2019 Marshall Pres

    17/42

  • 7/26/2019 Marshall Pres

    18/42

  • 7/26/2019 Marshall Pres

    19/42

  • 7/26/2019 Marshall Pres

    20/42

  • 7/26/2019 Marshall Pres

    21/42

  • 7/26/2019 Marshall Pres

    22/42

    Case Frequency Probability

    a 1 4%

    b1-6 6 25%

    c1-3 3 12.5%

    d1-4 4 17%

    e1-4 4 17%

    f 1 4%

    g1-3 3 12.5%

    h 1 4%

    i 1 4%

    Total 24 100%

  • 7/26/2019 Marshall Pres

    23/42

    More frequent(three variants)

    Typicality

    Less frequent(each a singular

    type)

    Comb Irregulars Fractal

    Type

  • 7/26/2019 Marshall Pres

    24/42

    The X-cell program.

    1)Each constituent element is identical.

    2) Each element is added to form structure one at atime.

    3) Each new element occupies a position on thestructure that is chosen at random.

    4) Each new element joins the existing structure atone or both of its ends (but not along its middle);the resulting structure can have multi-spoked nodesand form circuits.

  • 7/26/2019 Marshall Pres

    25/42

  • 7/26/2019 Marshall Pres

    26/42

  • 7/26/2019 Marshall Pres

    27/42

  • 7/26/2019 Marshall Pres

    28/42

  • 7/26/2019 Marshall Pres

    29/42

  • 7/26/2019 Marshall Pres

    30/42

    Deeply layered andbranching structures

    Meshes and spokes

    Long chain structures

  • 7/26/2019 Marshall Pres

    31/42

    Area whereactual streetpatterns aretypically

    found Grid-likestreetpatterns

    Inter-mediate

    Tree-like

    streetpatterns

    1.0

  • 7/26/2019 Marshall Pres

    32/42

    0.0

    0.1

    0.2

    0.3

    0.4

    0.5

    0.6

    0.7

    0.8

    0.9

    0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

    1st, 2nd 3rd generation Real (Figure 8.4)

    4th generation 4th gen. resultant

    2D

  • 7/26/2019 Marshall Pres

    33/42

    2D

    3D ti t

  • 7/26/2019 Marshall Pres

    34/42

    3D nesting, etc.

    Ziggurat of Ur

    I Di t l ti f

  • 7/26/2019 Marshall Pres

    35/42

    ba2a12a11

    a1 a2 b c d e f

    I. Direct selection ofwhole pattern single act of design

    II. Selection atsuccessivestages in design,directing growth

    towards desiredoutcomeProgram e1

    e2

    III. Selection of programs

    Conclusions and Implications

  • 7/26/2019 Marshall Pres

    36/42

    Conclusions and Implications

    1. Demonstration of a possible mechanism for the

    probabilistic generation of characteristic urban

    structure

    Conclusions and Implications

  • 7/26/2019 Marshall Pres

    37/42

    Conclusions and Implications

    1. Demonstration of a possible mechanism for the

    probabilistic generation of characteristic urban

    structure2. Neo-traditional patterns could be recreated using

    programs rather than pattern templates

    Conclusions and Implications

  • 7/26/2019 Marshall Pres

    38/42

    Conclusions and Implications

    1. Demonstration of a possible mechanism for the

    probabilistic generation of characteristic urban

    structure2. Neo-traditional patterns could be recreated using

    programs rather than pattern templates

    3. A program-based design approach can use simplerules yet generate complex patterns

    Conclusions and Implications

  • 7/26/2019 Marshall Pres

    39/42

    Conclusions and Implications

    1. Demonstration of a possible mechanism for the

    probabilistic generation of characteristic urban

    structure2. Neo-traditional patterns could be recreated using

    programs rather than pattern templates

    3. A program-based design approach can use simplerules yet generate complex patterns

    4. This approach could be an alternative toconventional town planning or master planning

    Conclusions and Implications

  • 7/26/2019 Marshall Pres

    40/42

    Conclusions and Implications

    1. Demonstration of a possible mechanism for the

    probabilistic generation of characteristic urban

    structure2. Neo-traditional patterns could be recreated using

    programs rather than pattern templates

    3. A program-based design approach can use simplerules yet generate complex patterns

    4. This approach could be an alternative toconventional town planning or master planning

    5. Possibility of practical application through synthesis

    of urban coding + road hierarchy

  • 7/26/2019 Marshall Pres

    41/42

    S T E P H E N M A R S H A L L

    s . m a r s h a l l @ u c l . a c . u k

    w w w . s p o n p r e s s . c o m

  • 7/26/2019 Marshall Pres

    42/42

    CITIESCITIESCITIES

    DESIGN &DESIGN &DESIGN &

    EVOLUTIONEVOLUTIONEVOLUTION

    S T E P H E N M A R S H A L L

    s . m a r s h a l l @ u c l . a c . u k

    w w w . s p o n p r e s s . c o m


Top Related