Edward Mitchell AIA; Yale University and Edward Mitchell Architects, New Haven CT

Transcription

1 Temporal Urbanisms Edward Mitchell AIA; Yale University and Edward Mitchell Architects, New Haven CT REPORT RESEARCH We studied the former manufacturing districts on the Brooklyn waterfront in Williamsburg and Greenpoint rezoned by the New York City Planning Commission and approved by the Board of Alderman in the spring of Principals of self-organization had already established a successful and complex urbanism in Williamsburg. Greenpoint s plan was more conventional, restricted by the crude tools of zoning. Using statistical analysis and genetic algorithms we hoped to establish a tool for future planning exercises. GENERAL ANALYTIC METHOD Games like Sim City organize information into coded graphic tiles that factor development costs, tax rates, and environmental factors into urban development. Our basic research involved analysis of land use and zoning in both Brooklyn and Manhattan, cataloguing statistical relationships between population and public and commercial uses. We then constructed graphic tiles to illustrate potential spatial relationships between urban typologies, FAR ratios, and the location of public open spaces and amenities as short hand descriptions for the genetic algorithms. STATISTICAL FINDINGS An addition of 15,000 residents should require a new public school, supplementary police substations and other local amenities and the new waterfront park similarly falls short of the average public space requirements for the city. Tax rebates for soil remediation were unfairly tied to the project costs rather than site improvements to encourage high density development. Statistical analysis is presented in the appendix as both graphic tile and spreadsheets. THE GENETIC CODE The statistical research was used to code the genetic algorithms. An 80 x 80 building block forms the basic graphic tile to simulate a grouping of four 20 townhouses, a developer tower, or rough envelope for a loft or bar building. The genetic algorithm randomly arranges the building blocks and scores each generation of the field according to local rule sets by evaluating adjacencies, proximity to amenities, targeted ratios for open space and overall population. The power of the algorithm is its ability to evaluate thousands of iterations of randomly sited blocks and then gradually settling to a best-fit pattern that would replicate a collective of self-organizing actors that resembles an organic pattern of development. Over fifteen successful iterations led to the final demonstration model. Models GA1 and GA2 used the town house typology and score respectively for zero or two neighbors preferences, leading to quasi-checkerboard or linear organizations. Subsequent models include the two tower types. Model GA5 scored for FAR ratios of 4.0 based on ratios determined for dense mixed-use development for North American cities. Later generations add public amenities based on population scores. Amenities were then given local areas of attraction so that residential blocks would gravitate towards the Boston Society of Architects Design Research Grants, 2005 Page 1

2 amenities. Sliding bars are supplied in the last model in order to shift relative values assumed in the preliminary studies. Models are included as a digital appendix to the report. CONCLUSION The general design strategy confirmed by the simulations indicated preference to develop new housing in proximity to mass transit and the existing commercial zones and public parks rather than at the water s edge. In the plan at the Newman Institute we also suggested finer grain details like locating parking garages under McCarren Park to leverage improvements to the public realm while connecting the open spaces on the high-rise park into the public space green network. Plans, typologies and hybrid buildings and the computer models are included in the appendix. Sim City took over 40,000 hours of programming and millions of dollars in research, but our basic codes establish the ground work for an open source library for future research and development. Our project will be in the public domain on a web site that will be released this January. A number of the models are included in the digital appendix to this report. By playing with the models we noticed that timing is critical to successful development and growth and that these kind of animated tools show promise for more sophisticated urban simulations. This start-up project has been presented publicly at conferences in Boston, New York, Helsinki, and Salzburg, Austria with considerable enthusiasm. We were encouraged to find that over the past summer Boston and Chicago announced that they would be using Sim City and GSA statistical data as tools for establishing long range planning strategies. In the future we hope to continue to develop more sophisticated modeling that will include development costs, financial models and site characteristics to enhance the power of the data driven logic of the algorithms. Boston Society of Architects Design Research Grants, 2005 Page 2

3 Boston Society of Architects Design Research Grants, 2005 Page 3

4 Boston Society of Architects Design Research Grants, 2005 Page 4

5 Boston Society of Architects Design Research Grants, 2005 Page 5

6 Boston Society of Architects Design Research Grants, 2005 Page 6

7 Boston Society of Architects Design Research Grants, 2005 Page 7

8 Boston Society of Architects Design Research Grants, 2005 Page 8

9 Boston Society of Architects Design Research Grants, 2005 Page 9

10 Boston Society of Architects Design Research Grants, 2005 Page 10

11 Boston Society of Architects Design Research Grants, 2005 Page 11

12 Boston Society of Architects Design Research Grants, 2005 Page 12

13 Boston Society of Architects Design Research Grants, 2005 Page 13

14 Boston Society of Architects Design Research Grants, 2005 Page 14

15 Boston Society of Architects Design Research Grants, 2005 Page 15