For the Smart Geometry Conference 2013 in London, I was working with the two PAD Cluster, Henrik Malm and John Stack Ross, helping to set up a parametric workflow in Generative Components.
PAD addresses the sg2013 challenge Constructing for Uncertainty by coupling probability theory with parametric architectural and urban design. Cluster participants will examine the relationship between probabilistic design methods and emerging urban environments. Through an application of statistical models, developed in partnership with the Bartlett’s Urban Design Masters program, we will estimate contextual parameters and environmental data sets to be used for computational modeling.
Development authorities and planning agencies often rely on issues such as, but not limited to, the physical constraints set by traditional by-laws, zoning restrictions, and easements to anticipate urban growth and infrastructural strategies. Many of these current strategies address trends regarding sustainability and environmental factors and are often reactive to current conditions within the built environment, which may, or may not, interrogate the volumetric implications for future developments.
Our aim is to further investigate these methods, through probabilistic design, as they relate to the volumetric implications within topics such as right to light, solar access, and solar envelopes for future urban development at the scale of the parcel, its adjacencies, and the structuring of districts.
Contextual parameters, such as future building heights in adjacent blocks, will be estimated using available data and statistical models of urban development. These estimations will result in probability distributions for each contextual parameter instead of single values or numbers. Each of the probability distributions will have an estimated mean xm and standard deviation σ, i.e. information about the probability that the actual future parameter value lies within a certain so-called confidence interval. By drawing a large amount of sample parameter sets from the distributions and constructing a compound model of all resulting models we receive a design proposal which encodes the uncertainties in the contextual parameters.
The probabilistic design approach reveals the possible range of variability of designs under the influence of uncertain parameters and can be applied to a variety of scales. Site specific locations within the Thames Gateway ‘corridor’ to the east of London will be used as areas for prototype developments. Contextual parameters will be identified at each site and estimated from available data. In this way we will be able to focus the investigations on specific parameters as they relate to massing heights, setbacks, and mass vs. void conditions for individual parcels and the overall propagation of urban / suburban environments.
The probabilistic design process will predict a flow of variability through the design, which will result in proposals that contain areas and volumes within which the final design proposal should be contained. Varying the design within these areas and volumes will guarantee that the final model complies with the actual future contextual parameter values with respect to a specified degree of confidence. Within these ‘malleable’ volumes, participants will explore an array of specific design proposals to be constructed both digitally and physically, which ultimately will be evaluated in tandem with proposals for adjacent sites developed by other cluster participants.