Explorations in Urban Design: An Urban Design Research Primer by Matthew Carmona

Author:Matthew Carmona [Carmona, Matthew]
Language: eng
Format: azw3
ISBN: 9781317137528
Publisher: Taylor and Francis
Published: 2017-05-14T16:00:00+00:00

Figure 15.2 Evaluating the four growth iterations against the spatial properties of Barcelona and a randomly generated structure. The generative code is written in Processing (Java). Spatial Structures are analysed using UCL Depthmap (Turner 2011) Figure 15.3 Binning data for correlations; storing data and spatial configurations in two overlapping grid reference layers and selecting the highest values in a third higher-resolution reference layer The first three invariants were identified in Al-Sayed et al. 2010, Al-Sayed et al. 2012 and Al-Sayed 2013. The fourth invariant was observed by Conroy Dalton (2001). What is yet to be investigated is whether these invariants can be a natural product of a local generative rule.

After generating four growth iterations, the structures were evaluated using the invariants and compared to an urban region in Barcelona and a randomly generated system. To evaluate against invariant 1, choice [SLW], a segment length weighted measure of the shortest angular paths across the whole street network, is calculated. A structure that has the highest 10 per cent values is then extracted and evaluated. The structure’s continuity is evaluated by measuring its normalised cumulative total depth values. Total depth is also an angular-based measure based on the sum of angular turns taken to reach any segment element in the street network. A normalised reciprocal of this measure defined as segmental integration can help estimating the potentials for a street segment to be a destination in the urban system.

To evaluate against invariant 2, metric mean depth analysis (MMD) is calculated. The measure here simply represents average physical distance from each street segment to the neighbouring segments within a metric radius of 1,000 metres (Hillier et al. 2007). Relating to invariant 3, the distribution of angular integration is evaluated using a KSL8 test to check whether the values of angular integration follow a lognormal distribution. In addition, the degree of skewness is compared to random and real systems; skewness being a statistical measure for asymmetry in the distribution of values. Considering invariant 4, the R2 coefficients of axial intelligibility are compared. Intelligibility is an axial graph measure that represents the relationship between streets that have high connections to other streets (connectivity) and streets that are more integrated in an axial system.

The evaluation measure of choice indicates that iteration 3 performs better than iterations 1, 2 and 4 (see Figure 15.2), whilst calculating MMD for different radii does not identify clear patchwork patterns in the background network of any of the three variations. On the basis of the KSL test, iteration 3 fits best with a log-normal distribution, it also reflects a well differentiated structure (Skewness = −1.46). Moreover, when considering the part-whole structural unity, the structure of iteration 3 is more intelligible than other iterations and reasonably similar in terms of synergy values. Yet, all three iterations present less competitive structures when compared to the deformed grid of Barcelona or even to a random system.

Figure 15.4 A neural network model applied to Barcelona, using normalised spatial measures of choice, integration and connectivity as factors and form-function attributes as response variables Figure 15.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.