The Making Of Functional Town Tissues: Morphogenetic Experiments In Coastal Low-lying Areas

How to define an appropriate form for turning a village into town in rapidly urbanizing coastal low lands poses a crucial question to urban designers. Since uncovering the truth of how a city is formed is usually not easy, current urban morphogenesis mainly uses the’ proof of concept’ model to simulate complex urban forms, in which simple rules are hypothesized and feedback loops are implied by the process. Cellular Automata (CA) can realize the above process by using simple local level rules, but are limited in predicting the final form, and cannot extract rules from the terminal form directly. Thus the CA rules are usually generated by the trial-and-error methods when being used in the parcel and building scale. In this paper we tried several methods of how to define the CA rules through terminal shapes and functions, and explored a series of approaches to generating town tissues under the pressure of the much higher density required by the current redevelopment.We mapped32coastal towns in Tianjin and extracted the key elements such as street networks, parcels and buildings, and classified the levels of complexity in patterns according to Salingaros. We also induced the possible driving forces of the transformation by comparing the historical and current satellite images. We summarized the courtyard buildings’ self-organizing rules and the adaptive rules under various constraints aided by Shape Grammar (SG).We performed three CA-based generative experiments in terms of predictability. To control the spatial configuration, first we used the random porosity as a sensitive variant to generate the town tissue. We found it can generate low-rise and high-density town tissues by simple transition rules. Next we used hierarchical CA (3m size cell nested into9m) constrained by the sunshine, ventilation, and traffic to gain a multi-story, mixed town tissue. Further, we applied SG to configure the residential floor plans based on the patterns generated by3m sized CA. We derived the CA rules from the SG patterns by using a high-density known form as a terminal object. Based on the SG-CA method, we employed the totalistic CA to generate a framework plan and used the general CA to gain the detailed form.The first and second experiments show that hierarchical CA helps express more architectural details than traditional CA through defining different rules across layers. If we only use sunshine spacing to define C A rules, the relevant output configuration appears a lack of abundance and diversity. The last experiment shows that deriving the CA rules from the SG rules will improve the morphological similarity between the generative and known terminal town tissues, and can effectively decrease the uncertainty in form generation. But it is short of repeatability in generative process, probably because that we have not derived rigorous mathematical expressions for SG rules--it is indeed a challenge for the entire field of urban morphogenesis.