Road Growth Modeling - The Development of a Road Growth Model That Co-Evolves with Land Use

Abstract

Land development and infrastructure growth are two co-evolving processes. When studying long-term urban evolution it is useful to model both pro- cesses in parallel but co-dependently. The focus in this thesis is a novel road growth model that extends and adapts an existing land use model. The model transforms land use into interactions—a form of abstract traffic-like quantities—between every pair land lots. Each interaction needs to find a path from its start lot to its destination. If it is more efficient to extend the existing road network then a new road is suggested. Though each land lot pair acts independently when finding the path between the pair, a level co-operation occurs in the model as suggestions may add up. The model formulation turns out to have a scale-invariant property and a possible con- nection to the fractal-like structures of real road networks is explored. The model is open-ended with regards to the underlying geography, but makes several simplifications; only one type of traffic is considered, all roads have infinite capacity and equal speed limits, and travel time is approximated by availability and Euclidean distance. Care has been taken to make the model implementable in a computer with a reasonable amount of working memory even when there are many land lots, at the cost of increased computing power. A large part of the model is however parallelizable and some applicable timesaving techniques are recognized. A brief overview of relevant aspects of infrastructure in general — and road networks in particular — is also presented, giving a basis for model evaluation.