Cities across the world are starting to recover space, previously devoted to cars, for other uses. The main purpose of this project was to better understand the removal of space in urban settings and to provide some analytical results showing that it is possible to remove streets from a city without worsening traffic excessively.
We have created an abstract grid network composed of 100 nodes and bidirectional streets. A simple demand model is applied to load the network effectively, emulating a dense urban environment. We have developed a static traffic assignment model using the Frank–Wolfe algorithm. The link removal strategies seek to represent city planning policies aiming at recovering space for other activities. Links are removed following three different strategies: in a total random manner, focusing on the center of the grid, and focusing on the perimeter of the grid. Up to 30% of the total links of the full grid are removed.
The results indicate that a certain number of links can be removed without affecting traffic considerably. This magnitude is very dependent on the link removal strategy. Our case covers a homogeneous grid with uniform demand where the central links carry higher flow. For that, the peripheral removal allows the highest rate of link removal, whereas the central removal is the most restrictive strategy. The restriction of capacity at intersections is the main bottleneck of the system when links are removed, driving most of the increase in delays. Under these conditions, the peripheral removal retains a higher connectivity in the center, allowing a better distribution of flows.