Network Aggregation for Transportation Planning (Arc Abstraction, Pseudo-Arcs, Sketch).
Kuo-Liang Ting · Deep Blue (University of Michigan) · 1985
Two original arc-abstraction methods for aggregating transportation planning networks were developed. They are useful in reducing the number of arcs and nodes in large transportation networks prior to traffic assignment. The methods aggregate a region's basic planning study subarea networks. After subareas are defined, the aggregation procedures are applied to the relatively square individual subareas independently of the rest of the network. Pseudo-arcs are formulated to represent reasonable paths for trips ending in a subarea and for trips going through the subarea. Depending on the purpose of analysis, the size of a subarea can vary from a single zone to a superdistrict consisting of a large number of zones. The boundaries of subareas can be natural, political or may coincide with major freeway or arterial transportation facilities which form boundary arcs. The first aggregation method retains all the boundary arcs and represents each subarea by a subarea centroid and a set of pseudo-arcs connecting the centroid with the nodes on the boundary. The second method of aggregation also represents each subarea by a centroid and a set of pseudo-arcs, but eliminates subarea boundary arcs by using diagonal pseudo-arcs which combine the boundary arcs with the diagonal arcs to the subarea corners. In the second method, the pseudo-arcs are also grouped in major directions of traffic movement within the subareas. Although using the same methodology, these two methods differ slightly based on the trade-off between savings in computer time and cost and loss of information. Compared with other aggregation techniques proposed for transportation planning practice, the aggregation methods developed in this research are more consistent with the user-optimal route-choice criterion. The aggregation process in these methods requires minimal computer time to find the minimum paths on individual subarea networks. The aggregation methods have been developed to reduce the errors due to aggregation at both the system and individual arc levels. The aggregated networks thus obtained always retain important information and sacrifice little accuracy. Besides, the reduction in the network size always results in savings in computer and analyst costs. The methodology was applied to a test two-subarea network and to the Detroit regional-planning network in a sketch-planning context. The results of traffic assigned on these aggregated networks are compared to traffic assigned on the detailed networks and are comparable in accuracy. The savings in computer time and storage needs of the aggregated networks over the detailed networks are significant.