6.2.3 Traffic Assignment Problem

A simple application of shortestt-path techniques occurs in the preliminary determination of the traffic loads to be expected on different segments of a transportation grid. To make this determination, an origin-destination traffic matrix is required for the network of interest. Passengers are then assumed to follow the least distance (or time or cost) path from origin to destination through the network. The levels of flow on all links of the network are then computed by:

  1. Using some algorithm to find the shortest paths between all pairs of nodes.

  2. Adding, for each origin-destination pair i-j, the traffic flow t(i, j) to the flow on every link in the shortest path from i to j.

Example 3

Consider again the network of Figure 6.6. Assuming that the origin-destination data are as shown below, find the traffic flow on the links of the graph.

Origin-destination matrix (in tens of vehicles per hour):


We have already obtained all shortest paths for this graph and recorded them in the matrix P(5) (see Section 6.2.2). We can then determine the traffic flow on each link of the graph, as shown in Figure 6.7. To better understand the computation of the various flows, consider the flow of 52 units on the undirected link (2, 4) and in the direction from 2 to 4. rne 1ink (2, 4) in the 2 -----> 4 direction is part of the shortest path from node 2 to node 4 and from node 5 to node 4. Therefore, the traffic for these two destination pairs (i.e., 12 and 40 units, respectively) will use the (2, 4) link in the 2 ----> 4 direction, hence the flow of 52 units.

Note that there are traffic flows in both directions through some undirected links and that other (directed or undirected) links remain completely unused as indicated by the absence of any traffic flow numbers next to them. Finally, note the conservation of traffic at each node of the network.

This type of traffic flow determination can be useful in determining the capacity requirements for an urban transportation network at the initial stages of planning or as long as the network is relatively free of congestion. It is much less useful and meaningful in predicting traffic flows for a transportation network that has already been built and is heavily utilized. This is so because the determination of traffic flows, as performed above, ignores the facts that:

  1. Shortest distance--or, shortest travel time--is not the sole criterion for route selection.
  2. The capacity of any link on a transportation system is finite.
  3. Because of 2, travel speeds are a function of the amount of congestion on a transportation link and, as a result, the determination of the shortest paths cannot be made independently of traffic-flow determination.

In particular, the effects of congestion call for more sophisticated approaches than the above in determining traffic flows. These approaches are generally referred to as "traffic assignment" techniques. Good general references about these techniques can be found in [FLOR 76]. (See also Problem 4.10 for the differences between "user-optimum" and "system-optimum" traffic assignment.)