INTRODUCING "FEEDBACK" INTO FOUR-STEP TRAVEL FORECASTING PROCEDURE VERSUS EQUILIBRIUM SOLUTION OF COMBINED MODEL

Abstract

The manner in which the solutions produced by various methods of introducing feedback into the four-step travel forecasting procedure compare with the equilibrium solution of a model combining the trip distribution, mode split, and assignment steps was examined. The comparisons were performed on a sketch-planning model of the Chicago region with about 300 zones and 3,000 highway links. From these comparisons one can learn that iterating the four-step procedure in an ad hoc manner does not produce the desired result. Instead one needs to apply an algorithm designed to converge to a well-defined equilibrium of the travel flows and the link times and costs determined by these flows. Progress in improving travel forecasts will not result from calls for solving the four-step procedure with feedback. Rather progress will be made as professional practitioners begin to understand the requirements of the desired equilibrium solutions. Then they must insist that their software developers correctly implement the algorithms required to compute these solutions. Finally they should insist that FHWA short courses introduce participants to contemporary solution methods that yield the desired equilibrium properties. Likewise university instructors and textbook authors should update their courses to produce a new generation of professionals who understand the principles of equilibrium travel models.