From Macro to Microscopic Trip Generation: Representing Heterogeneous Travel Behavior

Abstract

This research provides evaluations of the gravity model as an analytical tool for simulating present and forecasting future urban trip distribution patterns. The evaluations were made by com.paring gravity model trip interchanges with those found in home b1terview origin and destination surveys conducted in Washington, D. C., in 1948 and 1955. The 1955 survey data were used for calibrating the basic gravity model and for testing this model for its ability to simulate current travel patterns. The 1948 survey provided comprehensive data to analyze the forecasts made by the calibrated model. The gr avity model will give satisfactory results if properly calibrated and tested. The level of accuracy obtained by forecasting trip distribution patterns in 1948 was comparable to the level of model accuracy for the base year. •THE GRAVITY MODEL trip distribution formula has been used in transportation planning studies in many urban areas during the past few years. The theory of this formula and the general procedures used to simulate the present travel patterns have been docume nted to some extent in the litel'ature (1, 2, 3). The use of this model to forecast future travel patterns in several urban areas has also been reported (1, 2). To date, however, there are little published data available to illustrate factually the ability of the gravity model to either simulate existing travel patterns or to forecast future patterns. About tlu·ee years ag0 the Urban Planning Division of the U. S. Bureau of Public Roads in cooperation with the Wasbingto11 Metropolitru1 .Area Transportation Study (WMATS) began a research project to refine and document detailed procedures !or calibrating and testing a gravity model trip distribution formula for use in simulating present travel patterns and forecasting future t ravel patterns in an urban area. This project also included the development of a series of IBM 704/7090 electronic computer progr a ms for implementing the ru1alytical procedures devised. To accomplish such a project, adequate data on travel patterns for two time periods were required. At that time the Washington, D. C. , metropolitan area was the only large area in the country having complete and adequate home interview surveys for two separate time periods. During the summer of 1948, a comprehensive origin-destination survey was conducted in 5 percent of the dwelling units in the Washington metropolitan area (~). In 1955 a repeat origin-destination survey was conducted in the same area (5). Within the District of Columbia, occupants of 3 percent of the dwelling units were interviewed. Elsewhere in the area, occupants of 10 percent of the dwelling units were interviewed. Consequently, the Washington area provided an ideal situation for testing and evaluating the ability of the gravity model to simulate travel patterns for one period of time and also to forecast such patterns for a different period of time.