Estimation of origin-destination trip-tables based on a partial set of traffic link volumes

Knowledge of motorists' origin-destination (OD) travel information is necessary for a wide range of transportation planning activities. The high cost of manpower and other expenses associated with conventional survey techniques for OD estimation have motivated the development of models that can inexpensively estimate these flows from easily available traffic link volumes. In this paper, we develop an approach for synthesizing these OD flows based on only a partial set of link volume information. This consideration introduces nonlinearities in the cost function of the model because of the dependence of link travel costs on link volumes, and requires the determination of a fixed-point (rather than an optimal) solution to the proposed model. Such a fixed point is determined heuristically by iteratively approximating the nonlinear model using a sequence of linear programs. Computational results on three sample networks from the literature are presented to evaluate the method and to provide insights into its performance relative to some maximum entropy and bilevel programming approaches.