2

This study aimed at the development of procedures for the estimation of intrazonal trip distances based on data of origin-destination (O/D) surveys. In the application discussed here, the analysis was based on an O/D survey conducted in the city of São Carlos, Brazil. The urban area was subdivided several times so as to produce 471 traffic analysis zones (TAZ). The geometrical characteristics of the zones that might influence the distances of intrazonal trips were identified. Numerical indicators of the selected geometrical patterns were compared with the average values of intrazonal trip distances to search for evidences of correlation between these variables. Two analytical approaches were explored: i) continuous and ii) discontinuous. In the first case, all trips were considered a single dataset, whereas in the second the dataset was split into two subsets of homogeneous characteristics. The datasets with higher correlation were used to build global and individual models for the estimation of intrazonal trip distances. In the global models, walking, cycling, and auto trips were considered part of a common dataset. The three transport modes were separately taken into account in the stratified models. The values obtained with these models were then compared with the estimations of other models found in the literature. The discontinuous models calibrated in this study clearly outperformed the traditional ones, which may be an indication they can be used to replace the traditional models for the estimation of average intrazonal trip distances, at least in Brazilian medium-sized cities