Modeling traffic operations and drivers' behavioral parameters at not-conventional roundabouts. A theoretic-experimental approach

The inapplicability of current methods based on gap-acceptance theory for analyzing operational conditions at not-conventional roundabouts (very frequent in urban areas) seriously hampers the performance assessment; in practical applications, this also makes the choice of corrective design measures very uncertain. Starting from this considerations, the present paper shows the conceptual path followed for analyzing traffic operations at multilane-large-diameter not-conventional roundabouts. The research follows a theoretic-experimental approach that intends to put in a fair equilibrium the need both to match field observations and to have a general criterion to determine behavioral parameters, on which traffic performances depend. The main idea of the proposed approach derives from field observations at not-conventional roundabouts which show traffic operations following a pattern of a consensus of right-of-way alternating between vehicles entering from the approach and those streaming in the circulating lanes. A generalized model, similar to that one characterizing All-Way-Stop-Controlled (AWSC) intersections, was implemented, accounting for peculiarities revealed by an in-depth exploratory analysis of field data at not-conventional roundabouts. As basic behavioral parameters (e.g. saturations headways) elude direct observations, i.e. operational conditions in which they are observable rarely occur, in this paper a procedure to draw them from macroscopic observations of traffic conditions has been proposed. For this purpose, a regression analysis was carried out starting from observational data; the presence of response correlation required the regression parameters to be estimated through Generalized Estimating Equations models (GEEs), i.e. developing a marginal model for the unobservable - unknown - parameters.