Longitudinal and Lateral Throughput on an Idealized Highway

Highway automation has recently enjoyed renewed research interest as a method for solving highway congestion problems. To determine the feasibility of automated highways, it will be essential to estimate the automated highway benefits. A critical task in the analysis of automated highway benefits is to estimate the potential increase in highway capacity. This paper uses deterministic approximations to model highway throughput, accounting for both longitudinal requirements (i.e., lane flow) and lateral requirements (i.e., lane changing). The model also accounts for trip length distributions, and the effect of these distributions on the lane “flux” (i.e., the rate at which vehicles move between adjacent lanes, per unit length of highway). Based on these representations, the model identifies conditions under which lane changes have an appreciable effect on capacity, assuming certain idealized conditions are met. For typical highway parameters (mean trip length of 20 km, and mean speed of 30 m/s), and an increased nominal capacity (i.e., not factoring in lane changes) of 7200 vehicles per hour, the incremental effect of lane changes only appears to be significant (i.e., cause a capacity reduction in excess of 20%) when the time-space requirement exceeds 1000 m-s. As an example, this requirement could imply a lane change maneuver that lasts 10 seconds and requires 100 m of lane-space during execution. This calculation does not factor in queueing effects. To ensure that all or most vehicles succeed in exiting at their desired locations, the actual capacity will have to be decreased somewhat further.