Effects of high occupancy vehicle lanes on freeway congestion

Previous research on the effect of HOV (high occupancy vehicle) lanes on bottleneck flows is extended here to entire freeways using both theory and empirical evidence. The paper shows that if the flows of both high- and low-occupancy vehicles remain invariant before and after a freeway lane is converted to HOV use, then the freeway's overall traffic density upstream of its bottlenecks is reduced - albeit less than expected - if the HOV lane is underutilized. As a result, HOV lanes can extend queues over longer distances. These expansions can be problematic if the queues' expanded portions impede traffic on heavily traveled routes that do not pass through the bottleneck. To quantify this effect, the paper analyzes HOV lanes on long, multi-ramp freeways. Formulae are given for the changes in people-hours and vehicle-hours of travel induced by an HOV lane, both when there is uncongested freeway space upstream of the queue to accommodate its expansion, and when there is not. All the inputs to these formulae are either observable or easy to estimate. Hence, the recipes can help evaluate any freeway's existing, or planned, HOV lane installation. The HOV lanes at all the sites we have analyzed, which are quite typical, add less than 2% to vehicular delay and reduce people delay by more than 10%. These estimates assume no increase in car-pooling. More generally, the paper also suggests how to deploy HOV lanes on city-wide freeway systems and recommends steps to better plan city-wide systems of bus lanes.