Application of Relaxation Procedure for Lane Changing in Microscopic Simulation Models

Abstract

Lane-changing submodels in microscopic traffic simulation models require the implementation of a relaxation procedure. The basis of this requirement is the observation from field data that both a vehicle performing a lane change into a short gap and its new follower maintain persistently short following distances for a significant period (20 s or more), gradually increasing to equilibrium values. A proposed approach involves a temporary reduction in the vehicle sensitivity factor of the car-following component in the FRESIM submodel of CORSIM for a user-specified period. The relaxed lane-changing model was implemented on a spreadsheet, and a series of values of the relaxation time was used in a scenario in which an initially short value for the time headways between the leader, lane changer, and follower was assumed. The results for speed and time headways indicate qualitative agreement with the field data. The same scenario was run with the unrelaxed procedure in the existing FRESIM model, and the results for the speed and time headways were in disagreement with the field data. From these results, it was concluded that the lane-changing system currently imbedded in the FRESIM model was invalid. Subsequently, the relaxed lane-changing model was implemented in FRESIM, and a merge scenario was executed. It was seen that the value of the sensitivity factor required to reproduce any particular merge capacity in this revised model increased to physically realizable values as the relaxation time is increased, whereas non-physical values were required in the existing FRESIM model to produce equivalent merge capacities.