A multi-stage procedure for validating microscopic traffic simulation models

<title>A<sc>bstract</sc> </title> Validating microscopic traffic simulation models incorporates several challenges because of the inadequacy and rareness of validation data, and the complexity of the car following and lane-changing processes. In addition, validation data were usually measured in aggregate form at the link level and not at the level of the individual vehicle. The majority of model validation attempts in the literature use average link measurements of traffic characteristics. However, validation techniques based on averages of traffic variables have several limitations including possible inconsistency between the field observed and simulation-estimated variables, and as such the resulting spatial--temporal traffic stream patterns. Due to these inconsistencies, this paper introduces a novel approach to the validation of microscopic traffic simulation models. A three-stage procedure for validating microscopic simulation models is presented. The paper describes the field measurements, experimental setup, and the simulation-based analysis of the three stages. The purpose of the first stage is to validate a benchmark simulator (NETSIM) using limited field data. The second stage examines the spatial--temporal traffic patterns extracted from the benchmark simulator versus those extracted from the simulation model to be validated (I-SIM-S). Different traffic patterns were examined accounting for various factors, such as traffic flow, link speeds, and signal timing. The third stage compares the aggregate traffic measures extracted from the subject simulator against those extracted from the benchmark simulator.