Incorporating Local Truck Performance Characteristics into the Calibration of Microsimulation Models

摘要

Traffic microsimulation models incorporate default vehicle performance characteristics thatmight not be representative of local conditions. In spite of this, many users do not perform arigorous calibration of the model parameters and even when this is done, the vehicleperformance parameters are generally omitted (1, 2, 3). Where significant differences existbetween the default vehicle performance characteristics and those of the local traffic, results ofthe microsimulation analyses can be misleading. For example, the default acceleration anddeceleration curves for simulating truck movements in VISSIM, a widely used trafficmicrosimulation model, are based on data from Western Europe that might not be directlyapplicable to the U.S. truck fleet (4).This paper proposes an approach for incorporating the unique operating characteristics ofthe local truck fleet into the calibration process of microsimulation models. In particular, thedesired acceleration and deceleration functions for trucks are treated as unknowns that must beestimated as part of the model calibration process as opposed to using their default values. It isexpected that such explicit consideration of local truck characteristics will lead to more reliablemicrosimulation results. The applicability and potential usefulness of the procedure wasdemonstrated through a case study using VISSIM. The results indicated that calibrating theacceleration and deceleration function in addition to the driver behavior parameters reduced theaverage discrepancy between observed and simulated speeds by 42.5%. The methodology maybe of potential interest to many transportation agencies as it is based on data that are oftenreadily available.