The HCM2000 uses average travel speed (ATS) and percent time spent following (PTSF) as measures of effectiveness to assess the level of service on two-lane highways. While ATS is difficult to observe directly, it is virtually impossible to obtain PTSF from traffic observations; thus, the HCM2000's procedure uses ATS-flow and PTSF-flow functions derived from simulation to estimate the level of service. The highway environment in Brazil is sufficiently different from the North-American one to invalidate the use of the original HCM2000 ATS and PTSF functions. The objective of this study was to adapt HCM2000's ATS and PTSF functions for two-lane rural highways in Brazil. Traffic data were collected in 11 locations, capturing a wide range of road and traffic conditions. These data were used to calibrate and validate TWOPAS, using a genetic algorithm (GA). Synthetic traffic data, similar to those collected by in-road traffic detectors, were generated for a wide range of conditions. The generation of synthetic data uses solutions created by the GA during the calibration of TWOPAS that were just marginally worse than the optimal calibration parameter set. The synthetic data set obtained consists of distributions of flow-ATS-PTSF observations for a hypothetical road segment representing a two-lane rural highway with ideal conditions in Brazil. The HCM2000 functions were then recalibrated using these synthetic data. New mathematical relationships were also investigated: a concave ATS-flow function and an exponential function for PTSF-flow. Besides being capable of representing several curve shapes (including those in the HCM2000), these new models are more compatible with the experimental data and are thus able to better represent the behavior traffic streams on Brazilian roads. These models could potentially be used to develop an adaptation of the HCM for Brazil.
Read full abstract