Calibration of traffic stream models on high-speed urban roads with heterogeneous traffic

The basic objective of this work was to identify best-fitting traffic stream models for the varying traffic flows on high-speed urban roads under heterogeneous traffic in India. To this end, traffic data on ten-lane, eight-lane and six-lane divided roadways were collected. An iterative process was developed to fit traffic stream models by optimising constant parameters using optimisation techniques. Different single-regime models, multi-regime models and models based on the car-following approach were fitted to the field-observed data set. Among the single-regime models, the Northwestern model fitted the data set with the highest level of accuracy. The three-regime model was found to be more accurate than the other multi-regime models. Various multi-variate car-following models were also applied and found to fit uncongested and congested data with higher accuracy than the use of a single equation. Among all the models examined, the model proposed by Van Aerde and Rakha provided the highest accuracy. This observation was also validated for roadway sections with varying carriageway widths. The outcome of this research may be useful for the development of correct trends of speed–flow relationships for similar kinds of roadway facilities in the Asia-Pacific nations. Using the most suitable trend, level-of-service boundaries can be delineated more effectively.