Analysis of Traffic Flow Characteristics Based on Local Data: Speed-Density Relationship for Tehran- Qom Freeway

Characteristics of traffic flow depend on driver behavior, vehicle specifications, highway characteristics and the interaction between these factors. Despite all similarities between characteristics of passing vehicles and geometry of highways, the characteristics of traffic flow are not necessarily the same. The differences may be attributed to the differences in behavior of drivers. Since 1934, different models have been suggested to predict characteristics of traffic flow. However selecting the best model, indicating the relationship between different parameters of traffic flow has ever been challenging. This is mainly because of the difference in behavior of drivers in different conditions. This paper investigates the speed-density relationship according to behavior of local drivers. In this research, more than 4 million records of traffic data are studied and analyzed in Qom-Tehran Freeway. Information of traffic counting is gathered by means of inductive loop detectors in both directions per each lane. Based on this information, different Speed-Density models are calibrated, analyzed and compared with previous models. However a multi-regime linear model with different slopes at break point has been suggested. According to the results, weighted average of speed variance in densities less than 7 pc/km/ln is nearly 30% less than the corresponding value in densities more than or equal to 7 pc/km/ln. Analysis show that regression errors in proposed model and Green shields model are 0.8% and 9.6% respectively when the density is less than 7 pc/km/ln and 1.7% and 47.3% respectively when the density is more than or equal to 7 pc/km/ln. This suggests that presented models in this research, are better describing the flow situation in Qom Tehran freeway, than the Green shields model.