Next Generation Simulation (NGSIM)
-Improved Simulation of Stop Bar Driver Behavior at Signalized Intersections-
Los Angeles, CA and Atlanta, GA Data Sets

Working Papers

Some of the findings from a recent study of the queue discharge headway process are summarized. One outcome of the study was the development of a model of discharge headway at signalized junctions. The model is based on vehicle and driver capabilities, including driver reaction time, driver acceleration, and vehicle speed. To calibrate the model, data were collected at five signalized junctions. The discharge headway model developed in this research indicates that the minimum discharge headway of a traffic movement is not reached until the eighth or higher queue positions. Application of the model suggests that the minimum discharge headway of a traffic movement under ideal conditions may be shorter than 2.0 sec/veh and that its corresponding start-up lost time may be longer than 2.0 sec.

This paper appears in Transportation Research Record No. 1365, Highway Capacity and Traffic Flow [Year of Publication 1992].

Figures (8); References (15)

Summary

Introduction

The purpose of this paper is to suggest a model to simulate discharge headways at signalized junctions. This model is based on vehicle and driver capabilities.

Background

Discharge Headway: The HCM suggests that the headways of all vehicles from queue position 5 onwards be averaged to estimate the minimum discharge headway. It also recommends an ideal saturation flow rate of 1800vphplphg. This translates to an ideal minimum saturation headway of 2 s/veh. Some studies have found the ideal saturation headway to be slightly lower (1.92-1.97s/veh).

Headway Models: The headway model by Briggs (reference 9 in the paper) is calibrated by using data from five previous studies. The model uses the basic kinetics equation to calculate the distance traveled to reach a desired speed and headway of a queued vehicle. Based on the calibration data, the parameters giving the best fit for the model were a starting response time of 1.22s, a *constant* acceleration of 3.67ft/s² and a 19.65ft spacing between vehicles in the stopped queue. The model has two parts depending on whether the vehicle under consideration has reached the desired speed or not. If it hasn’t (i.e., for the first few vehicles), the headway is calculated as a function of vehicle acceleration and queue position. If the desired speed has been reached (i.e. for vehicles after the fourth or fifth queue position), the headways become dependent only on driver response time and desired speed. Thus, it supports the anticipated trend of decreasing headways with queue position.

Starting Response Time and Distance Between Queued Vehicles: The first few studies (Messer, Fambro; reference 10) on this subject found the driver response to be fairly constant at 1.0s, except for the very first driver in the queue, who had an additional delay of 2.0s. They found the average space headway to be 25ft. George and Heroy (reference 11) found driver response to be 1.3s for all queue positions. Herman et al. (reference 12) found that the driver response to disturbance was fairly constant as the platoon of queued vehicles increased its speed. They also found the average distance between stopped vehicles to be approx. 25.9ft, thereby calculating the startup response time to be 1.0s.

Model Development

Since Briggs’s model assumes a constant acceleration rate (which is not observed in reality; drivers have high rates of acceleration initially which decrease as they approach their desired speeds), an alternative headway model based on non-constant acceleration behavior was warranted. Studies/models by Buhr et al. (reference 13) and Herman et al. (reference 12) show a strong inverse linear relationship between acceleration and speed.

Based on the above findings, the author proposes a model for headway of leading vehicle based on the additional response time of the first driver, the driver starting response time, distance between vehicles in a stopped queue, stop-line speeds of a leading-following vehicle pair and assumed maximum values of acceleration and speed. The model conforms to the expectation of minimum discharge headway converging for vehicles approaching their desired speeds. The stop-line speeds, in turn are calculated with the help of an inverse exponential function dependent on queue position of the vehicle and maximum desired speed. Finally, the lost times which depend on minimum discharge headway (which in turn depend on stop line speeds) are calculated. Because of the underlying exponential relationship, this model assumes that *all* vehicles are responsible for some amount of lost time.

Statistical Analysis, Model Calibration (and Validation) and Conclusions

The data was collected at same locations as the previous paper, i.e., 3 SPUIs and 2 AGIs in locations in FL and TX. Sufficient data was collected to use half of it for model calibration and the other half for model validation.