Effects of driver characteristics and traffic composition on traffic flow

TR Number
Journal Title
Journal ISSN
Volume Title
Virginia Tech

This paper describes the development of simulation models for a variety of traffic flow scenarios. The major goal of the models was to evaluate the effects of driver characteristics and traffic composition on traffic flow. The five scenarios modeled and their respective objectives were as follows:

  1. Vehicles switching lanes to increase speed. Objectives were thruput and number of lane switches.
  2. Vehicles merging into an adjacent lane. Objectives were distance traveled before merging and number of collisions during lane switching.
  3. Vehicles switching from the left or right lane into the center lane. Objectives were number of collisions and number of new misses during lane switching.
  4. Vehicles passing on a two-lane bidirectional road. Objective was number of collisions during passing.
  5. Vehicles switching from the center lane to the left or right lane to avoid an impassible obstacle. Objectives were number of collisions during lane switching and number of collisions with obstacle.

Various driver characteristics were implemented in the models. The concept of preoccupation/attentiveness was factored into the models through the use of varied reaction times. 0ther driver characteristics were incorporated in the models via the assignment of vehicle speed. The models provided for a wide variety of driver types. Examples are as follows:

  1. Drivers in a hurry.
  2. Tourists or drivers unfamiliar with the area.
  3. Law-abiding drivers.
  4. Aggressive and passive drivers.
  5. Young, inexperienced drivers.
  6. Tired truck drivers.

The driver characteristics were varied via percentage allocations entered at run-time. The traffic composition for the models consisted of automobiles and multi-axle vehicles of fixed lengths. The percentages for each vehicle type were also entered at run-time.

The scope and level of detail for each model was delineated with assumptions. General assumptions made included the following:

  1. An automobile is 10 feet long, a multi-axle vehicle is 30 feet long.
  2. The width of a lane is such that only one vehicle can be accommodated at a time.
  3. A vehicle is considered to be entirely in one lane or another.
  4. A vehicle switches lanes instantaneously.
  5. The reaction time of an attentive driver is normally distributed with a mean of .5; the reaction time of a preoccupied driver is normally distributed with a mean of .7. Three standard deviations are included to ensure complete population coverage.
  6. A collision between two vehicles results in the termination of the vehicle causing the collision: the other vehicle continues.

Implementation of these models was performed using the student version of the simulation language GPSS/H. The models were validated. but not verified against their real world counterparts. Test results showed that select driver characteristics can affect Traffic flow; however, the effect of traffic composition was relatively unshown.