As the United States continues to build and repair the ageing highway infrastructure, the bearing of freeway work zones will continue to impact the capacity. To predict the capacity of a freeway work zone, there are several tools available for engineers to evaluate these work zones but only microsimulation has the ability to simulate the driver behavior. One of the limitations of current car-following models is that they only account for one overall behavioral condition. This dissertation hypothesizes that drivers change their driving behavior as they drive through a freeway work zone compared to normal freeway conditions which has the potential to impact traffic operations and capacity of work zones. Psycho-physical car-following models are widely used in practice for simulating car-following. However, current simulation models may not fully capture car-following driver behavior specific to freeway work zones. This dissertation presents a new multidimensional psycho-physical framework for modeling car-following based on statistical evaluation of work zone and non-work zone driver behavior. This new framework is close in character to the Wiedemann model used in popular traffic simulation software such as VISSIM. This dissertation used two methodologies for collecting data: (1) a questionnaire to collect demographics and work zone behavior data and (2) a real-time vehicle data from a field experiment involving human participants. It is hypothesized that the parameters needed to calibrate the multidimensional framework for work zone driver behavior can be derived statistically by using data collected from runs of an Instrumented Research Vehicle (IRV) in a Living Laboratory (LL) along a roadway. The design of this LL included the development of an Instrumented Research Vehicle (IRV) to capture the natural car-following response of a driver when entering and passing through a freeway work zone. The development of a Connected Mobile Traffic Sensing (CMTS) system, which included state-of-the-art ITS technologies, supports the LL environment by providing the connectivity, interoperability and data processing of the natural, real-life setting. The IRV and CMTS system are tools designed to support the concept of a LL which facilitates the experimental environment to capture and calibrate natural driver behavior. The objective is to have these participants drive the instrumented vehicle and collect the relative distance and the relative velocity between the instrumented vehicle and the vehicle in the front of the instrumented vehicle. A Phase I pilot test was conducted with 10 participants to evaluate the experiment and make any adjustments prior to the full Phase II driver test. The Phase II driver test recruited a group of 64 participants to drive the IRV through an LL set up along a work zone on I-95 near Washington D.C. in order to validate this hypothesis In this dissertation, a new framework was applied and it demonstrated that there are four different categories of car-following behavior models each with different parameter distributions. The four categories are divided by traffic condition (congested vs. uncongested) and by roadway condition (work zone vs. non-work zone). The calibrated threshold values are presented for each of these four categories. By applying this new multidimensional framework, modeling of car-following behavior can enhance vehicle behavior in microsimulation modeling. This dissertation also explored driver behavior through combining vehicle data and survey techniques to augment the model calibrations to improve the understanding of car-following behavior in freeway work zones. The results identify a set of survey questions that can potentially guide the selection of parameters for car-fallowing models. The findings presented in this dissertation can be used to improve the performance of driver behavior models specific to work zones. This in return will more acutely forecast the impact a work zone design has on capacity during congestion.
| Identifer | oai:union.ndltd.org:ucf.edu/oai:stars.library.ucf.edu:etd-5592 |
| Date | 01 January 2014 |
| Creators | Lochrane, Taylor |
| Publisher | STARS |
| Source Sets | University of Central Florida |
| Language | English |
| Detected Language | English |
| Type | text |
| Format | application/pdf |
| Source | Electronic Theses and Dissertations |
Page generated in 0.0024 seconds