Kyung-Soo Chon

Transportation Network Design Considering Travel Time Reliability

Fluctuations in roadway capacity due to non-recurrent traffic incidents is a source of uncertainty of travel time. Travel time reliability, as a transportation performance measure, which may reflect travel time uncertainty was applied to network analysis. Recently, research which applies network reliability concepts to transportation network design has been undertaken. In this study, a continuous network design problem (CNDP) with a travel time reliability constraint was formulated as a bi-level model. A simulation-based genetic algorithm with penalty method was used to obtain the optimal capacity enhancement of each link, combined with a routine which estimates travel time reliability under fluctuations of link capacity. The effect of the travel time constraint on a CNDP was analyzed through a numerical experiment using a simple network. Additionally, the effect of budget allocation and factors to be specified in the model on the results were analyzed.

Development of Potential Measures to Evaluate the Safety of Curves in Rural Highways

This research develops a safety evaluation method based on individual driving behavior on two-lane curved highway sections. A field survey was conducted to collect lateral placement and operating speed data from individual vehicles at various curved sections. The data were then used to estimate the vehicle path radius required for individual vehicles at different locations of curved sections. With these field data and estimated properties, two different models are proposed to predict the driving behavior that is significantly affected by a horizontal curves design conditions. In these models, the driving behavior is presented in terms of the Path Deviation Index (PDI) and the Path Trajectory Ratio (PTR). In addition, accident data were collected from the highway sections the functions of which are identical to those of the sections where the driving behavior data were collected. Accident rates were then estimated to evaluate how strongly they relate to each parameter. Based on peer reviews of previous researches and the results of this study, an accident prediction model for twolane highways is satisfactorily presented.