This study deals with the development and application of travel time functions based on signalized intersection delay in urban traffic assignments. The background problem of this study is that the existing assignment models can not effectively account for signalized intersection delay. The purpose o...
This study deals with the development and application of travel time functions based on signalized intersection delay in urban traffic assignments. The background problem of this study is that the existing assignment models can not effectively account for signalized intersection delay. The purpose of this study is divided into three parts: (1) to examine the characteristics and limitations of existing theories; (2) to develop a travel time model based on the signalized intersection delay and turn penalty; and (3) to apply the developed model to real networks and propose the most statistically significant model in urban traffic assignments. For the purpose of calibrating BPR parameters using real travel time data collected by a test car survey, regression analysis is carried out. However, the calibrated BPR model is not applicable to this study because the reliability of the model is very low(R2=0.003). It is thought that the BPR model does not contain variables reflecting signalized intersection delay although real travel time data contain those delays. Because it is judged that existing models have their limitations in expressing real travel time in urban signalized networks, this study gives particular attention to dividing the link travel time into link running time and stopped time at a node, making the model based on variables such as link travel (running and stopped)time, volume and geometry, and signal data from signalized intersections in Cheongju. For the purpose of applying the g/C of approaches to model building in the planning stage, in which it is difficult to collect concrete data on signalized intersections, this study substitutes revised percentage of lanes (considering the type of intersections) for g/C in calculating intersection delays. And this study develops a left turn penalty model for solving the problem of overlooking left turn delay time in estimating link travel time. This study applies BPR and divided travel time models to traffic assignments in the Cheongju network, and compares each model's result with observed real data. Traffic assignment is carried out using EMME/2 and the comparison of statistical indices with real data is programmed in Fortran. The main results are as follows. First, it is desired that the BPR model should be used in macroscopic transportation planning rather than urban signalized networks because it has limitations in expressing real travel time including stopped time at signalized intersections. Second, the study presented a revised percentage of lanes instead of g/C for calculating intersection delay, which was analyzed to be significant in the paired t-test. Third, the assigned results of applying these models to Cheongju network by EMME/2 were statistically compared with data observed from a test car survey in Cheongju. The analyses show that the BPR model does not consider intersection delay but the divided travel time model is fit relatively well to the observed data. Fourth, among the divided travel time models, Model 1, which does not include the v/c variable in estimating the link running time, was found to be closest to the observed data because the link running time model including the v/c variable gives rise to more errors. Finally, the travel time model showed a higher correlation coefficient with the observed turn volume or the turn ratio when turn penalty is applied to traffic assignment. Compared to the BPR model, the divided travel time model showed more stable estimation results even after the application of the turn penalty function. In addition, the form of the BPR model is recommended as the form of a turn penalty model because it does not decrease the accuracy of the model's result despite it is easily adapted to traffic assignments for its simple form.
This study deals with the development and application of travel time functions based on signalized intersection delay in urban traffic assignments. The background problem of this study is that the existing assignment models can not effectively account for signalized intersection delay. The purpose of this study is divided into three parts: (1) to examine the characteristics and limitations of existing theories; (2) to develop a travel time model based on the signalized intersection delay and turn penalty; and (3) to apply the developed model to real networks and propose the most statistically significant model in urban traffic assignments. For the purpose of calibrating BPR parameters using real travel time data collected by a test car survey, regression analysis is carried out. However, the calibrated BPR model is not applicable to this study because the reliability of the model is very low(R2=0.003). It is thought that the BPR model does not contain variables reflecting signalized intersection delay although real travel time data contain those delays. Because it is judged that existing models have their limitations in expressing real travel time in urban signalized networks, this study gives particular attention to dividing the link travel time into link running time and stopped time at a node, making the model based on variables such as link travel (running and stopped)time, volume and geometry, and signal data from signalized intersections in Cheongju. For the purpose of applying the g/C of approaches to model building in the planning stage, in which it is difficult to collect concrete data on signalized intersections, this study substitutes revised percentage of lanes (considering the type of intersections) for g/C in calculating intersection delays. And this study develops a left turn penalty model for solving the problem of overlooking left turn delay time in estimating link travel time. This study applies BPR and divided travel time models to traffic assignments in the Cheongju network, and compares each model's result with observed real data. Traffic assignment is carried out using EMME/2 and the comparison of statistical indices with real data is programmed in Fortran. The main results are as follows. First, it is desired that the BPR model should be used in macroscopic transportation planning rather than urban signalized networks because it has limitations in expressing real travel time including stopped time at signalized intersections. Second, the study presented a revised percentage of lanes instead of g/C for calculating intersection delay, which was analyzed to be significant in the paired t-test. Third, the assigned results of applying these models to Cheongju network by EMME/2 were statistically compared with data observed from a test car survey in Cheongju. The analyses show that the BPR model does not consider intersection delay but the divided travel time model is fit relatively well to the observed data. Fourth, among the divided travel time models, Model 1, which does not include the v/c variable in estimating the link running time, was found to be closest to the observed data because the link running time model including the v/c variable gives rise to more errors. Finally, the travel time model showed a higher correlation coefficient with the observed turn volume or the turn ratio when turn penalty is applied to traffic assignment. Compared to the BPR model, the divided travel time model showed more stable estimation results even after the application of the turn penalty function. In addition, the form of the BPR model is recommended as the form of a turn penalty model because it does not decrease the accuracy of the model's result despite it is easily adapted to traffic assignments for its simple form.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.