Reliability Distribution Modeling for Bus Dwell Time and its Contributing Factors: A Case Study in Selected Routes of Addis Ababa City
No Thumbnail Available
Date
2024-04
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
The initial point of contact between the passenger and the passenger service is at the bus stop. All buses in Addis Ababa City now operate in mixed traffic with no signal prioritization, with the exception of BDL (Bus Dedicated Lane) on a few chosen routes. Bus stops are significant components of the Public bus system, and the way they are run has a significant impact on the network's overall service level and transit effectiveness. The aim of this work was determining the most influential factor on dwell time and the likelihood of dwell time occurrence at stop. Quantitative and qualitative data were gathered for a study on bus dwell times at stops, with quantitative data obtained from bus stops and qualitative data from passengers. Manual field data collection was conducted due to the lack of Automatic Passenger Count (APC) and Automatic Vehicle Location (AVL) data. To ensure accuracy, data collection was repeated for three days at each stop and time frame, three bus type, and the averages were used for dwell time determination. Directionality was considered, categorizing bus stops into upstream, stop, and downstream sections. The lengths of these sections were determined based on observed bus maneuvers within a 10-meter radius during a pilot survey. Video recording and direct transcription onto paper sheets were used to collect data for a specific bus. Only buses stopping at designated stations within the specified range were considered for the study. The data were analyzed using a statistical model and the probabilistic method of analysis. The research successfully achieved its stated objectives, contributing to a novel approach for evaluating dwell time. The author developed two models to identify the most significant factors affecting bus dwell time specifically, the Gaussian and Weibull regression models. The Weibull regression model demonstrated superior accuracy and a better fit for predicting bus dwell time, with a significantly lower RMSE (0.075) and an Adjusted R-squared of 92.63%, compared to the Gaussian regression model (RMSE: 3.93, Adjusted R-squared: 93.00%). Moving to specific coefficients in the Weibull regression model, factors such as No_aligt. Weibull (Number of alighting), Alig.Weibull (Alighting time), Board.Weibull (Boarding time), Idle.Weibull (Idle time), No. Boa.Weibull(Numbers of Boarding), Odd.pen.Weibull (Odd. Penny Weibull), Re.boa.Webull(Re-boarding passengers), Accel.Weibull (Acceleration time), and Far side (Far side stop location) exhibited a statistically significant effect on bus dwell time at stops. However, the location of the far side stops and the time of day had no effect on dwell time. The combined effect of stop location and bus type did not show statistical significance (p-value: 0.845), surpassing the conventional threshold of 0.05. However, analyzing the variables individually revealed a significant difference between stop type location and bus type in relation to dwell time. Specifically, Alliance buses exhibited longer dwell times compared to single Sheger and Anbessa buses.
Description
Keywords
Bus stop, Dwell time, Gaussian regression, Regression models, Weibull distribution.