Modelling of segment level travel time on urban roadway arterials using floating vehicle and GPS probe data

Abstract

With the increasing traffic congestion levels on urban arterials, an essential step to tackling this challenge is to effectively quantify it and understand how it relates to its contributing factors. Although researchers have proposed models that relate travel time with several traffic and roadway factors for quantifying congestion especially in the Western world, most of these models lack the predictive accuracy for arterials in low-income countries due to differences in roadway and roadside interference factors, heterogeneous traffic flow, and others. A number of them do not incorporate delays from factors including on-street parking activities. Additionally, some existing models are complex in structure and require several parameters which may not be available in many low-income countries such as Ghana or may be expensive to collect. This study aimed at exploring a simple model for predicting travel time which will capture the contributing factors of congestion typical of low-income country arterial road environment and flow characteristics using a multiple linear regression model. Using moving observer method, traffic and roadway data were collected from eight arterial roadways in the Greater Kumasi Metropolitan Area for modeling travel time at the segment level. The fitted model that captures the impact of factors including on-street parking, access density, traffic density, and segment downstream bounding conditions on travel time will aid decision making by transport planners on the factors to consider to mitigate congestion. The model demonstrates how inadequate enforcement of on-street parking restrictions on arterial roadways exacerbates congestion during periods of high demand.