Date of Award

2016

Degree Type

Thesis

Degree Name

Master of Engineering Science

School

School of Engineering

First Advisor

Associate Professor Sanjay Kumar Shukla

Second Advisor

Dr Steven Richardson

Abstract

Analysis of the right turn lane length of urban roads in left-hand traffic countries, such as Australia, UK and India (left-turn lane length in right-hand traffic countries such as USA), at signalised intersections encounters two main geometric features namely, deceleration length and storage length. The literature shows that in routine practice, the deceleration length is generally estimated by using constant deceleration rate. Many researchers consider this assumption for all design speeds unrealistic as it does not reflect the influence of the pavement condition. Hence, it may be desirable to consider the pavement’s condition in terms of its longitudinal coefficient of friction in the design analysis. In regard to the storage length, a large number of the current guidelines and models estimate the storage length of right-turn lane at signalised intersections under split phase. Hence, there is a need to examine other phase types and timings and integrate the signal timing as a part of the geometric design

In this thesis, two analytical expressions have been analysed for the design of deceleration length. The first expression assumes a constant deceleration rate, and the second expression employs the concept of forces on a rotating wheel in which the coefficient of longitudinal friction between a vehicle’s tyres and the road surface is considered. The calculated deceleration lengths by these two expressions were compared with the recommended values in American and Australian standards as well as with the deceleration lengths that were obtained by a recent simulation study presented in the literature. It has been found that applying a constant deceleration rate of 2.74 m/s2 in the first expression provides the values of deceleration length comparable to most guidelines and studies. The second expression highlights the importance of using the pavement design in terms of the coefficient of friction to reduce the deceleration length in the case of limited space.

A MATLAB based simulation programme has been developed to provide an estimate of the right-turn lane storage length for different traffic volumes in order to avoid the problems associated with blocking and overflow of right turn vehicles in 95% of cycles. In established intersections that cannot be modified due to physical constraints, the model is flexible enough to examine different signal phase types and timings and provides other solutions to reduce overflow and/or blockage situations. The simulation model also takes into consideration the leftover queue. The model results have been compared against an available analytical method in which similar signal phases and timings were investigated. The outcomes are similar to those of the analytical model in most of the signal phase types. The simulation model provides the flexibility to estimate the right-turn lane length for different combinations of through lane and right-turn lane traffic volumes.

The developed simulation model has also been validated against the field data using three parameters, namely 95th percentile of maximum queue, overflow cycle percentage, and blockage cycle percentage. Comparing with the field observations yields a level of accuracy in the range of 78%-85%. Finally this simulation model has been used to optimise the green time in the case of split phase that demonstrates a large difference in traffic volumes of two opposite approaches; this could reduce the mean wait time by up to 28%.

Share

 
COinS