Coordinated Ramp Metering Strategies

Get Complete Project Material File(s) Now! »

Chapter 3. METHODOLOGY

This chapter discusses the research methodology incorporating study framework, data collection, preliminary analysis and simulation model development.

Research Framework

Figure 3.1 presents the methodological framework of this study. After identifying research gap, traffic data of a critical bottleneck section on State Highway 1 of Auckland Motorway is collected. Then the traffic occupancy and volume data is analyzed to investigate traffic behavior in the vicinity of the bottleneck section. Based on data analysis results, AIMSUN model is developed for the test bed. Two existing RM algorithms are assessed namely: ALINEA and HERO. Then, a logic tree based VSL control algorithm is also assessed. The results revealed that the existing logic tree based VSL control algorithm cannot improve significantly mobility performance of motorway systems and is rather rough for the speed control. Thus two alternative VSL control algorithms are proposed namely: a modified logic tree based controller and a fuzzy logic based controller. Both utilize optimized control logics and detector-controller configurations. In the next step, an integrated RM and VSL control strategy is proposed to address the issues confronted by RM. The integrated method is aimed to preserve the traffic flow at bottlenecks close to their capacity and avoid excessive delays at on-ramps. Finally, this research investigates the relationship between efficiency and equity using a modified HERO strategy.

Data Collection

The purpose of the data collection stage is to collect the attributes for model development. This research requires extensive data covering different traffic attributes from motorway network, which can be divided into two categories namely: static data and dynamic data.

Static Data

This category contains the physical and technical characteristics of the motorway network (e.g. the number of lanes, width, location of ramps and their control mechanism). This data can be obtained from GIS viewer provided by Auckland Council. A critical bottleneck section on State Highway 1 (SH1) of Auckland Motorway connecting Central Auckland with Northern Auckland is selected for this study. Figure 3.2 presents a layout of the study section, which has 5 on-ramps and 4 off-ramps in a direction towards Auckland city centre. Here O1 represents on-ramp from Esmonde Road while O5 represents on-ramp from Greville Road.

Dynamic Data

The traffic data used in this study is provided by New Zealand Transport Agency (NZTA) including loop detector measurements from the mainline, off- ramps and on-ramps and accumulated over a 30 seconds time period. Three months data is collected for the State Highway 1 starting from 5th of March 2012 to the 27th of May 2012.

Preliminary Analysis

The occupancy and volume data is analyzed to investigate traffic behavior in the vicinity of the bottleneck section D1. In this study a method proposed by Kianfar et al. (2013) is employed to identify the capacity and critical occupancy at the bottleneck section. Critical occupancies are determined using a two-step procedure. First, the scatter plots of flow versus occupancy data points are observed for any clear change in trends. The initial trend of increase in flow with increase in occupancy is observed. This trend reverses after the critical occupancy is reached. Nevertheless, the exact critical occupancy value is not easily discernible from the scatter plot. Thus, to determine the exact critical occupancy value, regression lines are fitted for free flow and congested regions for different possible critical occupancy values, and the root mean square error (RMSE) are calculated. The critical occupancy value that produces the least RMSE is selected. Then, the before breakdown flow and after breakdown flow values are obtained from best fit-lines at critical occupancies.
Data treatment is performed to validate the available data and remove data that is faulty or irrelevant for the current analysis. Daily data that provides little or no information regarding the effect of traffic congestion on traffic conditions is not taken into account in the analysis. This faulty or irrelevant data includes weekends, incidents and days with adverse weather. A total of 53 flow breakdown points are extracted for this particular location D1. Figure 3.3 presents a scatter plot of occupancy versus flow data points just before flow breakdowns for the bottleneck section. In this figure, a vertical line is drawn at an occupancy value of 17 that separates two different traffic flow conditions. Left side of the line represents cases with occupancy lower than 17% where flow breakdown points are distributed in a range from 2100 to 2400 veh/h/l with most of breakdown points located in a range from 2200 veh/h/l to 2300 veh/h/l. While right side of the line represents cases with occupancy greater than 17% where flow breakdown points are distributed evenly in a wider range from 2000 to 2350 veh/h/l.
Frequency distribution analysis is also performed for these flow breakdown points, which is presented in Figure 3.4 for flow and Figure 3.5 for occupancy data points just before flow breakdowns. Based on chi-square distribution test, both flow and occupancy data points just before flow breakdowns fit well in normal as well as lognormal distribution functions. For flow data points, the mean and standard deviation are recorded as 2228.5 veh/h/l and 75.8 veh/h/l; while chi-square values for normal and lognormal distribution functions are recorded as 6.9 and 7.4 respectively with a critical chi-square value of 14.1. For occupancy data points, the mean and standard deviation are recorded as16.5 % and 1.5%; while chi-square values for normal and lognormal distribution functions are recorded as 12.9 and 11.5 respectively with a critical chi-square value of 22.4.

READ  SIMULATION OF REAL SURFACES: CASE OF TEXTILE FABRICS

Gini Coefficient

Gini coefficient proposed by Gini (1936) is a widely accepted measure in economic studies to analyse the degree of inequality in income distribution. Figure 3.6 illustrates the concept of Gini coefficient. The Lorenz curve (Lorenz 1905) shows the proportion of X receiving a given proportion of Y. While 100% of the population receives 100% of the resource, the more unfortunate 50% may only receive 25% of the total resource. The Gini coefficient corresponding to this Lorenz curve can be computed as A1/(A1+ A2) in this graph. A zero value for Gini coefficient indicates perfect equality, while 1 indicates perfect inequality.

1. INTRODUCTION
1.1. Background
1.2. Research Gap
1.3. Aims and Objectives
1.4. Scope of the Research
1.5. Structure of the Thesis
2. LITERATURE REVIEW
2.1. Local Ramp Metering Strategies
2.2. Coordinated Ramp Metering Strategies
2.3. Variable Speed Limits
2.4. Combination of RM and VSL
2.5. Efficiency versus Equity
2.6. Simulation
3. METHODOLOGY
3.1. Research Framework
3.2. Data Collection
3.3. Preliminary Analysis
3.4. Performance Measures
3.5. Simulation in AIMSUN
4. ASSESSING EXISTING RM AND VSL CONTROL METHODS
4.1. Simulation in AIMSUN
4.2. Analysis Results
4.3. Summary
5. A MODIFIED LOGIC TREE BASED VSL CONTROLLER
5.1. Existing Logic Tree based Algorithm and Proposed Modifications
5.2. Analysis Results
5.3. Summary
6. A FUZZY LOGIC BASED VSL CONTROLLER 
6.1. Fuzzy Logic Algorithm
6.2. Simulation in AIMSUN
6.3. Analysis Results
6.4. Summary
7. AN INTEGRATED RM AND VSL CONTROLLER
7.1. Integrated Method
7.2. Simulation in AIMSUN
7.3. Analysis Results
7.4. Summary
8. RELATIONSHIP BETWEEN EFFICINENCY AND EQUITY
8.1. Modified HERO Strategy
8.2. Combined Index
8.3. Analysis Results
8.4. Summary
9. CONCLUSIONS AND RECOMMENDATIONS 
9.1. Conclusions
9.2. Recommendations for Future Research

GET THE COMPLETE PROJECT
Towards an Efficient and Equitable Motorway System using Ramp Metering and Variable Speed Limits

Related Posts