- : Ms Word Format
- : 100 Pages
- : ₦3000
- : 1-5 Chapters
- Click to DOWNLOAD Materials
MULTI-RESOLUTION ANALYSIS OF DISRUPTIONS IN DIFFERENT URBAN NETWORK CONFIGURATIONS UNDER VARYING INFORMATION SCENARIOS
ABSTRACT
Urban networks are suffering more congestion than ever. Aside from recurrent congestion resulting from bottlenecks and signal timing, it is estimated that about 40 percent of the congestion is caused by non-recurring disruptions such as crashes, work zones, and special events. Although many recent works have studied the capacity and efficiency of different network configurations, their performance under such non-recurring disruptions is currently missing in the literature.
This dissertation compares three street network configurations under both prior-knowledge and no-prior-knowledge disruptions, where drivers do and do not know about the disruptions ahead of time. Two-way (TW), two-way with prohibited left-turn movements (TWL), and one-way (OW) networks are studied and compared with three approaches at different levels of fidelity. First, a simplified analytical method is applied to networks under light traffic conditions, where drivers are assumed to take the shortest physical route with the fewest turning maneuvers. This step provides initial insights into the problem through paper-and-pen analysis. Second, the Link Transmission Model (LTM) is applied to the networks for light-to-medium traffic demands. Traffic signals will be added at this step and drivers will be routed based on experienced travel time on the links. As a numerical solution to the kinematic wave theory, the LTM can capture network dynamics including intersection spillbacks, which makes it an ideal intermediate model between simplified analytical approach and more comprehensive microscopic simulation. Finally, the networks are tested in a microscopic traffic simulation environment for more detailed and realistic outputs. In addition, mitigation strategies that provide early disruption notification to a subset of vehicles either locally or globally in the networks are also tested in the simulation environment.
Based on the results of the experiments, it is recommended that the TW network should be used when traffic demands are expected to be moderate. Because of its flexibility, the TW network can accommodate link disruptions that might occur as long as the network capacity is still higher than the traffic demand. However, the TW network has limited capacity so performs poorly under high traffic demands. For networks operating under high demands, the TWL network is recommended due to its large capacity that allows it to absorb the negative impacts of link disruptions. For all network configurations, it is important to provide road users with early notification to offset the negative impacts when disruptions occur within the most critical (central) part of the network. In the TW network, this is best done using VMSs placed near the disruption while in the TWL network it is more efficient to use a broadcasting strategy (e.g., using in-vehicle navigation systems). For disruptions near the network periphery, no prior information should be provided to road users so that the detour traffic is more likely to use links outside the central area with relatively low traffic, which provides a more even congestion distribution.
TABLE OF CONTENTS
List of Tables………………………………………………………………………………………………………. vii
List of Figures……………………………………………………………………………………………………. viii
Acknowledgments…………………………………………………………………………………………………. x
- Introduction……………………………………………………………………………………………………… 1
- Literature review………………………………………………………………………………………………. 5
2.1. A brief history of network configuration changes in American cities………………… 5
2.2. Operation of networks from a macroscopic perspective…………………………………… 7
2.2.1. Aggregated traffic models………………………………………………………………………. 7
2.2.2. Comparisons between different network configurations…………………………… 10
2.3. Resilience in a traffic network context…………………………………………………………. 12
2.4. Summary…………………………………………………………………………………………………. 14
3.1. Simplified analytical approach……………………………………………………………………. 16
3.2. Link Transmission Model (LTM)……………………………………………………………….. 17
3.3. Microscopic simulation……………………………………………………………………………… 18
4.1. Routing within TW and TWL networks……………………………………………………….. 20
4.2. Routing within OW networks……………………………………………………………………… 25
4.3. Results…………………………………………………………………………………………………….. 27
4.3.1. Networks without link disruptions…………………………………………………………. 27
4.3.2. Networks under prior-knowledge (PK) disruptions………………………………….. 30
4.3.3. Networks under no-prior-knowledge (NPK) disruptions…………………………… 35
4.3.4. Comparisons between prior-knowledge and no-prior-knowledge results…….. 38
4.4. Summary…………………………………………………………………………………………………. 39
5.1. Model set-up…………………………………………………………………………………………….. 43
5.1.1. Network structure………………………………………………………………………………… 43
5.1.2. Intersection modeling…………………………………………………………………………… 43
5.1.3. Vehicle routing……………………………………………………………………………………. 44
5.2. Network-level models for operation evaluation under higher traffic demands….. 46
5.3. Experiments under light traffic……………………………………………………………………. 47
5.4. Experiments under increasing traffic…………………………………………………………… 52
5.4.1. Networks without link disruptions…………………………………………………………. 53
5.4.2. Networks under disruptions – case studies and an overview……………………… 54
5.4.3. Networks under disruptions – a deeper look……………………………………………. 68
5.4.4. Networks under temporary no-prior-knowledge disruptions……………………… 82
5.5. Summary…………………………………………………………………………………………………. 89
6.1. Microsimulation set-up………………………………………………………………………………. 92
6.1.1. Network modeling……………………………………………………………………………….. 92
6.1.2. Vehicle routing……………………………………………………………………………………. 94
6.1.3. Parameters for routing model………………………………………………………………… 96
6.2. Networks without link disruptions………………………………………………………………. 99
6.3. Networks under link disruptions……………………………………………………………….. 102
6.4. Networks under temporary no-prior-knowledge disruptions…………………………. 106
6.5. Mitigation strategies………………………………………………………………………………… 109
6.6. Summary……………………………………………………………………………………………….. 114
7.1. Findings and contributions……………………………………………………………………….. 116
7.2. Future work……………………………………………………………………………………………. 120
References………………………………………………………………………………………………………… 123
Appendix A Additional figures for LTM results……………………………………………………. 127
Appendix B Additional figures for Simulation results……………………………………………. 128
1. INTRODUCTION
Urban traffic networks are experiencing more congestion than ever. The latest Urban Mobility Scorecard reveals that total commuter delay and fuel waste have steadily increased over the past two decades (1). As shown in Table 1-1, the total annual delay for auto commuters across the United States reached 6.9 billion hours in 2014, which resulted in 3.1 billion gallons of fuel waste and $160 billion worth of monetary loss.
Table 1-1 Congestion costs in America (1)
| 1982 | 2000 | 2010 | 2014 | |
| Annual delay per auto commuter (hours) | 18 | 37 | 40 | 42 |
| Fuel waster per auto commuter (gallons) | 4 | 15 | 15 | 19 |
| Congestion cost per auto commuter (2014$) | 400 | 810 | 930 | 960 |
| National commuter travel delay (billion hours) | 1.8 | 5.2 | 6.4 | 6.9 |
| National commuter fuel waste (billion gallons) | 0.5 | 2.1 | 2.5 | 3.1 |
| National commuter congestion cost (billion 2014$) | 42 | 114 | 149 | 160 |
The FHWA report Traffic Congestion and Reliability: Linking Solutions to Problem (2) categorized congestion across the country by source. The report estimated that approximately 60 percent of congestion was caused by recurring events—such as bottlenecks or poor signal timing— and inclement weather. Such congestion is expected by road users and thus can be considered as a routine in network operation and roadway users can plan for and accommodate such recurring congestion in their trip-making processes. In comparison, the other 40 percent congestion is caused by non-recurring sources that bring disruptions to a network, such as traffic crashes, work zones, or special events. The non-recurring congestion sources are more problematic as they are unexpected to road users and may significantly magnify congestion generated by recurring sources.
From the perspective of traffic engineers, there are various engineering options to address congestion due to recurring sources, such as bottlenecks and signal timing issues. However, unexpected disruptions are difficult to deal with. If some network configurations perform more robustly to such non-recurring disruptions (i.e., are able to accommodate these disruptions with fewer negative repercussions), a significant portion of urban congestion might be alleviated by the network itself.
There has been a recent wave of one-way to two-way street network conversions in American cities over the past two to three decades. Generally speaking, one-way streets are believed to better serve automobile traffics due to the higher flow capacities and vehicle operating speeds (3–6), but they pose more threats to vulnerable road users and cause confusion to visitors (7–9). Two-way streets, on the other hand, are more appealing to multi-modal traffic, communities and street-side business (10–12). Most of the previous research on the operational performance of two-way and one-way streets compared vehicle speeds, delays, and capacities at a corridor level and concluded one-way streets are better than two-way streets. However, higher speeds and lower delays in a one-way street network do not guarantee road users actually complete their trips faster because trips in one-way streets typically have longer travel distances than in two-way streets.
Several recent works looked into operation of street configurations at a network level using macroscopic metrics such as trip completion rate (rather than speed) or trip-serving capacity (rather than roadway capacity) (13–15). These metrics combine the additional vehicle flow capacity with the additional travel distances that one-way streets impose onto traffic. Contrary to previous beliefs, these works indicated that two-way street networks are not necessarily inferior to one-way street networks in terms of vehicle operation.
In a previous research work on comparison between one-way and two-way networks (15), a two-way network without left turns was found to offer the best balance between travel distance and intersection delay under low demand scenarios, but its performance degrades at high congestion levels due to lack of redundant shortest paths available. This provides an indication that two-way network without left turns might be less resilient to disruptions. However, these indications were obtained from analyses of complete grid networks without disruptive events.
Although a few works studied network operation when individual links or lanes were removed (16, 17), they focused on permanent removals from a planning perspective. A high proportion of links/lanes were removed in these works and the performance of networks were evaluated using equilibrium traffic assignments, which are not true in networks under temporary disruptions. In addition, only one specific network was studied in these works and no comparison was provided among different network configurations.
This dissertation proposes to quantify the resilience, or the ability to accommodate unexpected disruptions, of some common urban traffic network configurations. Specifically, it focuses on network configurations that are applied to grid networks, including one-way streets (OW) and two-way streets with (TW) and without (TWL) left-turning maneuvers in abstract grid networks. Scenarios where all road users do and do not have a priori knowledge of the disruption will be examined and compared. Three different approaches will be used to assess the network operation under disruptions with varying degrees of modeling fidelity and applicability to different traffic conditions. These methods are: a simplified analytical approach, a kinematic-wave-based traffic model, and a microscopic traffic simulation environment. The analytical approach is only viable for light traffic situations, but can provide insights into expected behavior in medium and heavy congestion. The latter two approaches can consider the full range of conditions but do so by modeling aggregate flows (the kinematic wave model) and individual vehicles (the microscopic simulations), respectively. The kinematic wave model will capture key features in urban networks such as queue spillback, but has limitations in vehicle routing due to its aggregate nature. The microscopic simulation approach allows more features in network geometry and vehicle dynamics, which brings more realistic outputs with higher computation requirements. It is hoped that general observations from a simpler approach can be replicated in more sophisticated models so that the network analysis can be done with lower computational requirement.
As prior knowledge is the key to minimize the negative impacts of disruptions and allow road users to make better routing decisions, mitigation strategies that provide early disruption notification to a subset of road users are examined and compared in the network configurations as well. Based on their coverage, the mitigation strategies can be categorized as local and global notifications, where the disruption information is sent to a sub set of road users near the disruption or across the network, respectively.
The rest of the dissertation is organized as follows. Chapter 2 reviews the existing literature on street network configurations, macroscopic traffic models, and resilience in traffic. Chapter 3 provides a brief introduction to the analytical approaches used in this dissertation. Chapters 4, 5, and 6 present the findings obtained from the three analytical approaches used to model network operation. Chapter 7 provides a summary of the findings and recommendations for future work.
MULTI-RESOLUTION ANALYSIS OF DISRUPTIONS IN DIFFERENT URBAN NETWORK CONFIGURATIONS UNDER VARYING INFORMATION SCENARIOS
