The Network Reliability of Transport

The objective of this paper is to give an overview of various reliability concepts that have been developed in the last decades. The paper first summarises various indicators that have been developed in order to measure the reliability of a network and then looks at techniques to calculate these indicators. The usefulness and limitations of the different indicators is discussed. The paper suggests that there is no single perfect indicator but that the choice of indicator and technique depends on several factors, including the viewpoint of the analyst and the type and range of interventions being considered. In order to assess the impact of incidents the authors propose to distinguish between three types of intervention, namely “benevolent”, “neutral” or random, and “malevolent”. Also discussed is why the provision of up-to-date information to the traveller has a central role to play when trying to minimise the impact of an incident.

The importance of the adverse impacts of network degradation has stimulated substantial international research interest in transport network reliability, that is, the ability of degraded transport networks to cope with travel demand. Most of the recent research effort has focused on the reliability of urban passenger transport networks, in terms of the probability that the network will deliver a required standard of performance. This situation is characterised by high levels of congestion, a dense road network, and quantifiable probability of degradation of the network. Outside major urban centres, the situation is very different. The main dominant consideration in transport network infrastructure provision is accessibility - linking urban centres, providing regional coverage, and basic levels of accessibility for the non-urban community and economy. The network is sparse, congestion is not a significant issue, and access to essential community services and to markets is the major driving force underlying network development. In this context, the vulnerability of the network is perhaps more important than ‘reliability’. This paper develops the concept of network vulnerability. It begins by reviewing the current state of research into network reliability, then proposes extensions and adaptations to the reliability concepts that are more appropriate for strategic-level multi-modal transport systems. Several alternative definitions for vulnerability are proposed. The paper also discusses the development of algorithmic and visualisation tools that may be used to identify specific ‘weak spots’ in a network, where failure of some part of the transport infrastructure would have the most serious effects on access to specific locations and on overall system performance. Finally, the paper describes potential applications of network vulnerability concepts, and proposes directions for further research.

Risk evaluation and management methods are used to assess the reliability of a New Zealand inter-urban road network, which is subject to closures due to snow and ice, earthquakes, volcanic activity and road accidents. Using the probabilities and consequences of closures of various durations, the expected annual costs of closures are determined for each hazard. The benefit-cost ratios for various risk mitigation options are also identified. The importance of considering both the probabilities and consequences of closures is discussed.

It is found that a ‘high-frequency, low-consequence’ hazard (snow and ice) has the highest expected annual cost, and higher than for the ‘low-frequency, high-consequence’ hazards (earthquakes and volcanic activity). It is also found that not allowing for elasticity in the demand for travel (e.g. trips being cancelled or postponed because of road closures) leads to inaccurate estimates of the costs of closure.

It is concluded that it is important to allow for the behavioural responses of network users when estimating the costs of closures, and that more attention should be given to ‘high-frequency, low-consequence’ hazards, in order to maximise the economic benefits of expenditure on risk mitigation.

This paper attempts to assess the parking reliability with taking into account drivers' travel choice behaviors under various traffic conditions. Apart from the well-known travel-time reliability, a parking reliability is newly defined as the probability that the drivers' average searching time for parking is less than a given threshold. This is particularly important under conditions of shortage of parking spaces in urban areas. A Monte Carlo simulation approach, which incorporates a combined trip distribution and assignment model with explicit elastic demand function, is proposed to estimate the two reliability measures of road network (i.e. travel-time and parking reliabilities). A numerical example is used to illustrate the applications of the reliability measures and the proposed approach.

In this paper, a model to estimate travel time reliability is proposed assuming a stochastic user equilibrium. Travel time reliability is defined as the probability that travel time between the origin and destination does not exceed a standard travel time corresponding to each service level. Also it is shown how to calculate travel time reliability in a large-scale network. This model is applied to the road network in the Hanshin area to evaluate new lines in the national road plan from the viewpoint of travel time reliability. Some important links and lines in the future road network are evaluated from the viewpoint of travel time reliability.

Travel time variability has generally been recognized as one of the most important attributes in travelers' route choice decisions. In fact, many empirical studies have indicated that both passengers and freight carriers are strongly averse to travel time variability, because it introduces uncertainty to their route choice decisions. In this chapter, we examine the effect of incorporating travel time variability and risk-taking behavior into the route choice models and its impact on the estimation of travel time reliability under demand and supply variations.

This paper attempts to assess the transit service reliability with taking into account the interaction between network performance and passengers' travel choice behaviors. Besides the well-known schedule reliability, a waiting-time reliability is newly defined as the probability that the passengers' average waiting time is less than a given threshold. A Monte Carlo simulation approach, which incorporates a stochastic user equilibrium transit assignment model with explicit capacity constraints and elastic frequencies, is proposed to estimate the above two reliability measures of transit service. A numerical example is used to illustrate the applicability of the reliability measures and the proposed approach.

How an individual makes a travel decision under uncertain conditions has been one of the critical issues in designing the information that is delivered by Intelligent Transportation Systems (ITS). This has been a difficult problem because a suitable mathematical framework that deals with the interaction among uncertainty, information, and traveler's attitude toward uncertainty is not available. This paper introduces a possibility theory framework, and demonstrates how this framework represents the uncertainty perceived by the traveler, and calculates the feasibility of achieving the travel objective under different degrees of specificity of information. We present this framework in the setting of selecting the time of departure in the face of a not well-defined desired arrival time and estimated travel time. Feasibility of arrival is measured by the possibility and necessity measures; these measures represent two views, optimistic or conservative, respectively. Anxiety that is associated with the departure time, whether to leave now or not is modeled by Yager's anxiety measures. The anxiety measure considers the conflict of impelling forces between “to leave now” and “not to leave yet”; possibility and necessity measures for these outcomes represent these forces. Thus, along the time axis of possible departure time, possibility and necessity measures of arrival and non-arrival are computed and accordingly, the anxiety measure associated with each departure time. The range of time within which a traveler decides to leave is indicated by the anxiety measure. The size of the range is sensitive to the specificity of information. The more specific the information, the smaller the range of time. While the purpose of this paper is to introduce the mathematical framework useful for the analysis of a traveler's decision under uncertainty, the analysis raises an interesting issue of the paradoxical effects of information accuracy also.

Travel behaviour depends on the travellers' perception of the trip characteristics. The uncertainty of the journey time is one of the determinants of the choice. Studies of decision behaviour in uncertain conditions show the limited validity of the common assumptions in traffic models. Decision making under uncertainty has to be considered as a contingent process, depending on the objectives and conditions with which the choice is made. Expected utility is one of the many possible criteria used by people to decide. There is a discrepancy between the objective and subjective value of this concept. This is due to the bias with respect to the perception of very likely and very unlikely events. In many cases the expected utility is a less applicable objective, e.g. because people want to arrive before a certain deadline and maximize the probability to achieve that. The fact that probabilities of travel times have to be learned from experiences introduces a dynamic character of choice behaviour.

Most of the reported properties of decision-making under uncertainty still have to be verified for travel behaviour.

In this study, a mode choice model explicitly considering travel time reliability is developed. This model quantifies travelers' attitudes towards travel time variability as well as average travel time. Data were collected from the morning commuters who have two or three alternative modes including some public transportation and private vehicles. The survey period includes both a normal period where all the transportation modes were available and an abnormal period where the main major public transportation service was closed. The model is applied to practical commuters' decision making, and one of the findings in the mode choice model is that they pay relatively large attention to the travel time variability. In this model, travel time variability is dealt with as the possibilities that the commuters arrive before or after their job starting time separately. The best-fit model indicates that the commuters pay more attention to early arrival and less to late arrival in the normal period. In the abnormal period, however, their attention shifts drastically to late arrival. This suggests that the commuters behave optimistically in the normal period and pessimistically in the abnormal period.

This study was performed to examine the relationship between dynamic traffic information and driver's route choice behavior when non-recurrent events cause traffic jams. To enhance the effectiveness and applicability of the information system as a traffic control measure, it is important to investigate the influence of the software aspects of a dynamic traffic information system; these software aspects include information content and the timing of information delivery. This study was undertaken to investigate whether providing drivers with dynamic traffic information might relieve traffic jams and improve travel time reliability. A questionnaire survey was conducted to obtain data regarding the respondents' attitudes to detours and their stated preference (SP) of route choice behavior. These data are used to analyze the characteristics of drivers who show negative and positive attitudes to detours, including driver information usage and knowledge about the alternative routes. Also, this study analyzes the relation between the way to provide drivers with information and driver's decision-making on route using the SP data. Especially, the analytical attentions are paid to the influences of both timing and contents of information provided upon the respondent' route choices.

In this paper, we discussed relations between drivers' cognition and the dynamics of traffic conditions. When we take into account drivers' cognition, many types of psychological impedance to behavioral change could be accounted for, including cognitive conservatism and habitual decision making. Based on psychological theories with respect to these impedance to behavioral change, we elucidated the process of changes in traffic conditions that might be induced by changes in road-network structure. As a result of the discussion, we concluded that traffic conditions are unlikely change as much as predicted by theories that assume a simple, and sometimes unique, relation between a traffic condition and a road-network structure, e.g., theories that assume user equilibrium.

This paper describes and extends the game theoretic approach to network vulnerability assessment. The basic idea is to set up a game between the network users who are trying to minimise their expected travel time by choice of route and a network tester who is trying to penalise the users most by degrading a link through capacity reduction leading to congestion. The method therefore finds the worst possible location for a link degradation, taking re-routing options into account (an upper, lower bound of impact). The original game identifies the weakest link for routes between an OD pair in the network. Two variations are introduced in this paper in order to determine the weak links for a specific origin or a specific destination and for the whole network. All three game variations are tested on a small network in Leicester and the results are presented.

An optimal network design model is formulated providing a set of link investment pattern for the most reliable network with highest network performance under uncertain conditions. The connectivity probability of a link is assumed to be improved by the investment to the link. The object function is represented as the expected performance measure. The formulated model is categorized in a group of stochastic network design problem in which the existence of a link is probabilistic. The characteristics of the gradient vector of the objective function are analyzed. The derivatives of the objective function can be approximately evaluated without enumerating all possible network state vectors. Numerical examples are calculated for analyzing the sensitivity of optimal investment policies.

Emergency vehicles should be planned to have smooth access to people's residences after a large scale disaster. This study proposes ANR, the Accessible Node Rate, which represents the ratio of the number of accessible nodes to total nodes in the network, as an indicator for evaluating the reliability of a local street network. Using the ANR, this study evaluates the urban districts of Nada Ward in Kobe City where the earthquake caused extensive damage, and proposes planning changes to provide the local street networks with higher reliability. Furthermore, in order to see if the ANR can be used for analyzing another city's reliability, a case study of Beppu City which is supposed to have the possibility of the same magnitude of earthquake, is examined. A system that supports traffic network editing and analysis for reducing the workload in a wider area is developed.

Street closure following an earthquake makes life-saving and rescue work more difficult, especially in congested urban areas. After the great Chi-chi earthquake, recovery and reconstruction work became necessary. In particular, we have to investigate the street network damages and other effects. Considering the road improvement plan, not only malfunction recovery but also a comprehensive evaluation should be considered in future urban planning. The reliable street network against earthquakes is the central concern of this survey.

This study attempts to implement a case study survey of damaged areas, and analyze the effecting factors on street closure. We also try to evaluate the road function against earthquake integrated with the street-damaged characteristics. Using a discriminant model of street closure, some design guidelines for future transportation planning decisions are proposed.

Nan-tou City network has 365 links and 66.6% of them are less than 6 meters in width. Ton-shi Town network has 248 links and 35% of them are less than 6 meters in width. The same definition of street closure (impassable by vehicles) of ‘street width less than 4 meters’ was applied to these two case studies. This research considered several street network alternatives, the inaccessible nodes for all alternatives and analysis in order to determine a more reliable street network for increased safety. Results of both surveyed areas suggest that when we decrease the interval distance of the 8-M street, the percentage of the differences of inaccessible nodes will also decrease.

The Chi-chi earthquake did great harm in the disaster area. Fortunately we find the street damages of our surveying area were not too severe to maintain their functions in times of earthquake. However, it is still meaningful for a transport planner to evaluate the reliability of residential streets because the main target of planning a residential street network is to restrain through-traffic in this area to ensure safe and comfortable conditions for pedestrians while meeting residential access requirements.

Road network performance when a large disaster happens depends on how the road traffic is regulated. To evaluate the performance of the road network, one should consider the traffic regulation in an emergent condition. Authors have proposed the idea of area traffic regulation when a large disaster occurs (Iida, et al., 2000). In the former paper, a bi-level optimisation model to calculate the optimal regulation ratio is proposed. This paper proposes an efficient and fast method to calculate regulation ratios for two-stage road traffic regulation. The simplified model adopts linear programming method instead of bi-level optimisation method, which is quicker and simpler. The simpler algorithm contributes to conduct various kinds of case studies, and helps finding a reliability of the road network considering traffic regulations after a disaster.

This paper presents integrated models of probabilistic vehicle routing and scheduling model with time windows and dynamic traffic simulation. The probabilistic model incorporates the uncertainty of travel times. After applying the models to test road network, incorporating the uncertainty of travel times using the models not only allows freight carriers to reduce their total costs, but also improves the environment in terms of decreasing CO₂ emissions. As well, the paper investigated the effects and profitability of co-operative freight transport systems using a simulation model based on forecast vehicle routing and scheduling model. Results showed cooperative freight transport systems can lead to a substantial reduction in total delivery costs and total travel times within the whole road network.

This paper provides a day-to-day analysis of the reliability of commuting time and trip scheduling under the Advanced Traveler Information System (ATIS). A simple network with parallel routes and bottleneck congestion is used to simulate the departure time and route choice decisions of commuters to minimize total travel time and scheduling delay cost. There are two major factors influencing the decisions of drivers in their departure time and route choices: their accumulated travel experience and information provided by ATIS. A simple experiment is carried for investigating trip-scheduling reliability of this network system.

This paper presents models for evaluating effects of implementing TDM schemes on road traffic with the consideration of uncertainty of travel times. The models are composed of drivers' behavioural model for identifying the optimal starting time and route with the designated time for arrival at the destination as well as dynamic traffic simulation model for estimating the change of travel times. A case study for applying the models on a test road network showed that TDM schemes of park and ride systems and road pricing enabled drivers to obtain benefits by improving the reliability of travel times as well as benefits by reducing travel times.

Computer models are often used for studying the effects of changing conditions in the road network. State-of-the-art macroscopic models generally take some kind of network equilibrium approach and therefore have difficulties in appropriately representing short-term capacity reductions, probably resulting in too low estimates of delays. Recently developed dynamic models may be more promising. The purpose of this paper is to investigate the implications of model choice further, as well as the possibilities to study effects of short-term incidents. Three different computer programs were used: TRACKS, SATURN, and Paramics. The results show that microsimulation is a feasible tool for studying short-term disturbances in the road transportation system.

Traffic accident on inter-city expressways might cause large-scale traffic congestion. It might increase travel times of many drivers and it produces a large social loss. This study aims to estimate the social loss of travel time of drivers caused by traffic accidents on inter-city expressway using traffic simulation model, and to evaluate the effects of outflow recommendations when an accident occurs on the expressway. The traffic simulation model on Tomei Expressway is constructed to estimate the dynamic traffic congestion. Travel time losses of drivers are estimated by the simulation results with hypothetical traffic accidents. It is understood that the total losses of travel times are depending on the positions of accident spots and the occurrence times of accidents, because it might influence to congestion at the bottlenecks of flow capacity. Moreover, the effect of traffic control in emergency situation is discussed. The influences of outflow recommendations for drivers are estimated using the traffic simulation model.

Ramp meters in the Twin Cities were turned off for 8 weeks in the Fall of 2000. This paper analyzes traffic data collected in this experiment on travel time variability with and without ramp metering for several representative freeways during the afternoon peak period. Travel time variability is generally reduced with metering. However, it is found that ramp meters are particularly helpful for long trips relative to short trips. The annual benefits from reducing travel time variability with meters are estimated to be $33.1 million, compared to the annual ramp metering costs of $2.6 million in the Twin Cities metro area. Thus, the impact on travel time variability should be captured in future ramp metering benefit/cost analysis.

This study reports on the effect of demand variation on the optimal location of road-pricing cordons. The optimal road-pricing cordon, in this study, aims to maximise the social welfare function. This optimisation program is categorised as Bi-level optimisation programming which is a NP hard problem. The paper first describes the method developed to solve the optimal toll problem for a given set of chargeable links. The tests were carried out with a small toy network and a larger scale network. For the small network, four single user class demand characteristics were varied individually; these were the elasticity of trip generation with respect to increases in travel cost, value of travel time, volume of traffic, and traffic distribution pattern. For the larger scale network, only elasticity, value of time, and trip volume were tested. The results of the larger scale network are also analysed by including the cost per toll point. The tests with the larger scale network were modified so that the constraint of uniform charge is applied. The results showed that demand variation could influence the best location of toll points. This finding raises the question of whether the implementation of the same cordon all day in an urban traffic network is the optimal approach under the existence of demand variations by time of day, and also whether the evaluation process of the cordon location should consider the effect of different time periods together.