Reliability-based Urban Transportation Network Studies

Urban traffic system is one of the most vital network systems which can maintaincity function work well. A stable and efficient traffic network is of crucial importanceto a city. However, for a real road network, traffic supply and traffic demand arealways uncertain, ignoring the uncertainties of traffic system will bring risks totravelers'trip choices and the construction and management of road networks. Thispaper aims to obtain a more reliable urban traffic network in an uncertain environment,at the same time, improve the trip quality of individual traveler. The main contents ofthe paper are described as follows.The researches on road network reliability evaluation index: we firstly apply theprobability theory to depict the characteristics of random supplies and demands andanalyze the probability feature of route travel time in a capacity degradable roadnetwork. Based on this, we propose the level of service reliability index for stochasticroad network and establish an evaluation model for Los reliability with the mostprobable states generation algorithm.The researches on road network reliability evaluation methods: firstly, a traveltime reliability evaluation method is proposed considering the correlative relationshipsbetween route flows and link travel times. Secondly, we establish a road networkcapacity reliability model by using bi-level programming, in which the OD pair traveltime reliability is introduced in the bi-level model as a constraint. Thirdly, we use theCornish Fisher asymptotic expansion obtain the route travel time budget with aspecified travel time reliability requirement, then set up a Los reliability calculationmodel considering travel time reliability based on the variational inequality technique.Finally, the mixed road choice model is adopted to analyze the influence of the marketpenetration and information quality of ATIS on road network reliability.The researches on trip choices under uncertain environment: a probabilistic userequilibrium traffic assignment model for capacity degradable road network is proposedwith a route choice criterion that the mean route travel time is minimized. Then, wedefine the link generalized travel cost from the point of view of decreasing travel time,travel cost and enhancing travel reliability, and further carry out some analyses of thedifferent route choice criteria on degradable network traffic flow distribution underATIS. Moreover, a stochastic traffic assignment model with multiple user classes andglobal elastic demands is formulated as an equivalent variational inequality. In the model, the travelers are differentiated into several classes by their own value of timeand different user class has different travel time reliability requirement. By introducingthe conditional value at risk theory in the process of route decision, we study thecombined travel route and travel mode choice model based on a hierarchical structureof the combined trip decision in conjunction with the discrete choice theory, sensitivityanalyses for model parameters on mode split rate are carried out by numericalexperiments.The researches on road network reliability optimization: a critical link recognitionalgorithm is established based on the failure probability and failure consequence byusing risk assessment theory. Then, we propose a measurement method for theimportance of network node with complex network theory, and further analyze theimpact of different travel network structure and different travel route choice model onnode importance. For uncertain road network design, we present a continuousequilibrium network design bi-level programming model considering stochastic supplyand demand simultaneously, in which the traffic demand is described as log-normaldistributed variable and the link capacity is assumed to follow a truncated normaldistribution. A Monte Carlo simulation-based genetic algorithm is provided to solvethe proposed network design model and numerical study on a small example roadnetwork is carried out with the algorithm.