Transportation Research Part B-Methodological最新文献

The proportionality condition is a standard approach to dealing with the non-uniqueness issue in the user equilibrium (UE) traffic assignment problems (TAP). Although the proportionality condition can reduce the degree of arbitrariness, it remains unclear how much arbitrariness remains and whether it can meaningfully affect model outcomes and relevant decisions that depend on them. The answers to these questions are impeded by the lack of an efficient algorithm that can find the exact maximum entropy UE path flow solution for networks of practical size. In this paper, we fill this gap by developing a high-performance augmented Lagrangian algorithm that effectively exploits the special problem structure. Our numerical results reveal that there are a considerable number of links with non-negligible arbitrariness in the solution generated by the proportionality condition, and that this problem becomes worse if the level of congestion increases in the network. Since about a decade ago, many practitioners have relied on state-of-the-art traffic assignment tools based on the proportionality condition to perform select link analysis, among other applications. The results reported herein are a reminder that their toolbox may need reevaluation and perhaps an upgrade.

The refueling station location problem with routing considers vehicles’ ranges and drivers’ preferences about their routes to improve the alternative fuel station infrastructure. Comprehensive planning is necessary for developing a mature infrastructure to overcome budgetary constraints and spatial limitations. Hence, adopting a multi-period planning approach becomes crucial when taking into account the evolving demand for alternative fuel vehicles over time. The evolution of demand can be dependent on exogenous and endogenous factors. Although it is typical to account for exogenous demand growth in multi-period planning, a few studies also take into account an endogenous factor which is the refueling opportunity of drivers on their paths. In this study, in addition to the refueling opportunities, we consider the proximity of each individual station to the flow-based demands. We draw attention to the significance of considering the effects of individual station locations on demand evolution, as these strategic locations can play an important role in reducing the drivers’ range anxiety and increasing their acceptance of the technology. Hence, we introduce a multi-period alternative fuel refueling station location problem with routing under different vehicle flow evolution dynamics that employ various weights for the factors where the natural growth rate is exogenous and the decisions of station locations and flow coverage are endogenous to the problem. We propose three mixed integer linear programming formulations for different evolution dynamics. We carry out computational experiments on the real road networks of Belgium, California, and Turkey and present our findings on the performances of the proposed mathematical models and the gains that can be obtained by considering multi-period planning and incorporating the effects of decisions into the vehicle flow evolution.

This paper addresses the ownership rationing issues of gasoline vehicles (GV) and electric vehicles (EV) in a given time horizon. A state-owned vehicle manufacturer is assumed to be the producer of these vehicles. Two typical rationing schemes for the GV and EV ownership that have been implemented in some Chinese megacities, namely a simultaneous lottery scheme for both GV and EV and a hybrid scheme consisting of lottery for GV and first-come-first-served (FCFS) for EV, are investigated. Multi-period social cost minimization models are proposed for determining the optimal auto quota (i.e., the number of vehicles to be produced) and the optimal frequency of ownership allocation for each scheme in the given time horizon. In the proposed models, residents’ heterogeneity is considered in terms of their values of time (VOTs). The properties of the proposed models are analytically explored, and a comparison between the lottery mode and the FCFS mode is made. The results show that the lottery mode outperforms the FCFS mode in terms of the total social cost. The VOT threshold of the subsequent participants in auto rationing schemes tends to be higher than that of the preceding participants. The decision of EV subsidy should be cautiously made due to a tug of war between the resultant decreased pollutant emissions and increased traffic congestion. The implementation of the auto ownership rationing schemes can cause inequity issue in terms of the changes of travel costs of different-income residents. A Pareto-improvement strategy is presented to balance equity and efficiency of the rationing schemes.

The railway line planning problem (LPP) plays a crucial role in determining the quality of services provided to passengers, as well as operation costs borne by railway companies. In periodic railway LPPs, it is common to consider passenger transfers between train lines to realize a general passenger travel cost setting in the railway system. While detecting passenger transfers requires incorporating passenger routing into mathematical formulations, thereby significantly complicating the problem. Studies on transfer-included LPPs are generally based on the Change&Go network that is constructed based on a pre-given line pool, which however is usually non-exhaustive due to computational intractability. To efficiently include passenger transfers in large-scale railway LPPs, this paper proposes a novel extended direct-service network representation of LPP, where lines are dynamically generated within the optimization process, and part of passenger transfers between lines can be precisely captured without the need for explicit modeling of passengers’ distribution on specific lines. A two-phase solution approach based on the representation is designed. The first phase formulates LPP with part of transfers as a path-based service network design model, solved using a branch-price-and-cut algorithm. The second phase conducts a neighborhood search around the first-phase solution to seek better ones when considering all passenger transfers. Numerical results showcase the good performance of the two-phase solution approach. It delivers optimal solutions in 18 out of 24 test instances for a small case and achieves optimality gaps within 2.85% across all small instances. The large case study of China’s Shandong high-speed railway network whose line pool size reaches millions demonstrates the scalability of the approach and its advantage over the traditional Change&Go method with partial line pools and an exact model developed in the paper.

This study addresses the sequencing problem in ship operations, with the aim of minimizing the quay crane operation time in a ship bay. By analysing realistic trajectories of the quay crane spreader in various situations, a mathematical model for expressing the transport times of the spreader is established and a method for determining the shortest-time trajectory for a given bay configuration is proposed. Positioning times of the spreader are analysed for various cases of ship operations. Based on the analysis of transport and positioning times, an exact algorithm to obtain the optimal sequence of the discharging and loading operations and a common rule to be applied in practice are developed. An extended discussion on sequencing reshuffling operations is presented. Numerical experiments with the data from a real-world port, show that this approach can determine a ship operation sequence with the shortest total operation time.

This paper proposes a weibit-based equilibrium choice model for investigating the effect of the emerging shared parking services, which have recently received increasing interest, on combined destination location and parking choice behaviors. The model considers the features of shared parking services, including the avoidance of cruising to search for a parking space and limited supply of shared parking spaces. The spatial similarity issues of destination and parking choices, i.e., the correlations among spatially adjacent destination locations and the parking spaces around them, are separately considered through the advanced spatially correlated weibit (SCW) model and parking-size weibit (PSW) model, respectively. Subsequently, an equivalent mathematical programming (MP) formulation of the equilibrium SCW-PSW model is developed, which guarantees the existence and uniqueness of the solutions. Based on the MP formulation, a partial linearization algorithm embedded with the iterative balancing direction-finding scheme and self-regulated averaging line search scheme is developed to solve the proposed equilibrium model. Numerical examples are presented to illustrate the properties of the proposed model and its applicability to analyzing planning scenarios with different shared parking supplies.

Lane changing, recognised as one of the most intricate manoeuvres in road traffic, has attracted extensive scholarly interest. To date, the concept of lane change has been categorised into two distinct classifications, namely mandatory and discretionary. Mandatory lane changes (MLCs) are often regarded as absolute, implying that the possibility of not executing the lane change is frequently disregarded. This paper questions this widely accepted proposition by evaluating the costs of neglecting an MLC. Specifically, we examine the costs associated with not making MLCs for exiting freeways, effectively quantifying the cost of missing such exits. The core of this study involves a dual approach comprising an analytical model for the costs of missing exits alongside an empirical analysis of two GPS datasets from the Minneapolis - St. Paul metropolitan area. The performance of the analytical model is validated by cross-referencing it against exit-missing costs from the top 50 metropolitan areas in the US. Regarding the empirical study, it was found that while both time and distance costs are associated with missing exits, the magnitudes of these costs are not substantial. The results obtained in this study offer novel insights into the nature of MLC, and we argue that future models should consist of discretionary (DLC), mandatory (MLC), and expedient (ELC) lane changes. Moreover, the analytical model developed in this study can be integrated into the trade-off function of an ELC model, enabling drivers to bypass their intended exit when needed.

This paper introduces the consistent vehicle routing problem with stochastic customers and demands. We consider driver consistency as customer-driver assignments that remain fixed when the realizations of the random variables are observed. We study the problem in a two-stage scenario-based stochastic programming framework. In the first stage, customers are assigned to drivers, while in the second stage, customers are selected and delivery routes are designed for each of the scenarios. We assume that the realization of the random variables becomes known before the vehicles depart from the depot. The routes are then optimized according to the observed customers and their demands. The first-stage driver-customer assignments can violate the consistency requirement, which is modeled as a desired maximum number of drivers assigned to each customer. This is modeled as a soft constraint with a penalty in the objective function. It is hence possible to assign multiple drivers to a specific customer in the first stage. In the second stage, a customer can only be visited by one of the preassigned drivers. Our problem, therefore, consists in finding assignments that minimize the consistency violation penalties, the expected routing costs, and the penalties for unserved customers when the uncertain parameters are revealed. We present a mathematical formulation and a sample average approximation (SAA) approach for the problem. We introduce a branch-and-cut and a Benders decomposition method to solve the sample problems in our SAA algorithm. Computational experiments show that SAA allows finding good-quality solutions for instances with large sets of scenarios. We also analyze the cost-consistency trade-offs and the impact of the uncertainty on the problem. In particular, we observe that consistency can be promoted through a flexible approach that does not compromise excessively on other operational metrics. Furthermore, we analyze the impact of not considering the problem uncertainties during the planning stage.

The Capacitated Team Orienteering Problem (CTOP) is a challenging combinatorial optimization problem, wherein a fleet of vehicles traverses multiple locations, each with distinct prizes, demand weights, and service times. The primary objective is to determine optimal routes for the vehicles that collectively accumulate the highest total prize within capacity and time constraints. The CTOP finds applications across various domains such as disaster response, maintenance, marketing, tourism, and surveillance, where coordinated teams are required to efficiently explore and gather prizes from different sites. The complexity of this problem is further compounded by uncertainties in predicting specific attributes of each location, making it hard to plan routes accurately in advance. In numerous scenarios in practice, subjective predictions for these parameters may exist, but their precise values remain unknown until a location is visited by one of the vehicles. Given the unpredictable nature of these parameters, there is a pressing need for innovative online optimization strategies that can adapt to new information, ensuring the strategic allocation of resources and route planning within set constraints. To address this challenging online optimization problem, we offer a detailed analysis through the lens of theoretical and empirical competitive ratios. We derive an exact tight upper bound on the competitive ratio of online algorithms, and we introduce three novel online algorithms, with two of them achieving optimal competitive ratios. The third algorithm is a polynomial time approximation-based online algorithm with a competitive ratio of $\frac{1}{3.53}$ times the tight upper bound. To evaluate our algorithms, we measure their empirical competitive ratios on randomly generated instances as well as instances from the literature. Our empirical analysis demonstrates the effectiveness of our solutions across a diverse range of simulation scenarios.

This article proposes a method to estimate disaggregated discrete choice models with errors in the variables. The objective is to estimate the discrete choice models' coefficients to compute the value of time and use it for cost-benefit analysis in transportation planning. The method is general, as it only requires a validation sample to estimate the conditional density of the error-free variables given the mismeasured variables. More specifically, we assume that the attributes of the chosen alternative are reported without error in revealed preference surveys, and we use this information as the validation sample. The mismeasured variables may be spatially aggregate service levels from mobility surveys or transportation network models. Monte Carlo simulations show that the proposed method substantially reduces bias in parameters. We validate the technique with two real data sets from Santiago, Chile.