EURO Journal on Transportation and Logistics最新文献

Providing punctual, reliable and performant services to customers is one main goal of railway network operators. The railway scheduling problem is to determine, ahead of time (timetabling), a plan describing the timing of the operations in a railway network, or updating such plan during operations (rescheduling). By optimization and automation, it is possible to operate more trains on the network, closer to the infrastructure capacity. Especially when the scale and complexity of the scheduling problem is increasing, for large-scale networks and multiple interconnected problems, this is of great value for network operators. When planning or adjusting railway operations becomes increasingly complex, modern scheduling algorithms can bring significant performance and economic benefits. In this survey we review approaches in the state of the art for the problems of railway scheduling. We show how the many different approaches of decomposition proposed in the literature of railway scheduling can be categorized into two general principles. We study different solution methods and identify a list of open topics for dealing with large-scale problems for future research.

Machine learning has shown great potential in various domains, but its appearance in inventory control optimization settings remains rather limited. We propose a novel inventory cost minimization framework that exploits advanced decision-tree based models to approximate inventory performance at an item level, considering demand patterns and key replenishment policy parameters as input. The suggested approach enables data-driven approximations that are faster to perform compared to standard inventory simulations, while being flexible in terms of the methods used for forecasting demand or estimating inventory level, lost sales, and number of orders, among others. Moreover, such approximations can be based on knowledge extracted from different sets of items than the ones being optimized, thus providing more accurate proposals in cases where historical data are scarce or highly affected by stock-outs. The framework was evaluated using part of the M5 competition’s data. Our results suggest that the proposed framework, and especially its transfer learning variant, can result in significant improvements, both in terms of total inventory cost and realized service level.

We study a setting in which a company not only has a fleet of capacitated vehicles and drivers available to make deliveries but may also use the services of occasional drivers (ODs) willing to make deliveries using their own vehicles in return for a small fee. Under such a business model, a.k.a crowdshipping, the company seeks to make all the deliveries at the minimum total cost, i.e., the cost associated with their vehicles plus the compensation paid to the ODs.

We consider a stochastic and dynamic last-mile delivery environment in which customer delivery orders, as well as ODs available for deliveries, arrive randomly throughout the day, within fixed time windows.

We present a novel deep reinforcement learning (DRL) approach to the problem that can deal with large problem instances. We formulate the action selection problem as a mixed-integer optimization program.

The DRL approach is compared against other optimization under uncertainty approaches, namely, sample-average approximation (SAA) and distributionally robust optimization (DRO). The results show the effectiveness of the DRL approach by examining out-of-sample performance.

Events that cause disruptions to air company flight schedules transporting passengers and cargo are commonplace in practice, and their operational and financial consequences are often relevant. The aircraft recovery problem (ARP) consists of recovering the flight schedules lost due to such events, determining new flight departure times and possible flight cancellations, as well as revising routes for different aircraft. In this study, the main characteristics considered in ARP papers to make the problem as realistic as possible are identified and a systematic literature review is carried out based on seminal studies of the problem in the 1980s. The literature papers are reviewed in terms of the ARP variants studied, the objectives chosen to be optimized and the practical constraints most often considered in real applications, as well as the networks and mathematical formulations to represent the problem and the solution approaches used to deal with it. The aim is to present an up-to-date review of the state-of-the-art of the ARP and to identify possible literature gaps and interesting opportunities for future research on this problem. It is noteworthy that, compared to the simplicity of the first ARP studies, the studies of the following decades tended to involve more complex variants of the problem as a way to become more adherent to current practical environments, requiring more elaborate formulations and solution methods to be properly addressed.

In the challenging environment of attended home deliveries, pricing delivery options can play a crucial role to ensure profitability and service quality of retailers. To differentiate between standard and premium delivery options, many retailers include time windows of various lengths and fees within their offer sets. Even though customers prefer short delivery time windows, longer time windows can help maintaining flexibility and profitability for the retailer. We classify pricing strategies along two dimensions: static versus dynamic price setting and whether an offer set can include one or multiple price points. For static pricing, we implement price configurations that reflect current business practice. For the dynamic pricing, we adapt routing mechanisms that consider the flexibility within the underlying route plan during the booking process and set delivery fees accordingly. To evaluate the pricing strategies under plausibly realistic conditions, we model customer behavior through a nested logit model. This model represents customer choice as sequential decisions between premium and standard time windows. We perform a computational study considering realistic travel and demand data to investigate the effectiveness of static and dynamic time window pricing. Finally, we offer managerial insights and an outlook into applying strategic analysis to decide on price setting strategies.

Collaborative delivery employing drones in last-mile delivery has been an extensively studied topic in recent years. In this paper, it is studied Truck–Drone Delivery Problems (TDDPs) in which a traditional delivery truck is gathered with a drone to cut delivery times and costs. The vehicles work together in a hybrid operation involving one drone launching from a larger vehicle that operates as a mobile depot and a recharging platform. The drone launches from the truck with a single package to deliver to a customer. Each drone must return to the truck to recharge batteries, pick up another package, and launch again to a new customer location. This work proposes a novel Mixed Integer Programming (MIP) formulation and a heuristic approach to address the problem. The proposed MIP formulation yields better linear relaxation bounds than previously proposed formulations for all instances, and was capable of optimally solving several unsolved instances from the literature. A hybrid heuristic based on the General Variable Neighborhood Search metaheuristic combining Tabu Search concepts is employed to obtain high-quality solutions for large-size instances. The efficiency of the algorithm was evaluated on 1415 benchmark instances from the literature, and over 80% of the best known solutions were improved.

In the context of on-demand ridehailing, we propose a heuristic matching algorithm where a passenger can share their ride with one more passenger while experiencing a high-quality service with a minimal increase in travel time. To evaluate the performance, we implemented the algorithm in a traffic microsimulator and compared it with a ride-matching algorithm developed by Simonetto et al. (2019) at IBM. Moreover, to enhance efficiency and reduce computational time, we proposed a decentralized version that is based on vehicle-to-infrastructure (V2I) and infrastructure-to-infrastructure (I2I) communication. Application on the downtown Toronto road network demonstrated that the service rate in the centralized version improved by 24%, compared to the IBM algorithm. The decentralized version demonstrated a 25.53 times speedup and the service rate improved by 19%, compared to the IBM algorithm. Furthermore, a sensitivity analysis was conducted over both centralized and decentralized versions to address how different variables and parameters can affect the system’s performance.

This investigation combines the aircraft maintenance routing problem (AMRP) with the aircraft maintenance facility location problem as they are interdependent. Unlike the classical fixed charge location model, aircraft maintenance facility location is not based on static customer demands but on how aircraft move around the airline network while undergoing periodic maintenance checks. Two important factors relevant to this analysis are: (a) aircraft overnight stays at maintenance facilities and airports away from their home bases and (b) the location and cost of maintenance facility construction. There are two possible scenarios concerning the status of maintenance facilities. The first is a tabula rasa case – for a start-up airline with no existing maintenance facilities (AMFLP). The second is an operating airline case with prepositioned maintenance facilities, which we denote as a maintenance facility location update problem (AMFUP). We formulate both as binary integer multicommodity flow problems whose aim is to find the cost-minimizing number, size, and location of the maintenance facilities. Experiments with flight schedules from two regional airlines indicate that total annualized costs are convex in the number of maintenance airports. Though new facility costs tend to dominate, costs associated with noncyclic routes become increasingly important as the number of maintenance airports decreases. Key contributions of this paper are that it (a) demonstrates the superiority of the integration of the aircraft maintenance routing and aircraft maintenance facility location problems, (b) provides a formulation and solution methodology dealing with this significant but under-researched problem, and (c) presents extensive computational experiments, cost and sensitivity analyses for two real life aircraft flight schedules.

Mobility-on-demand services continue to grow in popularity and could provide a cheap and resource-saving alternative to private vehicles. However, to be truly attractive to the general public, these services must be thoroughly optimized. In this paper, we consider a ride-hailing problem where available vehicles have to be assigned to dynamically arising customer requests and, furthermore, vacant vehicles have to be repositioned to other parts of the service area to balance supply and demand. We propose a novel repositioning strategy based on dynamically created, overlapping zones that addresses identified weaknesses of previous repositioning approaches. While most other ride-hailing studies only consider one specific setting for which a suitable ride-hailing strategy is developed, we further analyze which design decisions in the context of assignment and repositioning work best under different given problem characteristics. Our results show that the proposed repositioning approach outperforms the benchmark approaches in most of the relevant settings, independent of the underlying objective function. Additionally, we show that, especially for low-utilized fleets, the simple nearest-vehicle assignment strategy outperforms matching-based assignment approaches in many settings. The insights gained are analyzed and thoroughly discussed.

Due to the uncertain nature of the traffic system, it is not trivial for delivery companies to reliably satisfy customers’ time windows. To guarantee the reliability of the pickup and delivery service under stochastic and time-dependent travel times, we consider a pickup and delivery problem with hard time windows considering stochastic and time-dependent travel times. We propose a chance-constrained model where the operational cost and the service’s reliability are considered. To quantify the service reliability, every node is associated with a desired node service level, and there exists a global service level, both measured by success probabilities. We present an estimation method for arrival times and success probabilities under stochastic travel and service times. We propose an exact solution approach based on a branch-price-and-cut framework, where a labeling algorithm generates columns. Computational experiments are conducted to assess the effectiveness of the solution framework, and Monte Carlo simulations are used to show that the proposed method can generate routes that satisfy both node and global service levels.