EURO Journal on Transportation and Logistics最新文献

The inventory routing problem (IRP) is an optimisation problem that integrates transportation and inventory management decisions. When subjected to unexpected events such as demand changes, the a posteriori approach consists in re-optimising including the data related to this event; the challenge is to ensure that the obtained solution does not deviate too much from the original one, lest that creates important organisational issues. Therefore, a stability metric is needed when re-optimising IRP models. This article proposes a panel of stability metrics adapted from the scheduling, routing and inventory management literature to fit the requirements of the IRP and proposes mathematical formulations for the most relevant ones. A framework of comparison is proposed to validate and compare these metrics over a benchmark of 3000 instances generated from the literature. A strong correlation between the metrics is observed. Moreover, the results show that ensuring the stability of the re-optimised solutions has little impact on the initial objective, the total cost.

Overloaded axles not only lead to increased erosion on the road surface, but also to an increased braking distance and more serious accidents due to higher impact energy. Therefore, the load on axles should be already considered during the planning phase and thus before loading the truck in order to prevent overloading. Hereby, a detailed 2D or 3D planning of the vehicle cargo space is required. We model the Axle Weight Constraint for trucks with and without trailers based on the Science of Statics and provide flexible formulas for different axle configurations of trucks. We include the Axle Weight Constraint into the combined Vehicle Routing and Container Loading Problem (“2L-CVRP” and “3L-CVRP”). A hybrid heuristic approach is used where an outer Adaptive Large Neighbourhood Search tackles the routing problem and an inner Deepest-Bottom-Left-Fill algorithm solves the packing problem. Moreover, to ensure feasibility, we show that the Axle Weight Constraint must be checked after each placement of an item. The impact of the Axle Weight Constraint is also evaluated.

We consider the problem of assigning flights to baggage belts in the baggage reclaim area of an airport. The problem is originated by a real-life application in Copenhagen airport. The objective is to construct a robust schedule taking passenger and airline preferences into account. We consider a number of business and fairness constraints, avoiding congestion, and ensuring a good passenger flow. Robustness of the solutions is achieved by matching the delivery time with the expected arrival time of passengers, and by adding sufficient buffer time between two flights scheduled on the same belt. We denote this problem as the Baggage Belt Assignment Problem (BBAP). We first derive a general Integer Linear Programming (ILP) formulation for the problem. Then, we propose a Branch-and-Price (B&P) algorithm based on a reformulation of the ILP model tackled by Column Generation. Our approach relies on an effective dynamic programming algorithm for handling the pricing problems. We tested the proposed algorithm on a set of real-life data from Copenhagen airport as well as on a set of instances inspired by the real data. Our B&P scheme outperforms a commercial solver launched on the ILP formulation of the problem and is effective in delivering high quality solutions in limited computational times, making it possible to use the solution approach in daily operations in medium-sized and large airports.

Delivery options are at the heart of the generalized vehicle routing problem with time windows (GVRPTW) allowing that customer requests are shipped to alternative delivery locations which can also have different time windows. Recently, the vehicle routing problem with delivery options was introduced into the scientific literature. It extends the GVRPTW by capacities of shared locations and by specifying service-level constraints defined by the customers’ preferences for delivery options. The vehicle routing problem with delivery options also generalizes the vehicle routing problem with home roaming delivery locations and the vehicle routing problem with multiple time windows. For all these GVRPTW variants, we present a widely applicable matheuristic that relies on a large neighborhood search (LNS) employing several problem-tailored destruction operators. Most of the time, the LNS performs relatively small and fast moves, but when the solution has not been improved for many iterations, a larger destruction move is applied to arrive in a different region of the search space. Moreover, an adaptive layer of the LNS embeds two exact components: First, a set-partitioning formulation is used to combine previously found routes to new solutions. Second, the Balas-Simonetti neighborhood is adapted to further improve already good solutions. These new components are in the focus of our work and we perform an exhaustive computational study to evaluate four configurations of the new matheuristic on several benchmark instances of the above-mentioned variants. On all the benchmark sets, our matheuristic is competitive with the previous state-of-the-art methods. In summary, the four configurations provide 81 new best-known solutions.

The Vehicle Routing Problem (VRP) is one of the most studied topics in Operations Research. Among the numerous variants of the VRP, this research addresses the VRP with relaxed priority rules (VRP-RPR) in which customers are assigned to several priority groups and customers with the highest priorities typically need to be served before lower priority ones. Additional rules are used to control the trade-off between priority and cost efficiency. We propose a Mixed Integer Linear Programming (MILP) model to formulate the problem and to solve small-sized instances. A metaheuristic based on the Adaptive Large Neighborhood Search (ALNS) algorithm with problem-tailored components is then designed to handle the problem at larger scales. The experimental results demonstrate the performance of our proposed algorithm. Remarkably, it outperforms a metaheuristic recently proposed to solve the Clustered Traveling Saleman Problem with d-relaxed priority rule (CTSP-d), a special case of VRP-RPR, in both solution quality and computational time.

Service network design is an important optimization problem for intermodal freight transportation on a tactical level. It includes the decisions on choosing transportation modes and paths for commodities throughout the intermodal network. We present a stochastic service network design model with an integrated vehicle routing problem (SSND-VRP), which simultaneously covers transportation service choice and tour planning decisions for road transportation under consideration of uncertain transportation times. A sample average approximation approach is combined with an iterated local search in order to solve problem instances in a real-world case study for three intermodal road-rail networks in Central Europe. Results of the SSND-VRP are compared with its expected value model and a successive planning approach, demonstrating the possible cost reductions and the decrease in missed intermodal services that are achieved by the integrated stochastic model. In further parameter variation experiments we show that the attractiveness of rail transportation is highly sensitive to changes in intermodal costs, whereas the impact of delay reductions of the railway services is relatively low.

We present an approach to couple the resolution of Combinatorial Optimization problems with methods from Machine Learning. Specifically, our study is framed in the context where a reference discrete optimization problem is given and there exist data for many variations of such reference problem (historical or simulated) along with their optimal solution. Those variations can be originated by disruption but this is not necessarily the case. We study how one can exploit these to make predictions about an unseen new variation of the reference instance.

The methodology is composed by two steps. We demonstrate how a classifier can be built from these data to determine whether the solution to the reference problem still applies to a perturbed instance. In case the reference solution is only partially applicable, we build a regressor indicating the magnitude of the expected change, and conversely how much of it can be kept for the perturbed instance. This insight, derived from a priori information, is expressed via an additional constraint in the original mathematical programming formulation.

We present the methodology through an application to the classical facility location problem and we provide an empirical evaluation and discuss the benefits, drawbacks and perspectives of such an approach.

Although it cannot be used in a black-box manner, i.e., it has to be adapted to the specific application at hand, we believe that the approach developed here is general and explores a new perspective on the exploitation of past experience in Combinatorial Optimization.

The crew pairing problem (CPP) is generally modelled as a set partitioning problem where the flights have to be partitioned in pairings. A pairing is a sequence of flight legs separated by connection time and rest periods that starts and ends at the same base. Because of the extensive list of complex rules and regulations, determining whether a sequence of flights constitutes a feasible pairing can be quite difficult by itself, making CPP one of the hardest of the airline planning problems. In this paper, we first propose to improve the prototype Baseline solver of Desaulniers et al. (2020)2020) by adding dynamic control strategies to obtain an efficient solver for large-scale CPPs: Commercial-GENCOL-DCA. These solvers are designed to aggregate the flights covering constraints to reduce the size of the problem. Then, we use machine learning (ML) to produce clusters of flights having a high probability of being performed consecutively by the same crew. The solver combines several advanced Operations Research techniques to assemble and modify these clusters, when necessary, to produce a good solution. We show, on monthly CPPs with up to 50 ​000 flights, that Commercial-GENCOL-DCA with clusters produced by ML-based heuristics outperforms Baseline fed by initial clusters that are pairings of a solution obtained by rolling horizon with GENCOL. The reduction of solution cost averages between 6.8% and 8.52%, which is mainly due to the reduction in the cost of global constraints between 69.79% and 78.11%.