Search

In the UAV-based urban delivery setting, the battery capacity serves as one of the key factors affecting the delivery range of drones. For a typical multi-rotor unmanned aerial vehicle (UAV), the maximum effective working range of a lithium battery for a single flight leg is approximately 20 km. In practice, the lithium battery is replaced each time the UAV completes a delivery, even though the remaining power of the replaced lithium battery is still capable of supporting the following flight legs. Therefore, the one-order-one-swapping strategy significantly brings down the battery utilisation as well as the efficiency of the delivery system. In the present paper, the UAV-based delivery problem with battery swapping decision is considered, where multiple en route pickup and deliveries are allowed. To obtain optimal decisions in terms of flight route, time and location of each battery swapping, a branch-and-price-and-cut algorithm is introduced. First we formulate this problem as a mixed integer linear programming (MILP) model, and then decompose it into a master problem and a pricing sub-problem, where the master problem is considered as a set covering model, and the sub-problem as an elementary shortest path problem with resource constraints (ESPPRC). To speed up the process of column generation, efficient heuristics are also integrated to provide basic feasible solutions. In the pricing sub-problem, a combination of heuristic algorithm and dynamic programming is implemented to improve the computational efficiency. In the numerical experiments, the proposed algorithm is tested on various scales of UAV-based urban delivery instances, the performance of which is analysed and compared with the off-the-shelf commercial solver. [ABSTRACT FROM AUTHOR]

The purpose of this article is to investigate a practical scheduling problem in which a group of elective surgical cases are scheduled over time, while considering their unpredictable durations and potential delays in the sterilisation of surgical instruments. The primary objectives were to schedule the maximum number of surgeries and decrease overtime for the surgical staff, as well as limit the number of instruments requiring emergency sterilisation. The study was conducted in collaboration with the University Hospital of Angers in France, which also contributed historical data for the experiments. We propose two robust mixed integer linear programming models, which are then solved iteratively through a rolling horizon approach, in which the objective functions are taken into account in lexicographic order. Experiments on randomly generated instances indicated which of the two approaches had better performance. Comparison of the results for a real-world scenario involving actual planning at the hospital indicated a greater than 69% decrease in overtime, and a minimum of 92% fewer stressful situations in the sterilising unit. [ABSTRACT FROM AUTHOR]

This paper highlights the tight relationship between the picking and packing processes in warehouse management and the need to consider them as an integrated problem. The study describes and models this integrated problem as a mixed-integer programming model, to optimise overall labour costs by determining the assignment of the subsets of orders, i.e. batches, for picking and packing. To address the issue of model complexity, the paper presents a statistical-based framework for generating approximate models and selecting the optimal one through examination. Based on the examination results, a pair-swapping heuristic is additionally proposed to be combined as a hybrid algorithm. Numerical experiments based on a real-world case demonstrate the effectiveness of the framework-proposed and selected hybrid algorithm by comparison with other framework-proposed approximate models, a solver, and existing heuristics. Our findings indicate that the combined usage of integrated picking and packing processes planning and the hybrid algorithm proposed and selected within the statistical-based framework can effectively reduce the cost of warehouse management. [ABSTRACT FROM AUTHOR]

For over five decades, researchers have presented various assembly line problems. Recently, assembly lines with multiple workers at each workstation have become very common in the literature. These lines are often found in the manufacturing of large vehicles, where workers at a workstation may perform their assigned tasks at the same time. Most research on multi-manned assembly lines focuses on balancing tasks and workers among workstations and scheduling tasks for workers. This study, however, concentrates on assigning tasks to workers already assigned to a specific workstation, rather than balancing the entire line. The problem was identified through an industrial case study at a large vehicle manufacturing company. The study presents two methods, one using mixed integer linear programming and the other using constraint programming, to minimise the number of workers required on a multi-manned assembly line with sequence-dependent setup times. The results of the computational experiments indicate that the constraint programming method performs better than the mixed integer linear programming method on several modified benchmark instances from the literature. The constraint programming model is also tested on the real-world scenario of our industrial case study and leads to significant improvements in the productivity of the workstations. [ABSTRACT FROM AUTHOR]

In this paper, stochastic optimisation of CVaR is applied to maintain risk-averse viability and improve resilience of a supply chain under propagated disruptions. In order to establish the risk-averse boundaries on supply chain viability space, two stochastic optimisation models are developed with the two conflicting objectives: minimisation of Conditional Cost-at-Risk and maximisation of Conditional Service-at-Risk. Then, the risk-averse viable production trajectory between the two boundaries is selected using a stochastic mixed integer quadratic programming model. The proposed approach is applied to maintain the supply chain viability in the smartphone manufacturing and the results of computational experiments are provided. The findings indicate that when the decision-making is more risk-aversive, the size of the viability space between the two boundaries is greater. As a result, more room is available for selecting viable production trajectories under severe disruptions. Moreover, the larger is viability space, the higher is both worst-case and average resilience of the supply chain. Risk-neutral, single-objective decision-making may reduce the supply chain viability. A single-objective supply chain optimisation which moves production to the corresponding boundary of the viability space, should not be applied under severe disruption risks to avoid greater losses. [ABSTRACT FROM AUTHOR]

This work presents four mixed integer linear programming (MILP) models for the assembly flow shop problem in order to minimize the makespan. This production environment has two stages: production and assembly. The first stage consists of different machines designed to manufacture parts of a product. The second stage is intended for a final assembly. The performance measure considered is highly essential for industries from different segments, as it focuses on the best use of the time available for production. Statistical analysis with different tools was used to assess the performance and efficiency of mathematical models, emphasizing the analysis of performance profiles. Results showed that mathematical models are efficient, and the position-based model presented the best results for small and large instances during computational experimentation. All mathematical models can be used as direct tools in decision-making for the production sequencing problem in the approached environment. [ABSTRACT FROM AUTHOR]