With the surge in e-commerce, last-mile delivery (LMD) faces challenges in meeting the dynamic arrival of large-scale customer demand. Existing research considers traditional, static, single or two-echelon, and inflexible vehicle routing-based dispatch strategies, yielding inefficiencies while the sustainability impacts remain unclear. This paper proposes a novel tri-objective mathematical model for dispatch-steerability in sustainable LMD systems by integrating insourced and crowdsourced transport modes with heterogeneous capacities. Deliveries originate from main dispatch points (MDPs) or dynamically generated alternative dispatch points (ADPs), adapting to stakeholder, driver, and customer preferences via negotiable timing through temporary rejections or incentives. The model optimizes economic efficiency (minimizing routing, rejection, and negotiation costs), environmental impact (reducing underutilization of transport modes), and social responsibility (balancing driver workload and just-in-time deliveries). A data-driven density-based spatial clustering of applications with noise (DBSCAN) algorithm identifies ADP locations, enhancing the adaptability of the decisions. Scalability is achieved via a rolling horizon strategy, an improved augmented epsilon-constraint method, and multi-objective meta-heuristics, including a genetic algorithm (GA) and simulated annealing (SA) featuring an archive and grid mechanism with a tailored representation scheme. Results show a 200 % increase in Pareto-optimal solutions as demand rises, and that a dispatch steerable LMD system with ADPs offers 150 % more diverse delivery options than MDP-only systems. Pareto surface analysis underscores LMD adaptation trends and the need for multi-criteria decision-making in sustainable LMD systems.

In today’s competitive business environment, organizations must decide how to handle the processing of their logistics equipment economically. One of the vital logistical concerns is distribution planning that is especially crucial depending on the facilities and goods being used. When it comes to perishable goods, this problem assumes double the significance. The position of the warehouse and the route of the vehicles make up the distribution planning problem. These two problems are considered concurrently and solved in the location-routing mathematical model. This paper aims to provide a production and distribution strategy to serve clients and consumers better. This research attempts to produce as efficiently as possible while providing prompt customer service, which is crucial in today’s corporate environment. This study uses three-level supply chains for perishable goods to create a supply chain network that minimizes costs. In this case, time-dependent demands refer to requests that may be made when the vehicle will arrive. Places and routes in this area are designed to meet all needs. In general, it is desirable to have factories and distribution centers in known locations, know the service’s opening and closing hours, and know how to manage the flow of materials and goods as they are stored in distribution centers and for retailers (clients). Additionally, it is desirable to route vehicle that connects the various levels of the supply chain and ensures that vehicles travel on schedule overall. First, the supply chain model represented as non-linear programming is transformed into linear programming to solve it using the CPLEX solver of GAMS commercial software and the Lagrangian relaxation (LR) method. Then, this model is verified using numerical examples and related parameters to see how it impacts the variables and the objective function’s result. The results show the capability of the LR method.

Hubs act as intermediate points for the transfer of materials in the transportation system. In this study, a novel p-mobile hub location–allocation problem is developed. Hub facilities can be transferred to other hubs for the next period. Implementation of mobile hubs can reduce the costs of opening and closing the hubs, particularly in an environment with rapidly changing demands. On the other hand, the movement of facilities reduces lifespan and adds relevant costs. The depreciation cost and lifespan of hub facilities must be considered and the number of movements of the hub’s facilities must be assumed to be limited. Three objective functions are considered to minimize costs, noise pollutions, and the harassment caused by the establishment of a hub for people, a new objective that locates hubs in less populated areas. A multi-objective mixed-integer non-linear programming (MINLP) model is developed. To solve the proposed model, four meta-heuristic algorithms, namely multi-objective particle swarm optimization (MOPSO), a non-dominated sorting genetic algorithm (NSGA-II), a hybrid of k-medoids as a famous clustering algorithm and NSGA-II (KNSGA-II), and a hybrid of K-medoids and MOPSO (KMOPSO) are implemented. The results indicate that KNSGA-II is superior to other algorithms. Also, a case study in Iran is implemented and the related results are analyzed.

Abstract Here, a mixed‐integer linear programming model is developed to represent a transportation system of students traveling from/to a university campus. The concept of ridesharing is used and the mechanism of combinatorial auctions is incorporated within a routing‐based model. The mathematical model is based on the vehicle routing problem along with appropriate constraints accommodating features that express the auction clearing phase. A hybrid heuristic‐based optimization framework, that takes advantage of meta‐heuristic algorithms to improve an initial solution, is also developed to solve large‐sized instances of the problem. Three meta‐heuristics, namely particle swarm optimization, dragonfly algorithm, and imperialist competitive algorithm, are implemented in the proposed framework, whose performances are assessed and compared. Moreover, two improvement heuristic procedures that attempt to improve the outcomes of the foregoing meta‐heuristics are proposed and compared as well.

Abstract This paper addresses the multiobjective, multiproducts and multiperiod closed‐loop supply chain network design with uncertain parameters, whose aim is to incorporate the financial flow as the cash flow and debts' constraints and labor employment under fuzzy uncertainty. The objectives of the proposed mathematical model are to maximize the increase in cash flow, maximize the total created jobs in the supply chain, and maximize the reliability of consumed raw materials. To encounter the fuzzy uncertainty in this model, a possibilistic programming approach is used. To solve large‐sized problems, the multiobjective simulated annealing algorithm, multiobjective gray wolf optimization, and multiobjective invasive weed optimization are proposed and developed. The numerical results demonstrate that these algorithms solve the problems within about 1% of the required solving time for the augmented ε‐constraint and have similar performance and even better in some cases. The multiobjective simulated annealing algorithm with a weak performance takes less time than the other two algorithms. The multiobjective gray wolf optimization and multiobjective invasive weed optimization algorithms are superior based on the multiobjective performance indices.

Traditional production management approaches have paid less attention to integrate supply chain functions. Two main functions in supply chain management (i.e., production planning and distribution) are mutually dependent and important in today's real world applications. More of the time, they should be used simultaneously in an integrated manner to minimize costs in the whole chain. In this paper, a coordinated air transportation and production scheduling problem is studied to minimize the total cost of a supply chain. Additionally, time windows for the due date are considered; however, the due date for delivery time has been considered in the literature. Since this problem is NP-hard, two well-known meta-heuristics (i.e., genetic algorithm and simulated annealing) are used. Some new procedures and operators are developed in these algorithms. To evaluate the performance of the algorithms, different problem sizes are used and the results are compared. Finally, the impacts of increasing the problem size are studied.

This paper considers an open shop scheduling problem that minimizes bi-objectives, namely makespan and total tardiness. This problem, due to its complexity, is ranked in the class of NP-hard problems. In this case, traditional approaches cannot reach to an optimal solution in a reasonable time. Thus, we propose an efficient method based on multi-objective simulated annealing and ant colony optimization, in order to solve the given problem. Finally, we compare our computational results with a well-known multi-objective genetic algorithm, namely NSGA II. Comparisons are made in single objective case too. The outputs show the encouraging results in the form of the solution quality.

This paper presents a new mathematical model of mixed-model assembly lines (MMAL) to find the best sequences of product models in a just-in-time (JIT) production system. The objective is to minimize three criteria with their importance weights: (i) total utility work cost, (ii) total production rate variation cost, and (iii) total setup cost. Due to its NP-hardness, a memetic algorithm (MA) is proposed and its performance is compared with the Lingo 6 software. To validate the proposed model, a number of test problems are solved to verify the good ability of the proposed MA in terms of the solution quality and computational time. The results reveal that the MA finds promising results, especially in the case of large-sized problems.

This paper presents a special design of an evolutionary algorithm based on a genetic algorithm (GA) for solving a generalized model of a single-item resource-constrained aggregate production planning (APP) problem. We linearize a linear mixed-integer model of APP. Due to its NP-hardness one, we develop the proposed GA with effective operators for solving the above model with an integer representation. This model is optimally solved and validated in small-sized problems by an optimization software package, in which the obtained results are compared with the GA results. The results imply the efficiency of the proposed GA achieving to near optimal solutions within a reasonably computational time.