A three-echelon Multi-commodity supply chain including manufacturers, distribution centers (DCs) and customers is considered. Customers may be partially or fully covered by the DCs which should be opened in some candidate locations. A two-objective model is developed to find the locations of DCs and the flows of commodities in the whole supply chain considering a pre-determined number of DCs. The first objective function minimizes the total operation costs including transportation, inventory holding, production and site opening costs while the second objective maximizes the customers’,partial coverages. Since the presented problem is NP-hard in nature, three metaheuristic algorithms of NSGA-II, NRGA and MOPSO are developed to find the Pareto-optimal solutions and are compared using some standard criteria for Multi-objective algorithms. Numerical examples are designed to assess the performance of the model and the developed metaheuristic algorithms. Considering different criteria for comparing the algorithms, the superiority of some algorithms against others are reported.

This study presents a Multimodal hub location problem which has the capability to split commodities by limited-capacity hubs and transportation systems, based on the assumption that demands are stochastic for Multi-commodity network flows. In the real world cases, demands are random over the planning horizon and those which are partially fulfilled, are lost. Thus, the present study handles demands using a discrete chance constraint programming to make the model one step closer to the reality. On the other hand, commodity splitting makes it possible for the remaining portion of commodity flow to be transported by another hub or transportation system in such a way that demands are completely fulfilled as much as possible. The problem decides on the optimum location of hubs, allocates spokes to established hubs efficiently, adopts and combines transportation systems and then makes a right decision as to whether transportation infrastructure to be built at points lacking a suitable transportation infrastructure and having the potential for infrastructure establishment. A Mixed Integer Linear Programming (MILP) model is formulated with the aim of cost minimization. Also, the proposed sensitivity analysis shows that, the discrete chance constraint programming is a good approximation of the continuous chance constraint programming when an uncertain parameter follows a normal distribution. The results indicate the higher accuracy and efficiency of the proposed model comparing with other models presented in the literature.

Recently, the difference in the most effective competencies is considered the main competitive factor in organizations. To this end, organizations seek to improve a number of their functional capabilities, expertise, and capacities to enhance their operational area. Therefore, when an organization focuses on the quality of its services or products, it is trying to improve maintainability to gain a competitive advantage. In this study, a closed-loop, Multi-objective, Multi-level, Multi-commodity, and Multi-period mathematical model for a supply chain with producer, and distributor components is presented to locate and allocate items. The presented model can control environmental, economic, and social factors along the chain. One of the most important and unique aspects of the current study is considering different scenarios in the closed-loop supply chain (CLSC) so that the quality of the produced and transported products is paid attention to according to perishability. In addition, to control environmental effects, the model can minimize total CO2 emissions. The problem is solved on small, medium, and large scales using Epsilon Constraint and NSGA-II methods. According to the obtained results, the flow according to the boom scenario is more than the stagnation scenario. Finally, according to the sensitivity analysis, the number of centers established increases with an increase in demand. The results show that the non-dominated sorting genetic algorithm (NSGA-II) model can predict the behavior of the model well in the long term. For this purpose, Mean ideal distance (MID) index, has been used for evaluation of calculation. the value of standard MID is equal to 6.56 that shows the model accuracy is adequate.

Planning the freight flow from the plants to the customer zones is one of the most challenging problems in the field of supply chain management. Because of many traffic regulations, oversize/overweight vehicles often are not permitted to enter city boundaries. Therefore, intermediate facilities (city distribution centers) play a very important role in distribution networks. Accordingly, in this paper, transportation of goods from the plants to the customers is considered an integrated process containing two phases, namely, transportation from plant to distribution centers and distribution from city distribution centers to customers using small and environmentally-friendly vehicles. The Transportation Location Routing Problem (TLRP) studied can be considered as an extension of the two-echelon location routing problem. Minimizing the operational costs, and the workload balancing of the heterogeneous fleet in the distribution phase are considered as the two objective functions. A mixed integer programming (MIP) model, as well as two solution approaches, based on Multi-objective Particle Swarm Optimization Algorithm, and Non-dominated Sorting Genetic Algorithm, is presented for the problem. In order to illustrate the efficacy of the proposed methods, they have been implemented on test problems of different sizes. The results show the methods are able to produce efficient solutions in a reasonable amount of time.

In this paper, a new method is proposed for trip distribution forecasting in Multi-commodity transportation networks. In this new approach, real trip distributions at present are modeled by approximation functions for each commodity. In large-scale Multi-commodity transportation networks, these functions could be obtained by parts of network as training data, without losing generality of the algorithm. Of course, it will have an influence on the approximation. Then, a fuzzy model for trip distribution forecasting calibrates these functions. This model considers all adjustments simultaneously. In order to show the application of the new method, a numerical example is given and solved.