Solving the integrated inventory supply chain problems using meta-heuristic methods / Seyed Mohsen Mousavi

Nowadays, managing supply chain networks are really important for all companies especially those producing seasonal products. The novelties of this work are as follows. First, a novel multi-objective multi-item seasonal inventory control model was developed for known-deterministic variable demands where shortages in combination of backorder and lost sale were considered. Moreover, all unit discounts (AUD) for a number of products and incremental quantity discount (IQD) for some other items were considered. The main novelty of the first model is to minimize the required storage space for the first time in the literature in addition to minimizing the total inventory cost. While the weights of both objectives are considered to be fuzzy numbers due to their uncertainty, the model was formulated into a fuzzy multi-objective decision making (FMODM) framework called Fuzzy Weighted Sum Method (FWSM) and was shown to be a mixed-integer binary nonlinear programming type. In order to solve the model, a multi-objective particle swarm optimization (MOPSO) approach was applied. The efficiency of the algorithm was compared to a multi-objective genetic algorithm (MOGA) as well. The second novelty of the work is an extension of the proposed model, where a novel model of the integrated seasonal inventory-supply chain distributor–retailer network for a multiple-product location allocation problem in a planning horizon consisting of multiple periods was formulated. The distance between the distributors and retailers were assumed to be Euclidean and Square Euclidean. The retailers purchase the products from the distributors under both AUD and IQD policies. Furthermore, the products were delivered in packets of known size of items where the model was extended for both cases of with and without shortages. Besides, the distributors (vendors) stored the manufactured products in their own warehouses before delivering them to the retailers since the total warehouse spaces and the total available budget for purchasing the items from the distributors were limited. It was considered that the distributors manufacture the products under some production limitations. The aim of the problem was to find the optimal order quantities of the products purchased by the retailers from the distributors in different periods and determine the coordinates of the distributors’ locations to minimize the total inventory-supply chain cost. In fact, finding out the optimal order quantities of items in each period and the optimal locations of distributers among retailers are the main novelty of the second model proposed for the first time in the literature. As the mixed integer nonlinear model of the problem was complicated to solve using exact methods, several meta-heuristic algorithms were employed in to optimize the models under investigation. A Modified Particle Swarm Optimization (MPSO) algorithm, a Genetic Algorithm (GA), a modified fruit fly optimization algorithm (MFOA) and a simulated annealing (SA) algorithm were used to find the optimal solution. A design of experiment approach i.e. Taguchi was used to optimize the algorithms parameters. While there was no benchmark in the literature, some numerical examples were generated to show the performance of the algorithms for both Euclidean and Square Euclidean distances while some case studies were also considered.