Hasan Rabiee; Farhad Etebari
Abstract
In this study, a location routing model has been considered for the distribution network of multiple perishable food products in a cold supply chain, where vehicles can refuel at filling stations. In this model, the fuel consumption is expected to vary based on the amount of load transported between ...
Read More
In this study, a location routing model has been considered for the distribution network of multiple perishable food products in a cold supply chain, where vehicles can refuel at filling stations. In this model, the fuel consumption is expected to vary based on the amount of load transported between nodes using a fleet that utilizes unconventional fuels. The objective of the study is to minimize the production of Carbon Dioxide, and therefore, the problem has been formulated as an integer linear programming model. To validate the model, several numerical examples were solved using GAMS software. The results indicate that, on average, fuel consumption decreases by 14 percent in this particular case. Due to the complexity of the problem, genetic simulated annealing algorithms were developed to solve real-sized instances of the problem, and their performance was also evaluated.IntroductionThe negative impacts of vehicle transportation on the environment are undeniable due to its effects on land and resource consumption. Therefore, efficient management of logistics activities and their optimization lead to providing differentiated services to customers, fast service delivery, reduced waiting times, damage reduction, customer support, and more. Considering the significant role of transportation in the generation and dissemination of environmental pollutants, as well as its irreversible impacts on the environment, the health of living beings, and ozone layer depletion, incorporating the concept of sustainability and considering environmental constraints alongside economic requirements in modeling and optimizing transportation and distribution networks can greatly contribute to environmental preservation and risk reduction.Materials and MethodsThe perishable food location-routing problem, considering the dependency of fuel consumption on the cargo load, has been formulated and solved on a complete graph consisting of customers, potential warehouse locations, and virtual refueling stations. The proposed model aims to minimize environmental impacts by minimizing fuel consumption, correlating vehicle fuel consumption with the amount of cargo transported between nodes, and optimizing the fleet composition for delivering customer orders, in addition to economic considerations and food freshness preservation. In the literature of classical location-routing problems, it is considered a challenging problem, and finding an optimal solution in this domain is difficult even for relatively small dimensions, let alone large-scale problems, where exact methods are computationally hard or nearly impossible to apply. For this reason, most research conducted to solve such problems has focused on the development of heuristic and metaheuristic methods. In this study, the genetic algorithm and simulated annealing are briefly described, and then the performance of these algorithms in solving the introduced mathematical model will be evaluated.Discussion and ResultsIn order to validate the proposed algorithms, the results obtained from these algorithms were compared to the optimal solutions of small-sized problems solved using the GAMS software. Reported numerical examples with processing times exceeding 3 hours in GAMS software were disregarded, and a total of 200 iterations were considered for the convergence of the algorithms, based on multiple experimental runs, to achieve reasonably good solutions within a reasonable time. In the solved numerical examples with small dimensions, the maximum deviation of the simulated annealing and genetic algorithms compared to the exact solution is 4.3% and 6.3%, respectively. Considering the negligible deviation values, the performance of both algorithms is acceptable for solving the proposed model. However, by comparing the solutions obtained from each algorithm, it is found that the genetic algorithm provides better solutions compared to the simulated annealing algorithm. In order to enhance the credibility of the proposed algorithms, pairwise t-tests were performed, in addition to comparing the solutions and solution times obtained. Firstly, the hypothesis of the genetic algorithm outperforming the simulated annealing algorithm in terms of solution quality was tested, followed by the hypothesis regarding solution times. According to the results, the null hypothesis was rejected in both tests, and the alternative hypothesis was accepted. In other words, the values of the simulated annealing algorithm solutions were larger than those of the genetic algorithm, as can also be observed in the box plot. Conversely, the solution time of the genetic algorithm was greater than that of the simulated annealing algorithm.ConclusionIn the conducted research, in order to further approximate the problem to real-world conditions and make it more practical, a heterogeneous fleet with non-conventional fuels was utilized, and the fuel consumption was correlated with the amount of cargo transported between nodes, aiming to minimize environmental risks alongside economic considerations in the logistics process. This problem was initially modeled by considering constraints and assumptions, and then, due to the computational complexity of solving the problem in medium and large dimensions, genetic and simulated annealing algorithms were developed to solve the proposed problem. To validate the presented mathematical model, a small numerical example was first solved accurately using the GAMS software, where the model demonstrated an average 14% fuel consumption reduction. In the next section, 15 numerical examples were generated and the performance of the algorithms in solving the proposed model was evaluated and tested in three categories: small, medium, and large. The obtained results indicate the superior quality of the genetic algorithm in finding suitable solutions for the model compared to the simulated annealing algorithm in cases where the solution time was less of a concern
supply chain management
Shaghayegh Vaziri; Farhad Etebari; Behnam Vahdani
Abstract
In this paper, we study a novel pickup and delivery problem called carrier collaboration (CC). This problem includes a set of heterogenous (refrigerated and general type) vehicles with specific capacities for serving perishable products to several pickup and delivery nodes. Each carrier can have reserved ...
Read More
In this paper, we study a novel pickup and delivery problem called carrier collaboration (CC). This problem includes a set of heterogenous (refrigerated and general type) vehicles with specific capacities for serving perishable products to several pickup and delivery nodes. Each carrier can have reserved and selective requests which can be delivered before the products are corrupted. The fleet of vehicles must serve reserved requests, but the selective requests can be served or not. Products are corrupted at a constant rate and a rate of corrosion in general type vehicles is greater than referigrated type veicles and the cost of using general one is less than referegireted. For the mentioned features, we develop a nonlinear mathematical model. The purpose is to find routes to maximize profits and reduce costs while at the same time, enhance customer satisfaction which is dependent on the freshness of delivered products. A Gnetic Algorithm (GA) is proposed to solve this problem due to its NP-hard nature. In this study, Variable Neighborhood Search (VNS) method is developed for improving the quality of initial solutions. Several instances are generated at different scales to evaluate the algorithm performance by comparing the results of an exact optimal solution wih that of the proposed algorithm. The obtained results demonstrate the efficiency of the proposed algorithm in providing reasonable solutions within an acceptable computational time.
Abstract
One of the most important problems of logistic networks is designing and analyzing of the distribution network. The design of distribution systems raises hard combinatorial optimization problems. In recent years, two main problems in the design of distribution networks that are location of distribution ...
Read More
One of the most important problems of logistic networks is designing and analyzing of the distribution network. The design of distribution systems raises hard combinatorial optimization problems. In recent years, two main problems in the design of distribution networks that are location of distribution centres and routing of distributors are considered together and created the location-routing problem. The location-routing problem (LRP), integrates the two kinds of decisions. The classical LRP, consists in opening a subset of depots, assigning customers to them and determining vehicle routes, to minimize total cost of the problem. In this paper, a dynamic capacitated location-routing problem is considered that there are a number of potential depot locations and customers with specific demand and locations, and some vehicles with a certain capacity. Decisions concerning facility locations are permitted to be made only in the first time period of the planning horizon but, the routing decisions may be changed in each time period. In this study, customer demands depend on the products offering prices. The corresponding model and presented results related to the implementation of the model by different solution methods have been analysed by different methods. A hybrid heuristic algorithm based on particle swarm optimization is proposed to solve the problem. To evaluate the performance of the proposed algorithm, the proposed algorithm results are compared with exact algorithm optimal value and lower bounds. The comparison between hybrid proposed algorithm and exact solutions are performed and computational experiments show the effectiveness of the proposed algorithm.
