Ali Akbar Mohamadian; Masoud Simkhah
Abstract
Supplier selection is one of the most important problems in the field of management and optimization. It aims to optimize the cost of supply, quality of products and services, and the risk of non-supply, among others. However, existing models often overlook the risk of non-supply and the brand effect ...
Read More
Supplier selection is one of the most important problems in the field of management and optimization. It aims to optimize the cost of supply, quality of products and services, and the risk of non-supply, among others. However, existing models often overlook the risk of non-supply and the brand effect on demand. In this research, a supplier selection integer model is developed that considers lead time and the risk of non-supply. To solve this model, the LOKAD benchmark database is utilized, and a new adaptive variable neighborhood search algorithm is introduced, incorporating a scoring strategy to handle the model's complexity and obtain optimal or near-optimal solutions. The obtained Pareto solutions outperform traditional results, as confirmed by the Wilcoxon test. Sensitivity analysis of the solutions on the budget demonstrates that the final profit is more sensitive compared to lead time. Furthermore, distance from the ideal and diversity measures are used to quantitatively compare the results. IntroductionSupplier selection involves identifying, evaluating, and contracting with suppliers to meet an organization's requirements for raw materials and related infrastructure. It plays a crucial role in financial resource allocation and product/service quality. The main objectives of supplier selection include reducing purchasing risks, minimizing lead time, and enhancing quality. Many organizations face multiple criteria for selecting suppliers, such as receipt risk, green criteria, and lead time. Mathematical modeling and optimization techniques are commonly employed to achieve these objectives. However, existing models often lack real-case assumptions, motivating researchers to extend models in this environment. This research addresses these gaps by developing a mathematical model for supplier selection that considers the risk of non-supply and the substitution rate of inventory-less products. Materials and MethodsTechnically, the conducted modeling is a quantitative research in which data has been collected using library-based tools. Additionally, the statistical population corresponds to the LOKAD company, and a sample of its information is publicly available and will be utilized in the numerical computations phase of this research. This research develops a novel supplier selection multi-objective mathematical model that considers the risk of non-supply and substitution rate of demand when a product is inventory-less. Additionally, a modified adaptive variable neighborhood search (AVNS) algorithm is proposed to solve the model. While the focus of this paper is on the retail industry, the proposed model can be adapted to any industry. Discussion and ResultsThe developed model will be solved using data related to the LOKAD company, which was collected during a one-year period in 2018. This dataset includes sales and purchasing information of products related to the LOKAD company, encompassing details about the suppliers of the products and their brands. The utilization of this dataset is due to the fact that many articles have made use of it for its detailed information provided by the LOKAD company. In order to compare the method, the Wilcoxon test is used to compare two groups of variables, which can determine the presence of a difference between them. The observations reveal that the substitution rate of demand significantly affects the results and can alter the selection of final suppliers. Budget limitations are another important factor, where increasing the budget leads to higher profits by enabling the selection of more competent suppliers and high-quality products for customers. ConclusionsSupplier selection is a challenging problem in various industries. The experiments conducted in this research demonstrate that increasing the budget limitation results in higher profits, as customers prefer products or services with higher levels of quality. The developed mathematical multi-objective model incorporates real-case assumptions such as the risk of non-supply and substitution rate of demand. The model is solved using the proposed modified AVNS algorithm. The solutions obtained are analyzed using mean ideal distance and diversification metric measures to ensure their reliability. The analysis highlights the significance of budget limitations, which outweighs the impact of lead time in supplier selection problems. Ultimately, the model provides an optimal combination of suppliers. Additionally, the sensitivity analysis performed on the budget constraint reveals that changes in the budget have a greater impact on the final profit compared to lead time. The proposed model effectively determines the allocation of purchases from each supplier to enhance the final profit. In this regard, an initial estimation of future demand is considered as deterministic, although transforming this parameter into a probabilistic form can make the model more robust.
