Designing a hybrid model for clustering warehouse items and allocating them to storage locations

Abstract

Management and warehousing operations are essential parts of manufacturing and service organizations. Warehousing is a significant component of an organization's activities, which incurs high costs and deserves more attention from researchers in this field. The aim of this research is to investigate the storage problem based on item clustering while considering all factors that affect the storage of bulky and varied products in a warehouse. The main objective of this research is to reduce transportation costs for collecting and delivering orders and to achieve more efficient use of storage space. The K-means technique is employed to solve the clustering problem, and the Generalized Allocation mathematical programming model is used to address the assignment of item categories to storage locations. This model is an integer programming model that aims to minimize transportation costs for collecting and delivering orders. This research provides a comprehensive approach to clustering and item allocation by identifying and considering effective indexes and utilizing the generalized allocation mathematical planning model to formulate and solve the problem optimally. Company managers can utilize this model to reduce their inventory costs. The innovation of this research lies in the use of clustering for the allocation of storage sites to warehouse items, followed by mathematical modeling. The proposed model was implemented at Mashhad Housebuilding Company, and several simulated problems were solved using GAMS software for validation.IntroductionThe issue of storage and warehousing is one of the main axes of industries and companies. If the warehouse is properly managed, the efficiency and productivity of the organization can be increased optimally. Warehousing is one of the main costly components in the organization's activities and deserves more attention from researchers in this field. Therefore, the main goal of this research is to reduce transportation costs during the collection and delivery of orders and more effective use of warehouse space. For this purpose, the warehouse of the Mashhad house-building factory was studied. The warehouse of the Mashhad house building factory incurs a lot of costs to collect the orders, which is the result of the improper arrangement of the warehouse. Therefore, in the current research, to achieve a suitable deployment plan and reduce storage costs, the objectives are: 1) to identify the influential criteria in the clustering of items and the model for assigning clusters to their storage location according to the studied warehouse and clustering of warehouse items, 2) to provide a model for improvement The arrangement of the group of items is followed by considering the identified criteria, parameters, and limitations.Materials and MethodsIn the present research, first, according to ABC analysis, the items are divided into three groups (A, B, and C) based on their importance in terms of storage volume and circulation. Group A items are selected for clustering, while for groups B and C, a virtual cluster is considered in the allocation problem. Influential indicators for item clustering in the warehouse were determined through content analysis. The relationship of these indicators with the problem model and their importance were identified using a questionnaire. Cummins cluster analysis was employed for item clustering. Subsequently, a generalized allocation mathematical programming model was utilized to allocate groups of items to their storage locations. This model considered limitations such as warehouse access space, interdependence between groups, demand volume, physical dimensions of items and storage locations, and crane movement during order delivery. The objective of the model was to minimize transportation costs during order collection and delivery. The problem addressed in this research is commonly known as the Warehouse Location Allocation Problem (SLAP).Discussion and ResultsIn this research, 20 clusters were obtained based on 15 indicators, resulting in a total of 154 goods items. ANOVA analysis was conducted on the obtained clusters to examine the impact of each factor on warehouse arrangement. The F statistic value indicated a significant difference among all clusters and indicators The cluster analysis results revealed that the first cluster comprised various types of footings, with the "level of activity" index scoring higher than other indices. This cluster ranked second in terms of this index. The second cluster consisted of roof products, with a higher score in the "need for quick access" index compared to other indicators. Additionally, it obtained the highest scores in the indicators of quick access requirement, demand level, consumption similarity, product activity level, and average item references in the second step, the allocation problem was formulated with two dimensions: the item cluster and its storage location. The first dimension encompassed group A items, including five clusters of frequently used items and 20 clusters resulting from the clustering process, as well as groups B and C inventory items, represented by one cluster consisting of all items from these groups and virtual items. The second dimension consisted of storage locations, with the entire storage space divided into equal areas and a total of 360 storage locations considered in the model to validate the model, the problem was solved in smaller dimensions, and the warehouse manager manually arranged the clusters in the same dimensions. The objective function value was calculated in this case and compared to the value obtained from the mathematical allocation model. The results demonstrated that the researcher's mathematical model achieved over 70% improvement compared to manual arrangement. It should be noted that the actual warehouse conditions were less efficient than the manual arrangement provided by the warehouse manager, as the items had already been clustered, and the warehouse supervisor arranged the problem manually using the data from the warehouse clustering.ConclusionsThe grouping of goods, based on the results obtained from cluster analysis, ensures that items are placed together according to important criteria such as demand, access requirements, and employee safety, among others. This arrangement creates clusters of related products, forming families of goods. This approach minimizes the search time for requested products and enables timely order fulfillment in this research, the area of each cluster was determined by considering the maximum inventory of each product. Additionally, a confidence factor was applied to account for cluster area, allowing for sufficient space in case of additional inventory. This approach ensures efficient search and timely delivery of requested products. It is recommended that if a new product is introduced to the factory's product lineup, managers should conduct cluster analysis to determine its appropriate group. If the area of that category exceeds the area considered in the present research, the allocation issue should be revisited to accommodate the new addition.

Keywords

• : clustering
• storage
• allocation generalized mathematical programming model