Demand Forecasting and Inventory Control System for Industrial Valves

Abstract

Listen

Translate article

AbstractForecasting customer demand for preliminary products in an accurate way plays a vital role in increasing efficiency of inventory control systems, reducing total costs and meeting the requirements of customers on time. Considering this fact, the chief objective of the study is to develop a user-friendly decision support system (DSS) to be able to forecast demand for products and minimize the cost of total inventory control costs including ordering and holding costs. Due to the complexity of the problem of this study, the project is handled in two parts, namely, demand forecasting and inventory management. In the demand forecasting part, unlike the traditional methods which mostly ignore the statistical behaviour of demand distribution of products, we employed Holt-Winters and SARIMA techniques which minimize the error of forecasting by harnessing demand behaviour. In the second part, the forecasted demand values are used as inputs for the inventory control system. In this part, we developed a Mixed Integer Programming Model (MIP) where the total inventory cost, involving ordering and holding costs, is to be minimized. To solve the proposed mathematical model, IBM CPLEX OPTIMIZER coupled with Branch & Bound Algorithm (B&B) is employed. In addition to this exact solution technique, we also used the Benders Decomposition method which is suitable to solve MIP models in a more reasonable computational time with optimality, by decomposing the model into master and sub-problem. Besides these two exact-solution techniques, to determine the number of products to be ordered from a supplier in a shorter computational time when the problem size is larger, a heuristic solution was developed, adapted from the Silver Meal algorithm. The results obtained using the aforementioned techniques are compared concerning their solution quality and computational time.KeywordsForecastingInventory control systemDecision support systemSARIMAHolt-wintersBranch and boundSilver meal heuristicBenders decomposition