Aggregate production planning for batch-oriented discrete manufacturing system integration with forecasting

Abstract

Aggregate Production Planning (APP) involves the determination of company’s optimal production, inventory and employment levels with a given set of resources and constraints. Forecasted demand of products is one of the important inputs of APP and a more justified as well as realistic forecasting technique for prediction of market demand is very crucial for reducing unnecessary inventories, smoothing the production plan etc. Usually in APP process, economic planners of most of the manufacturing companies in Bangladesh use subjective and intuitive judgments to estimate future demand which leads the result to infeasibility or decreased performance. Nevertheless, aggregate plan is the basis of subsequent plan, and thus, accuracy in it leads to proportionate accuracy in master production schedule (MPS) and material requirements plan (MRP). This study develops a decision support model for multi-period multi-product aggregate production planning integration with forecasting technique aiming at minimizing the total relevant cost considering projected demand, production capacity and work forces, inventory control, backorder, and wastage reduction. In this study, different time series forecasting models are applied on the historical data of two product groups (Hooded jacket, Ladies cardigan). Then, error levels are compared with those obtained by subjective and intuitive judgements (company’s current practice). It is found that winter’s additive method and Holt’s method provide lower forecast errors for hooded jacket and ladies cardigan respectively. A multi-period multi-product mathematical model for APP problem is formulated which is solved by Linear Programming (LP) and Genetic Algorithm (GA) approaches. Finally, the results drawn from two different approaches are compared with company’s current production plan in terms of total cost to evaluate the best one for a situational APP decision. According to cost minimization objective of APP, linear programming seems to be satisfactory than genetic algorithm and company’s current practice. Practically, for simple linear optimization problems, linear programming (LP) approach is suitable to provide better result. The proposed framework is effective and easy to implement in practical management and supply chain systems. So, this study can be the roadmap for manufacturers as well as planners to minimize total cost.