Demand forecasting for production planning in remanufacturing

Abstract

Remanufacturing is effective for energy and material savings; however, production planning and control in remanufacturing are more complex than those in traditional manufacturing. Developing a reliable forecasting method is critical for facilitating effective production planning and control. This study examined the effectiveness of demand forecasting in remanufacturing by time series analysis. Most existing methods of demand forecasting in remanufacturing assume that the time distributions of new product sales are known and that the time distributions of the demands of remanufactured products are determined by adding the product lifespan to the time distribution of new product sales. In addition, most previous studies focused on relatively long-term demand trends without considering the seasonality of demands. In this study, we examined the Holt–Winters model and the autoregressive integrated moving average (ARIMA) model, both representative time series analysis methods. These methods do not require information regarding the time distributions of new product sales and can handle the seasonality of demands. To examine the effectiveness of these methods, the time series data of the sales of 160 types of remanufactured alternators and starters manufactured by an independent auto parts remanufacturer over a period of 12 years was used. The results of demand forecasting for 2 months yielded average errors of 26.7 % for alternators and 18.4 % for starters, which represent an average improvement of 6.5 points compared to the method involving referencing the demands of the same month of previous year. The implications of the results and future steps are also discussed.