Demand and Replacement Forecasting for Remanufactured Parts of Industrial Products

Abstract

Remanufacturing supply chains are complex due to the circular and interconnected nature of demand and supply. Good demand forecasts are critical for remanufacturers to optimize inventory (of cores, components and finished products) and production planning. Inventory shortages lead to lost sales, delayed fulfillment, or expensive substitutions with new parts, while excess inventory ties up working capital. We collaborated with a large industrial products manufacturer to improve demand forecasting for remanufactured parts requiring periodic replacement. The manufacturer's large global install base of products requires parts replacements at stipulated intervals as part of maintenance. However, the demand tends to be highly variable. We have developed an analytics-based approach to model this variability by considering equipment usage and customer behavior. For installed products, the duty cycles and hence the run-time vary considerably over time. We analyze historical run-time data from products in the field, modeling it as time-series and applying several time-series forecasting techniques to predict future usage more accurately. Additionally, we found that customers do not adhere to the stipulated replacement intervals. To account for such deviation, we characterized individual customer's replacement behavior. By combining each product unit's forecasted run-time with the customer's replacement pattern, we can more accurately predict the upcoming parts replacements. By aggregating forecasts across product units and regions, we generate insights into future demand at headquarters, regional, and dealer levels. Our approach enables manufacturers to optimize inventory across their entire network, streamline production processes, minimize operational costs, and enhance customer service by ensuring timely availability of replacement parts.