Improving spare parts forecasting: the case of Carmeuse

Details

Supervisors

Blome, Constantin

Faculty

Louvain School of Management

Degree label

Master [120] : ingénieur de gestion, à finalité spécialisée

Abstract

During my internship in the supply chain department of Carmeuse Americas, I focused on enhancing spare part usage forecasting within the organization’s purchasing department. The objective was to develop a user-friendly model to optimize spare part forecasting to choose the best minimum and maximum value of the quantity in stock for each spare part of the Black River plant of Carmeuse, aiming to create value by minimizing stock while ensuring operational needs are met. After a comprehensive analysis of spare parts prediction methods, we selected the Weighted Moving Average (WMA) and Exponential Smoothing techniques. These were chosen due to their compatibility with Carmeuse’s preference for an Excel-based model and their suitability for high-consumption items. However, upon analyzing Carmeuse’s data, it became evident that the accuracy of these methods was hindered by discrepancies between actual usage and the available inventory movement data. Consequently, we compared the WMA and Exponential Smoothing methods with a simple average. We found that the latter yielded the lowest average Root Mean Square Error (RMSE), indicating its superiority in predicting usage with the available data. Therefore, we proposed a model based on SAP data, recommending the average of the last ten years as the reorder point and the upper bound of the 95% confidence interval as the maximum stock level. In conclusion, our project aims to improve spare part management at Carmeuse Americas by providing a practical forecasting model that balances operational needs with inventory optimization, ultimately reducing storage costs while ensuring adequate inventory levels.