Optimizing Raw Wire Inventory Management: A Data-Driven Approach to Demand Forecasting and Supply Chain Decision Support

Abstract

This thesis investigates methods to improve demand forecasting and inventory management for raw wire. Challenges such as supply chain disruptions from the COVID-19 pandemic, operational variability, and loss of expertise exposed vulnerabilities in the existing manufacturing system, leading to shortages and inefficiencies. By leveraging extensive production data, this research develops and evaluates tools to mitigate these issues while aiming for a 100% service rate. The project leveraged extensive production data to predict future wire requirements, optimize inventory, and achieve a 100% service rate. Key contributions include: 1. A data-driven demand simulation model, reducing forecast error and surpassing baseline methods 2. Quantification of waste distributions and variability in wire consumption 3. An inventory simulation framework for policy evaluation and shortage mitigation 4. Clustering analysis to classify demand patterns and identify key wire categories 5. A decision support tool supporting real-time visibility into inventory levels and risks The models and tools developed through this project provide enhanced capabilities to predict future wire requirements and manage inventory more effectively through continued development. Though the initial results indicate potential business value, areas for future work include incorporating additional data sources, exploring advanced machine learning techniques, and conducting longer-term pilot studies to quantify business impact. This project demonstrates the value of leveraging data analytics and simulation modeling to enhance supply chain decision-making in complex manufacturing environments.