Optimizing Sustainable Production Scheduling and Inventory Management in a Mixed-Model Assembly Line Using Digital Twin and Real-Time Data Analytics

Project Abstract

This research presents a comprehensive framework for optimizing sustainable production scheduling and inventory management in a mixed-model assembly line by leveraging digital twin technology and real-time data analytics. The study addresses the increasing complexity of modern manufacturing environments where product variety, fluctuating demand, and resource constraints challenge traditional scheduling and inventory practices. A digital twin of the entire production system is developed to mirror physical processes, equipment health, and material flow, enabling closed-loop decision-making. Real-time data from shop-floor sensors, MES, ERP, and external sources are integrated to provide a holistic view of operations, enhancing visibility and responsiveness across planning horizons from tactical to operational levels. The core methodology combines advanced optimization, simulation, and machine learning to formulate and solve a multi-objective problem that simultaneously minimizes production lead times, energy consumption, and waste, while maximizing on-time delivery, throughput, and service level. A mixed-model sequencing and scheduling model is proposed to handle diverse product variants on shared resources, incorporating setup time reduction, line balancing, and takt-based pacing. Inventory policies are embedded within the digital twin to optimize buffer sizing, safety stock, and reorder points under demand variability, supplier reliability, and obsolescence risk. A sustainable emphasis is achieved through carbon footprint accounting, energy-aware routing, and circular economy considerations such as waste reduction and recycling opportunities at different stages of the value chain. Key contributions include (i) a digital twin-enabled decision support architecture that synchronizes the physical shop floor with a virtual counterpart for scenario analysis, what-if studies, and rapid replanning in response to disturbances; (ii) a hybrid optimization framework that integrates mixed-integer linear programming, heuristic search, and reinforcement learning to cope with combinatorial complexity while delivering near-optimal schedules in real time; (iii) a robust inventory management strategy that leverages predictive demand signals and lead-time variability to reduce stockouts and excess inventory; and (iv) an environmental performance module that tracks energy usage, emissions, and waste, and provides actionable insights to steer sustainable choices without compromising operational performance. The research adopts a multi-site case study across an automotive components manufacturer transitioning to a highly configurable product portfolio. Data-driven validation includes historical and live shop-floor data, synthetic stress tests, and cross-validation against benchmark policies. Performance metrics encompass throughput, cycle time, on-time delivery, total cost of ownership, energy consumption, and CO2 emissions. Sensitivity analyses examine the resilience of the proposed framework to demand volatility, equipment failures, and supplier disruptions. The outcomes demonstrate significant reductions in lead times and work-in-process, improved fill rates, lower energy consumption, and enhanced adaptability to demand shifts, while maintaining or improving environmental performance. The study also outlines deployment guidelines, change-management considerations, and potential scalability to other manufacturing contexts, offering a practical blueprint for integrating digital twins and real-time analytics to achieve sustainable optimization in modern mixed-model production environments.

Project Overview

What This Project Is About

---

A plain-language overview of how a manufacturing line with different product mixes can be scheduled efficiently and keep inventory in check using digital tools. The project looks at combining real-time data from production with a digital replica of the factory to plan tasks, manage stock, and reduce waste.

---

---

The Problem It Addresses

---

Many factories struggle with balancing multiple products on the same line, unpredictable delays, and excess or missing inventory. This leads to higher costs, longer lead times, and less sustainable operations. The project investigates ways to improve scheduling and stock control in a mixed-model environment.

---

---

Objectives of the Project

---

1. Understand current scheduling and inventory practices in a mixed-model line.
2. Develop a simple framework for integrating digital twin concepts with real-time data.
3. Propose methods to reduce waste, energy use, and stockouts.
4. Test the framework with a small-scale or simulated production scenario.
5. Provide guidelines for implementation in similar manufacturing settings.

---

---

What You Will Do Step by Step

---

1. Review basic concepts of production scheduling and inventory management. 2. Learn how digital twins and real-time data work at a high level. 3. Collect or simulate data on production times, scrap, and stock levels. 4. Build a simple digital model of a mixed-model line. 5. Run scenarios to compare traditional vs. digital-assisted schedules. 6. Analyze results for waste, lead time, and stock accuracy. 7. Summarize findings and limitations. 8. Prepare implementation tips for industry use.

---

---

Expected Outcome

Expect a clear, actionable framework that combines digital twin ideas with real-time data to improve scheduling and inventory management. Anticipated benefits include shorter lead times, lower waste, better stock control, and an approachable path for adoption in real factories.