| dc.description.abstract | The management of spare parts inventory in the automotive industry presents significant
challenges and opportunities. This research introduces an advanced operational planning
model for a multi-item two-echelon inventory system, aiming to address these challenges.
The model integrates both reactive and proactive interventions by leveraging real-time
supply chain data to enhance decision-making processes. Key interventions include lateral
transshipments between warehouses and emergency shipments from a central depot,
combined with strategic stock allocations. To further improve system efficiency and
accuracy, computer vision technology is incorporated for defect detection in spare parts.
This integration employs convolutional neural networks (CNN) to identify and categorize
defects, enabling precise demand forecasting and timely preemptive actions. A mixed
integer programming (MIP) approach is utilized to determine the optimal timing and size
of each intervention, aiming to minimize total downtime and associated shipment costs.
Additionally, a greedy heuristic is developed to provide a computationally efficient
alternative for real-time decision-making, ensuring practical applicability. The model is
evaluated using case data from a leading original equipment manufacturer in the automotive
sector. Results show a significant reduction in total downtime, primarily driven by
proactive emergency shipments and lateral transshipments. Despite higher transportation
costs, these proactive measures yield substantial savings in downtime costs, which are
particularly high in automotive manufacturing. | en_US |
