Using Simulation to Analyze the Predictive Maintenance Technique and its Optimization Potential

The semiconductor domain faces the rising complexity of manufacturing processes. It is characterized by high-cost pressure due to increasing competition and customer requirements in terms of quality. Therefore, it is important to explore and establish optimization technologies that increase productivity to remain competitive.

Driven by emerging Industry 4.0 technologies, Predictive Maintenance (PdM) offers a possibility to improve productivity. Predictive Maintenance is a technique that uses data analysis to detect problems in machine operations and fix them before failure. In other words, PdM optimizes machine utilization and therefore increases overall productivity.


Current research predominantly focuses on the technical implementation of the Predictive Maintenance technique, thus lacking the investigation of its operational. The researchers, therefore, apply discrete-event simulation to study the operational impact of maintenance strategies in wafer fabrication.

The model for maintenance optimization is designed in AnyLogic and it includes: two products, 83 tool groups, and 32 operator groups. Moreover, 48 units add up to one lot and each product passes more than 200 process steps. The re-entrant flow is considered deterministic, while the rework processing is modeled stochastically.

Due to the procedural nature of the model, the research team decided to use discrete-event simulation to represent the system. The team furthermore complement the model by adding certain evaluation metrics as well as maintenance and scrap data.

Related posts