A Simulation-Driven Framework for Risk-Based Maintenance Using Digital Twin

The increasing complexity of industrial systems requires maintenance strategies that strike a balance between reliability, cost-effectiveness, and uninterrupted production. This paper proposes a simulation-driven framework for risk-based maintenance (RBM) that integrates digital twins (DTs), process mining (PM), Discrete Event Simulation (DES) and real-time, sensor-based diagnostics. The framework is intended for maintenance planners who decide when to intervene and how to schedule maintenance activities while accounting for system-level impacts on production. It is particularly applicable in settings where assets provide diagnostic signals and operational and maintenance event logs are available from Manufacturing Execution Systems (MES) / Computerized Maintenance Management Systems (CMMS). Within the proposed architecture, DTs replicate the behavior of machines and processes using sensor data, while PM extracts operational patterns and activity dependencies from event logs to continuously calibrate model structure and parameters. These representations are then linked to a DES module that simulates various maintenance and fault scenarios and assesses their impact on production KPIs such as availability, cost, and downtime. The framework allows maintenance and production strategies to be evaluated dynamically through `what if' scenario simulations, providing a quantitative basis for decision-making in uncertain and evolving operational contexts. Simulations outcomes are then integrated into a CMMS, closing the loop between datadriven prediction, risk evaluation and operational execution. This approach supports both predictive and prescriptive maintenance, enhancing system resilience by combining continuous monitoring with model-based simulation. By facilitating scenario-based risk assessment and optimization, the framework serves as a valuable asset in transitioning industrial maintenance from reactive and preventive strategies to intelligent, simulation-assisted decision-making within the context of Industry 5.0.

Keywords: Risk-based maintenance, Discrete Event Simulation, Digital Twin, CMMS.