Dynamic Requirement Verification and Safety Assessment of Complex Systems using MBSE and Co-Simulation
Critical infrastructures, particularly nuclear energy systems, face growing complexity due to regulatory constraints, performance demands, and the transition toward low-carbon power. Model-Based Systems Engineering (MBSE) supports complexity management by linking system architecture with multiphysics simulation, ensuring traceability from requirements to behavior. However, most MBSE verification approaches remain limited to nominal operating conditions and do not explicitly integrate reliability-driven degradation into requirement evaluation. In practice, aging, wear-out mechanisms, and random faults progressively erode safety margins, limiting early detection of degraded performance. This paper proposes an integrated methodology combining MBSE with stochastic degradation modeling for dynamic requirement verification under both functional and degraded scenarios. A stochastic Petri net (SPN) with Weibull-distributed transitions is embedded into a Dymola physical model to represent time-dependent failure behavior. Safety and performance requirements are formalized using the Common Requirement Modeling Language (CRML) within a SysML architecture model developed in Catia Magic and continuously evaluated during co-simulation. Interoperability between architectural and behavioral domains is ensured via Functional Mock-up Units (FMUs) and UDP-based communication, preserving synchronization and bidirectional traceability. The framework is applied to the primary loop of a Molten Salt Reactor (MSR), focusing on stochastic degradation of the primary fuel pump over 5000 s simulation. Results show a 39% mass-flow reduction, core temperature rise beyond the 708 °C safety limit, and an 18% increase in thermal power, demonstrating significant safety margin erosion. The proposed approach transforms verification from static compliance checking into a dynamic, reliability-aware assessment of system behavior over time.