EMUS-TWIN

“EMUS-TWIN“ is an integrated system for condition-based integrity assessment of pipelines in nuclear power plants. It combines tomographic EMUS testing technology (electromagnetically excited ultrasound) with a physics-informed digital twin (DT) based on a finite element model (FEM) and physics-informed neural network (PINN) calibration. EMUS sensors arranged in a ring around the pipes selectively excite guided shear wave modes (SH0 non-dispersive, SH1 dispersive) and detect signals from multiple directions and paths (axial, tangential, helical). From changes in transit time and amplitude, both geometric defects (e.g., wall thickness losses or cracks) and operational stress fields (acoustoelastic effect) are volumetrically reconstructed in tomography. The high-resolution defect and stress maps obtained in this way serve as a real basis for DT to continuously identify material parameters, boundary conditions, and loads and to provide crack propagation forecasts under real operating conditions. The central innovation is the “virtual sensor“: the verified, physics-based DT extrapolates the conditions measured in an accessible pipe section to areas that are difficult to access and critical to safety (e.g., nozzles, complex geometries, areas with high radiation loads) where physical sensors are not practical. This enables continuous, location-independent condition assessment – the basis for predictive, risk-informed maintenance strategies. Implementation will take place in three phases:

1. Development and benchmarking of the EMUS tomography system, including signal processing and inversion algorithms.
2. Development of physics-informed DT (FEM and PINN) with data assimilation and uncertainty quantification.
3. Staged validation from laboratory (TRL4) to mechanical cyclic four-point bending tests (crack initiation and propagation, TRL5) to thermo-hydraulic tests with realistic temperature transients (e.g., thermal stratification, TRL6).

The material focus is on austenitic stainless steel 304L, which is commonly used in LWR primary and secondary circuits. The concept can be transferred to other materials in the nuclear engineering environment (e.g., 316L(N), 16MND5, Alloy 690, CASS) and degradation mechanisms (erosion/corrosion, SCC, creep).

Expected effects in operation:

• Early damage detection and growth prognosis
• Reduction of unnecessary ISI deployments and personal dose (ALARA)
• Lower OPEX (approx. 20% less maintenance costs)
• Fewer unplanned shutdowns (approx. 30–50%)
• Support for long-term operation (LTO) and approval of new concepts (SMR/AMR) through explainable, verifiable AI

EMUS-TWIN addresses CONNECT-NM RL4 (NDE & Materials Health Monitoring), provides a multi-parameter NDT and E solution with AI-supported degradation prediction, and advances the technology from TRL4 to TRL6.