In the nuclear industry, the regulator imposes periodical assessments of the Nuclear Power Plant’s (NPP) « good health » for everyone’s safety, and not only before commissioning or in case of an incident.

Structural integrity assessment codes provide a set of guidelines and methodologies to follow in addressing safety cases. Either French design codes RCC-MRx, RCC-M, and RSE-M, or the UK’s R5 and R6, rely on analytical and semi-analytical methods to assess fracture, fatigue, and creep-fatigue life. These guidelines are purposely defined to be fairly simple for industry use (elastic analysis and plastic correction) and to cover all NPP situations in a conservative way, so that all uncertainties are accounted for and margins ensured.

But in some cases, when applied to NPP component geometries and realistic transients, they can be overly conservative, which is not always appropriate and justified. For this reason, they are continuously evolving, but the addition of new methods requires robust technical justifications.

Particularly, material data and characteristics, criteria being obtained from tests on laboratory specimens (uniaxial specimens, CT specimens, Charpy, …), understanding their transferability from laboratory specimens to plant-relevant components has been a key issue for decades.

When addressing this issue combined with conditions and parameters as complex as environmental assisted fatigue, variable amplitude loading in biaxial conditions, weld residual stress effect, etc…, feature tests become the best candidates to provide a mechanistic understanding of the differences between small scale lab tests and full component, gain a better knowledge on particular uncertainties and questions remaining opened (mean stress, triaxiality effect, closure effect, stress in singularities, …).

While their design and realisation are a real challenge to ensure high-quality results, these tests bring precious knowledge to improve our abilities to handle safety cases accurately.

As a first illustration, tests on notched and tubular specimens combined with a complex experiment in biaxial cycling loading in a Pressurised Water Reactor environment brought essential data to highlight the reduction in environmental effect and overload effect in biaxial loading conditions compared to uniaxial ones.

A second example concerns weld residual stress effects on fatigue propagation, where component-scale pipes are tested to evaluate the difference in mean stress, stress re-distribution and closure effect between the pipe and CT specimens. In this case, the challenge is to accurately measure the stress field and the resulting surface crack geometry at every step of the propagation. This must be achieved using a non-destructive method such as neutron diffraction. This would generate the knowledge to account for residual stress more realistically in fatigue assessment of components.