STATISTICAL FRACTOGRAPHY: THE MISSING LINK BETWEEN FRACTURE MECHANICS AND FAILURE ANALYSIS
Statistical fractography is a new engineering technique, halfway between data science and failure analysis, that extracts mechanics data encrypted in the fracture surfaces from the statistical analysis of their roughness. Like standard fractographic tools, it provides the failure history (crack direction and initiation, failure mode…) and thus assists traditional expertise. More remarkably, it also measures previously inaccessible data, like (i) the applied mechanical load at failure (stress at initiation, quasi-static and fatigue load amplitude…) as well as (ii) the in-service mechanical properties of the failed material (toughness, failure strength, fatigue resistance…). This information, complementary to the classical metallurgic analysis, are precious to determine the failure root causes, but also for redesigning the failed part. As a result, the technique has now been used for a few years in Europe for determining the root causes of critical in-service failure in the aeronautics, train, automotive and energy sector.
In this presentation, I will show, through use cases, how statistical fractography changes the scope of failure analysis by providing not only the root causes of failure events with unprecedented details but also key information that make failure a keystone for corrective action, (re)design and development of new and more reliable products.