Defense of Clément MOREAU

Measurement uncertainty assesses the reliability of a result by taking into account the various sources of variability likely to affect it. In the industrial field, where physical quantities such as length, temperature or surface roughness are regularly measured, it is an essential element. Unlike the notion of error, uncertainty does not express a deviation from a value assumed to be true, but describes all possible fluctuations around a result. It thus makes it possible to estimate the variability of a measurement and to judge its comparability, conformity or validity.

The rigorous assessment of uncertainties in surface metrology requires a consistent, reproducible experimental approach adapted to the diversity of instruments and processes encountered. This approach is based on metrological standards and revolves around several axes: instrument validation, definition of experimental protocols (Morphomeca Monitoring plans), short- and long-term instrumental stability, selection of relevant parameters and prediction of fluctuations.

At Morphomeca Monitoring, we are able to assess the uncertainties associated with surface metrology.

Three complementary methods have been developed to structure a generalizable quantification of measurement uncertainties. The first, known as the "Pixel" method, is based on local analysis of height fluctuations from repeated measurements at the same location. It produces maps revealing correlations between fluctuations and surface gradients. A visual tool, the B2D plot, facilitates this relationship and identifies critical gradient thresholds.

The second method, called "Index", is based on statistical analysis of roughness parameters through four indices: Quality, Drift, Stability and Relevance Index. These indicators qualify sources of uncertainty according to instrument type, measurement strategy and environmental conditions. This approach also highlights the variable sensitivity of parameters to fluctuations, and the need to rule out anomalous measurements to guarantee the reliability of results.

The third method, known as "Mean/Standard Deviation", involves a series of repeated acquisitions at the same location, under controlled conditions, in order to study the intrinsic behavior of an instrument. Analysis of the {mean, standard deviation} pairs of roughness parameters reveals a relationship of proportionality in line with Taylor's law. The coefficients α and β can then be used to model the metrological behavior of the instrument, and to predict the uncertainty of a parameter from a single measurement, after experimental testing.

Application of these three approaches to TA6V abraded surfaces, then to other materials and configurations (stitched, multilayer, isotropic or anisotropic surfaces), has confirmed the robustness and generalizability of the proposed methods. Cross-analysis of the results shows that certain parameters, notably Sdr, are particularly well suited to time tracking and instrument stability monitoring. Finally, the synthesis of all the work leads to concrete recommendations on metrological practices, instrument choice, settings and measurement strategies, thus paving the way for a more systematic integration of uncertainty in the topographic characterization of industrial surfaces.

Reporters

• Stéphane Benayoun (Dr., Professor, LTDS - École Centrale de Lyon, France)
• Noël Brunetière (Dr., CNRS Research Director, Pprime, Poitiers, France)

Examiners

• Christopher A. Brown (Dr., Professor, Worcester Polytechnic Institute, Worcester, USA)
• Marie-Ange Bueno (Dr., Professor, LPMT, Mulhouse, France)
• Thierry Coorevits (Dr., Senior Lecturer, ENSAM, Lille, France)
• Julie Marteau (Dr., HDR Senior Lecturer, Roberval - UTC, Compiègne, France)

Thesis supervisor

• Maxence Bigerelle (Dr., Professor, LAMIH - INSA Hauts-de-France, Valenciennes, France)

Encadrant

• Julie Lemesle (Dr., Engineer, Valutec - LAMIH, Valenciennes, France)

Guest member

• François Blateyron (Research Director & Metrology, Digital Surf, Besançon, France)
  

Métrologie, Incertitudes, Rugosité, Topographies de Surface, Profilomètre 3D, Surfaces Manufacturées, Interféromètre, Confocal, Focus Variation, Paramètres de rugosité.