Organizational and Human factors for NDE reliability assessment – FOEHN

The methodology adopted consisted in working on three main aspects:
- Analysis and modeling of organizational and human factors in NDT
- Monitoring and gesture capture to observe control under controlled conditions
- Simulation to analyze influential factors and estimate POD.

The first stage involved the selection of relevant target cases in line with the project objectives and the requirements or specificities of the various disciplines involved. This first phase resulted in the selection of techniques (END) and case studies covering the various aspects. A real case proposed by EDF was used as a support case for the study and modeling of FOH. This involved a high-energy radiography NDT (RT by isotope) for which it was found that several successive inspections had failed to detect an existing defect. It was therefore decided to focus the scope of FOH modeling on radiographic inspection methods, and more specifically even on the radiogram interpretation phase.The human performance model chosen by CRC for this project is that of the CREAM method (Cognitive Reliability and Error Analysis Method) which, based on a hierarchical task analysis and a cognitive control model (COCOM), provides error probabilities. This modeling was based on a field survey campaign. The human factor can also be studied from the angle of analyzing operator gestures and their variability, using monitoring tools. Hand-held ultrasound was chosen for this aspect of the project. A study was carried out at the Institut de Soudure to observe operators under constrained conditions and analyze their performance. This study constituted the target case for the CEA's development of an operator gesture instrumentation system. In particular, the aim of such a system is to measure, under realistic conditions, the variability of parameters (such as sensor position, orientation and trajectory) related to the operator and not directly modelled by physics. The section on POD estimation and simulation covered both radiography and ultrasound. It should also be noted that a case of eddy current control was chosen by Airbus to develop a POD assessment methodology involving simulation. As mentioned above, simulation can be used to estimate PODs, based on a priori statistical distributions of the variability of influential parameters. The FOEHN project has also explored ways of combining instrumentation and simulation to determine the variability of parameters related to the human factor, and thus produce more realistic PODs.

The CREAM method has been adapted to the context of radiographic NDT to enable FOH analysis of radiogram interpretation activity. The method thus adapted can be used as a stand-alone analysis method, producing an estimate of the probability of error, depending on the context in which it is performed. The surveys and analyses carried out have enabled us to draw some initial conclusions about FOH and how it is taken into account in NDT. These lessons have been summarized in a document that will serve as a starting point for a working group launched in 2020 by COFREND (Confédération Française des Essais Non Destructifs). A virtual negatoscope has been developed within the CIVA simulation platform to approximate radiogram interpretation conditions and test POD estimates based on digital detectability criteria. Dedicated to manual ultrasonic inspection, an experimental demonstrator based on an optical tracking solution was developed and used to capture the trajectory of a sensor held in the operator's hand. The proof-of-concept of the operational simulation was carried out on a case of ultrasonic inspection of composite parts.

At the end of this work, we can confirm the importance of taking organizational and human factors into account when considering the reliability of NDT/CND processes. Observation campaigns and interviews have shed light on the sector's “cultural maturity”, and highlighted FOH issues in the fields of ergonomics, psychology, sociology and anthropology. The findings have enriched a body of knowledge that is currently under construction in the sector. The lessons learned and the prospects they open up for demonstrating performance have been discussed in a summary and recommendations document. One of the project's conclusions is the applicability of the CREAM approach to the NDT field. Limited within the framework of the project to a restricted perimeter (the interpretation of radiograms), the generalization of the approach to other operations and other methods would be desirable for its ability to provide a quantitative analysis of the risks associated with a task and the conditions under which it is implemented. At the same time, several avenues have been explored to make the processing of technical parameters directly related to the human factor more robust. These various avenues combine dedicated instrumentation for capturing operator gestures and digital simulation. The prototypes that have been developed show great potential, opening up prospects for innovation that are not limited to simply demonstrating performance, but could lead to better control and traceability of these factors.

Larouzée, J.; Martin, E.; Calmon, P. Human Reliability Assessment method applied to investigate human factors in NDT - The case of the interpretation of radiograms in the French nuclear sector. Internatonal conference Non-Destructive Examination (NDE) in Nuclear, Sustainable Nuclear Energy Technology Platorm (SNETP). Sheffield - UK, United Kingdom. Jun 2023.

Rodat, D.; Guibert, F.; Dominguez, N.; Calmon, P. Introduction of physical knowledge in kriging-based meta-modelling approaches applied to Non-Destructive Testing simulations. Simulation Modelling Practice and Theory. 2018, 87, 35-47.

CEA List : Michel Cardoso, Thomas Desrez, Sébastien Bey. Configuration d’un dispositif de contrôle non destructif. N°1859569. 16 octobre 2018.
Demande internationale WO2020/079355 A1, publiée le 23 Avril 2020

The project FOEHN fits the challenge 3 of the thematic axis Factory of the Future (Axis2) of the call. The context is the field of research dedicated to provide innovation for increasing the reliability of NDE (Non Destructive Evaluation) methods used by industry. More precisely the project proposes to develop new methodologies and simulation tools aiming at increasing the accuracy of the reliability assessment of NDE methods by taking into account influential factors including OHF (Organizational and Human Factors).
The reliability of NDE methods can be estimated by calculating probability of detection (POD) for the researched flaws. This approach allows to accounting for the variability of the parameters which impact the response of the inspection system through statistical analysis. Numerical simulation is more and more used in this context. Nowadays Organizational and Human Factors (OHF) are difficult to take into account in this context and often, the performances observed on site are lower that initially predicted. The variability of some “technical factors” also remains difficult to account for. The project FOEHN aims at analyzing and modelizing influence. The objective is to propose POD evaluation tools taking into account technical factors which influence the response of the inspection system and OHF. More specifically the project FOEHN aims at improving in thi sense the so-called MAPOD (Model Assisted POD) approach based on the use of simulation. The approach adopted consists in studying several cases of applications from which will be proposed more generic methodologic tools.
In order to fulfil these objective the consortium of the collaborative project FOEHN is constituted by:
• Three NDE industrial end-users : AIRBUS GROUP for the aircraft insudtry, EDF CEIDRE for the nuclear industry and the INSTITUT DE SOUDURE active in many industrial sectors.
• The Institute LIST of CEA Tech, RTO active in NDE which in particular develops the CIVA simulation plat-form and works on MAPOD approach.
• The CRC (Centre de recherche sur les Risques et les Crises) de l’école Mines Paristech, research center dedicated to organizational and Human Factors.
The expected results are at different levels. First of all the project will provide a formal framework allowing to better understanding, modeling and predicting the impact of OHF on the performances of inspections. In addition at the end of the project will be available sets of data obtained in well controlled conditions involving the monitoring of inspectors. Methodologies for the capture and exploitation of human factors in the context of a POD study. These tools will be implemented in a laboratory plat-form which will possibly be enriched and exploited beyond the project. In parallel, the MAPOD approach will be extended in order to provide more realistic estimated POD. The numerical tools will be implemented in CIVA. More generally it is expected a significant improvement of the estimation of statistical indicators related to POD and consequently new possibilities to optimize the design and assessment of NDE methods.