Pierre Calmon received in 1990 his PhD in Physics at the University of Paris Sud, France. He joined the Ultrasonic NDT laboratory of CEA and worked on ultrasonic modelling. Involved since the start in the CIVA project, he is in charge of the coordination of the software developments from 2000 to 2005. During the same period he is the head of the Simulation and Processing Ultrasonic Laboratory of CEA. Since 2008 he coordinates core research around CIVA. He is also involved in NDT standardization and teaching. His main fields of interest are modeling and processing of NDT data.

Abstract

Trends and stakes of NDT simulation

A major evolution in the NDT field during the last decade is the importance taken by numerical simulation. Indeed, in all the industrial sectors concerned by high-level inspection methods the role of the NDT simulation has been continuously increasing and diversifying. The last advances in the CIVA simulation platform developed at CEA LIST illustrate this evolution. The applications of simulation are numerous.

Historically, the interpretation of results and understanding of phenomena have been the first motivation for NDT modelling. This need remains relevant and simulation is a powerful tool to help the expert in its diagnosis. One promising research direction in this domain is the development of model-based inverse methods paving the way for automatic diagnosis algorithms. On the other hand, facing the real-time challenges, model-based imaging algorithms are emergent tools for ultrasonic array applications. An increasingly important type of applications concerns technical qualification and reliability assessment of inspections. The evaluation of performances of NDT techniques is a crucial issue to which numerical simulation can significantly contribute. Results issued from simulation can now be included in a technical justification as valuable information on inspection capability. A recent trend is to apply simulation in the context of statistical approaches taking into account fluctuation of inspection parameters and leading to the evaluation of probability of detection (POD) of flaws. By replacing and complementing experimental measurements, numerical simulation can contribute to drastically reduce costs of expensive experimental campaign and to improve reliability of POD determination. Numerical simulation is also currently used for the conception of new inspections and probes. In particular, simulation has accompanied the considerable development of ultrasonic array techniques by offering tools for optimizing the design of the arrays and computing adapted delay laws for the detection of a given defect in a given part. Latest advances in this domain concern the development of automated design tools connecting numerical simulation to optimization algorithms.

Aiming at fulfilling the strong industrial demand, the research is very active in the field. The domain of applicability of the codes is continuously widened to deal with more complex parts, defects or materials while the spectrum of NDT techniques addressed by the codes increases. Simulation is coupled to data processing algorithms, to inverse loops or to statistic methods in order to propose new tools as evoked above. These new uses of simulation involve intensive computations and the numerical efficiency of the algorithms remains a crucial challenge. Hybrid approaches coupling different numerical approaches appear as a satisfying solution in many intricate cases. The exploitation of new generation hardware architectures through the development of adapted parallel algorithms is also very promising and first studies prove significantly improved performances.

In this communication, this overview is illustrated by examples of applications and recent developments achieved in CIVA.