On model assisted probability of detection in eddy current evaluation

Doctoral thesis, 2015

The goal of this work has been to investigate the possibility to use mathematical modelling to characterize the capability and reliability of automated eddy current (EC) inspections. EC evaluation is one of the most applied methods for non-destructive evaluation (NDE) and is studied within this work. The nature of the method is complex and there is therefore a need for deeper understanding that may be gained from mathematical models. The finite element method is used for prediction of the EC interaction with defects within this work. Such models can have several objectives as for example procedure and equipment optimization or understanding capability and reliability of the method. This work focus on the model based estimation of method capability as the method is applied in realistic and automated procedures. The work has shown that tight fatigue cracks can be modelled together with the variations of realistic procedures with good accuracy. Several experimental validations are carried out. Probe-flaw interaction has been validated both regarding simple flaws such as notches but also fatigue and weld cracks. Validations have shown good accuracy but also that sever complexity is arising from the state of contact between crack faces. The bridging electrical contacts between the faces can be included in a finite element model and is important for a relevant description of the flaw. The influencing procedure parameters with their variability can also be included in the mathematical description resulting in a prediction of the probability of detection (POD) as a function of defect size. Validations have also been carried out on a procedure level and model based POD curves have been calculated and compared to experimental curves with good agreement. Within the framework of POD this work includes new optimization ideas and methods to increase the information in the procedures, for example by linking POD to the flaw severity. These developments have been accompanied with high focus on efficient models. This has lead to contributions regarding the combination of analytical and numerical models. Such models are shown to have both improved accuracy and decreased computational time compared to conventional approaches. It is also shown that several receiving elements can be included in the numerical model without any significant cost regarding the computational effort. This enables efficient optimizations as well as evaluation of complex array sensors. The hope is to see future use of this work in development of sensor design and inspection procedures resulting in improved, more efficient and reliable EC NDE. The results show that model based POD is possible to apply in industrial problems and can aid at various steps from capability estimations to signal interpretation, procedure development and training of personnel.