Pod Analysis Of Ultrasonic Testing For Inspecting Fatigue Crack

Due to the effects of fatigue loading and corrosion,cracks are a common defect in aging structures.For example,cracks usually occur in pipelines,cladding of pressure vessels,and other core equipment in nuclear power plants,the propagation of crack will lead to sudden and catastrophic failure.Therefore,ensuring the integrity of these equipment is critical to the safe operation of nuclear power plants.Ultrasonic testing is widely used by industries to detect cracks and other defects in components and equipment.However,the testing process is inevitably affected by various factors,such as geometry of structure,material properties,temperature,etc.The effect of influencing factors will introduce noise into the testing signal,thereby reducing the reliability of the testing results.In order to evaluate the detection capability of ultrasonic testing to cracks,its detection reliability needs to be quantified.In past related researches,artificial cracks with regular profile were usually fabricated in the specimens for reliability analysis of ultrasonic testing.However,the cracks in reality involve more complicated profile and other factors,therefore,using artificial cracks cannot reasonably evaluate the reliability of ultrasonic testing for crack detection.This paper uses the probability of detection(POD)to evaluate the reliability of ultrasonic testing to detect fatigue cracks in austenitic stainless steel based on numerical simulation and experiment.Three-dimensional simulation using ComWave,a numerical simulation software based on finite element method,was conducted.The model was firstly constructed,which considers ultrasonic testing of a plate with fatigue crack.And then the change of signal with crack dimensions is analyzed,which confirms that both depth and length of the crack have large impact on the signal.Therefore,the reliability analysis of ultrasonic testing needs to consider these two variables simultaneously.In the experiments,actual fatigue cracks were prepared by using the four-point bending test in the austenitic stainless steel plate samples.The surface of plate samples is processed to expose the fatigue crack whose dimensionsare subsequently measured by electronic microscope.The ultrasonic testing instrument UI-25 is used to inspect them to obtain the experimental signal.In the respect of evaluating the reliability of ultrasonic testing,the conventional univariate probability model is firstly used to build the relationship between POD and crack depth using simulation signals and experimental signals.The results show that,under the assumed decision threshold,the ultrasonic detection method can reliably detect cracks with a depth greater than 1.6 mm.In addition,it is found that the decision threshold also has a large influence on the reliability analysis results,thus,reasonable methods should be used to define the decision threshold.Due to the limitations of the univariate model which requires linear relation between signal and variable,it is not able to use this model to establish the relationship between POD and crack length.With the aid of multivariate POD model,the discrepancy between experimental signal and simulated signal is calibrated,and a POD contour is created.This contour can show the change of detection probability with the crack length and depth at the same time,so it can more comprehensively evaluate the reliability of ultrasonic detection.