The paper provides a model for multi-hit limited sample POD analysis for simulated raster scanning that is used for flaw detection in nondestructive evaluation. Raster scanning of the transducer is used in ultrasonic testing of parts. In multi-hit POD analysis system, resolution is taken into account. The scan data, known as C-scan, is represented as a 2D pixel grid. Typically, the pixel gray value is equal to signal amplitude. The data is taken at every step or index between the pixels. Each pixel represents an area of the part that is sampled by the transducer. The sampled area or the aperture provides the signal amplitude. The step size may be equal to or smaller than the aperture, creating either a non-overlapping or overlapping aperture scan pattern. Optimal scanning uses least number of steps or pixels that can provide reliable flaw detection, accurate flaw sizing and adequate spatial resolution for the target size and larger flaws. Here, the scan patterns are defined in relation to the target flaw size, transducer aperture, and step size. In this work, flaw field is simulated as symmetrical bivariate standard distribution and transducer as a receiver with square aperture. This arrangement is similar to a raster scan with an unfocused transducer. The optimal scan pattern is determined by simulating a number of scan patterns and comparing their signal amplitude, flaw sizing accuracy, probability of detection and probability of false positive in relation to contrast-to-noise ratio. If results of such simulation are corroborated in empirical data, the model can be used in assessing reliability of flaw detection in NDE as well as for choosing optimal scan pattern.
|