Determination of the Probability of Detection (POD) as a Function of Test Grid Size in Ultrasonic Testing of Large Forgings

Quality requirements for large forgings—which are used in particular in the power generation industry in the form of turbine and generator rotors—have risen steadily in recent years. For this reason, the forgings undergo automated, non-destructive ultrasonic testing during the manufacturing process. The inspection grid is defined by the choice of measurement point spacing (distance between adjacent inspection points in the scan direction) and the choice of track offset (distance between adjacent inspection tracks). Due to their manufacturing process and heat treatment, the forgings discussed here have a fine-grained microstructure, which in turn results in low sound attenuation for the ultrasonic waves within the forging. As a result, high-amplitude phantom indications—so-called “late returns”—are often observed. To avoid these, a low pulse repetition frequency must be selected for the inspection, which leads to a long inspection time. Optimizing the inspection grid therefore means optimizing both the inspection time and the inspection costs. However, current regulations contain various instructions for defining such an inspection grid, some of which are ambiguous and unsuitable for automated testing. For this reason, the DGZfP Subcommittee on “Automated Inspection Systems” a guideline was developed that describes the procedure for determining such an optimal inspection grid for complete 100% volumetric inspection of large forgings, taking into account the acoustic field geometry of the probes used. This paper presents the procedure developed in the guideline for determining the inspection grid based on simulation calculations and experiments. The effect of the inspection grid on the probability of defect detection is investigated. To this end, POD calculations were performed to determine the detectable defect size as a function of the inspection grid size.