A study of the sources of sizing error incurred in manual ultrasonic NDT

摘要

Errors are inevitably incurred when sizing defects by manual ultrasonic NDT, even when the defect echoes are clear and unambiguous, and no human blunders are made by the inspector. These sizing errors are due to small random fluctuations in the various calibration and measurement processes which constitute a manual ultrasonic inspection, and due to inherent errors in the sizing technique itself. This paper describes a study undertaken to quantify 11 of these random fluctuations: the so-called "primary errors". Examples are the error in calibrating the probe beam angle, the error in reading the range off the flaw detector screen, and the error due to coupling variations caused by roughness of the scanning surface. In most cases, careful experimental studies were performed to quantify the particular primary error in isolation from the others, using a number of qualified inspectors and processing their results statistically. The results can be used in existing algorithms to obtain a statistical estimate for the overall defect sizing error.