Engineering topics
Publications and source records attributed to Yee, B. G. W..
Program calculates probability of defects at selected confidence levels from nondestructive evaluation data. It provides alternate method of grouping sample data to obtain reasonable value for lower confidence limit with small sample size.
Twenty sets of relevant Nondestructive Evaluation (NDE) reliability data have been identified, collected, compiled, and categorized. A criterion for the selection of data for statistical analysis considerations has been formulated. A model to grade the quality and validity of the data sets has been developed. Data input formats, which record the pertinent parameters of the defect/specimen and inspection procedures, have been formulated for each NDE method. A comprehensive computer program has been written to calculate the probability of flaw detection at several confidence levels by the binomial distribution. This program also selects the desired data sets for pooling and tests the statistical pooling criteria before calculating the composite detection reliability. Probability of detection curves at 95 and 50 percent confidence levels have been plotted for individual sets of relevant data as well as for several sets of merged data with common sets of NDE parameters.
Computer system automatically analyzes and records ultrasonic weld inspection data. Unit can be operated in three modes: manual, automatic, and computer-controlled.
Twenty sets of relevant nondestructive test (NDT) reliability data were identified, collected, compiled, and categorized. A criterion for the selection of data for statistical analysis considerations was formulated, and a model to grade the quality and validity of the data sets was developed. Data input formats, which record the pertinent parameters of the defect/specimen and inspection procedures, were formulated for each NDE method. A comprehensive computer program was written and debugged to calculate the probability of flaw detection at several confidence limits by the binomial distribution. This program also selects the desired data sets for pooling and tests the statistical pooling criteria before calculating the composite detection reliability. An example of the calculated reliability of crack detection in bolt holes by an automatic eddy current method is presented.
Report describes a prototype computer-automated ultrasonic system developed for the inspection of weldments. This system can be operated in three modes: manual, automatic, and computer-controlled. In the computer-controlled mode, the system will automatically acquire, process, analyze, store, and display ultrasonic inspection data in real-time. Flaw size (in cross-section), location (depth), and type (porosity-like or crack-like) can be automatically discerned and displayed. The results and pertinent parameters are recorded.
The signal counting technique with the Delta ultrasonic method is evaluated and optimized for flaw detection in aluminum welds. A comparison is made between the counting and conventional amplitude-gate methods to detect flaws. No conclusion is drawn on the sensitivity of these two methods to detect flaws when the mismatch at the welds is small (25 mils or less). When the mismatch is 25 mils or more, the signal counting method is more sensitive. Of the 24 welded specimens, 50 flaws were found by X-ray inspection and 59 by the Delta method. On 1/4-inch thick welds, X-ray is equal to or slightly more sensitive than Delta method, but on the 1/2-inch thick welds Delta ultrasonic appears more sensitive in flaw detection.