GEOMETRICAL AND ULTRASONIC MODELING OF HARD-ALPHA DEFECT COMPONENTS IN TITANIUM ALLOYS Chien-Ping Chiou, Brian Boyd and R. Bruce Thompson Center for Nondestructive Evaluation Iowa State University Ames, Iowa 500 11 James Oliver Department of Mechanical Engineering Iowa State University Ames, Iowa 500 11 INTRODUCTION In the last few years, considerable research have been conducted in the Engine Titanium Consortium [1] in the study of naturally occurring hard-alpha flaws found in titanium alloys. One task in the Consortium is undertaking a new approach of bringing state-of-the-art geometrical and ultrasonic models together [2-6] to assess the improvements in the capabilities to detect hard-alpha inclusions, as quantified by the probability of detection (POD) [7-8]. In the past, such POD studies normally required statistical analysis of large amount of experimental data. However, the rare occurrence of naturally occurring hard-alpha defects and the infeasibility of manufacturing realistic synthetic flaws have made the data acquisition very difficult. To get around this obstacle, a new approach utilizes ultrasonic models to accurately calculate the flaw responses. These require, as input parameters, geometrical models containing the spatial and material descriptions of the flaws. By using models in such way, we are able to examine the effect of flaw morphology on the ultrasonic signals and to establish the maximal knowledge base from the limited naturally occurring flaws. Furthermore, a variety of inspection scenarios can be rapidly simulated to generate the necessary data for the POD evaluation. This approach was advanced recently, when a power failure in production caused a heat of titanium to contain many hard-alpha flaws whose properties were considered to be similar to those occasionally occurred in normal manufacturing processes [I]. This provided an unprecedented opportunity to gain extensive knowledge on naturally occurring hard-alpha defects. Since that time, various experimental data have been obtained by ultrasonic, metallurgical and chemical analyses to determine the detailed morphology and composition of these defects. Ten of the indications were chosen as representative sample defects for these studies. Most recently, the metallurgical sectioning of two of the ten Review of Progress in Quantitative Nondestructive Evaluation. Vol. 18 Edited by Thompson and Chimenti, Kluwer Academic/Plenum Publishers, 1999 1829