Inclusions can cause early and catastrophic fatigue failure in powder metallurgy alloys. They act as sites for premature crack initiation, bypassing the typically long period of slow crack growth. To study this phenomenon, researchers conducted a test program at the NASA Glenn Research Center involving specially prepared powder metallurgy alloy test specimens with controlled inclusions. Because inclusions tend to be small and relatively rare in current production powder, inclusions of known size distributions were introduced to the precursor powder to ensure that a prescribed average number would occur on the surface of the fatigue test samples. A geometric model was devised at Glenn so that the inclusions could be modeled as crack-initiating defects. It models the inclusions as ellipsoids that can have three different dimensions (semi-axes) and random orientation.
Projected area of an ellipsoidal representation of an inclusion. Inset photograph displays cracks propagating from noncoplanar asperities.
In examining the fracture surfaces of failed “seeded” specimens, we found that the expected maximum cross section of an inclusion was not always an accurate predictor of the starting crack size. Geometric simulations based on the maximum-cross-section assumption predicted size distributions that have smaller means and less scatter than the size distributions observed on fracture surfaces. We also observed that cracks tend to grow from the corners of inclusions and do not always initiate in the same plane (see the preceding figure).
These observations led to an alternate assumption that the projected area of the inclusion may be a more accurate predictor of the crack-initiation area. Consequently, a closed-form solution for the projected area of a randomly oriented arbitrary ellipsoid was derived. Simulations based on the projected-area assumption showed improved correlation with the observed fracture surface area distributions.
Comparison of fracture surface failure-initiating inclusion areas with simulated maximum cross section and projected area for two inclusion size distributions. Projected-area distributions agree with observed data for both surface and internal failures. Left: Ram 90-270+325 (mesh size). Right: Alcoa T64-140+170 (mesh size).
Long description of figure 2.
Last updated: October 11, 2006
Responsible NASA Official:
Gynelle.C.Steele@nasa.gov
216-433-8258
Point of contact for NASA Glenn's Research & Technology reports:
Cynthia.L.Dreibelbis@nasa.gov
216-433-2912
SGT, Inc.
Web page curator:
Nancy.L.Obryan@nasa.gov
216-433-5793
Wyle Information Systems, LLC
