| Abstract Scope |
Generating comprehensive fatigue test data is a costly and time intensive endeavor. These restrictions are exacerbated when additive manufacturing techniques are being investigated. Although the majority of specimens will fail as designed, a portion of them will have fatigue lives that exceed a pre-defined upper limit and are termed runout. Other specimens may fail in the shoulder or grip regions due to experimental designs or, more likely, at the unrepresentative discontinuities and microstructures that may form during build stoppages or interruptions. Normally, such data would represent a loss. However, by considering these types of tests as right censored in that the representative fatigue life may have occurred at a later cycle number, additional information and utility can be extracted through reliability analysis. By using reliability analysis techniques, such as Cox regression, runout and invalid fatigue test data can be leveraged to help better quantify the marginal contributions to failure likelihood and provide a means of recouping losses in highly experimental research. |