Semester

Summer

Date of Graduation

2019

Document Type

Thesis

Degree Type

MS

College

Statler College of Engineering and Mineral Resources

Department

Mechanical and Aerospace Engineering

Committee Chair

Terence Musho

Committee Co-Chair

Stefanos Papanikolaou

Committee Member

Stefanos Papanikolaou

Committee Member

Bruce Kang

Abstract

With recent advances in air breathable engines comes more extreme temperature environments that engine components must tolerate. During the design of these engines, it is necessary to understand how material fatigue failures occur at these new, higher operating temperatures. In providing understanding, the following fundamental study focuses on the statistical nature of crack jumps (changes in crack length over time) during fatigue in a polycrystalline nickel-based superalloy, Inconel 718 (IN718). In situ measurement of the crack length at several loading conditions were conducted using a direct current potential drop (DCPD) measurement method. Experimental data was collected at six different fatigue peak loads (R=0.15) for a statistically significant number of trials (n≥17). Calibration curves to relate electrical potential to crack length were derived from FEA and compared to analytical equations. It was determined that the mean normalized change in crack length over subsequent cycles increases with peak load. The standard deviation of the crack lengths remains constant for all loading cases. The signal-to-noise ratio was found to be best at or above a peak load of 1600N (29.65% of YS) for the given sample geometry. Results of the normalized change in crack length for a single case deviated from a Gaussian distribution. However, when all trials were considered at a single load, the distribution of the normalized change in crack length conformed to a Gaussian distribution. This lack of conformity for a single case can be explained by the history dependence of prior crack events on the crack growth for an individual specimen. This temporal information as the crack evolves, which is often overlooked in fatigue experiments, is hypothesized to be well suited for a machine learning approach that can better predict fatigue failures in superalloys.

Share

COinS