Corrosion-nucleated fatigue crack growth

Keith van der Walde, Purdue University

Abstract

The presence of corrosion is known to dramatically reduce the fatigue failure resistance of materials. Pitting, a particularly prevalent and insidious manifestation of corrosion, is the damage state addressed in the present work. First, attention is focused on developing an improved understanding of the physical process of fatigue crack growth in pre-corroded 2024-T3 aluminum. To this end, a large-scale quantitative fractographic study of specimens that failed due to pit-initiated fatigue crack growth was performed. Over half of the specimens were found to have failed due to crack growth from more than one nucleating pit. This led to subsequent experimental work consisting of interrupted fatigue testing of similar specimens. Here again multiple cracking was observed, as was early stage microcracking. These microcracks were found to bear resemblance to the grain structure in terms of size, shape, and orientation, and were noted to originate from pitting damage at numerous locations. It was further conclusively deduced from this work that crack nucleation is essentially instantaneous. The observed rapid formation of cracking is likely due to the stress-concentrating effects of pitting damage. Using finite element analysis, this was quantified in terms of magnitude and distribution for several pitting geometries of increasing complexity and realism. Here the complex flow of stress around such damage was successfully demonstrated and a simplified means of pit representation was assessed. This work further provided insight into the interrelation of the stress-concentrating effects of pitting, and the nucleation and growth of cracking. To allow for systematic quantification of corrosion damage, computer code was developed that is capable of measuring pitting from cross-sectional micrographs. This tool was applied to the material system analyzed above and found to effectively extract metrics bearing direct relevance to fatigue performance. A fracture mechanics-based fatigue life prediction model was developed that accepts as the input the aforementioned micrographs from which size-equivalent flaws are produced. By simulating multiple crack evolution scenarios, this model established distributions of potential fatigue lives for the material. The extreme values of these distributions were taken as limiting and were shown to agree well with the experimentally observed fatigue lives.

Degree

Ph.D.

Advisors

Hillberry, Purdue University.

Subject Area

Mechanical engineering

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS