Propagation of Small Cracks Caused by Underloading in AA7050-T7451

The fatigue design and management of aircraft is essential to ensure they safely and economically achieve their designed service lives. However, despite over 100 years of research into fatigue crack growth there are still significant challenges to accurately predicting the fatigue life of components. As a result, all commercial and military aircraft types are required to undergo full scale fatigue tests whereby they are repeatedly loaded in a manner representative of flight. To account for the statistical scatter in crack growth, they are often subject to repeated loads equivalent to at least two lifetimes of the aircraft design life. This results in expensive and time-consuming testing. In addition, these tests often reveal unexpected fatigue issues with the initial aircraft design, resulting in costly redesign processes. These redesigns are necessary for ensuring the safety of the aircraft and will usually increase the economic life of the aircraft. Ideally, however, the initial crack growth predictions should be accurate enough not to require such redesigns. Two reasons for this unpredictability include (i) the differences in the expected growth of small and long cracks, and (ii) the differences in the expected crack growth for constant amplitude (CA) and variable amplitude (VA) loading. It has been shown in many studies that small cracks, here defined as cracks less than 1 mm in length, will typically grow faster than expected compared to long cracks when subject to nominally identical stress intensity factor ranges. This means that the extensive crack growth data gathered from long crack tests during the past century cannot be directly used to predict small crack growth accurately. Additionally, it has been shown that coupons subject to VA loading, where loads are applied in a pseudorandom order, will nearly always experience faster crack growth than when the loads are applied in a blocked CA manner. As such, even when predicting long crack growth, where there is extensive historical data, if the sequence of the test loads is different to that in the prediction, then it is likely the results will deviate noticeably between the test and prediction. The difficulty in the predictability of VA loading growth compared to CA loading growth comes from ‘history effects’ associated with the loading and the material. These history effects are often explained in the literature as being caused by material surrounding the fatigue crack being plasticly deformed from the previous loads. In the case of a large tensile load, the material around the crack tip is stretched more than for smaller loads. Once the load is relaxed, this stretched material no longer fits within its original volume around the crack tip, meaning that it must be compressed by the neighbouring, non-plastically deformed, material. This has the effect of superimposing a compressive stress on all bulk stresses at the crack tip. The opposite can occur if the crack tip remains open when a large compressive load is applied, resulting in the superimposition of tensile stresses. This means that the true stress range at the crack tip is constantly changing in the case of VA loading and can be quite different at the crack tip compared to the far field stress ranges being applied. This is one reason why the literature has found that using linear summation of damage without accounting for history effects can lead to vastly incorrect predictions of fatigue life. The other above-stated reason for the unpredictability of fatigue crack growth rates stems from the small crack effect. This is a result of the fact that cracks will typically grow faster when they are small (shorter than 1 mm) compared to when they are long (greater than 1 mm), despite nominally identical crack tip stress intensity factor ranges being applied. This poses a problem for highly stressed structures, since they can often spend 70% or more of their lifespan in this small crack regime. This means that predictions made using long crack data can be highly unconservative. There are various reasons for this stated in the literature, including reduced crack closure, and a fundamental breakdown of the similitude concept that underpins linear elastic fracture mechanics. Regardless of the reason, this variation means that crack growth rate predictions of small cracks using the long crack data are typically slower than reality. The focus of this thesis is on better quantifying and describe small fatigue crack growth. The material chosen for the experimental phase of this study is AA7050-T7451 and the focus of the study will be the effect that underloads have on the crack growth rate. These underloads are individual load cycles where the minimum applied load is lower than the surrounding loads. These underloads may have the effect of superimposing a tensile residual stress on the load cycles following their application. When these tensile residual stresses are considered in conjunction with the known faster small crack growth rates, highly unconservative predictions are often made when modelling such crack growth conditions. These unconservative predictions may be one reason for unexpected fatigue cracking that commonly occurs in full-scale test articles. This study into underloads has had both a quantitative and qualitative focus. Firstly, multiple crack growth studies were conducted to gather data on crack growth rates of cracks in test coupons when subject to underloads. These underloads varied in magnitude, spacing, and consecutive number with the goal of understanding how each of these variables affected the proceeding crack growth rates. After the crack growth studies, detailed fractographic analysis was conducted to describe the fracture surface features common to underloads on both mating fracture surfaces. It was expected that these fractographic features would provide clues as to the underlying mechanisms driving the underload acceleration. In addition, a detailed study of the correlation between surface roughness and the growth associated with underload applications was made. This was examined since reduced surface roughness has been a quoted mechanism associated with underload acceleration reported in the literature. The study conducted here ultimately quantified the amount that underloads accelerate fatigue crack growth of the subsequent load cycles for AA7050-T7451 under the loading conditions presented here. It showed that underload magnitude, spacing, and their consecutive number all play an important role in dictating the amount of acceleration that occurs. Additionally, fracture surface fissures were found to be a common feature that was evident on the fracture surface when underloads were applied at all studied crack lengths, indicating that further investigation of these features may be key to understanding the physical mechanisms behind the underload effect. It was also found that small scale systematic surface roughness had no clear correlation with the application of underloads, indicating that this was not an important mechanism behind the accelerative effects of the underloads. The research presented here provides a step forward in our understanding of the role that underloads can have in small crack propagation. However, there is much scope for future research in this area, including investigating different materials, different cyclic load ratios, and further investigation of the potential underlying mechanisms that cause the underload accelerations. This will be discussed in more detail at the end of the thesis.<p></p>