Public defence in Engineering Physics, M.Sc. Juha Savolainen

Title of the doctoral thesis: Elastic interfaces in disordered media and their application to fatigue fractures

Doctoral student: Juha Savolainen
Opponent: Prof. Eduard Vives Santa-eulalia, University of Barcelona, Spain
Custos: Prof. Mikko Alava, Aalto University School of Science, Department of Applied Physics

Thesis available for public display 10 days prior to the defence at: https://aaltodoc.aalto.fi/doc_public/eonly/riiputus/

The failure of materials under repetitive stress is called fatigue. Fatigue is usually caused by a slowly growing crack, which eventually leads to the sudden rupture of the whole material. Slowly growing cracks have often been modelled using the framework of elastic interfaces in statistical physics. This dissertation looks at whether the model can be applied to fatigue. 

The work consists of three parts. Two look at the empirical Paris-Erdogan law, which describes the slow crack growth phase in fatigue. First, we studied crack growth experimentally and looked at the statistical distribution of crack growth jumps. The distribution was similar for cracks that advanced at different velocities and did not match distributions in previous literature. We also studied theoretically the connection of the elastic interface model and fatigue. Although the model has been used for crack growth under a constant loading, it clearly disagrees with fatigue, where the loading is varied. We found that the model works significantly better for fatigue if certain material parameters found in the literature have different values than what have been suggested previously. 

The third part of the work looks at the interface model and uses it to analyse random movement and noisy signals. It is common to leave out the smallest signals in experimental data, as they are difficult to separate from noise. Events in random movement, however, can consist of both large and small signals, and leaving out the small signals has been found to change the remaining events. We continued on previous work and found that several different event distributions change with a signal detection threshold. We also looked at the possible connection of the elastic interface model to earthquakes, which has been discussed in previous studies. We found that the connection results from statistical errors. However, ignoring small signals in data makes the data resemble more seismological behaviour.

Contact details:

Doctoral theses in the School of Science: https://aaltodoc.aalto.fi/handle/123456789/52