Determination of planar crack front geometry based on near-crack surface displacement field obtained from simulated digital image correlation measurements

Determination of planar crack front geometry based on near-crack surface displacement field obtained from simulated digital image correlation measurements

Slagmolen, Mark (TU Delft Mechanical, Maritime and Materials Engineering)

Kaminski, Mirek (mentor)
den Besten, Henk (mentor)
Janssen, Michael (graduation committee)
Synetos, Georgios (mentor)

Delft University of Technology

2017-09-14

Digital image correlation (DIC) is a surface measurement technique that can be applied whilst performing experimental tests. At Delft University of Technology, a fatigue test is to be conducted and DIC will be used to measure surface displacements around the developing crack. These surface displacements are directly related to the crack geometry parameters. These crack geometry parameters can be found numerically via a minimisation algorithm where the difference between the surface displacements measured via DIC and surface displacement found via finite element analysis is minimised. Aim of present study is to perform a preliminary analysis before the experiment is performed in order to assess what is to be expected. It is looked into whether the out-of-plane surface displacement information that has to be taken into account since this influences the DIC hardware to be used during the experiment. Secondly, it has been looked into the noise as consequence of using DIC, and the influence of the noise on the minimisation to be performed. Finally it has been looked at the influence of the DIC noise on the accuracy of the minimisation. Since present study describes a preliminary analysis, simulated DIC surface measurements have been used. A finite element (FE) model of the specimen to be tested has been used in order to perform the study for uniaxial tension only. An algorithm using the FE model to perform the minimisation has been build and tested for simple 2D cases and a simpler 3D first before performing the 3D analysis with geometry considered. The simulated DIC noise taken is 0%, 1%, 2%, 5% and 10% of the mean measured displacement field in order investigate the influence of the noise on the minimisation. Present study showed that neglecting the out-of-plane displacement information introduces an error of around 2x10^-2 millimetres for 1000 MPa uniaxial tension applied. It is assumed that the algorithm used finds the same crack geometry parameters as have been used to construct the simulated DIC surface displacement field in case no noise is applied. It is found that this is not the case in present study. Reason is the algorithm converging to local minima located very close to the global minimum. Present study showed the location of the global minimum becoming less defined when noise applied is increased, i.e. the area in which the global minimum is located becomes larger. Present study shows that the algorithm can convergence towards the target crack geometry parameters. Accuracy of the algorithm as function of the noise has been checked for the full displacement field and the displacement field where out-of-plane information is neglected. Confidence intervals increase for increasing DIC noise, as can be expected. Due to the local minima the algorithm might converge to, it is not possible to draw conclusions on whether the full displacement field should be taken into account in order to have the accuracy required. Therefore, it is strongly recommended to change the algorithm to take extra information into account to make the algorithm converge to the global minimum. Furthermore, it is recommended to improve the numerical model, in order to improve the speed of the algorithm.

simulated DIC
FEM
Minimisation

http://resolver.tudelft.nl/uuid:5ebdd5b3-7217-41c3-9152-e433e8108a6f

Student theses

master thesis

© 2017 Mark Slagmolen