Inverse analysis and microstructure effects in nanoindentation testing.

Inverse analysis of nanoindentation data has attracted increasing interest in industry due to its ability to estimate the bulk tensile properties of materials and potentially offers an alternative technique to conventional characterisation methods. Inverse analysis of nanoindentation data is particularly valuable in applications where conventional techniques are not suitable due to either the scale of characterisation (very small regions) or because the testing is expensive and time consuming. Despite using best practices to minimise sources of error in the experimental data, given the scale of the indentations, the heterogeneity of material microstructure can create significant variability in the data, ultimately affecting the reliability of the inverse analysis solution. This thesis proposes and discusses pragmatic approaches to mitigate the effects of material heterogeneity on the accuracy of the inverse problem solution as well as of nanoindentation data in general. The work has involved finite element analysis modelling, nanoindentation and tensile testing. One mitigation approach consisted in the implementation and verification of a new ‘multi-objective’ function inverse analysis methodology where the bias of selecting only one experimental nanoindentation curve as representative of the homogenised response of the material is overcome. The new approach uses all the experimental curves generated from a grid of nanoindentations and employs a weighted averaging procedure. This methodology was applied to S355 steel samples through recording nanoindentation and tensile test data. Despite the variation present in the experimental nanoindentation load-depth curves, this being in the order of 13%, the ‘multi-objective’ function approach was found to estimate the tensile parameters with an error margin as low as 3-6% compared to an error margin of 9-20% for the conventional method. A framework of activities was also undertaken to monitor the variation of the measured nanoindentation properties (e.g. hardness) as function of the indentation depth, in relation to the average grain size of the material. Commercial purity aluminium 1050 samples (with varying average grain sizes) and S355 steel were employed as test materials. These results in addition to those from other materials were used to construct a look-up plot of the hardness COV values as function of the normalised nanoindentation depths (normalised with respect to the average grain diameter). The plot is based on upper and lower bound curves and intends to provide guidance on the selection of the nanoindentation testing parameters to minimise the variability of the indentation response