Österreichische Akademie der Wissenschaften
The determination of materials properties with nanoindentation, exceeding Oliver-Pharr’s method, which allows the precise determination of elastic modulus, is non-trivial. Since the technique was introduced more than 30 years ago, many publications have provided mathematical frameworks to estimate more and more material properties. Many leverage dimensional analysis [1] and already, for more than two decades, machine learning (ML) [2, 3]. The models usually allow the determination of parameters such as yield stress and, depending on the used material model, a hardening exponent. However, most of these studies are based on finite element simulations of perfectly sharp indenters; therefore, the surrogate model will not capture the tip wear effects.
A method will be presented that allows the determination of the tip wear of a Berkovich nanoindenter tip solely from the load-displacement (P-h) curve. A Residual Multi-Fidelity Neural Network was used to learn from 2D- and 3D finite element simulation and experimental data. The experimental tip wear was monitored over 30,000 indents in a vast amount of materials. The experimental data showed that an indentation tip not only became blunt over its lifetime but also, under the right conditions, started to get sharper again. Features were based on parameters known from studies using dimensional analysis and extracted from the P-h curve. The surrogate model performed well on experimental data, enabling in-situ tip radius characterisation for the first time [4].
[1] M. Dao, N. Chollacoop, K.J. Van Vliet, T.A. Venkatesh, S. Suresh, Acta Materialia, 2001, Volume 49, Issue 19, Pages 3899-3918
[2] N. Huber, Ch. Tsakmakis, Journal of the Mechanics and Physics of Solids, 1999, Volume 47, Issue 7,
Pages 1569-1588,
[3] L. Lu, M. Dao, P. Kumar, U. Ramamurty, G.E. Karniadakis, S. Suresh, Proc. Natl. Acad. Sci. 117, 2020, 7052.
[4] C. O. W. Trost, S. Žák, S. Schaffer, C. Saringer, L. Exl, M.J. Cordill, The Journal of The Minerals, Metals & Materials Society (TMS), 2022, JOM 74, 2195–2205
Abstract
Erwerben Sie einen Zugang, um dieses Dokument anzusehen.
© 2025