Abstract
Indentation test is used with growing popularity for the characterization of various materials on different scales. Developed methods are combining the test with computer simulation and inverse analyses to assess material parameters entering into constitutive models. The outputs of such procedures are expressed as evaluation of sought parameters in deterministic sense, while for engineering practice it is desirable to assess also the uncertainty which affects the final estimates resulting from various sources of errors within the identification procedure. In this paper an experimental-numerical method is presented centered on inverse analysis build upon data collected from the indentation test in the form of force-penetration relationship (so-called 'indentation curve'). Recursive simulations are made computationally economical by an 'a priori' model reduction procedure. Resulting inverse problem is solved in a stochastic context using Monte Carlo simulations and non-sequential Extended Kalman filter. Obtained results are presented comparatively as for accuracy and computational efficiency.
Highlights
Indentation tests have been employed for decades for mechanical characterization of materials at diverse scales and in various technological fields
Later the assessment of mechanical parameters entering into constitutive models and necessary for overall inelastic structural analyses became the purpose of this test
Indentation instruments are presently equipped with tools capable to provide the relationship between applied force and the consequent penetration depth
Summary
Indentation tests have been employed for decades for mechanical characterization of materials at diverse scales and in various technological fields. The test is simulated, traditionally by the use of Finite Element Modeling (FEM), while the results of simulation are compared to those measured in the experiment in the form of a “discrepancy function” This function quantifies the difference between computed and measured data, and in a subsequent phase it is minimized with respect to sought parameters by a suitable mathematical programming technique. In order to use measurable quantities endowed by a covariance matrix quantifying their uncertainties, it is required to design an inverse analysis procedure capable to estimate the propagation of this uncertainty up to the assessed parameters To achieve such goal, a stochastic framework needs to be adopted for the resulting inverse problem.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have