engleski

Identification of the mechanical properties of nodular graphite cast iron via multiaxial tests

The majority of engineering structures is subjected to various and complex in-service loading regimes. In order to satisfy the construction demands and increase the reliability of structures resulted in the development of new engineering elements. Furthermore, new technologies and materials are also introduced. This trend has stimulated in the last few decades scientists to describe the material behaviour with more reliable and complex constitutive laws. The latter ones require advanced methods for the identification and validation of material parameters. The material response was initially observed with contact measurement techniques (e.g., strain gauges, extensometers ) yielding data locally at just few points on the loaded specimen/structure. Measurements performed with classical methods give a global overview of the material behaviour. Hence, localised phenomena (e.g., cracks, heterogeneous strain fields, shear bands) are not captured due to the lack of measured degrees of freedom. These limitations are successfully overcome with the development of noncontact full-field measurement techniques. Among different principles Digital Image Correlation (DIC) is the most widely used in the experimental mechanics community. Over the years DIC surpassed the initial application of measuring displacement/strain fields to become a powerful tool for identification of mechanical properties, verification of numerical simulations, detection and monitoring of local phenomena (e.g., onset of plasticity, crack initiation and propagation). Originally the applied principle of DIC was a local approach based on successive correlations over small zones of interest (ZOIs) where displacements are calculated independently from each other. Detached ZOIs are favourable regarding the time cost since the computations may be run in parallel. However, it is necessary to emphasize that the assumption of field continuity defined by a global DIC approach, outperforms the local ones in terms of displacement resolution. As opposed to local approach, global DIC considers the whole region of interest (ROI) using a large number of Degrees of Freedom (DOFs) and minimizes the difference between reference and deformed configurations. Hence, global DIC was chosen as a full-field measurement technique since lower displacement resolution is possible. Furthermore, coupling the latter with Finite Element analyses is straightforward for identification and validation purposes. Within this thesis, the identification and validation of nonlinear models describing the behaviour of nodular graphite cast iron is carried out. In general the material parameters are obtained by iterative or direct identification procedures. Finite Element Model Updating (FEMU) procedures are used herein as the identification tool. The aim is to analyse the damage law coupled with the elastoplastic behaviour of graphite cast iron in the identification process for simple and complex loading regimes. First, the identification is performed on uniaxial experiments with two different loading histories, namely monotonic and cyclic. A FEMU- UF procedure is used to determine parameters of isotropic and kinematic hardening laws. Furthermore, the parameters describing damage coupled with isotropic hardening are obtained. The identification results of the same constitutive model for different uniaxial loading histories yield different parameters. Force residuals as the criterion for estimation of the most reliable behaviour lead to the lowest values when coupled elastoplastic and damage law is used to model cyclic loading regime. In order to determine microdamage mechanisms an in-situ uniaxial cyclic experiment is monitored via X-ray tomography. For the investigated SG cast iron it is detected via digital volume correlation that debonding between the matrix and the nodules causes damage phenomena. Second, in-plane biaxial experiments are conducted. Cruciform samples are subjected to two different loading regimes, namely proportional (i.e., equibiaxial) and nonproportional (referred to as “snail”). Since the microstructure of cast iron is well suited to correlation techniques, experiments are monitored on two scales (i.e., macro- and mesoscales). The challenge related with mesoscale observations implies that conventional or even standard global DIC approaches fail to accurately capture the displacement fields. In order to overcome such issues a regularized DIC algorithm is introduced and validated. The results observed on the two scales show a very good agreement when two opposite sides of the sample are simultaneously analysed. The identification of isotropic and kinematic hardening parameters is performed for the three tests. The constitutive behaviour is better captured with kinematic hardening since the sample was subjected to cyclic loading histories. Higher force residuals are obtained in the “snail” experiments when compared with the equibiaxial tests. This shows that non-proportional loading regimes are more damaging and complex to analyse, and that even more advanced models need to be considered. Last, two series of fatigue experiments are presented. A first series of five experiments is performed with an equibiaxial history, and a second series of five “snail” loadings. The goal is to evaluate the fatigue lifetime for the two loading paths in order to evaluate the more damaging one. Two-scale DIC is performed to measure displacement fields of both sample sides. From the mesoscale measurements, the crack networks are detected and quantified with respect to the number of cycles. Correlation and mechanical residuals are used to monitor the crack network initiation and growth since the strain data could not be directly used. The analysed fatigue results show that lower lifetimes are to be expected from more complex loading histories.

nonlinear material behaviour ; full-field measurement ; identification ; cast iron ; multiaxial loading ; crack networks ; regularization ; tomography

Medunarodni dvojni doktorat znanosti