Universitätsbibliothek
Weimar Open Access

A novel parameter identification approach for buffer elements involving complex coupled thermo-hydro-mechanical analyses

This study contributes to the identification of coupled THM constitutive model parameters via back analysis against information-rich experiments. A sampling based back analysis approach is proposed comprising both the model parameter identification and the assessment of the reliability of identified model parameters. The results obtained in the context of buffer elements indicate that sensitive parameter estimates generally obey the normal distribution. According to the sensitivity of the parameters and the probability distribution of the samples we can provide confidence intervals for the estimated parameters and thus allow a qualitative estimation on the identified parameters which are in future work used as inputs for prognosis computations of buffer elements. These elements play e.g. an important role in the design of nuclear waste repositories.

Many structures in different engineering applications suffer from cracking. In order to make reliable prognosis about the serviceability of those structures it is of utmost importance to identify cracks as precisely as possible by non-destructive testing. A novel approach (XIGA), which combines the Isogeometric Analysis (IGA) and the Extended Finite Element Method (XFEM) is used for the forward problem, namely the analysis of a cracked material, see [1]. Applying the NURBS (Non-Uniform Rational B-Spline) based approach from IGA together with the XFEM allows to describe effectively arbitrarily shaped cracks and avoids the necessity of remeshing during the crack identification problem. We want to exploit these advantages for the inverse problem of detecting existing cracks by non-destructive testing, see e.g. [2]. The quality of the reconstructed cracks however depends on two major issues, namely the quality of the measured data (measurement error) and the discretization of the crack model. The first one will be taken into account by applying regularizing methods with a posteriori stopping criteria. The second one is critical in the sense that too few degrees of freedom, i.e. the number of control points of the NURBS, do not allow for a precise description of the crack. An increased number of control points, however, increases the number of unknowns in the inverse analysis and intensifies the ill-posedness. The trade-off between accuracy and stability is aimed to be found by applying an inverse multilevel algorithm [3, 4] where the identification is started with short knot vectors which successively will be enlarged during the identification process.

Polymer modification of mortar and concrete is a widely used technique in order to improve their durability properties. Hitherto, the main application fields of such materials are repair and restoration of buildings. However, due to the constant increment of service life requirements and the cost efficiency, polymer modified concrete (PCC) is also used for construction purposes. Therefore, there is a demand for studying the mechanical properties of PCC and entitative differences compared to conventional concrete (CC). It is significant to investigate whether all the assumed hypotheses and existing analytical formulations about CC are also valid for PCC. In the present study, analytical models available in the literature are evaluated. These models are used for estimating mechanical properties of concrete. The investigated property in this study is the modulus of elasticity, which is estimated with respect to the value of compressive strength. One existing database was extended and adapted for polymer-modified concrete mixtures along with their experimentally measured mechanical properties. Based on the indexed data a comparison between model predictions and experiments was conducted by calculation of forecast errors.

In the field of engineering, surrogate models are commonly used for approximating the behavior of a physical phenomenon in order to reduce the computational costs. Generally, a surrogate model is created based on a set of training data, where a typical method for the statistical design is the Latin hypercube sampling (LHS). Even though a space filling distribution of the training data is reached, the sampling process takes no information on the underlying behavior of the physical phenomenon into account and new data cannot be sampled in the same distribution if the approximation quality is not sufficient. Therefore, in this study we present a novel adaptive sampling method based on a specific surrogate model, the least-squares support vector regresson. The adaptive sampling method generates training data based on the uncertainty in local prognosis capabilities of the surrogate model - areas of higher uncertainty require more sample data. The approach offers a cost efficient calculation due to the properties of the least-squares support vector regression. The opportunities of the adaptive sampling method are proven in comparison with the LHS on different analytical examples. Furthermore, the adaptive sampling method is applied to the calculation of global sensitivity values according to Sobol, where it shows faster convergence than the LHS method. With the applications in this paper it is shown that the presented adaptive sampling method improves the estimation of global sensitivity values, hence reducing the overall computational costs visibly.