Polymorphic uncertainty modeling, evaluation and quantification for fluid saturated soil and earth structures

点击翻译按钮获取中文摘要

In the dimensioning and assessment of earth structures, the main difficulties lie less in providing a suitable deterministic physical calculation model, but in dealing with fuzzy data on material parameters, initial and boundary conditions. For this purpose efficient numerical and probabilistic tools are needed. In earth structures the coupled non-linear behavior of the soil and the time-varying microstructure, e. g. in erosion processes, additionally aggravate a reliable design. The consequences are unforeseen damage or uneconomic dimensions. The aim of the project is therefore the development of numerical and probabilistic methods for the improved design of structures with uncertain properties of soil and earth structures. The polymorphic uncertainties (PU) result from the scatter or incompleteness in the material parameters, the initial and boundary conditions and inaccuracy of the modeling and numerical calculation.The multi-physical soil response behavior is described in this project by means of the Theory of porous Media (TPM) and numerically evaluated within the framework of the finite element method (FEM) for specific initial boundary value problems. As an extension to the first funding period, the model is extended to describe typical mechanisms of soil failure, including the phenomena of non-linear expansion, plasticity and erosion processes. As representative benchmark studies tectonic disorders as well as the hydraulic ground failure will be investigated.Based on the results of the first funding period of the SPP1886, the variational sensitivity analysis (VSA) for the TPM and its well interpretable evaluation by the local and global impact analysis shall be used as a priori knowledge for probabilistic methods. Thus, an improvement of the Bayesian sensitivity analysis can be achieved. Furthermore, the parameter space to be investigated is to be considerably reduced in advance by the VSA in a Monte Carlo (MC) analysis.In addition, fuzzy arithmetic is used to adequately model epistemic uncertainties. The combination of all approaches should combine their strengths and provide a holistic model for PU quantification. The computational efficiency can be improved by the Kriging approach, a meta-model that reduces the number of evaluations necessary for an analysis. The individual evaluations are accelerated by a model reduction method (MOR) using the discrete empirical interpolation method (DEIM).Finally, a uniform digital tool will be created, which can be flexibly and efficiently used for decision making on uncertain data for the numerical design of soil and earth structures. For this, interface problems between the different programs and programming languages have to be solved.