State of the art geotechnical engineering and project designare nowadays rigidly connected to the mindful adaption of the Eurocode. However, questions can be raised on the critical selection of the characteristic values for parameters such as the soil compressibility and permeability. The experiment conducted at Leendert de Boerspolder, where a dyke was controllably brought to failure, poses as an excellent circumstance for investigating this issue. After setting up an FEM model in PLAXIS 2D, as well as orchestrating multiple ULS schemes and comparing the available soil constitutive models, the effects of these characteristic values on the response of the system are explored. Displacement profiles are collated, but more importantly, the safety factors and possible failure mechanisms of every analysis are put under comparison. The outcome is the evaluation of the characteristic value pick on a quantitative and qualitative level. Thus, certain concepts regarding the application of the Eurocode in geotechnical engineering can be tackled or validated, allowing for a critical implementation of the norms.

Insufficient information on soil parameters and their spatial variability pose as a main factors of uncertainty in geotechnical design. When soil response differs from the expected, the notion of inverse analysis becomes relevant; back-calculating the parameter set able to reproduce the monitored observations. Accordingly, its application attempts to clarify the effective soil conditions and allows for an update of the design based on in-situ measurements. The main goal of this thesis is assessing the adaption of available inverse analysis concepts in geotechnical engineering, as well as evaluating the reliability of their outcome. Specifically, the methods are employed on a slope construction problem in order to exhibit their function, compare their efficiency and point out the obstacles encountered in their application.
After selecting the three most competitive approaches (namely: the Markov Chain Monte Carlo, the Genetic Algorithm and the Regularized Particle Filter), a synthetic case of RFEM model simulating a staged slope excavation is composed to test them, thus includes both soil heterogeneity and a pseudo-time scheme. Horizontal displacements at specified mesh points, mimicking the readings of inclinometers, act as the observations. Subsequently, the soil parameters to be identified are the undrained shear strength and the Young's modulus, as they yield the heaviest impact on displacement generation. Essentially, the objective of the inverse analysis is to minimize the error between observations and their RFEM-simulated counterpart. Furthermore, since all methods produce stochastically defined approximations of the parameter fields, a reliability analysis is suitable for their appraisal. Utilizing a Monte Carlo framework, the confidence curves of the results are plotted and judged. In an effort to comply with modern design standards, the reliability level of 95% acts as one of the evaluation criteria. Moreover, a practice-driven example employs the reliability concept in predicting the displacements of the next stage.
Inverse methods are evaluated according to their computational performance, their efficiency in identifying soil parameters, as well as their behavior in the reliability analysis. In this study, the Genetic Algorithm is proven to be the most competent inverse analysis approach. It easily adapts to the problem, requires fewer RFEM runs and provides accurate and reliable results. Additionally, in the course of the project two main hindrances where encountered: the approximation of the strength field when applying a Linearly Elastic – Perfectly Plastic constitutive model and the combination of the Regularized Particle Filter with the RFEM model. Both are thoroughly analyzed and the conditions required for their mitigation are presented with numerical examples. Ultimately the thesis delivers on its research goals by providing insight on inverse method applications. It includes an RFEM setting, that employs authentic elements, and also addresses some practice-related aspects. All in all, this is an effort towards the development of the numerical inverse analysis and hopefully its adoption as part of an integrated geotechnical design.