Filtern
Dokumenttyp
- Vortrag (9)
- Beitrag zu einem Tagungsband (3)
- Zeitschriftenartikel (1)
Sprache
- Englisch (13)
Schlagworte
- Gaussian process regression (7)
- Geotechnical site-characterization (5)
- CPT (3)
- Monopile design (3)
- Probabilistic (3)
- Bayesian Inference (2)
- Boulder impact (2)
- CPT data (2)
- Cone Penetration Tests (2)
- Gaussian Process (2)
Organisationseinheit der BAM
- 7 Bauwerkssicherheit (13)
- 7.2 Ingenieurbau (13)
Eingeladener Vortrag (wissenschaftliche Konferenzen)
- nein (9)
The aim of the study is to infer the soil stratification from the provided CPT and borehole data. We infer the soil type at any location within the domain of interest from the SBT index Ic (Robertson, 2009). This index can be directly related to the CPT data through an empirical correlation model. In addition, the soil classes contained in the borehole logs can be expressed as bounds on Ic. A log-transformation was applied to Ic, Y = ln(Ic), and Y was modelled by a 3D Random Field, with a fully Bayesian hierarchical Gaussian Process model to explicitly capture uncertainties.
This work was presented in the two-day workshop ‘Reliability-based methods in geotechnics’ organized by BAM and DGGT AK 2.15 in cooperation with the DGGT Academy. The presentation provides an overview of the sources of uncertainty in geotechnical analyses. A probabilistic 3D ground model of the cone-tip resistance from CPTs is developed utilizing sparse CPT data from a real offshore wind farm site in the North Sea. The predictive cone tip resistance model is used to derive a design property, such as the friction angle, using geotechnical transformation models. The friction angle profiles are then used to show an example of monopile design under uncertainty.
Accurate prediction of subsurface stratigraphy and geotechnical properties, along with quantification of associated uncertainties, is essential for improving the design and assessment of geotechnical structures. Several studies have utilized indirect data from Cone Penetration Tests (CPTs) and employed statistical and Machine Learning methods to quantify the geological and geotechnical uncertainty. Incorporating direct borehole data can reduce uncertainties. This study proposes a computationally efficient multivariate Gaussian Process model that utilizes site-specific data and: (i) jointly models multiple categorical (USCS labels) and continuous CPT variables, (ii) learns a non-separable covariance structure leveraging the Linear Model of Coregionalization, and (iii) predicts a USCS based stratigraphy and CPT parameters at any location within the 3D domain. The results demonstrate that integrating geotechnical and geological data into a unified model yields more reliable predictions of subsurface stratification, enabling the parallel interpretation of both USCS classification and CPT profiles. Importantly, the model demonstrates its potential to integrate multiple variables from different sources and data types, contributing to the advancement of methodologies for the joint modeling of geotechnical, geological, and geophysical data.
Tip damage of monopiles from boulder collisions during installation has emerged as a critical design issue, primarily due to the combination of the widespread use of large-diameter piles and the complexity of the ground conditions at the available sites. Recently, a framework relating potential pile tip damage to cone tip resistance from Cone Penetration Tests (CPTs) has been proposed in the literature. Gaussian processes are powerful stochastic models that enable probabilistic spatial interpolation of soil data at any location within a site. On this basis, this study utilizes sparse CPT data from a site in the North Sea, to first develop an efficient Gaussian process regression model, which is used to derive a three-dimensional (3D) probabilistic predictive map of the cone tip resistance. Assuming deterministic loading conditions and a factual collision with a boulder of pre-defined properties, the cone tip resistance predictive model is subsequently used for a probabilistic preliminary assessment of potential pile tip damage. Results of the analysis are realistic 3D probability maps of potential damage that aim to support engineering judgment and contribute towards cost-effective site investigation planning and offshore wind farm design.
Tip damage of monopiles from boulder collisions during installation has emerged as a critical design issue, primarily due to the combination of the widespread use of large-diameter piles and the complexity of the ground conditions at the available sites. Recently, a framework relating potential pile tip damage to cone tip resistance from Cone Penetration Tests (CPTs) has been proposed in the literature. Gaussian processes are powerful stochastic models that enable probabilistic spatial interpolation of soil data at any location within a site. On this basis, this study utilizes sparse CPT data from a site in the North Sea, to first develop an efficient Gaussian process regression model, which is used to derive a three-dimensional
(3D) probabilistic predictive map of the cone tip resistance. Assuming deterministic loading conditions and a factual collision with a boulder of pre-defined properties, the cone tip resistance predictive model is subsequently used for a probabilistic preliminary assessment of potential pile tip damage. Results of the analysis are realistic 3D probability maps of potential damage that aim to support engineering judgment and contribute towards cost-effective site investigation planning and offshore wind farm design.
This work was presented at the the 2nd Workshop on Future of Machine Learning in Geotechnics (2FOMLIG) & the 5th Machine Learning in Geotechnics Dialogue (5MLIGD), in Chengdu, China. In this study, we propose an efficient multivariate Gaussian process regression model, utilizing the Linear Model of Coregionalization, stochastic variational inference and Dirichlet-based transformations, to jointly model continuous CPT and categorical (USCS) borehole variables. The predictive performance of the model is assessed using a real dataset from a site located in Christchurch, New Zealand.
This study utilizes cone penetration testing data data from a real offshore windfarm project in the North Sea and presents a method for incorporating both, statistical and spatial uncertainties, in a reliability-based assessment of monopile foundations. Initially, a Gaussian Process regression model is constructed to predict a 3D map of the cone tip resistance and the associated uncertainties in the predictions and the hyperparameters, leveraging Markov Chain Monte Carlo methods. The CPT-based prediction is combined with a correlation derived from data collected at a nearby site to predict the probability distribution of the friction angle at a test location, which is subsequently used to evaluate the probability of failure for a monopile foundation with a finite element model.
Quantifying uncertainties in subsurface properties and stratigraphy can lead to better understanding of the ground conditions and enhance the design and assessment of geotechnical structures. Several studies have utilized Cone Penetration Test (CPT) data and employed Bayesian and Machine Learning methods to quantify the geological uncertainty, based on the Robertson’s soil classification charts and the Soil Behaviour Type Index (Ic). The incorporation of borehole data can reduce the stratigraphic uncertainty. Significant challenges can arise, however, mainly due to the intrinsic differences between field and laboratory-based soil classification systems, which can potentially lead to inconsistent soil classification. To this end, this study proposes a multivariate Gaussian Process model that utilizes site-specific data and: i) jointly models multiple categorical (USCS labels) and continuous (Ic) variables, ii) learns a (shared) spatial correlation structure and the betweenoutputs covariance, and iii) produces two types of dependent classification outputs. The results indicate that the integration of geotechnical and geological information into a unified model can provide more reliable predictions of the subsurface stratification, by allowing simultaneous interpretation of USCS and Ic profiles. Importantly, the model demonstrates the potential to integrate multiple variables of different types, aiming to contribute to the development of a methodology for joint modeling of geotechnical, geological and geophysical data.
This presentation is part of the Engineering Risk Analysis group open seminars. It aims to showcase the results of an ongoing study centered on developing a novel probabilistic methodology for 3D geotechnical site characterization. This methodology integrates data from Cone Penetration Tests (CPTs) and categorical borehole data. The presentation covers the mathematical details of the proposed Multivariate Gaussian Process model and demonstrates its application to a real geotechnical site in New Zealand.
This presentation discusses the potentials of probabilistic methods in offshore foundation installation, from the perspective of probabilistic ground models and data-driven site characterization. We discuss about methodologies for utilizing site-specific geotechnical (CPT) and geological data, aiming to construct an integrated ground model that can predict stratigraphic profiles and useful for geotechnical design parameters at any location within a 3D domain. The predicted parameters and stratigraphy are then used to predict the probability of potential pile tip damage, upon collision with a boulder.
