Keywords: monopile, probabilistic analysis, Abaqus, failure probability, kriging.
The analysis and design of offshore monopile foundations are generally undertaken using deterministic approaches. In this paper, a probabilistic analysis is performed taking into account the soil spatial variability. The aim is to compute the failure probability against exceeding a threshold value on the pile head displacement. The mechanical model employed for the computation of the monopile head displacement is based on 3D numerical simulations making use of Abaqus finite-element software. The soil is assumed to be an elastic perfectly plastic material obeying Tresca failure criterion. The soil undrained cohesion and Young modulus are considered as log-normal random fields.
As it is well known, numerical 3D deterministic models of offshore monopile foundations are computationally-expensive and thus they present a great obstacle to the use of the conventional Monte Carlo Simulation (MCS) methodology for the probabilistic analysis. Furthermore, the study of spatially varying soils with small values of the autocorrelation distances significantly increases the computational effort. To overcome this shortcoming, a reliable and efficient probabilistic model called Global Sensitivity Analysis enhanced Surrogate (GSAS) modeling is proposed. This model is based on kriging metamodeling. The essential issues in the classical kriging-based approaches such as the Active learning method combining Kriging and Monte Carlo Simulations MCS (named AK-MCS method) are that both the choice of a ‘best new sample’ for the enrichment process and the stopping criterion concerning the addition of a new training sample are defined from the perspective of individual responses, which may lead to some extra evaluations of unnecessary added training samples. In the proposed GSAS method, both the convergence criterion and the strategy of selecting new training samples are defined from the perspective of reliability estimate instead of individual responses of MCS samples. Probabilistic numerical results are then presented and discussed.
This work was carried out within the framework of the WEAMEC, West Atlantic Marine Energy Community, and with funding from the CARENE.