Abstract

In this paper, we seek to minimize the cost of the anchoring system of a floating offshore wind turbine under reliability constraints. Taking into account the uncertainties on the model, on the resistance threshold and on the environmental conditions implies constraints expressed as probabilities depending on random vectors and a piecewise stationary Gaussian process. The main difficulty of the studied problem is to compute these probabilities since reliability methods require many calls to the simulator of the system. We propose in this paper a two-step methodology allowing to solve the optimization problem with a reasonable number of calls to the simulator. First, we exploit the properties of the problem to reformulate the constraints into easier to compute ones. Then we propose a new approach based on adaptive kriging well suited to the reformulated problem: AK-ECO.