A methodology for expert knowledge imbrication in mooring system design using Bayesian based optimization

Mooring system design optimisation is a complex problem requiring a specific technical expertise. Because of the large number of parameters influencing the design and their related uncertainty, efficient design methodologies and simplified cost models are unavoidable. This study proposes a methodology for the imbrication of expert knowledge on the design optimisation of mooring systems via Bayesian Optimisation (BO). A Gaussian Process Regression has been used as a surrogate model, which is able to estimate both the cost function and the uncertainty of its own predictions. The methodology has been applied to a simplified use case: the design of a three-line simple catenary mooring system. Results show that BO is able to effectively arrive at an optimum solution while providing valuable information about the whole design space, demonstrating potential of the methodology to deal with uncertainties and enable informed decision-making from early design stages.