Surrogate modeling of hydrodynamic forces between multiple floating bodies through a hierarchical interaction decomposition

Abstract

Efficient estimation of wave interactions between multiple floating bodies (diffraction forces and radiation coefficients) is of importance to many applications, with a key one being the layout optimization of arrays of wave energy converter. The computational complexity for this estimation dramatically increases for configurations with large number of bodies. To address this challenge, a data-driven surrogate modeling implementation is discussed in this paper. The foundation of the approach is an innovative application of the many-body expansion principle that overcomes the curse of dimensionality for the surrogate model development. Instead of using a single surrogate model to predict the hydrodynamic characteristics of the multi-body configuration, multiple surrogate models corresponding to clusters with fewer bodies are employed. These lower-order surrogate models can be developed at a substantial smaller computational cost, especially for the first terms of the many-body expansion that contribute dominantly to the total hydrodynamic characteristics. Additional enhancements for the surrogate modeling implementation pertain to the characterization of the surrogate model input and output, exploiting symmetry and invariance principles for the hydrodynamic problem. Using Kriging as surrogate model, the approach is demonstrated for predicting the hydrodynamic characteristics for an array of vertical axisymmetric bodies, but it is extendable to similar multi-body interaction problems or other surrogate modeling techniques. Several numerical case studies are finally presented, showcasing the computational efficiency and prediction accuracy of the proposed method, as well as its scalability to arrays of large size, and the advantages it can offer within the context of layout optimization for wave energy converters.