Multidisciplinary Risk-Adaptive Set-Based Design of Supercavitating Hydrofoils

Abstract

This paper presents a multidisciplinary framework for a design under uncertainty that leverages surrogate models trained using risk-adaptive statistical learning. The hydrodynamic and structural features of a supercavitating hydrofoil designed to support the displacement of ultrafast vessels are described. Surrogate models predicting superquantile risk of multiple competing quantities of interest are constructed, which adapt to the risk averseness level of a particular design stage. To accurately train surrogate models, simulation data provided by two computational frameworks are leveraged, predicting hydrodynamic and solid mechanics quantities of interest. Each framework provides data at low- and high-fidelity levels through a multiresolution approach for the hydrodynamic quantities (Reynolds-averaged Navier–Stokes at high and low resolutions), and a multifidelity approach for the solid mechanics predictions (full three-dimensional and a simple beam). The framework is demonstrated through the design of a complex supercavitating hydrofoil, but the method is generally applicable to the design of a complex physical system under uncertainty.