In this talk, we will present a new approach to develop an information-driven, learning-based framework for Design under Uncertainty, and optimal operation of ultra-high-speed marine vessels, doubling their speed compared to existing vessels. The framework is very general and scalable with respect to high number of decision variables and uncertain parameters, it integrates design of physical components with control algorithms, and it achieves holistic design that will replace traditional fragmented processes by set-based design. A key concept is the seamless integration of simulation and experimental data of variable fidelity. An equally important focus area is the risk assessment at all design stages through properly defined risk measures, especially super-quantile risk measures that allow us to account for parametric uncertainty, inaccuracies in surrogate and physical models, and uncertainty in the learning process. Taken together, these modern concepts lead to a new design paradigm, namely the risk-averse set based design that we will use to propose a new high-speed marine vessel based on super-cavitating hydrofoils that will travel at 120 knots, which is a record speed, twice higher than current limits.

This works is supported by the DARPA EQUiPS program on Uncertainty Quantification.