Multi-Fidelity MMG-Model for Digital Design of High-Speed Small Craft

This paper introduces advanced methods for ship maneuvering predictions in the context of digital design of high-speed craft. While these methods are not yet commonly used by naval architects, it is believed that the present paper provides guidance to naval architects as to the way forward for advanced tools in ship design. Specifically, the paper presents the integration of the ship Maneuvering Modeling Group (MMG) model with multi-fidelity (MF) computational fluid dynamics (CFD) input data, particularly tailored for the analysis of high-speed small craft within the digital design paradigm. An uncertainty propagation approach via Monte Carlo sampling is introduced and discussed, covering from CFD computations to final MF-MMG predictions, passing trough the MF surrogate model. The investigation focuses on the straight-ahead self-propulsion of a standardized hull form, namely a 40 ft Generic Prismatic Planing Hull (GPPH), offering the opportunity to test both the MMG and the MF approaches against a challenging real world application. The study demonstrates how multi-fidelity methods could effectively reduce computational costs while maintaining predictive accuracy. Results show how a deterministic MMG model trained by MF-CFD simulations predicts a speed at maximum engine power above the required threshold, whereas the stochastic model (i.e., MF-MMG with uncertainty prediction) indicates a 95% confidence of achieving only a speed below the threshold at maximum power. Similarly, the speed that the boat can achieve at maximum shaft speed is larger than the threshold if evaluated by the deterministic model, whereas falls below the threshold if we consider the 95% confidence band provided by the stochastic model.