Performance Assessment of Warped Bottom Planing Hulls Using Machine Learning Techniques

Determining resistance and trim of planing hulls in the early design phase has traditionally relied on semi-empirical prediction models. However, in the case of hulls with warped bottoms, resistance predictions frequently require validation with tank testing or numerical simulations. The present work focuses on predicting resistance and trim of planing hulls using five machine learning methods trained using data from available experimental model tests of warped bottom hulls. This database contains large systematic data from Series 50, US Coast Guard, and Naples Systematic Series which cover a wide range of ship particulars characteristics Furthermore, the dataset is analyzed following the standard ITTC methodology and incorporates speed-dependent wetted length to extrapolate predictions for vessels of varying full-scale sizes. Afterward, the models are validated against tests conducted on three warped planing hulls and one prismatic monohull developed by the University of Naples. Through analysis, the study identifies improvements in resistance predictions but with limitations when planing hulls have a transom deadrise between 5 and 12 degrees.