Slamming of High-Speed Craft: A Machine Learning and Parametric Study of Slamming Events

TR Number
Date
2022-05-27
Journal Title
Journal ISSN
Volume Title
Publisher
Virginia Tech
Abstract

Slamming loads are the critical structural design load for high-speed craft. In addition to damaging the hull structure, payload, and injuring personnel, slamming events can also significantly limit operating envelopes and decrease performance. To better characterize slamming events and the factors affecting their severity, a parametric study will be carried out in the Virginia Tech Hydroelasticity Lab. This thesis provides the groundwork for this longitudinal project through meticulous analysis of irregular wave tow tank experiments. Through the modification of machine learning techniques and taking inspiration from facial recognition algorithms, key parameters were identified to form an experimental matrix which captures intricacies of the complex interdependent relation of variables in the slamming problem. The independent effects of parameters to be evaluated include hull flexural rigidity, LCG location, heave and surge velocity, and impact trim, angular velocity and acceleration. In preparation for this parametric study, an innovative experimental setup was designed to simulate the impact of a deep-vee planing hull into waves, through a controlled motion slam into calm water. To provide a baseline to compare data from future controlled motion experiments to, a model drop experiment was completed to characterize the relationships of impact velocity and trim to slamming event severity. During this experiment, the position, acceleration, strain, and pressure were measured. These measurements illustrated a decrease in peak acceleration, pressure, and strain magnitude with an increase in impact trim. Additionally, as trim was increased a delay in the time of peak magnitude for all measurements was observed. These results are attributed to the change in buoyancy with the change in impact angle. At non-zero angles of trim, a pitching moment was generated by the misalignment of the longitudinal center of buoyancy and center of gravity. This moment caused racking in the setup which was observed in the acceleration time histories immediately after impact. This finding furthers the need to investigate the angular velocity and acceleration of the model at impact, through the proposed series of experiments, as they are crucial naturally occurring motions inherent to slamming events.

Description
Keywords
High-Speed Craft, Slamming, Machine Learning, Water Entry
Citation
Collections