Composite structures subjected to underwater explosive loadings: A comprehensive review

Abstract

Composites materials and structures are increasingly used to replace conventional materials in civilian and defence-related maritime transportation and infrastructure such as naval vessels, submarines, civilian ships, and oil platforms for its better performance-to-weight ratio and electro-magnetic signature control. However, when subjected to under water explosions (UNDEX), navel composite structures experience highly nonlinear deformations and damages. Such transient deformation phenomena of composites and associated multiscale damages have been a subject of research for many years. This review aims to provide historical and methodological overviews of significant research and contributions in this area over the last 20 years from experimental programs, modelling approaches, post-mortem analysis techniques, analytical approximation and recently emerging area of data-led predictive simulations. UNDEX event is often described by a series of events including (a) the formation of the arriving shock wave, (b) the attenuation of the initial shock wave, (c) development of cavitation due to the reflected tension wave from free surface or the structural obstacles, (d) fluid-structure interaction-induced deformation and associated (e) cavitation coalescence and collapse. Such interconnected dynamic events and their influences on the behaviours of composite structures are subjected to extensive research and therefore summarised in this review work to highlight state-of-the-art field and laboratory-scaled experimental programs including investigations on low temperature and cavitation’s influences. Furthermore, the ongoing increase in the computing power and the development of advanced numerical methods have made it possible for multiscale and multi-physics simulations capturing the complex fluid dynamics associated with UNDEX. Over ten different modelling approaches, hydrocodes and their hybrid combination are summarised and discussed for potential applications. Review on current computational approaches also reveals the shortcomings of predictive modelling due to unavoidable simplifications, empirical assumptions on limited experimental data. Therefore, this work also provides a brief discussion on how data-led modelling approach such as artificial neural networks or deep learning, which is based largely on experimental data, could provide powerful assistance to analytical and deterministic numerical analysis.