Date of Award

Spring 2023

Document Type

Open Access Thesis


Mechanical Engineering

First Advisor

Austin Downey


Naval ship structures such as support trusses, hull sections, driving machinery, and load-bearing beams are subjected to various damage states that develop on short-term (i.e. impact) and long-term (i.e. fatigue) timescales. Naval structures fitted with a structural health monitoring system with damage detection abilities will enable appropriate real-time adjustments to the ships’ posture and control policies and thus have increased survivability and lethality. A digital twin can provide real-time condition assessment of naval structures when conjoined with a decision-making framework will increase naval ship survivability through informed response management. A fundamental challenge for the development of digital twins is reliable methodology advancement that is able to distinguish short-term damage from long-term damage states. Moreover, the methodology advancement must efficiently update vast amounts of data into data-driven or physics-based models while efficiently computing on the naval ships resourced constrained environments and meeting stringent latency constraints. This work details the numerical and experimental validation of a particularly designed framework for multi-event model updating that meets stringent latency constraints while computing on a system with limited computational resources. The proposed framework tracks impact and fatigue structural damage through a particle swarm implementation that represents numerical models with various input parameters with set latency constraints and available computational resources. The proposed methodology is used to conduct experimental validation using data measured from a structural testbed designed to provide representative ship responses subjected to impact and fatigue events while considering a pre-determined wave loading condition. Results demonstrate that a structural physics-based model can be updated in real-time while differentiating plastic deformation caused by impact events from continuous fatigue crack growth. Latency effects, resource-constrained computation accuracy, parameter optimization, and process robustness of the proposed framework are quantified and discussed further in this work.