Vibration-based structural assessment with damage-detection-oriented models

Abstract

Most structural systems are susceptible to damage in their service lives. Structural damage generally produces changes in the dynamic characteristics of the structure. This fact has been widely used as the physical basis for the vibration-based non-destructive damage detection techniques. The basis for vibration-based damage detection appears intuitive, but its practical application still encounters significant technical challenges over the past few decades. The most important challenge is the fact that the measured global dynamic properties of the structure may not be sensitive enough to the local change in structural properties (stiffness, mass and damping) caused by damage. Since the damage may be caused by various factors such as operating loads, impact, fatigue, corrosion etc., and is of various patterns, and, in turn, affects the structural behavior very differently, another difficulty is how to model the damaged structural member so that modal uncertainty due to the modeling error is less than the modal perturbations caused by damage and it will not corrode the damage detection result. The main work presented in this dissertation includes: 1) a methodology of vibration-based damage detection for damage location indication and damage quantification is developed; 2) typical damage-detection-oriented models for skeletal structures and plate-like structures are proposed and incorporated with the developed method of damage identification. Damage indices are formulated based on modal flexibility and its derivatives and are generalized from the ones for one-dimensional structures to two-dimensional plate-like structures. The contributions in this part of research are: 1. To conduct a comprehensive comparison on the constructed damage indices through numerical examples with different support conditions, random noise, mode truncation and sensor sparsity. 2. To present a new curvature estimation approach based on the Chebyshev Polynomial approximation for the continuous distribution of Uniform Load Surface Curvature (ULSC) over a plate. 3. To develop a model-updating-based damage quantification method using the estimated ULSC. On the modeling of damaged structural members, the generic element modeling technique is reviewed and extended for damage detection in skeletal structures. To extend the current work to plate structures, damage-detection-oriented models for two typical damage patterns in plate-like structures, i. e. crack and delamination, are proposed. Contributions in this part of research are: 1. An eigen-parameterization strategy for decomposing elemental stiffness is derived from the generic element theory. It is incorporated with a previously developed identification method for the pattern-differential damage detection in skeletal structures. 2. Experimental study is carried out on a cantilever space frame with an attempt of correcting model errors and identifying damage simultaneously with the generic parameters and eigen-parameters. 3. An anisotropic stiffness model for a cracked plate is proposed for non-destructive vibration-based damage detection. Another non-model-based damage detection method is also developed based on the estimation of ULSC. 4. A delamination-detection-oriented model for a Fiber-Reinforced Polymer (FRP) bonded plate is proposed and incorporated with the developed damage identification method with ULSC. To have more insight into the features of the developed methods and models, numerical examples have been studied to show the ability of detecting various damages in different type of structures. Apart from numerical examples, experimental studies are also carried out to verify the feasibility of the methodology in real applications. By incorporating the proposed damage-detection-oriented models with the developed damage identification methods, a systematic methodology has been achieved for non-destructive damage detection based on vibration measurements from a structure.