Dynamic Methods for Damage Detection in Structures

Fundamental concepts for the characterization of the dynamical response of SDOF and NDOF systems are provided. A description is given of the main techniques to represent the response in the frequency domain and its experimental characterization. Two classical procedures of modal parameter identification are outlined and selected numerical and experimental examples are reported.

This chapter gives an overview of the use of inverse methods in damage detection and location, using measured vibration data. Inverse problems require the use of a model and the identification of uncertain parameters of this model. Damage is often local in nature and although the effect of the loss of stiffness may require only a small number of parameters, the lack of knowledge of the location means that a large number of candidate parameters must be included. This leads to potential ill-conditioning problems, and this topic is reviewed in this chapter. This chapter then goes on to discuss a number of problems that exist with the inverse approach to structural health monitoring, including modelling errors, environmental effects, damage localisation, regularisation, models of damage and sensor validation.

This paper presents a methodology that can be used for the identification of second-order models of structural systems using dynamic measurements of the input and of the structural response. This approach (and its variations) starts from an identified first order model of a structural system and obtain estimation of the structure’s mass, damping and stiffness matrices. For these approaches, both the full instrumentation option and the partial instrumentation option are presented. For the case of partial/limited instrumentation, five different model types have been considered showing, for each models, the limitations imposed on the identification by the lack of available data. Once these dynamic characteristics have been determined, structural damage can be assessed by comparing the undamaged and damaged estimation of such parameters. A damage representative parameter is introduced: this parameter uses the identified models to detect the location and amount of structural damage. This methodology has been tested on simulated numerical results and its effectiveness in determining structural damage is evaluated.

This chapter describes structural identification techniques based on parametric models. Focus is given to physical models, where the inertial and mechanical characteristics are the goals of identification. These models are more attractive than modal models as they potentially yield local information. To reduce the computational effort of this complex problem, a refined algorithm is implemented which identifies the parameters of a structure’s finite element model. Experimental data, usually represented by selected modal quantities, are used in the identification process.

This paper deals with a class of inverse problems in vibration concerning the identification of damages in elastic beams by dynamic data. A review of some recent results is presented.

A simple, pragmatic approach to the detection of damage is outlined with a closer look at the detailed effects of individual cracks. The approach focuses on the frequency changes caused by damage and on determining the location, but not the severity, of the damage whether locally or globally. Laboratory application illustrates the detection of fatigue cracks in a simple structure. A second application looks at crack closure that may develop in fatigue and its potential influence on detection.

A quantitative study of the reflection of the T(0,1) mode from defects in pipes in the frequency range 10-300 kHz has been carried out, finite element predictions being validated by experiments on selected cases. Both cracklike defects with zero axial extent and notches with varying axial extents have been considered. The results show that the reflection coefficient from axisymmetric cracks increases monotonically with depth at all frequencies and increases with frequency at any given depth. In the frequency range of interest there is no mode conversion at axisymmetric defects. With nonaxisymmetric cracks, the reflection coefficient is a roughly linear function of the circumferential extent of the defect at relatively high frequencies, the reflection coefficient at low circumferential extents falling below the linear prediction at lower frequencies. With nonaxisymmetric defects, mode conversion to the F(1,2) mode is generally seen, and at lower frequencies the F(1,3) mode is also produced. The depth and circumferential extent are the parameters controlling the reflection from cracks; when notches having finite axial extent, rather than cracks, are considered, interference between the reflections from the start and the end of the notch causes a periodic variation of the reflection coefficient as a function of the axial extent of the notch. The results have been explained in terms of the wave-number-defect size product, ka. Low frequency scattering behavior is seen when ka<0.1, high frequency scattering characteristics being seen when ka> 1.

The small sensitivity to local variations of mechanical characteristics turns out to be the major limit of indirect identification techniques based on frequency response measurements. To overcome this limit, the use of sensitivity enhancement techniques has recently proposed: the monitored structure is coupled to an auxiliary system, the constitutive parameters of which are then suitably tuned to enhance the sensitivities relevant to the identification process. The damage identification problem in these “augmented” structures is here introduced, and its main advantages and drawbacks are discussed. A beam-like structure coupled to a network of piezoelectric patches supplies an enlightening application of the proposed technique.