Geoffrey R. Tomlinson

Use of active constrained-layer damping for controlling resonant vibration

The work described in this paper is concerned with controlling the strain of the constraining layer of a composite structure in such a way as to enhance the shear generated in the viscoelastic material and hence improve the overall damping of the composite structure. The results have indicated that this concept of active damping produces very effective levels of vibration suppression. In the case of cantilever beams the vibration levels in the first two modes can be almost eliminated when velocity feedback of the beam tip is used. The results show that the addition of active control and passive damping in a single structure combines the advantages of passive damping in the higher modes and active control in the lower modes. In addition active damping as defined in this paper produces a fail-safe mechanism in case of instability occurring in the feedback loop since a considerable level of passive damping is always present.

Direct parameter estimation for linear and non-linear structures

Abstract A method is presented for directly estimating the physical parameter matrices, i.e., mass, stiffness and damping, of linear and non-linear structures by using measured time data. The structure is modelled by an equivalent lumped parameter system with symmetric parameter matrices. A least squares strategy which makes use of these symmetry relations is adopted to allow the estimation of all the parameters and requires excitation at only one location. The validity of the method is demonstrated by using both computer simulation and experimental procedures.

Use of the Hilbert transform in modal analysis of linear and non-linear structures

Abstract An initial study into the application of the Hilbert transform in modal analysis procedures is presented. It is shown that typical structural non-linearities such as non-linear damping and stiffness can be detected and identified directly without the need to generate explicit models. No assumptions regarding the degree of non-linearity are made, which is a restriction in the classical methods for dealing with non-linearities. The properties of the Hilbert transform are discussed with respect to linear and non-linear dynamical systems, and a discrete transform, developed from the continuous functions, is derived in the frequency domain and adapted to modal analysis data in the form of mobility transfer functions. Truncation effects arising from limited frequency ranges of the mobility transfer functions are accounted for by employing correction terms in the frequency domain. Several examples are studied of single and multi-mode systems with non-linearities such as friction, clearance and non-linear stiffness. These examples indicate that the Hilbert transform offers a new method for extending modal analysis to the domains of non-linear systems.

Wavelet signal processing for enhanced Lamb-wave defect detection in composite plates using optical fiber detection

A wavelet-transform-based technique to enhance defect de- tection in a carbon fiber composite plate interrogated using ultrasonic Lamb waves and incorporating an optical fiber receiver is described. Fundamental symmetric (S0) Lamb waves were introduced into the sample plates using a conventional piezoelectric transducer operating at a frequency of around 250 kHz. Coupling into the plates was achieved using a perspex phase-matching wedge. The propagating acoustic pulses were monitored using a simple embedded or surface-mounted singlemode optical fiber forming the signal arm of an optical fiber Mach- Zehnder interferometer. The direct Lamb wave reflections from delami- nations in the sample plates were of low amplitude, although a degree of defect visibility enhancement was achieved by correlating the received signals with the outgoing ultrasonic pulse. A considerable improvement in the defect visibility over the latter technique was found by using a wavelet-transform-based novelty technique to identify the defective plate zones. Using an orthogonal wavelet transform to compress the data, important structurally related features were extracted by setting appro- priate threshold levels on the wavelet coefficients. The reconstructed (uncompressed) data from defect-free portions of the plate were used to construct a template representing a normal condition. Defect location was achieved by analysis of the departure of signals arising from defec- tive plate regions from the no-fault condition template.

Determination of the complex moduli of viscoelastic structural elements by resonance and non-resonance methods

The successful prediction of the vibration transmissibility characteristics of viscoelastic structural elements is strongly dependent upon the use of fairly accurate estimates of the complex moduli, which are frequency and temperature dependent functions. Two methods, namely the standing wave resonance and the non-resonance (dynamic stiffness) methods, which are based on longitudinal forced vibration procedures, are used to investigate frequency and temperature dependent characteristics of the complex Youngs modulus of a composite viscoelastic pipe. It is shown that in the case of the standing wave resonance method, the use of the simple classical frequency equation for the determination of the complex modulus of a viscoelastic prismatic element from the modal values of frequency and transmissibility results in an absolute error of less than 8% for loss factors of up to 0·4. Also, it is shown that as the loss factor increases the number of modes for which the classical frequency equation is applicable decreases and criteria for establishing the range of validity of the classical frequency equation are described. Complex moduli data obtained from experimental tests are then used with the method of reduced variables to produce master curves and equations of reduced dynamic Youngs modulus and loss factor which cover many decades of frequency.

Developments in the use of the Hilbert transform for detecting and quantifying non-linearity associated with frequency response functions

Abstract This paper describes the developments in the use and application of the Hilbert transform for identifying and quantifying non-linearity associated with simulated and experimental frequency response functions. Calculation of the Hilbert transform is carried out in the time domain employing fast Fourier transform procedures and new correction terms are proposed which ameliorate the problems with zoomed data. Linearisation procedures, employing the causal Hilbert transform and random excitation methods, are applied to experimental data and it is shown that both procedures give similar trends in the extracted modal parameters, with consistently lower damping estimates being obtained from the causalisation procedure.

Advances in damping materials and technology

Abstract In the continual search for better damping materials and technologies, significant advances have been made of late. Functionally gradient materials, liquid crystal polymers, magnetostrictive materials and plasma deposited damping coatings are some of the novel materials and technologies being investigated in the Dynamics Research Group at the University of Sheffield. This article presents an overview of the work being carried out in these areas.

Crack Detection in Metallic Structures Using Broadband Excitation of Acousto-Ultrasonics:

This paper reports the application of piezoceramic sensors for crack detection in metallic structures. Rectangular aluminium plates with a crack initiated by spark erosion were used in a simple fatigue experiment. The plates were instrumented with piezoceramic devices bonded in a symmetrical configuration on both sides of the crack. One of the piezoceramics was used as an actuator and excited by a sine sweep and Gaussian white noise signals in order to exploit broadband excitation. The plates were subjected to static and dynamic tensile loading. The growing crack was monitored by the remaining piezoceramic sensors. The experimental data were analyzed using time, frequency and wavelet domain statistical parameters. The results show the potential of the low-frequency broadband excitation for structure-integrated damage detection systems based on acousto-ultrasonics in metallics.

An improved nonlinear model for an automotive shock absorber

A new physical model for a shock absorber is presented which provides a more realistic representation of the stiffness characteristics than previous simple models. The new model is validated on experimental data.

Spectral analysis for non-linear systems, part III: Case study examples

Abstract In this the third part of this paper, the algorithms derived in Part I are applied to estimate the generalised frequency response functions of two unknown real systems and the results are interpreted using the techniques developed in Part II of the study.