IJAV site directory

Published Articles

The Volume 5, No 4, December 2000

«« Back to the List of the Articles

Transfer Matrices of Single Volute Centrifugal Pumps

Fabrice Carta, Jacques Charley and Guy Caignaert


Over the last few years, many studies have been carried out in order to analyse the hydroacoustic behaviour of hydraulic systems in the low frequency range. With that aim in view, Dumont reported that the manufacturer as well as the pump user needs to know the transfer matrix and the associated source terms1 with maximum accuracy. Most of the time, the well-known two-dimensional matrix model is used to perform pressure and flow-rate fluctuation evolutions without taking into account the vibrations of the structure. For some components of hydraulic circuits, such as rotodynamic pumps in various operating conditions, this determination of the transfer matrix can only be accurately performed with the help of experimental investigations. The experimental methodologies, developed over many years, are now well established. They are used in applications for fans and IC-engine exhaust and inlet systems. If we look at the transfer matrix of a pump, we notice that the transfer matrix is apparently independent of the operating conditions, as long as no cavitation appears in the pump. It also appears that such a transfer matrix can be determined in standstill conditions, which is obviously very convenient as experiments become quicker and cheaper. This paper presents a few arguments about the quality criteria of such experimental devices. On the basis of experimental results, obtained with a single volute centrifugal pump tested in air, some transfer matrix characteristics are shown in order to know when the pump can be considered as a reciprocal or a quasi-reciprocal system according to Brennens classification.

Display the first page:

FRF-based Structural Damage Detection Using Kohonen Self-organising Maps

C. Zang and M. Imregun


This paper presents a structural damage detection method using measured frequency response functions (FRFs). The size of the raw vibration data was reduced via a principal component analysis. Kohonen neural networks, the so-called self-organising maps because of their ability to function without supervised training, were used for the actual damage detection. The procedure has two distinct steps. The first step consists of determining the principal components of a response variation matrix formed by including all available FRF data. Substantial data compression is then achieved by projecting the FRFs onto their most significant principal components. During the second step, the compressed data are fed into a Kohonen artificial neural network where the input layer is directly connected to the output layer. Using a winning neuron algorithm for each available measurement, the FRFs are automatically grouped into distinct clusters according to the dynamic properties of the structure they represent, a feature that provides a straightforward detection method. The methodology was applied to 100 measured FRFs obtained from healthy and damaged railway wheels, each response function having 4,096 spectral lines. The visualisation of a Kohonen output layer showed that the FRFs from 10 x 10 the damaged and healthy specimens were grouped in distinct clusters with no overlaps. It was concluded that the approach was well suited to on-line industrial applications, not only because of its ability to work with limited measurement data, but also because of the automatic damage detection capability of Kohonen networks, without requiring any supervised training.

Display the first page:

Prediction of Random Errors in Sound Intensity Measurement

Finn Jacobsen


It is well known that the random errors in sound intensity estimates can be much larger than the theoretical minimum value determined by the BT-product, in particular under reverberant conditions, in the near field of structural sources, and when there are several independent sources present. More than ten years ago it was shown that one can predict the random errors in estimates of the sound intensity in, say, one-third octave bands from the power and cross power spectra of the two signals from a sound intensity probe, determined with fine spectral resolution with a dual channel FFT analyser. This is not very practical, though. In this paper it is demonstrated that one can predict the random errors from the power and cross power spectra determined with the same spectral resolution as the sound intensity itself.

Display the first page:

The Application of Neural Networks to the Prediction of Traffic Noise

Shuoxian Wu, Jiping Zhang


Artificial neural networks are powerful when used in predictions. Based on the traffic noise data measured near a highway in China, an artificial neural network was formed to predict further traffic noise levels. It was proved that the predicted data are in good agreement with measured data. Therefore, the artificial neural network method provides a new method for traffic noise prediction.

Display the first page:

Minimum Variance Control for Acoustically-Compact Plants

Marek Pawelczyk


The objective of the paper is to present and thoroughly examine some concepts of noise control algorithms for acoustically-compact plants. Acoustically-compact plants deserve to be considered as a special group of noise control plants. They are distinguished by the fact that, due to the geometrical arrangement of the loudspeakers and microphones used in the digital control, the transmission times of the acoustic waves are shorter than the transmission times of the corresponding electrical signals. For other electro-acoustic plants, where a reference signal is accessible via a suitable mechanical, electrical or optical transducer and the controller can be causal, feedforward systems can be designed with much success. Otherwise, feedback systems are normally used. For acoustically-compact plants, no optimal controller can be causal since it has to be designed to predict signals with different delays. In this paper minimum variance algorithms based on different control structures: feedforward, feedback (classical and internal model control), and combined (three versions) are considered. First, performance limitations are described. Then the algorithms are analysed theoretically, verified by simulations and finally tested on a real-world device. Features of the algorithms employed for other electro-acoustic plants are also discussed.

Display the first page:

A Combination between Laplace Transform, Strip Method and Transition Matrix for Determination of Dynamic Response and Damping Effect of Plates

A.S. Ashour, A.M. Farag


The main objective of this paper is to assess the dynamic response of a plate under the action of a force having a general variation with time. An efficient and powerful numerical method for the determination of the dynamic response and the damping analysis of a rectangular plate subjected to lateral excitation is proposed. The present technique is accomplished by a combination of the finite strip, transition matrix, and Laplace transform with respect to time. The plate system is idealised as an assembly of a number of strip elements whose transition matrices are derived in the Laplace transformation domain. In the transformed domain, the problem is solved numerically by a finite strip transition matrix method, whereas the dynamic response is finally expressed in the time domain by a numerical inversion of the transformed solution. This study takes into account the effect of the internal viscoelastic damping and the partially restrained boundary conditions. The merits of the present methodology are demonstrated with suitable numerical examples.

Display the first page:

Benefits of Force Limited Vibration Testing

Mark E. McNelis, Terry D. Scharton


Force limited random vibration testing is used at NASA John Glenn Research Center (formerly NASA Lewis Research Center) for the purposes of qualifying hardware for flight. The benefit of force limited testing is that it limits overtesting of flight hardware, by controlling input force and acceleration from the shaker (dual control) to the test article. The purpose of force limiting is to replicate the test article resonant response for the flight mounting condition. The force limited testing technology has been implemented at the Jet Propulsion Laboratory for the past 10 years on various spacecraft testing programs. For example, the Cassini mission to Saturn utilised force limited vibration testing extensively for both the instrument and spacecraft system level vibration testing. NASA John Glenn Research Center is responsible for microgravity combustion and fluid science research on the Shuttle and the International Space Station. Qualification testing of delicate and vibration sensitive science instrumentation is particularly challenging to qualify successfully for flight. In order to facilitate the testing process, force limiting has been implemented. This paper addresses recent flight camera testing (qualification random vibration and strength testing) for the Combustion Module-2 mission and the semi-empirical method of specifying force limits.

Display the first page:

«« Back to the List of the Articles