1  Introduction

Over recent decades the volume of traffic in developed countries has increased at a dramatic rate. The potential for a corresponding increase in traffic noise pollution has been countered, in most countries, by legislation which restricts the noise output of new vehicle designs (Waters (1980)). The criterion for measuring the noise produced by a vehicle is the drive past test (Hassall and Zaveri (1979), Waters (1980)). In brief, this measures the A-weighted total sound pressure at a point a particular distance from the path of the vehicle. The test site is regulated and the vehicle is operating under a specific set of conditions for the duration of the test. Legislation generally places an upper limit on the measured noise for each different category of vehicle. With the clear aim of keeping traffic noise to a reasonable level, over the years it has been the tendency for legislators to systematically reduce these noise limits.
The pressure of keeping the noise from vehicles to a certain level and the need to satisfy the customers' personal demands for less noisy vehicles has meant that noise has become an important factor for automotive engineers to consider. The noise arises as a result of the vibration of the vehicle. Close inspection of any vehicle reveals that there are several discernible sources of vehicle vibration. Of these, the engine is found to be a major source under most operating conditions. Hence a large part of the burden of designing low-noise vehicles has fallen to the engine designer.
The relationship between the noise produced by engines, the vibration of the engine, the design of the engine structure and the excitation forces has interested engineers for several decades. Equipment exists for the measurement of each of these properties. Usually, the relationships between these properties are analysed only after they are resolved into their frequency components, although recent advances in the analysis of non-linear effects of joints and interaction through oil films have led to some work being performed in the time domain. The results of such experimental work help the engineer to understand how the noise arises and perhaps to decide how the design may be modified to reduce the noise. However, development in this way is costly and hence a computer-aided approach to the design of low-noise engines has been pursued over recent years.
Once the design of the basic powertrain structure itself has been finalised, shielding may be used to effect a further reduction in noise. Shielding may take the form of complete or substantial enclosure of the engine block or it may consist of one or more shields which together only partially enclose the engine. Both methods add bulk to the engine block and hence neither method is an easy option for designers. However, whereas complete or substantial enclosure is generally obstructive to routine maintenance operations, partial shielding need not be. For discussions on the practicalities of shielding, see Waters (1980) and Thien (1982). In this paper the partial enclosure of the engine using a set of close fitting shields is considered.
The steady improvement in both numerical techniques and computer equipment over recent decades has meant that it is possible to make computational predictions of both the vibratory and acoustic properties of engines. The object of this paper is to describe the advanced computational techniques that may be employed: the use of the finite element method (FEM) for the prediction of the vibratory properties, the boundary element method (BEM) for the prediction of the sound field around a vibrating engine and the use of the boundary and shell element method (BSEM) for the prediction of the effect of an engine noise shield. Results from the application of the FEM and BEM to engine-like structures are given. The BSEM is applied to a suitable test problem and the results from this are presented.
The ultimate aim of this work is to provide a range of software tools to aid the design development process. In other words, the aim is to develop an integrated computer-aided design approach to low-noise engine design. The design development of any system may ultimately be regarded as a repetitive process, each stage consisting of an analysis of the existing design followed by a decision on the modifications for the new design. In this paper we are concerned with computational techniques for the analysis stage of the design process. Such techniques are considered alongside standard measurement techniques and within the computer-aided design context.