Wave-based numerical methods for acoustics


Astley, R.J., Gamallo, P. and Gabard, G. (2008) Wave-based numerical methods for acoustics. In, 8th.World Congress on Computational Mechanics (WCCM8): 5th European Congress on Computational Methods in Applied Sciences and Engineeering (ECCOMAS 2008), Venice, Italy, 30 Jun - 05 Jul 2008.

Download

[img] PDF (Abstract)
Download (73Kb)

Description/Abstract

Recent developments in wave-based numerical methods are reviewed in application to problems in
acoustics where small perturbations of pressure and velocity propagate on a steady compressible mean
flow. In the most general case, such wave fields are represented by the solution of the linearized Euler
equations. In the case of steady time-harmonic solutions in a homogeneous medium and in the absence
of mean flow, these reduce to the solution of the Helmholtz equation. When irrotational mean flow is
present, they can be represented by a convected Helmholtz-like equation. When rotational mean flow
effects are significant, the time harmonic, coupled, first order linearized Euler equations must be solved.
Wave-based numerical methods have been applied to all three categories of problem.

The solution of the time-harmonic acoustic equations, particularly for large three-dimensional domains,
is computationally challenging. When traditional finite element methods are used with polynomial basis
functions within each element, approximability arguments alone require the use of many nodes per
wavelength to resolve the resulting spatially harmonic solutions. It is moreover well known that these
requirements are exacerbated when the computational domain spans many wavelengths of the solution.
In such cases the ’pollution’ effect [1] means that the global error increases with frequency, irrespective
of the number of nodes per wavelength that are used to resolve the solution. This effect is particularly
acute for problems which involve exterior scattering by objects whose geometric lengthscale is large
compared to a typical wavelength. A problem this type which is of particular interest to the authors
is acoustic radiation from aero-engine nacelles where the length scale of the acoustic disturbance at
peak frequencies is an order of magnitude smaller than the diameter of the nacelle. Another problem
which exhibits the same disparity of lengthscales and where wave-based methods have been recently
been applied is in the calculation of Head Related Transfer Functions for the human head and torso.
Here a similar relationship holds between the wavelength of the disturbance and the dimensions of the
torso over much of the audible range. To resolve either of these problems in three dimensions by using
conventional numerical methods requires many millions of node or grid points.

The use of non-polynomial, wave-like bases to represent such solutions more effectively with fewer
degrees of freedom, can be traced back to infinite element schemes developed over several decades,
in which a single outwardly propagating wave direction is used [2,3]. Such methods are now routinely
implemented in a number of commercial codes. More recently, the Partition of Unity method [4,5,6] has
been applied to a more general class of problem where no a priori wave direction is known. Here a continuous
wave-like trial solution is generated from a conventional Finite Element mesh by multiplying
the shape functions by a discrete set of wave solutions.

A number of discontinuous formulations have also been explored. In such methods, continuity at element
boundaries must be imposed on wavelike trial solutions defined within each element [7,8]. In the
case of the Helmholtz problem, such methods can also be regarded as Trefftz solutions [9], and have
been shown to embrace also the more esoteric Ultra-Weak Variational Approach [10].

All of the above methods suffer from poor conditioning as the number of wave directions increases.
The extent to which this is an impediment to their practical implementation is significant in assessing
the utility of each method, as are potential pre-conditioning strategies to reduce condition number.

While most of the methods noted above, have been developed for the homogeneous Helmholtz problem
in the absence of flow, some have been applied also to the case with non-zero mean flow [8]. Particular
attention will be given in the current review to the effectiveness of these formulations for the flow case.

Item Type: Conference or Workshop Item (Paper)
Additional Information: Keynote Lecture
Keywords: helmholtz problem, flow acoustics, wave basis, partition of unity method, discontinuous Galerkin method, Trefftz methods
Subjects: T Technology > TA Engineering (General). Civil engineering (General)
Q Science > QA Mathematics > QA76 Computer software
Divisions: University Structure - Pre August 2011 > Institute of Sound and Vibration Research > Fluid Dynamics and Acoustics
ePrint ID: 63665
Date Deposited: 23 Oct 2008
Last Modified: 27 Mar 2014 18:45
URI: http://eprints.soton.ac.uk/id/eprint/63665

Actions (login required)

View Item View Item