< back to main site


Finite element optimisation for problems in acoustics.

Esward, T J; Wright, L (2001) Finite element optimisation for problems in acoustics. In: National Measurement Conference (NMC) 2001, 6-8 November 2001, Harrogate, UK.

[img] Text

Download (62kB)


NPL uses finite element (F.E.) techniques to investigate the performance of transducers for applications in acoustics, and to assist designers and manufacturers of medical ultrasound devices. In both cases there is frequently the need to optimise the performance of the device against an objective function such as a required acoustic field characteristic or electrical response. Frequently these problems involve resonant systems that make measurement and modelling problematic due to the large responses near to resonant frequencies.
This paper describes an optimisation program developed at NPL, which optimises transducer design within given constraints. The program varies the material properties of components in the design, uses commercial finite element vibro-acoustics software to generate results, and minimises either the difference between results and desired behaviour of an ideal device, or the difference between results and measured response data of an existing device as an objective function. The objective function includes a weighting function to help avoid problems with resonance.
F.E. software is effectively a "Black box" where the relationship between the input data and the output results is complicated and hidden. This relationship cannot generally be written as an analytical formula and derivative information is not available, and this restricts the range of optimisation techniques that can be used. The paper explains the implementation used at NPL, and presents results which demonstrate the success of the method which was adopted. Similar techniques could be applied to many other optimisation problems involving "black box" type software.

Item Type: Conference or Workshop Item (UNSPECIFIED)
Subjects: Mathematics and Scientific Computing
Mathematics and Scientific Computing > Modelling
Last Modified: 02 Feb 2018 13:17
URI: http://eprintspublications.npl.co.uk/id/eprint/2155

Actions (login required)

View Item View Item