An optimised stiffness reduction method for simulating infinite elastic space using commercial Finite Elements codes

13th Anglo-French Physical Acoustics Conference

Pettit, J.R.^1,2 ; Walker, A.^1 ; Lowe, M.J.S.^2

Rolls-Royce Nuclear, PO BOX 2000, Derby

DE21 7XX, United Kingdom^1

Research Centre for NDE, Imperial College London, Exhibition Road, London

SW7 2AZ, United Kingdom^2

关键词: Absorbing layers;    Analytical and numerical solutions;    Industrial companies;    Material damping;    Perfectly Matched Layer;    Spatial domains;    Stiffness matrices;    Stiffness reduction;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/581/1/012005/pdf
DOI  :  10.1088/1742-6596/581/1/012005

来源: IOP

【 摘 要 】

A common goal when using Finite Element (FE) modelling in time domain wave scattering problems is to minimise model size by only considering a region immediately surrounding a scatterer or feature of interest. The model boundaries must simulate infinite space by minimising the reflection of incident waves. This is a significant and long-standing challenge that has only achieved partial success. Industrial companies wishing to perform such modelling are keen to use established commercial FE packages that offer a thorough history of validation and testing. Unfortunately, this limits the flexibility available to modellers preventing the use of popular research tools such as Perfectly Matched Layers (PML). Unlike PML, Absorbing Layers by Increasing Damping (ALID) have proven successful offering practical implementation into any solver that has representation of material damping. Despite good performance further improvements are desirable. Here, a Stiffness Reduction Method (SRM) has been developed and optimised to operate within a significantly reduced spatial domain. The technique is applied by altering damping and stiffness matrices, inducing decay of incident waves. Variables are expressed as a function of known model constants, easing implementation for generic problems. Analytical and numerical solutions have shown that SRM out performs ALID, with results approaching those of PML.