Efficient solution for the diffraction of elastic SH waves by a wedge: Performance of various exact, asymptotic and simplified solutions

The diffraction of horizontally polarized shear waves by a semi-infinite wedge in frequency and time domains is studied. In particular, this work focus on the performance of different solutions, including the classical contributions from Macdonald, Sommerfeld and Kouyoumjian & Pathak. In additio...

Full description

Autores:
Aristizabal Tique, Victor Hugo
Velez Hoyos, Francisco Javier
Jaramillo J.D.
Tipo de recurso:
Article of journal
Fecha de publicación:
2017
Institución:
Universidad Cooperativa de Colombia
Repositorio:
Repositorio UCC
Idioma:
OAI Identifier:
oai:repository.ucc.edu.co:20.500.12494/41604
Acceso en línea:
https://doi.org/10.1128/MRA.01719-18
https://hdl.handle.net/20.500.12494/41604
Palabra clave:
Computation theory
Diffraction
Elastic waves
Geometry
Program processors
Shear waves
Analytic solution
Computational speed
Diffracted waves
Frequency and time domains
Geometrical theory of diffraction
Graphics processor units
Multi-scale Modeling
Wedge diffraction
Shear flow
analytical framework
elastic wave
geometry
numerical model
SH-wave
wave diffraction
Rights
closedAccess
License
http://purl.org/coar/access_right/c_14cb
Description
Summary:The diffraction of horizontally polarized shear waves by a semi-infinite wedge in frequency and time domains is studied. In particular, this work focus on the performance of different solutions, including the classical contributions from Macdonald, Sommerfeld and Kouyoumjian & Pathak. In addition, two fully analytical, simplified solutions are proposed using arguments from the so-called geometrical theory of diffraction. The main advantage of the two proposed solutions is the fact that the resulting solutions can be scaled to problems with arbitrary and complex geometries. Moreover, it is found that one of the proposed new solutions is highly efficient in terms of accuracy and computational speed as compared to alternative formulations (approximately 1000 times faster than the Macdonald and Kouyoumjian & Pathak solutions), thus, this important characteristic renders this solution ideal for implementation in GPUs (Graphics Processor Units) for multiscale modeling applications. © 2017 Elsevier Ltd