Boundary Integral Equations and Boundary Elements Methods in Elastodynamics
Introduction
In several areas of physics, like solid mechanics, electromagnetism, fluid dynamics, or thermodynamics, the resolution of a problem can be reduced to an integral equation defined over the boundary of the domain studied. This expresses the fact that the totality of the information pertaining to the domain is carried by the boundary. Apparently this remarkable fact was discovered by G. Green although some attribute the finding to C.F. Gauss. In any event, the Boundary Integral Equations (BIE) formulate the solution of a problem in terms of values at the domain's boundary. BIE methods arise from potential theory and can be grouped into two families. “Direct” BIE relate the physical variables (the wavefield of displacements and tractions in elastodynamics) in a given domain to the values taken by these variables at the boundary. “Indirect” BIE, on the other hand, rely on an intermediate unknown, which is usually a distribution of fictitious (or apparent) sources along the boundary.
The resolution of the direct or indirect BIE must be done numerically. The first step consists in discretizing the boundary into a set of points or into a set of elements. In the latter case, the BIE method is often referred to as a Boundary Element Method (BEM) while the indirect BIE may be named as indirect BEM (IBEM). In elastodynamics, which is our domain of interest here, the second step is the evaluation of the medium Green's functions. In doing so, one needs to avoid the inherent singularities of having to evaluate the wavefield at the source location itself. To this effect, various formulations of the BIE have been developed which are un-conditionally stable. The third and final step is to solve the resulting linear system of equations.
Compared to domain methods, like finite-differences or finite-elements, BIE methods have an interesting conceptual advantage which is the reduction of one space dimension for both discretization and handling of the unknowns. For instance, two-dimensional (2-D) problems are reduced to integral representations along lines (thus, this is equivalent to 1-D problems). This advantage is enhanced because the domain effectively considered can be much larger than in domain methods which require to make explicit such size. Moreover, BIE methods (in frequency) do not need absorbing boundaries. Indeed, we have to mention that BIE are more accurate than domain methods. They match more easily the boundary conditions and do not suffer from grid dispersion. They easily fulfill radiation conditions at infinity and can handle complex boundary geometries for which domain methods would require complicated grid setting. On the other hand, the benefits are counter-balanced by the global character of boundary approaches, each equation involves the full wavefield (at least in the considered sub domain), while in domain methods, only the local field values are related by the governing equations. In the BIE the field at each point of the boundary is linked to all boundary field values and, as a consequence, the linear system that needs to be solved is dense. For this reason, sometimes they are more demanding in computer time and memory space, because they require solving non-sparse linear systems.
As the BIE methods directly express the physics of the problem, they also provide potentially valuable insight into the problems they tackle. They cannot be easily designed to handle non-linear problems. They are relatively well adapted to study acoustic or seismic wave propagation in the shallow Earth which can often be described as consisting of relatively homogeneous layers separated by sharp boundaries.
After describing the direct and indirect BIE methods, we will review and discuss some salient numerical aspects of these techniques. We will also present and discuss other boundary methods, in which approximate expressions of the wavefield, such as plane wave decompositions or basis function expansions are used to set up boundary equations from which relatively fast, approximate solutions can be obtained.
Excellent reviews of the available literature on BEM in elastodynamics up to the early 90s are those by Manolis and Beskos (1988), Brebbia and Domínguez (1992), and Domínguez (1993). These researchers focused on various structural engineering applications. Herein the emphasis is laid on seismological aspects.
Section snippets
The Direct BIE
The formulation of the direct BIE in elastodynamics can be traced back to the pioneering work of Somigliana (1886), more than a century ago. It stems from the theory of elastic potentials and states that field values of a problem of linear elasticity in a given domain are fully determined from the displacements and tractions along the domain's boundary. It can be seen as an application of the Maxwell–Betti reciprocity theorem. Somigliana's work is based upon the classical work by Stokes (1849)
The Indirect BIE
The indirect formulation of the elastodynamic problem expresses the wavefield as an integral over the boundary of elementary source radiations: where denotes a force density distribution applied at time τ at point ξ of the surface S. In this equation we have again assumed that the volumic body sources are null. If present, their contribution simply needs to be added to the integral term.
An example of this formulation is given in Vai et al. (1999);
Approximate Representations
Early applications of BIE methods in elastodynamics have dealt with the inherent singularities of the integrand kernels, as the Green's functions have to be evaluated at the very location of the sources. The pioneering works by Wong and Jennings (1975), and Sills (1978) are good examples of these efforts. However, in several practical instances a simplified approach was needed. The chosen strategy was to avoid these singularities and reliable methods have been developed where the sources are
Discretization and Inversion
The discretization of the boundaries leads to the discretization of the BIE and the application of the boundary conditions transforms the BIE into a system of linear equations which, in general, is not symmetric. This fact is sometimes undesirable, particularly for large size matrices. The resolution of this system is usually done implicitly, and in the frequency domain. The discretization of the boundaries is generally done by approximating the surfaces by a set of elements over which the
Time Domain Implementation
In order to overcome memory limitations, which arise from solving very large linear systems, BIE methods may be implemented in the time domain. The BIE then gives the expression of the displacement field u at time t as a function of the values of u and t at times . Its discretization leads to an implicit formulation which can be solved for each time step (Mansur and Brebbia, 1982, Mansur and Brebbia, 1985, Rice and Sadd, 1984, Banerjee).
It is, however, often more interesting to render this
Other Boundary Methods
One of the most successful boundary method in elastodynamics is the Aki–Larner method. This method, developed by Aki and Larner (1970), represents the harmonic wave field diffracted by a surface or interface as a superposition of plane waves with unknown coefficients propagating away from the diffracting boundary. Inhomogeneous waves are allowed. This discrete summation results from an assumed spatial periodicity of the medium. The equations, in displacement and stress, expressing the boundary
Hybrid Methods
In many ways, boundary and domain methods can be viewed as complementary techniques, and their combination may lead to more efficient solutions than relying on one technique only. Formulations combining BE and FE have been proposed by Mossessian and Dravinski (1987), Regan and Harkrider (1989), Fujiwara (1996b), Zhang et al. (1998), and Fu (2002). This combination is particularly relevant for media containing both irregular boundaries and volume heterogeneities (Fu, 2002). Figure 16, shows the
Domains of Application
BIE methods have been extensively applied to model or simulate seismic wave propagation in the shallow earth. They have helped understand and quantify the effect of irregular topography, like hills, mountains, or valleys on earthquake ground motion (Bouchon, 1973, Bouchon, 1985, Wong and Jennings, 1975, Sánchez-Sesma, 1978, Sánchez-Sesma, 1983, Sills, 1978, Sánchez-Sesma and Rosenblueth, 1979, Bard, 1982, Wong, 1982, Sánchez-Sesma, Géli, Kawase, 1988, Gaffet and Bouchon, 1989, Axilrod and
Concluding Remarks
This rapid passage over the BIE in elastodynamics essentially covers certain developments in the Earth sciences and seismology of the last three decades. There is a number of problems that may benefit from advances in other areas.
Variational, domain decomposition and subdomain hybrid formulations hold the promise to give the power to deal with more complex configurations…
Acknowledgements
The critical comments of V. Maupin and R.-S. Wu helped to sharpen the focus of this work. Thanks are given to M. Campillo, O. Coutant for their comments and suggestions. The multifarious help of E. Flores was crucial to finish this work. Our research benefited from various sources of support along the years. We want to thank the French CNRS, and the Mexican CONACYT and DGAPA-UNAM.
References (133)
- et al.
Modelling of dynamic crack propagation using time-domain boundary integral equations
Wave Motion
(1992) - et al.
Propagation of SH elastic waves in deep sedimentary basins with an oblique velocity gradient
Wave motion
(2003) - et al.
Numerical implementation of the boundary element method for two-dimensional transient scalar wave propagation problems
Appl. Math. Mod.
(1982) - et al.
Azimuth dependent wave amplification in alluvial valleys
Soil Dyn. Earthquake Engrg.
(1995) - et al.
Surface motion of a layered medium having an irregular interface due to incident plane SH waves
J. Geophys. Res.
(1970) - et al.
Quantitative Seismology, vol. 1
(1980) Dynamic plane-strain shear rupture with a slip-weakening friction law calculated by a boundary integral method
Bull. Seismol. Soc. Am.
(1985)- et al.
Spontaneous rupture propagation on a non-planar fault in 3-D elastic medium
Pure Appl. Geophys.
(2000) - et al.
The 1999 Izmit, Turkey, earthquake: Non-planar fault structure, dynamic rupture process, and strong ground motion
Bull. Seismol. Soc. Am.
(2003) - et al.
SH-wave scattering from a sinusoidal grating: An evaluation of four discrete wavenumber methods
Bull. Seismol. Soc. Am.
(1990)
Transient elastodynamic analysis of three-dimensional problems by Boundary Element Methods
Earthquake Engrg. Struct. Dyn.
Diffracted waves and displacement field over two-dimensional topographies
Geophys. J. R. Astron. Soc.
The seismic response of sediment-filled valleys. Part 1: The case of incident SH waves
Bull. Seismol. Soc. Am.
The seismic response of sediment-filled valleys. Part 2: The case of incident P and SV waves
Bull. Seismol. Soc. Am.
The two-dimensional resonance of sediment-filled valleys
Bull. Seismol. Soc. Am.
Boundary integral—Gaussian beam method for seismic wave scattering: SH waves in two-dimensional media
J. Acoust. Soc. Am.
Multiple scattering of SH waves in 2-D media with many cavities
Pure Appl. Geophys.
Regular boundary integral equations for three-dimensional finite or infinite bodies with and without curved cracks in elastodynamics
Symmetric Galerkin boundary element methods
Appl. Mech. Rev.
Effect of topography on surface motion
Bull. Seismol. Soc. Am.
A simple complete numerical solution to the problem of diffraction of SH waves by an irregular surface
J. Acoust. Soc. Am.
Diffraction of elastic waves by cracks or cavities using the discrete wavenumber method
J. Acoust. Soc. Am.
A numerical simulation of the acoustic and elastic wavefields radiated by a source in a fluid-filled borehole embedded in a layered medium
Geophysics
Full-wave acoustic logging in an irregular borehole
Geophysics
A boundary integral equation—discrete wavenumber representation method to study wave propagation in multilayered media having irregular interfaces
Geophysics
Calculation of synthetic seismograms in a laterally-varying medium by the boundary element—discrete wavenumber method
Bull. Seismol. Soc. Am.
Effect of 3-D topography on seismic motion
J. Geophys. Res.
Propagation of a shear crack on a non-planar fault: A method of calculation
Bull. Seismol. Soc. Am.
Ground motion on stratified alluvial deposits for incident SH waves
Bull. Seismol. Soc. Am.
Boundary Elements: An Introductory Course
The numerical solution of certain integral equations with non-integrable kernels arising in the theory of crack propagation and elastic wave diffraction
Philos. Trans. R. Soc. London A
Modeling of SH-wave propagation in an irregularly layered medium—Application to seismic profiles near a dome
Geophys. Prosp.
Lg wave propagation in a laterally varying crust and the distribution of the apparent quality factor in central France
J. Geophys. Res.
Synthetic seismograms in a laterally varying medium by the discrete wavenumber method
Geophys. J. R. Astron. Soc.
Destructive strong ground motion in Mexico City: Source, path, and site effects during great 1985 Mechoacan earthquake
Bull. Seismol. Soc. Am.
Attenuation of crustal waves across the Alpine range
J. Geophys. Res.
Effect of random thickness variations on the seismic response of a soft soil layer: Applications to Mexico City
Seismogram synthesis for multi-layered media with irregular interfaces by global generalized reflection/transmission matrices method. 1. Theory of two-dimensional SH case
Bull. Seismol. Soc. Am.
Seismogram synthesis for multi-layered media with irregular interfaces by global generalized reflection/transmission matrices method. 3. Theory of 2-D P-SV case
Bull. Seismol. Soc. Am.
Imaging the Earth's Interior
Dynamic faulting under rate-dependent friction
Pure Appl. Geophys.
Complexity of seismicity due to highly rate-dependent friction
J. Geophys. Res.
A numerical Boundary Integral Equation Method for elastodynamics, I
Bull. Seismol. Soc. Am.
Numerical study of the diffraction of elastic waves by fluid-filled cracks
J. Geophys. Res.
A numerical method for determination of source time functions for general three-dimensional rupture propagation
Geophys. J. R. Astron. Soc.
A numerical study of two-dimensional spontaneous rupture propagation
Geophys. J. R. Astron. Soc.
On the numerical boundary integral equation method for three-dimensional dynamic shear crack problems
J. Appl. Mech.
Boundary Elements in Dynamics
Borehole seismic-source radiation in layered isotropic and anisotropic media: boundary element modeling
Geophysics
Cited by (65)
A half-space TD-BEM model for a seismic corrugated orthotropic stratum
2023, Engineering Analysis with Boundary ElementsSeismic response of stratified rock slopes due to incident P and SV waves using a semi-analytical approach
2022, Engineering GeologyCitation Excerpt :With the development of elastodynamic theory and computing technology, analytical and numerical calculations are extensively adopted to solve the dynamic problems of slope topography. The solution approaches include closed-form analytical scheme (e.g., wave function expansion method (Eshraghi and Dravinski, 1989), semi-analytical scheme relying on frequency-domain Green's function (e.g., boundary integral equation method (Bouchon and Sánchez-Sesma, 2007)), and various numerical schemes based on time-domain algorithms (e.g., finite element method (Huang et al., 2015), finite difference method (Xiao et al., 2016), and discrete element method (Lu et al., 2018)). Of the outstanding contributions in this subject, Zhang et al. (2017), and Liu et al. (2019) give the analytical expressions for the scattering of elastic waves by a slope or other simple geometries using the wave function expansion method, with the incident waves covering SH-, P-, SV-, and Rayleigh-waves.
A three-dimensional indirect boundary integral equation method for the scattering of seismic waves in a poroelastic layered half-space
2022, Engineering Analysis with Boundary ElementsCitation Excerpt :Thus the boundary-type methods have been extensively used in wave motion simulation in an infinite domain. This method can be divided into boundary element method [20–29], boundary integral equation method [29–35], Aki-Larner method [36,37], and discrete-wave number method [38], etc. By using the boundary method, previous scholars have made many important achievements in the study of the effect of local site on ground motion in saturated soil, but most of these studies assume that these local site are located in saturated full or half-space.
A three-dimensional indirect boundary integral equation method for modeling elastic wave scattering in a layered half-space
2019, International Journal of Solids and Structures