8
Boundary Element Method

8.1 Introduction

In 1977, for the first time the term boundary element method (BEM) was coined in some publications and the first book on the BEM appeared in 1992 [1]. In the last 20 years the BEM has become one of the important numerical computational techniques along with the FDM (Chapter 5) and FEM (Chapter 6). It is used to solve certain classes of differential equations by formulating as integral equations 2,3]. It is worth mentioning that for certain classes of differential equations, integral equations' reformulation may not always be possible [ 3 ,4]. The advantage in using the BEM is that only the boundaries of the considered domain of the differential equations require subdivision to develop a boundary mesh [5,6]. Whereas, in other methods like FDM or FEM the whole domain of the differential equation requires discretization. In this chapter, we are presenting a brief introduction of the BEM along with a simple example problem for easy understanding of the method.

8.2 Boundary Representation and Background Theory of BEM

Detailed BEM formulation for handling boundary value problem (BVP) in a field or domain may be found in Refs. [ 6 ,7]. By using known data on the boundary S, one can get the rest of information on the boundary S as well as in the domain D in the BEM. Due to various difficulties [ 7 ,8], BVPs do not always have closed‐form solutions or analytic solutions. In general, this may be done by the facility of defining the boundary as a set of panels which have the same characteristic [8] . For example, a closed two‐dimensional boundary may be represented by a set of straight lines (panels), as shown Figure 8.1.

Image described by caption and surrounding text.

Figure 8.1 Closed two‐dimensional boundary represented by a set of straight lines (panels).

Let S be the original boundary and images (for j = 1, 2, …, n) be known as the panels that represent an approximation to S. If images is the surface described by the complete set of panels, then images is the approximation to images. The representation of the boundary in this way is the first step to discretize the integral operators that occur in the boundary integral formulations [ 8 ,9].

Next, we address linear differential operator, fundamental solution, and Green's integral theorems which play an important role in the BEM.

8.2.1 Linear Differential Operator

A linear differential operator, denoted by L, is a function [ 8 , 9 ] such that

  • In one dimension:
    8.1equation
  • In two dimensions:
    8.2equation

All linear operators have the property that L(αu + βv) = αL(u) + βL(v), where α and β are constants and u and v are functions over the domain. The adjoint operator of L represented as L* is the function defined as

8.3equation

and also due to the self‐adjoint property we have L = L*.

To use the BEM for solving BVPs, we must transform the problem into an equivalent boundary integral equation (BIE) problem. In this regard, the fundamental solution and Green's integral theorems are very useful tools. Accordingly, we discuss next the fundamental solution, Green's function, and integral theorems.

8.2.2 The Fundamental Solution

Before discussing the fundamental solution, a brief review of the Heaviside function (H) and Dirac delta function δ in R2 is needed.

8.2.2.1 Heaviside Function

The function H(x − c) is said to be the Heaviside function [10] and is defined as (Chapter 2)

equation

Heaviside functions are also called as step functions.

8.2.2.2 Dirac Delta Function

Let q(x, y) and p(x1, y1) represent moving and fixed points [ 10 ,11] in the domain D as presented in Figure 8.2. The set of functions denoted by δε(p, q) for ε > 0 is defined as

equation
Schematic of a two-dimensional domain D with boundary S, depicted by an ellipse containing a small circle, which has a northwest arrow r pointing to q(x,y). δ(x1 - x, y1 - y) is indicated inside the circle.

Figure 8.2 A two‐dimensional domain D with boundary S.

where ε is the area of the small circle (Figure 8.2 ).

Now, by taking the limit of δε(p, q) as ε → 0 the Dirac delta function δ may be defined as

equation

Further, the Dirac delta function can be viewed as the derivative of the Heaviside step function

equation

and the Dirac delta function has the fundamental property

equation

8.2.2.3 Finding the Fundamental Solution

The fundamental solution of the Laplace equation 9,12] will be described using the Dirac delta function, where the Laplace equation in two dimension is

8.4equation

The Dirac delta function will be used to derive the fundamental solution of Eq. (8.4), then it should satisfy

8.5equation

In polar coordinates the Laplacian operator can easily be written and due to the radial symmetric property [11] we have

8.6equation

where images

Furthermore, the Dirac delta in polar coordinates [ 12 ,13] may be obtained as

8.7equation

Hence, from Eqs. (8.5)–(8.7) we have

8.8equation

By applying multiple integration on both sides of Eq. (8.8) over the domain D, one may obtain the fundamental solution as

8.9equation

where c1 and c2 are integral constants. By considering c1 = c2 = 0, we may have

8.10equation

8.2.3 Green's Function

Green's function G(x, s) is described as integral kernel that can be used to solve differential equations [12]. Here, the Green's function G(x, s) may be considered as a fundamental solution for the differential equation

8.11equation

So, if G(x, s) is a fundamental solution of Eq. (8.11), then it should satisfy L[G(x, s)] =  − δ(x − s) and the solution u(x) of Eq. ( 8.11 ) can be expressed in the following form [ 12 , 13 ]:

8.12equation

From Eqs. ( 8.11 ) and (8.12), one can get

8.13equation

Since L is a linear operator and L[G(x, s)] =  − δ(x − s), from Eq. (8.13) we have

8.14equation

Further, by using the known property images for any c, d > 0, one can rewrite Eq. (8.14) as

8.15equation

8.2.3.1 Green's Integral Formula

Green's integral formula is a very useful tool [14] for finding the derivation of integral equations formed in the BEM. Suppose u and v are continuous functions in a domain D ⊂ R2 with continuous first and second derivatives in D. Then, functions u and v satisfy the Green's formula

8.16equation

where S is the boundary of the domain D and images and images are the outward normal derivatives of u and v, respectively, on the boundary of domain D.

8.3 Derivation of the Boundary Element Method

The BEM can be derived for differential equations for which we can find first a fundamental solution. Accordingly, here a Laplace equation is handled by using the BEM [15,16]

8.17equation

In the BEM the fundamental solution will be used as the weight function in the integral equation. To find a weak fundamental solution, we multiply Eq. (8.17) by a weight function (fundamental solution) w and integrating over the domain D we have

8.18equation

Using the Green–Gauss theorem [17] and Eq. (8.18), one may obtain

8.19equation

Integrating Eq. (8.19) by parts along with Green's formula, we have

8.20equation

It may be noted that in the FEM we chose simple piecewise polynomials as our weighting (test) functions. In the BEM we choose the fundamental solution (Eq. 8.5 ) so that the last term of Eq. (8.20) becomes

8.21equation

Assume that (x1, y1) ∈ D and not on the boundary. From Eqs. ( 8.18 ), ( 8.19 ), and (8.21), one may obtain

8.22equation

8.3.1 BEM Algorithm

This section discusses the solving procedure of the BEM.

  • Step 1: Find the fundamental solution of the governing equation (Eq. (8.10)).
  • Step 2: Derive a formula of two field variables which converts domain integrals into boundary integrals. Integrating by parts twice the given equation (Eq. ( 8.19 )).
  • Step 3: Obtain the BIE for a domain point (Eq. (8.22)).
  • Step 4: Assume that x approach to the boundary and obtain the BIE for a boundary point.
  • Step 5: Choose a finite number of boundary points, discretize the BIEs for the boundary points, and put the discretized BIEs in matrix form.
  • Step 6: In the final step, with respect to the boundary conditions solve the obtained matrix equation.

Next, we present a simple example problem to solve it using the BEM for easy understanding.

Exercise

  1. 1 Solve the following equations by using the BEM:
    1. Poisson equation
    2. Diffusion equation
    3. Wave equation
    4. Heat equation

References

  1. 1 Brebbia, C.A. and Abascal, J.D. (1992). Dominguez Boundary Elements: An Introductory Course. Boston: Computational Mechanics Publications.
  2. 2 Gaul, L., Kogl, M., and Wagner, M. (2003). Boundary Element Methods for Engineers and Scientists: An Introductory Course with Advanced Topics. Berlin: Springer.
  3. 3 Hunter, P. and Pullan, A. (1997). FEM/BEM Notes. New Zealand: University of Auckland.
  4. 4 Trevelyan, J. (1994). Boundary Elements for Engineers: Theory and Applications. Boston: Computational Mechanics Publications.
  5. 5 Hall, W.S. (ed.) (1994). Boundary element method. In: The Boundary Element Method, 61–83. Dordrecht: Springer.
  6. 6 Becker, A.A. (1992). The Boundary Element Method in Engineering: a Complete Course. Singapore: McGraw‐Hill.
  7. 7 Partridge, P.W. and Brebbia, C.A. (eds.) (2012). Dual Reciprocity Boundary Element Method. Southampton, Boston: Springer Science & Business Media.
  8. 8 Banerjee, P.K. and Butterfield, R. (1981). Boundary Element Methods in Engineering Science, vol. 17, 37. London: McGraw‐Hill.
  9. 9 París, F. and Cañas, J. (1997). Boundary Element Method: Fundamentals and Applications, vol. 1. Oxford: Oxford University Press.
  10. 10 Bracewell, R. (2000). The Fourier Transform and Its Applications, 61–65. New York: McGraw‐Hill.
  11. 11 Khelashvili, A. and Nadareishvili, T. (2017). Dirac's reduced radial equations and the problem of additional solutions. International Journal of Modern Physics E 26 (7): 1750043.
  12. 12 Berger, J.R. and Tewary, V.K. (2001). Greens functions for boundary element analysis of anisotropic bimaterials. Engineering Analysis with Boundary Elements 25 (4–5): 279–288.
  13. 13 Cheng, A.D., Abousleiman, Y., and Badmus, T. (1992). A Laplace transform BEM for axisymmetric diffusion utilizing pre‐tabulated Green's function. Engineering Analysis with Boundary Elements 9 (1): 39–46.
  14. 14 Bochner, S. (1943). Analytic and Meromorphic Continuation by Means of Green's Formula, Annals of Mathematics, 652–673. Providence, RI: American Mathematical Society.
  15. 15 Lesnic, D., Elliott, L., and Ingham, D.B. (1997). An iterative boundary element method for solving numerically the Cauchy problem for the Laplace equation. Engineering Analysis with Boundary Elements 20 (2): 123–133.
  16. 16 Delvare, F. and Cimetiere, A. (2008). A first order method for the Cauchy problem for the Laplace equation using BEM. Computational Mechanics 41 (6): 789–796.
  17. 17 Solecki, R. (1983). Bending vibration of a simply supported rectangular plate with a crack parallel to one edge. Engineering Fracture Mechanics 18 (6): 1111–1118.
..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.
Reset
18.117.185.169