FEM/BEM NOTES

Professor Peter Hunter
p.hunter@auckland.ac.nz

Associate Professor Andrew Pullan
a.pullan@auckland.ac.nz

Department of Engineering Science The University of Auckland New Zealand February 21, 2001

c Copyright 1997 : Department of Engineering Science

Contents
1 Finite Element Basis Functions 1.1 Representing a One-Dimensional Field . 1.2 Linear Basis Functions . . . . . . . . . 1.3 Basis Functions as Weighting Functions 1.4 Quadratic Basis Functions . . . . . . . 1.5 Two- and Three-Dimensional Elements 1.6 Higher Order Continuity . . . . . . . . 1.7 Triangular Elements . . . . . . . . . . . 1.8 Curvilinear Coordinate Systems . . . . 1.9 CMISS Examples . . . . . . . . . . . . 1 1 2 4 7 7 10 14 16 19 21 21 22 22 23 24 25 27 27 27 28 29 30 32 34 35 37 40 41 41 41 41 43

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

2 Steady-State Heat Conduction 2.1 One-Dimensional Steady-State Heat Conduction . . . . . 2.1.1 Integral equation . . . . . . . . . . . . . . . . . . 2.1.2 Integration by parts . . . . . . . . . . . . . . . . . 2.1.3 Finite element approximation . . . . . . . . . . . 2.1.4 Element integrals . . . . . . . . . . . . . . . . . . 2.1.5 Assembly . . . . . . . . . . . . . . . . . . . . . . 2.1.6 Boundary conditions . . . . . . . . . . . . . . . . 2.1.7 Solution . . . . . . . . . . . . . . . . . . . . . . . 2.1.8 Fluxes . . . . . . . . . . . . . . . . . . . . . . . . 2.2 An x-Dependent Source Term . . . . . . . . . . . . . . . 2.3 The Galerkin Weight Function Revisited . . . . . . . . . . 2.4 Two and Three-Dimensional Steady-State Heat Conduction 2.5 Basis Functions - Element Discretisation . . . . . . . . . . 2.6 Integration . . . . . . . . . . . . . . . . . . . . . . . . . . 2.7 Assemble Global Equations . . . . . . . . . . . . . . . . . 2.8 Gaussian Quadrature . . . . . . . . . . . . . . . . . . . . 2.9 CMISS Examples . . . . . . . . . . . . . . . . . . . . . . 3 The Boundary Element Method 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . 3.2 The Dirac-Delta Function and Fundamental Solutions 3.2.1 Dirac-Delta function . . . . . . . . . . . . . 3.2.2 Fundamental solutions . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

ii

CONTENTS

3.3 3.4 3.5 3.6 3.7 3.8

3.9 3.10 3.11 3.12

3.13 3.14 3.15 3.16

3.17 4

The Two-Dimensional Boundary Element Method . . . . . . . . . Numerical Solution Procedures for the Boundary Integral Equation Numerical Evaluation of Coefficient Integrals . . . . . . . . . . . The Three-Dimensional Boundary Element Method . . . . . . . . A Comparison of the FE and BE Methods . . . . . . . . . . . . . More on Numerical Integration . . . . . . . . . . . . . . . . . . . 3.8.1 Logarithmic quadrature and other special schemes . . . . 3.8.2 Special solutions . . . . . . . . . . . . . . . . . . . . . . The Boundary Element Method Applied to other Elliptic PDEs . . Solution of Matrix Equations . . . . . . . . . . . . . . . . . . . . Coupling the FE and BE techniques . . . . . . . . . . . . . . . . Other BEM techniques . . . . . . . . . . . . . . . . . . . . . . . 3.12.1 Trefftz method . . . . . . . . . . . . . . . . . . . . . . . 3.12.2 Regular BEM . . . . . . . . . . . . . . . . . . . . . . . . Symmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Axisymmetric Problems . . . . . . . . . . . . . . . . . . . . . . Infinite Regions . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix: Common Fundamental Solutions . . . . . . . . . . . . 3.16.1 Two-Dimensional equations . . . . . . . . . . . . . . . . 3.16.2 Three-Dimensional equations . . . . . . . . . . . . . . . 3.16.3 Axisymmetric problems . . . . . . . . . . . . . . . . . . CMISS Examples . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

46 51 53 55 56 58 58 59 59 59 60 62 62 62 63 65 67 70 70 70 71 71 73 73 74 77 79 81 83 84 86 86 89 90 93 95 97 97 97 97 99 100 101

Linear Elasticity 4.1 Introduction . . . . . . . . . . . . . . . . . . . 4.2 Truss Elements . . . . . . . . . . . . . . . . . 4.3 Beam Elements . . . . . . . . . . . . . . . . . 4.4 Plane Stress Elements . . . . . . . . . . . . . . 4.5 Navier’s Equation . . . . . . . . . . . . . . . . 4.6 Note on Calculating Nodal Loads . . . . . . . . 4.7 Three-Dimensional Elasticity . . . . . . . . . . 4.8 Integral Equation . . . . . . . . . . . . . . . . 4.9 Linear Elasticity with Boundary Elements . . . 4.10 Fundamental Solutions . . . . . . . . . . . . . 4.11 Boundary Integral Equation . . . . . . . . . . . 4.12 Body Forces (and Domain Integrals in General) 4.13 CMISS Examples . . . . . . . . . . . . . . . . Transient Heat Conduction 5.1 Introduction . . . . . . . . . . . . . . . . . . 5.2 Finite Differences . . . . . . . . . . . . . . . 5.2.1 Explicit Transient Finite Differences . 5.2.2 Von Neumann Stability Analysis . . . 5.2.3 Higher Order Approximations . . . . 5.3 The Transient Advection-Diffusion Equation

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

5

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

CONTENTS

iii

5.4 5.5

Mass lumping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 CMISS Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 109 109 109 111 112 113 115 115 116 116 117 118 118 119 120 122 133 133 133 135 136 137 138 141

6 Modal Analysis 6.1 Introduction . . . . . . 6.2 Free Vibration Modes . 6.3 An Analytic Example . 6.4 Proportional Damping 6.5 CMISS Examples . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

7 Domain Integrals in the BEM 7.1 Achieving a Boundary Integral Formulation . . . . . . . . . 7.2 Removing Domain Integrals due to Inhomogeneous Terms . 7.2.1 The Galerkin Vector technique . . . . . . . . . . . . 7.2.2 The Monte Carlo method . . . . . . . . . . . . . . . 7.2.3 Complementary Function-Particular Integral method 7.3 Domain Integrals Involving the Dependent Variable . . . . . 7.3.1 The Perturbation Boundary Element Method . . . . 7.3.2 The Multiple Reciprocity Method . . . . . . . . . . 7.3.3 The Dual Reciprocity Boundary Element Method . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

8 The BEM for Parabolic PDES 8.1 Time-Stepping Methods . . . . . . . . . . . . . . . . . . . . . 8.1.1 Coupled Finite Difference - Boundary Element Method . 8.1.2 Direct Time-Integration Method . . . . . . . . . . . . . 8.2 Laplace Transform Method . . . . . . . . . . . . . . . . . . . . 8.3 The DR-BEM For Transient Problems . . . . . . . . . . . . . . 8.4 The MRM for Transient Problems . . . . . . . . . . . . . . . . Bibliography

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

but if the polynomial order is increased further to improve the accuracy of fit a problem arises: the polynomial can be made to fit the data accurately. showing the unacceptable oscillation between data points. as shown in Figure 1. Figure 1.called elements. . c and d from a least-squares fit to the data. u is plotted as a function of this arclength in Figure 1. As the degree of the polynomial is increased the data points are fitted with increasing accuracy and polynomials provide a very convenient form of expression because they can be differentiated and integrated readily. b. measurements of temperature u against distance x along a bar.g..1: (a) Temperature distribution u x along a bar.Chapter 1 Finite Element Basis Functions 1.2a. u + + + + + + + + ++ + + + + + u + + + + + + + + ++ + + + + + x (b) x (a) F IGURE 1. For low degree polynomials this is a satisfactory approach. but it oscillates unacceptably between the data points. One approach would be to use a polynomial expression u x = a + bx + cx2 + dx3 + : : : and to estimate the values of the parameters a. (b) A least-squares polynomial fit to the data. For later generality we also introduce a parameter s which is a measure of distance along the bar. as shown in Figure 1.1b.2b shows three linear polynomials in s fitted by least-squares separately to the data in each element. we divide the bar into three subregions and use low order polynomials over each subregion . while retaining the advantages of low degree polynomials.1a. The points are the measured temperatures.1 Representing a One-Dimensional Field Consider the problem of finding a mathematical expression u x to represent a one-dimensional field e. To circumvent this.

e i.2 Linear Basis Functions A new problem has now arisen in Figure 1. It is convenient always to associate the nodal quantity un with element node n and to map the temperature U defined at global node  onto local node n of element e by using a connectivity matrix  n.2b: the piecewise linear polynomials are not continuous in u across the boundaries between elements.  '2   =  such that u   = '1   u1 + '2   u2 and refer to these expressions as the basis functions associated with the nodal parameters u1 and u2. e = global node number of local node n of element e. One solution would be to constrain the parameters a. We then define a linear variation between these two values by u   = 1 . '1   = 1 .   u1 + u2 where  0   We define  1 is a normalized measure of distance along the curve. which are the values of u at the two ends of that element.2: (a) Temperature measurements replotted against arclength parameter s.3. (b) The s domain is divided into three subdomains. and linear polynomials are independently fitted to the data in each subdomain. 1. to ensure continuity of u across the element boundaries. un = Un. c etc. The basis functions '1   and '2   are straight lines varying between 0 and 1 as shown in Figure 1.2 F INITE E LEMENT BASIS F UNCTIONS u + + + + + + + u + + + + + + + + + + + + + + + + + + + + + + + + + s (b) s (a) F IGURE 1. elements. This has the advantage that the . but a better solution is to replace the parameters a and b in the first element with parameters u1 and u2 ..e where  n. b.e.

2 L INEAR BASIS F UNCTIONS 3 '1   1 '2   1 1 0 1  0 1  F IGURE 1.2) = U2 and u2 = U3 .3: Linear basis functions '1   = 1 . u   = '1   u1 + '2   u2 with u1 = U1 and u2 = U2 .4.  and '2   =  .1. interpolation u   = '1   u1 + '2   u2 holds for any element provided that u1 and u2 are correctly identified with their global counterparts. Thus.4: The relationship between global nodes and element nodes. since the parameter U2 is shared between the first and second elements .1) u   = '1   u1 + '2   u2 with u1 (1. in the first element node 1 global nodes: U1 node 2 U2 node 3 U3 node 4 U4 x element nodes: u1 0 u2 element 1 1  u1 0 u2 element 2 1  u1 0 u2 element 3 1  F IGURE 1. In the second element u is interpolated by (1. as shown in Figure 1.

u node 1 + + + + node 2 + + + + + + node 3 + + + + node 4 + + element 1 element 2 element 3 s F IGURE 1. 1.6 shows the temperature field defined by the three interpolations (1. 0 u1 + 0u2 = u1 which is the value of u at the left hand end of the element and has no dependence on u2 1 1 = 1 .3 Basis Functions as Weighting Functions It is useful to think of the basis functions as weighting functions on the nodal parameters. in element 1 at  =0 u 0 = 1 .5: Temperature measurements fitted with nodal parameters and linear basis functions. The fitted temperature field is now continuous across element boundaries. in the third element u is interpolated by u   = '1   u1 + '2   u2 (1. 1 u + 1u = 1u + 1u at  = u 2 2 2 1 2 2 2 1 2 2 which depends equally on u1 and u2 3 at  = 4 3 1 3 u 4 = 1. 1 u + 1u = 3u + 1u at  = u 4 4 4 1 4 2 4 1 4 2 which depends on u1 and u2 . Thus. but is weighted more towards u1 than u2 1 1 = 1 .1)–(1. Similarly.4 F INITE E LEMENT BASIS F UNCTIONS the temperature field u is implicitly continuous. with the parameter U3 being shared between the second and third elements. Figure 1.3) with u1 = U3 and u2 = U4 .3). 3 u 1 + u2 = u1 + u2 4 4 4 4 .

.

.

.

.

3 .

.

u x.4) and similarly for the other two elements. respectively. w2 weights the global parameter U2 and the influence of U2 falls off linearly in the elements on either side of node 2. w1 (a) s w2 (b) s w3 (c) s w4 (d) s F IGURE 1. is therefore . Notice the linear fall off in the elements adjacent to a node. as shown in Figure 1. We now have a continuous piecewise parametric description of the temperature field u   but in order to define u x we need to define the relationship between x and  for each element. in element 1 x   = '1   x1 + '2   x2 (1.6.1. For example. the weighting functions are defined to be zero. 1 u1 + 1u2 = u2 which is the value of u at the right hand end of the region and has no dependence on u1 . where the weighting function wn associated with global node n is constructed from the basis functions in the elements adjacent to that node. these weighting functions can be considered as global functions.3 BASIS F UNCTIONS AS W EIGHTING F UNCTIONS 5 which depends on u1 and u2 but is weighted more towards u2 than u1 at  =1 u 1 = 1 . For example. A convenient way to do this is to define x as an interpolation of the nodal values of x. Outside the immediately adjacent elements. The dependence of temperature on x. Moreover.6: (a) : : : (d) The weighting functions wn associated with the global nodes n = 1 : : : 4.

6 F INITE E LEMENT BASIS F UNCTIONS defined by the parametric expressions u   = x  = X X n n 'n   un 'n   xn where summation is taken over all element nodes (in this case only 2) and the parameter  (the “element coordinate”) links temperature u to physical position x. The points at  = 0:2 are emphasized.7: Illustrating how x and u are related through the normalized element coordinate  . The values of x   and u   are obtained from a linear interpolation of the nodal variables and then plotted as u x. . u u1 u2 0  = 0: 2 1  u1 u2 u u x at  = 0:2  x x2 0 x1 x2 x x1 0  = 0: 2 1  F IGURE 1. as illustrated in Figure 1. x   provides the mapping between the mathematical space 0    1 and the physical space x1  x  x2 .7.

0:5  . Similarly for the other two basis functions. This ensures the linear independence of the basis functions. 0:5  .4  .4 QUADRATIC BASIS F UNCTIONS 7 1. u2 and u3 u   = '1   u1 + '2   u2 + '3   u3 (1. with their mathematical expressions. a quadratic variation of u over an element requires three nodal parameters u1 . 1. 1  1 0 0:5 1 (c) '3   = 2  . another factor is  . Notice that since '1   must be zero at  = 0:5 (node 2).and Three-Dimensional Elements Two-dimensional bilinear basis functions are constructed from the products of the above onedimensional linear functions as follows . Finally.8: One-dimensional quadratic basis functions. in Figure 1. It is also the key to establishing the form of the basis functions for higher order interpolation. since '1   is 1 at  = 0 (node 1) we have '1   = 2  . 0:5  F IGURE 1.5 Two. For example.8.5) The quadratic basis functions are shown. 0:5 and since it is also zero at  = 1 (node 3).4 Quadratic Basis Functions The essential property of the basis functions defined above is that the basis function associated with a particular node takes the value of 1 when evaluated at that node and is zero at every other node in the element (only one other in the case of linear basis functions). 1. '1   must have a factor  . 1 '3    0 0:5 1 (b) '2   = .1. '1   '2   1 1 0 0:5 1 (a) '1   = 2  . 1.

These four bilinear basis functions are illustrated in Figure 1. 2 '2 1. 2  is 1 at node n and zero at the other three nodes. 2 u1 + '2 1. Similarly. '2 1 . 2  = '2 1 '1 2  : : : etc. 2 = 1 . As before the geometry of the element is defined in terms of the node positions xn . 2  receives a contribution from each nodal parameter un weighted by 'n 1 . 2  and that when u 1 . 2 = 1 . 1 2 '4 1. 2 = 1 1 . 2 '3 1. '1 node 3 '3 2 node 4 2 0 1 node 1 '2 node 2 1 '4 1 2 0 1 0 2 1 1 1 F IGURE 1. 2 = 12 (1. n = . yn . 2 u3 + '4 1. Notice that 'n 1 .8 F INITE E LEMENT BASIS F UNCTIONS Let u 1.9: Two-dimensional bilinear basis functions. 1 1 . This ensures that the temperature u 1 . 2  is evaluated at node n it takes on the value un . 2 u2 + '3 1.6) Note that '1 1 . 2 u4 where '1 1. 2  = '1 1 '1 2  where '1 1  and '1 2  are the one-dimensional quadratic basis functions. 2 = '1 1.9.

a six-noded (see Figure 1. 2 = 2 1 . 2  1) and the physical space x. : : : . 2 un '1 1. y . 4 by x= y= X X n n 'n 1. Higher order 2D basis functions can be similarly constructed from products of the appropriate 1D basis functions.1.5 T WO - AND T HREE -D IMENSIONAL E LEMENTS 9 1. 2 2 . For example.10) quadratic-linear element (quadratic in 1 and linear in 2 ) would have u= where 6 X n=1 'n 1. 1 1 . 2  (where 0  1 . 2 yn which provide the mapping between the mathematical space 1 . 1 1 . 2 xn 'n 1.

2 = 21 1 . 2 = 41 1 . 1 2 2 1 . 2 1 . 1 '3 1. 2 '5 1.

1 1 . 2 = 41 1 . '2 1. 2 .

1 1 . 2 2 . 2 = 2 1 . 1 '4 1.

Three-dimensional basis functions are formed similarly.10: A 6-node quadratic-linear element.9) 0 0 0:5 1 1 F IGURE 1.. e. 2 2 (1.g. 2 = 21 1 . 1 '6 1.8) (1. a trilinear element basis has eight .7) (1.

2 3 '7 1. 3 '6 1.11: An 8-node trilinear element.11) (1.12) (1. 3 = 123 7 8 (1. 2. 1 1 . 2 1 . 3 '4 1. A convenient way to achieve this is by defining two additional nodal parameters . 3 '5 1. 1 23 3 5 6 1 2 '2 1. 2 1 . 2. Sometimes it is desirable to use basis functions which also preserve continuity of the derivative of u with respect to  across element boundaries. 2. 1. 3 = 1 1 .10) (1. 3 = 1 . 3 = 12 1 . 3 = 1 1 . 2. 3 '3 1. 2. 2.13) 3 2 4 1 F IGURE 1. 3 = 1 .11) with basis functions '1 1. 2. 1 2 1 . 3 = 1 . 2. 3 = 1 . 2 3 '8 1.6 Higher Order Continuity All the basis functions mentioned so far are Lagrange1 basis functions and provide continuity of u across element boundaries but not higher order continuity.10 F INITE E LEMENT BASIS F UNCTIONS nodes (see Figure 1. 1 1 .

du d n . The basis functions are chosen to ensure that du = du d =0 d .

1 = u01 and du = du d =1 d .

Since the number of element parameters is 4 the basis functions must be cubic in  . To derive these cubic 1 Joseph-Louis Lagrange (1736-1813). 2 = u02 and since un is shared between adjacent elements derivative continuity is ensured. .

3u1 . 2u2  3 or. 3u1 .12. b. c and d back into the original cubic then gives u   = u1 + u01 + 3u2 . d and impose the constraints u 0 = a = u1 u 1 = a + b + c + d = u2 du 0 = b = u01 d du 1 = b + 2c + 3d = u0 2 d These four equations in the four unknowns a.6 H IGHER O RDER C ONTINUITY 11 Hermite2 basis functions let u   = a + b + c 2 + d 3. It is much more useful to define a global node derivative ds n arclength and then use derivative . u02  2 + u01 + u02 + 2u1 . 2u2 Substituting a. 2u01 .14) where the four cubic Hermite basis functions are drawn in Figure 1. 2u01 . b. c and d are solved to give a = u1 b = u01 c = 3u2 . du = b + 2c + 3d 2. The defined at node n is dependent upon the element  -coordinate in the two adn du where s is jacent elements. rearranging. 1 0 0 1 0 u   = 0 1   u1 + 1   u1 + 2   u2 + 2   u2 (1. One further step is required to make cubic Hermite basis functions useful in practice. u02 d = u01 + u02 + 2u1 .1.

du d .

.

du d where 2 .

ds d n n = ds .

du n.e  d .

.15) is an element scale factor which scales the arclength derivative of global node Charles Hermite (1822-1901). ds n (1.

If u is cubic in 1 and cubic in 2 .  to the  -coordinate derivative of element node n. then @u is cubic in  and quadratic 1 @2 in 2 . Thus du ds is constrained to be continuous du across element boundaries rather than . @u u.dimensional bicubic Hermite basis requires four d derivatives per node @u . 1 0 0 1 1   slope = 1 F IGURE 1. 3 + 2 slope = 1 2 1 1   =   . 2  2 1 2   =   . 1 0 1  0 1  1 2 0 2   =  3 . Now consider the side 1–3 The of u with 2 is specified by . @ @ 1 2 and @2u @1 @2 The need for the second-order cross-derivative term can be explained as follows.12: Cubic Hermite basis functions. A two.12 F INITE E LEMENT BASIS F UNCTIONS 1 2 3 0 1   = 1 .

@uin Figure 1.13. .

@u cubic variation @u the four nodal parameters u1 . . @2 1. u3 and @2 3. But since @1 (the normal derivative) is also cubic in 2 along that side and is entirely independent of these four parameters.

@u four additional .

. Two of these are specified by @1 1 and @1 3. @u parameters are required to specify this cubic.

@2u .

is quadratic in 1 and cubic in 2 . and @u @1 . @2u and the remaining two by @1 @2 1 and @1 @2 3.

6 H IGHER O RDER C ONTINUITY 13 2 .1.

@u @1 3 node 3 node 4 .

The bicubic interpolation of these nodal parameters is given by 0 0 0 u 1. @u @1 1 node 1 F IGURE 1.13: Interpolation of nodal derivative @u @1 node 2 1 along side 1–3. 2 = 0 1 1  1 2  u1 + 2 1  1 2  u2 0 0 0 + 0 1 1  2 2  u3 + 2 1  2 2  u4 .

@u .

@u 1 0 1 0 + 2 1 1 2 + 1 1 1 2 @ 1 1 1 2 .

@u .

@ @u 1 0 1 0 + 1 1 2 2 @ + 2 1 2 2 @ .

@u1 3 .

@u1 4 1 1 + 0 + 0 1 1  1 2  @ 2 1  1 2  @ .

@u2 2 .

@u2 1 1 1 + 0 + 0 2 1  2 2  1 1  2 2  @ @ 2 3 .

@2u .

2 @42 u 1 1 + 1 + 1 1 1  1 2  2 1  1 2  @ @ 1 @2 2 .

@12 u 2 1 .

16) . @ 2u @ 1 1 1 1 + 1 1 2 2 @ @ + 2 1 2 2 @ @ 1 2 3 1 2 4 (1.

to preserve derivative continuity in physical x-coordinate space as well as in  -coordinate space the global node derivatives need to be specified with respect to physical arclength. measuring arclength along the 1-coordinate. 12  2 3 .14 F INITE E LEMENT BASIS F UNCTIONS where 0 1   1 1   0 2   1 2   = = = = 1 . There are now two arclengths to consider: s1 . 1 (1.17) are the one-dimensional cubic Hermite basis functions (see Figure 1. and s2. 3 2 + 2 3   . measuring arclength along the 2-coordinate. 2   2  . As in the one-dimensional case above.e  @1 n .12). Thus 1 @1 n = @s1 n.

@s .

@u .

@u 2 @2 n = @s2 n.e  @2 n .

@2u .

@2u .

ds .

e d1 n d2 n 2 . ds 1 2 =   @1 @2 n @s1 @s2 n.

@u .

@u .

14 shows a four element bicubic Hermite surface in 3D space where each node has the following twelve parameters where . @s (1. The bicubic Hermite basis is a powerful shape descriptor for curvilinear surfaces.18) d1 n d2 n are element scale factors which scale the arclength derivatives of global node  to the  -coordinate derivatives of element node n. Figure 1.

ds 1 and .

dimensional elements whose basis functions are formed as the product of onedimensional basis functions). 3 and P x. ds @x .7 Triangular Elements Triangular elements cannot use the 1 and 2 coordinates defined above for tensor product elements (i.. @ 2 x . The natural coordinates for triangles are based on area ratios and are called Area Coordinates . @s @s1 @s2 @s1 @s2 @s1 @s2 @s1 @s2 1 @s2 @s1 @s2 1. Consider the ratio of the area formed from the points 2. z. @z . @y . @y . two. y.and three. @x . @z and @ 2 z x.15 to the total area of the triangle L1 = Area Area x y P 23 = 1 1 1 x y2 = = a1 + b1x + c1y = 2 123 2 1 x2 3 y3 . y  in Figure 1. @ 2 y .e.

y2 L1 = 0 2 L1 = 3 L1 = 1 3 L1 = 1 F IGURE 1. y3 Area P 23 P(x.15: Area coordinates for a triangular element.14: A surface formed by four bicubic Hermite elements. .y ) 1 x1 .1. y1 2 x2 .7 T RIANGULAR E LEMENTS 15 z 12 parameters per node 2 y 1 x F IGURE 1. 3 x3 .

c3 = x2 . Notice that L1 + L2 + L3 = 1. Six node quadratic triangular elements are constructed as shown in Figure 1. x3 y2 .16: Basis functions for a six node quadratic triangular element. y = '1 x. c2 = x1 . y u2 + '3 x. x3 and a3 = x1 y2 . interpolation over the triangle is given by u x. Area coordinate L1 varies linearly from L1 = 0 when P lies at node 2 or 3 to L1 = 1 when P lies at node 1 and can therefore be used directly as the basis function for node 1 for a three node triangle.16 F INITE E LEMENT BASIS F UNCTIONS 1 x1 y1 where  = 1 x2 y2 is the area of the triangle with vertices 123. x1 . Similarly. area coordinates for the other two triangles containing P and two of the element vertices are 1 2 L2 = Area Area x y P 13 = 1 1 1 x y3 = = a2 + b2x + c2y = 2 123 2 1 x3 1 y1 x y P 12 = 1 1 1 x y1 = = a3 + b3x + c3y = 2 123 2 1 x1 2 y2 L3 = Area Area where a2 = x3 y1 . x2 y1 . x2 . b3 = y1 . c1 = x3 . b1 = 1 x3 y3 y2 . 1 '3 = L3 2L3 . x1 y3 . Notice that L1 is linear in x and y . Thus. '2 = L2 and '3 = L3 = 1 . y1 . y u3 where '1 = L1 . y3. L2 . 1 '4 = 4L1 L2 '5 = 4L2 L3 '6 = 4L3 L1 2 4 6 5 3 F IGURE 1.16. and a1 = x2 y3 . y2 . y u1 + '2 x. . L1 . 1 '2 = 2L2 . b2 = y3 . 1 '1 = L1 2L1 .

1.8 Curvilinear Coordinate Systems It is sometimes convenient to model the geometry of the region (over which a finite element solution is sought) using an orthogonal curvilinear coordinate system. . can be modelled geometrically using one element with cylindrical polar r. A 2D circular annulus. for example.8 C URVILINEAR C OORDINATE S YSTEMS 17 1.

g. -coordinates. e. . Notice that element vertices 1 and 2 in r. the first with r = r1 and the second with r = r2 . y 2 1 2 2 3 2 2 3 4 x 0 r1 1 4 r2 (b) r 1 1 (a) (c) F IGURE 1. the annular plate in Figure 1.17a has two global nodes..17: Defining a circular annulus with one cylindrical polar element.

Global nodes 1 and 2. 2 -space. shown in x. . each map to two element vertices r. as shown in (b) and (c). map onto the single global node 1 in xnode. Similarly. respectively. y -space in (a). element vertices 3 and 4 map onto global node 2.17a. -space or 1 . y -space in Figure 1.

as shown in Figure 1. 2-space. . The r.-space.17b. as shown in Figure 1.17c. and in 1.

 coordinates at any 1. 2 point are given by a bilinear interpolation of the nodal coordinates rn and .

n as in r = 'n 1. 2  rn .

= 'n 1. 2  .

2  are given by (1. . Cylindrical polar r. Three orthogonal curvilinear coordinate systems are defined here for use in later sections.n where the basis functions 'n 1 .6).

. z  : x = r cos .

y = r sin .

z=z (1.19) .

.18 Spherical polar r.

. F INITE E LEMENT BASIS F UNCTIONS : x = r cos .

cos y = r sin .

. . cos z = r sin Prolate spheroidal .

20) x = d cosh  cos  y = d sinh  sin  cos .  : (1.

z = d sinh  sin  sin .

(1.21) .

r z y  d  x F IGURE 1.18 and a single prolate spheroidal element is shown in Figure 1. .19. The coordinates . .18: Prolate spheroidal coordinates. The prolate spheroidal coordinates rae illustrated in Figure 1.

Only four global nodes are required provided the four global nodes map to eight element nodes as shown in Figure 1.19. .  are all trilinear in 1 . 3 . 2 .

8 C URVILINEAR C OORDINATE S YSTEMS 19 (a) 24 y (b) z 2 1 1 3 3 x (c)  90o 0 2 2 1 2 4 1 3 3 4 (d) 2 2 2 1 4 1 4 3 3  1 3 .1.

(c) in . . shown (a) in x. . y. z -coordinates. F IGURE 1.19: A single prolate spheroidal element.

. 3 -coordinates. 2 . (b) shows the orientation of the i -coordinates on the prolate spheroid.-coordinates and (d) in 1 .

21.20.22. To refine a mesh run the CMISS example 113. 3.9 CMISS Examples 1.23). To define a 2D cubic Hermite-linear finite element mesh run example 114.20: Node positions for example 111. To define a quadratic-linear element run the cmiss example 115. To define a triangular element mesh run CMISS example 116 (see Figure 1. The nodes should be positioned as shown in Figure 1. To define a bilinear mesh in cylindrical polar coordinates run CMISS example 122. After defining elements the mesh should appear like the one shown in Figure 1. 1 2 F IGURE 1. 4.20 F INITE E LEMENT BASIS F UNCTIONS 1. . To define a 3D trilinear element run CMISS example 121. 6. 7. To define a 2D bilinear finite element mesh run the CMISS example number 111. After the first refine the mesh should appear like the one shown in Figure 1. 5.21: 2D bilinear finite element mesh for example 111. 4 2 6 5 1 3 F IGURE 1. 2.

9 CMISS E XAMPLES 21 4 8 5 10 6 1 7 2 9 3 F IGURE 1.23: Defining a triangular mesh for example 116 .1.22: Refined mesh for example 113 3 4 1 2 F IGURE 1.

Chapter 2 Steady-State Heat Conduction
2.1 One-Dimensional Steady-State Heat Conduction
Our first example of solving a partial differential equation by finite elements is the one-dimensional steady-state heat equation. The equation arises from a simple heat balance over a region of conducting material: Rate of change of heat flux = heat source per unit volume or

d (heat flux) + heat sink per unit volume = 0 dx
or

d ,k du + q u; x = 0 dx dx where u is temperature, q u; x the heat sink and k the thermal conductivity (Watts/m/ Consider the case where q = u

du d , dx k dx + u = 0 0 x 1 subject to boundary conditions: u 0 = 0 and u 1 = 1. This equation (with k = 1) has an exact solution ,ex , e,x u x = e2 e ,1
with which we can compare the approximate finite element solutions. To solve Equation (2.1) by the finite element method requires the following steps: 1. Write down the integral equation form of the heat equation.

(2. C). Integrate by parts (in 1D) or use Green’s Theorem (in 2D or 3D) to reduce the order of derivatives.2) 2.1) (2. .

7. 6. Integrate over the elements to calculate the element stiffness matrices and RHS vectors.24 S TEADY-S TATE H EAT C ONDUCTION 3. Solve the global equations.1) directly. dx dx . Assemble the global equations.1. we form the weighted residual Z R!:dx = 0 (2. 8. Evaluate the fluxes. 4. Introduce the finite element approximation for the temperature field with nodal parameters and element basis functions. 2. 5.1 Integral equation Rather than solving Equation (2. Apply the boundary conditions.3) where R is the residual d k du + u R = .

Substituting Equation (2. we try to obtain an approximate solution u for which the residual or error (i. given that in real engineering problems this will not be the case.e.3) gives Z1 0 d k du ! + u! dx = 0 . the residual R would be zero everywhere. But.. (2. the amount by which the differential equation is not satisfied exactly at a point) is distributed evenly over the domain.4) into Equation (2. dx dx .4) for an approximate solution u and ! is a weighting function to be chosen below. If u were an exact solution over the whole domain.

(2.k dx . ! is chosen such that the residual is kept orthogonal to the space of functions used in the approximation of u (see step 3 below). equivalently for 2D problems.1. More precisely.see later). 2.2 Integration by parts A major advantage of the integral equation is that the order of the derivatives inside the integral can be reduced from two to one by integrating by parts (or. Thus. by applying Green’s theorem .5) This formulation of the governing equation can be thought of as forcing the residual or error to be zero in a spatially averaged sense. substituting f du into the integration by parts = ! and g = .

1 O NE -D IMENSIONAL S TEADY-S TATE H EAT C ONDUCTION 25 formula Z1 dg Z1 df 1 f dx dx = f:g 0 .2. g dx dx 0 0 gives  .

du 1 Z1 .

du d! Z1 d .

5) becomes  du 1 Z1 .k dx dx = ! . . du ! dx .k dx .k dx dx dx 0 0 0 and Equation (2.

We also choose ! = 'm (called the Galerkin1 assumption). Bubnov in 1913.  and '2   =  are the linear basis functions for both u and x.3 Finite element approximation We divide the domain 0 x 1 into 3 equal length elements and replace the continuous field variable u x within each element by the parametric finite element approximation u   = '1   u1 + '2   u2 = 'n   un x   = '1   x1 + '2   x2 = 'n   xn (summation implied by repeated index) where '1   = 1 . thereby ensuring that the residual.G. In many Russian texts the Galerkin finite element method is known as the Bubnov-Galerkin method. or error.1. The domain integral in Equation (2.6) can now be replaced by the sum of integrals taken separately over the three elements Z1 0 1  dx = Z 0 1 3  dx +  dx + 1 3 Z 2 3 Z1 2 3  dx Boris G. . is monotonically reduced as the finite element mesh is refined (see later for a more complete justification of this very important step) . He published a paper using this idea in 1915. Galerkin (1871-1945). The method was also attributed to I. This forces the residual R to be orthogonal to the space of functions used to represent the dependent variable u.6) 2. Galerkin was a Russian engineer who published his first technical paper on the buckling of bars while imprisoned in 1906 by the Tzar in pre-revolutionary Russia. du d! k + u! dx = k ! 0 dx dx dx 0 (2.

6) have the form Z1 . 2.4 Element integrals The element integrals arising from the LHS of Equation (2.26 and each element integral is then taken over  -space S TEADY-S TATE H EAT C ONDUCTION Zx 2 x1 where J  dx = Z1 0 J d = dx d is the Jacobian of the transformation from x coordinates to  coordinates.1.

Since 'n and 'm are both functions of  the derivatives with respect to x need to be converted to derivatives with respect to  . Thus Equation (2. du d! + u! J d k 0 dx dx (2.7) where u = 'n un and ! = 'm .7) becomes Z1 .

Emn Notice that un has been taken outside the integral because it is not a function of  . d'm d d'n d + 'm 'n J d un k 0 d dx d dx (2. The term multiplying the nodal parameters u is called n dx d 3 the element stiffness matrix. In this case x =  or 3 d = 3 and the Jacobian is J = dx = 1 . The term 1 Z1 .8) d is dx 1 evaluated by substituting the finite element approximation x   = 'n :Xn .

d'm d d'n d Z1 .

To evaluate Emn we substitute the basis functions 123 '1   = 1 .  '2   =  d'1 = . d'm d'n Emn = k + 'm'n J d = k 3 3 + 'm'n d 0 d dx d dx 0 d d 3 where the indices m and n are 1 or 2.1 d d'2 = 1 or d or .

The bar is shown at the top divided into three elements. ! 1  . Thus.2.1 O NE -D IMENSIONAL S TEADY-S TATE H EAT C ONDUCTION 27 x Node 1 Node 1 2 3 4 X Node 2 X X X 0 0 0 0 X X U2 U2 = U3 U4 X X X X X Node 3 X X X 0 Node 4 0 F IGURE 2.1: The rows of the global stiffness matrix are generated from the global weight functions.

1 Z Z1 , 2 1 d' 1 1 1 2 2 2 9k + '1 d = 9k ,1 + 1 ,   d = 9k + E11 =
3
0

d

3

0

3

3

and, similarly,

1 E12 = E21 = 1 , 9 k + 3 6

9k + 1 E22 = 1 3 , 3 ,  1 9k + 1 1 ,9k + 1  3 3 ,   Emn = 1 ,9k + 1 31 ,9k + 16
3 6 3 3

Notice that the element stiffness matrix is symmetric. in this particular case.1. is the same for all elements. For simplicity we put k = 1 in the following steps.1) overlaps . This process is illustrated in Figure 2.1 where rows 1. ::. 2. 4 of the global stiffness matrix (shown here multiplied by the vector of global unknowns) are generalised from the weight function associated with nodes 1. Note how each element stiffness matrix (the smaller square brackets in Figure 2. 4.5 Assembly The three element stiffness matrices (with k = 1) are assembled into one global stiffness matrix. ::. Notice also that the stiffness matrix.

since the basis functions are local to elements. 18 5 4U37 53 28 0 0 . 18 9 + 9 . 53 0 0 1 9 18 53 28 + 28 53 7 6 6 U . The RHS of Equation (2.containing many zeros .28 S TEADY-S TATE H EAT C ONDUCTION with its neighbour because they share a common global node.6) is  du x=1 . 18 9 U4 Notice that the first row (generating heat flux at node 1) has zeros multiplying U3 and U4 since nodes 3 and 4 have no direct connection through the basis functions to node 1. . 0 27 18 9 9 18 7 6 6 28 28 53 53 5 4 0 . The assembly process gives 3 2U 3 2 28 . Finite element matrices are always sparse matrices .

du = k dx ! k dx ! x=0 x=1 .

!1 jx=1 = 0. Thus Equation (2. k dx ! x=0 (2.9) for node 1 reduces to  du x=1 . For node 1 !1 is obtained from the basis function '1 associated with the first node of element 1 and therefore !1 jx=0 = 1.6).9) To evaluate these expressions consider the weighting function ! corresponding to each global node (see Fig. du . Also. since !1 is identically zero outside element 1.1.

 du x=1 =0 k dx !n x=0 (nodes 2 and 3) and  du x=1 . du = . Similarly. k dx k dx !1 x=0 x=0 = flux entering node 1.

18 9 + 9 . Putting these global equations together we get 2 28 . 53 18 9 2 . 18 6 4 0 . Note: k has been left in these expressions to emphasise that they are heat fluxes. 53 0 9 18 53 28 28 53 6 . 53 28 + 28 0 0 18 9 . du = k k dx !4 dx x=0 x=1 = flux entering node 4.

du 3 3 2U 3 6. k dx 7 0 1 x=0 7 6 7 6 6 7 0 7 6U2 7 0 7 6 7 = 53 7 . 18 5 4U3 5 6 0 6 7 .

10) k dx x=1 or Ku = f . 28 4 5 du U 9 4 (2.

2.8 Fluxes The fluxes at nodes 1 and 4 are evaluated by substituting the nodal solutions U1 U3 = 0:6089 and U4 = 1 into Equation (2. corresponding to an insulated boundary.6 Boundary conditions The boundary conditions u 0 = 0 and u 1 = 1 are applied directly to the first and last nodal values: i. 2. where the flux terms on the RHS are at present unknown 1st equation U1 56 53 2nd equation .1.as shown in the next section. since effectively no additional constraint is applied to the global equation. if a flux boundary condition had been applied. U2 = 0:2855.10) = 0. 18 U3 56 53 3rd equation . the vector of unknowns and the global “load” vector. Note that if the governing differential equation had included a distributed source term that was independent of u.1 O NE -D IMENSIONAL S TEADY-S TATE H EAT C ONDUCTION 29 where is the global “stiffness” matrix.7 Solution Solving these equations gives: U2 = 0:2885 and U3 = 0:6098. Moreover. the contribution from each element would be different .e.on the RHS of Equation (2. 53 18 U1 + 9 U2 .1.via its weighted integral . 18 U4 th 4 equation U4 =0 =0 =0 =1 Note that. An applied boundary flux of zero. the known value of flux would enter the appropriate RHS term and the value of U at that node would remain an unknown in the system of equations.1. this term would appear . is termed a natural boundary condition. 53 18 U2 + 9 U3 . The finite element solution is shown in Figure 2.2. 2.10). respectively. From Equation (2. if the source term was a function of x. These so-called essential boundary conditions then replace the first and last rows in the global Equation (2. K u f 2. At least one essential boundary condition must be applied.10) rather than on the LHS as here..2) the exact solutions at these points are 0:2889 and 0:6102. . rather than an essential boundary condition. U1 = 0 and U4 = 1.

exact solution 0:8509) dx x =0 . k = .0:8496 (k = 1. du flux entering node 1 = .

exact solution 1:3131) These fluxes are shown in Figure 2.2 as heat entering node 4 and leaving node 1. du flux entering node 4 = k dx x=1 = 1:1357 (k = 1. . consistent with heat flow down the temperature gradient.

1): d k du + u . dx dx Equation (2. x = 0 0 x 1 .30 S TEADY-S TATE H EAT C ONDUCTION u 1:3156 1 0:6098 0:2855 0:8406 0 1 3 2 3 1 x F IGURE 2.2: Finite element solution of one-dimensional heat equation.6) now becomes . 2.2 An x-Dependent Source Term Consider the addition of a source term dependent on x in Equation (2.

 du 1 Z1 Z1 .

12) . du d! k + u! dx = k ! + x! dx 0 dx dx dx 0 (2. Replacing the domain integral for this source term by the sum of three element integrals Z1 0 x! dx = Z 0 1 3 x! dx + Z 1 3 2 3 x! dx + Z1 2 3 x! dx and putting x in terms of  gives (with Z1 0 x! dx = 1 3 Z1  Z1 1 3 ! d + 3 0 0 dx = 1 for all three elements) d 3 1 +   ! d + 1 3 3 Z1 0 2 +   ! d 3 (2.11) 0 where the x-dependent source term appears on the RHS because it is not dependent on u.

Equation (2.12) corresponding to element 1 is '2 =  . Evaluating these expressions. . the contribution to the element 1 RHS vector from the source term is 54 1 . where '1 = 1 .  and Z1 1  1 . 27 1 Z1 0 1 1 +   1 .3 T HE G ALERKIN W EIGHT F UNCTION R EVISITED 31 where ! is chosen to be the appropriate basis function within each element.   d = 1 9 54 and Z1 0 1  2 d = 1 9 27 Thus.   d = 7 and 9 54 Z1 0 1 2 +    d = 5 gives 9 54 7 54 5 54 Assembling these into the global RHS vector. 5618 0 0 18 . Similarly. 0 1 9 Z1 0 'm d . Z1 0 1 2 +   1 . the first term on the RHS of (2.2. For example.   d = 2 and 9 27 Z1 0 1 1 +    d = 5 gives 9 54 2 27 5 54 and for element 3. for element 2. 53 18 9 6 53 4 0 .10) becomes 2 28 . 53 18 9 2 . 53 0 9 18 53 56 6 .

du 3 3 2U 3 6. k dx 7 2 1 3 0 1 54 x=0 7 6 1 + 27 7 6 7 6 6 0 0 7 U 2 27 27 7 7 76 7 6 +6 5 + 75 53 5 4U 5 = 6 7 4 0 . 18 3 54 54 7 6 28 5 4 .

In Figure 2.3 The Galerkin Weight Function Revisited A key idea in the Galerkin finite element method is the choice of weighting functions which are orthogonal to the equation residual (thought of here as the error or amount by which the equation fails to be exactly zero).3. The difference between the exact vector e and the approximate vector is the ' u u u u' u . du 5 U 9 4 k dx x=1 54 2.3a an exact vector e (lying in 3D space) is approximated by a vector = 1 1 where 1 is a basis vector along the first coordinate axis (representing one degree of freedom in the system). This idea is illustrated in Figure 2.

3b) is added.3c) is permitted in the approximation = u1 '1 + u2 '2 + u3 '3 with the result that the residual (now also orthogonal to '3 ) is reduced to zero and = e . The Galerkin technique minimises this residual by making it orthogonal to '1 and hence to the approximating vector . (shown by the broken line in Figure 2. If a second degree of freedom (in the form of another coordinate axis in Figure 2.32 S TEADY-S TATE H EAT C ONDUCTION (a) (b) '3 u3 (c) ue u u1 R '1 u2 '2 u R u u = u1'1 r  '1 = 0 u = u1'1 + u2'2 : : : + r  '2 = 0 u = u1'1 + u2'2 + u3'3 : : : + r  '3 = 0 F IGURE 2. but in the infinite dimensional vector space associated with a spatially continuous field u x we need to impose the equivalent orthogonality r u u u u u ' u ' . For a 3D vector space we only need three axes or basis vectors to represent the true vector . a third degree of freedom (a third axis in Figure 2.3c. error or residual = e . in Figure 2. Finally.3: Showing how the Galerkin method maintains orthogonality between the residual vector r and the set of basis vectors 'i as i is increased from (a) 1 to (b) 2 to (c) 3.3a). the approximating vector is = u1 1 + u2 2 and the residual is now also made orthogonal to '2 and hence to .

equivalently.4 Two and Three-Dimensional Steady-State Heat Conduction Extending Equation (2. The key point is that in this analogy the residual is made orthogonal to the current set of basis vectors . to the set of basis functions used to represent the dependent variable. in finite element analysis. @ k @u = 0 .1) to two or three spatial dimensions introduces some additional complexity which we examine here. @ k @u . Consider the three-dimensional steady-state heat equation with no source terms: @ k @u . This ensures that the error or residual is minimal (in a least-squares sense) for the current number of degrees of freedom and that as the number of degrees of freedom is increased (or the mesh refined) the error decreases monotonically.Z condition u x.or. @x x @x @y y @y @z z @z . R'dx = 0 u u u for every basis function ' used in the approximate representation of 2.

13).4). This is used (with g = ku and f = ! ) to reduce the derivative order from two to one as follows: Z r  . subject to boundary conditions on . ky and kz are the thermal diffusivities along the x. Solution region: F IGURE 2. 1993).15) (see p553 in Advanced Engineering Mathematics” by E. . Here we consider the solution of Equation (2. this can be written as y and z axes respectively.16) Z d .. Kreysig.r  kru ! d = 0 (2.14) The multi-dimensional equivalent of integration by parts is the Green-Gauss theorem: Z Z @g f r  rg + rf  rg d = f @n d. The weighted integral equation. this reduces to Laplace’s equation k r2 u = 0. 7th edition. Wiley. is Z . if k is spatially constant.r  kru = 0 (2.13) over the region . .2. Solution region boundary: .13) and. ky = kz = k.kru ! d = Z Z @u kru  r! d . corresponding to Equation (2. = .4: The region and the boundary . (see Figure 2. (2. k @n ! d.4 T WO AND T HREE -D IMENSIONAL S TEADY-S TATE H EAT C ONDUCTION 33 If kx where kx . (2.

k dx ! dx = k dx dx dx . x x 2 1 Using Equation (2. k dx ! x .14) gives the two-dimensional equivalent of Equation (2.6) . du Z du d!  du x cf.16) in Equation (2. Integration by parts is dx .

5 Basis Functions . . for a two-dimensional element. @ 1 @1 2.34 S TEADY-S TATE H EAT C ONDUCTION (but with no source term): Z Z @u kru  r! d = k @n ! d.Element Discretisation = i . i. @x @y @1 @2 @1 @2 @2 @1 2 @y @x 3 . 2 @1 @1 3 2 @x @x 3. In each i let i=1 u = 'nun = '1u1 + '2u2 + : : : + 'N uN and map each i to the 1.18) @u  @! = @u @i  @! @j ru  r! = @x @i @xk @j @xk k @xk where inverse matrix @i u = 'nun and ! = 'm. and the geometric terms @x k are found from the  @   @x .1 i = k @xk @i or. (2.e.1 1 6 @x @y 7 1 @2 7 4 @ 5=6 4 @ @y @y 5 = @x @y @x @y 2 @2 . @ 7 6 @ 2 4 @y @x 2 5 . 2 plane.5 shows an Let example of this mapping. Figure 2.17) is evaluated using @u being given on another part of the @n (2. as before. I .17) subject to u being given on one part of the boundary and boundary. the solution region is the union of the individual elements.. The integrand on the LHS of (2.

25) (2.   2 @1 @'1 = .27) (2.22) (2.28) (2.29) '2 = 11 . 1 @2 @'3 = .19) (2.24) (2.  2 @1 @'1 = .1 . 2  @'1 = .   1 @2 @'2 = 1 .20) (2.1 . 2 @1 @'3 = 1 .23) (2. the basis functions and their derivatives are: '1 = 1 .E LEMENT D ISCRETISATION 35 2 y 7 3 2 8 3 1 7 8 4 1 8 1 9 4 9 0 4 0 6 5 1 0 5 0 6 1 1 5 4 2 1 2 3 1 4 0 1 0 1 2 5 2 1 1 5 6 2 2 1 1 0 2 0 3 1 1 x F IGURE 2.5: Mapping each to the 1 .21) (2. 2 plane in a 2  2 element plane.26) (2.2. 12 '4 = 12 . 2 '3 = 1 .  1 @2 @'4 =  @1 2 @'4 =  @2 1 (2. 11 . For each element.5 BASIS F UNCTIONS .

e. .. Z . (2.36 S TEADY-S TATE H EAT C ONDUCTION 2.6 Integration The equation is Z Z @u ru  r! d = k @n ! d.30) i.

This gives X Z . (2. weight function is one of the basis functions used to approximate the dependent variable..e. @u @! @u @! k + d @x @x @y @y Z @u = k @n ! d.31) u has already been approximated by 'n un and ! is a weight function but what should this be chosen to be? For a Galerkin formulation choose ! = 'm i. .

. : : : . .32) where the stiffness matrix is Emn where m = 1.. y 2 1 0 1 x 1 F IGURE 2. i. @n . @'n @'m @'n @'m un k + d i @x @x @y @y Z @u = k 'm d. 4 and Fm is the (element) load vector. The names originated from earlier finite element applications and extension of spring systems. This yields the system of equations Emn un = Fm .g. F = kx where k is the stiffness of spring and F is the force/load. : : : .e. (2.6).6: Considering heat flow in a unit square. heat flow in a unit square (see Figure 2. e. 4 and n = 1.

6 3 3 6 7 6 2 7 = RHS k6 1 2 1 5 4u 5 4.3 . consider 4 elements (each of unit size) and nine nodes.1 3 6 6 2+2 1 6 . 1 6 . typically there is more than one element and so the next step is required.3 1 1 . In this case we would have to include the Jacobian of the transformation and @'i . 6 3 + 3 .6 . 6 3 2u 3 1 7 6 .. 2 2 .1 6 2 3 u9 (2.1 6. 3 7 6u57 7 2+2 1 1 7 6u 7 . 6 3 6 1 1 2 2 6 . 1 .1 . y2 + 1 . 6 u8 5 .6 1 . Each element has the same element stiffness matrix as that given above. .3 1 1 .33) u4 Note that the Galerkin formulation generates a symmetric stiffness matrix (this is true for self adjoint operators which are the most common). This is because each element is the same size. 1 . 6 .1 .1 .1 6. 6 . For example.3 2 3 2+2+2 +3 3 3 1 . 1 . . 3 6 3 3 6 1 1 1 2 . Note that if the element was not the unit square we would need to transform from x. 2. 1 7 6u 7 6 . 2 coordinates.3 6 6 3 6 6 6 1 1 6 . .6 3 (Right Hand Side) (2. 1 .7 A SSEMBLE G LOBAL E QUATIONS The first component E11 is calculated as 37 E11 = k Z1 Z1 0 0 1 . 6 3 3+3 6 1 . 3 6 6 6 . shape and interpolation.34) .6 u 3 7 6 7 1 1 7 6 7 .1 . 1 3 2u 3 1 3 6 6 3 1 2 . 1 .7 Assemble Global Equations Each element stiffness matrix must be assembled into a global stiffness matrix.6 1 .2 .1 6 .6 .1 6 3 3 3 6 1 2 6 . @'n = @'n @1 + @'n @2 = @'n @i (Refer to also use the chain rule to calculate @xj @x @1 @x @2 @x @i @x Assignment 1) The system of Emn un = Fm becomes 2 2 . e. Given that boundary conditions can be applied and it is possible to solve for unknown nodal temperatures or fluxes.6 .g.1 u 27 3 7 6 7 1 7 6 7 .1 u7 7 7 6 7 3 6 1 1 2 2 1 5 4 .1 6 6 1 4 .6 . x2 dxdy =2 3k and similarly for the other components of the matrix.6 . However.3 u 4 7 6 7 = RHS 1 1 1 1 1 1 7 6 . 3 . 6 1 .1 3 1 . 3 .2. 6 3 3 3 67 6 7 2 7 6 . y  to 1. 6 . .3 . 6 .

6 7 6u67 7 6 7 7 6 u 77 1 6 5 . . ui is known. The final system to solve is only as big as the number of unknown values of u.1 . 1 . It should also be clear that the matrix will be sparse if there is a larger number of elements.3 1 . . we can remove the ith equation and replace it with the known value of ui . From this system of equations.1 .3 1 8 1 .6 1 . 6 4u87 5 2 3 u9 Note that the matrix is symmetric. If at global node i.1 .3 6 .1 .6 3 4 1 .1 6 6 1 6 .1 6 .3 3 6 1 . 3 6 6 1 6 1 3 1 3 1 3 1 3 1 3 3 2u13 6 u27 7 6 7 7 6 u37 7 6 7 7 6 7 u 4 7 6 7 7 6 u 57 1 7 6 . 2 1 . This is because the RHS at node i is known but the RHS equation is uncoupled from other equations so the equation can be removed. . 3 6 6 6 6 6 4 . 1 . As an example to illustrate this consider fixing the temperature (u) at the left and right sides of the plate in Figure 2.1 4 . This yields the system of equations 22 3 6 .3 3 .38 S TEADY-S TATE H EAT C ONDUCTION y 7 3 4 5 8 4 6 2 2 3 9 global node numbering element numbering 1 1 x F IGURE 2. 6 3 3 3 1 2 . To solve. This means that .7: Assembling 4 unit sized elements into a global stiffness matrix.1 .1 6 6 6 6 6 .3 1 . . firstly boundary conditions are applied to reduce the size of the system.3 4 3 1 6 .1 . 3 3 3 6 3 1 1 . 3 7 6u67 = RHS 7 1 7 6 7 .3 3 3 3 1 4 1 . Therefore the size of the system is reduced.1 . .1 3 6 . . boundary conditions can be applied and the equations solved.7 and insulating the top (node 8) and the bottom (node 2).1 .1 .

or three-dimensional problems can seldom be evaluated analytically. therefore there is a 3  3 matrix to solve. We now choose the Gauss points and weights to exactly integrate a polynomial of degree 2I . For example.called Gauss points . Consider first the problem of integrating f   between the limits 0 and 1 by the sum of weighted samples of f   taken at points 1 .8). : : : .3): Z1 0 f   d = I X i=1 Wif i + E Here Wi are the weights associated with sample points i . = 0 at node 5.5 and 8). Numerical integration or quadrature is therefore required and the most efficient scheme for integrating the expressions that arise in the finite element method is Gauss-Legendre quadrature. @n 2. Correct way: on . 1 has 2I arbitrary coefficients and there are 2I unknown Gauss points and weights). does not pass through node 5 and each basis function that is not zero at 5 is zero @u is opposite in neighbouring elements so it cancels (see Figure 2. To find out what the RHS is at node 5 Z @u we need to examine the RHS expression @n ! d. Other way: . This is zero as flux is always 0 at internal nodes. 2 .and E is the error in the approximation of the integral. The RHS is known at these three nodes (see below).8: “Cancelling” of flux in internal nodes. 1 . 2 n n F IGURE 2. 1 (since a general polynomial of degree 2I .8 G AUSSIAN Q UADRATURE 39 there are only 3 unknown values of u at nodes (2. I (see Figure 2.8 Gaussian Quadrature The element integrals arising from two. We can then solve the 3  3 matrix and then multiply out the original matrix to find the unknown RHS values. This can be explained in two ways. The RHS is 0 at nodes 2 and 8 because it is insulated.2. with I = 2 we can exactly integrate a polynomial of degree 3: .

f   is sampled at I Gauss points 1 ..39) These four equations yield the solution for the two Gauss points and weights as follows: .35) Since a. b..36) 1 = W : + W :  d = 2 1 1 2 2 2 + W : 2  2 d = 1 = W1 :1 2 2 3 3 + W : 3  3 d = 1 = W1 :1 2 2 4 (2.40 S TEADY-S TATE H EAT C ONDUCTION f   .. Thus.9: Gaussian quadrature.. Then Z1 0 f   d = a Z1 0 d + b Z1 0  d + c Z1 0  d + d 2 Z1 0  3 d (2. 2 : : : I : Z1 Let 0 f   d = W1 f 1 + W2 f 2 and choose f   = a + b + c 2 + d 3 .38) Z1 0 (2.35 must be integrated exactly. I 1  F IGURE 2.37) (2. Z1 Z1 Z 0 0 1 0 d = 1 = W1 :1 + W2:1 (2.. 0 1 2 .. each integral on the RHS of 2. c and d are arbitrary coefficients.

and three@i terms which come from the dimensional problems the integral is not polynomial (owing to the @xj . 1 = 2 2 3 1 1 .g. 1 r 5 W1 = 18 W2 = 4 9 5 W3 = 18 (2. for 2x2 Gauss quadrature the 4 weights are all . 2 = 2 + p 2 3 1 W1 = W2 = 2 A similar calculation for a 5th degree polynomial using three Gauss points gives (2.41) 2 For two. 2 2 5 1. W1 = W2 = 1 2: Then.39) is satisfied identically.. Thus. substituting in (2. 21 1 3 giving 1 : p  1 = 1 2 2 3 Equation (2. p 1 . the two Gauss points are given by 1. 1 and.8 G AUSSIAN Q UADRATURE 41 From symmetry and Equation (2. The number of Gauss points chosen for each i -direction is 4 governed by the complexity of the integrand in the element integral (2.37). 2 = 2 r 1 1 3 = 2 + 2 3 5.8).38).or three-dimensional Gaussian quadrature the Gauss point positions are simply the values given above along each i -coordinate with the weights scaled to sum to 1 e.  2 = 2 1 1 3 2 .2. In general two. 2 + 1 = 0. 2 + 1 .40) 1 3. 2 = 1 . 1 = 1 . from (2.36).

if the two-dimensional basis is cubic in the 1 -direction and linear in the 2 -direction.9 CMISS Examples 1. For example. . To solve for the steady state temperature distribution inside a plate run CMISS example 311 2. To investigate the convergence of the steady state temperature distribution with mesh refinement run CMISS examples 3141. 2. 3143 and 3144. three Gauss points would be used in the 1-direction and two in the 2-direction. 3142.42 S TEADY-S TATE H EAT C ONDUCTION inverse of the matrix  @x  @j i ) and no attempt is made to achieve exact integration. To solve for the steady state temperature distribution inside an annulus run CMISS example 312 3. The quadrature error must be balanced against the discretization error.

around 1926-27. It was rigorously defined as a so-called generalised function by Schwartz in 1950-51.Chapter 3 The Boundary Element Method 3. Dirac (1902-1994) was awarded the Nobel Prize (with Erwin Schrodinger) in 1933 for his work in quantum mechanics. 3.1 Dirac-Delta function What we do here is very non-rigorous. To gain an intuitive feel for this unusual function. one of which involves the choice of weighting function (recall the Galerkin finite element method used as a weighting function one of the basis functions used to approximate the solution variable). Fundamental solutions are tied to the Dirac1 Delta function and we deal with both here. 3. and strictly speaking we should talk about the “Dirac Delta Distribution”. consider the following sequence of force distributions applied to a large plate as shown in Figure 3.2.2 The Dirac-Delta Function and Fundamental Solutions Before one applies the boundary element method to a particular problem one must obtain a fundamental solution (which is similar to the idea of a particular solution in ordinary differential equations and is the weighting function). Before launching into the boundary element method we must briefly develop some ideas that are central to the weighting function used in the boundary element method. we now develop the basic ideas of the boundary element method.1 Introduction Having developed the basic ideas behind the finite element method. There are several key differences between these two methods. as we will do here. .1 n jxj 2 wn x = 0 jxj 1 n 1 n 1 Paul A. Dirac introduced the idea of the “Dirac Delta” intuitively.M.

the “function” is zero everywhere except at the origin. It also has the property that for any continuous function h x .e. we have The Dirac delta “function” is not a function in the usual sense.1 x h x dx = h 0 (3. and it is more correctly referred to as the Dirac delta distribution. 2 w4 w3 w2 w1 3 2 1 1 2 1 4 1 3 1 2 1 F IGURE 3.1 wn x dx = 1 (i. where it is infinite.1 wn x dx = 1..1) . Z1 . given the symbol x. the magnitude of the force increases but the total force applied remains unity. in a nonrigorous sense we have x = nlim w x !1 n the Dirac Delta“function”. the total force applied is unity) but as n increases the area of force application decreases and the force/unit area increases. As n gets larger we can easily see that the area of application of the force becomes smaller and smaller.1 x dx = 1 since each Z1 . Z1 However. Thus. This is not a function that we are used to dealing with because we have x = 0 if x 6= 0 and “ 0 = 1” i..1: Illustrations of unit force distributions wn . If we imagine letting n ! 1 we obtain an idealised “point” force of unit strength. acting at x = 0.e.44 T HE B OUNDARY E LEMENT M ETHOD Each has the property that Z1 .

n n . where  2 .1 wn x h x dx 1 by definition of x = nlim !1 2 n nZ 1 .n h x dx by definition of wn x n h   2 = nlim !1 2 n = h 0 1 1 by the Mean Value Theorem. .3.2 T HE D IRAC -D ELTA F UNCTION AND F UNDAMENTAL S OLUTIONS 45 A rough proof of this is as follows Z1 .1 x h x dx = nlim !1 Z1 .

. n n and as n ! 1.  ! 0 . 1 1 since  2 .

4) 3.e. x h x dx = h   (3.2.)   . x.2 Fundamental solutions We develop here the fundamental solution (also called the freespace Green’s3 function) for Laplace’s Equation in two variables.) (3. Most of the common equations Oliver Heaviside (1850-1925) was a British physicist. but rather its integral behaviour. 2 . y =  . t = step function. Some properties of the Dirac delta are listed below Z1 .2)  . One does not usually talk about the values of the Dirac delta at a particular point. who pioneered the mathematical study of electrical circuits and helped develop vector analysis. x = H 0  .e. 3 George Green (1793-1841) was a self-educated miller’s son.1)) is often used as the defining property of the Dirac delta in more rigorous derivations.. the Dirac Delta function is the slope of the Heaviside2  . Most widely known for his integral theorem (the Green-Gauss theorem).3) t t (i. x  . the two dimensional Dirac delta is just a product of two one-dimensional Dirac deltas. The above result (Equation (3.. The fundamental solution of a particular equation is the weighting function that is used in the boundary element formulation of that equation.1 (Note:  . y (i. It is therefore important to be able to find the fundamental solution for a particular equation. t 0 1 if  if  (3. x is the Dirac delta distribution centred at x =  instead of x = 0) where H  . .

We expect the solution to be symmetric about the point . y = 0 @x2 @y2 (3.. . . from Section 1. We briefly illustrate here how to find a simple fundamental solution. . x.5) in 2 (i.46 T HE B OUNDARY E LEMENT M ETHOD have well-known fundamental solutions (see Appendix 3. This is not as difficult as it sounds. x. x2 +  .8 we have 2 q @ r @! + 1 @ 2 ! r!=1 r @r @r r2 @.e. @ 2 ! + @ 2 ! +  . The method is to try and find solution to r2 ! = 0 in 2 which contains a singularity at the point . y  is symmetric about this point. Let r =  . So we adopt a local polar coordinate system about the singular point .16). Consider solving the Laplace Equation The fundamental solution for this equation (analogous to a particular solution in ODE work) is a solution of @ 2 u + @ 2 u = 0 in some domain 2 @x2 @y2 2. .  since  . we solve the above without reference to the original domain or original boundary conditions). y2 Then.

2 For r 2 .

. x. y = 0 and owing to symmetry.6) 0.  . @ ! @. (3.

Thus Equation (3.2 is zero.6) becomes .

To find A and B we make use of the integral property of the Delta function. If D is a small disk of .5) we must have Z D r ! dD = . The solution is ! = A log r + B (3. 2 Z D dD = . We choose a simple domain to allow us to evaluate the above integrals.7) Note that this function is singular at r = 0 as required.1 (3.8) where D is any domain containing r = 0. From Equation (3. 1 @ r @! = 0 r @r @r This can be solved by straight (one-dimensional) integration.

2: Domain used to evaluate fundamental solution coefficients. 21  log r + B ! = . 21  log r .2) then from the Green-Gauss theorem @D is the surface of the disk D since D is a disk centred at r = =A " 2" = 2A @D @r dS = 0 so n and r are in the same direction from Equation (3.3. from Equation (3.8) A = .7). so that the fundamental solution for the twodimensional Laplace Equation is ! = .  D x F IGURE 3. radius " Z D Z @! r ! dD = @n dS @D Z @! 2 0 centred at r = 0 (Figure 3.2 T HE D IRAC -D ELTA F UNCTION AND F UNDAMENTAL S OLUTIONS 47 y " . and the fact that D is a disc of radius " Therefore. 21 : So we have B remains arbitrary but usually put equal to zero.

1 = 21 log  r (3.9) .

. using the property of the Dirac delta q ! = .  2 (3. . ru:r! d 2 (3.4 we have seen that 0= Z Z Z @u r u:! d = @n ! d. The fundamental solution for the three-dimensional Laplace Equation can be found by a similar technique.11) For the Galerkin FEM we chose ! . x2 +  . For the boundary element method we choose ! to be the fundamental solution of Laplace’s Equation derived in the previous section i. From Section 2. y d = .12) i. .1).. We firstly must form an integral equation from the Laplace Equation by using a weighted integral equation and then use the Green-Gauss theorem. ). 21  log r Z ur ! d = . y 2 (singular at the point . the domain integral has been replaced by a point value.. This gives Z @u Z 0 = @n ! d. where r =  .3 The Two-Dimensional Boundary Element Method We are now at a point where we can develop the boundary element method for the solution of r2 u = 0 in a two-dimensional domain . The basic steps are in fact quite similar to those used for the finite element method (refer Section 2. 2 Z u  . .11). 1 ! = 4r 3. y 2 (singular at the point .e. the weighting function. one of the basis functions used to approximate u.u . Then from Equation (3. to be 'm . + ur ! d 2 (3.10) @ This was the starting point for the finite element method. ru:r! d @ Z @! Z Z @u = @ @n ! d.48 T HE B OUNDARY E LEMENT M ETHOD where r =  .  . To derive the starting equation for the boundary element method we use the Green-Gauss theorem again on the second integral.e. . The result is q where r is now a distance measured in three-dimensions. x. x2 +  . @ u @n d.  2 ).

In order to be able to evaluate ur2! d in this case we enlarge to include a disk of radius " about P (Figure 3.  2 .  is on the boundary of the domain . .3). We call this enlarged region 0 and let . .  replaced by 1 u .0 = . The previous equation (Equation (3.  in Equation (3.13) the solution domain to integral expressions involving u and @n over the boundary of the solution domain. When .14) We must now investigate this equation as lim"0 .." . Equation (3.13)) holds if . y d = 0 since the integrand of the second integral is zero at every point except ..  and this point is outside the region of integration. Rather than having an expression relating the value of u at some point inside the domain to boundary integrals.  was inside the domain.3 T HE T WO -D IMENSIONAL B OUNDARY E LEMENT M ETHOD 49 Thus Equation (3. u P  + Z . a more useful expression would be one relating the value of u at some point on the boundary to boundary integrals." . since P is inside the enlarged region 0 . There are 4 integrals to consider. 2 Z u  . Now. (3. = . If . 2 2 Z Let P denote the point . It relates the value of u at some point inside @u Z ur ! d = . Rigorous details of where this coefficient 1 comes from are given below. We can see this in a non-rigorous way as 2 follows. and we look at each of these in turn. We derive such an expression below. This case also happens to be the most important for numerical work as we shall see. . ." .  is on the boundary we only have half of the singularity contained inside the domain.e. When .." Z @u @n ! d. so we integrate around one-half of the singularity to get 1 u . .3." u @! @n d. The integral expression we will ultimately obtain is simply Equation (3.13).11) becomes Z @u Z @! u . x...  + u @n d.  2 (i. The case which needs special consideration is when the singular point .e." .  2 @ @ (3. the singularity of Dirac Delta function is inside the domain). = @n ! d.  is outside then This equation contains only boundary integrals (and no domain integrals as in Finite Elements) and is referred to as a boundary integral equation.13) holds for this enlarged domain i. we integrated around the entire singularity of the Dirac Delta to get u .13) with u .

" . u @r 2 . 1 log r d.. Firstly consider Z @! Z u @n d." ." F IGURE 3.50 T HE B OUNDARY E LEMENT M ETHOD 0 " P .3: Illustration of enlarged domain when singular point is on the boundary." . 1 log r d. = . u @n 2 @ ." @ ." Z = .

by definition of ! since .

Thus Z 1 1 = . " @n @r since r = " on . 2 " u d. @ @ on ." 1 1 ! . by the mean value theorem for a surface with a unique tangent at P . 2 " u P  " Z u 1 = . 2 r d. ." .

15) By a similar process we obtain . 1 u P  u P  Z @! lim u @n d. 2 "0 "0 . 2 " " = ." (3. = lim .

" (3. 2 @n P  " log " = 0 "0 "0 .16) since lim" log "0 as lim"0 . . = lim . 1 @u Z @u lim ! @n d.

@ lim "!0 @n @n ...11) Thus we have 1 Z 0Z @u ! d.4: Illustration of internal angle .17) (3." . .18) Combining Equations (3. For “nice” integrals (which includes the integrals we are dealing with here) we have 0Z @ lim "0 1 Z (nice integrand) d.  2 @ (i. then the coefficient 1 is replaced by where 2 2 is the internal angle at P (Figure 3. = Z @u ! d.4). a point with a unique tangent) on the boundary of .e. . 2 @n @n . singular point is on the boundary of the region)." since .A = @! u d. (3..g.." 0Z 1 Z @! u d. Note: If the integrand is too badly behaved we cannot always replace . where P = .14)–(3.18) we get Z @u Z @! 1 u P  + u @n d.3 T HE T WO -D IMENSIONAL B OUNDARY E LEMENT M ETHOD 51 It only remains to consider the integrand over . P F IGURE 3." .A = . .. . ." ! . If P happens to lie at some nonsmooth point e. as lim"0 .e. = 2 u P  + @n ! d. (refer Section 4. . lim @ "!0 @n @n .. (nice integrand) d. in the limit and one must deal with Cauchy Principal Values.. or 1 u P  + Z u @! d.A = @u ! d. Note: The above is true if the point P is at a smooth point (i.." by .3. a corner.

.19) are then known. Thus we solve for the boundary data first. x2 +  .19) to find the surface distributions of u and @n and then (if required) use Equation (3.19) involves only the surface distributions of point P . Once the surface distributions of u and @n are known. smooth at P if P 2 . the overall size of the problem has been reduced by one dimension (from volumes to surfaces). and find the volume data as a separate step.52 T HE B OUNDARY E LEMENT M ETHOD Thus we get the boundary integral equation. The procedure is thus to use Equation (3. the value of u at any point P inside can be found since all surface integrals in Equation (3. and . with 1 ! = 4r c P  = q r =  .6. @u @u u and @n and the value of u at a @u . Thus the boundary element method offers some distinct advantages over the finite element method in certain situations. the boundary integral equation expression above is the same. x2 +  . and . . Also the effort required to produce a volume mesh of a complex three-dimensional object is far greater than that required to produce a mesh of the surface. This can result in huge savings for problems with large volume to surface ratios (i. Since Equation (3. Z @u Z @! c P  u P  + u @n d. y2 +  . = @n ! d. It also has some disadvantages when compared to the finite element method and these will be discussed in Section 3. not smooth at P if P For three-dimensional problems. and .19). (3. not smooth at P if P Equation (3. z2 8 1 if P 2 : inner solid angle 1 2 4 2 .19) where ! = . 21  log r c P  = q r =  .19) to find the solution at any point P 2 .e. problems with large domains). as opposed to volume integrals in a finite element formulation. and .19) only involves surface integrals. We now turn our attention to solving the boundary integral equation given in Equation (3. y2 8 1 if P 2 : internal angle 1 2 2 2 . smooth at P if P 2 . .

j (3.20) (a) (b) F IGURE 3.5: Schematic illustration of a boundary element mesh (a) and a finite element mesh (b).4 N UMERICAL S OLUTION P ROCEDURES FOR THE B OUNDARY I NTEGRAL E QUATION 53 3. but are generally taken to be the same (this is termed an isoparametric formulation).j j =1 . qj are values of u and q on element .j we introduce standard (finite element) basis functions uj = X ' uj and j qj @u @n = X ' qj (3. . we only have to interpolate the functions over a one-dimensional element). In general the basis functions used for u and q do not have to be the same (typically they are) and these basis functions can even be different to the basis functions used for the geometry.22) where uj . into some set of elements (hence the name boundary .j (3.4 Numerical Solution Procedures for the Boundary Integral Equation The first step is to discretise the surface elements).19) becomes N Z N Z X X @! @u ! d. qj are values of u and q at node on element . These basis functions for u and q can be any of the standard one-dimensional finite element basis functions (although we are dealing with a two-dimensional problem. = @n j =1 .3. . c P  u P  + u @n d.= N j =1 .j and uj .21) Over each element .j . Then Equation (3.

we must specify either a value of u or q (or some combination of these) to have a well-defined problem. then we have ciui + Z N X X j =1 uj .26) Equation (3. We now generate one equation per node by putting the point P to be at each node in turn. If P is at node i. We need to rearrange the above system of equations to get Au = Bq (3. while the matrix is dependent on the number of finite element nodes (which include nodes in the domain).j where !i is the fundamental solution with the singularity at node i (recall ! is . As x Cx = f A B A B C K (3. say.j ' ! d. The matrices and (and hence ) are fully populated and not symmetric (compare to the finite element formulation where the global stiffness matrix is sparse and symmetric). We can write Equation (3.54 T HE B OUNDARY E LEMENT M ETHOD This gives c P  u P  + Z N X X j =1 uj .j ' !i d. then we can generate L equations. This can be solved using standard linear equation solvers. = j =1 Z ' !i d.25) is for node i and if we have L nodes.27) (compare to the global finite element equations Ku = f ) where the vectors u and q are the vectors of nodal values of u and q . We therefore have L equations (the number of nodes) and have L unknowns to find.25) aij = Z .j N X X ' @! d . We can assemble these equations into the matrix system At each node. where 2 r is the distance from the singularity point). th where is the vector of unknowns. although specialist solvers are required if the problem is large (refer [todo : Section ???]).j i ' @! @n d.23) This equation holds for any point P on the surface .j N X X @! i qj ' @n d. and bij = Z .28) K . (3. Note that the ij component of the A matrix in general is not aij and similarly for B .24) .24) in a more abbreviated form as ciui + where N X X j =1 uj aij = N X X j =1 qj bij (3. 1 (3. (3. = qj @n j =1 Z . The size of the and matrices are dependent on the number of surface nodes. log r ..

then we can use Equation (3.e. These integrals typically must be evaluated numerically. The matrix is therefore also well conditioned and Equation (3.5 for more on this). . all boundary values of and are known.j Z @! i   i '   @n jJ  j d = '   @!   dri jJ  j d @ri dn 0 (3. The use of the fundamental solution as a weight function ensures that the and matrices are generally well conditioned (see Section 3.j i ' @! @n d.25) with the singular point P located at the required solution point i.30) 1 . If a solution is then required at a point inside the domain. Recall that aij = where Z . 21  log ri ri = distance measured from node i In terms of a local  coordinate we have bij = aij = Z1 Z 0 '   !i   jJ  j d (3.j !i = . it depends on the surface to volume ratio as to which method will generate the smallest and quickest solution.5 N UMERICAL E VALUATION OF C OEFFICIENT I NTEGRALS 55 mentioned earlier. Once this has been found.28) can be solved reasonably easily. 3. and require far more work and effort than the analogous finite element integrals. In fact the matrix is diagonally dominant (at least for Laplace’s equation). The vector contains the unknown values of and on the boundary.29) contains no unknowns and only involves evaluating the surface integrals using the fundamental solution with the singular point located at P . N X X j =1 uj aPj (3. x u q C u q A A B u P  = N X X j =1 qj bPj .31) .29) The right hand side of Equation (3. and bij = Z ' !i d..3.5 Numerical Evaluation of Coefficient Integrals We consider in detail here how one evaluates the aij and bij integrals for two-dimensional problems.

56 The Jacobian J   can be found by T HE B OUNDARY E LEMENT M ETHOD s.

2 .

2 J   = d. but the integrand becomes singular. This is one of the disadvantages of the BEM .j (or close to it) we see that ri approaches 0 and the fundamental solution !i tends to 1. yi2 where xi .the fact that singular integrands must be accurately integrated. y 0   ) by in the appropriate direction and then normalise. The integrals for which node i lies in element . and the integrals can therefore be evaluated numerically using some suitable quadrature schemes. The integral still exists. In such cases special care must be taken . large numbers of Gauss points or other special treatment.j then standard Gaussian quadrature can be used to evaluate these integrals. dri we note that To find !i = . It is therefore important to ensure that these integrals are calculated as accurately as possible since these terms will have most influence on the solution. 2 Thus every expression in the integrands of the aij and bij integrals can be found at any value of  . A relatively straightforward way to evaluate all the integrals is simply to use Gaussian quadrature with varying number of quadrature points. where The fundamental solution is can be found by straight differentiation of the ri   = x   . we simply rotate the tangent  vector (given by x0   . depending on how close or far the singular point is from the current element. For the case when node i is contained in the current element one can use special quadrature schemes which are designed to integrate log-type functions. xi 2 + y   . or also to subdivide the current element into two or more smaller elements and evaluate the integral over each n dri = rr  n i ^ dn (3. These are to be preferred when one is dealing with Laplace’s equation. However.j are in general the largest in magnitude and lead to the diagonally dominant matrix equation.32) dy s represents the arclength and dx and d d interpolation expression for x   and y  .either by using special quadrature schemes. Gaussian quadrature is still the norm. It is possible to incorporate adaptive integration schemes that keep adding more quadrature points until some error estimate is small enough.33) . if node i is in . 21  log ri   q dn where ^ is a unit outward normal vector. However. This is not very elegant or efficient. If node i is well removed from element . yi  are the coordinates of node i. = ds = dx + dy d d d d (3. these special log-type schemes cannot be so readily used on other types of fundamental solution so for a general purpose implementation. To find a unit normal vector. but has the benefit that it is relatively easy to implement.

8.11). As in Section 3. . As noted above. It is also possible to evaluate the “worst” integrals by using simple solutions to the governing equation. and this technique is the norm for elasticity problems (Section 4.6 T HE T HREE -D IMENSIONAL B OUNDARY E LEMENT M ETHOD 57 . 3.6: Illustration of the decrease in ri as node i approaches element .6 The Three-Dimensional Boundary Element Method The three-dimensional boundary element method is very similar to the two-dimensional boundary element method discussed above.j .j ri ri node i (a) (b) F IGURE 3. It should be noted that research still continues in an attempt to find more efficient ways of evaluating the boundary element integrals. The numerical solution procedure also parallels that given in Section 3. and the expressions given for aij and bij apply equally well to the three-dimensional case. with ! and c P  being defined as in Section 3.5 we illustrate how to evaluate each of the terms in the integrand of aij and bij . Details on each of these methods is given in Section 3. the three-dimensional boundary integral equation is the same as the two-dimensional equation (3. The only real difference between the two procedures is how to numerically evaluate the terms in each integrand of these coefficient integrals.4.3.3.j node i . subelement.19).

As before we use i dn dn The unit outward normal ^ is found by normalising the cross product of the two tangent vectors . 2 jJ 1 . 2  . 2 !i 1. yi 2 + z 1 . 2 = 4r 1 q i 1. 2 @!   1 . yi . 2  . zi  are the coordinates of node i.  j d d i ' 1. 2  1 2 1 2 @r dn i (3.35) Z Z1 0 The fundamental solution is !i 1. Z Z1 0 dri jJ  . 2 2 where ri 1 .58 T HE B OUNDARY E LEMENT M ETHOD The relevant expressions are aij = = Z . xi  + y 1 . where xi .j 1 0 i ' @! @n d. 2  . 2j d1 d2 (3. zi 2 dri = rr  ^ to find dri .j 1 0 Z . ' 1. 2  = x 1 .34) bij = = ' !i d.

@1 . @1 and t2 . @x @y @z n t1 = @1 .

whereas here we have a (possibly curved) two-dimensional surface in three-dimensional space. @x @y @z = @ . Comment: Because of the reduction in size of the mesh. 2  is given by kt1  t2 k (where t1 and t2 are the two tangent vectors). one often hears of people saying that the problem size has been reduced by one dimension. The integrals are evaluated numerically using some suitable quadrature schemes (see Section 3.7 A Comparison of the FE and BE Methods We comment here on some of the major differences between the two methods. This is one of the major pluses of . The Jacobian J 1 . @ . 1.8) (typically a Gauss-type scheme in both the 1 and 2 directions). Depending on the application some of these differences can either be considered as advantageous or disadvantageous to a particular scheme. @ (it relies on the user of any BEM code to 2 2 2 ensure that the elements have been defined with a consistent set of element coordinates 1 and 2 ). BEM: A mesh of the boundary only is required.in that case we are dealing with a two-dimensional surface in two-dimensional space. Note that this is different for the determinant in a two-dimensional finite element code . n 3. FEM: An entire domain mesh is required.

. BEM: Cannot even handle all linear problems. u q 4. There are many linear problems (e. Also the integrals that are the most difficult (those containing singular integrands) have a significant effect on the accuracy of the solution. Comment: The need to evaluate integrals involving singular integrands makes the BEM at least an order of magnitude more difficult to implement than a corresponding finite element procedure. 8. BEM: Much more difficult to implement.only in its approximations of the boundary conditions.. virtually any nonhomogeneous equation) for which fundamental solutions are not known. FEM: Entire domain solution is calculated as part of the solution. BEM: Both and of the same accuracy. is a very time consuming exercise. BEM: Solution on the boundary is calculated first. or are localised to a particular part of the domain. Comment: The matrices are generally of different sizes due to the differences in size of the domain mesh compared to the surface mesh. Handles nonlinear problems well. There are certain areas in which the BEM is clearly superior. FEM: Element integrals easy to evaluate. and then the solution at domain points (if required) are found as a separate step. 3. FEM: Relatively easy to implement. and some contain integrands that become singular. Comment: BEM integrals are far harder to evaluate. There are problems where either method can give rise to the smaller system and quickest solution . 5. FEM: Widely applicable. BEM: Only boundary conditions are being approximated. and hence an entire domain solution is not required. Comment: A fundamental solution must be found (or at least an approximate one) before the BEM can be applied. 6. FEM: Reactions on the boundary typically less accurate than the dependent variables. 7. so these integrals need to be evaluated accurately. Comment: The use of the Green-Gauss theorem and a fundamental solution in the formulation means that the BEM involves no approximations of the differential Equation in the domain . but it can be rather restrictive in its applicability. Comment: There are many problems where the details of interest occur on the boundary. For problems involving infinite or semi-infinite domains. 2. BEM: Fully populated nonsymmetric matrices generated.g.it depends partly on the volume to surface ratio. BEM: Integrals are more difficult to evaluate.7 A C OMPARISON OF THE FE AND BE M ETHODS 59 the BEM .construction of meshes for complicated objects. FEM: Sparse symmetric matrix generated. FEM: Differential Equation is being approximated.3. particularly in 3D. BEM is to be favoured.

then it is possible to obtain analytic expressions to most (if not all) of the integrals that will appear in the BEM (at least for two-dimensional problems). The expressions can become quite lengthy to write down and evaluate.e. These are obtained by approximating the integral as Z1 0 log   f   d  N X i=1 f i wi i. log quadrature schemes can be developed which will exactly integrate polynomial functions f  . a Gaussian scheme can be used of each variable of integration if the region of integration is rectangular. The basic tool for evaluation of these integrals is quadrature.. the log function has been factored out. integrals of the form Z1 0 log   f   d will be required. For a two-dimensional BEM solution of Laplace’s equation.8 a one-dimensional integral is approximated by a sum in which the integrand is evaluated at certain discrete points or abscissa Z1 0 f   d  N X i=1 f i wi where wi are the weights and i are the abscissa. 1 or less. It is relatively common to use logarithmic schemes for this.8. 3. As discussed in Section 2. Tables of these are given below It is possible to develop similar quadrature schemes for use in the BEM solution of other PDEs. In the same way as Gaussian quadrature schemes were developed in Section 2.60 T HE B OUNDARY E LEMENT M ETHOD 3. but benefit from the fact that they will be exact.1 Logarithmic quadrature and other special schemes Low order Gaussian schemes are generally sufficient for all FEM integrals. but that is not the case for BEM. This is generally not the optimal choice for the weights and abscissae but it allows easy extension to higher order integration.each new equation to be used requires its own special quadrature scheme. when one begins to use general curved elements and/or solve threedimensional problems then the integrals will not be available as analytic expressions.8. The weights and abscissa for the Gaussian quadrature scheme of order N are chosen so that the above expression will exactly integrate any polynomial of degree 2N .8 More on Numerical Integration The BEM involves integrals whose integrands in generally become singular when the source point is contained in the element of integration. If one uses constant or linear interpolation for the geometry and dependent variable. which use different fundamental solutions to the log function. The problem with this approach is the lack of generality . . However. For the numerical evaluation of two or three-dimensional integrals.

For example u = x is a solution to Laplaces’ equation (assuming the boundary conditions are set correctly).112009 0.what works well for one problem may not be so good for another problem. Preconditioners are generally problem dependent . y 2 ) can be used to find the other matrix entries (or just used to check the accuracy of the matrices).513405 4 0.. The conjugate gradient technique is generally regarded as a solution technique for (sparse) symmetric matrix equations. The current technique of favour in the BE community for solution of large BEM matrix equations is a preconditioned Conjugate Gradient solver.718539 0. although it is very dependent on the implementation of the BE method. the time taken for the matrix solution begins to dominate the matrix assembly stage.602277 0. 3. Thus if one sets both u and q in Equation (3. direct solution methods.3.245275 0.391980 0.766880 0.556165 0.2 Special solutions Another approach.383464 0.8.9 The Boundary Element Method Applied to other Elliptic PDEs Helmholtz.368997 0. based on LU factorisations.27) at every node according to the solution u = x. particularly useful if Cauchy principal values are to be found (see Section 4.041448 0.g.386875 0. As mentioned above the matrix is generally well conditioned. As the problem size increases.9 T HE B OUNDARY E LEMENT M ETHOD A PPLIED Abscissas = ri i . can be used. fully populated and nonsymmetric.11) is to use special solutions of the governing equation to find one or more of the more difficult integrals.190435 0.281461 TO OTHER E LLIPTIC PDE S 61 n 2 n 3 Weight Factors = wi i . For small problems. modified Helmholtz (CMISS example) Poisson Equation (domain integral and MRM.10 Solution of Matrix Equations The standard BEM approach results in a system of equations of the form = (refer (3.094615 0.848982 0.wi 0. Monte-carlo integration. 3.039225 TABLE 3. one can then use this to solve for some entry in either the or matrix (typically the diagonal entry since this is the most important and difficult to find). C Cx f .28)). DRM.1: Abscissas and weight factors for Gaussian integration for integrands with a logarithmic singularity. Further solutions to Laplaces equation (e. A B 3. u = x2 .wi n i .063891 0.wi 0. This usually occurs when there is between 500 and 1000 degrees of freedom.

There has been a lot of research on coupling FE and BE procedures .11 Coupling the FE and BE techniques There are undoubtably situations which favour FEM over BEM and vice versa.I = FEM region = BEM region = FEM boundary = BEM boundary = interface boundary The BEM matrices for B can be written as Au = Bq where is a vector of the nodal values of u and The FEM matrices for F can be written as (3. There are at least two possible methods. where f B .7.7: Coupled finite element/boundary element solution domain.we will only talk about the basic ideas and use Laplace’s Equation to illustrate this.the use of one or the other depends on the problem.36) u @u q is a vector of the nodal values of @n (3.B . 1. Often one problem can give rise to a model favouring one method in one region and the other method in another region eg. in the sense of which part is more dominant FEM or BEM) Consider the region shown in Figure 3.B . Treat the BEM region as a finite element and combine with FEM 2. in a detailed analysis of stresses around a foundation one needs FEM close to the foundation to handle nonlinearities. Treat the FEM region as an equivalent boundary element and combine with BEM Note that these are essentially equivalent . but to handle the semi-infinite domain (well removed from the foundation).f F IGURE 3.37) Ku = f .62 T HE B OUNDARY E LEMENT M ETHOD B f .f . BEM is better.I . 3.

e. replace K B by B 2 B results). as before.1A u = Mq = f KBu = f B where K B = MB . This can be assembled into the existing BEM system (using compatability conditions) and use existing BEM matrix solvers.38) to an equivalent load vector by weighting the nodal values of by the appropriate basis functions. = . ) B F @ nB @ nF Ku Mq To apply method 2 (i.e. Applying this to (3.. Notes: 1.e. Attempts have been made to “symmetricise” the K B 1 . but ..often yields inaccurate matrix .e.B. 2. This means that different matrix equation solvers must be used for solving the new combined FEM-type system (most solvers in FEM codes assume sparse and symmetric). B = Therefore Equation (3. (e. then we can convert in Equation (3. @ uF (q is continuous.38) q M f Mq M . On . One must make use of the following uIB = uIF (u is continuous) I @ uI B = ... Therefore we can assemble this together with original FEM matrix to produce an FEM-type system for the entire region B . to treat the FEM region as an equivalent BEM region) we firstly note that. K B is in general not symmetric and not sparse.1 A B i. an equivalent stiffness matrix obtain from BEM..1Au = q q (3.11 C OUPLING THE FE AND BE TECHNIQUES 63 where is the stiffness matrix and is the load vector.g.3.. producing a matrix i. To apply method 1 (i.I nodal values of u and q are unknown. treating BEM as an equivalent FEM region) we get (from Equation (3.37) contained.of doubtful quality.37) yields = an equivalent BEM system.36)) K f If we recall what the elements of in Equation (3. = .38) becomes f B. This approach does not require any matrix inversion and is hence easier (cheaper) to implement 2. Notes: f Mq 1.K . Existing BEM solvers will not assume symmetric or sparse matrices therefore no new matrix solvers to be implemented . K T  .

This leads to singular integrands. e.64 T HE B OUNDARY E LEMENT M ETHOD 3. p @ . 21  log r @ @ Z @u Z @!p u @n d. = @n ! d.12. 3. @ with The usual procedure is to put point P at each solution variable node .this yields ! = . = @n d. Method is not so popular . Consider Laplaces Equation in a (bounded) domain Z @u Z @! weighted residuals  ! @n d.12 Other BEM techniques What we have mentioned to date is the so-called singular (direct) BEM.calculated the value (numerical) of the contraction coefficient of a round jet issuing from an infinite tank . Must find a (complete) set of functions (If you just use usual approximations for system is not diagonally dominant so not so good) u matrix 3.. Doesn’t introduce singular functions so integrals are easy to evaluate 2.1 Trefftz method Trefftz was the first person to perform a BEM calculation (in 1917 . Notes: 1. Given a BIE there are other ways of solving the Equation although these are not so widely used. @ @ if r2 ! =0 The procedure is to express u as a series of (complete) functions satisfying Laplace’s equation with coefficients which need to be numerically determined through utilisation of the boundary conditions. Another possibility is to put point P outside of the domain .a nonlinear free surface problem).12.Green’s functions more widely available that complete systems 3.g. = u @n d. This method basically uses a “complete” set of solutions instead of a Fundamental Solution.creating an equation for each node.2 Regular BEM Consider the BIE for Laplace’s equation Z @! Z @u c P  u P  + u @n d.

: : : . There is considerable choice for the location of the point P .e. This method does not involve singular integrands. By placing the point P (the singular point) at other distinct points outside as many equations as there are unknowns (or more if required).g.e.. Thus the solution H x.. = j=1 qj !i d.an equation involving u and q at each surface node. . . 3. For an acoustic problem (governed by Helmholtz Equation r2 u + k 2 u = 0) in an unbounded region the system of Equations produced by the usual (singular) BEM approach is singular for certain “fictitious” frequencies (i. z  = H . certain values of k ). so that integrals are inexpensive to calculate. This method is not very popular. 2. z has the property that H x.. i. N (3. Exterior Acoustic Problems. N is even) constant elements) Z @!i Z N N X 1u + X 2 i j=1 uj @n d. The idea of placing the singularity point P away from the solution variable node is often of use in other situations e.39) .13 S YMMETRY 65 Following discretisation as before gives Z N X X j =1 uj .e. 4. The Boundary Element Equation is (with N = 2M (i. = qj @n j =1 Z .j N X X p ' @! d . The distance along each node is often found by experimentation .j ' !p d. To overcome this further equations are generated (by placing the singular point P at various locations outside ). This behaviour can be found in many problems and we can make use of this as follows. z  8x. Often the set of Equations generated are ill-conditioned unless P chosen carefully.j i = 1. 3. In practise P is chosen along the unit outward normal of the surface at each solution variable node.3.x..various research papers suggesting “ideal” distances (Patterson & Shiekh). Both the boundary conditions and the governing Equation exhibit symmetry about the vertical axis.13 Symmetry Consider the problem given in Figure 3.8 (the domain is outside the circle).j . Notes: one can generate 1. The system of equations are then overdetermined and are solved in a least squares sense. putting x to .x makes no difference to the problem formulation.

+ @!i d. ui = un+1. M ). 2 i j=1 j : @n @n . M (3. M and solve as before.j M +1 +1 (3. + !i d. (3.j .41) for nodes i = 1.j .8: A problem exhibiting symmetry. = X q 1u + X u !i d. (The Equations for nodes i = M +1. : : : . .j . : : : . .41) as M M X 1u + X 2 i j=1 aij uj = j=1 bij qj i = 1. + . j=1 j : !i d.j then we can write Equation (3.N . N are the same as the Equations for nodes i = 1. If we define Z @!i Z @!i aij = @n d. M . : : : .N +1. We have N Equations and N unknowns (allowing for the boundary conditions).i So we can write i = 1.j .44) .40) 8 9 8 9 Z Z Z Z M = = @!i d. (This procedure has halved the number of unknowns.42) bij = !i d.j .sz r2H = s2H x F IGURE 3. + @n d.N .43) .66 T HE B OUNDARY E LEMENT M ETHOD z H = e.N .j Z Z +1 (3.9). From symmetry we know that (refer to Figure 3. The above M Equations have only M unknowns. : : : . : : : .) (3.

e. 3. .N will never contain a singularity arising from the fundamental solution. u r.N . An alternative approach to the method above arises from the implied no flux across the z axis.14 Axisymmetric Problems If a three-dimensional problem exhibits radial or axial symmetry (i. Note: Since i = 1. except possibly on the axis of symmetry if linear or higher order elements are used.9: Illustration of a symmetric mesh.14 A XISYMMETRIC P ROBLEMS 67 z .1 F IGURE 3.N +1. M this means that the integrals over the elements . Further examples of how symmetry can be used (e..i . : : : . radial symmetry) are given in the next section.g.i x .M +1 to .3.. However now the surface to be discretised extends to infinity in the positive and negative z directions and the resulting systems of equations produced is much larger. This approach ignores the negative x axis and considers the half plane problem shown.

. z  = u r.1 .

. The first step in such a procedure is to write the standard boundary integral equation in terms of cylindrical polars r. z ) it is possible to reduce the two-dimensional integrals appearing in the standard boundary Equation to one-dimensional line integrals and thus substantially reduce the amount of computer time that would otherwise be required to solve the fully three-dimensional problem.2 .

.. 1 Z 0Z2 @!p 1 Z 0Z2 c P  u P  + u @ @n d. z  i.e.

q A rq d. = q @ !p d.

.q A rq d. 0 . 0 (3. .45) where rp .

.p . zp  and rq .

and .q . and . is the intersection of . zq  are the polar coordinates of P and Q respectively.

= 0 semi-plane (Refer Figure 3.10). (n.) .b. Q is a point on the surface being integrated over.

11 2 = r 2 + r 2 .11: The distance from the source point (P ) to the point of interest (Q) in terms of cylindrical polar coordinates. r F IGURE 3. From Figure 3. z Q r2 r P rp r1 rq r F IGURE 3. for an axisymmetric problem. For three-dimensional problems governed by Laplace’s equation where rp is the distance from P to Q. 2r r cos .10: Illustration of surface .68 T HE B OUNDARY E LEMENT M ETHOD z . .

. .

 r1 p q p q p q q 2 + r2 r2 = r1 2 1 !p = 4r q 2 + r 2 . 2rp rq cos .

.p .

b cos .q  + zp . zq 2 r = rp q q 2 = a .

.p .

zq  (3.46) .q  + zp .

3.15 I NFINITE R EGIONS 69 We define Z 1 p m !p = 4 !p d.

q K  a+b 2 0 where m = 2b a+b (3. the axisymmetric fundamental solution cannot be written as simply a function of the distance between two points P and Q. but it also depends upon the distance of these points to the axis of revolution. i. rP  is the dirac delta centered on the ring r = rp .and three-dimensional cases..e. We also define Z @!p 1 p  qp = 4 @n d. rp = 0 instead of (3. Unlike the two. !p is called the axisymmetric fundamental solution and is the Green’s function for a ring source as opposed to a point source.47) and K m is the complete elliptic integral of the first kind.49) where p is the dirac delta centered at the point P and r . ! p is a solution of r2! + r .48) r2 ! + p = 0 (3.

50) we can write Equation (3. 3. b E m nz Q a. r2 + zp .45) as (3.15 Infinite Regions The boundary integral equations we have been using have been derived assuming the domain is bounded (although this was never stated). K m nr Q + a .52) and the solution procedure for this Equation follows the same lines as the solution procedure given previously for the two-dimensional version of boundary element method.50) becomes q = p  1 r2 .47) and Equation (3.b  a + b 2rq Z @!p Z c P  u P  + u @n d.q @! @n 2 0 (3. zq   1 z .50) For Laplace’s equation Equation (3. z p q p q p E m . . = q!p d. Using Equation (3. . However all concepts presented thus far are also . (3.51) where E m is the complete elliptic integral of the second kind.

. (3. . + q!P d. (see Figure 3. = q!P d. + u @n d. The boundary integral equations for the bounded domain R can be written as Z @!P Z Z Z @!P c P  u P  + u @n d. F IGURE 3. is the centre of a circle (or sphere in three dimensions) of radius centred at some point x0 on . and surrounding .53) will only be valid for the points on .e.12: Derivation of infinite domain boundary integral equations.. R R x0 . . valid for infinite regular (i. ! 1..53) If we let the radius R ! 1 Equation (3. Consider the problem of solving r2 u = 0 outside some surface . if Z .12).70 T HE B OUNDARY E LEMENT M ETHOD . . nice) regions provided the solution and its normal derivative behave appropriately as . .

the boundary integral Equation for Z @!P Z c P  u P  + u @n d. @!P lim u @n . . will be as expected i.e. q!P d. (3..55) . (3.54) If this is satisfied. . = 0 R!1 . = q!P d.

53) will be satisfied if u behaves at most as O R.e.3.1  (i. d.1 so that q = O R. These are the regularity conditions at infinity and these ensure that each term in the integral Equation (3.53) behaves at most as O R. = jJ j d.2.1 .. each term will ! 0 as R ! 1) For two-dimensional problems with !  = O log R we require u to behave as log R so that q = O R. In this case Equation (3.15 I NFINITE R EGIONS 71 For three-dimensional problems with !  where jJ j is the Jacobian and O  represents the asymptotic behaviour of the function as R ! 1. For almost all well posed infinite domain problems the solution behaves appropriately at infinity.

 ! = O R.R.  .d .1 @! = O .2 @n 1 = 4r where jJ j = O R2 .

p Laplace Equation Solution @ 2 u + @ 2 u + = 0 0 @x2 @x2 1 2.16.16 Appendix: Common Fundamental Solutions 3.1 Two-Dimensional equations Here r 2 = x2 1 + x2 .72 T HE B OUNDARY E LEMENT M ETHOD 3.

1 log u = 21  r @ 2 u + @ 2 u + 2u + = 0 0 @x2 @x2 1 2 1 H 2 r u = 4 i 0 where H is the Hankel funtion. 2 Helmholtz Equation Solution Wave Equation .

@2u @2u @2 u c @x2 + @x2 . r2 where c is the wave speed. @t2 + 0 t = 0 H ct . 1 2 Solution Diffusion Equation Solution @ 2 u + @ 2 u . r u = . 1 @u = 0 @x2 @x2 k @t 1 2 where k is the diffusivity. 2c c2t2 . .

8 1 . exp . r2 1  u =. 4kt 4kt 3 2 Navier’s Equation Solution  @ jk @xj + l = 0 for a point load in direction l.  2  r2 .  p i = pjiej 1 p ji = .

nir. @r @n 1 .ir. p .j  ej for a traction in direction k where  is Poisson’s ratio.j + 1 . 3.16.i . 2  ij + 3r.2 Three-Dimensional equations Here r 2 2 = x2 1 + x2 + x3 . 2  nj r.

17 CMISS E XAMPLES 73 Laplace Equation Solution @ 2 u + @ 2 u + @ 2 u + = 0 0 @x2 @x2 @x2 1 2 3 1 u = 4r @ 2 u + @ 2 u + @ 2 u + 2 u + = 0 0 @x2 @x2 @x2 1 2 3 1 exp .ir u = 4r Helmholtz Equation Solution Wave Equation .3.

 u k = ulk el . @2u @2u @2 u @2u c @x2 + @x2 + @x2 . @t2 + 1 2 3 where c is the wave speed. r 2 t =0 Solution Navier’s Equation Solution u =  @ jk @xj + l = 0 for a isotropic homogenenous Kelvin solution for a point load in direction l.

c 4r 3.16.51) 3. 2D steady-state heat conduction inside an annulus To determine the steady-state heat conduction inside an annulus run the CMISS example 324. To determine the steady-state heat conduction inside a sphere run the CMISS example 328. CMISS comparison of 2-D FEM and BEM calculations To determine the CMISS comparison of 2-D FEM and BEM calculations run examples 324 and 312. 2. CMISS biopotential problems C4 and C5. 3. 4 @r @r 1  ulk = 16G 1 .3 Axisymmetric problems Laplace For u see Equation (3. 4.47) and for q  see Equation (3.   r lk + @x1 @x2 for a displacement in direction k where  is Poisson’s ratio and G is the shear modulus. 3D steady-state heat conduction inside a sphere.17 CMISS Examples 1. 3 . . t.

.

and finally three-dimensional elasticity. this leads to the same algebraic equations as would be obtained by the Galerkin method (now equivalent to virtual work) but the physical assumptions made (in neglecting certain strain energy terms) are exposed more clearly in the Rayleigh-Ritz method. 4.a technique known as the Rayleigh-Ritz. Evaluate the potential energy from the sum of total strain energy for all elements together with the work done by applied boundary forces. 5. as we will show later. 8. e.g. In fact. Evaluate the strain energy for each element by integrating the products of stress and strain components over the element volume. 7.Chapter 4 Linear Elasticity 4. by fixing nodal displacements. Evaluate the stresses and strains using the nodal displacements and element basis functions.1 Introduction To analyse the stress in various elastic bodies we calculate the strain energy of the body in terms of nodal displacements and then minimize the strain energy with respect to these parameters . . 6. then two-dimensional plane stress and plane strain elements. We will first consider one-dimensional truss and beam elements. 2. Minimize the potential energy with respect to the unconstrained nodal displacements. In all cases the steps are: 1. Solve the resulting system of equations for the unconstrained nodal displacements. Evaluate the boundary reaction forces (or moments) at the nodes where displacement is constrained. Evaluate the components of stress from strain using the elastic material constants.. 3. Apply the boundary conditions. 9. Evaluate the components of strain in terms of nodal displacements.

2 Truss Elements Consider the one-dimensional truss of undeformed length L in Figure 3.76 L INEAR E LASTICITY 4. 0 and x. y  and making an angle .1 with end points 0.

Y  u l L . relative to the left hand end. y X + u. Y + v v X. with the x-axis. Under the action of forces in the x.and y directions the right hand end of the truss displaces by u in the x-direction and v in the y -direction.

x F IGURE 4. Displacements of the right hand end relative to the left hand end are u and v in the x.directions. respectively.and y . The new length is l with axial strain X = cos .1: A truss of initial length L is stretched to a new length l.

and Y = sin .

Neglecting second order terms in the binomial expansion L p1 +L 1 " = 1 + 2 " + O "2..1 = r L u2 + v2 u v = 1 + 2 cos .1= .1 X + Y pL2 + 2 Xu + Y v + u2 + v2 . the strain for small displacements u and v is using X + u2 + Y + v2 l p 2 2 e= L .

: + sin .

1 L L L q u v e = cos .: + 2 .

: L + sin .

2) .1) The strain energy associated with this uniaxial stretch is Z 1 SE = 2 Z 1 e dV = 2 A 0 Z 1 2 e dx = 2 EAe2 dx = 1 ALEe 2 0 (4.: L L L (4.

where u1.1) into Equation (4.4. v1 .2 T RUSS E LEMENTS 77 where = Ee is the stress in the truss (of cross-sectional area A). We now substitute for e from Equation (4. linearly related to the strain e via Young’s modulus E . u1 and v = v2 . v2 are the nodal displacements of the two ends of the truss u2 .2) and put u = u2 . v1 and u2. u1 + sin .

v1 1 SE = ALE cos .: v2 .

: 2 L L .

Node 3 is a pivot and therefore has zero displacement in both x. A force of 100 kN is applied in the x-direction at node 1. consider the system of three trusses shown in Figure 4. Node 2 is a sliding joint and has zero displacement in the y-direction only. The problem is to find all nodal displacements and the stress in the three trusses. The strain in truss 1 (joining nodes 1 and 3) is u1 cos 30 + v1 sin 30 = 3 u1 + 1 v1 L L 2 L 2L The strain in truss 2 (joining nodes 1 and 2) is u1 . The Rayleigh-Ritz approach is to minimize this potential energy with respect to the nodal displacements once all displacement boundary conditions have been applied.3) The potential energy is the combined strain energy from all trusses in the structure minus the work done on the structure by external forces.directions. u2 cos 90 + v1 sin 90 = v1 L L L The strain in truss 3 (joining nodes 2 and 3) is u2 cos 30 = 3 u2 L 2 L p p .2. For example. node 1 1 30 30 node 3 3 2 node 2 100 kN F IGURE 4. 2 (4.and y .2: A system of three trusses.

3 m2.2 = 116 kN (tensile).6) gives u2 = 0 Equation (4. 53 u1 Solving p these last two equations gives u1 = 3:34 mm and v1 = . 100u1 2 L 2 2 2 L Minimizing the potential energy with respect to the three unknowns u1 ..0:115 and in truss 3 is zero.57:5 kN (compressive) and in truss 3 is zero. 1 2 is . in truss The tension in truss 1 is A = AEe = 5  10.3 m2 107 kPa=m = 5  104 kN m.78 L INEAR E LASTICITY Since a force of 1 kN acts at node 1 in the x-direction. 100 = 0 2 (4. 2 1:15  10 = 0:232. E = 10 GPa and L = 1 m (e. 100 mm  50 mm timber AE = 5  10. truss) then L Equation (4. in truss 2 is .1:15 mm.6) = 5  10.3. p . Thus the strain in truss .3 1 is  23 3:34 .4) " p 3u + 1v 1 + v = 0 2 1 2 1 2 1  (4. 100:u1 = 1 AE 4 3 u1 + 1 v1 + 3 u2 + v1 5 . v1 and u2 gives @ PE = AE @u1 L @ PE = AE @v1 L p 3u + 1v 2 1 2 1 !p ! 3 .  v1 = .5) @ PE = AE @u2 L If we choose A p ! p 3 3 2 u2 2 =0 (4.4) gives 3u1 + 3v1 = 4  102= 5  104 Equation (4.5) gives for two dimensions p .1. The nodal reaction forces are shown in Figure 4.g.3 m2 107 kPa  0:232  10. the potential energy is PE = X1 trusses 2 2 p !2  p !2 23 ALEe2 .

Thus = Z x z dA .3: Reaction forces for the truss system of Figure 4.2. along the beam (assumed here to be in the x-direction).8) = Z z2 dA is the second moment of area of the beam cross-section.4. The axial strain is given by ex = R where z is the lateral distance from the neutral axis in the plane of the bending and R is the radius of curvature in that plane.7) M= where I Z Z E 2 dA = EI z x z dA = R R (4. Thus.7) becomes The slope of the beam is dw = .3 Beam Elements Simple beam theory ignores all but axial strain ex and stress x = Eex (E = Young’s modulus) z . E =M R I x and Equation (4.3 B EAM E LEMENTS 79 100 kN 100 kN 57:7 kN 57:7 kN F IGURE 4. where A is the beam x z = Eex = E R (4. The bending moment is given by M crossectional area. 4.

and the rate of change of slope is the curvature dx d.

10) . = d2w = 1 K = dx dx2 R = Mz I (4.9) (4.

11) and a force balance gives the shear force d EIw00 = .13) . V = . balancing the loading forces p with the axial stresses associated with beam flexure d2 EIw00  = . dx 2 dx2 (4. dM dx dx and the normal force (per unit length of beam) (4. p = dV dx dx2 d2 EI d2w = p .80 L INEAR E LASTICITY Thus the bending moment is 2w 00 M = EI d dx2 = EIw (4.12) This last equation is the equilibrium equation for the beam.

Thus. but this latter is ignored in the simple beam theory considered here. pw dx: 0 0 (4. The potential energy is therefore L L Z Z 1 2 00 PE = 2 EI w  dx . The external work associated with forces p acting normal to the beam and moving through a L transverse displacement w is Z 0 pw dx. A linear basis function has a zero second derivative and therefore cannot represent the flexural strain.14) or the equa- .14) The elastic strain energy stored in a bent beam is the sum of flexural strain energy and shear strain energy. (4. The boundary conditions associated with the 4th order equilibrium Equation (4. the (flexural) strain energy is Z 1 SE = 2 =2 1Z L Z x=0 A L x=0 Z z 2 ZL 1 00 2 E R dA dx = 2 EI w  dx A x=0 1 x ex dA dx = 2 ZL Z x=0 E A e2 x dA dx where x is taken along the beam and A is the cross-sectional area of the beam.15) The finite element approximation for the transverse displacement w must be able to represent the second derivative w 00 . The natural choice of basis function for beam deflection is in fact cubic Hermite because the inter-element slope continuity of this basis ensures transmission of bending moment as well as shear force across element boundaries.

Three possible combinations of boundary condition with their associated reactions are Boundary conditions Reactions (i) (ii) Simply supported zero displacement w = 0 zero moment M = EIw 00 Cantilever zero displacement w = 0 zero slope .4 P LANE S TRESS E LEMENTS 81 tions arising from minimum potential energy Equation (4.4.15) (which contain the square of 2nd derivative terms) are more complex than the simple temperature or flux boundary conditions for the (second order) heat equation.

= w 0 = 0 zero shear force V =0 (iii) Free end = . d EIw00  = 0 dx zero moment M = EI 00 = 0 shear force V slope .

= w 0  shear force V moment M slope .

 E e + e  y= x 1 . 0 0 3 5.4 Plane Stress Elements For two-dimensional problems. displacement w 4. we define the displacement vector 2e 3 4 ex y 5 and stress vector exy u = v . 21  E 4 1 2 gradients by 0 0 1. The stress-strain relation for two-dimensional plane stress: xy u = E 2 ex + ey  1. 2 y E e  xy = 1 +  xy x can be written in matrix form (4.16) where E = 1. strain vector e = 2 3 x = 4 y 5. = Ee The strain components are given in terms of displacement ex = @u @x ey = @v @y .

17) . @u @v 1 exy = 2 @y + @x (4.

v 2 dV Ee dV = 2 1 . uT f dA T A Z (4.  2 e2 + e x y x y xy V Z 1 e dV = 2 ex x + ey y + exy V xy  dV The potential energy is PE = SE .1 we approximate the displacement field finite element basis u = 'n un . Following the steps outlined in Section 4.82 L INEAR E LASTICITY The strain energy is Z 1 SE = 2 = 2 1Z V V T e T Z E h i 1 2 + 2ve e + 1 .18) where represents the external forces acting on the elastic body. v = 'n vn and calculate the strains f u with a ex = @u = @'n un @x @x ey = @v = @'n vn @y @y . external work Z 1 =2 V e Ee dV .

.

uT f dA Z uT f dA . 2 Z 1 T = u 2 T B EB dV:u .18) the potential energy is therefore Z Z 1 T PE = Bu E Bu dV .20) From Equation (4. uT f dA Z 1 T = 2 u Ku .19) 3 2 @'n 0 7 2e 3 6 @x 6 7  x @' n 7 un 6 4 5 0 = Bu e = ey = 6 7 @y v 7 6 n exy 4 1 @'n 1 @'n 5 2 @y 2 @x (4. @' 1 @u @v 1 @' n n exy = 2 @y + @x = 2 @y un + @x vn or (4.

.21) Z We next minimize the potential energy with respect to the nodal parameters un and vn giving 4.23) where the internal work done due to the stress field is equated to the work due to internal body forces and external surface forces. Equation (2.i uj + tij uj.tij.i = fj (tij are the components of stress) and by Cauchy’s formula. Let be the stress vector acting over the surface S enclosing a volume V of material of mass density and let be the equailibrium external force vector per unit area of surface (i. The equilibrium equation in V is . = tij ni j .5 Navier’s Equation The Galerkin finite element equations for linear elasto-statics can be derived from a physically appealing argument. Now.4. Let = e and interpolate the virtual displacements u i S .15): Z Z @g f r  rg + rf  rg d = f @n d. in moving through a virtual displacement = uj j to the work done by the stress vector = tj j in moving through a compatible set of virtual diaplcements .22) is now used to replace the right hand surface integral by a volume integral.5 NAVIER ’ S E QUATION 83 where K = BT EB dV is the element stiffness matrix. = ).i work equation = . Thus. where ni is a component of the unit normal to S .i dV Substituting the equilibrium relation tij. the principle of virtual work. giving Z S sj uj dS = Z V tij uj dV = Z V tij. t t s t s i u i s Z S sj uj dS = Z S tj uj dS = Z S u t i i tij ni uj dS The Green-Gauss theorem.fj into the right hand integral yields the virtual Z V tij uj. Ku = f (4. (4..i dV = Z V fj uj dV + Z S sj uj dS (4.e. the principle of virtual work equates the work done by the surface forces = sj j .

e = @ d ij 'n.e Since virtual dispalcements are arbitrary we get XZ e We now have e Z XZ @ k ij 'n.84 L INEAR E LASTICITY from their nodal values.k @xj d = e bi'n d + ti'n d. i. e is the global node number of local node n on element e  where un i  = Ui This gives 0 X @Z e e 1 0Z  @k d A U 4n.24) 4n.k i @xj e e+ Z @ 1  ti'n d. e @ Z e Z @ k ij 'n.k @xj d ij 'M.e.j = @x ui  j @k un = 'n..e and 4n. so n u i = 'n ui  @'n n  u i.k @x i j (4.m 1 .A Ui4n.

i 2 @xi N j 2 @n @xj i N @'N N ekk = uk.25) We wish to find EMNij . ij = ekk ij + 2eij N =1 2 ui. isotropic.27) Note that a coefficient of uN j also comes from eji (the first term) .' uN  = 1 @'N @'N uN 2 j.m @xj J  d @m d @xj (4.e. the coefficient of uN j (i.26) where  and  are Lam´ es constants..k = @' @ @x uk n k (4. homogeneous materials we have the generalised Hookes Law.i  we put uj = 'N uj this gives (4. Also eij 1 u = 1 @ . ZZZ 0 e @ m ij 'M. the displacement at node N in direction i) For linear.j + uj.

4.6 N OTE ON C ALCULATING N ODAL L OADS 85 So @'N @'N uN + 2 1 @'N @'N uN + 1 @'N @'N uN ij =  ij @n @xk k 2 @n @xj i 2 @n @xi j So the coefficient of uN j can be calculated by .

1 .

@' @ @' @ ZZZ @' N @n @'M @m N n M m + 2 J  d EMNij =  ij 0 @n @xk @m @xk @n @xj @m @xi Note: there is no sum for ij . The expression for EMNij can be simplified to give EMNij = where g mn 1 .

  d = 1 2 pL  d = 1 pL 2 . The nodal load vector in Equation (4.m 'N.ng @xi @xj (4.28) m @n = @ @x @x .29) where  is the normalized element coordinate along the side of length L loaded by the constant stress p kN m. If the element side has a linear basis. For example.4a.e.1 applied to the edge of the plane stress element in Figure 4.29) gives f1 = pL f2 = pL Z Z '1 d = pL '2 d = pL Z Z 1 . consider a uniform load p kN m.6 Note on Calculating Nodal Loads If a normal boundary stress is known it is necessary to compute the equivalent nodal load forces to represent the distributed load. the metric tensor resulting from the inner product of basis vectors. k k 4.21) has components f fn = Z p'n dx = pL Z 'n d (4.n @ ij 'M. ZZZ 0 m @n mn J  d +  2'M. i.m'N. Equation (4.1 ..

If the element side has a quadratic basis.29) gives f1 = pL f2 = pL f3 = pL Z '1 d = pL '2 d = pL '3 d = pL Z .86 L INEAR E LASTICITY as shown in Figure 4.4b. Equation (4.

1 Z Z .

2. Two adjacent quadratic elements both contribute to a common node in (d). p kN m. 2.   d = 2 3 pL 2  .4: A uniform boundary stress applied to the element side in (a) is equivalent to nodal 1 1 2 1 loads of 1 2 pL and 2 pL for the linear basis used in (b) and to 6 pL. j = 1. 1 d = 1 pL 2 6 as shown in Figure 4. and bi is the body force/unit volume (e.g.L 2 2p .   d = 1 6 pL 4 1 . 3 (4.L 3 2 .7 Three-Dimensional Elasticity Recall that if a body is in equilibrium then we have ij.L 3 2 1p . Z Z 2 2 .4c.j + bi =0 i.4d. as shown in Figure 4.i i. j = 1. 4. 3 pL and 6 pL for the quadratic basis used in (c).L 3 2 2p .  1p L p 2 (a) (b) F IGURE 4.30) where ij are the components of the stress tensor ( ij is the component of the traction or stress vector in the ith direction which is acting on the face of a rectangle whose normal is in the j th direction).. = ) Recall also that the components of the (small) strain tensor are b g eij = 1 2 ui. A node common to two elements will receive contributions from both elements.  1 . where the element length is now L 2.1 L 1 2 pL 1 6 pL 2 3 1 6 pL 1 2 pL (c) 1 6 pL (d) 6 .j + uj. 3 (4.31) .

uj. .31) and Equation (4.kj + bj = 0 These are the 3 equations for displacements. An equation of state (stress-strain relation or constitutive law) is required. Note: .32) eij = 21   2 3 + 2 kk ij where .. is the difference between the final and initial positions of a material point in question).kk +  +  uk. u u ij = cijklekl where cijkl are the 81 components of a 4th order tensor.  are related to Young’s modules E and Poisson’s ratio  by  + 2 E =  3 +  = 2  +  As long as the displacements are continuous functions of position then Equation (4. (4. If we start dealing with materials such as rubber or living tissue then we need to use the exact finite strain tensor).30).  are Lam´ es constants. Equation (4.32) are sufficient to determine the 15 unknown quantities.31) yields strains and Equation (4.e. Symmetry reduces the number of unknowns to 21. The object of solving an elasticity problem is to find the distributions of stress and displacement in an elastic body. This can often work with some smaller grouping of Equation (4. Currently we have 15 unknowns (6 stresses.31) into Equation (4.e. For a linear elastic material we usually propose that ij depend linearly on eij . In the general three-dimensional case.30).31) and Equation (4.g.30) yields Navier’s equation. If all boundary conditions are expressed in terms of displacements Equation (4. subject to a known set of body forces and prescribed stresses or displacements at the boundaries. Equation (4.32) yields stresses. the material response is independent of orientation of the material element) then we have the generalized Hook’s Law.4.7 T HREE -D IMENSIONAL E LASTICITY 87 where is the displacement vector (i. If the material is isotropic (i. i.32) then into Equation (4. this means finding 6 stress components ij = ji  and 3 displacements ui each as a function of position in the body..32) e.. (Note: We are assuming here that the displacement gradients are small compared to unity which is the usual situation in solid mechanics. ij or inversely = ekk ij + 2eij ij . Equation (4. 6 strains and 3 displacements) and only 9 equations.e..

j ui + ij ui. .j d  r u d .8 Integral Equation Using weighted residuals as before we can write Z  ij. This statement Equation (4.34) Thus combining Equation (4. The principle of virtual work equates the internal work due to the stress field (left hand side integral) to the work due to internal body forces and external surface forces. we have essentially used the Green-Gauss theorem once to move from Equation (4.33)  where  = u i  is a (vector) weighting field. 4.. The ui are usually interpreted as a consistent set of virtual displacements (hence we use the notation u instead of w ). This statement is independent of the constitutive law of the material.j d  (4.35) is the starting point for the general finite element formulation (Section 4.9 Linear Elasticity with Boundary Elements Equation (4.j d ij i u .7).j d u = = Z Z b u d i i i i + + @ @ ij nj ui  d.34) we have Z ij i. un Z Z Z ij ui. In the above derivation. Now u  ij u i .j i u d = = = Z Z Z  ij i .88 L INEAR E LASTICITY 4. ti = ij nj ). where ti are the (surface) tractions (i.35) b u d tiu i d. Z Z ij ui.e.j = Therefore.33) and Equation (4.33) to Equation (4. (4.j  Z ij.35) (as was done for the derivation of the FEM equation for Laplace’s .35) is more usually derived from considering virtual work (we use weighted residuals to tie in to Chapter 3).j d ij ui.j + bi  ui d =0 (4. by the divergence theorem ij.  @ ij i j d.

36) This is known as Betti’s second reciprical work theorem or the Maxwell-Betti reciprocity relationship between two different elastic problems (the starred and unstarred variables) established on the same domain. Thus from the virtual work statement.j d u = Z = = = Z Z ij eij d  ekk ekk e e ij ij d kk d + 2 + 2 Z Z eij e ij d Z eij e ij d d eij ij due to symmetry.36) above (after noting the reciprocity between ij and eij ). (4. Since the body forces.32)) we have Z ij i.i 2 2 1 1  = 2 ij u i. ( ij ij i i Note: What we have essentially done is use integration of parts to get Equation (4. the second domain integral on the left hand side of Equation (4. = i i Z b u @ i id + Z @ t i ui d. Using the constitutive law for linearly elastic materials (Equation (4.j ui d + Z b u d i i = Z t u @ i i d. Equation (4.37) does not introduce any unknowns into the problem (more about this later).37)  .. . bi .j 2 2 i. ij. .e. t represents the equilibrium state corresponding to the virtual displacements u ). then use it again to get Equation (4.i = 1 ij u + 1 ij u i. (4.j (i.   ij.j  = ij ui.9 L INEAR E LASTICITY WITH B OUNDARY E LEMENTS 89 equation).j + 2 ji uj.j + ij uj.35) and the above symmetry we have Z b u d i i + Z t u d.j where e ij are the virtual strains corresponding to the virtual displacements. we firstly note that ij eij  1   =1 ij ui.36) can be written as Z . Z @ ti u i d.j + bi = 0). are known functions.35). e .    Note that b i = . To continue. Therefore Equation (4. ij.4.

j ui d  =. Substituting these into Equation (4.b i + ei = 0). x tj x d. j Z t u @ j j d. 1 Somigliana was an Italian Mathematician who published this result around 1894-1902. where ei is the i component of a unit vector in the i direction and ei = ei  .90 L INEAR E LASTICITY The first domain integral contains unknown displacements in remove. x ei (4.39)  If each point load is taken to be independent then u j and tj can be written as  u j = uij P.41) th  where u ij P..j + ei  =0 (4. P . This is known as Somigliana’s 1 identity for displacement. x uj x d. We choose the virtual displacements such that and it is this integral we wish to ij.37) becomes ui P  ei = Z @ tj u d. Therefore x Z ij.40) (4. + Z bj u jd P2 (4. x ei  t j = tij P. . x represent the displacements and tractions in the j direction at x th corresponding to a unit point force acting in the i direction (ei ) applied at P . x + Z u ij P. the volume integral is replaced with a point value (as for Laplace’s equation). Z @ t ij P. x .38) th th (or equivalently . Therefore.uiP ei i. x bj x d x (4.e.39) (and equating components in each ei direction) yields ui P  = Z u @ ij P. Equation (4. x and tij P.42) where P 2 (see later for P 2 @ ). . P  eiui d = . Z x . We can interpret this as the body force components which correspond to a positive unit point load applied at a point P 2 in each of the three orthogonal directions.

j ni  where = 1. . xi P  x due to a unit point load applied at P in the i jk + th direction are .kk + 1 . x.j g (4. ri = xi x .k g Lord Kelvin (1824-1907) Scottish physicist who made great contributions to the science of thermodynamics .4.ki + bi = 0 G  G u i.ki + x .j r.1 u ij P.kk + 1 . 2  ij + r. x = 16 1 . 4  ij + r.j ik  .45) for three-dimensions and for two-dimensional plane strain problems.j g and (4. x = 8 1 . 2 uk.i 2 r. 1 .k ij + r. P  Navier’s equation for the displacements u i is where G = shear Modulus.ir. r.10 Fundamental Solutions  satisfied Recall from Equation (4. Here r In addition the strains at an point given by ri .   r 1 . 4  ij log r .1 e jki P. = 2.38) that ij ij.1 @r t (4. 3 for two-dimensional plane strain and three-dimensional problems respecand r.   G f3 . = @x@r = r i x r P. P  ei = 0 (4.44) The solutions to the above equation in either two or three dimensions are known as Kelvin 2 ’s fundamental solutions and are given by 1 u ij P. Thus u i satisfy G   G u i.i tively.46) . the distance between load point (P ) and field point (x).i nj . 2  r. r. x = 8  1 .ir.   Gr f3 .j + or equivalently  x .j  @n .43) b i = ei x .47) ij P. x = 4  1 . P  ei = 0 (4.10 F UNDAMENTAL S OLUTIONS 91 4. 2 uk.   Gr f1 . r. 2. 2  r.ir.ir.

there are obviously physical differences.1 x = 4  1 .92 L INEAR E LASTICITY and the stresses are given by ijk P. The plane strain expressions are valid for plane stress if  is replaced by  mathematical equivalence of plane stress and plane strain .  . Note that in three dimensions u and for two dimensions ij =O r . 2  r.all we have to do is modify the values of the elastic constants).k g =  1+ (This is a where and are defined above. r.j r. What the mathematical equivalence allows us to do is to use one program to solve both types of problems .k ij + r.i jk  + r.   r f1 .j ki .ir.

1 t ij = O r2 =O .

1 .

42)) is a continuous representation of displacements at any point P 2 . one can find the stress at any P 2 firstly by combining derivatives of (4..e. If P 2 @ we enlarge to 0 as shown. x tk x d. . Details can be found in Brebbia. t ijk P. x uk x d.42) to produce the strains and then substituting into Hooke’s law.1 u ij = O log r t ij r : Somigliana’s identity (Equation (4. x . Consequently.11 Boundary Integral Equation Just as we did for Laplace’s equation we need to consider the limiting case of Equation (4.  Expressions for the new tensors u ijk and tijk are on page 191 in (Brebbia et al. Telles & Wrobel (1984b) pp 190–191.) We use the same procedure as for Laplace’s equation but here things are not so easy. 4. we need to find the equivalent of c P  (in section 3) . x bk x d x Note: One can find internal stress via numerical differentiation as in FE/FD but these are not as accurate as the above expressions. 255–258.42) as P is moved to @ . x . (i. Z u ijk P. 1984b). Z ijk P.called here cij P . This yields ij P  = Z u + .

x uj x d. x (4." x uj x d.. x + . Then Equation (4.42) can be written as ui P  = Z u ij P.4. x ." t ij P. x uj x ..11 B OUNDARY I NTEGRAL E QUATION 93 0 " P . x (4." Z t ij P. x uj x d.. .. The only integral that presents a problem is the second integral.48) We need to look at each integral in turn as " 0 (i. x uj x d." +. This can be written as Z t ij P. x = t ij P. x d. uj P  d.5: Illustration of enlarged domain when singular point is on the boundary." | z 0 uj x Z by continuity of Z + uj P  t ij P." .x ." x uj x d. x = t ij P." Z 0 u ij P..." F IGURE 4. x ." Z + .49) Z The first integral on the right hand side can be written as t ij P. Z . . x tj x d." +.50) .e. x bj x d x (4." +." . " ! 0 from above).

" 1 . In practical applications cij and the principal value integral can be found indirectly from using Equation (4. if an integral exists in the proper sense. u and @n = .52) to represent rigid-body movements." by Z ." " But if we replace . x tj x d. = u ij tj d. and we write the second integral of Equation (4. Z (4." ! . t u ij j d. t ij P. in brief (no body force).... x d. x (4.) where the integral on the left hand side is interpreted in the Cauchy Principal sense. x uj x d.. x bj x d Z .2 x dx + Z2 " x 1 dx = ln jxjj. x 1 dx = Z2 . i." + ln jxjj2 ."2 x 1 1 dxA(by definition of improper integration) which does NOT exist. but it does in the Cauchy Principal Value sense.2.2 "  "lim !0 Z f x dx =0 Z This is the Cauchy Principle Value of .2. ln " = 0 8" 0 .52) (or. 2 Thus it is What is a Cauchy Principle Value? Consider f x = on ." t ij P. whereas with Laplace’s equation the unknowns were 3 @u (scalar quantities) here the unknowns are vector quantities.. x cij P  uj P  + Z . t ij P.2 x 1 0 Z2 1 dxA + @ lim " !0 2 ." = ." ". x + u ij P.e. u ij P. However. However.51) As " 0. then 1 . then it exists in the Cauchy Principal Value sense and the two values are the same.g. Thus as " 0 we get the boundary integral equation Z .52) is similar to the numerical implementation of an elliptic equation (e. the integral does not exist in the proper sense.49) as where we interpret this in the Cauchy Principal Value3 sense. 2 = . .." = ln " . The numerical implementation of Equation (4.2 x 0 Z" 1 dx = @ lim " !0 1 lim !0. Laplace’s Equation).. ln 2 + ln 2 . cij uj + Z . x = Z . .. .94 L INEAR E LASTICITY Let cij P  = ij + lim "0 Z ." x Then 1 Z f x dx = Z. x uj x d. f x dx .

. and t = 7 6 5 4 . There is a similar 4. A (the cij ’s) do not need to be determined explicitly. at each node (each node has 6 unknowns . tu d.e. use 2u 3 1 u = 4u25 .the fundamental solution . (4.e.53) We can discretise the boundary as before and put P . If we have a rigid-body displacement of a finite body in any one direction then we get Au = Bt (4.52) as cu + Z .. u3 2t 3 1 t = 4t25 t3 2u u u 3 11 12 13  5 u = 4u 21 u22 u23 .12 B ODY F ORCES ( AND D OMAIN I NTEGRALS IN G ENERAL ) 95 more convenient to work with matrices instead of indicial notation.3 displacements and 3 tractions .54) ( l = vector defining a rigid body displacement in direction l) i Ail = 0  aii = . 5 4 ..4. the singular point.we get 3 equations per node). X i6=j aij (no sum on i) i.. 7 un tn The diagonal elements of the A matrix in Equation (4.12 Body Forces (and Domain Integrals in General) The body force gives rise to a domain integral although it does not give rise to any further unknowns in the system of equations. = Z . 6u2 7 6 where u = 6 .54) (for three-dimensions. ut d.    u31 u32 u33 2t t t 3 11 12 13  5 t = 4t 21 t22 t23    t31 t32 t33 Then (in absence of a body force) we can write Equation (4.. i. a 3 x 3 matrix) contains principal value components. The overall matrix equation 2 3 2 3 u1 t1 7 6 6 t2 7 7 where n is the number nodes. the diagonal entries of result for an infinite body. (This is because the body force is known .

e. or a centrifugal load due to rotation about a fixed axis or the effect of a steady state thermal load can all be transformed to a boundary integral.52) is still classed as a Boundary Integral Equation. a domain mesh nullifies one of the advantages of BEM .  Firstly. a body force arising from a constant gravitational load.. However.(n. let G ij (the Galerkin tension) be related to uij by 1   u ij = Gij.96 L INEAR E LASTICITY was chosen so that it removed all unknowns appearing in domain integrals). In some standard situations a domain integral can be transformed to a boundary integral. In some cases domain integrals must be used but there are techniques developing to avoid many of them.   Gik.b.that of having to prepare only a boundary mesh. the Poisson equation r2 u = f yields a domain integral involving f .kj 8 1 r ij .g. Integrals over the domain containing known functions (eg body force integral) appear in many situations e. Since the integral also contains the fundamental solution and may not be a “nice” region it is unlikely that it can be evaluated analytically).. 2 1 .g.kk . The question is how do we evaluate domain integrals such as those appearing in the boundary integral forumalation of such equations? Since the functions are known a coarse domain mesh may work. Thus Equation (4.

1  Gij = 8G 1 2 : 8G r log r ij Then (3D) (2D) Bi = Z u b ij j d = Z .

G ij.j .   1 G ik.kk . 2 1 . bj = gj  Bi = gj = gj Z .kj bj d Under a constant gravitational load g = gj  G ik.

developed around 1988. Z  . . The two integrals of note are the DRM. 2 1 . dual reciprocity method.j  d which is a boundary integral. developed around 1982 and the MRM.  g G ik.   1 G ik. There has been a considerable amount of research on domain integrals in BEM which has produced techniques for overcoming some domain methods.kj 1 nk d. 2 1 . G ij. Unless the domain integrand is “nice” the above simple application of Green’s theorem won’t work in general. multiple reciprocity method.k .

2.4.2.13 CMISS Examples 1.13 CMISS E XAMPLES 97 4. To solve a truss system run CMISS example 411 This solves the simple three truss system shown in Figure 4. . To solve stresses in a bicycle frame modelled with truss elements run CMISS example 412.

.

The temperature at the grid point i. : : : (for the t-direction). t = u1 and the initial conditions u x. subject to the boundary conditions u 0.2 to solve the transient one dimensional heat equation.a slight generalization of the heat equation.1. nt = un i: (5. t = u ix. t 0 @t @x2 where D is the conductivity and u = u x. There is. t = u0 and u L.Chapter 5 Transient Heat Conduction 5. In time-dependent problems the solution proceeds from an initial solution at t = 0 and it is almost always convenient to calculate each new solution at a constant time (t 0) throughout the entire spatial domain . I (for the x-direction) and n = 0.1 ui . 0 = 0.1) 0 x L. ui.2) n n+1 Finite difference equations are derived by writing Taylor Series expansions for un i+1 .1 Introduction In the previous discussion of steady state boundary value problems the principal advantage of the finite element method over the finite difference method has been the greater ease with which complex boundary shapes can be modelled.N u x. A finite difference approximation of this equation is obtained by defining a grid with spacing x in the x-domain and t in the time domain. 1. therefore. as shown in Figure 5.3 to solve the transient advection-diffusion equation . Grid points are labelled by the indices i = 0. n is therefore labelled as . (5. 1. 5.1 Explicit Transient Finite Differences Consider the transient one-dimensional heat equation @u = D @ 2 u . t is the temperature.2. no need to use the greater flexibility (and cost) of finite elements to subdivide the time domain: finite difference approximations of the time derivatives are usually preferred. A combination of finite elements for the spatial domain and finite differences for the time domain is used in Section 5.2 Finite Differences 5. : : : . Finite difference techniques are introduced in Section 5.

n T RANSIENT H EAT C ONDUCTION n 2 1 x3 : @ 3 u n + O .100 about the grid point i.x4  + 1 x2 : @ u + @x2 i 6 @x3 i i 2 .

@x .

.

1 = ui . x: @x + 2 x : @x2 + 6 x : @x3 + O x4 i i .  @u 1 @ 1 @ n n 2 3 ui. n n 2u 3 u n .

 +1 n un + O t2 i = ui + t: @t i un i+1 = un i + x: . @u ni .

@u n .

.

3) (5.1 = 2ui + x : or .5) where O x4  and O t2  represent all the remaining terms in the Taylor Series expansions.4) gives n n 2 un i+1 + ui. Adding Equations (5.3) and (5. (5.4) (5.

 . @ 2 u n @x2 i + O x4 .

x2  .1 + O . = i+1 @x2 i x2 (5. @2 u n un .6) which is a “central difference” approximation of the second order spatial derivative. Rearranging Equation (5. 2un + un i i.5) gives a “difference” approximation of the first order time derivative .

2ui + ui. 2.1). un i + O t : = i @t i t (5. ui + O t = D ui+1 . @u n un+1 . 2ui + ui.x2  t x2 +1 th which is rearranged to give an expression for un i in terms of the values of u at the n time step .8) i = ui + D x2 ui+1 . the values of ui for the next time step are found from Equation (5. 2 2 t . n n n  +1 n un (5. 2. t = 0). yields the time dependent temperatures at the grid points (see Figure 5.8) iteratively for time steps n = 1.6) and Equation (5. 2. This is an explicit finite difference formula because the value of un i depends only on the values of +1 n+1 n ui i = 1. : : : .7) Substituting Equation (5. : : : .1) gives the finite difference approximation +1 n n n n un i .1 + O t . I .9) .8) with i = 1. The accuracy of the solution depends on the chosen values of x and t and in fact the stability of the scheme depends on these satisfying the Courant condition: t  1 : D x2 2 (5.. : : : etc.1 at the latest time step. I  at the previous time step and not on the neighbouring terms un i+1 and ui. x : n+1 Given the initial values of un i at n = 0 (i. Applying Equation (5.1 + O .e.7) into the transient heat equation Equation (5.

we can rearrange this to be of the form. An k e L we obtain. The equation is centred at grid point i. we obtain.... n shown by the .1 (5. +1 n n n un i = ui+1 + 1 . O 5.2 Von Neumann Stability Analysis The concept behind the Von Neumann analysis is that all Fourier components decay as time is advances or as they are processed by an iterative solver. Considering Equation (5. With central differences in x and a forward difference in t an explicit finite difference formula gives the solution at time step n + 1 explicitly in terms of the solution at the three points below it at step n.11) by.8). i.2.5. By subsituting the general Fourier component An e k x +1 i An k e kj 4x L i k j L  = An k e h  +1 4x  1 . 2 ei kj . I x F IGURE 5. as indicated by the dark shading. 2 + ei kj L  + e.11) i kj4x  If divide Equation (5.1 i i+1 .i L  n Ak .1: A finite difference grid for the solution of the transient 1D heat equation. +1 4x kj 4x An k = 1 . The lightly shaded region shows where the solution is known at time step n.10) 4x 4t i kj L  . where  = D 2 .14x L  + ei kjL4x  i (5.2 F INITE D IFFERENCES 101 t n+1 n : x x x x 1 00 1 2 . 2ui + ui..

k4x = 1 . 2 + 2cos .

k4L x 2 = 1 .12) Equation (5.12) predicts the growth of any component (specified by k or j ) admitted by the . 4 + sin L (5.

15) 4t  2xD 2 (5.13) As the sin2 term in Equation (5.  t rather than ix. we effective have the stablity (5. the time step should be at least half the size of the x2 term 2D @u @t 5.14) 1  j1 .2. if a central difference approximation is used for ..12) is always between criteria that. For example. nt we get centering the equation at ix. If all components are to decay. +1 An k 1 An k for stability (5. 0 and 1. (5.102 T RANSIENT H EAT C ONDUCTION system. 4j This condition will always hold if.3 Higher Order Approximations An improvement in accuracy and stability can be obtained by using a higher order approximation for the time derivative. t1  = D4 2 x 2 This can be rearranged to be of the form.e. n + 1 2 .16) i.

@u n+ un+1 .17) in place of Equation (5.  1 2 by @t i = i t i + O t2 (5. ui = D 1 @ u t 2 @x2  . un .7) and Equation (5.1) is approximated with the “Crank-Nicolson”formula +1 n 2 un i .

n+1 i 2u 1 @ + 2 @x2 .

18) 1 in which the spatial second derivative term is weighted by 1 2 at the old time step n and by 2 at the new time step n + 1. Thus each new time step requires the solution of a system of coupled equations. The price paid for the better accuracy (for a given t) and unconditional stability (i.1 . Notice that the finite difference time derivative has not changed . n  i (5. stable for any t) is that Equation (5.only the time position at which it is centred.the +1 equations for the new time step are now coupled in that un i depends on the neighbouring terms +1 n+1 un i+1 and ui.18) is an implicit scheme .e. ..

as well as x... The lightly shaded region shows where the solution is known at time step n.2: An implicit finite difference scheme based on central differences in t.5. i. I x 1 which tie together the 6 points shown by .3 T HE T RANSIENT A DVECTION -D IFFUSION E QUATION 103 t n+1 n : x x x x x x 1 00 1 2 . O x A generalization of (5.18) is +1 n 2 un i . ui = D . n + ) shown 2 by the . The equation is centred at the point (i.. F IGURE 5. The dark shading shows the region of the coupled equations.1 i i+1 ..

@ u t @x2 i .

n+1 + 1 . .

 .

19) in which the spatial second derivative of Equation (5.1) has been weighted by . @ 2 u n  @x2 i (5.

. at the new time step and by 1 .

8) is recovered when .  at the old time step. The original explicit forward difference scheme Equation (5.

An implicit backward difference scheme is obtained when .19) when 1 . = 0 and the implicit central difference (Crank-Nicolson) scheme (5.

= 1. .

the advection-diffusion equation..20) v v v . 5. The ratio of advective to diffusive transport is characterised by the Peclet number V L=D where v = k k2 and L is a characteristic length).  ru then represents advective transport by a velocity field . where u is concentration or temperature.=2 In the following section the transient heat equation is approximated for numerical analysis by using finite differences in time and finite elements in space. We also generalize the partial differential equation to include an advection term and a source term. @u + v  ru = Dr2u + f @t (5.g. D is the diffusivity and f is source term.3 The Transient Advection-Diffusion Equation Consider a linear parabolic equation where u is a scalar variable (e.

20) with weight Z .104 T RANSIENT H EAT C ONDUCTION Applying the Galerkin weighted residual method to Equation (5.

f ! d = ! gives =0 or  Z . @u 2 @t + v  ru . Dr u .

2.21) can be represented by a system of first order ordinary differential equations. Equation (5.24) can be re0.22) dt where M is the global mass matrix.23) Z1 @'m @'n @i @j Z1 n @i Kmne = D @ @  @x @x J d + vj 'm @' @i @xj J d i j k k 0 0 (5. Md (5. u1 = 0. n = 0.24) If the time domain is now discretized placed by t = nt.21) where @n is the normal derivative to the boundary @ . @ (5. @u @t + v  ru ! + Dru  r! d @ Z Z @u f! d + @n ! d. Putting = 'n un and = 'm and summing the element contributions to the global equations. : : : Equation (5. 1. u w u + K u . K the global stiffness matrix and u a vector of global nodal unknowns with steady state values (t ! 1) u1 . The element contributions to M and K are given by Mmne = and Z e 'm'nJ d (5.

1 1 (5.25) n+1 un   + K .

.un+1 + 1 .

t where . un = Ku1 M u .

is a weighting factor discussed in Section 5.2. Note that for .

25) as M + . Rearranging Equation (5. = the method is known 2 as the Crank-Nicolson-Galerkin method and errors arising from the time domain discretization are O Dt2.

tK un+1 = M . . 1 .

The stability of the above scheme can be examined by expanding (assumed to be smoothly .26) gives a set of linear algebraic equations to solve at the new time step n + 1 t from the known solution n at the previous time step nt. tK un + tKu1 u u (5.

the set of ordinary differential equations Equation (5. can be written as the recursion formula u I + . bi  ei t s i (5.27) The time-difference scheme Equation (5. with now replaced by a set of discrete values n at each time step nt.5. Writing the initial conditions 0 = ai i and steady state solution 1 = i s u u= u X i bi + ai .3 T HE T RANSIENT A DVECTION -D IFFUSION E QUATION 105 X A M K u s X bi si .26) on the other hand.22) has solution i continuous in time) in terms of the eigenvectors i (with associated eigenvalues i ) of the matrix = .1 .

1 . .tA with solution un+1 = I .

28) u= X i  1 . bi  (5. t 1 . tA un + tAu1 bi + ai . .

  n 1 + t.

27) by the approximation As i s e.29) are indeed the solutions of Equation (5.i t in Equation (5.i i si (5.25). respectively.29) shows that replacing the ordinary differential equations (5.27) and Equation (5.) Comparing Equation (5.25) is equivalent to replacing the exponential e.29) (You can verify that Equation (5. with t = nt.22) by the finite difference approximation Equation (5.it or. by substituting and using i = i i .27) and Equation (5.22) and Equation (5. .   1 . t 1 .

  n 1 + t.

i (5. .30) t 1 .

1 + t. ti e.it  1 . i = 1 .

 1 + t.

For stability we require ti  1 .1  1 . 1 + t. i i (5.31) The stability of the numerical time integration scheme can now be investigated by examining the behaviour of this approximation to the exponential.

since t.32) is trivially satisfied.32) since this term appears in Equation (5.29) raised to the power n. i (5. i and . The right hand inequality in Equation (5.

and the left hand inequality gives ti  2 1 + t. are all positive.

i or ti 1 . 2.

33) is that the scheme is unconditionally stable if 1 2  .33) A consequence of Equation (5.  2 (5.

 1. .

106 For .

T RANSIENT H EAT C ONDUCTION 1 2 the stability criterion is ti If the exponential approximation given by Equation (5. This oscillatory noise can be avoided by choosing 1 .31) is negative for any i the solution will contain components which change sign with each time step n. 2.

.34) t 1 . 1 . 2 (5.

These theoretical results are explored numerically with a Crank-Nicolson-Galerkin scheme 1 ) in Figure 5. max (5. where the one-dimensional diffusion equation (. but in practice this imposes a limit which is too severe for t and a small amount of oscillatory noise. Alternatively the oscillatory noise can be filtered out by averaging.35) where max is the largest eigenvalue in the matrix . is tolerated. associated with the high frequency vibration modes of the system.3.

consider the mass M . = 2 A @u = D @ 2 u @t @x2 subject to initial conditions and boundary conditions on 0x1 (5. t = x +  . Figure 5. Figure 5. t = 1 is solved for various time increments (t) and element lengths (x) for both linear and cubic Hermite elements.n n=1 2 2 (5. Decreasing x from 0:25 to 0:1 with linear elements produces more oscillation because the system has moredegrees of freedom and leads to greater oscillation.4 Mass lumping A technique known as mass lumping is sometimes used in which the mass matrix is replaced by a diagonal matrix having diagonal terms equal to the row sums. 0 = 0 u 0.n  t sin nx u x. At a sufficiently small t the oscillations are negligible (bottom right. With this value of t (0:01 s) the numerical results agree well with the exact solution (top.37) 5.3). For example. t = 0.36) u x.3) given by 1 X 2 1n e. u 1.

t = 0:01 1:0 0:5 t=1 0 u t = 0:02 t = 0:1 x x x x x x x x t = 0:05 x x x x x x x x x x x x 1 u x (b) x = 0:75 x x x x x x x x (c) x = 0:75 x x x x x x x t : x x x x linear elements 0 x u  = 0 25  =01 x t : : cubic elements 1 t 0 u  = 0 25  =01 : 1 t x (d) x = 0:9 x x x x x x x x (e) x = 0:9 x x x x x x x x 0  =01  =01 x : t : linear elements x t x linear elements 1 0  =01  = 0 01 x : t : 1 t F IGURE 5.4 M ASS LUMPING 107 ux. The top graph shows the exact and approximate solutions as functions of x at various times.3: Analytical and numerical solutions of the transient 1D heat equation showing the effects of element size x and time step size t. The lower graphs show the solution through time at the specified x positions and with various choices of x and t as indicated. . t x x Exact solution Linear CNG with x = 0:1.5.

M11 = M22 = M12 = M14 = ZZ ZZ ZZ 1 .6)).  2   1 =3 2 1 2 33 44 3 9 . 2 2 12 ZZ M13 = ZZ 1  1 = 1 and similarly M and M . 2 1 .23)) for a bilinear element (see Figure 1. 12 1 .108 T RANSIENT H EAT C ONDUCTION matrix ((5.9 and (1.

31 3 0 1 . 1 1 . 1 . 1 . 212 = 36 23 = . 12 2 1 . 1 . 3  3 = 18 1 and similarly M and M . Such a scheme has computational advantages when . 1 2 1 .! 40 mass lumping 0 0 0 1 7 4 0 07 1 0 4 05 0 0 0 1 4 3 The element mass is effectively lumped at the element vertices. 212 = 18 34 24 1 and similarly M . 2 . 1 1 1 1 2 1 1 . 1 12 = .32 3 0 11=1 1= 3 3 9 21 9 1 6 18 6 therefore M = 4 1 18 1 36 1 18 1 9 1 136 1 18 1 18 1 36 1 9 1 18 1 36 1 18 1 18 1 9 3 21 4 7 6 0 7 6 5 . 1 1 .

f = 0 are compared for the propogation and dispersion of an initial unit mass pulse at x = 0. To investigate the stability of time integration schemes run CMISS examples 3321 and 3322. t = p e 4Dt 4t (5. The exact solution is a Gaussian distribution whose variance increases with time: u x . . solution also using centered time differences.38) The finite element solution.see below. the finite element and finite differences (lumped f. The length of the solution domain is sufficient to avoid reflected end effects. To solve for the transient heat flow in a plate run CMISS example 331 2. is only conditionally stable (see (5.34)) and suffers from phase lag errors . using the Crank-Nicolson-Galerkin technique.e.4. D = 0:1 m2 s. For evenly spaced elements the finite element scheme with mass lumping is equivalent to finite differences with central spatial differences. V t2 . reproduces the amplitude of the pulse very well but shows a slight phase lag. In Figure 5.26) because each component of the vector n+1 is obtained directly without the need to solve a set of coupled equations. mass matrix) solutions of the one-dimensional advection-diffusion equation (5. however. This explicit time integration scheme.1 . or lumped mass. M u x. The finite difference. shows excellent amplitude and phase characteristics when compared with the exact solution. = 0 in Equation (5. 5.20) with V = 5 mps.5 CMISS Examples 1.

1 D = 0:1 m2 s. t t = 0:01 s t = 0:05 s x V = 5 m s. 0:05 s.1 Exact solution x x Finite element solution o o Finite difference solution x x x x x x x t = 0:2 s x x x x t = 0:4 s ox o o x o x oxx x o x o x ox x o x x F IGURE 5.5.01 for t  0:01 s.1. 0:2 s and 0:4 s) and finite difference solutions (at t=0:4 s only) are compared with the exact solution.5 CMISS E XAMPLES 109 u x. x= 0. The finite element solutions (at t=0:01 s. .4: Advection-diffusion of a unit mass pulse. t = 0:001 s for 0 t 0:01 s and t = 0.

.

2 Free Vibration Modes Consider an extension of Equation (6. 6. In direct time integration methods u t and _ t are replaced by finite differences and the resulting system of algebraic equations is solved at u f u t = Px t _ t + P T KPx t = P T f P T MP x t + P T CP x _ t + Ku t = f t M u t + C u (6. In fact. a suitable transformation (6. can be chosen to diagonalize and and thereby uncouple the equations entirely.1) successive time steps. nodal point accelerations) M .or analysed by mode superposition..g. This transformation (which is still applicable when damping is P M K .as in the last section for parabolic equations . the time-dependent solution is expressed as the superposition of the natural (or resonant) modes of the system. if a solution is required over a long time period.2) applied to Equation (6. f t is the external load vector and u t is the vector of n nodal unknowns.3) having a much smaller bandwidth than in the original system and hence being more economical to solve. To find these modes requires the solution of an eigenvalue problem. damping and stiffness matrices. That is. For a small number of steps this is the most economical method of solution but. C and K are the mass. respectively.1 Introduction The system of ordinary differential equations which results from the application of the Galerkin finite element (or other) discretization of the spatial domain to linear parabolic or hyperbolic equations can either be integrated directly .Chapter 6 Modal Analysis 6. or for a large number of different load vectors t.3) which includes second order time derivatives (e. if damping is neglected.1) can result in the matrices of the transformed system (6.

2 !n ST MS = I (6.8) (6. Substituting Equation (6. : : : .5) into Equation (6.112 M ODAL A NALYSIS included but does not then result in an uncoupled system unless further simplications are made) is found by solving the free vibration problem M u t + Ku t = 0 Proof: Consider a solution to Equation (6..6) 2 s  .whose columns are the M -orthonormalised eigenvectors of satisfying Equation (6.rewriting Equation (6. (6.10) or ST KS = ST MS  = I  =  P (6. : : : .5) when the original partial differential operator is self-adjoint) the eigenvectors are orthogonal and can be “normalized ” such that Ks = !2Ms (6. Combining the n eigenvectors into a matrix = 1.4) where ! and t0 are constants and is a vector of order n.4) gives the generalized eigenproblem s u t = s sin ! t .the modal matrix .12) . ! 2 s ..11) (i.9) where 3 7 7 7 5 (6. 2 . If K is a symmetric matrix (as is the case having n eigensolutons !1 1 2 2 n n siT Msj = 1 i=j 0 i= 6 j (6. Thus the modal matrix .6) becomes KS = MS 2!2 1 2 6 !2 6 =6 4 0 0 .6)) .4) of the form (6. (6. . ! 2 s  .7) S s s where (the eigenvectors are said to be M -orthonormalised ).can be used as the transformation matrix in Equation (6.7) as s I is the identity matrix. t0 .1) to the canonical form K _ t + x t = S T f t x t + ST CS x (6. n .e.2) required to reduce the original system of equations (6.

!2 = 0 or !4 . when used as the transformation matrix.e. n (6.p 7 6 6 giving the modal matrix S = 4 13 2 5 which. s p 3 K M  p 6 . To find the magnitude of the eigenvectors we use Equation (6.  d + i sin !it + i cos !id i t 0 where the constants i and i are determined from the initial conditions 1 xi 0jt=0 = siT M u 0jt=0 x _ i 0jt=0 = xi 0jt=0 = siT M u 0jt=0 siT M u 0jt=0 (6.2  2 s=0 ..2 4 and f = 10 Ks = !2Ms 6 .12) becomes a system of uncoupled equations xi t + !i2xi t = ri t i = 1.! To find the solution by modal analysis we first solve the generalised eigenproblem i.3 An Analytic Example As an example. This characteristic polynomial 2 T has solutions ! = 2. 2. The M -orthognormalised eigenvectors are now s1 = p 3 3 6 6 3 21 1 p .1 = 0). 5 with corresponding eigenvectors T 1 = a 1 1 .e. K = .6. 7!2 + 10 = 0.13) where xi is the ith component of and ri is the ith component of the vector this system is given by the Duhamel integral x S T f . p p .2 4. s   2 01 2 0.7).3 A N A NALYTIC E XAMPLE 113 With damping neglected equation Equation (6. The solution of (6. i. 2!2 .. (Notice that the orthogonality condition is satisfied: ab 1 1 1 2  1 01   1 2 T T p and s2 = .1 2 0 1 2 = 1  b = p 3 3      2 0 . has a solution when det .14) Z 1 xi t = ! ri   sin !i t .1     1 1 2 2 a 1 1 0 1 1 =1a= p b . 1 = b .15) 6.1 2 . : : : .2 0 M = 0 1 . consider the equilibrium equations M u + Ku = f where 2 0  6 .

p p 6 n X i=1 6 p by means of the Duhamel integral (6. t0 p 3 and 2 13 . t00 : p 6 The solution of the non-homogeneous system.16) as the superposition of the natural modes (eigenvectors) of the homogeneous equations. from Equation (6.16) Notice that the solution is expressed in Equation (6.12) is then not a system of uncoupled equations.if it coincides then resonance will occur. the higher frequency modes can be ignored and the solution adequately represented by superposition of the p eigenvectors associated with the p lowest eigenvalues. : : : .14) (in this case with constant) and then. 6 . One simplification often made in order to retain the diago- S .114 M ODAL A NALYSIS reduces the stiffness matrix to 2 1 13 21 13     p p p p .1) the transformation to a lower bandwidth system is still based on the model matrix but Equation (6. 2 .p 7 p u2 t = 6 4 2 6 5 sin 5 t . 3 37 2 0 6 3 67 1 0 S T MS = 6 4 1 25 0 1 41 2 5= 0 1 =I . n s r 6 u t = Sx t = sixi t (6. 6. 4 1 2 67 5= 2 2 2 4 0 5 p p . where p n. subject to given initial conditions. If the forcing function (load vector) is close to one of these modes the corresponding coefficient xi will be large and will dominate the response .p p 6 6 and the mass matrix to p 3 p 6 Thus the natural modes of the system are given by 213 p7 p u1 t = 6 4 13 5 sin 2 t . Very often it is unnecessary to evaluate all n eigenvectors of the system. is found by solving the uncoupled equations x t + P S s s 2 0 0 5 2 1 13 2 10 3   p p p7 0 6 7 6 3 3 x t = 4 1 2 5 10 = 4 203 5 .p p 6 6 3 6 21 2 1 13 13     p p p p .2 6 3 0 = 3 37 ST KS = 6 4 1 5.2) with =  1 .4 Proportional Damping When element damping terms are included in the original dynamic equations (6.

 dt + e.6.12) is to approximate the overall energy dissipation of the finite element system with proportional damping siT Csj = 2!ii ij . To analyse a clamped beam modal analysis run CMISS example 452 3.5 CMISS E XAMPLES 115 nal nature of Equation (6.18) with solution (the Duhamel integral) s i and i are calculated from the initial conditions Equation (6.5 CMISS Examples 1.15). To analyse a plane stress modal analysis run CMISS example 451 2. Equation (6. To analyse a steel-framed building modal analysis run CMISS example 453 .17) where i is a modal damping parameter and ij is the Kronecker delta.19) (or alternative time integration methods applied to (6.  : sin !i t . 2 x t = r t xi t + 2!iix _ i t + !1 1 i (6.12) now reduces to n equations of the form (6.i!it f i sin !it + i cos !itg i t 0 (6. Z 1 xi t = ! ri   :ei!i t.18)). p where ! i = !i 1 .16). i2 . the solution t is expressed as a superposition of the mode shapes i by Equation (6.19) u 6. Once the components xi t have been found from Equation (6.

116 M ODAL A NALYSIS .

No specific boundary conditions are prescribed but in some cases regularity conditions at infinity need to be satisfied. Telles & Wrobel 1984a).1) where * indicates the adjoint of the operator L and is the Dirac delta function. have known fundamental solutions. Domain integrals will also arise for inhomogeneous equations.Chapter 7 Domain Integrals in the BEM 7. This property is advantagous as it reduces the size of the solution system leading to improved computational efficiency. in general. . To achieve a boundary integral formulation it is necessary to find an appropriate reciprocity relationship for the problem and to determine an appropriate fundamental solution. the diffusion equation and the wave equation (Brebbia. The most common difficulty in applying the BEM is in determining an appropriate fundamental solution. To achieve this reduction of dimension it is necessary to formulate the governing equation as a boundary integral equation. This reduction of dimension also eases the burden on the engineer as it is only necessary to construct a boundary mesh to implement the BEM. The fundamental solution for this equation is a solution of L! x.  +  = 0 (7. The mathematical theory required to determine the fundamental solution of a constant coefficient PDE is well-developed and has been used successfully to determine the fundamental solutions for a wide range of constant coefficient equations (Brebbia & Walker 1980) (Clements & Rizzo 1978) (Ortner 1987). is an inhomogeneous source term and u is the dependent variable.1 Achieving a Boundary Integral Formulation The principal advantage of the BEM over other numerical methods is the ability to reduce the problem dimension by one. A linear differential equation can be expressed in operator form as Lu = where L is a linear operator. In particular. by no means can it be guaranteed that the fundamental solution to a specific differential equation is known. PDEs with variable coefficients do not. If the fundamental solution to an operator cannot be found then domain integrals cannot be completely removed from the integral formulation. However. If either of these requirements cannot be satisfied then a boundary integral formulation cannot be achieved. Fundamental solutions are known and have been published for many of the most important equations in engineering such as Laplace’s equation. The fundamental solution is a Green’s function which is not required to satisfy any boundary conditions and is therefore also commonly termed the free-space Green’s function.

1984a). This allows a problem to be subdivided into a number of zones for which the BEM can be applied individually. the presence of domain integrals in the BEM formulation negates one of the principal advantages of the BEM in that the problem dimension is no longer reduced by one.2 Removing Domain Integrals due to Inhomogeneous Terms Inhomogeneous PDEs occur for a large number of physical problems. The BEM generates a formulation involving both the dependent variable u and the flux q. Gq = . This approach. however. This method is simple and has been shown to produce accurate results (Brebbia et al. This zoning approach is suited to problems with significantly different length scales or different properties in different areas. a body force term.1 The Galerkin Vector technique For some particular forms of the inhomogeneous function the domain integral can be transformed directly into boundary integrals. This allows flux boundary conditions to be applied directly which cannot be achieved in either the finite element or finite difference methods. Domain integration can be simply and accurately performed in the BEM. Several methods have been developed which allow domain integrals to be expressed as equivalent boundary integrals. requires a domain discretisation and a numerical domain integration procedure which reduces the attraction of the BEM over domain-based numerical methods. What distinguishes these phenomena is the boundary conditions of the problem.2) The domain integral in this formulation does not involve any unknowns so domain integration can be used directly to solve this equation. This requires discretising the domain into internal cells in much the same way as for the finite element method. He argued that for problems such as flow problems a wide range of phenomena are described by the same governing equations. In this section these methods will be discussed. Z !d (7. 7. For this reason accurate description of the boundary conditions is vital for solution accuracy. An inhomogeneous linear PDE can be expressed in operator form as Lu = where is a known function of position or a non-zero constant. However. Another advantage of the BEM over other numerical methods is that it allows an explicit expression for the solution at an internal point. Consider the Poisson equation r2 u = . or due to initial conditions in time-dependent problems. An inhomogeneous term may arise due to a number of factors including a source term. the resulting BEM formulation will be Hu . If the fundamental solution is known for the operator L.118 D OMAIN I NTEGRALS IN THE BEM Wu (1985) argued that the BEM has several advantages over other numerical methods which justify its use for many practical problems . Applying the BEM gives an equation of the form of . As the domain integral does not involve any unknown values accurate results can generally be achieved using a fairly coarse mesh.even in cases where domain integration is required.2. 7.

 r2 v .7.2 R EMOVING D OMAIN I NTEGRALS DUE TO I NHOMOGENEOUS T ERMS 119 Equation (7. Using Green’s second identity Z .2). vr2 d Z .

(7. then for the specific case of being harmonic (r2 = 0) Green’s second identity can be reduced to domain integration can be avoided for certain forms of . If a v can be found which satisfies Z Z . where ! is the fundamental solution of Laplace’s equation. . @v @ = @n .3) r2v = !. v @n d.

4) Therefore if a Galerkin vector can be found and is harmonic the domain integral in Equation (7. as this method removes the burden of preparing a domain mesh. is required to be harmonic). yi  is the value of the integrand at random point xi . it only applies to specific forms of the inhomogeneous term (i. @v @ !d = @n ..2 The Monte Carlo method Domain discretisation could be avoided by using a Monte Carlo technique. a domain integral I is approximated as N X A I  N f xi . rotation about a fixed axis or steady-state thermal loading. yi i=1 (7. Fairweather. yi . Unfortunately this method often proves to be computationally expensive as a large number of integration points are needed for accurate domain integration. This approximation allows a domain integral to be approximated by a summation over a set of random points so domain integration can be performed without requiring a domain mesh. The method was proposed by Gipson (1987). Shippy & Wu (1979) determined the Galerkin vector for the two-dimensional Poisson equation and Monaco & Rangogni (1982) determined the Galerkin vector for the threedimensional Poisson equation. This method has the secondary advantage of allowing the integration to be performed over a simple geometry enclosing the problem domain . Unfortunately. . . (7. This Galerkin vector approach provides a simple method of expressing domain integrals as equivalent boundary integrals. Gipson has successfully applied this method to a number of Poisson-type problems. Gipson argues however that. This technique approximates a domain integral as a sum of the integrand at a number of random points. 7. Specifically. He considered the practical cases where the body force term arose due to either a constant gravitational load. N is the number of random points used and A is the area of the region over which the integration is performed. v @n d. Danson (1981) showed how this method can be applied successfully for a number of physical problems involving linear isotropic problems with body forces.2) can be expressed as equivalent boundary integrals.e. in two dimensions.5) where f xi .2.if a random point is not in the problem domain its contribution is ignored. In each of these cases the domain integral can be expressed as equivalent boundary integrals. Rizzo.

This is especially likely if L involves variable coefficients for which case it has been shown that it is particularly difficult to find ^ a fundamental solution. This approach can be applied in a situation where an analytic expression for a particular solution can be found. Instead. a BEM formulation can be derived based on a related operator L ^ with known fundamental solution. L u! d (7. . Applying BEM to the governing equation using the expansion u = uc + up gives Hu .8) ^ . Coleman & Phan-Thien (1991) proposed a method where a particular solution is determined by approximating the inhomogeneous source term using a global interpolation function.3 Domain Integrals Involving the Dependent Variable Consider the linear homogeneous PDE Lu = 0. Consider the linear problem Lu = . 7.2). Zheng. A BEM formulation for Lu = 0 based on the operator L will be of the form Hu . Gq = .2. Alternatively an approximate particular solution could be calculated numerically. Unfortunately particular solutions are generally only known for simple operators and for simple forms of . all values on the right-hand-side of Equation (7.120 D OMAIN I NTEGRALS IN THE BEM the extra computational expense is justified. Gq = Hup . and a particular solution up which satisfies Lup = but is not required to satisfy the boundary conditions of the problem. For many operators the fundamental solution to the operator L may be unobtainable or may be in an unusable form. u can be considered as the sum of the complementary function uc. This integral equation is where ! is the fundamental solution corresponding to the operator L similar to Equation (7. Gqp Hu .6) If a particular solution up can be found. This approach is a special case of a more general method known as the dual reciprocity boundary element method. This problem could be solved using domain integration where the internal nodes are treated as formal problem unknowns.7) where is a vector of known values. Z ^ L .reducing the problem to (7.3 Complementary Function-Particular Integral method A more general approach can be developed using particular solutions. d 7. However in this case the domain integral term involves the dependent variable u. which is a solution of the homogeneous equation Luc = 0. Gq = d (7. This linear system can be solved by applying boundary conditions.6) are known .

He considered the two-dimensional generalised Laplace equation r   x.9) can be recast as a heterogeneous r2 u + f x.boundary element method to the general second-order variable coefficient PDE solution to Equation (7. y @u = h x.12) into Equation (7. y u = 0 (7. Rangogni treated this equation as a perturbation about Laplace’s equation. y @u + g x. y  Equation (7. Rangogni reported that in practice this r2 u + f x.9) 1 Using the substitution V x.  .3. Each successive uj can then be found by solving a Poisson equation with homogeneous boundary conditions as uj . y u = 0 for which he sought a solution of the form 2 where 0  "  1 (7. This allows Equation (7. Rangogni (1991) extended this coupled perturbation .3 D OMAIN I NTEGRALS I NVOLVING THE D EPENDENT VARIABLE 121 7.13) to be treated as an infinite series of distinct problems which can be solved using the boundary element method.10) is therefore given by 1 X uj .14) . Rangogni used a domain discretisation to solve these Poisson problems. y = 0 .11) u = u0 + "u1 + " u2 + : : : = 1 X j =0 uj "j (7.12) Substituting Equation (7.1 The Perturbation Boundary Element Method Rangogni (1986) proposed solving variable coefficient PDEs by coupling the boundary element method with a perturbation method.7. He considered the class of equations r2 u + "f x.  (7. y  =  2 u x.13) A solution will only exist for all values of " if the terms at each power of " equal zero. u0 can be found by solving r2 u0 = 0 which Rangogni assumes will satisfy the boundary conditions of the original problem. The j =0 series converged rapidly and in his numerical examples he achieved accurate results using only u0 and u1 .1 has been previously determined. Helmholtz equation (7.11) and grouping powers of " gives r2 u0 + " r2u1 + fu0 + "2 r2 u2 + fu1 + : : : = 0 .10) where f is a known function of position. y rV x.10) is a particular member of this family of equations for which " = 1. y @x @y (7. Equation (7.

y @u + g  x. Reible & Savant (1987) considered a class of hyperbolic and elliptic problems which can be transformed into an inhomogeneous Helmholtz equation.15) Applying the perturbation method to this family of equations allows Equation (7. y 0  "  1 r u + " f x. + b0 !0 d . Instead of using one higher-order fundamental solution. Lafe & Cheng investigated the convergence of the perturbation method. Since then the method has been successfully applied to a wide range of problems. They used the perturbation method to recast this as an infinite sequence of Poisson equations.16) where b0 = b0   is a known function of position. They avoided domain discretisation by using a Monte Carlo integration technique (Gipson 1987) to evaluate the required domain integrals.17) needs to be expressed as equivalent  . They showed the method produced accurate results for simply varying hydraulic conductivities with convergence after two or three terms. to convert the remaining domain integrals to equivalent boundary integrals a series of higher-order fundamental solutions is used. The MRM can be viewed as a generalisation of the Galerkin vector approach. To avoid domain discretisation the domain integral in Equation (7. They found that for rapidly varying hydraulic conductivity convergence is not guaranteed. gives x Z @!0 Z c  u  + u @n d. using the fundamental solution to the Laplace operator. Again Rangogni used an domain mesh to solve these Poisson equations. 7. This process could become computationally expensive.17) where !0 is the known fundamental solution to Laplace’s equation applied at point . = Z . the Galerkin vector. as the solution of multiple subproblems will be required within each subregion. If the hydraulic conductivity variation is more significant they recommend using the perturbation method in conjunction with a subregion technique so that the variation of conductivity within each subregion satisfies their requirements. Lafe & Cheng (1987) used the perturbation method to solve steady-state groundwater flow problems in heterogeneous aquifers.122 D OMAIN I NTEGRALS IN THE BEM He considered the family of equations 2 @u = h x. @u d. y  @x @y   (7. Gipson. !0 @n (7. From this investigation they concluded that accurate results can be obtained so long as the hydraulic conductivity does not vary by more than one order of magnitude within the solution domain. particularly if convergence is not rapid. Consider the Poisson equation r2u = b0 (7. Applying BEM to this equation.2 The Multiple Reciprocity Method The multiple reciprocity method (MRM) was initially proposed by Nowak (1987) for the solution of transient heat conduction problems.3.15) to be expressed as an infinite series of distinct Poisson equations which can be solved using the boundary element method. Rangogni found that in practice convergence was rapid and accurate results were produced.

19) or Z Z Z .3 D OMAIN I NTEGRALS I NVOLVING THE D EPENDENT VARIABLE 123 boundary integrals. Using MRM this is achieved by defining a higher-order fundamental solution !1 such that r2!1 = !0 (7.17) can be written as Z b0 !0 d = Z b0 r2 !1 d (7.18) Using this higher-order fundamental solution the domain integral in Equation (7.7.

2.21) giving Z !1 r b0 d = 2 Z !1b1 d (7.25) Using these recurrence relationships gives the boundary integral formulation Z . 2. b1 = r2 b0 (7. 1.23) bj+1 = r2 bj r2 !j+1 = !j for j for j = 0. : : : = 0. : : : (7. 1. @!1 @u b0 !0 d = u @n .22) This process can be repeated by introducing a new higher-order fundamental solution !2 such that r2!2 = !1 and continuing until convergence is reached.20) This formulation has generated a new domain integral. !1 @n d. This procedure is based on the recurrence relationships (7. + !1r2 b0 d .24) (7. b0 is a known function so we can introduce a new function b1 which can be determined analytically from the relationship (7.

@!0 @u 1 Z .

Errors are only introduced at the stage of boundary discretisation. X @! @b j +1 j c  u  + u @n .26) which is an exact formulation if the infinite series converges. . = 0 j =0 . + bj @n . !0 @n d. . !j+1 @n d. (7.

2j u In this case the boundary integral formulation will be (7.2 u and the recurrence relationship defined by Equation (7. The MRM can be applied based on operators other than the Laplace operator.28) c  u  + 1 Z X j =0 . t. This approach relies on knowledge of the higher-order fundamental solutions necessary for application of the method. . u.27) where J +1 and J +1 are influence coefficient matrices corresponding to the higher-order fundamental solutions and j and j contain the nodal values of bj and its normal derivative.25) . These solutions have been determined and successfully used for the Laplace operator in both two and three dimensions but the extension of the method to other equation types needs further research.24) and (7.can be employed.24) becomes simply x uj+1 = r2 uj = . GJ +1rj  (7. Brebbia & Nowak (1989) have applied the MRM to the Helmholtz equation r2 u + 2 u = 0 where b0 = .124 D OMAIN I NTEGRALS IN THE BEM Introducing interpolattion functions and discretising the boundary gives the matrix system H0u . G0q = H G p r 1 X j =0 HJ +1pj . The MRM can be extended to other equations by allowing the forcing function b0 to be a general function such that b0 = b0  . The MRM will be restricted to cases where the recurrence relationships .Equations (7. Itagaki & Brebbia (1993) have determined the higher order fundamental solutions for the two-dimensional modified Helmholtz equation.

3. Applying the BEM to Equation (7. !j @n d.30) will x x x . Consider a second-order PDE which can be expressed in the form r2u = b (7.29) 7. The most common choice is the Laplace operator (Partridge. u. @! @u j  u @n . For potential problems b = b  . Brebbia & Wrobel 1992) and in this chapter the DR-BEM will be illustrated for this choice. u and for transient problems b = b  . If b = b   then b is a known function of position and the differential equation described is simply the Poisson equation.30) The forcing function b can be completely general. Instead the DR-BEM is applied using an appropriate related operator with known fundamental solution. t.3 The Dual Reciprocity Boundary Element Method Equation Derivation The dual reciprocity boundary element method (DR-BEM) was developed to avoid the need for domain integration in cases where the fundamental solution of the governing differential equation is unknown or is impractical to apply. = 0 2j (7.

30) the forcing function b is approximated by a weighted summation of particular solutions to the Poisson equation.34) The DR-BEM essentially constructs an approximate particular solution to the governing PDE as a summation of localised particular solutions. r2u = ^j jr u 2 (7.called poles .35) is obtained.7. The aim of the DR-BEM is to express the domain integral due to the forcing function b as equivalent boundary integrals. A global approximation to b of the form b= M X j =1 j fj (7. Green’s theorem is applied twice and the fundamental solution of the Laplacian is used to remove the remaining domain integrals. b! d Z (7.33) By substituting Equations (7. The approximating functions fj are applied at M different collocation points . : : : . Gq = . of which are located on the boundary of the problem domain.32) and (7.34). which may be difficult to determine. The DR-BEM uses the idea of approximating b using interpolation functions.3 D OMAIN I NTEGRALS I NVOLVING THE D EPENDENT VARIABLE 125 give Hu . r2 u ^j = fj j = 1. Equation (7. the DR-BEM employs a series of particular solutions u ^j which are related to the approximating functions fj as shown in Equation (7. The name dual reciprocity BEM is derived from the application of reciprocity relationships to both sides of Equation (7.33). M M X j =1 (7. After applying these steps Equation (7.33) into Equation (7.32) is proposed.34) the standard boundary element approach can be applied. With the governing equation rewritten in the form of Equation (7. Instead of using a single particular solution.31) where ! is the known fundamental solution to Laplace’s equation. but not all.34) is multiplied by a weighting function ! and integrated over the domain.generally most. j are unknown coefficients and fj are approximating functions used in the interpolation and are generally chosen to be functions of the source point and the field point of the fundamental solution. As discussed in the previous chapter the solution to a linear PDE Lu = can be constructed as the sum of a complimentary function uc (which satisfies the homogeneous equation Luc = 0) and a particular solution up to the equation Lup = . where the fundamental solution pole is applied at .

Z .126 D OMAIN I NTEGRALS IN THE BEM point .

@! @u c  u  + u @n . . 0 1 . ! @n d.

A = u ^j @! . F = F . If the matrix is nonsingular this expression can be rearranged to give Equation (7.36) to be rewritten as u q U Q ^ ^ Hu .36) where the M poles were chosen to be the N boundary nodes plus I internal points so that M = N + I . ! j c  uj  + @n @n j =1 .37) b F where is a vector containing the nodal values of . The internal points can be chosen to be locations where the solution is of interest. The ^ j and ^j vectors can be treated as columns of the matrices ^ and ^ respectively. Gq = N +I j =1 X j Hu ^j .32) which.35) In implementing a numerical solution of this equation similar steps are taken as for the standard BEM. GQ (7.it is only necessary to specify the coordinates of the internal collocation points. In general using internal points is likely to improve the solution accuracy as it increases the number of degrees of freedom. 1992) (Huang & Cruse 1993) using boundary points only in this procedure is insufficient to define the problem.  (7. It has been shown that for many problems (Partridge et al.33) once the approximating functions fj have been defined. The form of each u ^j is known from Equation (7.35). No theory has been developed of how many internal collocation points should be used for optimal accuracy. This allows Equation (7. for the nodal values. It is not necessary to use interpolation functions to approximate each u ^j .1b (7. The boundary is discretised into elements and interpolation functions are introduced to approximate the dependent variable within each element. The application of this method results in the system H G Hu . Gq = H U . Although it is not generally necessary to include poles at internal points it has been found that in general improved accuracy is achieved by doing so (Nowak & Partridge 1992). Using internal poles in this interpolation does not require domain discretisation . However by using the same interpolation functions to approximate u and u ^j the numerical implementation will generate the same matrices and on both sides of Equation (7.38) . 1992). can be expressed in matrix form as = . Z M X @ @hatuj d. To solve this system it is necessary to evaluate . Gq ^j  (7. The error generated by approximating each u ^j in this manner has been found to be small and can be justified by the improved computational efficiency of the method (Partridge et al.38) which provides an explicit expression for . is defined by Equation (7. or where these points should be positioned within the problem domain.

Some work has been conducted into investigating what form of fj should be used in a given situation to provide the highest accuracy and computational efficiency. 1992) but it has been found that in general the most accurate results were generated using some radial function. to specify u ^ F and q ^. applying Equation (7. Therefore the choice of the approximating functions fj is a key consideration when implementing the DR-BEM. Brebbia & Nardini (1986) recommended choosing fj from the series x x 1 x.41) Using f d2u ^ + 1 du ^ =f 2 dr r dr defined by Equation (7.e.  = . for simplicity. if f = f r  then the relationship r2 u ^ = f r can be reduced to the ordinary differential equation (7. fj = fj r .40) where rj is the distance between the field point (node j ) and the DR-BEM collocation point (node i). For two-dimensional problems. The fundamental solution of Laplace’s equation is !  space and !  .33). Several other options for fj have been tried (Partridge et al.39) The approach taken to solve this equation will depend on the form of b. 21 ln r in two-dimensional  fj = 1 + rj + rj2 + : : : + rjm (7. They showed that accurate results can be achieved using some combination of terms from this series. The Approximating Function f The accuracy of the DR-BEM hinges on the accuracy of the global approximation to the forcing function b (defined by Equation (7.. the form of approximating functions fj will be referred to by a single f .7.where r is the Euclidean distance between the field point and the source point of the fundamental solution.  = 4r in three-dimensional space . Gq = H U .1 Hu .37) gives (7. The only requirement so far prescribed on the form of the approximating functions fj is that the matrix generated should be nonsingular and that the related particular solutions u ^j can be determined and can be expressed in a practical closed form. Generally the same approximating function fj is used at all the collocation points so in this thesis. Usually a form of fj is defined and this can be used.40) the corresponding forms of (7. For both two and three-dimensional problems Wrobel. GQ F b in Equation (7. Due to the dependence of this fundamental solution only on r the approximating function is generally chosen to be some radial function i.32)). Choosing f to be a function of only one variable simplifies the process of determining u ^ and q ^.42) u ^ and q ^.3 D OMAIN I NTEGRALS I NVOLVING THE D EPENDENT VARIABLE 127 Including this explicit expression for ^ ^ . for two-dimensional .

can be shown to be 2 3 m+2 u ^ = r4 + r9 + : : : + r 2 m + 2 .128 D OMAIN I NTEGRALS IN THE BEM problems.

@x @y .

The theory of using radial basis functions for multi-dimensional approximation is fairly advanced. it seems that choosing f to be a radial function is a logical extension for two or three-dimensional problems.3. Recently some more formal analysis of the use of approximating functions has been undertaken. If a different operator is used as the basis of the DR-BEM then it is likely a different form of f will be more appropriate. choosing f to be a radial basis function ensures convergence of the DR-BEM. It has been found that in general including higher-order terms leads to little improvement in accuracy (Partridge et al. 1992). The performance of the DR-BEM hinges on the choice of the approximating function f . Therefore. It has been shown that f = r is optimal for three-dimensional problems which justifies the use of f = 1 + r when applying the DR-BEM to three-dimensional problems . for the Poisson equation. The choice of f in this case will be discussed in Section 7. Equation (7. The theory of how to determine the best approximating function is therefore a vital component of the DR-BEM. Unfortunately the approximating function has generally been chosen and used in a rather ad-hoc manner. (7. However for two-dimensional problems it has been shown that optimal approximation is attained using the thin plate spline f = r 2 logr . Golberg & Chen showed that.40) can be used for specifying f . In this case it is not necessary to invert the matrix as can simply be calculated from = using Gaussian elimination. This observation lead Golberg & Chen to suggest that choosing f to be a thin plate spline may improve the accuracy of the DR-BEM in two dimensions. F Inhomogeneous Equations If the forcing function b is a function of position only then the differential equation under consideration is simply Poisson’s equation. Cubic splines are known to be optimal for one-dimensional interpolation. xi and ry = yj .43) (7. rather than being an arbitrary choice.3. Recently Golberg (1995) has published a review of the DR-BEM concentrating on developments since 1990 concerning the numerical evaluation of particular solutions.39) can be rewritten b F F .the constant is included to ensure a non-zero diagonal for . yi . Golberg & Chen (1994) argued that a formal analysis of the approximating function f can be undertaken using the theory of radial basis functions.44) Any combination of terms from Equation (7.40) was recommended as a basis for the approximating function f due to the particular form of the fundamental solution of Laplace’s equation and its dependence on r only. Radial basis functions are a generalisation of cubic splines in multi-dimensions. Equation (7. m 1 r r q ^ = rx + ry @n @n 2 + 3 + : : : + m + 2 where rx = xj . They also demonstrated that f = 1 + r is a specific member of the group of radial basis functions. The most commonly used form is f = 1 + r as this approximation will generally give accurate results with greater computational efficiency than other choices.

49) Again. suggested several alternative ways of determining the unknown coefficients j for inhomogeneous equations. the linear second-order differential equation r2 u + u = 0 (7.48) which can be rearranged to give ^ ^ . based on the fundamental solution to Laplace’s equation.3 D OMAIN I NTEGRALS I NVOLVING THE D EPENDENT VARIABLE 129 as ^ ^ Hu .46) using singular value decomposition. Coleman et al. GQ F u (7. gives F . Instead the solution can be treated as a coupled problem and the solutions at boundary and internal nodes are generated simultaneously. for example. GQ F F (7.7. Consider. However. N X j =1 j fj rm  ! (7. Elliptic Problems If b is a function of the dependent variable then will also be a function of the dependent variable. For large systems they found the computational efficiency could be improved by employing the conjugate gradient method.1 . by applying boundary conditions Equation (7.49) can be reduced to a linear system = which can be solved to determine the unknown nodal values. (1991) and Coleman. Zheng et al. (1991) successfully solved inhomogeneous potential and elasticity problems which are governed by operators other than the Laplacian.1 H + S  u = Gq where S = H U . for example. Dassios & Beskos (1994) this method is essentially equivalent to the DR-BEM. Zheng et al. 1992).45) can be reduced to a linear system = which can be solved to give the unknown nodal values of u and q . (1991) used a least-squares method where they minimised the sum of squares S= M  X m=1 b rm  . As discussed by Polyzos.47). and Coleman et al. Gq = .45) By applying boundary conditions Equation (7. Due to the presence of the fully-populated .1 Hu .1 matrix in Equation (7. Tullock & Phan-Thien (1991) have proposed a method which uses a global shape function to construct an approximate particular solution. Derivative Terms The DR-BEM can also be applied for elliptic problems where b involves derivatives of the dependent variable (Partridge et al. GQ Ax (7.49) it is not possible to solve the boundary problem and internal problem separately. Gq = d where d = H U . Zheng et al. the differ- AX . Consider.u so = .u ^ ^ . Applying the DR-BEM to Equation (7. H U .47) In this case b = .

For example.56) can be reduced to a linear system which can be solved to give the unknown nodal values.1u @x @X Equation (7. using the Laplace fundamental solution.52) can be used to approximate u where j are the chosen interpolation functions and j are the unknown coefficients.53) gives u= (7. evaluating .130 D OMAIN I NTEGRALS IN THE BEM ential equation r2u + @u @x = 0 In this case applying DR-BEM.1 R = HU .32).53) Although it is not necessary. GQ F @X F (7.39) can now be rewritten as (7. As mentioned earlier. This can lead to numerical problems if derivative terms are included in the forcing function b.1 @F . H U .50) ^ ^ . equating  to F improves the computational efficiency of the method F @u = @  @x @X (7. Gq = .1 @u Hu .55) H + R u = Gq where ^ ^ . y  j (7. GQ F @x u (7. the approximating function f is generally chosen to be f = 1 + r .56) By applying boundary conditions Equation (7.54) to be rewritten as @u = @F F . As shown in Equation (7.55) derivative terms require derivatives of f to be evaluated. This is achieved by using interpolation functions to approximate in a similar manner as was used to approximate b in Equation (7. Differentiating Equation (7.54) allows Equa- Choosing  = and inverting Equation (7.53) to give an explicit expression for tion (7.51) To solve this problem it is necessary to relate the nodal values of u to the nodal values of @x . gives (7. A global approximation function of the form @u u= M X j =1 j x. In system form this can be expressed as as only one matrix inversion procedure is required.

. 0 0  .e. .60) where M is the number of collocation points used in applying the DR-BEM. These approximating functions produce accurate results and can be simply applied for both linear and nonlinear problems. .  xM .3 D OMAIN I NTEGRALS I NVOLVING THE D EPENDENT VARIABLE 131 the @F matrix requires calculation of @f . y i. .53) to allow nodal values of u to be associated to its derivatives can be applied to extend the DR-BEM to cases involving higher-order derivatives or cross derivatives of the dependent variable.57) This derivative function can become singular.significant numerical error may result. Zhang (1993) suggested two possibilities for avoiding this problem.as shown by Zhang (1993) . This will especially be the case in problems where collocation points are located close together. yM  0 0 3 7 7 7 5 (7. The first suggestion involved using a mapping procedure to map the governing equation to an equation without convective terms.. . Appropriate approximating functions need to be chosen to avoid the problem of singularities. . If the DR-BEM is applied using the known fundamental solution to the Laplace operator then the forcing function is b = .u. GQ F b Ku K (7. 4 ..58) where  is a function of position . . y  in two dimensions. This method was shown to produce accurate results but is somewhat cumbersome and can only be applied to linear problems.. so . The relationship b = . y  0    1 1 6 0  x2 .1 Hu . y2    K=6 6 . Zhang recommended use of either f = 1 + r3 or f = 1 + r2 + r3. Consider the variable coefficient Helmholtz equation r2u +  x u = 0 (7. Gq = H U . .u can be written in matrix form as = . Zhang recommended the adoption of these approximating functions for all use of the DR-BEM. Using the approximating function f = 1 + r gives @x @X @f = @f @r = @f rx @x @r @x @r r (7.59) where is a vector of the nodal values of the forcing function b. Variable Coefficients The DR-BEM can be readily extended to equations with variable coefficients. The same idea of using Equation (7. A simpler approach is to choose an approximating function which does not lead to singularities for convective terms. Applying the DR-BEM gives ^ ^ . =  x.7. b b 2 x . where is a diagonal matrix containing the nodal values of  x.

Formulating the DR-BEM for a General Elliptic Problem In this section it has been shown how the DR-BEM can be applied for elliptic problems with varying forms of b.64) K .62) Consider a two-dimensional equation of the form @u + n x. ^ ^ .63) H .1 H + SK  u = Gq whereS = H U . GQ F (7. y = k x. The DR-BEM Using Other Operators The DR-BEM has been presented in this chapter based on the Laplace operator. GQ F  . n is a vector containing the nodal values of n. l and m respectively. The DR-BEM can be applied in cases where b involves a sum of terms due to the basic property Z b1 + b2  d = Z b1 d + Z b2 d (7. The problem with applying the DRBEM based on another operator is in choosing the approximating function f . y + m  x. This should improve the accuracy of the method.61) which is a boundary-only expression for the variable coefficient Helmholtz equation. R u = Gq + Sn where (7. y  r2 u x. However the DRBEM can be be applied using essentially any operator of appropriate order with known fundamental solution. A choice of f which produces accurate results is required but it is also necessary to choose an f for which a particular solution u ^ can be determined. This method is general and can easily be extended to accommodate variable coefficient derivative terms and a sum of variable coefficient terms. y @u @x @y Applying the DR-BEM to this equation gives a matrix system of the form (7. L and M are diagonal matrices where the diagonals contain the nodal values of k.132 D OMAIN I NTEGRALS IN THE BEM Using this matrix expression for b Equation (7. If an appropriate operator can be found the complexity of the forcing function b can be reduced.1 S = HU .59) can be rearranged to give ^ ^ . y u + l x.

66) . @F  @F R = S K + L @X + M @Y F .65) (7.1 (7.

v @u . vx . Partridge et al. vx @u . vx @x y @y @t @u . Zhu chose an approximating function of the form f = r m where m is a positive integer.68) which can be achieved using a variation of coefficients method. ku = 0 Dr2u . (1986).70) @u ^ . Although somewhat ad-hoc this approach was found to produce accurate results. vx @x y @y ^ = f (7.7. chose to define u ^ and use Equation (7. Along the lines of Wrobel et al. .69) where the material parameters D . vy and k are all assumed to be homogeneous. (1992) applied the DR-BEM to the transient convection diffusion equation (7. r2 u + 2 u = b x.71) to determine the corresponding approximating function.3 D OMAIN I NTEGRALS I NVOLVING THE D EPENDENT VARIABLE 133 Zhu (1993) has determined the particular solutions necessary for applying the DR-BEM based on the two-dimensional Helmholtz operator. ku = @u Dr2u . u. This analysis requires the determination of a particular solution u ^ which satisfies (7. ku Dr2u ^ . v @u ^ . Determining the particular solution u ^ requires solving the ordinary differential equation d2u ^ + 1 du ^ + 2u = rm 2 dr r dr @u . t (7.71) Instead of defining a form of the approximating function f and solving for u ^ Partridge et al. v y @x @y which has a known fundamental solution. y. They applied the DR-BEM based on the steady-state convection-diffusion operator (7.67) Radial functions have generally been used when applying the DR-BEM.

.

These two methods will be outlined in this section for the diffusion equation where the diffusivity  is a material parameter which can be a constant or a function of position. tm+1  = u x. 1984b) is to assume that.1. or a semi-analytic technique is used which can directly calculate a solution at a specified time. The most common approach (Brebbia et al. 1 @u x.Chapter 8 The BEM for Parabolic PDES 8.3) Using this finite difference approximation the original parabolic equation has been reduced to an elliptic equation.1) 8. for sufficiently small t. t r2 u x. In this section time-stepping procedures will be considered.  .2) 1 u x. Consider the variation between a time tm and a time tm+1 = tm + t. Using the weighted residuals method an integral equation can be generated from . The two most commonly used time-stepping methods are the coupled finite difference .Boundary Element Method This approach discretises the time-domain in a finite difference form.1 Coupled Finite Difference .  (8. tm+1 . the time derivative can be approximated using a first-order fully implicit finite difference scheme @u x.1 Time-Stepping Methods Several approaches have been proposed for applying the BEM to parabolic problems. tm  @t t which allows the diffusion equation in this time-range to be approximated as (8. Either some form of time-stepping procedure is used to advance the solution in time.BEM and the direct time integration method. u x. t =  @t (8. tm+1 + 1 u x. tm = 0 r2u x. These methods can be broadly classified into two main approaches. Time-stepping approaches discretise the time domain in some manner and use some form of time marching scheme to advance the solution from one discrete time to the next.  t t . tm+1  .

4) where um+1 = u  . If an internal solution is required at a specific time this can be determined explicitly from Equation (8. The time-stepping procedure can be repeated using the internal solution at t = t0 +t as pseudo-initial conditions for the next time-step. t = u x. Curran et al.6) the solution can be advanced in time.7) . They approximated the time variation within an interval as B Hum+1 .  . tm  . They found that this method will only produce accurate results if Equation (8. um+1 @! d.  t (8. and can be calculated once and stored. solution can be calculated at t = t0 + t. Unfortunately Equation (8. tm.5)  x where is a matrix containing the influence coefficients due to the domain integral. Using Equa0 is known from the initial conditions so a tion (8. Unfortunately it lead to a deterioration in convergence behaviour. Cross & Lewis (1980). Applying the BEM to Equation (8. c   um+1 + Z . This coupled finite difference . If a problem has time-dependent boundary conditions the solution system needs to be reformed at each time-step. This integral is generally evaluated by using a domain mesh (Brebbia et al.4) where the fundamental solution is applied at internal point  and c   = 1. It was implemented and investigated by Curran. 1984b). investigated the use of a higher-order approximation to the time-derivative.  u x.boundary element method (FD-BEM) was first proposed by Brebbia & Walker (1980) for the diffusion equation. A solution at internal nodes can then be calculated. tm+1 + 1 . Gqm+1 = Bum U (8. Tanaka.3). The domain integral does not include any problem unknowns so a fairly coarse domain mesh will generally suffice.  +  = 0 r2 ! x. = @n = u x. For a problem with time-independent boundary conditions at each time-step it is only necessary to update and solve the system for m+1 .4) contains a domain integral. m+1 = where The boundary conditions can be applied to form a solution system of the form m+1 is the vector of unknown nodal values at time tm+1 and is a vector constructed from known nodal values from the previous time-step. This will generally require small time-steps to be adopted. Matsimoto & Yang (1994) proposed a generalised version of this time-stepping scheme. tm+1  and um modified Helmholtz equation x The fundamental solution ! is a solution of the applied at some source point  . If a constant time-step is used the matrices . Z qm+1 !d. They found that this improved the accuracy of the method.6) HG B Ax x u x.136 T HE BEM FOR PARABOLIC PDES Equation (8. + 1 Z um! d t (8.2) accurately approximates the time derivative. The fundamental solution of the modified Helmholtz equation is known in both two and three dimensions.  (8.4) gives 1 ! x.

1. The variation of u and q with time is unknown so it is still necessary to step in time. t d dt = 0 r u x. Similarly to the FD-BEM. termed the time-scheme parameter.11) This time dependent fundamental solution is known in two and three dimensions. They found that the best results were achieved using a Crank-Nicolson scheme.  F @t 2  (8. An integral equation can be derived from Equation (8. t0 ! d t0 . The weighted residual statement using this approach is ZtF Z  t0 1 @u x. implemented this method and found it gave accurate results for a range of diffusion problems. tF . They tested the 1 ). t + @! . x. t0 . t . Substituting this approximation and a first-order finite difference approximation of the time derivative into the diffusion equation gives If = 1 this approximation of the diffusion equation is equivalent to the standard FD-BEM discussed earlier. Physically this fundamental solution represents the effect at a field point at time t of a unit point source applied at a point  at time tF .@t x (8. accurate results can be achieved using larger time-steps than with the FD-BEM. These two time-stepping approaches are discussed in detail in Brebbia et al. is a constant in the range 0  1.8).10) with c   = 1.2 Direct Time-Integration Method Instead of converting the original parabolic equation to an elliptic equation the problem can be treated directly in the time domain by directly integrating over both time and space.8. Two different time-stepping schemes can be used. Tanaka et al.u r2 um+1 + 1 . However. t .9) Integrating in time once and in space twice gives ZtF Z @! ZtF Z Z c   u . t ! .10) x. where the fundamental solution ! satisfies (8. as the time dependence is included in the fundamental solution. tF  +  u @n d. tF .8) 8. x. dt + u x. Alternatively the time integration process can be restarted at t0 with increasing numbers of intermediate steps used. a Galerkin scheme ( = 2 ) and a fully implicit accuracy for a Crank-Nicolson scheme ( = 2 3 scheme ( = 1). The first method requires a new domain integral to be calculated after each time-step due to .1 T IME -S TEPPING M ETHODS 137 where .  r2um = u  t m+1 m (8. dt =  q! d. If an internal solution is required at a specific time this can be determined from Equation (8. t +  tF  = 0 r2! . . each time-step can be treated as a new problem so that an internal solution is constructed at the end of each time-step to be used as pseudo-initial conditions for the next time-step. (1984b).

This inversion procedure requires solutions to be generated for several values of the transform parameter . This requires the procedure to be restarted at some time where an internal solution is constructed and used as pseudo-initial conditions to repeat the process. H G 8. The parabolic PDE is thus converted to an elliptic problem for which the boundary element method has been shown to generally produce accurate results. This includes many practical cases such as constant initial temperature or an initial linear temperature profile. The Laplace transform of u  bolised as U  . u0   is the initial conditions of u. u x r2 U x. As mentioned by Brebbia et al. Therefore.138 T HE BEM FOR PARABOLIC PDES the updated pseudo-initial conditions. in practice.  = Z1 0 e. Unfortunately. This method was first proposed by Rizzo & Shippy (1970) and has since been successfully 1 U x. Once the solution is determined in Laplace transform space this solution can be inverted to give a solution in the time-domain. in many practical cases the domain integral can be avoided.t u x.14) is an elliptic PDE which can be readily solved using the boundary element method. ideally.  and is defined as 1 @u x.  =  0 (8. (1984b). however. in practice it is not always feasible to restart the integration process at t0 . Equation (8.12) x x.  . At each time-step new and matrices are required so if many time-steps are required the storage capacity of the computer is likely to be exceeded. will still require the user to create a domain mesh. This. t dt Applying Laplace transforms to Equation (8. Consider the diffusion equation with appropriate boundary and initial conditions. t will be sym(8.13) U x. If the initial conditions are u0 = 0 throughout the body the domain integral equals zero. Once the solution to the elliptic problem is determined in the transform space a solution in the original space can be attained using an inverse transform procedure.12) gives with transformed boundary conditions. a domain integral only needs to be calculated once. If the initial conditions satisfy Laplace’s equation r2 u0 = 0 then a Galerkin vector can be found and the domain integral can be expressed as equivalent boundary integrals.14) x . t =  @t (8. The second time-stepping procedure involves only a domain integral at t0 so.2 Laplace Transform Method An alternative approach which avoids time-stepping is to solve the problem in a transform domain which removes the time dependence of the problem. The most appropriate transform approach for parabolic problems is the Laplace transform. t r2u x. both time-stepping methods are likely to require domain integration.

giving a matrix system of the form . (1994) proposed using the DR-BEM to convert this domain integral term to equivalent boundary integrals.14) this means that the DR-BEM will be used to convert the right-hand-side to equivalent domain integrals. One limitation of the Laplace transform method is that Equation (8. They noted that the direct method is simpler to apply. Satravaha & Lu 1994). They chose to apply the DR-BEM based on the known fundamental solution to the Laplace operator. 1984b). recently Zhu et al.15) The DR-BEM can now be applied to Equation (8. Traditionally this domain integral has been calculated by using a domain discretisation (Brebbia et al.14) is inhomogeneous so that applying the standard BEM will generate a domain integral involving the initial conditions. due to its greater efficiency. u x = N fj 0  j =1 j (8. However. Therefore the required DR-BEM approximation is +L X 1 U x.3 T HE DR-BEM F OR T RANSIENT P ROBLEMS 139 used by other practitioners (Moridis & Reddell 1991) (Zhu.14).8. Considering Equation (8. Liggett & Liu (1979) compared the Laplace transform method with the time-dependent Green’s function method. However. they recommended the Laplace transform method for solving diffusion problems.  .

u r2 u x.17) in which case a matrix expression of the form . t + b x. Once the solution is determined for this elliptic equation in the transform space a solution at a given time can be constructed using an inversion process. H.16) which can be reduced to a square system by applying boundary conditions. t =  @t (8. S 1 Su u .  0 (8. This Laplace transform dual reciprocity method (LT-DRM) can easily be extended to equations of the form 1 @u x. Gq = .

Zhu and his colleagues have successfully extended the LT-DRM to solve diffusion problems with nonlinear source terms. 8.19) . the diffusion equation 1 @u r2 u =  @t (8. Gq = .  0 (8.3 The DR-BEM For Transient Problems The DR-BEM can also be applied to parabolic problems. for example.  1 Su u . S H . Consider.R.18) is generated.

1 .24) (8.39) becomes (8.20) 1 H U ^ .23) u = 1 . q  qm + q qm+1 (8.140 T HE BEM FOR PARABOLIC PDES where the thermal diffusivity. Partridge & Brebbia (1990) recommended using a first-order finite difference approximation to the time derivative and linear approximations to u and q within a time-step.1 @u Hu . um @t t (8. GQ (8. u um + uum+1 q = 1 .4 The MRM for Transient Problems The MRM can be applied to the diffusion equation r2 u of Laplace’s equation. 1 and the time-step is between times tm and tm+1 = tm + t. Gq =  @t where (8.25) where u and q are weighting parameters with values in the range 0. q = 1:0. 1 t C + uH  c . u H um+1 .1 C = . Using this two-level time-integration scheme the most common choice of time-scheme parameters is u = 0:5. In this case the global approximation of b implies a separation of variables such that M @u = X @t j=1 fj x j t Using Equation (8.22) can be solved using a standard direct time-integration method. .22) Equation (8. q Gqm+1 =  t CH G h i um + 1 . 1 @u using the fundamental solution = @t 1 @u and the recurrence In this case the forcing function becomes b0 =  @t . ^ .21) or C @u @t + Hu = Gq 1 H u ^ F . @u = um+1 . GQ ^ F . Equation (7. If a constant time-step is used the matrices . q  Gqm (8. 8. Substituting these approximations into Equation (8. and only need to be constructed once.26) The values of u0 and q 0 are known from the initial conditions so a time-stepping procedure can be used.22) an expression at tm+1 can be derived in terms of values at tm . is a constant.20).

If a constant time-step is used the matrices L .30) (8.24) becomes ju uj+1 = r2 uj = @ @tj (8.32) (8.33) HLum+1 . This approach is termed a first-order approach as it removes all but the H H G G .  um + um+1 q = 1 . R . This allows Equation (8. @! c   u   + d . (8. A second order approach can be formulated by using quadratic interpolation of u and q within the time-range. 1 0 R 0 L tm 1 0 R tm tm 1 .28) The standard BEM numerical procedure can be applied to this boundary integral equation. This requires some interpolation between the two time-levels marked by m and m + 1.8. GLqm+1 = . 1 tm G1 + 1 . This gives the matrix system fundamental solutions. The most common approach is to solve this system numerically by discretising the time domain and using a time-stepping procedure. Using Equation (8.29) to be simplified to (8. q q_ = tm and all the other derivatives vanish.31) m+1 um u _ = u t. In this case the MRM formulation becomes 1 Z 1 Z ju X X @ @ j q d.4 T HE MRM FOR T RANSIENT P ROBLEMS 141 relationship defined by Equation (7.  H . This equation can be solved using a time-integration procedure. H = . L and R only need to be constructed once outside the time-stepping loop.  G0 . If the boundary conditions are not time-dependent the boundary conditions only need to be applied 1 H + 1 .29) where the matrices H1 . This most common approach is to use a linear approximation to u and q in this time-range H 0 u + H1 u _ + H2u + : : : = G0q + G1q_ + G2q + : : : (8.HRum + GRqm where (8. = ! @n @tj @tj j =0 . G = HL = 1 H + H . Differentiating these linear approximations gives (8.34) first derivatives. m m +1 m q . G1 etc are the influence coefficient matrices relating to the higher-order u = 1 .34) the solution can be advanced in time.  qm + qm+1 where has a value in the range 0 to 1.27) The higher-order fundamental solutions are known for Laplace’s equation. G = 1 G + G . j =0 .

142 T HE BEM FOR PARABOLIC PDES once. .

(1981). Cross. J. S. A. Applied Mathematical Modelling 4. Shaw. C. J. Reible. A boundary element formulation of problems in linear isotropic elasticity with body forces. (1987). M. Curran. A. 105–122. & Rizzo. A. Gipson. Computational Mechanics Publications. & Nowak. S. Y. L. (1987). (1984a). J. B. Comput. D. W. Gipson. J. Newnes– Butterworths. Periaux & R. Boundary Element Techniques In Engineering. L. D. & Wrobel. S. & Lewis. L. in R.. Brebbia. in C.. 31. A.. A. pp. SpringerVerlag. ’A method for the numerical solution of boundary value problems governed by second-order elliptic systems’. P. Springer-Verlag. New York. J. Computational Mechanics Publications. Brebbia. J. Boundary element techniques: Theory and applications in engineering. D. (1984b). J. pp. N. ’Solution of parabolic differential equations by the boundary element method using discretisation in time’. L. C.. & Wu. C. 398–400. D. (1980). Danson. eds. F. Coleman. Springer-Verlag. ’Boundary Element Methods’. Rizzo. ’Proceedings of the Fifth International Symposium on Numerical Methods in Engineering’. A. 96–112. C. (1989). S. G. (1979). (1991).Basic Concepts And Recent Developments In The Poisson Equation. Wendland & . J. Computational Mechanics Publications and Springer-Verlag.. ’An effective boundary element method for inhomogeneous partial differential equations’. A. F. Clements. Brebbia. S. & Phan-Thien. A new approach for transforming domain integrals to the boundary. & Walker. Gruber. ed. Brebbia.Bibliography Brebbia. Telles.. (1978). Boundary Element Techniques. in C. Boundary elements and perturbation theory for certain classes of hyperbolic and parabolic problems. ’On the numerical solution of two-dimensional problems by an improved boundary integral equation method’. F. Shippy. & Wrobel. Telles. D. C. Phys. C. Tullock. D. A. L. (1980). 197–202. A. G. C. A. C. G.. 73–84. C. Boundary Element Fundamentals .. & Savant. Fairweather. D. F. 730–745. Brebbia. S. Journal of Applied Mathematics and Physics (ZAMP) 42. Journal of the Institute of Mathematics Applications 22.

29–19. D. A. A. C. Partridge. Engineering Analysis With Boundary Elements 11(1). ’A perturbation boundary element code for steady state groundwater flow in heterogeneous aquifers’. Nowak. & Cruse. 861–866. L. ’Topics In Boundary Elements Research’. pp. & Cheng. L. T. G. (1979). Q. ’Boundary Elements IX Vol 3’. P. pp. G. & Liu. Brebbia. A. (1987). 1079–1084. L. Water Resources Research 23(6). D. Ortner. & Beskos. & Rangogni. Zielke. Berlin and New York. Lafe. Polyzos. E. Computational Mechanics Publications and Springer-Verlag. (1993). G. Nowak. H. (1987). Solution of transient heat conduction problems using boundary-only formulation. eds. (1995). (1994). O.a review’. eds. W. pp. BE Communications 5(6). C.. ed. (1990). A. Holz. Monaco. . (1987). S. C. ’Unsteady flow in confined aquifers: A comparison of two boundary integral methods’. pp. Gray. eds. (1982). Meissner. 265–276. Dassios. R. ’The theory of radial basis functions applied to the BEM for inhomogeneous partial differential equations’. in ’Computational Methods In Water Resources VIII’.144 BIBLIOGRAPHY G.-D. D. The Laplace transform boundary element (LTBE) method for the solution of diffusion-type problems. Springer-Verlag. in K. A.-F. M. ’The numerical evaluation of particular solutions in the BEM . Brebbia. & Chen. Moridis. Liggett. A. U. C. A. ’Finite Elements In Water Resources IV’. P. (1993). 57–61. Golberg. A. P. 115–127. E. W. Kuhn. ’Generation of higher order fundamental solutions to the twodimensional modified Helmholtz equation’. J. in ’Boundary Elements XIII’. (1994). Partridge. A. A. A.. W. 285–288. in C. in C. ’Boundary Elements IX’. (1991). Computational Mechanics Publications and SpringerVerlag. Golberg. BE Communications 5(2). Computational Mechanics Publications and Elsevier Applied Science. D. Huang. M. 155–160. C. & Partridge. & Reddell. & Brebbia. Computational Mechanics Publications and Springer-Verlag. Boundary element method: Processing of the source term of the Poisson equation by means of boundary integrals only. ’On the equivalence of dual reciprocity and particular integral approaches in the BEM’. Engineering Analysis With Boundary Elements 10. 87–90. J. & Brebbia. A. 19. Computational Mechanics Publications and Springer-Verlag. Brebbia. P. Brebbia.36. 68–73. L. ’Some remarks on particular solution used in BEM formulation’. & Wrobel. The BEM dual reciprocity method for diffusion problems. N. A. A. Water Resources Research 15(4). C. P. 397–403. pp. Computational Mechanics 13. J. (1992). Construction of fundamental solutions. Itagaki. (1992). W. M.. Springer-Verlag. ’Comparison of the dual reciprocity and the multiple reciprocity methods’. Kuhn. 83–97. The Dual Reciprocity Boundary Element Method. Boundary Element Communications . Wendland & G. Pinder & W. L. W.

R.. ’A boundary element approach for nonhomogeneous potential problems’. Master’s thesis. J. Engineering Analysis With Boundary Elements 13. & Lu. C. Solution of variable coefficients PDEs by means of BEM and perturbation technique. Zheng. (1970). F.. 279–288. Wu. M. On the dual reciprocity boundary element method. Brebbia & B. pp. Applied Mathematical Modelling 18. AIAA Journal 8(11). (1986). Matsimoto. Coleman. in ’Boundary Elements XIII’. C. R. in C. & Phan-Thien. Tanaka. Applied Mathematical Modelling 10. ’Particular solutions associated with the Helmholtz operator used in DRBEM’. (1994). (1994). (1991). Boundary element methods and inhomogeneous parabolic equations. eds. in ’Finite Elements In Water Resources VI’. Brebbia. Computational Mechanics Publications and Springer-Verlag. ’Numerical solution of the generalized Laplace equation by coupling the boundary element method and the perturbation method’. Q. Wrobel. (1986). 569–576. BE Abstracts 4(6). & Yang. (1991). 266–270. Rangogni. J. Zhu. ’Solving linear diffusion equations with the dual reciprocity method in Laplace space’. A. Computational Mechanics 7. Computational Mechanics Publications and Springer-Verlag. C.BIBLIOGRAPHY 145 Rangogni. 231–233. (1993). ’A method of solution for certain problems of heat conduction’. J. (1985). (1993).. L. D. 1–10. Zhang. Satravaha. R. N. J. & Shippy. S. Zhu. J. Y. T. Noye. ’Time-stepping boundary element method applied to 2-D transient heat conduction problems’. X. The dual reciprocity boundary element formulation for transient heat conduction. Rizzo. C. . F. Springer-Verlag.. P. S. D. A. 2004–2009. The University of Wollongong. 19–30. & Nardini. ’BETECH 85’.

.

14–16 Truss elements. 81–83 Potential energy. 82 Trefftz method. 117 diffusion equation. 18 Spherical polar. 12 Basis functions. 53 Hermite. 10–14 bicubic. 29. 10 Lagrange. 17 Prolate spheroidal. 75. 79 Boundary conditions application of. 31 Galerkin vector. 119 Gaussian quadrature. 64 Triangular elements. 12. 91 Laplace. 26 Fundamental solution. 54 Coupled finite difference . 124. 106 Multiple reciprocity method. 77 Regular BEM. 72. 124 Poisson equation. 133 element stiffness matrix. 46 Navier. 10 bilinear. 135 Curvilinear coordinate systems. 139 variable coefficients. 138 Mass lumping. 39–42. 131 Dual reciprocity BEM. 24 Isoparametric formulation. 122 Perturbation BEM. 7–9 linear. 25.Index Advection-diffusion equation. 43. 24 Weighting function. 45–48. 82 Rayleigh-Ritz method. 76 Weighted residual. 72 Helmholtz. 56 Global stiffness matrix. 14 cubic. 24 .boundary element method. 2–4 quadratic. 137 Dual reciprocity BEM approximating function. 129 transient problems. 17–19 Cylindrical polar. 53 Laplace transform method. 72 wave equation. 14 Basis functions Hermite cubic. 64 Strain energy. 122–124 diffusion equation. 43–45 Direct time-integration method. 140 Helmholtz equation. 121 Plane stress elements. 27 integration by parts. 72 Kelvin. 103 Area Coordinates. 127 derivative terms. 18 Dirac-Delta function. 2–16. 129 elliptic problems. 72 Galerkin formulation. 7 Beam elements.

Sign up to vote on this title
UsefulNot useful