Ph.D. Thesis of Farzana Hussain

Efficient Integration Schemes and
Algorithms for Finite Element

Domain Integrals - An
Application to Computational
Engineering Science
Farzana Hussain
Registration No: 6012009, Session: 2009-10,
Department of Mathematics
Shahjalal University of Science and Technology, Sylhet,
Bangladesh.
A thesis submitted for the degree of
Doctor of Philosophy
January 2014
Efficient Integration Schemes and Algorithms for Finite Element

Domain Integrals An Application to Computational Engineering
Science
Submitted by
FARZANA HUSSAIN
Registration No: 6012009, Session: 2009-10
A Dissertation Submitted for the Degree of
DOCTOR OF PHILOSOPHY
Dr. Md. Shajedul Karim

Supervisor and Professor
Department of Mathematics
SUST, Sylhet
(Signature)
External member
(Signature)
External member
(Signature)
Dedicated
to
my loving parents
and
my kids
Declaration
I hereby declare that the matter embodied in this thesis is the result
of investigation carried out by me in the Department of Mathematics,
Shahjalal University of Science and Technology, Sylhet, Bangladesh
under the supervision of Dr. Md. Shajedul Karim, Professor, Department of Mathematics, Shahjalal University of Science and Technology, Sylhet, Bangladesh. This thesis has not been submitted for
the Award of any type of Degree, Diploma, Associateship, Fellowship
etc. of any other University or Institute.
Farzana Hussain
Certified
Dr. Md. Shajedul Karim

Supervisor and Professor,
Department of Mathematics,
Shahjalal University of Science and Technology,
Sylhet - 3114, Bangladesh.
Acknowledgements
It gives me immense pleasure to express my deep sense of respect and
gratitude to my guide Dr. Md. Shajedul Karim, Professor of
Mathematics, Shahjalal University of Science and Technology, Sylhet
for his enthusiastic guidance, encouragement and all kind of technical
support. He has always given me suggestions, instructions, inspirations in my academic progress as well as during the course of my
Ph.D. work.
I am also thankful to the SPM, Entity of the sub-project Postgraduate
Research Enhancement Schemes in Mathematics and Statistics under HEQEP, Higher Educational Quality Enhancement Project, UGC,
Bangladesh for the much needed financial support during the research
work.
I am grateful to Dr. Md. Rashed Talukder, Professor of Department of Mathematics, Shahjalal University of Science and Technology,
Sylhet for helping me in preparation of this thesis in LATEX. I am
deeply thankful to all of my colleagues for their help, kind cooperation and encouragement in my academic progress. Thanks to all the
office staffs of the Department of Mathematics, SUST. I would like to
thank The University Authority also for their collaboration and
cooperation.
My loving mother Mrs. Farida Parvin, My husband Mr. Icramul
Jalil, son Shadwan Sabah Jalil and my daughter Nujhat Tanjim
Jalil had always been my source of strength and inspiration to accomplish my Ph.D. work. I express my heartiest thanks to all of my family
members and friends for their great contribution during my research
period.
Farzana Hussain
Abstract
This thesis is concerned with accurate numerical schemes to evaluate
the integrals of rational functions of a n-th degree bivariate or trivariate polynomial numerator with linear or nonlinear denominator over a
triangular or tetrahedral domain in the local parametric space in two
dimensions or three dimensions respectively. These integrals encountered in finite element application to higher order partial differential
equations governing diverse physical phenomena employing the most
popular isoparametric, subparametric and superparametric finite elements. To derive a suitable, highly accurate technique to evaluate
such integrals numerically, this thesis studied on different approaches
to establish the new formulae.
Many realistic problems involving multiple integration frequently occur in a large number of engineering applications in Computer aided
design (CAD), Computer aided Engineering (CAE), Computer aided
Manufacturing (CAM), in geometric modelling as well as in robotics.
Particularly for irregular, complicated or distorted type of domain
such integrals cannot be evaluated analytically or explicitly and numerical integration schemes are the only alternative. Finite element
method (FEM) gained more importance due to its flexibility and versatility to obtain the solution of many complicated problems which
cannot be evaluated accurately by any other methods. In practical finite element analysis, complex geometrical domains are discretized by
different types of curvilinear isoparametric finite elements, which are
advantageously mapped into n-simplex (a square or a unit triangle)
or n-hyper (a cube or an orthogonal tetrahedron) in a local parametric space. If the domain consist a polygonal boundary, elements with
straight sides, usually triangle, quadrilateral, tetrahedron, cube are
perfectly satisfactory. But if some part of the boundary is curved,

however, elements with at least one curved side or face are desirable.
When dealing with curved or irregular boundaries, the side of the element should closely approximate the boundary which requires many
elements in order to reduce the error in the solution. Many researchers
strongly suggested that a large number of lower order elements should
be used in the portion of the domain where the function varies rapidly
to get the desired accuracy, which generally requires much more computational effort, computing memory and time. Alternatively, a fewer
number of higher order elements are suitable to achieve better accuracy
with less computational effort. Hence, this thesis considered general
types of elements:
1. triangle with three straight sides or triangular element with one
curved and two straight sides,
2. tetrahedron with four plane surfaces or tetrahedral element with
one curved and three plane surfaces.
Using shape functions and mathematical transformation the curved
elements are mapped to straight sided elements. The versatility of
these elements enhanced in this thesis by improved numerical integration schemes.
Among various numerical integration schemes, Gaussian quadrature,
which can evaluate the (2n-1)th order polynomials with n Gaussian
integration points, is mostly used in view of the accuracy and efficiency of calculation. However, the integrands in practical situations
are either polynomials or rational expressions. To get the desired accuracy for the polynomial of higher degree it is necessary to increase
the Gauss points with corresponding weights. On the other hand, if
the integrands are rational functions of bivariate or tri-variate polynomial numerator with linear or nonlinear denominator then more Gauss
points are needed. It is well known that the first step of the Gaussian
quadrature rule is to evaluate the integrands at all Gauss points which
require more computing time before the application of the weighted

sum. In addition, specially for two and three dimensional cases the
higher order (more Gauss points and weights) formulae become erroneous due to crowding of Gauss points and that effects the accuracy
and efficiency of the calculations. Hence, it is an important task to
make a proper balance between accuracy and efficiency of calculations.
Therefore, the main aim of this Thesis is to derive numerical integration formulae that employ less number of Gaussian points and weights
in the calculation process avoiding the said crowding provided the
accuracy and efficiency is assured. Finally, as demanded, it presents
higher order Gaussian quadrature formulae to evaluate the integrals of
(i) polynomial functions, (ii) non-polynomial functions and (iii) rational functions of n-th degree bivariate or trivariate polynomial numerator with linear/nonlinear denominator over triangular or tetrahedral
domain in two or three dimensions respectively. Further, for such
derivations it utilizes the property of definite integrals and uses shape
functions for geometric transformations.
Very few explicit formulae are developed so far for triangular domain
integrals which are not sufficient for satisfactory degree of accuracy and
for some element geometry these formulae are not at all reliable. This
thesis at first derived Gaussian quadrature formula for accurate evaluation of the triangular domain integrals using the Gaussian quadrature formula for square domain integrals as they are highly developed
from lower order to higher order. It is expected that these developed
techniques will be the best alternatives to test the accuracy of higher
order (Gaussian) quadrature formulae developed later in this thesis.
Through application examples it is demonstrated that the available
numerical integration methods for triangular surfaces becomes uncertain and not at all reliable for some integrals of rational functions. But
the developed integration formulae are always consistent, reliable and
provides considerable savings in computational effort, in comparison
with existing numerical methods that employs Gaussian quadrature
procedures.
It is astounding to note here that the Thesis is fully concerned with the
systematic gradual development of quadrature formulae to evaluate
the domain integrals and its contribution is directed in the area of
FEM and other numerical methods. Thus:
Chapter 1 describes the objectives and scope of the Thesis, important
applications and relevant literature review.
Chapter 2 contains some basic concepts and a brief historical background of FEM, elements in FEM, Interpolation functions. Some problems and some mathematical preliminaries are also discussed.
Chapter 3 is concerned with an accurate numerical integration scheme
to compute the triangular domain integrals. At first, the two dimensional domain of integration is divided into finite number of triangular
elements. Using mathematical transformation based on shape functions, each of the arbitrary triangle is transformed into a unit triangle
{(u, v)| 0 u, v 1, 0 u + v 1}, then the unit triangle is
subdivided into three quadrilaterals. Through isoparametric transformation each quadrilateral is again transformed into its equivalent
2-square given by {(, )| 1 , 1}. Then Gaussian quadrature
formula is derived for accurate evaluation of the triangular domain integrals employing the Gaussian quadrature formula for square domain
integrals.
In Chapter 4, each triangular element of the original domain is transformed into a standard triangular element {(s, t)| 1 s, t 1, s +
t 0} which is then subdivided into three quadrilaterals. Through
isoparametric transformation each quadrilateral is transformed into
its equivalent 2-square {(, )| 1 1, 1 1}. Then
quadrature formulae are obtained for accurate evaluation of the triangular domain integrals employing the Gaussian quadrature formulae
for square domain integrals. These developed techniques of Chapter
3 and 4 will be the best testing tools for the accuracy of higher order
(Gaussian) quadrature formulae developed in Chapter 5, 6, 7.
Chapter 5, 6, 7 concerned with different approaches for accurate nu-
merical integration schemes to compute two dimensional integrals. In

these chapters numerical schemes are derived to evaluate the triangular domain integrals with higher order precision employing the lesser
number of quadrature points and weights. At first, the arbitrary two
dimensional domain is divided into finite number of triangles. Using
step by step mathematical transformation, each of these triangles is
transformed into equivalent 2-square. Then schemes are presented to
evaluate the triangular domain integrals ensuring higher order precision with significant reduction of total Gauss points and weights. The
derived new formulae are also free of crowding of Gaussian points. Different types of elements are used for the search of an accurate method.
Chapter 5 deals with unit triangle, Chapter 6 concerned with standard triangle and Chapter 7 deals with a triangle which is symmetric
about the line of symmetry. All these studies are done to obtain a
highly accurate, efficient numerical scheme to evaluate the triangular
domain integrals. Comparison between the results of Chapter 5, 6,
7 and the existing methods are also shown for clear understanding.
Quadrature schemes so derived up to this stage are different and the
accuracy of one scheme is tested by the other schemes. Almost all
the schemes are found efficient in view of accuracy and efficiency of
calculations.
Chapter 8 is devoted with the development of efficient method to
evaluate the three dimensional integrals for tetrahedral element with
reduced number of Gaussian points. At first, the three dimensional
domain of integration is divided in to finite number of tetrahedron.
Using mathematical transformation depending on shape functions,
each tetrahedron is transformed into its equivalent unit tetrahedron
given by {(u, v, w)|0 u, v, w 1, 0 u + v + w 1} whose
base triangle is an unit triangle in w = 0 plane. The unit tetrahedral element is then transferred into its equivalent 2-cubic element
{(, , )| 1 , , 1} through isoparametric transformations
using shape functions. Then two sets of Gaussian quadrature formulae are derived for the evaluation of tetrahedral domain Integrals. The
first set of quadrature formulae involves a huge crowding of Gaussian

points and weights within the domain of integration whereas the second set of formulae are completely free of such crowdings.
Chapter 9 presents efficient method to evaluate the three dimensional
integrals using tetrahedral element {(s, t, r)|1 s, t, r 1, s+t+r
1}, whose base triangle is a standard triangle in r = 1 plane. Here,
also two sets of quadrature formulae are developed of which the first set
of quadrature formulae involves a huge crowding of Gaussian points
and weights within the domain of integration and the second set of
formulae are completely free of such crowdings. Comparison between
the obtained results of Chapter 8, 9 and by the present methods are
also shown to verify the efficiency and accuracy. All the quadrature
schemes so presented in Chapters 8 and 9 for tetrahedral domain
integrals are completely different. Through regorous investigation it
is observed that their accuracy and efficiency are almost same.
The chapter 10, devoted to focus on some applications of the derived
formulae on different fields.
Finally, Chapter 11 focus on the main conclusions of the present research work. Computer codes (in FORTRAN) based on the presented
integration schemes are presented in Appendix B.
Contents
Contents
xi
List of Figures
1 Introduction
1.1 Objectives and Scope .
1.2 Important Applications
1.3 Literature Review . . .
1.4 Plan of The Thesis . .
xvi
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
2 Some Preliminaries
2.1 Finite Element Method Preliminaries . . . . . . . . . . . . .
2.1.1 Some Basics of the Finite Element Method . . . . . .
2.1.2 History of the Finite Element Method . . . . . . . .
2.1.3 Weighted Residual Method . . . . . . . . . . . . . . .
2.1.4 Finite Element Formulation of Problems . . . . . . .
2.1.5 Elements in FEM . . . . . . . . . . . . . . . . . . . .
2.1.6 Shape Functions (Interpolation Functions) . . . . . .
2.1.7 Isoparametric Elements . . . . . . . . . . . . . . . . .
2.1.8 Some Practical Problems . . . . . . . . . . . . . . . .
2.2 Mathematical Preliminaries . . . . . . . . . . . . . . . . . .
2.2.1 Properties of Multivariate Polynomial of n-th Degree
2.2.2 Integration of Monomials . . . . . . . . . . . . . . . .
2.2.3 Gauss Quadrature Formula . . . . . . . . . . . . . .
2.3 Finite Element Formulation of Physical Problems . . . . . .
xi
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
2
3
7
8
12
.
.
.
.
.
.
.
.
.
.
.
.
.
.
16
16
16
19
20
21
22
23
24
25
28
28
29
30
30
CONTENTS
2.3.1
2.3.2
2.3.3
Torsion Problem . . . . . . . . . . . . . . . . . . . . . . .
Elasticity Problem . . . . . . . . . . . . . . . . . . . . . .
Heat Conduction Problem . . . . . . . . . . . . . . . . . .
3 Evaluation of Triangular Domain Integrals by use

Quadrature for Square Domain Integrals
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . .
3.2 General Description and Formulation of the Problem
3.3 Numerical Evaluation Procedure . . . . . . . . . . . .
3.3.1 Gaussian Quadrature for Triangle (GQT) . . .
3.3.2 Integration Over Unit Triangle (IOUT) . . . .
3.4 Application Examples . . . . . . . . . . . . . . . . .
3.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . .
4 An Accurate Evaluation Procedure of Triangular
grals
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . .
4.2 Formulation of the Problem . . . . . . . . . . . . .
4.3 Integration Over Standard Triangle (IOST) . . . . .
4.4 Application Examples . . . . . . . . . . . . . . . .
4.5 Conclusions . . . . . . . . . . . . . . . . . . . . . .
31
33
37
of Gaussian
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
41
41
43
46
46
46
51
52
Domain Inte.
.
.
.
.
55
55
57
59
61
62
.
.
.
.
.
.
.
.
65
66
68
70
71
73
76
80
83
6 Accurate Evaluation Schemes for Triangular Domain Integrals

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
99
99
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
5 Appropriate Gaussian Quadrature Formulae for Triangles

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2 General Description and Formulation of the Problem . . . .
5.3 Transformation to Square Domain . . . . . . . . . . . . . . .
5.3.1 Gauss Quadrature for Unit Triangle (GQUTS) . . . .
5.3.2 Gauss Quadrature for Unit Triangle (GQUTM) . . .
5.4 Application Examples . . . . . . . . . . . . . . . . . . . . .
5.4.1 Two Dimensional Elasticity Problem . . . . . . . . .
5.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . .
xii
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
CONTENTS
6.2
6.3
6.4
6.5
Problem Statement . . . . . . . . . . . . . . . . .
Numerical Evaluation Procedures . . . . . . . . .
6.3.1 Gauss Quadrature using Standard Triangle
6.3.2 Gauss Quadrature using Standard Triangle
Application Examples . . . . . . . . . . . . . . .
6.4.1 Two Dimensional Elasticity Problem . . .
Conclusions . . . . . . . . . . . . . . . . . . . . .
. . . . . . .
. . . . . . .
(GQSTS) .
(GQSTM)
. . . . . . .
. . . . . . .
. . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
102
103
104
106
110
112
115
7 A Symmetrical Extended Gaussian Quadrature Formula for Evaluation of Triangular Domain Integrals
128
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
7.2 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . 130
7.3 Integration Over Arbitrary Triangle (IOAT) . . . . . . . . . . . . 131
7.4 Transformation of Triangular Domain . . . . . . . . . . . . . . . . 134
7.4.1 Symmetric Gauss Quadrature for Triangle (SGQTS) . . . 138
7.4.2 Symmetric Gauss Quadrature for Triangle (SGQTM) . . . 140
7.5 Application Examples . . . . . . . . . . . . . . . . . . . . . . . . 143
7.6 Error Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
7.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
8 Efficient Gauss Quadrature Rules for Tetrahedral Domain Integrals
170
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
8.2 Problem Statement over Tetrahedral Domain . . . . . . . . . . . . 172
8.3 Integration Over Arbitrary Tetrahedron . . . . . . . . . . . . . . . 173
8.4 Integration over Orthogonal Tetrahedron and Quadrature Formulae
(GQTTS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
8.5 Efficient Gaussian Quadrature formula for Tetrahedron (GQTTM) 181
8.6 Application Examples . . . . . . . . . . . . . . . . . . . . . . . . 186
8.7 Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . 191
8.8 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
9 Accurate Gaussian Quadrature Rules for Numerical Integration
over Tetrahedron
202
xiii
CONTENTS
9.1
9.2
9.3
9.4
9.5
9.6
9.7
9.8
Introduction . . . . . . . . . . . . . . . . . . . . . . . . .
Problem Statement over Tetrahedral Domain . . . . . . .
Integration over Arbitrary Tetrahedron . . . . . . . . . .
Integration Over Standard Tetrahedron . . . . . . . . . .
Numerical Evaluation Procedures . . . . . . . . . . . . .
9.5.1 Gaussian Quadrature for Tetrahedron (GQTSS) .
9.5.2 Gaussian Quadrature for Tetrahedron (GQTSM)
Application Examples, Results and Discussions . . . . .
Error Analysis . . . . . . . . . . . . . . . . . . . . . . . .
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
203
204
205
207
211
211
213
217
221
224
10 Appropriate Gauss Quadrature and Recursive Schemes to Obtain

Element Matrices for General Triangular Finite Elements
241
10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
10.2 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . 243
10.3 Recursive Scheme to Compute the Integrals for Calculation of Stiffness Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
10.3.1 Jacobian is a Linear Function . . . . . . . . . . . . . . . . 244
10.3.2 Jacobian is a Nonlinear Function . . . . . . . . . . . . . . 245
10.4 Example-1 (Torsion of the Elliptic/circular Cross Sections) . . . . 247
10.4.1 Quadratic Curved Triangular Element (NP = 6) . . . . . . 247
10.4.2 For Cubic Curved Triangular Element (NP = 10) . . . . . 248
10.4.3 Calculation Procedure . . . . . . . . . . . . . . . . . . . . 249
10.4.4 Computed Results . . . . . . . . . . . . . . . . . . . . . . 250
10.5 Example-2 (Torsion of the Trapezoidal Cross Sections) . . . . . . 251
10.6 Example-3 (Torsion of the Part of an Elliptical Cross Section) . . 252
10.7 Example-4: Integration (A Lunar Model) . . . . . . . . . . . . . . 253
10.8 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
11 Conclusions
266
Appendix A
269
Appendix B
283
xiv
CONTENTS
References
332
List of Publications
341
xv
List of Figures
1.1
1.2
Finite Element Discretization of an Aircraft. . . . . . . . . . . . .

Finite Element Discretization of a Bridge. . . . . . . . . . . . . .
2.1
2.2
2.3
2.4
2.5
2.6
2.7
Connection between elements of Discretized domain.

Discretized domain in 2D . . . . . . . . . . . . . . . .
A single 2D element. . . . . . . . . . . . . . . . . . .
Discretized domain in 3D. . . . . . . . . . . . . . . .
A single complex 3D element. . . . . . . . . . . . . .
Thin Body in Plane Stress. . . . . . . . . . . . . . . .
Thick Body in Plane Strain. . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
17
22
22
23
23
34
34
3.1
3.2
3.3
3.4
3.5
3.6
Triangulation of the domain of integration. . . .

Original and Transformed Unit Triangle. . . . .
Unit Triangle Divided into three Quadrilaterals.
Transformation of Quadrilateral into Square. . .
Distribution of 3 7 7 points using IOUT. . .
Distribution of 3 10 10 points using IOUT. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
44
45
47
48
51
52
4.1
4.2
4.3
Original and Transformed Standard Triangle. . . . . . . . . . . .

Standard Triangle Divided into three Quadrilaterals. . . . . . . .
Distribution of 3 8 8 points using IOST. . . . . . . . . . . . .
58
60
62
5.1
5.2
5.3
5.4
Original and Transformed Unit Triangle. . . . . . . . .

Distribution of 10 10 Gauss Points using GQUTS. . .
Distribution of 15 15 Gauss Points using GQUTS. . .
Gauss Points for m =n =10 (54 points) using GQUTM.
69
73
74
77
xvi
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
5
LIST OF FIGURES
5.5
5.6
Gauss Points for m =n =15 (119 points) using GQUTM. . . . . .

Transformation of Quadratic Curved Triangular element A0 into
Unit triangle A. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
83
6.1
6.2
6.3
6.4
6.5
Transformation of Arbitrary triangle into Standard triangle. . .

Distribution of 10 10 Gauss Points using GQSTS. . . . . . . .
Distribution of Gauss Points for m=6, n=8 (30 points) GQSTM.
Distribution of Gauss Points for m=n=10 (54 points) GQSTM.
Mapping of Cubic Curved element A0 to Standard triangle S. . .
7.1
7.2
7.3
7.4
7.5
7.6
7.7
7.8
7.9
7.10
7.11
7.12
7.13
7.14
7.15
7.16
7.17
7.18
7.19
7.20
Discretization of the Domain of Integration into Triangles. . . . . 131

Arbitrary Triangle Subdivided into Quadrilaterals . . . . . . . . . 133
Distribution of 3 9 9 points using IOAT. . . . . . . . . . . . . 135
Original and Transformed triangle. . . . . . . . . . . . . . . . . . 136
Gauss Points for m = n = 10 using SGQTS (100 points, formula 2). 140
Gauss Points for m = n = 15 using SGQTS (225 points, formula 1). 141
Gaussian points for m = 10 (54-points, formula 1) using SGQTM. 143
Gaussian points for m = 16 (135-points, formula 2) using SGQTM. 144
Absolute Error in Polynomial of order 15. . . . . . . . . . . . . . 163
Percentage Error in I1 . . . . . . . . . . . . . . . . . . . . . . . . . 165
Percentage error in Example - 1 (r = 2). . . . . . . . . . . . . . . 167
Percentage error in Example - 2 (r = 4). . . . . . . . . . . . . . . 167
Percentage error in Example - 3. . . . . . . . . . . . . . . . . . . . 168
Percentage error in Example - 4. . . . . . . . . . . . . . . . . . . . 168
8.1
8.2
8.3
The Arbitrary Tetrahedron and the Transformed Unit Tetrahedron. 173

Distribution of Gauss Points for GQTTS (m =5, 125 points). . . . 181
Distribution of Gauss Points for GQTTS (m =10, 1000 points). . 182
xvii
.
.
.
.
.
78
102
107
109
110
113
LIST OF FIGURES
8.4
8.5
8.6
8.7
Distribution of Gauss Points for GQTTM,(m =5, 75 points). .

Distribution of Gauss Points for GQTTM,(m =10, 350 points).
Calculation of Total Gauss Points for k =3 (28 points). . . . .
Cubic tetrahedron V 0 with one curved surface. . . . . . . . . .
.
.
.
.
185
186
187
190
9.1
9.2
205
9.3
9.4
9.5
9.6
9.7
9.8
9.9
9.10
9.11
9.12
9.13
9.14
9.15
9.16
Discretization of a Three Dimensional Domain. . . . . . . . . . .

The Arbitrary Tetrahedron and The Transformed Standard Tetrahedron. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Distribution of Gauss Points for GQTSS, m =15 (3375 points). . .
Distribution of Gauss Points for GQTSM, m =15 (950 points). . .
Percentage Error in I1 . . . . . . . . . . . . . . . . . . . . . . . . .
Percentage error in I52,1,0 . . . . . . . . . . . . . . . . . . . . . . . .
2,2,2
Error Precision in I1,1,1
. . . . . . . . . . . . . . . . . . . . . . . .
0,2,2
Error Precision in I1,1,1 . . . . . . . . . . . . . . . . . . . . . . . .
2,2,2
Error Precision in I1,0,1
. . . . . . . . . . . . . . . . . . . . . . . .
2,2,0
Error Precision in I1,1,0 . . . . . . . . . . . . . . . . . . . . . . . .
206
214
217
234
234
235
235
236
236
237
237
238
238
239
239
10.1
10.2
10.3
10.4
10.5
10.6
Curved Domain of Example-1 . . . . . . . . . . . . . . . . .

Quadratic and Cubic Curved Elements . . . . . . . . . . . .
Discretization of the Trapezoidal Cross Sections . . . . . . .
Discretization of Part of an Elliptical Cross Section . . . . .
Gaussian Points within the Lunar Domain, SGQTS, m = 10
Gaussian Points within the Lunar Domain, SGQTM, m = 10
248
249
252
253
256
257
xviii
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Chapter 1
Introduction
Chapter 1
Introduction
For numerical solution of continuum mechanics problem using finite element method
(FEM) an extra and important task is involved in the calculation of element matrices. Among all the numerical rules, Gaussian quadrature rule occupies a central
role for such calculations. Complication arise from two main sources, firstly the
large number of integrations that need to be performed and secondly in methods
which use isoparametric or subparametric elements, the presence of the determinant of Jacobian matrix in the denominator of the stiffness matrix for which the
integrands are rational functions. Among all the recently developed numerical
integration schemes, the technique based on symbolic manipulation can reduce
the computation time of the usual numerical integration and give adequate integration accuracy by increasing integration points. It is highly expected that the
expressions for the exact values of the integrals have to be evaluated with care
and hence the higher order numerical integration techniques are the best choice.
Explicit formulae are available by which a rational integral of monomial numerators with linear denominators can be evaluated. But when the denominator is
bilinear or expression of higher order then Gaussian quadrature formula is the
only alternative for the evaluation of these integrals.
Although, such integrals of rational functions occurs in the calculation of stiffness matrix of higher order partial differential equation employing curved triangular or tetrahedral elements, very few explicit formulae are developed so far.
Many researchers had also shown that the most accurate 7 - points and 13 points Gaussian quadrature formulae for the triangular domain integrals are not
2
1. Introduction
sufficient for satisfactory degree of accuracy and for some element geometry these
formulae are not at all reliable. Similar complication arises for the high order
transformation method as in this case the denominators are linear, bilinear or of
higher order. This thesis originated with the development of the integration formulae to compute the corresponding domain integrals encountered in the arena
of computational engineering science and to present fast algorithm to compute
element matrices in order to solve physical problems.
This chapter includes the objectives and scope of this thesis, important applications, relevant literature survey and plan of the Thesis.
1.1
Objectives and Scope
The procedure of structural analysis, which afterward came to be known as the

Finite Element Method (FEM), was formerly developed as an extension of standard structural analysis procedures and was intended for application to the design
of advanced airplane structures. Thus the motivation of the early development
was the solution of practical engineering problems, and the implication of the
method at first was recognized only in this limited context. However, others soon
recognized the resourcefulness of the method and its underlying rich mathematical basis for application in nonstructural areas. Hence, as its range of application
was being extended rapidly in the early 1960s, it became apparent that the finite element method is essentially a special discretization procedure which can be
employed in the solution of a wide range of field problems, and thus it has much
greater importance than just as a tool for structural analysis.
Basically, extension of the mathematical basis to nonlinear and nonstructural
problems was achieved through the method of weighted residuals, originally conceived by Galerkin [1915] in the early 20th century. Since then, Finite Element
research efforts have followed two rather distinct paths. On the one hand, engineers interested in the design and analysis of complex structural systems have continued to develop different and more efficient elements for idealizing a wide range
of actual structural systems; they also have improved the assembly and solution
techniques employed in the analysis so as to increase the capability and reduce
the cost of practical usage. At the same time mathematically oriented researchers
3
1. Introduction
Figure 1.1: Finite Element Discretization of an Aircraft.

of the finite Element Method began making detailed studies of convergence, error analysis, alternative formulation procedures and other factors pertinent to a
complete understanding of the capabilities and limitations of the method in the
solution of field problems.
Both of these research avenues continue to offer opportunities for important
improvements in the use of the Finite Element Method. Consequently, tremendous
advances have been made in last 50 years both on the mathematical foundations
and generalization of the method to solve field problems in various areas of engineering analysis. It is appropriate at this stage of development that detailed
studies are made of the mathematical basis of the method so that it may have a
rigorous foundation. Looking back over the development of the Finite Element
Method, it is astounding to note its dramatic increase in popularity and in the
range of its applications during the past 50 years: it is now considered to be the
standard analytical technique for nearly all areas of structural design and a powerful tool for the approximate solution of differential equations governing diverse
physical phenomena, as discussed in the work of Clough and Peterson [1976]
1. Introduction
and Zienkiewicz and Morgan [1983].
Figure 1.2: Finite Element Discretization of a Bridge.

In numerical solution of practical problems using FEM, stiffness matrix calculation is one of the most important task. Among all the numerical integration
rules, Gaussian quadrature rule occupies a central role for such calculations. Complications arise from two main sources, firstly the large number of integrations that
need to be performed and secondly, the presence of the determinant of the stiffness matrix for which the integrands are rational functions in methods which use
isoparametric or subparametric elements. Yagawa and Yashimara [1990] have
outlined that the usual Gauss quadrature cannot evaluate exactly such rational
integrals related to linear quadrilateral elements as it can exactly evaluate the
(2n-1)th order polynomials with n Gaussian integration points. They have developed a numerical integration scheme based on symbolic manipulation, which
can reduce the computation time of the usual numerical integration and give adequate integration accuracy by increasing quadrature points. Still an attention
is required to select the order of the integration rule since it is not yet totally
worked out. Since it is expected that the expressions for the accurate values of
5
1. Introduction
the integrals have to be evaluated with care and hence the accurate integration
techniques are the best choice. For this reason most emphasis has been placed on
the four node isoparametric planar quadrilateral elements and exact integration
formulae are given by Babu and Pinder [1984]; Griffths [1994]; Griffths and
Mustoe [1995]; Hacker and Schreyer [1989]; Okabe [1981]; Rathod [1988];
Rathod and Islam [1998]; Rathod and Karim [2002]; Videla [1996].
Although integrals of rational functions occurs in the calculation of stiffness
matrix for the second or higher order partial differential equations employing
curved elements as well as in axisymmetric case with linear triangular or tetrahedral elements, such development has not been made so far. On the other hand
for the numerical integration method of integrating an arbitrary function of two
variables over the surface of a triangle very few integrating coefficients have been
published. According to the Gaussian quadrature formulae given by Cowper
[1973] the 13-point rule is the most accurate method. But Lague and Baldur
[1977] practically investigated that the method is not sufficient for a satisfactory
degree of accuracy and for some element geometry it is not at all reliable. Karim
[2001]; Rathod and Karim [2002] presented integration formulae to compute
the integrals of rational functions of a n-th degree bivariate polynomial numerator
with linear denominator over a unit triangle in a local two dimensional parametric
space. They also presented methods based on analytical integration schemes for
the calculation of stiffness matrix for the linear quadrilateral elements.
Triple integral formulae have always been of massive curiosity in all sides of
applied science and engineering applications. An idea about methods for evaluating volume integrals has given by Lee and Requicha [1982]. A hypothetical
advancement to the evaluation of volume integrals by transforming the volume
integral to a surface integral over the boundary of the domain of integration
was discussed by Timmer and Stern [1980]. Bernardini [1991]; Cattani and
Paoluzzi [1990a,b]; Lien and Kajiya [1984]; Zienkiewicz and Taylor [2000]
have presented formulae and algorithms for computing integrals of polynomials
over tetrahedron. Rathod et al. [2004, 2005] proposed formulae based on GaussLegendre quadrature rule for the numerical integration of an arbitrary function
over the standard tetrahedron. It is clearly noticed that they used Gaussian
quadrature schemes applicable for square or cubes to evaluate triangular or tetra6
1. Introduction
hedral domain integrals respectively and that exhibits, enforces severe crowd of
Gaussian points involved in the calculation process. These higher order (more
Gauss points and weights) formulae become erroneous due to the crowding of
Gaussian points and that effects the accuracy and efficiency of the calculations.
The main objective of the present Thesis is to develop efficient numerical integration schemes using less number of Gaussian points and weights for the desired
accuracy of both triangular and tetrahedral domain integrals of different types of
functions, e.g. (i) polynomial functions, (ii) non-polynomial functions and (iii)
rational functions of bivariate or trivariate polynomial numerator and a linear
or nonlinear denominator. These integrals encountered in many areas of science
and engineering. Finally, by algebraic manipulation and geometric consideration
the crowding of Gaussian points is totally removed from the scheme. Hence with
considerable savings in computational effort, numerical evaluation of the said domain integrals as well as the element matrices will be possible without resorting
to uncertain and expensive integration schemes. Further the Thesis enhances its
objectives by presenting computer codes (FORTRAN) for such higher order elements compatible with the element formulations. The Thesis will be (hopefully)
directed to contribute on this important topic and related basic developments in
the area of Finite Element Method and other numerical methods employed in
computational engineering.
1.2
Important Applications
Since the advent of freely available powerful computers in the mid-1960s, the finite element method has been a standard technique for the numerical analysis of
a wide range of continuum mechanics problems. In FEM solution procedure, the
high degree of accuracy which can be achieved with the higher order elements
means that a small number of such elements can often be used to obtain practically adequate solutions. Unfortunately, the simple shapes of elements restricts
severely their application in the analysis of practical problems, where often quite
complex geometrical boundaries have to be modeled. This restriction would be
removed if we could map a simple element in the local coordinate system, into
a more complex shape in the global coordinate system. Once a particular form
7
1. Introduction
of mapping is adopted and the coordinates are chosen for every element so that
these map into contiguous spaces, then shape functions written in the local space
can be used to represent the function variation over the element in the global
space without upsetting the inter element continuity requirements. This has led
to the development of elements of more arbitrary shape and are called isoparametric elements. The isoparametric elements are widely used for the analysis of
continuum mechanics problems of complex shapes involving curved boundaries.
Complication that arise for these elements is the presence of the determinant of the
Jacobian matrix in the denominator of the stiffness matrix, i.e., each component
of the stiffness matrix is an integral of rational function of bivariate or trivariate
polynomial numerator with a denominator, which is an expression of more than
one local variables. In the evaluation of these integrals normally a high order
and relatively too expensive numerical integration scheme is required. In axisymmetric case some integrands are rational functions of product of shape functions
in the numerator with nonlinear denominator in local variables. Such integrals
could not evaluated analytically and efficient numerical integration schemes are
demanded.
This Thesis mainly concentrates on the development of efficient numerical
integration formulae to compute the domain integrals encountered in many areas
computational engineering. Presented computer codes in this Thesis are straight
forward to evaluate such integrals.
1.3
Literature Review
A brief discussion to the literature concerned with this work is very necessary and
essential to build up a proper frame work for the research undertaken in this study.
Particularly in the finite element applications to problems involving an enclosed
region in R2 or R3 , if the original domain has a polygonal boundary, elements with
straight sides usually triangles, quadrilaterals, tetrahedrons, cubes are perfectly
satisfactory. If some part of the boundary of the region is curved, however elements
with at least one curved side are desirable, Mcleod and Mitchel [1972]. The two
principal methods of dealing with curved boundaries in the finite element method
are:
8
1. Introduction
The direct method: In this case basis functions are constructed to match
the curved boundaries and the integrations carried out directly in the original plane. This method originally developed with some success by Mcleod
and Mitchel [1975]; Wachspress [1972, 1973] for triangular elements. It
is shown that the basis functions are no longer polynomials and so the numerical work is necessarily more involved.
The use of isoparametric transformations: This is the most popular
method where each element involving part of the curved boundary in the
global (x, y, z) space is mapped into a standard straight sided element in
local (u, v, w) space by a formula of the same form as that used for the
interpolation on the standard element. Isoparametric elements were first introduced by Ergatoudis et al. [1968] and later mathematically analyzed by
Ciarlet and Raviart [1972]; Strang and Fix [1973]; Zlamal [1974]. According to Mcleod and Mitchel [1972, 1975] the only element, which can
deal adequately with most plane problems involving curved boundaries and
interface is the triangular element with two straight sides and one curved
side. For this element the isoparametric transformation enables the integration to be carried out in the normalized triangles.
An alternative method to rational basis function method and the isoparametric
method is given by Mcleod [1978]. This method, called a high order transformation method, allows any degree of approximation like the rational basis function
method, while retaining the simplicity of the isoparametric method. High order
basis functions, as well as Isoparametric functions, have been in the search for exact analytical expressions of integrals of rational function of bivariate polynomial
numerator with linear denominator over a unit triangle in local space.
Lague and Baldur [1977]; Rathod [1988] have shown that such integrals
are also encountered in axisymmetric finite element solution employing triangular
elements in a local coordinate system. In their study Lague and Baldur [1977]
clearly shown that the Gauss quadrature formulae developed by Cowper [1973]
are not satisfactory in respect of integration accuracy. Consequently they have
developed a technique based on a transformation of the triangular surface into a
square for which the Gaussian quadrature formulae have been extensively ana9
1. Introduction
lyzed and are readily available. It is also demonstrated that for some element the
proposed technique is very inefficient to obtain the sufficient accuracy and little
assurance is guaranteed if any divergence is evident. These integrals were considered by Andersen and Mcleod [1979]; Baart and Mcleod [1983]; Mcleod
[1978] extending the work of Mcleod and Mitchel [1972]. They employed recursive and reduction algorithms to derive analytical, closed-form formulae. They
also considered in detail the propagation of errors in the associated recursive formulae. The expressions derived by them ultimately depend upon the evaluation
of several integrals, which in turn require a recursive procedure in obtaining the
final result. Therefore the evaluation procedure is very lengthy even for one final
integral. In their study only one type of integrals with rational integrands had
been discussed.
The approach of Rathod and Karim [2000] differs from that of Andersen
and Mcleod [1979]; Baart and Mcleod [1983]; Mcleod [1978]; Mcleod and
Mitchel [1972] in the sense that they have identified four distinct types of rational integrands depending on the position of intermediate node(s) along the
curved side. Accordingly integration formulae they have derived and their application have shown in the calculation of the components of element stiffness
matrices for the second order partial differential equation employing the curved
triangular elements and in axisymmetric case with linear triangular elements. A
method for exact integration of the polynomials in the convex and concave circular sectors is provided in Silva and Mote [1988]. Because of the circular sector
these composite boundary elements have got limited applications in modelling
arbitrary curved domains. Several Authors considered a convex linear quadrilateral element and presented integration formulae. Babu and Pinder [1984];
Griffths [1994]; Griffths and Mustoe [1995]; Hacker and Schreyer [1989];
Okabe [1981]; Rathod [1988]; Videla [1996] have successfully attempted to derive the explicit finite element relations for general convex linear quadrilateral
finite elements. A good overview of numerical integration schemes which employ
Gaussian quadrature is given by Yagawa and Yashimara [1990]. They have
developed a numerical integration scheme based on symbolic manipulation, which
can dramatically reduce the computing time of the usual numerical integration
and give adequate integration accuracy by increasing integration points. They
10
1. Introduction
have shown that the evaluation of the coefficient of the bivariate polynomial in
numerator with linear denominator requires relatively large amount of time in
the total estimation of the stiffness matrix. The computation of these coefficient has to be improved in order to reduce the total computation time. Karim
[2001]; Rathod and Karim [2002] have presented methods for computing the
components of element stiffness matrix in an efficient way employing the exact
integration formulae encountered for planar quadrilateral and curved triangular
elements. Karim [2001] mainly contributed with special attention on the fournode, six-node and ten-node triangular elements with two straight sides and one
curved side.
Triple integral formulae have always been of great interest in all sides of applied science and engineering applications. Lee and Requicha [1982] have given a
good inkling of existing methods for evaluating volume integrals. A hypothetical
advancement to the evaluation of volume integrals by transforming the volume
integral to a surface integral over the boundary of the integration domain was discussed by Timmer and Stern [1980]. Bernardini [1991]; Cattani and Paoluzzi
[1990a,b]; Lien and Kajiya [1984] has presented formulae and algorithms for computing integrals of polynomials over tetrahedron. Rathod et al. [2004] derived
explicit integration formulae for computing volume integrals of trivariate polynomials over an arbitrary tetrahedron in Euclidean space. They evaluated this
volume integral as a sum of four integrals over the unit triangle. In Zienkiewicz
and Taylor [2000] a numerical integration formulae for tetrahedral regions with
a degree of precision up to 3 are listed. Rathod et al. [2005] proposed product
formulae based on Gauss-Legendre quadrature rule for the numerical integration
of an arbitrary function over the standard tetrahedron.
The serious drawbacks of such higher order (more Gauss points and weights)
formulae for two and three dimensional problems developed so far is that, they
become erroneous due to crowding of Gaussian points and weights in the calculation process that effects the accuracy and efficiency of the calculations and also
require much computing time and memory. The existing formulae for triangular
domain integrals are not sufficient for satisfactory degree of accuracy and for some
element geometry these formulae are not at all reliable.
11
1. Introduction
At first, in this thesis Gaussian quadrature formula is derived for accurate evaluation of the triangular domain integrals using the Gaussian quadrature formula
for square domain integrals as they are widely developed (from lower order to
higher order). Secondly, the developed technique is used as standard testing tools
for the accuracy of higher order (Gaussian) quadrature formulae so developed in
this thesis. Curved boundary can be accurately modeled by curved finite elements.
Because fewer elements are required, the effort needed to obtain a solution is usually reduced. This Thesis also presents methods for computing the components of
element stiffness matrix in an efficient way employing the numerical integration
scheme using triangular elements for two dimensional and tetrahedral elements
for three dimensional problems. This thesis considered general types of elements:
1. triangle with three straight sides or triangular element with two straight
sides and one curved side,
2. tetrahedron with four plane surfaces or tetrahedral element with one curved
and three plane surfaces.
The versatility of these elements enhanced in this thesis by presenting efficient
numerical integration schemes.
1.4
Plan of The Thesis
To achieve the objectives of this Thesis the plan of work is as follows. The work
in the thesis is spread over the following chapters:
Chapter 1 gives the objectives and scope of the Thesis, important applications
and relevant literature review.
Chapter 2 contains some basic concepts and a brief historical background of
FEM., elements in FEM, Interpolation functions, some problems are also discussed
and some mathematical preliminaries are described.
Chapter 3 concerned with accurate numerical integration schemes to compute
triangular domain integrals employing the Gaussian quadrature for square domain
integrals. At first, the arbitrary domain of integration is divided into finite number
of triangular elements. The arbitrary triangle is transformed into a unit triangle
12
1. Introduction
{(u, v)| 0 u, v 1, 0 u + v 1}, then the unit triangle is subdivided into
three quadrilaterals. Through isoparametric transformation each quadrilateral is
again transformed into its equivalent 2-square given by {(, )| 1 , 1}.
Then Gaussian quadrature formula for the triangular domain integrals is derived
employing the Gaussian quadrature formula for square domain integrals.
In Chapter 4, each triangular element of the original domain is transformed
into a standard triangular element {(s, t)| 1 s, t 1, s + t 0}, which
is then subdivided into three quadrilaterals. Each quadrilateral is transformed
into its equivalent 2-square {(, )| 1 1, 1 1}. Then quadrature
formula is obtained for accurate evaluation of the triangular domain integrals using
the Gaussian quadrature formula for square domain integrals. These developed
techniques of Chapter 3 and 4 are used as standard testing tools for the accuracy
of higher order (Gaussian) quadrature formulae developed later in this thesis.
Chapter 5, 6, 7 concerned with accurate numerical integration schemes to
compute two dimensional domain integrals of arbitrary functions. In these chapters numerical schemes are presented to evaluate the integrals with higher order
precision with lesser number of Gauss points in the calculation process. Different
types of elements are used for the search of an accurate method. Chapter 5 deals
with unit triangle, Chapter 6 concerned with standard triangle and Chapter 7
deals with a triangle which is symmetric about the line of symmetry. All these
studies are done to obtain a highly accurate, efficient numerical scheme to evaluate the triangular domain integrals. Comparison between the results of Chapter
5, 6, 7 and by the present methods are also shown.
Chapter 8 is devoted with development of efficient methods to evaluate the
tetrahedral domain integrals in three dimension with reduced number of Gaussian
points. At first, the three dimensional domain of integration is divided in to
finite number of tetrahedron. Using mathematical transformation depending on
shape functions, each of the tetrahedron is transformed into its equivalent unit
tetrahedron {(u, v, w)|0 u, v, w 1, 0 u + v + w 1} whose base triangle
is an unit triangle in w = 0 plane. Using mathematical transformations the
unit tetrahedral element is then transferred into its equivalent 2-cubic element
{(, , )| 1 , , 1}. Finally, two sets of Gaussian quadrature formulae
are derived for the evaluation of tetrahedral domain Integrals. The first set of
13
1. Introduction
quadrature formulae involves a huge crowding of Gaussian points and weights
within the domain of integration whereas the second set of formulae are completely
free of such crowding.
Chapter 9 devoted with development of efficient method with tetrahedral
elements {(s, t, r)| 1 s, t, r 1, s + t + r 1}, whose base triangle is a
standard triangle in r = 1 plane. Here also two sets of quadrature formulae are
developed of which the first set of quadrature formulae involves a huge crowding
of Gaussian points and weights within the domain of integration and the second
set of formulae are completely free of such crowding. All these studies are under
taken for obtaining a highly accurate, efficient numerical scheme to evaluate the
said triple integrals. Comparison between the obtained results of Chapter 8, 9
are also shown to verify the accuracy and efficiency of the presented schemes.
Chapter 10 is devoted to focus on some applications of the derived formulae
on some real world problem of Applied Science and Engineering.
Finally, Chapter 11 focus on the main conclusions of the present research
work. Computer codes based on the presented integration schemes are also appended.
The development of the numerical integration rules in this thesis is believed
to provide the way of accurate evaluation of integrals encountered in solving continuum mechanics problems with less computational effort.
14
Chapter 2
Some
Preliminaries
Chapter 2
Some Preliminaries
The primary intension of this chapter is to introduce the essentials of Finite Element Method, its adaptability, popularity and related information. The second
step is to bring in some well known theorems or formulae from a range of Mathematical origins - particularly from Algebra, Advanced Calculus and Numerical
Analysis those are essential to develop new and powerful techniques pertaining to
the problems investigated in this Thesis for ready reference.
2.1
Finite Element Method Preliminaries
In this section, brief discussion of the Finite Element Method, its history and
relevant definitions are narrated in a concise form.
2.1.1
Some Basics of the Finite Element Method
The basic idea in the finite element method (FEM) is to find the solution of a
complicated problem by replacing it by a simpler one. Since the actual problem
is replaced by a simpler one in finding the solution, we will be able to find an
approximate solution rather than the exact solution. The majority of practical
design problems fall outside the reach of closed form solutions due to the complex
and irregular form of structures, complexity of loading conditions, non-linearity
and in homogeneity in properties of materials. For this reason, there is a growing
interest in numerical methods for the solution of continuum mechanics problems.
16
2. Some Preliminaries
In such instances Finite difference (FD) Method were used extensively in the
1960s. In the 1970s Finite Element began to replace Finite Differences as the
dominant technique. Now a day the FEM is a powerful tool for the approximate
solution of differential equations governing diverse physical phenomena. Its use
in industry and research is extensive, and indeed without it many problems in
science and engineering would be incapable of solution.
Figure 2.1: Connection between elements of Discretized domain.

Finite element method requires a problem defined in a geometrical space (or
domain) to be subdivided into finite number of smaller region (an element). In
finite difference, the element or mesh consists of rows and columns of orthogonal
lines; in finite elements, each subdivision is unique and need not be orthogonal. For example, triangles or quadrilaterals can be used in two dimensions and
tetrahedrons or hexahedrons in three dimensions. Over each finite element, the
unknown variables are approximated using known functions. These functions can
be linear or higher order polynomial expansions that depend on the geometrical
locations (nodes) used to define the finite element shape. In contrast to finite
difference procedures, the governing equations in the finite element method are
integrated over each finite element and the solution summed (assembled) over
17
the entire problem domain. As a consequence of these operations, a set of finite
linear equations is obtained in terms of a set of unknown parameters over each
element. Solution of these equations is achieved using the techniques of linear
algebra.
The advantages of finite elements are:
1. It is easier to conform the physical geometry where as with finite differences
the geometry is usually adjusted to fit the grid spacing,
2. The boundary conditions are easier to apply for geometrical reasons,
3. It is easier to use variable grid spacing by using small elements in regions of
interest and large elements in regions of little interest,
4. It is easier to write universal programs which can be applied to any geometry and a large number of physically different situations without having to
change of the code for each individual case.
The primary disadvantages of finite element calculations are:
a. The time consuming data input of the grid although the grid generation programs may substantially shorten this process;
b. Long computing times, specially for complex problems and three-dimensional
problems;
c. The requirement, that the analyst must find an element size such that the
governing differential equation is satisfactorily approximated everywhere.
With the advent of computers and its use in FEM such disadvantages are
considerably reduced and the method with above mentioned advantages become
a powerful numerical procedure for solving physical problems governed by a differential equation or an energy theorems in the realm of science and engineering
analysis. This method has principal characteristics that distinguish it from other
numerical procedures:
1. The method utilizes an integral formulation to generate a system of algebraic
equations;
18
2. The method uses continuous piece wise smooth functions for approximating
the unknown quantity or quantities.
Actually, the second characteristic distinguishes the finite element method
from other numerical procedures that utilize also an integral formulation. The
implementation of the FEM can be subdivided into three broad steps:
1. Establishing the element interpolation properties;
2. Evaluating the element matrices; and
3. Solving an actual problem.
2.1.2
History of the Finite Element Method
The history of finite element method is particularly interesting, especially since

the method has only been in existence since the mid 1950s. Early work of numerical solution of boundary value problems can be traced to the use of finite
difference schemes; the use of such methods was discussed in the book Southwell
[1946]. The actual coining of the term finite element appeared in Clough [1960].
The beginnings of the finite element method actually stemmed from those early
numerical methods and the frustration associated with attempting to use finite
difference methods on more difficult, geometrically irregular problems, Roach
[1972].
Beginning in mid 1950s, efforts to solve continuum problems in elasticity using
small, discrete elements to describe the overall behavior of simple elastic bars
began to appear. Argyris [1954] was the first to publish on the use of such techniques for the aircraft industry. The early use of finite elements was restricted
to the application of such techniques for structural related problems. However,
others soon recognized the versatility of the method and its underlying rich mathematical basis for application in nonstructural areas. Zienkiewicz and Cheung
[1965] were among the first to apply the finite element method to field problems
involving solution of Laplace and Poisson equations. An early application of the
finite element technique to viscous fluid flow is given by Martin [1968]. Much of
the early work on nonlinear problems can be found in Oden [1972]. Efforts to
19
model heat transfer problems with complex boundaries are discussed by Huebner
[1975]; a comprehensive three-dimensional finite element model for heat conduction is described by Heuser [1975].
Recent and ongoing research in finite element method indicates a profuse literature describing the versatility and reach mathematical abstraction of the subject.
Excellent reviews and description of the method given in the texts by Bickford
[1990]; Desai [1979]; Finlayson [1972]; Fletcher [1984]; Hughes [1987]; Reddy
[1984]; Segerlind [1984]; Zienkiewicz and Taylor [1989]. A short monograph
of the development of the FEM is given by Owen and Hinton [1980]. Programming of the finite element method is described by Smith [1982] and a rigorous
mathematical discussions is given in text book by Johnson [1987].
2.1.3
Weighted Residual Method
The underlying mathematical basis of the finite element method first lies with the
classical Rayleigh-Ritz method and variational calculus procedures introduced by
Rayleigh [1877] and Ritz [1909]. These theories provided the reasons why the
finite element method worked well for the class of problems in which variational
statements could be obtained. However, as interest expanded in applying the finite
element method to more areas, especially in mechanics, classical theory could no
longer be used to describe such problems.
Extension of the mathematical basis to nonlinear and nonstructural problems
was achieved through the method of weighted residuals, originally conceived by
Galerkin [1915] in the early 20th century. The method of weighted residuals was
found to provide the ideal theoretical basis for a wide set of problems including
those to which the Rayleigh-Ritz method can be applied. Basically, the method
requires the governing differential equation to be multiplied by a set of predetermined weights and the resulting product to be integrated over the problem domain; this integral is then required to vanish. The number of weighting functions
equals the number of unknown coefficients in the approximate solution. There are
several choices for the weighting functions, and some of the more popular choices
are: (1) Collocation method, (2) Subdomain method, (3) Least square method
and (4) Galerkins method. A rigorous description of the method of weighted
20
residuals can be found in Finlayson [1972]. Technically, Galerkins method is a
particular case of the general weighted residual procedure where various types of
weights can be utilized; in the case of Galekins method, the weights are chosen
to be the same as functions used to define the unknown variables.
The underlying principle of the finite element method resides in the method
of weighted residuals. The two most commonly used procedures are the RayleighRitz and Galerkins methods. The Rayleigh-Ritz method is based on calculus of
variation; however the method is difficult to use on complicated equations. Since
the variational method has a major disadvantage: it is not applicable to any
differential equation containing a first derivative term. The Galerkins method
is simple to use and is guaranteed to yield a compatible approximation to the
governing differential equation. For the linear and elliptic equations, this method
yields results identical to the Rayleigh-Ritz approximations. Hence, Galerkins
method is the most preferable method to establish the approximations to the
governing equations.
2.1.4
Finite Element Formulation of Problems
In the realm of science and engineering many problems are amenable to solution
using two and three dimensional analysis. The best way to solve any such physical
problem governed by a differential equation is to obtain the analytical solution.
There are many situations, however, where the analytical solution is not at all
possible to obtain.
A numerical method can be used to obtain an approximate solution when an
analytic solution cannot be developed. All numerical solutions produce values at
discrete points for one set of the independent parameters. The calculated values
provide important information about the physical process even though they are
at discrete points for one set of the independent parameters. Hence, numerical
solutions, however, are now desirable with great importance than no solution at
all.
A primary advantage of the finite element method is the ease with which it
can be generalized to solve problems composed of several different materials and
having irregular boundaries. In fact it is in the problem domains that the strength
21
of the finite element method becomes evident. The concept of a mesh and the
choices for available element discretizations become significantly important.
Figure 2.2: Discretized domain in 2D .
Figure 2.3: A single 2D element.
For the solutions of the problems, the physical domain is subdivided into sub
regions or elements. Different type of shapes can be used to define the elements.
Specifically, various types of element shapes and node numbering are the special
feature of the finite element method among all other numerical methods. Since
the FEM is based on the use of unstructured grids i.e. calculations are performed
on element-by-element basis, we are free to place our elements where we wish to
connect them to other elements without regard for orthogonally or sequential
node numbering.
2.1.5
Elements in FEM
Finite elements can normally be classified into three groups according to their
polynomial order: simplex, complex and multiplex, Oden [1972]. In a simplex
element, the number of coefficients in the polynomial is equal to the problems
dimensional space, plus one, the two dimensional three node triangular element
is the simplex element utilize a polynomial that contains constant and linear
terms. Similarly, the three dimensional four node tetrahedral element is the simplex element utilize a polynomial that contains constant and linear terms. A
complex element utilizes a polynomial that contains constant and linear terms,
plus higher-order terms. While the shape may be identical to a simplex element,
the complex element has additional boundary nodes and can have internal nodes.
The multiplex elements also use polynomials with higher order terms; however,
22
the boundaries of the elements must be parallel to the co-ordinate axes. An example of a multiplex element is a rectangular or quadrilateral element, cubic or
cuboid element.
Figure 2.4: Discretized domain in 3D. Figure 2.5: A single complex 3D element.
Historically, the finite element method first employed triangular elements to
model structural problems in two dimension. The generation of the simple mesh
using linear triangular elements is relatively easy for simple geometries, and in
some instances may not require sophisticated mesh-generation techniques. Use
of quadratic elements would more accurately approximate the curved boundaries.
The quadrilateral element, a four-sided polygon is just as versatile as the triangle
in discretization a region. For three dimensional problems tetrahedron, cubic elements are very widely used. Many commercially available finite element codes for
structural analysis use a mixture of triangles and quadrilaterals; cubes and tetrahedrons for two and three dimensional problems respectively. A more extensive
discussion on accuracy and convergence properties of various elements are given
by Oden and Reddy [1976] and Carey and Oden [1983].
2.1.6
Shape Functions (Interpolation Functions)
In finite element analysis using the displacement model, one assumes the variation
of displacement within an element since the true variation of displacement is not
known. In general, in higher mathematics, it is necessary in many situations to
deal with functions whose analytical form is either totally unknown or else is of
23
such a nature that the function can not easily be subjected to such operations as
may be required. In either case, it is desirable to replace the given function by
another function, which can be more easily handled. This operation of replacing
or representing a given function by simpler one is known as interpolation in a
board sense. In finite element literature interpolating functions are referred to
as Shape Functions.
There are two types of interpolation functions: (1) Lagrange interpolation and
(2) Hermite interpolation. In the Lagrange interpolation, which is widely used in
practice, the assumed function takes on the same values as the given function
at specified points. In the Hermite type of function, the slopes of the function
also take the same value as the given function at specified points. Using the
Lagranges interpolation formula the Lagrange type shape function can be derived
more directly. An easy and systematic method of generating shape functions can
be achieved by simple products of appropriate polynomials in two or three coordinates. Now a days lower and higher shape functions for all finite elements are
available.
For any specified element the values of shape function Ni (say) gives a unit
value at the node i and zero value at the other nodes of the element. The variation
of field variable u (say) in the element can be described as the sum of all shape
functions each multiplied by the corresponding model displacement i.e.
u=
n
X
ui Ni
i=1
Hence, shape function plays the most important role in the finite element solution
procedure.
2.1.7
Isoparametric Elements
For the analysis of physical problems of complex shapes involving curved boundaries, simple elements are no longer sufficient. This has led to the development of
elements of more arbitrary shapes are called isoparametric elements. The concept
of isoparametric elements is based on the transformation of the parent element
into local or natural coordinate system to an arbitrary shape in the global coordinate system. A convenient way of expressing the transformation is to make use of
24
the shape functions of the rectilinear elements in their natural coordinate system
and the nodal values of the coordinates. Thus the global x, y, z coordinates of a
point in an element may be expressed as
x=
n
X
xi Ni0 ,
y=
i=1
n
X
yi Ni0 ,
z=
i=1
n
X
zi Ni0
i=1
Ni0
Where,
are the shape functions of the parent rectilinear element and xi , yi , zi
are nodal coordinate of the element. Thus, the shape functions Ni0 used in the
above transformation help us to define the geometry of the element in the global
coordinate system. If this shape functions Ni0 are the same as the shape functions
Ni used to represent the variation of displacement i.e.
u=
n
X
ui Ni
i=1
in the element, these elements are called isoparametric elements. In addition,

in cases where the geometry of the elements is defined by shape functions of order
higher than that for representing the variation of displacements, the elements are
called Superparametric. Similarly if more nodes are used to define displacement
compared to the nodes used to represent the geometry of the elements, then they
referred to as Subparametric elements. Such elements generally found to be more
often of use in practice.
2.1.8
Some Practical Problems
The general field equation

Dx
2
2
+
D
G + Q = 0
y
x2
y 2
(2.1)
has many important applications in the physical sciences, which are known as
two dimensional problems. A few of these are discussed in this section.
1. The first application area is the torsion of non-circular sections. With
Dx = Dy = g1 , G = 0, Q = 2, the field equation reduces to the torsion
problem
25
1 2 2
(
+
+ 2g) = 0
g x2 y 2
(2.2)
Where g is the shear modules of the material and is the angle of twist.
2. Several fluid mechanics problems are embedded within the field equation.
The streamline and potential formulations for an ideal irrotational fluid are
respectively governed by
2 2
+
=0
x2 y 2
(2.3)
2 2
+ 2 =0
x2
y
(2.4)
3. The seepage of water under a dram or retaining wall and within a confined
aquifer is governed by
2
2
Dx 2 + Dy 2 = 0
x
y
(2.5)
Where Dx , Dy are the permeability of the earth material and represents

the piezometric head.
4. The water level around a well during the pumping process is governed by
Dx
2
2
+
D
+Q=0
y
x2
y 2
(2.6)
Where Q is the point sink term, Dx , Dy are the permeability and represents
the piezometric head and aquifer is assumed to be confined.
5. The heat transfer from a two-dimensional fin to the surrounding fluid by
convection is governed by
Dx
2T
2T
2h
2hTf
+
D
T+
=0
y
2
2
x
y
t
t
26
(2.7)
the coefficients Dx , Dy represents thermal conductivities in the x and y directions respectively, h is the convection coefficients, t is the thickness of
the fin, Tf is the ambient temperature of the surrounding fluid and T is the
temperature of the fin. If the fin is thin and the heat loss from the edges is
neglected.
When the body is very long in the z-direction and the temperature is a
function of only the x- and y- co-ordinate directions the heat transfer is
governed by
Dx
2T
2T
+
D
=0
y
x2
y 2
(2.8)
6. Seiche motion, which describes the standing waves on the bounded shallow
body of water, is governed by
h
2 w 4 2
2w
+
h
+
w=0
x2
y 2
gT 2
(2.9)
Where h is the water depth at the quiescent state, w is the wave height above
the quiescent level, g is the gravitational constant, and T is the period of
oscillation.
7. A fluid vibrating within closed volume is governed by
2 p 2 p w2
+
+ 2p=0
x2 y 2
c
(2.10)
Where p is the pressure excess above ambient pressure, w is the wave frequency, and C is the wave velocity in the medium.
Above different physical problems are contained within the general differential equation. Finite element formulation of these equation is given in detail by
Segerlind [1984].
27
2.2
Mathematical Preliminaries
This section briefly introduces some of the mathematical results, well known theorems and formulae needed to develop the new and efficient techniques, which
highlight the lemmas and theorems pertaining to the problems investigated in the
thesis for ready reference.
2.2.1
Properties of Multivariate Polynomial of n-th Degree
A bivariate polynomial g(x, y) of n-th degree is written as

g(x, y) =
n X
m
X
fmk, k xmk y k
m=0 k=0
A trivariate polynomial g(x, y, z) of n-th degree is written as

g(x, y, z) =
k
n X
m X
X
fmk, kl, l xmk y kl z l
m=0 k=0 l=0
Property 1. For each m 1, if we define

Hm (x, y) =
m
X
fmk,k xmk y k
k=0
then Hm (x, y) is a homogeneous function in (x, y) of degree m and g(x, y)

contains one constant and n homogeneous function in (x, y) of degree 1, 2,
..., n.
Property 2. In the bivariate polynomial F (x, y) there are exactly
terms.
(n+1)(n+2)
2
the proof of the results follows Karim [2001]; Rathod and Karim [2000]. Similar
properties exists for trivariate polynomials also.
28
2.2.2
Integration of Monomials
(a) The integral of a monomial of the type f (p, q) = pm q n (where m, n are nonnegative integers) over a unit triangle T = {(0, 0), (1, 0), (0, 1)} in the local
p, q coordinate system is denoted by ITm,n and given by
ITm,n
1p
pm q n dq dp =
=
p=0
q=0
m! n!
(m + n + 2)!
(b) The integral of a monomial of the type f (p, q) = pm q n (where m, n are nonnegative integers) over a 2-square S = {(1, 1), (1, 1), (1, 1), (1, 1)} in
the local p, q coordinate system is denoted by ISm,n and given by
ISm,n
p=1
0 if p or q or both are odd

4
otherwise.
(m+1)(n+1)
pm q n dq dp =
=
q=1
(c) The integral of a monomial of the type f (p, q, r) = pm q n rl (where m, n, l are

non-negative integers) over a 2-cube V = {(p, q, r)| 1 p, q, r 1)} is
denoted by IVm,n,r is given by
1
IVm,n,r
=
p=1
q=1
pm q n rl dr dq dp
r=1
0 if al least one of p, q, r are odd

8
otherwise.
(m+1)(n+1)(r+1)
(d) The integral of a monomial of the type f (p, q, r) = pm q n rl (where m, n, l are

non-negative integers) over a Unit Tetrahedron T R = {(p, q, r)|0 p, q, r
1; p + q + r 1)} in the local p, q, r coordinate system is denoted by ITm,n,l
R
and given by
ITm,n,l
R
1p
1pq
=
p=0
q=0
pm q n rl dr dq dp =
r=0
29
m! n! l!
(m + n + l + 3)!
2.2.3
Gauss Quadrature Formula
The m-points Gauss Quadrature formula for numerical integration of a function

f over interval I is of the form
Z
f dx =
m
X
wim f (im )
i=1
I
(m)
(m)
where i is the i-th sampling point and wi is the weight of the i-th sampling
point of m-th order.
The error in the m-th order one dimensional Gaussian quadrature formula
is given by
(22m+1 [m!]4 )
f 2m ()
=
(2m + 1)[(2m)!]3
Where f 2m () is the (2m)-th partial derivative of the function with respect to
x and is a point somewhere in the interval [1, 1].
The Gauss Quadrature formula for numerical integration of a function f over
domain D is of the form
ZZZ
f dD =
D
N
X
wi f (i , i , i )
i=1
where (, , ) are the coordinate of the i-th sampling point and wi is the weight
of the i-th sampling point.
2.3
Finite Element Formulation of Physical Problems
This section presents Finite Element formulations for some two dimensional field
problems. The underlying mathematical basis of the finite element method first
lies with the classical theories introduced by Rayleigh [1877] and Ritz [1909].
Extension of the ideas, originally conceived by Galerkin [1915] in the early 20th
century and the method was found to provide the idea theoretical basis for wide
set of problems encountered in the realm of engineering science. Basically this
30
method requires the governing differential equations to be multiplied by a set
of predetermined weights and the resulting product to be integrated over the
problem domain, this integral is then required to vanish. The number of weighting
functions equal to the number of unknown coefficients in the approximate solution.
A rigorous description of the method can be found in Finlayson [1972]. Here we
wish to use Galerkins techniques to present finite element problem equations of
some physical problems in a general manner.
2.3.1
Torsion Problem
Usually, the torsion problem, with simply connected cross-sections is written as

2 2
+
+h=0
x2 y 2
= 0 on C
and
in A
(2.11)
= 0 on C
Where C and C constitute the cross section boundaries. (h = 2g)

Using of weighted residuals
ZZ
W R d = 0
(2.12)
2 2
+
+h
x2 y 2
(2.13)
where
R=
is the residual function and W is the weighting function.

From the generic formula for differentiation we have
W
=
{W }
2
x
x
x
x x
2
W
W 2 =
{W }
y
y
y
y y
W
31
Hence from Eq.(2.12) and (2.13) we obtain
ZZ
[ (W ) +
(W )]d
x
x
y
y
ZZ
W W
(
+
)d +
x x
y y
ZZ
W hd = 0
(2.14)
We known from the Greens theorem in plane
ZZ
F
G
(
+
)d =
x
y
Z
(F dy Gdx)
Which gives
ZZ
[ (W ) +
(W )]d =
x
x
y
y
Z
(W
dy W dx)
x
y
(2.15)
Where F = C + C , the right hand side of Eq.(2.15) is equated to zero owing

to the boundary condition = 0 on C* and
= 0 on C.
Thus we have
ZZ
(W )]d = 0
[ (W ) +
x
x
y
y
and for which form the Eq.(2.14) we obtain

ZZ
W W
+
)d +
(
x x
y y
ZZ
W hd = 0
(2.16)
We now approximate the function, using shape functions as

=
M
X
Nj j
(2.17)
j=1
Where M is the number of nodes in the domain . In Galerkins formulation

we set Wj = Nj . Now we have
ZZ
M ZZ
X
W W
{
(
+
)d}j =
hNi d = 0 1 i m
x x
y y
j=1
32
(2.18)
It is clear that Eq.(2.18) can be written as
[K]{} = {F }
(2.19)
where the components of [K] and {F } are given by

ZZ
(
Kij =
Ni Nj Ni Nj
+
)dxdy
x x
y y
(2.20)
ZZ
Fi =
hNj dxdy
(2.21)
Matrix [K] usually known as the stiffness matrix which is symmetric and {F } is
called load vector. Similarly, employing weighted residual Galerkins formulation
for other problems the stiffness matrix can be obtained as in the same form.
From Eq.(2.20), it is clear that the evaluation of the stiffness matrix [K] requires
to integrate the product of the global derivatives of shape functions.
2.3.2
Elasticity Problem
Two dimensional elasticity is generally categorized into two modes: plane strain
and plane stress. When the thickness of a solid object is large, a state of plane
strain is considered to exist. If this thickness is small compared to its overall
dimensions (x, y), the condition of plane stress is assumed. Both cases are subsets
of general three dimensional elasticity problems. In this instance, body forces (or
loads) cannot have components in the z-direction, nor vary in the direction of the
body thickness.
The governing equations which described two dimensional elastic stress are
defined as Reddy [1984]
x xy
+
+ fx = 0
x
y
xy y
+
+ fy = 0
x
y
(2.22)
where x and y are the normal stress components in the x and y directions,
33
Figure 2.6: Thin Body in Plane Stress.
Figure 2.7: Thick Body in Plane Strain.
respectively; xy is the shear stress which acts in the xy plane; and fx and fy are
the body force terms. The strain displacement relations are defined from Eq.(2.22)
as
u
v
u v
(C11
+ C12 ) C33 (
+
) = fx
x
x
y
y y x
u v
u
v
C33 (
+
)
(C12
+ C22 ) = fy
x y x
y
x
y
(2.23)
(2.24)
with the boundary traction given by

v
u v
u
+ C22 )x C33 (
+
)y
x
y
y y x
u v
u
v
= C33 (
+
)x + (C12
+ C22 )y
y x
x
y
tx = (C11
(2.25)
ty
(2.26)
The u and v displacements are approximated over an element e as

e
u =
n
X
Ni ui
v =
i=1
n
X
Ni vi
i=1
By Galerkin formulation, the residual equations are
34
(2.27)
ZZ
[
u
v
u v
(C11
+ C12 ) + C33 (
+
) + fx ]Ni dxdy = 0
x
x
y
y y x
ZAZ
[C33
(2.28)
u v
u
v
(
+
)+
(C12
+ C22 ) + fy ]Ni dxdy = 0
x y x
y
x
y
From Greens theorem in the plane we know that

ZZ
G F
(
)dxdy =
x
y
Z
(F dx + Gdy)
If we choose
F = C33 (
u v
+
)Ni
x y
G = (C11
u
v
+ C12 )Ni
x
y
then from Eq.(2.28) we have
ZZ
[
u
v
u v
(C11
+ C12 ) + C33 (
+
)]Ni dxdy
x
x
y
y y x
(2.29)
ZZ
=
Ni
u
v
Ni u v
(C11
+ C12 ) + C33
(
+
)]dxdy
x
x
y
y y x
Z
Ni {C33 (
u v
u
v
+
)dx + (C11
+ C12 )dy}
y x
x
y
By using the Eq.(2.25), we get
u v
u
v
Ni {C33 (
+
)dx + (C11
+ C12 )dy} =
y x
x
y
Z
Ni tx ds
S1
Now from Eq.(2.29) and Eq.(2.30), we get,
35
(2.30)
ZZ
[
u
v
u v
(C11
+ C12 ) + C33 (
+
)]Ni dxdy
x
x
y
y y x
(2.31)
ZZ
=
Ni
u
v
Ni u v
(C11
+ C12 ) + C33
(
+
)]dxdy
x
x
y
y y x
Z
Ni tx ds
+
S1
Now from Eq.(2.28), we get,
ZZ
Ni
u
v
Ni u v
[
(C11
+ C12 ) + C33
(
+
) Ni fx ]dxdy
x
x
y
y y x
Z
Ni tx ds = 0
S1
(2.32)
Similarly, we have,
ZZ
Ni
u
v
Ni u v
( + )+
(C12 +C22 )Ni fy ]dxdy
[C33
x y x
y
x
y
Z
Ni ty ds = 0 (2.33)
S2
Now from Eq.(2.27), we have,

v e X Nj
=
uj ,
x
x
j
v e X Nj
=
vj ,
x
x
j
ue X Nj
=
uj
y
y
j
v e X Nj
=
vj
y
y
j
Now from Eq.(2.32), we get,

n ZZ
X
j=1
A
Z
n
X Z
j=1
uj (C11
Ni Nj
Ni Nj
+ C33
)dxdy
x x
y y
Ni Nj
Ni Nj
vj (C12
+ C33
)dxdy =
x y
y x
ZZ
Z
Ni fx dxdy +
Ni tx ds
S1
(2.34)
36
for i = 1, 2, 3, ..., n. Similarly,
n ZZ
X
uj (C33
j=1
A
Z
n
X Z
j=1
Ni Nj
Ni Nj
+ C12
)dxdy
x y
y x
Ni Nj
Ni Nj
vj (C33
+ C22
)dxdy =
x x
y y
ZZ
Ni ty ds
Ni fy dxdy +
A
S2
(2.35)
Eq.(2.34) and (2.35) can be written in matrix form as
K11 u + K12 v = F1
(2.36)
K21 u + K22 v = F2
(2.37)
where
ZZ
K11 =
Ni Nj
Ni Nj
+ C33
)dxdy
x x
y y
ZZ
Ni Nj
Ni Nj
=
(C12
+ C33
)dxdy
x y
y x
(C11
A
K12 = K21
(2.38)
(2.39)
ZZ
Ni Nj
Ni Nj
+ C22
)dxdy
x x
y y
ZAZ
Z
=
Ni fx dxdy + Ni tx ds
K22 =
F1
(C33
Ni fy dxdy +
A
2.3.3
(2.41)
Z1
ZZ
F2 =
(2.40)
Ni ty ds
(2.42)
S2
Heat Conduction Problem
The steady state heat conduction in two dimension is of the form:

kx
2
2
+
k
+ Q = 0 in
y
x2
y 2
(2.43)
where kx and ky are the thermal conductivities and Q is an internal heat source
37
or sink. The differential Eq.(2.43) is embedded within Eq.(2.1) in page 25. The
parameters for Eq.(2.1) Dx = kx , Dy = ky , G = 0, Q = Q with standard boundary
conditions
= on
= q
on q
(2.44)
where + q = . Now the weighted residual statement, using the trial functions themselves as weighting functions, can be deduced as the similar procedure
in the previous cases. We obtain,
Z
Z
Z
Ni b Ni b
kx
+
ky )dxdy = QNi dxdy qNi d
(
x x
y y
q
(2.45)
Eq.(2.45) can be written in matrix form as

[K]{} = {f }
(2.46)
in which the components of the matrices K and f are determined by summing

the individual element contributions
Kije
Z
=
Nie Nje Nie Nje

kx
+
ky
)dxdy
x
x
y
y
Z
Z
e
e
fi = QNi dxdy qNie d
e
(2.47)
(2.48)
eq
Here e is the surface of the element e and eq is the position of the element
boundary which lies on, or approximates to, a portion of q . The integral over
eq in Eq.(2.47) will thus only appear for elements that are adjacent to q . This
finite element formulation, is completely general. We wish to note that one can
use only the general triangular or the convex quadrilateral elements as well as
both type of elements if necessary for discretization of the problem domain. At
a glance, it can be said that in Finite Element equation of the physical problems
following integrals are needed to evaluate to form the element matrices:
38
ZZ
A
ZZ
ZZ
Ni Ni
Ni Nj
Ni Ni
dxdy,
dxdy,
dxdy,
x x
y y
x y
A
Z
Z A
ZZ
ZZ
Ni Ni
Ni ds
Ni dxdy, Ni ds,
dxdy,
y x
A
S1
S2
for i, j = 1, 2, 3,..., N, and N is the number of nodes specified over the element
and s1 , s2 part of the boundary of the domain A.
39
CHAPTER 3
Evaluation of Triangular
Domain Integrals by use
of Gaussian Quadrature
for Square Domain
Integrals
Chapter 3
Evaluation of Triangular Domain
Integrals by use of Gaussian
Quadrature for Square Domain
Integrals
This chapter presents a technique to evaluate the integrals over the triangular surfaces using readily available Gaussian quadrature for the square domain integrals.
In order to perform the integration we transformed the arbitrary triangular domain in (x, y)-space to a unit triangular domain {(u, v)| 0 u, v 1, u + v 1},
then the unit triangle is divided into three quadrilaterals. Each of the quadrilaterals is transformed into its equivalent 2-square {(, )| 1 1, 1 1}.
As the technique suitably can employ higher order Gaussian quadrature for square
domain integrals, higher degree of accuracy of the integrals is possible to achieve
without resorting to inefficient quadrature formulae for triangles.
3.1
Introduction
Most of the domain integrals encountered in several areas of science and engineering are not amenable to evaluate analytically or tedious in calculation. In
recent and ongoing researches, we have been witnessing finite element method
41
3. Use of Gaussian Quadrature for Square Domain Integrals

(FEM) gained importance due to the most obvious reason that it can provide solutions to many complicated problems that would be intractable by other numerical methods, Abramowitz and Stegun [1974]; Bathe [1996]; Cowper [1973];
Hughes [1987]; Lannoy [1977]; Laurie [1977]; Lyness and Jespersen [1975];
Rathod and Karim [2002]; Reddy [1978]; Reddy and Shippy [1981]; Rogers
and Adams [1990]. It is well known that the evaluation of the integrals is the pivotal task to form the element matrices in FEM solution procedure, Hammer and
Stroud [1956, 1958]; Hammer et al. [1956]; Irons and Razzaque [1972]; Keller
[1964]; Keller and McKean [1973]; Molyneux [1968]; Strang and Fix [1973];
Wachspress [1975]; Zienkiewicz and Taylor [1989]. Such integrals encountered
for employing linear elements in the discretization are simple and may be evaluated analytically. But large numbers of integrals are needed to be evaluated,
Hillion [1977]; Lague and Baldur [1977]; Laurie [1977]; Laursen and Gellert
[1978]; Lether [1976]. However, encountered integrals for employing higher order
elements or for some distorted elements are too complicated and their evaluation
culminated the use of numerical integration schemes. Among all the numerical
integration schemes Gaussian quadrature which can exactly evaluate the (2n-1)th
order polynomial with n Gaussian points is most commonly used in view of accuracy and efficiency of calculations, Abramowitz and Stegun [1974]; Rathod
and Karim [2002]; Reddy [1978]; Reddy and Shippy [1981]. It is not possible
to list all the publications in this field, they are so numerous.
Though the triangular elements (lower order to higher order) are widely used
in finite element analysis, very few Gaussian quadrature formula are available for
triangular domain integrals and among such formulae the Gaussian 7-point and
13-point formulae are not sufficient for satisfactory degree of accuracy for the
evaluation of the integrals, Lague and Baldur [1977]. It is also investigated by
Rathod and Karim [2002] that for some element geometry, such formulae are
not at all reliable. Same fact is found in Dunavant [1985]; Taylor et al. [2000];
Wandzura and Xiao [2003].
The versatility of the popular triangular elements can be further enhanced by
improved numerical integration schemes and hence evaluation of the triangular
domain integrals with desired accuracy by other technique is preferable. It is
notable that the high order Gaussian quadrature formulae available only for the
42

square domain integrals and the same is demanded for the triangular domain integrals. But, the derivation of the higher order Gaussian quadrature for triangular
domain integrals is not so easy and indeed very difficult task. As an alternative
in this chapter the triangular domain integral is transformed into square domain
integrals in order to utilize the readily available (lower order to higher order)
Gaussian quadrature for square domain integrals and hence the evaluation of the
integrals over the triangular surface with desired accuracy is now possible. The
efficiency and accuracy of the present technique is demonstrated through application examples given by Rathod and Karim [2002]; Rathod et al. [2004].
Thus, the present technique will find immense application in many areas in science and engineering specially in FEM solution procedure of continuum mechanics
problems.
3.2
General Description and Formulation of the

Problem
Consider the integral over the arbitrary two dimensional domain as:
ZZ
I=
f d;
: original domain
(3.1)
To solve a practical problem using FEM, we divide the irregular domain into
finite number of elements. Observe that I can be calculated as a sum of integrals
evaluated over simplex divisions i :
=
i ;
i : completely covers
(3.2)
i are triangles in two dimensional domain (see Figure 3.1).

Now Eq.(3.1) can be written as
ZZ
I=
f d =
X ZZ
i
f di
(3.3)
To evaluate the integral I in Eq.(3.3), it is now required to evaluate the trian43
Figure 3.1: Triangulation of the domain of integration.

gular domain integral
ZZ
II =
f (x, y) dx dy;
: triangular domain
(3.4)
When some complex geometrical boundaries have to be modeled, the simple

shapes of the elements restrict severely their applications in the analysis of practical problems. This restriction can be removed by mapping a simple element in the
local (, ) elements coordinates into a more complex shape in the global (x, y)
coordinate system. Once a particular form of mapping is adopted and the coordinates are chosen for every element so that these map into contiguous space, then
shape functions written in the local element (, ) space can be used to represent
the function variation over the element in the global (x, y) space without upsetting
the inter-element continuity requirements. Therefore, integration over triangular
domains is usually carried out in normalized co-ordinates. In this chapter we are
transforming each arbitrary triangular element into its equivalent Unit triangle in
transformed domain. To perform the integration, first map one vertex (vertex 1)
to the point (1, 0), the second vertex (vertex 2) to the point (0, 1) and the third
vertex (vertex 3) to the origin, (see Figure 3.2).
This transformation is most easily accomplished by use of shape functions as:
x
y
!
=
x1 x2 x3
y1 y2 y3
44
N1
N2
N3
(3.5)
Figure 3.2: Original and Transformed Unit Triangle.

where
N1 (u, v) = u,
N2 (u, v) = v,
N3 (u, v) = 1 u v
(3.6)
The original and the transformed triangles are shown in Figure 3.2. From
Eq.(3.5) and Eq.(3.6), we obtain
x(u, v) = x3 + (x1 x3 )u + (x2 x3 )v

y(u, v) = y3 + (y1 y3 )u + (y2 y3 )v
(3.7)
Jacobian of the transformation is then

(x, y)
= (x1 x3 )(y2 y3 ) (x2 x3 )(y1 y3 ) = J (say)
(u, v)
(3.8)
Finally Eq.(3.4) reduces to

1
1u
II = J
u=0
f (x(u, v), y(u, v)) dv du
(3.9)
f (x(u, v), y(u, v)) du dv
(3.10)
v=0
One can simply verify that

Z
1v
II = J
v=0
u=0
45

Here, we wish to mention that the integrals II in Eq.(3.9) or in Eq.(3.10) is
generally evaluated by the existing Gaussian quadrature for triangle. But, depending upon the integrand the evaluation is either acceptable with desired accuracy
or highly erroneous. This is one of the serious drawbacks of such integration
schemes.
3.3
Numerical Evaluation Procedure
Our goal is to obtain an appropriate method to evaluate the integrals of Eq.(3.9)

or Eq.(3.10). In this section, we will describe briefly two procedures to evaluate
the integral II of which one procedure is the available Gaussian Quadrature for
Triangles and another method is newly proposed, in order to use extensively the
Gaussian quadrature formulae for square domain integrals.
3.3.1
Gaussian Quadrature for Triangle (GQT)
It is well known that the Gaussian quadrature formula for triangle to evaluate the
integral in Eq.(3.9) or in Eq.(3.10) is
II = J
N
GP N
GP
X
X
i=1
Wi Wj f (x(ui , vj ), y(ui , vj ))
(3.11)
j=1
where (ui , vj ) are the ij-th sampling points; Wi , Wj are corresponding weights
and NGP denotes the number of Gauss points in the formula and NGP 13.
It is thoroughly investigated by several authors that in some cases e.g., specially
for the nonpolynomial functions (integrand) available Gaussian quadrature for
triangle can not evaluate the integral II accurately and hence either higher order
formulae are required or a different method is time demanded.
3.3.2
Integration Over Unit Triangle (IOUT)
Integration over the normalized unit triangle can be calculated as a sum of integrals evaluated over three quadrilaterals (Figure 3.3). Now, Eq.(3.9) can be
written as
46
Figure 3.3: Unit Triangle Divided into three Quadrilaterals.
II = J
Zu=0
Z
= J{
1u
f (x(u, v), y(u, v))dv du

ZZ
f (x(u, v), y(u, v)) dv du +
v=0
e1
e2
ZZ
+
f (x(u, v), y(u, v)) dv du}

e3
= J
3 ZZ
X
i =1
(3.12)
ei
Each quadrilateral e1 , e2 , e3 can be transformed into its equivalent 2-square in

{(, )| 1 1, 1 1} by using isoparametric transformation and
shape functions ni given by
47
u=
4
X
u i ni ,
v=
i=1
4
X
vi ni
(3.13)
i=1
where the shape functions to transform a quadrilateral to its equivalent 2square is given by
(1 + )(1 )
,
4
(1 + )(1 + )
n3 (, ) =
,
4
n1 (, ) =
(1 )(1 + )
,
4
(1 )(1 )
n4 (, ) =
4
n2 (, ) =
(3.14)
Figure 3.4: Transformation of Quadrilateral into Square.

Using Eq.(3.13) and Eq.(3.14) for a single quadrilateral element, we have
u = uA0 nA0 + uB 0 nB 0 + uC 0 nC 0 + uD0 nD0

1
{(uA0 + uB 0 + uC 0 + uD0 ) + (uA0 + uB 0 uC 0 uD0 )
=
4
+ (uA0 + uB 0 + uC 0 uD0 ) + (uA0 + uB 0 uC 0 + uD0 )}
v = vA0 nA0 + vB 0 nB 0 + vC 0 nC 0 + vD0 nD0
1
=
{(vA0 + vB 0 + vC 0 + vD0 ) + (vA0 + vB 0 vC 0 vD0 )
4
+ (vA0 + vB 0 + vC 0 vD0 ) + (vA0 + vB 0 vC 0 + vD0 )}
48
(3.15)

The jacobian of the transformation is given by

u v u v

Jei =

For element e1 using the coordinates of A0 (1, 0), B 0 (1/2, 1/2), C 0 (1/3, 1/3), D0 (1/2, 0)
we have
u = (14 + 4 4 2)/24,
v = (5 + + 5 + )/24
Using these values of u, v in Eq.(3.7) we can obtain (X1 , Y1 ) for element e1

1
and the jacobian is J = 96
(4 + ). Then the integral over element e1 become
ZZ
1
f (x(u, v), y(u, v)) dv du =
96
f (X1 , Y1 )(4 + )dd

=1
e1
=1
Similarly, for quadrilateral element e2 we have

ZZ
1
f (x(u, v), y(u, v)) dv du =
96
f (X2 , Y2 )(4 + + )dd

=1
e2
=1
and for quadrilateral element e3 we have

ZZ
e3
1
f (x(u, v), y(u, v)) dv du =
96
f (X3 , Y3 )(4 )dd

=1
=1
Combining the results for all the three quadrilateral elements we get
Z 1Z 1
J
II =
[f (X1 , Y1 )(4 + ) + f (X2 , Y2 )(4 + + )
96 1 1
+ f (X3 , Y3 )(4 )]d d
(3.16)
Eq.(3.16) is obtained after transforming each quadrilaterals into a square in

(, ) space where
49
X1 =
1
[a
24 11
+ a12 + a13 + a14 ] Y1 =
1
[b
24 11
+ b12 + b13 + b14 ]
X2 =
1
[a
24 21
+ a22 + a23 + a24 ],
Y2 =
1
[b
24 21
+ b22 + b23 + b24 ]
X3 =
1
[a
24 31
+ a32 + a33 + a34 ],
Y3 =
1
[b
24 31
+ b32 + b33 + b34 ]

(3.17)
and
a11 = 14x1 + 5x2 + 5x3
b11 = 14y1 + 5y2 + 5y3
a12 = 4x1 + x2 5x3
b12 = 4y1 + y2 5y3
a13 = 4x1 + 5x2 x3
b13 = 4y1 + 5y2 y3
a14 = 2x1 + x2 + x3
b14 = 2y1 + y2 + y3
a21 = 5x1 + 14x2 + 5x3
b21 = 5y1 + 14y2 + 5y3
a22 = x1 + 4x2 5x3
b22 = y1 + 4y2 5y3
a23 = 5x1 + 4x2 + x3
b23 = 5y1 + 4y2 + y3
a24 = x1 + 2x2 x3
b24 = y1 + 2y2 y3
a31 = 5x1 + 5x2 + 14x3
b31 = 5y1 + 5y2 + 14y3
a32 = 5x1 x2 4x3
b32 = 5y1 y2 4y3
a33 = x1 + 5x2 4x3
b33 = y1 + 5y2 4y3
a34 = x1 x2 + 2x3
b34 = y1 y2 + 2y3
Now right hand side of Eq.(3.16) with Eq.(3.17) can be evaluated by use of
available higher order Gaussian quadrature for square domain. Since the unit
triangle is divided into three quadrilaterals and each quadrilateral is transformed
into equivalent 2-square, for m n points Gaussian quadrature formula, total
number of points within the triangle is 3 m n. Figure 3.5 and 3.6 shows
the distribution of Gauss points within the Unit triangle for m = n = 7 and
50

m = n = 10.
Figure 3.5: Distribution of 3 7 7 points using IOUT.
3.4
Application Examples
To show the accuracy and efficiency of the derived formulae, following examples
with known results are considered:
1y
I1 =
(x + y) 2 dx dy = 0.4
y=0
1
x=0
1y
I2 =
(y) 2 dx dy = 0.266
y=0
x=0
51
Figure 3.6: Distribution of 3 10 10 points using IOUT.

Numerically evaluated values of these integrals by the existing Gauss 7-points,
13-points rules for triangle and the proposed method are given in Table 3.1.
It is clear from the Table 3.1 that the existing 7-point and 13-point rules
for triangles produces more erroneous results. The results obtained by the new
method IOUT are satisfactory in view of accuracy.
3.5
Conclusions
This chapter presented a new technique to use readily available Gaussian quadrature formulae for square domain integrals to evaluate the integral over the triangular surfaces. To do so, the original triangle is first transformed into a unit
triangle and then the unit triangle subdivided into three quadrilaterals. Finally,
all the quadrilaterals individually transformed into 2-square {(, )| 1
52
Table 3.1: Calculated values of the integrals I1 , I2 .

Method
Gauss Points
77
GQT
13 13
355
Proposed Method
366
IOUT
377
388
399
3 10 10
I1
0.4001498818
0.4000451564
0.4000032328
0.4000013845
0.4000006727
0.4000003534
0.4000002024
0.4000001234
I2
0.2682188451
0.2674752175
0.2668388764
0.2667701218
0.2667336604
0.2667125320
0.2666994391
0.2666908966
1, 1 1} and consequently the original integral over the triangle reduces as

resulted integral over the 2-square domain. Hence, it is possible to use higher order
one dimensional Gaussian quadrature formulae for the square domain integrals to
evaluate the triangular domain integrals with satisfactory degree of accuracy.
The technique is demonstrated through application examples and it is recommended that the technique will be suitable alternative method to evaluate
triangular domain integrals instead of using the usual Gaussian quadratures for
triangle which are not always reliable in view of accuracy. Thus, the versatility
of the triangular finite elements is now enhanced with proposed method. Further
more, present method is efficient and suitable for computer coding. A complete
program (Program -1: IOUT.for) in FORTRAN is developed and appended for
ready use and reference. It is expected now that the developed technique will be
the best alternative to test the accuracy of higher order (Gaussian) quadrature
formulae for triangular domain integrals.
53
Chapter 4
An Accurate Evaluation
Procedure of
Triangular Domain
Integrals
Chapter 4
An Accurate Evaluation
Procedure of Triangular Domain
Integrals
This chapter mainly concentrates to present quadrature schemes to evaluate accurately the triangular domain integrals with satisfactory degree of accuracy. In
this chapter, the original triangular domain in (x, y) space is first transformed to
Standard triangular domain {(s, t)| 1 s, t 1, s + t 0}. Then the standard triangle is divided into three quadrilaterals. Each of the quadrilaterals are
transformed into a square domain in {(, )| 1 1, 1 1}. The
technique successfully accommodates the readily available Gaussian quadrature
formulae for square and therefore it guaranties the high accuracy in evaluating
such integrals. Hence, the technique instead of the usual Gaussian quadrature
formula for triangle will find better place in application when high accuracy is
desired.
4.1
Introduction
Most of the domain integrals encountered in real problems defies our analytical
skills and we are resort to numerical integration methods, Bathe [1996]; Hughes
[1987]; Irons and Razzaque [1972]; Rogers and Adams [1990]; Strang and
Fix [1973]; Wachspress [1975]; Zienkiewicz and Taylor [1989]. Among vari55
4. An accurate evaluation procedure of triangular domain integrals

ous numerical methods Gaussian quadrature formulae are widely used and there
are respectively quadrature formulae for triangular domain integrals as well as for
square domain integrals. In many practical situations, with the advancement of
research higher order quadrature formulae for triangle domain integrals are highly
demanded. Finite element method (FEM) gained importance due to the most obvious reason that it can provide solutions to many complicated problems that
would be intractable by other numerical methods, Cowper [1973]; Hammer and
Stroud [1956, 1958]; Hammer et al. [1956]; Keller [1964]; Keller and McKean
[1973]; Lether [1976]; Lyness and Jespersen [1975]; Molyneux [1968]; Strang
and Fix [1973]. The evaluation of the domain integrals is the main and pivotal
task to form the element matrices in FEM solution procedure. Such encountered
integrals for employing higher order elements or for some distorted elements are
too complicated and their evaluations culminated the use of numerical integration
schemes. Enormous number of publications exists for this field, and the existing
formulae are not sufficient for a satisfactory degree of accuracy of the triangular
domain integrals. It is also investigated that for some element geometry, such
formulae are not at all reliable Lague and Baldur [1977]; Rathod and Karim
[2002]; Reddy [1978]; Reddy and Shippy [1981] and in case of the integral of
nonpolynomial functions these are almost useless in view of accuracy and efficiency. Similar limitations are found in Dunavant [1985]; Taylor et al. [2000];
Wandzura and Xiao [2003].
Improved numerical integration schemes can further enhance the versatility of
the popular triangular elements. The high order Gaussian quadrature formulae
available only for the square domain integrals and the same are demanded for
the triangular domain integrals, Abramowitz and Stegun [1974]; Lague and
Baldur [1977]; Rathod and Karim [2002]; Reddy [1978]; Reddy and Shippy
[1981]. But, the derivation of the higher order Gaussian quadrature for triangular
domain integrals is very difficult task. To do so, in this chapter the arbitrary
triangular domain of integration is transformed into a Standard triangle. The
standard triangle is then subdivided into three quadrilateral, each of which is
then transformed into its equivalent 2-square. Now the readily available Gaussian
quadrature for square domain integrals can be utilized to evaluate the integrals
over the triangular surface with desired accuracy. The efficiency and accuracy of
56

the present technique is demonstrated through examples studied in Rathod and
Karim [2002]; Rathod et al. [2004]. Thus, the present technique will be highly
applicable in many areas of computational engineering science specially in FEM
solution procedure of continuum mechanics problems.
4.2
Formulation of the Problem
In many practical problems, the complications arises with the numerical evaluation
of the integral over the irregular surface as:
ZZ
f (x, y) dx dy
I=
: irregular domain
(4.1)
the irregular domain can be divided over simplex divisions i where i are
triangles in two dimensional domain (see Figure 3.1 in page 44). Then I can be
calculated as a sum of integrals over 4i :
=
4i ;
4i : completely covers
(4.2)
To evaluate the integral I in Eq.(4.1), it is now required to evaluate the triangular domain integral
ZZ
II =
4 : arbitrary triangle
f (x, y) dx dy;
(4.3)
In FEM, integration over triangular domains is usually carried out in normalized co-ordinates. In this chapter, we are transforming each arbitrary triangles
into a normalized Standard triangle {(s, t)| 1 s, t 1, s + t 0}. To perform
this we first map one vertex (vertex 1) to the (1, -1), the second vertex (vertex 2)
to point (-1, 1) and the third vertex (vertex 3) to point (-1, -1) (as seen in Figure
4.1).
This transformation by using shape functions can be easily accomplished as:
57
Figure 4.1: Original and Transformed Standard Triangle.
x
y
!
=
x1 x2 x3
y1 y2 y3
N1
N2
N3
(4.4)
1
N3 (s, t) = (s + t)
2
(4.5)
where
1
N1 (s, t) = (s + 1),
2
1
N2 (s, t) = (t + 1),
2
From Eq.(4.4) and Eq.(4.5), we obtain
1
1
1
x(s, t) = (x1 + x2 ) + (x1 x3 )s + (x2 x3 )t
2
2
2
1
1
1
y(s, t) = (y1 + y2 ) + (y1 y3 )s + (y2 y3 )t
2
2
2
(4.6)
and hence the Jacobian of this transformation is given by

(x, y)
1
J
= ((x1 x3 )(y2 y3 ) (x2 x3 )(y1 y3 )) =
(s, t)
4
4
(4.7)
The original and the transformed triangles are shown in Figure 4.1. Eq.(4.3)
58

then reduces to
J
II =
4
s = 1
f (x(s, t), y(s, t)) dt ds
(4.8)
f (x(s, t), y(s, t)) ds dt
(4.9)
t = 1

J
II =
4
t = 1
s = 1
The integrals II in Eq.(4.8) or Eq.(4.9) is generally evaluated by the existing

Gaussian quadrature (say 7-point and 13-point) for triangle (GQT). The results
obtained using the existing methods are highly erroneous or acceptable with desired accuracy depending upon the integrand, which is a serious disadvantage of
such integration schemes.
4.3
Integration Over Standard Triangle (IOST)
The main interest of this chapter is to obtain a suitable, highly accurate, stable
technique to evaluate the triangular domain integrals of type given in Eq.(4.8) or
Eq.(4.9). In this section, an alternative procedures is proposed to evaluate the
integral II in order to use extensively the Gauss quadrature for square domain
integrals.
Integration over the normalized Standard triangle can be calculated as a sum
of integrals evaluated over three quadrilaterals (Figure 4.2). Following similar
procedure of calculation described in section 3.3.2 in page 46 we have
II =
=
f (x(s, t), y(s, t))

s=1 t=1
3 ZZ
X
J
4
i =1
ei
1
J
[f (X1 , Y1 )(4 + ) + f (X2 , Y2 )
96 1 1
(4 + + ) + f (X3 , Y3 )(4 )]d d
(x, y)
dt ds
(s, t)
59
(4.10)
Figure 4.2: Standard Triangle Divided into three Quadrilaterals.

Eq.(4.10) is obtained after transforming each quadrilaterals into a square in
(, ) space where
X1 =
1
[a
24 11
+ a12 + a13 + a14 ] Y1 =
1
[b
24 11
+ b12 + b13 + b14 ]
X2 =
1
[a
24 21
+ a22 + a23 + a24 ],
Y2 =
1
[b
24 21
+ b22 + b23 + b24 ]
X3 =
1
[a
24 31
+ a32 + a33 + a34 ],
Y3 =
1
[b
24 31
+ b32 + b33 + b34 ]

(4.11)
and
a11 = 14x1 + 5x3 + 5x2
b11 = 14y1 + 5y2 + 5y3
a12 = 4x1 + x2 5x3
b12 = 4y1 + y2 5y3
a13 = 4x1 + 5x2 x3
b13 = 4y1 + 5y2 y3
a14 = 2x1 + x2 + x3
b14 = 2y1 + y2 + y3
60

a21 = 5x1 + 14x2 + 5x3
b21 = 5y1 + 14y2 + 5y3
a22 = x1 + 4x2 5x3
b22 = y1 + 4y2 5y3
a23 = 5x1 + 4x2 + x3
b23 = 5y1 + 4y2 + y3
a24 = x1 + 2x2 x3
b24 = y1 + 2y2 y3
a31 = 5x1 + 5x2 + 14x3
b31 = 5y1 + 5y2 + 14y3
a32 = 5x1 x2 4x3
b32 = 5y1 y2 4y3
a33 = x1 + 5x2 4x3
b33 = y1 + 5y2 4y3
a34 = x1 x2 + 2x3
b34 = y1 y2 + 2y3
Each quadrilaterals in Figure 4.2 is transformed into 2-square in {(, ) | 1

1, 1 1} space through isoparametric transformation to evaluate the
integral II in Eq.(4.10). Now right hand side of Eq.(4.10) with Eq.(4.11) can
be evaluated by use of available higher order Gaussian quadrature for square
domain. Since the standard triangle is divided into three quadrilaterals and each
quadrilateral is transformed into equivalent 2-square, for m n points Gaussian
quadrature formula, total number of points within the triangle is 3 m n.
Figure 4.3 shows the distribution of Gauss points within the Standard triangle for
m = n = 8.
4.4
1y
I1 =
y=0
1
x=0
y
I2 =
y=0
(x + y) 2 dx dy = 0.6666667
1
(x2 + y 2 ) 2 dx dy = 0.881373587
x=0
Numerically evaluated values of these integrals by the existing Gauss 7 - point,

13 - point rules for triangles (Explained in Section 3.3.1 in page 46) and the
61
Figure 4.3: Distribution of 3 8 8 points using IOST.

proposed method are given in Table 4.1.
It is clear from the Table 4.1 that the existing 7- point and 13 - point rules
for triangles produce more erroneous results but the results obtained by IOST are
satisfactory in view of accuracy.
4.5
Conclusions
An alternative technique to use readily available Gaussian points and corresponding weights for square domain integrals to evaluate the integral over the triangular surfaces is derived in this chapter. To do so, each triangle is first transformed into a standard triangle then the standard triangle is subdivided into three
quadrilaterals. Finally, all the quadrilaterals individually transformed into square
{(, ) | 1 1, 1 1} and consequently the original integral over the
62
Table 4.1: Calculated values of the integrals I1 , I2

Method
GQT
Proposed Method
IOST
Gauss Points
77
13 13
355
366
377
388
399
3 10 10
Test example
I1
I2
0.6606860757 0.8315681219
0.6637058258 0.8501738309
0.6660540844 0.8704939556
0.6662960549 0.8735856484
0.6664256193 0.8755247201
0.6665011716 0.8768203350
0.6665481694 0.8777286724
0.6665789279 0.8783900003
arbitrary domain reduces as resulted integral over the square domain. Hence, it is
possible to use higher order Gaussian quadrature formulae for the square domain
integrals to evaluate the triangular domain integrals with satisfactory degree of
accuracy. The technique is demonstrated through application examples and it is
recommended that the technique will be suitable alternative method to evaluate
triangular domain integrals instead of using the usual Gaussian quadratures for
triangles. Thus, the versatility of the triangular finite elements is now enhanced
with the proposed method.
It can be be seen that the final form of the techniques IOST in Eq.(4.10) for
standard triangular element and IOUT in the Eq.(3.16) for unit triangular element
are identical. Thus any choice of triangular domain will produce the same result by
these two methods. It is expected now that these developed techniques will be the
better alternatives to obtain the desired accuracy of triangular domain integrals
using higher order Gauss quadrature formulae. Also, the developed technique can
be used as the best alternative to test the accuracy of higher order (Gaussian)
quadrature formulae developed later in this thesis.
63
Chapter 5
Appropriate Gaussian
Quadrature Formulae
for Triangles
Chapter 5
Appropriate Gaussian
Quadrature Formulae for
Triangles
This chapter mainly presents higher order Gaussian quadrature formulae for numerical integration over the triangular surfaces. The arbitrary triangular domain
in (x, y)-space is transformed to a normalized Unit triangle in (u, v)-space, then
the Unit triangle is transformed into its equivalent 2-square. In order to show
the exactness and efficiency of derived quadrature formulae, it shows the effective
use of available Gaussian quadrature for square domain integrals to evaluate the
triangular domain integrals. Finally, it presents n n points (for n > 1) and
n(n+1)
1 points (for n > 2) Gaussian quadrature formulae for triangle utilizing
2
n-point one dimensional Gaussian quadratures. By use of simple but straightforward algorithms, Gaussian points and corresponding weights are calculated and
presented for clarity and reference. The proposed n(n+1)
1 points formulae com2
pletely avoids the crowding of Gaussian points and overcomes all the drawbacks
in view of accuracy and efficiency for the numerical evaluation of the triangular
domain integrals of any arbitrary functions encountered in the realm of science
and engineering. To verify the efficiency and accuracy, the method is applied to
evaluation of triangular domain integral of different types of functions and to the
evaluation of integrals for element stiffness matrices in two dimensional elasticity
problem.
65
5. Appropriate Gaussian Quadrature Formulae for Triangles
5.1
Introduction
The integration theory extends from real line to the plane and three dimensional
spaces by the introduction of multiple integrals. Integration procedures on finite
domains underlie physically acceptable averaging process in engineering. In probabilistic estimations and in spatially discretized approximations, e.g., FEM and
boundary element methods (BEM), evaluation of integrals over arbitrary shaped
domain are the pivotal task. In practice, most of the encountered integrals
either cannot be evaluated analytically or the evaluations are very lengthy and
tedious. Thus, for simplicity numerical integration methods are preferred and the
methods extensively employ the Gaussian quadrature technique that was originally designed for one dimensional cases and the procedure naturally extends to
two and three dimensional rectangular domains according to the notion of the
Cartesian product. Gaussian quadratures are considered as the best method of
integrating polynomials because they guarantee that they are exact for polynomials less than a specified degree. In order to obtain the result with the desired
accuracy, Gaussian integration points and weights necessarily increase and there
is no computational difficulty except time in evaluating any domain integral when
the two and three dimensional regions are bounded respectively, by systems of
parallel lines and parallel planes.
Analysts cannot ignore at all the randomness in material properties and uncertainty in geometry that are frequently encountered in complex engineering systems. This entails the evaluation of an integral of the function over the domain
of the component. In general, the -shape-class is very irregular in two and
three dimensional geometry. For non-parallelogram quadrilateral, very frequent
in finite element modeling, there is no consistent procedure to select the sampling
point to implement a Gaussian quadrature on the entire element. Special integration schemes, e.g., reduced integration over quadrilaterals have been successfully
developed in Hughes [1987] and are widely used in commercial programs. There
is no methodical way to design such approximate integration schemes for polygons with more than four sides. An attempt to distribute the sampling points
according to the governing perspective transformation fails to assure the error
order germane to the quadrature formula. The reason can be traced to the crowd-
66

ing of quadrature points and this numerical computational difficulty persists in
all non-parallelogram polygonal finite elements, Rogers and Adams [1990]. A
considerable amount of research has been performed to attain perfect results of
domain integration for plane quadrilateral elements where numerical quadrature
techniques are employed, Bathe [1996]. The accuracy of a selected quadrature
strategy is indicated by compliance with the patch test proposed in Irons and
Razzaque [1972].
The overall error in a finite element calculation can be reduced by not relying
so heavily on artificial tessellation, which requires the deployment of elements with
large number of sides. An elegant systematic procedure to yield shape functions
for convex polygons of arbitrary number of sides developed in Wachspress [1975]
by which the energy density can be obtained in closed algebraic form in terms of
rational polynomials. However, a direct Gaussian quadrature scheme to numerically evaluate the domain integral on n-sided polygons cannot be constructed to
yield the exact results, even on convex quadrilaterals. In two dimension, n-sided
polygons can be suitably discretized with linear triangles rather than quadrilaterals and hence triangular elements are widely used in finite element analysis.
Another advantage is to be mentioned that there is no difficulty with triangular
elements as the exact shape functions are available and the quadrature formulas
are also exact for the polynomial integrands, Zienkiewicz and Taylor [1989].
Integration schemes based on weighted residuals are prone to instability since
the accuracy goal cannot be controlled. In deterministic cases the underlying averaging process may be inconsistent, which was stated as a variational crime Strang
and Fix [1973]. In stochastic differential equation such averaging processes are
termed dishonest Keller [1964]; Keller and McKean [1973]; Molyneux [1968].
Thus, the high accuracy integration method is demanded and it is meaningful
when the shape functions are the very best. Therefore, there has been considerable interest in the area of numerical integration schemes over triangles Cowper
[1973]; Hammer and Stroud [1956, 1958]; Hammer et al. [1956]; Hillion [1977];
Lague and Baldur [1977]; Lannoy [1977]; Laurie [1977]; Laursen and Gellert
[1978]; Lether [1976]; Lyness and Jespersen [1975]; Rathod and Karim [2002];
Reddy [1978]; Reddy and Shippy [1981]. It is explicitly shown in Lague and
Baldur [1977]; Rathod and Karim [2002] that the most accurate 7-point and
67

13-point rules are not sufficient to evaluate the triangular domain integrals and
for some element geometry these rules are not reliable also. And in case of the
integral of nonpolynomial functions these are almost useless in view of accuracy
and efficiency.
To address all these short comings, to make a proper balance between accuracy
and efficiency and to avoid the crowding of quadrature points we have proposed
1 points higher order Gaussian quadrature formulae
n n points and n(n+1)
2
to evaluate the triangular domain integrals. It is thoroughly investigated that
1 point formulae are appropriate in view of accuracy and efficiency
the n(n+1)
2
and hence the formulae are believed to find better place in numerical solution
procedure of continuum mechanics problems.
5.2
General Description and Formulation of the

Problem
One of the main task of this thesis is to evaluate the integral of an arbitrary
function f for two dimensional problems:
ZZ
II =
f (x, y) dx dy;
: triangle (arbitrary)
(5.1)
These type of integrals arise in axisymmetric finite element method with linear
triangular element as well as in finite element formulations of second order linear
differential equations by use of quadratic triangular element with two straight
sides and one curved side. In FEM, the original domain of integration is usually
discretized into finite number of elements. Unfortunately, the simple shapes of the
elements restrict severely their applications in the analysis of practical problems,
where often quite complex geometrical boundaries have to be modeled. This
restriction can be removed by mapping a simple element in the local coordinates
into a more complex shape in the global coordinate system. Once a particular
form of mapping is adopted and the coordinates are chosen for every element so
that these map into contiguous space, then shape functions written in the local
element space can be used to represent the function variation over the element
68

in the global space without upsetting the inter element continuity requirements.
If the domain contains a curved boundary then it can be modeled accurately by
curved finite elements.
In this chapter, each arbitrary triangle in (x, y)-space is transformed to a unit
triangle in (u, v)-space. To serve this purpose, first map one vertex (vertex 1) to
the point (1, 0), the second vertex (vertex 2) to point (0, 1) and the third vertex
(vertex 3) to origin, (see Figure 5.1). This transformation to a unit triangular
domain by use of shape functions can be easily accomplished as:
x
y
!
=
x1 x2 x3
y1 y2 y3
N1
N2
N3
(5.2)
where
N1 (u, v) = u,
N2 (u, v) = v,
N3 (u, v) = 1 u v
(5.3)
Figure 5.1: Original and Transformed Unit Triangle.

From Eq.(5.2) and Eq.(5.3), we obtain
x(u, v) = x3 + (x1 x3 )u + (x2 x3 )v

y(u, v) = y3 + (y1 y3 )u + (y2 y3 )v
69
(5.4)

and hence the jacobian of this transformation is given by
(x, y)
= (x1 x3 )(y2 y3 ) (x2 x3 )(y1 y3 ) = 4( say)
(u, v)
(5.5)
where 4 = 2(Area of the original triangle). Finally, Eq.(5.1) reduces to

1
1u
II = 4
u=0
(5.6)
f (x(u, v), y(u, v)) du dv
(5.7)
v=0

Z
1v
II = 4
v=0
u=0
The evaluation of integrals II in Eq.(5.6) and in Eq.(5.7) will yield the same
results. Any one of these two can be evaluated numerically. Influences of these
integrals will be investigated later to present new quadrature formulae for triangles.
5.3
Transformation to Square Domain
In this section, we wish to describe a new procedures to evaluate the integral

II numerically using Gaussian quadrature formulae for square domain integrals
and The new Gaussian Points and weights will be presented as the results of this
chapter. Consider the transformation equations:
1+
1+ 1+
1
, v = (1
)(
) = (1 )(1 + )
(5.8)
2
2
2
4
Using Eq.(5.8) the integral II of Eq.(5.6) is transformed into an integral over
the surface of the standard square {(, )| 1 1, 1 1} and the
Eq.(5.4) reduces to
u=
x = x3 + 21 (x1 x3 )(1 + ) + 41 (x2 x3 )(1 )(1 + )

y = y3 + 12 (y1 y3 )(1 + ) + 41 (y2 y3 )(1 )(1 + )
70
(5.9)

The determinant of the Jacobian and the differential area are:
u v u v
1
(u, v)
=
= (1 )
(,

8
du dv = dv du =
(u, v)
1
d d = (1 ) d d
(,
8
(5.10)
(5.11)
Using Eq.(5.9)and Eq.(5.11) into Eq.(5.6), we get
1 + (1 )(1 + )
1 + (1 )(1 + ) 1
,
), y(
,
)}
dd
2
4
2
4
8
1 1
Z 1Z 1
1 + (1 )(1 + ) 1
,
}
dd
(5.12)
= 4
f{
2
4
8
1 1
Z
II = 4
f {x(
In order to evaluate the square domain integral II in Eq.(5.12) efficient Gaussian quadrature coefficient (points and weights) are readily available so that any
desired accuracy can be obtained.
5.3.1
Gauss Quadrature for Unit Triangle (GQUTS)
In this section, Gaussian quadrature formula for triangles are computed in a

straight forward manner (GQUTS). The Gauss points are calculated simply for i
= 1, m and j = 1, n. Thus, the m n points Gaussian quadrature formula for
Eq.(5.12) is given by
II = 4
m X
n
X
i=1
mn
X
= 4{
(m)
(m)
1 + i
f{
2
j=1
(1 i
f (ur , vr )Gr }
(n)
)(1 + j ) 1 i(m)
(m)
(n)
}(
)Wi Wj
4
8
(5.13)
r=1
Here, (ur , vr ) are the new Gaussian points, Gr is the corresponding weights
(m)
(n)
for triangles where i , i = 1, m; j , j = 1, n are the m-th and n-th order
(m)
(n)
Gaussian quadrature points respectively. Wi , Wj are the corresponding m-th
and n-th order weights respectively.
71

Again, if we consider the integral II of Eq.(5.7) and substitute
u = (1
1+ 1+
)(
),
2
2
v=
1+
2
Then one can obtain (on the same lines of Eq.(5.12) and Eq.(5.13))
II = 4
=
f(
(1 + )(1 ) 1 + 1
,
)
dd
4
2
8
1 1
mn
X
4{
G0r f (u0r , vr0 )}
r=1
(5.14)
where G0r and (u0r , vr0 ) are respectively weights and Gaussian points for triangle.
All the Gaussian points and corresponding weights can be calculated simply using
the following algorithm:
step 1. r 1
step 2. i = 1, m
step 3. j = 1, n
(m)
(1 i
Gr =
8
(m)
(m)
(n)
(m)
Wi Wj ,
1 + i
ur =
2
(m)
(n)
(1 j ) (m) (n)
Wi Wj ,
=
8
step 4. compute step 3
G0r
u0r
(1 + i
vr =
(1 i
(n)
)(1 j )
,
4
(n)
)(1 + j )
4
(n)
vr0
1 + j
=
2
For clarity and reference, computed Gauss points and weights for different
values of m = n based on the above algorithm are listed in Table 5.1. Figure 5.2
and Figure 5.3 shows the distribution of Gaussian points for m = n = 10 and
72
Figure 5.2: Distribution of 10 10 Gauss Points using GQUTS.

m = n = 15. In Figure 5.2 and Figure 5.3, it is seen that there are crowdings
of Gauss points at terminal points in the u-direction within the triangle and that
is one of the major causes of error germane in the calculation. To avoid this
crowding further modification is done in the next section.
5.3.2
Gauss Quadrature for Unit Triangle (GQUTM)
It is clearly noticed in the Eq.(5.13) that for each i (i = 1, 2,..., m) j varies

from 1 to n and hence at the terminal value i = m there are n crowding points
(same situation is found for Eq.(5.14)). Table 5.1 and Figure 5.2 and Figure 5.3
shows the same feature. This situation can be prevailed over by using algebraic
73
Figure 5.3: Distribution of 15 15 Gauss Points using GQUTS.

manipulation, in Eq.(5.13) by making j dependent on i for the calculation of new
Gauss points and corresponding weights. To do so, we wish to calculate Gauss
points and weights for i = 1, m-1 and j = 1, n + 1 - i. The number of total points
is given by
TP =
n(n + 1) (n m + 1)(n m + 2)
;
2
2
nm
The new Gaussian quadrature formulae for Eq.(5.6) can be written as:
74
II = 4
m1
X n+1i
X
i=1
j=1
(m1)
(m1)
1 + i
f{
2
(1 i
(m1)
1 i
8
(m1)
)Wi
(n+1i)
Wj
= 4{
(n+1i)
)(1 + j
4
TP
X
f (pr , qr )Lr }
(5.15)
r=1
where (pr , qr ) are the new Gaussian points, Lr is the corresponding weights
for triangles. Similarly, the new Gaussian quadrature formulae for Eq.(5.7) can
be written as:
f{
II = 4
1 1
T
P
X
(1 + )(1 ) 1 + 1
,
}
d d
4
2
8
f (p0r , qr0 )L0r }
= 4{
(5.16)
r=1
where (p0r , qr0 ) and L0r are respectively the Gaussian quadrature points and
weights for triangle. All the Gaussian points and corresponding weights can be
calculated simply using the following algorithm:
step 1. r 1
step 2. i = 1, m 1
step 3. j = 1, n + 1 i
(m1)
(n+1i)
(m1)
(1 i
)(1 + j
1 + i
, qr =
pr =
2
4
(m1)
(1 i
) (m1) (n+1i)
Lr =
Wi
Wj
8
75

step 4. j = 1, m 1
step 5. i = 1, n + 1 j
(n+1j)
(m1)
(m1)
)(1 j
)
1 + j
=
, qr0 =
4
2
(m1)
(1 j
) (n+1j) (m1)
L0r =
Wi
Wj
8
r =r+1
p0r
(1 + i
step 6. compute step 3, step 2

Computed Gauss points and weights for different values of m(= n) based on
above algorithm listed in Table 5.2, Figure 5.4 and Figure 5.5 shows the distribution of Gaussian points for n = m = 10 i.e. 54-points and n = m = 15 i.e.
119-points formulae. Thus, the new Gaussian quadrature formulae is now obtained which is completely crowding free (as seen in Figure 5.4 and Figure 5.5)
and the total number points reduces significantly. In these figures, there are only
two points at the terminal side along u -direction, more points can be taken by
choosing n > m. For n = m, the technique GQUTS computes m m Gauss
1 points.
points whereas technique GQUTM calculates m(m+1)
2
5.4
1y
I1 =
(x + y) 2 dx dy
y=0
x=0
1y
y=0
(x + y) 2 dx dy
I2 =
x=0
76
Figure 5.4: Gauss Points for m =n =10 (54 points) using GQUTM.
Z
(x2 + y 2 ) 2 dx dy
I3 =
y=0
x=0
exp|x + y 1|dx dy
I4 =
y=0
x=0
The computed results are summarized in Table 5.3 for m m points GQUTS
and m(m+1)
1 points GQUTM formulae. In order to compare the results obtained
2
by methods GQT and IOUT are also included for all examples.
Some important remarks from the Table 5.3 are:
77
Figure 5.5: Gauss Points for m =n =15 (119 points) using GQUTM.
Usual Gauss quadrature (GQT) for triangles e.g. 7-point and 13point rules cannot evaluate the integral of non-polynomial functions
accurately.
Splitting unit triangle into quadrilaterals (IOUT) provides a way of
using Gaussian quadrature for square. But the convergence rate is
slow (as it needs the function to be calculated at more points) but
satisfactory in view of accuracy.
New Gaussian quadrature formulae for triangle (GQUTS and GQUTM)
are exact in view of accuracy and efficiency. GQUTM scheme needs
78

very less number of Gauss points in the calculation and the rate of
convergence is faster.
Again, we consider the following integrals of rational functions due to Rathod
and Karim [2002] to test the influences of formulae in Eq.(5.13), Eq.(5.14),
Eq.(5.15) and Eq.(5.16). These integrals arise in axisymmetric finite element
method with linear triangular element as well as in finite element formulations of
second order linear differential equations by use of quadratic triangular element
with two straight sides and one curved side. Consider
p,q
y=0
Example-1: I
r,0
Example-2: I
x=0
Example-3: I
1y
=
y=0
0,0
x=0
1y
y=0
0,r
1y
x=0
1y
=
y=0
x=0
1y
xp y q
dx dy
+ x + y
xr
dx dy
0.375 0.375 x
6= = 0
yr
dx dy
0.375 0.375 y
6= = 0
1
dx dy
12 + 21.53679831x 8.821067231y
6= 0, 6= 0
Example-4: I
0,0
=
y=0
x=0
1
dx dy
12 + 9.941125498(x + y)
= 6= 0
Results are summarized in Tables 5.4, 5.5, 5.6 and 5.7 for mm points GQUTS
and m(m+1)
1 points GQUTM formulae.
2
In Tables (5.4 - 5.7) for method GQUTS, Formula 1 is for Eq.(5.13) and
Formula 2 is for Eq.(5.14); for method GQUTM, Formula 1 is for Eq.(5.15) and
Formula 2 is for Eq.(5.16). These tables substantiated the influences of numerical
79

evaluation of the integrals as described before. Some important comments may
be drawn from the Tables (5.4 - 5.7).
xr
with 6= = 0 formula 1 in Eq.(5.13) and
For the integrand +x+y
Eq.(5.15) are more accurate and rate of convergence is higher. But the formula
in Eq.(5.15) requires very less computational effort.
yr
Similarly for the integrand +x+y
with 6= = 0 formula 2 in Eq. (5.14)
and Eq.(5.16) are more accurate and convergence is higher. Here also the formula
in Eq.(5.16) requires very less computational effort.
Similar influences of these formulae may be observed for different conditions
on , .
General Gaussian quadrature e.g. 7-point and 13-point rules and IOUT
cannot evaluate all of these type of integrals of rational functions accurately.
It is evident that the new formulae GQUTM given by Eq.(5.15) and Eq.(5.16)
are very fast and accurate in view of accuracy and equally applicable for any
geometry that is for different values of , and . We recommend this is appropriate quadrature scheme for triangular domain integrals encountered in science
and engineering.
We know the m-th order Gaussian quadrature formula can evaluate exactly
the integral of polynomial of degree 2m-1. The new methods are also tested
on the integral of all monomials xi y j where i , j are non-negative integers such
that i + j 28 = N . In Table 5.8 and 5.9, we present the absolute error over
1 points
corresponding monomials integrals for mm points GQUTS and m(m+1)
2
GQUTM formulae respectively. Table 5.10 and 5.11 presents the absolute error
over corresponding polynomial integrals for m m points GQUTS and m(m+1)
1
2
points GQUTM formulae respectively. It is observed that the new methods are
always accurate in view of both accuracy and efficiency and hence a proper balance
is observed.
5.4.1
Two Dimensional Elasticity Problem
From the principle of virtual work, the stiffness matrix of an arbitrary element is
given by
80
[B]T [D] [B] dA0
[K]e =
(5.17)
A0
Where [D] is a material property matrix, [B] is the strain displacement matrix
and the integration is performed over the element A0 in global coordinates. The
symmetric material property matrix [D] for anisotropic materials can be written
as
D11
D12
D13
D14
D12
D22
D23
D24
D13
D23
D33
D34
D14
D24
D34
D44
The element stiffness matrix of Eq.(5.17) can be written as

Z
[K]e = t
[B1 B2 B3 ...Bne ]T [D][B1 B2 B3 ...Bne ]dA0
We can visualize the [K]e as consisting of ne ne sub-matrices, [Kij ]e signifying

the stiffness relationship between nodes i and j, where ne is the number of nodes
in the element. Any such submatrix is given by
Z
[Kij ]e =
[Bi ]T [D][Bj ]dA0
A0
For a two dimensional element, the matrix [Bi ] has the form
Ni
x0
Ni
y 0
Ni
y 0
Ni
x0
In which Ni are the appropriate element shape functions. The element stiffness
relating nodes i and j for an anisotropic material under two dimensional loading
is given by:
[Kij ]e =
Kij11 Kij12
Kij21 Kij22
81

Where Kij11 , Kij12 , ...Kij22 are double integrals over domain A0 and they are double
integrals of the product of global derivatives of shape functions. Hence we need
to evaluate
ZZ
Ni Nj
dx0 dy 0
t s
t, s = x0 , y 0
A0
We shall evaluate these double integrals, where A0 is the triangle element with
two straight and one curved side. It can be shown that the evaluation of integrals
now amounts to the evaluation of integrals of the type
II
pq
xp y q
dx dy
J(x, y)
A
Z 1 Z 1x p q
xy
=
dx dy
x=0 y=0 J(x, y)
ZZ
(5.18)
Where J is the Jacobian of transformation and A is the unit triangle: {(x, y)|0
x, y 1, x + y 1}. The transformation which maps a one side curved triangle
A0 in to an orthogonal unit triangle A is given by
t=
m
X
Ni (x, y)ti ,
t = x0 , y 0 and
ti = (x0i , yi0 )
i=1
Where Ni are the corresponding shape functions for each nodes. Then for
quadratic curved element (m = 6, Fig 5.6) we have
x0 = x03 + (x01 x03 )x + (x02 x03 )y + (4x04 2x01 2x02 )xy
y 0 = y30 + (y10 y30 )x + (y20 y30 )y + (4y40 2y10 2y20 )xy
The shape functions Ni which transforms a quadratic curved triangular element to a orthogonal unit triangle are given in Appendix A. If the coordinates
are given 1(1, 0), 2(0, 1), 3(0, 0) and 5, 6 are at mid points of two straight sides,
then we have
82
x0 = x + 4xy(x04 0.5) y 0 = y + 4xy(y40 0.5)

and jacobian J(x, y, z) = 1 + 4x(y40 0.5) + 4y(x04 0.5)
Figure 5.6: Transformation of Quadratic Curved Triangular element A0 into Unit

triangle A.
By choosing x04 = y40 = 12 , we have computed the values of the integrals of the
1
Eq.(5.18) for different values of p, q using m m points GQUTS and m(m+1)
2
points GQUTM formulae and the results are summarized in Table 5.12 and 5.13.
These tables also proves the efficiency of the new quadrature formulae of this
chapter.
5.5
Conclusions
Though Gaussian quadrature formulae for the triangular domain integrals are
widely used in FEM and BEM, they are not so developed comparing the square
domain integrals. To achieve the desired accuracy of the triangular domain integral it was necessary to increase the number of points and corresponding weights.
Therefore, it is an important task to make a proper balance between accuracy and
efficiency of the calculations.
83

For the necessity of accurate evaluation of the triangular domain integrals,
this chapter presented new techniques to derive quadrature formulae utilizing the
Gaussian quadrature formulae for square domain integrals and that overcomes all
the difficulties pertinent in the derivation of higher order formulae. The first technique (GQUTS) presented m n points (for m, n>1) quadrature formulae utilizing the one dimensional m-point Gaussian quadrature formula using Unit triangle
in the transformation of geometry. Finally, in the second technique (GQUTM)
m(m+1)
1 points (for m= n>2) quadrature formula is derived utilizing the m2
point one dimensional Gaussian quadrature formula using Unit triangle for the
same geometrical transformation. It is observed that this scheme is appropriate
for the triangular domain integrals as it requires significantly less computational
effort for desired accuracy and removes all kinds of crowding of Gaussian points
in the calculation process. Through practical application examples, it is demonstrated that the new appropriate Gaussian quadrature formula for triangles are
accurate in view of accuracy and efficiency and hence a proper balance is observed.
Thus, we believe that the newly derived appropriate quadrature formulae for
triangles will ensure the accurate evaluation of the integrals in an efficient manner
and enhance the further utilization of triangular elements for numerical solution of
field problems in science and engineering. Further more, present methods are efficient and suitable for computer coding and complete computer program (program
-2: GQUTS.FOR and program -3: GQUTM.FOR) in FORTRAN are developed
and is also appended.
84
Table 5.1: Computed weights G and corresponding Gauss points (u, v) for n n
point method (GQUTS).
n
G
0.684643776713535D-01
0.109543004274166D+00
0.684643776713535D-01
0.617283950617284D-01
0.987654320987654D-01
0.617283950617284D-01
0.869611615580696D-02
0.139137858492911D-01
0.869611615580696D-02
0.408492975558259D-02
0.882401185479183D-02
0.120457917319127D-01
0.131855763377203D-01
0.120457917319127D-01
0.882401185479183D-02
0.408492975558259D-02
0.788426937775230D-02
0.170311096195916D-01
0.232494247307213D-01
0.254493080585855D-01
0.232494247307213D-01
0.170311096195916D-01
0.788426937775230D-02
0.868834290475827D-02
0.187680193729915D-01
0.256205064439922D-01
0.280447438447181D-01
0.256205064439922D-01
0.187680193729915D-01
0.868834290475827D-02
0.676492884474095D-02
0.146131796369908D-01
0.199486720263899D-01
0.218362349021241D-01
0.199486720263899D-01
0.146131796369908D-01
0.676492884474095D-02
0.367196988822002D-02
0.793196156673469D-02
0.108280404231932D-01
0.118526002080616D-01
0.108280404231932D-01
0.793196156673469D-02
0.367196988822002D-02
0.117014141080756D-02
0.252766688745083D-02
0.345055620900489D-02
0.377705121539608D-02
0.345055620900489D-02
0.252766688745083D-02
0.117014141080756D-02
0.106659360355737D-03
0.230398933768031D-03
0.314521061065587D-03
0.344281351761558D-03
0.314521061065587D-03
0.230398933768031D-03
0.106659360355737D-03
u
0.112701665379258D+00
0.112701665379258D+00
0.112701665379258D+00
0.500000000000000D+00
0.500000000000000D+00
0.500000000000000D+00
0.887298334620742D+00
0.887298334620742D+00
0.887298334620742D+00
0.254460438286208D-01
0.254460438286208D-01
0.254460438286208D-01
0.254460438286208D-01
0.254460438286208D-01
0.254460438286208D-01
0.254460438286208D-01
0.129234407200303D+00
0.129234407200303D+00
0.129234407200303D+00
0.129234407200303D+00
0.129234407200303D+00
0.129234407200303D+00
0.129234407200303D+00
0.297077424311301D+00
0.297077424311301D+00
0.297077424311301D+00
0.297077424311301D+00
0.297077424311301D+00
0.297077424311301D+00
0.297077424311301D+00
0.500000000000000D+00
0.500000000000000D+00
0.500000000000000D+00
0.500000000000000D+00
0.500000000000000D+00
0.500000000000000D+00
0.500000000000000D+00
0.702922575688699D+00
0.702922575688699D+00
0.702922575688699D+00
0.702922575688699D+00
0.702922575688699D+00
0.702922575688699D+00
0.702922575688699D+00
0.870765592799697D+00
0.870765592799697D+00
0.870765592799697D+00
0.870765592799697D+00
0.870765592799697D+00
0.870765592799697D+00
0.870765592799697D+00
0.974553956171379D+00
0.974553956171379D+00
0.974553956171379D+00
0.974553956171379D+00
0.974553956171379D+00
0.974553956171379D+00
0.974553956171379D+00
85
v
0.100000000000000D+00
0.443649167310371D+00
0.787298334620742D+00
0.563508326896291D-01
0.250000000000000D+00
0.443649167310371D+00
0.127016653792583D-01
0.563508326896291D-01
0.100000000000000D+00
0.247985426820927D-01
0.125945902810518D+00
0.289517979151782D+00
0.487276978085690D+00
0.685035977019597D+00
0.848608053360861D+00
0.949755413489286D+00
0.221575394388360D-01
0.112532875195889D+00
0.258684799487838D+00
0.435382796399849D+00
0.612080793311860D+00
0.758232717603808D+00
0.848608053360861D+00
0.178865986691016D-01
0.908417823768389D-01
0.208822428275864D+00
0.351461287844349D+00
0.494100147412834D+00
0.612080793311860D+00
0.685035977019597D+00
0.127230219143104D-01
0.646172036001514D-01
0.148538712155651D+00
0.250000000000000D+00
0.351461287844349D+00
0.435382796399849D+00
0.487276978085690D+00
0.755944515951914D-02
0.383926248234639D-01
0.882549960354370D-01
0.148538712155651D+00
0.208822428275864D+00
0.258684799487838D+00
0.289517979151782D+00
0.328850438978473D-02
0.167015320044137D-01
0.383926248234639D-01
0.646172036001514D-01
0.908417823768389D-01
0.112532875195889D+00
0.125945902810518D+00
0.647501146528088D-03
0.328850438978473D-02
0.755944515951914D-02
0.127230219143104D-01
0.178865986691016D-01
0.221575394388360D-01
0.247985426820927D-01
Table 5.2: Computed Gauss points (p, q) and corresponding weights L for
{ n(n+1)
1} point method GQUTM.
2
n
n=5
n=9
p
0.694318442029737D-01
0.694318442029737D-01
0.694318442029737D-01
0.694318442029737D-01
0.694318442029737D-01
0.330009478207572D+00
0.330009478207572D+00
0.330009478207572D+00
0.330009478207572D+00
0.669990521792428D+00
0.669990521792428D+00
0.669990521792428D+00
0.930568155797026D+00
0.930568155797026D+00
0.198550717512319D-01
0.198550717512319D-01
0.198550717512319D-01
0.198550717512319D-01
0.198550717512319D-01
0.198550717512319D-01
0.198550717512319D-01
0.198550717512319D-01
0.198550717512319D-01
0.101666761293187D+00
0.101666761293187D+00
0.101666761293187D+00
0.101666761293187D+00
0.101666761293187D+00
0.101666761293187D+00
0.101666761293187D+00
0.101666761293187D+00
0.237233795041836D+00
0.237233795041836D+00
0.237233795041836D+00
0.237233795041836D+00
0.237233795041836D+00
0.237233795041836D+00
0.237233795041836D+00
0.408282678752175D+00
0.408282678752175D+00
0.408282678752175D+00
0.408282678752175D+00
0.408282678752175D+00
0.408282678752175D+00
0.591717321247825D+00
0.591717321247825D+00
0.591717321247825D+00
0.591717321247825D+00
0.591717321247825D+00
0.762766204958164D+00
0.762766204958164D+00
0.762766204958164D+00
0.762766204958164D+00
0.898333238706813D+00
0.898333238706813D+00
0.898333238706813D+00
0.980144928248768D+00
0.980144928248768D+00
q
0.436530238707252D-01
0.214742881469342D+00
0.465284077898513D+00
0.715825274327684D+00
0.886915131926301D+00
0.465186775265609D-01
0.221103222500738D+00
0.448887299291690D+00
0.623471844265867D+00
0.371926177849334D-01
0.165004739103786D+00
0.292816860422638D+00
0.146726751310273D-01
0.547591690719464D-01
0.156037898816279D-01
0.803566392721822D-01
0.189476014677302D+00
0.331164789916112D+00
0.490072464124384D+00
0.648980138332656D+00
0.790668913571466D+00
0.899788288976586D+00
0.964541138367140D+00
0.178364709110403D-01
0.913306309413408D-01
0.213115003430640D+00
0.366773901111335D+00
0.531559337595478D+00
0.685218235276173D+00
0.807002607765473D+00
0.880496767795773D+00
0.194093822823562D-01
0.985756383301930D-01
0.226600619520678D+00
0.381383102479082D+00
0.536165585437487D+00
0.664190566627971D+00
0.743356822675808D+00
0.199794790791376D-01
0.100234137152044D+00
0.225261107830170D+00
0.366456213417655D+00
0.491483184095780D+00
0.571737842168687D+00
0.191525719105520D-01
0.942174931981956D-01
0.204141339376088D+00
0.314065185553979D+00
0.389130106841623D+00
0.164715798970249D-01
0.782894009149582D-01
0.158944394126877D+00
0.220762215144811D+00
0.114580133114576D-01
0.508333806465933D-01
0.902087479817289D-01
0.419587036543941D-02
0.156592013857925D-01
86
L
0.191734646470685D-01
0.387333412614470D-01
0.460377090452794D-01
0.387333412614470D-01
0.191734646470685D-01
0.379971476479502D-01
0.712356204997401D-01
0.712356204997401D-01
0.379971476479502D-01
0.298908447599701D-01
0.478253516159522D-01
0.298908447599701D-01
0.603805085320831D-02
0.603805085320831D-02
0.201598349766349D-02
0.448091604484195D-02
0.646435948463322D-02
0.774766276990871D-02
0.819147462543485D-02
0.774766276990871D-02
0.646435948463322D-02
0.448091604484195D-02
0.201598349766349D-02
0.505566374507052D-02
0.111063912872568D-01
0.156674725751545D-01
0.181135411193860D-01
0.181135411193860D-01
0.156674725751545D-01
0.111063912872568D-01
0.505566374507052D-02
0.774594695636761D-02
0.167323141055649D-01
0.228415344658821D-01
0.250028228175863D-01
0.228415344658821D-01
0.167323141055649D-01
0.774594695636761D-02
0.919182785685142D-02
0.193554244975459D-01
0.251043168357703D-01
0.251043168357703D-01
0.193554244975459D-01
0.919182785685142D-02
0.877088559745420D-02
0.177185350308217D-01
0.210599120522939D-01
0.177185350308217D-01
0.877088559745420D-02
0.647199750524138D-02
0.121334570276227D-01
0.121334570276227D-01
0.647199750524138D-02
0.314010549248654D-02
0.502416878797847D-02
0.314010549248654D-02
0.502474962829402D-03
0.502474962829402D-03
87
Exact Value
GQUTM
GQUTS
IOUT
GQT
Method
77
13 13
377
388
399
3 10 10
77
88
99
10 10
44
54
77
90
Points
I1
0.4001498818
0.4000451564
0.4000006727
0.4000003534
0.4000002024
0.4000001234
0.4000037510
0.4000020035
0.4000011460
0.4000006929
0.4000016184
0.4000009417
0.4000003700
0.4000002468
0.4
I2
0.6606860757
0.6637058258
0.6664256193
0.6665011716
0.6665481694
0.6665789279
0.6659893927
0.6662009982
0.6663329015
0.6664193644
0.6662596411
0.6663718426
0.6664974532
0.6665339400
0.6666667
I3
0.8315681219
0.8501738309
0.8755247201
0.8768203350
0.8777286724
0.8783900003
0.8696444210
0.8722479925
0.8740714951
0.8753981854
0.8725338709
0.8742865042
0.8765237986
0.8772635781
0.881373587
Test example
Table 5.3: Calculated values of the integrals I1 , I2 , I3 , I4 .

I4
0.6938790083
0.7238717079
0.7178753416
0.7179654297
0.7180285699
0.7180745324
0.7184323903
0.7185688310
0.7181265336
0.7182531970
0.7180036731
0.7175459725
0.7179128710
0.7180958213
0.71828183
Points
r=2
77
0.7288889289
GQT
13 13
0.7883351445
355
0.8536515855
366
0.8636423810
377
0.8699174296
IOUT
388
0.8741141888
399
0.8770583628
3 10 10
0.8792029273
Formula 1
Formula 2
55
0.8888888936 0.8189709473
66
0.8888888912 0.8386859037
77
0.8888888942 0.8511113215
GQUTS
88
0.8888888822 0.8594404172
99
0.8888888894 0.8652928433
10 10
0.8888888916 0.8695606956
14
0.8888888591 0.7979759423
44
0.8888888822 0.8620172475
GQUTM
77
0.8888888823 0.8738937177
90
0.8888889030 0.876131041
Exact
Value
0.8888888
Method
r=6
0.2209523767
0.2803986370
0.3457150818
0.3557058569
0.3619809270
0.3661776763
0.3691218499
0.3712664246
Formula 1
Formula 2
0.3809523857 0.3110344248
0.3809523670 0.3307493958
0.3809523811 0.3431748166
0.3809523803 0.3515039099
0.3809523861 0.3573563424
0.3809523859 0.3616241943
0.3809523785 0.2900394411
0.3809523803 0.3540807426
0.3809523815 0.365957215
0.3809523782 0.368194523
0.3809523
Computed value of I r,0

r=4
0.3733333349
0.4327795803
0.4980960396
0.5080868226
0.5143618757
0.5185586402
0.5215028018
0.5236473748
Formula 1
Formula 2
0.5333333316 0.4634153949
0.5333333289 0.4831303515
0.5333333215 0.4955557672
0.5333333368 0.5038848662
0.5333333261 0.5097372915
0.5333333260 0.5140051414
0.5333333287 0.4424203912
0.5333333368 0.5064616972
0.5333333366 0.5183381644
0.5333333378 0.5205754802
0.5333333
Table 5.4: Computed results of Example -1 for r=2, r=4, r=6.

88
Points
r=2
77
0.7288889289
GQT
13 13
0.7883350849
355
0.8536515855
366
0.8636423810
377
0.8699174296
IOUT
388
0.8741141888
399
0.8770583628
3 10 10
0.8792029273
Formula 1
Formula 2
55
0.8189709473 0.8888888936
66
0.8386859037 0.8888888912
77
0.8511113215 0.8888888942
GQUTS
88
0.8594404172 0.8888888822
99
0.8652928433 0.8888888894
10 10
0.8695606956 0.8888888916
14
0.7979759423 0.8888888591
44
0.8620172475 0.8888888822
GQUTM
77
0.8738937177 0.8888888823
90
0.8761310414 0.8888889030
Exact
Value
0.8888888
Method
r=6
0.2209523618
0.2803986370
0.3457150818
0.3557058569
0.3619809270
0.3661776763
0.3691218499
0.3712664246
Formula 1
Formula 2
0.3110344248 0.3809523857
0.3307493958 0.3809523670
0.3431748166 0.3809523811
0.3515039099 0.3809523803
0.3573563424 0.3809523861
0.3616241943 0.3809523859
0.2900394411 0.3809523785
0.3540807425 0.3809523803
0.3659572152 0.3809523814
0.3681945230 0.3809523782
0.3809523
Computed results of I 0,r .

r=4
0.3733333349
0.4327795803
0.4980960396
0.5080868226
0.5143618757
0.5185586402
0.5215028018
0.5236473748
Formula 1
Formula 2
0.4634153813 0.5333333316
0.4831303515 0.5333333289
0.4955557672 0.5333333215
0.5038848662 0.5333333368
0.5097372915 0.5333333261
0.5140051414 0.5333333260
0.4424203912 0.5333333287
0.5064616972 0.5333333368
0.5183381644 0.5333333366
0.5205754802 0.5333333378
0.5333333
Table 5.5: Computed values of Example-2 for r=2, r=4, r=6.

89
Table 5.6: Computed results of Example -3
GQUTM
GQUTS
IOUT
GQT
Method
Points
77
13 13
355
366
377
388
399
3 10 10
55
66
77
88
99
10 10
14
44
77
90
Computed results of I 0,0 .

0.03669412062
0.03688941523
0.03694724295
0.03694789912
0.03694799350
0.03694800728
0.03694800931
0.03694800972
Formula 1
Formula 2
0.0369262034 0.0369568441
0.0369429074 0.0369500918
0.0369467828 0.0369485115
0.0369477085 0.0369481329
0.0369479348 0.0369480402
0.0369479907 0.0369480173
0.0368514313 0.0369856322
0.0369477086 0.0369481216
0.0369480048 0.0369480110
0.0369480083 0.0369480100
Exact
Value
0.03694801040
90
GQUTM
GQUTS
IOUT
GQT
Method
Points
77
13 13
355
366
377
388
399
3 10 10
55
66
77
88
99
10 10
14
44
77
90

0.02731705643
0.02731722965
0.02731723353
0.02731723339
0.02731723359
0.02731723343
0.02731723344
0.02731723331
Formula 1
Formula 2
0.0273172332 0.0273172332
0.0273172336 0.0273172336
0.0273172332 0.0273172332
0.0273172333 0.0273172333
0.0273172334 0.0273172334
0.0273172333 0.0273172333
0.02731722858 0.02731722858
0.02731723355 0.02731723355
0.02731723345 0.02731723345
0.02731723357 0.02731723357
Exact
Value
0.02731723349
91
Table 5.8: The absolute error over corresponding monomials integrals for GQUTS.
N
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
17
19
20
21
22
23
24
25
26
27
28
i
1
0
1
2
3
3
4
3
3
7
6
7
8
10
3
11
4
6
6
7
8
15
14
11
11
6
j
0
2
2
2
2
3
3
5
6
3
5
5
5
4
12
6
15
14
15
15
15
9
11
15
16
22
m
2
2
3
3
5
5
6
6
7
7
7
8
8
9
9
10
11
11
12
12
13
13
14
14
15
15
Formula 1
0.11102230246251565404D-15
0.69388939039072283776D-16
0.93675067702747583098D-16
0.25153490401663702869D-16
0.24286128663675299322D-16
0.11926223897340548774D-16
0.69931040125315035993D-17
0.45265440701269810120D-17
0.26834003769016234742D-17
0.76300727888667374543D-17
0.13145951341386741262D-17
0.18821072087990553534D-17
0.10693790959085541781D-17
0.34711410178481227895D-17
0.97019153778507560837D-17
0.22703129964375418776D-18
0.14525556254227182940D-17
0.72970564916977889366D-19
0.62256921623191074922D-19
0.16136639389684463881D-19
0.67353181377130278426D-20
0.26850282683471278326D-20
0.86559281224845889752D-21
0.43010806913343210882D-21
0.28096998962953544725D-21
0.87461942068294381203D-20
92
Formula 2
0.13877787807814456755D-15
0.55511151231257827021D-16
0.62450045135165055399D-16
0.25153490401663702869D-16
0.32959746043559334794D-16
0.11817803680091998331D-16
0.80773061850170080334D-17
0.35507621148900270214D-17
0.19244588561617703704D-17
0.75487576259303246218D-17
0.14196272195982073683D-17
0.22107559923337238850D-17
0.12730905197232134096D-17
0.40056188075655863035D-17
0.99340024053984343766D-17
0.21292290711604974773D-18
0.12598556299223649668D-17
0.82638651916497676830D-19
0.69430231895250930918D-19
0.19764653556342043263D-19
0.77750841676921836300D-20
0.23076271138698089560D-20
0.82382019131453099994D-21
0.46376398030668342234D-21
0.29547149974335571368D-21
0.87056623568143397642D-20
Table 5.9:
GQUTM.
N
1
2
3
4
5
6
7
9
10
11
12
13
14
15
17
19
20
21
22
23
24
24
25
25
26
26
27
27
28
28
i
0
0
1
2
3
3
4
4
7
6
7
8
10
3
11
4
6
6
15
9
15
15
17
17
11
11
11
11
6
6
j
1
2
2
2
2
3
3
5
3
5
5
5
4
12
6
15
14
15
7
14
9
9
8
8
15
15
16
16
22
22
The absolute error over corresponding monomials integrals for
m
3
3
4
6
7
6
8
8
8
8
10
9
10
11
13
16
23
17
17
23
15
23
17
25
15
21
16
23
16
23
Formula 1
0.52735593669694935670D-15
0.31918911957973250537D-15
0.57176485768195561832D-14
0.23210600108569678923D-14
0.81141690588815151841D-15
0.35247412627503749150D-15
0.84947240214239272404D-16
0.22020029571447052863D-12
0.24570731733952744236D-16
0.20913072536459919454D-12
0.82127035546026606883D-14
0.36685927411276306534D-13
0.19663415267925790988D-12
0.13295543797322249835D-12
0.64774877142554824301D-14
0.96761867520779067109D-17
0.17882321354416373486D-18
0.89397710137790976567D-18
0.88817939246452559470D-17
0.54887569545956151366D-20
0.37408630675467081079D-16
0.80600478887167016639D-20
0.97180131397516505553D-17
0.95833009867331021272D-20
0.52781280261133280729D-17
0.99993211360589904835D-21
0.81391527216587962374D-18
0.25275537592323451890D-21
0.45687021572263586205D-17
0.10446463955932186506D-19
93
Formula 2
0.33306690738754696213D-15
0.97144514654701197287D-16
0.13357370765021414627D-14
0.23210600108569678923D-14
0.32916377956659914616D-15
0.35247412627503749150D-15
0.35517455946806886224D-11
0.33732556136388536228D-11
0.89810173187210295964D-11
0.15063337289479821618D-10
0.15025672712483265428D-12
0.23860034588321524327D-11
0.49008535029954492651D-12
0.85906081510531140388D-17
0.13019009425257495457D-14
0.60896692589019075365D-15
0.87985547396040447721D-18
0.91583551450074603259D-16
0.65983039410497443386D-18
0.96866985633022305866D-20
0.99060573491742491661D-17
0.32247636180379783905D-20
0.33752484509766879502D-18
0.46148406374333413981D-20
0.11767352280171164067D-16
0.74633663237304036071D-20
0.26984091943001830982D-17
0.93141829443178030296D-21
0.18615208877889429484D-15
0.73087735050746025737D-18
Table 5.10: Absolute error in M point formula for Polynomial of order N

(GQUTS).
N
2
4
6
8
10
12
14
16
18
20
23
26
30
3
5
7
9
10
12
14
16
18
20
22
23
25
27
29
30
M
2
3
4
5
6
7
8
9
10
11
12
13
14
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
15
Formula 1
0.13877787807814456755D-15
0.61062266354383609723D-15
0.56996074526693973894D-13
0.81185058675714572018D-15
0.70429773124658368033D-15
0.10373646386341306425D-14
0.23835100559921329477D-14
0.43966566498632175808D-14
0.65156213757688874466D-14
0.83275400464266624567D-14
0.49274820335121205517D-14
0.56465249143045070923D-14
0.34867941867133822598D-15
0.38172243144174444751D-12
0.22305768343500176343D-12
0.14502982148556498032D-12
0.10154724283673033369D-12
0.86659845965897375208D-13
0.64941108046667750386D-13
0.50034629217599047024D-13
0.39316640221276344391D-13
0.31355126828280788231D-13
0.25294870364955812647D-13
0.20589432936368723404D-13
0.18630062770252919790D-13
0.15331919761552015302D-13
0.12692104311984309106D-13
0.10559261798270824784D-13
0.96363889090511634095D-14
94
Formula2
0.13877787807814456755D-15
0.61062266354383609723D-15
0.56996074526693973894D-13
0.81532003370909933437D-15
0.70082828429463006614D-15
0.10338951916821770283D-14
0.23835100559921329477D-14
0.43957892881252291772D-14
0.65164887375068758502D-14
0.83284074081646508603D-14
0.49274820335121205517D-14
0.56473922760424954959D-14
0.34954678040932662952D-15
0.38172243144174444751D-12
0.22300217228377050560D-12
0.14503676037946888755D-12
0.10154377338977838008D-12
0.86659845965897375208D-13
0.64941108046667750386D-13
0.50029425047171116603D-13
0.39316640221276344391D-13
0.31356861551756765039D-13
0.25291400918003859033D-13
0.20588999255499729202D-13
0.18630062770252919790D-13
0.15332353442421009504D-13
0.12692971673722297510D-13
0.10559261798270824784D-13
0.96350878664441808041D-14

(GQUTM).
N
2
3
3
5
7
7
11
16
23
29
1
3
5
7
9
10
12
14
16
18
20
21
23
24
26
27
28
29
30
M
3
4
6
8
9
10
11
13
14
15
16
17
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
16
Formula 1
0.52735593669694935670D-15
0.34833247397614286456D-13
0.33834046675451645569D-13
0.40010356139319469548D-11
0.64681315858905463756D-11
0.12284062655965044542D-11
0.90869672897397890665D-12
0.87030209428018423523D-13
0.88775341244851091460D-13
0.96405956134804071667D-13
0.60945692936797968287D-12
0.28793634143653434876D-12
0.17497808757482857800D-12
0.11922407505693399798D-12
0.86691070988464957736D-13
0.75148220979315283330D-13
0.57768026473503653051D-13
0.45427203665404647381D-13
0.36325976948692328961D-13
0.29435655302112451182D-13
0.24149952870811119965D-13
0.21992824228433960343D-13
0.18524244638218334558D-13
0.17214962094724839403D-13
0.15642001582882869570D-13
0.15574347367319774094D-13
0.16233108607321966588D-13
0.17945714358980069392D-13
0.21206127132078478326D-13
95
Formula2
0.52735593669694935670D-15
0.34833247397614286456D-13
0.33834046675451645569D-13
0.40010356139319469548D-11
0.64681315858905463756D-11
0.12284062655965044542D-11
0.90869846369745488346D-12
0.87030209428018423523D-13
0.88775341244851091460D-13
0.96405956134804071667D-13
0.60945692936797968287D-12
0.28793634143653434876D-12
0.17497114868092467077D-12
0.11922407505693399798D-12
0.86691070988464957736D-13
0.75148220979315283330D-13
0.57768026473503653051D-13
0.45428938388880624188D-13
0.36325976948692328961D-13
0.29434787940374462778D-13
0.24149952870811119965D-13
0.21992824228433960343D-13
0.18524244638218334558D-13
0.17214962094724839403D-13
0.15642001582882869570D-13
0.15574347367319774094D-13
0.16233108607321966588D-13
0.17945714358980069392D-13
0.21206127132078478326D-13
M
2
3
4
5
6
7
8
9
10
11
12
13
14
15
POINTS
4
9
16
25
36
49
64
81
100
121
144
169
196
225
I22 2
0.274994667299803D-02
0.325156766835848D-02
0.326147362431669D-02
0.326151351705471D-02
0.326151369867892D-02
0.326151374374570D-02
0.326151362742850D-02
0.326151380595353D-02
0.326151374714880D-02
0.326151374399900D-02
0.326151376422602D-02
0.326151371351188D-02
0.326151391384561D-02
0.326151371024150D-02
I32 4
0.177094483171524D-03
0.223692365249628D-03
0.230577406022251D-03
0.227783881738088D-03
0.227781427793912D-03
0.227781433987425D-03
0.227781420418245D-03
0.227781441200401D-03
0.227781439860638D-03
0.227781433283100D-03
0.227781436303126D-03
0.227781431639375D-03
0.227781448332771D-03
0.227781430014434D-03
I52 3
0.546336186031229D-04
0.128392351031521D-03
0.111603259040227D-03
0.113382200902182D-03
0.113363432955252D-03
0.113363393267416D-03
0.113363383742831D-03
0.113363402109816D-03
0.113363382009822D-03
0.113363396969313D-03
0.113363395315162D-03
0.113363385540465D-03
0.113363408869501D-03
0.113363387592995D-03
2
Table 5.12: The values of integrals Ipq
using GQUTS
I82 8
0.130239293785839D-07
0.206906550387037D-06
0.116184871832071D-06
0.149778923092289D-06
0.140874043849123D-06
0.142527331787087D-06
0.142336737470806D-06
0.142346694835904D-06
0.142346752435195D-06
0.142346764992089D-06
0.142346772794248D-06
0.142346743090658D-06
0.142346775552142D-06
0.142346747743832D-06
96
M
3
4
5
6
7
8
9
10
11
12
13
14
15
16
POINTS
5
9
14
20
27
35
44
54
65
77
90
104
119
135
I22 2
0.275485757590948D-02
0.325158889113563D-02
0.326147298020953D-02
0.326151343707490D-02
0.326151371810622D-02
0.326151376345027D-02
0.326151359160585D-02
0.326151374400892D-02
0.326151364663877D-02
0.326151364910581D-02
0.326151370828131D-02
0.326151374783817D-02
0.326151391876362D-02
0.326151371009277D-02
I32 4
0.179182452053709D-03
0.223687133177605D-03
0.230575241486151D-03
0.227783627665424D-03
0.227781398791806D-03
0.227781436381915D-03
0.227781415052045D-03
0.227781429756894D-03
0.227781415126699D-03
0.227781414262975D-03
0.227781424787914D-03
0.227781435253389D-03
0.227781450366441D-03
0.227781424592687D-03
I52 3
0.545678818403744D-04
0.128396027303604D-03
0.111603887708712D-03
0.113382277228293D-03
0.113363444919348D-03
0.113363396118669D-03
0.113363381437413D-03
0.113363399280595D-03
0.113363379195043D-03
0.113363391444485D-03
0.113363391961321D-03
0.113363387385158D-03
0.113363411069091D-03
0.113363391435240D-03
2
using GQUTM
I82 8
0.171783947167217D-07
0.206903489038316D-06
0.115385158234197D-06
0.149694890639606D-06
0.140862213129016D-06
0.142525165812834D-06
0.142336270369235D-06
0.142346588190201D-06
0.142346683853969D-06
0.142346730172438D-06
0.142346767351738D-06
0.142346747351104D-06
0.142346795007845D-06
0.142346742313464D-06
97
Chapter 6
Accurate Evaluation
Schemes for Triangular
Domain Integrals
Chapter 6
Accurate Evaluation Schemes for
Triangular Domain Integrals
In this chapter an alternative higher order Gaussian quadrature formula for numerical integration over the triangular surfaces is derived using Standard Triangle
in the geometrical transformation. It presents n n points (for n > 1) and
n(n+1)
1 points (for n > 2) Gaussian quadrature formulae for triangle utilizing
2
1
n-point one dimensional Gaussian quadrature formulae. The proposed n(n+1)
2
points formulae reduces the total number of Gaussian points in the calculation
process and completely avoids the crowding of Gaussian points as well. To verify
the efficiency and accuracy, the method is applied to evaluate triangular domain
integrals of different types of functions and to the evaluation of integrals for element stiffness matrices in two dimensional elasticity problem. Thus, it overcomes
all the drawbacks in view of accuracy and efficiency for the numerical evaluation
of the triangular domain integrals of any arbitrary functions encountered in the
realm of science and engineering.
6.1
Introduction
Integration of a function over an arbitrary domain is one of the major concern

in FEM solution procedure. In the solution of practical problems most of the
encountered integrals either cannot be evaluated analytically or the evaluations
are very lengthy and tedious. Thus, numerical integration methods are preferred
99
6. Accurate evaluation Schemes for triangular domain integrals

where the methods extensively employ the Gaussian quadrature technique that
was originally designed for one dimensional cases. The Gaussian Integration procedure naturally extends to two and three dimensional rectangular/cuboid domains according to the notion of the cartesian products. Gaussian quadratures
are considered as the best method of integrating polynomials because they guarantee that they are exact for polynomials less than a specified degree. To obtain
the desired accuracy of the result, Gaussian integration points and weights necessarily increase for which computational difficulty arise in evaluating an arbitrary
domain integral. Since, the Gaussian quadrature formulae enforces the evaluation
of the function values at each Gaussian points, increasing the number of Gauss
points causes a severe problem due to the crowding of Gauss points within the
transformed domain.
In FEM solution procedure, irregular, distorted or complex type of domains
are subdivided into finite number of small and simple elements such as: quadrilateral, triangle, tetrahedron etc. The serious and notable difficulties of integrating an arbitrary function of two variables over the surface of the triangle were
first discussed by Hammer and Stroud [1956, 1958]; Hammer et al. [1956].
Bernardini [1991]; Cowper [1973]; Dunavant [1985]; Hillion [1977]; Lannoy
[1977]; Laurie [1977]; Laursen and Gellert [1978]; Lether [1976]; Lyness and
Jespersen [1975]; Reddy [1978] and Reddy and Shippy [1981] derived some
numerical integration formulae for triangles with precision limited up to degree
ten. Lague and Baldur [1977] proposed the product formulae based only on the
sampling points and weight coefficients of Gauss-Legendre quadrature rules. By
Lague and Baldur [1977] one can obtain numerical integration rules of very high
degree of precision as the derivation rely on standard Gauss-Legendre quadrature
rules. However, Lague and Baldur [1977] have not worked out explicitly weight
coefficients and sampling points for application to integrals over a triangular surface. Special integration schemes, e.g., reduced integration over quadrilaterals
have been successfully developed in Hughes [1987] and are widely used in commercial programs. There is no methodical way to design such approximate integration schemes for polygons with more than four sides. An attempt to distribute
the sampling points according to the governing perspective transformation fails
to assure the error order germane to the quadrature formula. The reason can be
100

traced to the crowding of quadrature points and this numerical computational
difficulty persists in all non-parallelogram polygonal finite elements (Rogers and
Adams [1990]). Rathod and Rao [1996, 1998]; Rathod et al. [2004] provided
this information in a systematic manner in their work.
The accuracy of a selected quadrature strategy is indicated by compliance with
the patch test proposed in Irons and Razzaque [1972]. An elegant systematic
procedure to yield shape functions for convex polygons of arbitrary number of sides
developed in Wachspress [1975]. An advantage is to be mentioned that there
is no difficulty with triangular elements as the exact shape functions are available and the quadrature formulas are also exact for the polynomial integrands
(Zienkiewicz and Taylor [1989]). A considerable amount of research has been
performed to attain perfect results of domain integration for plane quadrilateral
elements where numerical quadrature techniques are employed (Bathe [1996]).
Therefore, there has been considerable interest in the area of numerical integration schemes over triangles (Hammer et al. [1956]; Rathod and Karim [2002]).
It is explicitly shown in Lague and Baldur [1977]; Rathod and Karim [2002]
that the existing rules are not sufficient to evaluate the triangular domain integrals and for some element geometry these rules are not reliable also. Thus, the
high accuracy integration method for triangular element is demanded and it is
meaningful when the shape functions are the very best.
To fulfill all these demands with a proper balance between accuracy and ef 1 points higher order
ficiency this chapter proposed n n points and n(n+1)
2
Gaussian quadrature formulae to evaluate the triangular domain integrals using
standard triangle in the geometrical transformation. It is thoroughly investigated
that the n(n+1)
1 point formulae are appropriate in view of accuracy and effi2
ciency. This new method is totally free of crowding of quadrature points and it
uses significantly less number of Gauss points in the calculation process to obtain
the desired accuracy and that ensures the proper balance between accuracy and
efficiency. Thus, we believe that the formulae so presented will find better place
in numerical solution procedure of continuum mechanics problems.
101
6.2
Problem Statement
The solution of any physical phenomena using finite element method requires the
accurate evaluation of components of element matrices. This thesis intents to
derive accurate methods to evaluate the associated domain integrals for the accurate evaluation of the element matrices. Consider the two dimensional triangular
domain integral:
ZZ
f (x, y) dx dy;
I=
(6.1)
To perform the integration, we transform the arbitrary triangle to a orthogonal

Standard triangle. To do so, first map one vertex (vertex 1) to the (1, -1), the
second vertex (vertex 2) to point (-1, 1) and the third vertex (vertex 3) to point
(-1, -1). This geometrical transformation to Standard triangle (see Figure 6.1) is
done using shape functions given by:
Figure 6.1: Transformation of Arbitrary triangle into Standard triangle.
x
y
!
=
x1 x2 x3
y1 y2 y3
102
N1
N2
N3
(6.2)

where
1
N2 (s, t) = (t + 1),
2
1
N1 (s, t) = (s + 1),
2
1
N3 (s, t) = (s + t)
2
(6.3)
From Eq.(6.2) and Eq.(6.3), we obtain the transformation equation from the
arbitrary triangle into a standard triangle as:
1
1
1
x(s, t) = (x1 + x2 ) + (x1 x3 )s + (x2 x3 )t
2
2
2
1
1
1
y(s, t) = (y1 + y2 ) + (y1 y3 )s + (y2 y3 )t
2
2
2
(6.4)
and hence the Jacobian of this transformation is

1
J
(x, y)
= ((x1 x3 )(y2 y3 ) (x2 x3 )(y1 y3 )) =
(s, t)
4
4
(6.5)
where J = 2(Area of the original triangle). Using Eq.(6.4) and Eq.(6.5),

Eq.(6.1) reduces to
J
I=
4
s = 1
(6.6)
f (x(s, t), y(s, t)) ds dt
(6.7)
t = 1

J
I=
4
t = 1
s = 1
The evaluation of integrals I in Eq.(6.6) and in Eq.(6.7) by the existing Gaussian quadrature (i.e. 7-point and 13-point) will yield the same results. Thus, any
one of these two can be evaluated numerically. Influences of these integrals will
be studied later to present new quadrature formulae for triangles.
6.3
Numerical Evaluation Procedures
In this section, we wish to derive alternative techniques to evaluate the integral I

numerically and the new Gaussian quadrature formulae for triangles as the final
103

result of this chapter.
Using mathematical transformation equations:
(1 )(1 + )
1
(6.8)
2
the integral I of Eq.(6.6) is transformed into an integral over the surface of
the standard square {(, )| 1 , 1} and the Eq.(6.4) reduces to
s = ,
t=
1
1
1
(x1 + x2 ) + (x1 x3 ) + (x2 x3 )( 1)
2
2
4
1
1
1
y=
(y1 + y2 ) + (y1 y3 ) + (y2 y3 )( 1)
2
2
4
x=
(6.9)
Now the determinant of the Jacobian and the differential area are:
s t
s t
1
(s, t)
=
= (1 )
(,

2
ds dt = dt ds =
(s, t)
1
d d = (1 ) d d
(,
2
(6.10)
(6.11)
Now using Eq.(6.9)and Eq.(6.11) into Eq.(6.6), we get
(1 )(1 + )
(1 )(1 + )
1
1), y(,
1)}
dd
2
2
8
1 1
Z 1Z 1
(1 )(1 + )
1
1}
dd
(6.12)
= J
f {,
2
8
1 1
Z
I = J
f {x(,
In order to evaluate the integral I in Eq.(6.12) efficient Gaussian quadrature

coefficient (points and weights) are readily available so that any desired accuracy
can be readily obtained Lague and Baldur [1977]; Rathod and Karim [2002].
6.3.1
Gauss Quadrature using Standard Triangle (GQSTS)
In this section we are straightly computing Gaussian quadrature formulae for

standard triangles (GQSTS). The Gauss points are calculated simply for i = 1, m
and j = 1, n. Thus the m n points Gaussian quadrature formulae for Eq.(6.12)
104

gives
I = J
m X
n
X
(m)
(m)
f {i ,
(1 i
i=1 j=1
= J{
mn
X
(n)
(m)
)(1 + j )
1 i
(m)
(n)
1}(
)Wi Wj
2
8
f (sr , tr )Gr }
(6.13)
r=1
where (sr , tr ) are the new Gaussian points, Gr is the corresponding weights for
(m)
(n)
standard triangle where i , i = 1, m; j , j = 1, n are the m-th and n-th order
(m)
(n)
Gaussian quadrature points respectively. And Wi , Wj are the corresponding
m-th and n-th order weights respectively.
Again, if we consider the integral I of Eq.(6.7) and substitute
s=
(1 + )(1 )
1,
2
t=
Then one can obtain (on the same line of Eq.(6.13))
I = J
= J{
f(
1 1
mn
X
(1 + )(1 )
1
1, )
d d
2
8
f (s0r , t0r )G0r }
(6.14)
r=1
where G0r and (s0r , t0r ) are respectively weights and Gaussian points for triangle.
step 1. r 1
step 2. i = 1, m
step 3. j = 1, n
105

(m)
(n)
(m)
(1 i )(1 + j )
(1 i ) (m) (n)
(m)
Gr =
Wi Wj , sr = i ,
tr =
1
8
2
(n)
(m)
(n)
(1 j ) (m) (n)
(1 + i )(1 j )
(n)
0
0
Gr =
Wi Wj , sr =
1, t0r = j
8
2
r =r+1
Computed Gauss points and weights for different values of n( = m) based on
above algorithm listed in Table 6.1 and the distribution of Gaussian points for
m = n = 10 is plotted in Figure 6.2. It is seen that there are crowding of Gauss
points at the terminal points in the s-direction within the triangle and that is
one of the major causes of error germane in the calculation. We know, the Gauss
quadrature formulae enforces the evaluation of the function values at each Gauss
points within the transformed domain. Thus, increasing number of Gauss points
causes a severe problem due to crowding of Gaussian points. To avoid such type
of crowding further modification is needed and that will be described in the next
section.
6.3.2
Gauss Quadrature using Standard Triangle (GQSTM)
It is clearly noticed in the Eq.(6.13) that for each i (i = 1, 2,..., m) j varies from
1 to n and hence at the terminal value of i = m-1, m(say) there are respectively
n-1, n crowding points as shown in Table 6.1 and Figure 6.2. To overcome this
situation, algebraic manipulation is done in this section. We can use the advantage
of Eq.(6.12) by making j dependent on i for the calculation of new Gauss points
and corresponding weights. To do so, we wish to calculate Gauss points and
weights for i = 1, m-1 and j = 1, n + 1 - i. The number of total Gaussian
quadrature points within the standard triangle is then given by
n(n + 1) (n m + 1)(n m + 2)
;
nm
2
2
The new Gaussian quadrature formulae for Eq.(6.12) is then
TP =
106
Figure 6.2: Distribution of 10 10 Gauss Points using GQSTS.
I = J
m1
X n+1i
X
i=1
(m1)
(m1)
f {i
,
(1 i
j=1
(m1)
1 i
(
8
(m1)
(n+1i)
)Wi
Wj
(n+1i)
)(1 + j
2
= J{
TP
X
1}
f (pr , qr )Lr }
(6.15)
r=1
where (pr , qr ) are the new Gaussian points, Lr is the corresponding weights
for triangles.
Similarly, for Eq.(6.7) the new Gaussian quadrature formulae can be written
as
107
I = J
m1
X n+1j
X
j=1
(n+1j)
f{
(m1)
)(1 j
2
(1 + i
i=1
(m1)
1 j
{
8
(m1)
(n+1j)
Wj
Wi
= J{
(m1)
1, j
TP
X
f (p0r , qr0 )L0r }
(6.16)
r=1
where L0r and (p0r , qr0 ) are respectively weights and Gaussian points for triangle.
step 1. r 1
step 2. i = 1, m 1
step 3. j = 1, n + 1 i
(m1)
pr =
(m1)
,
i
qr =
(1 i
(n+1i)
)(1 + j
2
(m1)
) (m1) (n+1i)
(1 i
Wi
Wj
8
step 4. j = 1, m 1
Lr =
step 5. i = 1, n + 1 j
(n+1j)
(m1)
)(1 j
)
=
1,
2
(n+1j)
) (n+1j) (m1)
(1 i
L0r =
Wi
Wj
8
r =r+1
p0r
(1 + i
(m1)
qr0 = j

For clarity and reference, computed Gauss points and weights for different
values of n (= m) based on above algorithm listed in Table 6.2. The distribution
of Gaussian points within the standard triangle for m = 6, n = 8 i.e. 30-points are
108

plotted in Figure 6.3 and m = n = 10 i.e. 54-points are plotted in Figure 6.4. In
Figure 6.4, there are only two points at the terminal side along s -direction, more
points can be taken by choosing n > m as in Figure 6.3. For n > m, instead of
(nm+1)(nm+2)
calculating mn quadrature points we are now calculating n(n+1)
2
2
quadrature points, e.g. for n = 10, m = 8 we are calculating 49 instead of 80
quadrature points. For m = n, instead of calculating m m quadrature points
we are now calculating m(m+1)
1 quadrature points, e.g. for m = 10 we are
2
calculating 54 instead of 100 quadrature points. Thus, the obtained new Gaussian
quadrature formulae are now crowding free and number of points are significantly
reduced.
Figure 6.3: Distribution of Gauss Points for m=6, n=8 (30 points) GQSTM.
109
Figure 6.4: Distribution of Gauss Points for m=n=10 (54 points) GQSTM.
6.4
To show the accuracy, efficiency of the derived method and to compare the results
between the derived formulae of this thesis we are following the same examples
(of previous Chapter) with known results:
1y
I1 =
(x + y) 2 dx dy
y=0
1
x=0
1y
y=0
(x + y) 2 dx dy
I2 =
x=0
110

1
(x2 + y 2 ) 2 dx dy
I3 =
y=0
1
x=0
y
exp|x+y1| dx dy
I4 =
y=0
x=0
Computed values are summarized in Table 6.3 for m m points GQSTS and
1 points GQSTM formulae.
Some important remarks from the Table 6.3 are:
m(m+1)
2
Usual Gauss quadrature (GQT) for triangles e.g. 7-point and 13point rules cannot evaluate the integral of non-polynomial functions
accurately.
Splitting standard triangle into quadrilaterals (IOST) provides a way
of using Gaussian quadrature for square. But it need comparatively
a large number of Gauss points whereas the convergence rate is slow
but satisfactory in view of accuracy.
New Gaussian quadrature formulae for triangle (GQSTS and GQSTM)
are exact in view of accuracy and efficiency. Also (GQSTM) is faster
and needs significantly lesser number of Gauss point.
Again, we consider the following integrals of rational functions to test the
influences of formulae in Eq.(6.13), Eq.(6.14), Eq.(6.15) and Eq.(6.16). Consider
I
p,q
1y
y=0
Example-1: I
r,0
x=0
Example-2: I
0,r
Example-3: I
1y
=
y=0
x=0
1y
=
y=0
1y
=
y=0
0,0
xp y q
dx dy
+ x + y
x=0
x=0
xr
dx dy
0.375 0.375 x
yr
dx dy
0.375 0.375 y
1
dx dy
12 + 21.53679831x 8.821067231y
111
Example-4: I
0,0
1y
=
y=0
x=0
1
dx dy
12 + 9.941125498(x + y)
Results are summarized in Tables 6.4, 6.5, 6.6, 6.7 for m m points GQSTS
and m(m+1)
1 points GQSTM formulae.
2
In Tables 6.4, 6.5, 6.6, 6.7 for method GQSTS, Formula 1 is for Eq.(6.13) and
Formula 2 is for Eq.(6.14), for method GQSTM, Formula 1 is for Eq(6.15) and
Formula 2 is for Eq.(6.16). These tables substantiated the influences of numerical
evaluation of the integrals as described before. Some important comments may
be drawn from the Tables 6.4, 6.5, 6.6, 6.7.
xr
with 6= = 0 first formulae described in
Eq.(6.13) and Eq.(6.15) are more accurate and rate of convergence is higher. But
the new formula in Eq.(6.15) requires very less computational effort.
yr
with 6= = 0 second formula described
in Eq.(6.14) and Eq.(6.16) are more accurate and convergence is higher. Here also
the new formula in Eq.(6.16) requires very less computational effort.
Similar influences of these formulae in procedure-3 may be observed for
different conditions on , .
General Gaussian quadrature e.g. 7-point and 13-point rules and IOST
cannot evaluate all these integrals of rational functions accurately.
It is evident that the new formulae e.g. Eq.(6.15) and Eq.(6.16) are very fast
and accurate in view of accuracy and equally applicable for any geometry that is
for different values of , and . We recommend this is appropriate quadrature
scheme for triangular domain integrals encountered in science and engineering.
6.4.1
Two Dimensional Elasticity Problem
The two dimensional elasticity problem explained in section 5.4.1 of page 80 is

also tested with the present method. The evaluation of double integral over A0
(the triangle element with two Straight and one curved side) now amounts to the
evaluation of integrals of the type
112
II
pq
xp y q
dx dy
J(x, y)
U
Z 1 Z 1x p q
xy
=
dx dy
x=0 y=0 J(x, y)
ZZ
(6.17)
Where J is the Jacobean of transformation and U is the unit triangle: {(x, y)|0
x, y 1, x + y 1}. For cubic curved triangular element (m=10, Fig 6.5) the
transformation which maps a one side curved triangle A0 in to an orthogonal unit
triangle U is given by
t=
10
X
Ni (x, y)ti ,
t = x0 , y 0 and
ti = (x0i , yi0 )
i=1
We need to solve the integral of Eq.(6.17) using standard triangle element
Figure 6.5: Mapping of Cubic Curved element A0 to Standard triangle S.

For cubic curved triangular element (m=10, Fig 6.5)
t=
10
X
Ni (x, y)ti ,
t = x0 , y 0 and
i=1
113
ti = (x0i , yi0 )

This implies
9
x0 = x03 + (x01 x03 )x + (x02 x03 )y + (x01 x02 2x03 x04 x05 + 6x010 )xy
2
9 0
9
+
(x + 2x03 + 3x04 6x010 )x2 y + (x01 + 2x03 + 3x05 6x010 )xy 2
2 2
2
9
0
y 0 = y30 + (y10 y30 )x + (y20 y30 )y + (y10 y20 2y30 y40 y50 + 6y10
)xy
2
9 0
9
0
0
+
(y2 + 2y30 + 3y40 6y10
)x2 y + (y10 + 2y30 + 3y50 6y10
)xy 2
2
2
If the coordinates are given 1(1, 0), 2(0, 1), 3(0, 0) and nodes 6, 7, 8 and 9 are
at the trisection points of two straight sides of the curved triangle in A0 , then
9
x0 = x03 + (x01 x03 )x + (x02 x03 )y + (x01 x02 + x04 + x05 )xy
4
9
y 0 = y30 + (y10 y30 )x + (y20 y30 )y + (y10 y20 + y40 + y50 )xy
4
Using Appropriate transformation we have the relation between A0 and the
standard triangle S with vertices 1(1, 1), 2(1, 1), 3(1, 1) given by
x+1
y+1
(1 + R(
))
2
2
y+1
x+1
y0 =
(1 + R(
))
2
2
9
R =
(t1 t2 + t4 + t5 )
4
1
R
J(x, y) =
(1 + (x + y + 2))
4
2
x0 =
By choosing x04 = 0.8660254, y40 = 0.5 and x05 = 0.5, y50 = 0.8660254, we have
computed the values of the integrals of the Eq.(6.17) for different values of p, q
using m m points GQSTS and m(m+1)
1 points GQSTM formulae and the
2
results are summarized in Table 6.8 and 6.9.
We know the m-th order Gaussian quadrature formula can evaluate the integral
of polynomial of degree up to 2m-1 accurately. The presented methods of this
114

chapter are verified for the integral of all monomials xi y j where i, j are nonnegative integers such that i + j N = 30. In Table 6.10 and 6.11, we present the
absolute error over corresponding polynomial integrals for m m points GQSTS
1 points GQSTM formulae. The results are compared with the
and m(m+1)
2
results of GQUTS and GQUTM and it is observed that the new methods are
always accurate for monomial/polynomial functions in view of both accuracy and
efficiency and hence a proper balance is observed. Also, it is seen from the Tables
6.10 and 6.11 that for higher order polynomial functions the error decreases with
the increase of Gauss points.
6.5
Conclusions
This Chapter presented new techniques to derive quadrature formulae utilizing the
one dimensional Gaussian quadrature points and weights and that overcomes all
the difficulties pertinent in the derivation of the higher order formulae. The first
technique (GQSTS) presents m n point quadrature formulae for the triangles
utilizing the one dimensional quadrature points. Finally, in the second technique
(nm+1)(nm+2)
; n m points quadrature formulae are de(GQSTM), n(n+1)
2
2
rived utilizing the one dimensional Gaussian quadrature formulae. For m = n,
1 quadrature points are calculated by the formulae GQSTS
m m and m(m+1)
2
and GQSTM respectively. It is observed that the scheme GQSTM is appropriate
for the triangular domain integrals as it requires less computational effort for desired accuracy. Through practical application examples, it is demonstrated that
the new appropriate Gaussian quadrature formulae for triangles are accurate in
view of accuracy and efficiency and hence a proper balance is observed. Computer
codes in FORTRAN for these two formulae are appended (Program 4: GQSTS.for
and Program 5: GQSTM.for).
Thus, we believe that the newly derived appropriate quadrature formulae for
triangles will ensure the accurate evaluation of the integrals in an efficient manner
and enhance the further utilization of triangular elements for numerical solution
of field problems in science and engineering.
115
Table 6.1: Computed Gauss points (u, v) and corresponding weights G for n n
point method (GQSTS).
n
u
-0.77459666924148340428D+00
-0.77459666924148340428D+00
-0.77459666924148340428D+00
0.00000000000000000000D+00
0.00000000000000000000D+00
0.00000000000000000000D+00
0.77459666924148340428D+00
0.77459666924148340428D+00
0.77459666924148340428D+00
-0.94910791234275848627D+00
-0.94910791234275848627D+00
-0.94910791234275848627D+00
-0.94910791234275848627D+00
-0.94910791234275848627D+00
-0.94910791234275848627D+00
-0.94910791234275848627D+00
-0.74153118559939446008D+00
-0.74153118559939446008D+00
-0.74153118559939446008D+00
-0.74153118559939446008D+00
-0.74153118559939446008D+00
-0.74153118559939446008D+00
-0.74153118559939446008D+00
-0.40584515137739718416D+00
-0.40584515137739718416D+00
-0.40584515137739718416D+00
-0.40584515137739718416D+00
-0.40584515137739718416D+00
-0.40584515137739718416D+00
-0.40584515137739718416D+00
0.00000000000000000000D+00
0.00000000000000000000D+00
0.00000000000000000000D+00
0.00000000000000000000D+00
0.00000000000000000000D+00
0.00000000000000000000D+00
0.00000000000000000000D+00
0.40584515137739718416D+00
0.40584515137739718416D+00
0.40584515137739718416D+00
0.40584515137739718416D+00
0.40584515137739718416D+00
0.40584515137739718416D+00
0.40584515137739718416D+00
0.74153118559939446008D+00
0.74153118559939446008D+00
0.74153118559939446008D+00
0.74153118559939446008D+00
0.74153118559939446008D+00
0.74153118559939446008D+00
0.74153118559939446008D+00
0.94910791234275848627D+00
0.94910791234275848627D+00
0.94910791234275848627D+00
0.94910791234275848627D+00
0.94910791234275848627D+00
0.94910791234275848627D+00
0.94910791234275848627D+00
v
-0.80000000000000004441D+00
-0.11270166537925829786D+00
0.57459666924148344869D+00
-0.88729833462074170214D+00
-0.50000000000000000000D+00
-0.11270166537925829786D+00
-0.97459666924148335987D+00
-0.88729833462074170214D+00
-0.80000000000000004441D+00
-0.95040291463581461073D+00
-0.74810819437896391992D+00
-0.42096404169643553139D+00
-0.25446043828620812377D-01
0.37007195403919368459D+00
0.69721610672172240619D+00
0.89951082697857298598D+00
-0.95568492112232794611D+00
-0.77493424960822177017D+00
-0.48263040102432486211D+00
-0.12923440720030276996D+00
0.22416158662371921118D+00
0.51646543520761634127D+00
0.69721610672172240619D+00
-0.96422680266179672248D+00
-0.81831643524632213804D+00
-0.58235514344827121747D+00
-0.29707742431130146343D+00
-0.11799705174331709401D-01
0.22416158662371921118D+00
0.37007195403919368459D+00
-0.97455395617137918762D+00
-0.87076559279969723004D+00
-0.70292257568869853657D+00
-0.50000000000000000000D+00
-0.29707742431130146343D+00
-0.12923440720030276996D+00
-0.25446043828620812377D-01
-0.98488110968096176379D+00
-0.92321475035307232204D+00
-0.82349000792912596669D+00
-0.70292257568869853657D+00
-0.58235514344827121747D+00
-0.48263040102432486211D+00
-0.42096404169643553139D+00
-0.99342299122043054016D+00
-0.96659693599117268992D+00
-0.92321475035307232204D+00
-0.87076559279969723004D+00
-0.81831643524632213804D+00
-0.77493424960822177017D+00
-0.74810819437896391992D+00
-0.99870499770694387554D+00
-0.99342299122043054016D+00
-0.98488110968096176379D+00
-0.97455395617137918762D+00
-0.96422680266179672248D+00
-0.95568492112232794611D+00
-0.95040291463581461073D+00
116
G
0.27385751068541380748D+00
0.43817201709666231402D+00
0.27385751068541380748D+00
0.24691358024691342776D+00
0.39506172839506170646D+00
0.24691358024691342776D+00
0.34784464623227831503D-01
0.55655143397164555386D-01
0.34784464623227831503D-01
0.16339719022330379677D-01
0.35296047419167315973D-01
0.48183166927650789013D-01
0.52742305350881341086D-01
0.48183166927650789013D-01
0.35296047419167315973D-01
0.16339719022330379677D-01
0.31537077511009203867D-01
0.68124438478366217131D-01
0.92997698922885035921D-01
0.10179723223434195001D+00
0.92997698922885035921D-01
0.68124438478366217131D-01
0.31537077511009203867D-01
0.34753371619033066042D-01
0.75072077491965857887D-01
0.10248202577596891849D+00
0.11217897537887257242D+00
0.10248202577596891849D+00
0.75072077491965857887D-01
0.34753371619033066042D-01
0.27059715378963785509D-01
0.58452718547963133933D-01
0.79794688105559577473D-01
0.87344939608496471872D-01
0.79794688105559577473D-01
0.58452718547963133933D-01
0.27059715378963785509D-01
0.14687879552880067288D-01
0.31727846266938745934D-01
0.43312161692772820076D-01
0.47410400832246554770D-01
0.43312161692772820076D-01
0.31727846266938745934D-01
0.14687879552880067288D-01
0.46805656432302384165D-02
0.10110667549803324661D-01
0.13802224836019571369D-01
0.15108204861584317857D-01
0.13802224836019571369D-01
0.10110667549803324661D-01
0.46805656432302384165D-02
0.42663744142294604891D-03
0.92159573507212338376D-03
0.12580842442623490877D-02
0.13771254070462312334D-02
0.12580842442623490877D-02
0.92159573507212338376D-03
0.42663744142294604891D-03
1 point method GQSTM.
n
n=5
n=9
L
0.76693858588273891574D-01
0.15493336504578786306D+00
0.18415083618111743879D+00
0.15493336504578786306D+00
0.76693858588273891574D-01
0.15198859059180083730D+00
0.28494248199896055063D+00
0.28494248199896055063D+00
0.15198859059180083730D+00
0.11956337903988054072D+00
0.19130140646380897063D+00
0.11956337903988054072D+00
0.24152203412833226498D-01
0.24152203412833226498D-01
0.80639339906539760777D-02
0.17923664179367811145D-01
0.25857437938532890770D-01
0.30990651079634820575D-01
0.32765898501739394721D-01
0.30990651079634820575D-01
0.25857437938532890770D-01
0.17923664179367811145D-01
0.80639339906539760777D-02
0.20222654980282091242D-01
0.44425565149027380907D-01
0.62669890300618077972D-01
0.72454164477543922529D-01
0.72454164477543922529D-01
0.62669890300618077972D-01
0.44425565149027380907D-01
0.20222654980282091242D-01
0.30983787825470423399D-01
0.66929256422259622572D-01
0.91366137863528443241D-01
0.10001129127034502631D+00
0.91366137863528443241D-01
0.66929256422259622572D-01
0.30983787825470423399D-01
0.36767311427405684199D-01
0.77421697990183765969D-01
0.10041726734308135949D+00
0.10041726734308135949D+00
0.77421697990183765969D-01
0.36767311427405684199D-01
0.35083542389816799179D-01
0.70874140123286688553D-01
0.84239648209175455040D-01
0.70874140123286688553D-01
0.35083542389816799179D-01
0.25887990020965513660D-01
0.48533828110490902230D-01
0.48533828110490902230D-01
0.25887990020965513660D-01
0.12560421969946170689D-01
0.20096675151913882124D-01
0.12560421969946170689D-01
0.20098998513176078752D-02
0.20098998513176078752D-02
p
-0.86113631159405257254D+00
-0.86113631159405257254D+00
-0.86113631159405257254D+00
-0.86113631159405257254D+00
-0.86113631159405257254D+00
-0.33998104358485631282D+00
-0.33998104358485631282D+00
-0.33998104358485631282D+00
-0.33998104358485631282D+00
0.33998104358485631282D+00
0.33998104358485631282D+00
0.33998104358485631282D+00
0.86113631159405257254D+00
0.86113631159405257254D+00
-0.96028985649753628717D+00
-0.96028985649753628717D+00
-0.96028985649753628717D+00
-0.96028985649753628717D+00
-0.96028985649753628717D+00
-0.96028985649753628717D+00
-0.96028985649753628717D+00
-0.96028985649753628717D+00
-0.96028985649753628717D+00
-0.79666647741362683899D+00
-0.79666647741362683899D+00
-0.79666647741362683899D+00
-0.79666647741362683899D+00
-0.79666647741362683899D+00
-0.79666647741362683899D+00
-0.79666647741362683899D+00
-0.79666647741362683899D+00
-0.52553240991632899082D+00
-0.52553240991632899082D+00
-0.52553240991632899082D+00
-0.52553240991632899082D+00
-0.52553240991632899082D+00
-0.52553240991632899082D+00
-0.52553240991632899082D+00
-0.18343464249564978008D+00
-0.18343464249564978008D+00
-0.18343464249564978008D+00
-0.18343464249564978008D+00
-0.18343464249564978008D+00
-0.18343464249564978008D+00
0.18343464249564978008D+00
0.18343464249564978008D+00
0.18343464249564978008D+00
0.18343464249564978008D+00
0.18343464249564978008D+00
0.52553240991632899082D+00
0.52553240991632899082D+00
0.52553240991632899082D+00
0.52553240991632899082D+00
0.79666647741362683899D+00
0.79666647741362683899D+00
0.79666647741362683899D+00
0.96028985649753628717D+00
0.96028985649753628717D+00
117
n(n+1)
q
-0.91269395225854965670D+00
-0.57051423706131632052D+00
-0.69431844202973769242D-01
0.43165054865536878204D+00
0.77383026385260200719D+00
-0.90696264494687817148D+00
-0.55779355499852401046D+00
-0.10222540141661973223D+00
0.24694368853173420675D+00
-0.92561476443013324111D+00
-0.66999052179242812866D+00
-0.41436627915472301620D+00
-0.97065464973794535197D+00
-0.89048166185610722057D+00
-0.96879242023674416728D+00
-0.83928672145563554263D+00
-0.62104797064539618567D+00
-0.33767042016777559787D+00
-0.19855071751231800903D-01
0.29796027666531199607D+00
0.58133782714293258387D+00
0.79957657795317182980D+00
0.92908227673428056548D+00
-0.96432705817791941616D+00
-0.81733873811731838632D+00
-0.57376999313871901442D+00
-0.26645219777733009092D+00
0.63118675190956929910D-01
0.37043647055234596444D+00
0.61400521553094544736D+00
0.76099353559154647719D+00
-0.96118123543528766195D+00
-0.80284872333961398994D+00
-0.54679876095864454744D+00
-0.23723379504183550459D+00
0.72331170874973427232D-01
0.32838113325594298075D+00
0.48671364535161654175D+00
-0.96004104184172489589D+00
-0.79953172569591102192D+00
-0.54947778433966076861D+00
-0.26708757316468956233D+00
-0.17033631808439197997D-01
0.14347568433737456495D+00
-0.96169485617889594042D+00
-0.81156501360360888686D+00
-0.59171732124782483453D+00
-0.37186962889204100424D+00
-0.22173978631675383966D+00
-0.96705684020595017447D+00
-0.84342119817008365246D+00
-0.68211121174624533836D+00
-0.55847556971037892737D+00
-0.97708397337708474684D+00
-0.89833323870681347501D+00
-0.81958250403654209215D+00
-0.99160825926912121719D+00
-0.96868159722841506998D+00
118
Exact Value
GQSTM
GQSTS
IOST
GQT
Method
Test example
I1
I2
I3
I4
0.4001498818 0.6606860757 0.8315681219 0.6938790083
0.4000451564 0.66370582580 0.85017383098 0.72387170791
0.4000006727 0.6664256193 0.8755247201 0.7178753416
0.4000003534 0.6665011716 0.8768203350 0.7179654297
0.4000002024 0.6665481694 0.8777286724 0.7180285699
0.4000001234 0.6665789279 0.8783900003 0.7180745324
0.4000037510 0.6659893927 0.8696444210 0.7184323903
0.4000020035 0.6662009982 0.8722479925 0.7185688310
0.4000011460 0.6663329015 0.8740714951 0.7181265336
0.4000006929 0.6664193644 0.8753981854 0.7182531970
0.4000016184 0.6662596411 0.8725338709 0.7180036731
0.4000009417 0.6663718426 0.8742865042 0.7175459725
0.4000003700 0.6664974532 0.8765237986 0.7179128710
0.4000002468 0.6665339400 0.8772635781 0.7180958213
0.4
0.6666667
0.881373587
0.71828183
Points
77
13 13
377
388
399
3 10 10
77
88
99
10 10
44
54
77
90
Table 6.3: Calculated values of the integrals I1 , I2 , I3 , I4
Points
r=2
77
0.7288889289
GQT
13 13
0.7883351445
355
0.8536515855
366
0.8636423810
377
0.8699174296
IOST
388
0.8741141888
399
0.8770583628
3 10 10
0.8792029273
Formula 1
Formula 2
55
0.8888888936 0.8189709473
66
0.8888888912 0.8386859037
77
0.8888888942 0.8511113215
GQSTS
88
0.8888888822 0.8594404172
99
0.8888888894 0.8652928433
10 10
0.8888888916 0.8695606956
14
0.8888888591 0.7979759423
44
0.8888888822 0.8620172475
GQSTM
77
0.8888888823 0.8738937177
90
0.8888889030 0.876131041
Exact
Value
0.8888888
Method
r=6
0.2209523767
0.2803986370
0.3457150818
0.3557058569
0.3619809270
0.3661776763
0.3691218499
0.3712664246
Formula 1
Formula 2
0.3809523857 0.3110344248
0.3809523670 0.3307493958
0.3809523811 0.3431748166
0.3809523803 0.3515039099
0.3809523861 0.3573563424
0.3809523859 0.3616241943
0.3809523785 0.2900394411
0.3809523803 0.3540807426
0.3809523815 0.3659572152
0.3809523782 0.3681945230
0.3809523
Computed value of I r,0

r=4
0.3733333349
0.4327795803
0.4980960396
0.5080868226
0.5143618757
0.5185586402
0.5215028018
0.5236473748
Formula 1
Formula 2
0.5333333316 0.4634153949
0.5333333289 0.4831303515
0.5333333215 0.4955557672
0.5333333368 0.5038848662
0.5333333261 0.5097372915
0.5333333260 0.5140051414
0.5333333287 0.4424203912
0.5333333368 0.5064616972
0.5333333366 0.5183381644
0.5333333378 0.52057548020
0.5333333

119
Points
r=2
77
0.7288889289
GQT
13 13
0.7883350849
355
0.8536515855
366
0.8636423810
377
0.8699174296
IOST
388
0.8741141888
399
0.8770583628
3 10 10
0.8792029273
Formula 1
Formula 2
55
0.8189709473 0.8888888936
66
0.8386859037 0.8888888912
77
0.8511113215 0.8888888942
GQSTS
88
0.8594404172 0.8888888822
99
0.8652928433 0.8888888894
10 10
0.8695606956 0.8888888916
14
0.7979759423 0.8888888591
44
0.8620172475 0.8888888822
GQSTM
77
0.8738937177 0.8888888823
90
0.8761310414 0.8888889030
Exact
Value
0.8888888
Method
r=6
0.2209523618
0.2803986370
0.3457150818
0.3557058569
0.3619809270
0.3661776763
0.3691218499
0.3712664246
Formula 1
Formula 2
0.3110344248 0.3809523857
0.3307493958 0.3809523670
0.3431748166 0.3809523811
0.3515039099 0.3809523803
0.3573563424 0.3809523861
0.3616241943 0.3809523859
0.2900394411 0.3809523785
0.3540807425 0.3809523803
0.3659572152 0.3809523814
0.3681945230 0.3809523782
0.3809523

r=4
0.3733333349
0.4327795803
0.4980960396
0.5080868226
0.5143618757
0.5185586402
0.5215028018
0.5236473748
Formula 1
Formula 2
0.4634153813 0.5333333316
0.4831303515 0.5333333289
0.4955557672 0.5333333215
0.5038848662 0.5333333368
0.5097372915 0.5333333261
0.5140051414 0.5333333260
0.4424203912 0.5333333287
0.5064616972 0.5333333368
0.5183381644 0.5333333366
0.5205754802 0.5333333378
0.5333333

120
GQSTM
GQSTS
IOST
GQT
Method
Points
77
13 13
355
366
377
388
399
3 10 10
55
66
77
88
99
10 10
14
44
77
90

0.03669412062
0.03688941523
0.03694724295
0.03694789912
0.03694799350
0.03694800728
0.03694800931
0.03694800972
Formula 1
Formula 2
0.0369262034 0.0369568441
0.0369429074 0.0369500918
0.0369467828 0.369485115
0.0369477085 0.0369481329
0.0369479348 0.0369480402
0.0369479907 0.0369480173
0.0368514313 0.0369856322
0.0369477086 0.0369481216
0.0369480048 0.0369480110
0.0369480083 0.0369480100
Exact
Value
0.03694801040
121
GQSTM
GQSTS
IOST
GQT
Method
Points
77
13 13
355
366
377
388
399
3 10 10
55
66
77
88
99
10 10
14
44
77
90

0.02731705643
0.02731722965
0.02731723353
0.02731723339
0.02731723359
0.02731723343
0.02731723344
0.02731723331
Formula 1
Formula 2
0.0273172332 0.0273172332
0.0273172336 0.0273172336
0.0273172332 0.0273172332
0.0273172333 0.0273172333
0.0273172334 0.0273172334
0.0273172333 0.0273172333
0.02731722858 0.02731722858
0.2731723355 0.2731723355
0.02731723345 0.02731723345
0.02731723357 0.02731723357
Exact
Value
0.02731723349
122
M POINTS
I22 2
I32 4
I52 3
I82 8
3
9
0.325947151477D-02 0.224256520359D-03 0.128714495542D-03 0.207440393085D-06
4
16
0.326936162780D-02 0.231153940689D-03 0.111884992164D-03 0.116484848575D-06
5
25
0.326940113312D-02 0.228354076715D-03 0.113667897705D-03 0.150166858783D-06
6
36
0.326940130900D-02 0.228351635765D-03 0.113649141988D-03 0.141238492500D-06
7
49
0.326940135410D-02 0.228351642100D-03 0.113649102677D-03 0.142896199680D-06
8
64
0.326940123749D-02 0.228351628498D-03 0.113649093133D-03 0.142705081674D-06
9
81
0.326940141646D-02 0.228351649332D-03 0.113649111547D-03 0.142715068831D-06
10
100
0.326940135750D-02 0.228351647989D-03 0.113649091396D-03 0.142715126336D-06
11
121
0.326940135435D-02 0.228351641395D-03 0.113649106393D-03 0.142715138925D-06
12
144
0.326940137462D-02 0.228351644422D-03 0.113649104734D-03 0.142715146747D-06
13
169
0.326940132378D-02 0.228351639748D-03 0.113649094935D-03 0.142715116967D-06
14
196
0.326940152460D-02 0.228351656481D-03 0.113649118323D-03 0.142715149511D-06
15
225
0.326940132050D-02 0.228351638118D-03 0.113649096993D-03 0.142715121632D-06
2
using GQSTS
123
M POINTS
I22 2
I32 4
I52 3
I82 8
4
9
0.325949245527D-02 0.224251251839D-03 0.128718170385D-03 0.207437326110D-06
5
14
0.326936098818D-02 0.231151779323D-03 0.111885618075D-03 0.115682730958D-06
6
20
0.326940105419D-02 0.228353823362D-03 0.113667973621D-03 0.150082546398D-06
7
27
0.326940132861D-02 0.228351606865D-03 0.113649153897D-03 0.141226615819D-06
8
35
0.326940137387D-02 0.228351644526D-03 0.113649105522D-03 0.142894025164D-06
9
44
0.326940120159D-02 0.228351623122D-03 0.113649090819D-03 0.142704612880D-06
10
54
0.326940135436D-02 0.228351637861D-03 0.113649108710D-03 0.142714961834D-06
11
65
0.326940125673D-02 0.228351623193D-03 0.113649088574D-03 0.142715057565D-06
12
77
0.326940125921D-02 0.228351622326D-03 0.113649100854D-03 0.142715104012D-06
13
90
0.326940131853D-02 0.228351632878D-03 0.113649101372D-03 0.142715141290D-06
14
104
0.326940135820D-02 0.228351643370D-03 0.113649096785D-03 0.142715121238D-06
15
119
0.326940152954D-02 0.228351658522D-03 0.113649120528D-03 0.142715169019D-06
16
135
0.326940132035D-02 0.228351632683D-03 0.113649100845D-03 0.142715116188D-06
2
using GQSTM
124
Table 6.10: Absolute error in M point formula for Polynomial of order N (GQSTS).
N
2
4
6
8
10
12
14
16
18
20
23
26
30
3
6
8
10
11
14
17
19
21
23
25
27
29
30
M
2
3
4
5
6
7
8
9
10
11
12
13
14
15
15
15
15
15
15
15
15
15
15
15
15
15
15
Formula 1
0.13877787807814456755D-15
0.61062266354383609723D-15
0.56996074526693973894D-13
0.81185058675714572018D-15
0.70429773124658368033D-15
0.10356299151581538354D-14
0.23835100559921329477D-14
0.43966566498632175808D-14
0.65156213757688874466D-14
0.83275400464266624567D-14
0.49274820335121205517D-14
0.56465249143045070923D-14
0.34867941867133822598D-15
0.38172243144174444751D-12
0.17785078965104617055D-12
0.12044532038402167018D-12
0.86659845965897375208D-13
0.74709335939893151135D-13
0.50034629217599047024D-13
0.35053557279063340957D-13
0.28127673801225938632D-13
0.22797735921287198835D-13
0.18630062770252919790D-13
0.15331919761552015302D-13
0.12692104311984309106D-13
0.10559261798270824784D-13
0.96363889090511634095D-14
125
Formula 2
0.13877787807814456755D-15
0.61062266354383609723D-15
0.56996074526693973894D-13
0.81532003370909933437D-15
0.70082828429463006614D-15
0.10338951916821770283D-14
0.23835100559921329477D-14
0.43957892881252291772D-14
0.65164887375068758502D-14
0.83284074081646508603D-14
0.49274820335121205517D-14
0.56473922760424954959D-14
0.34954678040932662952D-15
0.38172243144174444751D-12
0.17785078965104617055D-12
0.12045919817182948464D-12
0.86659845965897375208D-13
0.74712805386845104749D-13
0.50029425047171116603D-13
0.35053557279063340957D-13
0.28127673801225938632D-13
0.22797735921287198835D-13
0.18630062770252919790D-13
0.15332353442421009504D-13
0.12692971673722297510D-13
0.10559261798270824784D-13
0.96350878664441808041D-14

(GQSTM).
N
2
3
3
5
5
7
7
11
16
18
19
20
21
22
23
24
25
26
27
28
29
21
26
27
28
29
30
M
3
4
6
8
9
10
11
11
13
14
15
15
15
15
15
15
15
15
15
15
15
17
18
21
22
22
22
Formula 1
0.52735593669694935670D-15
0.34833247397614286456D-13
0.33834046675451645569D-13
0.40010356139319469548D-11
0.92428148468215454159D-12
0.64681315858905463756D-11
0.31802338540387609100D-12
0.90869672897397890665D-12
0.87030209428018423523D-13
0.36290415117434804415D-13
0.25661764380124907348D-13
0.23849845709467132338D-13
0.22624263573689518125D-13
0.22124663212608197682D-13
0.22604314253715784844D-13
0.24489958672102574155D-13
0.28469414325993369630D-13
0.35621679217445745280D-13
0.47618159415563354742D-13
0.66877926807595855507D-13
0.96405956134804071667D-13
0.23984286778855334887D-13
0.19719469113166354646D-13
0.19230277092940895045D-13
0.18625292280693983571D-13
0.17502926191736989381D-13
0.16464260510495876133D-13
126
Formula 2
0.52735593669694935670D-15
0.34833247397614286456D-13
0.33834046675451645569D-13
0.40010356139319469548D-11
0.92428842357605844882D-12
0.64681315858905463756D-11
0.31802338540387609100D-12
0.90869846369745488346D-12
0.87030209428018423523D-13
0.36290415117434804415D-13
0.25661764380124907348D-13
0.23849845709467132338D-13
0.22624263573689518125D-13
0.22124663212608197682D-13
0.22604314253715784844D-13
0.24489958672102574155D-13
0.28469414325993369630D-13
0.35621679217445745280D-13
0.47618159415563354742D-13
0.66877926807595855507D-13
0.96405956134804071667D-13
0.23984286778855334887D-13
0.19719469113166354646D-13
0.19230277092940895045D-13
0.18625292280693983571D-13
0.17502926191736989381D-13
0.16464260510495876133D-13
Chapter 7
A Symmetrical
Extended Gaussian
Quadrature Formula
for Evaluation of
Triangular Domain
Integrals
Chapter 7
A Symmetrical Extended
Gaussian Quadrature Formula for
Evaluation of Triangular Domain
Integrals
This chapter is devoted to present symmetrical quadrature formulae for triangular
1
domain integrals. As a result, it presents n n points (for n > 1) and n(n+1)
2
points (for n > 2) quadrature formulae in which the last one is totally free of
crowding of Gaussian quadrature points and weights. By the suitable transformation of a triangle in global space into its contiguous space, Gauss points and
weights are computed which are symmetric about the line of symmetry. For clarity
and reference, Gaussian integration points and weights for different values of n are
presented in tabular form. The efficiency and accuracy of the schemes are tested
through application examples. Finally, an error formula also presented to evaluate
the error in monomial/polynomial integration using m n points method successfully. The error calculated by the new error formula and the error in calculation
of integrals by the proposed methods are found in good agreement.
128
7. Symmetrical Gaussian Quadrature for Triangular Domain Integrals
7.1
Introduction
The basic problems of integrating an arbitrary function of two variables over the
surface of the triangle were first described by Hammer and Stroud [1956, 1958];
Hammer et al. [1956]. In FEM solution of real word problems, the integrands
in practical situations are either polynomials or rational expressions. To get the
desired accuracy for the polynomial of higher degree it is necessary to increase
the Gaussian quadrature points with corresponding weights. If the integrands are
rational functions of bivariate or trivariate polynomial numerator with linear or
nonlinear denominator then more Gaussian points are needed. It is well known
that the first step of the Gaussian quadrature rule is to evaluate the integrands
at all Gaussian points which require more computing time before the application
of the weighted sum. In addition, specially for two and three dimensional cases
the higher order (more Gauss points and weights) formulae become erroneous due
to crowding of Gauss points and that effects the accuracy and efficiency of the
calculations.
A table of Gaussian quadrature formulae with symmetrically placed integration points is provided by Cowper [1973]. A detailed study of symmetric quadrature rules by formulating the problem in polar coordinates is made by Lyness and
Jespersen [1975]. Some numerical integration formulae for triangles with precision limited up to degree ten are derived in Dunavant [1985]; Hillion [1977]; Lannoy [1977]; Laurie [1977]; Laursen and Gellert [1978]; Lether [1976]; Reddy
[1978]; Reddy and Shippy [1981] and it is not likely that these techniques can be
extended much further to give a greater accuracy which may be demanded in future. Lague and Baldur [1977] proposed the product formulae based only on the
sampling points and weight coefficients of Gauss-Legendre quadrature rules. So,
numerical integration rules of very high degree of precision can be obtained as the
derivation rely on standard Gauss-Legendre quadrature rules. However, Lague
and Baldur [1977] have not worked out explicitly the weight coefficients and sampling points for the higher order quadrature rules for triangles solely. Bernardini
[1991]; Cools [2003]; Cools and Rabinowitz [1993]; Lyness and Cools [1994];
Rathod and Rao [1996, 1998]; Rathod et al. [2004] provided these informations
in a systematic manner. But, they have not considered the symmetry as well as
129

the crowding of the Gauss points of the quadrature rules. The principal drawback
in the symmetric quadrature scheme of Wandzura and Xiao [2003] was that,
one has to manually adjust the annealing parameters several times, before the
process yields a satisfactory initial approximation of weights and abscissae, also
it provides only 6 type of quadrature rules of order up to 30 over triangles. Similar disadvantages are also found in the quadrature formulae given by Dunavant
[1985]; Taylor et al. [2000].
The main aim of this chapter is to present symmetrical extended Gaussian
quadrature formulae avoiding the crowd of Gaussian integration points and weights
in the calculation process. First, it propose a numerical integration scheme to evaluate the triangular domain integral employing Gaussian quadrature schemes for
square domain integrals. This scheme is used as a tool for testing the accuracy
for the derived numerical integration formulae for triangular domain integrals.
Secondly, we presents two types of extended quadrature n n point (for n > 1)
1 point (for n > 2) formulae for which Gauss points are symmetrical
and n(n+1)
2
about the line of symmetry. It is easy to observe that n n point formulae give
rise to huge crowding of Gauss points but n(n+1)
1 point formulae are totally
2
free of such crowding. Through application examples it is demonstrated that the
1 point
formulae so presented are accurate in view of accuracy and the n(n+1)
2
formula is faster as it utilizes minimum number of Gauss points and weights in
the calculation process.
An error formula is described in this chapter to calculate the error in two
dimensional domain integral. The error calculated by the new error formula and
the error in the resultant integrals of the proposed methods are found in good
agreement. Therefore, a proper balance between accuracy and efficiency is ensured
for the presented quadrature scheme.
7.2
Problem Statement
It is well known that the domain integrals frequently encountered in many areas of
continuum mechanics. In two dimensional case, any arbitrary domain can be subdivided using triangles as shown in Figure 7.1. Thus, an arbitrary domain integral
renders to triangular domain integrals which may be evaluated numerically.
130
Figure 7.1: Discretization of the Domain of Integration into Triangles.

Mathematically, the integral
ZZ
f (x, y) dx dy
D
can be expressed as
ZZ
f (x, y) dx dy =
n ZZ
X
i=1
f (x, y) dx dy
where i are the triangles used in the mesh. Therefore, derivation of numerical
integration scheme is now required to evaluate the integral
ZZ
f (x, y) dx dy;
I=
(7.1)
accurately in view of accuracy and efficiency. In the following sections, we

wish to present such numerical integration schemes.
7.3
Integration Over Arbitrary Triangle (IOAT)
By use of the property of definite integrals, the triangular domain integrals can be
expressed as the sum of three quadrilateral domain integrals as shown in Figure
7.2. In this chapter we are dividing the original shape of the triangle without any
131

type of transformation. Hence, the integral of Eq. (7.1) is expressed as
I=
3 ZZ
X
i=1
f (x, y) dx dy = I1 + I2 + I3
(7.2)
ei
where
ZZ
f (x, y) dx dy
I1 =
e
Z Z1
f (x, y) dx dy
I2 =
e
Z Z2
I3 =
f (x, y) dx dy
e3
By using the transformation equations

x=
4
X
xi i (, )
y=
i=1
4
X
yi i (, )
i=1
where i (, ) are the standard shape functions for the quadrilateral elements.
Now, the integrals I1 , I2 , I3 reduces to
I1
I2
I3
Z Z
J 1 1
=
f (X1 , Y1 )(4 + ) dx dy
96 1 1
Z Z
J 1 1
f (X2 , Y2 )(4 ) dx dy
=
96 1 1
Z Z
J 1 1
=
f (X3 , Y3 )(4 + ) dx dy
96 1 1
(Above derivation are on the same line of Eq.(3.12) to Eq.(3.16) in page number
47 to 49.)
Each quadrilaterals in Figure 7.2 is now transformed into its equivalent 2square in {(, ) | 1 , 1} in (, ) space and that results the equivalent
integral I in Eq.(7.3).
132
Figure 7.2: Arbitrary Triangle Subdivided into Quadrilaterals
I =
3 ZZ
X
i =1
f (x, y)dxdy
ei
1
Z 1
J
=
[f (X1 , Y1 )(4 + ) + f (X2 , Y2 )(4 + + )
96 1 1
+ f (X3 , Y3 )(4 )]d d
Z
(7.3)
where J = (x1 x3 )(y2 y3 ) (x2 x3 )(y1 y3 ) = 2 Area of the original triangle

and
X1 =
1
[a
24 11
+ a12 + a13 + a14 ] Y1 =
1
[b
24 11
+ b12 + b13 + b14 ]
X2 =
1
[a
24 21
+ a22 + a23 + a24 ],
Y2 =
1
[b
24 21
+ b22 + b23 + b24 ]
X3 =
1
[a
24 31
+ a32 + a33 + a34 ],
Y3 =
1
[b
24 31
+ b32 + b33 + b34 ]

(7.4)
133
a11 = 14x1 + 5x3 + 5x2
b11 = 14y1 + 5y2 + 5y3
a12 = 4x1 + x2 5x3
b12 = 4y1 + y2 5y3
a13 = 4x1 + 5x2 x3
b13 = 4y1 + 5y2 y3
a14 = 2x1 + x2 + x3
b14 = 2y1 + y2 + y3
a21 = 5x1 + 14x2 + 5x3
b21 = 5y1 + 14y2 + 5y3
a22 = x1 + 4x2 5x3
b22 = y1 + 4y2 5y3
a23 = 5x1 + 4x2 + x3
b23 = 5y1 + 4y2 + y3
a24 = x1 + 2x2 x3
b24 = y1 + 2y2 y3
a31 = 5x1 + 5x2 + 14x3
b31 = 5y1 + 5y2 + 14y3
a32 = 5x1 x2 4x3
b32 = 5y1 y2 4y3
a33 = x1 + 5x2 4x3
b33 = y1 + 5y2 4y3
a34 = x1 x2 + 2x3
b34 = y1 y2 + 2y3
Now, the right hand side of Eq.(7.3) with Eq.(7.4) can be evaluated by use of
available higher order Gaussian quadrature for square. Since the triangle is divided
into three quadrilaterals and each quadrilateral is transformed into equivalent 2square, for m n points Gaussian quadrature formula, total number of points
within the triangle is 3 m n. Figure 7.3 shows the distribution of Gauss points
within the triangle for m = n = 9. This scheme will ensure the desired accuracy
of the evaluation of the triangular domain integral derived later in this chapter.
7.4
Transformation of Triangular Domain
In FEM, integration over triangular domains is usually carried out in normalized

co-ordinates. To perform the integration, first map one vertex (vertex 1) to the
(1,0), the second vertex (vertex 2) to point (-1,1) and the third vertex (vertex
134
Figure 7.3: Distribution of 3 9 9 points using IOAT.

3) to point (-1, -1)), (see Fig 7.4). This geometrical transformation of domain is
done using shape functions as:
x
y
!
=
x1 x2 x3
y1 y2 y3
N1
N2
N3
(7.5)
where
1
N1 (u, v) = (1 + u),
2
1
N2 (u, v) = (1 u + 2v),
4
1
N3 (u, v) = (1 u 2v) (7.6)
4
The original and the transformed triangles are shown in Figure 7.4.
From Eq.(7.5) using Eq.(7.6), we obtain the transformation equations
135
Figure 7.4: Original and Transformed triangle.
1
{(2x1 + x2 + x3 ) + (2x1 x2 x3 )u + (2x2 2x3 )v}
4
1
y(u, v) =
{(2y1 + y2 + y3 ) + (2y1 y2 y3 )u + (2y2 2y3 )v}
4
x(u, v) =
(7.7)
and the Jacobian of this transformation is hence

1
J
(x, y)
= (x1 x3 )(y2 y3 ) (x2 x3 )(y1 y3 ) =
(u, v)
4
4
(7.8)
Finally Eq.(7.1) reduces to

J
I=
4
(1u)/2

u = 1
(7.9)
v = (u1)/2
Here, we wish to present a new technique to calculate Gaussian points and corresponding weights for triangle. Consequently, new Gaussian quadrature formulae
will be presented as the final result of this study.
Using mathematical transformation equations:
136
(1 )
(7.10)
2
the integral I of Eq.(7.9) is transformed into an integral over the surface of
the standard square {(, )| 1 , 1} and the Eq.(7.7) reduces to
u = ,
v=
1
{(2x1 + x2 + x3 ) + (2x1 x2 x3 ) + (x2 x3 )(1 )}
4
1
y=
{(2y1 + y2 + y3 ) + (2y1 y2 y3 ) + (y2 y3 )(1 )} (7.11)
4
x=
Now the Jacobian and the differential area are:

(u, v)
u v u v
1
=
= (1 )
(,

2
du dv = dv du =
(u, v)
1
d d = (1 ) d d
(,
2
(7.12)
(7.13)
Now using Eq.(7.11) and Eq. (7.13) into Eq.(7.9), we get
(1 )
(1 ) 1
), y(,
)}
dd
2
2
8
1 1
Z 1Z 1
(1 ) 1
}
dd
= J
f {,
2
8
1 1
Z
I = J
f {x(,
(7.14)
Again, if we use the shape functions

1
N1 (u, v) = (1 + 2u t),
4
1
N2 (u, v) = (1 + v),
2
1
N3 (u, v) = (1 2u v)
4
then using mathematical transformation

1
{(x1 + 2x2 + x3 ) + 2(x1 x3 )u + (x1 + 2x2 x3 )v}
4
1
y(u, v) =
{(y1 + 2y2 + y3 ) + 2(y1 y3 )u + (y1 + 2y2 y3 )v} (7.15)
4
x(u, v) =
The original triangle transforms to a triangle with vertices 1(1, -1), 2(0, 1),
137

3(-1, -1). For this triangle, if we consider the transformation
(1 )
,
2
The integral I in Eq. (7.1) becomes
u=
v=
(1 )
(1 )
1
, ), y(
, )}
dd
2
2
8
1 1
Z 1Z 1
1
(1 )
, )}
dd
f {(
= J
2
8
1 1
Z
I = J
(7.16)
f {x(
(7.17)
Our aim is to derive suitable and efficient quadrature formulae to evaluate

numerically the integrals given in Eq. (7.14) or Eq. (7.17).
7.4.1
Symmetric Gauss Quadrature for Triangle (SGQTS)
Here, our strive is to compute Gaussian points and corresponding weights for
the symmetric quadrature formulae for triangles. Conventionally, the numerical
schemes for the integral in Eq.(7.14) is expressible as
I = J
= J
m X
n
X
(1 im )jn
1 im
(
)Wim Wjn f (im ,
)
8
2
i=1 j=1
mn
X
Gr f (ur , vr )
(7.18)
r=1
where (ur , vr ) are the new Gaussian points, Gr is the corresponding weights for
triangles. Here im , i = 1, m; jn , j = 1, n are the m-th and n-th order Gaussian
quadrature points respectively; Wim , Wjn are the corresponding m-th and n-th
order weights respectively. For Eq. (7.17) we will get
138
I = J
= J
n X
m
X
1 jn
(1 jn )im n
(
)Wim Wjn f (
, j )
8
2
j=1 i=1
mn
X
G0r (u0r , vr0 )
(7.19)
r=1
where (u0r , vr0 ) are the new Gaussian points, G0r is the corresponding weights for
triangles. In order to calculate such Gaussian points and corresponding weights,
we present the following simple algorithm:
step 1. r 1
step 2. i = 1, m
step 3. j = 1, n
(1 im ) m n
Wi Wj ,
Gr =
8
(1 jn ) m n
Wi Wj ,
8
r =r+1
G0r =
(1 im )jn
vr =
2
ur =
im ,
u0r =
(1 jn )im
,
2
vr0 = jn

Computed Gauss points and weights for different values of n(= m) based on
above algorithm listed in Table 7.1. The distribution of Gaussian points for m, n
= 10 and m, n =15 for both type of formulae are plotted in Figure 7.5 and 7.6.
In these figures formula 1 stands for Eq.(7.18) and formula 2 stands for Eq.(7.19).
Similarly, distribution of Gaussian points can be shown for higher values of m, n.
It is evident that due to symmetry, we are in a advantageous position because only
Gauss points and corresponding weights in any one side of the line of symmetry
are needed to compute. This is the substantial reduction in the computational
effort. But it is also clear that a huge crowding occurs for higher values of m, n
139

which is one of the major causes of error germane in the calculation.
Figure 7.5: Gauss Points for m = n = 10 using SGQTS (100 points, formula 2).
7.4.2
Symmetric Gauss Quadrature for Triangle (SGQTM)
As it is seen in the previous section that quadrature formulae so presented have

drawbacks due to huge crowding of Gaussian points. The reason behind this can
be explained with the help of Eq.(7.18) or Eq.(7.19) that for each i (i = 1, 2,...,
m) j varies from 1 to n and hence at the terminal value of i = m (say) there are
n crowding points as shown in Table 7.1 and Figure 7.5, 7.6. To overcome this
situation, we are taking the advantage of Eq.(7.18) by making j dependent on i
for the calculation of new Gauss points and corresponding weights. To do so, we
wish to calculate Gauss points and weights for i = 1, m-1 and j = 1, n + 1 - i
140
Figure 7.6: Gauss Points for m = n = 15 using SGQTS (225 points, formula 1).
that is
m(m+1)
2
I2 = J
1 points Gaussian quadrature formulae for n = m are as:
m1
X m+1i
X
i=1
j=1
m1
m+1i
(1 i )j
1 im1
)Wim1 Wjm+1i f (im1 ,
8
2
m(m+1)
1
2
= J
Lr f (pr , qr )}
(7.20)
r=1
where (pr , qr ) are the new Gaussian points, Lr is the corresponding weights for
the quadrature rules for triangles.
For Eq.(7.17) we can write
141
I2 = J
m1
X m+1j
X
j=1
i=1
(1 jm1 )im+1j m1
1 jm1
m+1j
m1
)Wi
Wj f (
, j )
(
8
2
m(m+1)
1
2
= J
L0r f (p0r , qr0 )}
(7.21)
r=1
where (p0r , qr0 ) are the new Gaussian points, L0r is the corresponding weights
for the quadrature rules for triangles. The Gaussian points and corresponding
weights can be calculated using the following algorithm:
step 1. r 1
step 2. i = 1, m 1
step 3. j = 1, m i + 1
pr =
im1 ,
(1 im1 )jm+1i
,
qr =
2
Lr =
(1 im1 ) m1 m+1i
Wi Wj
8
r =r+1
step 6. k 1
step 7. j = 1, m 1
step 8. i = 1, m j + 1
(1 jm1 )im+1j
=
,
2
k =k+1
p0r
qr0
jm1 ,
L0r
(1 jm1 ) m+1j m1
=
Wi
Wj
8

Thus, the new m(m+1)
1 points Gaussian quadrature formulae is now obtained
2
which are crowding free. For clarity and reference, computed Gauss points (p, q)
and weights L for different values of m( =n) based on above algorithm listed in
142

Table 7.2 and Figure 7.7 and 7.8 exhibits the distribution of Gaussian points for
m= n= 10 i.e. 54-points formula and m= n= 16 i.e. 135-points formula. In these
figures formula 1 stands for Eq.(7.20) and formula 2 stands for Eq.(7.21). In these
figures only two Gauss points are seen along the terminal side of the triangle,
more points can be taken by choosing n > m. If we interchange p and q then we
obtain (q, p) = (p0 , q 0 ) and G = G0 .
Figure 7.7: Gaussian points for m = 10 (54-points, formula 1) using SGQTM.
7.5
To show the accuracy and efficiency of the derived formulae some examples with
known results are considered. To compare the results we have taken the results
obtained by available Gaussian 7-points and 13-points rules, the quadrature rule
143
Figure 7.8: Gaussian points for m = 16 (135-points, formula 2) using SGQTM.

of Taylor et al. [2000]; Wandzura and Xiao [2003] and Dunavant [1985] for
triangles. The quadrature formulae of Taylor et al. [2000] is well known as Fekete
quadrature formulae.
1y
(x + y) 2 dx dy = 0.4
I1 =
y=0
1
x=0
1y
I2 =
y=0
(x + y) 2 dx dy = 0.6666667
x=0
144

Z
(x2 + y 2 ) 2 dx dy = 0.881373587
I3 =
y=0
1
x=0
y
exp|x + y 1|dx dy = 0.71828183
I4 =
y=0
x=0
Computed values are summarized in Table 7.3. Some important remarks from
the Table 7.3 are:
Splitting the original triangle into quadrilaterals without any transformation (IOAT) provides the way of using Gaussian quadrature for
square domain integrals and the convergence rate is slow (as it needs
large number of points) but satisfactory in view of accuracy.
New Gaussian quadrature formulae for triangle (SGQTS and SGQTM)
are exact in view of accuracy and efficiency and the formula (SGQTM)
is faster.
Usual Gauss quadrature (GQT) for triangles e.g. 7-point and 13point rules cannot evaluate the integral of nonpolynomial functions
accurately.
The developed integration formulae of this chapter are always consistent, reliable and provides considerable savings in computational effort, in comparison with existing numerical quadrature rule of Dunavant [1985]; Taylor et al. [2000]; Wandzura and Xiao [2003].
The Fekete quadrature rule for triangle have some quadrature
points within the domain of integration where the integrands becomes
undefined and the models fail in that case, approximations are done
by simply excluding that points.
Again, we consider the following integrals of rational functions due to Rathod
and Karim [2002] to test the influences of formulae in Eq.(7.18), Eq.(7.19),
Eq.(7.20) and Eq.(7.21);
p,q
1y
=
y=0
x=0
xp y q
dx dy
+ x + y
145
Example-1: I
Example-2:I
r,0
r,0
Z
=
y=0
x=0
1y
=
y=0
Example-3: I
0,0
0,0
x=0
1y
x=0
1y
=
y=0
x=0
xr
dx dy, 6= = 0
0.375 0.375 x
xr
dx dy, = 0, 6= 0
1y
=
y=0
Example-4: I
1y
1
dx dy, , 6= 6= 0
12 + 21.53679831x 8.821067231y
1
dx dy = 6= 0
12 + 9.941125498(x + y)
Results are summarized in Tables 7.4, 7.5 and 7.6. In these tables formula
1 stands for Eq.(7.18) and Eq.(7.20) whereas formula 2 stands for Eq.(7.19) and
Eq.(7.21).
These tables substantiated the influences of numerical evaluation of the integrals as described in equations Eq.(7.18), Eq. (7.19), Eq.(7.20) and Eq. (7.21).
r
x
with 6= = 0 first formulae described in
Eq.(7.18) (SGQTS) and Eq.(7.20) (SGQTM) are more accurate and
rate of convergence is higher. But the formula in Eq.(7.20) requires
very less computational effort.
r
y
with 6= = 0 second formulae
described in Eq.(7.19) (SGQTS) and Eq.(7.21) (SGQTM) are more
accurate and convergence is higher. Here also the formula in Eq.(7.21)
requires very less computational effort.
Similar influences of these formulae may be observed for different

conditions on , .
The general Gaussian quadrature eg. 7-point and 13-point rules
cannot evaluate the integral of rational functions accurately. The
146

Fekete quadrature rule of Taylor et al. [2000] for triangle fails to
calculate integrals of type in Example -1 and example -2 accurately.
Results of example - 3 and 4 are satisfactory but convergence is slow.
Where as the developed integration formulae of this chapter are always consistent, reliable and provides considerable savings in computational effort, in comparison with existing numerical methods Dunavant [1985]; Wandzura and Xiao [2003].
It is evident that the new formulae e.g. SGQTS and SGQTM are very fast and
accurate in view of accuracy and equally applicable for any geometry that is for
different values of , and . We recommend these are appropriate quadrature
scheme for triangular domain integrals encountered in science and engineering.
7.6
Error Analysis
The n-point Gauss quadrature formula can evaluate exactly the integral of polynomial of order up to 2n-1. The total error in n-point Gauss quadrature formula
to evaluate the integral of polynomial of high order is given by
=
22n+1 [n!]4
f 2n (x1 )
(2n + 1)[(2n)!]3
Where f 2n (x1 ) is the 2n-th derivative of the function at a point x1 in the

interval [-1, 1] (Chapra and Canale [2006]).
In this chapter the triangular domain integral is evaluated by converting it to
a equivalent square domain integral. Now we consider the square domain integral
Z
I=
f (x, y)dydx
x=1
y=1
Integrating with respect to x using Gauss quadrature formula, keeping y fixed

we have
Z
I=
y=1
m
X
f (xi , y)W (m, i)}dy +
i=1
147
22m+1 [m!]4
fx2m
3
(2m + 1)[(2m)!]

Where fx2m is the (2m) th partial derivative of the function with respect to x,
m points are taken along x direction, W(m, i) are the corresponding weight. Now
integrating with respect to y we have
I =
m X
n
X
f (xi , yj )W (m, i)W (n, j) +
i=1 j=1
m
22m+1 [m!]4
f 2m
(2m + 1)[(2m)!]3 x
XX
22n+1 [n!]4
2n
+
f
=
f (xi , yj )W (m, i)W (n, j) + E
(2n + 1)[(2n)!]3 y
i=1 j=1
The total error in the 2-square domain integral is found to be given by the
error formula
2(2n+2) [n!]4
22m+2 [m!]4
2m
fx (x1 , y1 ) +
fy2n (x2 , y2 )
E=
3
3
(2m + 1)[(2m)!]
(2n + 1)[(2n)!]
(7.22)
Where fx2m (x1 , y1 ) is the 2m-th partial derivative of the function with respect
to x, fy2n (x2 , y2 ) is the 2n-th partial derivative of the function with respect to y
and (x1 , y1 ) and (x2 , y2 ) are points somewhere in the domain [1, 1] [1, 1].
Consider the square domain polynomial integral
Z
J=
1
N
X
(
xi y N i )dydx
1 i=0
The computed errors in the integral J using m m Gauss quadrature method

and the error formula given by Eq.(7.22) are given in Table 7.7. The table 7.7
shows that the error formula of Eq.(7.22) gives exactly same results for the error
calculated using the m m point method.
We know, the one dimensional n-point Gaussian Quadrature rule gives exact
results for polynomials of degree at most 2n-1. Thus with n = 2, we have a rule
with 4 nodes which is exact for any monomial of degrees at most 3 in x and y separately, so the total degree of this monomial is at most 6. But this rule is not exact
for all monomials of degree at most 6, which includes x6 , x5 y, x4 y 2 , x2 y 4 , xy 5 , y 6 .
Let the maximum value of p+q be 2N-1 for all monomial xp y q . Then N N -point
148

rule can calculate all the monomials of order up to 2(p + q) 1 = 2N 1.
Now consider the triangular domain integral
Z
(1x)/2
xp y q dy dx
JJ =
x = 1 y = (x1)/2
Z 1
1
p (1
Z
=
= 1
=
=
1
2q+1
1
2q+1
= 1
1
= 1
= 1
) q 1
}
(1 )d d
2
2
p (1 )q+1 q d d
= 1
p+q+1
Al d
l=p
q d
= 1
Let p+q = 2N-1. Then N N -point rule can calculate the monomial over the
triangular domain, if p+q is odd. If p+q is even, then (N + 1) (N + 1)-point rule
can calculate the monomials over the triangular domain. The methods SGQTS
and SGQTM are verified for the integral of all monomials xi y j where i , j are
non-negative integers such that i + j M = 30 over unit triangle. In Table 7.8
and 7.9 , we present the absolute error over corresponding monomial integrals. In
Table 7.10 and 7.11 we present the absolute error over corresponding polynomial
integrals. All these Tables are in good agreement with the above statement. It
is observed that the new methods are always accurate for monomials/polynomial
functions in view of both accuracy and efficiency and hence a proper balance is
observed.
Figure 7.9, 7.10, 7.11 and 7.12 shows the absolute error in the integral of
polynomial of order 15, 20, 25 and 30 using the methods SGQTS (m m points)
1 points). It is seen from the figures that, the error
and SGQTM ( m(m+1)
2
decreases with increase in number of total Gaussian integration points. These
figures verifies that, if SGQTS gives accurate results for a value N, SGQTM gives
the results of same accuracy for N+1 to evaluate the polynomials of degree at
most 2N 1 accurately. Table 7.8, 7.9, 7.10, 7.11 and Figure 7.9, 7.10, 7.11, 7.12
proves the efficiency of the method SGQTM to calculate the polynomial integral
over triangle with significantly less number of Gaussian integration points and
weights. It is seen from the figures that, the method SGQTM which is completely
149

free of crowding of Gaussian integration points and can evaluate the integrals
successfully with desired accuracy and efficiency.
Figure 7.13, 7.14, 7.15 and 7.16 shows the percentage error in calculated results
of the integrals I1 , I2 , I3 and I4 using SGQTS (m m points) formulae and
SGQTM ( m(m+1)
1 points) formulae. These figures also show that, if SGQTS
2
gives accurate results for a value N, SGQTM gives the results of same accuracy
for N+1.
Figure 7.17, 7.18, 7.19 and 7.20 shows the percentage error in calculated results
of Example-1 (for r=2), example 2 (for r=6), example - 3 and 4. Figures show
that, the percentage error in SGQTS and SGQTM fluctuates within a very small
range that is about 2E 06. The percentage error decreases with the increases
in number of Gauss points. As m increases the range of fluctuation also become
smaller. All these figures show the accuracy and efficiency of the methods SGQTS
and SGQTM to calculate the integral of a polynomial, non-polynomial or rational
functions.
7.7
Conclusions
Three numerical schemes IOAT, SGQTS and SGQTM to evaluate the triangular
domain integrals are presented in this chapter. The scheme IOAT is derived based
on the subdivision of an arbitrary triangular domain by three quadrilaterals. This
scheme is different from IOUT, IOST of previous chapters. The main advantage of
this scheme is that the readily available Gauss quadrature formulae for the square
domain is employed and that grantees the highest accuracy of the evaluation.
The scheme further is used as the testing tools for the accuracy of the developed
new quadrature formulae SGQTS and SGQTM. Both the quadrature schemes so
derived are symmetric and that substantially reduces the computational effort.
Here, the quadrature scheme SGQTS and SGQTM are respectively m m points
1 points formulae (for n=m). Figures shown that, in SGQTM there
and m(m+1)
2
are only two points at one terminal side of the triangle, more points can be taken
by choosing n > m. It is clearly shown that SGQTS suffers due to crowding of
Gauss points whereas the SGQTM is totally free from such crowding. As far as
the accuracy is concern, both the schemes are so accurate but the last one i.e.
150

SGQTM is very much faster. Therefore, it can be said that the scheme SGQTM
guarantees the proper balance between the accuracy and efficiency for evaluating
the triangular domain integrals.
Through several application examples, the computed results clearly exhibits
the best performances of the derived quadrature formulae. The computer codes
in FORTRAN for each scheme are also developed and appended. The comparison
between the computed results of Chapter 5, 6 and 7 and by the present methods
are also done and it is found that the proposed new methods of this chapter
are always accurate. It is astounding to note here that the scheme SGQTM is
consistent in view of accuracy for all cases as described in previous chapters.
Thus, gradual development of the quadrature schemes for the triangular domain
integrals is now satisfactory and that will find better applications in the area of
FEM, BEM and other numerical methods.
151
Table 7.1: Computed Gauss points (u, v) and corresponding weights G for n n
point method (SGQTS).
n=8, 64 points
n=3, 9 points
u
-0.7745966692D+00
-0.7745966692D+00
0.0000000000D+00
0.0000000000D+00
0.7745966692D+00
0.7745966692D+00
-0.9602898565D+00
-0.9602898565D+00
-0.9602898565D+00
-0.9602898565D+00
-0.7966664774D+00
-0.7966664774D+00
-0.7966664774D+00
-0.7966664774D+00
-0.5255324099D+00
-0.5255324099D+00
-0.5255324099D+00
-0.5255324099D+00
-0.1834346425D+00
-0.1834346425D+00
-0.1834346425D+00
-0.1834346425D+00
0.1834346425D+00
0.1834346425D+00
0.1834346425D+00
0.1834346425D+00
0.5255324099D+00
0.5255324099D+00
0.5255324099D+00
0.5255324099D+00
0.7966664774D+00
0.7966664774D+00
0.7966664774D+00
0.7966664774D+00
0.9602898565D+00
0.9602898565D+00
0.9602898565D+00
0.9602898565D+00
v
G
0.6872983346D+00 0.6846437767D-01
0.0000000000D+00 0.1095430043D+00
0.3872983346D+00 0.6172839506D-01
0.0000000000D+00
0.9876543210D-01
0.8729833462D-01 0.8696116156D-02
0.0000000000D+00
0.1391378585D-01
0.9412232325D+00 0.2510939335D-02
0.7808486073D+00 0.5516085752D-02
0.5150979262D+00 0.7781386411D-02
0.1797925345D+00 0.8996247611D-02
0.8626602969D+00 0.5055663745D-02
0.7156719768D+00 0.1110639129D-01
0.4721032318D+00 0.1566747258D-01
0.1647854365D+00 0.1811354112D-01
0.7324766495D+00 0.6055613217D-02
0.6076702656D+00 0.1330310188D-01
0.4008583619D+00 0.1876631018D-01
0.1399177461D+00 0.2169618166D-01
0.5682201415D+00 0.5431069819D-02
0.4714013539D+00 0.1193109146D-01
0.3109666298D+00 0.1683085382D-01
0.1085414553D+00 0.1945855410D-01
0.3920697150D+00 0.3747417313D-02
0.3252651235D+00 0.8232407275D-02
0.2145657801D+00 0.1161322448D-01
0.7489318721D-01 0.1342632758D-01
0.2278132070D+00 0.1883402929D-02
0.1889962118D+00 0.4137500225D-02
0.1246740480D+00 0.5836654737D-02
0.4351689638D-01 0.6747896641D-02
0.9762955961D-01 0.5721629090D-03
0.8099450059D-01 0.1256939834D-02
0.5342917807D-01 0.1773129531D-02
0.1864920601D-01 0.2049957612D-02
0.1906662400D-01 0.5086480501D-04
0.1581787007D-01 0.1117409020D-03
0.1043448371D-01 0.1576297352D-03
0.3642107988D-02 0.1822395206D-03
152
1 point method (SGQTM).
n
m=3, 5 points
m=4, 9 points
m=5, 14 points
m=8, 35 points
p
-0.5773502692D+00
-0.5773502692D+00
0.5773502692D+00
-0.7745966692D+00
-0.7745966692D+00
0.0000000000D+00
0.0000000000D+00
0.7745966692D+00
-0.8611363116D+00
-0.8611363116D+00
-0.8611363116D+00
-0.3399810436D+00
-0.3399810436D+00
0.3399810436D+00
0.3399810436D+00
0.8611363116D+00
-0.9491079123D+00
-0.9491079123D+00
-0.9491079123D+00
-0.9491079123D+00
-0.7415311856D+00
-0.7415311856D+00
-0.7415311856D+00
-0.7415311856D+00
-0.4058451514D+00
-0.4058451514D+00
-0.4058451514D+00
0.0000000000D+00
0.0000000000D+00
0.0000000000D+00
0.4058451514D+00
0.4058451514D+00
0.7415311856D+00
0.7415311856D+00
0.9491079123D+00
m(m+1)
q
L
0.6109051324D+00 0.1095382131D+00
0.0000000000D+00 0.1752611410D+00
0.1220084679D+00 0.5283121635D-01
0.7640848152D+00 0.4286819789D-01
0.3016646138D+00 0.8036768192D-01
0.3872983346D+00 0.6172839506D-01
0.0000000000D+00
0.9876543210D-01
0.6506833684D-01 0.1565300908D-01
0.8432621081D+00 0.1917346465D-01
0.5010823929D+00 0.3873334126D-01
0.0000000000D+00
0.4603770905D-01
0.5769531667D+00 0.3799714765D-01
0.2277840768D+00 0.7123562050D-01
0.2556242426D+00 0.2989084476D-01
0.0000000000D+00
0.4782535162D-01
0.4008649394D-01 0.6038050853D-02
0.9358542787D+00 0.3193509426D-02
0.7763944673D+00 0.7015570468D-02
0.5121596892D+00 0.9896667159D-02
0.1787669565D+00 0.1144177446D-01
0.8264505139D+00 0.7884269378D-02
0.6456998424D+00 0.1703110962D-01
0.3533959938D+00 0.2324942473D-01
0.0000000000D+00
0.2544930806D-01
0.6554538727D+00 0.1149574334D-01
0.4647790050D+00 0.2420682761D-01
0.1677308129D+00 0.3139666970D-01
0.4530899230D+00 0.1237822093D-01
0.2692346551D+00 0.2500590605D-01
0.0000000000D+00
0.2972154195D-01
0.2558241574D+00 0.9864562309D-02
0.1010006927D+00 0.1849370967D-01
0.1001045414D+00 0.5020494508D-02
0.0000000000D+00
0.8032791213D-02
0.1469128025D-01 0.8237200311D-03
153
Table 7.3: Calculated values of the integrals I1 , I2 , I3 , I4

Points
Test example
I1
I2
I3
77
0.4001498818 0.6606860757 0.8315681219
GQT
13 13
0.4000451564 0.6637058258 0.8501738309
54
0.4000012535 0.6663130315 0.8737747596
Wandzura
85
0.4000004580 0.6664725905 0.8762870188
126
0.4000001581 0.6665642484 0.8770618865
55
0.3999728246 0.6633657190 0.8540687388
Fekete
91
0.3999935906 0.6653406650 0.8666482684
136
0.3999985466 0.6661290912 0.8731521253
25
0.4000107345 0.6653960001 0.8615849602
Dunavant
48
0.4000017306 0.6662365284 0.8727339088
79
0.4000005196 0.6664581214 0.8760476828
3 7 7 0.4000006727 0.6664256193 0.8755247201
3 8 8 0.4000003534 0.6665011716 0.8768203350
IOAT
3 9 9 0.4000002024 0.6665481694 0.8777286724
3 10 10 0.4000001234 0.6665789279 0.8783900003
77
0.4000037510 0.6659893927 0.8696444210
88
0.4000020035 0.6662009982 0.8722479925
SGQTS
99
0.4000011460 0.6663329015 0.8740714951
10 10
0.4000006929 0.6664193644 0.8753981854
54
0.4000009417 0.6663718426 0.8742865042
SGQTM
77
0.4000003700 0.6664974532 0.8765237986
90
0.4000002468 0.6665339400 0.8772635781
104
0.4000001695 0.6665606288 0.8778458328
Exact Value
0.4
0.6666667
0.881373587
154
I4
0.6938790083
0.7238717079
0.7165792295
0.7162337687
0.7173750469
0.7164511237
0.7171579302
0.7173863082
0.7153629197
0.7166547182
0.7152268938
0.7178753416
0.7179654297
0.7180285699
0.7180745324
0.7184323903
0.7185688310
0.7181265336
0.7182531970
0.7175459725
0.7179128710
0.7180958213
0.7182831374
0.71828183
N.B.: For the highest accuracy, the scheme SGQTM require less number of points.
N.B.: * - Method fails, approximation done excluding points of discontinuity.
Method

Method
GQT
Wandzura
Fekete
Dunavant
IOAT
SGQTS
SGQTM
Exact
Value
Computed value ofI r,0

r=2
r=4
r=6
77
0.7288889289
0.3733333349
0.2209523767
13 13
0.7883351445
0.4327795803
0.2803986370
54
0.8643341060
0.5087785538
0.3563976047
85
0.8724169890
0.5168614388
0.3644804928
126
0.8781372388
0.5225816740
0.3702007151
54
0.8095840242
0.4558851351
0.3053608511
91
0.8460916001
0.4911054664
0.3392939554
136
0.8651135795 0.5097291161 0.3575192530
25
0.8314373633
0.4758817940
0.3235008513
48
0.8608934610
0.5053379103
0.3529569503
79
0.8716316464
0.5160761035
0.3636951444
355
0.8536515855
0.4980960396
0.3457150818
366
0.8636423810
0.5080868226
0.3557058569
377
0.8699174296
0.5143618757
0.3619809270
388
0.8741141888
0.5185586402
0.3661776763
399
0.8770583628
0.5215028018
0.3691218499
3 10 10 0.8792029273
0.5236473748
0.3712664246
Results obtained using formula 1
55
0.8888888936
0.5333333316
0.3809523857
66
0.8888888912
0.5333333289
0.3809523670
77
0.8888888942
0.5333333215
0.3809523811
88
0.8888888822
0.5333333368
0.3809523803
99
0.8888888894
0.5333333261
0.3809523861
10 10
0.8888888916
0.5333333260
0.3809523859
44
0.8888888822
0.5333333368
0.3809523803
77
0.8888888823
0.5333333366
0.3809523814
90
0.8888889030
0.5333333378
0.3809523782
104
0.8888889011
0.5333333300
0.3809523797
Points
0.8888888
0.5333333
0.3809523
155

Method
GQT
Wandzura
Fekete
Dunavant
IOAT
SGQTS
SGQTM
Exact
Value

r=2
r=4
r=6
77
0.1108333394
0.04000000283
0.02035714500
13 13
0.1110424771
0.03999926149
0.02040804736
54
0.1111099898
0.0400000020
0.0204081629
85
0.1111107663
0.0400000001
0.0204081632
126
0.1111110190
0.0400000001
0.0204081632
55
0.1111115101
0.0400000244
0.0204081649
91
0.1111109809
0.0400000033
0.0204081632
136
0.1111110973
0.0400000004
0.0204081631
25
0.1111191026
0.0400000359
0.0204081637
48
0.1111094171
0.0399999999
0.0204081633
79
0.1111107485
0.0399999998
0.0204081633
355
0.1111078131
0.03999999048
0.02040816293
366
0.1111099001
0.03999999791
0.02040816328
377
0.1111105972
0.03999999965
0.02040816323
388
0.1111108684
0.03999999955
0.02040816329
399
0.1111109861
0.03999999964
0.02040816318
3 10 10
0.1111110426
0.03999999981
0.02040816318
Results obtained using formula 2
55
0.1111111124
0.04000000014
0.0204081636
66
0.1111111103
0.04000000016
0.0204081631
77
0.1111111106
0.04000000038
0.0204081631
88
0.1111111121
0.04000000059
0.0204081634
99
0.1111111110
0.03999999975
0.0204081633
10 10
0.1111111107
0.03999999995
0.0204081630
44
0.1111111119
0.04000000020
0.0204081633
77
0.1111111112
0.04000000003
0.0204081634
90
0.1111111111
0.03999999995
0.0204081632
104
0.1111111115
0.04000000005
0.0204081633
Points
0.1111111111111
0.04
0.0204081632653
156
Table 7.6: Computed results of I 0,0 Example -3 and 4

Method
GQT
Wandzura
Fekete
Dunavant
IOAT
SGQTS
SGQTM
Exact
Value
Points
77
13 13
54
85
126
55
91
136
25
48
79
355
366
377
388
399
3 10 10
55
66
77
88
99
10 10
44
77
90
104
Example -3
0.03669412062
0.03688941523
0.036947991050
0.036948008095
0.036948009595
0.03694928293
0.036948026227
0.036948013918
0.036948615464
0.036947887278
0.036948007107
0.03694724295
0.03694789912
0.03694799350
0.03694800728
0.03694800931
0.03694800972
formula 2
0.036956844135
0.036950091822
0.036948511561
0.036948132905
0.036948040297
0.036948017307
0.036948121605
0.036948011096
0.036948010012
0.036948009928
Example -4
0.02731705643
0.02731722965
0.027317233499
0.027317233545
0.027317233566
0.027317233923
0.027317233402
0.027317233345
0.027317233460
0.027317233409
0.027317233437
0.02731723353
0.02731723339
0.02731723359
0.02731723343
0.02731723344
0.02731723331
formula 1 / 2
0.027317233290
0.027317233662
0.027317233227
0.027317233349
0.027317233489
0.027317233318
0.027317233552
0.027317233459
0.027317233575
0.027317233635
0.03694801040
0.02731723349
157
Table 7.7: Error calculated using the new error formula and the m m point
method
Points
22
33
44
55
66
77
88
99
10 10
N Error calculated using error formula Error using m m points

4
0.71111066
0.71111066
6
0.18285666
0.18285666
8
0.04643863
0.04643863
10
0.01172679
0.01172679
12
0.00295302
0.00295302
14
0.00074186
0.00074186
16
0.00018619
0.00018619
18
0.00004669
0.00004669
20
0.00001170
0.00001170
158
Table 7.8: The absolute error over corresponding monomials integrals for SGQTS.
N
1
2
3
3
4
5
5
6
7
7
8
9
10
11
12
13
14
16
17
19
20
21
22
23
24
25
26
27
28
29
30
30
30
30
30
i
1
0
1
2
2
3
3
3
4
4
3
3
7
6
7
8
10
3
11
4
6
6
7
8
15
14
11
11
6
15
15
30
30
30
30
m
2
2
3
3
3
4
5
5
5
6
6
7
7
7
8
8
9
10
10
11
11
12
12
13
13
14
14
15
15
15
15
16
18
19
20
Formula 1
0.11102230246251565404D-15
0.69388939039072283776D-16
0.93675067702747583098D-16
0.62450045135165055399D-16
0.25153490401663702869D-16
0.51451898297472098420D-14
0.24286128663675299322D-16
0.12034644114589099217D-16
0.51499603193061460615D-17
0.70473141211557788210D-17
0.45536491244391186228D-17
0.12834243216797158738D-16
0.76300727888667374543D-17
0.13145951341386741262D-17
0.18812601758518010531D-17
0.10693790959085541781D-17
0.34711410178481227895D-17
0.12001609829646181504D-16
0.22705776942335588464D-18
0.14525556254227182940D-17
0.72970564916977889366D-19
0.62256921623191074922D-19
0.16136639389684463881D-19
0.67353181377130278426D-20
0.26850282683471278326D-20
0.86559281224845889752D-21
0.43010806913343210882D-21
0.28096998962953544725D-21
0.87461942068294381203D-20
0.89334258922459623774D-19
0.41875729320400059259D-19
0.63853002746361298136D-14
0.72730450134672608442D-14
0.74376269032505604173D-14
0.74504204888858893696D-14
159
Formula 2
0.13877787807814456755D-15
0.55511151231257827021D-16
0.62450045135165055399D-16
0.97144514654701197287D-16
0.25153490401663702869D-16
0.46187012547882488889D-14
0.33393426912553536567D-16
0.11817803680091998331D-16
0.63425827090402009389D-17
0.80773061850170080334D-17
0.35507621148900270214D-17
0.12387009820646888159D-16
0.75487576259303246218D-17
0.14196272195982073683D-17
0.22107559923337238850D-17
0.12730905197232134096D-17
0.40056188075655863035D-17
0.10882679306323250756D-16
0.21292290711604974773D-18
0.12598556299223649668D-17
0.82638651916497676830D-19
0.69436849340151355139D-19
0.19766307917567149319D-19
0.77750841676921836300D-20
0.23076271138698089560D-20
0.82382019131453099994D-21
0.46373813091254114022D-21
0.29547149974335571368D-21
0.87056623568143397642D-20
0.83832869692064102146D-19
0.41875731743780760098D-19
0.46824523425303965496D-14
0.69096204452501197579D-14
0.79038338374193273239D-14
0.87811702353945975119D-14
Table 7.9: The absolute error over corresponding monomials integrals for SGQTM.
N
1
2
3
4
5
5
6
7
7
9
11
12
14
15
17
19
19
20
21
22
23
24
25
26
27
28
29
30
i
0
0
1
2
3
3
3
4
4
4
6
7
11
12
11
4
4
6
6
15
9
15
14
15
6
7
13
11
m
3
3
4
4
5
7
7
7
8
8
8
10
11
12
13
14
16
16
17
17
17
17
17
17
17
17
17
17
Formula 1
0.52735593669694935670D-15
0.31918911957973250537D-15
0.57176485768195561832D-14
0.65832755913319829233D-15
0.42149443657546470376D-14
0.81141690588815151841D-15
0.19320482713691689014D-15
0.11703962451981020365D-15
0.84947240214239272404D-16
0.22020029571447052863D-12
0.20913072536459919454D-12
0.82127035546026606883D-14
0.10281285913772697516D-16
0.77063057541196244848D-17
0.64774877142554824301D-14
0.34968188216441369043D-15
0.96761867520779067109D-17
0.53413037386019124233D-17
0.89397710137790976567D-18
0.88817939246452559470D-17
0.52169371335045647124D-18
0.32294409108116594855D-17
0.96615005738923339810D-18
0.85244149353887245280D-18
0.63794981254172174580D-18
0.38053348066190405421D-18
0.95259153317372358323D-19
0.78654016277078905607D-19
160
Formula 2
0.33306690738754696213D-15
0.97144514654701197287D-16
0.13357370765021414627D-14
0.65832755913319829233D-15
0.42500725161431773813D-14
0.32916377956659914616D-15
0.19320482713691689014D-15
0.18841499974803221917D-10
0.35517455946806886224D-11
0.33732556136388536228D-11
0.15063337289479821618D-10
0.15025672712483265428D-12
0.11435428782559115685D-12
0.17804363194808166260D-13
0.13019009425257495457D-14
0.30886005041982282464D-14
0.60896703176930916043D-15
0.24019216008850158479D-15
0.91583551450074603259D-16
0.65983039410497443386D-18
0.34613012908714982488D-17
0.38533092141955847647D-18
0.33492863534759454854D-18
0.23097206563013923845D-18
0.77370757148537395401D-16
0.75016567210806363964D-17
0.20251223810865608639D-18
0.51832691377395518820D-18
Table 7.10: Absolute error in M point formula for Polynomial of order N for
SGQTS.
N
2
4
6
8
10
12
14
16
18
20
23
26
30
3
6
8
10
11
14
17
19
21
23
25
27
29
30
M
2
3
4
5
6
7
8
9
10
11
12
13
14
15
15
15
15
15
15
15
15
15
15
15
15
15
15
Formula 1
0.13877787807814456755D-15
0.61062266354383609723D-15
0.56996074526693973894D-13
0.81185058675714572018D-15
0.70429773124658368033D-15
0.10356299151581538354D-14
0.23835100559921329477D-14
0.43966566498632175808D-14
0.65156213757688874466D-14
0.83275400464266624567D-14
0.49274820335121205517D-14
0.56465249143045070923D-14
0.34867941867133822598D-15
0.38172243144174444751D-12
0.17785078965104617055D-12
0.12044532038402167018D-12
0.86659845965897375208D-13
0.74709335939893151135D-13
0.50034629217599047024D-13
0.35053557279063340957D-13
0.28127673801225938632D-13
0.22797735921287198835D-13
0.18630062770252919790D-13
0.15331919761552015302D-13
0.12692104311984309106D-13
0.10559261798270824784D-13
0.96363889090511634095D-14
161
Formula2
0.13877787807814456755D-15
0.61062266354383609723D-15
0.56996074526693973894D-13
0.81532003370909933437D-15
0.70082828429463006614D-15
0.10338951916821770283D-14
0.23835100559921329477D-14
0.43957892881252291772D-14
0.65164887375068758502D-14
0.83284074081646508603D-14
0.49274820335121205517D-14
0.56473922760424954959D-14
0.34954678040932662952D-15
0.38172243144174444751D-12
0.17785078965104617055D-12
0.12045919817182948464D-12
0.86659845965897375208D-13
0.74712805386845104749D-13
0.50029425047171116603D-13
0.35053557279063340957D-13
0.28127673801225938632D-13
0.22797735921287198835D-13
0.18630062770252919790D-13
0.15332353442421009504D-13
0.12692971673722297510D-13
0.10559261798270824784D-13
0.96350878664441808041D-14
Table 7.11: Absolute error in M point formula for Polynomial of order N for
SGQTM.
N
2
3
3
5
5
7
7
11
16
18
19
20
21
22
23
24
25
26
27
28
29
21
26
27
28
29
30
M
3
4
6
8
9
10
11
11
13
14
15
15
15
15
15
15
15
15
15
15
15
17
18
21
22
22
22
Formula 1
0.52735593669694935670D-15
0.34833247397614286456D-13
0.33834046675451645569D-13
0.40010356139319469548D-11
0.92428148468215454159D-12
0.64681315858905463756D-11
0.31802338540387609100D-12
0.90869672897397890665D-12
0.87030209428018423523D-13
0.36290415117434804415D-13
0.25661764380124907348D-13
0.23849845709467132338D-13
0.22624263573689518125D-13
0.22124663212608197682D-13
0.22604314253715784844D-13
0.24489958672102574155D-13
0.28469414325993369630D-13
0.35621679217445745280D-13
0.47618159415563354742D-13
0.66877926807595855507D-13
0.96405956134804071667D-13
0.23984286778855334887D-13
0.19719469113166354646D-13
0.19230277092940895045D-13
0.18625292280693983571D-13
0.17502926191736989381D-13
0.16464260510495876133D-13
162
Formula2
0.52735593669694935670D-15
0.34833247397614286456D-13
0.33834046675451645569D-13
0.40010356139319469548D-11
0.92428842357605844882D-12
0.64681315858905463756D-11
0.31802338540387609100D-12
0.90869846369745488346D-12
0.87030209428018423523D-13
0.36290415117434804415D-13
0.25661764380124907348D-13
0.23849845709467132338D-13
0.22624263573689518125D-13
0.22124663212608197682D-13
0.22604314253715784844D-13
0.24489958672102574155D-13
0.28469414325993369630D-13
0.35621679217445745280D-13
0.47618159415563354742D-13
0.66877926807595855507D-13
0.96405956134804071667D-13
0.23984286778855334887D-13
0.19719469113166354646D-13
0.19230277092940895045D-13
0.18625292280693983571D-13
0.17502926191736989381D-13
0.16464260510495876133D-13
Figure 7.9: Absolute Error in Polynomial of order 15.

163

164
Figure 7.13: Percentage Error in I1 .

165

166
Figure 7.17: Percentage error in Example - 1 (r = 2).
Figure 7.18: Percentage error in Example - 2 (r = 4).

167
Figure 7.19: Percentage error in Example - 3.
Figure 7.20: Percentage error in Example - 4.

168
Chapter 8
Efficient Gauss
Quadrature Rules for
Tetrahedral Domain
Integrals
Chapter 8
Efficient Gauss Quadrature Rules
for Tetrahedral Domain Integrals
This chapter concentrates to present crowding free Gaussian Quadrature formulae
to evaluate tetrahedral domain integrals in Euclidean three dimensional space
(x, y, z). Through mathematical transformations, the tetrahedron T in (x, y, z)
space is transformed to a unit tetrahedron V : {(u, v, w)| 0 u, v, w 1, u + v +
w 1} whose base triangle is an unit triangle in w = 0 plane. It presents two
set of quadrature formulae and each set contains 6 formulae, for which Gaussian
points and corresponding weights are different. Among them one such set of
formulae are crowding free but the other one is not. Then it demonstrates the
efficiency and accuracy evaluating the integrals of polynomial, non-polynomial and
rational functions. In all the test cases the efficiency and accuracy of the derived
formulae are found satisfactory. Finally, it correctly evaluates all the integrals to
form element matrices for tetrahedral finite elements employing less number of
Gaussian points and weights. Thus, it ensures the balance between accuracy and
efficiency of evaluation of tetrahedral domain integrals encountered in science and
engineering.
8.1
Introduction
Many realistic problems involving triple integration frequently occur in a large

number of engineering applications in Computer Aided Design (CAD), Computer
170
8. Gauss Quadrature Rules for Tetrahedral Domain Integrals

Aided Engineering (CAE), Computer Aided Manufacturing (CAM), Geometric
modelling as well as in robotics. Particularly, for irregular, complicated or distorted type of domain such integrals cannot be evaluated analytically or explicitly
and numerical integration scheme are the only choice (Gauss [1814]; Hughes
[1987]; Zienkiewicz and Taylor [2000]). Finite element method gained more attraction due to its advantage to obtain the solution of many complicated problems
which cannot be evaluated such accurately by any other methods. Finite element
methods extensively use the hexahedral and tetrahedral elements for three dimensional problems. The versatility of these elements can be further enhanced by
the improvement of numerical integration schemes. Mathematically, the problem
in three dimensions can be defined as the evaluation of the triple integral of a
function f (x, y, z) over the arbitrary tetrahedral domain T in Euclidean ThreeDimensional Space (x, y, z).
Triple integral formulae have always been of enormous curiosity in computer
applications, a good impression of existing methods for evaluating volume integrals is given by Lee and Requicha [1982]. Numerical integration formulae for
tetrahedral regions with a degree of precision up to 3 are listed in Zienkiewicz and
Taylor [2000]. Timmer and Stern [1980] discussed a hypothetical advancement
to the evaluation of volume integrals by transforming it to a surface integral over
the boundary of the integration domain. Lien and Kajiya [1984] gave an outline
of a closed form formula for volume integration by decomposing the solid into
a set of solid tetrahedral. Bernardini [1991]; Cattani and Paoluzzi [1990a,b]
presented formulae and algorithms for computing integrals of polynomials over
tetrahedron. Rathod et al. [2004] derived explicit integration formulae for computing volume integrals of trivariate polynomials over an arbitrary tetrahedron
in Euclidean space. They evaluated this volume integral as a sum of four integrals over the unit triangle. Rathod et al. [2005] proposed product formulae
based on Gauss-Legendre quadrature rule for the numerical integration of an arbitrary function over the standard tetrahedron. The main disadvantages in those
methods was a huge crowding of Gauss points at one side of the tetrahedron. In
previous chapters, integration schemes totally free of crowding of Gaussian integration points for triangular domain integrals are presented. Such developments
are not yet done in case of tetrahedral domain integrals.
171

In this chapter, we present quadrature schemes for the evaluation of numerical
integration of the triple integrals of a function f (x, y, z) over any arbitrary tetrahedron T in Euclidean Three Dimensional Space (x, y, z) having privileged precisions
based on abscissa and weight coefficients of the Gauss quadrature convention. Using Mathematical transformation based on shape functions, the tetrahedron T in
(x, y, z) space is mapped into a unit tetrahedron V : {(u, v, w)| 0 u, v, w
1, u + v + w 1} in (u, v, w) space. Using isoparimetric transformation this unit
tetrahedron is then mapped into its equivalent 2-cube {(, , )| 1 , , 1}
in (, , ) space. Then the Gauss weight coefficients and abscissas are used over
the standard 2-cube to obtain well-organized quadrature rules for tetrahedron.
Further, a technique based on algebraic manipulation is developed for higher
order schemes so that they are now free from crowding of Gauss points. The
efficiency and accuracy of the derived formulae are verified by applying them to
the integration of different types of functions and to the evaluation of integrals
for element stiffness matrices in three-dimensional elasticity problem. The results
are found accurate in view of accuracy and efficiency. Thus we believe that the
formulae so presented will find better application in numerical solution procedure
of real problems in science and engineering.
8.2
Problem Statement over Tetrahedral Domain
The finite element solution for three dimensional problems with tetrahedron element requires the numerical integration of terms containing product of shape
functions and the product of their global derivatives over an arbitrary tetrahedron T with coordinates (x1 , y1 , z1 ), (x2 , y2 , z2 ), (x3 , y3 , z3 ) and (x4 , y4 , z4 ) in the
natural coordinate space (x, y, z). The numerical integration of an arbitrary function f , over the tetrahedron T is given by
ZZZ
III =
f (x, y, z) dx dy dz
(8.1)
where, T is a tetrahedron in Euclidian three dimensional spaces R3 . The

integral can be a trivariate monomial, polynomial, a rational function or of any
type depending on the problem under consideration.
172
8.3
Integration Over Arbitrary Tetrahedron
In this section, we first obtain the volume integral of a function f over an arbitrary
tetrahedron T in the Cartesian coordinate system (x, y, z) by transforming it to a
normalized tetrahedron V . The transformation from Cartesian coordinate system
(x, y, z) to tetrahedron V = {(1, 0, 0), (0, 1, 0), (0, 0, 1), (0, 0, 0)} is comfortably
accomplished by using shape functions:
Figure 8.1: The Arbitrary Tetrahedron and the Transformed Unit Tetrahedron.
x
x1 x2 x3 x4

y = y1 y2 y3 y4
z
z1 z2 z3 z4
N1
N2
N3
N4
(8.2)
where
N1 = u,
N2 = v,
N3 = w,
N4 = 1 u v w
where (xi , yi , zi ) refer to the Cartesian coordinates of vertex i of the tetrahedron. We can rewrite the relations (8.2) as:
173
x(u, v, w) = x4 + u(x1 x4 ) + v(x2 x4 ) + w(x3 x4 )

y(u, v, w) = y4 + u(y1 y4 ) + v(y2 y4 ) + w(y3 y4 )
z(u, v, w) = z4 + u(z1 z4 ) + v(z2 z4 ) + w(z3 z4 )
(8.3)
If we consider the mapping (see Fig. 8.1) between the three dimensional spaces
(x, y, z) and (u, v, w) by the Eq.(8.3), we have the Jacobian of transformation and
the volume element

(x1 x4 ) (x2 x4 ) (x3 x4 )

Jacobian = (y1 y4 ) (y2 y4 ) (y3 y4 )

(z1 z4 ) (z2 z4 ) (z3 z4 )

= J (say)

(8.4)
dx dy dz = |J| du dv dw . So, using Eq.(8.3) and Eq.(8.4), Eq.(8.1) becomes
ZZZ
III =
Z ZTZ
|J| f {x(u, v, w), y(u, v, w), z(u, v, w)} du dv dw
=
V
ZZZ
= |J|
f (u, v, w) du dv dw
(8.5)
where V is the orthogonal tetrahedron with vertices (1,0,0), (0,1,0), (0,0,1),

(0,0,0).
8.4
Integration over Orthogonal Tetrahedron and

Quadrature Formulae (GQTTS)
Generally, the integral of Eq.(8.5) is evaluated directly by numerical integration

schemes. Through several inspections and rigorous study of the nature of the
integrands, it is observed that due to ineffective contribution of Gauss points the
schemes become inefficient in view of accuracy. The main objective of this section
174

is to present suitable Gauss quadrature schemes in the sense that the scheme
ensures the effective use of all the Gauss points. To do so, without detailed
derivation, by use of calculus and geometrical concept the integral of Eq.(8.5) can
be written as:
1
1u
1uv
III = |J|
u=0
or
v=0
1u
f (u, v, w) dw dv du
(8.6)
f (u, v, w) dv dw du
(8.7)
f (u, v, w) dw du dv
(8.8)
f (u, v, w) du dw dv
(8.9)
f (u, v, w) dv du dw
(8.10)
(8.11)
w=0
1uw
III = |J|
u=0
or
Z
w=0
v=0
1v
1uv
III = |J|
v=0
or
Z
u=0
w=0
1v
1vw
III = |J|
v=0
or
w=0
u=0
1w
1uw
III = |J|
w=0
or
Z
u=0
v=0
1w
1vw
III = |J|
w=0
v=0
u=0
It is now clear that either all or at least one form of the integral III described in
Eq.(8.6) to Eq.(8.11) will ensure the effective use of Gauss points of the quadrature
formulae. Thus, considering the presence or absence of the independent variables
of the integrands we need six quadrature formulae for the integrals mentioned
in Eq.(8.6) to Eq.(8.11). Using isoparametric transformation, the domain of the
integrals can be transformed into equivalent 2-cube given by {(, , )| 1
, , 1} in (, , ). Now, the task is to present quadrature formulae one by
one.
Formula - 1: For the integral III
Here, we are required to evaluate the integral of Eq.(8.6) given as
Z
1u
1uv
III = |J|
u=0
v=0
w=0
175

Using transformation
u=
1+
2
(1 )(1 + )
4
v=
w=
(1 )(1 )(1 + )
8
the determinant of the Jacobian and the differential area are:

(u, v, w)

(, , )

u v w v w
u v w v w
u v w v w

= {
}
{
}+
{
}

(1 )2 (1 )
=
64
and
(u, v, w)
(1 )2 (1 )
d d d =
d d d
(, , )
64
dw dv du =
Using all these, the Eq.(8.6) becomes

Z
III = |J|
f (, , )
=1
=1
=1
(1 )2 (1 )
d d d
64

Here we required to evaluate the integral of Eq.(8.7)
Z
1u
1uw
III = |J|
u=0
w=0
v=0
u=
1+
2
v=
(1 )(1 + )(1 )
8
w=
(1 )(1 + )
4
the determinant of the Jacobian and the differential area are:

(u, v, w) (1 )2 (1 )

(, , ) =
64
176
(8.12)

and
(u, v, w)
(1 )2 (1 )
d d d =
d d d
(, , )
64
dv dw du =
Then Eq.(8.7) becomes

Z
III = |J|
f (, , )
=1
=1
=1
(1 )2 (1 )
d d d
64
(8.13)

By using the following transformation equations,
u=
(1 + )(1 )
4
1+
2
v=
w=
(1 )(1 )(1 + )
8
the integral in Eq.(8.8) reduces to

Z
III = |J|
f (, , )
=1
=1
=1
(1 )(1 )2
d d d
64
(8.14)

By using the following transformation equations,
u=
(1 + )(1 )(1 )
8
v=
1+
2
w=
(1 )(1 + )
4
the integral of Eq.(8.9) reduces to

Z
III = |J|
f (, , )
=1
=1
=1
(1 )2 (1 )
d d d
64

Similarly, the transformation equations
u=
(1 + )(1 )(1 )
8
v=
177
(1 + )(1 )
4
w=
1+
2
(8.15)

reduces the integral of Eq.(8.10) to
1
III = |J|
f (, , )
=1
=1
=1
(1 )(1 )2
d d d
64
(8.16)

Using transformation equations
u=
(1 + )(1 )
4
(1 )(1 + )(1 )
8
v=
w=
1+
2
the integral of Eq.(8.11) becomes

Z
III = |J|
f (, , )
=1
=1
=1
(1 )(1 )2
d d d
64
(8.17)
The m n p point conventional Gauss quadrature formulae for the integrals

III in Eq.(8.12) to Eq.(8.17) is given by
III = |J|
p
m X
n X
X
f (u(im , jn , kp ), v(im , jn , kp ), w(im , jn , kp ))
i=1 j=1 k=1
Wim Wjn Wkp

mnp
X
(u, v, w)
= |J|{

f (ur , vr , wr )Gr }
(, , )
r=1
(8.18)
Here (ur , vr , wr ) are the new Gauss points and Gr are the corresponding
weights for the integral over unit tetrahedron where im , i = 1, m; jn , j =
1, n; kp , k = 1, p are the m-th, n-th and k-th order one dimensional Gaussian
quadrature points respectively and Wim , Wjn , Wkp are the corresponding weights
respectively. The new Gauss points (ur , vr , wr ) and the corresponding weights
Gr can be computed by the following algorithm:
178
step 1. r 1
step 2. i = 1, m
step 3. j = 1, n
step 4. k = 1, p
For Formula 1.
(1 im )(1 + jn )
1 + im
, vr =
,
ur =
2
4
(1 im )2 (1 jn ) m n p
Gr =
Wi Wj Wk
64
(1 im )(1 jn )(1 + kp )
wr =
,
8
For Formula 2.
(1 im )(1 + jn )(1 kp )
1 + im
ur =
, vr =
,
2
8
(1 im )2 (1 kp ) m n p
Wi Wj Wk
Gr =
64
wr =
(1 im )(1 + kp )
,
4
For Formula 3.
(1 + im )(1 jn )
1 + jn
, vr =
,
4
2
(1 im )(1 jn )2 m n p
Gr =
Wi Wj Wk
64
ur =
wr =
(1 im )(1 jn )(1 + kp )
,
8
For Formula 4.
(1 + im )(1 jn )(1 kp )
1 + jn
, vr =
,
ur =
8
2
(1 jn )2 (1 kp ) m n p
Gr =
Wi Wj Wk
64
179
(1 jn )(1 + kp )
wr =
,
4

For Formula 5.
(1 + im )(1 jn )(1 kp )
(1 + jn )(1 kp )
ur =
, vr =
,
8
4
(1 jn )(1 kp )2 m n p
Gr =
Wi Wj Wk
64
1 + kp
wr =
,
2
For Formula 6.
(1 im )(1 + jn )(1 kp )
(1 + im )(1 kp )
ur =
, vr =
,
4
8
(1 im )(1 kp )2 m n p
Wi Wj Wk
Gr =
64
r=r+1
wr =
1 + kp
,
2

In order to evaluate the integral III in Eq.(8.6) to Eq.(8.11) new Gaussian
quadrature formulae are presented using Eq.(8.12) to Eq.(8.17). Since, for all the
presented Gauss quadrature formulae new Gauss points and corresponding weights
are calculated by simple algorithms that utilizes Gauss points and weights of onedimensional quadrature formulae. Hence, higher order quadrature formula for the
tetrahedral domain integral are now obtained.
Here, we mention that the algorithms are different for the six different formulae and for which computed Gauss points and weights are also different. But,
computed Gauss points and weights for one formula can be used for the other ones
with simple manipulation in computer code. For clarity and reference, computed
Gauss points and weights for (m = n = p = 5) based on the algorithm for the
formula -1 are listed in Table 8.1 and Fig. 8.2 and 8.3 exhibits the arrangement
of Gaussian points for 5 5 5 = 125 points and 10 10 10 = 1000 points.
It is seen in Fig. 8.2 and 8.3 that there are crowding of Gauss points and such
crowding will be severe for increasing Gauss points in the higher order formulae.
Ultimately, such crowding seriously causes of error germane in the evaluation of
the tetrahedral domain integral. To overcome this situation further improvement
180

is required.
Figure 8.2: Distribution of Gauss Points for GQTTS (m =5, 125 points).
8.5
Efficient Gaussian Quadrature formula for

Tetrahedron (GQTTM)
It is clearly noticed in Eq.(8.12) to Eq.(8.17) as well as in all algorithm that for

each i (i = 1, 2, ..., m); j varies from 1 to n and k varies from 1 to p hence at the
terminal value i = m (say) i.e. near the corner side of the tetrahedron there are
crowding of n k Gauss points as shown in Table 8.1 and Fig. 8.2 and 8.3. To
avoid such crowding, keeping the generality, we can select the terminal values of
the inner dummy indices depending on the present and terminal value of its outer
indices. Therefore, the new Gaussian quadrature formulae now can be written as:
181
Figure 8.3: Distribution of Gauss Points for GQTTS (m =10, 1000 points).
III = |J|
m (n=m+2i)
X
X (p=n+2j)
X
i=1
j=1
f (u(i0 , j0 , j0 ), v(i0 , j0 , j0 ), w(i0 , j0 , j0 ))
k=1

TP
X
(u, v, w)

= |J|{
f (u0r , vr0 , wr0 )G0r }
Wi Wj Wk
(, , )
r=1
(8.19)
Where (u0r , vr0 , wr0 ) are the new Gaussian points, G0r is the corresponding weights.
For a given value of m, total Gaussian points TP can be expressed as
m
X
(r + 2)(r + 3)
m3 + 9m2 + 20m
m(m + 4)(m + 5)
TP =
{
1} =
=
2
6
6
r=1
182

Now six cases arise for the evaluation of Eq.(8.19). All the Gaussian points
and corresponding weights now can be computed by the following algorithm:
step 1. r 1
step 2.
For Formula 1.
i = 1, m; j = 1, (n = m i + 2); k = 1, (p = n j + 2)
(1 i0 )(1 + j0 )
1 + i0
, vr0 =
,
2
4
(1 i0 )2 (1 j0 )
G0r =
Wi Wj Wk
64
u0r =
wr0 =
(1 i0 )(1 j0 )(1 + k0 )
,
8
For Formula 2.
i = 1, m; k = 1, (n = m i + 2); j = 1, (p = n k + 2)
(1 i0 )(1 + j0 )(1 k0 )
1 + i0
, vr0 =
,
2
8
(1 i0 )2 (1 k0 )
G0r =
Wi Wj Wk
64
u0r =
wr0 =
(1 i0 )(1 + k0 )
4
For Formula 3.
j = 1, m; i = 1, (n = m j + 2); k = 1, (p = n i + 2)
1 + j0
(1 + i0 )(1 j0 )
; vr0 =
,
4
2
(1 i0 )(1 j0 )2
G0r =
Wi Wj Wk
64
u0r =
183
wr0 =
(1 i0 )(1 j0 )(1 + k0 )
8

For Formula 4.
j = 1, m; k = 1, (n = m j + 2); i = 1, (p = n k + 2)
(1 + i0 )(1 j0 )(1 k0 )
1 + j0
0
=
, vr =
,
8
2
(1 j0 )2 (1 k0 )
0
Wi Wj Wk
Gr =
64
u0r
wr0
(1 j0 )(1 + k0 )
=
4
For Formula 5.
k = 1, m; j = 1, (n = m k + 2); i = 1, (p = n j + 2)
(1 + j0 )(1 k0 )
(1 + i0 )(1 j0 )(1 k0 )
, vr0 =
,
8
4
(1 j0 )(1 k0 )2
G0r =
Wi Wj Wk
64
u0r =
wr0 =
1 + k0
2
wr0 =
1 + k0
2
For Formula 6.
k = 1, m; i = 1, (n = m k + 2); j = 1, (p = n i + 2)
(1 i0 )(1 + j0 )(1 k0 )
(1 + i0 )(1 k0 )
, vr0 =
,
4
8
(1 i0 )(1 k0 )2
0
Gr =
Wi Wj Wk
64
r=r+1
u0r =

One can verify that T P = m(m+4)(m+5)
is greater than (m m m) for m = 2
6
and 3. But T P is about 50 percent less than (m m m) for 3 < m < 10 and it
is too less for m 10. Therefore, it is established that the new scheme GQTTM
substantially reduces the computational effort as it requires the evaluation of the
integrands at fewer number of Gauss points. As far as the accuracy is concern,
still the scheme GQTTM will give the same or better accuracy of the scheme
GQTTS because Gauss points and weights are calculated accordingly. For all
different quadrature scheme Gauss points are different but Gauss points of one
quadrature formula can be obtained from the others by simple manipulation.
Computed Gauss points and weights (for m = 5) based on above algorithm
184

listed in Table 8.3 and Fig. 8.4 and 8.5 exhibits the arrangement of Gaussian
points for m = 5 (75 points) and m = 10 (350 points). Influences of these cases on
the results of the integrals will be investigated in the result section. An explanation
for the total number of points for k = 1, 3 (for formula-6) is shown in Fig 8.6.
Suppose m = 3. For k = 1(i = 1, 4; j = 1, 5), number of points is 5(5+1)
1; for
2
k = 2(i = 1, 3; j = 1, 4) number of points is 4(4+1)
1; for k = 3(i = 1, 2; j = 1, 3)
2
3(3+1)
number of points is 2 1. The total number of point is then 14 + 9 +
5 = 28. Thus, the new Gaussian quadrature formula is now obtained which is
completely crowding free with significantly reduced number of Gauss points. For
m = 10, GQTTS presents mmm = 1000 Gaussian points and weights, whereas
= 350 Gaussian points and weights.
GQTTM presents m(m+4)(m+5)
6
Figure 8.4: Distribution of Gauss Points for GQTTM,(m =5, 75 points).
185
Figure 8.5: Distribution of Gauss Points for GQTTM,(m =10, 350 points).
8.6
To show the accuracy and efficiency of all the quadrature formula so presented,
we consider some typical integrals thoroughly studied by Rathod and Rao [1996,
1998]; Rathod et al. [2004, 2005]; Reddy and Shippy [1981].
Consider
Z
1x
1xy
p
(x + y + z) dz dy dx
I1 =
x=0
1
x=0
1
y=0
1x
z=0
1xy
y=0
1x
I2 =
Z
z=0
1xy
I3 =
dz dy dx
p
(x + y + z)
sin(x + 2y + 4z) dz dy dx
x=0
y=0
z=0
186
Figure 8.6: Calculation of Total Gauss Points for k =3 (28 points).

Z
1x
1xy
(1 + x + y + z)4 dz dy dx
I4 =
x=0
y=0
z=0
The results are summarized in Table 8.5.

Again, consider the monomial integrals
I5,,
ZZZ
=
x y z dz dy dx
Where V is tetrahedron in (x, y, z) space with spanning vertices points (5,

5, 0), (10, 10, 0), (8, 7, 8) and (10, 5, 0). The results are summarized in
Table 8.6.
Now, consider the following rational integrals for different values of a, b, c,
,,.
,,
Ia,b,c
1x
1xy
=
x=0
y=0
z=0
x y z
p
(ax + by + cz)
dz dy dx
The results are summarized in Table 8.7.

Linear Three Dimensional Elasticity Problem: From the principle of
virtual work, the stiffness matrix of an arbitrary element is given by
187
[B]T [D] [B] dV
[K]e =
V
(8.20)
Where [D] is a material property matrix, [B] is the strain displacement matrix and
the integration is performed over the volume element V 0 in global coordinates. The
symmetric material property matrix [D] for anisotropic materials can be written
as
D11
D12
D13
D14
D15
D16
D12
D22
D23
D24
D25
D26
D13
D23
D33
D34
D35
D36
D14
D24
D34
D44
D45
D46
D15
D25
D35
D45
D55
D56
D16
D26
D36
D46
D56
D66
The element stiffness matrix of Eq.(8.20) can be written as

Z
[K]e = t
[B1 B2 B3 ...Bne ]T [D][B1 B2 B3 ...Bne ]dV 0
We can visualize the [K]e as consisting of ne ne sub-matrices, [Kij ]e signifying

the stiffness relationship between nodes i and j, where ne is the number of nodes
in the element. Any such sub matrix is given by
Z
[Kij ]e =
[Bi ]T [D][Bj ]dV 0
V0
For a three dimensional element, the matrix [Bi ] has the form
Ni
x0
Ni
z0
0
0
Ni
y 0
0
Ni
z 0
0
Ni
y 0
188
Ni
0
z
Ni
y 0
Ni
x0
Ni
x0

In which Ni are the appropriate element shape functions. The element stiffness
relating nodes i and j for an anisotropic material under three dimensional loading
is given by:
Kij11 Kij12 Kij13
[Kij ]e = Kij21 Kij22 Kij23

Kij31 Kij32 Kij33
Where Kij11 , Kij12 , ...Kij33 are volume integrals over domain V 0 and they are triple
integrals of the product of global derivatives of shape functions. Hence we need
to evaluate
ZZZ
Ni Nj
dx0 dy 0 dz 0
t s
t, s = x0 , y 0 , z 0
V0
We shall evaluate these triple integrals, where V 0 is the tetrahedral element

with three plane surface and one curved surface. It can be shown that the evaluation of integrals now amounts to the evaluation of integrals of the type
IJpqr
xp y q z r
dx dy dz
J(x, y, z)
V
Z 1 Z 1x Z 1xy p q r
xy z
dx0 dy 0 dz 0
=
J(x,
y,
z)
x=0 y=0
z=0
ZZZ
(8.21)
Where J is the Jacobian of transformation and V is the tetrahedron: {(x, y, z)|0

x, y, z 1, x + y + z 1}. The transformation which maps a cubic curved tetrahedral element V 0 in to an orthogonal tetrahedral element V is given by
t=
20
X
Ni (x, y, z)ti ,
t = x0 , y 0 , z 0 and
i=1
189
ti = (x0i , yi0 , zi0 )

That is
9
t(x, y, z) = t4 + (t1 t4 )x + (t2 t4 )y + (t3 t4 )z + (t1 t2
4
9
9
+
t11 + t12 )xy + (t2 t3 + t13 + t14 )yz + (t1 t3
4
4
9
+
t15 + t16 )xz + {2(t1 + t2 + t3 ) 3(t11 + t12
4
+
t13 + t14 + t15 + t16 ) + 12t20 }xyz
For the 20 nodal data of Fig 8.7 we have
x0
= x + Rx(y + z) + Sxyz,
z0
= z + Rz(x + y) + Sxyz,
y 0 = y + Ry(x + z) + Sxyz,
J(x, y, z) = 1 + 2R(x + y + z) + R2 (x2 + y 2 + z 2 )

+(2R2 + s)(xy + yz + zx) +
+RS(x2 y + xy 2 + yz 2 + zy 2 + x2 z + xz 2 )
+4R3 xyz + R2 S(x2 yz + y 2 xz + z 2 xy)
Figure 8.7: Cubic tetrahedron V 0 with one curved surface.
190

Where R = 0.842329219213245 and S = 1.534481952840430. We have computed the values of the integrals of the Eq.(8.21) for different values of p, q, r.
Results are summarized in Table 8.8.
8.7
Results and Discussions
For clarity and reference, Gaussian points and weights for the schemes GQTTS
and GQTTM are presented in Tables 8.1 and 8.3 respectively. The scheme
GQTTM is so faster as it require function values at less number of Gauss points,
e.g. for m = 10, the scheme GQTTS require function values at 101010 = 1000
points where as the scheme GQTTM require function values at 10(10+4)(10+5)
= 350
6
Gauss points.
Computed results of Tables 8.5, 8.6 and 8.7 clearly establishes the best accuracy of the schemes so presented. At the same time the scheme GQTTM is always
in good agreement with the result of the scheme GQTTS. For example, in table
8.5 for the same accuracy GQTTS utilizes 1000 points and GQTTM utilizes 350
points. Further it is observed that the absence of any independent variable in the
explicit form of the integrand enhances faster convergence of one among the six
quadrature formulae for both GQTTS and GQTTM. For all case studies in Table
8.8 for IJpqr in Eq.(8.21) the method GQTTS require more Gaussian points than
the method GQTTM. Results are compared with the results obtained by Rathod
and Rao [1996, 1998]; Rathod et al. [2004, 2005]; Reddy and Shippy [1981].
It is found that the developed integration formulae are always consistent, reliable
and provides considerable savings in computational effort, in comparison with the
existing numerical methods that employs Gaussian quadrature procedures.
8.8
Conclusions
This chapter presented suitable Gaussian Quadrature formulae to evaluate tetrahedral domain integrals in Euclidean three dimensional space. It is also demonstrated that there are two set of quadrature formulae and each set contains six
quadrature formulae for which Gaussian points and corresponding weights are
different. Derived formulae substantially reduced the number of Gaussian in191

tegration points for higher order schemes and the efficiency and accuracy of the
formulae in evaluating various types of tetrahedral domain of integration is demonstrated practically. In all the test cases the efficiency and accuracy of the derived
formulae are found satisfactory. Thus, it ensures the balance between accuracy
and efficiency of evaluation of tetrahedral domain integrals encountered in science and engineering. Complete computer coding using FORTRAN compatible
with the formulation is also appended (program 8: GQTTS.for and program 9:
GQTTM.for).
192
Table 8.1: Computed Gauss points (u, v, w ) and corresponding weights G for
n n n point method (GQTTS).
u
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
v
0.044709521703645
0.044709521703645
0.044709521703645
0.044709521703645
0.044709521703645
0.219940124839679
0.219940124839679
0.219940124839679
0.219940124839679
0.219940124839679
0.476544961484666
0.476544961484666
0.476544961484666
0.476544961484666
0.476544961484666
0.733149798129653
0.733149798129653
0.733149798129653
0.733149798129653
0.733149798129653
0.908380401265687
0.908380401265687
0.908380401265687
0.908380401265687
0.908380401265687
0.036084856923188
0.036084856923188
0.036084856923188
0.036084856923188
0.036084856923188
0.177512700518577
0.177512700518577
0.177512700518577
0.177512700518577
0.177512700518577
0.384617327526421
0.384617327526421
0.384617327526421
0.384617327526421
0.384617327526421
0.591721954534264
0.591721954534264
0.591721954534264
0.591721954534264
0.591721954534264
0.733149798129653
0.733149798129653
0.733149798129653
0.733149798129653
0.733149798129653
0.023455038515334
0.023455038515334
0.023455038515334
0.023455038515334
0.023455038515334
0.115382672473579
0.115382672473579
0.115382672473579
0.115382672473579
0.115382672473579
w
0.042612194596523
0.209622716641314
0.454190200632844
0.698757684624373
0.865768206669165
0.034392113505281
0.169185566063329
0.366574899064827
0.563964232066324
0.698757684624372
0.022354760851822
0.109970062419839
0.238272480742333
0.366574899064827
0.454190200632844
0.010317408198364
0.050754558776350
0.109970062419839
0.169185566063329
0.209622716641314
0.002097327107122
0.010317408198364
0.022354760851822
0.034392113505281
0.042612194596523
0.034392113505281
0.169185566063329
0.366574899064827
0.563964232066324
0.698757684624372
0.027757722467940
0.136548920950906
0.295860977267132
0.455173033583358
0.563964232066324
0.018042428461594
0.088756350259289
0.192308663763210
0.295860977267132
0.366574899064827
0.008327134455248
0.040963779567671
0.088756350259289
0.136548920950906
0.169185566063329
0.001692743417907
0.008327134455248
0.018042428461594
0.027757722467940
0.034392113505281
0.022354760851822
0.109970062419839
0.238272480742333
0.366574899064827
0.454190200632844
0.018042428461594
0.088756350259289
0.192308663763210
0.295860977267132
0.366574899064827
193
G
0.001439311118758
0.002907629360183
0.003455952678910
0.002907629360183
0.001439311118758
0.002346734777065
0.004740764418080
0.005634781968811
0.004740764418080
0.002346734777065
0.001813025505580
0.003662589777903
0.004353284200533
0.003662589777903
0.001813025505580
0.000704005022098
0.001422198192795
0.001690397586996
0.001422198192795
0.000704005022098
0.000070841369555
0.000143110438979
0.000170098332251
0.000143110438979
0.000070841369555
0.001894039243550
0.003826249962299
0.004547807567383
0.003826249962299
0.001894039243550
0.003088149395941
0.006238535737859
0.007415004329977
0.006238535737859
0.003088149395941
0.002385822920682
0.004819728467296
0.005728637128330
0.004819728467296
0.002385822920682
0.000926424538886
0.001871519752855
0.002224452604614
0.001871519752855
0.000926424538886
0.000093222606464
0.000188323972528
0.000223838273981
0.000188323972528
0.000093222606464
0.000951130350813
0.001921429284706
0.002283774120164
0.001921429284706
0.000951130350813
0.001550777064587
0.003132807678149
0.003723595323417
0.003132807678149
0.001550777064587
Table 8.2: Continuation of Table 8.1

u
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
v
0.250000000000000
0.250000000000000
0.250000000000000
0.250000000000000
0.250000000000000
0.384617327526421
0.384617327526421
0.384617327526421
0.384617327526421
0.384617327526421
0.476544961484666
0.476544961484666
0.476544961484666
0.476544961484666
0.476544961484666
0.010825220107480
0.010825220107480
0.010825220107480
0.010825220107480
0.010825220107480
0.053252644428581
0.053252644428581
0.053252644428581
0.053252644428581
0.053252644428581
0.115382672473579
0.115382672473579
0.115382672473579
0.115382672473579
0.115382672473579
0.177512700518577
0.177512700518577
0.177512700518577
0.177512700518577
0.177512700518577
0.219940124839679
0.219940124839679
0.219940124839679
0.219940124839679
0.219940124839679
0.002200555327023
0.002200555327023
0.002200555327023
0.002200555327023
0.002200555327023
0.010825220107480
0.010825220107480
0.010825220107480
0.010825220107480
0.010825220107480
0.023455038515334
0.023455038515334
0.023455038515334
0.023455038515334
0.023455038515334
0.036084856923188
0.036084856923188
0.036084856923188
0.036084856923188
0.036084856923188
0.044709521703645
0.044709521703645
0.044709521703645
0.044709521703645
0.044709521703645
w
0.011727519257667
0.057691336236790
0.125000000000000
0.192308663763210
0.238272480742333
0.005412610053740
0.026626322214291
0.057691336236790
0.088756350259289
0.109970062419839
0.001100277663512
0.005412610053740
0.011727519257667
0.018042428461594
0.022354760851822
0.010317408198364
0.050754558776350
0.109970062419839
0.169185566063329
0.209622716641314
0.008327134455248
0.040963779567671
0.088756350259289
0.136548920950906
0.169185566063329
0.005412610053740
0.026626322214291
0.057691336236790
0.088756350259289
0.109970062419839
0.002498085652232
0.012288864860910
0.026626322214291
0.040963779567671
0.050754558776350
0.000507811909116
0.002498085652232
0.005412610053740
0.008327134455248
0.010317408198364
0.002097327107122
0.010317408198364
0.022354760851822
0.034392113505281
0.042612194596523
0.001692743417907
0.008327134455248
0.018042428461594
0.027757722467940
0.034392113505281
0.001100277663512
0.005412610053740
0.011727519257667
0.018042428461594
0.022354760851822
0.000507811909116
0.002498085652232
0.005412610053740
0.008327134455248
0.010317408198364
0.000103228219901
0.000507811909116
0.001100277663512
0.001692743417907
0.002097327107122
194
G
0.001198089532334
0.002420324733784
0.002876751714678
0.002420324733784
0.001198089532334
0.000465222935413
0.000939821730277
0.001117054144152
0.000939821730277
0.000465222935413
0.000046813628964
0.000094570715294
0.000112404944503
0.000094570715294
0.000046813628964
0.000170456069101
0.000344347420568
0.000409284761972
0.000344347420568
0.000170456069101
0.000277921277830
0.000561443635575
0.000667321173390
0.000561443635575
0.000277921277830
0.000214714662336
0.000433756571553
0.000515554769800
0.000433756571553
0.000214714662336
0.000083374558238
0.000168429403669
0.000200191969715
0.000168429403669
0.000083374558238
0.000008389667270
0.000016948415501
0.000020144562700
0.000016948415501
0.000008389667270
0.000003486737214
0.000007043744304
0.000008372059841
0.000007043744304
0.000003486737214
0.000005684974827
0.000011484521661
0.000013650282922
0.000011484521661
0.000005684974827
0.000004392061882
0.000008872639079
0.000010545849211
0.000008872639079
0.000004392061882
0.000001705455115
0.000003445281079
0.000004094995235
0.000003445281079
0.000001705455115
0.000000171613514
0.000000346685638
0.000000412063922
0.000000346685638
0.000000171613514
Table 8.3: Computed Gauss points (u, v, w) and corresponding weights G for
method GQTTM.
u
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.046910077030668
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
v
0.032181312753100
0.032181312753100
0.032181312753100
0.032181312753100
0.032181312753100
0.032181312753100
0.032181312753100
0.161448959877800
0.161448959877800
0.161448959877800
0.161448959877800
0.161448959877800
0.161448959877800
0.362832190643147
0.362832190643147
0.362832190643147
0.362832190643147
0.362832190643147
0.590257732326185
0.590257732326185
0.590257732326185
0.590257732326185
0.791640963091532
0.791640963091532
0.791640963091532
0.920908610216232
0.920908610216232
0.036084856923188
0.036084856923188
0.036084856923188
0.036084856923188
0.036084856923188
0.036084856923188
0.177512700518577
0.177512700518577
0.177512700518577
0.177512700518577
w
0.023433480857716
0.119013078326949
0.273581157949139
0.460454305108116
0.647327452267094
0.801895531889283
0.897475129358516
0.026729949407128
0.134100263792109
0.301370120404256
0.490270842687276
0.657540699299423
0.764911013684404
0.027689035691369
0.136211029207980
0.295128866163093
0.454046703118206
0.562568696634817
0.025192108132559
0.119738061911055
0.243094128732092
0.337640082510588
0.018195566651977
0.080724479938900
0.143253393225823
0.006800711586111
0.025380601166989
0.024754981014778
0.124192134960240
0.279103095011448
0.454046703118206
0.608957663169414
0.708394817114875
0.027757722467940
0.136548920950906
0.295860977267132
0.455173033583358
195
G
0.000576651691559
0.001245647212280
0.001700451805538
0.001861350220022
0.001700451805538
0.001245647212280
0.000576651691559
0.001381103480046
0.002908218533639
0.003772009211444
0.003772009211444
0.002908218533639
0.001381103480046
0.001847056770230
0.003731338155492
0.004434997208027
0.003731338155492
0.001847056770230
0.001666971143698
0.003125174680918
0.003125174680918
0.001666971143698
0.000913358455696
0.001461373529114
0.000913358455696
0.000155625745624
0.000155625745624
0.001369601055914
0.002883997638157
0.003740594295469
0.003740594295469
0.002883997638157
0.001369601055914
0.003088149395941
0.006238535737859
0.007415004329977
0.006238535737859

u
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.230765344947158
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.500000000000000
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.769234655052841
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
0.953089922969332
v
0.177512700518577
0.384617327526421
0.384617327526421
0.384617327526421
0.384617327526421
0.591721954534264
0.591721954534264
0.591721954534264
0.733149798129653
0.733149798129653
0.034715922101487
0.034715922101487
0.034715922101487
0.034715922101487
0.034715922101487
0.165004739103786
0.165004739103786
0.165004739103786
0.165004739103786
0.334995260896214
0.334995260896214
0.334995260896214
0.465284077898513
0.465284077898513
0.026007638687364
0.026007638687364
0.026007638687364
0.026007638687364
0.115382672473579
0.115382672473579
0.115382672473579
0.204757706259795
0.204757706259795
0.009913265714653
0.009913265714653
0.009913265714653
0.036996811316015
0.036996811316015
w
0.563964232066324
0.026704690362579
0.126927363566585
0.257689963959836
0.357912637163842
0.020005976974413
0.088756350259289
0.157506723544164
0.007625627532458
0.028459229390730
0.021826511935363
0.107371440734671
0.232642038949257
0.357912637163842
0.443457565963151
0.023259338763280
0.110551611250369
0.224443649645845
0.311735922132934
0.018596308892467
0.082502369551893
0.146408430211319
0.007336337565514
0.027379584535973
0.014216705160388
0.067571983801774
0.137185722458021
0.190541001099406
0.013003819343682
0.057691336236790
0.102378853129897
0.005496060745114
0.020511577942250
0.004169602249037
0.018498405658008
0.032827209066978
0.002094919542875
0.007818346171778
196
G
0.003088149395941
0.003502853052757
0.006567016899386
0.006567016899386
0.003502853052757
0.002172316996694
0.003475707194710
0.002172316996694
0.000393465715983
0.000393465715983
0.001363446374903
0.002754370934147
0.003273792643220
0.002754370934147
0.001363446374903
0.002702019388299
0.005065644124426
0.005065644124426
0.002702019388299
0.002125571182931
0.003400913892690
0.002125571182931
0.000429372505117
0.000429372505117
0.000546317509560
0.001024215479118
0.001024215479118
0.000546317509560
0.000786674148254
0.001258678637206
0.000786674148254
0.000199483844893
0.000199483844893
0.000028555008472
0.000045688013556
0.000028555008472
0.000013772324628
0.000013772324628
197
Exact
Value
GQTTM
GQTTS
Method
8
27
64
125
216
343
512
729
1000
14
28
48
75
110
154
208
273
350
Points
0.14285714285714
I1
0.143127410953799
0.142875312759849
0.142860037924269
0.142857834882225
0.142857355360141
0.142857220464426
0.142857175072351
0.142857157618562
0.142857150174568
0.143058065243440
0.142868320929028
0.142858821339166
0.142857541720436
0.142857267261819
0.142857189383667
0.142857162683365
0.142857152181787
0.142857147595743
0.2
I2
0.197660776240555
0.199583323221216
0.199881018522955
0.199956079429135
0.199980842555110
0.199990593978398
0.199994953918813
0.199997099779897
0.199998238575602
0.198325920889004
0.199730676421398
0.199924597609756
0.199971858116752
0.199987455481037
0.199993687508278
0.199996530092608
0.199997959524722
0.199998734389269
I4
0.020103982733156
0.020798626362385
0.020831943224904
0.020833282126123
0.020833331531397
0.020833333271718
0.020833333331268
0.020833333333265
0.020833333333331
0.020334811873381
0.020811918661132
0.020832494969141
0.020833301701749
0.020833332126155
0.020833333282340
0.020833333330381
0.020833333333016
0.020833333333269
0.131902326890181 0.020833333333333
I3
0.131949528497795
0.131902664864685
0.131902289803505
0.131902327570123
0.131902326884088
0.131902326890216
0.131902326890182
0.131902326890182
0.131902326890182
0.133142199123991
0.131877204546620
0.131902589866991
0.131902325527761
0.131902326955921
0.131902326902415
0.131902326892685
0.131902326890580
0.131902326890160
Value of I
Table 8.5: Calculated values of the integrals I1 ,I2 ,I3 , I4

198
Exact
Value
GQTTM
GQTTS
Method
8
27
64
125
216
343
512
729
1000
14
28
48
75
110
154
208
273
350
Points
15721.6666666667
I52,1,0
15550.9773662551233
15721.6666666665187
15721.6666666666715
15721.6666666666587
15721.6666666666897
15721.6666666666751
15721.6666666667060
15721.6666666666787
15721.6666666667078
15550.1862778507402
15721.6666928552531
15721.6666681218612
15721.6666668061934
15721.6666666861729
15721.6666666702421
15721.6666666674991
15721.6666666669025
15721.6666666667388
109662.063492063
426917356.623377
Value of I
I52,2,0
I54,4,0
107484.1792409692279 387905448.6299038529396
109657.4916666656208 425756672.2764838933944
109662.0634920635493 426910389.2180011272430
109662.0634920634620 426917352.2770764231681
109662.0634920636221 426917356.6233774423599
109662.0634920635347 426917356.6233763694763
109662.0634920638694 426917356.6233769655227
109662.0634920634765 426917356.6233772635459
109662.0634920637239 426917356.6233774423599
107434.4429454763012 384070078.9281456470489
109657.4927840148156 425742213.6549844145774
109662.0635520677169 426910391.3103920221328
109662.0634976892179 426917352.7276153564453
109662.0634928344079 426917356.6618708372116
109662.0634922022436 426917356.6281496286392
109662.0634920942829 426917356.6241724491119
109662.0634920715237 426917356.6235436201095
109662.0634920659504 426917356.6234186291694
330565.079365081
I51,2,3
281238.2292672184994
329105.6422399957082
330562.9232264332240
330565.0793650789419
330565.0793650801642
330565.0793650793493
330565.0793650807463
330565.0793650799896
330565.0793650812120
278927.4392087566666
329272.2495105185662
330562.7560054035275
330565.0694366566604
330565.0784725357079
330565.0792490325402
330565.0793439878034
330565.0793599335593
330565.0793634889996
Table 8.6: Calculated values of the integrals I52,1,0 , I52,2,0 , I54,4,0 and I51,2,3
GQTTM
Formula
1-6
GQTTS
199
Formula 6
Formula 5
Formula 4
Formula 3
Formula 2
Formula 1
Case
Method
Points
8
27
64
125
216
343
512
729
1000
110
208
350
544
110
208
350
544
110
208
350
544
110
208
350
544
110
208
350
544
110
208
350
544
222
I111
0.0000269860345093
0.0000202862454377
0.0000231837802848
0.0000233422437574
0.0000233426594248
0.0000233426697419
0.0000233426703367
0.0000233426703927
0.0000233426704000
0.0000233426460487
0.0000233426700306
0.0000233426703855
0.0000233426704001
0.0000233426460487
0.0000233426700306
0.0000233426703855
0.0000233426704001
0.0000233426460487
0.0000233426700306
0.0000233426703855
0.0000233426704001
0.0000233426460487
0.0000233426700306
0.0000233426703855
0.0000233426704001
0.0000233426460487
0.0000233426700306
0.0000233426703855
0.0000233426704001
0.0000233426460487
0.0000233426700306
0.0000233426703855
0.0000233426704001
022
I111
0.0006015651986381
0.0008488381955233
0.0008546314116172
0.0008546969203429
0.0008547004410159
0.0008547007916240
0.0008547008422886
0.0008547008517505
0.0008547008538885
0.0008547006761305
0.0008547008477305
0.0008547008541789
0.0008547008546419
0.0008547006761305
0.0008547008477305
0.0008547008541789
0.0008547008546419
0.0008547003467509
0.0008547008418419
0.0008547008538989
0.0008547008546175
0.0008547001755466
0.0008547008358275
0.0008547008535749
0.0008547008545902
0.0008547001755466
0.0008547008358275
0.0008547008535749
0.0008547008545902
0.0008547003467509
0.0008547008418419
0.0008547008538989
0.0008547008546175
222
I101
0.0000333182575959
0.0000258267656879
0.0000288727421023
0.0000292327112870
0.0000292495089719
0.0000292513882514
0.0000292517007873
0.0000292517681974
0.0000292517857518
0.0000292495203673
0.0000292516825132
0.0000292517802690
0.0000292517912085
0.0000292484738197
0.0000292516351558
0.0000292517791266
0.0000292517918411
0.0000292511271472
0.0000292517169499
0.0000292517800403
0.0000292517906566
0.0000292511271472
0.0000292517169499
0.0000292517800403
0.0000292517906566
0.0000292495203673
0.0000292516825132
0.0000292517802690
0.0000292517912085
0.0000292484738197
0.0000292516351558
0.0000292517791266
0.0000292517918411
Table 8.7: Calculated results of Iabc

220
I110
0.0009230975391085
0.0008883907833675
0.0009316174070870
0.0009323393056688
0.0009323923274447
0.0009323992493688
0.0009324005189310
0.0009324008123282
0.0009324008927173
0.0009323918072337
0.0009324004766320
0.0009324008890532
0.0009324009261585
0.0009323939970271
0.0009324005524678
0.0009324008837150
0.0009324009227078
0.0009323918072337
0.0009324004766320
0.0009324008890532
0.0009324009261585
0.0009323939970271
0.0009324005524678
0.0009324008837150
0.0009324009227078
0.0009323981782032
0.0009324006198537
0.0009324008787228
0.0009324009203554
0.0009323981782032
0.0009324006198537
0.0009324008787228
0.0009324009203554
GQTTM
GQTTS
Method
Points
27
64
125
216
343
512
729
1000
14
28
48
75
110
154
208
273
350
M
3
4
5
6
7
8
9
10
2
3
4
5
6
7
8
9
10
IJ0,0,0
0.057883923504
0.057917371950
0.057918451577
0.057918486772
0.057918487951
0.057918487991
0.057918487993
0.057918487993
0.057106321695
0.057890655765
0.057917552866
0.057918455127
0.057918486659
0.057918487915
0.057918487985
0.057918487992
0.057918487993
Value of the Integrals

IJ2,1,2
IJ1,2,1
0.000246800681 0.000668334789
0.000278299062 0.000673911632
0.000277145835 0.000674438418
0.000277101003 0.000674419407
0.000277102668 0.000674419323
0.000277102679 0.000674419323
0.000277102679 0.000674419323
0.000277102679 0.000674419323
0.000329956379 0.000912787302
0.000246998166 0.000666851269
0.000278276097 0.000673943150
0.000277145427 0.000674441631
0.000277101005 0.000674419712
0.000277102668 0.000674419357
0.000277102679 0.000674419328
0.000277102679 0.000674419324
0.000277102679 0.000674419324
Table 8.8: Calculated values of the integrals IJp,q,r

IJ3,3,3
0.000008025783
0.000011177365
0.000010628963
0.000010581796
0.000010595557
0.000010594735
0.000010594742
0.000010594743
0.000015749900
0.000007736142
0.000011200562
0.000010613468
0.000010581637
0.000010595651
0.000010594748
0.000010594743
0.000010594743
200
Chapter 9
Accurate Gaussian
Quadrature Rules for
Numerical Integration
over Tetrahedron
Chapter 9
Accurate Gaussian Quadrature
Rules for Numerical Integration
over Tetrahedron
This chapter presents an accurate Gaussian Quadrature formulae to evaluate
tetrahedral domain integrals in Euclidean three dimensional space (x, y, z). The
tetrahedral element T in (x, y, z) space is transformed into a standard tetrahedron
V : {(u, v, w)| 1 u, v, w 1, u+v +w 1} whose base triangle is a standard
triangle in w = 1 plane. The derivation is based on the transformation of the arbitrary tetrahedron in global space into its contiguous standard tetrahedron in the
local space. As a result, it presents two quadrature schemes GQTSS and GQTSM
and each scheme contains six Gaussian quadrature formulae. It is shown that the
scheme GQTSS suffers from huge crowding of Gaussian points whereas GQTSM
is free of such crowdings. The efficiency and accuracy of the schemes are tested by
evaluating the integral of some polynomial and nonpolynomial functions. Finally,
element matrices for tetrahedral finite elements are evaluated through the newly
derived formulae. In all the test cases the results are found satisfactory and the
scheme GQTSM is faster. Thus, it ensures the balance between accuracy and
efficiency of evaluation of tetrahedral domain integrals encountered in practice.
202
9. Numerical Integration over Tetrahedron
9.1
Introduction
Different types of complicated problems concerning triple integration regularly

arise in a large number of engineering applications. Such integrals cannot be
evaluated analytically or explicitly for unbalanced, intricate or distorted type of
domain and we are resort to numerical integration scheme (Gauss [1814]; Hughes
[1987]; Zienkiewicz and Taylor [2000]). The extensively used cubic and tetrahedral elements and the utilization of such elements can be further enhanced by
improvement of numerical integration schemes. Mathematically, the problem in
three dimensions can be defined as the estimation of the triple integral of a function
f (x, y, z) over the arbitrary tetrahedral domain T in Euclidean Three-Dimensional
Space (x, y, z).
The basic difficulty of integrating an arbitrary function of two variables over
the surface of the triangle were first discussed by Hammer and Stroud [1956,
1958]; Hammer et al. [1956]. Lee and Requicha [1982] have given a good
inkling of existing methods for evaluating volume integrals. In Zienkiewicz and
Taylor [2000] a numerical integration formulae for tetrahedral regions with a
degree of precision up to 3 are listed. The evaluation of volume integrals by
transforming it to a surface integral over the boundary of the integration domain
was discussed by Timmer and Stern [1980]. Bernardini [1991]; Cattani and
Paoluzzi [1990a,b]; Lien and Kajiya [1984] have presented formulae and algorithms for computing integrals of polynomials over tetrahedral. Rathod et al.
[2004] derived explicit integration formulae for computing volume integrals of
trivariate polynomials over an arbitrary tetrahedron in Euclidean space. They
evaluated this volume integral as a sum of four integrals over the unit triangle. Rathod et al. [2005] proposed product formulae based on Gauss-Legendre
quadrature rule for the numerical integration of an arbitrary function over the
standard tetrahedron. The main problem in those methods was a huge crowding
of Gauss points within the tetrahedron. In previous chapters of this thesis integration scheme totally free from the crowding of Gaussian integration points for
triangular domains are nicely presented. Such developments are also needed for
the tetrahedral domain integrals.
Triple integral formulae have a great application in the FEM solution of prob-
203

lems in applied science and engineering. In the current study, we suggest an
alternative techniques for the evaluation of numerical integration of the triple
integral of a function f (x, y, z) over any arbitrary tetrahedron T in Euclidean
Three-Dimensional Space (x, y, z) having privileged precisions based on abscissa
and weight coefficients of the Gauss quadrature convention. Using Mathematical
transformation the tetrahedron T in (x, y, z) space is mapped into a standard
tetrahedron V : {(u, v, w)| 1 u, v, w 1, u + v + w 1} in (u, v, w)
space. Using isoparametric transformation this standard orthogonal tetrahedron
is mapped to a 2-cube {(, , )| 1 , , 1} in (, , ) space. Then
the Gauss weight coefficients and abscissas are used over the standard 2-cube to
obtain a well-organized quadrature rule. This anticipated new method removed
all kind of crowding of quadrature points and also reduce the total number of
points in the calculation process. The proposed method is easy for writing computer codes in programming language. The accomplishment of the formulae is
verified by applying them to the integration of different types of functions and
to the evaluation of integrals for element stiffness matrices in three-dimensional
elasticity problem. The obtained results are found accurate in view of accuracy
and efficiency. Hence it is believed that the formulae so derived will find better
place in numerical solution procedure of many practical problems.
9.2
Problem Statement over Tetrahedral Domain
Most computational studies of triple integrals deal with problems in which the
domain of integration is very simple, like a cube or a sphere, but the integrand
is complicated. However, in real applications, we confront the mixed problems:
both of the integrating function f (x, y, z) and the domain are very complicated.
In finite element method we divide a complex domain into finite number of simple
element. The finite element method for three dimensional problems with tetrahedron element requires the numerical integration of expressions containing product
of shape functions and their global derivatives over an arbitrary tetrahedron T
with vertices (x1 , y1 , z1 ), (x2 , y2 , z2 ), (x3 , y3 , z3 ) and (x4 , y4 , z4 ) in the natural coordinate space (x, y, z).
The numerical integration of an arbitrary function f , over the tetrahedron T
204
Figure 9.1: Discretization of a Three Dimensional Domain.

is given by
3
ZZZ
I =
(9.1)
where, T is a tetrahedron in Euclidian three dimensional spaces R3 . The

integrand function may be a trivariate monomial, polynomial, a rational function
or any other type of function.
9.3
Integration over Arbitrary Tetrahedron
In this section, we first obtain the volume integral of a function f over an arbitrary tetrahedron T in the cartesian coordinate system (x, y, z) by transforming
it to a normalized orthogonal tetrahedron S. The transformation from arbitrary
tetrahedron in (x, y, z)-space into a standard tetrahedral domain S with vertices
{(1, -1, -1), (-1, 1, -1), (-1, -1, 1), (-1, -1, -1)} is effortlessly accomplished by using
shape functions:
x1 x2 x3 x4
x

y = y1 y2 y3 y4
z1 z2 z3 z4
z
205
N1
N2
N3
N4
(9.2)
Figure 9.2: The Arbitrary Tetrahedron and The Transformed Standard Tetrahedron.
where (xi , yi , zi ) refer to the Cartesian coordinates of vertex i of the tetrahedron
and
N4
u+1
,
2
w+1
,
2
1 u v w
= 1 N1 N2 N3 =
2
N1 =
N2 =
v+1
,
2
N3 =
We can rewrite the relations given by Eq.(9.2) as:
x(u, v, w) = 12 {(x1 + x2 + x3 x4 ) + u(x1 x4 ) + v(x2 x4 ) + w(x3 x4 )}

y(u, v, w) = 12 {(y1 + y2 + y3 y4 ) + u(y1 y4 ) + v(y2 y4 ) + w(y3 y4 )}
z(u, v, w) = 21 {(z1 + z2 + z3 z4 ) + u(z1 z4 ) + v(z2 z4 ) + w(z3 z4 )}
(9.3)
If we consider the mapping (see Fig. 9.2) between the three dimensional spaces
(x, y, z) and (u, v, w) by the Eq.(9.3), we have the Jacobian and the volume element
206

(x1 x4 ) (x2 x4 ) (x3 x4 )
1
Jacobian = (y1 y4 ) (y2 y4 ) (y3 y4 )
8
(z1 z4 ) (z2 z4 ) (z3 z4 )

vol

=

8

(9.4)
and
vol
|du dv dw
8
So, using Eq.(9.3) and Eq.(9.4), Eq.(9.1) becomes
dx dy dz = |
ZZZ
=
Z ZTZ
|
=
Z ZSZ
=
vol
| f {x(u, v, w), y(u, v, w), z(u, v, w)} du dv dw
8
vol
8
(9.5)
where S is the orthogonal tetrahedron with vertices {(1,-1,-1),(-1,1,-1),(-1,-1,1),

(-1,-1,-1)}.
9.4
Integration Over Standard Tetrahedron
Through several inspections and rigorous study of the nature of the integrand,
it is observed that due to ineffective contribution of Gauss points the scheme
become inefficient in view of accuracy. The main objective of this chapter is to
present suitable alternative Gauss quadrature schemes in the sense that the scheme
ensures the effective use of all the Gauss points. By use of calculus and geometrical
concept the integral of Eq.(9.5) can be calculated in six ways (as shown in previous
chapter) by rearranging the order of variables in the integration process. Using
isoparametric transformations the domain of these integrals can be transformed
to a 2-cube given by {(, , )| 1 , , 1} in (, , ) space. These six cases
are as follows:
Case - 1: For the integral I 3
207

We are to evaluate the integral of Eq.(9.5) as
Z
vol
I =
8
3
1uv
u=1
v=1
w=1
u=
v=
(1 )(1 + )
(1 )(1 )(1 + )
1 w =
1
2
4

(u, v, w)

(, , )

u v w v w
u
v
w
v
w
u
v
w
v
w
}
{
}+
{
}
= {

2
(1 ) (1 )
=
8
and
(u, v, w)
(1 )2 (1 )
dw dv du =
d d d =
d d d
(, , )
8
Now Eq.(9.5) becomes
I = vol
f (, , )
=1
=1
=1
(1 )2 (1 )
d d d
64

We are to evaluate
vol
I =
8
3
1uw
u=1
w=1
v=1
u=
v=
(1 )(1 + )(1 )
(1 )(1 + )
1 w =
1
4
2
208
(9.6)


(u, v, w) (1 )2 (1 )

(, , ) =
8
and
(u, v, w)
(1 )2 (1 )
d d d =
d d d
(, , )
8
dv dw du =
Now Eq.(9.5) becomes
I = vol
f (, , )
=1
=1
=1
(1 )2 (1 )
d d d
64
(9.7)

We are required to evaluate
Z
vol
I =
8
3
1uv
f (u, v, w) dw du dv
v=1
u=1
w=1
Using the following transformation:

u=
(1 + )(1 )
1 v =
2
w=
(1 )(1 )(1 + )
1
4
Eq. 9.5 becomes

3
I = vol
f (, , )
=1
=1
=1
(1 )(1 )2
d d d
64

Considering the transformations
u=
(1 + )(1 )(1 )
1 v =
4
209
w=
(1 )(1 + )
1
2
(9.8)

the integral
Z
vol
I =
8
3
1vw
f (u, v, w) du dw dv
can be evaluated as
Z 1
Z
3
I = vol
=1
v=1
w=1
u=1
f (, , )
=1
=1
(1 )2 (1 )
d d d
64
(9.9)

similarly, Eq.(9.5) can be written as
Z
vol
I =
8
3
1vw
w=1
v=1
u=1
With transformation
u=
(1 + )(1 )(1 )
(1 + )(1 )
1 v =
1 w =
4
2
we have
Z
I = vol
f (, , )
=1
=1
=1
(1 )(1 )2
d d d
64

the integral I 3 of Eq.(9.5) can be written as
vol
I =
8
3
1uw
f (u, v, w) dv du dw
w=1
u=1
v=1
Considering the transformation

u=
(1 + )(1 )
(1 )(1 + )(1 )
1 v =
1 w =
2
4
210
(9.10)

we get
3
I = vol
f (, , )
=1
=1
=1
(1 )(1 )2
d d d
64
(9.11)
In order to evaluate the integral I 3 in Eq. 9.6 - 9.11 available Gaussian quadrature coefficients (points and weights) are computed in this study so that preferred
precision can be readily obtained.
9.5
Numerical Evaluation Procedures
As discussed before, we wish to present two sets of numerical quadrature schemes

in this chapter.
9.5.1
Gaussian Quadrature for Tetrahedron (GQTSS)
In this section, we are computing Gaussian quadrature formulae for a 2-cube in a

straightforward manner. The Gauss points are designed simply for i = 1, m, j =
1, n; and k = 1, p. Thus the m n p points Gaussian quadrature formula is as
follows
vol X X X
f (u(i , j , j ), v(i , j , j ), w(i , j , j ))
=
8 i=1 j=1 k=1

mnp
X
(u, v, w)

Wi Wj Wk
=
vol{
f (ur , vr , wr )Gr }
(, , )
r=1
(9.12)
Where (ur , vr , wr ) are the new Gaussian points, Gr is the corresponding weights.
All the Gaussian points and analogous weight can be calculated simply by using
211
step 1. r 1
step 2. i = 1, m
step 3. j = 1, n
step 4. k = 1, p
For Case 1:
(1 i )(1 + j )
1,
2
(1 i )2 (1 j )
Gr =
Wi Wj Wk
64
ur = i ,
wr =
vr =
(1 i )(1 j )(1 + k )
1,
4
For Case 2:
(1 i )(1 + j )(1 k )
1,
4
(1 i )2 (1 k )
Gr =
Wi Wj Wk
64
ur = i ,
vr =
wr =
(1 i )(1 + k )
1,
2
For Case 3:
(1 + i )(1 j )
1, vr = j ,
2
(1 i )(1 j )2
Gr =
Wi Wj Wk
64
ur =
wr =
(1 i )(1 j )(1 + k )
1,
4
For Case 4:
(1 + i )(1 j )(1 k )
1,
4
(1 j )2 (1 k )
Gr =
Wi Wj Wk
64
ur =
vr = j ,
212
wr =
(1 j )(1 + k )
1,
2

For Case 5:
(1 + i )(1 j )(1 k )
1,
4
(1 j )(1 k )2
Wi Wj Wk
Gr =
64
ur =
vr =
(1 + j )(1 k )
1,
2
wr = k ,
For Case 6:
(1 + i )(1 k )
(1 i )(1 + j )(1 k )
1, vr =
1,
2
4
(1 i )(1 k )2
Wi Wj Wk
Gr =
64
r=r+1
ur =
wr = k ,

For different values m, n, p of we have different Gauss quadrature formulae. For
clarity and reference, computed Gauss points and weights for (m = n = p = 5)
based on the algorithm for Case-1 listed in Table 9.1 and Fig. 9.3 shows the
distribution of Gaussian points for 15 15 15 = 3375 points. In Figure 9.3 it is
seen that there are huge crowding of Gauss points within the tetrahedron which
is one of the foremost causes of error germane in the calculation. For higher order
formulae the crowding become severe. To avoid this crowding further improvement
is shown in the next section. Influences of these cases on the results of the integrals
will be investigated in the result fragment.
9.5.2
Gaussian Quadrature for Tetrahedron (GQTSM)
In Eq.(9.6) to Eq.(9.11) it can be seen that for each i (i = 1, 2, ..., m); j varies
from 1 to n and k varies from 1 to p hence at the terminal value i = m there
are n k crowding points as shown in Table 9.1 and Fig. 9.3. To avoid such
crowding, keeping the generality, we can select the terminal values of the inner
dummy indices depending on the present and terminal value of its outer indices.
Therefore, the new Gaussian quadrature formulae now can be written as:
213
Figure 9.3: Distribution of Gauss Points for GQTSS, m =15 (3375 points).
n(i) p(j)
vol X X X
=
Wi Wj Wk f (u(i0 , j0 , j0 ), v(i0 , j0 , j0 ), w(i0 , j0 , j0 ))
8 i=1 j=1 k=1

tp
X
(u, v, w)

(9.13)
= vol{
G0r f (u0r , vr0 , wr0 )}

(, , )
r=1
Where (u0r , vr0 , wr0 ) are the new Gaussian points, G0r is the corresponding weights.
Now six cases arise for the evaluation of Eq.(9.13). All the Gaussian points and
corresponding weight for the six cases can be calculated simply by using the following algorithm:
214
step 1. r 1
step 2.
For Case 1.
i = 1, m; j = 1, (m i + 2 = n); k = 1, (n j + 2 = p)
(1 i0 )(1 + j0 )
1,
2
(1 i0 )2 (1 j0 )
G0r =
Wi Wj Wk
64
u0r = i0 ,
wr0 =
vr0 =
(1 i0 )(1 j0 )(1 + k0 )
1,
4
For Case 2.
i = 1, m; k = 1, (m i + 2 = n); j = 1, (n k + 2 = p)
(1 i0 )(1 + j0 )(1 k0 )
1,
4
(1 i0 )2 (1 k0 )
Wi Wj Wk
G0r =
64
u0r = i0 ,
vr0 =
wr0 =
(1 i0 )(1 + k0 )
1
2
For Case 3
j = 1, m; i = 1, (m j + 2 = n); k = 1, (n i + 2 = p)
(1 + i0 )(1 j0 )
1; vr0 = j0 ,
2
(1 i0 )(1 j0 )2
G0r =
Wi Wj Wk
64
u0r =
wr0 =
(1 i0 )(1 j0 )(1 + k0 )
1
4
For Case 4.
j = 1, m; k = 1, (m j + 2 = n); i = 1, (n k + 2 = p)
(1 + i0 )(1 j0 )(1 k0 )
1,
4
(1 j0 )2 (1 k0 )
0
Gr =
Wi Wj Wk
64
u0r =
vr0 = j0 ,
215
wr0 =
(1 j0 )(1 + k0 )
1
2

For Case 5.
k = 1, m; j = 1, (m k + 2 = n); i = 1, (n j + 2 = p)
(1 + i0 )(1 j0 )(1 k0 )
=
1,
4
(1 j0 )(1 k0 )2
0
Gr =
Wi Wj Wk
64
u0r
vr0
(1 + j0 )(1 k0 )
=
1,
2
wr0 = k0
For Case 6.
k = 1, m; i = 1, (m k + 2 = n); j = 1, (n i + 2 = p)
(1 i0 )(1 + j0 )(1 k0 )
(1 + i0 )(1 k0 )
, vr0 =
,
4
8
(1 i0 )(1 k0 )2
Wi Wj Wk
G0r =
64
r=r+1
u0r =
wr0 =
1 + k0
2

The total number of points in this process is given by
m
X
(r + 2)(r + 3)
m(m + 4)(m + 5)
TP =
{
1} =
2
6
r=1
Computed Gauss points and weights (for m = 5) based on above algorithm is

listed in Table 9.3 and Fig. 9.4 shows the distribution of Gaussian points for m
= 15 (950 points). Thus, the new Gaussian quadrature formula is now obtained
which is completely crowding free with significantly reduced number of Gauss
is greater than (m m m) for m
points. One can verify that T P = m(m+4)(m+5)
6
= 2 and 3. But T P is about 50 percent less than (m m m) for 3 < m < 10 and
it is too less for m 10. Therefore, it is established that the new scheme GQTSM
substantially reduces the computational effort as it requires the evaluation of the
integral at fewer number of Gauss points. As far as the accuracy is concern, still
the scheme GQTSM will give the same or better accuracy of the scheme GQTSS
because Gauss points and weights are calculated accordingly. For all different
quadrature scheme Gauss points are different but Gauss points of one quadrature
formula can be obtained from the others by simple manipulation. Influences of
216

these cases on the results of the integrals will be investigated in the result section.
Figure 9.4: Distribution of Gauss Points for GQTSM, m =15 (950 points).
9.6
Application Examples, Results and Discussions
To verify the accuracy and efficiency of the derived quadrature formulae and to
compare the results of the projected techniques, we reflect on the same typical
integrals with known correct values following Rathod and Rao [1996, 1998];
Rathod et al. [2004, 2005]; Reddy and Shippy [1981].
217

Consider the following nonpolynomial functions:
1
1x
1xy
p
(x + y + z) dz dy dx
I1 =
x=0
1
y=0
1x
z=0
1xy
I2 =
y=0
x=0
1x
z=0
dz dy dx
p
(x + y + z)
1xy
sin(x + 2y + 4z) dz dy dx
I3 =
x=0
Z 1
y=0
Z 1x
z=0
Z 1xy
I4 =
y=0
x=0
(1 + x + y + z)4 dz dy dx
z=0
The results are summarized in Table 9.5. It is seen from the Table that both
of the methods GQTSS and GQTSM can successfully evaluate the results of the
considered integrals but the method GQTSM calculate the integrals with significantly less number of points, the results of 350 points GQTSM is approximately
equal or better than 1000 points GQTSS.
Consider the monomial integrals
I5,,
ZZZ
=
x y z dz dy dx
Where V is tetrahedron in (x, y, z) space with spanning vertices points (5, 5,

0), (10, 10, 0), (8, 7, 8) and (10, 5, 0). The results are summarized in Table 9.6.
Both of the methods GQTSS and GQTSM can evaluate the monomial integrals
accurately but the method GQTSM calculate the integrals with significantly less
number of points, the same fact is found for higher order formulae also. This
Table also verify the efficiency of GQTSM to calculate the monomial integrals.
Consider the following rational integrals for different values of a, b, c, , , .
,,
Ia,b,c
1x
1xy
=
x=0
y=0
z=0
218
x y z
p
(ax + by + cz)
dz dy dx

These are very complex type of integrals. There are very few formulae available
which can evaluate these type of integrals accurately and the evaluation of these
type of integrals are very lengthy and tedious also. Obtained results using GQTSS
and GQTSM are summarized in Table 9.7. This table shows that GQTSM can
evaluate the integrals with less number of Gauss points.
Linear Three Dimensional Elasticity Problem: Example 8.6 in page
187 is also tested for the new technique. Hence we need to evaluate
ZZZ
Ni Nj
dx0 dy 0 dz 0
t s
t, s = x0 , y 0 , z 0
V0
We shall evaluate these triple integrals, where V 0 is the tetrahedral element with
three plane surface and one curved surface. It can be shown that the evaluation
of integrals now amounts to the evaluation of integrals of the type
IJpqr
xp y q z r
dx0 dy 0 dz 0
J(x, y, z)
V
Z 1 Z x Z 1xy p q r
xy z
=
dx dy dz
J(x, y, z)
x=1 y=1 z=1
ZZZ
(9.14)
Where J is the Jacobean of transformation and V is the standard tetrahedron:

{(x, y, z)| 1 x, y, z 1, x + y + z 1}. The transformation which maps a
curved tetrahedron V 0 in to an orthogonal tetrahedron V is given by
t=
20
X
Ni (x, y, z)ti ,
t = x0 , y 0 , z 0 and
i=1
219
ti = (x0i , yi0 , zi0 )

That is
1
1
1
t(x, y, z) = t4 + (t1 t4 )(x + 1) + (t2 t4 )(y + 1) + (t3 t4 )(z + 1)
2
2
2
9
+
(t1 t2 + t11 + t12 )(x + 1)(y + 1)
16
9
+
(t2 t3 + t13 + t14 )(y + 1)(z + 1)
16
9
+
(t1 t3 + t15 + t16 )(x + 1)(z + 1)
16
9
+
{2(t1 + t2 + t3 ) 3(t11 + t12 + t13 + t14 + t15 + t16 ) + 12t20 }
32
(x + 1)(y + 1)(z + 1)
For the 20 nodal data of Fig 8.7 in page 190 we have
x0 = u + Ru(v + w) + Sxyz,
y 0 = v + Rv(u + w) + Suvw,
z 0 = w + Rw(u + v) + Suvw
x+1
y+1
z+1
where u =
, v=
, w=
2
2
2
1
J(x, y, z) =
[1 + 2R(u + v + w)
8
+ R2 (u2 + v 2 + w2 ) + (2R2 + s)(uv + vw + wu)
+ RS(u2 v + uv 2 + vw2 + wv 2 + u2 w + uw2 )
+ 4R3 uvw + R2 S(u2 vw + v 2 uw + w2 uv)]
Where R = 0.842329219213245 and S = 1.534481952840430. We have computed the values of the integrals of the Eq.9.14 for different values of p, q, r. Results
are summarized in Table 9.8. For all case studies in Table 9.8 for IJpqr in Eq.(9.14)
the method GQTSS require more Gaussian points than the method GQTSM. Further, it is observed that the absence of any independent variable in the explicit
form of the integrand enhances faster convergence of one among the six quadrature
formulae for both GQTSS and GQTSM.
Comparison between the results of Chapter 8, 9 and existing methods of
Rathod et al. [2004, 2005] are also done. It is found that all the methods
220

GQTTS, GQTTM, GQTSS and GQTSM can evaluate these integrals successfully but GQTTM and GQSTM needs very less number of quadrature
points. Thus, we believe that any one of the newly derived appropriate
quadrature formulae for tetrahedral domain integrals will ensure the accurate evaluation of the integrals in an efficient manner and enhance the further
utilization of tetrahedral elements for numerical solution of field problems
in science and engineering.
9.7
Error Analysis
The integral of polynomial of order up to 2n-1 can be evaluated exactly by the

n-point Gauss quadrature formula. The total error in n-point Gauss quadrature
formula to evaluate the integral of polynomial of 2n or high order is given by
=
22n+1 [n!]4
f 2n (x1 )
(2n + 1)[(2n)!]3
Where f 2n (x1 ) is the 2n-th derivative of the function at a point x1 in the

interval [-1, 1] (Chapra and Canale [2006]).
In this thesis the tetrahedral domain integral is evaluated by converting it to
a cubic domain integral. So, we consider the integral
Z
I=
f (x, y, z)dzdydx
z=1
y=1
x=1
Integrating with respect to x using Gauss quadrature formula, keeping y, z

fixed we have
I=
z=1
y=1
m
X
{
f (xi , y, z)W (m, i) +
i=1
22m+1 [m!]4
f 2m (x1 , y, z)}dydz
(2m + 1)[(2m)!]3 x
Where fx2m (x1 , y, z) is the 2m-th partial derivative of the function with respect
to x, m points are taken along x direction, W(m,i) is the corresponding weight.
Now integrating with respect to y and z successively, we have
221
I =
m X
n
X
f (xi , yj , z)W (m, i)W (n, j) +
z=1
i=1 j=1
2n+1
4
22m+1 [m!]4
f 2m (x1 , y1 , z)
(2m + 1)[(2m)!]3 x
2
[n!]
f 2n (x2 , y2 , z)}dz
(2n + 1)[(2n)!]3 y
p
m X
n X
X
=
f (xi , yj , zk )W (m, i)W (n, j)W (p, k) +
i=1 j=1 k=1
The total error in the integration of polynomial function of order 2(m+n+p)

or high may be found given by the new error formula
2(2n+3) [n!]4
22m+3 [m!]4
2m
f
(x
,
y
,
z
)
+
fy2n (x2 , y2 , z2 )
1
1
1
x
3
3
(2m + 1)[(2m)!]
(2n + 1)[(2n)!]
(2p+3)
4
2
[p!]
f 2p (x3 , y3 , z3 )
+
(9.15)
(2p + 1)[(2p)!]3 z
Where fx2m (x1 , y1 , z1 ), fy2n (x2 , y2 , z2 ) and fz2p (x3 , y3 , z3 ) are the 2m-th, 2n-th
and 2p-th partial derivative of the function with respect to x, y and z respectively,
(x1 , y1 , z1 ), (x2 , y2 , z2 ) and (x3 , y3 , z3 ) are points somewhere in the cubic domain
{(x, y, z)| 1 x, y, z 1}.
Consider the cubic domain monomial integrals
Z
JJ =
z=1
y=1
N
X
(
xi y ji z N j )dzdydx
x=1 i=0
The computed errors for the integral JJ using mmm Gaussian quadrature
formula and the error formula given by Eq.(9.15) are given in Table 9.9. The table
9.9 shows that the error formula gives exactly same results for the error calculated
using the m m m Gaussian quadrature method. Thus, this error formula can
be used to evaluate the error in the integral over cubic domain.
We know, the n-point Gaussian Quadrature rule gives exact results for polynomials of degree at most 2n-1. Thus with n = 2, we have a rule with 8 nodes
which is exact for any polynomial of degrees at most 3 in x, y and z separately, so
222

the total degree of this monomial is at most 9. But this rule is not exact for all
monomials of degree at most 9, which includes x9 , x8 y, x5 yz 3 , x4 y 5 , x2 z 5 , xzy 7 , y 9
and so on.
If m + n + p = 2M-1 (odd), i varies from 0 to 2M+1, j varies from 0 to 2M and
k varies from 0 to 2M-1. That is we have to choose (M +1)(M +1)M th order
Gauss quadrature to evaluate the integral of xm y n z p . Now if m + n + p = 2M-2
(even) i varies from 0 to 2M, j varies from 0 to 2M-1 and k varies from 0 to 2M-2.
That is we have to choose (M +1)M M th order Gauss quadrature to evaluate
the integral of xm y n z p . Which conclude that a precision d = 2M-1 or 2M-2 can
be obtained by Gauss quadrature rule of order (M + 1) (M + 1) (M + 1).
Integral I5 and Table 9.6 can be considered for the verification of the formulae.
Now, consider the tetrahedral domain integral of the monomial functions
Z
1xy
xm y n z p dz dy dx
It =
x=1
y=1
z=1
The absolute error of the integral of all monomials xi y j z k where i , j and k are
non-negative integers are calculated with the methods GQTSS and GQTSM. In
Table 9.10 we present the absolute error over corresponding polynomial integral
of order up to m + n + p = 20. It is observed that the method GQTSM is
always accurate for the integration of monomial /polynomial functions and hence
a proper balance is observed.
Figure 9.5, 9.6, 9.7 and 9.8 shows the percentage error in the calculated results
of I1 , I2 , I3 , I4 . The figures proves the efficiency of the method GQTSM with
very less number of Gauss points to obtain an accurate results. It is seen from
these figures that, the error decreases with increase in number of total Gaussian
integration points. These figures also show that, if SGQTS gives accurate results
for a value N, SGQTM gives the results of same accuracy for N or N+1.
Figure 9.9, 9.10, 9.11 and 9.12 shows the percentage error in the calculated
results of the monomial integrals I5,, . Same observation is found in these figures.
Figure 9.13, 9.14, 9.15 and 9.16 shows the error precisions calculated for the
,,
. These figures show the sucresults of the integrals of rational functions Ia,b,c
cess of the newly derived method GQTSM to calculate the integral of rational
functions.
223
9.8
Conclusions
This chapter presented crowding free Gaussian Quadrature formulae to evaluate

tetrahedral domain integrals in Euclidean three dimensional space. It is also
demonstrated that there are six possibilities to derive quadrature formulae for
which Gaussian points and corresponding weights are different. Also it condensed
the total number of points in the calculation process; it then demonstrates the
efficiency and accuracy evaluating the integrals of polynomial, nonpolynomial,
rational functions. In all the test cases the efficiency and accuracy of the derived
formulae are found so satisfactory. The quadrature schemes so derived in chapter
8 and in the present are different but accurate in view of accuracy and efficiency
of evaluations. Thus, it completes the successful derivation of quadrature schemes
for tetrahedral domain integrals.
224
Table 9.1: Computed Gauss points (u, v, w ) and corresponding weights G for
n n n point method (GQTSS).
u
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
v
-0.910580956592710D+00
-0.910580956592710D+00
-0.910580956592710D+00
-0.910580956592710D+00
-0.910580956592710D+00
-0.560119750320643D+00
-0.560119750320643D+00
-0.560119750320643D+00
-0.560119750320643D+00
-0.560119750320643D+00
-0.469100770306681D-01
-0.469100770306681D-01
-0.469100770306681D-01
-0.469100770306681D-01
-0.469100770306681D-01
0.466299596259307D+00
0.466299596259307D+00
0.466299596259307D+00
0.466299596259307D+00
0.466299596259307D+00
0.816760802531374D+00
0.816760802531374D+00
0.816760802531374D+00
0.816760802531374D+00
0.816760802531374D+00
-0.927830286153624D+00
-0.927830286153624D+00
-0.927830286153624D+00
-0.927830286153624D+00
-0.927830286153624D+00
-0.644974598962845D+00
-0.644974598962845D+00
-0.644974598962845D+00
-0.644974598962845D+00
-0.644974598962845D+00
-0.230765344947159D+00
-0.230765344947159D+00
-0.230765344947159D+00
-0.230765344947159D+00
-0.230765344947159D+00
0.183443909068528D+00
0.183443909068528D+00
0.183443909068528D+00
0.183443909068528D+00
0.183443909068528D+00
0.466299596259307D+00
0.466299596259307D+00
0.466299596259307D+00
0.466299596259307D+00
0.466299596259307D+00
-0.953089922969332D+00
-0.953089922969332D+00
-0.953089922969332D+00
-0.953089922969332D+00
-0.953089922969332D+00
-0.769234655052841D+00
-0.769234655052841D+00
-0.769234655052841D+00
-0.769234655052841D+00
-0.769234655052841D+00
-0.500000000000000D+00
-0.500000000000000D+00
-0.500000000000000D+00
-0.500000000000000D+00
-0.500000000000000D+00
w
-0.914775610806955D+00
-0.580754566717371D+00
-0.916195987343129D-01
0.397515369248745D+00
0.731536413338329D+00
-0.931215772989438D+00
-0.661628867873342D+00
-0.266850201870347D+00
0.127928464132649D+00
0.397515369248745D+00
-0.955290478296355D+00
-0.780059875160321D+00
-0.523455038515334D+00
-0.266850201870347D+00
-0.916195987343129D-01
-0.979365183603272D+00
-0.898490882447301D+00
-0.780059875160321D+00
-0.661628867873342D+00
-0.580754566717371D+00
-0.995805345785755D+00
-0.979365183603272D+00
-0.955290478296355D+00
-0.931215772989438D+00
-0.914775610806955D+00
-0.931215772989438D+00
-0.661628867873342D+00
-0.266850201870347D+00
0.127928464132649D+00
0.397515369248745D+00
-0.944484555064120D+00
-0.726902158098188D+00
-0.408278045465736D+00
-0.896539328332845D-01
0.127928464132648D+00
-0.963915143076812D+00
-0.822487299481423D+00
-0.615382672473579D+00
-0.408278045465736D+00
-0.266850201870347D+00
-0.983345731089503D+00
-0.918072440864658D+00
-0.822487299481423D+00
-0.726902158098188D+00
-0.661628867873342D+00
-0.996614513164185D+00
-0.983345731089503D+00
-0.963915143076812D+00
-0.944484555064120D+00
-0.931215772989438D+00
-0.955290478296355D+00
-0.780059875160321D+00
-0.523455038515334D+00
-0.266850201870347D+00
-0.916195987343129D-01
-0.963915143076812D+00
-0.822487299481423D+00
-0.615382672473579D+00
-0.408278045465736D+00
-0.266850201870347D+00
-0.976544961484666D+00
-0.884617327526421D+00
-0.750000000000000D+00
-0.615382672473579D+00
-0.523455038515334D+00
225
G
0.143931111875796D-02
0.290762936018300D-02
0.345595267890959D-02
0.290762936018300D-02
0.143931111875796D-02
0.234673477706453D-02
0.474076441808048D-02
0.563478196881094D-02
0.474076441808048D-02
0.234673477706453D-02
0.181302550558012D-02
0.366258977790325D-02
0.435328420053336D-02
0.366258977790325D-02
0.181302550558012D-02
0.704005022097696D-03
0.142219819279516D-02
0.169039758699556D-02
0.142219819279516D-02
0.704005022097696D-03
0.708413695548069D-04
0.143110438979225D-03
0.170098332250659D-03
0.143110438979225D-03
0.708413695548069D-04
0.189403924354977D-02
0.382624996229895D-02
0.454780756738284D-02
0.382624996229895D-02
0.189403924354977D-02
0.308814939594078D-02
0.623853573785862D-02
0.741500432997716D-02
0.623853573785862D-02
0.308814939594078D-02
0.238582292068215D-02
0.481972846729571D-02
0.572863712833023D-02
0.481972846729571D-02
0.238582292068215D-02
0.926424538886224D-03
0.187151975285516D-02
0.222445260461427D-02
0.187151975285516D-02
0.926424538886224D-03
0.932226064642663D-04
0.188323972528047D-03
0.223838273981466D-03
0.188323972528047D-03
0.932226064642663D-04
0.951130350812901D-03
0.192142928470617D-02
0.228377412016423D-02
0.192142928470617D-02
0.951130350812901D-03
0.155077706458653D-02
0.313280767814902D-02
0.372359532341708D-02
0.313280767814902D-02
0.155077706458653D-02
0.119808953233352D-02
0.242032473378445D-02
0.287675171467764D-02
0.242032473378445D-02
0.119808953233352D-02

u
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
v
-0.230765344947159D+00
-0.230765344947159D+00
-0.230765344947159D+00
-0.230765344947159D+00
-0.230765344947159D+00
-0.469100770306681D-01
-0.469100770306681D-01
-0.469100770306681D-01
-0.469100770306681D-01
-0.469100770306681D-01
-0.978349559785040D+00
-0.978349559785040D+00
-0.978349559785040D+00
-0.978349559785040D+00
-0.978349559785040D+00
-0.893494711142838D+00
-0.893494711142838D+00
-0.893494711142838D+00
-0.893494711142838D+00
-0.893494711142838D+00
-0.769234655052841D+00
-0.769234655052841D+00
-0.769234655052841D+00
-0.769234655052841D+00
-0.769234655052841D+00
-0.644974598962845D+00
-0.644974598962845D+00
-0.644974598962845D+00
-0.644974598962845D+00
-0.644974598962845D+00
-0.560119750320643D+00
-0.560119750320643D+00
-0.560119750320643D+00
-0.560119750320643D+00
-0.560119750320643D+00
-0.995598889345954D+00
-0.995598889345954D+00
-0.995598889345954D+00
-0.995598889345954D+00
-0.995598889345954D+00
-0.978349559785040D+00
-0.978349559785040D+00
-0.978349559785040D+00
-0.978349559785040D+00
-0.978349559785040D+00
-0.953089922969332D+00
-0.953089922969332D+00
-0.953089922969332D+00
-0.953089922969332D+00
-0.953089922969332D+00
-0.927830286153624D+00
-0.927830286153624D+00
-0.927830286153624D+00
-0.927830286153624D+00
-0.927830286153624D+00
-0.910580956592710D+00
-0.910580956592710D+00
-0.910580956592710D+00
-0.910580956592710D+00
-0.910580956592710D+00
w
-0.989174779892520D+00
-0.946747355571419D+00
-0.884617327526421D+00
-0.822487299481423D+00
-0.780059875160321D+00
-0.997799444672977D+00
-0.989174779892520D+00
-0.976544961484666D+00
-0.963915143076812D+00
-0.955290478296355D+00
-0.979365183603272D+00
-0.898490882447301D+00
-0.780059875160321D+00
-0.661628867873342D+00
-0.580754566717371D+00
-0.983345731089503D+00
-0.918072440864658D+00
-0.822487299481423D+00
-0.726902158098188D+00
-0.661628867873342D+00
-0.989174779892520D+00
-0.946747355571419D+00
-0.884617327526421D+00
-0.822487299481423D+00
-0.780059875160321D+00
-0.995003828695537D+00
-0.975422270278180D+00
-0.946747355571419D+00
-0.918072440864658D+00
-0.898490882447301D+00
-0.998984376181768D+00
-0.995003828695537D+00
-0.989174779892520D+00
-0.983345731089503D+00
-0.979365183603272D+00
-0.995805345785755D+00
-0.979365183603272D+00
-0.955290478296355D+00
-0.931215772989438D+00
-0.914775610806955D+00
-0.996614513164185D+00
-0.983345731089503D+00
-0.963915143076812D+00
-0.944484555064120D+00
-0.931215772989438D+00
-0.997799444672977D+00
-0.989174779892520D+00
-0.976544961484666D+00
-0.963915143076812D+00
-0.955290478296355D+00
-0.998984376181768D+00
-0.995003828695537D+00
-0.989174779892520D+00
-0.983345731089503D+00
-0.979365183603272D+00
-0.999793543560198D+00
-0.998984376181768D+00
-0.997799444672977D+00
-0.996614513164185D+00
-0.995805345785755D+00
226
G
0.465222935413471D-03
0.939821730277485D-03
0.111705414415182D-02
0.939821730277485D-03
0.465222935413471D-03
0.468136289636074D-04
0.945707152938267D-04
0.112404944502805D-03
0.945707152938267D-04
0.468136289636074D-04
0.170456069100769D-03
0.344347420567744D-03
0.409284761972482D-03
0.344347420567744D-03
0.170456069100769D-03
0.277921277830243D-03
0.561443635574862D-03
0.667321173390394D-03
0.561443635574862D-03
0.277921277830243D-03
0.214714662336039D-03
0.433756571552638D-03
0.515554769799693D-03
0.433756571552638D-03
0.214714662336039D-03
0.833745582383381D-04
0.168429403668663D-03
0.200191969714883D-03
0.168429403668663D-03
0.833745582383381D-04
0.838966726974746D-05
0.169484155008367D-04
0.201445626995949D-04
0.169484155008367D-04
0.838966726974746D-05
0.348673721407182D-05
0.704374430430755D-05
0.837205984069912D-05
0.704374430430755D-05
0.348673721407182D-05
0.568497482726903D-05
0.114845216605656D-04
0.136502829220052D-04
0.114845216605656D-04
0.568497482726903D-05
0.439206188153587D-05
0.887263907855757D-05
0.105458492105091D-04
0.887263907855757D-05
0.439206188153587D-05
0.170545511491618D-05
0.344528107924556D-05
0.409499523510995D-05
0.344528107924556D-05
0.170545511491618D-05
0.171613514481643D-06
0.346685637877664D-06
0.412063922372617D-06
0.346685637877664D-06
0.171613514481643D-06
Table 9.3: Computed Gauss points (p, q, r) and corresponding weights L for
method GQTSM.
u
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.906179845938664D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
-0.538469310105683D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
v
-0.935637374493801D+00
-0.935637374493801D+00
-0.935637374493801D+00
-0.935637374493801D+00
-0.935637374493801D+00
-0.935637374493801D+00
-0.935637374493801D+00
-0.677102080244399D+00
-0.677102080244399D+00
-0.677102080244399D+00
-0.677102080244399D+00
-0.677102080244399D+00
-0.677102080244399D+00
-0.274335618713707D+00
-0.274335618713707D+00
-0.274335618713707D+00
-0.274335618713707D+00
-0.274335618713707D+00
0.180515464652371D+00
0.180515464652371D+00
0.180515464652371D+00
0.180515464652371D+00
0.583281926183063D+00
0.583281926183063D+00
0.583281926183063D+00
0.841817220432465D+00
0.841817220432465D+00
-0.927830286153624D+00
-0.927830286153624D+00
-0.927830286153624D+00
-0.927830286153624D+00
-0.927830286153624D+00
-0.927830286153624D+00
-0.644974598962845D+00
-0.644974598962845D+00
-0.644974598962845D+00
-0.644974598962845D+00
-0.644974598962845D+00
-0.230765344947159D+00
-0.230765344947159D+00
-0.230765344947159D+00
-0.230765344947159D+00
0.183443909068528D+00
0.183443909068528D+00
0.183443909068528D+00
0.466299596259307D+00
0.466299596259307D+00
-0.930568155797026D+00
-0.930568155797026D+00
-0.930568155797026D+00
-0.930568155797026D+00
-0.930568155797026D+00
-0.669990521792428D+00
-0.669990521792428D+00
-0.669990521792428D+00
-0.669990521792428D+00
-0.330009478207572D+00
-0.330009478207572D+00
-0.330009478207572D+00
227
-0.694318442029738D-01
w
-0.953133038284567D+00
-0.761973843346101D+00
-0.452837684101723D+00
-0.790913897837677D-01
0.294654904534188D+00
0.603791063778566D+00
0.794950258717031D+00
-0.946540101185744D+00
-0.731799472415783D+00
-0.397259759191488D+00
-0.194583146254489D-01
0.315081398598846D+00
0.529822027368808D+00
-0.944621928617262D+00
-0.727577941584041D+00
-0.409742267673815D+00
-0.919065937635887D-01
0.125137393269633D+00
-0.949615783734883D+00
-0.760523876177890D+00
-0.513811742535817D+00
-0.324719834978824D+00
-0.963608866696046D+00
-0.838551040122200D+00
-0.713493213548354D+00
-0.986398576827778D+00
-0.949238797666023D+00
-0.950490037970444D+00
-0.751615730079521D+00
-0.441793809977105D+00
-0.919065937635887D-01
0.217915326338827D+00
0.416789634229750D+00
-0.944484555064120D+00
-0.726902158098188D+00
-0.408278045465736D+00
-0.896539328332845D-01
0.127928464132648D+00
-0.946590619274843D+00
-0.746145272866830D+00
-0.484620072080328D+00
-0.284174725672316D+00
-0.959988046051174D+00
-0.822487299481423D+00
-0.684986552911672D+00
-0.984748744935084D+00
-0.943081541218540D+00
-0.956346976129275D+00
-0.785257118530658D+00
-0.534715922101487D+00
-0.284174725672316D+00
-0.113084868073699D+00
-0.953481322473439D+00
-0.778896777499262D+00
-0.551112700708310D+00
-0.376528155734133D+00
-0.962807382215067D+00
-0.834995260896214D+00
-0.707183139577362D+00
-0.985327324868973D+00
G
0.576651691559072D-03
0.124564721228047D-02
0.170045180553786D-02
0.186135022002191D-02
0.170045180553786D-02
0.124564721228047D-02
0.576651691559072D-03
0.138110348004552D-02
0.290821853363949D-02
0.377200921144364D-02
0.377200921144364D-02
0.290821853363949D-02
0.138110348004552D-02
0.184705677022988D-02
0.373133815549181D-02
0.443499720802718D-02
0.373133815549181D-02
0.184705677022988D-02
0.166697114369810D-02
0.312517468091849D-02
0.312517468091849D-02
0.166697114369810D-02
0.913358455696345D-03
0.146137352911416D-02
0.913358455696345D-03
0.155625745624003D-03
0.155625745624003D-03
0.136960105591409D-02
0.288399763815690D-02
0.374059429546915D-02
0.374059429546915D-02
0.288399763815690D-02
0.136960105591409D-02
0.308814939594078D-02
0.623853573785862D-02
0.741500432997716D-02
0.623853573785862D-02
0.308814939594078D-02
0.350285305275700D-02
0.656701689938599D-02
0.656701689938599D-02
0.350285305275700D-02
0.217231699669361D-02
0.347570719470980D-02
0.217231699669361D-02
0.393465715983046D-03
0.393465715983046D-03
0.136344637490265D-02
0.275437093414734D-02
0.327379264321987D-02
0.275437093414734D-02
0.136344637490265D-02
0.270201938829868D-02
0.506564412442597D-02
0.506564412442597D-02
0.270201938829868D-02
0.212557118293120D-02
0.340091389268994D-02
0.212557118293120D-02
0.429372505117035D-03

u
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.000000000000000D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.538469310105683D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
0.906179845938664D+00
v
-0.930568155797026D+00
-0.930568155797026D+00
-0.930568155797026D+00
-0.930568155797026D+00
-0.930568155797026D+00
-0.669990521792428D+00
-0.669990521792428D+00
-0.669990521792428D+00
-0.669990521792428D+00
-0.330009478207572D+00
-0.330009478207572D+00
-0.330009478207572D+00
-0.694318442029738D-01
-0.694318442029738D-01
-0.947984722625272D+00
-0.947984722625272D+00
-0.947984722625272D+00
-0.947984722625272D+00
-0.769234655052841D+00
-0.769234655052841D+00
-0.769234655052841D+00
-0.590484587480411D+00
-0.590484587480411D+00
-0.980173468570694D+00
-0.980173468570694D+00
-0.980173468570694D+00
-0.926006377367970D+00
-0.926006377367970D+00
228
w
-0.956346976129275D+00
-0.785257118530658D+00
-0.534715922101487D+00
-0.284174725672316D+00
-0.113084868073699D+00
-0.953481322473439D+00
-0.778896777499262D+00
-0.551112700708310D+00
-0.376528155734133D+00
-0.962807382215067D+00
-0.834995260896214D+00
-0.707183139577362D+00
-0.985327324868973D+00
-0.945240830928054D+00
-0.971566589679223D+00
-0.864856032396452D+00
-0.725628555083959D+00
-0.618917997801187D+00
-0.973992361312636D+00
-0.884617327526421D+00
-0.795242293740205D+00
-0.989007878509772D+00
-0.958976844115500D+00
-0.991660795501926D+00
-0.963003188683985D+00
-0.934345581866044D+00
-0.995810160914250D+00
-0.984363307656444D+00
G
0.136344637490265D-02
0.275437093414734D-02
0.327379264321987D-02
0.275437093414734D-02
0.136344637490265D-02
0.270201938829868D-02
0.506564412442597D-02
0.506564412442597D-02
0.270201938829868D-02
0.212557118293120D-02
0.340091389268994D-02
0.212557118293120D-02
0.429372505117035D-03
0.429372505117035D-03
0.546317509560028D-03
0.102421547911849D-02
0.102421547911849D-02
0.546317509560028D-03
0.786674148253923D-03
0.125867863720628D-02
0.786674148253923D-03
0.199483844892810D-03
0.199483844892810D-03
0.285550084723027D-04
0.456880135556845D-04
0.285550084723027D-04
0.137723246280313D-04
0.137723246280313D-04
229
Exact
Value
GQTSM
GQTSS
Method
8
27
64
125
216
343
512
729
1000
14
28
48
75
110
154
208
273
350
Points
0.14285714285714
I1
0.143127410953799
0.142875312759849
0.142860037924269
0.142857834882225
0.142857355360141
0.142857220464426
0.142857175072351
0.142857157618562
0.142857150174568
0.143058065243440
0.142868320929028
0.142858821339166
0.142857541720436
0.142857267261819
0.142857189383667
0.142857162683365
0.142857152181787
0.142857147595743
0.2
I2
0.197660776240555
0.199583323221216
0.199881018522955
0.199956079429135
0.199980842555110
0.199990593978398
0.199994953918813
0.199997099779897
0.199998238575602
0.198325920889004
0.199730676421398
0.199924597609756
0.199971858116752
0.199987455481037
0.199993687508278
0.199996530092608
0.199997959524722
0.199998734389269
I4
0.020103982733156
0.020798626362385
0.020831943224904
0.020833282126123
0.020833331531397
0.020833333271718
0.020833333331268
0.020833333333265
0.020833333333331
0.020334811873381
0.020811918661132
0.020832494969141
0.020833301701749
0.020833332126155
0.020833333282340
0.020833333330381
0.020833333333016
0.020833333333269
0.131902326890181 0.020833333333333
I3
0.131949528497795
0.131902664864685
0.131902289803505
0.131902327570123
0.131902326884088
0.131902326890216
0.131902326890182
0.131902326890182
0.131902326890182
0.133142199123991
0.131877204546620
0.131902589866991
0.131902325527761
0.131902326955921
0.131902326902415
0.131902326892685
0.131902326890580
0.131902326890160
Value of I
Table 9.5: Calculated values of the integrals I1 ,I2 ,I3 , I4
230
Exact
Value
GQTSM
GQTSS
Method
8
27
64
125
216
343
512
729
1000
14
28
48
75
110
154
208
273
350
440
544
663
798
950
Points
15721.6666666667
15613.2458847736452
15721.6666666665296
15721.6666666666770
15721.6666666666642
15721.6666666666770
15721.6666666666661
15721.6666666666879
15721.6666666666661
15721.6666666667024
15612.4547963692658
15721.6666928552586
15721.6666681218594
15721.6666668061935
15721.6666666861638
15721.6666666702495
15721.6666666674864
15721.6666666668807
15721.6666666667461
15721.6666666666024
15721.6666666665897
15721.6666666665933
15721.6666666665951
15721.6666666665969
I52,1,0
109662.063492063
17357405.82972
I54,0,4
19959073.9239162914455
17447147.0802131444216
17356864.7296515144408
17357404.7323902957141
17357405.8297258540988
17357405.8297257982194
17357405.8297258764505
17357405.8297258913517
17357405.8297258727252
15916008.4708074014634
17395025.3200442716479
17356769.8780943527818
17357410.5770655572414
17357406.3276728168130
17357405.8860965371132
17357405.8378492966294
17357405.8311539664865
17357405.8300211951137
17357405.8297953382134
17357405.8297439366579
17357405.8297308571637
17357405.8297272846103
17357405.8297261968255
Value of I
107754.6467764059344
109657.4916666657664
109662.0634920635057
109662.0634920633602
109662.0634920636367
109662.0634920634475
109662.0634920636076
109662.0634920634766
109662.0634920636221
107739.9768428653770
109657.4923773875344
109662.0635323561437
109662.0634959668678
109662.0634926121420
109662.0634921646124
109662.0634920864541
109662.0634920696466
109662.0634920655139
109662.0634920636658
109662.0634920631564
109662.0634920630109
109662.0634920630400
109662.0634920631128
I52,2,0
Table 9.6: Calculated values of the integrals I52,1,0 , I52,2,0 , I54,4,0 and I51,2,3
330565.079365081
I51,2,3
380177.2290809325059
331027.3599999966100
330568.5208940719021
330565.0793650789419
330565.0793650799314
330565.0793650795240
330565.0793650801061
330565.0793650801061
330565.0793650796986
329118.0069088268210
330824.6705146196764
330568.9540617458988
330565.1115077321883
330565.0829795532045
330565.0799310719012
330565.0794796128175
330565.0793933367240
330565.0793732203892
330565.0793677304173
330565.0793660327327
330565.0793654504814
330565.0793652340653
330565.0793651472777
GQTSS
GQTSM
Case
1-6
231
Case 6
Case 5
Case 4
Case 3
Case 2
Case 1
Case
Method
Points
8
27
64
125
216
343
512
729
1000
110
208
350
544
110
208
350
544
110
208
350
544
110
208
350
544
110
208
350
544
110
208
350
544
222
I111
0.0000269860345093
0.0000202862454377
0.0000231837802848
0.0000233422437574
0.0000233426594248
0.0000233426697419
0.0000233426703367
0.0000233426703927
0.0000233426704000
0.0000233426460487
0.0000233426700306
0.0000233426703855
0.0000233426704001
0.0000233426460487
0.0000233426700306
0.0000233426703855
0.0000233426704001
0.0000233426460487
0.0000233426700306
0.0000233426703855
0.0000233426704001
0.0000233426460487
0.0000233426700306
0.0000233426703855
0.0000233426704001
0.0000233426460487
0.0000233426700306
0.0000233426703855
0.0000233426704001
0.0000233426460487
0.0000233426700306
0.0000233426703855
0.0000233426704001
022
I111
0.0006015651986381
0.0008488381955233
0.0008546314116172
0.0008546969203429
0.0008547004410159
0.0008547007916240
0.0008547008422886
0.0008547008517505
0.0008547008538885
0.0008547006761305
0.0008547008477305
0.0008547008541789
0.0008547008546419
0.0008547006761305
0.0008547008477305
0.0008547008541789
0.0008547008546419
0.0008547003467509
0.0008547008418419
0.0008547008538989
0.0008547008546175
0.0008547001755466
0.0008547008358275
0.0008547008535749
0.0008547008545902
0.0008547001755466
0.0008547008358275
0.0008547008535749
0.0008547008545902
0.0008547003467509
0.0008547008418419
0.0008547008538989
0.0008547008546175
222
I101
0.0000333182575959
0.0000258267656879
0.0000288727421023
0.0000292327112870
0.0000292495089719
0.0000292513882514
0.0000292517007873
0.0000292517681974
0.0000292517857518
0.0000292495203673
0.0000292516825132
0.0000292517802690
0.0000292517912085
0.0000292484738197
0.0000292516351558
0.0000292517791266
0.0000292517918411
0.0000292511271472
0.0000292517169499
0.0000292517800403
0.0000292517906566
0.0000292511271472
0.0000292517169499
0.0000292517800403
0.0000292517906566
0.0000292495203673
0.0000292516825132
0.0000292517802690
0.0000292517912085
0.0000292484738197
0.0000292516351558
0.0000292517791266
0.0000292517918411
Table 9.7: Calculated results of Iabc

220
I110
0.0009230975391085
0.0008883907833675
0.0009316174070870
0.0009323393056688
0.0009323923274447
0.0009323992493688
0.0009324005189310
0.0009324008123282
0.0009324008927173
0.0009323918072337
0.0009324004766320
0.0009324008890532
0.0009324009261585
0.0009323939970271
0.0009324005524678
0.0009324008837150
0.0009324009227078
0.0009323918072337
0.0009324004766320
0.0009324008890532
0.0009324009261585
0.0009323939970271
0.0009324005524678
0.0009324008837150
0.0009324009227078
0.0009323981782032
0.0009324006198537
0.0009324008787228
0.0009324009203554
0.0009323981782032
0.0009324006198537
0.0009324008787228
0.0009324009203554
GQTSM
GQTSS
Method
Points
27
64
125
216
343
512
729
1000
14
28
48
75
110
154
208
273
350
440
544
663
798
950
3
4
5
6
7
8
9
10
2
3
4
5
6
7
8
9
10
11
12
13
14
15
IJ0,0,0
0.057883923504
0.057917371950
0.057918451577
0.057918486772
0.057918487951
0.057918487991
0.057918487993
0.057918487993
0.057106321695
0.057890655765
0.057917552866
0.057918455127
0.057918486659
0.057918487915
0.057918487985
0.057918487992
0.057918487993
0.057918487993
0.057918487993
0.057918487993
0.057918487993
0.057918487993
Value of I
IJ2,1,2
IJ1,2,1
0.002450805315 0.003530383143
0.002403948085 0.003416615059
0.002414474857 0.003423823551
0.002413862879 0.003424492584
0.002413804476 0.003424516105
0.002413802396 0.003424516819
0.002413802334 0.003424516841
0.002413802332 0.003424516841
0.001951020091 0.002545145635
0.002450792220 0.003530376620
0.002403953091 0.003416615436
0.002414475280 0.003423823630
0.002413862920 0.003424492594
0.002413804481 0.003424516106
0.002413802396 0.003424516819
0.002413802334 0.003424516841
0.002413802332 0.003424516841
0.002413802332 0.003424516842
0.002413802332 0.003424516842
0.002413802332 0.003424516842
0.002413802332 0.003424516842
0.002413802332 0.003424516842
Table 9.8: Calculated values of the integrals IJp,q,r

IJ3,3,3
0.000746365433
0.000896058708
0.000879518117
0.000881385239
0.000881244360
0.000881245407
0.000881246357
0.000881246323
0.001123016446
0.000746332974
0.000896096293
0.000879521003
0.000881385287
0.000881244349
0.000881245406
0.000881246357
0.000881246323
0.000881246318
0.000881246318
0.000881246318
0.000881246318
0.000881246318
232
Table 9.9: Error calculated using the new error formula and the method GQTSS
Points N Error calculated using error formula Error using m m m points
22 4
2.133333333
2.133333333
33 6
0.548571428
0.548571428
44 8
0.139319727
0.139319727
5 5 10
0.035181749
0.035181749
R 1 R 1x R 1xy
Table 9.10: Calculated Errors in the integral x=0 y=0 z=0 xi y j z k dz dy dx for
GQTSS
i, j, k
3, 4, 5
5, 7, 8
7, 4, 9
8, 7, 5
6,10, 4
6, 6, 6
TP
73
103
93
103
103
103
GQTTS
Absolute error
0.407419326993121D-12
0.693087437300628D-12
0.693691154321356D-16
0.295766425612876D-16
0.634844425226502D-16
0.818942693206072D-17
233
TP
208
350
440
350
273
544
GQTTM
Absolute error
0.465243339007592D-12
0.693061853399468D-12
0.668925855145236D-16
0.941456653102749D-16
0.588036204043632D-16
0.854762493675243D-17

234

235
Figure 9.9: Percentage error in I52,1,0 .

236

237
2,2,2
Figure 9.13: Error Precision in I1,1,1
.
0,2,2
.
238
2,2,2
.
2,2,0
.
239
Chapter 10
Appropriate Gauss
Quadrature and
Recursive Schemes to
Obtain Element Matrices
for General Triangular
Finite Elements
Chapter 10
Appropriate Gauss Quadrature
and Recursive Schemes to Obtain
Element Matrices for General
Triangular Finite Elements
This chapter concentrates to show the applications of the derived appropriate
Gaussian quadrature formulae along with the recursive scheme to form the element matrices for the general (straight sided or curved sided) triangular finite
elements. Also it presents a recursive formula to evaluate the integrals when the
Jacobean is a nonlinear function of special type. The formulation of element matrices is one of the most important, time consuming and sensitive task in the
FEM solution procedure. To demonstrates the application, this chapter presents
the Prandtle stresses and torsional constant k for different types of cross sections. Also, the curved triangular finite elements are applied to integrate some
complicated functions over the curved domain in the shape of a lunar model. Obtained results are compared with the results obtained by others and found in good
agreement.
241
10. Element Matrices for General Triangular Finite Elements
10.1
Introduction
The finite element method has proved superior to other numerical methods due to
its better adaptability to any complex geometry. If the domain of a problem consist a polygonal boundary, element with straight sides, usually triangle (in 2D)
or tetrahedron (in 3D) are perfectly satisfactory to represent the domain. But
the domain of real problems often contains curved boundaries, however, elements
with at least one curved side or face are desirable to represent the domain accurately. In classical finite element applications curved boundaries are discretized
by extremely refined meshes because simplifying the curved domains by polygonal domains may cause global changes in the physical solution of the problem.
When dealing with curved or irregular boundaries, the side of the element should
closely approximate the boundary which requires many elements in order to reduce the error in the solution. Many researchers strongly suggested that a large
number of lower order elements should be used in the portion of the domain where
the function varies rapidly to get the desired accuracy, which generally requires
much more computational effort, computing memory and time. Alternatively, a
fewer number of higher order elements are suitable to achieve better accuracy with
less computational effort. For irregular, complicated or distorted type of domain
the integrals cannot be evaluated analytically or explicitly and numerical integration scheme are the only alternative. Complication arise from two main sources:
firstly the large number of integrations that need to be performed and secondly
in methods which use isoparametric or subparametric elements, the presence of
the determinant of Jacobean matrix in the denominator of the stiffness matrix for
which the integrands are rational functions.
Among all the numerical rules, Gaussian quadrature rule occupies a central
role for such calculations. To get the desired accuracy for the polynomial of higher
degree it is necessary to increase the Gauss points with corresponding weights.
In addition specially for two and three dimensional cases the higher order (more
Gauss points and weights) formulae become erroneous due to crowding of Gaussian
points and that effects the accuracy and efficiency of the calculations. Hence, it
is an important task to make a proper balance between accuracy and efficiency of
calculations. The previous chapters of this Thesis successfully derived numerical
242

integration formulae that employ less number of Gauss points and weights in the
calculation process of stiffness matrices in two and three dimensional problems
avoiding the said crowding provided the accuracy and efficiency is assured.
Rathod and Karim [2002] showed that for a rational integral of n-degree
bivariate polynomial numerator with a linear denominator there are (n+1)(n+2)
ra2
tional integral of monomial numerator with the same linear denominator. Rathod
integrals whenand Karim [2001] presented a recursive scheme to compute n(n+1)
2
ever (n + 1) integrals of order 0 to n in one of the variables are known. For a
rational integral of n-degree bivariate polynomial numerator with a nonlinear denominator there are (n+1)(n+2)
rational integral of monomial numerator with the
2
same nonlinear denominator. In this chapter we presented a recursive scheme to
integrals whenever (2n + 1) integrals of order 0 to n are known.
compute n(n1)
2
To demonstrates the application, this chapter presents the Prandtle stresses
and torsional constant k for different types of cross sections. Obtained results are
compared with the results obtained by existing methods: (i) Gaussian 77 points
and 13 13 points and (ii) the analytic method derived by Karim [2001]. In this
chapter we considered general form of elements: (1) triangles (three straight or two
straight and one curved side, (2) tetrahedron (four plane surfaces or three plane
and one curved surfaces). Also, the curved triangular finite elements are applied to
integrate some complicated functions which are very recently studied by Rathod
et al. [2013] over the curved domain in the shape of a lunar model. Obtained
results are found accurate and efficient compared with the results obtained by
other methods.
10.2
Problem Statement
In the FEM solution procedure of physical problems, formulation of element matrices is one of the most important, time consuming and sensitive task. By using
the Galerkins weighted residual FE procedure (Discussed in Section 2.3 in page
30) for any problem we achieve the following FE equations:
[K]{u} = {F }
243

where the stiffness matrix [K] and load vector {F } are given by
kije
ZZ
{
Ni Nj Ni Nj
ij
ij
+
}dxdy = kxx
+ kyy
x x
y y
(10.1)
Fie
ZZ
f (x, y)Ni dxdy
(10.2)
The Lagrange interpolation function for the Prandtl stress function u(x, y) is
u=
NP
X
ui Ni (, )
i=1
where Ni = corresponding shape functions, i =1, NP (Number of points in

an element). The corresponding shape functions for six node quadratic curved
element (NP=6) and ten node cubic curved element (NP=10) are appended.
10.3
Recursive Scheme to Compute the Integrals for Calculation of Stiffness Matrix
ij
ij
For curved elements to calculate kxx
and kyy
we have to calculate the integrals of
rational function of a monomial numerator with a linear or nonlinear denominator.
Rathod and Karim [2002] showed that for a rational integral of n-degree bivariate polynomial numerator with a linear denominator there are (n+1)(n+2)
rational
2
integral of monomial numerator with the same linear denominator. Rathod and
Karim [2001] presented a recursive scheme to compute n(n+1)
integrals whenever
2
(n + 1) integrals of order 0 to n in one of the variables are known. In this section we wish to present numerical integration schemes to compute the integrals
in Eq.10.1 and Eq.10.2 in an efficient manner. Here, two cases depending on the
Jacobian of transformations will be discussed.
10.3.1
Jacobian is a Linear Function
To compute kxx and kyy for a curved element when the Jacobian is a linear function
of the type J(, ) = 0 + 1 + 2 we need to calculate
244
ij
ktz
ZZ
=
Z
Ni Nj
dtdz
t z
Ni Nj
|J|dd
t z
=0 =0
Z 1 Z 1
n X
m
X
mk k
=
dd
Cij
0
1
2
=0
=0
m=0 k=0
n X
m
X
Cij Imk,k
(10.3)
m=0 k=0
where
Z
Imk,k =
=0
=0
mk k
d d
(0 + 1 + 2 )
m=0 to n; k = 0 to m. Now if Ir,0 or I0,r for r = 0 to n i.e. (n+1) integrals

are known or calculated by the derived methods of this thesis, then for m = 0 to
integrals can be calculated accurately by using the
n; k =0 to m the rest n(n+1)
2
recursive relation given by Rathod and Karim [2001]:
1 (m k)! k!
{
0 Imk,k 2 Imk,k+1 }
1 (m + 2)!
1 (m k)! k!
= {
0 Imk,k 1 Imk+1,k }
2 (m + 2)!
Imk+1,k =
if 1 2
Imk,k+2
if 2 > 1
(10.4)
These (n+1) integrals of order 0 to n in one of the variables can be computed by

any one of the derived integration formulae GQUTS, GQUTM, GQSTS, GQSTM,
SGQTS or SGQTM. In this chapter these (n + 1) integrals are computed by
using the numerical integration formulae SGQTS, SGQTM presented in previous
chapters.
10.3.2
Jacobian is a Nonlinear Function
In this section a new recursive scheme is developed to calculate integrals when

the jacobian is a nonlinear function of the type J(, ) = 0 + 1 + 2 + 3 .
To compute kxx and kyy we need to calculate
245
ij
ktz
ZZ
=
Ni Nj
dtdz
t z
Ni Nj
|J|dd
t z
Z 1 Z 1
=
=0
=0
n X
m
X
m=0 k=0
n X
m
X
Cij
=0
=0
mk k
dd
0 + 1 + 2 + 3
Cij IImk,k
(10.5)
m=0 k=0
where
Z
IImk,k =
=0
=0
mk k
d d
(0 + 1 + 2 + 3 )
m = 0, n; k = 0, m.
(10.6)
Now, we know
Z
=0
p q d d =
=0
p! q!
(p + q + 2)!
(10.7)
then
=0
mk
d d =
=0
=0
=0
mk k (0 + 1 + 2 + 3 )
dd
(0 + 1 + 2 + 3 )
By using Eq.10.6 and 10.7 which implies

0 IImk,k + 1 IImk+1,k + 2 IImk,k+1 + 3 IImk+1,k+1 =
(m k)! k!
(m + 2)!
(10.8)
From this we can write

IImk+1,k+1 =
1 (m k)! k!
{
0 IImk,k 1 IImk+1,k 2 IImk,k+1 } (10.9)
3 (m + 2)!
Now if IIr,0 and II0,r for r = 0 to n ie. (2n+1) integrals are known or calculated
by the derived quadrature formulae (so presented in previous chapters) then the
246

rest n(n1)
integrals can be calculated accurately by using the recursive relation
2
given in Eq.10.9.
10.4
Example-1 (Torsion of the Elliptic/circular

Cross Sections)
To show the application of the derived quadrature formula so presented in previous chapters we consider the following two dimensional boundary value torsion
problem following Lewis and Ward [1991]; Segerlind [1984].
2u
x2
2u
y 2
+2=0
u = 0 on C1 ,
and
within A
u
n
= 0 on C2
2
(10.10)
where A is the interior of the quarter ellipse x62 + y22 = 1 and the boundary of
A = C1 + C2 for which the finite element equation is of the form
[k]{{u} = {F }.
and the components of element matrices are given by Eq.10.1 and Eq.10.2.
The domain of the problem is shown in Figure 10.1 and FE models using the
quadratic and the cubic curved triangular elements are shown in Fig 10.2a and
10.2b respectively.
10.4.1
Quadratic Curved Triangular Element (NP = 6)
If nodes 5, 6 are at the mid points of two straight sides (Fig 10.2a), then the
transformation formulae from quadratic curved element to unit triangular element
is
x = x3 + (x1 x3 ) + (x2 x3 ) + (4x4 2x1 2x2 )

y=
y3 + (y1 y3 ) + (y2 y3 ) + (4y4 2y1 2y2 )
247
(10.11)
Figure 10.1: Curved Domain of Example-1

Now the Jacobian for this transformation is of the form
J = 0 + 1 + 2
(10.12)
where 0 , 1 , 2 are constants in terms of xi , yi and for (i, j = 1, 2, ...6).
10.4.2
For Cubic Curved Triangular Element (NP = 10)
If nodes 6, 7, 8 and 9 are at the trisection points of two straight sides (Fig 10.2b)
and following Karim [2001] if chosen
x5 = x4 31 (x1 x2 ),
x10 =
1
(x1
12
+ x2 + 4x3 + 3x4 + 3x5 ),
1
y5 = y4 (y1 y2 )
3
1
y10 = (y1 + y2 + 4y3 + 3y4 + 3y5 )
12
then the transformation formulae from cubic curved element to unit triangular
element is
248
Figure 10.2: Quadratic and Cubic Curved Elements
x = x3 + (x1 x3 ) + (x2 x3 ) + 94 (x4 + x5 x1 x2 )

y = y3 + (y1 y3 ) + (y2 y3 ) + 49 (y4 + y5 y1 y2 )
(10.13)
Now the Jacobian for this transformation is of the form

J = 0 + 1 + 2
where 0 , 1 , 2 are constants in terms of xi , yi .
10.4.3
Calculation Procedure
The calculation process consist of the following steps:

1. For each element the components of element matrices Kij and Fi are obtained.
2. The global FE equations for the whole system is obtained by assembling
element equations.
3. Boundary conditions are imposed and then the global system of equation is
solved the for the generalized stress vectors of the whole system.
249

4. The torsional constant k is calculated for which analytical result is
ZZ
u dxdy
k=2
A
5. A measure of error Ek is provided with the exact solution of the torsional

constant k. Where
Ek = 100|1
10.4.4
k
kexact
Computed Results
The geometry and FE model for this problem are shown in Figures 10.1, 10.2a,
10.2b. To show the application of the quadratic and cubic curved triangular
element we shall illustrate the solution by use of a single element.
We have four cases on account of the numerical values of 1 and 2 in the
Jacobian, which are
Case 1. 1 6= 2 6= 0,
Case 2. 1 = 2 6= 0,
Case 3. 1 = 0, 2 6= 0,
Case 4. 1 6= 0, 2 = 0
The cases can be attributed to a choice of point along the curved side of the
triangle. For six node element considering a = 6, b = 2.
Case-1: For x4 = 3.6, y4 = 3.6, we have 0 = 12, 1 = 21.53679831, 2 =

8.821067231. That is 1 6= 2 6= 0.
Case-2: For x4 = 3 2, y4 = 2, we have 0 = 12, 1 = 2 = 9.941125498.

That is 1 = 2 6= 0.
Case-3: For x4 = 3 3, y4 = 1, we have 0 = 12, 1 = 0, 2 =

17.56921938. That is 1 = 0, 2 6= 0.
Case-4: For x4 = 3, y4 = 3, we have 0 = 12, 1 = 17.56921938, 2 = 0.

That is 1 6= 0, 2 = 0.
250

Obtained results for Prandtl stress values (i = 3, 5, 6), torsional constant k,
percentage error Ek by use of a single six node quadratic curved triangular element
with (a = 6, b = 2) are summarized in Table 10.1. The values of M for which
the M M points formulae SGQTS and M (M2 +1) 1 points formulae SGQTM
matched with the results of Karim [2001] are also given in Table 10.1. Elements of
corresponding stiffness matrix and load vectors are given in Table 10.2 and 10.3.
These tables shows that SGQTM need less number of Gauss-points for desired
results. For both of the formulae, results are always convergent as the number of
points increases. Similar observation is found for the derived formulae GQUTS,
GQUTM, GQSTS and GQSTM also (results are not shown).
For ten node element, we solve the equation for the quarter elliptic element
according to the following choice of x4 , y4 , x5 , y5 as mentioned in Eq.(10.13), we
obtain
x10
b
a
x4 = (1 + 17), y4 = (1 + 17)
6
6
a
b
x5 = (1 + 17), y5 = (1 + 17)
6
6
1
1
= (a + 3x4 + 3x5 ), y10 = (b + 3y4 + 3y5 )
12
12
Computed results of element of stiffness matrix and load vectors are given in
Table 10.4. It is found that the calculated results with SGQTS and SGQTM
completely matched with results of the analytical integration formula given by
Karim [2001] for M = 4 (16 points) and M = 5 (14 points) respectively.
10.5
Example-2 (Torsion of the Trapezoidal Cross

Sections)
Finite element models with straight sided quadratic and cubic triangular elements
are shown in the Figure 10.3. The cross section is modelled by eight six node
element and four ten node element as shown in Fig 10.3. Results are tabulated
in Table 10.5. Nguyen [1992] has shown that the approximate result for k is
erroneous, it should be greater than 0.1406 and finally the computed value of k
251

converges to 0.1682. From Table 10.5 it is evident that the computed value for
k has very good agreement with Nguyen [1992] and the obtained FE results are
consistent. Here AB=1.2, CD= BD =1.
Figure 10.3: Discretization of the Trapezoidal Cross Sections
10.6
Example-3 (Torsion of the Part of an Elliptical Cross Section)
Here we consider the Poissons equation in two dimension given by

2u 2u
+
+ f (x, y) = 0 (x, y)
x2 y 2
(10.14)
u = 0 on the boundary of .
For this problem we have to determine
ZZ
Fi =
f (x, y)Ni dxdy
and other calculation are the same as before.

For this example, the center of the conic does not lie at the origin of the
global (x, y) coordinate system. The domain is the interior part of the conic
252

(ellipse) 2x2 4xy + 5y 2 x + 5y 4 = 0 as shown in the figure 10.4 and
f (x, y) = (8x2 42xy + 8y 2 10x + 2y).
Figure 10.4: Discretization of Part of an Elliptical Cross Section

Choosing intermediate nodal points are x4 =1.572685023, y4 =0.5057195085;
x5 =1.010638135, y5 = 0.6806178673 in this example and after imposing the boundary condition we have only one equation k10,10 u10,10 = f10 . Computed results using formulae SGQTS and SGQTM for M=10 are listed in Table 10.6.
Exact solution is u(x, y) = (4 + x 5y 2x2 + 4xy 5y 2 )xy Which gives
u10 = u(x10 , y10 ) = 0.626042176 where x10 = 0.786342511, y10 = 0.340308933.
10.7
Example-4: Integration (A Lunar Model)
The curved triangular finite elements are applied to integrate some complicated
functions which are very recently studied by Rathod et al. [2013] over the curved
domain in the shape of a lunar model. The theoretical developments of the preRR
vious chapters used to compute the integral f (x, y)dxdy over the domain in
the shape of a lunar model. The boundary of the domain is composed of two
circular arcs. The outer circular arcs satisfies the equations
(x 0.5)2 + (y 0.5)2 = 0.25
253

and the inner arcs satisfies the equations
x2 + y 2 = 0.25
The lunar domain is subdivided into a finite number of curved triangles. Each
of the curved triangles is transformed into unit triangles using the Eq. 10.13 described in section 10.4.2. The numerical integration formula SGQTS and SGQTM
developed in the previous chapters are used to evaluate Integrals of the following
functions over this lunar domain:
f1 (x, y) = (x + y)19
p
(x 0.5)2 + (y 0.5)2
f2 (x, y) =
2 +(y0.5)2
f3 (x, y) = e(x0.5)
f4 (x, y) = e100(x0.5)
2 +(y0.5)2
f5 (x, y) = cos[20(x + y)]

f6 (x, y) = 1
and the exact values of integrals are reported below:
ZZ
ZZ
f1 (x, y)dxdy =
ZZ
ZZ p
(x 0.5)2 + (y 0.5)2 dxdy = 0.20646770293563
f2 (x, y)dxdy =
ZZ
ZZ
f3 (x, y)dxdy =
(x + y)19 dxdy = 638.55743274702
2 +(y0.5)2
e(x0.5)
254
dxdy = 0.57263720432530

ZZ
ZZ
f4 (x, y)dxdy =
ZZ
2 +(y0.5)2
dxdy = 0.03137185199242
ZZ
cos[20(x + y)]dxdy = 0.0062895812195655
f5 (x, y)dxdy =
ZZ
ZZ
1dxdy = 0.6426990816987241
f6 (x, y)dxdy =
e100(x0.5)
Using the numerical scheme SGQTS, SGQTM results are summarized in Table
10.7. The distribution of Gaussian points (for m =10) in the Lunar domain for
these two methods SGQTS, SGQTM are shown in Fig 10.5 and 10.6. Fig 10.5
shows that there are a huge crowding of Gaussian points with in the lunar model
domain, whereas Fig 10.6 is free of such crowding. The present schemes are
shown efficient to evaluate the integrals of these complicated test functions using
significantly less number of Gaussian points and weights with desired accuracy.
10.8
Conclusions
Two recursive schemes to evaluate the integrals of rational functions of bivariate

polynomial numerator with linear/ nonlinear denominator are successfully presented in this chapter. It is clearly shown that the application of the recursive
schemes require the evaluation of few integrals of rational functions of monomial
numerators with linear/ nonlinear denominators by the so developed quadrature
schemes in previous chapters. Thus a substantial reduction in computational effort is achieved. The application of the quadrature rules along with the recursive
schemes so presented is shown to compute the Prandtls stress function values and
torsion constant k for different type cross sections. Obtained results are compared
with the results obtained by existing methods: (i) Gaussian 7 points and 13 points
quadrature rule for triangles and (ii) the analytical method presented by Karim
[2001]. The number of quadrature points for which the computed results of the
new methods matched with the results of the Analytical method of Karim [2001]
are shown for different examples.
It is seen that the method SGQTM with less number of quadrature points
evaluate the integrals accurately and the method is always convergent. Hence, it
255
Figure 10.5: Gaussian Points within the Lunar Domain, SGQTS, m = 10

can be said that the developed quadrature schemes along with recursive schemes
remove all the difficulties pertinent to form the element matrices in FEM solution
procedure. It is found that the developed integration formulae of this thesis are
always consistent, reliable and provides considerable savings in computational
effort, in comparison with existing numerical methods.
Accordingly, a computer code (Program 12: NIPTE.FOR) in FORTRAN to
calculate the component of element matrices for the general triangular element is
appended. In the program one can call the subroutine NIUTS, NIUTM, NISTS,
NISTM, NISYS and NISYM for the quadrature schemes GQUTS, GQUTM,
GQSTS, GQSTM, SGQTS and SGQTM respectively as desired. Furthermore,
the application of the curved triangular finite elements are shown to integrate
256
Figure 10.6: Gaussian Points within the Lunar Domain, SGQTM, m = 10

some complex functions over the curved domain in the shape of a Lunar model.
Through such demonstrations the versatility of the curved/straight sided triangular elements are now enhanced by the developed numerical integration schemes.
257
Table 10.1: Computed Prandtl stress values ui , torsional constant k, error Ek for
Example -1.
SGQTM
SGQTS
Analytic
13 13
77
Method
u3
u5
u6
k
u3
u5
u6
k
u3
u5
u6
k
u3
u5
u6
k
Ek
M
u3
u5
u6
k
Ek
M
Computed FE results for the condition

1 6= 2 6= 0
1 = 2 6= 0
1 = 0, 2 6= 0
1 6= 0, 2 = 0
3.560457776
3.490275770
3.186710601
3.738911961
3.214656567
2.933900869
2.545774511
3.234029318
2.070848118
2.703925599
2.294990780
2.529061082
103.8036322
130.4159926
108.3265740
126.6798967
3.546356650
3.492068111
3.189393678
3.737759703
3.194019614
2.934362916
2.547020027
3.231648451
2.071170347
2.703223660
2.292496376
2.529410073
103.4720605
130.4078708
108.30343330
126.6408518
3.54196547
3.492101285
3.189536668
3.737705041
3.187950666
2.934370961
2.547081760
3.231540848
2.071201396
2.703211030
2.292375269
2.529426400
103.37315710
130.40771690
108.30232690
126.63910420
3.541965399681 3.491994911356 3.189536629912 3.737705004328
3.187950631942 2.934340181661 2.547081746172 3.231540832514
2.071201388872 2.703254324973 2.292375270109 2.529426396964
103.3731367
130.4077183
108.3023263
126.6391035
23.832
3.912
20.1999
6.689
9 (81 points)
4 (16 points)
5 (25 points)
5 (25 points)
3.541971258344 3.491996937155 3.189523128574 3.737717446838
3.187958674601 2.934340301911 2.547074980460 3.231556794684
2.071201358230 2.703253708652 2.292389141276 2.529423091712
103.3731546
130.4077183
108.3023263
126.6391035
23.832
3.912
20.1999
6.689
10 (54 points)
5 (14 points)
6 (20 points)
6 (20 points)
Exact solution is u1 = 3.6, u5 = 2.7, u6 = 2.7, k = 135.7168026.

Here results of SGQTS and SGQTM match with the analytic method for given value of M.
Shown results of SGQTS and SGQTM are calculated using M=10.
258
259
SGQTM
SGQTS
Method
condition
condition
condition
condition
condition
condition
condition
condition
1
2
3
4
1
2
3
4
Condition
k33
1.408954947029
1.340335082143
1.398016972353
1.345959227348
1.408954687936
1.340336710677
1.398009859097
1.345946297006
Computed Values for

k35
k36
-1.374207110430 -0.396626857069
-1.801856111014 0.101893718454
-2.078285484105 0.236312248423
-1.548525761453 -0.126311101503
-1.374205877846 -0.396626966276
-1.801858026482 0.101892792910
-2.078279590917 0.236316267634
-1.548515532641 -0.126302076437
six node FE model

k55
k56
2.762088187625 0.155290312142
4.549403989170 -0.395523785596
5.77450845300 -0.757648554857
3.457220409217 -0.084183172761
2.762082037855 0.155290833906
4.549406217992 -0.395522672844
5.774487559547 -0.757624152697
3.457195255396 -0.084180461410
Table 10.2: Calculated elements of corresponding stiffness matrix for Example -1
k66
3.473000140410
2.512902376386
2.768895479951
2.801709347663
3.473000094361
2.512902878264
2.768845175747
2.801697361016
Table 10.3: Computed Values for six node FE model for example -1 (same for
both SGQTS and SGQTM)
F3
F5
F6
Condition 1
Condition 2
Condition 3
Condition 4
-0.211928851253 -0.331370866666 -0.292820323027 -0.292820323027
4.259644256269 5.988225173333 6.342562584220 5.171281292110
6.283501958777 5.988225226666 5.171281292110 6.342562584220
260
K
3
6
7
8
9
10
261
6
1.0336408565
6.6949714454
-4.6536505668
-0.3055757768
0.1409903483
0.0887848905
7
-1.8618026534
-4.6536505668
5.6678980122
-0.4448290735
0.0233451488
0.0786123665
8
0.1834227902
-0.3055757768
-0.4448290735
3.4970038177
-0.2907481188
-3.6294305476
9
0.0014101027
0.1409903483
0.0233451488
-0.2907481188
2.7984437166
-1.4636074881
10
-0.2181688406
0.0887848905
0.0786123665
-3.6294305476
-1.4636074881
1.0176103061
u3 = 3.565524275402, u6 = 1.956740306259, u7 = 3.145996809511,

u8 = 3.176043558653, u9 = 1.989711140650, u10 = 2.282662995131;
Result shown for M=10.
3
1.2104154852
1.0336408565
-1.8618026534
0.1834227902
0.0014101027
-0.2181688406
F
0.5684658437
1.65809628
0.8999999999
0.8999999999
1.6580963137
8.4323851875
Table 10.4: Computed elements of Stiffness matrix of TEN node cubic FE model using SGQTM.
8 six node
4 ten node
262
SGQTM
Analytical M.
SGQTM
Analytical M.
FE Model
i
7
8
9
7
8
9
5
6
7
9
10
5
6
7
9
10
Computed Stress
ui
i
0.0960802472
12
0.1236618019
13
0.0903051441
14
0.09608024972341 12
0.12366180644874 13
0.09030514551905 14
0.055549792
12
0.089792221
13
0.0550452768
14
0.1241786087
16
0.1238791629
17
0.059023000487
12
0.096295652764
13
0.058562968131
14
0.129507227925
16
0.129106499063
17
function values ui
ui
0.123643778
0.1621862722
0.1209024325
0.123643783156
0.162186275133
0.120902433437
0.094800726
0.1620487414
0.0923284681
0.1318099076
0.1343235984
0.094725716001
0.163109710489
0.092548242198
0.136907410754
0.134420821735
at node i
i
ui
17
0.0972297317
18
0.1233425114
19
0.100282778
17 0.097297319135
18 0.123342514484
19 0.100282779360
19
0.056065628
20
0.101048889
21
0.06697125
19 0.056777163927
20 0.101949643739
21 0.066973467265
0.16780884
0.167526745
0.1605573640
0.1605573601
Torsional
Constant k
Table 10.5: Computed Stress function values ui , the torsional constant k for example-2
Table 10.6: Computed Values for six node FE model for example-3, m=10
Method
Gauss 7-point
Gauss 13-point
Analytical M.
SGQTS
SGQTM
K10,10
7.811317529
8.262921381
8.274075346
8.274075342
8.274075300
263
F10
u10
5.368507707 0.687273009
5.385059531 0.651713756
5.385091615 0.6508390835
5.372745751 0.6493469698
5.302350861 0.6408390870
264
f1
f2
f3
f4
f5
f6
f1
f2
f3
f4
f5
f6
f1
f2
f3
f4
f5
f6
f
f1
f2
f3
f4
f5
f6
SGQTS
Ea
SGQTM
Ea
0.6385586779312916D+03 0.1245184271624566D-02 0.6386421161691412D+03 0.8468342212120206D-01
0.2064676940601683D+00 0.8875461671165041D-08 0.2064676832778503D+00 0.1965777970958094D-07
0.5726372020008313D+00 0.1645301561659096D-08 0.5726372042263089D+00 0.3870779141657010D-08
0.3115537787347361D-01 0.2164741189463852D-03 0.3165191572921538D-01 0.2800637367953787D-03
0.5290947148691215D-02 0.9986340708742856D-03 0.1046493985035890D-01 0.4175358630793395D-02
0.6426990807542486D+00 0.9444754978105152D-09 0.6426990807542489D+00 0.9444752757659103D-09
Number of triangle = 200, (m=5) SGQTS with 25 points, SGQTM with 14 points
0.6385586866716534D+03 0.1253924633374481D-02 0.6386421544879713D+03 0.8472174095129503D-01
0.2064676999357412D+00 0.2999888781829441D-08 0.2064676970158540D+00 0.5919775947216266D-08
0.5726372015421841D+00 0.1186654330354031D-08 0.5726372029562166D+00 0.2600686777043393D-08
0.3115544111954633D-01 0.2164108728736729D-03 0.3165178537247875D-01 0.2799333800587514D-03
0.5290956980767337D-02 0.9986242387981636D-03 0.1046457069760205D-01 0.4174989478036545D-02
0.6426990816396912D+00 0.5903288968767129D-10 0.6426990816396914D+00 0.5903277866536882D-10
0.6385574325131282D+03 0.2338917965971632D-06 0.6385574316997154D+03 0.1047304635903856D-05
0.2064677035093190D+00 0.5736890129615091D-09 0.2064677023774669D+00 0.5581630713180630D-09
0.5726372045185323D+00 0.4163002498280832D-08 0.5726372035271892D+00 0.3171659379574976D-08
0.3137186096244889D-01 0.8970028893273696D-08 0.3137186784505314D-01 0.1585263313674812D-07
0.6289581650472803D-02 0.4309073028951893D-09 0.6289566892772273D-02 0.1432679322746488D-07
0.6426990816396914D+00 0.5903277866536882D-10 0.6426990816396915D+00 0.5903266764306636D-10
0.6385574323024241D+03 0.4445959120857879D-06 0.6385574325546615D+03 0.1923584704854875D-06
0.2064677030899757D+00 0.1543457028851947D-09 0.2064677035359275D+00 0.6002975061480953D-09
0.5726372033211775D+00 0.2965647727393161D-08 0.5726372045429263D+00 0.4187396540622501D-08
0.3137185186267821D-01 0.1297417936974021D-09 0.3137185184658459D-01 0.1458354062511269D-09
0.6289581219460250D-02 0.1052500100962028D-12 0.6289581346177302D-02 0.1266118017720363D-09
0.6426990816396916D+00 0.5903255662076390D-10 0.6426990816396912D+00 0.5903288968767129D-10
Table 10.7: Computed Integrals and Absolute Error for ten node Lunar model
Chapter 11
Conclusions
Chapter 11
Conclusions
The main concern of this thesis was to develop suitable quadrature schemes to
evaluate the triangular/tetrahedral domain integrals encountered in many areas
of science and engineering.
The employment of general triangular/ tetrahedral finite elements in FEM
solution procedure give rise to huge number of triangular or tetrahedral domain
integrals of rational functions of n-th degree bivariate or trivariate polynomial
numerators with linear or nonlinear denominators. Generally, such integrals as
well as domain integrals of nonpolynomial, irrational functions are evaluated by
numerical integration schemes that employ Gaussian quadrature.
In some recent studies, many researchers have also shown that the most accurate 7 - points and 13 - points Gaussian quadrature rules for the triangle are not
satisfactory in view of accuracy and for some element geometry these formulae are
not at all reliable. Hence, among the others Nguyen [1992] first proposed a technique to evaluate the triangular domain integrals using the quadrature rules for
square domain integrals to achieve the desired accuracy. This technique seriously
affected by the huge crowding of quadrature points and that leads to computational inefficiency. Similar technique for tetrahedral domain integrals lead to
severe computational inefficiency. Thus, the higher order quadrature rules for triangles and tetrahedrons which ensures the efficiency and accuracy of calculations
are highly demanded.
Therefore, this thesis was fully devoted to develop higher order quadrature
schemes for triangular and tetrahedral domain integrals ensuring the accuracy and
efficiency of evaluations. Furthermore, it is concentrated to present the recursive
266
11. Conclusions
schemes to compute the components of element matrices efficiently. The gradual
development of the research work, relevant topics and demonstrations are shown
through out the thesis.
In Chapter 1 the objectives and scope of the Thesis, important applications
and relevant literature review are described. In Chapter 2 some basic concepts
and a brief historical background of FEM, elements, Interpolation functions are
discussed. Some problems and some mathematical preliminaries are also included.
The numerical integration scheme IOUT that can evaluate the triangular domain integrals using the quadrature schemes for square domain integrals is presented in Chapter 3. This technique is almost similar to the extended Gaussian
quadrature for triangles. It is demonstrated that the scheme IOUT so developed
is always reliable and appropriate in view of accuracy.
Chapter 4 presented the numerical integration schemes IOST by transforming
the triangular domain integrals into square domain integrals using the transformation of arbitrary triangle first to standard triangle, then standard triangle is
subdivided into three quadrilaterals. The reliability and accuracy of the scheme
is demonstrated through application examples.
The developed quadrature schemes are shown sufficient to obtain the desired
accuracy of evaluation of the triangular domain integrals. As it is so, such schemes
so developed in Chapter 3 and 4 are used to test the accuracy of the higher
order quadrature schemes developed in other chapters.
The higher order Gaussian quadrature schemes GQUTS, GQUTM; GQSTS,
GQSTM; SGQTS and SGQTM are respectively presented in Chapter 5, 6 and
7. It is thoroughly investigated that the developed quadrature schemes based
on different technique/approaches are also different but their accuracy is almost
same. Any one of the higher order quadrature schemes so presented may be used
for obtaining the exact value of the triangular domain integrals efficiently and
hence its direct implication enhanced the versatility of triangular finite elements.
Chapter 8 developed two sets of quadrature formulae to evaluate the tetrahedral domain integrals. It is shown that the quadrature formulae of one set GQTTS
exhibits crowding of Gauss quadrature points but the quadrature schemes of the
other set GQTTM are completely free of such crowding. The accuracy and efficiency of the schemes are investigated through practical application examples.
267
11. Conclusions
Chapter 9 also derived quadrature schemes for the tetrahedral domain integrals using a standard tetrahedral element but this concept is different from the
concept of Chapter 8. Here also, two sets of quadrature formulae GQTSS and
GQTSM are presented among which the scheme GQTSM is totally free of crowding of quadrature points. The efficiency and accuracy of the presented quadrature
schemes of Chapter 8 and 9 are thoroughly investigated. It is observed that all
the schemes are accurate in view of accuracy but the schemes free of crowdings
of quadrature points are so efficient as they employ very less number of Gauss
points.
It is our firm belief that the thesis may be considered as the complete reference
for the higher order quadrature schemes to evaluate the triangular or tetrahedral
domain integrals encountered in many areas of continuum mechanics. The computer codes in FORTRAN for such schemes enhances the contribution of the
thesis. So, researchers may use such computer codes whenever needed.
268
Appendix A
Shape functions for a quadratic curved triangular element:
The shape functions Ni which transforms a quadratic curved triangular element to a orthogonal unit triangle are listed below:
N1 (, ) = + 2 2 ,
N2 (, ) = + 2 2 ,
N3 (, ) = 1 3 3 + 2 2 + 2 2 + 4,
N4 (, ) = 4,
N5 (, ) = 4 4 4 2 ,
N6 (, ) = 4 4 4 2
269
Appendix A
Shape functions for a cubic curved triangular element:
The shape functions Ni which transforms a cubic curved triangular element to

a orthogonal unit triangle are listed below:
N1 (, ) =
N2 (, ) =
N3 (, ) =
N4 (, ) =
N5 (, ) =
N6 (, ) =
N7 (, ) =
N8 (, ) =
N9 (, ) =
N10 (, ) =
1
(2 9 2 + 9 3 ),
2
1
(2 9 2 + 9 3 ),
2
1
(2 11 11 + 18 2 + 18 2 + 36 9 3 9 3 27 2 27 2 ),
2
9
( + 3 2 ),
2
9
( + 3 2 ),
2
9
( + + 4 2 3 2 9 3 ),
2
9
(2 5 5 2 + 3 2 + 6 2 + 3 3 ),
2
9
(2 5 5 2 + 3 2 + 6 2 + 3 3 ),
2
9
( + 4 2 + 3 3 3 2 ),
2
27( 2 2 )
270
Appendix A
Table 1: Computed Gauss points (p, q) and corresponding weights L for

m(m+1)
1 point method (SGQTM) for triangles.
2
n
m=6, 20 points
m=7, 27 points
m=8, 35 points
p
0.888727297463133
0.630192003213731
0.227425541683039
0.697064941206465
0.414209254015687
0.430568155797026
0.169990521792428
0.178750067572431
2.708354560136226E-002
0.000000000000000E+000
0.000000000000000E+000
0.917061053145689
0.716493205000875
0.392141691261992
0.774513554793957
0.549203619608679
0.198198215856174
0.561205871610772
0.333479209306941
0.327826332907397
0.129427521840461
0.131213040406755
1.949437207665818E-002
0.000000000000000E+000
0.000000000000000E+000
0.000000000000000E+000
0.935854278720920
0.776394467312567
0.512159689180237
0.178766956543018
0.826450513922025
0.645699842407919
0.353395993824022
0.655453872674869
0.464779005004411
0.167730812890342
0.453089922969332
0.269234655052842
0.255824157429295
0.101000692742857
0.100104541368752
1.469128025426521E-002
0.000000000000000E+000
0.000000000000000E+000
0.000000000000000E+000
q
-0.906179845938664
-0.906179845938664
-0.906179845938664
-0.538469310105683
-0.538469310105683
0.000000000000000E+000
0.000000000000000E+000
0.538469310105683
0.906179845938664
-0.538469310105683
0.538469310105683
-0.932469514203152
-0.932469514203152
-0.932469514203152
-0.661209386466265
-0.661209386466265
-0.661209386466265
-0.238619186083197
-0.238619186083197
0.238619186083197
0.238619186083197
0.661209386466265
0.932469514203152
-0.932469514203152
-0.238619186083197
0.661209386466265
-0.949107912342758
-0.949107912342758
-0.949107912342758
-0.949107912342758
-0.741531185599394
-0.741531185599394
-0.741531185599394
-0.405845151377397
-0.405845151377397
-0.405845151377397
0.000000000000000E+000
0.000000000000000E+000
0.405845151377397
0.405845151377397
0.741531185599394
0.949107912342758
0.000000000000000E+000
-0.741531185599394
0.741531185599394
271
L
9.671808407443765E-003
2.036612959907326E-002
2.641521865037473E-002
2.180780247074805E-002
4.405510797397064E-002
2.473634454310782E-002
4.637476656800330E-002
1.534040420130362E-002
2.778564607155512E-003
5.236305923555275E-002
2.454464672208581E-002
5.358724942779149E-003
1.157558519997316E-002
1.580200602497686E-002
1.283435698778468E-002
2.702557440373014E-002
3.505263253647306E-002
1.716438073492758E-002
3.467468341195214E-002
1.549087599565812E-002
2.904171054783806E-002
8.487682963359328E-003
1.446203274907969E-003
1.729720730431097E-002
4.121366590559383E-002
1.358029274137493E-002
3.193509426162615E-003
7.015570468089016E-003
9.896667159022517E-003
1.144177445787316E-002
7.884269377752301E-003
1.703110961959155E-002
2.324942473072126E-002
1.149574334237841E-002
2.420682760510148E-002
3.139666969662116E-002
1.237822093354783E-002
2.500590605057915E-002
9.864562308726572E-003
1.849370967345216E-002
5.020494508311843E-003
8.237200310701331E-004
2.972154195011338E-002
2.544930805858548E-002
8.032791213298952E-003
Appendix A
Table 2: Computed Gauss points (p, q) and corresponding weights L for

m(m+1)
1 point method (SGQTM) for triangles.
2
n
m=9, 44 points
m=10, 54 points
p
0.948937348485512
0.819431649704404
0.601192898894164
0.317815348416544
0.862660296884733
0.715671976824132
0.472103231845533
0.164785436484144
0.723947440393452
0.565614928297778
0.309564965916809
0.551758363089550
0.391249046943736
0.141195105587486
0.369977534931071
0.219847692355784
0.204290635247786
8.065499321191914E-002
7.875073467027130E-002
1.146333102035305E-002
0.0
0.0
0.0
0.0
0.958402053212199
0.851291661485933
0.668593449333590
0.426495791355461
0.146504277333313
0.888786158306404
0.767489559872218
0.563084515391895
0.297669686013274
0.774652110792637
0.642659468024678
0.423939488558651
0.147974105985064
0.628429701049770
0.490987605545342
0.268720915541678
0.466234757101576
0.330604693233132
0.119309593041598
0.306173964336758
0.181934396453014
0.166469949200553
6.572319189508555E-002
6.350487906200106E-002
9.191347145602645E-003
0.0
0.0
0.0
0.0
q
-0.960289856497536
-0.960289856497536
-0.960289856497536
-0.960289856497536
-0.796666477413627
-0.796666477413627
-0.796666477413627
-0.796666477413627
-0.525532409916329
-0.525532409916329
-0.525532409916329
-0.183434642495650
-0.183434642495650
-0.183434642495650
0.183434642495650
0.183434642495650
0.525532409916329
0.525532409916329
0.796666477413627
0.960289856497536
0.183434642495650
-0.960289856497536
-0.525532409916329
0.796666477413627
-0.968160239507626
-0.968160239507626
-0.968160239507626
-0.968160239507626
-0.968160239507626
-0.836031107326636
-0.836031107326636
-0.836031107326636
-0.836031107326636
-0.613371432700590
-0.613371432700590
-0.613371432700590
-0.613371432700590
-0.324253423403809
-0.324253423403809
-0.324253423403809
0.0
0.0
0.0
0.324253423403809
0.324253423403809
0.613371432700590
0.613371432700590
0.836031107326636
0.968160239507626
0.836031107326636
0.324253423403809
-0.324253423403809
-0.836031107326636
272
L
2.015983497663494E-003
4.480916044841954E-003
6.464359484633224E-003
7.747662769908705E-003
5.055663745070523E-003
1.110639128725685E-002
1.566747257515452E-002
1.811354111938598E-002
7.745946956367606E-003
1.673231410556490E-002
2.284153446588211E-002
9.191827856851423E-003
1.935542449754594E-002
2.510431683577034E-002
8.770885597454200E-003
1.771853503082168E-002
6.471997505241379E-003
1.213345702762273E-002
3.140105492486543E-003
5.024749628294020E-004
2.105991205229387E-002
8.191474625434849E-003
2.500282281758626E-002
5.024168787978471E-003
1.333102027244705E-003
2.988298775027559E-003
4.380659742804402E-003
5.384022372785824E-003
5.909044540081589E-003
3.369591872501191E-003
7.489574345999887E-003
1.080477752203405E-002
1.294973968319760E-002
5.320342238493575E-003
1.168784271708034E-002
1.648770968862596E-002
1.906183693488799E-002
6.694804681384016E-003
1.446170176936692E-002
1.974188730358828E-002
7.072261232401939E-003
1.489220864908214E-002
1.931544949367340E-002
6.250944716822827E-003
1.262786770048660E-002
4.381211780382315E-003
8.213730741263730E-003
2.056991588376092E-003
3.234696324495897E-004
3.291186541401748E-003
1.500924217028327E-002
2.160988400642072E-002
1.369154378180941E-002
Appendix A
Table 3: Computed Gauss points (p, q, r) and corresponding weights L for M =

10 method GQTTM for tetrahedron.
p
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
0.013046735741414
q
0.010743648455019
0.010743648455019
0.010743648455019
0.010743648455019
0.010743648455019
0.010743648455019
0.010743648455019
0.010743648455019
0.010743648455019
0.010743648455019
0.010743648455019
0.010743648455019
0.055731967907878
0.055731967907878
0.055731967907878
0.055731967907878
0.055731967907878
0.055731967907878
0.055731967907878
0.055731967907878
0.055731967907878
0.055731967907878
0.055731967907878
0.133163679476162
0.133163679476162
0.133163679476162
0.133163679476162
0.133163679476162
0.133163679476162
0.133163679476162
0.133163679476162
0.133163679476162
0.133163679476162
0.237314822536963
0.237314822536963
0.237314822536963
0.237314822536963
0.237314822536963
0.237314822536963
0.237314822536963
0.237314822536963
0.237314822536963
0.360463383316429
0.360463383316429
0.360463383316429
0.360463383316429
0.360463383316429
0.360463383316429
0.360463383316429
0.360463383316429
0.493476632129293
0.493476632129293
0.493476632129293
0.493476632129293
0.493476632129293
0.493476632129293
0.493476632129293
0.626489880942157
0.626489880942157
0.626489880942157
0.626489880942157
0.626489880942157
0.626489880942157
0.749638441721623
0.749638441721623
0.749638441721623
0.749638441721623
0.749638441721623
0.853789584782424
0.853789584782424
0.853789584782424
0.853789584782424
0.931221296350708
0.931221296350708
0.931221296350708
0.976209615803566
0.976209615803566
r
0.009000343078836
0.046800828160385
0.112311611011103
0.201432090647445
0.308564484743052
0.426977779097407
0.549231836706159
0.667645131060515
0.774777525156121
0.863898004792463
0.929408787643181
0.967209272724731
0.010136968592262
0.052584856125216
0.125644099593486
0.223913962990053
0.340108484621152
0.465610648175354
0.591112811729556
0.707307333360655
0.805577196757222
0.878636440225491
0.921084327758446
0.011139167091428
0.057603746063275
0.136858385783597
0.241880555591099
0.363341112359269
0.490448472423154
0.611909029191324
0.716931198998827
0.796185838719148
0.842650417690996
0.011934154220146
0.061458692597240
0.144915818357696
0.253282805339145
0.374819220860811
0.496355636382478
0.604722623363926
0.688179749124382
0.737704287501476
0.012439001537527
0.063693197178343
0.148624572011216
0.255784966802195
0.370704914139962
0.477865308930941
0.562796683763814
0.614050879404630
0.012557028009562
0.063774160020431
0.146600766830786
0.246738316064646
0.346875865298507
0.429702472108862
0.480919604119731
0.012171133693667
0.061060805395110
0.137224952088334
0.223238431228095
0.299402577921320
0.348292249622762
0.011132456605728
0.054764036883816
0.118657411268482
0.182550785653147
0.226182365931235
0.009245799846884
0.043945276380129
0.089218403096034
0.123917879629279
0.006281085598081
0.027865983953939
0.049450882309797
0.002270400063718
0.008473248391302
273
L
0.000021086958459
0.000047800933658
0.000071553598188
0.000090814044281
0.000104369100374
0.000111366527676
0.000111366527676
0.000104369100374
0.000090814044281
0.000071553598188
0.000047800933658
0.000021086958459
0.000053546428804
0.000120793127271
0.000179188652286
0.000224304198222
0.000252786187266
0.000262520924912
0.000252786187266
0.000224304198222
0.000179188652286
0.000120793127271
0.000053546428804
0.000087221974090
0.000195517907109
0.000286617065143
0.000352264905689
0.000386615967301
0.000386615967301
0.000352264905689
0.000286617065143
0.000195517907109
0.000087221974090
0.000116860652101
0.000259745564195
0.000374719964555
0.000449109293105
0.000474835765027
0.000449109293105
0.000374719964555
0.000259745564195
0.000116860652101
0.000137086819800
0.000301155285992
0.000424831258156
0.000491157615021
0.000491157615021
0.000424831258156
0.000301155285992
0.000137086819800
0.000143441629306
0.000309853709416
0.000422985974226
0.000463009321186
0.000422985974226
0.000309853709416
0.000143441629306
0.000133493245590
0.000281099524076
0.000364590893663
0.000364590893663
0.000281099524076
0.000133493245590
0.000107845482009
0.000217864425396
0.000258949491601
0.000217864425396
0.000107845482009
0.000070977335331
0.000133065633519
0.000133065633519
0.000070977335331
0.000031981545160
0.000051170472256
0.000031981545160
0.000004919349119
0.000004919349119
Appendix A
Table 4: Continuation of Table 3

p
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.067468316655508
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
q
0.012166494443092
0.012166494443092
0.012166494443092
0.012166494443092
0.012166494443092
0.012166494443092
0.012166494443092
0.012166494443092
0.012166494443092
0.012166494443092
0.012166494443092
0.062916342903180
0.062916342903180
0.062916342903180
0.062916342903180
0.062916342903180
0.062916342903180
0.062916342903180
0.062916342903180
0.062916342903180
0.062916342903180
0.149480367469124
0.149480367469124
0.149480367469124
0.149480367469124
0.149480367469124
0.149480367469124
0.149480367469124
0.149480367469124
0.149480367469124
0.264188373451698
0.264188373451698
0.264188373451698
0.264188373451698
0.264188373451698
0.264188373451698
0.264188373451698
0.264188373451698
0.396850822703577
0.396850822703577
0.396850822703577
0.396850822703577
0.396850822703577
0.396850822703577
0.396850822703577
0.535680860640916
0.535680860640916
0.535680860640916
0.535680860640916
0.535680860640916
0.535680860640916
0.668343309892794
0.668343309892794
0.668343309892794
0.668343309892794
0.668343309892794
0.783051315875368
0.783051315875368
0.783051315875368
0.783051315875368
0.869615340441313
0.869615340441313
0.869615340441313
0.920365188901401
0.920365188901401
0.013367999605811
0.013367999605811
0.013367999605811
0.013367999605811
0.013367999605811
r
0.010018792579021
0.051971825849235
0.124179350182252
0.221303590942994
0.336143525628120
0.460182594450700
0.584221663273280
0.699061597958407
0.796185838719148
0.868393363052166
0.910346396322380
0.011345641543418
0.058671483157382
0.139395178702936
0.246364028614955
0.370075965732413
0.499539374708900
0.623251311826357
0.730220161738377
0.810943857283931
0.858269698897895
0.012466083175355
0.064198028585252
0.151375004189405
0.264572123000961
0.391525657937684
0.518479192874407
0.631676311685963
0.718853287290116
0.770585232700013
0.013270004372377
0.067948299748769
0.158553619796689
0.272872996889125
0.395470313003669
0.509789690096105
0.600395010144026
0.655073305520417
0.013630958658022
0.069228398473477
0.159138690332064
0.267840430320458
0.376542170308851
0.466452462167439
0.522049901982894
0.013399764423026
0.067224666852556
0.151077301196801
0.245773521506776
0.329626155851020
0.383451058280551
0.012393096949226
0.060965521130610
0.132094186725849
0.203222852321088
0.251795276502472
0.010378697585519
0.049329938070762
0.100150429398362
0.139101669883605
0.007090776624761
0.031458171451590
0.055825566278419
0.002571082800639
0.009595411642452
0.010780997661352
0.055751551843689
0.132457833243631
0.234103114061445
0.351658220984276
274
L
0.000059509032276
0.000134243912621
0.000199142006864
0.000249281344617
0.000280934914111
0.000291753652722
0.000280934914111
0.000249281344617
0.000199142006864
0.000134243912621
0.000059509032276
0.000150952285656
0.000338376599161
0.000496039003275
0.000609653624843
0.000669103910383
0.000669103910383
0.000609653624843
0.000496039003275
0.000338376599161
0.000150952285656
0.000242902915463
0.000539899048033
0.000778881259316
0.000933504603098
0.000986978847182
0.000933504603098
0.000778881259316
0.000539899048033
0.000242902915463
0.000317364752095
0.000697193740583
0.000983511523017
0.001137061279546
0.001137061279546
0.000983511523017
0.000697193740583
0.000317364752095
0.000357101778493
0.000771389109565
0.001053034913256
0.001152674107609
0.001053034913256
0.000771389109565
0.000357101778493
0.000350036453042
0.000737079092838
0.000956004198305
0.000956004198305
0.000737079092838
0.000350036453042
0.000293619012570
0.000593155468982
0.000705013252414
0.000593155468982
0.000293619012570
0.000198459095465
0.000372063633308
0.000372063633308
0.000198459095465
0.000091004790589
0.000145607664942
0.000091004790589
0.000014131170069
0.000014131170069
0.000102967932626
0.000230814251757
0.000338359306367
0.000415858382636
0.000456410753006
Appendix A

p
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.160295215850488
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
q
0.013367999605811
0.013367999605811
0.013367999605811
0.013367999605811
0.013367999605811
0.068842731814761
0.068842731814761
0.068842731814761
0.068842731814761
0.068842731814761
0.068842731814761
0.068842731814761
0.068842731814761
0.068842731814761
0.162326928825093
0.162326928825093
0.162326928825093
0.162326928825093
0.162326928825093
0.162326928825093
0.162326928825093
0.162326928825093
0.283713816620238
0.283713816620238
0.283713816620238
0.283713816620238
0.283713816620238
0.283713816620238
0.283713816620238
0.419852392074756
0.419852392074756
0.419852392074756
0.419852392074756
0.419852392074756
0.419852392074756
0.555990967529274
0.555990967529274
0.555990967529274
0.555990967529274
0.555990967529274
0.677377855324419
0.677377855324419
0.677377855324419
0.677377855324419
0.770862052334751
0.770862052334751
0.770862052334751
0.826336784543702
0.826336784543702
0.014230084199161
0.014230084199161
0.014230084199161
0.014230084199161
0.014230084199161
0.014230084199161
0.014230084199161
0.014230084199161
0.014230084199161
0.072864333686846
0.072864333686846
0.072864333686846
0.072864333686846
0.072864333686846
0.072864333686846
0.072864333686846
0.072864333686846
0.170024914572384
0.170024914572384
0.170024914572384
0.170024914572384
r
0.474678563559426
0.592233670482257
0.693878951300071
0.770585232700013
0.815555786882350
0.012272031559499
0.063198698562623
0.149018645439834
0.260453696446561
0.385431026167376
0.510408355888190
0.621843406894918
0.707663353772128
0.758590020775252
0.013449385920162
0.068866812722558
0.160696919295911
0.276561645299257
0.400816210025162
0.516680936028508
0.608511042601861
0.663928469404257
0.014147770528067
0.071853163097369
0.165172364573945
0.277995483764637
0.390818602955329
0.484137804431906
0.541843197001207
0.014176417999888
0.071121024752307
0.159833778001397
0.260018614073359
0.348731367322450
0.405675974074868
0.013309036992320
0.065471316758644
0.141856908310119
0.218242499861594
0.270404779627918
0.011270658032131
0.053569425080607
0.108757503744486
0.151056270792962
0.007758690524781
0.034421365907380
0.061084041289980
0.002824990717434
0.010543008888376
0.011183200273643
0.057591418310226
0.135797023368205
0.237345042281768
0.351233806432731
0.465122570583695
0.566670589497258
0.644876194555237
0.691284412591820
0.012783357625703
0.065456452867118
0.152739032168659
0.262866010269902
0.380967353107876
0.491094331209119
0.578376910510660
0.631050005752075
0.013910659583212
0.070648932977925
0.162404142163887
0.273336391246120
275
L
0.000456410753006
0.000415858382636
0.000338359306367
0.000230814251757
0.000102967932626
0.000260265082353
0.000578489846161
0.000834553981011
0.001000229461794
0.001057525927395
0.001000229461794
0.000834553981011
0.000578489846161
0.000260265082353
0.000410939770525
0.000902761361728
0.001273499961506
0.001472323874040
0.001472323874040
0.001273499961506
0.000902761361728
0.000410939770525
0.000517102354726
0.001117012428919
0.001524850521750
0.001669133370859
0.001524850521750
0.001117012428919
0.000517102354726
0.000546256852972
0.001150264500013
0.001491912742958
0.001491912742958
0.001150264500013
0.000546256852972
0.000482818900053
0.000975368279414
0.001159304092992
0.000975368279414
0.000482818900053
0.000338401945391
0.000634423214647
0.000634423214647
0.000338401945391
0.000158880691017
0.000254209105627
0.000158880691017
0.000024984583805
0.000024984583805
0.000139415696920
0.000309878545114
0.000447043928551
0.000535790995223
0.000566482882934
0.000535790995223
0.000447043928551
0.000309878545114
0.000139415696920
0.000349625324427
0.000768064462517
0.001083486849256
0.001252645153975
0.001252645153975
0.001083486849256
0.000768064462517
0.000349625324427
0.000535672337832
0.001157126154460
0.001579610373778
0.001729074653693
Appendix A

p
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.283302302935376
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.425562830509184
0.574437169490816
0.574437169490816
0.574437169490816
0.574437169490816
0.574437169490816
0.574437169490816
0.574437169490816
0.574437169490816
0.574437169490816
0.574437169490816
0.574437169490816
0.574437169490816
q
0.170024914572384
0.170024914572384
0.170024914572384
0.292615255613059
0.292615255613059
0.292615255613059
0.292615255613059
0.292615255613059
0.292615255613059
0.424082441451564
0.424082441451564
0.424082441451564
0.424082441451564
0.424082441451564
0.546672782492239
0.546672782492239
0.546672782492239
0.546672782492239
0.643833363377778
0.643833363377778
0.643833363377778
0.702467612865463
0.702467612865463
0.014617153391652
0.014617153391652
0.014617153391652
0.014617153391652
0.014617153391652
0.014617153391652
0.014617153391652
0.014617153391652
0.074237047072965
0.074237047072965
0.074237047072965
0.074237047072965
0.074237047072965
0.074237047072965
0.074237047072965
0.170652314741006
0.170652314741006
0.170652314741006
0.170652314741006
0.170652314741006
0.170652314741006
0.287218584745408
0.287218584745408
0.287218584745408
0.287218584745408
0.287218584745408
0.403784854749810
0.403784854749810
0.403784854749810
0.403784854749810
0.500200122417850
0.500200122417850
0.500200122417850
0.559820016099163
0.559820016099163
0.014369232340683
0.014369232340683
0.014369232340683
0.014369232340683
0.014369232340683
0.014369232340683
0.014369232340683
0.072088346222680
0.072088346222680
0.072088346222680
0.072088346222680
0.072088346222680
r
0.384268640328352
0.476023849514315
0.532762122909028
0.014319246644569
0.071837575264130
0.161444117220108
0.262638324231456
0.352244866187434
0.409763194806995
0.013726604181157
0.067525460398349
0.146307627806530
0.225089795214711
0.278888651431902
0.011805143379214
0.056109833340320
0.113915081232065
0.158219771193171
0.008211931753257
0.036432166843423
0.064652401933588
0.003007170628092
0.011222913571069
0.011115266587425
0.056915087943902
0.132808226959586
0.228564815792052
0.331255200307111
0.427011789139577
0.502904928155262
0.548704749511739
0.012728114238126
0.064643066302190
0.148598164008093
0.250100061208925
0.351601958409758
0.435557056115660
0.487472008179724
0.013633893699332
0.068399259338159
0.153717020678344
0.250067834071466
0.335385595411650
0.390150961050477
0.013473445935047
0.066280095784009
0.143609292372704
0.220938488961399
0.273745138810361
0.011848704929974
0.056316881342594
0.114335433398412
0.158803609811032
0.008366638837962
0.037118523536483
0.065870408235004
0.003088967973098
0.011528185418554
0.010463250321044
0.053140360903866
0.122156335037195
0.205596799084250
0.289037263131306
0.358053237264635
0.400730347847457
0.011935151820329
0.059876918699973
0.134564345272441
0.218910139014064
0.293597565586532
276
L
0.001579610373778
0.001157126154460
0.000535672337832
0.000635662488365
0.001338527819612
0.001736093490551
0.001736093490551
0.001338527819612
0.000635662488365
0.000606552151637
0.001225328437748
0.001456401959262
0.001225328437748
0.000606552151637
0.000447572137223
0.000839091376250
0.000839091376250
0.000447572137223
0.000217154553172
0.000347447285074
0.000217154553172
0.000034748745319
0.000034748745319
0.000155709948310
0.000342066976851
0.000482544225219
0.000557880961558
0.000557880961558
0.000482544225219
0.000342066976851
0.000155709948310
0.000384423220178
0.000830407193042
0.001133601389558
0.001240863862768
0.001133601389558
0.000830407193042
0.000384423220178
0.000560512390214
0.001180282683454
0.001530846840634
0.001530846840634
0.001180282683454
0.000560512390214
0.000603540458013
0.001219244354405
0.001449170534097
0.001219244354405
0.000603540458013
0.000480978848727
0.000901721020056
0.000901721020056
0.000480978848727
0.000244790553608
0.000391664885772
0.000244790553608
0.000040163151675
0.000040163151675
0.000143400918949
0.000309765769426
0.000422865926019
0.000462877913901
0.000422865926019
0.000309765769426
0.000143400918949
0.000343450840604
0.000723211630750
0.000938017862639
0.000938017862639
0.000723211630750
Appendix A

p
0.574437169490816
0.574437169490816
0.574437169490816
0.574437169490816
0.574437169490816
0.574437169490816
0.574437169490816
0.574437169490816
0.574437169490816
0.574437169490816
0.574437169490816
0.574437169490816
0.574437169490816
0.574437169490816
0.574437169490816
0.716697697064624
0.716697697064624
0.716697697064624
0.716697697064624
0.716697697064624
0.716697697064624
0.716697697064624
0.716697697064624
0.716697697064624
0.716697697064624
0.716697697064624
0.716697697064624
0.716697697064624
0.716697697064624
0.716697697064624
0.716697697064624
0.716697697064624
0.716697697064624
0.716697697064624
0.716697697064624
0.839704784149512
0.839704784149512
0.839704784149512
0.839704784149512
0.839704784149512
0.839704784149512
0.839704784149512
0.839704784149512
0.839704784149512
0.839704784149512
0.839704784149512
0.839704784149512
0.839704784149512
0.839704784149512
0.932531683344492
0.932531683344492
0.932531683344492
0.932531683344492
0.932531683344492
0.932531683344492
0.932531683344492
0.932531683344492
0.932531683344492
0.986953264258586
0.986953264258586
0.986953264258586
0.986953264258586
0.986953264258586
q
0.072088346222680
0.162007687132911
0.162007687132911
0.162007687132911
0.162007687132911
0.162007687132911
0.263555143376274
0.263555143376274
0.263555143376274
0.263555143376274
0.353474484286505
0.353474484286505
0.353474484286505
0.411193598168501
0.411193598168501
0.013289732853664
0.013289732853664
0.013289732853664
0.013289732853664
0.013289732853664
0.013289732853664
0.065376353661207
0.065376353661207
0.065376353661207
0.065376353661207
0.065376353661207
0.141651151467688
0.141651151467688
0.141651151467688
0.141651151467688
0.217925949274170
0.217925949274170
0.217925949274170
0.270012570081712
0.270012570081712
0.011129592453413
0.011129592453413
0.011129592453413
0.011129592453413
0.011129592453413
0.052898940541990
0.052898940541990
0.052898940541990
0.052898940541990
0.107396275308498
0.107396275308498
0.107396275308498
0.149165623397075
0.149165623397075
0.007603791647411
0.007603791647411
0.007603791647411
0.007603791647411
0.033734158327754
0.033734158327754
0.033734158327754
0.059864525008097
0.059864525008097
0.002757099674531
0.002757099674531
0.002757099674531
0.010289636066883
0.010289636066883
r
0.341539332466176
0.012363392077610
0.060819393573824
0.131777571688137
0.202735749802450
0.251191751298664
0.011248492492696
0.053464072296347
0.108543614836563
0.150759194640214
0.008124476673733
0.036044173111340
0.063963869548947
0.003036576090371
0.011332656250313
0.009117040014437
0.045738862139902
0.102791195188291
0.167221374893421
0.224273707941810
0.260895530067275
0.010222923067433
0.050289756857191
0.108962974637085
0.167636192416979
0.207703026206737
0.009835100679876
0.046746222583354
0.094904928884335
0.131816050787812
0.007368023934041
0.032688176830603
0.058008329727165
0.002808451006571
0.010481281847093
0.006997370883884
0.034422256537484
0.074582811698537
0.114743366859591
0.142168252513190
0.007456721455199
0.035441788775994
0.071954486532504
0.099939553853299
0.005961798695881
0.026449470270995
0.046937141846109
0.002351959627232
0.008777632826181
0.004156504373647
0.019755860661066
0.040108664347031
0.055708020634450
0.003801895823705
0.016867079163877
0.029932262504048
0.001606870246458
0.005996921400953
0.001159659120884
0.005144818033441
0.009129976945999
0.000582643717629
0.002174455956902
277
L
0.000343450840604
0.000459323439224
0.000927903842541
0.001102888770595
0.000927903842541
0.000459323439224
0.000414540002858
0.000777163975548
0.000777163975548
0.000414540002858
0.000227132676091
0.000363412281746
0.000227132676091
0.000038700788121
0.000038700788121
0.000104510657004
0.000220070280072
0.000285434919692
0.000285434919692
0.000220070280072
0.000104510657004
0.000235648564158
0.000476046266093
0.000565819492374
0.000476046266093
0.000235648564158
0.000267293510290
0.000501111800218
0.000501111800218
0.000267293510290
0.000165763800754
0.000265222081206
0.000165763800754
0.000030024334684
0.000030024334684
0.000053966849766
0.000109021319164
0.000129580656044
0.000109021319164
0.000053966849766
0.000106949181923
0.000200504295923
0.000200504295923
0.000106949181923
0.000084132667633
0.000134612268213
0.000084132667633
0.000016995080924
0.000016995080924
0.000014581587902
0.000027337011495
0.000027337011495
0.000014581587902
0.000020996870944
0.000033594993510
0.000020996870944
0.000005324360227
0.000005324360227
0.000000621553620
0.000000994485792
0.000000621553620
0.000000299780623
0.000000299780623
Appendix A
Table 8: Computed Gauss points (u, v, w) and corresponding weights G for M =

10 method GQTSM for tetrahedron.
u
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.9739065285171717D+00
v
-0.9785127030899614
-0.9785127030899614
-0.9785127030899614
-0.9785127030899614
-0.9785127030899614
-0.9785127030899614
-0.9785127030899614
-0.9785127030899614
-0.9785127030899614
-0.9785127030899614
-0.9785127030899614
-0.9785127030899614
-0.8885360641842432
-0.8885360641842432
-0.8885360641842432
-0.8885360641842432
-0.8885360641842432
-0.8885360641842432
-0.8885360641842432
-0.8885360641842432
-0.8885360641842432
-0.8885360641842432
-0.8885360641842432
-0.7336726410476753
-0.7336726410476753
-0.7336726410476753
-0.7336726410476753
-0.7336726410476753
-0.7336726410476753
-0.7336726410476753
-0.7336726410476753
-0.7336726410476753
-0.7336726410476753
-0.5253703549260738
-0.5253703549260738
-0.5253703549260738
-0.5253703549260738
-0.5253703549260738
-0.5253703549260738
-0.5253703549260738
-0.5253703549260738
-0.5253703549260738
-0.2790732333671420
-0.2790732333671420
-0.2790732333671420
-0.2790732333671420
-0.2790732333671420
-0.2790732333671420
-0.2790732333671420
-0.2790732333671420
-0.0130467357414141
-0.0130467357414141
-0.0130467357414141
-0.0130467357414141
-0.0130467357414141
-0.0130467357414141
-0.0130467357414141
0.2529797618843137
0.2529797618843137
0.2529797618843137
0.2529797618843137
0.2529797618843137
0.2529797618843137
0.4992768834432453
0.4992768834432453
0.4992768834432453
0.4992768834432453
0.4992768834432453
0.7075791695648470
0.7075791695648470
0.7075791695648470
0.7075791695648470
0.8624425927014152
0.8624425927014152
0.8624425927014152
278
w
-0.9819993138423283
-0.9063983436792293
-0.7753767779777941
-0.5971358187051090
-0.3828710305138969
-0.1460444418051857
0.0984636734123185
0.3352902621210299
0.5495550503122417
0.7277960095849270
0.8588175752863625
0.9344185454494611
-0.9797260628154768
-0.8948302877495671
-0.7487118008130278
-0.5521720740198941
-0.3197830307576960
-0.0687787036492924
0.1822256234591111
0.4146146667213093
0.6111543935144430
0.7572728804509823
0.8421686555168919
-0.9777216658171440
-0.8847925078734500
-0.7262832284328062
-0.5162388888178013
-0.2733177752814612
-0.0191030551536918
0.2238180583826483
0.4338623979976532
0.5923716774382970
0.6853008353819912
-0.9761316915597071
-0.8770826148055193
-0.7101683632846072
-0.4934343893217102
-0.2503615582783774
-0.0072887272350445
0.2094452467278523
0.3763594982487646
0.4754085750029524
-0.9751219969249455
-0.8726136056433137
-0.7027508559775688
-0.4884300663956099
-0.2585901717200765
-0.0442693821381174
0.1255933675276275
0.2281017588092593
-0.9748859439808757
-0.8724516799591379
-0.7067984663384281
-0.5065233678707071
-0.3062482694029861
-0.1405950557822763
-0.0381607917605384
-0.9756577326126661
-0.8778783892097810
-0.7255500958233327
-0.5535231375438092
-0.4011948441573610
-0.3034155007544759
-0.9777350867885435
-0.8904719262323680
-0.7626851774630369
-0.6348984286937057
-0.5476352681375303
-0.9815084003062328
-0.9121094472397426
-0.8215631938079326
-0.7521642407414426
-0.9874378288038375
-0.9442680320921216
-0.9010982353804058
G
0.0000210869584592
0.0000478009336582
0.0000715535981885
0.0000908140442808
0.0001043691003742
0.0001113665276756
0.0001113665276756
0.0001043691003742
0.0000908140442808
0.0000715535981885
0.0000478009336582
0.0000210869584592
0.0000535464288036
0.0001207931272715
0.0001791886522862
0.0002243041982222
0.0002527861872664
0.0002625209249121
0.0002527861872664
0.0002243041982222
0.0001791886522862
0.0001207931272715
0.0000535464288036
0.0000872219740904
0.0001955179071091
0.0002866170651435
0.0003522649056890
0.0003866159673009
0.0003866159673009
0.0003522649056890
0.0002866170651435
0.0001955179071091
0.0000872219740904
0.0001168606521007
0.0002597455641951
0.0003747199645547
0.0004491092931053
0.0004748357650267
0.0004491092931053
0.0003747199645547
0.0002597455641951
0.0001168606521007
0.0001370868198001
0.0003011552859919
0.0004248312581563
0.0004911576150206
0.0004911576150206
0.0004248312581563
0.0003011552859919
0.0001370868198001
0.0001434416293063
0.0003098537094155
0.0004229859742259
0.0004630093211860
0.0004229859742259
0.0003098537094155
0.0001434416293063
0.0001334932455900
0.0002810995240760
0.0003645908936630
0.0003645908936630
0.0002810995240760
0.0001334932455900
0.0001078454820094
0.0002178644253955
0.0002589494916014
0.0002178644253955
0.0001078454820094
0.0000709773353307
0.0001330656335193
0.0001330656335193
0.0000709773353307
0.0000319815451597
0.0000511704722556
0.0000319815451597
Appendix A
Table 9: Continuation of Table 8.

u
-0.9739065285171717D+00
-0.9739065285171717D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.8650633666889845D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
v
0.9524192316071329
0.9524192316071329
-0.9756670111138167
-0.9756670111138167
-0.9756670111138167
-0.9756670111138167
-0.9756670111138167
-0.9756670111138167
-0.9756670111138167
-0.9756670111138167
-0.9756670111138167
-0.9756670111138167
-0.9756670111138167
-0.8741673141936402
-0.8741673141936402
-0.8741673141936402
-0.8741673141936402
-0.8741673141936402
-0.8741673141936402
-0.8741673141936402
-0.8741673141936402
-0.8741673141936402
-0.8741673141936402
-0.7010392650617517
-0.7010392650617517
-0.7010392650617517
-0.7010392650617517
-0.7010392650617517
-0.7010392650617517
-0.7010392650617517
-0.7010392650617517
-0.7010392650617517
-0.4716232530966042
-0.4716232530966042
-0.4716232530966042
-0.4716232530966042
-0.4716232530966042
-0.4716232530966042
-0.4716232530966042
-0.4716232530966042
-0.2062983545928467
-0.2062983545928467
-0.2062983545928467
-0.2062983545928467
-0.2062983545928467
-0.2062983545928467
-0.2062983545928467
0.0713617212818314
0.0713617212818314
0.0713617212818314
0.0713617212818314
0.0713617212818314
0.0713617212818314
0.3366866197855889
0.3366866197855889
0.3366866197855889
0.3366866197855889
0.3366866197855889
0.5661026317507365
0.5661026317507365
0.5661026317507365
0.5661026317507365
0.7392306808826250
0.7392306808826250
0.7392306808826250
0.8407303778028012
0.8407303778028012
-0.9732640007883790
-0.9732640007883790
-0.9732640007883790
-0.9732640007883790
-0.9732640007883790
-0.9732640007883790
-0.9732640007883790
-0.9732640007883790
279
w
-0.9954591998725648
-0.9830535032173966
-0.9799624148419588
-0.8960563483015299
-0.7516412996354959
-0.5573928181140126
-0.3277129487437590
-0.0796348110985994
0.1684433265465601
0.3981231959168137
0.5923716774382970
0.7367867261043313
0.8206927926447598
-0.9773087169131647
-0.8826570336852367
-0.7212096425941288
-0.5072719427700896
-0.2598480685351741
-0.0009212505822009
0.2465026236527146
0.4604403234767538
0.6218877145678618
0.7165393977957897
-0.9750678336492901
-0.8716039428294952
-0.6972499916211889
-0.4708557539980772
-0.2169486841246318
0.0369583857488138
0.2633526233719254
0.4377065745802315
0.5411704654000264
-0.9734599912552455
-0.8641034005024621
-0.6828927604066217
-0.4542540062217496
-0.2090593739926616
0.0195793801922106
0.2007900202880510
0.3101466110408346
-0.9727380826839559
-0.8615432030530465
-0.6817226193358712
-0.4643191393590843
-0.2469156593822975
-0.0670950756651221
0.0440998039657872
-0.9732004711539487
-0.8655506662948876
-0.6978453976063979
-0.5084529569864489
-0.3407476882979591
-0.2330978834388980
-0.9752138061015480
-0.8780689577387805
-0.7358116265483021
-0.5935542953578239
-0.4964094469950564
-0.9792426048289611
-0.9013401238584765
-0.7996991412032752
-0.7217966602327908
-0.9858184467504782
-0.9370836570968202
-0.8883488674431620
-0.9948578343987214
-0.9808091767150953
-0.9784380046772969
-0.8884968963126230
-0.7350843335127386
-0.5317937718771106
-0.2966835580314484
-0.0506428728811481
0.1844673409645141
0.3877579026001421
G
0.0000049193491194
0.0000049193491194
0.0000595090322757
0.0001342439126211
0.0001991420068639
0.0002492813446167
0.0002809349141111
0.0002917536527217
0.0002809349141111
0.0002492813446167
0.0001991420068639
0.0001342439126211
0.0000595090322757
0.0001509522856561
0.0003383765991611
0.0004960390032748
0.0006096536248433
0.0006691039103832
0.0006691039103832
0.0006096536248433
0.0004960390032748
0.0003383765991611
0.0001509522856561
0.0002429029154634
0.0005398990480330
0.0007788812593161
0.0009335046030977
0.0009869788471822
0.0009335046030977
0.0007788812593161
0.0005398990480330
0.0002429029154634
0.0003173647520954
0.0006971937405828
0.0009835115230168
0.0011370612795458
0.0011370612795458
0.0009835115230168
0.0006971937405828
0.0003173647520954
0.0003571017784935
0.0007713891095648
0.0010530349132563
0.0011526741076089
0.0010530349132563
0.0007713891095648
0.0003571017784935
0.0003500364530424
0.0007370790928380
0.0009560041983049
0.0009560041983049
0.0007370790928380
0.0003500364530424
0.0002936190125703
0.0005931554689816
0.0007050132524139
0.0005931554689816
0.0002936190125703
0.0001984590954645
0.0003720636333079
0.0003720636333079
0.0001984590954645
0.0000910047905887
0.0001456076649420
0.0000910047905887
0.0000141311700691
0.0000141311700691
0.0001029679326262
0.0002308142517569
0.0003383593063674
0.0004158583826363
0.0004564107530060
0.0004564107530060
0.0004158583826363
0.0003383593063674
Appendix A

u
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.6794095682990244D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
v
-0.9732640007883790
-0.9732640007883790
-0.8623145363704783
-0.8623145363704783
-0.8623145363704783
-0.8623145363704783
-0.8623145363704783
-0.8623145363704783
-0.8623145363704783
-0.8623145363704783
-0.8623145363704783
-0.6753461423498142
-0.6753461423498142
-0.6753461423498142
-0.6753461423498142
-0.6753461423498142
-0.6753461423498142
-0.6753461423498142
-0.6753461423498142
-0.4325723667595235
-0.4325723667595235
-0.4325723667595235
-0.4325723667595235
-0.4325723667595235
-0.4325723667595235
-0.4325723667595235
-0.1602952158504878
-0.1602952158504878
-0.1602952158504878
-0.1602952158504878
-0.1602952158504878
-0.1602952158504878
0.1119819350585480
0.1119819350585480
0.1119819350585480
0.1119819350585480
0.1119819350585480
0.3547557106488386
0.3547557106488386
0.3547557106488386
0.3547557106488386
0.5417241046695025
0.5417241046695025
0.5417241046695025
0.6526735690874035
0.6526735690874035
-0.9715398316016786
-0.9715398316016786
-0.9715398316016786
-0.9715398316016786
-0.9715398316016786
-0.9715398316016786
-0.9715398316016786
-0.9715398316016786
-0.9715398316016786
-0.8542713326263087
-0.8542713326263087
-0.8542713326263087
-0.8542713326263087
-0.8542713326263087
-0.8542713326263087
-0.8542713326263087
-0.8542713326263087
-0.6599501708552311
-0.6599501708552311
-0.6599501708552311
-0.6599501708552311
-0.6599501708552311
-0.6599501708552311
-0.6599501708552311
-0.4147694887738811
-0.4147694887738811
-0.4147694887738811
-0.4147694887738811
-0.4147694887738811
280
w
0.5411704654000264
0.6311115737647004
-0.9754559368810017
-0.8736026028747540
-0.7019627091203327
-0.4790926071068780
-0.2291379476652488
0.0208167117763804
0.2436868137898351
0.4153267075442564
0.5171800415505041
-0.9731012281596763
-0.8622663745548832
-0.6786061614081772
-0.4468767094014855
-0.1983675799496759
0.0333618720570159
0.2170220852037219
0.3278569388085149
-0.9717044589438657
-0.8562936738052630
-0.6696552708521097
-0.4440090324707260
-0.2183627940893423
-0.0317243911361890
0.0836863940024137
-0.9716471640002230
-0.8577579504953869
-0.6803324439972055
-0.4799627718532823
-0.3025372653551008
-0.1886480518502648
-0.9733819260153596
-0.8690573664827117
-0.7162861833797618
-0.5635150002768118
-0.4591904407441639
-0.9774586839357379
-0.8928611498387868
-0.7824849925110273
-0.6978874584140763
-0.9844826189504376
-0.9311572681852391
-0.8778319174200406
-0.9943500185651311
-0.9789139822232480
-0.9776335994527140
-0.8848171633795486
-0.7284059532635898
-0.5253099154364650
-0.2975323871345371
-0.0697548588326091
0.1333411789945156
0.2897523891104743
0.3825688251836399
-0.9744332847485946
-0.8690870942657640
-0.6945219356626813
-0.4742679794601965
-0.2380652937842477
-0.0178113375817628
0.1567538210213200
0.2621000115041505
-0.9721786808335768
-0.8587021340441509
-0.6751917156722266
-0.4533272175077608
-0.2314627193432952
-0.0479523009713708
0.0655242458180552
-0.9713615067108625
-0.8563248494717401
-0.6771117655597831
-0.4747233515370886
-0.2955102676251314
G
0.0002308142517569
0.0001029679326262
0.0002602650823529
0.0005784898461613
0.0008345539810105
0.0010002294617935
0.0010575259273953
0.0010002294617935
0.0008345539810105
0.0005784898461613
0.0002602650823529
0.0004109397705245
0.0009027613617285
0.0012734999615058
0.0014723238740403
0.0014723238740403
0.0012734999615058
0.0009027613617285
0.0004109397705245
0.0005171023547262
0.0011170124289186
0.0015248505217504
0.0016691333708590
0.0015248505217504
0.0011170124289186
0.0005171023547262
0.0005462568529719
0.0011502645000127
0.0014919127429584
0.0014919127429584
0.0011502645000127
0.0005462568529719
0.0004828189000526
0.0009753682794138
0.0011593040929919
0.0009753682794138
0.0004828189000526
0.0003384019453905
0.0006344232146466
0.0006344232146466
0.0003384019453905
0.0001588806910169
0.0002542091056270
0.0001588806910169
0.0000249845838053
0.0000249845838053
0.0001394156969197
0.0003098785451142
0.0004470439285509
0.0005357909952225
0.0005664828829345
0.0005357909952225
0.0004470439285509
0.0003098785451142
0.0001394156969197
0.0003496253244272
0.0007680644625167
0.0010834868492557
0.0012526451539752
0.0012526451539752
0.0010834868492557
0.0007680644625167
0.0003496253244272
0.0005356723378323
0.0011571261544599
0.0015796103737784
0.0017290746536926
0.0015796103737784
0.0011571261544599
0.0005356723378323
0.0006356624883655
0.0013385278196120
0.0017360934905505
0.0017360934905505
0.0013385278196120
Appendix A

u
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.4333953941292472D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
-0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
0.1488743389816312D+00
v
-0.4147694887738811
-0.1518351170968717
-0.1518351170968717
-0.1518351170968717
-0.1518351170968717
-0.1518351170968717
0.0933455649844783
0.0933455649844783
0.0933455649844783
0.0933455649844783
0.2876667267555559
0.2876667267555559
0.2876667267555559
0.4049352257309258
0.4049352257309258
-0.9707656932166957
-0.9707656932166957
-0.9707656932166957
-0.9707656932166957
-0.9707656932166957
-0.9707656932166957
-0.9707656932166957
-0.9707656932166957
-0.8515259058540692
-0.8515259058540692
-0.8515259058540692
-0.8515259058540692
-0.8515259058540692
-0.8515259058540692
-0.8515259058540692
-0.6586953705179881
-0.6586953705179881
-0.6586953705179881
-0.6586953705179881
-0.6586953705179881
-0.6586953705179881
-0.4255628305091844
-0.4255628305091844
-0.4255628305091844
-0.4255628305091844
-0.4255628305091844
-0.1924302905003809
-0.1924302905003809
-0.1924302905003809
-0.1924302905003809
0.0004002448357003
0.0004002448357003
0.0004002448357003
0.1196400321983269
0.1196400321983269
-0.9712615353186331
-0.9712615353186331
-0.9712615353186331
-0.9712615353186331
-0.9712615353186331
-0.9712615353186331
-0.9712615353186331
-0.8558233075546404
-0.8558233075546404
-0.8558233075546404
-0.8558233075546404
-0.8558233075546404
-0.8558233075546404
-0.6759846257341786
-0.6759846257341786
-0.6759846257341786
-0.6759846257341786
-0.6759846257341786
-0.4728897132474524
-0.4728897132474524
-0.4728897132474524
-0.4728897132474524
-0.2930510314269907
-0.2930510314269907
-0.2930510314269907
281
w
-0.1804736103860092
-0.9725467916376855
-0.8649490792033028
-0.7073847443869405
-0.5498204095705783
-0.4422226971361956
-0.9763897132415725
-0.8877803333193609
-0.7721698375358703
-0.6835604576136586
-0.9835761364934850
-0.9271356663131544
-0.8706951961328238
-0.9939856587438157
-0.9775541728578627
-0.9777694668251506
-0.8861698241121969
-0.7343835460808280
-0.5428703684158955
-0.3374895993857777
-0.1459764217208451
0.0058098563105240
0.0974094990234775
-0.9745437715237478
-0.8707138673956205
-0.7028036719838149
-0.4997998775821498
-0.2967960831804849
-0.1288858877686792
-0.0250559836405518
-0.9727322126013357
-0.8632014813236818
-0.6925659586433119
-0.4998643318570689
-0.3292288091766991
-0.2196980778990452
-0.9730531081299070
-0.8674398084319827
-0.7127814152545922
-0.5581230220772018
-0.4525097223792776
-0.9763025901400513
-0.8873662373148126
-0.7713291332031754
-0.6823927803779368
-0.9832667223240769
-0.9257629529270346
-0.8682591835299924
-0.9938220640538042
-0.9769436291628915
-0.9790734993579121
-0.8937192781922685
-0.7556873299256109
-0.5888064018314990
-0.4219254737373872
-0.2838935254707294
-0.1985393043050859
-0.9761296963593420
-0.8802461626000544
-0.7308713094551189
-0.5621797219718718
-0.4128048688269363
-0.3169213350676487
-0.9752732158447804
-0.8783612128523528
-0.7364448566237263
-0.5945285003950997
-0.4976164974026720
-0.9775030150146072
-0.8930718554073052
-0.7829127703268735
-0.6984816107195715
-0.9837510466525349
-0.9279116537773202
-0.8720722609021055
G
0.0006356624883655
0.0006065521516366
0.0012253284377478
0.0014564019592622
0.0012253284377478
0.0006065521516366
0.0004475721372230
0.0008390913762495
0.0008390913762495
0.0004475721372230
0.0002171545531715
0.0003474472850744
0.0002171545531715
0.0000347487453190
0.0000347487453190
0.0001557099483103
0.0003420669768513
0.0004825442252185
0.0005578809615585
0.0005578809615585
0.0004825442252185
0.0003420669768513
0.0001557099483103
0.0003844232201780
0.0008304071930423
0.0011336013895582
0.0012408638627684
0.0011336013895582
0.0008304071930423
0.0003844232201780
0.0005605123902143
0.0011802826834540
0.0015308468406342
0.0015308468406342
0.0011802826834540
0.0005605123902143
0.0006035404580134
0.0012192443544051
0.0014491705340967
0.0012192443544051
0.0006035404580134
0.0004809788487273
0.0009017210200564
0.0009017210200564
0.0004809788487273
0.0002447905536076
0.0003916648857722
0.0002447905536076
0.0000401631516755
0.0000401631516755
0.0001434009189485
0.0003097657694262
0.0004228659260195
0.0004628779139008
0.0004228659260195
0.0003097657694262
0.0001434009189485
0.0003434508406038
0.0007232116307498
0.0009380178626390
0.0009380178626390
0.0007232116307498
0.0003434508406038
0.0004593234392237
0.0009279038425407
0.0011028887705954
0.0009279038425407
0.0004593234392237
0.0004145400028581
0.0007771639755478
0.0007771639755478
0.0004145400028581
0.0002271326760911
0.0003634122817458
0.0002271326760911
Appendix A

u
0.1488743389816312D+00
0.1488743389816312D+00
0.4333953941292472D+00
0.4333953941292472D+00
0.4333953941292472D+00
0.4333953941292472D+00
0.4333953941292472D+00
0.4333953941292472D+00
0.4333953941292472D+00
0.4333953941292472D+00
0.4333953941292472D+00
0.4333953941292472D+00
0.4333953941292472D+00
0.4333953941292472D+00
0.4333953941292472D+00
0.4333953941292472D+00
0.4333953941292472D+00
0.4333953941292472D+00
0.4333953941292472D+00
0.4333953941292472D+00
0.4333953941292472D+00
0.4333953941292472D+00
0.6794095682990244D+00
0.6794095682990244D+00
0.6794095682990244D+00
0.6794095682990244D+00
0.6794095682990244D+00
0.6794095682990244D+00
0.6794095682990244D+00
0.6794095682990244D+00
0.6794095682990244D+00
0.6794095682990244D+00
0.6794095682990244D+00
0.6794095682990244D+00
0.6794095682990244D+00
0.6794095682990244D+00
0.8650633666889845D+00
0.8650633666889845D+00
0.8650633666889845D+00
0.8650633666889845D+00
0.8650633666889845D+00
0.8650633666889845D+00
0.8650633666889845D+00
0.8650633666889845D+00
0.8650633666889845D+00
0.9739065285171717D+00
0.9739065285171717D+00
0.9739065285171717D+00
0.9739065285171717D+00
0.9739065285171717D+00
v
-0.1776128036629980
-0.1776128036629980
-0.9734205342926717
-0.9734205342926717
-0.9734205342926717
-0.9734205342926717
-0.9734205342926717
-0.9734205342926717
-0.8692472926775869
-0.8692472926775869
-0.8692472926775869
-0.8692472926775869
-0.8692472926775869
-0.7166976970646236
-0.7166976970646236
-0.7166976970646236
-0.7166976970646236
-0.5641481014516603
-0.5641481014516603
-0.5641481014516603
-0.4599748598365756
-0.4599748598365756
-0.9777408150931738
-0.9777408150931738
-0.9777408150931738
-0.9777408150931738
-0.9777408150931738
-0.8942021189160209
-0.8942021189160209
-0.8942021189160209
-0.8942021189160209
-0.7852074493830036
-0.7852074493830036
-0.7852074493830036
-0.7016687532058506
-0.7016687532058506
-0.9847924167051783
-0.9847924167051783
-0.9847924167051783
-0.9847924167051783
-0.9325316833444923
-0.9325316833444923
-0.9325316833444923
-0.8802709499838063
-0.8802709499838063
-0.9944858006509373
-0.9944858006509373
-0.9944858006509373
-0.9794207278662345
-0.9794207278662345
282
w
-0.9939268478192584
-0.9773346874993747
-0.9817659199711265
-0.9085222757201959
-0.7944176096234181
-0.6655572502131575
-0.5514525841163795
-0.4782089398654491
-0.9795541538651344
-0.8994204862856184
-0.7820740507258301
-0.6647276151660418
-0.5845939475865258
-0.9803297986402473
-0.9065075548332930
-0.8101901422313307
-0.7363678984243763
-0.9852639521319173
-0.9346236463387935
-0.8839833405456696
-0.9943830979868571
-0.9790374363058146
-0.9860052582322312
-0.9311554869250322
-0.8508343766029254
-0.7705132662808184
-0.7156634949736194
-0.9850865570896014
-0.9291164224480115
-0.8560910269349921
-0.8001208922934022
-0.9880764026082388
-0.9471010594580104
-0.9061257163077820
-0.9952960807455358
-0.9824447343476380
-0.9916869912527055
-0.9604882786778677
-0.9197826713059386
-0.8885839587311007
-0.9923962083525891
-0.9662658416722462
-0.9401354749919031
-0.9967862595070837
-0.9880061571980946
-0.9976806817582317
-0.9897103639331173
-0.9817400461080028
-0.9988347125647420
-0.9956510880861953
G
0.0000387007881208
0.0000387007881208
0.0001045106570038
0.0002200702800715
0.0002854349196915
0.0002854349196915
0.0002200702800715
0.0001045106570038
0.0002356485641582
0.0004760462660934
0.0005658194923739
0.0004760462660934
0.0002356485641582
0.0002672935102895
0.0005011118002183
0.0005011118002183
0.0002672935102895
0.0001657638007540
0.0002652220812064
0.0001657638007540
0.0000300243346837
0.0000300243346837
0.0000539668497656
0.0001090213191645
0.0001295806560438
0.0001090213191645
0.0000539668497656
0.0001069491819233
0.0002005042959233
0.0002005042959233
0.0001069491819233
0.0000841326676332
0.0001346122682131
0.0000841326676332
0.0000169950809241
0.0000169950809241
0.0000145815879016
0.0000273370114952
0.0000273370114952
0.0000145815879016
0.0000209968709439
0.0000335949935103
0.0000209968709439
0.0000053243602271
0.0000053243602271
0.0000006215536202
0.0000009944857924
0.0000006215536202
0.0000002997806231
0.0000002997806231
Appendix B
Computer Codes for The Formulae of The Thesis
This section includes the complete computer codes for the derived formulae of
the thesis using FORTRAN for readily use and reference. Thus, it enhance the
use of the derive formulae of the thesis in the solution of practical problems in
real world.
283
C (PROGRAM-1) IOUT.FOR
C COMPLETE PROGRAM TO DETERMINE THE INTEGRAL OVER THE UNIT
C TRIANGULAR, STANDARD OR THE ORIGINAL TRIANGULAR SURFACES
C USING THE NEW TECHNIQUE. THAT IS THE TRIANGLE IN (X,Y) SPACE IS
C EITHER IN ORIGINAL SHAPE OR TRANSFORMED INTO STANDARD OR
C UNIT TRIANGLE WHICH IS AGAIN SUB-DIVIDED INTO THREE
QUADRILATERALS.
==============================================================
PARAMETER (NP=20)
CHARACTER*8,OUTFILE, OUTF1
INTEGER TYPE
DOUBLE PRECISION S(2:NP,NP),W(2:NP,NP),INT(2:NP,2:NP),F1,F2,F3
1 ,X1,Y1,X2,Y2,X3,Y3,AREA,U(3),V(3),X,Y,A11,A12,A13,A14,
2 B11,B12,B13,B14,A21,A22,A23,A24,B21,B22,B23,B24,A31,A32,
3 A33,A34,B31,B32,B33,B34
C HERE THE INTEGRAND SHOULD BE GIVEN AS STATEMENT FUNCTION
F(U,V)=(U+V)**0.5D0
PRINT*,'TYPE THE FILE NAME FOR OUTPUT OF THE INTEGRAL'
READ*,OUTFILE
OPEN( UNIT =2, FILE= OUTFILE)
PRINT*,'TYPE THE FILE NAME FOR OUTPUT OF THE GAUSSPOINTS AND WTS'
READ*,OUTF1
OPEN( UNIT =3, FILE= OUTF1)
C
GAUSS OR SAMPLE POINTS AND CORRESPONDING WEIGHT POINTS
CALL GQPW(NP,S,W)
C
DO 2 JJ=1,3
WRITE(*,3) JJ
3 FORMAT('TPYE X AND Y COORDINATES OF NODE:-',I2)
READ*,U(JJ),V(JJ)
2
CONTINUE
AREA=(U(1)-U(3))*(V(2)-V(3))-(U(2)-U(3))*(V(1)-V(3))
C
A11=14.D0*U(1)+5.D0*U(2)+5.D0*U(3)
A12=4.D0*U(1)+U(2)-5.D0*U(3)
A13=-4.D0*U(1)+5.D0*U(2)-U(3)
A14=-2.D0*U(1)+U(2)+U(3)
B11=14.D0*V(1)+5.D0*V(2)+5.D0*V(3)
B12=4.D0*V(1)+V(2)-5.D0*V(3)
B13=-4.D0*V(1)+5.D0*V(2)-V(3)
B14=-2.D0*V(1)+V(2)+V(3)
A21=5.D0*U(1)+14.D0*U(2)+5.D0*U(3)
A22=U(1)+4.D0*U(2)-5.D0*U(3)
A23=-5.D0*U(1)+4.D0*U(2)+U(3)
284
A24=-U(1)+2.D0*U(2)-U(3)
B21=5.D0*V(1)+14.D0*V(2)+5.D0*V(3)
B22=V(1)+4.D0*V(2)-5.D0*V(3)
B23=-5.D0*V(1)+4.D0*V(2)+V(3)
B24=-V(1)+2.D0*V(2)-V(3)
A31=5.D0*U(1)+5.D0*U(2)+14.D0*U(3)
A32=5.D0*U(1)-U(2)-4.D0*U(3)
A33=-U(1)+5.D0*U(2)-4.D0*U(3)
A34=-U(1)-U(2)+2.D0*U(3)
B31=5.D0*V(1)+5.D0*V(2)+14.D0*V(3)
B32=5.D0*V(1)-V(2)-4.D0*V(3)
B33=-V(1)+5.D0*V(2)-4.D0*V(3)
B34=-V(1)-V(2)+2.D0*V(3)
C THIS PART FOR THE SUM IN THE GAUSS INTEGRATION FORMULA
DO 7 M=2,NP
N=M
INT(M,N)=0.D0
DO 9 I=1,M
DO 10 J=1,N
X1=(A11+A12*S(M,I)+A13*S(N,J)+A14*S(M,I)*S(N,J))/24.D0
Y1=(B11+B12*S(M,I)+B13*S(N,J)+B14*S(M,I)*S(N,J))/24.D0
F1=F(X1,Y1)*(4.D0+S(M,I)-S(N,J))
F2=F(X2,Y2)*(4.D0+S(M,I)+S(N,J))
F3=F(X3,Y3)*(4.D0-S(M,I)-S(N,J))
INT(M,N)=INT(M,N)+(F1+F2+F3)*W(M,I)*W(N,J)
10 CONTINUE
9
CONTINUE
WRITE(2,11) M,N,INT(M,N)*AREA/96.D0
11 FORMAT(2X,'I(',I2,',',I2,')=',D18.10)
7
CONTINUE
15 PRINT*, 'YOUR DATA FILE NAME ', OUTFILE
STOP
END
C================================****===========================
285
C SUBROUTINE GQPW AND NEWTON FOR CALCULATION OF GAUSS

POINTS AND WEIGHTS
SUBROUTINE GQPW(NP,S,W)
DOUBLE PRECISION S(2:NP,NP),W(2:NP,NP),Z1,Z,XM,XL,PP,P3,P2,P1,
EPS,X1,X2
EPS=0.0000000000000003D0
X1=-1.D0
X2=1.D0
C
DO 70 N=2,NP
M=(N+1)/2
XM=0.5D0*(X2+X1)
XL=0.5D0*(X2-X1)
DO 80 I=1,M
Z=DCOS(3.141592654D0*(I-0.25D0)/(N+0.5D0))
82
CALL NEWTON (N,P1,P2,P3,PP,Z1,Z)
IF(DABS(Z-Z1).GT.EPS) GO TO 82
S(N,I)= XM-XL*Z
S(N,N+1-I)=XM+XL*Z
W(N,I)=(2.D0*XL)/((1.D0-Z*Z)*PP*PP)
W(N,N+1-I)=W(N,I)
80
CONTINUE
70
CONTINUE
RETURN
END
C==============================================================
SUBROUTINE NEWTON (N,P1,P2,P3,PP,Z1,Z)
DOUBLE PRECISION Z1,Z,PP,P3,P2,P1
P1=1.D0
P2=0.D0
DO 90 J= 1,N
P3=P2
P2=P1
P1=((2.D0*J-1.D0)*Z*P2-(J-1.D0)*P3)/J
90 CONTINUE
PP=N*(Z*P1-P2)/(Z*Z-1.D0)
Z1=Z
Z=Z1-P1/PP
RETURN
END
C==============================================================
286
C
PROGRAM-2 GQUTS.FOR
C COMPLETE PROGRAM TO EVALUTE THE INTEGRAL OVER THE UNIT
C TRIANGULAR SURFACES USING THE NEW WEIGHTS AND GAUSSIAN
C POINTS. THAT IS THE TRIANGLE IN (X,Y) SPACE IS TRANSFORMED INTO
C A UNIT TRIANGLE WHICH IS AGAIN TRANSFORMED INTO A 2-SQUARE.
C NP= NUMBER OF GAUSSIAN POINTS, ITP= TOTAL NEW GAUSSIAN POINTS
FOR TRIANGLES.
===============================================================
PARAMETER(NP=11, ITP=NP*NP)
INTEGER R
DOUBLE PRECISION S(2:NP,NP),W(2:NP,NP),XC(3),YC(3),X,Y,
1 G(ITP),U(ITP),V(ITP),U1(ITP),V1(ITP),G1(ITP),XR,YR,XR1,YR1,
2 INT1,INT2,AREA
F(U,V)=1.D0/(12.D0-8.821067231*U+21.53679831D0*V)
OPEN( UNIT =1, FILE= '11.DAT')
C
DO 2 JJ=1,3
WRITE(*,3) JJ
READ*,XC(JJ),YC(JJ)
2 CONTINUE
AREA=DABS((XC(1)-XC(3))*(YC(2)-YC(3))-(XC(2)-XC(3))*(YC(1)-YC(3)))
C
CALL GQPW(NP,S,W)
DO 7 M=2,NP
C FOR MN POINTS N=2,NP; FOR MM POINTS N=M
C DO 7 N=2,NP
N=M
INT1=0.D0
INT2=0.D0
R=1
DO 9 J=1,N
DO 10 I=1,M
G(R)=(1.D0-S(M,I))*W(M,I)*W(N,J)/8.D0
U(R)=(1.D0+S(M,I))/2.D0
V(R)=(1.D0-S(M,I))*(1.D0+S(N,J))/4.D0
XR=XC(3)+(XC(1)-XC(3))*U(R)+(XC(2)-XC(3))*V(R)
YR=YC(3)+(YC(1)-YC(3))*U(R)+(YC(2)-YC(3))*V(R)
INT1=INT1+F(XR,YR)*G(R)
G1(R)=(1.D0-S(N,J))*W(M,I)*W(N,J)/8.D0
U1(R)=(1.D0+S(M,I))*(1.D0-S(N,J))/4.D0
V1(R)=(1.D0+S(N,J))/2.D0
287
10
9
XR1=XC(3)+(XC(1)-XC(3))*U1(R)+(XC(2)-XC(3))*V1(R)
YR1=YC(3)+(YC(1)-YC(3))*U1(R)+(YC(2)-YC(3))*V1(R)
INT2=INT2+F(XR1,YR1)*G1(R)
R=R+1
CONTINUE
CONTINUE
12
7
INT1=INT1*AREA
INT2=INT2*AREA
WRITE(1,12) M,N,INT1,INT2
FORMAT(2X,2I4,4D25.15)
CONTINUE
PRINT*, 'COMPUTATION OVER'

STOP
END
C================================****===========================
C
PROGRAM-3 GQUTM.FOR
C COMPLETE PROGRAM TO EVALUTE THE INTEGRAL OVER THE UNIT
C TRIANGULAR SURFACES USING THE NEW WEIGHTS AND GAUSSIAN
C POINTS. THAT IS THE TRIANGLE IN (X,Y) SPACE IS TRANSFORMED INTO A
C UNIT TRIANGLE WHICH IS AGAIN TRANSFORMED INTO A 2-SQUARE. NP=
C NUMBER OF GAUSSIAN POINTS, TP= TOTAL NEW GAUSSIAN POINTS FOR
C TRIANGLES WHICH IS COMPLETELY CROWDING FREE WITH LESS
C NUMBER OF POINTS.
===============================================================
PARAMETER(NP=11, TP=NP*(NP+1)/2)
INTEGER R
DOUBLE PRECISION S(2:NP,NP),W(2:NP,NP),XC(3),YC(3),X,Y,
1 G(ITP),U(ITP),V(ITP),U1(ITP),V1(ITP),G1(ITP),XR,YR,XR1,YR1,
2 INT1,INT2,AREA
F(U,V)=1.D0/(12.D0-8.821067231*U+21.53679831D0*V)
C
DO 2 JJ=1,3
WRITE(*,3) JJ
READ*,XC(JJ),YC(JJ)
2 CONTINUE
C
CALL GQPW(NP,S,W)
C
DO 17 M=2,NP-1
288
INT1=0.D0
INT2=0.D0
R=1
DO 19 I=1,M
N=M-I+2
DO 20 J=1,N
G(R)=(1.D0-S(M,I))*W(M,I)*W(N,J)/8.D0
U(R)=(1.D0+S(M,I))/2.D0
V(R)=(1.D0-S(M,I))*(1.D0+S(N,J))/4.D0
C
XR=XC(3)+(XC(1)-XC(3))*U(R)+(XC(2)-XC(3))*V(R)
YR=YC(3)+(YC(1)-YC(3))*U(R)+(YC(2)-YC(3))*V(R)
C
G1(R)=(1.D0-S(M,I))*W(N,J)*W(M,I)/8.D0
U1(R)=(1.D0+S(N,J))*(1.D0-S(M,I))/4.D0
V1(R)=(1.D0+S(M,I))/2.D0
C
XR1=XC(3)+(XC(1)-XC(3))*U1(R)+(XC(2)-XC(3))*V1(R)
YR1=YC(3)+(YC(1)-YC(3))*U1(R)+(YC(2)-YC(3))*V1(R)
R=R+1
20
CONTINUE
19
CONTINUE
INT1=INT1*AREA
INT2=INT2*AREA
WRITE(1,11) M+1,-1+(M+2)*(M+1)/2,INT1,INT2
11 FORMAT(2X,2I5,4D25.15)
17 CONTINUE
STOP
END
C================================****===========================
C
PROGRAM-4 GQSTS.FOR
C COMPLETE PROGRAM TO EVALUTE THE INTEGRAL OVER THE
C STANDARD TRIANGULAR SURFACES USING THE NEW WEIGHTS AND
C GAUSSIAN POINTS. THAT IS THE TRIANGLE IN (X,Y) SPACE IS
C TRANSFORMED INTO A STANDARD TRIANGLE WHICH IS AGAIN
C TRANSFORMED INTO A 2-SQUARE. NP= NUMBER OF GAUSSIAN POINTS,
C ITP= TOTAL NEW GAUSSIAN POINTS FOR TRIANGLES.
==============================================================
PARAMETER (NP=15,ITP=NP*NP)
INTEGER R
DOUBLE PRECISION S(2:NP,NP),W(2:NP,NP),AREA,
1 XC(3),YC(3),X,Y,G(ITP),U(ITP),V(ITP),U1(ITP),V1(ITP),
2 G1(ITP), XR,YR,XR1,YR1,INT,INT1
289
C
C
HERE THE INTEGRAND SHOULD BE GIVEN AS STATEMENT FUNCTION

F(X,Y)=1.D0/(12.D0+21.53679831D0*Y-8.821067231*X)
OPEN( UNIT =1, FILE= 'P41.DAT')
3
2
DO 2 JJ=1,3
WRITE(*,3) JJ
FORMAT(' TPYE X AND Y COORDINATES OF NODE:-',I2)
READ*,XC(JJ),YC(JJ)
CONTINUE
AREA=AREA/4.D0
CALL GQPW(NP,S,W)
C
DO 7 M=2,NP
C
DO 7 N=2,NP
N=M
INT=0.D0
INT1=0.D0
R=1
DO 9 I=1,M
DO 10 J=1,N
G(R)=(1.D0-S(M,I))*W(M,I)*W(N,J)/2.D0
U(R)=S(M,I)
V(R)=(1.D0-S(M,I))*(1.D0+S(N,J))/2.D0-1.D0
XR=(XC(1)+XC(2)+(XC(1)-XC(3))*U(R)+(XC(2)-XC(3))*V(R))/2.D0
YR=(YC(1)+YC(2)+(YC(1)-YC(3))*U(R)+(YC(2)-YC(3))*V(R))/2.D0
INT=INT+F(XR,YR)*G(R)
C
G1(R)=(1.D0-S(N,J))*W(M,I)*W(N,J)/2.D0
U1(R)=(1.d0-S(N,J))*(1.d0+S(M,I))/2.D0-1.D0
V1(R)=S(N,J)
XR1=(XC(1)+XC(2)+(XC(1)-XC(3))*U1(R)+(XC(2)-XC(3))*V1(R))/2.D0
YR1=(YC(1)+YC(2)+(YC(1)-YC(3))*U1(R)+(YC(2)-YC(3))*V1(R))/2.D0
R=R+1
10 CONTINUE
9 CONTINUE
INT=INT*AREA
INT1= INT1*AREA
WRITE(1,11) M, N, INT, INT1
7
CONTINUE
STOP
END
C================================****===========================
290
C
PROGRAM-5 GQSTM.FOR
C
COMPLETE PROGRAM TO EVALUTE THE INTEGRAL OVER THE
STANDARD TRIANGULAR SURFACES USING THE NEW WEIGHTS AND
GAUSSIAN POINTS. THAT IS THE TRIANGLE IN (X,Y) SPACE IS
TRANSFORMED INTO A STANDARD TRIANGLE WHICH IS AGAIN
TRANSFORMED INTO A 2-SQUARE. NP= NUMBER OF GAUSSIAN POINTS, TP=
TOTAL NEW GAUSSIAN POINTS FOR TRIANGLES WHICH IS COMPLETELY
CROWDING FREE WITH LESS NUMBER OF POINTS.
===============================================================
PARAMETER (NP=15, TP=NP*(NP+1)/2)
INTEGER R
DOUBLE PRECISION S(2:NP,NP),W(2:NP,NP),AREA,
1 XC(3),YC(3),X,Y,L(ITP),P(ITP),Q(ITP),P1(ITP),Q1(ITP),
2 L1(ITP),XR,YR,XR1,YR1,INT,INT1
C
C
HERE THE INTEGRAND SHOULD BE GIVEN AS STATEMENT FUNCTION
F(X,Y)=1.D0/(12.D0+21.53679831D0*Y-8.821067231*X)
OPEN( UNIT =1, FILE= 'P41.DAT')
C
DO 2 JJ=1,3
WRITE(*,3) JJ
3 FORMAT(' TPYE X AND Y COORDINATES OF NODE:-',I2)
READ*,XC(JJ),YC(JJ)
2 CONTINUE
AREA=AREA/4.D0
CALL GQPW(NP,S,W)
C
DO M=2,NP-1
INT=0.D0
INT1=0.D0
R=1
DO I=1,M
N=M-I+2
DO J=1,N
L(R)=(1.d0-S(M,I))*W(M,I)*W(N,J)/2.D0
P(R)=S(M,I)
Q(R)=(1.D0-S(M,I))*(1.D0+S(N,J))/2.D0-1.D0
XR=(XC(1)+XC(2)+(XC(1)-XC(3))*p(R)+(XC(2)-XC(3))*q(R))/2.D0
YR=(YC(1)+YC(2)+(YC(1)-YC(3))*p(R)+(YC(2)-YC(3))*q(R))/2.D0
INT=INT+F(XR,YR)*L(R)
C
L1(R)=(1.D0-S(M,I))*W(N,J)*W(M,I)/2.D0
291
P1(R)=(1.D0-S(M,I))*(1.D0+S(N,J))/2.D0-1.D0
Q1(R)=S(M,I)
XR1=(XC(1)+XC(2)+(XC(1)-XC(3))*P1(R)+(XC(2)-XC(3))*Q1(R))/2.D0
YR1=(YC(1)+YC(2)+(YC(1)-YC(3))*P1(R)+(YC(2)-YC(3))*Q1(R))/2.D0
INT1=INT1+F(XR1,YR1)*L1(R)
R=R+1
ENDDO
ENDDO
INT=INT*AREA
INT1=INT1*AREA
WRITE(1,11) M+1,-1+(M+2)*(M+1)/2,INT,INT1
11 FORMAT(2X,2I5,5D25.15)
ENDDO
STOP
END
C================================****===========================
C
PROGRAM-6 SGQTS.FOR
TRIANGULAR SURFACES USING THE NEW WEIGHTS AND GAUSSIAN
POINTS. THAT IS THE TRIANGLE IN (X,Y) SPACE IS TRANSFORMED INTO A
SYMMETRIC TRIANGLE WHICH IS AGAIN TRANSFORMED INTO A 2SQUARE. NP= NUMBER OF GAUSSIAN POINTS, ITP= TOTAL NEW GAUSSIAN
POINTS FOR TRIANGLES
===============================================================
PARAMETER(NP=17, ITP=NP*NP)
INTEGER R,K
DOUBLE PRECISION S(2:NP,NP),W(2:NP,NP),XC(3),YC(3),X,Y,AREA,
1 G(ITP),U(ITP),V(ITP),XR,YR,XR1,YR1,INT1,INT2,L(TP),P(TP),Q(TP)
F(U,V)=U**6/(1.D0-1.D0*V)
OPEN( UNIT =1, FILE= 'NFA11.DAT')
C
DO 2 JJ=1,3
WRITE(*,3) JJ
READ*,XC(JJ),YC(JJ)
2 CONTINUE
C
CALL GQPW(NP,S,W)
DO 7 M=2,NP
N=M
INT1=0.D0
INT2=0.D0
R=1
292
DO 9 I=1,M
DO 10 J=1,N
G(R)=(1.D0-S(M,I))*W(M,I)*W(N,J)/8.D0
U(R)=S(M,I)
V(R)=(1.D0-S(M,I))*S(N,J)/2.D0
XR= ((2.D0*XC(1)+XC(2)+XC(3))+((2.D0*XC(1)-XC(2)-XC(3))*U(R))+
1
((2.D0*XC(2)-2.D0*XC(3))*V(R)))/4.D0
YR= ((2.D0*YC(1)+YC(2)+YC(3))+((2.D0*YC(1)-YC(2)-YC(3))*U(R))+
1
((2.D0*YC(2)-2.D0*YC(3))*V(R)))/4.D0
L(R)=(1.D0-S(N,J))*W(M,I)*W(N,J)/8.D0
P(R)=(1.D0-S(N,J))*S(M,I)/2.D0
Q(R)=S(N,J)
XR1= ((XC(1)+2.D0*XC(2)+XC(3))+2.D0*(XC(1)-XC(3))*P(R)+
1
(-XC(1)+2.D0*XC(2)-XC(3))*Q(R))/4.D0
YR1= ((YC(1)+2.D0*YC(2)+YC(3))+2.D0*(YC(1)-YC(3))*P(R)+
1
(-YC(1)+2.d0*YC(2)-YC(3))*Q(R))/4.d0
INT2=INT2+F(XR1,YR1)*L(R)
R=R+1
10 CONTINUE
9 CONTINUE
INT1=INT1*AREA
INT2=INT2*AREA
WRITE(1,12) M, INT1, INT2
12 FORMAT(2X,I4,2F25.12)
7
CONTINUE
STOP
END
C=============================****==============================
C
PROGRAM-7 SGQTM.FOR
TRIANGULAR SURFACES USING THE NEW WEIGHTS AND GAUSSIAN
POINTS. THAT IS THE TRIANGLE IN (X,Y) SPACE IS TRANSFORMED INTO A
SYMMETRIC TRIANGLE WHICH IS AGAIN TRANSFORMED INTO A 2SQUARE. NP= NUMBER OF GAUSSIAN POINTS, TP= TOTAL NEW GAUSSIAN
POINTS FOR TRIANGLES. CROWDING FREE FORMULA.
===============================================================
PARAMETER(NP=17, TP=NP*(NP+1)/2)
INTEGER R,K
DOUBLE PRECISION S(2:NP,NP),W(2:NP,NP),XC(3),YC(3),X,Y,AREA,
1 G(ITP),U(ITP),V(ITP),XR,YR,XR1,YR1,INT1,INT2,L(TP),P(TP),Q(TP)
F(U,V)=U**6/(1.D0-1.D0*V)
293
3
2
DO 2 JJ=1,3
WRITE(*,3) JJ
FORMAT('TPYE X AND Y COORDINATES OF NODE:-',I2)
READ*,XC(JJ),YC(JJ)
CONTINUE
C
CALL GQPW(NP,S,W)
DO 17 M=2, NP-2
INT1=0.D0
INT2=0.D0
R=1
K=1
DO 19 I=1,M
N=M-I+2
DO 20 J=1,N
G(R)=(1.D0-S(M,I))*W(M,I)*W(N,J)/8.D0
U(R)=S(M,I)
V(R)=(1.D0-S(M,I))*S(N,J)/2.D0
1
((2.D0*XC(2)-2.D0*XC(3))*V(R)))/4.D0
1
((2.D0*YC(2)-2.D0*YC(3))*V(R)))/4.D0
R=R+1
20 CONTINUE
19 CONTINUE
C
N=M
DO 21 J=1,N
NN=N-J+2
DO 22 I=1,NN
L(K)=(1.D0-S(N,J))*W(NN,I)*W(N,J)/8.D0
P(K)=(1.D0-S(N,J))*S(NN,I)/2.D0
Q(K)=S(N,J)
XR1= ((XC(1)+2.D0*XC(2)+XC(3)) +2.D0*(XC(1)-XC(3))*P(K)+
1
(-XC(1)+2.D0*XC(2)-XC(3))*Q(K))/4.D0
YR1= ((YC(1)+2.D0*YC(2)+YC(3)) +2.D0*(YC(1)-YC(3))*P(K)+
1
(-YC(1)+2.D0*YC(2)-YC(3))*Q(K))/4.D0
INT2=INT2+F(XR1,YR1)*L(K)
K=K+1
22 CONTINUE
21 CONTINUE
INT1=INT1*AREA
294
INT2=INT2*AREA
WRITE(1,12) -1+(m+1)*(m+2)/2,INT1,INT2
17 CONTINUE
STOP
END
C=============================****==============================
C
PROGRAM 8 GQTTS.FOR
C ARBITRATY TETRAHEDRON IS TRANSFORMED TO UNIT TETRAHEDRON
C THEN TO 2-CUBE M*N*P POINTS
PARAMETER(NP=10, ITP=NP*NP*NP)
INTEGER R,M,N,P
DOUBLE PRECISION S(2:NP,NP),W(2:NP,NP),X,Y,G(ITP),U(ITP),V(ITP),
1T(ITP),XR,YR,ZR,INT1,XC(4),YC(4),ZC(4)
C
OPEN( UNIT =11, FILE= 'N_1.DAT')
C
DO 2 JJ=1,4
WRITE(*,3) JJ
3 FORMAT('TPYE X,Y AND Z COORDINATES OF NODE:-',I2)
READ*,XC(JJ),YC(JJ),ZC(JJ)
2 CONTINUE
VOL=DABS((XC(1)-XC(4))*((YC(2)-YC(4))*(ZC(3)-ZC(4))-(YC(3)-YC(4))*
1(ZC(2)-ZC(4)))-(XC(2)-XC(4))*((YC(1)-YC(4))*(ZC(3)-ZC(4))2(YC(3)-YC(4))*(ZC(1)-ZC(4)))+(XC(3)-XC(4))*((YC(1)-YC(4))*
3(ZC(2)-ZC(4))-(YC(2)-YC(4))*(ZC(1)-ZC(4))))
CALL GQPW(NP,S,W)
C
C CASE-1, XYZ
DO 14 M=2,NP
N=M
P=M
INT1=0.D0
C
R=1
DO 11 I=1,M
DO 12 J=1,N
DO 13 K=1,P
G(R)=((1.D0-S(M,I))**2)*(1.D0-S(N,J))*W(M,I)*W(N,J)*W(P,K)/64.D0
U(R)=(1.D0+S(M,I))/2.D0
295
V(R)=(1.D0-S(M,I))*(1.D0+S(N,J))/4.D0
T(R)=(1.D0-S(M,I))*(1.D0-S(N,J))*(1.D0+S(P,K))/8.D0
XR=XC(4)+(XC(1)-XC(4))*U(R)+(XC(2)-XC(4))*V(R)+(XC(3)-XC(4))*T(R)
YR=YC(4)+(YC(1)-YC(4))*U(R)+(YC(2)-YC(4))*V(R)+(YC(3)-YC(4))*T(R)
ZR=ZC(4)+(ZC(1)-ZC(4))*U(R)+(ZC(2)-ZC(4))*V(R)+(ZC(3)-ZC(4))*T(R)
INT1=INT1+F1(XR,YR,ZR)*G(R)*VOL
R=R+1
13 CONTINUE
12 CONTINUE
11 CONTINUE
WRITE(11,*) M*M*M,INT1
14 CONTINUE
C CASE-2, XZY
DO 24 M=2,NP
N=M
P=M
INT1=0.D0
c
R=1
DO 21 I=1,M
DO 23 K=1,P
DO 22 J=1,N
G(R)=((1.D0-S(M,I))**2)*(1.D0-S(P,K))*W(M,I)*W(N,J)*W(P,K)/64.D0
U(R)=(1.D0+S(M,I))/2.D0
V(R)=(1.D0-S(M,I))*(1.D0+S(N,J))*(1.D0-S(P,K))/8.D0
T(R)=(1.D0-S(M,I))*(1.D0+S(P,K))/4.D0
R=R+1
22 CONTINUE
23 CONTINUE
21 CONTINUE
24 CONTINUE
C CASE-3, YXZ
DO 34 M=2,NP
N=M
296
P=M
INT1=0.D0
C
R=1
DO 32 J=1,N
DO 31 I=1,M
DO 33 K=1,P
G(R)=((1.D0-S(N,J))**2)*(1.D0-S(M,I))*W(M,I)*W(N,J)*W(P,K)/64.D0
U(R)=(1.D0+S(M,I))*(1.D0-S(N,J))/4.D0
V(R)=(1.D0+S(N,J))/2.D0
R=R+1
33 CONTINUE
31 CONTINUE
32 CONTINUE
34 CONTINUE
c CASE-4, YZX
DO 44 M=2,NP
N=M
P=M
INT1=0.D0
c
R=1
DO 42 J=1,N
DO 43 K=1,P
DO 41 I=1,M
G(R)=((1.D0-S(N,J))**2)*(1.D0-S(P,K))*W(M,I)*W(N,J)*W(P,K)/64.D0
U(R)=(1.D0+S(M,I))*(1.D0-S(N,J))*(1.D0-S(P,K))/8.D0
V(R)=(1.D0+S(N,J))/2.D0
T(R)=(1.D0-S(N,J))*(1.D0+S(P,K))/4.D0
R=R+1
297
41
43
42
CONTINUE
CONTINUE
CONTINUE
44 CONTINUE
c CASE-5, ZYX
DO 54 M=2,NP
P=M
N=M
INT1=0.D0
c
R=1
DO 53 K=1,P
DO 52 J=1,N
DO 51 I=1,M
G(R)=((1.D0-S(P,K))**2)*(1.D0-S(N,J))*W(M,I)*W(N,J)*W(P,K)/64.D0
V(R)=(1.D0+S(N,J))*(1.D0-S(P,K))/4.D0
T(R)=(1.D0+S(P,K))/2.D0
R=R+1
51 CONTINUE
52 CONTINUE
53 CONTINUE
54 CONTINUE
C CASE-6, ZXY
DO 64 M=2,NP
P=M
N=M
INT1=0.D0
C
R=1
DO 63 K=1,P
DO 61 I=1,M
DO 62 J=1,N
G(R)=((1.D0-S(P,K))**2)*(1.D0-S(M,I))*W(M,I)*W(N,J)*W(P,K)/64.D0
U(R)=(1.D0+S(M,I))*(1.D0-S(P,K))/4.D0
298
T(R)=(1.D0+S(P,K))/2.D0
R=R+1
62 CONTINUE
61 CONTINUE
63 CONTINUE
64 CONTINUE
STOP
END
C============================****===============================
FUNCTION F1(X,Y,Z)
DOUBLE PRECISION X,Y,Z,F1
F1=(X**2.D0)*Y
C F1=DSQRT(X+Y+Z)
RETURN
END
C============================****===============================
C
PROGRAM-9 GQTTM.FOR
C COMPLETE PROGRAM TO EVALUATE INTEGRALS OVER THE ARBITRARY
C TETRAHEDRON WHERE THE ARBITRARY TETRAHEDRON IS FIRST
C TRANSFORMED TO UNIT TETRAHEDRON THEN THE UNIT TETRAHEDRON
IS AGAIN TRANSFORMED TO A 2-CUBE. TOTAL NUMBER OF POINTS
ITP = M*(M+4)*(M+5)/6 (GQTTM)
===============================================================
INTEGER R,M,N,P
1 T(ITP),XR,YR,ZR,INT1,INT2,INT3,INT4
C
DO 2 JJ=1,4
WRITE(*,3) JJ
FORMAT('TPYE X,Y AND Z COORDINATES OF NODE:-',I2)
299
CONTINUE
VOL=DABS((XC(1)-XC(4))*((YC(2)-YC(4))*(ZC(3)-ZC(4))-(YC(3)-XC(4))*
1 (ZC(2)-ZC(4)))-(XC(2)-XC(4))*((YC(1)-YC(4))*(ZC(3)-ZC(4))2 (YC(3)-XC(4))*(ZC(1)-ZC(4)))+(XC(3)-XC(4))*((YC(1)-YC(4))*
3 (ZC(2)-ZC(4))-(YC(2)-XC(4))*(ZC(1)-ZC(4))))
PRINT*,VOL
CALL GQPW(NP,S,W)
C
c CASE-1, GQTTM
DO 14 L=2,NP-2
TT=0
INT1=0.D0
INT2=0.D0
INT3=0.D0
INT4=0.D0
c
R=1
M=L
DO 11 I=1,M
N=M-I+2
DO 12 J=1,N
P=N-J+2
DO 13 K=1,P
U(R)=(1.D0+S(M,I))/2.D0
V(R)=(1.D0-S(M,I))*(1.D0+S(N,J))/4.D0
ENDIF
13
12
R=R+1
CONTINUE
CONTINUE
TT=TT-1+(I+3)*(I+2)/2
300
11
9
14
CONTINUE
WRITE(11,9) TT, INT1,INT2,INT3,INT4
FORMAT(2X,I5,3D25.15)
CONTINUE
c CASE-2, GQTTM
DO 24 L=2,NP-2
TT=0
INT1=0.D0
INT2=0.D0
INT3=0.D0
INT4=0.D0
c
R=1
M=L
DO 21 I=1,M
P=M-I+2
DO 23 K=1,P
N=P-K+2
DO 22 J=1,N
U(R)=(1.D0+S(M,I))/2.D0
T(R)=(1.D0-S(M,I))*(1.D0+S(P,K))/4.D0
22
23
R=R+1
CONTINUE
CONTINUE
TT=TT-1+(I+3)*(I+2)/2
CONTINUE
CONTINUE
21
24
c CASE-3, GQTTM
DO 31 L=2,NP-2
TT=0
301
INT1=0.D0
INT2=0.D0
INT3=0.D0
INT4=0.D0
c
R=1
N=L
DO 32 J=1,N
M=N-J+2
DO 33 I=1,M
P=M-I+2
DO 34 K=1,P
U(R)=(1.D0+S(M,I))*(1.D0-S(N,J))/4.D0
V(R)=(1.D0+S(N,J))/2.D0
34
33
32
31
R=R+1
CONTINUE
CONTINUE
TT=TT-1+(J+3)*(J+2)/2
CONTINUE
CONTINUE
c CASE-4, GQTTM
DO 44 L=2,NP-2
TT=0
INT1=0.D0
INT2=0.D0
INT3=0.D0
INT4=0.D0
c
R=1
N=L
DO 42 J=1,N
P=N-J+2
302
DO 43 K=1,P
M=P-K+2
DO 41 I=1,M
V(R)=(1.D0+S(N,J))/2.D0
T(R)=(1.D0-S(N,J))*(1.D0+S(P,K))/4.D0
41
43
R=R+1
CONTINUE
CONTINUE
TT=TT-1+(J+3)*(J+2)/2
CONTINUE
CONTINUE
42
44
c CASE-5, GQTTM
DO 54 L=2,NP-2
TT=0
INT1=0.D0
INT2=0.D0
INT3=0.D0
INT4=0.D0
c
R=1
P=L
DO 53 K=1,P
N=P-K+2
DO 52 J=1,N
M=N-J+2
DO 51 I=1,M
V(R)=(1.D0+S(N,J))*(1.D0-S(P,K))/4.D0
T(R)=(1.D0+S(P,K))/2.D0
303
51
52
R=R+1
CONTINUE
CONTINUE
TT=TT-1+(K+3)*(K+2)/2
CONTINUE
CONTINUE
53
54
c CASE-6 GQTTM
DO 64 L=2,NP-2
TT=0
INT1=0.D0
INT2=0.D0
INT3=0.D0
INT4=0.D0
c
R=1
P=L
DO 63 K=1,P
M=P-K+2
DO 61 I=1,M
N=M-I+2
DO 62 J=1,N
U(R)=(1.D0+S(M,I))*(1.D0-S(P,K))/4.D0
T(R)=(1.D0+S(P,K))/2.D0
304
62
61
63
64
R=R+1
CONTINUE
CONTINUE
TT=TT-1+(K+3)*(K+2)/2
CONTINUE
CONTINUE
STOP
END
C=============================****==============================
PROGRAM-10 GQTSS.FOR
C
INTEGER R,M,N,P
1T(ITP),XR,YR,ZR,INT1,VOL,XC(4),YC(4),ZC(4)
c
OPEN( UNIT =11, FILE= 'tT3d.dat')
OPEN( UNIT =12, FILE= 'sq3d.dat')
C
DO 2 JJ=1,4
WRITE(*,3) JJ
2 CONTINUE
VOL=(XC(1)-XC(4))*((YC(2)-YC(4))*(ZC(3)-ZC(4))-(YC(3)-yC(4))*
1 (ZC(2)-ZC(4)))-(XC(2)-XC(4))*((YC(1)-YC(4))*(ZC(3)-ZC(4))2(YC(3)-yC(4))*(ZC(1)-ZC(4)))+(XC(3)-XC(4))*((YC(1)-YC(4))*
3(ZC(2)-ZC(4))-(YC(2)-yC(4))*(ZC(1)-ZC(4)))
CALL GQPW(NP,S,W)
C
c CASE-1, xyz
DO 14 M=2,NP
N=M
P=M
INT1=0.D0
INT2=0.D0
INT3=0.D0
INT4=0.D0
c
R=1
DO 11 I=1,M
DO 12 J=1,N
305
DO 13 K=1,P
G(R)=((1.d0-S(M,I))**2)*(1.d0-S(N,J))*W(M,I)*W(N,J)*W(P,K)/64.D0
U(R)=S(M,I)
V(R)=((1.d0-S(M,i))*(1.d0+S(N,J))/2.D0)-1.D0
T(R)=((1.d0-S(M,i))*(1.d0-S(N,J))*(1.d0+S(P,K))/4.D0)-1.D0
XR=(XC(1)+XC(2)+XC(3)-XC(4)+(XC(1)-XC(4))*U(R)+(XC(2)-XC(4))*V(R)+
1 (XC(3)-XC(4))*T(R))/2.d0
YR=(YC(1)+YC(2)+YC(3)-YC(4)+(YC(1)-YC(4))*U(R)+(YC(2)-YC(4))*V(R)+
1 (YC(3)-YC(4))*T(R))/2.d0
ZR=(ZC(1)+ZC(2)+ZC(3)-ZC(4)+(ZC(1)-ZC(4))*U(R)+(ZC(2)-ZC(4))*V(R)+
1 (ZC(3)-ZC(4))*T(R))/2.d0
R=R+1
13 CONTINUE
12 CONTINUE
11 CONTINUE
14 CONTINUE
c CASE-2, XZY
DO 24 M=2,NP
N=M
P=M
INT1=0.D0
c
R=1
DO 21 I=1,M
DO 23 K=1,P
DO 22 J=1,N
G(R)=((1.d0-S(M,I))**2)*(1.d0-S(P,K))*W(M,I)*W(N,J)*W(P,K)/64.D0
U(R)=S(M,I)
V(R)=((1.d0-S(M,i))*(1.d0+S(N,J))*(1.d0-S(P,K))/4.D0)-1.D0
T(R)=((1.d0-S(M,i))*(1.d0+S(P,K))/2.D0)-1.D0
1 (XC(3)-XC(4))*T(R))/2.d0
1 (YC(3)-YC(4))*T(R))/2.d0
1 (ZC(3)-ZC(4))*T(R))/2.d0
306
R=R+1
22 CONTINUE
23 CONTINUE
21 CONTINUE
WRITE(11,*) M*M*M,int1
24 CONTINUE
c CASE-3, YXZ
DO 34 M=2,NP
N=M
P=M
INT1=0.D0
c
R=1
DO 32 J=1,N
DO 31 I=1,M
DO 33 K=1,P
G(R)=((1.d0-S(N,J))**2)*(1.d0-S(M,I))*W(M,I)*W(N,J)*W(P,K)/64.D0
U(R)=((1.d0+S(M,i))*(1.d0-S(N,J))/2.D0)-1.D0
V(R)=S(N,J)
T(R)=((1.d0-S(M,i))*(1.d0-S(N,J))*(1.d0+S(P,K))/4.D0)-1.D0
1 (XC(3)-XC(4))*T(R))/2.d0
1 (YC(3)-YC(4))*T(R))/2.d0
1 (ZC(3)-ZC(4))*T(R))/2.d0
R=R+1
33 CONTINUE
31 CONTINUE
32 CONTINUE
34 CONTINUE
c CASE-4, YZX
DO 44 M=2,NP
N=M
P=M
INT1=0.D0
c
R=1
307
DO 42 J=1,N
DO 43 K=1,P
DO 41 I=1,M
T(R)=(1.d0-S(N,J))*(1.d0+S(P,K))/4.D0
G(R)=((1.d0-S(N,J))**2)*(1.d0-S(P,K))*W(M,I)*W(N,J)*W(P,K)/64.D0
U(R)=((1.d0+S(M,i))*(1.d0-S(N,J))*(1.d0-S(P,K))/4.D0)-1.D0
V(R)=S(N,J)
T(R)=((1.d0-S(N,J))*(1.d0+S(P,K))/2.D0)-1.D0
1 (XC(3)-XC(4))*T(R))/2.d0
1 (YC(3)-YC(4))*T(R))/2.d0
1 (ZC(3)-ZC(4))*T(R))/2.d0
R=R+1
41 CONTINUE
43 CONTINUE
42 CONTINUE
44 CONTINUE
c CASE-5, ZYX
DO 54 M=2,NP
P=M
N=M
INT1=0.D0
c
R=1
DO 53 K=1,P
DO 52 J=1,N
DO 51 I=1,M
G(R)=((1.d0-S(P,K))**2)*(1.d0-S(N,J))*W(M,I)*W(N,J)*W(P,K)/64.D0
U(R)=((1.d0+S(M,i))*(1.d0-S(N,J))*(1.d0-S(P,K))/4.D0)-1.D0
V(R)=((1.d0+S(N,J))*(1.d0-S(P,K))/2.D0)-1.D0
T(R)=S(P,K)
1 (XC(3)-XC(4))*T(R))/2.d0
1 (YC(3)-YC(4))*T(R))/2.d0
308
(ZC(3)-ZC(4))*T(R))/2.d0
R=R+1
51 CONTINUE
52 CONTINUE
53 CONTINUE
54 CONTINUE
c CASE-6, ZXY
DO 64 M=2,NP
P=M
N=M
INT1=0.D0
c
R=1
DO 63 K=1,P
DO 61 I=1,M
DO 62 J=1,N
U(R)=((1.D0+S(M,I))*(1.D0-S(P,K))/2.D0)-1.D0
V(R)=((1.D0-S(M,I))*(1.D0+S(N,J))*(1.D0-S(P,K))/4.D0)-1.D0
T(R)=S(P,K)
1 (XC(3)-XC(4))*T(R))/2.d0
1 (YC(3)-YC(4))*T(R))/2.d0
1 (ZC(3)-ZC(4))*T(R))/2.d0
R=R+1
62 CONTINUE
61 CONTINUE
63 CONTINUE
64 CONTINUE
STOP
END
309
C================================****===========================
PROGRAM -11 GQTSM.FOR
C COMPLETE PROGRAM TO EVALUATE INTEGRALS OVER THE ARBITRARY
C TETRAHEDRON WHERE THE ARBITRARY TETRAHEDRON IS FIRST
C TRANSFORMED TO STANDARD TETRAHEDRON THEN THE STANDARD
C TETRAHEDRON IS AGAIN TRANSFORMED TO A 2-CUBE. TOTAL NUMBER
OF POINTS ITP = M*(M+4)*(M+5)/6(GQTSM)
INTEGER R,M,N,P,TT
1T(ITP),XR,YR,ZR,INT1,INT2,INT3,INT4,VOL,XC(4),YC(4),ZC(4)
C
OPEN( UNIT =1, FILE= 'TT3D.DAT')
OPEN( UNIT =2, FILE= 'SQ3D.DAT')
OPEN( UNIT =11, FILE= 'TT1.DAT')
C
DO 2 JJ=1,4
WRITE(*,3) JJ
2
CONTINUE
VOL=DABS((XC(1)-XC(4))*((YC(2)-YC(4))*(ZC(3)-ZC(4))-(YC(3)-YC(4))
1 *(ZC(2)-ZC(4)))-(XC(2)-XC(4))*((YC(1)-YC(4))*(ZC(3)-ZC(4))2 (YC(3)-YC(4))*(ZC(1)-ZC(4)))+(XC(3)-XC(4))*((YC(1)-YC(4))*
3 (ZC(2)-ZC(4))-(YC(2)-YC(4))*(ZC(1)-ZC(4))))
PRINT*,VOL
CALL GQPW(NP,S,W)
C
c CASE-1, xyz
DO 14 L=2,NP-2
TT=0
INT1=0.D0
INT2=0.D0
INT3=0.D0
INT4=0.D0
c
R=1
M=L
DO 11 I=1,M
N=M-I+2
DO 12 J=1,N
P=N-J+2
DO 13 K=1,P
310
U(R)=S(M,I)
V(R)=((1.D0-S(M,I))*(1.D0+S(N,J))/2.D0)-1.D0
T(R)=((1.D0-S(M,I))*(1.D0-S(N,J))*(1.D0+S(P,K))/4.D0)-1.D0
1 (XC(3)-XC(4))*T(R))/2.D0
1 (YC(3)-YC(4))*T(R))/2.D0
1 (ZC(3)-ZC(4))*T(R))/2.D0
C
13
12
11
R=R+1
CONTINUE
CONTINUE
TT=TT-1+(I+3)*(I+2)/2
CONTINUE
FORMAT(I5,4F25.16)
CONTINUE
9
14
C
c CASE-2, XZY
DO 24 L=2,NP-2
TT=0
INT1=0.D0
INT2=0.D0
INT3=0.D0
INT4=0.D0
c
R=1
M=L
DO 21 I=1,M
P=M-I+2
DO 23 K=1,P
N=P-K+2
DO 22 J=1,N
C
U(R)=S(M,I)
T(R)=((1.D0-S(M,I))*(1.D0+S(P,K))/2.D0)-1.D0
311
1 (XC(3)-XC(4))*T(R))/2.D0
1 (YC(3)-YC(4))*T(R))/2.D0
1 (ZC(3)-ZC(4))*T(R))/2.D0
22
23
21
24
R=R+1
CONTINUE
CONTINUE
TT=TT-1+(I+3)*(I+2)/2
CONTINUE
CONTINUE
c CASE-3, YXZ
DO 34 L=2,NP-2
TT=0
INT1=0.D0
INT2=0.D0
INT3=0.D0
INT4=0.D0
c
R=1
N=L
DO 32 J=1,N
M=N-J+2
DO 31 I=1,M
P=M-I+2
DO 33 K=1,P
C
U(R)=((1.D0+S(M,I))*(1.D0-S(N,J))/2.D0)-1.D0
V(R)=S(N,J)
T(R)=((1.D0-S(M,I))*(1.D0-S(N,J))*(1.D0+S(P,K))/4.D0)-1.D0
C
1 (XC(3)-XC(4))*T(R))/2.D0
1 (YC(3)-YC(4))*T(R))/2.D0
312
33
31
32
(ZC(3)-ZC(4))*T(R))/2.D0
R=R+1
CONTINUE
CONTINUE
TT=TT-1+(J+3)*(J+2)/2
CONTINUE
CONTINUE
34
C
c CASE-4, YZX
DO 44 L=2,NP-2
TT=0
INT1=0.D0
INT2=0.D0
INT3=0.D0
INT4=0.D0
c
R=1
N=L
DO 42 J=1,N
P=N-J+2
DO 43 K=1,P
M=P-K+2
DO 41 I=1,M
U(R)=((1.D0+S(M,I))*(1.D0-S(N,J))*(1.D0-S(P,K))/4.D0)-1.D0
V(R)=S(N,J)
T(R)=((1.D0-S(N,J))*(1.D0+S(P,K))/2.D0)-1.D0
1 (XC(3)-XC(4))*T(R))/2.d0
1 (YC(3)-YC(4))*T(R))/2.d0
1 (ZC(3)-ZC(4))*T(R))/2.d0
R=R+1
313
41
43
42
CONTINUE
CONTINUE
TT=TT-1+(J+3)*(J+2)/2
CONTINUE
CONTINUE
44
c CASE-5, ZYX
DO 54 L=2,NP-2
TT=0
INT1=0.D0
INT2=0.D0
INT3=0.D0
INT4=0.D0
c
R=1
P=L
DO 53 K=1,P
N=P-K+2
DO 52 J=1,N
M=N-J+2
DO 51 I=1,M
U(R)=((1.D0+S(M,I))*(1.D0-S(N,J))*(1.D0-S(P,K))/4.D0)-1.D0
V(R)=((1.D0+S(N,J))*(1.D0-S(P,K))/2.D0)-1.D0
T(R)=S(P,K)
C
1 (XC(3)-XC(4))*T(R))/2.d0
1 (YC(3)-YC(4))*T(R))/2.d0
1 (ZC(3)-ZC(4))*T(R))/2.d0
C
C
51
52
53
R=R+1
CONTINUE
CONTINUE
TT=TT-1+(K+3)*(K+2)/2
CONTINUE
314
54
CONTINUE
c CASE-6, ZXY
DO 64 L=2,NP-2
TT=0
INT1=0.D0
INT2=0.D0
INT3=0.D0
INT4=0.D0
c
R=1
P=L
DO 63 K=1,P
M=P-K+2
DO 61 I=1,M
N=M-I+2
DO 62 J=1,N
C
U(R)=((1.D0+S(M,I))*(1.D0-S(P,K))/2.D0)-1.D0
T(R)=S(P,K)
C
1 (XC(3)-XC(4))*T(R))/2.D0
1 (YC(3)-YC(4))*T(R))/2.D0
1 (ZC(3)-ZC(4))*T(R))/2.D0
C
R=R+1
62 CONTINUE
61 CONTINUE
TT=TT-1+(K+3)*(K+2)/2
63 CONTINUE
64 CONTINUE
STOP
END
C================================****===========================
PROGRAM 12 NIPTE.FOR
315
C (NUMERICAL INTEGRATION PROGRAM FOR TRIANGULAR C ELEMENT)

C COMPLETE PROGRAM TO DETERMINE THE STIFFNESS MATRIX FOR THE
BOTH
C ST.SIDED AND WITH ONE CURVE SIDED QUADRATIC AND CUBIC
TRIANGULAR C ELEMENT. NE- NUMBER OF ELEMENTS, NN=2 FOR SIX
NODED AND NN=3 FOR
C TEN NODED TRIANGULAR ELEMENT
PARAMETER(NE=1,NN=2)
PARAMETER(N=2*NN,NP=(NN+1)*(NN+2)/2)
INTEGER CHOICE,TE
DOUBLE PRECISION X(NP),Y(NP),AX(NP,NP),AY(NP,NP),KXX(NP,NP)
1 ,KYY(NP,NP),K(NP,NP),B1,B2,B3,C1,C2,C3,E(NP-1,NP),AL0,AL1,AL2,
2 AII(0:N,0:N),SUM(0:N),SUM1(0:N,0:N),CX(NP,NP),CY(NP,NP),SKX,
3 SKY,NFACT(0:NP),D1(NP,NP),B(NP,NP),F(NP),TOL,XX,YY
OPEN( UNIT =1, FILE= 'XY.DAT')
OPEN( UNIT =2, FILE= 'T11.DAT')
OPEN( UNIT =3, FILE= 'T22.DAT')
C****FORMATION OF D1 MATRIX ,KRONEKER DELTA TO DETERMINE
MATRICES A AND B****
DO 5 IC = 1, NP
DO 5 JC = 1, NP
IF(IC.EQ.JC) THEN
D1(IC,JC) =1.D0
ELSE
D1(IC,JC) =0.D0
ENDIF
5 CONTINUE
C**********************************************************************
DO 14 N1 = 1,NE
WRITE(*,6) N1
6 FORMAT('CHOICE TYPE FOR ELEMENT NO :- ',I2,/)
PRINT*,'1 FOR ST. SIDED,2 FOR WITH ONE CURVE SIDED ELEMENT'
READ*,TE
IF(TE.EQ.1) THEN
JNODE = 3
PRINT*,'THE ELEMENT IS STRAIGHT SIDED'
B3 = 0.D0
C3 = 0.D0
ELSE
JNODE = NN+2
PRINT*,'ELEMENT IS CURVED '
END IF
DO J1 =1,JNODE
WRITE(*,18) J1
18 FORMAT('TYPE X,Y CO-ORDINATES FOR NODE ',I2)
316
READ(1,*)X(J1),Y(J1)
PRINT*, X(J1),Y(J1)
ENDDO
IF(TE.NE.1) THEN
IF(NN.EQ.2) THEN
YY = 2.D0*Y(4) - Y(1) - Y(2)
XX = 2.D0*X(4) - X(1) - X(2)
B3 =-2.D0*YY
C3 = 2.D0*XX
ELSE
YY = Y(4) + Y(5) - Y(1) - Y(2)
XX = X(4) + X(5) - X(1) - X(2)
B3 =-9.D0*YY/4.D0
C3 = 9.D0*XX/4.D0
END IF
END IF
B1 = Y(2) - Y(3)
B2 = Y(3) - Y(1)
C1 = X(3) - X(2)
C2 = X(1) - X(3)
AL0 = B1*C2 - B2*C1
AL1 = B2*C3 - B3*C2
AL2 = B1*C3 - B3*C1
C==============================================================
IF(NN.EQ.2) THEN
DO 15 J=1,NP
E(1,J)= -1.D0*(D1(1,J)+3.D0*D1(3,J)-4.D0*D1(6,J))
E(2,J)= -1.D0*(D1(2,J)+3.D0*D1(3,J)-4.D0*D1(5,J))
E(3,J)= 4.D0*(D1(1,J)+D1(3,J)-2.D0*D1(6,J))
E(4,J)= 4.D0*(D1(3,J)+D1(4,J)-D1(5,J)-D1(6,J))
E(5,J)= 4.D0*(D1(2,J)+D1(3,J)-2.D0*D1(5,J))
15
CONTINUE
F(1) = (2.D0*AL1- AL2)/60.D0
F(2) =(2.D0*AL2- AL1)/60.D0
F(3) =(AL1 + AL2)/(-60.D0)
F(4) =(5.D0*AL0 + 2.D0*AL1 + 2.D0*AL2)/15.D0
F(5) =(5.D0*AL0 + AL1 + 2.D0*AL2)/15.D0
F(6) =(5.D0*AL0 + 2.D0*AL1 + AL2)/15.D0
C****FOR THE GLOBAL DERIVATIVES W.R.TO X AS IN EQUATION
(*)**********
CALL GDQT(NP,B1,B2,B3,E,AX)
C****FOR THE GLOBAL DERIVATIVES W.R.TO Y AS IN EQUATION
(*)**********
CALL GDQT(NP,C1,C2,C3,E,AY)
ELSE
DO 16 J=1,NP
317
E(1,J)=0.5D0*(2.D0*D1(1,J)-11.D0*D1(3,J)+18.D0*D1(8,J)-9.D0
1 *D1(9,J))
E(2,J)=0.5D0*(2.D0*D1(2,J)-11.*D1(3,J)-9.D0*D1(6,J)+18.D0*D1(7,J))
E(3,J)= -9.D0*(D1(1,J)-2.D0*D1(3,J)+5.D0*D1(8,J)-4.D0*D1(9,J))
E(4,J)= 4.5D0*(4.D0*D1(3,J)-D1(4,J)-D1(5,J)+D1(6,J)-5.D0*D1(7,J)
1
- 5.D0*D1(8,J)+D1(9,J)+6.D0*D1(10,J))
E(5,J)= -9.D0*(D1(2,J)-2.D0*D1(3,J)-4.D0*D1(6,J)+5.D0*D1(7,J))
E(6,J)= 13.5D0*(D1(1,J)-D1(3,J)+3.*D1(8,J)-3.D0*D1(9,J))
E(7,J)= 13.5D0*(D1(3,J)-D1(4,J)-D1(7,J)-2.D0*D1(8,J)+D1(9,J)
1
+2.D0*D1(10,J))
E(8,J)= -13.5D0*(D1(3,J)-D1(5,J)+D1(6,J)-2.D0*D1(7,J)-D1(8,J)
1
+2.D0*D1(10,J))
E(9,J)= 13.5D0*(D1(2,J)-D1(3,J)-3.D0*D1(6,J)+3.D0*D1(7,J))
16
CONTINUE
F(1) = (4.D0*AL0 + 2.D0*AL1 + AL2)/120.D0
F(2) = (4.D0*AL0 + AL1 + 2.D0*AL2)/120.D0
F(3) = (4.D0*AL0 + AL1 + AL2)/120.D0
F(4) = (AL0 + AL1)*(3.D0/40.D0)
F(5) = (AL0 + AL2)*(3.D0/40.D0)
F(6) = (AL0 + AL2)*(3.D0/40.D0)
F(7) = (3.D0/40.D0)*AL0
F(8) = (3.D0/40.D0)*AL0
F(9) = (AL0 + AL1)*(3.D0/40.D0)
F(10)= (9.D0*AL0 + 3.D0*AL1 + 3.D0*AL2)/20.D0
C*********FOR THE GLOBAL DERIVATIVES W.R.TO X AS IN EQUATION
(*)*******
CALL GDCT(NP,B1,B2,B3,E,AX)
C*********FOR THE GLOBAL DERIVATIVES W.R.TO Y AS IN EQUATION
(*)*******
CALL GDCT(NP,C1,C2,C3,E,AY)
END IF
C*****OUTPUT FOR AX AND AY WILL BE PRINTED BY THE FOLLOWING
FRAGMENTS****
WRITE( 2, 20) NP,N1
WRITE( 3, 20) NP,N1
20 FORMAT(/,8X,'FOR ',I2,'-NODED ELEMENT NO :- ',I2,/)
WRITE(2,21)((I,J,AX(I,J),I,J,AY(I,J),J=1,NP),I=1,NP)
21 FORMAT(2X,'AX(',I2,',',I2,')=',D18.10,2X,'AY(',I2,',',I2,')=',
1 D18.10)
C*******THIS PART FOR THE OUTPUT OF AL0,AL1,AL2 AND
F(I),I=1,2,3,...,NP***
WRITE(2,30) AL0,AL1,AL2
WRITE(3,30) AL0,AL1,AL2
30 FORMAT(2X,'AL0 =',D18.10,2X,'AL1 =',D18.10,2X,'AL2 =',D18.10)
WRITE(3,31) (II,F(II),II=1,NP)
31 FORMAT(3X,'F(',I2,')=',D18.10,2X,'F(',I2,')=',D18.10)
318
WRITE(3,32)
32 FORMAT(3X,'====================******================== ')
C******SUBROUTINE
'FACT'
CALLED
HERE
TO
OBTAIN
FACTORIAL(NP)*******
CALL FACT(NP,NFACT)
IF(TE.EQ.2) THEN
C******FOR ALL INTEGRALS SUBROUTINES ARE CALLED HERE****
RP=10
ITP=RP*RP
C FOR MM POINTS UNIT TRIANGLE ELEMENT
CALL NIUTS(N,RP,ITP,AL0,AL1,AL2,NFACT,AII)
C FOR M(M+1)/2-1 POINTS UNIT TRIANGLE ELEMENT
C
CALL NIUTM(N,RP,ITP,AL0,AL1,AL2,NFACT,AII)
C FOR MM POINTS STANDARD TRIANGLE ELEMENT
C
CALL NISTS(N,RP,ITP,AL0,AL1,AL2,NFACT,AII)
C FOR M(M+1)/2-1 POINTS STANDARD TRIANGLE ELEMENT
C
CALL NISTM(N,RP,ITP,AL0,AL1,AL2,NFACT,AII)
C FOR MM POINTS SYMMETRIC TRIANGLE ELEMENT
C
CALL NISYS(N,RP,ITP,AL0,AL1,AL2,NFACT,AII)
C FOR M(M+1)/2-1 POINTS SYMMETRIC TRIANGLE ELEMENT
C
CALL NISYM(N,RP,ITP,AL0,AL1,AL2,NFACT,AII)
WRITE(2,33) N1
33
FORMAT(3X,'FOR THE CURVED ELEMENT :-',I2,/,3X,'
1 CALCULATED INTEGRAL VALUES ARE',/)
DO 34 LM=1,N+1
DO 34 LK=1,LM
WRITE(2,35) LM-LK,LK-1,AII(LM-LK,LK-1)
35 FORMAT(3X,'AII(',I2,',',I2,')=',D18.10)
34 CONTINUE
ENDIF
CALL BMT(TE,NN,NP,N,AL0,NFACT,AII,B)
C*****THIS PART TO CALCULATE AND PRINTS THE COMPONENTS OF
STIFFNESS MATRICES*******
DO 36 I = 1,NP
DO 37 J = I,NP
KXX(I,J) = 0.0
KYY(I,J) = 0.0
IF(I.EQ.1) THEN
DO 40 II = 1,NP
CX(II,J) = 0.0
CY(II,J) = 0.0
DO 45 K1 = 1,NP
CX(II,J) = CX(II,J) +B(II,K1)*AX(K1,J)
CY(II,J) = CY(II,J) +B(II,K1)*AY(K1,J)
45
CONTINUE
319
40
CONTINUE
END IF
DO 50 L = 1,NP
KXX(I,J) = KXX(I,J) +AX(L,I)*CX(L,J)
KYY(I,J) = KYY(I,J) +AY(L,I)*CY(L,J)
50
CONTINUE
K(I,J) = KXX(I,J) +KYY(I,J)
C********OUTPUT OF KXX(I,J),KYY(I,J) AND K(I,J)************************
WRITE(3,51) I,J
51 FORMAT(3X,'FOR I =',I2,2X,'AND J =',I2)
WRITE(3,52) I,J,KXX(I,J),I,J,KYY(I,J),I,J, K(I,J)
52 FORMAT(3X,'CALCULATED',/,3X,'KXX(',I2,',',I2,')=',
1 D18.10,3X,'KYY(',I2,',',I2,')=',D18.10,/,3X,'AND HENCE',/
2 ,3X,'K(',I2,','I2,')=',D18.10)
WRITE(3,53)
53 FORMAT(3X,'=====================****====================',/)
37 CONTINUE
36 CONTINUE
WRITE (3, 60) N1
60 FORMAT(/,3X, 'GIVEN COORDONATES FOR ELEMENT NO:-',I2,/)
DO 61 IN = 1, JNODE
WRITE( 3,62) IN,X(IN),IN,Y(IN)
62 FORMAT(3X,'X(',I2,')=',D18.10,3X,'Y(',I2,')=',D18.10)
61 CONTINUE
WRITE( 3,63) B1,C1,B2,C2,B3,C3
63 FORMAT(/,3X,'CALCULATED',/,3X,'B1 =',D18.10,3X,'C1 =',D18.10,/,
1 3X,'B2 =',D18.10,3X,'C2 =',D18.10,/,3X,'B3 =',D18.10,3X,
1 'C3=',D18.10,/)
14 CONTINUE
PRINT*,'COMPUTATION OVER'
STOP
END
C============================================================
SUBROUTINE FACT(NP,NFACT)
DOUBLE PRECISION NFACT(0:NP)
DO 100 NFF = 1, NP+1
NF = NFF-1
IF(NF.EQ.0) THEN
NFACT(NF) = 1.D0
ELSE
NFACT(NF) = NF* NFACT(NF-1)
ENDIF
100
CONTINUE
RETURN
END
C==============================================================
320
SUBROUTINE GDQT(NP,S1,S2,S3,E,AT)
DOUBLE PRECISION S1,S2,S3,AT(NP,NP),E(NP-1,NP)
DO 100 I = 1,NP
AT(1,I) = S1 * E(1,I) + S2 * E(2,I)
AT(2,I) = S1 * E(3,I) + S2 * E(4,I) - S3 * E(1,I)
AT(3,I) = S1 * E(4,I) + S2 * E(5,I) + S3 * E(2,I)
AT(4,I) = -S3 * E(3,I)
AT(5,I) = 0.D0
AT(6,I) = S3 * E(5,I)
100 CONTINUE
RETURN
END
C==============================================================
SUBROUTINE GDCT(NP,S1,S2,S3,E,AT)
DOUBLE PRECISION S1,S2,S3,AT(NP,NP),E(NP-1,NP)
DO 105 I = 1,NP
AT(1,I) = S1*E(1,I)+S2*E(2,I)
AT(2,I) = S1*E(3,I)+S2*E(4,I)-S3*E(1,I)
AT(3,I) = S1*E(4,I)+S2*E(5,I)+S3*E(2,I)
AT(4,I) = S1*E(6,I)-S2*E(7,I)-S3*E(3,I)
AT(5,I) = -2.*(S1*E(7,I)-S2*E(8,I))
AT(6,I) = S1*E(8,I)+S2*E(9,I)+S3*E(5,I)
AT(7,I) = -S3*E(6,I)
AT(8,I) = S3*E(7,I)
AT(9,I) = S3*E(8,I)
AT(10,I)= S3*E(9,I)
105 CONTINUE
RETURN
END
C==============================================================
SUBROUTINE BMT(TE,NN,NP,N,AL0,NFACT,AII,B)
DOUBLE PRECISION B(NP,NP),NFACT(0:NP),AII(0:N,0:N),AL0
INTEGER TE
I=0
DO 102 IRR=1,NN+1
IR=IRR-1
DO 103 ISS=1,IRR
IS=ISS-1
I=I+1
J=0
DO 104 LL=1,NN+1
L=LL-1
DO 105 KK=1,LL
K=KK-1
J=J+1
IF(TE.EQ.1) THEN
321
B(I,J)=NFACT(L+IR-IS-K)*NFACT(K+IS)/(NFACT(L+IR+2)*AL0)
ELSE
B(I,J)=AII(L+IR-IS-K,K+IS)
END IF
105
CONTINUE
104
CONTINUE
103 CONTINUE
102 CONTINUE
RETURN
END
C==============================================================
SUBROUTINE NIUTS(N,NP,ITP,AL0,AL1,AL2,NFACT,AII)
DOUBLE PRECISION AII(0:N, 0:N),NFACT(0:NP),
1 AL0,AL1,AL2,U(ITP),V(ITP),G(ITP),XR,YR,Z,F,SUM(0:N,0:N),XR1,YR1,
2S(2:NP,NP),W(2:NP,NP),TEMP(N+1),PT,ETOL,U1(ITP),V1(ITP),G1(ITP)
REAL P
ETOL = 0.0000001D0
CALL GQPW(NP,S,W)
DO 7 M=2,NP
NN=M
PT=0.D0
DO P =0,N
AII(P,0)=0.D0
AII(0,P)=0.D0
R=1
DO 9 J=1,NN
DO 10 I=1,M
G(R)=(1.D0-S(M,I))*W(M,I)*W(NN,J)/8.D0
U(R)=(1.D0+S(M,I))/2.D0
V(R)=(1.D0-S(M,I))*(1.D0+S(NN,J))/4.D0
XR=U(R)
YR=V(R)
G1(R)=(1.D0-S(NN,J))*W(M,I)*W(NN,J)/8.D0
U1(R)=(1.D0+S(M,I))*(1.D0-S(NN,J))/4.D0
V1(R)=(1.D0+S(NN,J))/2.D0
XR1=U1(R)
YR1=V1(R)
IF((AL2.NE.0.).AND.(AL1.EQ.0.)) THEN
Z=F(P,0,AL0,AL1,AL2,XR1,YR1)
AII(P,0)=AII(P,0)+Z*G1(R)
ELSEIF((AL1.NE.0.).AND.(AL2.EQ.0.)) THEN
Z=F(0,P,AL0,AL1,AL2,XR,YR)
AII(0,P)=AII(0,P)+Z*G(R)
322
ELSEIF((AL1.NE.0.).AND.(DABS(AL2).GE.DABS(AL1))) THEN
Z=F(P,0,AL0,AL1,AL2,XR,YR)
AII(P,0)=AII(P,0)+Z*G(R)
ELSEIF((AL2.NE.0.).AND.(DABS(AL1).GT.DABS(AL2))) THEN
Z=F(0,P,AL0,AL1,AL2,XR1,YR1)
AII(0,P)=AII(0,P)+Z*G1(R)
ENDIF
R=R+1
10 CONTINUE
9 CONTINUE
ENDDO
7 CONTINUE
IF(DABS(AL1).GE.DABS(AL2))THEN
DO 178 A=0,N-1
DO 180 B=0,N-A-1
SUM(A,B)= (NFACT(A)*NFACT(B))/(NFACT(A+B+2))
AII(A+1,B)=(1./AL1)*(SUM(A,B)-AL0*AII(A,B)
1 -AL2*AII(A,B+1))
180
CONTINUE
178
CONTINUE
ELSE
DO 188 B=0,N-1
DO 190 A=0,N-B-1
AII(A,B+1)=(1./AL2)*(SUM(A,B)-AL0*AII(A,B)
1 -AL1*AII(A+1,B))
190
CONTINUE
188
CONTINUE
ENDIF
RETURN
END
C================================****===========================
SUBROUTINE NIUTM(N,NP,ITP,AL0,AL1,AL2,NFACT,AII)
1 AL0,AL1,AL2,U(ITP),V(ITP),G(ITP),XR,YR,Z,F,SUM(0:N,0:N),XR1,YR1,
2S(2:NP,NP),W(2:NP,NP),TEMP(N+1),PT,ETOL,U1(ITP),V1(ITP),G1(ITP)
REAL P
ETOL = 0.0000001D0
CALL GQPW(NP,S,W)
DO 7 M=2,NP
PT=0.D0
DO P =0,N
AII(P,0)=0.D0
AII(0,P)=0.D0
R=1
323
DO 9 I=1,M
NN=M-I+2
DO 10 J=1,NN
G(R)=(1.D0-S(M,I))*W(M,I)*W(NN,J)/8.D0
U(R)=(1.D0+S(M,I))/2.D0
V(R)=(1.D0-S(M,I))*(1.D0+S(NN,J))/4.D0
XR=U(R)
YR=V(R)
G1(R)=(1.D0-S(M,I))*W(M,I)*W(NN,J)/8.D0
U1(R)=(1.D0+S(NN,J))*(1.D0-S(M,I))/4.D0
V1(R)=(1.D0+S(M,I))/2.D0
XR1=U1(R)
YR1=V1(R)
10
9
7
ENDIF
R=R+1
CONTINUE
CONTINUE
ENDDO
CONTINUE
DO 178 A=0,N-1
DO 180 B=0,N-A-1
1 -AL2*AII(A,B+1))
180
CONTINUE
178
CONTINUE
ELSE
DO 188 B=0,N-1
DO 190 A=0,N-B-1
324
-AL1*AII(A+1,B))
CONTINUE
CONTINUE
ENDIF
RETURN
END
C================================****===========================
SUBROUTINE NISTS(N,NP,ITP,AL0,AL1,AL2,NFACT,AII)
1 AL0,AL1,AL2,U(ITP),V(ITP),G(ITP),XR,YR,Z,F,SUM(0:N,0:N),
2S(2:NP,NP),W(2:NP,NP),TEMP(N+1),PT,ETOL,XC(3),YC(3),
3U1(ITP),V1(ITP),G1(ITP),XR1,YR1
REAL P
DATA XC(1)/1.D0/, XC(2)/0.D0/,XC(3)/0.D0/,YC(1)/0.D0/, YC(2)/1.D0/
1 ,YC(3)/0.D0/
ETOL = 0.0000001D0
CALL GQPW(NP,S,W)
1
190
188
DO 7 M=2,NP
NN=M
PT=0.D0
DO P =0,N
AII(P,0)=0.D0
AII(0,P)=0.D0
R=1
DO 9 J=1,NN
DO 10 I=1,M
G(R)=(1.D0-S(M,I))*W(M,I)*W(NN,J)/8.D0
U(R)=S(M,I)
V(R)=(1.D0-S(M,I))*(1.D0+S(NN,J))/2.D0-1.D0
U1(R)=(1.D0-S(NN,J))*(1.D0+S(M,I))/2.D0-1.D0
V1(R)=S(NN,J)
325
ENDIF
R=R+1
10 CONTINUE
9 CONTINUE
ENDDO
7 CONTINUE
DO 178 A=0,N-1
DO 180 B=0,N-A-1
1 -AL2*AII(A,B+1))
180
CONTINUE
178
CONTINUE
ELSE
DO 188 B=0,N-1
DO 190 A=0,N-B-1
1 -AL1*AII(A+1,B))
190
CONTINUE
188
CONTINUE
ENDIF
RETURN
END
C==============================================================
SUBROUTINE NISTM(N,NP,ITP,AL0,AL1,AL2,NFACT,AII)
REAL P
1,YC(3)/0.D0/
ETOL = 0.0000001D0
CALL GQPW(NP,S,W)
326
DO 7 M=2,NP
PT=0.D0
DO P =0,N
AII(P,0)=0.D0
AII(0,P)=0.D0
R=1
DO 9 I=1,M
NN=M-I+2
DO 10 J=1,NN
G(R)=(1.D0-S(M,I))*W(M,I)*W(NN,J)/8.D0
U(R)=S(M,I)
V(R)=(1.D0-S(M,I))*(1.D0+S(NN,J))/2.D0-1.D0
G1(R)=(1.D0-S(M,I))*W(M,I)*W(NN,J)/8.D0
U1(R)=(1.D0+S(NN,J))*(1.D0-S(M,I))/2.D0-1.D0
V1(R)=S(M,I)
ENDIF
R=R+1
10 CONTINUE
9 CONTINUE
ENDDO
7 CONTINUE
DO 178 A=0,N-1
DO 180 B=0,N-A-1
327
-AL2*AII(A,B+1))
CONTINUE
CONTINUE
ELSE
DO 188 B=0,N-1
DO 190 A=0,N-B-1
1 -AL1*AII(A+1,B))
190
CONTINUE
188
CONTINUE
ENDIF
RETURN
END
C================================****===========================
SUBROUTINE NISYS(N,NP,ITP,AL0,AL1,AL2,NFACT,BJJ,AII)
DOUBLE PRECISION AII(0:N, 0:N),NFACT(0:NP),BJJ(0:N, 0:N),
REAL P
1 ,YC(3)/0.D0/
ETOL = 0.0000001D0
CALL GQPW(NP,S,W)
1
180
178
DO 7 M=2,NP
NN=M
PT=0.D0
DO P =0,N
AII(P,0)=0.D0
AII(0,P)=0.D0
R=1
DO 9 J=1,NN
DO 10 I=1,M
G(R)=(1.D0-S(M,I))*W(M,I)*W(NN,J)/8.D0
U(R)=S(M,I)
V(R)=(1.D0-S(M,I))*S(NN,J)/2.D0
XR=((2.D0*XC(1)+XC(2)+XC(3))+((2.D0*XC(1)-XC(2)-XC(3))*U(R))+
1
((2.D0*XC(2)-2.D0*XC(3))*V(R)))/4.D0
1
((2.D0*YC(2)-2.D0*YC(3))*V(R)))/4.D0
328
U1(R)=(1.D0-S(NN,J))*S(M,I)/2.D0
V1(R)=S(NN,J)
XR1= ((XC(1)+2.D0*XC(2)+XC(3))+2.D0*(XC(1)-XC(3))*U1(R)+
1
(-XC(1)+2.D0*XC(2)-XC(3))*V1(R))/4.D0
YR1= ((YC(1)+2.D0*YC(2)+YC(3))+2.D0*(YC(1)-YC(3))*U1(R)+
1
(-YC(1)+2.D0*YC(2)-YC(3))*V1(R))/4.D0
ENDIF
R=R+1
10 CONTINUE
9 CONTINUE
ENDDO
7
CONTINUE
DO 178 A=0,N-1
DO 180 B=0,N-A-1
1 -AL2*AII(A,B+1))
180
CONTINUE
178
CONTINUE
ELSE
DO 188 B=0,N-1
DO 190 A=0,N-B-1
1 -AL1*AII(A+1,B))
190
CONTINUE
188
CONTINUE
ENDIF
RETURN
END
C================================****===========================
329
SUBROUTINE NISYM(N,NP,ITP,AL0,AL1,AL2,NFACT,AII)
REAL P
1,YC(3)/0.D0/
ETOL = 0.0000001D0
CALL GQPW(NP,S,W)
DO 7 M=2,NP
PT=0.D0
DO P =0,N
AII(P,0)=0.D0
AII(0,P)=0.D0
R=1
DO 9 I=1,M
NN=M-I+2
DO 10 J=1,NN
G(R)=(1.D0-S(M,I))*W(M,I)*W(NN,J)/8.D0
U(R)=S(M,I)
V(R)=(1.D0-S(M,I))*S(NN,J)/2.D0
1
((2.D0*XC(2)-2.D0*XC(3))*V(R)))/4.D0
1
((2.D0*YC(2)-2.D0*YC(3))*V(R)))/4.D0
G1(R)=(1.D0-S(M,I))*W(NN,J)*W(M,I)/8.D0
U1(R)=(1.D0-S(M,I))*S(NN,J)/2.D0
V1(R)=S(M,I)
XR1= ((XC(1)+2.D0*XC(2)+XC(3)) +2.D0*(XC(1)-XC(3))*U1(R)+
1
(-XC(1)+2.D0*XC(2)-XC(3))*V1(R))/4.D0
YR1= ((YC(1)+2.D0*YC(2)+YC(3)) +2.D0*(YC(1)-YC(3))*U1(R)+
1
(-YC(1)+2.D0*YC(2)-YC(3))*V1(R))/4.D0
330
ENDIF
R=R+1
10 CONTINUE
9 CONTINUE
ENDDO
7 CONTINUE
DO 178 A=0,N-1
DO 180 B=0,N-A-1
1 -AL2*AII(A,B+1))
180
CONTINUE
178
CONTINUE
ELSE
DO 188 B=0,N-1
DO 190 A=0,N-B-1
1 -AL1*AII(A+1,B))
190
CONTINUE
188
CONTINUE
ENDIF
RETURN
END
C====================================================
DOUBLE PRECISION FUNCTION F(P,Q,AL0,AL1,AL2,X,Y)
DOUBLE PRECISION AL0,AL1,AL2,X,Y
F=(X**P)*(Y**Q)/(AL0+AL1*X+AL2*Y)
RETURN
END
C==============================================================
331
References
M. Abramowitz and Stegun. Handbook of mathematical functions. Dover, 1974.
42, 56
C. M. Andersen and R. J. Mcleod. Integration technique for isoparametric
higher order bases on finite elements with a curved side. Comp. Math. Appl.,
5:285295, 1979. 10
J. H. Argyris. Recent Advances in Matrix Methods of Structural Analysis. Pergamon Press, Almsford, N.Y., 1954. 19
M. L. Baart and R. J. Mcleod. Recursive calculation of curved finite elements
stiffness matrices. Report: CSIR, Pretoria, South Africa, 1983. 10
D. Babu and G.F. Pinder. Analytical integration formulae for linear isoparametric. International journal on Numerical Meth. Engneering, 20:11531166,
1984. 6, 10
K. J. Bathe. Finite Element Procedures, 4th Ed. Prentice-Hall, Englewood Cliffs,
N.J., 1996. 42, 55, 67, 101
F. Bernardini. Integrating of polynomials over n-dimensional polyhedra. Comput. Aid. Des., 23:5158, 1991. 6, 11, 100, 129, 171, 203
W. B. Bickford. A First Course in Finite Element Method. Richard D. Irwin
Inc., 1990. 20
G. F. Carey and T. J. Oden. Finite Elements: A Second Course. Prentice Hall,
Englewood Cliffs, N. J., 1983. 23
332
REFERENCES
C. Cattani and A. Paoluzzi. Boundary integration over linear polyhedra.
Comput-Aid. Des., 22:130135, 1990a. 6, 11, 171, 203
C. Cattani and A. Paoluzzi. Symbolic analysis of linear polyhedra. Enrng
Comput., 6:1729, 1990b. 6, 11, 171, 203
S. C. Chapra and R. P. Canale. The Numerical Methods for Engineers, 5th ed.
McGraw-Hill Publishing Company, 2006. 147, 221
P. G. Ciarlet and R. A. Raviart. General lagrange and hermite interpolation
in rn application to finite element method. Arc. Ret. Mech. Anal, 4:177199,
1972. 9
R. W. Clough. The finite element method in plane stress analysis. Proc., 2nd
Conf. Electronic Computations, ASCE, pages 345378, 1960. 19
R. W. Clough and Peterson. Application of finite element method. The mathematics of Finite elements and Application, 2:1520, 1976. 4
Ronald Cools. An encyclopaedia of cubature formulas. Journal of Complexity,
19(3):445453, 2003. 129
Ronald Cools and Philip Rabinowitz. Monomial cubature rules since Stroud
- a compilation. Journal of Computational and Applied Mathematics, 48(3):
309326, 1993. 129
G. R. Cowper. Gaussian quadrature formulas for triangles. International journal
on numerical methods and engineering, 7:405408, 1973. 6, 9, 42, 56, 67, 100,
129
C. S. Desai. Elementary Finite Element Method. Prentice Hall Inc., Englewood
Cliffs, N. J., 1979. 20
D. A. Dunavant. High degree efficient symmetrical gaussian quadrature rules for
the triangle. International Journal for Numerical Methods in Engineering, 21:
11291148, 1985. 42, 56, 100, 129, 130, 144, 145, 147
333
REFERENCES
J. G. Ergatoudis, B. M. Irons, and O. C. Zienkiewicz. Curved isoparametric
quadrilateral element for finite analysis. Int. J. Solids and Struct., 4:3142,
1968. 9
B. A. Finlayson. The Method of Weighthed Residuals and Variation Principles.
Academic Press, N. Y., 1972. 20, 21, 31
C. A. J. Fletcher. Computational Galerkins Method. Springer-Verlag, Berlin,
1984. 20
B. G. Galerkin. Series occuring in some problems of elasticstability of rods and
plates. Engineering Bulletine, 19:897908, 1915. 3, 20, 30
C. F. Gauss. Methodus nova integratiam valores per approximationem inveniendi
worke. Int. J., 3:163196, 1814. 171, 203
D. V. Griffths. Stiffness matrix of the four -node quadrilateral element in closed
form. Int. J. Num. Methods and Eng., 165:10271038, 1994. 6, 10
D. V. Griffths and G. G. W. Mustoe. Selective reduced integration of four-node
plane element in closed form. J.A.Mech, pages 725729, 1995. 6, 10
W. L. Hacker and H. L. Schreyer. Eigen value analysis of compatible and
incompatible finite element. Int. J. Num. Methods and Engineers, 28:687703,
1989. 6, 10
P. C. Hammer and A. H. Stroud. Numerical integration over simplexes. Math
Tables Other Aids Computation, 10:137139, 1956. 42, 56, 67, 100, 129, 203
P. C. Hammer and A. H. Stroud. Numerical evaluation of multiple integrals.
Math Tables Other Aids Computation, 12:272280, 1958. 42, 56, 67, 100, 129,
203
P. C. Hammer, A. H. Stroud, and O. J. Marlowe. Numerical integration over
simplex and cones. Math Tables Other Aids Computation, 10:130136, 1956.
42, 56, 67, 100, 101, 129, 203
J. Heuser. Finite Element Method for Thermal Analysis. NASA Technical Note
TN D-7274, Goddard Space Flight Center, Greenbelt, MD, 1975. 20
334
REFERENCES
P. Hillion. Numerical integration on a triangle. International journal on numerical methods and engineering, 11:797815, 1977. 42, 67, 100, 129
K. H. Huebner. The Finite Element Method for Engineers. John Wiley and
Sons, N.Y., 1975. 20
T. J. R. Hughes. The Finite Element Method. Prentice-Hall, Englewood Cliffs,
N.J., 1987. 20, 42, 55, 66, 100, 171, 203
B. M. Irons and A. Razzaque. Experience with the patch-test for convergence of
finite element method. Academic press, New York, 1972. 42, 55, 67, 101
C. Johnson. Numerical Solution of Partial Differential Equation by Finite Element Method. Cambridge University Press, Cambridge, U. K., 1987. 20
M. S. Karim. Integration of Some Bivariate Polymials with Rational Denumerators -An Application to Finite Element Method. A Thesis Submitted for the
Degree of DOCTOR OF PHYLOSOPHY, Department of Mathematics, Bangalore University, Bangalore, India, 2001. 6, 11, 28, 243, 248, 251, 255
J. B. Keller. Stochastic equations and wave propagation in random media. American Mathematical Society, Providence, R.I., 16, 1964. Proc., Symposium on
Applied Mathematics. 42, 56, 67
J. B. Keller and H. P. McKean. siam-ams proceedings - stochastic differential
equations. American Mathematical Society, Providence, R.I., VI, 1973. 42, 56,
67
G. Lague and R. Baldur. Extended numerical integration method for triangular
surfaces. International journal on numerical methods and engineering, 11:388
392, 1977. 6, 9, 42, 56, 67, 100, 101, 104, 129
F. G. Lannoy. Triangular finite element and numerical integration. Computers
Struct., 7:613, 1977. 42, 67, 100, 129
D. P. Laurie. Automatic numerical integration over a triangle. National Institute
for Mathematical Science, Pretoria, CSIR Spec. Rep. WISK 273, 1977. 42, 67,
100, 129
335
REFERENCES
M. E. Laursen and M. Gellert. Some criteria for numerically integrated matrices and quadrature formulas for triangles. International journal for numerical
methods in engineering, 12:6776, 1978. 42, 67, 100, 129
Y. T. Lee and A. A. G. Requicha. Algorithms for computing the volume and
other integral properties of solids i: known methods and open issues. Commun.
ACM, 25:635641, 1982. 6, 11, 171, 203
F. G. Lether. Computation of double integrals over a triangle. Journal comp.
applic. Math., 2:219224, 1976. 42, 56, 67, 100, 129
P. E. Lewis and J. P. Ward. The Finite element Method. Addison-Wesley
Publishing Company, 1991. 247
S. Lien and J.T. Kajiya. A symbolic method for calculating the integral properties of arbitrary nonconvex polyhedra. IEEE Comput. Graph. Applic., 4:3541,
1984. 6, 11, 171, 203
J. N. Lyness and Ronald Cools. A survey of numerical cubature over triangles. Proceedings of Symposia in Applied Mathematics, American Mathematical
Society, pages 127150, 1994. 129
J. N. Lyness and D. Jespersen. Moderate degree symmetric quadrature rules
for triangle. Journal of the Institute of Mathematics and its Applications, 15(1):
1932, 1975. 42, 56, 67, 100, 129
H. C. Martin. Finite element analysis of fluid flow. Proc., 2nd Conf. Matrix
Method in Structural Mechanics, Wright Patterson Air Force Bare, 1968. 19
R. Mcleod. High order transformation method for curved finite elements. J. Int.
Math. Appl., 21:419428, 1978. 9, 10
R. Mcleod and A. R. Mitchel. The construction of basic function for curved
element in the finite element method. J. Int. Math. Application, 10:382393,
1972. 8, 9, 10
R. Mcleod and A. R. Mitchel. Use of parabolic arc in matching curved boundaries in the finite element method. J. Int. Math. Appl., 16:239246, 1975. 9
336
REFERENCES
J. E. Molyneux. Analysis of dishonest methods in the theory of wave propagation in a random medium. J. Opt. Soc. Am., 58:951957, 1968. 42, 56,
67
S. H. Nguyen. An accurate finite element formulation of linear elastic torsion
calculations. Computers and structures, 42(5):111121, 1992. 251, 252, 266
J. T. Oden. Finite Elements of Nonlinear continua. McGraw Hill Book Co, N.Y.,
1972. 19, 22
T. J. Oden and C. T. Reddy. An Introduction to Mathematical Theory of Finite
Elements. John Wiley and Sons, N. J., 1976. 23
M. Okabe. Analytic integration formula related to convex quadrilateral finite
elements. Computer Methods Appl. Mechan. Engng, 29:201218, 1981. 6, 10
D. R. J. Owen and E. Hinton. A Simple Guide for Finite Elements. Pineridge
Press, U. K., 1980. 20
H. T. Rathod. Some analytical integration formulae for a four-node isoparametric
element. Computers and Structures, 30:11011109, 1988. 6, 9, 10
H. T. Rathod and M. S. Islam. Integration of rational functions of bivariate
polymial numerators with linear denumerators over a square in local parametric
space. Computers Methods in Applied Mechanics and Engineering, 161:195213,
1998. 6
H. T. Rathod and M. S. Karim. Synthetic division based integration of rational
functions of bivariate polymial numerators with linear denumerators over a unit
triangle. Computers Methods in Applied Mechanics and Engineering, 181:191
235, 2000. 10, 28
H. T. Rathod and M. S. Karim. An explicit integration scheme based on recursion for the curved triangular finite elements. Computer and Structures, 80(1):
7393, 2001. 243, 244, 245
H. T. Rathod and M. S. Karim. An explicit integration scheme based on recursion for curved triangular finite elements. Computer structure, 80:4376, 2002.
6, 11, 42, 43, 56, 57, 67, 79, 101, 104, 145, 243, 244
337
REFERENCES
H. T. Rathod and H. S. G. Rao. Integration of polynomials over an arbitrary
tetrahedron in euclidean three dimensional space. Computers and Structures,
59(1):5565, 1996. 101, 129, 186, 191, 217
H. T. Rathod and H. S. G. Rao. Integration of trivariate polynomials over linear
polyhedra in euclidean three dimensional space. J. Austral. Math. Soc. Ser B,
39:355385, 1998. 101, 129, 186, 191, 217
H. T. Rathod, K. V. Nagaraja, B. Venkatesudu, and N. L. Ramesh. Gauss
legendre quadrature over a triangle. Journal of Indian Institute of Science, 84:
183188, 2004. 6, 11, 43, 57, 101, 129, 171, 186, 191, 203, 217, 220
H. T. Rathod, B. Venkatesudu, and K. V. Nagaraja. Gauss legendre quadrature
formulas over tetrahedron. Numerical Methods for Partial Differential Equations, 22(1):197219, 2005. 6, 11, 171, 186, 191, 203, 217, 220
H.T. Rathod, A. S. Hariprasad, K.V. Vijayakumar, Bharath Rathod, and C.S.
Nagabhushana. Numerical integration over curved domains using convex quadrangulations and gauss legendre quadrature rules. International Journal Of
Engineering And Computer Science, 2(11):32903332, 2013. 243, 253
J. W. S. Rayleigh. Theory of Sound. Dover Publisher, N. Y., 1877. 20, 30
C. T. Reddy. Improved three point integration schemes for triangular finite
elements. International journal on numerical methods and engineering, 12:1890
1896, 1978. 42, 56, 67, 100, 129
C. T. Reddy and D. J. Shippy. Alternative integration formulae for triangular
finite elements. International journal for numerical methods in engineering, 17:
133139, 1981. 42, 56, 67, 100, 129, 186, 191, 217
J. N. Reddy. An Introduction to finite Element Method. McGraw Hill, N. Y.,
1984. 20, 33
W. Ritz. Uber eine neue method zur losung gewisses variations - problem der
mathematischen physik. Journal Reine Angew. Math, 135:161, 1909. 20, 30
338
REFERENCES
P. J. Roach. Computational Fluid Mechanics. Hermosa Publishers, Albuquerque,
N.M., 1972. 19
D. F. Rogers and J. A. Adams. Mathematical Elements of Computer Graphics.
McGraw-Hill, New York, 1990. 42, 55, 67, 101
L. J. Segerlind. Applied Finite Element Analysis, 2nd Ed. John Wiley and Sons,
N. Y., 1984. 20, 27, 247
P. J. Silva and C. D. Mote. Curved composite boundary elements for second
order two dimensional problems. International Journal on Numerical Methods
for Eng., 26:12651279, 1988. 10
I. M. Smith. Programming the Finite Element Method. John Wiley and Sons, N.
Y., 1982. 20
R. V. Southwell. Relaxation Methods in Theoretical Physics. Oxford University
Press, Clarendon, 1946. 19
G. Strang and G. J. Fix. An analysis of the finite element method. Prentice-Hall,
Englewood Cliffs, N.J., 1973. 9, 42, 55, 56, 67
M. A. Taylor, B. A. Wingate, and R. E.Vincent. An algorithm for computing
fekete points in the triangle. SIAM Journal on Numerical Analysis, 38(5):1707
1720, 2000. 42, 56, 130, 144, 145, 147
H. G. Timmer and J. M. Stern. Computation of global geometric properties of
solid objects. Comput. -Aid. Des., 12:301304, 1980. 6, 11, 171, 203
L. Videla. Explicit integration of the stiffness matrix of a four node plane elastic
finite element. AIAAJ, 6:11741176, 1996. 6, 10
E. L. Wachspress. A rational finite element basis. J. Int. Math. Appl., 8:5781,
1972. 9
E. L. Wachspress. A rational finite element basis. J. Int. Math. Appl., 11:83104,
1973. 9
339
REFERENCES
E. L. Wachspress. A rational finite element basis. Academic Press, San Diego,
1975. 42, 55, 67, 101
S. Wandzura and Xiao. Symmetric quadrature rules on a triangle. Computers
and mathematics with applications, 45:18291840, 2003. 42, 56, 130, 144, 145,
147
G. Ye. Yagawa and S. Yashimara. A numerical integration schemefor finite
element method based on symbolic manipulation. Int J. Numer. Method Engng,
29:153159, 1990. 5, 10
O. C. Zienkiewicz and Y. K. Cheung. Finite elements in the solution of field
problems. The Engineers, 220:507510, 1965. 19
O. C. Zienkiewicz and R. L. Taylor. The finite element method. Maidenhead,
McGraw Hill, U. K., 1989. 20, 42, 55, 67, 101
O. C. Zienkiewicz and R. L. Taylor. The finite element method, volume Vol. 1,
5th Ed. Butterworth, Heinemann. Elsevier Science, Burlington, Mass., 2000. 6,
11, 171, 203
O.C. Zienkiewicz and K. Morgan. The finite element and Approximation. John
wiley and sons, New york, 1983. 5
M. Zlamal. Curved element in finite element method. SIAM journal of Numerical
Analysis, 11(2):347362, 1974. 9
340
List of Publications
1. Hussain Farzana and Karim, M. S. (2010), An Accurate Evaluation Procedure of Triangular Domain Integrals, SUST Studies, Vol. 12(1), 15-20.
(Included in Chapter 3)
2. Hussain Farzana and Karim, M. S. (2010), Evaluation of Triangular Domain
Integrals by Use of Gaussian Quadrature for Square domain Integrals, SUST
Studies, Vol. 12(1), 52-57. (Included in Chapter 4)
3. Hussain Farzana, Karim, M. S. and Rezwan Ahmed (2012), Appropriate
Gaussian Quadrature formulae for Triangles, International Journal of Applied Mathematics and Computation, 2012, Vol. 4(1), 24-38. (Included in
Chapter 5)
4. Hussain Farzana and Karim, M. S. (2012), Accurate Evaluation Schemes
for Triangular Domain Integrals, IOSR Journal of Mechanical and Civil
Engineering, Vol. 2(6), 38-51. (Included in Chapter 6)
5. Hussain Farzana and Karim, M. S. (2013),A Symmetrical Extended Gaussian Quadrature Formula for Evaluation of Triangular Domain Integrals,
Communicated. (Included in Chapter 7)
6. Hussain Farzana and Karim, M. S. (2013), Efficient Gauss Quadrature Rules
for Tetrahedral Domain Integrals., Communicated. (Included in Chapter 8)
7. Hussain Farzana and Karim, M. S. (2013),Extended Gauss Quadrature Rules
for Numerical Integration over a Tetrahedron, Communicated. (Included in
Chapter 9)
341

Ph.D. Thesis of Farzana Hussain

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Ph.D. Thesis of Farzana Hussain

Hochgeladen von

Copyright:

Verfügbare Formate

Efficient Integration Schemes and

Algorithms for Finite Element

Efficient Integration Schemes and Algorithms for Finite Element

Dr. Md. Shajedul Karim

Dr. Md. Shajedul Karim

perfectly satisfactory. But if some part of the boundary is curved,

require more computing time before the application of the weighted

merical integration schemes to compute two dimensional integrals. In

first set of quadrature formulae involves a huge crowding of Gaussian

3 Evaluation of Triangular Domain Integrals by use

6 Accurate Evaluation Schemes for Triangular Domain Integrals

5 Appropriate Gaussian Quadrature Formulae for Triangles

10 Appropriate Gauss Quadrature and Recursive Schemes to Obtain

Finite Element Discretization of an Aircraft. . . . . . . . . . . . .

Connection between elements of Discretized domain.

Triangulation of the domain of integration. . . .

Original and Transformed Standard Triangle. . . . . . . . . . . .

Original and Transformed Unit Triangle. . . . . . . . .

Gauss Points for m =n =15 (119 points) using GQUTM. . . . . .

Transformation of Arbitrary triangle into Standard triangle. . .

Discretization of the Domain of Integration into Triangles. . . . . 131

The Arbitrary Tetrahedron and the Transformed Unit Tetrahedron. 173

Distribution of Gauss Points for GQTTM,(m =5, 75 points). .

Discretization of a Three Dimensional Domain. . . . . . . . . . .

Curved Domain of Example-1 . . . . . . . . . . . . . . . . .

Objectives and Scope

The procedure of structural analysis, which afterward came to be known as the

Figure 1.1: Finite Element Discretization of an Aircraft.

Figure 1.2: Finite Element Discretization of a Bridge.

Plan of The Thesis

Finite Element Method Preliminaries

Some Basics of the Finite Element Method

Figure 2.1: Connection between elements of Discretized domain.

History of the Finite Element Method

The history of finite element method is particularly interesting, especially since

Weighted Residual Method

Finite Element Formulation of Problems

Figure 2.2: Discretized domain in 2D .

Figure 2.3: A single 2D element.

Shape Functions (Interpolation Functions)

in the element, these elements are called isoparametric elements. In addition,

Some Practical Problems

The general field equation

Where Dx , Dy are the permeability of the earth material and represents

Properties of Multivariate Polynomial of n-th Degree

A bivariate polynomial g(x, y) of n-th degree is written as

A trivariate polynomial g(x, y, z) of n-th degree is written as

fmk, kl, l xmk y kl z l

m=0 k=0 l=0

Property 1. For each m 1, if we define

then Hm (x, y) is a homogeneous function in (x, y) of degree m and g(x, y)

0 if p or q or both are odd

(c) The integral of a monomial of the type f (p, q, r) = pm q n rl (where m, n, l are

0 if al least one of p, q, r are odd

(d) The integral of a monomial of the type f (p, q, r) = pm q n rl (where m, n, l are

Gauss Quadrature Formula

The m-points Gauss Quadrature formula for numerical integration of a function

Finite Element Formulation of Physical Problems

Usually, the torsion problem, with simply connected cross-sections is written as

Where C and C constitute the cross section boundaries. (h = 2g)

is the residual function and W is the weighting function.

Where F = C + C , the right hand side of Eq.(2.15) is equated to zero owing

and for which form the Eq.(2.14) we obtain

We now approximate the function, using shape functions as