Professional Documents
Culture Documents
CONTINUUM PROBLEMS
4.1 INTRODUCTION
Problems in engineering and science fall into two fundamentally different categories,
depending on which point of view we adopt. One point of view is that all matter consists of
discrete particles that retain their identity and nature as they move through space. Their
position at any instant is given by coordinates in some reference system, and these
coordinates are functions of time ─ the only independent variable for any process. This
viewpoint, known as the rigid body viewpoint, is the basis for Newtonian particle mechanics.
The other viewpoint, the one that we shall use, stems from the continuum rather than the
discrete molecular or particle approach to nature. In the continuum viewpoint we say that all
bodies of interest are continuous at all points in space to the extent that all field quantities
describing the state of the body are sufficiently differentiable in the independent variables of
the continuum, space, and time. This allows us to focus our attention on one point in space
and time and observe the phenomena occurring there.
Continuum problems are concerned with the fields of temperature, stress, mass concentration,
displacement and potentials, to name just a few examples. These problems arise from
phenomena in nature that are approximately characterized by partial differential equation and
their boundary conditions.
Continuum problems of mathematical physics are often called boundary value problems
because their solution is sought in some domain defined by a given boundary on which
certain conditions, called boundary conditions, are specified. Except for free boundary
problems, the shape of the boundary and its location are always known. The boundary may
be defined by a curve or a surface in n-dimensional space, and the domain it defines may be
finite or infinite, depending on the extremities of the boundary. The boundary is said to be
closed if conditions affecting the solution of the problem are specified everywhere on the
boundary, and open if part of the boundary extends to infinity and no boundary conditions are
specified at the part at infinity. It is important to note that our definition of a boundary
problem departs from the usual one. The usual definition distinguishes between boundary
value problems and initial value problems, where time is an independent variable.
FEM 86 CHAPTER 4
2. Approximate Solutions
(a) Perturbation
(b) Power series
(c) Probability schemes
(d) Method of weighted residuals
(e) Ritz method
(f) Finite difference techniques
(g) Finite element method
For a few problems it is possible to obtain an exact solution by direct integration of the
differential equation. This is accomplished sometimes by an obvious separation of variables
or by applying a transformation that makes the variables separable and leads to a similarity
solution. Occasionally a Fourier or Laplace transformation of the differential equation leads
to an exact solution. However, the number of problems with exact solutions is severely
limited, and most of these have already been solved. They are the classical problems.
Because regular and singular perturbation methods are applicable primarily when the
nonlinear terms in the equation are small in relation to the linear terms, their usefulness is
limited. The power series method is robust and has been employed with success, but since the
method requires the generation of a coefficient for each term in the series, it is relatively
tedious. Also, it is difficult, if not impossible, to demonstrate that the series converges.
The probability schemes, usually classified under the heading of Monte Carlo methods, are
used for obtaining a statistical estimate of a desired quantity by random sampling. These
methods work best when the desired quantity is a statistical parameter and sampling is done
from a selective population.
With the advent of computers the currently outstanding methods for obtaining approximate
solutions of high accuracy are the method of weighted residuals, the Ritz method, the finite
difference method, and the finite element method.
In almost all approximate methods used to determine the solution of differential and/or
integral equations, we seek a solution in the form
FEM 87 CHAPTER 4
of homogeneous part and a nonhomogeneous part , we require to satisfy the
specified essential boundary conditions while the homogeneous part vanishes at the same
boundary point where the essential boundary condition is specified. This follows from
If all specified essential boundary conditions are homogeneous (i.e., the specified value is
zero), then is taken to be zero and must still satisfy the same conditions,
. Note that the requirement that be zero at the boundary points where the
essential boundary conditions are specified is satisfied by the choice of .
The weighted-residual method is a generalization of the Ritz method in that the weight
functions can be chosen from an independent set of functions, and it requires only the
weighted-integral form to determine the parameters. Since the latter form does not include
any of the specified boundary conditions of the problem, the approximation functions must be
selected such that the approximate solution satisfies both the natural and essential boundary
conditions. In addition, the weight functions can be selected independently of the
FEM 110 CHAPTER 4
approximation functions, but are required to be linearly independent (so that the resulting
algebraic equations are linearly independent).
The method of weighted residuals can be described in its generality by considering the
operator equation
The function is not only required to satisfy the operator equation (7.70), it is also required
to satisfy the boundary conditions associated with the operator equation.
In the weighted residual method, the solution is approximated, in much the same way as in
the Ritz method, by the expression
except that the requirements on and for the weighted-residual method are more
stringent than those for the Ritz method. Substitution of the approximate solution into the
left hand side of equation (7.70) gives a function that, in general, is not equal to the
specified function . The difference , called the residual of the approximation, is
nonzero:
where is a two-dimensional domain and are weight functions, which, in general, are not
the same as the approximation functions . The must be a linearly independent set;
otherwise the equations provided by (7.75) will not be linearly independent and hence will
not be solvable.
The requirements and for the weighted-residual method are different from those for
the Ritz method, which is based on the weak (integral) form of the differential equation. The
differentiability requirement on in the weighted-residual method is dictated by the integral
statement (7.75), as opposed to the weal form in the Ritz method. Thus, must have nonzero
derivatives up to the order appearing in the operator equation (7.70). Since the weighted-
integral form (7.75) does not include any of the specified (either essential or natural)
boundary conditions, we must also require in (7.74) to satisfy all specified boundary
conditions of the problem. Consequently, is required to satisfy the homogeneous form of
all specified boundary conditions of the problem. These requirements on and will
increase the order of the polynomial expressions used for the weighted-residual method. In
general, the used in this method are higher order functions than those used in the Ritz
method, and the functions used in the latter may not satisfy the continuity (i.e.,
differentiability) requirements of the weighted-residual method. Various special cases of the
weighted-residual method are explained below.
or
In general, the Galerkin method is not the same as the Ritz method. This should be clear from
the fact that the former uses the weighted-integral form whereas the latter uses the weak form
to determine the coefficients . Consequently, the approximation functions used in the
Galerkin method are required to be of higher order than those in the Ritz method.
If the equation permits, and one wishes, differentiation from the dependent variable(s) can be
transferred to the weight function , thereby obtaining the weak form. Then there is no
difference the Galerkin method and the Ritz method. Thus, Ritz and Galerkin methods yields
the same solutions in two cases: (a) when the specified boundary conditions of the problem
are all of the essential type, and therefore the requirements on in the two methods become
the same and the weighted-integral form reduces to the weak form; and (b) when the
approximation functions of the Galerkin method are useful in the Ritz method.
or
With this choice of weight functions, the weighted-residual statement (7.75) becomes
or
………. (7.85)
Example 4.1
Find two parameter approximate solution of the equation
Solution
For a weighted residual method, and should satisfy the following conditions:
: satisfy actual boundary conditions.
: satisfy homogeneous form of the specified boundary
conditions.
Selection of :
Let us assume that . Substituting the actual boundary conditions, we get
, and . Hence, we obtain
Selection of :
Let us assume that . Substituting the specified boundary conditions
and assuming , we get , and . Hence, we obtain
Figure E4.6
Use (a) the Galerkin method, (b) the Petrov-Galerkin method, (c) the least squares
method and (d) the point collocation method.