Next: Overview of the Simplex
Up: Solving Structural Mechanics Problems
Previous: Introduction
The key underlying idea is to find a finite-dimensional embedding
for the global equilibrium path. Commonly the global path is embedded
into an infinite-dimensional function space, which is theoretically
perfect but practically useless. We appeal to the Uniqueness Theorem
of Peano [2] stating that trajectories of
explicit ODEs satisfying the Lifschitz condition
are uniquely determined by the Cauchy-type initial conditions.
Although our ODEs do satisfy these requirements, Peano's
result can not be applied immediately since we are interested
in the solution of BVPs instead of IVPs. However, let us consider
that the boundary conditions prescribe Cauchy-type conditions
on both ends. Let us regard only those BC which apply at the
origin (s=0). These conditions will fix some of the initial
conditions. The space spanned by the remaining initial
conditions serves us as a finite-dimensional embedding space
for the global equilibrium path: The map from the solution set
of the BVP onto this space is unique. Our method is based on
the piece-wise linearization of this finite-dimensional space.
The points of the n-dimensional space corresponding to BVP
solutions (this is the point set we are looking for) is specified by n-1
scalar conditions of the form
,
where the
-s denote the variable initial conditions at
s=0 and the
-s the conditions imposed at the far end (s=1).
We illustrate these ideas on a simple example. Let us regard the
ODE describing the static equilibrium of a slender, linearly elastic beam.
This equation was first described by Euler :
where EI denotes the (constant) bending stiffness of the beam,
P denotes the load parameter (compressive force applied at the end),
denotes the angle between the bar axis and the line of
application of P and ' denotes differentiation with respect
to the arc-length.
The IVP of this ODE is completely determined by the initial conditions
. If we choose the boundary conditions
to be
(cantilever beam) then the embedding
space for the global equilibrium path is spanned by
.
There are more complicated cases where this space is more than 2
dimensional, this depends on the ODE and the boundary conditions.
Since n=2, we have a single condition to meet:
.
Next: Overview of the Simplex
Up: Solving Structural Mechanics Problems
Previous: Introduction
Szeberenyi Imre
Fri Apr 26 22:27:24 METDST 1996