Sunteți pe pagina 1din 14

SOLVING FEM EQUATIONS

By
S. Ziaei Rad
Equation Solving
Direct Methods (Gauss Elimination):
• Solution time proportional to N*B^2 (N is the dimension of
the matrix, B the bandwidth)
• Suitable for small to medium problems, or slender
structures (small bandwidth)
• Easy to handle multiple load cases
Iterative Methods:
• Solution time is unknown beforehand
• Reduced storage requirement
• Suitable for large problems, or bulky structures (large
bandwidth, converge faster)
• Need solving again for different load cases
Equation Solving
Gauss Elimination
Step1 : Forward elimination
Suppose that in doing Gauss-Jordan elimination, we at each stage
subtract away rows only below the then-current pivot element.
Then, when we have done this for all the pivots, we will be left
with a reduced equation that looks like this (in the case of a single
right-hand side vector):

Step2: Backward substitution


The procedure defined as follows and called
backsubstitution.
Equation Solving
LU Decomposition
Suppose we are able to write the matrix A as a product of two
matrices, L U = A (1)
where L is lower triangular (has elements only on the diagonal
and below) and U is upper triangular (has elements only on
the diagonal and above). For the case of a 4*4 matrix A,
for example, equation (1) would look like this:

(2)
(3)
Equation Solving
LU Decomposition
Thus, equation (2) can be solved by forward substitution as
follows,

while (3) can then be solved by backsubstitution exactly as in


equations
Equation Solving
Iterative Methods
Iterative Method
The Gauss-Seidel Method
Ax = b (A is symmetric)

Start with an estimate and then iterate using the following:


Equation Solving
In vector form,

where is the diagonal matrix of A,

is the lower triangular matrix of A,


such that
Iterations continue until solution x converges, i.e.

where ε is the tolerance for convergence control.


Computer Resources Req'd
by FEM Solver
Storage & Solution Time for a Fully-Stored Stiffness Matrix

time numbers last adjusted in 1998


Typical Stiffness Matrix
Sparsity Pattern
Survey of Sparse Matrix
Storage Formats
If the coefficient matrix A is sparse, large-scale linear systems of the
form Ax=b can be most efficiently solved if the zero elements of A
are not stored. Sparse storage schemes allocate contiguous storage in
memory for the nonzero elements of the matrix, and perhaps a limited
number of zeros. This, of course, requires a scheme for knowing
where the elements fit into the full matrix.
There are many methods for storing the data:
1- Compressed Row and Column Storage
2- Block Compressed Row Storage
3- Diagonal Storage
4-Jagged Diagonal Storage
5-Skyline Storage
Computer Resources Req'd
by FEM Solver
Storage & Solution Time for a Skyline Stored Stiffness
Matrix Assuming
B= N

time numbers last adjusted in 1998


Skyline Storage

This array is complemented by a (N + 1) integer array p that contains


addresses of diagonal locations. The array has N +1 entries.
The (i +1)th entry of p has the location of the i th diagonal entry of K in s.
For the example matrix:
Skyline Storage
(aka Profile or Variable-Band Storage)
Marking Displacement BCs
Equations for which the displacement component is known
or prescribed are identified by a negative diagonal location
value. For example, if u3 and u5 are prescribed displacement
components in the sample system,

S-ar putea să vă placă și