We are left with a matrix inversion problem. The matrix
has to be inverted. The Z matrix groups together the degrees of
freedom for the boundary mode shapes for all the elements. The
remaining interior degrees of freedom are collected in one group per
element. The matrix now has the form shown in figure 4.18.
![]() |
We can take advantage of the bandedness of the A boundary-boundary matrix, and the decoupled C matrix. The system can be rewritten as:
![\begin{displaymath}
\left[\begin{array}
{cc} A-B C^{-1} B^T & 0 \ B^T & C \end...
...begin{array}
{c} f_b - B C^{-1} f_i \ f_i \end{array}\right] \end{displaymath}](img172.gif)
where
are an array of the unknown global boundary coefficients and
are the unknown local interior coefficients.