Bandwidth Optimisation |
 
|
Bandwidth Optimisation |
|
Finite element matrices are typically symmetric and sparsely populated. The symmetric nature of these matrices means that the entire matrix need not be stored. Computational efficiency may be increased significantly if the matrix is “banded”, meaning that non-zero terms are located close to the matrix diagonal, as the number of equations to be solved can be significantly reduced. Bandwidth optimisation essentially involves an internal renumbering of the finite element model, such that the original sparse matrix is transformed into an equivalent one which is heavily banded.
Flexcom provides an option to specify whether or not bandwidth optimisation is to be performed. This option is only relevant for non-restart analyses, for which bandwidth optimisation is never performed, as the solution profile (whether optimised by bandwidth optimisation or not) is taken from the preceding analysis. For initial (non-restart) analyses, bandwidth optimisation is performed by default, and this is recommended for the vast majority of models. Very occasionally, particularly if your model is quite complex, the actual optimisation can take a long time. If you want to quickly see results from, say, an initial static analysis, you can use the bandwidth disabling option to suppress the optimisation, but that would be an unusual move.
•*NO OPTIMISE is used to specify whether or not bandwidth optimisation is to be performed.