Share
 
 

Iterative Solvers: Best Practices

Methods of iteration preconditioning

Apply the following methods of iteration preconditioning when NoRHS > 2 or nModes > 5.

  • If checked, use Multilevel with ICCF preconditioning
  • If not checked, use Non-multilevel with ICCF preconditioning

NoRHS denotes a number of load cases in linear statics. nModes is the number of eigenmodes for modal analysis.

  • Apply the ICCF approach if available RAM is sufficient.
  • Apply the Vinget regularization for a trial solution when an analysis is performed for the first time after model preparation.
  • If none of the iterative methods show satisfactory convergence, it indicates that an ill-conditioned problem has been met. Usually this occurs due to errors in the generated calculation model, such as unrestricted structure parts, unreal geometrical or physical parameters, incorrect units, or ill mesh.

To Reduce the Calculation Time

To Reduce the Calculation Time with Iterative Solvers

  • Reduce the number of aggregation levels.
  • Increase the number of internal iterations.
  • Enable the Analysis of the diagonal option.
  • Implement Method 1 or preferably Method 2 in cases of more than one level of aggregation.

You can lock preconditioning for a multilevel method leading to slow convergence. If the number of iterations exceeds the bound value bv=40*lg(tol) and the desired tolerance is not obtained, it indicates slow convergence for multilevel methods. This criterion is incorrect for non-multilevel methods. To remedy such cases:

  • Increase the number of internal iterations to 8
  • Reduce the number of aggregation levels, but do not exceed the top value for the coarsest level
  • Select a ICCF or another non-multilevel method

See also:

Iterative Solver Parameter Dialog

Available Solvers

Was this information helpful?