6.12 Quasi-Newton solution technique

Products: ABAQUS/Standard  ABAQUS/CAE  

Benefits: The quasi-Newton solution technique has been enhanced in ABAQUS/Standard. This option may offer significant performance advantages over the default Newton method for some nonlinear static and dynamic simulations.

Description: The quasi-Newton solution technique controls the iterative equilibrium search in nonlinear static and dynamic simulations. Computational cost is saved by utilizing an inexpensive stiffness matrix update for many of the iterations, rather than forming and factorizing a new stiffness matrix for every iteration as in the default Newton technique. As a trade-off, quadratic convergence is usually not obtained for quasi-Newton iterations, resulting in relatively more iterations per increment. In large models the stiffness matrix factorization can dominate the cost of an iteration, making the quasi-Newton option an attractive choice. The technique can also be effective in mildly nonlinear static and dynamic analyses, where the stiffness matrix does not change significantly from one increment to the next.

The line search method is now activated automatically in steps where the quasi-Newton method is selected. The line search improves convergence behavior by automatically identifying and scaling back solution corrections that cause divergence. These unproductive corrections can occur more frequently in quasi-Newton iterations because of the approximate nature of the stiffness matrix update.

ABAQUS/CAE Usage: 
Step module:
   StepCreate: Procedure type: Dynamic, Implicit; Geostatic; Heat transfer; Soils; 
      Static, General; or Visco: Continue
   Step editor:   Other: Solution technique: Quasi-Newton
References:

ABAQUS Analysis User's Manual

ABAQUS Keywords Reference Manual

ABAQUS Example Problems Manual

ABAQUS Theory Manual