6. Conclusion

This document presents some general rules to facilitate the use of transient dynamic methods for simulating nonlinear systems: see [bib3] _, [biB6] _, [bib8] _, [bib8] _.

The first step is the adaptation of the model to dynamic analysis methods. The main aim is to ensure the good consistency of the conditions imposed, the correct sampling of the excitations, the correct definition of density and global damping (Rayleigh, modal); it is recommended to start with a linear transient analysis.

Then, it is recommended to start by using an implicit schema (« DYNA_NON_LINE » with a time schema like « NEWMARK » for relatively regular problems or « HHT », and « THETA_SCHEMA » for problems with collisions). Indeed, the implicit time schemes are the most efficient and the most general in Code_Aster in so-called « slow » dynamics.

Finally, for certain applications such as fast dynamics, calculation on a modal basis or certain cases of calculations in slow evolution (cf. [U2.04.07]), the user has the possibility of using explicit time diagrams. The performance in time CPU of Code_Aster in explicit is quite low, compared to a dedicated code like Europlexus [bib2] _. Moreover, all the functionalities available implicitly are not available explicitly (such as, for example, for contact where only penalization is authorized).

In order to qualify the quality of the numerical solution obtained, it is essential to conduct certain parametric studies:

• as for quasistatic calculations, by playing on spatial discretization,

• by testing different amortization settings,

• by testing different time steps,

• by testing different time schemes, initially non-dissipative and if necessary with digital dissipation at high frequencies.

To validate this choice of parameters, it is relevant to analyze the displacement and acceleration responses, as well as to use energy balance functionalities.