General

General

The default 0.1 numerical damping value used in transient analysis must have a reason. Is there any test correlation to validate the damping values? Or it’s just a random number the program picks?

    • FAQFAQ
      Participant

      Numerical damping, it is numerical – not physical – so it’s not related to test correlation. Please see Section 5.6.3 “Transient Dynamic Analysis Settings Based on Application” below: https://ansyshelp.ansys.com/account/secured?returnurl=/Views/Secured/corp/v192/ans_str/Hlp_G_STR5_12.html There, one will see different defaults. MAPDL actually defaults to gamma=0.005 while Mechanical defaults to gamma=0.1. The setting for Mechanical is the same as “Moderate Speed Simulation” case. To provide some background, when we have transient structural analyses, we can often get additional ‘noise’ – for example, spurious high-frequency content can be excited. Models may not be set up ‘perfectly’ by the user (not through any fault of the user, but to get an efficient solution, there may be some modeling short-cuts taken, such as using a coarse mesh). Using numerical damping can reduce such ‘noise’. It can also have a different effect – faster convergence. These benefits come at the expense of some numerical dissipation (larger numerical damping means larger numerical dissipation). The defaults in Mechanical of gamma=0.1 are a tradeoff between these effects (pros and cons) for structural simulations. Note that for purely acoustic simulations, it’s a bit different – we don’t have contact/impact in purely acoustic simulations, and everything is always linear. Thus, using no numerical damping (no numerical dissipation) is probably desired for this situation Albert has where amplitude is lower than expected. In short, numerical damping comes about mainly for structural analyses, and defaults are with structural analyses in mind. For purely acoustic simulations, numerical damping can add unwanted numerical dissipation, so user may wish to set it to zero.