HHT Solver, Hmin Error

Hi All,

I am running the HHT solver for an ADAMS/View model. The simulation errors because the minimum step size is reached. Up to that point the step size is reasonable around 1E-5 / 1E-6 but then hits the minimum.

Reducing Hmin does not solve the issue as the solver reaches the new Hmin value and errors.

The only work around I have found is adjusting the solver error. I've never encountered this with GSTIFF. The model is contact heavy.

Does anyone know what causes this with HHT? Any suggestions to avoid this?

joe.

0 Alexis Del Rio over 1 year ago

Hi Joe,

Have you tried running the model up to a fraction of a second before it fails?

How do you define the contacts, are you using parasolids, native parts or something else?

Thanks,

Alexis
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 jsph_inc over 1 year ago

Hi Alexis,

Thanks for the suggestion. i'll try running the model just before it fails...what will that help with?

Contact is defined using parasolids. Curiously when I substitute 3 of the parasolids with native solids (washer and two bushings) the model is even more unstable and the contact forces are noisier (counter to what I was expecting according to MSC contact presentations).
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Alexis Del Rio over 1 year ago

Ji Joe,

Stopping just a fraction before, most times it shows you something gets out of hand hence the failure.

Native solids getting out of hand may be due to the faceting_tolerance that you are using (possibly default??). This is heavily dependant on what you are trying to represent, but I noticed in the past that if one surface is very complex there may be gaps in the faceting (even if it is considered a solid). In my case this made the contact to reverse and the parts were jammed together, very weird I know.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Jesper Slattengren over 1 year ago

Hi Joe, which version of Adams are you using? In 2023.2 (I think) adaptivity was introduced in the C++ solver.

Setting the parameter ADAPTIVITY=1e-9 (a parameter under the INTEGRATOR statement) typically prevents this from happening without any ill effects.

Adaptivity defaults to 100*HMIN but as HMIN is so small for HHT, it barely helps without explicitly setting it to something meaningful (like typical_timestep/1000).
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 jsph_inc over 1 year ago

Hi JSlat,

I'm using ADAMS 2022.2. Thanks for the suggestion I was not aware of this. I will look into this feature and see if it is available for ADAMS 2022.2 or if I need to upgrade versions.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Jesper Slattengren over 1 year ago

No, it is not available in any 2022 version. Upgrade to 2023.3 which is more stable and have improved solver features.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Jesper Slattengren over 1 year ago

It is available in the Fortran solver in earlier versions, but with a different implementation that frankly doesn't work very well.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 jsph_inc over 1 year ago

OK will do. Thanks again for the help.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel