search

thelfer / MFrontGenericInterfaceSupport

This project aims at providing support for MFront generic behaviours. This project can be embedded in open-source and propriary sofware
35 stars 34 forks source link

Struggling with SpeedOfSound management ... #120

Open pbouda95 opened 11 months ago

pbouda95 commented 11 months ago

Hello,

As a user of the MaterialDataManager class through fortran bindings, I am trying to handle the @SpeedOfSound from the solver point of view.

First, I did not manage to find a crystal clear explanation about its features in the documentation so I looked at the Doxygen (in include/MGIS/Behaviour/Behaviour.hxx). It says that you need to activate a flag trough the first component of the tangent operator array (must be > 50). However, my implementations did not succeed so I am wondering if I am doing something wrong here.

Does anybody have a clue on it ?

Cheers,

thelfer commented 11 months ago

Hi @pbouda95,

It says that you need to activate a flag trough the first component of the tangent operator array (must be > 50).

This is true when using MGIS to integrate a mechanical behaviour on a single integration point.

For several integration points (managed by the MaterialDataManager class), the speed of sound can be computed in C++ and python using overloaded versions of the integrate function which takes a data structure named BehaviourIntegrationOptions as argument (in place of the IntegrationType enumeration type used so for).

Unfortunately, this data structure (BehaviourIntegrationOptions) and the associated overloaded versions of the integrate function are currently not wrapped in C nor in fortran.

pbouda95 commented 11 months ago

Hi @thelfer,

Thanks for the quick answer. Aside from the fact that some C and Fortran wrappers are missing for a few objects/functions, do you see any additional loopholes ? If not, I am pretty much motivated to contribute and write them.

Thanks,

thelfer commented 11 months ago

do you see any additional loopholes

I don't see any :)

pbouda95 commented 11 months ago

Hello @thelfer,

I will provide the missing bindings ASAP. However there is still something which bothers me: What about the way to provide the mass density at the reference configuration ? I did not see any information about that.

One solution might be to push it through an ExternalStateVariable but I am not sure if it is what it was scheduled ...

Thanks,

thelfer commented 11 months ago

You're right. The mass density is not passed right now !

thelfer commented 10 months ago

@pbouda95 branch 121-pass-the-mass-density-from-the-materialdatamanager contains a first implementation that may provide the wanted feature. It seems that a memory fault may happen from time to time, although valgrind does not report anything useful.

So it's mostly a work in progress, but you can start playing with it and consider porting the tests ComputeSpeedOfSoundTest and ComputeSpeedOfSoundTest2 to fortran