Convergence problem using contact

hoi Dcomment88,

your problem is the MODI_MAILAGE command:

mesh2 = MODI_MAILLAGE(reuse=mesh2,
MAILLAGE=mesh2,
ORIE_PEAU_2D=_F(GROUP_MA=('ground', 'masterU', 'masterD', 'slaveU', 'slaveD', 'load', 'sym_y'),),
INFO=1,)

vous avez utilise le mot clé ORIE_PEAU_2d alors que le problème est 3d. utilisez ORIE_PEAU_3d

I didnot look into detail into the problem, but I guess you have to use ORIE_PEAU_3D=... because your problem, as stated, is 3D (mess file line 473 and further that reoccurs at the end of the file, line 635).

Uhm, assuming you've attached the right files, the error is right there in plain sight: vous avez utilise le mot clé ORIE_PEAU_2d alors que le problème est 3d. utilisez ORIE_PEAU_3d

Use ORIE_PEAU_3d indstead of 2d

/C

Hi Dany,

To help you we need more details: please run a simulation with orie_peau_3d and post the message files for discrete and continue methods.
Non convergence can be caused by many different sources (integration of material law, friction stick slip, contact state oscillation, instabilities, too many constraints, mesh distortions leading to singular tangent matrix.)
Without message file it is impossible to tell what could go wrong.

As an advice, I found the following combination to be usually most robust:
discrete contact with Algo_CONT=LAGRANGIEN , REAC_GEOM=CONTROLE, ALGO_FROT=PENALISATION with E_T=100000
and for non Lin solver:
PREDICTION=ELASTIQUE, MATRICE=TANGENTE, PAS_ELAS_MINI =1E-3, Solveur MUMPS, MATRICE_DISTRIBUEE=oui, PCENT_PIVOT=70 (set mpi_nb_cpu=4 in asyl)

Another route to follow with continue method is to use penalty (start with penalty stiffnesses corresponding to Youngs modulus) for both friction and contact and point fixe for all algorithms.

Good luck and keep us posted.

Joel