Hello,
I have some problems using REST_SPEC_TEMP command.
Actually, I want to chain following instructions :
- TRANS_RESU_GENE=DYNA_VIBRA (BASE_CALCUL='GENE', TYPE_CALCUL='TRANS')
- TRANS_RESU=REST_GENE_PHYS (TRANS_RESU_GENE)
- HARMO_RESU=REST_SPEC_TEMP (TRANS_RESU)
It works well when I perform REST_GENE_PHYS on full model.
But If I use GROUP_MA/GROUP_NO in REST_GENE_PHYS (cause I don't need to compute FFT on complete model) , the REST_SPEC_TEMP give results but wrong results (It seems to me like an error on dof numbering, because the result seems to correspond to a dof which is not the one I ask for):
l_inst = DEFI_LIST_REEL(DEBUT=0.0,
INTERVALLE=_F(JUSQU_A=(Nfft-1)/Fs,
PAS=1/Fs))
TransRespGENE = DYNA_VIBRA(AMOR_MODAL=F(AMOR_REDUIT=lamort.getValues(),),
BASE_CALCUL='GENE',
COMPORTEMENT=(F(DIST_1=0.0,
DIST_2=r1_timbre-r2_timbre,
GROUP_NO_1=('NchocMarteau', ),
GROUP_NO_2=('NchocTimbre', ),
NORM_OBST=(0.0, 0.0, 1.0),
OBSTACLE=BICERCINT,
RELATION='DIS_CHOC',
RIGI_NOR=Kchoc),
F(RIGI_NOR=10.0,
AMOR_NOR=0.25,
JEU=0.5e-3,
DIST_1=0,
GROUP_NO_1=('NchocButee', ),
NORM_OBST=(0.0, 0.0, 1.0),
ORIG_OBST=(-0.0128005224746,0.00616148110636,0.00815),
OBSTACLE=CERCLE,
RELATION='DIS_CHOC',),),
ETAT_INIT=F(VITE=VITINI),
INCREMENT=F(INST_FIN=(Nfft-1)/Fs,
INST_INIT=0.0,
PAS=2.0e-6,),
ARCHIVAGE=F(LIST_INST=l_inst),
MATR_MASS=matMge,
MATR_RIGI=matKge,
SCHEMA_TEMPS=_F(SCHEMA='DIFF_CENTRE'),
TYPE_CALCUL='TRAN')
EnvMecaTransResp = REST_GENE_PHYS(GROUP_NO=('GlaceAV', 'Carrure', 'CarrureSupTimb' ),
INTERPOL='LIN',
LIST_INST=l_inst,
#TOUT_ORDRE='OUI',
NOM_CHAM=('VITE',),
RESU_GENE=TransRespGENE)
EnvMecaHarmResp = REST_SPEC_TEMP(INFO=1,
METHODE='PROL_ZERO',
NOM_CHAM=('VITE', ),
N_PUIS=0,
RESULTAT=EnvMecaTransResp,
SYMETRIE='NON')
VZdef_CentreGlaceAV= RECU_FONCTION(GROUP_NO=('CentreGlaceAV', ),
NOM_CHAM='VITE',
NOM_CMP='DZ',
RESULTAT=EnvMecaHarmResp,
TOUT_ORDRE='OUI')
VZdet_CentreGlaceAV= RECU_FONCTION(GROUP_NO=('CentreGlaceAV', ),
NOM_CHAM='VITE',
NOM_CMP='DZ',
RESULTAT=EnvMecaTransResp,
TOUT_ORDRE='OUI')
VZdef2_CentreGlaceAV = CALC_FONCTION(FFT=_F(FONCTION=VZdet_CentreGlaceAV,
METHODE='PROL_ZERO'),)
In this case, VZdef_CentreGlaceAV is different from VZdef2_CentreGlaceAV.
If I comment GROUP_NO in REST_GENE_PHYS, VZdef_CentreGlaceAV become the same result as VZdef2_CentreGlaceAV.
So the solution should be to apply REST_GENE_PHYS on complete model, without GROUP_MA/GROUP_NO restriction. But the problem is that my model is quite big (in terms of dof and time/frequency sampling) and whithout using the GROUP_MA/GROUP_NO restriction, the calculation (especially REST_SPEC_TEMP) becomes too expensive in memory and calculation time !
So I've tried several things to get around this problem especially to use REST_SPEC_TEMP on full RESU_GENE (I guess it is less computing expensive on GENE than on PHYS).
But I didn't succeed, REST_SPEC_TEMP lead to a SIGSEGV error :
HarmRespGene = REST_SPEC_TEMP(ACCELERATION_MPI='OUI',
INFO=1,
METHODE='PROL_ZERO',
N_PUIS=0,
RESU_GENE=TransRespGENE,
SYMETRIE='NON',
TOUT_CHAM='OUI')
SIGSEGV
Traceback returned by pdbhelp:
[0] print_trace_
[1] seh_filter_exe
[2] ???
[3] _C_specific_handler
[4] chkstk
[5] RtlRaiseException
[6] KiUserExceptionDispatcher
[7] mdarch
[8] ecresu_
[9] op0181_
[10] ex0000_
[11] execop_
[12] expass_
[13] Z9call_operRN8pybind116objectEi
[14] ZZN8pybind1112cpp_function10initializeIRPFvRNS_6objectEiEvJS3_iEJNS_4nameENS_5scopeENS_7siblingEA205_cNS_3argESB_EEEvOT_PFT0_DpT1_EDpRKT2_ENUlRNS_6detail13function_callEE1_4_FUNESP_
[15] ZN8pybind1112cpp_function10dispatcherEP7_objectS2_S2_
[16] PyMethodDef_RawFastCallKeywords
[17] PyMethodDef_RawFastCallKeywords
[18] PyEval_EvalFrameDefault
[19] PyMethodDef_RawFastCallKeywords
[20] PyEval_EvalFrameDefault
[21] PyMethodDef_RawFastCallKeywords
[22] PyEval_EvalFrameDefault
[23] PyEval_EvalCodeWithName
[24] PyFunction_FastCallDict
EXECUTION_CODE_ASTER_EXIT_8718=0
I also tried to use EXTR_RESU with RESTREINT option before apply REST_SPEC_TEMP but it did also fail.
So what is the solution or what am I doing wrong ? How to pass properly REST_SPEC_TEMP on a restricted area of the model and/or on a full RESU_GENE ?
Last but not least, I use code_aster V16.5.0 with the windows version of salome_meca. I think that my problems may come from there, I mean from the Windows version...
Best regards,
Cyrille Dodard