Cancel
Showing results for 
Search instead for 
Did you mean: 

rigid body modes with linear contacts

N/A

Hello,
I'm trying to perform a check on a FEM, to see if a modal analysis without constraints
produces the first 6 rigid body modes with a frequency < 0.001 Hz (this is a requirement).
This FEM contains linear contacts, part glued and part contact properties. I cannot
obtain 6 modes within requirements: if I use contact properties I get 3 rigid body
modes only, and if I set all contacts as glued I get 6 rigid body modes but the
ones from 4 to 6 are in the order of magnitude of 1E-1 Hz. What is the reason?
Thanks
Marco

5 REPLIES

Re: rigid body modes with linear contacts

N/A

Dear Vernon,
I'm using NX.Nastran 6.1 included in FEMAP 10.0.2. Thanks for the advice, I'll try
to change contact parameters, usually I leave them at default values given by FEMAP.
It's been a problem using contacts, because I always have to provide math check
results, and they are often not good with contacts, especially for rigid body modes
frequencies. Usually I perform modal analysis setting all contacts as glued.
Regards
Marco
"Vernon McKenzie" wrote:
>Marco,
>
>The modes performance of NX/Nastran is sensitive to the glue parameters (the
>
>normal and tangetial penalty factors). You will need to experiment with
>these to produce factors which are satisfactory for both structural and
>modes performance. Which version of NX Nastran are you using? Also, if you
>
>have contacts, are you expecting the modes when the parts are in the
>contacted or non-contacted consdiition?
>
>Regards,
>
>Vernon McKenzie.
>
>"Marco" wrote in message news:4ab34741$1@bbsnotes.ugs.com...
>>
>> Hello,
>>
>> I'm trying to perform a check on a FEM, to see if a modal analysis without

>
>> constraints
>> produces the first 6 rigid body modes with a frequency < 0.001 Hz (this

>is
>> a requirement).
>> This FEM contains linear contacts, part glued and part contact properties.

>
>> I cannot
>> obtain 6 modes within requirements: if I use contact properties I get 3

>
>> rigid body
>> modes only, and if I set all contacts as glued I get 6 rigid body modes

>
>> but the
>> ones from 4 to 6 are in the order of magnitude of 1E-1 Hz. What is the
>> reason?
>>
>> Thanks
>>
>> Marco
>>
>>
>>

>
>

Re: rigid body modes with linear contacts

N/A
Marco,
The modes performance of NX/Nastran is sensitive to the glue parameters (the
normal and tangetial penalty factors). You will need to experiment with
these to produce factors which are satisfactory for both structural and
modes performance. Which version of NX Nastran are you using? Also, if you
have contacts, are you expecting the modes when the parts are in the
contacted or non-contacted consdiition?
Regards,
Vernon McKenzie.
"Marco" wrote in message news:4ab34741$1@bbsnotes.ugs.com...
>
> Hello,
>
> I'm trying to perform a check on a FEM, to see if a modal analysis without
> constraints
> produces the first 6 rigid body modes with a frequency < 0.001 Hz (this is
> a requirement).
> This FEM contains linear contacts, part glued and part contact properties.
> I cannot
> obtain 6 modes within requirements: if I use contact properties I get 3
> rigid body
> modes only, and if I set all contacts as glued I get 6 rigid body modes
> but the
> ones from 4 to 6 are in the order of magnitude of 1E-1 Hz. What is the
> reason?
>
> Thanks
>
> Marco
>
>
>


Re: rigid body modes with linear contacts

N/A
Dear Marco,
I can give you only my opinion, only nx nastran developpers knows the reason. According your comments one point could be in the stabilization methods of the stiffness matrix used by the surface-to-surface contact algorithms. Fortunately the user can control the contact parameters, edit the SOL103 solution > Case Control > there you have both the "Contact Parameters" and "Glue Parameters" modelling objects (see image attached). You will have to play with "penalty" values and see how different values affect to the rigid body frequency results.

Also, I strongly suggest to avoid the use of Glue contact when possible, the work-around here is to use the classical method of merging nodes (making geometry coincident and performing splitting), don't pass the "brown" to the nx nastran solver if a solution to the problem at mesh level is available. Also, you will pay a ticked in time during nx nastran solving refining source region mesh, etc...
Best regards,
Blas.
--
~~~~~~~~~~~~~~~~~~~~~~
Blas Molero Hidalgo
Ingeniero Industrial
Director

IBERISA
Edificio Ercilla
Rodríguez Arias 23, 3º - Dpto. 19
48011 BILBAO (SPAIN)
Tel. (+34) 94 410 65 50
Fax. (+34) 94 470 26 34
E-mail: info@iberisa.com
WEB: http://www.iberisa.com
"Marco" escribió en el mensaje de noticias news:4ab34741$1@bbsnotes.ugs.com...
>
> Hello,
>
> I'm trying to perform a check on a FEM, to see if a modal analysis without constraints
> produces the first 6 rigid body modes with a frequency < 0.001 Hz (this is a requirement).
> This FEM contains linear contacts, part glued and part contact properties. I cannot
> obtain 6 modes within requirements: if I use contact properties I get 3 rigid body
> modes only, and if I set all contacts as glued I get 6 rigid body modes but the
> ones from 4 to 6 are in the order of magnitude of 1E-1 Hz. What is the reason?
>
> Thanks
>
> Marco
>
>
>

Re: rigid body modes with linear contacts

N/A

Dear Blas,
thanks for the answer. I agree with you for glued contacts, I set them just to spend
less time but then this advantage is overcome by the problems in analysis and post-processing...
Anyway I'm trying to change parameters (penalty factors) but frequency results are
always identical.
Marco

"Blas Molero Hidalgo" wrote:
>
>
>
>------=_NextPart_001_0025_01CA3854.6E671C60
>
>Dear Marco,
>I can give you only my opinion, only nx nastran developpers knows the =
>reason. According your comments one point could be in the stabilization =
>methods of the stiffness matrix used by the surface-to-surface contact =
>algorithms. Fortunately the user can control the contact parameters, =
>edit the SOL103 solution > Case Control > there you have both the =
>"Contact Parameters" and "Glue Parameters" modelling objects (see image =
>attached). You will have to play with "penalty" values and see how =
>different values affect to the rigid body frequency results.
>
>
>
>Also, I strongly suggest to avoid the use of Glue contact when possible, =
>the work-around here is to use the classical method of merging nodes =
>(making geometry coincident and performing splitting), don't pass the =
>"brown" to the nx nastran solver if a solution to the problem at mesh =
>level is available. Also, you will pay a ticked in time during nx =
>nastran solving refining source region mesh, etc...
>
>Best regards,
>Blas.
>
>--=20
>~~~~~~~~~~~~~~~~~~~~~~
>Blas Molero Hidalgo
>Ingeniero Industrial
>Director
>=20
>IBERISA
>Edificio Ercilla
>Rodr=EDguez Arias 23, 3=BA - Dpto. 19
>48011 BILBAO (SPAIN)
>Tel. (+34) 94 410 65 50
>Fax. (+34) 94 470 26 34
>E-mail: info@iberisa.com
>WEB: http://www.iberisa.com
>
>"Marco" escribi=F3 en el mensaje de noticias =
>news:4ab34741$1@bbsnotes.ugs.com...
>>=20
>> Hello,
>>=20
>> I'm trying to perform a check on a FEM, to see if a modal analysis =

>without constraints
>> produces the first 6 rigid body modes with a frequency < 0.001 Hz =

>(this is a requirement).
>> This FEM contains linear contacts, part glued and part contact =

>properties. I cannot
>> obtain 6 modes within requirements: if I use contact properties I get =

>3 rigid body
>> modes only, and if I set all contacts as glued I get 6 rigid body =

>modes but the
>> ones from 4 to 6 are in the order of magnitude of 1E-1 Hz. What is the =

>reason?
>>=20
>> Thanks
>>=20
>> Marco
>>=20
>>=20
>>

>------=_NextPart_001_0025_01CA3854.6E671C60
>
>
>
><meta http-equiv="3DContent-Type" content="3D"text/html;" />>charset=3Diso-8859-1">
><meta content="3D"MSHTML" 6.00.6001.18294="" />
>
>
>
>
Dear Marco,

>
I can give you only my opinion, only =
>nx nastran=20
>developpers knows the reason. According your comments one point =
>could be in=20
>the stabilization methods of the stiffness matrix used by the =
>surface-to-surface=20
>contact algorithms. Fortunately the user can control the contact =
>parameters,=20
>edit the SOL103 solution > Case Control > there you have both the =
>"Contact=20
>Parameters" and "Glue Parameters" modelling objects (see image =
>attached). You=20
>will have to play with "penalty" values and see how different values =
>affect to=20
>the rigid body frequency results.

>
 

>

>
 

>
Also, I strongly suggest to avoid the =
>use of Glue=20
>contact when possible, the work-around here is to use the classical =
>method of=20
>merging nodes (making geometry coincident and performing splitting), =
>don't pass=20
>the "brown" to the nx nastran solver if a solution to the problem at =
>mesh level=20
>is available. Also, you will pay a ticked in time during nx nastran =
>solving=20
>refining source region mesh, etc...

>
 

>
Best regards,

>
Blas.

>

>size=3D2>--=20
>
~~~~~~~~~~~~~~~~~~~~~~
Blas Molero Hidalgo
Ingeniero=20
>Industrial
Director
 
IBERISA
Edificio =
>Ercilla
Rodr=EDguez=20
>Arias 23, 3=BA - Dpto. 19
48011 BILBAO (SPAIN)
Tel. (+34) 94 410 =
>65=20
>50
Fax. (+34) 94 470 26 34
E-mail:
>href=3D"mailto:info@iberisa.com">>size=3D2>info@iberisa.com
>size=3D2>WEB: >href=3D"http://www.iberisa.com">>size=3D2>http://www.iberisa.com

>
 

>>
> =
>Hello,
>=20
>
> I'm trying to perform a check on a FEM, to see if a modal =
>analysis=20
>without constraints
> produces the first 6 rigid body modes with a =
>
>frequency < 0.001 Hz (this is a requirement).
> This FEM =
>contains=20
>linear contacts, part glued and part contact properties. I =
>cannot
> obtain=20
>6 modes within requirements: if I use contact properties I get 3 rigid=20
>body
> modes only, and if I set all contacts as glued I get 6 =
>rigid body=20
>modes but the
> ones from 4 to 6 are in the order of magnitude of =
>1E-1 Hz.=20
>What is the reason?
>
> Thanks
>
> =
>Marco
>=20
>
>
>

>
>------=_NextPart_001_0025_01CA3854.6E671C60--
>

Re: rigid body modes with linear contacts

N/A

Marco,
If the two faces that you are gluing together are not coincident, the original glue
formulation can lead to artificial rotational energy since it is similar in concept
to a CELASi element created between two non-coincident grids.
NX Nastran 6.1 introduces a new "weld like" glue algorithm that models a glue connection
more like a CWELD element so that it can transfer moment in situations where the
two faces are not coincident.
See the "Surface to Surface Glue Enhancements" chapter in the NX Nastran 6.1 Release
Guide. Be sure to set GLUETYPE=2 to use this method.
Regards,
Jim

"Marco" wrote:
>
>Dear Vernon,
>
>I'm using NX.Nastran 6.1 included in FEMAP 10.0.2. Thanks for the advice,
>I'll try
>to change contact parameters, usually I leave them at default values given
>by FEMAP.
>It's been a problem using contacts, because I always have to provide math
>check
>results, and they are often not good with contacts, especially for rigid body
>modes
>frequencies. Usually I perform modal analysis setting all contacts as glued.
>
>Regards
>
>Marco
>
>"Vernon McKenzie" wrote:
>>Marco,
>>
>>The modes performance of NX/Nastran is sensitive to the glue parameters (the
>>
>>normal and tangetial penalty factors). You will need to experiment with

>
>>these to produce factors which are satisfactory for both structural and
>>modes performance. Which version of NX Nastran are you using? Also, if

>you
>>
>>have contacts, are you expecting the modes when the parts are in the
>>contacted or non-contacted consdiition?
>>
>>Regards,
>>
>>Vernon McKenzie.
>>
>>"Marco" wrote in message news:4ab34741$1@bbsnotes.ugs.com...
>>>
>>> Hello,
>>>
>>> I'm trying to perform a check on a FEM, to see if a modal analysis without

>>
>>> constraints
>>> produces the first 6 rigid body modes with a frequency < 0.001 Hz (this

>>is
>>> a requirement).
>>> This FEM contains linear contacts, part glued and part contact properties.

>>
>>> I cannot
>>> obtain 6 modes within requirements: if I use contact properties I get 3

>>
>>> rigid body
>>> modes only, and if I set all contacts as glued I get 6 rigid body modes

>>
>>> but the
>>> ones from 4 to 6 are in the order of magnitude of 1E-1 Hz. What is the

>
>>> reason?
>>>
>>> Thanks
>>>
>>> Marco
>>>
>>>
>>>

>>
>>
>
>
>