Cancel
Showing results for 
Search instead for 
Did you mean: 

Model check/contact constraints compatibility

N/A

Hello,
I have a question about FE modelling in Nastran. Our customers always ask us as
requirement to provide FE models that pass several mathematical checks, described
also in Nastran user's manual. A condition to pass these checks is to avoid using
rigid elements (RBE) unless their nodes are coincident. I saw that I cannot use
contact conditions either, because if I apply them I don't pass checks any longer.
Is there any method to use contacts (and rigids in general) but passing mathematical
checks?
Thanks and Merry Christmas to everyone!
Marco
4 REPLIES

Re: Model check/contact constraints compatibility

N/A

Hello Marco,
I will separate my response for rigid elements and contact ....
Rigid elements (RBE)
It is perfectly acceptable for the connected nodes to not be coincident. Therefore
we do not have any model verifications that check for this condition - otherwise
we would warn or fatal on valid conditions. However I would agree that there may
be some modeling situations where you want the RBE nodes to be coincident. In this
type of situation where the modeling approach is acceptable in some cases but not
others, or is just not a preferred approach, the ideal would be to provide some
sort of user customizable set of model verifications. We have this to some degree
for element qualtiy checks, but there could be a whole host of other checks that
could be implemented too - like your mention - that we don't have anything for.
But we will keep it in mind for future enhancements. Alternatively, we know that
some customers have written their own routines to read an input file and do their
specific model verifications and that could be done in this case as well.
Contact
I am afraid I do not follow what the problem is with contact. It sounds like you
are setting up a contact condition, but one of our model checks is saying that it
is invalid. Is that correct? If so could you be more specific on what the error
message is and where it occurs. Or even send along the model if that is ok and it
is not too large.

Mark








"Marco" wrote:
>
>Hello,
>I have a question about FE modelling in Nastran. Our customers always ask
>us as
>requirement to provide FE models that pass several mathematical checks, described
>also in Nastran user's manual. A condition to pass these checks is to avoid
>using
>rigid elements (RBE) unless their nodes are coincident. I saw that I cannot
>use
>contact conditions either, because if I apply them I don't pass checks any
>longer.
>Is there any method to use contacts (and rigids in general) but passing mathematical
>checks?
>
>Thanks and Merry Christmas to everyone!
>
>Marco

Re: Model check/contact constraints compatibility

N/A

Dear Mark,
I'm sorry if my answer was too generic, I'll try to be more specific as requested.
One of the checks my model has to pass is the thermal expansion elastic stress check,
where all the parts of the model have to be set to a single fictitious material
with E 10^11 Pa, ni=0.3 and a coefficient of thermal expansion 10^-5 K^-1. Imposing
a delta temperature of 100 K, I have to get as result a max. Von Mises stress of
100 Pa and a max nodal rotation of 10^-4 rad everywhere.
Now, from my experiente this conditions cannot be guaranteed if for example I use
rigid elements with non coincident nodes. The same happens if I apply any contact
condition (glued or contact). How can I model a contact passing at the same time
this check? If you want I can build a little model and send it.
Greetings,
Marco

"Mark Donley" wrote:
>
>Hello Marco,
>
>I will separate my response for rigid elements and contact ....
>
>Rigid elements (RBE)
>It is perfectly acceptable for the connected nodes to not be coincident. Therefore
>we do not have any model verifications that check for this condition - otherwise
>we would warn or fatal on valid conditions. However I would agree that there
>may
>be some modeling situations where you want the RBE nodes to be coincident.
>In this
>type of situation where the modeling approach is acceptable in some cases
>but not
>others, or is just not a preferred approach, the ideal would be to provide
>some
>sort of user customizable set of model verifications. We have this to some
>degree
>for element qualtiy checks, but there could be a whole host of other checks
>that
>could be implemented too - like your mention - that we don't have anything
>for.
>But we will keep it in mind for future enhancements. Alternatively, we know
>that
>some customers have written their own routines to read an input file and do
>their
>specific model verifications and that could be done in this case as well.
>
>Contact
>I am afraid I do not follow what the problem is with contact. It sounds like
>you
>are setting up a contact condition, but one of our model checks is saying
>that it
>is invalid. Is that correct? If so could you be more specific on what the
>error
>message is and where it occurs. Or even send along the model if that is ok
>and it
>is not too large.
>
>
>Mark
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>"Marco" wrote:
>>
>>Hello,
>>I have a question about FE modelling in Nastran. Our customers always ask
>>us as
>>requirement to provide FE models that pass several mathematical checks, described
>>also in Nastran user's manual. A condition to pass these checks is to avoid
>>using
>>rigid elements (RBE) unless their nodes are coincident. I saw that I cannot
>>use
>>contact conditions either, because if I apply them I don't pass checks any
>>longer.
>>Is there any method to use contacts (and rigids in general) but passing mathematical
>>checks?
>>
>>Thanks and Merry Christmas to everyone!
>>
>>Marco

>

Re: Model check/contact constraints compatibility

N/A

Hello Marco,
I assume that this test is to make sure that you have no grounding of your model
and that it will have unconstrained thermal distortion. You do not mention boundary
conditions, but they would need to be such as to allow unconstrained thermal distortion.
As far as rigid elements, we implemented an option that allows them to expand under
a thermal load. You need to use the LAGRANGE option for the RIGID case control and
define a CTE on the RBE2, RBAR, or RBE3. Then it should not matter if your nodes
on the rigid element are coicident or not.
Glue connections should also not cause a problem. They should be free to grow unconstrained
also.
But I don't understand what you are doing with contact. I can't envision a physical
situation where you allow unconstrained thermal distortion and have contact at the
same time. Please send a simple model if that explains better.
Mark


"Marco" wrote:
>
>Dear Mark,
>
>I'm sorry if my answer was too generic, I'll try to be more specific as requested.
>One of the checks my model has to pass is the thermal expansion elastic stress
>check,
>where all the parts of the model have to be set to a single fictitious material
>with E 10^11 Pa, ni=0.3 and a coefficient of thermal expansion 10^-5 K^-1.
>Imposing
>a delta temperature of 100 K, I have to get as result a max. Von Mises stress
>of
>100 Pa and a max nodal rotation of 10^-4 rad everywhere.
>
>Now, from my experiente this conditions cannot be guaranteed if for example
>I use
>rigid elements with non coincident nodes. The same happens if I apply any
>contact
>condition (glued or contact). How can I model a contact passing at the same
>time
>this check? If you want I can build a little model and send it.
>
>Greetings,
>
>Marco
>
>
>
>"Mark Donley" wrote:
>>
>>Hello Marco,
>>
>>I will separate my response for rigid elements and contact ....
>>
>>Rigid elements (RBE)
>>It is perfectly acceptable for the connected nodes to not be coincident.

>Therefore
>>we do not have any model verifications that check for this condition - otherwise
>>we would warn or fatal on valid conditions. However I would agree that there
>>may
>>be some modeling situations where you want the RBE nodes to be coincident.
>>In this
>>type of situation where the modeling approach is acceptable in some cases
>>but not
>>others, or is just not a preferred approach, the ideal would be to provide
>>some
>>sort of user customizable set of model verifications. We have this to some
>>degree
>>for element qualtiy checks, but there could be a whole host of other checks
>>that
>>could be implemented too - like your mention - that we don't have anything
>>for.
>>But we will keep it in mind for future enhancements. Alternatively, we know
>>that
>>some customers have written their own routines to read an input file and

>do
>>their
>>specific model verifications and that could be done in this case as well.
>>
>>Contact
>>I am afraid I do not follow what the problem is with contact. It sounds like
>>you
>>are setting up a contact condition, but one of our model checks is saying
>>that it
>>is invalid. Is that correct? If so could you be more specific on what the
>>error
>>message is and where it occurs. Or even send along the model if that is ok
>>and it
>>is not too large.
>>
>>
>>Mark
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>"Marco" wrote:
>>>
>>>Hello,
>>>I have a question about FE modelling in Nastran. Our customers always ask
>>>us as
>>>requirement to provide FE models that pass several mathematical checks,

>described
>>>also in Nastran user's manual. A condition to pass these checks is to avoid
>>>using
>>>rigid elements (RBE) unless their nodes are coincident. I saw that I cannot
>>>use
>>>contact conditions either, because if I apply them I don't pass checks any
>>>longer.
>>>Is there any method to use contacts (and rigids in general) but passing

>mathematical
>>>checks?
>>>
>>>Thanks and Merry Christmas to everyone!
>>>
>>>Marco

>>
>

Re: Model check/contact constraints compatibility

Creator
Creator

Hi Mark,

 

I have the same problem as Marco had.

I need to do the thermal check and show that isothermal expansion show no stresses. I did a simple example shown in the picture attached. I used a shell plate and connected the plate to a beam with RBE2 elements. I turned on the LAGRANGE mode in Rigid case control and in the displacement post-processing it looks like everything is expanding consistently. But then the von-mises stresses show quite high stresses at the elements connected to RBE2 nodes.

Do you know if I need to change anything else in the settings in order to have no stresses during expansion?

 

Jintin