Welcome to the forums. Please post in English or French.

You are not logged in.

#1 2019-07-16 09:10:47

pironman
Member
Registered: 2018-05-15
Posts: 50

Strange behaviour of K_TR_D_N

3 weeks ago I posted a topic about the usage of this element. I doing some simulation, but I'm not convinced of the correctness of that usage, thus I made a test simulation (files attached).
I want to relax the rotational degree of freedom of the end of a beam, by using that element connected to the face of the beam with LIAISON_SOLIDE.

1st analysis: The element have 10.0E9 stiffness on all 6 degree of freedom, and I locked all 6 d.o.f. with DDL_IMPO.

01.png

As expected, the beam is full-locked.


2nd analysis: Removed the DRY d.o.f. in DDL_IMPO.

02.png

No differences compared to before, but should be rotate!


3rd Analysis: Added again DRY = 0 in DDL_IMPO, but stiffness along that d.o.f., in AFFE_CARA_ELEM, now is set to 0.0

03.png

Again, no differences compared to the first analysis, but in my opinion should rotate.


4th Analysis:  Removed the DRY d.o.f. in DDL_IMPO and stiffness set to 0.0

04.png

Finally, now it works as expected.
But in my opinion the 2nd, 3rd and 4th analysis must give the same results. Why this behaviour? If I want to set a finite stiffness, for example if I have a suspension from the component to the ground, could be a problem.


Attachments:
test.zip, Size: 20.35 KiB, Downloads: 17

Offline

#2 2019-07-16 12:01:00

mecour
Member
From: Ostrava (Czech)
Registered: 2011-04-04
Posts: 112

Re: Strange behaviour of K_TR_D_N

Hello,

This behaviour is not strange. If you want to operate with nodal stiffness you did it right way. If you use discrete element in some node, the stiffness of this element is primary in this node. So if you set the nodal dof to zero there is still the stiffness of discrete element which hold the system.

Offline

#3 2019-07-16 14:45:49

pironman
Member
Registered: 2018-05-15
Posts: 50

Re: Strange behaviour of K_TR_D_N

I didn't know that. I assume that the discrete element remain "in the air" untill I fix it to the ground.
Then the nodal dof are redundant, and if one node must be free because attached to a component that transmit force, the stiffness must be set to zero.

Thank you

Offline

#4 2019-07-16 15:01:48

jeanpierreaubry
Guru
From: nantes (france)
Registered: 2009-03-12
Posts: 3,569

Re: Strange behaviour of K_TR_D_N

hello

discrete element on a node are real night mare to use in static analysis
my guess is that they are only to be used on dynamic problem particularly with damping

if yo want to set a condition where one point is fixed, here the ground and a nearby one can rotate the best way is to use a very short (could even be 0 length) line element wit a K_TR_D_L on it

this is discussed in my book in the appendix about discrete element

in your case a node A end of the LIAISON_SOLIDE
one node B very close carrying the ground fixation
between A and B a K_TR_D_L with the wanted stiffness

jean pierre aubry

Offline

#5 2019-07-16 19:49:39

mecour
Member
From: Ostrava (Czech)
Registered: 2011-04-04
Posts: 112

Re: Strange behaviour of K_TR_D_N

I use teh nodal discrete elements in two cases. When I need rotation dof of 3D system and when I need to hold some free body in contact analysis before the contact process initialize.

When you need simulate ground stifness the better is to use the line discrete element as Jean Pierre wrote (see example SSNL130). And also when you need elbow in poutre elements (example in his book)

The nice example of nodal discrete element is on caelinux wiki. caelinux.org/wiki/images/4/46/ApplyingTorque.pdf

Offline

#6 2019-07-17 07:09:27

pironman
Member
Registered: 2018-05-15
Posts: 50

Re: Strange behaviour of K_TR_D_N

I will follow your advice.
Just one question, why it is better to prefer K_TR_D_L? the other one could be unstable?

Offline

#7 2019-07-17 09:43:49

jeanpierreaubry
Guru
From: nantes (france)
Registered: 2009-03-12
Posts: 3,569

Re: Strange behaviour of K_TR_D_N

pironman wrote:

Just one question, why it is better to prefer K_TR_D_L? the other one could be unstable?

because it has two nodes that moves independantly!!!!!!

Offline

#8 2019-07-17 15:00:32

pironman
Member
Registered: 2018-05-15
Posts: 50

Re: Strange behaviour of K_TR_D_N

I understand that, but if I want to fix one end to the ground, I can do it both with K_TR_D_N and K_TR_D_L, Why I must choose the second one?

Offline

#9 2019-07-17 15:39:35

jeanpierreaubry
Guru
From: nantes (france)
Registered: 2009-03-12
Posts: 3,569

Re: Strange behaviour of K_TR_D_N

just try out to see the difference!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Offline