06-26-2018 09:24 AM - edited 06-26-2018 09:26 AM
Hello experts,
here is a second approach to investigate the distribution of a single CONM2-element by means of RBE3-spider onto a native mesh in modal analysis (SOL103) in NX10.0.3. First attempt was discussed without results in
Simcenter 3D-Forum => Method-of-Mass-Distribution-by-CONM2-with-RBE3-in-SOL103
Description:
I try to find out what's the difference in defining a structure with geometrically equally distributed mass elements (CONM2) at the rim of shell elements in comparison to an approach with only one CONM2-element containing the summarized mass and mass moments of inertia at the center of gravity of formerly distributed mass elements and connected to shell structure by means of RBE3-spider element.
Solution method:
- Rigid element method: Linear elimination
- Eigenvalue method: Lanczos
Summary:
I've got differences in eigenmodes so I must assume that the RBE3 element DOES stiff the structure!!! - Although documentation says that RBE3 elements does not create additional stiffness in model (in contrast to RBE2).
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
Here my models:
All models have no mass in shell elements as density is 1e-9 g/cm³.
Dimensions: r.a = 100 mm, r.i = 50mm, t = 1 mm E = 2.1e5 MPa, nue = 0.3
Boundaries: Fixed outer rim in all 6 DOFs and fix 3-6 DOF on inner nodes to get a 2D-problem,
Displacement-CS of nodes is cylindrical around Z-axis.
1. Model with geometrically equal distributed CONM2-elements (A):
2. Model with on CONM2-element equipped with summarized mass and mass moment of inertia around Z-axis (B):
Additionally I did create two models with RBE3-element and decoupled summarized mass and mass moment of inertia.
3. No mass moment of inertia, only summarized mass at center of gravity (C):
4. only mass moment of inertia, no summarized mass at center of gravity (4):
I've chosen to calculate the first ten modes of structure.
Results:
1. Only model with distributed mass elements (A) could get all 10 eigenmodes.
2. Model with summarized mass and mass moment of inertia (B) provides only three modes. I could find them in first model (A), too. The modes are suggestive of acting like a single mass oscillator, keeping the inner structure (nearly) circle-like: mode 1 and 2 look like translation modes and mode 3 looks like rotation mode around Z-axis.
3. Model with summarized mass only (C) provides the both translation modes.
4. Model with mass moment of inertia only (D) provides only one mode: the rotational mode around Z-axis.
Next pictures show first modes of model with distributed mass (A) with corresponding modes in other model (if exist).
Upper left: model with distributed mass (A)
Upper right: model with CONM2+RBE3 - Summ. mass + MMoI (B)
Lower left: model with CONM2+RBE3 - Summ. mass (C)
Lower right: model with CONM2+RBE3 - MMoI (D)
It can be seen that
- mode 3 and 4 of model with distributed mass (A) have partners in (B: mode 1 and 2) and (C: mode 1 and 2) and
- mode 5 of model with distributed mass (A) have partners in (B: mode 3) and (D: mode 1).
All other modes of (A) are orphaned.
That's not what I did expect. - My expectations are (A) and (B) are the same. Why isn't it so?
Who can explain?
Any suggestions are welcome, best wishes, Michael
06-26-2018 09:40 AM
I already answered in the other post.
I would say that stiffness is not added by RBE3 because the static analyses that you performed were ok. If I remember well.
I would say that some expert/developer from Siemens should answer. I am FEMAP user not NX but the issue is interesting to deal with.
In theory, as you are saying, should be “the same”.
Sorry not helping more
06-27-2018 07:18 AM - edited 06-27-2018 07:22 AM
@orzewalla, thank you for answering in
Simcenter 3D-Forum => Method-of-Mass-Distribution-by-CONM2-with-RBE3-in-SOL103.
Please, don't keep the eye too much onto the reason of introducing COMN2-element with RBE3-spider there.
In this thread therefore, I did create a more simplified structure to find hints for correct interpretation of behavior of modal model (mass only in CONM2-elements not in shell structure, flat problem with really equal connections to nodes on rim).
I would like to ask if we could discuss the matter here with having focus on the structure described here.
I will copy my answer from this Simcenter-forum-thread to this forum, too.
Best wishes, Michael
06-27-2018 07:56 AM - edited 06-28-2018 05:19 AM
@orzewalla, thank you for answering.
Here I must admit that I do not understand the way a CONM2 from core node is transformed onto leg nodes of RBE3-elements.
I did imagine that RBE3 with "linear elimination" delivers the mass onto the independent nodes in a way like having CONM2-elements defined there directly. Isn't that correct?
In a static solution I could identify that a single mass related loading is transferred only if the center node itself has the suited property.
Example: I did create two load cases:
1. with rotational (angular) acceleration of 1000 radians/s² around Z-axis and an angular velocity of zero.
2. with translational gravity in X-direction with 1000 m/s².
I could find out that in my special model with only summarized mass and without mass moment of inertia (C) the 1. load case (rota acc) didn't produce deformations or MPC-forces.
And in the model without summarized mass and only mass moment of inertia I could see that gravity does not have an effect in deformation or MPC forces. In my model (B) I find both effects.
=> I believe that the mass related loading
* rot.acc: M.z = MMoI * alpha.z
* grav_X: F.x = m * a.x
is calculated fist and afterward transformed onto leg nodes.
"Having no MMoI" yields in having no rotational moment, and "having no mass" yields in having no gravity force.
In that static cases there is no mass matrix to get transferred mass elements on independent nodes. Therefore, I could understand this behavior.
START - EDIT 2018-06-28 >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
Here are the calculation results for static load cases for the for models
Upper left: model with distributed mass (A)
Upper right: model with CONM2+RBE3 - Summ. mass + MMoI (B)
Lower left: model with CONM2+RBE3 - Summ. mass (C)
Lower right: model with CONM2+RBE3 - MMoI (D)
I put MPC-forces and displacements into article. It can be seen that MPC-forces only act in models with RBE3 (not A) AND CONM2-parameters corresponding to loading (B,C: mass <=> Grav_X; B,D: MMoI <=> rotational acc). Don't miss the differences in decimal power (power of ten) in displacements and MPC-forces in the cases where no results are derived.
Load case: Rotational acceleration around Z-axis
Load case: Gravity +X
END - EDIT 2018-06-28 <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
But in a modal model the mass matrix exists directly and is important for eigenvalue problem. So, I thought the transferring yields in mass contributions at the leg nodes in a "physical" way. And a mass moment of inertia which represents a mass distribution which is corresponding to the radius of nodes at rim (here 50 mm)
MMoI = SUMM (m.i) * r²
yields in a distribution of exactly that mass at the independent nodes by means of Steiner's theorem (of course not knowing how does it is managed correctly).
But apparently it's not the case. Who can explain it? I can't, can you?
Best wishes, Michael