I am using NX 8.5- Advanced simulation for my work. I try to find natural@resonance frequency for double layer material (steel). Currently I am using SOL103- Response simulation.
1. For my first case, Clamped all edges for the double layer steel and no adhesion between the two layer.I am using 3D tetrahedral mesh.
What condition should I use between the touch surface for top and bottom plate. Is it correct surface-to surrface contact ? how about mesh mating, do I need to set anything about it?
2. For the same case as above, I am using edge to edge gluing for every edge top and bottom plate that touch each other? is it correct to represent my target case? however, I cant run the simulation if I am using this condition - 'BEDGE region is empty'
3. For my second case, clamped all edges for double layer steel with hard glue@adhesion between the two layer.
Is it correct i am using surface-to-surface gluing for the touch surface? Do I need to set anyting with mesh mating also?
4. What is mesh mating? and what is the different between glue coincident,free coincident and glue non coincident type.
Thanks a lot
Contact in a modal solution is supported, but it assumes that the contact state does not change through the frequency range. Essentially the solution determines the contact state from a static perspective, then converts closed contact conditions into glue conditions for the modal solution. The normal modes solution treats the contact condition as if it is defining an equilibrium state that the normal modes need to be calculated about. Where you have faces touching, you should have closed contact, therefore those regions should be glued together during the normal modes calculation.
Mesh mating is a feature in NX that lets you define relationships between adjacent solid meshes. The most commonly used mesh mating condition is glue coincident in the context of hex meshing. You can split a block into two blocks and define a glue coincident condition between them. NX forms a single face from the 2 coincident faces between the blocks. Each block uses the single face, so the meshes are connected directly at the nodes/element faces of the shared face. Free coincident will keep the two bodies separate, but force the mesh of the coincident faces to have the same locations for nodes and element faces. It produces coincident nodes and element faces. This is often used in contact solutions to improve the contact results and make the solve more efficient. A glue non-coincident mesh mating condition will maintain the two blocks as separate entities, and the user can put distinct meshes on the blocks. The connection between the blocks is made with a set of rigid and constraint elements.
For all mesh mating conditions, the geometry of the participating faces is edited. The source faces are imprinted onto the target faces and vice versa. This produces a common boundary at the intersection of the source and target faces. How the faces of the common boundary are treated depends on the type of mesh mating condition.
By mesh mating relative to this problem, I assume you mean glue coincident mesh mating or glue non-coincident mesh mating. Both of these options connect the two solid meshes together where the faces are touching. Using mesh mating conditions is more efficient when you can assume that there will be no opening of the contact regions. It’s more efficient because you are defining the connection in the mesh and not relying on the solver to determine the open/closed state of the contact condition.
Glue is a very good solution for this type of model when you do not need to have the solid meshes matching and connected directly (i.e. glue coincident mesh mating) and you can assume that there will be no opening of the contact regions. The same assumptions relative to using mesh mating conditions apply to glue. Glue is advantageous because the solids can be meshed independently of one another, yet connected with stiffness based upon the material properties of the connected meshes.
I think it would be good for you to set up multiple solutions exercising the different connection options described above. The glue and contact solutions shouldn't be much different for two plates touching in a plane or completely flush with one another.
Simulation Product Management
Simulation and Test Solutions
Siemens Industry Sector
Siemens Product Lifecycle Management Software Inc.
However I still unclear about few things, (let say my work consists of two plate with 1 mm thickness each).
1. (Experimental test) Both of the layer have a good adhesion and really stick well at the contact surface . (Nastran mode) - I am using surface-to-surface-glue at contact surface of the plate. Assumes that the contact state does not change through the frequency range.
Results- for resonant frequency,the combination two layer 1mm plate will excite as 1 layer of 2mm plate excited. - I have done this case.
2. (Experimental test) Both layer only contact at the edge, clamped condition. There are no adhision between the plate. Possibility very tiny and small air gap layer occure between the plate during excitation?-not sure.
(Nastran mode) - how should i do it?
i. assumes that the contact state does not change through the frequency range - not sure. If the contact surface changes through frequency, how can i solve it?
ii. do I need to glued it at the contact surface? I cannot put surface-to surface-glue contact because case no.1 is not the same as case no.2.
iii. do I need to use surface-to-surface-contact ? - i have done it with nastran, two almost identical mode shape occure at every resonant frequency.Is it correct?
Nastran assumes that the glue/contact state determined from a static pre-load step remains the same through the entire frequency range of a modal step. The modal solution is linear and assumes vibration is about a steady state condition (i.e. the state of the model after applying a pre-load). Glue and contact are represented with additional stiffness applied to the model, and the model’s stiffness remains constant through the frequency range. This is fine for your first experiment.
Your second experiment can’t make the assumption that the vibration is about a steady state condition. Since it is possible to have changes in the contact state over the frequency range, the stiffness of the model is changing with respect to frequency. You’d have to solve this problem using a nonlinear transient solution. A modal solution isn't valid here.
Thanks for the feedbacks. I assumed i have a little bit understanding on my first case.
Back to the second case, I need to look into more on nonlinear part since i have no experience before this.
i. I have try using SOL 129 nonlinear transient response but I did not manage to get the resonant frequency.All the other setup are the same as previous (when using SOL 103 response simulation). Fixed-clamped boundary condition. With no load- the plate doesnt change anything.
ii. Additional step, using the SOL 129, I also add a location of force at the bottom of the lower plate. The plate deflection react- around the force location only.
iii. Do i need to creat a GAP elements between the contact area of the two plates?
iv. Can you briefly explain on how can I solve using the nonlinear solution for this case to get mode shape and the resonant frequency.?
I don't think it is possible to briefly explain nonlinear transient. I don't have much experience with it either. SOL 129 doesn't support contact, but it does support GAP elements. SOL 601, 129 does support contact. The time step that the transient takes is inversely related to the maximum frequency that the analysis can hope to capture.
The nonlinear transient solution won't output mode shapes or resonant frequencies. It will output the transient response to transient excitations. It doesn't use mode shapes or resonant frequencies as part of its solution (unlike linear modal transient and modal frequency response solutions). Beyond that, you'll need to read the Nastran documentation related to nonlinear transient response, as I would have to do the same.
Thanks for your answer, I will look more on the suggestion.
One last question, I saw one of the thesis using ANSYS to solve on constraint layer damping (more or less similar to my second case), which can produce response vs freq result.
(credit to the original auhor).
If I am plan to use NX 8.5, do it have the capability to run like the same case above? which solution SOL that should i use?
Thanks a lot Mark for your feedback.