Simulation on stair: Model unstable

[TheDutchman]

Hallo,

I'm trying to simulate the impact of a human standing on the side of my staircase. You can see an image in the attachment. This is my first simulation so please have some patience .

The problem is, when I try to Run the study for results I get the error about my model not being stable. Something about my restrains. As you can see, I fixed the 2 faces where my staircase is connected (fixed) so I cannot think of any other missed restrain. The rest of the assembly is mated togheter. As you can see, it is fully defined because I fixed 1 component. Hower changing this doesn't give me a different outcome.

I'm doubting that mating is not enough ? In real life my stair is a bend sheet metal with 2 plates welded on the side. The V shape below the stair is also welded to the sheet metal. But is mating not enough ? I tried weldments but only received more errors.

Can somebody give me an Eng-tip in the right direction ?

Thanks in advance

 

[GregLocock]

Suppress the loads and the stair and the bracket.

Apply a force and a torque directly to the square tube.

Does that run?



Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[BlasMolero]

Hello!,
You have rigid body motions in your model, then the stifness matrix is singular because some portion of the model can move freely, i.e., coincident mesh nodes are not properly merged. Without having the FE model is dificult to see where the error is located. The trick here is to run a Frequency analysis to see the rigid body motions and realize where the meshing error is located.

For the picture I see you have walls with very small thickness compared with length. Please avoid at all to mesh the model with solid TETRA10 elements, do not fall in the typical error I see in many cases, this is a Shell model and then you must create midsurfaces and mesh with shell plate element. Unfortunately neither COSMOS/DesignSTAR nor COSMOS/Works supports 4-node Quad Shell elements, this is basic for a modern FE code, then you will have to mesh with poor 3-nodes Shell elements.

Meshing with solid 3D Tetra10 elements (please note I state clearly TETRA10, not low-order TERTRA4 that should be avoided at all) require to create two solid elements in the thickness, then you can imagine that model size will have millions and millions of nodes, not possible to manage. Instead SHELL elements are perfect in this type of problems, you will get a very accurate solutions, a very small model size and a fast solver computing time, this is the target, OK?.

Best regards,
Blas.

 

[kellnerp]

Are there any gaps between parts that would cause bonding to fail?

To find out, turn on soft spring support in the options. When you solve any "loose" parts will fly into space.

TOP
CSWP, BSSE

http://www.engtran.com

http://www.niswug.org

"Node news is good news."

 

[TheDutchman]

Hello,

First of all I want to thank you guys for the replies!
@BlasMolero:

I'm from Belgium and it took a while to understand what you posted there, but this is for me the picture:

If I don't use shell elements, the meshing will be done troughout the thickness and this will cost me precious recources. Hower when I use Shell elements, the thickness is set and the simulator works with a mesh that's very thin but acts like the right thickness?

However it was pretty hard to translate it to SolidWorks environment. I didn't find any settings for Tetra10 or Tetra4. Where is it exactly located? And working with Shell elements requires me to change every part if i'm correct?

@Kellnerp
The trick with the soft springs pointed me on some bonding issues. However I think it is kind of weird when I put a global contact of "Bonding" in it and I solve the bonding issue with a contact set of type "Bonding".

Now the next question. Am I simulating correct? I Want to simulate the displacement of my stair when a (man or woman ) with a weight of 100Kg takes the stair. So I made a plate with the dimensions of about a foot and put on the top face a Force load of 250N per item (I saw 4 arrows)
F=m*a 1000N = 100kg * (+/-) 10m/s²

In attachment I uploaded my displacement results. I think they are somewhat trough the roof?

As you see, I could use some help with this
Thanks in advance

 

[PeteDB]

If I remember correctly, you have to define all the contacts manually when using shell elements. I would add splitlines where surfaces intersect. This help the nodes match at the intersection. Use the "high quality" elements not "draft" elements.

 

[kellnerp]

.6mm is not so bad. 100kg is kind of low. I am 140kg plus dynamic loading would make 280kg on the model plus a factor of safety as required.

Is there no deflection in the tube? It should have torsion.

This type of thing is well suited to a shell model. In SolidWorks create a new part inside the assembly and join all the pieces into one body. Then in that new part use the mid surface command to derive a shell model at the midplane of the material.

Making a shell model is a little extra work but it has a big advantage. In a shell model you can change the thickness of the material by just editing one number. So you can easily search for the right thickness for all the members by just editing a few shell definitions. Much easier than remodeling the whole thing each time.

Finally, to change from a Tetra10 to a Tetra4 requires going into the meshing options and changing to draft meshing. Not recommended for getting "real" answers, but good for checking the model.

TOP
CSWP, BSSE

http://www.engtran.com

http://www.niswug.org

"Node news is good news."