Simple Patran Model error

[Songa]

Hi everyone,

I made a simple model of a car in PATRAN using BAR elements, QUADS, and a few TRIAs. When I run the entire model in NASTRAN, i get the following fatal message:

^^^ USER FATAL MESSAGE 9050 (SEKRRS)
^^^ RUN TERMINATED DUE TO EXCESSIVE PIVOT RATIOS IN MATRIX KLL.

The model is statically determined, and if I run only the bar elements in NASTRAN I get no fatal errors, so I know that the constraints of the model (at 3 of the suspension points) are ok. (a force is applied on the 4th suspension point).

If I look in the .f06 file and plot the nodes listed above the fatal message I see the following:

The problem seems to be the shell elements, which are somehow not constraint to each other. On a surface the edges are not displayed as giving errors but all the inner nodes are. By the way, the model was equivalenced.

When I select ANALYSIS -> SOLUTION TYPE -> SOLUTION PARAMETERS -> Automatic Constraints, the model runs without fatals but somehow the results look strange, because the deformed plate is not a plate anymore as some of the (inner) plate nodes moved way to much compared to the rest of the plate.

Does somebody have an idea of what is going on here?

 

[ThomasH]

Hi

It might be that a part of the model is "free" from the rest of the model. That means that you have a free body causing the excessive pivot ratios.

Try running an analysis for eigenfrequencies. If there is a free body (or not enough constraints) it will result in zero (0 Hz) frequencies. Then you can by animating the results see what's going on.

Good Luck

Thomas

 

[Songa]

Thank you Thomas for your quick respond.

I have tried a eigenfrequency analysis and indeed it looks like a part of the model is free (see attached pic 01.jpg)

http://files.engineering.com/getfile.aspx?folder=a0a3407f-a509-430e-98f2-d11186b6269f&file=01.JPG

As I said in my post above i've checked the .f06 file and the nodes in above the fatal error are plotted in attached picture 02.jpg. As you can see it looks like the nodes at the side of the panel are constraint but the inside nodes are not connected for some reason.

http://files.engineering.com/getfile.aspx?folder=43ea306f-1be0-4202-9cc9-c2543d033182&file=02.JPG

Without this panel the model runs ok, without errors. How can it be that the nodes at the inside of the panel are not automatically connected to other nodes using the quad elements, like it works on every other panel?

Maybe some of you have seen this before and know if there is a simple solution?

 

[ThomasH]

Might have someting to do with the meshing in Patran. I don't know Patran so I really can't say.

Can't you simply merge the nodes together?

Thomas

 

[alin09]

Try "equivalence" in ELEMENT of PATRAN. Maybe there are redundant nodes in the model automatic meshing, which could be the cause.

 

[BlasMolero]

Dear Songa,
You have a rigid body movement, some elements of you model a free to move in the space, the frequency plot shows where the problem is located. Reasons: you simply forgot to merge coincident ones, or your shell mesh is not coincident with base CBAR, then when merging nodes some elements remains free, a simply inspection of free-edges in your model will tell you how to solve the problem.
Best regards,
Blas.