Results Calibration Example

Results Calibration Example

Anonymous
Not applicable
460 Views
2 Replies
Message 1 of 3

Results Calibration Example

Anonymous
Not applicable

Hi folks,

 

This time I would like to ask about your opinion.

 

Below figure is self-explaining, you may notice 150 MPa applied to the tip of smaller solid object (lets assume it is mild steel) and below, larger box is providing supports. Assuming it is converged enough, in this simple example it is obvious real structure will "see" stress about 150 MPa in zone abutting to the larger object. But if the geometry would be more complicated, how would you be so sure how to get rid off this singularity? In other words, how to convince somebody that looking at stress range singularities are not real? I prefer to use stress linearization. What is your opinion? Please share.

 

@John_Holtz I will appreciate your input.

 

 

Test1.PNG

Reply
Reply
0 Likes
461 Views
2 Replies
Replies (2)
Message 2 of 3

John_Holtz
Autodesk Support
Autodesk Support

Hi,

 

Can you elaborate on what you mean by "I prefer to use stress linearization."?

  • If this mean that you are currently using stress linearization to show that the singularity is a false result, please provide some more details about how you do that.
  • If this means that you want to use the stress linearization, then hopefully someone else will have some suggestions because I do not know how stress linearization would be able to show that it is a false result.

At least for your simple model, the von Mises Precision ("Results Contours > Stress > von Mises Precision") can be used to show that the corners have the least accuracy. And since the precision cannot be improved by refining the mesh, this demonstrates that the result is a singularity. (Note that the von Mises precision is 0 by definition at nodes that are not shared by other elements within the part, but the other nodes on the same element should have a poor von Mises precision.)

 

Another approach is to have results for an existing design that is functioning. If the results shows a high stress but the design has been "operating for years without a problem", that can provide support that the stress result is artificial (a singularity).

 

Keep in mind to understand how the model differs from reality. For example, most models use bonded contact between the parts. This is true only if parts are glued together. In many cases, parts are welded around the perimeter, so the real results are much different than the modeled results. (Maybe you do not want to go into these details with the customer to explain how you did not model the real situation Smiley Wink.) It does not make too much sense to me to make a more elaborate model, such as including the welds, because there are many effects of welding that you cannot simulate accurately -- like the thermal stress, the localized change in the material properties, and so on.

 

What ideas do other readers have?



John Holtz, P.E.

Global Product Support
Autodesk, Inc.


If not provided, indicate the version of Inventor Nastran you are using.
If the issue is related to a model, attach the model! See What files to provide when the model is needed.
Reply
Reply
0 Likes
Message 3 of 3

Anonymous
Not applicable

Thank you for contribution.

 

By stress linearisation I mean I am doing this already, and by definition of this I understand presenting how abruptly normal stress is hitting infinity at a geometry change. This would be certainly stress concentration, but since I do not have welds in the model, just sharp edges, value is increasing at each mesh sizing down, hence the singularity. Fatigue is not the case because of static load, not dynamic.

 

 

However, to show/explain what is going on, to underline that infinity is infinity, one have to depict this using some stress plot which I find very useful.

In the stress peak at a geometry change the total normal stress value consist of: 150 MPa normal stress (A) + Stress concentration due to geometry (B) + Singularity peak due to sharp edges (C).

 

My question is what is your good practice for extracting those (B) and (C) values?

Further discussion will be appreciated.

Reply
Reply
0 Likes