Comparing 2D model with 3D 1

[harshitbhatt]

Happy New Year to all....

I have a 2D problem in Matlab where I have found the deformation of the problem profile. Now I need to compare the same thing i.e. the same profile but in 3D using Nastran. What I have done is, I have taken 0.1 degree of the profile and carried out the simulation.

The problem is that I am getting difference in the answers [0.5 to 1 mm]. If the answer in Matlab is say 0.3mm than in Nastran I am getting 0.8mm. I am getting little bit difference in both the answers. How can I reduce this difference. I tried reducing / increasing the mesh size but it gives the same result.

Waiting for the valuable replies.

 

[aerostress82]

[ul]
[li]Could you provide a sample file for the Matlab and the Nastran codes/decks? That would help detect the problem better.[/li]
[li]What is the size and type of the structure? (Length, Width and Thickness - if it is a simple plate structure)[/li]
[li]Are you performing a linear static (Nastran SOL101) analysis? And what are your boundary conditions (ie, constraints, force/moment/pressure/distributed force etc)[/li]
[/ul]


Just to make sure:
Are you constraining the 3D edge at all nodes when you switch to the 3D model? Or did you use "only the same nodes" from your previous Matlab analysis?

Spaceship!!


Aerospace Engineer, M.Sc. / Aircraft Stress Engineer with 7 years of experience
(United States)

 

[harshitbhatt]

@aerostress82 thank you for the reply. Below is the model whose, I need to find the deformation and compare the results with Matlab.

In NX, I have only taken 0.1 degree not the whole 360 degree. It is circular. I have given a fixed constraint to the center of gravity of the model[Black dot in the center of figure]. Also I have given symmetric constraint to front and back surface as it is a 360 degree part. The red line shows high pressure, blue line low pressure and green line shows pressure field. Yes I am using SOL101 for solving the problem

In Matlab, this whole thing is done in 2D. There also I have fixed the center of gravity and gave same pressures as acting on the model that I applied in NX.

But still there is difference in the result as I mentioned above.

 

[aerostress82]

I'm not an NX Nastran user, so don't want to affect judgement on your problem.
I'll just say this and you can check if it is related.
1) Is pressure field accurate in both models?
2) You said it is circular but you only modeled 0.1 degrees in NX. So, in Matlab, are you accounting for the circular shape somewhere? The circular effect will add more stiffness to your design in NX, but if this is not represented in Matlab (2D) that might be affecting your results.

If you are already sure of these 2 above, then I'll pass the problem to another NX experts here.

Yes, NX Experts?

Spaceship!!


Aerospace Engineer, M.Sc. / Aircraft Stress Engineer with 7 years of experience
(United States)

 

[harshitbhatt]

Yes field are the same in both the model. In matlab its totally 2D. No circular shape added or considered. Yes if possible you can pass on.

I am thank full for that.

 

[aerostress82]

Just to be clear:
- If you have no circular shape consideration in Matlab (2D),
- And if you are modeling this with 0.1 degree in NX (3D),
Then you will get different results I think. NX model is circular, Matlab model is not - so they will have structurally different results and deformations. In Matlab, you should be adding the circular structure effect into your formula. I don't know how, but something should resemble your circular shape in your equations there.

[ul]
[li]If you model your NX (3D) model with only 1 mm thickness, and not model it as 0.1degree (not circular this time), do you get similar results then?[/li]
[/ul]

I'm thinking this may be the problem. If you are still getting different results "with no circular shape, but only with 1mm thickness", then yes, the problem might be something else..

Spaceship!!


Aerospace Engineer, M.Sc. / Aircraft Stress Engineer with 7 years of experience
(United States)

 

[harshitbhatt]

@aerostress82 yes I get totally different results if I am considering 1 mm thickness.