Different results in linear static analysis between SimLab and HyperWorks (all version 2022.1)

James Lewis

Hi,

 

I did a simple linear static analysis in both SimLab and HyperWorks.

They are same with all BCs, but there a significant difference in Von mises stress.

Any can help explaining with this issue?

 

Many thanks!

Adriano_Koga

James Lewis said:

Hi,

Could I ask you one more question?

When I compare the stress concentration factor between: looking up at table and using SimLab linear static analysis, there is difference.

1. With parameters of the model as below:

The stress concentration factor = 1.5, according this table:

reference link to the table: https://topdogengineer.com/stress-concentration-factors-a-fundamental-example/

2. When using SimLab, I've received different Kt

If you have free time, please give me a look, I attached model below.

Hi,

 

in order to get the Kt closer to the analytical value, you would need to refine the mesh at the fillet region to get a better representation of stresses.

I've done it roughly in Inspire, as you can see below.

0.02MPa as far field stress and close to .03MPa at the fillet area.

 

In Inspire you can also choose the meshless SimSolid solver, and get similar results.

The advantage is that you don't need to worry about the mesh.

Adriano_Koga

Hi,

 

have you checked the total reaction forces in both models?

Are you sure the loads are the same?

In HW, for exxample, you could be applying the total load multiple times, in each node, instead of applying the load divided by the node count.

SimLab by default distributes the force in all the face equally.

 

I would start checking reaction forces (SPCF) to make sure you have the correct loading, first.

Jason_Craanen

Also make sure you are comparing the same results (average or not. Element stress or nodal stress)...

Each run should have a fem file and out file. Compare those as well.

James Lewis

Adriano A. Koga_21884 said:

Hi,

 

have you checked the total reaction forces in both models?

Are you sure the loads are the same?

In HW, for exxample, you could be applying the total load multiple times, in each node, instead of applying the load divided by the node count.

SimLab by default distributes the force in all the face equally.

 

I would start checking reaction forces (SPCF) to make sure you have the correct loading, first.

Hi,

Thank you, I got the problem.

In SimLab by default, loads are distributed to each nodes by shape function.

And in HW, by default, loads applied to each nodes equally, so that make the this difference.

James Lewis

Adriano A. Koga_21884 said:

Hi,

 

have you checked the total reaction forces in both models?

Are you sure the loads are the same?

In HW, for exxample, you could be applying the total load multiple times, in each node, instead of applying the load divided by the node count.

SimLab by default distributes the force in all the face equally.

 

I would start checking reaction forces (SPCF) to make sure you have the correct loading, first.

Hi,

Could I ask you one more question?

When I compare the stress concentration factor between: looking up at table and using SimLab linear static analysis, there is difference.

1. With parameters of the model as below:

The stress concentration factor = 1.5, according this table:

reference link to the table: https://topdogengineer.com/stress-concentration-factors-a-fundamental-example/

2. When using SimLab, I've received different Kt

If you have free time, please give me a look, I attached model below.

Adriano_Koga

James Lewis said:

Hi,

Could I ask you one more question?

When I compare the stress concentration factor between: looking up at table and using SimLab linear static analysis, there is difference.

1. With parameters of the model as below:

The stress concentration factor = 1.5, according this table:

reference link to the table: https://topdogengineer.com/stress-concentration-factors-a-fundamental-example/

2. When using SimLab, I've received different Kt

If you have free time, please give me a look, I attached model below.

Hi,

 

in order to get the Kt closer to the analytical value, you would need to refine the mesh at the fillet region to get a better representation of stresses.

I've done it roughly in Inspire, as you can see below.

0.02MPa as far field stress and close to .03MPa at the fillet area.

 

In Inspire you can also choose the meshless SimSolid solver, and get similar results.

The advantage is that you don't need to worry about the mesh.

Adriano_Koga

Adriano A. Koga_21884 said:

Hi,

 

in order to get the Kt closer to the analytical value, you would need to refine the mesh at the fillet region to get a better representation of stresses.

I've done it roughly in Inspire, as you can see below.

0.02MPa as far field stress and close to .03MPa at the fillet area.

 

In Inspire you can also choose the meshless SimSolid solver, and get similar results.

The advantage is that you don't need to worry about the mesh.

in Inspire, with refined mesh, running with second order tetra mesh, I got stresses around 0.034MPa at the fillet.

Adriano_Koga

Adriano A. Koga_21884 said:

in Inspire, with refined mesh, running with second order tetra mesh, I got stresses around 0.034MPa at the fillet.

reference model

James Lewis

Adriano A. Koga_21884 said:

reference model

Hi,

 

You've really widen my mind, thank you very much.