Linear Transient Stress (Direct Integration) vs. Non-Linear MES

[Anonymous]

Happy V-Day,

    While most are celebrating I am calculating haha. I am working through a modeling problem and am at an all-stop. I am hoping someone can help with this delimma. I am running an analysis on a foundation that recieves a static load from the equipments weight and a dynamic load through the equipments operating forces. I modeled the foundation and used three different analysis types to look at the maximum von-Mises stresses and deflections. First Imodeled all the loads as static in the worst case scenerio, Then modeled the same loads in the transient stress (Direct Integration) with time dependent loads using the known load curve for the equipment, the modeled the same loads in the non-linear MES with the same known time dependent load curves. The results are as follows:

Linear Static Stress:

     Max Stress = 8643.4 psi

     Max Displacement = 0.0205 in

Linear Transient Stress (Direct Integration):

     Max Stress = 24284.55 psi

     Max Displacement = 0.1161857 in

Non-Linear MES:

     Max Stress = 8526.458 psi

     Max Displacement = 0.01728682 in

     All linear element types are plates, the non-linear element type is of course shell. Why are the results for the linear transient (Direct Integration) Analysis so different from the other two analysis types? I've modeled and re-run the analysis several times to rull out errors. The results are the same. Does the transient analysis environment not handle dynamic loads to plate elements well? Thank you for your time and consideration!

- H. Adams

[AstroJohnPE]

Hi Horace,

Did you include midside nodes for the shell elements in the MES analysis? I am actually surprised that MES gave about the same result as the linear static analysis. Depending on what type of shell elements were used, they can behave too stiff, so I suspect the linear transient stress analysis is showing a real time-dependent maximum that will be shown with the MES when midside nodes are included. (The different types of shell elements are chosen on one of the tabs under the Element Definition.)

Also, are the maximum stresses in the same location in all of the analyses? In particular, the transient stress and MES should have similar results, so if the maximum is at a different location, that would indicate that something is different between the two.

[Anonymous]

John,

     Thank you for the very imformative reply. I did not include midside nodes in the MES but I am now going to include them and run the same simulation. To answer the second question, the Linear Static Analysis and the MES show the maximum deflections and maximum stress in the same location. The Linear Transient analysis however shows the maximums in a completely different location.

     When including the midside nodes what exactly does that mean from an analysis standpoint? Is it allowing for another point to integrate from? Will it prevent "shear-locking"? Thank you for your time, it is very much appreciated!

-H. Adams

[Anonymous]

John,

     I ran the same analysis with midside nodes enabled and the stress increased from 8526.5 psi to 14334.4 psi in the same location. The elements are shell with midside nodes and large displacement selected. The transient analysis still shows displacements and stresses in different locations than what the linear analysis and the nonlinear MES analysis show. I even loaded the transient analysis and loaded the identical load curves from the non-linear analysis. What analysis method is accurate? Is there anything I can change to correct the transient analysis? Thank you.

-H. Adams

[AstroJohnPE]

It's hard to say why the results of the transient stress are different. Assuming that everything is the same, the results should be pretty much the same. (I think that Transient Stress is an older analysis type, so it does not have the capability as far as loads that MES has.)

If you can attach the model, someone can take a look to see if the two analysis types give different results. (See the thread "Create, Post, or Provide an Archive of your model". In order to get the archive small enough to attach, you may need to "Save As" the model to a temporary name, then delete everything except for the transient stress analysis.)

If you can't provide the model, then some figures or sketches showing the layout, loads, load curves, etc. would enable someone to create a simple model in Transient Stress and MES to confirm that both analysis types give the same result.