Request for FEA Analysis

[rsterne]

This is a request for anyone who has the ability to run an FEA Stress Model for a very simple cylindrical HPA reservoir....



The material is 2024-T3 Aluminum, Yield Strength 50,000 psi (345 MPa) and Tensile Strength 70,000 psi (483 MPa).... It is magically machined in a single piece from a 1.250" OD with a 1.000" ID, so the walls are 0.125" thick.... The flat ends are 0.250" thick.... One end has a square internal corner, the other has a 0.125" radius (the same as the wall thickness).... The Working Pressure is 300 Bar (4,350 psi) and the FEA is to be run at 900 Bar (13,050 psi)....

I know that several GTA members have the ability to do this, and I would appreciate it if one or more of you could please run this test for me.... I put a square inside on one end and a radius on the other, so that only one FEA run would be required.... Thanks in advance....

Bob

[WobblyHand]

This is a request for anyone who has the ability to run an FEA Stress Model for a very simple cylindrical HPA reservoir....



The material is 2024-T3 Aluminum, Yield Strength 50,000 psi (345 MPa) and Tensile Strength 70,000 psi (483 MPa).... It is magically machined in a single piece from a 1.250" OD with a 1.000" ID, so the walls are 0.125" thick.... The flat ends are 0.250" thick.... One end has a square internal corner, the other has a 0.125" radius (the same as the wall thickness).... The Working Pressure is 300 Bar (4,350 psi) and the FEA is to be run at 900 Bar (13,050 psi)....

I know that several GTA members have the ability to do this, and I would appreciate it if one or more of you could please run this test for me.... I put a square inside on one end and a radius on the other, so that only one FEA run would be required.... Thanks in advance....

Bob

Not to be a joker, but I think you could do this.  You have FreeCAD and the FEM workbench.  If you need help, I'd be glad to help you get going.

This is a nice test case, since it gets rid of the threads.  1/8" walls at 900 bar!  I think one has to do a nonlinear simulation for this.  Do you know of a place to find the stress vs strain for 2024T3 all the way out to fracture?  I found the MIL-HBK-5J, which has curves for this for 2024T3, but they have two curves, one for longitudinal L and one for LT, do you know which curve is more appropriate? 

When I simulated the "other tube" with 3.5mm walls, the simulator had convergence problems at 75 MPa (750 bars).  At 70 MPa, the simulator converged. 

I can make a model and do a linear sim, that one will be easy.  For a nonlinear sim, could use some guidance...

[rsterne]

I had a brief look, and decided  it was beyond me at present, as I am still having problems constraining simple drawings.... If you can do this, it would be much appreciated.... The calculator I use, compliments of Lloyd Sikes, tells me the safety margins for the Hoop Strength are 2.5:1 to Yield and 3.5:1 using Tensile, at 300 bar.... It will do threaded or pinned (thick) end plugs, but not machined, and has no ability to add an internal radius.... The 2X wall thickness for the ends is a guess, if the stress in the middle (where it should be the highest) is greater than the hoop stress, they will need to be thicker (eg. 3X wall thickness = 0.375")....

While we're talking units, remember that 900 bar is "only" 90 MPa.... 

Bob

[WobblyHand]

I had a brief look, and decided  it was beyond me at present, as I am still having problems constraining simple drawings.... If you can do this, it would be much appreciated.... The calculator I use, compliments of Lloyd Sikes, tells me the safety margins for the Hoop Strength are 2.5:1 to Yield and 3.5:1 using Tensile, at 300 bar.... It will do threaded or pinned (thick) end plugs, but not machined, and has no ability to add an internal radius.... The 2X wall thickness for the ends is a guess, if the stress in the middle (where it should be the highest) is greater than the hoop stress, they will need to be thicker (eg. 3X wall thickness = 0.375")....

Bob

Honestly, I think this is still a bit beyond me, but I have stubbornly kept plodding ahead.  I don't know if you want to follow along or not, but here is the quick model I made of 1/4 of the cylinder.  You may want to see how I did it.  It's really just a 1/4 circle sketch, with lines to close the area, with the correct radius, then padded to length of 5.5".  Then I made a sketch of the inside, using the symmetry.  So the height is equal to the ID/2, and the length is 5.5-2x0.25 = 5".  Then I added a constraint preserving fillet, and set the radius to 1/8".  Once the sketch was constrained. I used the grooving tool, and revolved the sketch about the correct axis.  My model is converted to metric, because when I was learning the tool, all the videos were metric.  So all the imperial dimensions are metric in my model.  This is what a 1/4 of your magically machined model looks like.  Let's see if I can attach the model.  It's small.  I will change the extension to zip.  Just change it back to FCStd. 

[rsterne]

You had me at "L" vs "LT", so I had to look it up.... The Longitudinal (L) direction is the direction the material is rolled or extruded in.... The Long Transverse (LT) direction is across the width of the plate (and I would assume around the circumference of a tube).... The Short Transverse (ST) direction is across the thickness of the plate (radially, across the wall of a tube?).... It the part is made in such a way you cannot tell the direction (eg. a round rod), then they just use "T" for Transverse.... Here is a chart on 2024-T3 showing the stress/strain relationships at different temperature.... I would only use the room temperature data....

https://icme.hpc.msstate.edu/mediawiki/index.php/Aluminum_2024-T3_Stress-Strain_and_Fatigue_Life_Data.html

Bob

[rsterne]

That model looks perfect, it is exactly what I want.... I am primarily interested in two thing.... How the stress on the ends compares to the hoop stress, and if a radius equal to the wall thickness reduces the stress below the hoop stress.... If the stress on the ends is greater than the hoop stress, we will have to try one with thicker ends.... 

I tried to download that file, and all I got was a .zip folder with a bunch of files in it.... I could not get either of the .xml files to open in my FreeCad....

Bob

[WobblyHand]

Yes, we all have been tripped up on units!  Myself, more than once. 

Yes, the previous FEM sim of the other tube, failed to converge on the nonlinear sim at pressures over 75 MPa, but did converge at 70 MPa.

Next step is to add the FEA stuff, the model is complete.  So we select the FEM Workbench.  This is selected from the drop down menu where Parts Designer is normally displayed.  If FEM is not installed you can go to Tools > Addon Manager and install it.  Once in FEM Workbench, we add an Analysis Container, by selecting the yellow capital A.  On my display it is the first icon on the left, in the second row.

Then add the material, the icon for this is a big yellow circle.  If you hover over the icon, it says "Material for Solid".  Clicking this gets us an FEM material.  Unfortunately, there is no "material card" for 2024T3.  I had to create one.  I call it "Aluminum-2024T6".  I made a mistake in the name!  I used the generic aluminum card and added in the material properties that I found at https://asm.matweb.com/search/SpecificMaterial.asp?bassnum=MA2024T3 overwriting the generic values.  So I changed density, Youngs Modulus, and a few other properties.  I saved this material card into the same directory as where the other materials are.

After that, one needs to add in some constraints to the model.  We put in some displacement constraints on the quarter sections, so it behaves as if it were a full cylinder.  We also must pressurize the inside faces.  In our case, lets start out with something like, say 60 MPa.  We can increase this later!  So first lets do pressure.  Click the pressure constraint.  Edit the pressure to 60 MPa.  Then click Add, and control click faces that are under pressure.  Make sure we get all of them.  Of course, check that the pressure is in the right direction!

The displacement constraints are a little weird, but all we are trying to do is make sure the 1/4 section behaves exactly like a full section would.  Basically we line up the faces to the axes and figure out which dimension is pointing out of the screen and fix that.

Finally we need to mesh the solid.  We select the solid, in my case, called CavityGroove then click the icon Gmsh.  I typically give it a maximum element size, so the mesh makes sense for the part.  We'd like to have 3 tetrahedrons fit in the minimum wall thickness.  So I pick a number around 1mm.  Element dimension: From Shape, 2nd order element.  Click apply.  If we are lucky, we get a good mesh.  Sometimes we don't.  If it says "Clean run of Gmsh" we have been successful.  Sometimes playing with the max and min element sizes helps.  Then click ok.  For this mesh I used max size = 1mm, min size = 10um.  Not sure I like the mesh behaviour, but that is what I tried.

One last thing, before getting to the solver.  There's an incorrect setting in FreeCAD, that will utterly screw up your simulations.  We can fix that!  Took me a couple of days to find this...  At the top menu bar, select Tools > Edit Parameters.  Navigate the BaseApp to select Preferences > Mod > Fem > General.  Click on General in the group.  On the other pane, at the top is a parameter called "AnalysisGroupMeshing".  It will be set to true.  This is wrong, it must be set to false.  Double click true, and a pop up window will appear.  Change the value to false, then click ok.  Then click save to disk.  And then close.  Can't tell you how much grief this caused me...

Right click Analysis, in the Combo View and Activate Analysis.  Then go to the top menu and select Solve > CalculiX standard.  This adds a solver to the Analysis container.  Double click the solver (in the Tree in the Combo View).  The window says Analysis type = Static.  Click "Write .inp file".  If successful, it will state Write completed.  Then click "Run Calculix".

Then you get to explore all the linear simulation data at 60 MPa.  For what it is worth at 60 MPa, the maximum sheer stress in the inside corner is 340.9 MPa.  The von Mises stress is 600 MPa!  It appears the whole tube surface is very near to or at yield.

Once the file is meshed, the file grows a lot!  Put in your simulation, and it gets even bigger.  Just doing this exercise, has the model size now at 42.8 MB.  To reduce model size, I believe one needs to clear the simulation results and the mesh.  I did that, and the model is back down to 25KB again.

For anyone following along, the model with the linear problem set up for FEA analysis is attached as a zip.  You will need to have FEM workbench set up and you MUST do the Edit Parameters thing.  Other than that, just follow the instructions above, and you might find some interesting things to explore.

For the record, I am using Version 0.21.0 of FreeCAD.  I believe one can get a prebuilt AppImage to run on any platform.  I'm not a professional CAD person, I only learned FreeCAD in January of 2022.  Prior to that, I had no 3d CAD experience.  Likewise, I am not a professional licensed engineer.  Just a person with some interest in the field...
Prior to oh, three weeks ago, I had no idea how to do anything with FEM.  So I am telling you that I am not an expert, far from it...  Just trying to add to my skill set...

If you still want to play along, here is the set up file.  Change the extension from zip back to FCStd.  You need to do the mesh.  It is set up, but you need to Apply to remesh.  Likewise, you need to recreate the input file for the simulator and then run the simulator per the instructions above.

Hope no one is bored by all this.  If you are, well, there's lots of other gates and threads to explore on GTA!

Bob, saw your message while I was typing this up.  You should only have to do a file extension name change.  A .FCStd file is really a zip file.  So you should simply be able to open the whatevericalledthefile.FCStd file from FreeCAD.  I must ask, what version of FreeCAD are you using?

[WobblyHand]

That model looks perfect, it is exactly what I want.... I am primarily interested in two thing.... How the stress on the ends compares to the hoop stress, and if a radius equal to the wall thickness reduces the stress below the hoop stress.... If the stress on the ends is greater than the hoop stress, we will have to try one with thicker ends.... 

I tried to download that file, and all I got was a .zip folder with a bunch of files in it.... I could not get either of the .xml files to open in my FreeCad....

Bob

Rename the file extension, from xxxxxxy1.zip to xxxxxxy1.FCStd and open the main file in FreeCAD.  Don't try to look inside the file, it makes no sense to normal people!  What version of FreeCAD are you using?

At least on my computer, I changed the extension from zip back to FCStd.  I then used the File > Open to open the file.  Worked perfectly for me.  Hope you can get this part to work.  The version information can be found by Help > About.  It even has a button to copy to clipboard.  Copy it and paste in a reply message.

[WobblyHand]

You had me at "L" vs "LT", so I had to look it up.... The Longitudinal (L) direction is the direction the material is rolled or extruded in.... The Long Transverse (LT) direction is across the width of the plate (and I would assume around the circumference of a tube).... The Short Transverse (ST) direction is across the thickness of the plate (radially, across the wall of a tube?).... It the part is made in such a way you cannot tell the direction (eg. a round rod), then they just use "T" for Transverse.... Here is a chart on 2024-T3 showing the stress/strain relationships at different temperature.... I would only use the room temperature data....

https://icme.hpc.msstate.edu/mediawiki/index.php/Aluminum_2024-T3_Stress-Strain_and_Fatigue_Life_Data.html

Bob

Thanks for interpreting that.  As far as that stuff, I am admittedly out of my league.

[WhatUPSbox?]

Bob, I think you will need to pick a radius for the sharp corner. See paragraph "A Common Case of Ignoring Convergence " and many similar discussions for FEM analysis of sharp corners
https://www.nafems.org/publications/knowledge-base/the-importance-of-mesh-convergence-part-1/

[WobblyHand]

Bob, I think you will need to pick a radius for the sharp corner. See paragraph "A Common Case of Ignoring Convergence " and many similar discussions for FEM analysis of sharp corners
https://www.nafems.org/publications/knowledge-base/the-importance-of-mesh-convergence-part-1/

Good point.  Sharpest inside corner I could make is 0.002" with a very sharp triangular insert with a 0.002" nose radius.  Great for low deflection boring, but kind of fragile.  TPCB-2200-CM1.  Most other inserts have far blunter nose.  What's a typical sharp but not too sharp radius?  Don't have my insert charts nearby. 
CCGT21.51 inserts are (1/64") those are common, but that's 0.4mm.   0.4mm nose radius would be common.  That's also what Marko said was in the other tube.  I can add in a 0.4mm radius to the model.  It will take me maybe a minute or two.  The beauty of CAD...

Done.  Took 12 seconds.

[rsterne]

Yep, 0.4mm will work for me.... I would expect it to fail in the corners with that!.... I have FreeCAD Ver. 0.20.2 which appears to be the latest "stable release"....

When I click to download the .zip file, it ends up a folder in Windows 11, and I cannot rename it to .FCStd.... so I am kind of hooped at this point....

Bob

[WobblyHand]

Yep, 0.4mm will work for me.... I would expect it to fail in the corners with that!.... I have FreeCAD Ver. 0.20.2 which appears to be the latest "stable release"....

When I click to download the .zip file, it ends up a folder in Windows 11, and I cannot rename it to .FCStd.... so I am kind of hooped at this point....

Bob

No worries, we'll get it figured out.  Didn't originally understand that Win11 was playing around and unhelpfully helping you.  In a little while I will try renaming it as a text file, which I hope you can rename.  If all else fails, I can email the FCStd file directly to you.

You are not hooped!  The cylinder might be, but not you.

Bruce

[WobblyHand]

Ok, one last try, before I send it direct.  Don't know if this will work, but it is worth a try.  File is renamed as a .txt file.  FCStd files are actually zipped.  Don't know if Win11 will have heartburn with this.  So download and rename the extension back to FCStd.  Let me know if it works.  I was hoping that others could have access to the file, so they could possibly learn as well.

I have version 0.21.0, so it is possible that the file is not compatible.  I have an AppImage with 0.20.2, I will try to open the file.  0.20.2 opens the FCStd file.  It also did the mesh.  So I think you can use it.

This fiddling around with files to satisfy the forum filters isn't much fun.  Do you know if it is possible for GTA to allow FCStd files?  On a different forum that I am a member, I made a request to allow these files as attachments.  There is no problem attaching and using FreeCAD model files on that forum.  Might be worth looking into.

If this attachment proves to be an issue for you Bob, let me know.  There has to be a way to do this...

Sent you the files directly, just so you can get going.  I would hope we can solve the attachment issue here, so others might be able to benefit.

Bruce

[rsterne]

I was able to download the file as a .txt file, save it in my FreeCAD folder for this project (good suggestion) and then by clicking "View" and then "Show" and then "File name extensions" (which are normally hidden in Windows) I could change the extension from .txt to .FCStd and it opened in FreeCAD.  So, for those using Windows 11, that is how you can do it!....

Bob

[WobblyHand]

I was able to download the file as a .txt file, save it in my FreeCAD folder for this project (good suggestion) and then by clicking "View" and then "Show" and then "File name extensions" (which are normally hidden in Windows) I could change the extension from .txt to .FCStd and it opened in FreeCAD.  So, for those using Windows 11, that is how you can do it!....

Bob

Terrific!  Let the questions begin...  If there's anything you need help with, let me know.  Other good references are the FreeCAD documentation files, and the FreeCAD forum.  I have had a lot of help there, and believe me, I have asked lots of odd questions.

There's a lot to digest, but go slow, review how the solid was made and constrained, both in Part Designer and the FEM Workbench Analysis section.  Hope this helps lift some of the fog.

Bruce

[WobblyHand]

I was able to download the file as a .txt file, save it in my FreeCAD folder for this project (good suggestion) and then by clicking "View" and then "Show" and then "File name extensions" (which are normally hidden in Windows) I could change the extension from .txt to .FCStd and it opened in FreeCAD.  So, for those using Windows 11, that is how you can do it!....

Bob

When I was using Windows (at work) I always had show extensions on in File Manager.  Made things easier, to know what kind of file I was dealing with. 

[rsterne]

Bruce, I was able to add the material from the list, where I have added 2024 (although it doesn't show my choice in the model table), and I added the pressure to the inside faces (I assume the arrows should be pointing inside to outside).... I don't understand how to add the displacement contstraints....

Bob

[WobblyHand]

Bruce, I was able to add the material from the list, where I have added 2024 (although it doesn't show my choice in the model table), and I added the pressure to the inside faces (I assume the arrows should be pointing inside to outside).... I don't understand how to add the displacement contstraints....

Bob

Good, some progress!  Can I assume you have the FEM Workbench active?  Can you show me the whole FreeCAD application screen?  Does it look similar to this?
See up by the top left, a little drop down box?  If it says Part Designer, change it to FEM.
If you see an A in the Combo View, you can click it to open the Analysis section.  It should look similar to mine.  This file has all the constraints, and the set up for the mesher.  You shouldn't have to change the constraints, but you can double click on them to see how they are set up.  If you got this far, good.  If not, please show a picture of where you are in the process.
If you double click the MaterialSolid, what does it say in the box: Material card.  It should say Aluminum-2024T3.  If it doesn't then you need to navigate to find the card.

[WhatUPSbox?]

Bruce, I'm following your cookbook but when I try to do the "Preferences > Mod > Fem > General" fix, there is nothing in the right pane. Is there anything I need to install?
This is FreeCad 0.20.2