Mesh is correct on one computer and completely different on another?

[Anonymous]

Hello!

I'm currently teaching my students how to use and understand Inventor 2013+2014 Stress Analysis. One of my students is getting a totally different mesh result (Different amount of nodes and elements), than myself and his other classmates. To describe the situation more specific, we're testing a steel tube with the dimentions 75x75x7,5mm, with an extrude of 1500 mm.

As you can see in the pictures below (Named Teacher Results and Student Result), the students results is nearly 2 times as much as me (the teacher).

When we look at the mesh settings, both settings are appearently set to the same values:

Average Element Size = 0,100

Average Element Size in Shells = 0,050

Minimum Element Size = 0,200

Grading Factor = 1,500

Maximum Turn Angle = 60°

(Also see this in the attached file "Meshing Defaults")

Am I completely missing something here? Is it just a simple thing I need to edit?

Hope I can get some help in this. The only solution I did with my student, was him changing computers, but he also have this problem on his computer at home.

Regards,

Peter

[JDMather]

Attach the *.ipt files here.

The most obvious quesstion, are units the same?  (in, mm)?

If you continue this discussion I can share a lab with you that demonstrates why diffferent students can get different results on the "same" geometry.

[Anonymous]

Hey JDMather.

I think attaching the .ipt files to you, will give the same results as the "Teacher" files. I have already tried doing so with my student, transfering my file over to his computer, where my correct mesh, gave a completely different result on his computer. 

Another odd thing, is that all the students computers are a ghost of mine. So that -should- rule out, that Inventor has been installed differently on my students computers. 

As another note, all the computers are using millimeters (mm) as units. We have already been looking basically everywhere in setups and checked if things were correct.

Anyways, I've sent you my file, so you can check it out yourself, but as already mentioned, I've already tried transfering it to my students computer, and his elements and nodes were nearly 2-3 times as many as on my computer.

Besides, you also need to know what materials I'm using, where I'm using constrains and where + what my load is.

Material is Steel, Mild

Constrain is on the end face of the hollow section

Load is on top face of the hollow section, with a 6kN load.

Anyway, hope that helps you out a bit, and I hope you could demonstrate why different students can get different results on the same geometry. 🙂

Regards,

Peter

[JDMather]

---

@Anonymous wrote:

Besides, you also need to know what materials I'm using, where I'm using constrains and where + what my load is.

I hope you could demonstrate why different students can get different results on the same geometry. 🙂

---

Your problem description indicates that all confounding variables (same file) have been removed, but all the evidence you have supplied indicates otherwise.

1. (this isn't really the problem) the part is modeled incorrectly the wall thickness is not uniform.  The inside radius should be outside radius-thickness.

2. The material, constraint and load information is (or should be) stored in the file.  The file you had attached here did not have any of that information.  No material applied to the part.  No constraints applied to the part. No load applied to the part.

3. The images you attached of Instructor and Student show that they are not the same file.  I need these two different files represented by the images to demonstrate why the differences.

  This is from your original post.

[Anonymous]

Hey JDMather.

Thanks for the quick reply.

However, I think you're misunderstanding what I'm trying to point out here. I apologize if my two pictures has confused you, since I can see where you're pointing out the errors. (Teacher's picture got materials, constrains and loads, whereas students picture has not). That's a mistake from my part, but the dimension should be the same. The hollow section with 75x75x7,5 mm. Meaning the geometry should be the same. Meaning, if you created it, the geometry should be the same, even though the name of the file is different. It is the same profile, all around, so technically, it shouldn't have much meaning if I sent you the teachers file and the students file, as they are totally the same.

So, basically, when you create the profile with the 75x75x7,5, you then extrude it 1500 mm. That's the profile we're going through in our FEA compendium, which we downloaded from an technical academy's website (Which usually educates engineers, productiontechnicians, etc). I'm educating basic technicians, called Technical Designers, which translated to an american study, could be technical drawers, or something of the sort.

I've just asked my student to take his file (even though the geometry on his and mine are the same), and make a screenshot + saving the file and sending it to my computer, so I could run it through the same process. And correctly enough, his mesh (elements and node amount) is 2-3 times as much as when I ran through the same proces on my computer.

I've attached the files, so you can see for yourself. Even though my file is called "Bjælke Stress Test" and his is called "opgave 1 stress analyse", I'd like to point out once again, that the geometry is the same. Please note that I've also added in the Von Mises Result, so you can see my students MPa amount is higher than mine, even though his displacement is the same as mine. (around 7,5 mm).

So, what I'm actually thinking of asking you, is that you try and create the profile yourself and see if your mesh is closest to me (teacher) or the student, and then explain to me why it's like that, and if you have any solution to it (Which I can feel that you got. 🙂 )

Regards,

Peter

[JDMather]

AI am going to demonstrate why there can be differences and in the end I will post a lab document that I have my students go through in several different FEA programs to demonstrate differences.

I have to leave for a while, but I will be back.

Notice that the profile sketches for the two files are different.

 I know, you are going to tell me, "But the geometry is the same."  We'll get to that.

[JDMather]

I'll try to get back to this discussion as soon as I can, but I thought you might be interested in my list of topics for my FEA class.

I have been where you are now and experienced the same struggle.

But once you understand why you are seeing different results - it all starts to come together and the students will see equivalent results.

[cieslak.maciek]

JDMather, I read this topic back and forth like four times and still no solid anwser.

The algorithm of building the mesh on a model is based on geometry of the part, nothing to do with the constraints. Are you suggesting that if I build my model one way I will get one kind of results, and the other way the results will be different? What about STEP files, where there are NO constraints or sketch data?

@wakka3d - I hope and think that you know, that on the same geometry the FEA results and mesh will vary let's say +/-10% or even more which is normal. I understand you are concerned why the two computers with the same mesh settings created meshes that are almost 4000 elements apart. My first question is did the versions of Inventor on your and students computers vary. If yes, did you have the 2014 version updated with the update 2 service pack 1?

Users of Inventor FEA discovered that 2013 and 2014 versions produce totally different results (the topic was on that forum sometime in October/November), and the issue was resolved with Update 2 Service Pack 1.

Here are my rusults for your part and data:

Unfortunatelly I don't have the non service pack Inventor installed anywhere. But as you can see my nodes/elements # is pretty much the same, so that is good enough for an automatic algorithm.

If it's a wrong idea why it is so different, then you will have to give us those two files, not only the one you have.

Note - for Update 2 SP1, you need to have installed SP1 first.

[JDMather]

---

@cieslak.maciek wrote:

JDMather, I read this topic back and forth like four times and still no solid anwser.

---

I haven't even gone near the solid answer yet.

I am trying to guide critical thinking about confounding variables - there is something different - there are differences - I have already clearly demonstrated differences.  All of them must be identified to understand the problem.

[JDMather]

---

@cieslak.maciek wrote:

Are you suggesting that if I build my model one way I will get one kind of results, and the other way the results will be "different"?

 I hope and think that you know, that on the same geometry the FEA results and mesh will vary let's say +/-10% or even more which is normal.

---

Yes, that is what I am suggesting (and will prove) and I note that you state (not just suggest) the same thing.

Although I will have to define "different".

[cieslak.maciek]

But I don't see any point in stating obvious stuff. I can compare meshes and results from Ansys, Abaqus, Simulation Mechanical and Inventor FEA...but that's not the point.

The base problem was that wakka3d got 4000 nodes on one machine and 8000 on the other with exactly the same settings on exactly the same geometry. And this is not normal, if you get 50% less dense mesh. IMHO this should not happen.

[JDMather]

---

@cieslak.maciek wrote:

The base problem was that wakka3d got 4000 nodes on one machine and 8000 on the other with "exactly the same settings" on "exactly the same geometry". And this is not normal.....

---

Hmmm, what would that seem to indicate?

[cieslak.maciek]

What do you mean?

[JDMather]

---

@cieslak.maciek wrote:

 If yes, did you have the 2014 version updated with the update 2 service pack 1?

Users of Inventor FEA discovered that 2013 and 2014 versions produce totally different results (the topic was on that forum sometime in October/November), and the issue was resolved with Update 2 Service Pack 1.

---

Neither file that the OP attached has been saved in 2014.

As far as I know, the 2013/2014 difference you describe as resolved by Update 2 to Service Pack 1 was only relevant to assembly FEA, not individual part FEA.  But it might be (indirectly) relevant since neither of these files (attached here by the OP, can't make any determination about whether these files accurately represent the images the OP posted) was migrated to 2014.  (there is another HINT in there)

[Anonymous]

@ cieslak.maciek - Yes, I know that it should normally swing with about 10% +-. However, as I already mentioned before, the computers on the school is all ghosted from one machine, which means that all the other machines have installed the programs in the exact same way, meaning there's no way that a service pack can be installed (I think the computers have service pack 1).

@ JDMather - I thank you for the replies. While I'm still a bit sceptical with how you're gonna show me why the meshes can be different, since the student did everything I did, I'm still eagerly awaiting your guide, so you can show me why it does that. 🙂

[JDMather]

This will take a few more steps to get to answer to your original question, but open these two files and run the simulations without making any changes. 

The geometry is identical, correct?

The constraints, material and load is identical, correct?

All analysis settings identical, correct.?

Results identical?

[Anonymous]

@ JDMather - I've now downloaded your files, ran a mesh on both of them, and now see, that the differences are 3 for nodes and 5 for elements, on my computer. I've attached some screenshots, so you can see. And this was the ONLY thing I did.

Don't know if this is intended, but I eagerly await the next explaination. 🙂

[JDMather]

You got the same results that I got in both 2013 and in 2014.
Now have your student run the same two files and report back the results.

[cieslak.maciek]

Could you just enlighten us, what's all that for?

[JDMather]

• ---

  @cieslak.maciek wrote:
  Could you just enlighten us, what's all that for?

  ---

• No.  I don't skip steps.  You can ignore this thread if it bothers you.

Uhmm, actually - I thought that is exactly what I am attempting to do.