When can you trust Inventors Stress Analysis?

[ozitag]

Hi guys,

Here is a really weird result. Can anyone tell me how this can happen in real life. Admitedly I haven't added gravity, but otherwise it should all be possible. Two Pin Constraints in the two holes and a Force in the direction of the arrow of 300N. The bar is 30x10 x 376mm. The holes are 22mm and spaced 260mm apart. The force was placed on an area 30x10mm on the end of the bar. As indicated it is mild steel.  There are threaded holes along the top and certain intervals.

It shows the bar bending in the complete opposite direction!!!

I Finished the analysis, suppressed the threaded holes and tried it again, worked ok. Unsupressed the holes again and it worked ok. All this has happened directly after restarting Inv as before it was showing that there was no reaction what so ever.  Even to a force of 600000N. I'm sure it would bend a little with this sort of loading.

It should come up with some error when it can't get a result rather that just something. I'm worried about people who use the software without any idea of real life. Under design something that is critical.

Regards,

[JDMather]

Attach the file here.

[ozitag]

Here you go. As I said, I got a decent answer after I had supressed and the unsupressed the threaded holes.

Regards,

[JDMather]

There is no study in that file?

I am using 2012 - so I don't know if I could reproduce anyhow, but from your description I assume you had only the two pin constraints.  Did you set it to Detect and Eliminate Rigid Body Modes?

I set to test convergence at 5%.
This the result I get (with Detect and Eliminate Rigid Body Modes turned on) in 2012.

[IgorMir]

I am getting pretty much the same results Jeffrey got.

Here is IV 2010 file.

Regards,

Igor.

[henderh]

Hi ozitag,

  If you look closely at the image you provided, the holes are visible in the wireframe (original model) but do not appear in the results display of the deformed body.

  There was something definitely out of sync going on.

-Hugh

[blair]

I would guess, that you have really applied you load to the full top face of the bar. Either select the edge or do a Split-Face as in JD's model. If you look close at the end were the load is applied you can see a line running across the top face of the bar.

Even though your load indicator arrow is shown beyond the last pin, unless you have defined a small area for your load to be applied, Inventor will by default apply the load to the full face of your bar.

When every possible, suppress small features such as the holes unless you are really going to need them in your analysis. They just create smaller/finer mesh and increases your analysis time.

[ozitag]

As Hugh pointed out, there was something out of sync here. I had done a split face as suggested here, and the arrow was showing the force in the centre of the split off area. And I am sure if everyone does it on their own system they will get the correct result. (maybe, who'd know) The problem I have is that it clearly gave a wrong result to me when I first did it, and I can't see any reason for it other than inventor got out of sync somehow.  As I mentioned before, I closed the stress analysis environment, supressed the holes on top, tried the stress analysis again and it worked (without changing anything), so I went out of the environment, unsupressed the holes again, did the analysis again (without changing anything) and it worked. I do agree that an FEM result is only as good as the info put into it, but when everything put in is correct, it should give a correct result (or at least an error saying the model needs to be updated). How can someone know if the model is out of sync or not? Yes you can use your own intuition, but I'm scared of the backyard builders who have downloaded some pirated copy of inventor that don't have much intuition. The biggest problem I see here is I clearly got incorrect results when everything put in to the system should have given correct results. After all, it wasn't a very complicated part!!!

[JDMather]

---

@ozitag wrote:

 The biggest problem I see here is I clearly got incorrect results when everything put in to the system should have given correct results. After all, it wasn't a very complicated part!!!

---

You were fortunate that the results were clearly wrong.  Often it is not so obvious and not a fault of the software but an error on the setup for analysis.

In my class we solve the same problems in Inventor, Algor, SolidWorks and Creo.  Sometimes when going from one program to another we discover a fundamental error in previous analysis that we didn't realize had occured.  Also I have fifteen students running the same analysis - so when something comes out different the result immediatly come under question.  Between using different softwares and different students making errors we come to appreciate how easy it is to make an error and how hard it is sometimes to even realize there was an error or find the error.

I believe this stuff should become as accessable to the "casual" and as easy to use as speel check in MS Word, but at this time it still requires a lot of caution and like spell check we might correctly spell word (enter data) but use entirely wrong word (data).

When I talk about casual users using FEA the old-school classically trained engineers roll their eyes.