Bearing vs Force

[RichardOliver4462]

I have a model, and when I apply a bearing load to a lug get very different global results to a force of the same magnitude applied. I believe that this is only the case since the 2018 update...the bearing load seems to apply forces radially and not just sinusoidal distribution in the direction of the load.

I suspect that the application of a bearing load is not what I was expecting. For instance if one has a beam with a lug in the middle and fixed at the ends, applying a vertical bearing load  to the lug, to simulate a shackle pin does not give vertical reactions at each end of half the applied load. In fact the reactions vertically are very small and there is a large transverse load at each constraint.

Can you confirm, if either the bearing load application is not working correctly, or I have misunderstood the type of loading applied.

[John_Holtz]

Hi Richard (@RichardOliver4462)

The bearing load was changed in version 2018 so that the forces would be radial (or normal to the surface).

However, it looks like it is not working. The reaction forces (SPC Summation) are definitely not equal to the entered load. I am checking on the status of this.

[John_Holtz]

Hi Richard,

The developers are looking into this issue, assuming that what I tested is similar to what you are seeing.

I applied a bearing load in the Y direction to a hole. The reaction forces (the "SPC Summation" results) should equal the applied load but were only about 2/3 of the applied load (depending on how fine the mesh is). Whatever is causing the difference in the reaction forces would also affect the displacement and stress.

[John_Holtz]

Hi Richard,

Thanks for reporting this problem. The developers are working on a fix. I am not sure when the fix will be ready and released.

[Anonymous]

Hi John,

last week I stumbled upon the bearing load problem also. But it seems to be more serious than I thought at first sight.

Not only the component of force in the load direction is to small, as Richard reported. The distribution of the bearing load is completely wrong, especially near the edges of the cylinder face. Therefore the deformation result (and hence strain and stress) can be extremely unexpected.

Attached there are two results of the same model. One is from Release 2017, the other form release 2018. You see the direction of the main deformation in 2018 is opposite to the expected and plausible result in 2017. Furthermore the model gets twisted, although it is fully symmetric.

And yes, the whole thing is very mesh sensitive. With a slightly finer or coarser mesh, you get a total different result.

I can send you the model and some documentation if necessary.

--

Michael Puschner
Mensch und Maschine
Germany

[John_Holtz]

Nice test model Michael. Thanks!

I created a similar model and was able to get the part to displace to the wrong direction in 2018. As you mentioned, this behavior was totally dependent on the mesh size. (In fact, I could not reproduce the problem until I read your email more closely to see that it is mesh dependent.) I have provided my test models to the developers.

[Anonymous]

Hi John,

good to know that the developers are aware of this issue. Hopefully they are working on the problem and we will see a fix soon. Bearing load is useless as it is now in 2018.

Thanks

Michael

[Anonymous]

Any idea when we will see a fix for this issue?

Michael

[John_Holtz]

Hi,

It is on the schedule, but it is too early to know when it will be released.

[mrkwint96]

Can you give us an update on this?

Also, is this issue also present in Nastran 2018 (not In-CaD)?

Thanks

[John_Holtz]

Hi all,

The update on the status will only be known once the software release. Before it is released, any work that is done on improving the bearing load is not guaranteed to be released. (Another way of stating it is that if it is "fixed" this week, testing may reveal something that prevents it from being released next week.) So I cannot speculate on when it will be ready.

For @mrkwint96 second question, the source of the problem is within In-CAD. In-CAD calculates the loads on the elements and writes them to the Nastran file (.NAS). The Nastran solvers (stand-alone Nastran) is not doing any calculations for the bearing load distribution. So stand-alone Nastran is not related to the problem.

[Anonymous]

Good morning and a happy new year,

is the problem now solved with update 2018.2 ?

Can't find a hint in the regarding readme.

Thx!

[David.Kind]

@John_Holtz: I'd like to formally ask what kind of QA program Autodesk has for Nastran/Nastran-In-CAD. I'm disappointed that such an error (and others that have been documented on this forum) have slipped through the cracks. We (the end-users) rely on these tools for real-life problems. Can Autodesk provide any evidence that they take this seriously; or should I start looking for an alternative product?

I'm aware of the Nastran In-CAD verification manual. But at 264 pages it doesn't seem to be enough. The verification manual doesn't address the pre- and post-processing at all. And bugs such as this bearing load slipped through without any problem. 

Autodesk should take a look at the development of CODE_ASTER. That group takes quality very seriously; and it's for a product they release for free. The CODE_ASTER validation (verification) manual alone has 1257 documents in it

We can take this discussion to another medium (email or whatever else) if necessary.

[John_Holtz]

Hi @David.Kind

I received your post, and we are working on a reply.

[Rory_M]

I'm also very interested to find out if bearing load is fixed - We're in the process of rolling out PDMC 2018 and Nastran is one of the reasons we moved across to subscription. We use bearing load quite often (currently using the FEA within Inventor Pro).

[Anonymous]

Hello All,

I just wanted to check for an update on this as well.  I was going to make a separate post for this but I might as well keep things together.

I have tried running the same analysis of a bearing load on a bearing support ring that is tabbed, piloted, and bolted to a constrained ring gear.

 

The basic Inventor FEA package (Stress Analysis) seems to interpret the reaction reasonably.

When I try the same thing in Nastran, the load is completely uneven and gives a distorted result.

I read above that the defect was recreated by adjusting mesh refinements.  Does that mean that this can be avoided with sweet spot for mesh size?  I have tried both coarse and fine mesh and both have resulted in a distorted result so far.  Any advice or ETA of the patch is appreciated.

Derek

[Anonymous]

Hello,

I must say I'm really disappointed how Autodesk handles the whole FEA product line!

Now we got 3 (or even more) FEA solutions:

1. Sim Mechanical, which will not be continued.

    Therefore a no go for the future

2. Nastran Incad is not nearly a finished product

    Reported bugs are not fixed and ignored

    functions which worked in earlier release versions don't work anymore (e.g. bearing loads, multi select contacts)

    cloudy stress distributions

    more random mesh size over a surface

3. fusion 360

    very poor working perfomance !

What is the Autodesk roadmap regarding FEA?

Shall we End-User look for other products, what is your recommandation for valid future orientated FEA tool which does already work?

[John_Holtz]

Hi everyone,

The bearing load is not fixed in In-CAD 2018.2 (but multi-select in the browser that Mortiz_R mentioned is fixed). Other fixes in 2018.2 are described in the What's New.

Here is some background information about the bearing load:

In the 2018 version, we changed the behavior of how the bearing load worked from applying a force in the direction of the load to making it normal to the surface.  In doing so, it makes the applied load very dependent on the face orientation of the elements the load is applied to.  We are currently looking into reworking the entire approach so that it can provide the load in the normal direction as requested and still provide high quality results in all meshing scenarios.  We cannot provide a firm estimate on a release date as it is still in active development and must go through the QA checks.  If all goes well, this should be included in the next release of In-CAD.

[Rory_M]

Thanks @John_Holtz for keeping us informed. I have to say that I'm a little disappointed (as I'm sure you are) that such a major feature is not functioning correctly, and the only solution is to wait it out till the next release.

We use bearing load often and I'm thinking that we'll have to wait till 2019 is released before using the software in anger.

Talking of anger, my boss won't be happy when he finds out that one of the major reasons we moved from perpetual to subscription licenses is broken.

On a purely technical level, why was the change made from the old method to this new implementation? Is the new version (when fixed) proven to give more reliable/accurate results?

Also, is a patch that reverts to the old method feasible, just to get things going while the big fix is worked on? I guess not but it is worth asking about.

[RichardOliver4462]

I see there is some disappointment at the speed to which issues like this are addressed. There were other items which worked in earlier releases but don't now - for instance load arrows moving with the deformed shape.

Why doesn't In-CAD have any Beta testing by users? This used to exist and has for some reason been dropped and would have certainly picked up the issue with the Bearing Load.