Re: Connection/Idealization in Modal Analysis

ptc-5183982 · ‎Dec 16, 2014

I have a modal system that contains two parts, each constrained to ground. I would like to insert a zero displacement condition between points on each of the parts' extremeties. This is for validation of another software, which is why it is very simplified. It seems like a rigid link should do the trick, although I don't have an advance license and can't use advance rigid links, which I assume would have the ability for the link vector not to change. However, even when I try inserting a rigid link, the analysis fails. Also, a spring idealization doesn't affect the system at all. Is there something about modal analysis that is incompatible with these tools?

As always, I've been super disappointed with the Creo knowledge base. I don't even know where to begin looking for info.

sdensberger · ‎Dec 16, 2014

A simple rigid link should work. What error message do you get for the failed analysis? What do you mean by, "a spring idealization doesn't affect the system at all"?

Would it be possible for you to post the model itself as well as additional information regarding the setup and what you're trying to do?

ptc-5183982 · ‎Dec 17, 2014

Hi Shaun,

Thanks for the reply. When using a point-point simple rigid link, I get a fatal error that a solid element has singular mapping. That seems understandable, I think. I managed to get surface-surface rigid links to work on non-parallel surfaces (that one alludes me). Unfortunately, this is not entirely physical, or rather does not represent the nonphysical model I am trying to validate. I essentially want to connect points that will have zero linear displacement but not constraint in rotation relative to eachother. a surface-surface locks all six degrees of freedom between the two surfaces. Not only this, but I was hoping that not only the vector magnitude between the two points remained constant, but also that the unit vector stayed constant.

For some reason, the modal results were the exact same with and without a very stiff spring, but it seems like the rigid link is a better solution to pursue.

I have the model attached. Forgive me if there is something wrong, I am not very comfortable with the file architecture of Creo yet. I had some Library parts, so I'm not sure if they will transfer correctly. There are two points embedded in the parts I am trying to connect (RUN and HSS). This is a validation for AutoPIPIE idealized support and beam models. See below for a screengrab as well.

Cheers,

Curran

sdensberger · ‎Dec 17, 2014

When using a point-point simple rigid link, I get a fatal error that a solid element has singular mapping. That seems understandable, I think.

The issue is that solid element do not have rotational DOF. Given the thin nature of your geometry, you can use shell elements, which have all 6 DOF; using an Advanced Rigid Link will allow your to connect the two points while also releasing the rotation DOF (but still preventing rigid motion of the link). I've made modifications to you CAD model that reflect the best way to go about this (I deleted some geometry and replaced it with surfaces to resolve a Hard Point mesh control issue).

For some reason, the modal results were the exact same with and without a very stiff spring, but it seems like the rigid link is a better solution to pursue.

Not surprising. A rigid link and an infinitely stiff spring are the same. Even though the spring you've define has finite stiffness, it's most likely much large that the local stiffness associated with the elements it's connected to. You're also only looking at 5 modes and your analysis is set to Quick Check (which means it solves the entire model at a uniform p = 3 and does not escalate any element orders; I suggest running an MPA analysis with max p = 9 and 1% convergence). Also keep in mind that the lower frequencies are more accurate if a higher number of modes are included.

With all that out of the way, you should also consider whether it's realistic for connect the two parts together via two points. It's almost certainly more appropriate to connect them via two small surfaces. If you do need to do this, you can use an Advanced Spring or a beam element along with either rigid links or weighted links to capture the kinematic nature of the connect. If you run the model I uploaded you'll notice it's hard to get the first mode to converge; this is due to the point-to-point connection (which looks like it's converging to something in the low 6 Hz range). If you were to make a small surface-to-surface (using the method mentioned) you'd get better results. Even a small surface (0.2" diameter) causes a significant change in the first mode (around 29 Hz).

EDIT: I'm having issues uploading the CAD model. If you private message me an email address I can send it to you.

ptc-5183982 · ‎Dec 18, 2014

I mean tto say that the stiff spring modal was the same as having no connection at all, not compared to a rigid link. I had played around with the analysis method to see if that was the cause of my problems, but it seemed my results were fairly consistent regardless of method. Although since the method you described was the same as what Mats returned to me in his assembly, it runs muuuch faster.

I agree that what I am doing is not realistic, but we are trying to validate a simpler, 1D software. It is not the actual connection in question I am looking at, so I want to model it similarly, not realistically.

You guys have given me a good framework for discussing this, and gotten me to a point where I am almost there. If I go with this beam model, I would have all rotations released on both ends, but have the beam axis itself locked to a global axis (i.e. its original trajectory). In this situation, the points in question would always overlap when projected vertically. The more I explore, the less I think this is possible.

sdensberger · ‎Dec 18, 2014

If I go with this beam model, I would have all rotations released on both ends...

I don't really see a reason to release both end, but if you do make sure you have rigid mode search on for the modal analysis; the beam will undergo rigid body rotation.

In this situation, the points in question would always overlap when projected vertically. The more I explore, the less I think this is possible.

Let me preference this by saying that I could be misinterpreting what you're trying to do. With that out of the way, I think you've defined a contradictory condition. You cannot have two points that have a unit vector between them remain constant while also allowing rotations to exist at either point, especially when these points are part of a deformable surface.

ptc-5183982 · ‎Dec 18, 2014

I'm not sure if its a contradictory condition, or just one that can't be defined in creo. As I calculate it, the beam model would have 1 rotational degrees of freedom (fixed to WCS) around its axis, and translational DOF would be fixed to both points. I could fix that one rotational DOF to one of the points to get rid of it.

Regardless, I think I may have found an acceptable solution. I have the beam fixed to the pipe and all rotation released at the HSS. I had to insert an extra constraint on the pipe to lock both horizontal rotations, but that keeps the beam element vertical. It would have been ideal if I could have put that constraint directly on the beam, but at least this is something I can alter in AutoPIPE for a good comparison.

Thanks Shaun!

mlindqvist · ‎Dec 18, 2014

Here's my suggestion. Just as Shaun suggested with a beam element plus links at the beam endpoints. It should work with both rigid and weighted links, I used rigid, in case you don't have the advanced license.

I have also used beam release on each beam endpoint so I can control wether the beam does/does not transmit bending moment. In the attached model I have set the beam releases so that none of the endpoints can transmit any bending moment. Note that at least one of the beam endponts should have the rot-x fixed, otherwise the beam can spin around its own axis which would give you the "fatal error - the model is insufficiently constrained" and this modeling error can be difficult to identify.

I hope this helps...

/Mats L

ptc-5183982 · ‎Dec 18, 2014

Thanks for the model! I've always avoided shell models because it has never made 100% sense to me, but im starting to understand it more given that my model has been converted. Thanks! Funnily enough, when I released all rotations, I did not receive a fatal error and instead, it seems to have generated identical results. This is very close to what I am looking for, but unfortunately still not completely representative. To repeat what I said in my reply to Shaun, A similar beam connection, where all rotation at both ends is released, but where the beam itself remains parallel to the global vertical axis would be an exact representation of AutoPIPE (which I am starting to realise is way oversimplified if Creo can't even simplify a connection enough to match it).

I did some additional exploring down this path, and it is starting to look impossible, at least with this solution.

EDIT:

As per my reply to Shaun, I modified your solution and the AutoPIPE model to a point where I think they coincide. Thanks for your help!

mlindqvist · ‎Dec 19, 2014

I think this model does what you describe... The beam is now divided in two, and the midpoint is constrained so that is is free to move in x-, y- and z directions, but restricted from rotating.