Simple tipping analysis

[Anonymous]

Hello all,

Trying to run (what I would think) is a simple tipping analysis in MES with nonlinear material models. Scaffolding sitting on a slab, using a prescribed displacement to lift one side up so two feet will come off the ground while pivoting on the opposite two feet. At one point I was able to pull up on the one side but it acted as if the feet were bonded to the slab. All contacts are surface contacts with friction. I've tried both low & high speed contact. I've tried automatic, surface surface, point surface, point point contact types. Now everytime I try to run an analysis it won't even start & gives me a bright red "analysis failed" at the top of the design scenario tree. Whenever this happens I have to import the model from Inventor again & start over. Eventually the red failure note will appear again. It's like the file gets corrupted after so many times changing settings & rerunning the analysis. This should be a simple analysis correct? What am I missing? See attached.

I appreciate any/all help!

[AstroJohnPE]

Hi drpalo,

Is there a gap between the scaffold and the floor? Keep in mind that there are FIVE surfaces on each foot pad in contact with the floor: the bottom (which you specified as surface contact) and the four sides of the pad. So it might be that the side surfaces are bonded to the floor while the bottom surface is in surface contact. (I have not tested this.)

My suggestion is to create a small gap between the floor and feet (0.1 inch). You can then change the "Contact Distance" to compensate for the gap.

[Andrew.Sartorelli]

Hi Drpalo,

I've love to take a look at the analysis and provide you with some tips on the setup.

As far as the analysis failing with no warning messages, are you running any kind of endpoint protection? Where is the file saved to?

[Anonymous]

Good morning John & thanks for the reply. I played around with this more on Saturday morning. I removed the prescribed displacement & placed a vertical force (200lb) on one side of the scaffolding. This works, tilting the one side, but then it lifts the whole thing vertically like there's no gravity. I do have gravity on. Is prescribed displacement the issue? It can't be used in this situation?

Thanks.

[Anonymous]

Good morning Andrew & thanks for the reply. I received an open case email on this from Autodesk. Do you want me to reply to that email with the archive file? Or how can I get this to you?

Thanks.

[Andrew.Sartorelli]

Hi Dave,

If the archive is under 10mb you should be able to reply to that case open email with the archive is attached. I think based on the screenshot the model should be fairly small.

Andrew

[Anonymous]

Hello...


There is no too much information on what specific results of this excercise you are interested in.. so I will guess that you are just trying to (as the title states) tip over the structure and watch it fall. So, here is my take on that assumption and how I would approach it.


First, I wouldn't use a "part" to represent the slab. I would rather use a general impact plane to avoid extra elements an contact issues.

Second, I would guess that neither an upward prescribed motion nor an upward force would do the trick easily. I would guess that the prescribed displacement approach (If the analysis finally run as you said it didn't) would first lift the structure up and it would eventually start to turn (if properly constraint) as the structure's feet are at some distance from the "floor". The upward force magnitude would have to be carefully determined (maybe making it time dependant would help) to avoid the results you mention. So, instead I would use a lateral force and constraint the displacement on two of the structure's feet edges in the force direction. That combination will produce a "tipping" effect over the structure

Again, this is the simplest way I can think of setting up a "simple tipping excercise" but If there is some other phenomena that you are interested to see, there might be some other approaches that fit your requirements better.


I did a quick run with some generic geometry and in fact the structure tipped over. I attached some images.


Hope It helps in some way.


Bye

[martin_madaj]

Hi everybody,

regarding the red "Analysis failed" error: I've also seen this before. Closing and reopening the simulation file helped to run the analysis without error.

Martin

[Anonymous]

Thanks for the reply,

I tried using a general impact plane & immediately the "analysis is aborted". Also, I need the force or described displacement to be vertical. We are recreating an accident where a crane is lifting up on the one side of the structure. The ultimate goal is to have some beams laying on the top of this structure & when it starts tipping, the beams on top will slide off. I'm trying to get this simple version to work first.

Thanks.

[Anonymous]

Hello

Ok.. so the vertical force is a must.
But let me ask you something: in this scenario, are you guys assuming that the crane's hook (or cables, or ... ) is going to release the frame structure at some point (or for that matter that the frame is going to "slip away" from this "attachment") so that the frame structure will actually be completely tipped over on it's side? I ask this thinking in what I said in my last post about the upward force being time dependant in order to allow the frame structure to fall. Otherwise, thinking about the real scenario, I would guess, if the frame structure is strong enough, you would end up with a "hanging frame"

On the other hand, and just out of curiosity, I don't quite get your "ultimate goal" scenario. Are you intrested in the effects of the impact forces over the falling beams? Are you trying to see the falling beams "banging" the frame structure?

Have a nice day.

[Anonymous]

We are not looking to tip the structure over to the ground. With the beams laying across the top (see attached) we want to tip the structure to the point where the beams start to slide off. We want to find out at what angle or what distance the structure has to be lifted to make the beams start to slide. We're not looking for anything to fall to the ground.

[Anonymous]

We're not looking for stresses, so I changed everything to kinematic elements. I got the structure to tip but the beams fall straight down (see attached). The beams have a surface contact with the structure they are sitting on. What am I missing? Previously using brick elements, I ran it with only gravity & the beams stayed on top of the structure. Are contacts different with kinematic elements? Does mesh size matter in this situation?

Thanks.

[Anonymous]

I just ran the kinematic model with gravity only, no tipping. The beams stay where they should be on top of the structure, they don't fall straight down through the structure. What is it about the tipping motion that ignores the surface contacts?

Thanks.

[AstroJohnPE]

I can think of two possibilities as to why the beam falls through the structure.

1. Some type of bug in which the contact doesn't work in some circumstance.
2. The contact stiffness is not high enough to lift the beam as the structure is lifted. As a result, the beam "breaks through" the contact and then falls. Try increasing the contact stiffness for the beam-to-structure contact.

In other words, surface contact acts like springs between the two surfaces, so it has a stiffness associated with it. But if a spring compress too much, the software will "break" the spring. Depending on how quickly the structure is being lifted, the structure could be passing through the beam and breaking the contact.

[Anonymous]

Hello,

As you can probably see from your attempts and from the various responses that have been posted, sometimes MES is a fairly complex analysis to set up...or to put it in another words it has a considerable number of parameters to set up and considerations to be made. From your attached picture I would venture to guess that the time step might also be a factor to be revised... but it is just another guess.

Just a thought: if you only need to see the angle at which the resting beams will fall I would probably start with an approximation in the Dynamic simulation environment in Autodesk Inventor, taking into account that you don't need stress results or anything that you can only get from MES. It is probably easier to set up and you can get some pretty useful information from that analysis.

Good luck

[Anonymous]

Thanks for the recommendation John. I went into the advanced settings & changed the contact distance from automatic to user defined. The default contact distance is 0.01". I ran the analysis with that & when the structure tipped, the beams went with it like it should. After 1.47" of displacement, it got hung up. I didn't get any errors, but it kept running overnight without any more time steps. Looking at a straight end view of tha last time step that converged, you can see that the beams are no longer square with the structure by a very very small amount, which tells me they are starting to slide. Could this be the reason the next time step is taking so long? Because the beams are really starting to move/slide now & it's harder to computate?

The contact stiffness was also on automatic. I will change that to user defined & play around with it.

Thanks again.

[Anonymous]

I got the structure to tip. I got the beams to travel down the length of the structure. But, instead of sliding off the structure & falling downward, the beams come off the surface of the structure in an upward trajectory. No settings concerning contacts helps this. I am so confused.

So, to make things as simple as possible, I bring the model into MS with the structure already at an angle & I fix the feet. Set up surface contact between the structure & the beams, friction, gravity, etc. So all the beams have to do is slide down the structure. But, the beams fall down through the structure. Don't matter what I do as far as contact stiffness or contact distance. This is set up as simple as it gets isn't it? Should this really be this difficult? What am I missing?

Any help would be great.

Thanks.

[AstroJohnPE]

I do not have any thoughts on the first problem (the beam comes off the structure in an upward trajectory).

For the second problem (beam falls through stationary structure), yes: it either has to be difficult (meaning the user needs to setup the model) or it has to be a really, really long analysis runtime (meaning the software applies the most complicated settings to every option in every analysis.)

My thought for the second problem is related to the mesh size. "Surface to Surface" contact is a bit of a misnomer: it is really detecting when the nodes on surface A make contact with the element faces on surface B. This is the "Contact Type" of "Point to Surface" which can be controlled under the "Options" for the surface-to-surface contact. If the user or the software has chosen the  "Point to Surface" method, it is possible that the nodes on surface A do not touch the element faces on surface B when the mesh on surface A is too large; as a result, the bodies pass through each other. See the illustration on the page "Nonlinear Analyses > Loads and Constraints > Surface-to-Surface Contacts > Options" in the documentation. (This link goes to the page in the 2014 Help, but it should be close enough for whatever version of Simulation you are using.)

One of these may be the solution to the second problem:

1. Use a smaller mesh in the area of contact so that there is always a node on each surface making contact with the adjacent element faces. Then it doesn't matter (as much) which surface uses the "point" and which one uses the "surface".
2. If the software is using point to surface contact, you can switch the order of the parts so that which body uses the "points" and which uses the "surfaces" are reversed. This may help depending on the mesh size. (Which contact pairs are using "Points to Surface" and which are using "Surface to Surface" is shown in one of the log files, where it lists how many elements are in each part, just before the iterations begin.)
3. Force the software to use the "Contact Type" of "Surface to Surface". This really just detects contact between the nodes on surface A with the element faces on surface B and the nodes on surface B with the element faces on surface A.

If you want to post an archive of your model (especially the model with first problem), someone may be able to take a look at it. See "Create, Post, or Provide an Archive of your model".

[Anonymous]

Thanks for the reply John.

I will play around with the mesh size & see how that helps the situation. In the meanwhile, I will post the archive file here for anyone who wants to help figure out the puzzle. I guess not. It's too large.

Thanks.

[Anonymous]

Challenge!

Ok, so if anyone gets a few minutes (that's how long this SHOULD take) please set up this test model (attached). It don't matter what size the parts are. Two rails at 45 degrees & a larger beam sliding down the rails using gravity only. This is about as simple as it gets & ultimately what I'm trying to do in my other model, yet it don't work, for me anyways. It starts to slide & then gets kicked up & starts to tumble. I ask again, what am I doing wrong? Why is this so diffucult? Is this something MS can't be used for?

Thanks again.