Help with Parachute Modelling in LS-DYNA

[oreillsi]

Hi all, this is my first post so hopefully I'm not breaking some house-rules, apologies if I am somehow,

I'm new to using DYNA and am currently working on my first project which is to model a parachute. I'm modeling the post-inflation state of the parachute so really all I want to do for the first step is apply a pressure to the parachute and see what sort of stresses arise because of this.

The approach I've been told to take is to use the *AIRBAG card in DYNA to model the parachute but the problem is there are only two tutorials online for airbag related work it seems, so as a result I've used the card before but don't really understand how to apply it to a new problem such as this one. For the tutorials a fully sealed airbag is modeled, i.e. one with two sides joined together so that as the fluid is allowed to flow into the airbag it expands because of this. For the parachute analysis I've imported a geometric model of one gore/section of the parachute from CATIA into the preprocessor (LS PrePost) and meshed it, the mesh is coarse but seems to be ok and I've applied a cyclic boundary condition to represent how the parachute section would fit into the entire structure. I've also defined the sections and materials needed for each part in the model. However this is as far as I've gotten, when I run the model through DYNA I get a few warnings (it seems DYNA can run despite giving warnings? If so I can live with them for now) but I've also got some errors relating to "out-of-range velocities" of numerous nodes in the model which results in LS-DYNA terminating. I don't understand what's going on with this problem, I tried applying a fixed boundary condition to the two nodes at the top of the model but im not sure if that's worked correctly, the dyna message gives a warning regarding it. I've attached the keyword file to this, if anyone more knowledgeable on working with LS-DYNA than I am would be so kind as to look at it I'd really appreciate it. Any tips on how to get this initial model running would be fantastic!

If there's any more information needed that I've left out from here please say so and I'll try my best to provide it!

Cheers everyone

 

[GregLocock]

No you aren't breaking any rules. I've never used dyna in anger, so all I can say is your first priority is to get a run to a state where you can start debugging it. Most FEA type programs issue warnings which you can ignore sometimes, but which may be giving you clues.

The usual advice would be to start with a model that works, say your airbag run, and modify it and rerun it until it resembles your desired model. Then you'll find out what breaks it. So, cut a large hole in the airbag. reduce the inflation pressure, change the container, etc etc.

Sounds like a fun project.



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[missil3]

Hey,

I had earlier done an analysis on parachute using ALE to see if I a full parachute deployment simulation could be done, I have attached the input deck to this post and here is an animation of the flow around the parachute : [URL unfurl="true"]http://tech4server.com/parachute.gif[/url]. There are a few complexities that occur but those can be resolved and a good result can be obtained with some work. I dint continue it after this point though because of other work.

Coming to your deck:
The NaN - out of range velocities could be due to a variety of reasons. It is because you are using the Airbag_Hybrid card.
I doubt you can use this for your purpose, your part needs to enclose a volume so that it can calculate the pressure based on mass flow rate.
There are ways to enclose the volume with null elements but an easier option is to use LOAD_SEGMENT card to apply a pressure to the shell elements. You need to get your normals correct for this though. There is some inconsistency in your normals in current model.
Or if you want to simulate the flow as well as pressure you can use either the particle approach (*AIRBAG_PARTICLE)
or ALE methods (Arbitrary Lagrangian Eulerian) where you have a fluid domain and a coupling between fluid and structure.

Anyways if you need help with it let me know.

 

[VoyageofDiscovery]

Back check it with this.

http://www.dhondt.de/

 

[oreillsi]

Thanks for the input everyone. I think I've eliminated a good few of the errors now, turned out a lot of it was to do with inconsistent units. I now have a running model which will hopefully be able to gradually debug. I'll post back on this with results or more problems, which ever I get first!!

Cheers

 

[oreillsi]

Hi all, another question regarding this project.

I've made a lot of progress with modeling the parachute, it now starts from a semi-inflated shape then collapses under gravity to allow for the fabric to fold and crease "pseudo-randomly" and then the pressure load is applied so it inflates and pops into shape.

The problem I'm having now is actually to do with the outdated version of LS-Dyna I'm using (it says it was released on August 13th 2007, I can't get a more up to date version as i only have the limited cheap license). Because of this it seems this version doesn't recognise the input card *BOUNDARY_SPC_BIRTH_DEATH, is this correct that this version is not compatible with this card?

Because the parachute gets folded under gravity it is fixed at the top of the model at the beginning of the simulation and allowed to "sag" from this point, what I am now trying to do is remove the boundary conditions that build these nodes in once the inflation pressure is applied so that the top of the parachute is free to move. However without *BOUNDARY_SPC_BIRTH_DEATH card the only way I can find to do this is by doing a Restart analysis and making the restart deck not have those nodes built in. But when I run the simulation the results show the nodes remain built in from start to finish. So basically my question is: Does anyone know if it is possible to remove boundary conditions from a node/nodes via a restart deck and if so how to do this? I've seen that it's possible to add boundary conditions using the *CHANGE_BOUNDARY_CONDITION card but the keyword manual says that this can't be used to take away bc's or free previously constrained nodes which is what i am trying to achieve.

More of your extremely helpful advice would be appreciated!

 

[missil3]

Which release do you have ?any lsdyna active license allows you to have access to the newest version. Have you tried downloading one of their latest solver files must also download the lstc_client file or you will get license compatibility error.

another alternative I can think of is to split your simulation into different parts so you can remove the spc in the part.where you want it to sag under gravity. Read about springback analysis to split your analysis.

 

[oreillsi]

Thanks for the reply missil3. The info I get from DYNA regarding the release is as follows:
"Version: LS-DYNA_971", "Version Date: 08/13/2007", "Revision: 7600.1224", "Feature: SMP Enabled". I am running DYNA on a limited license, one that is limited to 10,000 elements or less and I've been told that it can't be updated to the latest version without paying for a full license?

Anyway I managed to solve the problem, the way I got around it was I applied a load to the nodes instead of restraining them, so while gravity pulls the entire model down an equal and opposite force pushed the nodes back up to hold them in place without restraining them. Then once the parachute had finished collapsing I set the load curve for the nodes' force to reduce to zero. Now the nodes are free to move but the parachute can still collapse, without the *BOUNDARY_SPC_BIRTH_DEATH card I think this is the best way to implement this but I'll look into springback analysis too.

I do have a new question now if anyone might know how to help with this one, I've been trying to find a way of doing this in LS-PrePost for a while now but I'm not sure if there is a way, it's a purely "cosmetic" problem: I modelled only a single "slice" of the parachute and have my results for this model, is there a way to revolve this slice about an axis of symmetry along the parachute's centre line so that I get a visualisation of the entire parachute structure? I know there is the option to reflect the model about the XY, YZ and XZ planes but does anyone know if it's possible to revolve one around a line or otherwise define a user's own reflection?

 

[missil3]

Hey,

Sounds like you have an academic license. But it doesn't mean you have to use the old version, in fact LSTC only supports the latest versions (R6 / R6.1).

You can find the download info for the new versions here:

http://feainformation.com/m-pdf/limited.pdf

Also, another keyword you can use is Boundary_prescribed_motion , it already has the option to set Birth and Death times. You just set the velocity of nodes to 0 via a load curve to stop them from moving.

I dont think its possible to revolve your model in LS-Prepost, try asking in the lsprepost google groups:

https://groups.google.com/forum/#!forum/ls-prepost

 

[oreillsi]

Thanks again missil3. I ended up changing from my method of restraining the nodes and opted for yours of using BOUNDARY_PRESCRIBED_MOTION, it's much tidier way of doing it. Looks like you were correct about the revolving issue too, there doesn't seem to be a way of doing it in prepost which is a shame, hopefully i'll find a way to run the results through another post processor that can do it.

Cheers again for all the help!

 

[missil3]

Hyper view is a good post processor although I am not sure if it can revolve. I think it's the most feature rich post processor.

 

[Sparweb]

Only 4 rigging lines?
You can see the edges billowing in the sim.

Really cool, though

STF

 

[missil3]

You are right, it needs more rigging lines for better stress distribution. I was just testing the capability of the software at the time though so its not the most accurate simulation.

 

[frusso110]

OP:
Where did you go to school and what did you get a degree in? I want your job!

(I am working toward my M.S. in ME at night... So, hopefully soon )

 

[GregLocock]

In a non linear event like this I think relying on symmetry is rather dangerous.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?