Simulating Chains in NX

[MrK159]

I know this is a very vague question, but what 1D element/connection would be best suited for modeling tie-down chains? Obviously chains can not take compression but tension only. Any suggestions?

Thanks

 

[MrK159]

I also assume a non-linear analysis will have to be performed.

 

[BlasMolero]

Dear Afree,
The element CROD of NX NASTRAN may have material nonlinear extensional properties, with linear torsion. You may supply plastic or nonlinear elastic material properties. Since the stress-strain curve for compression need not be the same as for tension, this element can, for example, be used to model cables which cannot carry compression. Of course, the analysis should be nonlinear (SOL106).
Best regards,
Blas.

~~~~~~~~~~~~~~~~~~~~~~
Blas Molero Hidalgo
Ingeniero Industrial
Director

IBERISA
48004 BILBAO (SPAIN)
WEB:

http://www.iberisa.com

Blog de FEMAP & NX Nastran:

http://iberisa.wordpress.com/

 

[MrK159]

Great thank you. I assume these tension/compression properties are in the materials themselves. Since the stress is also dependent on the cross-sectional area, how would this apply to a chain for a CROD element?

Thanks.

 

[adfergusson]

The cross-sectional area will affect the response of the chain, but so will the fact that the chain links are 3D structures themselves. If you want to capture the non-linear response of a chain with a 1D element (like Blas suggests) then you'll need a load vs displacement curve to represent the 1D response of your 3D chain. One way that you could do this, depending on how much work/time you've got for this, would be to use a (sort of) local-global modelling approach; i.e create a model of parts of two chain links in contact, displace the ends of the links (use symmetry) and capture the displacement vs load response. Use the displacement vs load response from this 3D model to define the tensile response of your 1D elements (you'll still need to calibrate these FEA results to input properties for the sections for your 1D elements) . Once this modelling approach is working, you could then change the cross section of your local 3D model, rerun and use the new results. Not familiar with CROD but it looks like it does torsion too so you'd need to do another model to capture rotational/'torsional' response. However, I'd expect the effective torsional stiffness of the chain to be strongly dependent on the tensile load it is under so this if neglecting chain rotation/torsion is a reasonable approximation for what would happen in reality then I' probably just ensure the element I use only responds to axial load

PS I'd probably go with the chain links in contact at the start and use displacement control, rather than load control, to avoid contact stabilization matters.
PPS Obviously I'm assuming you have a solver capable of running this sort of model.

 

[MrK159]

I am sorry for bringing this back up but I am still having issues with how to go about doing this analysis. I will give an example: I am trying to analyze cargo being transported with accelerations in combined load cases. The cargo is held down in various configurations so that the in some instances some of the chains are in tension, and some are in compressions (which theoretically should be non-existent). Instead of going through each load case, running the model and figuring out with chains are in tension, the removing the chains that are in compression, then running it again with the tension-only chains; I would like to be able to run all the load cases as once with the correct chain properties. Would anyone have the best solution for this analysis? The best results I have gotten so far involves using CGAP elements with specified stiffnesses but this approach seems very crude.

Thanks.

 

[rb1957]

i'm not so sure it's as simple as you're making it. in the initial state all cables are tensioned against the crate, so there's a unitform pressure between the crate and the ground. a cable that goes into compression doesn't "just" disappear, it reacts load untill the prelod is exceeded (and then it becomes slack). if you remove the slack cable from the initial problem, then you've changed the initial situation.

i think you need to say the cable becomes slack when x% of the loadcase is applied, and 1-x% is rected by the remaining three cables. maybe easier to see by superposition ...
1) initially all cables have tension, Ti,
2) apply x% of the load, reacted by four cables, get cable loads (one should be -Ti),
3) apply 1-x% of the load reacted by the other three cables,
4) sum the cable loads (one will be zero).

you can calculate x by applying 100% load seeing the compression load in the cable, C, then x = Ti/(Ti-C) ... C being -ve.

instead of a myriad of cases, can you examine the corners of the envelop, 8 cases ? how much precision do you need ?

another day in paradise, or is paradise one day closer ?

 

[rb1957]

other than this would be a non-linear analysis, using either gap element or material property.

some codes have "cable" elements that'd work as you want them to (tension only).

another day in paradise, or is paradise one day closer ?

 

[MrK159]

I am not so concerned with the initial aspects of the cargo as I am with the final results with the induces load cases. In the final 'position' I assume the initial tensions in the chains has little to no influence on the final results.

 

[rb1957]

well that is a much simpler approach.

how about the envelop idea (instead of considering every combination); that would give you fewer combinations, no?
another idea would be a unit load approach?

and once you eliminate a cable, the remaining three are statically determinate ? well, maybe ... how do you account for the deck reaction ? a vertical load under each (effective) corner ?

another day in paradise, or is paradise one day closer ?