modelling of thread

[rob768]

I need to make a fem model of a part where the properties and dimensions of the thread are of importance. I was thinking of modelling the contact of threads by gap elements. I am also considering using a slide line. However, i have little experience in this field. Anyone got any suggestions or useful links? Much aprreciated.

Using Femap 10 and NX-Nastran by the way.

 

[rb1957]

well, after you sort thru the practical difficulties of projected a thread profile along a 3D spiral spline, and meshing the begeepers out of it, then applying preload and plasticity, i've got to ask "why?" ...

we know how strong threads are, though i suspect you've got something exotic/unusual in mind, there are so many real world things that are really hard to model that i suspect your conclusion will either be horribly conservative or unconservative (unfortunately you won't know which 'till its too late).

how difficult is it to test the component ?

how unrealistic is it to model the component as a "welded" assembly and analyze the thread interface by hand ?

 

[rob768]

I am not going to model the pitch. I am using a section which i assume to be symmetrical. In fact, I will probably use axis-symmetric elements for this as loading and geometry is symmetrical (with the above assumprion).
The aplication is not to special but will need to be proven correct and up to a full spectrum of loads, including shock. There will be elaborate testing involved, to.

 

[MichelM]

I am also involved with parts where the properties and dimensions of the thread are of importance. For me the dimensions of the threaded joint are not standard like they are in a typical bolt/nut arrangement. I used ANSYS software with axis-symetric element and with contact elements for the threads. The same thing is probably feasible with Femap 10 and NX-Nastran.
After doing the modeling you will have to analyze and use the results. From stress plots, you will have to determine if the strength is sufficient to resist the load. Furthermore, if you are involved with repeating impacts or load cycles will have to consider the fatigue of the joint. This will be the difficult portion of the analysis.

 

[rb1957]

rob,
maybe using cyclindrical solid elements would give you what you're looking for ... load transfered per unit (thread?) thru the thickness ??

 

[rob768]

actually, fatigue is one of the issues so i feel most for axis-symetric elements. That allows for a fine mesh. Contact is the thing i am not too sure about. Either i use gaps, or line contact.

 

[kellnerp]

Unless you are doing an academic study this is an unnecessary effort. The manufacturers of fasteners will rate them for fatigue, ultimate and provide pre-load recommendations. I have worked on very high end aerospace stuff that had critical bolted joints and nobody modeled threads. We tested.

Here are some of the difficulties:

Bolts, especially those taking a big fatigue load are torqued to very high levels where local yielding just might occur.

Just getting a pre-load in the field that you assume in the analysis will be a huge achievement.

Many bolts have rolled threads. This means the local material properties at the thread root will be significantly different than the bulk material. Are you prepared to model that and have metallurgists tell you what kind of localized hardness and strength numbers you can use?

There can be big changes in stress gradient from small changes in geometry so you will likely have to run many anaylses to get a statistically meaningful set of data.



TOP
CSWP
BSSE

http://www.engtran.com

http://www.niswug.org

"Node news is good news."

 

[gwolf2]

I agree with kellnerp on this. I too have done a lot of 3D joint modelling in aerospace engines and we never model the threads. We use statistically based test data to get the minimum reliably obtainable pre-load for a given torque, and if the bolt loads exceed that, the design is rejected.

Don't forget the bolt holes. These often see significant fatigue stresses if thermal loads and flange packs are involved. The threads are often the least of your worries.

Lastly, don't use Nastran for this - it's contact, friction, pre-load, and nonlinear capabilities are just not up to this sort of continuum problem.

gwolf2

 

[rob768]

it is not about bolts, it is about threads. There is no pre-stress and the thread is cut. the item is used in positioning heavy machinery, and has to withstand large static and dynamic loads. I cant go in too much detail but all i am asking is tips on how to best model the contact in a thread.

 

[gwolf2]

OK, point taken, no bolts.

OK, if you really want to model threads and get a half reasonable answer, don't use Nastran because it's contact and friction capabilities are not really up to it.

I also have a vague memory (Nastran experts help me here) that the only axisymmetric elements in Nastran are 6 node triangles which will give absolutely appalling contact stresses and must also be used with extreme caution for stress concentrations at the tooth roots.

I think it's a bad idea in general to lift heavy machinery with threaded holes if large dynamic loads are involved. Even if dynamic loads were not present I would want to see a static margin on thread pull-out of at least 5-10. Is it a through hole or a blind hole?

Try a pull-out hand calc assuming half the hole shear area is effective at the mean thread radius in combination with the known dynamic loads. What reserve factor on shear does that give you?

 

[rob768]

i can use mostly quads when modeling. It is not really lifting but more positioning an di am afraid i am bound to Nastran, or CAEfem.
I will just give it a go with plain contact elements and see what rolls out the pc

 

[GBor]

I have done thread analysis using 2 different software packages...2D and 3D....not aerospace applications. Unfortunately, neither of these packages were Nastran-based codes, so I can't help you with specifics.

Contact elements really won't work because of the required connectivity and the way they "load up". You need some type of surface contact algorithm that takes into account penetration of one surface through another and the resulting force required to make sure that it doesn't penetrate. Ideally, you would like it to account for friction, but this isn't an easy calculation.

2D-axisymmetric works pretty well to get the load path, but will likely over-predict stress by up to, in general, about 30% (please note in general...just a "ballpark").

The jump to 3-D is difficult and requires some serious computing power. It is easier if you are careful with your modeling approach. Depending on your software, you want to minimize how many surfaces you have to designate for contact and rotate your thread through a minimum distance if you are checking the fit-up stresses. If you are just looking at in-service and ignoring pre-stress (probably not a good assumption, but sometimes necessary) then you just need a lot of computer power.

 

[MichelM]

From the information that I see, the geometry is more like a typical bolt/nut arrangement with cut threads and no preload.

To obtain a life estimate of typical bolt/nut arrangement and cut threads you can refer to the following ESDU papers where the life is evaluated as a function of the alternate nominal stress and the UTS of the material. There will be no finite elements required.
ESDU 84037 - Fatigue strength of external and internal steel screw threads under axial loading. (Standard forms not greater than 1.0 inch diameter.)

ESDU 80028 - Fatigue strength of large steel bolts and threaded connections under axial loading. (Thread diameter equal to or greater than 1.5 in)

For the preload, I recommend you to improve the design by having a good preload.

I wish you good success.

 

[kellnerp]

Consider this:

It is not really lifting but more positioning

Positioning is the key word here. Think injection moulding press or the lead screw on a machining center or a lead screw used to extend landing gear. Preload is not possible. This is not a fastener analysis.

If the nut is moving then fatigue would likely be an issue in the nut, not the screw if the loading can happen anywhere in the travel. Lubrication would likely be assumed and a more consistent friction coefficient than in a fastening application.



TOP
CSWP
BSSE

http://www.engtran.com

http://www.niswug.org

"Node news is good news."