difficult one I think. I use spring elements along threaded areas in contact to distribute the load. But I to am open (and looking for) to a better method.
In reality this will probably be a non-linear problem, which is most easily represented by a linear condition. Just set up a linear "bonded" type contact over the area in which the nut and the bolt interact. Anything else is just overkill and probably won't represent the "real" conditions (requiring correct material properties, geometry, appropriate friction etc) at the thread interface.
------------
See faq569-1083 for details on how to make best use of Eng-Tips.com
Hi,
I agree with Drej. It is extremely difficult to have complete knowledge over all the factors involved in the nut/bolt threads contact: material properties of course, + friction (is it oiled? greased? dry? "moderately-oiled"? and so on...!) + tolerances (tolerance on nominal diameter, tolerance on flanks inclination, tolerance on core diameter, tolerance on pitch, etc...!). Geometrical factors will make "random" knowing how many threads are really engaged, and which of them; friction factors will make it rather vague to know the relationship btw normal and tangential forces...
In my opinion this is a case where the uncertainties on the input data are greater than the precision which FE can ensure. Thus, I'd rather go with some normated (UNI/EN, VDI, JIS or others depending on what country you live in) analytical method and apply reasonable uncertainty factor.
I am the abaqus user, in version 6.6, they develop the efficient finite-sliding, surface to surface contact function. This is better for the contact analysis involving corners ( interior-to-exterior) corner contact between the threaded connections. This is better tahn node-to-surface contact with poor contact resolution.
Hi,
Johnsmith2, yes, for sure very sophisticated algorithms exist in order to model a thread-to-thread contact; there are many possibilities also in ANSYS. But in my opinion it is almost useless to build up such extremely sophisticated analyses when the inputs from the "real world" are subjected to very high uncertainties.
Hi,
It depends on whether you are studying the bolted connection itself, or you just have bolts in a wider structural problem.
When I am looking at structures that are bolted together, I normally bond the threaded area to threaded area, and inside of surface of bolt head to pressure surface. If there is a shoulder (unthreaded section) I normally leave that free. That gives you way of looking at how the bolts may stretch.
Also you can use analytical methods to relate bolt torque to tension in the bolt to give you an idea of the loads present due to the initial preload. Torque T = K x Fi x D, where empirical results give the constant K = 0.2 and friction Fi =0.15 (from Shigley Mechanical Engineering Design and widely available elsewhere). You might want to take account of these loads as well and apply them to your model.
But then if you were a bolt manufacturer then maybe you might want to do fea on thread interaction, in which case there has been lots of work done, the link posted previously is very interesting, and there has been a massive amount of work done in photoelasticity of threads which must be worth studying. The stuff I have read (aeons ago) showed how, in tension, only a minority of threads carried load. Good luck solving it!
cbrn,
The ABAQUS 6.6 feature that johnsmith2 refers to may be sophisticated, but is very easy to use. It takes input that you *already* have (pitch, thread angle, bolt diameter, etc) and makes the appropiate modeling assumptions for you. Furthermore it uses their surface to surface formulation which does not require conformal meshes.
Compared to other common techniques (e.g. tied nodes) it is far more appropriate from an engineering point of view... Especially if you are looking at the stress distribution around the threaded area. Of course, anything you do in FEA is an assumption (by definition), so it is up to you to decide when and where to use particular methods.
I've also run across this situation. I used surface contact elements in ANSYS to model the contact, but of course what you find is that local yielding typically occurs in the thread roots and under the head of the bolt.
That affects the load distribution as well as the lead error in the threads, and etc.
The question of what friction coefficient to use is also difficult.
Another question comes up when you get into this. Usually people look at the axial loads on a bolt, and sometimes the shear loads, but there can also be a bending moment applied to the head of a bolt! And I think there are real moments applied in many applications. Why is this ignored?
