Pre-tension in bolts 3

[johnsmith2]

Anyone has experience in the modeling of the bolts connecting two parts? What is the difference if the pre-tension on the bolts is included or not? I mean, the stress and displacement for the two parts will be affected so much with the presence of the pre-tension? Thanks a lot.

 

[RPstress]

Ignoring pretension is usually conservative, and it is often not modelled.

However, one way to model it is to use a finite length little bar element for the fastener, and to apply a thermal load which will generate the required force in the bolt. Careful modelling of the substrates being clamped up is needed to get reasonable behavior (if solid elements are not used for the substrates, then a spring element represeting their stiffness in compression can be added, another source of variability). Also, what preload to use is problematic: usually a fastener is simply torqued up, and not only is there a range of possible torque applied, the various ways of calculating the preload from the torque have a wide band of accuracy (5T/D? 7? 10?). NB: a linear analysis will not catch the behavior where the preload is overcome and the joint starts to open up.

Another way is to go material non-linear and to give the bolt element an initial modulus which will give the required high joint stiffness until the point at which the joint opens up and then the usual material stiffness after that. If your FE code has non-linear springs then this can be an easy way to do it. However, not only do you have to do the preload calculation, you have to do calculations on the stiffness of the substrates in compression under the bolt head and nut.

Whatever you do, you have to decide whether you want the minimum preload and stiffness, the maximum or some sort of average. Since preload can fade away over time, especially with cyclic loads, the basic case of no preload is often the one chosen.

 

[johnsmith2]

thanks for reply, RPstress,

I don't understand why the FEA is conservative if no pre-tension is modelled. Because I think pre-tension might increase the stress level in the bolt hole of the parts. Can I say that, if the objective of FEA is not to test the strength of the bolts, critical region is somewhere in the part, no preload is needed. But, if the study is to test the strength of bolts which can strongly hold the parts/components or not, preload is needed.

You are right to point out what preload need to be modelled in FEA. Usually, we know only the torque value of the fastners, 1/4", 1/2". But, we have no idea about the real case scenario of the axial loadings for the bolts.

 

[rb1957]

it depends on what you're analyzing for ...

statically it is a reasonable assumption that the joint gaps so that the load in the bolt is the externally applied load (so preload doesn't enter into the analysis). if, in fact, the joint doesn't gap, the load in the bolt is less ('cause some of the load is carried by the compression between the joint faces), so everything is ok.

but fatigue-wise preload has a hugh impact on the fatigue life of the bolt. if you model without preload, then the model is telling you the externally applied load, and you can do a hand calc to determine the loads in the bolt (basically preload + a fraction of the applied load).

if you include preload in the model, i think you're into a contact type of analysis that pretty much no-one did untill recently ('cause it was too difficult).

 

[killerbee]

I think your problem is that you are thinking of modelling it at all. The question indicates that you really need to look into a design text and discover why joints are installed with preload.

A model is the last thing that you want to create. I would hope that before even thinking of making a model you ask yourself what you expect to get out of it and could a hand calc suffice. Do you have a soild understanding of why the joint is/should be preloaded ?

Its harsh but models are the last solution to a problem and usually not the best.

 

[051174]

Pre-tension not simulated in fem is a good question.
The real loads and stress will be higher than those seen as results.in fem you can simulate it only in one way.in fem you can use a preload at both nodes at each side.tension load.but in general is useful to proceed analitically.Use ESDU.pay attention that you need alsoinformation of installation as wet or dry and torque moment applied.

 

[johnsmith2]

High preload tension helps keep bolts tight, increase joint strength, creates friction between parts to resist shear and improves the fatigue resistance of bolts connections. In reality, preload must be sufficint to hold the joint members in contact.

 

[rb1957]

"In reality, preload must be sufficint to hold the joint members in contact."

i'd challenge that one. i'd say that joint gapping is a little like skin buckling, in that everyone can decide for themselves what's acceptable (some places say no skin skin at all (very conservative), most i think say no skin buckling at limit load, most fuselages would probably buckle at limit without pressure).

you can design a joint to gap at ultimate load (most would probably accpet that) with all the loads being carried by the bolt. gapping under fatigue loads should probably be avoided.

 

[crackman]

I wholeheartedly agree with several of the last comments posted. I can give you some real world examples as I have just finished doing several failure analyses and correlations to tension fittings and the corresponding bolts which failed in fatigue but which even some of the most experienced stress engineers disagreed with me on.

Now, keeping in mind we are talking about true highly loaded tension joints, if preload is not maintained on the bolt, both the fitting and bolt can and will fail prematurely and in some cases at very low fatigue lives. There are some good tension bolt Sn curves with and without preload in the ESDU sheets. The fittings are tougher to analyze.

Just recently, we implemented replacements for a whole series of tension bolts (MS20000 series) because no proper record or mandated torque check had ever been required. Many older engineers balked at this and said that they had never seen such failures. Even the operators had the same opinion. Low and behold, upon removal of the bolts (difficult process as they were frozen-in since there was no requirement to ever replace them) fittings were found cracked. In addition, the fatigue cracks were located exactly were we predicted, on the mating surface were the end-pad bending would be highest without proper torque. A location that would never have been caught by simple visual inspection.

Needless to say, ....bolts are cheap but their failure can cost plenty. If there is no torque record, replace them, inspect the fittings for damage and mandate annual torque checks. Sorry, tension fittings and bolts are one of my pet peeves.

As for modelling the preload and various failures modes, the best work I ever saw was by an old friend of mine who worked for MACAIR. They tried to come up with a replacement for the Lockheed bathtub fitting method and it seems to work pretty well. Mind you though, it was done without lots of fancy modelling and based more on loads and constraint theory than on modelling contact surfaces, fine meshes, etc. Very practical and efficient method. Without going into details, the basic difference is that without preload, the endpad sees significant bending, with preload the load is transferred to the sidewalls in pure shear (same argument people have with shear clips). For the case without preload, they came up with endpad bending stresses based on various ratios of endpad to side wall thicknesses, lengths, etc.

Good luck

 

[gurmeet2003]

Johnsmith2,

I am a little late with my comments. If preload is not modeled, I fail to see the point of modeling the bolt at all. One way of modeling the bolt is to use a 3d solid model of the bolts with contact under the head and a tie joint on the threads. Both ABAQUS and ANSYS have the ability to apply bolt load/tension.

When the joints are subjected to cyclic external loads, seperation of the joint would be considered a failure of the joint. Because as long as the contact is maintained in the joint, only a small part of external cyclic load is passed on to the bolts. Most of the external load is taken up by the joint. The ratio of loads is inversely prorportional to the ratio of stiffnesses. Please refer to Bickford's book on bolted joint design. However if seperation takes place in the joint, all the extenal load is taken up by the bolt. A bolt is a member which is inherently weak in fatigue due to large no. of notches present (threads). Due to this reason, joint seperation of bolted joints is generally avoided (when subject to cyclic loads).

Gurmeet

 

[johnsmith2]

Gurmeet,

thanks for your information.

So, pre-tension is important to maintain the contacts between the parts. Without pre-tension, the bolts will take most of the external load, instead of joint?

 

[Qshake]

johnsmith -

that is correct. In my line of work, heavy infrastructure such a truss brides wherein you see a lot of cyclic loading due to traffic we use slip-crtical connections wherein the joint surfaces (faying surfaces) are meticulously prepared to meet roughness/friction requirements and the bolts pre-tensioned to provide the necessary normal force (clamping force).

Of course, because we're also concerned with ultimate failure of that joint (slip) we also make sure the bolts can withstand the load in shear too. Obviously this is due largely to the safety of public.

Regards,
Qshake

Eng-Tips Forums:Real Solutions for Real Problems Really Quick.

 

[EdDanzer]

It would be helpful if you described the type of bolted joint better. If you are using the bolts to hold parts together that are subjected to tension, the modeling method may need to be different.
We did an analysis on a failed part that had FEA done prior to manufacture. The part goes on the end of an excavator boom and has a large bearing bolted to it. The failures were breaking bolts, and it bent the plate the bolts went through.
To model this we fixed the boom pin and applied a load to the washer area equal to the breaking strength of the bolt. This showed that the plate was too thin.

 

[torpen]

Hello,

This technical paper presents some methods for modeling bolts :

http://ansys.net/ansys/papers/nonlinear/methods_for_modeling_bolts.pdf

Regards,

Torpen.