Bolted Connections in Cosmos Works 3

[Joest]

Does anyone have any suggestions on how to deal with components that are bolted together in Cosmos Works? Assuming I want to analyze an assembly and included the fasteners, what sort of contact definitions are most appropriate? The problem I have is that one of the part in my assembly acts as a stiffener for another part, but I don't want to assume the two are bonded along the entire length of the stiffener. I am trying to represent the fastener connections as realistically as possible since it is far from a welded connection. Thanks.

-Joest

 

[macPT]

Maybe you can sketch circles concentric to the bolt holes.

With the split curve feature, create surfaces representing the contact area that results from bolted connection.

For these surfaces choose the bonded contact/gap option.
For the other surfaces choose node to node.

Regards

 

[JNR]

We do this kind of analysis. I'll ask for some pointers for you. I know that our stress engineers allow for interface friction and tightening torque in their calcs.

John Richards Sr. Mech. Engr.
Rockwell Collins Flight Dynamics

A hobbit's lifestyle sounds rather pleasant...... it's the hairy feet that turn me off.

 

[djw2k3]

Hi Joset,

Its been a while now, but this could be useful for the case were the two parts are loaded so bolts are in axial tension only.

Make concentric circles around the bolt holes where the washers would be like macPT suggested. Add a rigid link to these areas for each bolt - this models bolt as rigid. It will also model the 2nd part acting as a stiffener because load is transferred through rigidly connected surfaces.

As for a friction type joint and other cases, I look forward to what JNR comes back with.

Dave

PS - There is an FEA forum that might help too.

 

[Joest]

Great input! I am also looking forward to the info JNR finds. It would be very useful to include the effects of bolt torque/preload in the analysis.

 

[JNR]

I have not forgotten you - our expert has been out (new baby).

John Richards Sr. Mech. Engr.
Rockwell Collins Flight Dynamics

"Heck, there are so many rules around here, it's hard to accomplish anything!" - Me.

 

[EdDanzer]

Bolted features are difficult to accurately model. We had a customer bring a part from a Cat feller buncher that was damaged. The fasteners went through 1” plate and went into a slewing ring bearing. The part was deformed by the bolts. My experience suggested thicker plate, but Cat said it was thick enough by there FEA. Since the force was high enough to break the bolts we modeled the plate with a force equal to the yield load of the bolt pushing on an area the size of the washer.
This method showed the plate needed to be thicker by at least .150”.
An inaccuracy is created by not using gap or some other element type to simulate the friction between the plate and washer as well as the washer and bolt.
You could search this on the Finite Element Analysis engineering forum for more ideas.

 

[JNR]

Our analysis was initally developed due to a customer requirement and in consultation with them. As I recall it accurately models the bolts, hole clearances and friction forces holding the parts together due to the bolt torque.

Our expert is back now so I will ask him on Monday as we have both been real busy today (we're off Friday).

John Richards Sr. Mech. Engr.
Rockwell Collins Flight Dynamics

Forget rich and famous - I want to be rich and unknown.

 

[JNR]

OK, here it is... and you are going to love this! You can't do it directly as you might think. You have to model the bolt and figure the preload due to the tightening torque. You then preload the bolt by giving it the appropriate temperature. Sneaky!

John Richards Sr. Mech. Engr.
Rockwell Collins Flight Dynamics

Forget rich and famous - I want to be rich and unknown.

 

[macPT]

Sorry JNR, but I'm not following you.

Can you explain with more detail your procedure? You simulate the preload using a temperature variation?

Regards

 

[djw2k3]

Nice solution, I take it the bolt is cooled for thermal contraction?

Joset, do you have the thermal module in CW?

 

[bobsheets]

I don't know if Cosmos can do it, but most FEA codes have a reference temperature that can be assigned to elements. This feature can be used to create "shrink" fits or applied loads when the model temperature is different than the "stress free" reference temp. This is done by artificially "heating" or "cooling" an element that is constrained in contact with adjacent elements. This creates a strain/stress that if you are careful can simulate a torqued bolt. I use "beam" elements with artificially high stiffness to simplify the model of the bolt. As in all FEA your modeling technique should be "proofed" by correlation to real world tests before any use of results. Great care should be taken to define the type of forces the bolted connection will see. My limited experience has shown that "traction" or "friction" created by the bolted connection is a more efficient solution. But that is for my narrow field of design. Sorry I can't give you more direct COSMOS command structure, but I use Algor.

 

[JNR]

bobsheets is correct - you just set the reference temperature. You do not use the thermal package. Again, he is correct in the need to use the friction force created at the interface. If you do it right it is quite accurate - we use it for analysis on aircraft structural mountings. Boeing uses a different FEA package and our results agree very closely from independently modelled analyses.

John Richards Sr. Mech. Engr.
Rockwell Collins Flight Dynamics

There are only 10 types of people in the world - those who understand binary and those who don't.

 

[macPT]

Thanks for the tip JNR.

It seems to be very useful for certain load simulations.

Regards

 

[TDrax]

I have been trying to model friction in a bolted connection in designstar, they say there is no difference between it and cosmosworks. I've been working with a very simple model consisting of two long bars with a single bolt connecting them. The bolt is pulled one way and the nut is pushed in the other direction to clamp the part together. (I never would have thought of using temperature to shrink the bolt, pretty clever) The assembly is simply supported at the ends and a small force pulls the bolt downwards. In the real world this is a stable system, the downward load is many times smaller than the force required to overcome friction (cof = 0.4). The model does not work, designstar continually told me to turn down the coefficient of friction. The reply from tech help is as follows.

"This is an unstable mechanism where you have rigid body displacements for the applied loads. This free rigid body rotations and displacements are not allowed if FE because of unique solution. You need to restrain the model somehow, so it doesn't exhibit mechanism."

I'm not too proud to admit that I don't follow everything in their reply.

In the past when I fastened parts together with more than one fastener and used a coefficient of friction, I assumed that some of the load was being resisted by the clamping force. Now I am not sure.

Any help is greatly appreciated,
Thank you
Tom Draxler

 

[JNR]

One thing to remember while you are talking about friction coef. Unless you are absolutely sure you have one-shot assembly with pristine cleaned interfaces, you need to use a value for lubricated surfaces due to dirt, finger grease, etc. It makes quite a difference to the required bolt torque values.

John Richards Sr. Mech. Engr.
Rockwell Collins Flight Dynamics

There are only 10 types of people in the world - those who understand binary and those who don't.

 

[TDrax]

Back to the problem of 2 bars connected with a single bolt in DesignStar. Take two bars, any size will do but lets say 1/4 x 1 x 12in long. Put a 3/8in hole 1/2inch in from the end, bolt the bars together using an appropriate initial tension for a 3/8in fastener. Simply support the now 23in long beam orientated so that friction is keeping the beam stable (if the bolt was loose it would act as a hinge and fall to the ground). Now model this in DesignStar or Cosmosworks. Can you get the model to be stable? If you can please tell me how you did it.

It is true that eventually a 'good' coefficient of friction will have to be determined. P166-167 in Vince Adams and Abraham Askenaze book provides a method. The above physical model is possibly another although without a spring the model will fail due to rigid body motion as an applied force overcomes friction and the two bars start to slide. And yes these values should be adjusted (usually downwards) for surface conditions.