Pipe Clamp Under Bolt Pre-Load study

[LDanielXVWBH]

Hi all I am looking for some help with a study I am currently conducting.

For context I have two clamps which clamp around a pipe with a rubber liner between the clamp faces and pipe under a specified bolt pre-load using connectors (image attached). I am then applying a moment on the clamp to try and establish what bolt pre-load is required to prevent the clamp from slipping around the pipe.

My current study settings are:

Analysis Type : Non-Linear Static

Connectors : Bolts

Surface Contacts : Shrink Fit/Sliding

Mesh Element Size : 15 mm

Load : Moment

Constraints : Pipe = Fixed Constraint

I am currently unsure on whether I am using the correct constraints and contacts and if anyone has any suggestions on how to improve my study.

When I am currently running my non-linear static analysis because of the shrink fit/sliding contacts currently applied, I encounter an E5076 error so any help with my study on how to correctly set it up would be appreciated.

I have attached images with all my study settings.

Thanks

[John_Holtz]

Hi @LDanielXVWBH . Welcome to the Inventor Nastran forum.

I think the first issue is that any type of "sliding" contact means the faces are free to slide. Friction is not included when the contact is free to slide. Therefore, you model is not statically stable.

Secondly, I recommend not using friction unless there is absolutely no alternative. The issues with friction are as follows:

1. It make the analysis much more difficult to converge.
2. If friction is overcome and the parts try to slide, the part will usually become statically unstable. When that happens, the analysis does not converge. When that happens, you do not know why it fails to converge: is it due to friction or any number of other reasons that a nonlinear analysis fails to converge.

Assuming you want the fastest answer with the least amount of hassle, the easiest approach is to assume the friction will hold the parts. Therefore, use bonded contact. From the results, check the contact normal force and the contact force in the shear direction. A hand calculation of (friction coefficient)*(normal force) will indicate if the joint will slip.

In your model, you may not even need to apply a moment load! The bolt preload creates the normal contact force. You can do a hand calculation to determine what shear force is created by the moment.

Also, I want to confirm that all the parts have an initial interference (without considering the bolt preload), and that is why all the contacts are a shrink fit.

Let us know what happens next in the analysis.

John

[LDanielXVWBH]

Hi John,

Thanks for the reply and suggestions.

In response to your interference query there is interference between the liner and pipe and between the liner and inside faces of the clamps.

When I tried your suggestion of using bonded contacts I find that this method is not a true reflection of my scenario as the this method bonds/welds the liner to the pipe whereas in the practical scenario the liner can slide around the pipe. When I use bonded contacts I find that no matter how much I increase the moment the results don't really change because of this bonded interaction between all of the parts. 

I have also done a test run where I excluded the pipe and liner from the analysis, placed an offset bonded contact between the clamp faces and when the moment was applied there was rotation that could not be stopped no matter how much bolt pre-load I specified. I assume this is because despite the pre-load there is nothing for the inside cylindrical faces of the clamps to clamp against.

Thanks

[John_Holtz]

Hi @LDanielXVWBH 

Let's think about how you would do the calculation if you did not have a computer.

1. Knowing the preload F, calculate the pressure p on the clamp. (p*D=2F)
2. From the pressure, calculate the normal force N on the clamp. (N=pi*D*L*p for both halves of the clamp of length L)
3. From the normal force, calculate how much friction force Ff occurs before it starts to slip. (Ff=mu*N)
4. From the friction force, calculate the moment that can be developed before it starts to slip. (M=Ff*D/2).

Note that there is no movement or sliding in the calculation. 🙂

The Nastran analysis can be performed in the same way.

1. Use a nonlinear static analysis. Since the parts needs to be statically stable, you need to use shrink fit/no sliding.
2. Apply the bolt preload. No other load required (other than a dummy load so that Inventor thinks the model is setup properly.)
3. Run the analysis.
4. Get the calculated normal contact force at all the nodes (on either the inside or outside of the liner). Sum the normal forces.
5. Calculate what friction force can be supported (Ff=mu*N)
6. Calculate what moment can be supported (M=Ff*D/2)
7. Note that because there is no load, there is no reason for anything to slip (ignoring the motion that may occur due to the shrink fit and preload). Therefore, the results are the same for bonded contact or sliding.

The alternative is to allow slipping which requires a transient analysis. When the part starts to accelerate, you know the friction force has been exceeded.

1. Use a nonlinear transient analysis. Output velocity and acceleration results.
2. Is shrink fit supported? Probably not. Would need to modify the contact and other parameters to include the effects of the interference. See How to perform a press-fit analysis in Inventor Nastran and Nastran In-CAD.
3. Need a fine mesh on the contact faces in order to get a more accurate friction calculation.
4. Probably need a small time step.
5. Define a time history curve for the load. After a dead time of X seconds to allow the bolt preload and interference to stabilize, start to ramp the load from 0 to 100%.
6. Finer mesh, small time steps, implies a longer analysis runtime.
7. See when the parts start to slip. Based on the time, determine what the load is.

I do not understand the model of the pipe clamps only. Constraints on the inside of the clamp should have been applied to simulated the pipe/liner which generate the radial force that balances the bolt preload. From the sum of the radial forces times the friction coefficient, you would know what moment would cause slipping. If you want to provide the model, see the last line of my signature.

P.S. I may have been wrong about friction and shrink fit. The contact type of "Sliding/No Separation" definitely does not use friction. I mistakenly assumed that "Shrink fit/sliding" used the same type of sliding, but it is really shrink fit/separation, and the contact type of "Separation" does support friction. So "Shrink fit/Sliding" is really "Shrink fit/Separation", so friction should be supported.

John

[LDanielXVWBH]

Hi John,

At present we have hand calculated values we are just aiming to use the FEA study as a method of validating our hand calculated results. 

However, when I attempt the non-linear analysis you have instructed I encounter the same issue 'E5076 : maximum number of bisections reached'. I have attached a model if you have a minute free to look at the setup and see if you can diagnose where the issue might be caused.

Thanks Lewis

[John_Holtz]

Hi Lewis. The assembly file (.iam) is not the complete file. See the last line of my signature for a link to the page describing how to create a pack and go file to get what we need.

My guess from seeing the input is the Poisson's ratio for rubber should be changed to 0.48 or so. Although 0.5 is technically accurate, simulation is not technically accurate and may not handle a value of 0.5. 🙂

John