Bolt post-processing

[jball1]

I am doing a transient analysis of a rectangular plate bolted around it's four edges being loaded by a transient pressure load. The plate is about 50" X 75" X 1.0" THK. It is attached to a fixture by 75 1.0" evenly spaced fasteners. I am modeling the fasteners with beam elements. I have post processed the beam forces and moments as shown in the attachment.

The attachment also includes a time trace of the axial stress, bending stress, shear stress and von Mises stress in the worst case bolt. The stresses seem to match the general behavior of the panel - initially several ripples form, which then coalesce into one concave "dish". This occurs around 0.035 sec, where you see the peak axial stress (green), due to bolt prying (contact is modeled at the interface between the plate and fixture bolting flange). The plate then rebounds and dishes outward. Peak displacement in this mode occurs at about 0.13 sec, where you see the peak bending stress (red).

My question is this: would you post-process the beam moments in the same way I did? Or am I overstating the bending stress?

I have found that at my company, the methods used to model fasteners, and the methods used to post-process fastener results vary widely. I am still trying to get a handle on when different methods should be used, and I also am very interested in getting input from analysts outside my company. Hence the forum post! Thank you for your time.

 

[rb1957]

how did you model preload ? axial stiffness of the bolt ? Huth ??

another day in paradise, or is paradise one day closer ?

 

[jball1]

Thank you for your response. Preload is not modeled. This is an analysis of a test that is going to be performed, and the actual preload has not yet been determined but will probably be minimal.

If the preload is not minimal, I will probably model it by increasing the stiffness of the bolt to account for it. The particular software package I am using (in house software) does not have the capability of accounting for preload in the ways that other packages are able to. If I were using Abaqus (the other package we often use here), I would either use the temperature method or Abaqus's built in preload option (where I think they just iteratively shrink the bolt until the preload you specify is achieved?).

I am not familiar with the Huth method, but I have learned from a quick Google search that I should probably get a copy of ASTM STP 927 and read it.

So, assuming minimal preload in the real world and no preload modeled, what are your thoughts on the way I post-processed the beam moments?

 

[MGZmechanical]

Just a couple of comments, I think:
- I = pi·r^4/4
- Svonmisses = sqrt(St^2+3Sv^2)
And although in a beam you should't add bending and shear for this cases I think it should be ok.

Also I'd make some hand calculations to check some figures.

 

[jball1]

ManoloGalarraga, thank you! The errors in the von Mises calculation were just dumb transcription errors - I didn't actually have those errors in my Matlab script. But the error in the moment of inertia calculation was in my Matlab script. Correcting that error sends the bending stress through the roof. So I guess I will have to figure out how I am going to deal with that.

Would you mind elaborating on why you would not include bending and shear in this case?

 

[jball1]

Does anyone else have any thoughts on this? I talked to one coworker who had a hazy rule of thumb that he didn’t consider bending unless the ratio of the thickness of the clamped members to the diameter of the bolt was greater than about 5. He couldn’t give me a source for that rule though. Does anyone else follow a similar rule? If so, do you have a source for it?

 

[jball1]

So... I can't figure out if I stumped eng-tips.com... or if I just asked such a stupid question that nobody is going to bother to respond. I'm usually inclined to believe my own stupidity is the issue - it rarely lets me down

But seriously... Does anyone care to chime in?

 

[MGZmechanical]

Sorry some days out. In long beams bending is just in outer fibers and shear in the center so when you have bending you don't have shear and viceversa. In short beams the stress distribution is not that clear but still is zero in outer fibers (where max bending).

For bolts I use to make hand calculations and as far as you don't have prying effects dur to lack of rigidity in the union is enough and much simpler to simulate than the model with the bolts.

Best regards,

 

[rb1957]

I guess I wouldn't combine shear and axial stresses together, since they don't occur at the same point in the section.

another day in paradise, or is paradise one day closer ?

 

[jball1]

Here's a schematic of the connection, if that's helpful.

That makes sense to me that you don't combine shear and bending, because they don't occur at the same location on the cross-section. I guess we do it because it's conservative. But, from talking to a few coworkers, it sounds like we often just ignore bending for short beams (although I'm still trying to find a source for what is considered a "short" beam in the case of fasteners - input on that would be welcome), or we use moments to calculate prying tensile stresses.

Thanks all for the input. I figured people would have more thoughts on this, since it's probably not an uncommon situation, and it doesn't seem to be very cut and dried (i.e. for this connection, you always use method X to assess bolt stresses). Or maybe it is cut and dried, and I'm just missing it... If anyone has any more thoughts, they would definitely be welcome.