Bolted connection issues 1

[Magos]

Hello all!
I have a couple of issues regarding the FEA analysis of a bolted connection and I was hoping you could help out. This is my first time posting so please let me know if this is the right place or if my post lacks information.

I've been recently tasked with validating the design of an assembly that includes bolted connections. As it isn't something I've done before, I decided to start simple by following the this guide (Link). Here's a picture of the assembly. I've also attached the .modfem files.

The most relevant parameters are as follows:
[ul]
[li]Plates are 200x50x5 mm[/li]
[li]Connection is done via M12 bolt and nut [/li]
[li]Material is 1020 carbon steel for all elements, imported directly from the software database[/li]
[li]The assembly is loaded with 5000 N in the X+ direction at one end and fixed in place at the other[/li]
[li]Meshing is done with a hexahedral elements with size set at 1.5 mm [/li]
[li]Bolt is meshed as it's done in the guide: by removing a section in the middle, creating two RBE2 elements and linking them with a bar element[/li]
[li]Bolt preload is 10 kN (Arbitrary)[/li]
[li]Contact connection (Frictionless) between the two plates. Default settings except for the contact iterations being raised[/li]
[li]Glue connection between the bolt head and the area where it touches the plate. Same for the nut. Default settings [/li]
[/ul]

My issues start with that last glue condition. While not explicitly stated, I assume that the guide defined all surface to surface connections as contact conditions that prevent solid penetration while allowing sliding. The problem is that when I did that the model ended up not being completely constrained, with the normal modes analysis showing that the parts were free to move everywhere. I was only able to solve this by applying that glue connection between the bolt head/nut and the plate. This obviously isn't a good representation of reality.

Moving on, once I perform the static analysis (SOL 101) the results just don't look right. First, only one of the plates is bending. While I understand the results won't be exactly symmetrical given the boundary conditions, shouldn't there be at least some bending in the second plate?
Second, I checked the axial force in the bar element using List>Output>Query and it outputs 9393.154 N. Shouldn't this be at least 10 000 N given the specified pre load?

Thank you for your time. I hope that with your help I'll be able to model a bolted connection that actually reflects reality.

 

[SWComposites]

What exactly are you trying to "validate"? If it is the strength of the joint, then the task is near hopeless, as there are a large number of intricate variables that influence strength of a bolted joint.

 

[Magos]

SWComposites - Yes, the idea would be to verify stress values in and around the bolted connection. While it's true that the behavior of a bolted joint is very complex, the same can be said about about any physical system. Engineering is about finding the correct idealizations that allow predictions to a reasonable degree of accuracy and make design decisions around them.

 

[SWComposites]

Local stress values around the hole in a bolted joint are highly complex, are highly influenced by factors such as clamp-up, tolerances, contact surfaces, local yielding, etc., and then using those local stresses predict joint strength is very difficult at best. Far better to use FEMs to only get fastener load distribution (and don't need 3D models for that) and then calculate joint strength using simple material strength values (tension, bearing, etc.).

 

[SWComposites]

By the way, I have been down this rat hole several times in research projects, and have reviewed the results of several other similar research projects, and the result has always been the same - its a complete waste of time.

 

[rb1957]

you should be able to understand why the strap is bending ... there is an offset (small though it may be) between the load and the reaction.
You could minimise this by loading the face of the strap (load just the edge nodes, yes ?) and react on the upper surface edge of the other strap.

the loaded strap, being is tension only ... yeah, it's just feeling the load until, all of a sudden, the load goes through the interface into the other strap ... so there's localised bending. The other strap reacts to an offset load.

"Hoffen wir mal, dass alles gut geht !"
General Paulus, Nov 1942, outside Stalingrad after the launch of Operation Uranus.

 

[Magos]

SWComposites - I understand what you're saying. In your experience, what would be the most efficient workflow when analyzing a bolted assembly? (i.e. how to get the most amount of useful data from the simplest model possible).

rb1957 - You're right. I was able to achieve what I had in mind by imposing a movement restriction on the face being pulled, being only able to move in the x axis. I neglected to add it in my first run

That being said, I still have the issue of the axial force in the bar element being lower that than the bolt preload. What gives?

 

[SWComposites]

Use a global FEM of the assembly to determine fastener loads; typically with shell elements; do not model fastener holes. Use spring/fastener type elements to represent the fasteners at each fastener location with appropriate fastener stiffness values (do not use rigid connections). Then extract the fastener forces at each fastener and analyze each fastener location for all failure modes using hand analysis (or in a spreadsheet). The detail analysis checks include fastener shear, fastener pull-thru, bearing in each joint member, net section stress in each joint member (and if composite materials are used then bearing-bypass interaction).

 

[rb1957]

I reiterate what SWC posts. The new generation of modelers/analysts seem to have an overriding concern/focus/interest in modeling discrete fasteners. This is not wrong but is very difficult to get "right" or "close to right", and would require a lot of testing and validation of the models, and would require a lot of budget and schedule.

The olde ways (as SWC posts) are proven and efficient.

Back to the model ... yes, now behaving more naturally. I'm not sure why the fastener tension would change for any reason. The bolt preload is creating compression around the fastener, this is transferring the bending between the straps ... it depends on how you've modelling the preload.

"Hoffen wir mal, dass alles gut geht !"
General Paulus, Nov 1942, outside Stalingrad after the launch of Operation Uranus.