Simulation Stresses Too High

Discussion created by Fahad Kafrawi on Jan 11, 2021
Latest reply on Jan 13, 2021 by Fahad Kafrawi

Like • Show 0 Likes0
Comment • 8

Hi,

I am not an expert at simulations but understand somewhat of how to do things.

I'm having issues with my simulations as I have never done an assembly like this before and I keep getting errors and would appreciate help from people to give proper results.

If I use bonded contact I get somewhat decent results, but I want no penetration.

Scenario: I am rotating (torque) a shaft in the middle hole clockwise, which will cause the jaws on the side brackets to create most of the force on the mount/jaw contact. I want to see if the mount will bend due to this force.

I have tried running lots of different simulations with "soft spring models" and different contact scenarios (currently running surface to surface) and like the results from bonded contact but I don't think "bonded contact" is appropriate for this application.

I have run two different cases (one with a bench top and one without) with both giving different results and being too high. I'd expect stresses to be around 500MPa or less, not 5000+ MPa. The benched case has given good results of roughly 250MPa except for the singularity but I wanted to do a simulation without the bench.

*When I iso-clip, I can see there is no stress on the force which there should be, indicating this simulation is wrong?

Also, is there a way to remove the singularity max value from my results and find the next real max (otherwise I'll continue to ISO clip).

I would appreciate as much help as possible as I have been stuck for a while.

Thanks

Outcomes

Matt Juric Jan 12, 2021 9:27 AM

Hard to tell without better pictures and or models, but here's a guess.

It looks to me like you have an "L" gib in there that you want to see if it will bend when the torque is applied to the square area.

If this is the case you are correct that "Bonded" is the incorrect choice as this will act as if the "L" gib is essentially the same part as the rail it is sliding on, in other words it is one big piece.

What you want, going from memory and I'm not an expert either, is to set a special contact set for the "L" gibs surfaces to the rail it is sliding on. You will set that to "No penetration" and there are other options like "Node to Node" etc etc. I'd look up those options and see which one makes the most sense to use.

Once you set up those contact sets you should see stress at the contact points between the "L" gib and rail.

As to why you're seeing 5000mps instead of 500 mpa. Could be all sorts of reason more than likely all of them specious and caused by sharp corners, to large a mesh etc. You'd have to hunt down the reasons for the high numbers and fix them as needed.

If that doesn't help, post your models and let some of the more expert people take a look at them :-)

Actions

Fahad Kafrawi @ Ryan Navarro on Jan 13, 2021 4:09 AM

Thank you, I normally look at that and do total, just must have missed it in this setup but should really help cut my results in a 1/4

Actions

Frederick Law Jan 12, 2021 9:51 AM

All you need to worry is when torque try to pull the gib and L-holder up.

So mate the L-holder to the gib and apply force up.

You can keep them bonded.

Only need 2 parts in assembly.

Keep it simple.

Actions

Matt Juric @ Frederick Law on Jan 12, 2021 10:09 AM

Frederick Law wrote:

All you need to worry is when torque try to pull the gib and L-holder up.
So mate the L-holder to the gib and apply force up.
You can keep them bonded.
Only need 2 parts in assembly.
Keep it simple.

I've done bonded and unbonded for gibs like this and it gives two very different results. Bonded works if you're only worried about the entire system and are pretty sure that the gib and the rail are fairly equal.

Non penetration gives you a better idea of which component, rail or gib will fail. non bonded also gives you an idea of stress points between the gib components themselves which will tell you where you'll most like see wear.

Actions

Frederick Law @ Matt Juric on Jan 12, 2021 10:14 AM

From the assembly, The leg should be weak point.

Unless it plastic gib and steel leg.

Would the leg fly away with no-pen, if analysis 2 parts only? Pulling the leg up.

Actions

Matt Juric @ Frederick Law on Jan 12, 2021 10:44 AM

Frederick Law wrote:

From the assembly, The leg should be weak point.
Unless it plastic gib and steel leg.

Would the leg fly away with no-pen, if analysis 2 parts only? Pulling the leg up.

I agree the gib would be the week point and if it's not the rest is easy to "Fix" by comparison.

We just shipped a machine that had two vises on them both were "L"gib and rail. This was a rebuild and the existing system had horrific wear in the gibs. So when I redesigned it I wanted to find wear points and redesign the gibs to take the loads it would see etc.

The assembly I used for the simulation ended up being 5 parts total. (2) L-gibs, (1) rail, (1) base and (1) jaw that the gibs mounted to. Rail mounted to the base. The "L"-gib was a slip fit with the rail and the base.

For all the surfaces between the "L"- gib and rail I used no penetration and remodeled it so I had .0005" clearance. I then applied a force at the point that the vise jaw was closing from and had the other side pushing against a virtual wall representing the part it was clamping against. The Rail and base had fixed mountings on them so didn't move.

By doing the above the gib was entirely constrained in the simulation and did not fly away. You could do something similar with two parts by making some portion of either the rail or gib fixed and appliy the force to the other along with a virtual wall(s) to keep it from flying away. You would need at least two walls if you only did one gib.

Actions

Fahad Kafrawi @ Matt Juric on Jan 13, 2021 4:12 AM

Thank you both so much for your educated responses. My computer is having issues atm so I can't rectify and test my simulation until it is fixed but you have both been a huge help and I can't wait to try these solutions.

Actions

8 Replies

Matt Juric Jan 12, 2021 9:27 AM
Hard to tell without better pictures and or models, but here's a guess.

It looks to me like you have an "L" gib in there that you want to see if it will bend when the torque is applied to the square area.

If this is the case you are correct that "Bonded" is the incorrect choice as this will act as if the "L" gib is essentially the same part as the rail it is sliding on, in other words it is one big piece.

What you want, going from memory and I'm not an expert either, is to set a special contact set for the "L" gibs surfaces to the rail it is sliding on. You will set that to "No penetration" and there are other options like "Node to Node" etc etc. I'd look up those options and see which one makes the most sense to use.

Once you set up those contact sets you should see stress at the contact points between the "L" gib and rail.

As to why you're seeing 5000mps instead of 500 mpa. Could be all sorts of reason more than likely all of them specious and caused by sharp corners, to large a mesh etc. You'd have to hunt down the reasons for the high numbers and fix them as needed.

If that doesn't help, post your models and let some of the more expert people take a look at them :-)
Like • Show 1 Like1
Actions
Ryan Navarro Jan 12, 2021 9:41 AM
Like • Show 1 Like1
Actions
- Fahad Kafrawi @ Ryan Navarro on Jan 13, 2021 4:09 AM
  Thank you, I normally look at that and do total, just must have missed it in this setup but should really help cut my results in a 1/4
  Like • Show 0 Likes0
  Actions
Frederick Law Jan 12, 2021 9:51 AM
All you need to worry is when torque try to pull the gib and L-holder up.
So mate the L-holder to the gib and apply force up.
You can keep them bonded.
Only need 2 parts in assembly.
Keep it simple.
Like • Show 0 Likes0
Actions
- Matt Juric @ Frederick Law on Jan 12, 2021 10:09 AM
  Frederick Law wrote:
  
  All you need to worry is when torque try to pull the gib and L-holder up.
  So mate the L-holder to the gib and apply force up.
  You can keep them bonded.
  Only need 2 parts in assembly.
  Keep it simple.
  I've done bonded and unbonded for gibs like this and it gives two very different results. Bonded works if you're only worried about the entire system and are pretty sure that the gib and the rail are fairly equal.
  
  Non penetration gives you a better idea of which component, rail or gib will fail. non bonded also gives you an idea of stress points between the gib components themselves which will tell you where you'll most like see wear.
  Like • Show 1 Like1
  Actions
  - Frederick Law @ Matt Juric on Jan 12, 2021 10:14 AM
    From the assembly, The leg should be weak point.
    Unless it plastic gib and steel leg.
    
    Would the leg fly away with no-pen, if analysis 2 parts only? Pulling the leg up.
    Like • Show 0 Likes0
    Actions
    - Matt Juric @ Frederick Law on Jan 12, 2021 10:44 AM
      Frederick Law wrote:
      
      From the assembly, The leg should be weak point.
      Unless it plastic gib and steel leg.
      
      Would the leg fly away with no-pen, if analysis 2 parts only? Pulling the leg up.
      I agree the gib would be the week point and if it's not the rest is easy to "Fix" by comparison.
      
      We just shipped a machine that had two vises on them both were "L"gib and rail. This was a rebuild and the existing system had horrific wear in the gibs. So when I redesigned it I wanted to find wear points and redesign the gibs to take the loads it would see etc.
      
      The assembly I used for the simulation ended up being 5 parts total. (2) L-gibs, (1) rail, (1) base and (1) jaw that the gibs mounted to. Rail mounted to the base. The "L"-gib was a slip fit with the rail and the base.
      
      For all the surfaces between the "L"- gib and rail I used no penetration and remodeled it so I had .0005" clearance. I then applied a force at the point that the vise jaw was closing from and had the other side pushing against a virtual wall representing the part it was clamping against. The Rail and base had fixed mountings on them so didn't move.
      
      By doing the above the gib was entirely constrained in the simulation and did not fly away. You could do something similar with two parts by making some portion of either the rail or gib fixed and appliy the force to the other along with a virtual wall(s) to keep it from flying away. You would need at least two walls if you only did one gib.
      Like • Show 1 Like1
      Actions
      - Fahad Kafrawi @ Matt Juric on Jan 13, 2021 4:12 AM
        Thank you both so much for your educated responses. My computer is having issues atm so I can't rectify and test my simulation until it is fixed but you have both been a huge help and I can't wait to try these solutions.
        Like • Show 0 Likes0
        Actions

Simulation Stresses Too High

Outcomes

Related Content

Recommended Content