8 Replies Latest reply on Apr 1, 2014 5:48 AM by Mark Scholten

bolt connectors on sheetmetal parts

Mark Scholten Mar 26, 2014 7:51 AM

Hi everyone,

i am sort of stuck on a project and hoping that someone can get me going again. I am doing a static study of a sheetmetal duct loaded by internal fluid pressures. The loading is such that membrane stresses will be significant. The sheetmetal panel flanges are connected using bolts. I am interested in plate deflections, plate stresses (in the webs and flanges at the bolt holes) and bolt stresses.

So initially I used "no penetration" contact conditions between the panels and bolt connectors, used shell elements created by SW simulation (so did not create them myself), and ran the study using the large displacement option to account for membrane stresses.

As it turns out, the simulation fails (inspite of http://help.solidworks.com/2013/English/SolidWorks/cworks/r_connector_bolt.htm) on the bolt connectors in combination with shell elements (see https://forum.solidworks.com/thread/49609), i get an error stating that the bolt length to diameter ratio is less than 5 and solids elements should be used.

After reading the post 49609 and KB S-058593 I decided to first try running the study using solid elements. Besides taking a very long time (even though the mesh only has 1 element in the through-thickness direction), it fails in the first load iteration (at 5% load) without stating what the problem is.

What is the smart road to take? Manually creating faces, replacing the bolt connectors and running the study without further changes? I am confused by the remark in

KB S-058593 stating that shell thickness should be SMALL enough to obtain desired accuracy, while the problem seems to be that the bolt length-diameter ratio should be INcreased... Does that mean that creating the shells manually would not remove the problem? Doing a validation study as KB S-058593 suggests is something i would like to avoid as we are pressed for time.

Thanks for your ideas,

Mark

Re: bolt connectors on sheetmetal parts

Mark Scholten Mar 26, 2014 10:03 AM (in response to Mark Scholten)

Ok, so what i am trying now is to first:
-delete all bolt connectors
-use shell elements
-use global BONDING instead of global NO PENETRATION conditions
-and use large displacement option.

Then:
-duplicate the study and run it WITHOUT large displacement and
-WITH "compute free body forces" option.

So that in the first run I get to assess stresses and deflections within the panels, and in the second run I hopefully find the forces between the flanges. Then i can use those to size the bolts.

This approach is necessary, because large displacement option and compute free body forces option cannot be used simultaneously.

Anyone who agrees / disagrees?
Mark
Like Show 1 Likes(1)

Actions
- Re: bolt connectors on sheetmetal parts
  
  Jared Conway Mar 26, 2014 4:24 PM (in response to Mark Scholten)
  
  are you expecting large displacements?
  
  Like Show 0 Likes(0)
  
  Actions
  - Re: bolt connectors on sheetmetal parts
    
    Mark Scholten Mar 26, 2014 7:37 PM (in response to Jared Conway)
    
    yes
    
    Like Show 0 Likes(0)
    
    Actions
    - Re: bolt connectors on sheetmetal parts
      
      Jared Conway Mar 27, 2014 3:18 AM (in response to Mark Scholten)
      
      Then you'll need nonlinear.
      
      But going back to your question, I'm not sure what you're stuck on. Your approach seems reasonable relative to the limitation with bolts and shells if you don't want to use solids.
      
      Like Show 0 Likes(0)
      
      Actions
      - Re: bolt connectors on sheetmetal parts
        
        Mark Scholten Mar 27, 2014 5:58 AM (in response to Jared Conway)
        
        Hi Jared,
        
        actually i was under the assumption that the large displacement option should be sufficient as I am not interested in stresses beyond yield stress. i need the large displacement option only to account for membrane stresses BELOW yield. the SW sim book says it is suited for this type of analysis.
        
        nonlinear is suited specifically for nonlinear material behaviour, right? i am dealing a lot with plate fields loaded by fluid pressures, and try to avoid any yielding. for what aspect is nonlinear required?
        
        where i was stuck is:
        the limitation of using bolts and shells
        not being able to get the study to complete using solids and large displacement. it would terminate at the 33% load iteration without stating the nature of the failure.
        
        Like Show 0 Likes(0)
        
        Actions
        
        Re: bolt connectors on sheetmetal parts
        
        Jared Conway Mar 27, 2014 1:00 PM (in response to Mark Scholten)
        
        linear is good to about 20-30% before yield, if you're beyond that, want nonlinear materials, large displacements, large displacement option failed, changing contact locations > nonlinear is the right solution
        
        if large displacement is failing on you, you may need nonlinear or you may need someone to look over your analysis setup to make sure the LD is caused by actual LD and not a setup issue
        
        bolts are what they are
        
        Like Show 0 Likes(0)
        
        Actions
Re: bolt connectors on sheetmetal parts

Shaun Densberger Mar 27, 2014 1:31 PM (in response to Mark Scholten)

Ok, so what i am trying now is to first:
-delete all bolt connectors
-use shell elements
-use global BONDING instead of global NO PENETRATION conditions
-and use large displacement option.

Agreed, although I'd start with with small displacement formulation to make sure everything is connected up properly.

Then:
-duplicate the study and run it WITHOUT large displacement and
-WITH "compute free body forces" option.

So that in the first run I get to assess stresses and deflections within the panels, and in the second run I hopefully find the forces between the flanges. Then i can use those to size the bolts.

This approach is necessary, because large displacement option and compute free body forces option cannot be used simultaneously.

Be careful with this. If small displacement formational is not valid, then you could see a different between the load paths calculated when using small and large displacement formulation. It'd be better to get your model working with the fasteners modeled.

i get an error stating that the bolt length to diameter ratio is less than 5 and solids elements should be used.

Does this prevent you from running the analysis? If you change the diameter of the hole (such that the ratio is greater than 5), does this warning go away?

What is the smart road to take? Manually creating faces, replacing the bolt connectors and running the study without further changes?

Generally speaking, it is better to manually create your contact interfaces.

I am confused by the remark in
KB S-058593 stating that shell thickness should be SMALL enough to obtain desired accuracy, while the problem seems to be that the bolt length-diameter ratio should be INcreased... Does that mean that creating the shells manually would not remove the problem?

Shell elements are an idealization and the assumptions in shell element theory have limitations. A general rule of thumb for shell elements is that the ratio of your smallest non-thickness dimension (on your part) to the thickness of the part should be 10 or more.

Doing a validation study as KB S-058593 suggests is something i would like to avoid as we are pressed for time.

I don't have access to the KB at the moment, but typically it's rather risky to avoid doing a validation portion of the analysis.
Like Show 2 Likes(2)

Actions
- Re: bolt connectors on sheetmetal parts
  
  Mark Scholten Apr 1, 2014 5:48 AM (in response to Shaun Densberger)
  
  Hi Shaun,
  
  thanks for your detailed remarks.
  
  I am aware that load paths can differ between large and small displacement analysis, but the bolt error is indeed preventing me from running the analysis.
  The KB S-058593 states that when using bolt connectors in combination with shell elements and when the bolts have a length/diameter ratio smaller than 5, the bolt forces become dependent on the mesh density. It then recommends a validation study to find the appropriate mesh density that yields the best bolt forces. Because the study won't complete using solid elements (it fails on the 5% load iteration), the only option I have to assess bolt forces is to do a small displacement run while computing body forces and dividing the flange interface forces by the number of bolts...
  
  Anyways, I think I'm alright now. The first results look reasonable and indicate the most critical regions of the structure. To keep displacements within allowable ranges (stresses are allowable) I am adding reinforcements which reduce the differences between the large and small displacement results. Bolt forces also seem reasonable.
  
  Thanks for your support, if I run into any new problems I will check back in!
  Mark
  
  Like Show 1 Likes(1)
  
  Actions

Go to original post