AnsweredAssumed Answered

No Penetration Troubleshoot

Question asked by Robert Fichera on Jul 1, 2014
Latest reply on Jul 3, 2014 by Shaun Densberger

Like • Show 0 Likes0
Comment • 20

Hi there,

I am running an assembly simulation. Initially, I used a globally bonded contact for my assembly - and the simulation was succesful as expected. Now, the only thing I altered was a no penetration contact set between two bodies. I know that local contact sets will override the global contact - so I know that my contact sets are valid. The ONLY thing I changed was switching from a bonded contact, to a no pentratino contact. Understand that where I applied this setting I am using a bolted connection - therefore solidworks requires a no penetration contact between those two faces. Now when I try to run the study, the solver will never finish the simulation. I will usually be prompted with "Cannot create master control file" or more recently "solver encountered numerical difficulties". I have no idea why this is happening. The study will run for well over an hour before I prompt with anything.

If it makes my question easier to answer, I using a curvature based mesh (relatively coarse for a good baseline study) with a max size of 15mm, min size of 3mm(more of less, cant remember), # of tetrahedrons per hole of 10, and a growth ratio of 1.3 I have no issues meshing the assembly.

Please help!

Thank you!

No one else has this question

Outcomes

20 Replies

Shaun Densberger Jul 1, 2014 2:46 PM
Mark Correct
Correct Answer
Can you post your model?
Like • Show 1 Like1
Actions
Robert Fichera Jul 1, 2014 6:59 PM
Mark Correct
Correct Answer
Of course, here you go.
Like • Show 0 Likes0
Actions
- Shaun Densberger @ Robert Fichera on Jul 1, 2014 7:17 PM
  Mark Correct
  Correct Answer
  Sorry, I should have be more specific; can you post the model itself (assembly and part files)? This issue will be easier to diagnose if I can look at the model setup.
  Like • Show 0 Likes0
  Actions
  - Robert Fichera @ Shaun Densberger on Jul 1, 2014 8:17 PM
    Mark Correct
    Correct Answer
    Send me your email and I will forward them to you. I cannot attach them to the forum
    Like • Show 0 Likes0
    Actions
    - Shaun Densberger @ Robert Fichera on Jul 1, 2014 8:47 PM
      Mark Correct
      Correct Answer
      shaun.densberger@sanmina.com
      Like • Show 0 Likes0
      Actions
      - Robert Fichera @ Shaun Densberger on Jul 1, 2014 9:17 PM
        Mark Correct
        Correct Answer
        Okay, I sent them your way.
        Like • Show 0 Likes0
        Actions
    - Jared Conway @ Robert Fichera on Jul 1, 2014 9:17 PM
      Mark Correct
      Correct Answer
      you can post to the forum using the advanced editor at the top right corner of the window.
      
      have you tried running the analysis with just the contact no external loads and also with no bolt preload?
      
      i suspect you're trying to do too much at one time and need to isolate what is causing the issue first.
      Like • Show 0 Likes0
      Actions
      - Robert Fichera @ Jared Conway on Jul 1, 2014 9:19 PM
        Mark Correct
        Correct Answer
        Jared, I agree with your thinking. I do have bolt pre-load.
        
        I will try those suggestions. In the meantime, would you know why bolt pre-load may hinder the solution process?
        Like • Show 0 Likes0
        Actions
        Jared Conway @ Robert Fichera on Jul 2, 2014 1:33 AM
        Mark Correct
        Correct Answer
        i've seen a lot of people put too high of a preload which makes it challenging to converge the problem with the contact also included
        Like • Show 0 Likes0
        Actions
Nicholas Luyster Jul 1, 2014 10:45 PM
Mark Correct
Correct Answer
Hi Robert,

If you switch to a course mesh, you'll be able to get to the error message much faster. This should save you some time while you're troubleshooting.

All the best,

Nick
Simulation Trainint and Tutorials at gosimulation.com
Like • Show 1 Like1
Actions
Shaun Densberger Jul 2, 2014 1:02 PM
Mark Correct
Correct Answer
Your model came in with some parent-child issues that caused some simulation definitions to require a redefine. I redefined them as best as I could for how I understand your model (disk brake under braking).

That being said, I was able to get your model to run with your set up, so I'm not too sure why you're having issues (I'm using SW 2014 though). I would try what Jared suggested with regards to the bolt pre-load value, as well as Nick's suggestion on the mesh size (although too course of a mesh on the contact surfaces can cause convergence issues, so keep that in mind). I also think it'd be worth some of your time to clean up the geometry a bit to remove small features to improve the robustness of your mesh.

One other thing. Once you get a stable model running, I'd suggest that you reconsider some of your bonded interfaces in your model for accuracy sake.
Like • Show 1 Like1
Actions
- Robert Fichera @ Shaun Densberger on Jul 2, 2014 1:22 PM
  Mark Correct
  Correct Answer
  I have taken away the bolt pre-load. Also, you cannot run the study without any external loads. So what I did instead was run the study one at a time with each of the three loads. What I found was that running the study with both brake torques prompted a warning for large displacement. Now, I understand that stressing the material beyond its yield is an indication that a nonlinear study should be done instead of a static study, yes?
  Like • Show 0 Likes0
  Actions
  - Robert Fichera @ Robert Fichera on Jul 2, 2014 1:24 PM
    Mark Correct
    Correct Answer
    Could you possible upload the assembly file with your changes?
    Like • Show 0 Likes0
    Actions
    - Shaun Densberger @ Robert Fichera on Jul 2, 2014 2:38 PM
      Mark Correct
      Correct Answer
      I can, but they'd be SW 2014 and a non-educational version at that. Is that OK?
      Like • Show 1 Like1
      Actions
  - Shaun Densberger @ Robert Fichera on Jul 2, 2014 2:37 PM
    Mark Correct
    Correct Answer
    Also, you cannot run the study without any external loads.
    
    Correct. You could add a dummy load to get the analysis to run though (something like 0.001 g of gravity).
    
    What I found was that running the study with both brake torques prompted a warning for large displacement. Now, I understand that stressing the material beyond its yield is an indication that a nonlinear study should be done....
    
    Generally speaking, yes. However, if the above-yield stresses are due to singularities, then you don't need to run a nonlinear analysis; you'll just need to handle the singularities in the appropriate manner. If you do have above-yield stresses that aren't due to singularities, then you'll need to define the appropriate elasto-plastic material model (perfectly plastic, linear hardening, etc.) Be aware that if you decide do go down the elasto-plastic route, your displacements will get larger due to the decrease in the Young's Modulus when you go into the plastic region. I'm not sure what the purpose of your project is, but if it's to design a spindle to handle the applied loads, then I'd stay away from a elasto-plastic material model and large-displacement formulation and modify the geometry of the model to bring the stresses to a sub-yield level.
    
    ...instead of a static study, yes?
    
    Nonlinear and static are two different categories. Linear and Nonlinear refer to the assumptions used in the formulation of the model. Linear assumes purely constant material properties (no plasticity, temperature dependent CTE, etc.), small strains and displacements (no second order terms), no contact (path dependent solution), etc. Nonlinear doesn't necessarily make these assumptions (it depends on how you set up the model), and can account for some or all of these effects. Static and Dynamic refer to whether your solution is time-independent or time-dependent. As such, you have four analysis types:
    Linear Static
    Linear Dynamic
    Nonlinear Static
    Nonlinear Dynamic.
    Like • Show 1 Like1
    Actions
  - Jared Conway @ Robert Fichera on Jul 2, 2014 3:25 PM
    Mark Correct
    Correct Answer
    mroe than likely you have a load issue, probably too high rather than needing nonlinear
    nonlinear is only needed for large displacement (doubt you want this) or plastic deformation (doubt you want this either)
    
    suggest you sit down and figure out how you came up with your loads
    what your expectations are regarding those loads
    if your loads are right, your design is poor and needs adjustment
    Like • Show 0 Likes0
    Actions
    - Robert Fichera @ Jared Conway on Jul 2, 2014 4:33 PM
      Mark Correct
      Correct Answer
      Shaun, then in order to determine whether there are any stress singularities, would you recommend refining the mesh and re-running the study?
      I am designing my material to stay within a reasonable safety factor of its yield stress, therefore, would you agree I should steer clear of elasto-plastic properties (as Jared mentioned).
      
      I also agree with Jared's logic. There are two braking torques appied at the brake caliper. Applying them individually works without issue - when together, soldiworks detects large displacement. So I think we are all getting on the same page here - I just wanted to clarify.
      
      Also, I am installing SW 2014 soon... You might as well attach it and I'll see if it works...
      Like • Show 0 Likes0
      Actions
      - Jared Conway @ Robert Fichera on Jul 2, 2014 8:07 PM
        Mark Correct
        Correct Answer
        what needs clarification still?
        
        linear material is fine as long as your material isn't significantly nonlinear, you don't care about plastic deformation, which i think is your situation
        large displacements are probably real large displacements and due to a setup issue
        Like • Show 0 Likes0
        Actions
      - Shaun Densberger @ Robert Fichera on Jul 3, 2014 3:48 PM
        Mark Correct
        Correct Answer
        I am designing my material to stay within a reasonable safety factor of its yield stress, therefore, would you agree I should steer clear of elasto-plastic properties (as Jared mentioned).
        
        Yes.
        
        I also agree with Jared's logic. There are two braking torques appied at the brake caliper. Applying them individually works without issue - when together, soldiworks detects large displacement.
        
        I'd double the thickness (or more; I forget how much over yield the stresses were) of the support member on the spindle for the caliper to resolve this.
        
        Here's a zip of the files I was using.
        Attachments
        BrakeAssemblySW14.zip1.3 MB
        Like • Show 1 Like1
        Actions
      - Shaun Densberger @ Robert Fichera on Jul 3, 2014 3:56 PM
        Mark Correct
        Correct Answer
        I am designing my material to stay within a reasonable safety factor of its yield stress, therefore, would you agree I should steer clear of elasto-plastic properties (as Jared mentioned).
        
        Yes.
        
        I also agree with Jared's logic. There are two braking torques appied at the brake caliper. Applying them individually works without issue - when together, soldiworks detects large displacement.
        
        I'd double the thickness (or more; I forget how much over yield the stresses were) of the support member on the spindle for the caliper to resolve this.
        
        I've emailed you the files that I was using.
        Like • Show 1 Like1
        Actions

No Penetration Troubleshoot

Outcomes

Attachments

Related Content

Recommended Content