Bouncing Ball ? COSMOS Motion

Discussion created by 1-C0XRUK on Sep 2, 2008
Latest reply on Sep 24, 2008 by Shrikant Savant

Like • Show 0 Likes0
Comment • 15

Ok, I have a 4" diameter rubber ball in free space 2 inches above a .5 inch thick steel plate. I apply gravity and drop the ball. Everything I've tried setting all frame rates to max, 3-d contact resolution to max etc. and still the ball passes right through the plate.

I've left the properties as acrylic and/or chaged them to rubber and steel. THe only way I can make the 3-d contact work is Physical simulation. What I really want to is have the ball bounce as it should!

Any suggestions?

Outcomes

15 Replies

Shrikant Savant Sep 2, 2008 4:17 PM
Hi Tom,

You need to reduce the "Maximum Integrator Step Size" (Motion Study Properties-> Advanced Options).

Thanks,
Shrikant
Like • Show 0 Likes0
Actions
- Devi Prasad Samal @ Shrikant Savant on Sep 9, 2008 10:31 AM
  Hi
  
  I am curious to know is only giving contact and gravity between the ball & plate and then changing their material properties to rubber & steel. can the ball bounce?
  
  please let me know?
  
  i use spring option but it is not at all perfect for the bounce option. So please help.
  Like • Show 0 Likes0
  Actions
1-D88IXC Sep 9, 2008 12:32 PM
Bounce
Like • Show 0 Likes0
Actions
- 1-D88IXC @ 1-D88IXC on Sep 9, 2008 12:37 PM
  Settings
  Like • Show 0 Likes0
  Actions
  - 1-D88IXC @ 1-D88IXC on Sep 9, 2008 12:46 PM
    Another Example, Steel to Steel contact
    Like • Show 0 Likes0
    Actions
    - Devi Prasad Samal @ 1-D88IXC on Sep 10, 2008 6:33 AM
      Hi,
      
      Thanks for the setting. I have done the setting and it really working.
      
      but some time if i change gravity value or any change in the assembly create the ball to penetrate in to the box.
      
      what is the reason behind it? is it due to setting or any other reason behind this.
      
      here i attach the assembly. while motion study 1 works perfectly, motion study2 does not work.
      Attachments
      bouncing ball assembly.zip381.2 KB
      Like • Show 0 Likes0
      Actions
      - 1-D88IXC @ Devi Prasad Samal on Sep 10, 2008 10:58 AM
        
        Devi Prasad Samal wrote:
        
        Hi,
        
        Thanks for the setting. I have done the setting and it really working.
        but some time if i change gravity value or any change in the assembly create the ball to penetrate in to the box.
        what is the reason behind it? is it due to setting or any other reason behind this.
        here i attach the assembly. while motion study 1 works perfectly, motion study2 does not work.
        
        Hello Devi:
        
        You forgot to turn on precision contact in study2, so there is nothing do with changing the gravity value.
        I turn on precision contact and set accuracy to 0.001, the calculation is correct.
        Like • Show 0 Likes0
        Actions
1-D88IXC Sep 10, 2008 11:03 AM
One thing I notice that the plot of Y position (ball origin) looks like totally in-correct for me.

If you did a plot of acceleration of ball, it will show some spikes which is extraordinarily high, this is an issue since SW2007. And it has not been fixed yet in SW2009.
If you plot a trace path of ball , it will not show the correct path, this is really sucks.

I hope some expert can explain the plot results.
Like • Show 0 Likes0
Actions
- Shrikant Savant @ 1-D88IXC on Sep 10, 2008 1:34 PM
  When dealing with 3D contacts involving a thin body, like a plate as in this example, if you see that the bodies pass through each other without recognizing the 3D contact it is usually due to too large a value for the "Maximum Integrator Step Size". The integrator will take as large a step size as is allowed by this setting, so it may take one step just before the contact occurs and the next step just after the contact. In such a case, since there is no change in force before or after, it has no way to know that it missed a contact event between the bodies. The way to circumvent this problem is to slow the integrator down, by reducing the max step size. But there is a trade-off, too small a value will slow down the integrator and the overall simulation. So, you need to play with this setting depending upon your problem.
  
  The other parameter to play with is the integrator itself. SI2_GSTIFF integrator is more accuracte in controlling errors in velocities and accelerations than GSTIFF and WSTIFF. If you see that with GSTIFF or WSTIFF, the simulation fails then try SI2_GSTIFF. The trade-off here is that SI2_GSTIFF is usually slower than the other integrators.
  
  I changed the max integrator step size to 0.001 and changed the integrator to SI2_GSTIFF and was able to solve both motions studies.
  
  Cheers,
  Shrikant
  Like • Show 0 Likes0
  Actions
1-D88IXC Sep 10, 2008 1:47 PM
Shrikant

Thanks for the good explanation. do you mind to post a plot result of acceleration and Y displacement of the ball and a screen shot of the trace path base on your study result?
And it would be best to post your setting here as well.

regards
Like • Show 0 Likes0
Actions
- Shrikant Savant @ 1-D88IXC on Sep 10, 2008 9:38 PM
  Hi Shaodun,
  
  The files are attached.
  
  Regards,
  Shrikant
  Attachments
  - AdvMotionStudyProps.JPG35.5 KB
  - MotionStudyProps.JPG18.1 KB
  - plot_acc_disp.JPG55.5 KB
  Like • Show 0 Likes0
  Actions
  - 1-D88IXC @ Shrikant Savant on Sep 10, 2008 11:20 PM
    
    Shrikant Savant wrote:
    
    Hi Shaodun,
    
    The files are attached.
    
    Regards,
    
    Shrikant
    
    Hi Shrikant:
    
    Thank you very much for your reply.
    
    Surprisingly see you did not turn on Precise Contact and did not use High 3D Contact resolution, I thought it won't work in the first place.
    
    For the acceleration plot , I saw the same spikes as I my plots which is extremely high, could you explain why CWMotion can not produce a " Physical correct " result?
    Like • Show 0 Likes0
    Actions
    - Devi Prasad Samal @ 1-D88IXC on Sep 11, 2008 4:21 AM
      Hey
      
      Can u have some good examples on cosmoss motion?
      
      any thing....
      Like • Show 0 Likes0
      Actions
    - Shrikant Savant @ 1-D88IXC on Sep 24, 2008 12:47 PM
      You need to look at your results in the context of your modeling assumptions. SWMotion assumes all parts to be perfectly rigid. This is clearly an approximation of the reality. Under this assumption, when a rigid ball impacts a rigid surface, the entire force of contact goes towards accelerating the bodies involved. The acceleration is calculated by the solver to be simply force at the moment of contact divided by the mass of the body (there are no other constraint forces in this example). In real life, this force will be reduced by what is absorbed by the deforming surface and so the net acceleration of the bodies involved will be less at the moment of contact. Though the real equations are a more complicated, conceptually this can be explained by :
      a = (F_impact - k_surface*surface_deflection - c_surface*deflection_vel)/mass.
      The last 2 terms in the numerator are absent in the rigid body case, thus causing the acceleration to be high.
      
      The precise contact option uses the CAD geometry kernel to compute intersection of bodies. If this option is not checked, the geometry is tesselated and the intersection is computed based on the tesselated geometry. The higher the 3D contact resolution, the finer the tesselation. The precise option is usually more accurate but can be a lot slower.
      
      Hope this helps,
      Regards,
      Shrikant
      Like • Show 0 Likes0
      Actions
Shaodun Lin Sep 11, 2008 6:17 AM
There is a nice mouse trap assembly which demonstrated the functionality of CMotion very well, maybe you can find this somewhere in SW website. Or you can ask Shrikant give you a link to download it.
Like • Show 0 Likes0
Actions

Bouncing Ball ? COSMOS Motion

Outcomes

Attachments

Attachments

Related Content

Recommended Content