Bouncing Ball

AliMosayyebiAliMosayyebi Member Posts: 31

Hi there,

 

I am trying to simulate a bouncing rubber ball on a flat surface.

I have attached what I thought should be done.

Could someone please have a look at it and let me know if this is correct and what I am missing? Since its not really showing what I was expecting as in animation when I press the play button in "Total Deformation".

 

Thanks for the help in advance.

Best,

Ali 

Comments

  • peteroznewmanpeteroznewman Member Posts: 11,071
    edited January 2019

    Hi Ali,

    Your model attempts to simulate dropping a ball from a height of 5 mm toward a fixed surface. Standard gravity accelerates the ball downward. You can use simple physics equations to determine that it takes 32 milliseconds for the ball to hit the surface. You can use those same simple equations to calculate how far the ball will have dropped after 1 second: it is 4.9 meters (half of 9.8). When you play the animation of the simulation, it multiplies the result by a factor of 8.7e-7 to fit such a large displacement on the screen. You should set the Result field to 1.0 (True Scale) and you will see that the ball is 900 meters below the plate.

    One problem with your model is you had an end time of 14 seconds and a Time Step of 1 second. What you want to see the ball falling is a time step of 2 ms to get 16 frames of the fall till impact. I changed your model to have Auto Time Stepping On and set a Maximum Time Step of 2 ms, but I didn't get what I expected.

    Another problem with your model is you have Bonded Contact which is like glue. What you want is Frictionless contact.

    After spending 32 ms falling toward the surface, the ball has reached impact velocity, which you can calculate with the simple physics equations. Please reply with your answer for the impact velocity.

    The problem that you haven't even seen yet is that the impact event is on a much shorter time scale than the 2 ms time step that was sufficient for the fall. It is orders of magnitude shorter. In order to help the solver see the impact, you have to use a second step that starts just before impact, and you have to use a much shorter time step to catch the impact and have several frames of the ball compressing and decompressing.

    Most people studying impact don't bother with the ball falling. They can calculate the impact velocity, so they use the geometry editor and they place the ball in contact with the plate, but they assign an initial velocity to the ball, the impact velocity calculated from the simple physics equations, and start the simulation on the very short time step required to see the ball compress and decompress over several time steps.

    If the ball is perfectly elastic and the end time is 32 ms, you can simulate the ball rebounding to its original drop height.

    Regards,
    Peter

  • jj77jj77 Member Posts: 843
    edited January 2019

    As peteroznewman says, you need first of all to have a frictional or friction less contact, and the simulate, say up to 0.07 s, to be able to see the rebound.

     

    Also one would place the ball closer to surface (almost touching), and then assign and initial velocity.

     

    If you want to model the whole drop, one would split this event up into two steps (or three). 

    Step1. Prior to contact. Ball falling freely almost touching (use a 0.01 s f or time step, simulating up to say 0.034 s). This with these large time steps, the first free fall part runs very fast.

    Step2. Contact. Ball getting in contact, use auto time steeping on and defined by time, with a much finer time step than the first step in order to capture the contact, say initial time step of 1E-6 s, and smallest, 1E-5, and maximum 5E-6, or even smaller time steps, in order to get enough time steps capturing the contact. End time for step 2 being say 0.08 s. If you have 3 steps then end time for this step could be 0.037 s, or until the contact is finished and the ball is moving up.

     If you have a 3rd one (after contact) so that would be after the contact is done (after say 0.037 s), and the ball is moving upwards simulating up to 0.08 s, use the same time steps as in step1 that is ~0.001 s. See attached image for details.

     

    As peteroznewman said use a scale of 1 only.

     

    I would also change the boundary condition on the plate, it does not make sense to have it all fixed (means it is an infinitely rigid body).

     

    Finally consider quarter symmetry to reduce model size, and a better structured hex mesh.

  • AliMosayyebiAliMosayyebi Member Posts: 31
    edited January 2019

    Hi both,

    thanks for the explanation and the information.

    I have now tried the instructions you kindly sent and the attached is the outcome.

    Attachments are as follow:

    Workbench files: Bouncing ball - Community copy v2

    Drawing for the main idea: object in the tube.png

    Regarding the simulation, basically this is part of a bigger idea I am exploring. In the main idea, there is a pipe (bent) with some angles in which fluid is being injected from its inlet and is coming out from the outlet. Then a small object is being released close to the inlet and I am hoping to see the movement of the object due to the fluid and its gravitational force.

     

    Regarding the time step, I have now tried changing the contact to frictionless, update the fixed surfaces and modify the timing. For the timing thought initially focusing on the step one out of three steps. (step1: falling before contact, step 2: contact and step 3: after contact). To keep it simple for now, I focused on the initial 0.0031s before contact.

     

    But, still making all these changes, nothing happened. could you please help me on what I am not doing write?

    as soon as this starts working, I will work on modifying the system by going to the 3 steps, then reducing the size etc etc.

     

    Kind regards,

    Ali

  • jj77jj77 Member Posts: 843
    edited January 2019

    You still have only one step (solves only for 0.0031, and you need to solve for 0.08 s, so it does not even reach the plate). As shown in the image I attached you need to have 3 steps, with the settings shown in that image (shown also below). Thus set number of steps to 3. You do that in number of steps, and step controls and under analysis settings. Then you need to change the current step numbers one at a time and set time step settings as shown below. Hope this explains it.

  • AliMosayyebiAliMosayyebi Member Posts: 31
    edited January 2019

    thanks very much

    But could you send another copy of the image since it is not readable. the easiest way is to attach is separately rather than putting the picture in the text. 

  • jj77jj77 Member Posts: 843
    edited January 2019

    I have attached in my previous post (not very good perhaps ), but here goes again.

     

  • AliMosayyebiAliMosayyebi Member Posts: 31
    edited January 2019

    sorry but none of them are readable. I have attached the image of how I can see it for your reference.

  • jj77jj77 Member Posts: 843
    edited January 2019

    One more try, this time it is a powerpoint so that should be fine.

  • AliMosayyebiAliMosayyebi Member Posts: 31
    edited January 2019

    Thanks very much, I got the file and it was fine.

    So, I have now changed the simulation with the steps you sent me and still not showing the expected result!! I assume it worked at your side, right? am I missing something?! I have uploaded the files for the reference.

  • jj77jj77 Member Posts: 843
    edited January 2019
    Yes it works. I would not spend 3 posts on something that is not solving. I thought that you might be new to Ansys and say steps and substeps, so perhaps this will help you understand this see http://www.padtinc.com/blog/the-focus/you-dont-wanna-step-to-this-breaking-down-loadsteps-and-substeps-in-ansys-mechanical.

    I will have a look tomorrow if I have time and see what is wrong, but if you followed the power point and defined 3 steps with the setting showed it should work.
  • peteroznewmanpeteroznewman Member Posts: 11,071
    edited January 2019

    Ali,

    The problem with your v2 model is that you set the step end time to .0031 s or 3.1 ms when I said that the time to impact was .032 s or 32 ms. 

    Attached is a Workbench 19.2 archive file of a bouncing ball. Follow these directions to Restore Archive.

    The Automatic Time Stepping logic is very good, so I used a small Initial Time to begin Step 2, just before impact, and the time step size increases automatically, so I don't have too many substeps before the end time is reached and there is no need for a Step 3.

    If this post answers your original question, please click Is Solution below to close the discussion or ask a follow-up question. I recommend starting a New Discussion on the topic of an object in fluid in a tube being carried along with flow and gravity acting. That is a very different topic than a bounding ball.

    Regards,
    Peter

  • AliMosayyebiAliMosayyebi Member Posts: 31
    edited January 2019

    Thanks very much to both of you.

    you indeed helped me as being  new to this. I will play with it before posting it at the fluid section for the tube.

    Best,

    Ali

  • jj77jj77 Member Posts: 843
    edited January 2019

     Actually the model (bouncing ball copy v3), which I assume is the latest worked. The only thing is that contact takes place during 2-4 time steps. so it is hard to spot (use the contact tool a s shown/highlighted in red below to detect the contact event, see blue marking ). Also you were probably not using a displacement scale of 1, see yellow marking. Use 1 (true scale) always for nonlinear analysis. All the best,

    jj77.

  • AliMosayyebiAliMosayyebi Member Posts: 31
    edited January 2019

    Thanks very much for the help and the patient, very helpful. being new in this, I will need to play more until I am confident with all the bits.

  • bozhengbozheng Member Posts: 8
    edited January 22

    Hello all,

    I wonder how to simulate a ball move in a U tunnel? I use explicit dynamic, the initial velocity is about 3m/s, The bottom is fixed. But, it always could not pass through the curve. Can you give me some advice? Thank you very much for your help.


  • peteroznewmanpeteroznewman Member Posts: 11,071

    @bozheng

    Make the ball smaller?

Sign In or Register to comment.