Ansys Products

Ansys Products

Discuss installation & licensing of our Ansys Teaching and Research products.

2 way coupled FSI for ball bearing

    • ashishkumar.gupta
      Subscriber

      This is how it is modeled for 2- way FSI. I guess this is the right approach as due to the Forces from Oil it will cause some Displacement/movement of Races/Ball and Vice-versa.

      I have prepared a model with a ball, IR and OR.

      IR rotates (by giving circumferential Disp.) and radial force also applied bcoz of shaft and OR is fixed --- Transient

      Lub. inlet on left side and outlet on another. Isothermal, incompressible fluid and SST k-w model used with Dynamic Meshing.

      But the model does not run as there is no direct contact between ball and races.

      any leads will be highly apprecitated to simplify the model. Thanks!

    • palmagoldner105
      Subscriber

       

       

      I really hope you achieve the effect through this idea.

      Slice Master

       

       

    • Rahul
      Ansys Employee

      Is it throwing any error? 

       

       

    • ashishkumar.gupta
      Subscriber

      Mech BC:

       

      SC:

      Fluent:

      Mech: 

      I do not understand what exactly is happening.

    • Rahul
      Ansys Employee

      Negative cell volume detected in Fluent. You can open the Fluent results and check the exact location of failure. I guess it is in the lubricant thickness region between the ball and the race. Have a sufficient number of cells of good quality in that region. 

      You can refer 'Mesh Folds within the First Coupling Steps' in the user guide: 42.6. FSI Setup Recommendations for Fluent-Mechanical Couplings (ansys.com)

    • ashishkumar.gupta
      Subscriber

      Added a fine mesh with inflation layer to the Lubricant body. Still Same error.

      Atleast the model should run right? Like I am not looking for accuracy of the results but rather the workable model...It is giving me a hard time...please look into it Thank you!!

       

    • Rahul
      Ansys Employee

      How many cells do you have in that thin region and what are the dynamic mesh settings? Are you using remeshing?

    • ashishkumar.gupta
      Subscriber

      Smoothing - Diffusion &

      Remeshing - Unified Meshing

    • Rahul
      Ansys Employee
      1. After failure, check the location of the mesh fold in Fluent and improve the mesh at that location using local sizing. Also, check the suggestions from 'Mesh Folds within the First Coupling Steps' in the user guide: 42.6. FSI Setup Recommendations for Fluent-Mechanical Couplings (ansys.com).
      2. Try applying the gradual load with ramping of data transfers to avoid the mesh failure at the start of the solution.
      3. Set lubricant zone to 'Deforming' in Dynamic mesh zones settings. You can refer to tutorial Reed Valve FSI Co-Simulation with Partial Setup Export from Workbench (Fluent-Mechanical) to understand the Two-way coupling setup.
    • ashishkumar.gupta
      Subscriber

       

      Okay I will try this now and revert back to you.

       But you sure it is not because of Rigid body motion in Mech. but due to Coarse Meshing in Fluid? Because if we do fine meshing it is really time consuming, sometime taking 2-3 days and that is the reason for going for decent mesh over finer mesh.

       

      • Rahul
        Ansys Employee

        You can create a plane through the point of contact and investigate the mesh in that thin region.

        To avoid negative volume, try defining the contact offset in Mechanical between those surfaces. You can find more information about it in Geometric Modification.

        Negative volume generation issue could be due to several reasons:

        1. First, check your mesh and ensure its quality is good. If not, please consider improving the mesh in the region where mesh motion is expected.

        2. Consider reducing the time step. (change time in system coupling).

        3. Check if unreasonable forces were received by Fluent leading to negative mesh in Fluent (Check BC in both participants and use data transfer ramping)

        • ashishkumar.gupta
          Subscriber

          1. Hex Dominant finer mesh with section plane:

           

          2. Smaller time steps in SyC/Transient:

           

          3. Gradual Radial load in +Y dir:

           

          None of this worked out...it solves upto 12% and then fails. 

        • ashishkumar.gupta
          Subscriber

          Hi Rahul, could you please look into the updates and help me out. Thank you!

    • ashishkumar.gupta
      Subscriber

       

      Hi Rahul,

      I tried whatever you said but still the same error persists. 

       

    • ashishkumar.gupta
      Subscriber

      Hi Rahul, Luckily with try & error, negative cell volume issue was resolved.

      But I encountered new problem. Could you tell me is this because of some weird setting in fluent?

    • Rahul
      Ansys Employee

      Hello,
      Check your boundary conditions here to see if the forces/displacement applied to the correct boundary.

      • ashishkumar.gupta
        Subscriber

        Hey, Yes in the Transient Structural it is applied Correctly.

        If I do not use FSI and keep frictional contacts in Transient Structural, The model works perfectly fine as shown in the LEFT image above.

        If I use FSI with same setup but COULD NOT provide contacts in Transient Structural (as there is 0.1 mm Lubricant ), The model behaves as shown in the RIGHT image above.

      • ashishkumar.gupta
        Subscriber

        Hi, any update? Thank you!!

    • Rahul
      Ansys Employee

      Check if the forces from fluent are transferred correctly. Have you tried running a Fluent alone case? Is it working well? 

      Also, share the contact settings and BCs used in Mechanical participant.

    • ashishkumar.gupta
      Subscriber

       

      How do we check the forces from Fluent are transferred correctly or not? 

      I simulated Fluent alone..It ran .. Since it is just the flow of fluid, I do not think it will be any problem.

      The attached image below is of fluid pressure around the ball. 

       

      Below is the Mech. alone setup with frictional contact:

       

    • Rahul
      Ansys Employee

      By run fluent alone, I mean run Fluent alone with a representative load on the FSI boundary to see the mesh deformation and assess if it is solving correctly.

      • ashishkumar.gupta
        Subscriber

        But how do I calculate that load or pressure? Like the forces are transferred from races to Lubricant to ball...

        • ashishkumar.gupta
          Subscriber

          Can u tell me how can I assess if the forces are being transferred fromn Fluent to Mech properly?

    • Rahul
      Ansys Employee

      Have you removed this contact definition in the FSI case?

      • ashishkumar.gupta
        Subscriber

        Yes, in case of FSI, since the ball and races are not in contact mechanically, there is NO contact definition assigned.

    • Rahul
      Ansys Employee

      Try keeping the contact definition as is. It models a true gap as you are not providing any offset here. Using this setting will not close gaps.

      Check 9.6.3.5. Geometric Modification for more info. 

       

    • ashishkumar.gupta
      Subscriber

      But the contact was automatically read in Mech. as the parts were touching each other.

      So in FSI, u want me to MANUALLY provide Frictional contact (0.2) with interafce treatment as: 

      Am I right? will that solve the issue?

    • Rahul
      Ansys Employee

      Change it to frictionless contact type as they not touching each other.

      • ashishkumar.gupta
        Subscriber

        ok I will Manually assign Fritionless contact

    • 220512119
      Subscriber

      Negative volume error, make sure the mesh mesh in fluent has an orthogonal mass of more than 0.2, and then try to lower the step size again

      • ashishkumar.gupta
        Subscriber

        Also Rahul, in reality the lubricant thickness will not be 0.1 mm but 0.1 um (micrometer). 

        In that case, I am worried how will I be able to do meshing of Lubricant in Fluent?

        Is there any tricks or it would not be feasible to do meshing of objects with 0.1 micro-meter (1E-06 meter) thickness?

    • Rahul
      Ansys Employee

      Okay. In that case, try using the gap model in Fluent with Flow modeling as discussed in this tutorial: Reed Valve FSI Co-Simulation with Partial Setup Export from Workbench (Fluent-Mechanical)

      You can find info about gap model in user guide here: 6.8. Controlling Flow in Narrow Gaps for Valves and Pumps

      • ashishkumar.gupta
        Subscriber

         

        Hi Rahul, I tried your suggestions and yet the same problem. I am very confused now. Is it this difficult to set-up FSI?

         

        Also, Which Fluid Flow is recommended for me out of the 4? Any major difference?

         

         

        • ashishkumar.gupta
          Subscriber

          Hi Rahul, any insights on this?

    • Rahul
      Ansys Employee

      Same problem? negative volume error or inner race not moving correctly?

      As you have a very small gap between the race and the ball, it makes the setup difficult. Also, the mesh is moving in that zone, so it must be handled very carefully. 

      This setting specifies the minimum distance to be maintained between the faces of the moving bodies equal to the current gap size. This will prevent the fluid mesh from collapsing by preventing the ball from interpenetrating the race (you can try specifying the offset = gap size).

      Here Fluid flow(CFX) uses the CFX solver

               Fluid Flow(Fluent with Fluent Meshing) uses the Fluent solver and Fluent Meshing mesher is used to mesh the geometry

               Fluid Flow(Fluent) uses the Fluent solver and geometry is meshed with Ansys WB Mesher

               Fluid Flow(Polyflow) is not for case.

      You can use Fluid Flow(Fluent with Fluent Meshing) or Fluid Flow(Fluent).

      • ashishkumar.gupta
        Subscriber

        This problem of the inner race not moving. 

        Ok, I will continue using the Fluid Flow (Fluent). Gap offset i can make 0.1 mm now for my case.

        But what about this issue of my model not working? What might be the problem here?

         

        • ashishkumar.gupta
          Subscriber

          So, it seems the meshes in the Fluent are not letting the IR surface (blue colored in contact with the Fluid) move/deform and also therefore getting multi-layered movement as seen in this picture.

          Any idea how can I correct this mistake? A quick reply would be appreciated please....

    • Rahul
      Ansys Employee

      Again, I would recommend you check the BCs. 
      There should be 3 surfaces (Coupling Interfaces) across which Fluent and Mechanical would exchange the data i.e. inner race, outer race, and ball surface.

      • ashishkumar.gupta
        Subscriber

         

        I already told you that the BCs before and I guess you said it looks fine.

        Mech BC:

         

      • ashishkumar.gupta
        Subscriber

        Hi Rahul, Any update? I have checked my BCs but I do not seem to find any error.

        Please help me out with this!!

    • ashishkumar.gupta
      Subscriber

       

      .

Viewing 24 reply threads
  • You must be logged in to reply to this topic.