Dynamic Meshing using User Defined Function for Gearbox Sloshing analysis   (ANSYS Fluent, Skill Lync, Medium)
Objective: Aim: To learn simulation of dynamic mesh by using user defined function. Introduction: Many of the CFD applications require a transient study on the object. Usually, the object itself moves from its position in time. To simulate this,…
- Houssam Eddine Ben Salahhello please i have little problem about understanding dynamic mesh parameter , can you help me please .
- Raj RathodYes. Tell me...
- Houssam Eddine Ben Salahthank you Raj for your reply , i didn't understand how you calculate total time step , why 1 tooth shift will require 0.000682 and why u divided in 10 divisions ? can you explain it to me please
- Raj Rathod1) About 1 tooth shift time :- I calculated the time needed for 1 complete rotation of gear. Since there are 46 teeth on the gear, each gear tooth tip and crest would require 0.0314/46 = 0.000682 seconds. 2) About total number of time steps :- Since, I need to simulate 5 complete rotations of gear and each gear requires 0.0314 sec to rotate once, total time would be 0.0314 x 5 = 0.15708 sec. Now, I chose the time step as 5e-5 sec. Therefore, total time steps required would be 0.15708/5e-5 = 3140 time steps 3) About 10 divisions :- There are 2 regions where significant mesh motion will be there, in gap between teeth and outside the gear on its tips. The remesh option would create mesh in the gap because the boundary will surely move, thus destroying and creating cells simultaneously. So, I need to figure out mesh settings near the tip. I have considered an extreme case where only 1 cell next to tip moves. In simulation it would not be possible, because the spring smoothing would stretch the next cells also. So, this assumption acts a safety factor. After knowing how much time 1 tooth will take to move, I assumed that splitting this movement in 10 time steps would do the job. I calculated the time step required for capturing motion of 1/10 of teeth and gap. I assumed the size of triangles as 0.75 mm. Since these settings gave me skewness under control for 2 time steps, which I thought can be considered as good. So, I went forward with resolving 1/10 of tooth and gap. Had it been that my skewness was going out of control in 1 time step only, I would have considered 20 divisions of teeth and gap or considered reducing mesh size. If someone would have an academic version of ANSYS, he might not increase the mesh cell count, so in that case refine the motion in more parts would be the solution. So in one sentence :- I have assumed 10 divisions and I was lucky that it worked out in one go, otherwise I would have taken 20 divisions. I was sure about it because I took the previously mentioned safety factor and also I rounded off the time step to a lower value. Hope this answers all your doubts regarding dynamic mesh settings...
- Raj Rathod<b><i>Please ignore the previous comment I wrote. The whole comment got published without new lines even when I had them. I have this new comment where I have inserted HTML tags to enters new lines and bold letters. Now you can read it</b></i> <b>1) About 1 tooth shift time :-</b><br> I calculated the time needed for 1 complete rotation of gear. Since there are 46 teeth on the gear, each gear tooth tip and crest would require 0.0314/46 = 0.000682 seconds. <br><br> <b>2) About total number of time steps :-</b><br> Since, I need to simulate 5 complete rotations of gear and each gear requires 0.0314 sec to rotate once, total time would be 0.0314 x 5 = 0.15708 sec. Now, I chose the time step as 5e-5 sec. Therefore, total time steps required would be 0.15708/5e-5 = 3140 time steps <br><br> <b>3) About 10 divisions :-</b><br> There are 2 regions where significant mesh motion will be there, in gap between teeth and outside the gear on its tips. The remesh option would create mesh in the gap because the boundary will surely move, thus destroying and creating cells simultaneously. <br><br> So, I need to figure out mesh settings near the tip. I have considered an extreme case where only 1 cell next to tip moves. In simulation it would not be possible, because the spring smoothing would stretch the next cells also. So, this assumption acts a safety factor. <br><br> After knowing how much time 1 tooth will take to move, I assumed that splitting this movement in 10 time steps would do the job. I calculated the time step required for capturing motion of 1/10 of teeth and gap. <br><br> I assumed the size of triangles as 0.75 mm. Since these settings gave me skewness under control for 2 time steps, which I thought can be considered as good. So, I went forward with resolving 1/10 of tooth and gap. <br><br> Had it been that my skewness was going out of control in 1 time step only, I would have considered 20 divisions of teeth and gap or considered reducing mesh size. If someone would have an academic version of ANSYS, he might not increase the mesh cell count, so in that case refine the motion in more parts would be the solution. <br><br> <b>So in one sentence</b> :- I have assumed 10 divisions and I was lucky that it worked out in one go, otherwise I would have taken 20 divisions. <br><br> I was sure about it because I took the previously mentioned safety factor and also I rounded off the time step to a lower value. <br><br> Hope this answers all your doubts regarding dynamic mesh settings...
- Houssam Eddine Ben SalahThank you for your reply i have understand what you did about total time step but about 10 divisions and skewness honestly i didn't understand you well , because i always getting negative cell i dont know to prevent that , please last can you explain that to me , sorry
- Raj RathodIf you are getting error of negative cell volume, it means that you have moved the boundary so fast in 1 time step that cells actually got destroyed. To understand this statement look at this example.<br><br> <b>Example:-</b> A boundary is moving and it is deforming a cell<br> <b>Case 1:</b> The moving boundary has moved a small distance in a time step. It moved by such a small distance that it deformed the cell and the new skewness value was below our specified limit. So the remesh program would continue for next time step without remeshing the cell. If in second time step, the cell gets more deformed and it goes beyond our remesh criteria, then remeshing would happen. <br><br> <b>Case 2:</b> The moving boundary has moved a medium distance in a single time step. It deformed the cell so much that its skewness got out of our specified limit in a single time step. So the remesh program would remesh that particular cell. <br><br> <b>Case 3:</b> The moving boundary has moved a large distance in single time step. It moved so far that the entire cell got destroyed in the process. The remesh program would see this as an error and give a negative cell volume message. <br><br> So if you are getting this error, try following tips:<br> 1) Use smaller time steps.<br> 2) If mesh is much more refined than required, try coarsening the mesh.<br> 3) Try using structured mesh instead of unstructured tetras. <br><br>You can have a look at this video: https://youtu.be/PqBDEsOEvBk
- Akash Kodarkarhappy to check