Jump to content


  • Content Count

  • Joined

  • Last visited

  • Days Won


acupro last won the day on September 9

acupro had the most liked content!


About acupro

  • Rank
    Super User

Profile Information

  • Are you University user?

Recent Profile Visitors

2615 profile views
  1. With thermal shells there is no physical surface between layers. The layers are just defined to compute the overall thermal resistance for the entire set. Conduction is solved in all directions - but again it would be using the overall thermal resistance - not a layer-by-layer conduction model.
  2. This sounds like a perfect candidate for multi-layered thermal shells. These are basically zero-thickness volume elements where the number and properties of layers are specified in the element definition.
  3. Correct - the Ideal Gas density model would be the other choice. Suggest you start with constant density and establish a physical flowfield based on absolute temperature and pressure - then restart with Ideal Gas density. I would also suggest running with S-A turbulence model. You could check results differences with realizable k-epsilon.
  4. Are you trying to run AcuSolve or ultraFluidX (uFX) from VWT? If uFX, that requires high-end GPUs to run, it doesn't run on CPUs. Note - I'm not a user of uFX, so I won't be much help beyond that...
  5. Sorry - I'm not an OptiStruct user. You may need to ask that question in the OptiStruct forum. Describe the two load cases, and indicate you don't want to write output from the first load case (Linear Static) to the OP2 file. You only want output from the Frequency/Modal load case written to the OP2 file. Probably include the .hm file again, too.
  6. It may be that the OP2 file for Eigenmode Manager can only contain output from the modal analysis subcase - so maybe make sure there is no output requested for the linear static subcase.
  7. Does the result get any better at all? If you look at both air and oil particles together - does the void decrease in size? What if you increase ref_vel_factor even more - say to 6.0? Is there additional decrease in the size of the 'void'? Are many fluid particles being removed - 'escaping' through the walls and/or moving inside the gears? The fluid particles 'feel' increased pressure due to the teeth being close together. Sometimes they get moving so fast that they pass through the walls - especially with not much space between the tooth contact region and the outer walls. You mentioned 6 mm between the gears and the housing (I'm assuming from the larger faces of the gears). How much space between the gear teeth and the 'curved' housing walls? What is the distance across a tooth tip? How many layers of particles do you have for the housing and the gears themselves?
  8. I would expect in this case that the reference velocity based purely on the rotating bodies (X 1.5 by default) is not enough due to likely large regions of acceleration with the walls so close. If you go really high, say ref_vel_factor 3.0 instead of default 1.5, how much of a difference does it make?
  9. You are correct in that as the gear teeth come together, the fluid particles feel increased pressure and move away from those areas. It will typically move laterally away from the contact areas. With the typical resolution (dx) in the mm range - looks like 0.7 mm in your case - nFX will not capture the flow between the teeth themselves, as this is of a much smaller length scale. There should also be a small number of these hot/fast particles compared to the total. You should still be able to see fluid on the left side of the gear contact area, but it would most likely then be air. Do you see the air phase there instead? You can see if reducing the time step makes a difference. There are a few ways you could do this: 1. Increase ref_vel_factor 2. Decrease dt_factor 3. Find the computed reference velocity in the output file itself, then set ref_vel to a higher value than that in the 'domain' section 4. Find the computed time step in the output file, and specify it manually to a lower value - this is probably the least desirable manner... It looks like there is not much clearance between the gears and the housing in the screenshot provided. If you look from the other direction, is the housing also very close to the gears? If yes, where the gears are very tight to the housing on all sides, it may take a fairly low time step to get good stability as there is not much volume for the fluids to stabilize much.
  10. Just curious - which AcuSolve version are you using, and on which OS?
  11. I would suggest two options. 1. Install the current version (2020 instead of 2017) and use SURFACE_SET commands to define the different surfaces. Then in SURFACE_OUTPUT you can list the various surface sets in one surface output to use with acuLiftDrag. 2. Convert the entire mesh to tets in order to obtain just one surface for the plane with all trias and attached only to tets.
  12. acuGetNodeSubset should be in the /script/ directory of the installation.
  13. The impact of the 'reference length' for a standard gearbox is probably not that great when it comes to computing the time step to be used. The reference velocity will likely have a larger impact on the time step. I would probably start with the initial oil 'column height'.
  14. If there are areas/regions where you expect high accelerations, you may want to specify the reference velocity manually rather than letting nFX decide it based on the tip velocity of the gears. If you also look at the number of air particles with density 1.1 and/or 1.2 X the reference density, that will help to indicate if the higher velocity represents a serious issue. If there are only a percent or two of the total air particles with high velocity/density, then you may be OK. There will always be higher velocity in the air than in the oil, probably part of the numerical approach - SPH.
  15. If you're just doing single-phase flow, I would probably use slip on the surface attached to the fluid volume. Then you still may need to add Element Boundary Condition of type heat-flux to assign the convective HTC.
  • Create New...