Jump to content

acupro

Members
  • Content Count

    238
  • Joined

  • Last visited

  • Days Won

    13

Reputation Activity

  1. Nadzrin liked a post in a topic by acupro in Shear Flow   
    1.  For the inlet NBC, you may want to use cubic-spline rather than linear.  It depends on if you want linear or cubic interpolation between your given data points.
    2.  For the NIC, you have two choices, since piecewise linear is not available in AcuConsole.  You can either use a tool like Excel to determine the equation for the velocity based on Z - and use 'Script' in AcuConsole.  Or you'll need to edit the input file once written, and use piecewise-linear (or cubic spline), as described in the Command Reference Manual under Nodal Data Commands > Nodal Initial Condition, which will then be similar to the NBC.
  2. whj liked a post in a topic by acupro in How to set the moment center in acusolve   
    If you issue the command:     acuTrans  -h      you'll see one of the options is  -mc  for moment_center:
     
    acuTrans -h | grep center
    acuTrans:   -mc <str>     center about which moment is computed
    acuTrans:                   moment_center= 0,0,0 [default]
     
    If you're using Linux, you can add    -mc  0.0,2.5,-2.3   to the acuTrans command directly.  If you're using Windows, you'll need to add the option to your local Acusim.cnf file - as the Windows parser typically doesn't work well with the commas.  (This also works for Linux if you don't want to type it as part of the acuTrans command.)  For example, in the Acusim.cnf file add:
     
    moment_center  =  0.0,2.5,-2.3
  3. Duc_Hoang liked a post in a topic by acupro in Calculation of Pressure rise   
    My guess would be that static pressure would be typical, as that is simpler to measure, and doesn't obstruct the flow.  Simply drill a hole in the side of the inlet and outlet pipes and put a tube there - flush to the inner/wetted surface.  A total pressure measurement requires a pitot probe - where you install a device that protrudes into the flow and bends such that it points directly into the oncoming flow.  I assume there's no harm in reporting both, but I would guess static pressure rise is typically reported.
  4. Duc_Hoang liked a post in a topic by acupro in Calculation of Pressure rise   
    The equation given in the tutorial compares static pressure.  If you want to compare total pressure, then your expression would be correct.  It depends on what you want to compare - static or total.
  5. Livil Lyle liked a post in a topic by acupro in error occur executing AcuSolve   
    Based on my quick look at the database, you do not have the interface surface attached to the Rotor volume - only that for the Fixed/main volume.  Right-click 'Air-rotor' under Volumes and select Mesh Op. > Find Missing Surfaces.  This will generate the other set of surfaces, attached to the Rotor Volume.  It appears you already have two sets of nodes there, but the surface definition was missing.  This new surface set should have the same treatment as 'interface' > deactivate Simple BC, and activate Interface Surface with 0 for gap and gap_factor.
     
    If you know or can estimate the angular velocity, it's simpler to use rotation mesh motion rather than rigid body.  It's difficult to estimate the resistance to rotation due to the gearbox, etc, so the angular velocity / rotation from Rigid Body motion is probably going to be higher than the actual would be.
  6. Duc_Hoang liked a post in a topic by acupro in error occur executing AcuSolve   
    Based on my quick look at the database, you do not have the interface surface attached to the Rotor volume - only that for the Fixed/main volume.  Right-click 'Air-rotor' under Volumes and select Mesh Op. > Find Missing Surfaces.  This will generate the other set of surfaces, attached to the Rotor Volume.  It appears you already have two sets of nodes there, but the surface definition was missing.  This new surface set should have the same treatment as 'interface' > deactivate Simple BC, and activate Interface Surface with 0 for gap and gap_factor.
     
    If you know or can estimate the angular velocity, it's simpler to use rotation mesh motion rather than rigid body.  It's difficult to estimate the resistance to rotation due to the gearbox, etc, so the angular velocity / rotation from Rigid Body motion is probably going to be higher than the actual would be.
  7. Duc_Hoang liked a post in a topic by acupro in Vortice Structure in AcuFieldView   
    The image does represent Q Criterion, colored by something like velocity magnitude or pressure.  You should create in isosurface of Q Criterion, then select a value for Q Criterion that gives you the 'look' you want.  In the function calculator, you would put the velocity function - the vector - inside the parenthesis.
  8. Duc_Hoang liked a post in a topic by acupro in Acusolve :Compile for UDF   
    When the user function is compiled, it includes the libraries necessary for the type of message passing to be used by the solver itself.  If you issue
    acuMakeDll  -h
    you'll see the option '-mp' with the default being  impi.  So if you're going to use impi at the solver level, the compiled library will be correct.  However, in the cases you show above, the message passing used by the solver is openmp, thus the user library needs to be recompiled.  The second system doesn't have the 64-bit compiler, so it fails.  If you want to compile for openmp to be used on the second system you would add
    -mp  openmp
    to the acuMakeDll command you used for the first attempt.  Then the compiled dll should be usable by the solver on both systems, without recompiling.
  9. marcyborges liked a post in a topic by acupro in Mesh 2D   
    You will need to extrude the surface mesh to one volume element thick to run in AcuSolve.  There are a few tutorials in the AcuSolve Help system covering flow over airfoils, using various turbulence model approaches, with and without transition models.
  10. Duc_Hoang liked a post in a topic by acupro in (Solved) Turbulence model and Y+ value   
    You'll achieve better accuracy with lower Y+ values - at the cost of more mesh and longer simulation runtime.  For example, you'll typically get a more accurate solution with Y+ = 20 than you would with Y+ = 100.  However, maybe the level of accuracy at Y+ = 100 is enough for your purposes.  That's where it becomes 'engineering' instead of just running a simulation.  You can run a mesh sensitivity study to see how the solution of interest changes as you modify the meshing parameters.
  11. StefanoColosio liked a post in a topic by acupro in AcuSolve meshing problems   
    What is the source of the CAD - original software used to create the geometry, and the format exported from the CAD software and imported to AcuConsole?  The initial warning mentions 'zero length edge'.  Please share the CAD model if you are able - native geometry model plus the model imported to AcuConsole - and maybe even the AcuConsole database (.acs file) to check mesh settings.
  12. StefanoColosio liked a post in a topic by acupro in AcuSolve meshing problems   
    The issues I am seeing are in the locations where the tubes 'kink' and the surface edges meet together with essentially zero angle and zero space between the surface edges.  With the current geometry, you'll probably need to create the mesh in HyperMesh, and toggle those edges so they're not meshed explicitly.  Otherwise, if you can modify the geometry such that those surface edges don't come together like that, then AcuConsole may work better.
    If you right-click 'Model' and select 'Query', add a row under the 'Point' tab and input the coordinates.  You may need to increase the point size and toggle visibility of things to determine where the point is - but this can help you locate problem areas.  You can also view surfaces/volumes in Solid & Outline mode so you can see where the surface edges are.
  13. whj liked a post in a topic by acupro in Compile the UDF   
    Looks like you should be good to go, since the dll was created.
  14. whj liked a post in a topic by acupro in Compile the UDF   
    You need to go into the directory where the C code user-function is on your machine.
     
    The compiler itself as you've shown in the Window above will not be aware of the libraries in the AcuSolve installation.  But when you are in the AcuSolve Command Prompt, the appropriate paths are set (PATH, LD_LIBRARY_PATH, etc), and the necessary libraries are available to acuMakeDll.
  15. whj liked a post in a topic by acupro in ACUSOLVE UDF   
    Already in other thread...
  16. whj liked a post in a topic by acupro in ACUSOLVE UDF   
    If you're issuing the acuMakeDll command from the AcuSolve Command Prompt, the libraries should all be readily available.  Please describe the process you used, and share any images of the window showing the errors and your process.
  17. whj liked a post in a topic by acupro in Compile the UDF   
    Try just typing
     
    acuMakeDll
     
    and return, without adding "-src gravity.c" - and see which errors are returned.  As mentioned, you do need to have a compiler installed.  It seems Visual Studio Community is the latest free version from Microsoft.
  18. mkress liked a post in a topic by acupro in Kinetic Energy   
    If we assume you have an element set named "flow" - like
     
    ELEMENT_SET( "flow" ) {
    ...
    ...
    }
     
    then
    ELEMENT_OUTPUT( "sample" ) { element_set = "flow" type = all integrated_output_frequency = 5 integrated_output_time_interval = 0.0 num_user_output = 0 } will give you integrated information for that element set every five time steps.  One of the integrated quantities will be turbulent-kinetic-energy.  But I'm not sure how one would integrate total kinetic energy.  Does you other solver give the equation used for what is reported as kinetic energy?
     
  19. alpha liked a post in a topic by acupro in How to run Acusolve on Linux commands?   
    You'll be better off using the   acuSub   command to submit through PBS.  It will create the script for you and submit the job.  See
     
    acuSub  -h
     
    for usage and options.  Again, you'll need to source the appropriate acusim.sh script first, in order to use acuSub.
     
    In your script, it looks like you want to use 24 cores from a single compute node.  (select=1:ncpus=24).  You don't need to include anything about mem.  Then in your acuRun command, you don't need quotes around the name of the .inp file (and I suggest you not start the name of the .inp file with a number), then you have -nt 14.
     
    Let's say your input file is test_run1.inp instead.  What about:
     
    $ALTAIR_HOME/ Acusolve/Linux64/bin/acuRun  -inp  test_run1.inp  -np  24  -nt  _auto  -do  all  -pbs_attach  -pbs
     
    Still - I suggest you try acuSub, and then look at the submit script it generates.
     
    acuSub  -pb  test_run1  -np  24  -ppn  24  -sched  pbspro  -pbs_attach
     
    again, assuming your inp file is test_run1.inp, you want to use 24 physical cores on a single compute node, and you're using the licensed pbs
     
  20. alpha liked a post in a topic by acupro in How to run Acusolve on Linux commands?   
    Right.  You need to establish the AcuSolve environment.  It looks like you're running bash.  As explained in the HyperWorks Installagion Guide, you should include the following in your ~/.profile file
     
    source  $INSTALL_DIR/altair/acusolve/linux64/script/acusim.sh
     
    $INSTALL_DIR is the chose installation directory.  Then you can issue the acuRun command with options from your problem directory.  (You can use     which  acuRun     to make sure you've correctly established the environment.)  You can also add that line to set the ALTAIR_LICENSE_PATH environment variable to your ~/.profile file.
     
    We typically use the local (in the problem directory) Acusim.cnf file to specify the problem name and other options.
     
  21. mkress liked a post in a topic by acupro in Animate water in AcuFieldView   
    I think you'll create an isosurface of something like 'levelset', then use a value of 0.5.  It will be something along those lines.
  22. mkress liked a post in a topic by acupro in Error import model   
    If you're doing geometry manipulations in HM, it's probably better to do at least the surface meshing there, then export Fluent/Nastran mesh to import to AcuConsole for the volume meshing.  Or you could do the entire surface/volume meshing in HM.  The parasolid export after manipulation in HM is not that robust.  What is the source of the original CAD?  Or - is it built entirely in HM?
    Also - that image above indicates a size of 1.7 km in X.  Is that correct?
  23. Duc_Hoang liked a post in a topic by acupro in Compile the UDF   
    Try just typing
     
    acuMakeDll
     
    and return, without adding "-src gravity.c" - and see which errors are returned.  As mentioned, you do need to have a compiler installed.  It seems Visual Studio Community is the latest free version from Microsoft.
  24. mkress liked a post in a topic by acupro in Error import model   
    That message "There are zero region sets in the model" typically means there are no 'solid' bodies represented in the parasolid file.  If the parasolid file definitely has 'solids' represented (which could also be the fluid volume, but from a solid modeler), there could be an issue with the CAD.  If the file only has surfaces, you'll need to enable the 'Fill Voids' option on the import geometry panel.  This will then create the volumes which are enclosed by surfaces, during the import process.
  25. Duc_Hoang liked a post in a topic by acupro in How to export surface pressure from AcuFieldView or AcuSolve to FEA solvers   
    If you first use 'cd' to go into the directory where you placed the acuGetNodeSubset python script itself, you should be able to use
     
    acuPython  acuGetNodeSubset  –h
     
    in that directory, rather than issuing it from the 'Documents' directory as in your snapshot.
×
×
  • Create New...