Jump to content

ydigit

Moderators
  • Content Count

    432
  • Joined

  • Last visited

  • Days Won

    17

Reputation Activity

  1. Simon Neuner liked a post in a topic by ydigit in Determination of cell reaching solar flux/temperature with AcuSolve   
    However, I did not define as air as a material or "solid". Are these conditions still adequate to simulate the heat exchange (conduction) with an ambient medium or do I have to model air especially?
    It is adequate to simulate without external air, if you use the BCs above. 
     
    And should I have to model a cuboid with air as static fluid, does the program automatically recognize that where there is a solid, there can't be fluid, or do I have to
    define that specifically?
    You have to define explicitly mediums for all volumes. 
     
  2. Simon Neuner liked a post in a topic by ydigit in Determination of cell reaching solar flux/temperature with AcuSolve   
    Short answers for now, as I cannot open the HM right now. 
    1. Yes you can use AcuSolve without fluid. You can define Material_Type= solid for Air. This will essentially create a "solid" with properties of air, however only conduction will take place. 
     
    2. Contact definition between two surface is default Flux = 0.  This means that there is prefect contact between two surfaces and conduction will take place without any thermal resistant (like air gaps or glue).
    If you need to define free convection, then all AcuSolve preProcessors  have options to define those as well. 

  3. Junta liked a post in a topic by ydigit in New Virtual Wind Tunnel 2020 for AcuSolve?   
    The last release of VWT for AcuSolve was with 2018. You download that from connect.altair.com
     
    I am not aware of any method to use VWT 2020 for AcuSolve. 
  4. Junta liked a post in a topic by ydigit in Meshing in AcuConsole vs HW CFD X   
    Correctly mentioned above, that both will very likely give the "same" results. 
     
    Also, AcuSolve is a node based solver. Hence the way these nodes are connected is secondary for the solver, important is to have enough nodes in the region of interest. I am not aware of any setting in AcuConsole to get the "nice" looking mesh. 
  5. EnDF liked a post in a topic by ydigit in Problem with CFD analysis   
    There are two issues with your models. 
     
    1. Scaling. If using acuConsole, you can scale your model by 0.001 in all directions, using Mesh Ops > Transform Coordinates. 

     
    2. You are using mixed topology (Tri3 and Quad4 for Inlet) in AcuConsole for mass_flux inlet. This is not supported by AcuConsole (HyperMesh supports it though).
    https://altairuniversity.com/6266-did-you-know-cfd-tips-tricks/
    You have two options either to use pure Tri3 and Tetra4 mesh for entire domain or to use the BC type Constant velocity as stated in the link. The easiest would be to use split elements in AcuConsole.
     
    Also HyperMesh supports mixed topology. So starting AcuSolve from HM should be OK (I did not check your HM file though). 
     
    Hope this helps. 
  6. EnDF liked a post in a topic by ydigit in Job Launcher in Acusolve doesn't work   
    The AcuSolve Job Launcher should be available in your Start > Program > Altair 2019> AcuSolve folder, like the screenshot below. 
     
    OR even using the Search  for Job Launcher on Windows 10 should reveal it. 
     

  7. Nadzrin liked a post in a topic by ydigit in Frontal area   
    One easy way to the frontal area calculation is import the surface mesh (.Nas or Bdf format) in Virtual Wind Tunnel. It calculates the frontal area automatically. 
     
     
  8. marcyborges liked a post in a topic by ydigit in Mesh 2D   
  9. SuyashS liked a post in a topic by ydigit in result output (mass flow rate) & conservation of mass   
    The easiest way to monitor mass flux is using AcuProbe.

    Surface Output > "surface name" > mass_flux

     

    You could also create a UDF in AcuProbe

    value= Get( "Surface Output - Outlet - mass_flux" ) + Get( "Surface Output - Inlet - mass_flux" )

     

    Note the sign is "plus" in the above snippet, due to sign conventions.

  10. llyle liked a post in a topic by ydigit in (Solved) Automotive Aerodynamic_Wind Tunnel   
    The virtual wind tunnel you show in your screenshot if very small for any reasonable aerodynamics results. You might get colorful picture, but probably no useful interpretation. 
     
     
    Wind tunnel dimensions:
    o 5 car lengths ahead of the car
    o 10 car lengths behind the car
    o 5 car widths on each side of the car
    o 10 car heights above the car
    o i.e., wind tunnel dimensions LWT x WWT x HWT = 16Lc x 11Wc x 11Hc.
    o The mesh farther away from the car will be coarse and increasing the wind tunnel domain would not affect the element count too much but it would still affect the B.Cs if they are too close to the car.

     Boundary Layer details
    o First layer height will depend on the freestream velocity to obtain a maximum y+<100. A first guess would be 0.2mm<y<0.5mm. Ideally y+max<30.
    o For some applications y+max<100 should still give good results
    o growth rate of 1.3
    o number of layers 5
    o all tets in boundary layers
  11. Junta liked a post in a topic by ydigit in (Solved) import solutions   
    Did NIC method work for you? 
    Based on the log file, it seems that command was not correct. It seems an issue with path. What command did you try to run and please copy the content of your .rst file. 
  12. Junta liked a post in a topic by ydigit in (Solved) Automotive Aerodynamic_Wind Tunnel   
    The virtual wind tunnel you show in your screenshot if very small for any reasonable aerodynamics results. You might get colorful picture, but probably no useful interpretation. 
     
     
    Wind tunnel dimensions:
    o 5 car lengths ahead of the car
    o 10 car lengths behind the car
    o 5 car widths on each side of the car
    o 10 car heights above the car
    o i.e., wind tunnel dimensions LWT x WWT x HWT = 16Lc x 11Wc x 11Hc.
    o The mesh farther away from the car will be coarse and increasing the wind tunnel domain would not affect the element count too much but it would still affect the B.Cs if they are too close to the car.

     Boundary Layer details
    o First layer height will depend on the freestream velocity to obtain a maximum y+<100. A first guess would be 0.2mm<y<0.5mm. Ideally y+max<30.
    o For some applications y+max<100 should still give good results
    o growth rate of 1.3
    o number of layers 5
    o all tets in boundary layers
  13. ydigit liked a post in a topic by Junta in Compile the UDF   
    Thank you guys,
    I have found the issue.
    My computer didn't have any C/C++ compiler.
     
    Altair university also has a post about this problem.
     
    Share to others who have faced problem like me : https://altairuniversity.com/11523-acusolve-search-for-installed-compiler-tips-tricks/
  14. mkress liked a post in a topic by ydigit in Solver-Error   
    If the script had a problem, the solver would not have started. It would have ended in AcuPrep stage.
  15. mkress liked a post in a topic by ydigit in Solver-Error   
    Generally I try to avoid elements with small angles <10°. Also check for any really small elements (cannot give an absolute value though). 
     
    Say you average element size is ~10, then search for elements <1 to see which are smallest (sliver) elements. These are typically going to create big trouble for boundary layer meshing. 
    In a way both are related issues. Small angles lead to small elements. 
     
     
  16. Junta liked a post in a topic by ydigit in (Solved) import solutions   
    I guess that Project Solutions is not yet implemented for HM-CFD 2017.3.
     
    What issue did you face with restarting from the steady state case?
     
    An alternative would be to export velcoity and pressure nodal value from steady case (use AcuTrans or AcuOut).
    The files can be read as NIC (Nodal Initial Condition) files in HM2017.3
    See screenshot.
     

     
  17. Junta liked a post in a topic by ydigit in Compile the UDF   
    AcuSolve installation does not have a C/C++ Compiler. So you would need to install a compiler  seprately and then link the compiler to AcuSolve.
     
    Windows system would probably work well with Microsoft Visual Studio.
    Linux systems generally have GCC preinstalled.
  18. Junta liked a post in a topic by ydigit in natural-convection using HM-CFD   
    The AcuConsole model uses periodicity in Z direction and has only one element "thickness" in Z-direction. The model generated in HyperMesh is not using periodicity, and it has two elements in the Z-direction, as seen in the screenshot above. 
     
  19. ydigit liked a post in a topic by Matthieu Pupat in [SOLVED] Meshing problem   
    Meshing errors are usually due to issues in geometry. Here is the process to find which geometry needs to be corrected:
     
    Run Part only for solidification Add cores and run for solidification
    Add risers and run for solidification Add sleeves and run for solidification Add chillers and run for solidification
    Add Running systems and run for solidification Add Gate and run for filling  
     
    By going progressively you can find out which component is creating the issue. Then the issue can come from either bad geometry itself or interface between components i.e. small gaps between components that cause the mesh to fail for the mold.
     
  20. Junta liked a post in a topic by ydigit in (Solved) HM-CFD-AcuSolve transient heat analysis   
    Multiplier function is available for Simple Boundary Conditions. Advanced option needs to be activated first.  
    Could you please explain what issue did you exactly encounter while selecting mulitplier function? 

     
     
     
     
  21. Mrt liked a post in a topic by ydigit in HyperMesh 2D Mesh Import Error in AcuSolve   
    Please import the mesh_intersections.nas in HyperMesh. This will give you an idea of where the problem region is this. As the name tells, there are intersecting elements in this region. 
     
  22. exx077 liked a post in a topic by ydigit in Changing direction of flow in VWT   
    Each AcuSolve installation has this folder, whether Linux or Windows. You can find all PDFs there.
     
  23. TNT6789 liked a post in a topic by ydigit in Temperature Setting in Abaqus   
    If you are talking about Abaqus setup in HyperMesh, then probably HyperMesh forum is more appropriate. 
  24. Rajendra Prasad liked a post in a topic by ydigit in ACUSOLVE   
    CFD solvers in general are mesh dependent. Depending on the mesh (or BL), the solver ends up capturing or ignoring certain phenomenon. Since a lot is happening in BL region and you do not have enough nodes there, solver is unable to simulate those and ends up simplifying that and hence the convergence. 
    In case you solve with BL, solver is capturing more physics and hence the issue with convergence. 
     
    As per your input file, you are solving till 5000 iterations for steady state. Default for AcuSolve is 100 iterations for steady state.  So 5000 is obviously an overkill. If you are not getting converged solution in 200 iterations, then you are unlikely to get converged solution anyways. AcuSolve is FEM based solver and has much more happening in the background per iteration than FVM based solvers. Hence the default 100 and I would recommend typically not going beyond 200 iterations. 
     
    There are bc_warnings being generated. That needs to be checked it those are for fluid nodes or for solid. Solids are ok, fluid nodes need BC assigned to them. 
     
    Not sure why used absolute_temperature_offset         = 353.15        # K
        That would offset all values of by 353.15K. Remove that unless you are sure. 
     
    By specifying, the inlet and outlet temperature, you are creating an overdefined system and any solver is bound to be instable for such a case.
    SIMPLE_BOUNDARY_CONDITION( "GasOutlet" ) {
        mass_flux                           = -0.0166666666667# -60.0 kg/hr
        temperature                         = 0.0           # K
     
    There are lots of issue with your case. It would be best to probably talk to your local Altair support and arrange a webmeeting to clarify all these issues. 
×
×
  • Create New...