https://linen.dev logo
Join Discord
Powered by
# les-ras
  • u
    I got that kind of situation with wrong BCs in general
  • u
    On the other hand, in which layer does your first cell lie?
  • u
    Hope that your first cell does not have y+ in buffer layer
  • t
    uhh im pretty sure its been refined; so i took out layer generation and it seemed to ignore the problem for a bit but it came back now; it manifests as a rapidly increasing bounded omega (and velocity field) even tho the residual (final) are extremely low
  • t
    Honestly, i no longer have any idea where to look on this issue as i think my BC are correct
  • t
    Copy code
    boundaryField
{
    HVAC_inlet
    {
        type                flowRateInletVelocity;
        volumetricFlowRate        0.5663; // m^3/s Mass Flow Rate

    }

    emptyClassroom_back_outlet
    {
        type                flowRateInletVelocity;
    volumetricFlowRate    0.0472;
    }

    door
    {
    type        zeroGradient;
    }
    
    door_outlet
    {
    type        zeroGradient;
    }

    circulation_in
    {
    //type            noSlip;
       type                    flowRateInletVelocity;
       volumetricFlowRate        0.0944; // m^3/s Mass Flow Rate
    }
     
    circulation_out
    {
    // type            noSlip;
        type                flowRateInletVelocity;
        volumetricFlowRate        -0.0944; // m^3/s Mass Flow Rate
    }

    t_in
    {
    // type            noSlip;
    type                flowRateInletVelocity;
        volumetricFlowRate        0.0472; // kg/s Mass Flow Rate
    }

    t_out
    {
    // type            noSlip;
    type                flowRateInletVelocity;
        volumetricFlowRate        -0.0472; // kg/s Mass Flow Rate
    }

    ceiling
    {
        type            noSlip;
    } 

    floor
    {
        type            noSlip;
    }

    walls
    {
        type            noSlip;
    }
}
  • t
    for U and for omega:
  • t
    Copy code
    dimensions      [0 0 -1 0 0 0 0];

internalField   uniform 1e-5;

boundaryField
{
    HVAC_inlet
    {
        type            fixedValue;
        value           $internalField;
    }

    emptyClassroom_back_outlet
    {
        type            zeroGradient;
    }
    ceiling
    {
        type            omegaWallFunction;
    value           $internalField;
    }
    circulation_in
    {
        type            fixedValue;
        value           $internalField;
    }
    circulation_out
    {
        type            zeroGradient;
    }
    door
    {
        type            inletOutlet;
        inletValue      $internalField;
        value           $internalField;
    }
    door_outlet
    {
    type            zeroGradient;
    }
    floor
    {
        type            omegaWallFunction;
    value           $internalField;
    }
    t_in
    {
        type            fixedValue;
        value           $internalField;
    }
    t_out
    {
        type            zeroGradient;
    }
    walls
    {
        type            omegaWallFunction;
    value           $internalField;
    }
}
  • t
    and 0/nut:
  • t
    Copy code
    dimensions      [0 2 -1 0 0 0 0];

internalField   uniform 0;

boundaryField
{
    HVAC_inlet
    {
        type            calculated;
        value           uniform 0;
    }
    emptyClassroom_back_outlet
    {
        type            calculated;
        value           uniform 0;
    }
    ceiling
    {
        type            nutkWallFunction;
        value           uniform 0;
    }
    circulation_in
    {
        type            calculated;
        value           uniform 0;
    }
    circulation_out
    {
        type            calculated;
        value           uniform 0;
    }
    door
    {
        type            calculated;
        value           uniform 0;
    }
    door_outlet
    {
    type            calculated;
        value           uniform 0;
    }
    floor
    {
        type            nutkWallFunction;
        value           uniform 0;
    }
    t_in
    {
        type            calculated;
        value           uniform 0;
    }
    t_out
    {
        type            calculated;
        value           uniform 0;
    }
    walls
    {
        type            nutkWallFunction;
        value           uniform 0;
    }
}
  • u
    If you save time step files, could you visualize and tell where omega goes far beyond? In this way, you may see where the fault is. I still think, the problem is in BCs.
  • u
    And by writing:
    Copy code
    type fixedValue;
value $internalField;
    I think that, you are implicitly applying zeroGradient boundary condition at inlets. This may be problem as well, at least, this is my opinion.
  • t
    Interesting; i tried ur suggestion and it seems to have not changed; however i did also ru na RAS with turbulence turned off (to test my U and p fields) ands it seems those are fine; i also ame getting better bounds on my k altho my omega remains ridicoulously high
  • z
    Couple of notes: - I believe there's a flowRateOutletVelocity patch you can use for outlets. I haven't taken a look at the code to see if it's equivalent to using inlet with a negative value, but might be worth a look - what's the difference between door and door_outlet? If you're expecting inflow through either, then a pressureInletOutletVelocity BC will generally be a better idea than zeroGradient. - What are your pressure boundary conditions? It's important they be appropriate for your velocity conditions. - That seems like quite a low value to initialise omega with, what happens if you change it from 1e-5 to 1? For that matter, what did you set k as? FYI, omega should be root(k)/l, where l is your length scale (could just take something like half the width of your duct, it's not too important)
  • t
    the flowRateOutletVelocity was not working for me, not sure why; the door_outlet is a fan based outlet whreas the door is just a standard door but i will implement the pressure... outlet the door htne; ye so i redid my Omega values to be 100 as starting and it was still erroring; i set k as 0.375 which was what cmae with the PitzDaily case but havent calculated it for my explicit case (with velocity vlaues in between 0-8,9 m/s); the length scale is a little tricky as this is a room so im working iwth about 0.25m per cell and a standard lenght of around 0.5 m (not sure how else i would approximate this)
  • d
    Hi everyone! I'd very much appreciate some help with understanding wall functions 🙂
  • d
  • d
    I am solving some 1D equations for k and omega and I derived an outer and inner problem for it, with matching conditions between them
  • d
    In the picture I was solving the outer problem in (delta, 1-delta), where delta is the thickness of the boundary layer, which I just kept decreasing until my solution (in the picture) no longer changed
  • d
    I'm not getting a very good match with the Fluent data, but I'm more worried about k starting to decrease much sooner
  • d
    I'd very much appreciate if someone could help me understand that 🙂
  • s
    Are you scaling correctly? Are we seeing normalized velocities/fields? Why is k not zero at the wall for the blue line?
  • d
    sorry, just saw this!
  • d
    I was looking at this wrongly. I, of course, need to form a composite solution y_c = y_inner + y_outer - intersection 🙂 Then I assume the composite solution would match the Fluent data (crosses)
  • s
    There are still problems with the solution; mind you
  • k
    i'm finding many articles and forum posts claiming it's OK to use upwind schemes for turbulent quantities. some of my cases are quite sensitive to turbulence (flow separation and stuff), is it really OK to use upwind? currently i'm getting explosions with linearUpwind so i can't really 'compare' the two. i'm using k-omega-SST on a low-Re mesh
  • a
    did you try another scheme except of linearUpwind, e.g. vanLeer? Maybe this does not "explode". My general observation is, that upwind for k and omega is fine, though....
  • s
    What do you mean? The use of upwind-biased schemes for k,epsilon,omega?? Or you refer to div(phi,U)?
  • s
    Are you talking about LES? Or About RAS?
  • k
    RAS, div(phi,omega) and div(phi,k)
1...202122...52Latest