The geometry is highly complicated and contains a hundred tiny passages and complicated curved shapes. Automatic generation with SALOME is all I could manage. How do you suggest I can improve the mesh quality? Does changing to Hexahedral help?

Mesh elements are currently tetrahedral

Yes hexahedra might help

Thanks. I will try that

Are initial or boundary conditions the only causes of epsilon blowup?

My U:

cpp
    Inlet
    {
        type            fixedValue;
        value           uniform (0 0 -0.1);
    }

    Outlet
    {
        type            zeroGradient;     
    }

    Walls
    {
        type            fixedValue;
        value           uniform (0 0 0);        
    }

My p:

Inlet
    {
        type            zeroGradient;
    }

    Outlet
    {
        type            fixedValue;
        value           uniform 0;
    }

    Walls
    {
        type            zeroGradient;
    }

Output:

time step continuity errors : sum local = 4.91717e-09, global = -6.25394e-12, cumulative = -8.98157e-09
smoothSolver:  Solving for epsilon, Initial residual = 0.999856, Final residual = 9.45248e-06, No Iterations 153
bounding epsilon, min: -7729.8 max: 4.03935e+09 average: 602658
smoothSolver:  Solving for k, Initial residual = 1, Final residual = 6.87711e-07, No Iterations 2
ExecutionTime = 53.2955 s  ClockTime = 54 s

time step continuity errors : sum local = 3215.73, global = 0.0387011, cumulative = 160751
smoothSolver:  Solving for epsilon, Initial residual = 1, Final residual = 0.149008, No Iterations 1000
bounding epsilon, min: -3.60506e+16 max: 1.0231e+24 average: 1.36127e+20
smoothSolver:  Solving for k, Initial residual = 1, Final residual = 0.479932, No Iterations 1000
bounding k, min: -1.23483e+16 max: 3.766e+18 average: 2.34329e+13
ExecutionTime = 172.315 s  ClockTime = 174 s

time step continuity errors : sum local = 7.31619e+36, global = 3.64242e+36, cumulative = 1.97806e+36
smoothSolver:  Solving for epsilon, Initial residual = 1, Final residual = 0.875552, No Iterations 1000
bounding epsilon, min: -5.33999e+80 max: 4.72258e+89 average: 1.3096e+84
smoothSolver:  Solving for k, Initial residual = 1.42391e-19, Final residual = 1.42391e-19, No Iterations 0
ExecutionTime = 258.096 s  ClockTime = 260 s

Both Time step continuity errors and epsilon blowup

Even on a good mesh?

I could not get Salome to mesh Hexahedral elements for the geometry

The mesh remains as bad as before

So then that is a likely cause of your error

What is your geometry like?

A flattish cylinder with a hundred-ish tiny tubules and some thin passages. Unfortunately I cannot share the design itself due to IP confidentiality

That's a shame coz otaolafr is the wizard of meshing funky designs with tiny tubes on Salome 😂

Well I was looking for a challenge 😅😂 to see what I could do

I assume you don't want to.use snappy or cfmesh.... Try body fitting algo in Salome to see what it achieves.

The geometry is so complicated that it probably cannot be done manually. By body fitting, is it a sort of meshing algorithm? Does it have tetrahedral or hexahedral elements?

Thanks for your time

Mmm I can not say at all from your description if it could be done manually... (in any case the master it is not me by any means it is kandelabr 😅). No body fitting is a 3d algo of Salome similar to snappy.

Oh. I have a 3D model (Solidworks, or even STEP) for the above mentioned geometry... Generating this from primitives is probably going to be an enormous waste of time. I have another component (a pump) which I would agree can be generated parametrically... Do I have any better automatic meshing alternatives?

Body fitting is an automatic algo. Your options in automatic world are Tetra you have a hundred options Hexa or mostly snappy cfmesh body fitting in salome Poly cfmesh, polydualmesh or Salome with a plugging to transform tet to poly.

Why not snappy? A few levels of refinement and you'll be there 😄

I simplified the mesh to a solid region like below. The Inlet is the tiny top circular face which looks like a bottle's mouth and the outlet is the bottom tiny circular face which also looks like a bottle's mouth. The Z direction points from outlet to the inlet (bottom to top in the diagram)

Velocity(U) BCs:

boundaryField
{
    Inlet
    {
        type            fixedValue;
        value           uniform (0 0 -0.1);
    }

    Outlet
    {
        type            zeroGradient;     
    }

    Walls
    {
        type            noSlip;
    }   
}

Pressure(P) BCs:

boundaryField
{
    Inlet
    {
        type            zeroGradient;
    }

    Outlet
    {
        type            fixedValue;
        value           uniform 1.013e5;
    }

    Walls
    {
        type            zeroGradient;
    }
}

You can see below that

k

and

epsilon

are blowing up: ``` GAMG: Solving for p, Initial residual = 0.284828, Final residual = 7.15948e-06, No Iterations 1 GAMG: Solving for p, Initial residual = 0.670365, Final residual = 7.15309e-06, No Iterations 1 GAMG: Solving for p, Initial residual = 0.302612, Final residual = 6.85635e-07, No Iterations 4 time step continuity errors : sum local = 5.25594e+83, global = 2.98324e+74, cumulative = 3.91295e+74 smoothSolver: Solving for epsilon, Initial residual = 1, Final residual = 9.96455e-06, No Iterations 580 bounding epsilon, min: -1.43447e+79 max: 5.13934e+172 average: 4.54879e+169 smoothSolver: Solving for k, Initial residual = 1, Final residual = 8.93145e-08, No Iterations 1 bounding k, min: -4.2093e+55 max: 3.69306e+127 average: 2.28031e+124 ExecutionTime = 0.84259 s ClockTime = 1 s Time = 0.00027s Courant Number mean: 3.00317e+85 max: 7.97363e+89 smoothSolver: Solving for Ux, Initial residual = 1, Final residual = 9.5775e-06, No Iterations 28 smoothSolver: Solving for Uy, Initial residual = 1, Final residual = 8.80905e-06, No Iterations 38 smoothSolver: Solving for Uz, Initial residual = 1, Final residual = 9.59237e-06, No Iterations 36 #0 Foam::error::printStack(Foam::Ostream&) at ??:? #1 Foam::sigFpe::sigHandler(int) at ??:? #2 ? in "/usr/lib/libc.so.6"```What am I doing wrong?

Can you post a image from front of it? Not angled

Furthermore it would affect a lot if I add fillers between the small cylinders and the main body?