[chegdan]

Hello,

I've been racking my brain trying to figure this one out and its stumped me. I'm using a really complicated 3d Mesh over the domain below:



and it has been decomposed in scotch like so:



All works fine (no errors) with simpeFoam and directMapped patches and I get convergence plots looking like



I'm monitoring point values around the domain in various locations and everything just keeps increasing linearly. Nothing seems to be topping off and evening out. On courser meshes, I do get some flattening of values...but im interested in the fine mesh. I am looking at changing my pRefCell to someplace different...but Im not sure if that will help. I also tried to use cyclic boundary conditions and make a simpleChannelFoam hybrid of channelFoam and simpleFoam with no luck on convergence (in the channel solver, i get pretty crazy gradP within 10 iterations). Any thoughts? Could it be my discretization using linearUpwind? Sorry for the large pictures in the post. Patch summary is :

Time = 2001

Valid fields:
    volScalarField	nut
    volVectorField	U
    volScalarField	k
    volScalarField	p
    volScalarField	epsilon

wall: particles
    scalar		nut		calculated
    scalar		k		zeroGradient
    scalar		p		zeroGradient
    scalar		epsilon		zeroGradient
    vector		U		fixedValue

wall: tube
    scalar		nut		calculated
    scalar		k		zeroGradient
    scalar		p		zeroGradient
    scalar		epsilon		zeroGradient
    vector		U		fixedValue

patch: outlet
    scalar		nut		zeroGradient
    scalar		k		zeroGradient
    scalar		p		fixedValue
    scalar		epsilon		zeroGradient
    vector		U		zeroGradient

symmetryPlane: zyplane
    scalar		nut		symmetryPlane
    scalar		k		symmetryPlane
    scalar		p		symmetryPlane
    scalar		epsilon		symmetryPlane
    vector		U		symmetryPlane

symmetryPlane: zxplane
    scalar		nut		symmetryPlane
    scalar		k		symmetryPlane
    scalar		p		symmetryPlane
    scalar		epsilon		symmetryPlane
    vector		U		symmetryPlane

directMappedPatch: inlet
    scalar		nut		zeroGradient
    scalar		k		directMapped
    scalar		p		zeroGradient
    scalar		epsilon		directMapped
    vector		U		directMapped

End

[chegdan]

Just thought I would provide an answer in case anyone else is looking for an answer. I ended up using an inletOutlet condition for velocity at the exit with

    outlet
    {
        type            inletOutlet;
        inletValue      uniform ( 1e-20 1e-20 1e-20 );
        value           uniform ( 0 0 1 );
    }

and I used a fixedMeanValue boundary condition for pressure that can be found at (http://www.cfd-online.com/Forums/ope...condition.html). The condition can be used with

    outlet
    {
        type            fixedMeanValue;
        value           uniform 1;
        meanValue       1e-20;
    }

using this with limited schemes and celllimited gradient schemes on a polyhedral mesh....the rise in pressure eventually leveled out. I monitored the pressure drop between the inlet and outlet and once that was constant, 10 probe points were constant, and the residuals dropped to a level lower than 1e-4 it was an acceptable answer.